Time and Space

Time and Space

Second edition
Barry Dainton
acumen
Barry Dainton, 2001, 2010
Second edition 2010
This book is copyright under the Berne Convention.
No reproduction without permission.
All rights reserved.
First published in 2001 by Acumen
Reprinted 2009
Acumen Publishing Limited
4 Saddler Street
Durham
DH1 3NP
www.acumenpublishing.co.uk
ISBN: 978- 1-84465-190-0 (hardcover)
ISBN: 978- 1- 84465-191-7 (paperback)
British Library Cataloguing- in- Publication Data
A catalogue record for this book is available from the British Library.
Printed in the UK by the MPG Books Group.
For Gwynneth
Contents
Preface to the second edition ix
Preface to the first edition xi
1 Preliminaries 1
1.1 Ontology: the existence of space and time 1
1.2 Questions of structure 3
1.3 Physics and metaphysics 4
1.4 Time: the great divide 6
1.5 Two frameworks 10
1.6 Matters terminological 11
2 McTaggart on times unreality 13
2.1 Could time be unreal? 13
2.2 Change as the essence of time 13
2.3 McTaggarts A- paradox 15
2.4 Other routes to the same place 17
2.5 The nature of A- properties 18
2.6 The overdetermination problem 20
2.7 Consequences 25
3 The Block universe 27
3.1 Time without passage 27
3.2 Passage and experience 28
3.3 A- truth in a B- world 31
3.4 Another A- paradox 34
3.5 The indispensability of the A- framework 35
3.6 Questions of attitude 35
3.7 B- theories of change 38
3.8 Emergent time 41
4 Asymmetries within time 44
4.1 The direction of time 44
4.2 Content- asymmetries: a fuller picture 45
4.3 Entropy 47
4.4 The causal route 51
4.5 Causation in question 51
vi CONTENTS
4.6 Time in reverse 53
4.7 Fundamental forks 55
5 Tensed time 63
5.1 Tense versus dynamism 63
5.2 Taking tense seriously 63
5.3 McTaggart revisited 65
5.4 Is tense enough? 65
6 Dynamic time 68
6.1 The Growing Block 68
6.2 Overdetermination 70
6.3 Dynamism without tense 72
6.4 The thinning tree 73
6.5 How can a block grow? 75
6.6 The eternal past 79
6.7 The varieties of Presentism 81
6.8 Solipsistic Presentism 83
6.9 Many- Worlds Presentism 84
6.10 Dynamic Presentism 87
6.11 Compound Presentism 95
7 Time and consciousness 103
7.1 The micro- phenomenology of time 103
7.2 Memory based accounts 105
7.3 The pulse theory 106
7.4 Awareness and overlap 107
7.5 The two- dimensional model 109
7.6 The overlap theory 112
7.7 The phenomenal arrow 116
7.8 Further consequences 117
8 Time travel 121
8.1 Questions of possibility and paradox 121
8.2 Misconceptions and multidimensions 122
8.3 Self- defeating loops 127
8.4 Global consistency constraints 129
8.5 Bilking 131
8.6 Quantum retroaction 133
8.7 The inexplicable 136
8.8 Voyaging in dynamic time 138
8.9 Real time 141
9 Conceptions of void 145
9.1 Space as void 145
9.2 The unseen constrainer 147
9.3 Connection in question 149
CONTENTS vi i
9.4 Substantivalism: a closer look 151
9.5 Relationism: a closer look 153
9.6 Two concepts of distance 158
9.7 Two conceptions of motion 161
9.8 Matters terminological 162
10 Space: the classical debate 164
10.1 The last of the magicians 164
10.2 Galileo 167
10.3 Descartes 172
10.4 Leibniz 175
10.5 The argument from indiscernibility 177
10.6 The argument from sufficient reason 178
10.7 The methodological argument 180
11 Absolute motion 182
11.1 Inertial motion 182
11.2 The argument for real inertial motions 182
11.3 The argument from inertial effects 185
11.4 Stalemate? 188
11.5 The Leibnizian response 189
11.6 The Machian response 190
11.7 The Sklar response 192
12 Motion in spacetime 194
12.1 Newtonian spacetime 194
12.2 Neo- Newtonian spacetime 197
12.3 The only reasonable view? 199
12.4 A threat vanquished 201
12.5 The charge of explanatory impotence 201
12.6 A rebuttal 203
12.7 Newtonian spacetime relationism 205
12.8 Neo- Newtonian spacetime relationism 207
12.9 Relationism redux 209
13 Curved space 213
13.1 New angles on old problems 213
13.2 Flat and curved spaces 213
13.3 The fifth postulate 218
13.4 Intrinsic curvature 222
13.5 Topology 225
13.6 Conventionalism 228
13.7 Realism versus anti- realism 231
14 Tangible space 233
14.1 Manifestations of curvature 233
14.2 The detachment thesis 235
vi i i CONTENTS
14.3 The explanatory challenge 237
14.4 A solitary hand 240
14.5 Global structures 241
15 Spatial anti- realism 245
15.1 Foster on matter and space 245
15.2 The intrinsic and the inscrutable 246
15.3 Modes of deviancy 250
15.4 Intrinsic versus functional geometry 254
15.5 The nomological thesis 255
15.6 Nomological contingency 258
15.7 Realism rejected 259
15.8 Geometrical pluralism 262
16 Zeno and the continuum I 267
16.1 Motion and the continuum 267
16.2 Numbering the continuum 269
16.3 The Dichotomy 272
16.4 The paradox of plurality 275
16.5 Cantors continuum 277
16.6 Plurality, measure and metric 280
16.7 The Dichotomy revisited 283
Appendix 285
17 Zeno and the continuum II 289
17.1 The Arrow 289
17.2 Velocity as intrinsic 290
17.3 The Stadium 294
17.4 Could our spacetime be discrete? 299
17.5 The standard continuum: basic concerns and further puzzles 301
17.6 Are more points the answer? 306
17.7 Extension as fundamental 309
18 Special relativity 313
18.1 Time, space and Einstein 313
18.2 Lightspeed 314
18.3 Compensation or revolution? 317
18.4 Simultaneity 319
18.5 Minkowski spacetime 322
19 Relativity and reality 328
19.1 Reality unconfined 328
19.2 Compatibilism 331
19.3 Time fragmented 332
19.4 Absolute simultaneity: the quantum connection 336
CONTENTS i x
20 General relativity 343
20.1 The limits of STR 343
20.2 Equivalence 344
20.3 Spacetime curvature 346
20.4 Feeling the grip of spacetime 349
20.5 Evidence 351
20.6 Equations 352
20.7 Relativistic cosmology 353
21 Spacetime metaphysics 368
21.1 Substantival spacetime 368
21.2 Machs Principle 369
21.3 The hole argument 371
21.4 Metrical essentialism 375
21.5 An outmoded debate? 379
21.6 GTR and time 381
22 Strings 387
22.1 Higher dimensions 387
22.2 KaluzaKlein theory 390
22.3 The standard model 391
22.4 Strings 393
22.5 CalabiYau space 396
22.6 Branes and bulk 399
22.7 Shards 402
Notes 407
Glossary 427
Web resources 441
Bibliography 443
Index 455
Preface to the second edition
In the preface to the first edition I drew attention to the absence of any mention
or treatment of Zenos paradoxes. Given that much of the book was (and is still)
largely concerned with two questions Does time pass? Does space exist? this
omission was justifiable, but also regrettable. Several chapters are devoted to ques-
tions relating to motion, and Zenos paradoxes raise serious questions about how
motion can occur, even though we are confident that it does. These same para-
doxes are also closely bound up with various issues relating to infinite divisibility
and the microstructure of the continuum; these issues may pertain more to the
nature of time and space than their very existence, but they are no less fascinating
for that. I have taken advantage of the opportunities offered by a second edition
to remedy this omission: Zenos paradoxes are now discussed. Mindful of the
dangers of covering difficult material in only a superficial manner infinity is a
more perplexing topic than most the treatment is quite detailed, and extends
over two substantial chapters.
Although these new chapters constitute the most significant divergence from
the first edition, there are a number of others. Several of these are to be found
in Chapter 6, where various dynamic conceptions of time are discussed. One
of the main developments over the past decade has been the growth of interest
in Presentism (the doctrine that only what is present is real), and my treatment
of this topic has expanded accordingly. One of the main problems confronting
the would- be Presentist is the past: how can there be facts about what happened
prior to the present if the past is wholly unreal? A good deal of ingenuity has been
expended on this problem in the recent literature, and the main proposals are
outlined in 6.10. This expanded treatment has led to changes elsewhere: in my
discussion of whether Presentism is compatible with time travel (in 8.8), and the
difficulties posed for Presentism by our immediate experience of change (in 7.8).
For unrelated reasons my treatment of the doctrine of Compound Presentism in
6.11 and a difficulty facing the Growing Block model in 6.6 have also been
modified.
The only other significant additions reflect recent work in cosmology and phys-
ics. In Chapter 22 there is a new section, Branes and Bulk, devoted to some
intriguing recent speculations originating from research in string theory, including
Steinhardt and Turoks cyclic model of the universe. The section on relativistic
cosmology in Chapter 20 has also been significantly extended. In the index to
the first edition the terms dark matter and dark energy are nowhere to be
found: another regrettable omission, at least with the benefit of hindsight. Since
xi i PREFACE TO THE SECOND EDITION
2003 increasingly precise data regarding the temperature variations in the cosmic
microwave background has been emerging, thanks to the activities of the Explorer
80 satellite (also known as WMAP). These new data have powerfully confirmed
previous suspicions that the vast bulk of the cosmos is not composed of ordinary
matter: it now seems likely that some 95 per cent of the universes mass and energy
is of the mysterious dark variety. The new data have also led cosmologists to
conclude that it is very likely that our spacetime is flat, rather than curved: a
second very significant result. In 20.7 the route to this new consensus on the
composition and shape of the cosmos is explained in some detail.
In addition to these various additions to the main body of the text, there are
some stylistic and terminological alterations throughout: most minor, but one that
is more significant, as explained en route. I have updated the list of web resources,
and made some additions to the bibliography; although there are well over a hun-
dred new entries, these are not intended (by any stretch of the imagination) to be
a complete record of all that has been published in the field over the past decade.
A number of people have supplied help and encouragement while I have been
working on these additions and revisions, and I am grateful to them all. Particular
thanks to Steven Gerrard for suggesting the new edition, Kate Williams for her
skilful handling of the many changes and additions to the text, Daniel Hill, for
his painstaking and useful scrutiny of the new Zeno chapters, and Graham Ner-
lich, for many fruitful and enjoyable conversations on various matters spatial and
temporal.
Preface to the first edition
Is space an ingredient of reality in its own right, or simply nothingness? What does
the passage of time amount to? Is only the present real, or is the past real as well?
And what of the future? Are space and time finite or infinite? For anyone with an
interest in the large- scale composition of the cosmos, no questions are more press-
ing than these; likewise for those concerned to understand the framework within
which we live our lives. But few questions are more challenging. Trying to think
clearly about space is not easy. How does one go about thinking about nothing?
But time is harder still. While it may seem clear that the past is real in a way the
future is not, it also seems clear that the present is real in a way the past is not, so
what sort of reality does the past possess? It may seem obvious that time passes,
that the present is steadily advancing into the future, but just what does the passage
of time involve? These are simple questions about the most fundamental features
of our world, yet no obvious answers spring to mind.
Answers to all these questions have been proposed; many of them are fascinat-
ing, and many far from obvious. The philosophical literature, both ancient and
modern, is both large and wide- ranging, embracing as it does topics as diverse as
semantics, causation, modality and phenomenology. The relevant scientific and
mathematical litera ture is no less vast, and the relevant theories are diverse and dif-
ficult: the physics of motion; the nature of the continuum; the geometry of flat and
curved spaces; relativity theory; and quantum mechanics. This range and diversity
make the study of space and time uniquely rewarding. Few other subjects intro-
duce as many unfamiliar and exotic ideas, and perhaps no other subject stretches
the imagination so far. But it also makes the subject unusually difficult for anyone
approaching it for the first time. The aim of this book is to make this task easier.
I have tried to provide an introduction to the contemporary philosophical
debate that presupposes little or nothing by way of prior exposure to the subject,
but that will also take the interested and determined reader quite a long way.
Anyone completing the book should be in a position to make a start on some of
the more advanced philosophical work, as well as coming away with a reasonable
(if elementary) under standing of the current state of play in physics. The book is
intended primarily for students of philosophy doing courses in metaphysics or
philosophy of science (second year and above), but most of it should be intelligible
to the interested reader who is new to philosophy: philosophical jargon is used,
but sparingly, and often explained en route; a good many other terms can be found
in the Glossary. Since they will already be familiar with many of the basic themes,
avid consumers of popular science should have little difficulty with most of the
xi v PREFACE TO THE FIRST EDITION
book, especially those whose appetite for mind- bending ideas from the frontiers
of physics is temporarily sated, and who are starting to wonder what the implica-
tions of these ideas might be. Much contemporary philosophy of space and time
is devoted to precisely this question.
Given the complexity of the subject matter, I have narrowed my focus and
concen trated on just two key questions:
Does time pass?
Does space exist?
Since these questions cannot be addressed in isolation a good many other topics
are discussed as well, but there are some that are not treated at all (e.g. Zenos
paradoxes). These omissions are regrettable, but a more comprehensive treatment
would neces sarily have been more superficial, and ultimately less interesting, so I
had no qualms about making them.
There are three interconnected parts to the book. The first of these, taking up
Chapters 28, is devoted exclusively to time. Although a good deal of the ana-
lytic philosophy of time over the past half century has been devoted to linguistic
and semantic issues, I decided not to give prominence to these. The significance
of these intricate debates is far from obvious to the uninitiated, and their interest
has diminished as the prospect of their having any decisive metaphysical impact
has receded. More recently, thanks to the appearance of works such as Mihcael
Tooleys Time, Tense and Causation (1997) and Huw Prices Times Arrow and
Archimedes Point (1996), the climate has changed as metaphysical and scientific
issues have come more to the fore. I have tried to reflect this development in my
treatment here.
Although my approach to time is largely ahistorical, I begin with McTaggarts
paradox (1908), which has exerted a significant influence on recent discussions.
Although some believe that J. M. E. McTaggarts argument (properly interpreted)
reveals the untenability of all dynamic conceptions of time, I argue for a more
moderate verdict: some dynamic conceptions perish, but others survive intact.
The surviving dynamic conceptions are examined in Chapters 5 and 6 (the latter
includes discussions of the growing universe model advocated by C. D. Broad
and Tooley, and Presentism, the doctrine that only what is present is real). Before
this, in Chapters 3 and 4, the alternative Block (or eternalist) conception of time
is considered: the (initially astonishing) doctrine that time is essentially like space,
in that it does not pass. Of especial concern here is saving the appearances: why
is it that time seems to pass if in reality it doesnt? This topic is taken up again in
Chapter 7, which is entirely devoted to the phenomenology of time; how time is
manifest in our streams of consciousness. This puzzling and important issue is too
often treated very superficially or not at all. To round off this opening stage, the
possibility (or otherwise) of time travel is examined in Chapter 8.
My approach to space follows a different kind of trajectory. After a general
overview in Chapter 9, we take a step back into the past and look in some detail at
the sixteenth- and seventeenth- century debates concerning space and motion, the
main protagonists being Descartes, Galileo, Leibniz and Newton (with the latter
pair having starring roles). This historical detour provides a change of atmosphere
after the largely a priori investigation into time, but is also justified on other
PREFACE TO THE FIRST EDITION xv
grounds. Since purely metaphysical investigations into the nature of space do not
carry one very far, empirical considerations have to be introduced at an earlier
stage, and the LeibnizNewton controversy provides a useful point at which to
start, not to mention the fact that this controversy is still being discussed in the
contemporary literature.
Chapter 12 takes the debate about motion and dynamics a step further by exam-
ining how the transition to neo- Newtonian spacetime affects the LeibnizNewton
controversy. The move towards the present day is continued in Chapters 13 and
14, where the impact of non- Euclidean geom etries is considered, together with
Kants hand argument and Poincars conventionalism. Chapter 15 is purely
metaphysical: the topic is Fosters powerful but complex argument for adopting
an anti- realist stance to physical space.
The final part of the book is devoted entirely to contemporary scientific theories
and their interpretation. I provide an informal introduction to Einsteins special
theory of relativity and Minkowski spacetime in Chapter 18, and consider their
metaphysical implications in Chapter 19. It is some times assumed that the special
theory, by virtue of the way it relativizes simultaneity (and hence the present)
establishes the Block view of time beyond all doubt, but this is not so. There
are dynamic models that are compatible with the special theory, and while these
models have some counter intuitive consequences, so, too, does the Block concep-
tion, not to mention the special theory itself. Some of the rudiments of Einsteins
general theory are expounded in Chapter 20, and its metaphysical implications
(mainly concerning substantivalism) are examined in Chapter 21. I close by briefly
considering some of the more speculative developments in recent physics.
The books three parts could be seen as being devoted to time, space and
spacetime respectively for a while I considered separating and labelling them
as such but this would have been misleading. The Block universe considered
in Chapters 3 and 4 is a four- dimensional spacetime continuum, and spacetime
returns again (albeit in neo- Newtonian guise) in Chapter 12. On reflection it
seemed better to acknowledge the interrelated character of the issues by not
imposing a formal division into parts.
A problem facing any book such as this concerns the balance between philoso-
phy and science. How much science should be included? How advanced should
it be? Where should it go? Needless to say, there is no perfect solution. I have
opted for a policy of progressive integration: as the story unfolds the balance tilts
from philosophy + a little physics to physics + philosophy of physics, but
the latter stage is reached only in the last few chapters, and the level of exposition
remains elementary. A more difficult choice was what physics to include, and here,
too, hard choices had to be made: there is no separate or systematic treatment of
quantum theory. This is a deficiency, for it may well be that developments in this
area will have a decisive impact in years to come, when quantum gravity theory
comes into its own. But we are not there yet, and given the plethora of competing
interpretations of quantum theory in its current guise, it is impossible to know
what impact it will eventually have. Hence I came to the conclusion that the
omission was justifiable; remedying it would have made a long work even longer,
and sacrificing relativity was never an option. Quantum theory does, however,
make an occasional unsystematic appearance: in 8.6, in the context of backward
xvi PREFACE TO THE FIRST EDITION
causation, and in 19.4, where the focus is on the tension between quantum entan-
glement and special relativity.
Throughout the book I am less concerned with reaching final conclusions than
with revealing interconnections. Given the number of important questions in
physics and cosmology that have yet to be answered, such conclusions would be
premature at best. But, as we shall see, if our understanding of space and time is to
improve, progress also needs to be made on certain philosophical problems. Some
doubt this is possible, but I am more optimistic. Progress in metaphysics is not
impossible, but it does not come easily. When it comes it often involves uncovering
patterns of dependency, connections between doctrines not previously suspected,
or at least not obvious. By revealing what must stand or fall together, these patterns
shed new light on what the coherent metaphysical options might be. I have tried
to bring some of these patterns into clearer view.
Acknowledgements
I am indebted to a number of people for ideas and assistance: the Liverpool
students who have served as a testing ground for much of this material over the
past two years; Stephen Clark, Nicholas Nathan and Howard Robinson for helpful
comments on the chapters concerning time; Duncan Cryle; John Earman; Gerard
Hurley; Jenny Cordy for material assistance in rendering the third dimension; and
Jay Kennedy, who read an earlier draft with great care, and whose many comments
assisted me enormously. I am also grateful to Kate Williams, for her design skills
and excellent copy- editing, and Steven Gerrard, for much sound advice and his
enthusiasm for the project especially after it became clear quite what it would
involve.
1 Preliminaries
By way of setting the stage, this opening chapter introduces some of the most
important metaphysical issues concerning time and space. Although many of the
questions that can be asked about time can also be asked about space, and vice
versa, time raises distinctive issues all of its own, and I will be devoting special
attention to these. Some of the distinctive issues concerning space are discussed
in Chapter 9.
1.1 Ontology: the existence of space and time
Do space and time exist? This is an obvious question, and one that is frequently
addressed, but it can mean different things. Since philosophers raise the question
of whether the physical world as a whole exists, it is not surprising that they ask
the same question of space and time. However, the question Do space and time
exist? is usually not asked in the context of a general scepticism. Those who pose
this question generally assume that the world is roughly as it seems to be: objects
are spatially extended (a planet is bigger than an ant), objects exist at different
places and things happen at different times. So what is usually meant by the ques-
tion is this: Are space and time entities in their own right, over and above things
such as stars, planets, atoms and people?
The answer is not obvious, and for an obvious reason: both space and time are
invisible. We can see things change and move, and by observing changes we can
tell that time has elapsed, but of time itself we never catch a glimpse. The same
applies to space. We can see that things are spatially extended, and we can observe
that two objects are separated by a certain distance, but that is all. Indeed, the very
idea that space exists can seem nonsensical.
When two objects are separated by an expanse of empty space there appears to
be strictly nothing between them (ignoring what appears behind them). If space
is just nothing, how can it possibly be something? But, of course, whether empty
space really is just nothing is precisely what is at issue, and invisibility is an
unreliable guide to non- existence. Science posits lots of invisible entities (neut-
rinos, force- fields) that many believe to exist; perhaps space and time fall into the
same category. Clearly, the fact that space and time are not directly observable does
complicate matters. If we are to believe that space and time exist as entities in their
own right, we need to be given compelling reasons; the situation is different with
observable things such as rabbits and houses.
2 TIME AND SPACE
There are two opposed views on this ontological issue that have a variety of
names, but that for now we can call substantivalism and relationism. Substan-
tiv al ists maintain that a complete inventory of the universe would mention every
material particle and also mention two additional entities: space and time. The
relationist denies the existence of these entities. For them the world consists of
material objects, spatiotemporal relations and nothing else. Figure 1.1 illustrates
the difference between these two competing views; to simplify the picture only a
small volume of space is shown, and time is omitted altogether.
It is important to note that in denying the existence of space and time as objects
in their own right relationists are not claiming that statements about where things
are located or when they happen are false; they readily admit that there are spatial
and temporal relations or distances between objects and events. What they claim
is that recognizing the existence of spatial and temporal relations between things
doesnt require us to subscribe to the view that space and time are entities in their
own right. The relationist is, of course, required to give a convincing account of what
these relations are, and how they manage to perform the roles required of them.
The substantivalist is sometimes said to regard space and time as being akin
to containers, within which everything else exists and occurs. In one sense this
charac teriz ation of substantivalism is accurate, but in another it is misleading. An
ordinary container, such as a box, consists of a material shell and nothing else. If
all the air within an otherwise empty box is removed, the walls of the box are all
that remain. Contemporary substantivalists do not think of space and time in this
way; they think of them as continuous and pervasive mediums that extend every-
where and everywhen. The air can be removed from a box, but the space cannot,
Figure 1.1 The substantival and relational conceptions of space. A pair of space- fish go about their
business. The shaded block on the left represents space (or at least a small portion of it) as construed by the
substantivalist. On this view, the void between material objects is far from empty: it is filled by an extended
(if invisible) entity. Relationists reject this entity and so regard empty space to be just that but they
also maintain that material objects (likewise their constituent parts) are connected to one another: by
spatial relations. These relations are usually taken to be basic ingredients of the universe, which connect
things directly, i.e. without passing through the intervening void (in this respect the picture on the right is
misleading).
Substantivalism
universe =things +space
Relationism
universe =things +relations
PRELIMINARIES 3
and if space is substantival, a so- called empty box remains completely filled with
substanti val space, as does the so- called empty space between the stars and
galaxies. Roughly speaking, a substan tiv al ist holds that space and time contain
objects in the way that an ocean contains the solid things that float within it.
It is plain that the verdict we reach on the ontological question concern ing
space and time has momentous implications for what the universe contains. If
the substantivalist is correct, space and time are the biggest things there are. The
relationist view is much more economical: the vast pervasive mediums posited by
the substantivalist simply do not exist.
We will start to explore this issue in Chapter 9, and it will remain centre stage for
much of the remainder of the book. Although the initial focus will be on space, the
ontological status of time will not be ignored, even though it will not receive separate
treat ment (in Chapters 28, where time occupies centre stage, the substantival-
ismrelationism controversy will not feature at all). Since any spatial point also has
a temporal location, it is more properly viewed as a point in space and time, i.e. as
a spacetime point. The substan tival ist takes the totality of these spacetime points
to be a real entity: spacetime. Figure 1.2 shows a small portion of a substantival
space time the translucent block over a period of three days (one spatial dimen-
sion is sup pressed). Clearly, if this conception is correct, time is just as real as space.
Figure 1.2 Spatiotemporal (spacetime)

substantivalism. Material objects are distributed
through an extended medium composed of spacetime
points. Since one spatial dimension has been
suppressed, each two- dimensional slice of the block
represents a three- dimensional volume of space.
1.2 Questions of structure
Irrespective of how the debate between substantivalists and relationists is resolved,
there are questions to be addressed concerning what can loosely be called the
structural characteristics of space and time. Whereas substantivalists take these
questions to be about actual entities, relationists take them to be about the ways
material things are related in spatial and temporal ways. To get some impression
of the flavour of these structural questions consider the following claims:
Space is infinite in size, three- dimensional, Euclidean (so travellers moving
along parallel trajectories will never meet), isotropic (there is no privileged
spatial direction) and continuous (infinitely divisible).
Time is infinite, one- dimensional, linear, anisotropic (there is a privileged
temporal direction) and continuous.
4 TIME AND SPACE
There is just one space, and one time.
Time and space are absolute, in that they are unaffected by the presence or
absence (or distribution) of material things.
Some readers may find this characterization quite plausible. Those with some
knowledge of ancient and medieval cosmologies will realize that it is similar to
one defended by the ancient atomists but soon rejected in favour of the Aristote-
lian system (which posited, among other things, a finite, non- homogeneous and
anisotropic space), and which returned to favour only in the 1500s. Those with
some scientific (or science fiction) background will realize that this essentially
Newton ian conception was overturned by Einsteins theories of special and gen-
eral relativity in the early years of the twentieth century. More recently, some
quantum theorists have argued that time is branching rather than linear, while
others maintain that time does not exist; relativistic cosmolo gists have argued that
there may be a multiplicity of spacetimes (or baby universes) sprouting from the
other end of black holes, and superstring theorists argue for the astonishing view
that there are nine (or more) spatial dimensions.
Fascinating as many of these recent theories are, this is not the place for a
thorough discussion of them, although their implications are such that they can
scarcely be ignored either. Where space is concerned, my main focus will be on the
substantivalistrelationist controversy, and I will only be looking at the structural
issues that are relevant to it. Since it turns out that a good many structural issues
are relevant to it, we shall have occasion to enter into at least the shallows of these
perplexing waters.
1.3 Physics and metaphysics
Are there any facts about space and time that are necessary (hold in all possible
worlds), and that can be established by a priori reasoning? Or are the answers to
all of the interesting questions about space and time contingent and a posteriori,
and hence to be answered only by the relevant sub- branches of science physics
and cosmology? If we answer yes to the latter question, the philosophy of space
and time reduces to a branch of the philosophy of science and we must confine
ourselves to examining the content of contemporary scientific theories, in particu-
lar relativity and quantum theory.
This has something to be said for it for, as we have just seen, scientists have
made claims about space and time that provide answers to many of the questions
philosophers have struggled with; many (if not all) of the relevant theories are well
confirmed and so cannot be ignored. It would also be fair to say that science and
mathematics have provided us with new ways of thinking about space and time,
ways that philosophers may never have stumbled across if left to their own devices.
The following simple but influential argument dates back to ancient Greek times:
The idea that space is finite can be shown to be absurd. Suppose that space
were finite. If you travelled to the outermost boundary of space and tried
to continue on, what would happen? Surely you could move through this
supposed barrier. What could stop you, since by hypothesis there is nothing
PRELIMINARIES 5
on the other side? In which case, the supposed limit isnt really the end of
space. Since this reasoning applies to any supposed edge of space, space must
be infinite.
There are replies the metaphysician can make. For example:
By definition, to move is to change your place in space. Accordingly, it makes
no sense at all to suppose you could move beyond the edge of space, since by
hypothesis there are no places beyond this limit for you to occupy. Conse-
quently, the assumption that space is finite does not result in an absurdity.
This may seem reasonable enough, but there are further moves and countermoves
available to the metaphysician. Does the reply tacitly assume a substantivalist view
of space? A relationist would reject the very idea that motion consists of mov-
ing on to pre- existing spatial locations; if motion consists instead of an object
changing its spatial relations to other objects, what is to prevent an object from
moving ever further from other objects? And so it proceeds. But as we shall see in
Chapter 13, the assumptions underlying this debate were completely transformed
in the nineteenth century by the discovery of non- Euclidean geometries and the
possibility that a space could be curved. If our space were positively curved a
three- dimensional counter part of the surface of a sphere someone could set off
on a journey, always travel in a straight line (never deviating to the right or left),
and still end up just where they started. Space can thus be finite without possess-
ing any limits, edges or obstacles to movement. This is just one example of how
philosophical discussions of space and time have been influenced if not rendered
redundant by developments in physics and mathematics; there are others.
However, while the science is important, there are several reasons why it would
be a mistake to suppose that only science matters, and that considera tions of a
philo sophical sort are dispensable. First of all, physics and metaphysics are to some
degree interdependent. When evaluating a theory, when inventing a theory, scien-
tists are themselves influenced, in part, by metaphysical considerations: by what it
makes sense to say the world is like. Moreover, the metaphysical implica tions of a
scientific theory i.e. what implications the mathematical formalism has for how
the world is are often unclear. This is especially true of relativity and quantum
theory. Hence there is a need for metaphysical inquiry, even if this inquiry does
not take place independently of scientific theorizing.
But metaphysics also has its own distinctive domains and methods of inquiry.
While it may well be that to discover the answers to some questions about the
space and time in our world we will have to listen to the physicist, there are other
questions for which this is not so. Some of these questions are about topics that
lie altogether outside the domain of physics: for example the role space plays in
our ordinary modes of thought; the way time manifests itself in human conscious-
ness; our different attitudes to the past, present and future, and their rationality
or otherwise. But even if we confine ourselves to talking about time and space
themselves, rather than how we think about or experience them, there are concep-
tual issues of a distinctively philosophical sort that remain of interest. Assuming
(as most do) that the laws of nature are contingent, there are possible worlds of
a recognizably physical sort, worlds that resemble ours in superficial (observable)
6 TIME AND SPACE
ways, but where the laws of nature are different. Many of these possible worlds
are spatiotemporal (e.g. those where Newtons laws hold, those where Aristotles
laws hold). The fact that these worlds are spatiotemporal means that our concepts
of space and time are to some degree independent of particular scientific theo-
ries, and so we can investigate these concepts independently of any one scientific
theory. What must a world be like in order for it to be spatial and temporal? What
are the different forms that space and time can take? In their attempts to answer
these questions philosophers have introduced many novel and counterintuitive
ideas, a good many of which are stranger than anything seriously entertained by
scientists engaged in the struggle to ascertain the particular forms that space and
time take in this world.
Finally, there is this point. We know that our current fundamental physical
theories are imperfect: quantum theory and general relativity have yet to be fully
reconciled. It may well be that the theory that emerges from this eventual marriage
will have very different implications for the nature of space and time than those
of currently accepted theories, so it would be very short- sighted to take current
scientific theories to be the last word on space and time in our universe (let alone
other possible universes).
1.4 Time: the great divide
The issues that have been introduced so far, concerning ontology and structure,
concern both space and time equally. But there is one issue concerning time that
has no spatial counterpart. If asked How does time differ from space?, after a
few moments reflection you might say Its not so easy to travel through, it has
one dimension rather than three, and it has a direction. I want to focus on the last
answer. Why do we say that time has a direction whereas space doesnt? Since there
are three dimensions of space and one dimension of time, someone in an awkward
frame of mind might point out that space has more directions than time, and in one
sense they would be right. But what this response omits to mention is that none of
these spatial directions is singled out as in any way special or distinctive. When-
ever we draw a diagram with an axis representing time we unthink ingly insert a
single- headed arrow pointing towards the future. Although we may also insert
arrows indicating various spatial directions, no one spatial direction is privileged;
up, down, right, left, north, south, east, west, irre spec tive of what we call them,
each direction gets its own individual arrow (see Figure 1.3). This diagram -
matic convention has a linguistic counterpart: we sometimes talk of the arrow
of time, but never of the arrow of space. But what exactly does times arrow
involve?
The pessimist who says Things always get worse, never better, a senti ment
often taken to be confirmed by the second law of thermodynamics, according
to which entropy (or, loosely speaking, disorder) will always increase, provides
an illustration of one sort of temporal arrow: certain sorts of process favour one
temporal direction over the other. But there is, it seems, more to it: the arrow of
time runs deeper, there is a directedness that belongs to time itself, rather than to
anything in time. Or so many of us are inclined to think.
PRELIMINARIES 7
We speak of time passing, the march of time, the flow of time, and there
are no spatial analogues of these locutions (we dont speak of space passing, the
march of space, the flow of space, etc.). As for what we mean when we say that
time passes, the following formulation encapsulates the essentials:
What is future will become present; what is present will become past; what
is past was once present.
Your birth was once present, but now lies in the past. Your next birthday lies in
the future, but it is getting closer every day. Getting closer to what? The present,
of course. We can think of the present as advancing into the future, or the future
advancing towards the present; either way the distance between the two is always
decreasing. What applies to your next birthday applies to all times and all events:
they start off in the future, become present and then recede into the past. This
process is often called temporal passage, and has no obvious spatial counterpart.
This is how we talk, but what is there in reality that corresponds to this talk? It is
here that we reach the great divide in the philosophy of time. There are those who
believe that temporal passage is a real phenomenon, an objective feature of reality
that is independent of the perspective that conscious beings such as ourselves have
on things. Anyone who believes this subscribes to a dynamic view of time. Some
dynamists hold that passage involves a special property of present ness moving
along the timeline. Others explain passage in terms of the non- existence of the
future: only the past and present are real, but since reality is growing new times
are coming into being as future possibilities crystallize into present actualities no
time remains present for long (see Figure 1.4). Others deny reality to both the past
and the future: time consists of a succession of ephemeral presents.
In the opposing camp are the many philosophers who reject the claim that
temporal passage is a real and objective feature of the world. These philoso phers
subscribe to a conception of time known variously as the Block view, four-
dimensionalism or eternalism. According to this doctrine all moments of time
(and all events) are equally real, and there is no moving or changing present; noth-
ing becomes present and then ceases to be present. The differences between past,
present and future are simply differences of perspective. If someone in Australia
Figure 1.3 A natural way of thinking about space and time. Nothing distinguishes the various spatial
directions; in outer space (if not on Earth) we can move equally freely in any direction. Time, however,
does possess an inherent direction or directedness, running from past to future.
FUTURE
PAST
PAST
up
left
front
down
right
back
8 TIME AND SPACE
says Its sunny here, they are saying something true if it is sunny at the place
where they are speaking; likewise for someone in New Zealand who says Its
raining here. We do not believe that either of these places is special or distinctive
simply because they are referred to by the term here. According to proponents of
the four-dimensional conception, the same applies in the case of now. Someone
in 1801 who says Its snowing now is saying something true if it is snowing at the
time they are speaking; likewise for someone who says the same thing in 2001. On
this view, time and space are ontologically equivalent in this crucial respect: space
consists of a three- dimensional arrangement of coexisting locations (places), none
of which is special; time consists of a linear one- dimensional series of coexisting
locations (times), none of which is special. Talk of coexisting times (or events)
in this context must be properly understood. The claim is not that the times in
question all exist at once, or simultaneously if they did, history would be reduced
to a single instant but rather that all times in our universe exist together, to just
the same extent as any two simultaneous locations in space, despite the fact that
they occur in succession, and so are non-simultaneous.
It is not without reason that this conception of time is often called the Block
view.
1
Think of a long, solid crystalline block embedded within which are small
plastic figures in various poses. If the substantivalist view is correct, this is essen-
tially our condition, save that the block we are embedded in extends through time
as well as space, and so is a four- dimensional rather than three- dimensional entity.
All the people frozen in different poses at different places and times in this block
are equally real; indeed, some of these people are you, as you were yesterday
and at still earlier times, and tomorrow and at still later times (see Figure 1.2).
Although all the contents of the block are permanent fixtures it is a mistake to
suppose that the block is a thing that endures. The block does not exist in time as
we do; it would be more accurate to say that time is in the block, for time is simply
one of the dimensions that the block possesses. Viewed as a whole, the block is an
eternal (or sempiternal) entity (see Figure 1.5).
If, on the other hand, the relationist view is correct, we are not embedded in
anything analogous to a block of crystal; the world consists of nothing but material
bodies in spatial and temporal relations. But these interrelated things nonethe-
less constitute a four- dimensional ensemble; all that is missing is the surrounding
medium posited by the substantivalist.
These two views of reality, the eternal and the dynamic, are as old as West-
ern philosophy. Heraclitus (approx. 540480BC) held that the world was entirely
dynamic, an endless process of change, creation and annihila tion. Parmenides
(approx. 520430BC) maintained that the opposite was true: A single story of
PAST
NOW
FUTURE
FUTURE
FUTURE
FUTURE
FUTURE
Figure 1.4 Perhaps the most popular of
the dynamic views of time: the past is fixed
and real; the future unreal and unfixed
(there are many ways things may unfold);
the present (or now) is real, distinctive
and advancing into the future. Despite its
appeal, this model of time is not without its
difficulties. For example, one might ask: Just
how is the present different from the past?
PRELIMINARIES 9
a road is left being is ungenerated and undestroyed, whole, of one kind and
motionless and balanced nothing is or will be other than what is since that has
Fate fettered to be whole and motionless.
2
And the issue is one that still resonates.
When first encountering the Block view many people find it utterly absurd. That
we live out our lives in a constantly advancing present, that past and future events
are different from those that are present what could be more obvious? But as
Block theorists are fond of pointing out, there are plenty of other cases where our
everyday experience seems to provide overwhelming support for beliefs that are
false. Can you remember how you felt when, as a child, you were told that there
are folk in other countries for whom our down is their up? Even now, are you
entirely comfortable with the thought that the Earth is shooting through space at
a rate of many miles per second?
Others find the Block view more repellent than absurd. Most of us tend to
assume that, even if the past is fixed, the future is open: what it will bring is as yet
undeter mined. If the Block view is true this is wrong, for the future is just as real,
solid and immutable as the past. How our lives will unfold from now until the
moment of our deaths is (in a manner of speaking) already laid down. How could
it be otherwise if the future stages of our lives are just as real as the past stages?
This is not to say that we have no power over the ways our lives will unfold, for
we do. We will all make choices, and the choices we make will contribute to the
ways our lives will turn out. But if the Block view is true, the choices we will make
are inscribed in the fabric of reality in precisely the same way as the choices we
have already made. While those who find this tempo ral determinism unsettling
will be inclined to reject the conception of time from which it derives, it would
be a mistake to suppose that the Block view must be false simply because it has
unpalatable consequences. The universe is under no obligation to conform to our
preconceptions or preferences.
3
The issue of whether or not time is dynamic may impact on how we think of
our lives, but it also has consequences of a less parochial sort. A universe in which
the sum total of reality is limited to what is (momentarily) present is far smaller
than a universe in which the sum total of reality includes both the past and future
(a) (b)
Figure 1.5 The Block universe. As (a) makes clear, to view the block as something that exists within time
is quite wrong. Image (b) is an attempt to put flesh on the idea that dynamic features of the universe, such
as change and movement, are merely patterns running along the block. Although depictions such as these
can be useful, they are also highly schematic. The block-universe is really a four-dimensional entity (one
dimension of space has been suppressed here). If time were infinite, the block would be infinitely long; if
space were infinite in all directions, the block would be infinitely wide and tall. It is only if time were finite
and space cube-shaped that the universe would have the shape depicted.
10 TIME AND SPACE
as well, and likewise, albeit to a lesser degree, for a universe in which only the
past and present are real. The smallness in question concerns temporal rather
than spatial extent but, as Figure 1.6 makes clear, it is no less significant for that.
The substantivalism relationism issue has significant ontological ramifica-
tions since, if space is a thing, it is a uniquely large thing, extending as it does
everywhere. But the ontological ramifications of the eternaldynamic dispute are
vast in comparison.
1.5 Two frameworks
Common sense may tell us that time is dynamic rather than eternal or passage-free,
but our ordinary ways of talking about time reflect both perspectives. There are
two families of temporal terms and concepts that we commonly use, one of which
embodies the dynam ic view and the other the opposing non-dynamic (or Block)
view. McTaggart is credited with clearly distinguishing these two ways of think-
ing. Since the distinction itself and McTaggarts terminology continue to play an
important part in contemporary debates, the sooner they are introduced the better.
4
Imagine finding the following message- in- a- bottle: The bomb will explode
under the Eiffel Tower four days from now. The message is annoyingly unin-
formative: ignorant as you are of when the message was written, you do not know
which time the word now refers to, and so you have no way of knowing whether
the bomb was intended to explode a hundred years ago, in two days time, or fifty
years in the future. The same would apply if the message read the day after
tomorrow or even today. It would be quite different if the message had read:
The bomb will explode under the Eiffel Tower on 1 June 1987.
This simple example provides an illustration of a more general point. We can
think of events as being situated at some distance or direction from the present
time: for example, E lies in the past (or future), E is occurring now, E will
occur three days from now, or E occurred fifty years ago. Alternatively, we
can think of events in terms of when they occurred in relation to other events
(or times), without making any reference to the present at all: for example, E
occurred seven years after the Battle of Trafalgar, E happened three days before
the first Moon landing, or E took place in 1977.
McTaggart called the time series that runs from the far past up to the present
now and then on into the future the A- series, and the time series that is
Figure 1.6 Three conceptions of
the sum total of reality.
PRELIMINARIES 11
entirely independent of the present the B- series. Each is associated with its own
range of distinctive temporal characteristics: pastness, presentness and futurity
are the basic A- properties, whereas the basic B- properties are simultaneous with,
earlier than and later than. Subsequent writers also talk of A- and B- concepts,
A- and B- sentences, A- and B- statements (what particular A- and B- sentences are
used to express on particular occasions of use) and of A- and B- facts. For the
sake of convenience I will include this diverse range of items under the rubrics
A- framework and B- framework.
The ways of thinking about time enshrined in the A- and B- frameworks are
sometimes called the tensed and tenseless conceptions. Those who adopt this
terminology speak of tensed properties, tenseless facts and so forth. Although
the concept of tense is usually associated with grammar and linguistics, this more
liberal use has a clear rationale. The basic grammatical tenses are past, present and
future the very notions that underlie the A- framework and that are conspicu-
ously absent from the B- framework.
Although the A- series and the B- series involve different ways of thinking about
time, it would be a mistake to suppose that the two are wholly independent. The
basic B- relationships of being simultaneous with, earlier than and later than, are all
to be found within the A- framework: for example, today is one day later than
yesterday, and a week ago is seven days earlier than today. Moreover, both
frameworks order the same events in the same ways. Suppose it is now 1 Decem-
ber. Then clearly today and 1 December refer to the same time, but so do
yesterday and 30 November, tomor row and 2 December, and likewise
for the remainder of the two series. Evidently, although we think of time in terms
of the A- and B- frameworks, we dont ordinarily think that there are two sets of
distinct times that run in parallel; there is just one sequence of times.
The two frameworks may be intertwined, but they are also very different. The
B- series is a static ordering of times and events. The first Moon landing in 1969
took place 24 years after the end of the Second World War, and 31 years earlier
than the end of the second millennium. These B- facts not only obtain now, they
obtain at other times; they will never cease to obtain. If a B- statement is true it
remains true. A- statements and A- facts could not be more different. If at the start
of December I say There are now 25 days until Christmas I am speaking the
truth; if I make the same statement a few days later I am not. Why? Because the
time referred to by now in the two statements is different. The A- series is a
dynamic, constantly changing, temporal ordering, and the fact that the present
is never still but always advancing into the future renders A- facts transitory and
dooms A- statements to changes in truth- value.
1.6 Matters terminological
The following characterizations are rough and ready but they will serve for present
purposes.
The Block view: all times and events timelessly coexist, and all are equally real;
temporal passage is unreal.
12 TIME AND SPACE
The dynamic view: temporal passage is real, and time is not an ensemble of
coexisting times and events. In short, the block-universe view is false.
The tensed theory: A- concepts such as past, present and future have an essen-
tial and ineliminable role to play in any metaphysically adequate account of
the nature of time.
The tenseless theory: B- concepts such as earlier than, later than and simultane-
ous with are all we need for a metaphysically adequate account of time.
The dispute between dynamists and eternalists is over the nature of time, how time
is in and of itself; the tensedtenseless dispute is over the terms or concepts needed
to describe time and its contents. These are distinct issues, and we should not at
this point make any assumptions about how they will be resolved. Consequently,
we need to be careful how we use and understand terms like A- theorist and
B- theorist. As I will be using the terms, a B- theorist is someone who subscribes
to the Block view, and an A- theorist is someone who doesnt.
2 McTaggart on
times unreality
2.1 Could time be unreal?
In an attempt to get us to take seriously the idea that reality might be very different
from how it seems, Descartes introduced an all- powerful demon. This demon, we
are to suppose, is causing us to hallucinate, to have experiences that do not corre-
spond to anything real. The experiences in question are the perceptual experiences
that we are having at the moment and those we remember having in the past: the
experiences that we unthinkingly take to be revealing the world to us. The point
of this scenario is to lead us to appreciate that our experience does not guarantee
that the world is as we normally believe it to be. Let us vary the scenario. The all-
powerful demon informs us that it has not been providing us with hallucinatory
experiences the world really is pretty much as it seems but it has instead taken the
decision to abolish time.
What would you make of this announcement? Should you care? What difference
would it make? Perhaps the world would cease to exist: when the demon acts the
present winks out of existence, and the past along with it. Or perhaps everything would
come to a stop. People and things wouldnt cease to exist, they would simply stop
chang ing. They would be frozen in mid- sentence, in mid- air, and (time lessly) remain
in this condition. These suggestions should not be taken too seriously, but most of us
would surely agree that the absence of time would make a very significant difference.
McTaggart had no truck with demons, but he did maintain that in actual fact time
is unreal. His claim is all the more remarkable (and hard to believe) because he did
not also claim that we are being provided with delusory perceptions. McTaggart
maintains that time would be unreal even if the world were exactly like it seems
to us to be (except for the fact that time is unreal). You may think you have never
heard of anything more ridiculous, but by way of a defence for his claim McTaggart
deploys a metaphysical argument to the effect that time could not possibly exist. This
intriguing and (at times infuriating) argument continues to be discussed and taken
seriously, and many hold that, even if it does not establish what McTaggart thought
it did, it still tells us something important about the nature of time.
2.2 Change as the essence of time
McTaggarts argument revolves around the A- series (which runs from the past,
on to the present and into the future) and the B- series (which runs from earlier
14 TIME AND SPACE
to later), which we encountered in 1.5. He suggests that we never observe time
except as forming both these series (1908: 458), and that anyone who believes
in the reality of time should believe that both series are real. His overall argument
for the unreality of time has this form:
1. If the A- series does not exist, time does not exist.
2. The A- series is metaphysically incoherent, and therefore does not exist.
3. Therefore time does not exist.
As is plain, if 1 and 2 are both true, his conclusion 3 follows. We will be looking
at his arguments for 2 in the next section; for now I want to focus on 1.
McTaggarts argument for 1 involves establishing two claims. The first is (on
the face of it at least) uncontroversial: time is the dimension of change. This isnt
to say that in a temporal world there is always something changing; there are
powerful arguments to the effect that temporal vacua periods of time during
which nothing changes are conceptually speaking possible.
1
Even if temporal
vacua are (conceptually) possible, it still seems true that time is the dimension in
which changes do and can occur. A world without time would certainly be a world
without change. McTaggarts distinctive (and undeniably plausible) claim is that a
world in which change cannot possibly occur is a world without time.
The second claim is far more contentious: change requires the A- series. Why
believe this? McTaggart asks us to consider a poker that is red hot at one end and cool
at the other. The poker possesses different properties at different places; it is spatially
extended and one part of it is hot whereas another part is cool. Does this variation
in properties over space amount to a change? Clearly not. A hot poker that cools
down over a period of a few minutes involves change, involving as it does a change
in properties over time, or so common sense suggests. Now just suppose that the
B- theorist is right and that time could exist without an A- series, and consider a
poker that is hot at t
1
and cold at t
2
. The B- theorist will say that this is an instance of
change, but McTaggart objects: what is it that changes here? Everything about the
B- series is permanent so it is always true that the poker is hot at t
1
and cold at t
2
. These
B- facts obtain at all times. Variations in properties across space at a given moment
of time do not amount to change because all the places in question coexist. The
B- theorist who holds that all moments of time are equally real is, in effect, viewing
time as we ordinarily view space: as an extended dimension of coexistent locations.
As a result there are no essential differences between spatial variation at a time and
the B- theorists account of change over time: in both cases we have an object pos-
sessing different properties at different locations in a static dimension. It is quite
different if we introduce the A- series into the equation. We still have the permanent
B- facts the poker is hot at t
1
and cold at t
2
but in addition we have ever- changing
A- facts. The pokers being hot at t
1
is first in the distant future, then the near
future, then the present and thereafter the ever more distant past.
Hence McTaggarts conclusion: a world without ever- changing A- facts, an
eternal B- world, is a world where change does not and cannot occur, and so is
a world without time. McTaggart does not disagree with those who hold that a
poker that is hot at one time and cold at another time has changed; his claim is
that in the absence of changing A- facts there are no times at all, and so change
cannot occur.
McTAGGART ON TIMES UNREALITY 15
McTaggarts line of argument here may fail to convince fully, but it is also true
that there is something to what he is saying. B- theorists undoubtedly do regard time
as being more closely akin to space than we are accustomed to supposing it to be,
and to this extent the B- theory account of change runs counter to common sense.
Let us turn now to McTaggarts reasons for thinking that the A- series cannot be real.
2.3 McTaggarts A- paradox
The objection to the A- series for which McTaggart is best known begins with a
statement of the obvious: past, present and future are incompatible attributes.
If an event M is present, for example, it cannot also be either past or future. No
event can possess any two of these attributes without generating a contradiction.
However, since everything starts off as being future, then becomes present before
sinking into the past, it follows that M does possess all three A- properties, and the
same applies to all events (or times). Since the A- properties are incompatible, this
is impossible; but if the A- series were real, it would be the case. The A- series is
thus paradoxical, and cannot exist.
This reasoning may well seem suspect, and McTaggart is not blind to the
obvious reply. The A- theorist will insist, quite properly, that while it is true that
nothing can possess incompatible properties at the same time, this is not how
A- properties behave:
It is never true, the answer will run, that M is present, past and future. It is
present, will be past, and has been future. Or it is past, and has been future
and present, or again, is future and will be present and past. The character-
istics are only incompatible when they are simultaneous, and there is no
contradiction to this in the fact that each term has all of them successively.
(McTaggart 1908: 468, original emphasis)
However, McTaggart maintains that this plausible reply fails. He starts by claim-
ing that the reply involves a vicious circle: confronted with an apparent paradox
involving the A- series, the A- theorist is employing the A- series itself to get around
the problem. However, he again anticipates a plausible response: Our ground
for rejecting time, it may be said, is that time cannot be explained without assum-
ing time. But may this not prove not that time is invalid, but rather that time
is ultimate? (1908: 470). It may well be impossible to explain notions such as
goodness or truth in completely different terms, but this does not lead us to reject
the notions in question; perhaps time is also a primitive concept, one that cannot
be explicated in other terms. But McTaggart has a further objection up his sleeve.
He maintains that the proposed solution introduces a vicious infinite regress.
The solution that McTaggart has in mind runs thus. An event M doesnt have
the incompatible characteristics of being past, present and future at the same time,
it has them successively; for example, an event is present, will be past and has been
future. McTaggart says that this, in effect, introduces a second A- series, and that
this A- series will suffer from the same difficulty as the first. What does he mean?
In the 1908 article he doesnt explain. In his 1927 restatement he goes into more
detail. In what follows I expand still further.
16 TIME AND SPACE
McTaggart starts by providing a way of reformulating A- statements such as:
(A) M is present, will be past and has been future.
He suggests that when we say X has been F, we are asserting X to be F at a
moment of past time; when we say that X will be F, we are asserting X to be F
at a moment of future time; when we say that X is F, we are asserting X to be F
at a moment of present time.
2
We can now restate (A) thus:
(A)' M is present at a moment of present time, past at a moment of future
time and future at a moment of past time.
Or, more succinctly:
(A)'' M is present in the present, past in the future and future in the past.
In using these formulations we are introducing compound or second- level tem-
poral predicates. The availability of these second- level predicates may seem to
provide an escape route from the initial paradox, but this is an illusion generated
by an incomplete grasp of the real situation. There are in fact nine second- level
predicates, not just three:
First- level Second- level
is past in the past [1]
is past is past in the present [2]
is past in the future [3]
is present in the past [4]
is present is present in the present [5]
is present in the future [6]
is future in the past [7]
is future is future in the present [8]
is future in the future [9]
Assuming that time passes, the position of an event in the A- series is constantly
changing, so every event falls under each and every one of these second- level
predicates. Take an event such as the death of Kennedy and consider predicate
1, is past in the past. This was true yesterday, since yesterday is now past, and
Kennedys death occurred before it. Take predicate 2, is past in the present.
This is true now, while predicate 3, is past in the future, will be true tomorrow;
predicate 4, is present in the past, is true at the date Kennedy died, as is predicate
5, is present in the present. Predicate 6, is present in the future, applies to the
event before it happened, as do predicates 7, is future in the past, 8, is future
in the present and 9, is future in the future.
Now, some of these second- level temporal predicates are compatible with one
another; for example, predicates 3, 5 and 7, which were introduced to avoid the
initial paradox. But some are not compatible; for example, predicates 2, 5 and 8,
which together state that M is past in the present, present in the present and future
in the present. McTaggart can thus reasonably claim that introducing second- order
tenses fails to eradicate the initial paradox since precisely the same paradox exists
at the higher level of temporal predication.
McTAGGART ON TIMES UNREALITY 17
In reply, the A- theorist might insist that events dont have all nine second- level
properties simultaneously, rather they have them successively. In describing this
succession third- level temporal predicates are required. But once again, the pas-
sage of time ensures that each and every third- level predicate applies to everything,
and not all of them are compatible. So we have to move to fourth- level predicates,
but the same applies here, and so we have embarked on an endless regress through
ever more complex temporal predicates. It is easy to see that the difficulty is not
removed by ascending in the hierarchy, no matter how many levels we add. The
three predicates:
past
present in the present in the present in the present
future
are equivalent to the first- level predicates past, present and future, so if
there is an inconsistency connected with the first level, the inconsistency is also
present at each and every level in the (infinite) hierarchy.
This regress is undeniably vicious. We start with a paradoxical situation, and
to eliminate the problem we appeal to a higher level of temporal predicates. But
since each successive level harbours essentially the same problem, the inconsistency
is never removed. McTaggart thus concludes that the A-series is inherently para-
doxical, and so cannot correspond with anything in reality. It doesnt make sense to
describe things as present, past or future, if we understand these terms as the tensed
theorist does, as properties that all events eventually possess. Given that he also
maintains that without A- change the relationships of earlier than and later than
cannot be real, he concludes that time does not exist. What does exist, according to
McTaggart, is an entirely atemporal C- series: an eternal four-dimensional block
of events, whose contents are ordered but not in a temporal way.
2.4 Other routes to the same place
What are we to make of McTaggarts A- paradox? If you find yourself thinking
Interesting, but Im not convinced; theres something fishy going on, but Im not
quite sure what! you would not be alone. Broad observed of the A- paradox: I
suppose that every reader must have felt about it as any healthy- minded person
feels about the Onto logical Argument for the existence of God, viz., that it is
obviously wrong some where, but that it may not be easy to say precisely what is
wrong with it (1998: 77). But then again, there are contemporary philosophers
of time who disagree: Horwich, Le Poidevin, Mellor, Oaklander and Schlesinger
have all argued that McTaggarts paradox is genuine.
For those inclined to resist McTaggarts reasoning, one plausible diagnosis is to
deny that the alleged vicious regress ever gets off the ground. McTaggart simply
assumes that a tensed theorist who believes that time passes is committed to the
claim that, for any event E, E is past and present and future. Are tensed theo-
rists really obliged to use this unnatural and obviously problematic formulation in
describing what is involved in the passing of time? Smith, himself an advocate of
the tensed theory, rejects this formulation in favour of the following: for any event
18 TIME AND SPACE
E, E will be past, is now present, and was future; or E is now past, was present, and
was (still earlier) future; or E is now future, will be present, and will (still later) be
past (1993: 171, original emphasis). Not only is this disjunctive formulation more
natural than McTaggarts somewhat strained constructions, but it seems free from
paradox. (See 5.2 for further discussion of this disjunctive approach.)
However, the fact that McTaggarts argument is flawed does not mean that his
conclusion is false: as we are about to see, the particular dynamic model of time
he was seeking to discredit is deeply problematic, even if not for the reason he
proposed.
McTaggarts target is a theory of time that could aptly be called the transitory
tensed property view. In its simplest guise, this theory posits a static time- line (the
B- series), and a moving present, with the latter typically construed as a distinctive
property, presentness, which is successively instantiated at different times. In a
more graphic vein, this conception is sometimes known as the Moving Spotlight
model. C. D. Broad used slightly different terms in describing this position:
We are naturally tempted to regard the history of the world as existing eter-
nally in a certain order of events. Along this, and in a fixed direction, we
imagine the characteristic of presentness as moving, somewhat like the spot
of light from a policemans bulls-eye traversing the fronts of the houses in a
street. What is illuminated is the present, what has been illuminated is the past,
and what has not yet been illuminated is the future. (Broad 1923: 59)
Whatever the label, the model faces two problems. First, what is the nature of the
transitory property? When objects and events take their turn under the moving
spotlight of the present, how are they affected or influenced? Second, does it really
make sense to suppose that this distinctive property of presentness is transitory? I
will deal with these in turn.
2.5 The nature of A- properties
Properties can be divided into those that are intrinsic and those that are relational.
A sugar cubes shape, size and composition are among its intrinsic properties. Its
relational properties include being smaller than Mount Everest, being to the left of
a teacup, being heavier than a flea and being above the floor but beneath the ceil-
ing. The cube has many of its relational properties in virtue of having the intrinsic
properties it does (if it were much, much bigger, it would be larger than Mount
Everest), but the two sorts of property are nonetheless quite different. Having a
relational property involves being related in some way or other to one or more
other objects. An objects intrinsic properties are a matter of how the object itself
is, irrespective of its relationships with other objects.
The doctrine that A- properties are intrinsic (or qualities in McTaggarts
terms) can take different forms. There could be just one intrinsic A- property of
presentness; pastness and futurity would then be defined relationally, in terms
of changing B- distances from the present. Or there could be three intrinsic
A- properties: pastness, presentness and futurity. Or, alternatively, just two, with
the third defined relationally. The important point is that, on this view, an object
McTAGGART ON TIMES UNREALITY 19
that possesses an intrinsic A- property has an additional feature, akin to size, shape
and mass, that does not depend on its relationships with other things. If, on the
other hand, A- properties are all relational, then possessing any given A- property
makes no difference to how an object is in itself; it is merely a matter of being
related in some way to something else.
McTaggart suggests that A- properties are more likely to be relational rather than
intrinsic (1908: 467). He does not explain why, but it is certainly not easy to see
how they could be intrinsic, and to this extent at least his position is understandable.
Suppose we try to explain the meanings of the A- predicates in the way that McTag-
gart himself suggests: Here is a hand. The hand you see before you is present, your
dinner this evening is future, your breakfast this morning is past. What intrinsic
feature does the hand you are now looking at have that a hand in the past lacks?
Does the hand situated in the past lack colour, mass, shape or size? If so, why is this
ghostly residue a hand at all? But if the past hand possesses colour, mass, shape and
size, what property does it lack that the present hand possesses? Saying the property
of presentness is of no assistance for, even if we suppose this is true, we have no
idea what difference possession of this property makes to an object.
So should an A- theorist say that A- properties are relational, the option McTag-
gart thought most plausible? Almost certainly not. We know that all B- type rela-
tions within the B- series are permanent; since events are always changing their
A- properties, if the latter are relational it follows that an events A- properties must
involve a relationship with something external to the B- series call it X and
hence external to the time- series itself. But what could this X be? As McTaggart
observes, it is difficult to say (1908: 468). But even if we could say, the rela-
tionist is not out of the woods. We are now supposing that the passage of time
consists of X moving along the B- series, momentarily bestowing presence on the
times it touches: first X confers presentness on t
1
, then on t
2
, then on t
3
, and so
on. The trouble is, since we are also assuming that being present does not involve
any alteration in a things intrinsic properties, coming to be related to X in the
presence- conferring way makes no difference to how a thing is in itself. Conse-
quently, if X were moving in the wrong direction towards what we call the
past we would be none the wiser. The passage of time, thus construed, is simply
not worth having: any serious A- theorist will want to insist that times passing
makes a significant difference of some sort to the things in time.
In which case, the A- theorist is obliged to return to the first option, and
hold that A- properties are intrinsic. As for the problem of the character of the
A- properties, one option is to say something like this:
It is a mistake to think that presentness is a single intrinsic property, an ingre-
dient to be found in every presently existing thing; there is no such property.
The movement of the present consists of a distinctive and transitory altera-
tion in at least some of the intrinsic properties objects possess.
By way of an analogy, suppose a reel of a black and white movie film is
unwound and glued to a long glass wall. As you walk along the wall, looking
at the film, you see a long series of frames or stills, each of which depicts
the action in the film at a different instant. This static series of images cor-
responds to the B- series. Suppose you now see, far to your right, a single
20 TIME AND SPACE
image shining brightly in full colour. An instant later the image has returned
to black and white, but now the next image is illuminated, and then the
next. This fleeting change in the intrinsic properties of individual images
corresponds to the present moving along the B- series. Not all the intrinsic
properties of the image change (e.g. shapes do not), but others do.
Our own experience confirms this general picture. We know perfectly well
what some intrinsic properties are like when they are present: the properties
we are aware of in our own conscious experience. We all know what pain
feels like. A pain that exists in the past is not being felt, and so lacks this
phenomenal quality, and the same applies to sounds, colours and textures
as they feature in experience. Of course, we do not know what the intrinsic
character of a past experience is like, simply because as soon as an experience
ceases to be present it is no longer experienced. But we can be confident that,
whatever this intrinsic character is like, it is different from that of present
experiences. As for material things, although we know quite a lot about their
causal and structural properties (e.g. the shape, size and mass of an elec-
tron, how an electron reacts with other particles), we have no knowledge of
their non- structural intrinsic properties (e.g. what the intrinsic nature of an
electron is like in itself). Given this ignorance, we certainly cannot rule out
the possibility that the non- structural intrinsic properties of material things
undergo changes when they become present.
I am not suggesting that this is a viable position, but it may be the best available.
In any event, even if we can make sense of A- properties as intrinsic, a problem
remains.
3
2.6 The overdetermination problem
The difficulty I have in mind arises when the following three claims are combined.
1. All actual events have permanent locations somewhere along the static B- series.
2. A- properties are intrinsic properties of events.
3. A- properties are transitory: every event changes from being future to present
and then past.
Here (as elsewhere) I use the term event in a broad way, to refer to what obtains
at a given time, or over an interval of time, irrespective of whether the objects
involved undergo changes. To simplify, let us return to the idea that there is a
single intrinsic property of presentness which objects possess only when they
are present. This simplification will not affect the main point.
Consider some event E at t. On the assumption that A- properties are both intrin-
sic and transitory, E gains and then loses presentness. An enduring object, such as a
lump of clay, can gain and lose intrinsic properties: one and the same lump could be
spherical at noon and non- spherical at midnight, or red all over on Monday, black
all over on Tuesday and green all over on Wednesday (see Figure 2.1). But this sort
of change consists in possessing different properties at different times. The sort of
change we are envisaging E undergoing is not like this: E both possesses and lacks
McTAGGART ON TIMES UNREALITY 21
presentness at precisely the same time, t, for the only time that E exists at is the time
at which it occurs: t. This seems quite absurd; just as absurd as supposing that our
lump of clay could be both spherical and not spherical at precisely the same time, or
be differently coloured (all over) at the same time (see Figure 2.2).
It is true that some parts of an extended event can be present when others are
past and future, but the problem remains. Just focus on any single instantaneous
(or very brief) part of this event; this part, let us suppose, occurs at t
1
. Can this
event- slice both possess and lack presentness at t
1
? It is hard to see how it could if
presentness is an intrinsic property: to suppose otherwise would be akin to think-
ing that the same material object could be both spherical and cubic at the same
time. So- called substantial change occurs when an object loses an essential prop-
erty and so ceases to exist (e.g. an ice cube melts into liquid water). But we cannot
think of Es gaining and losing presentness in this way, for E does not cease to exist:
it exists at the same location in the static B- series when it is both future and past.
I will call this the overdetermination problem. It may seem insuper able, but in
fact there are responses to it available to the A- theorist. Whether these responses
really help is another matter.
First response: enduring times
Our problem is that, while one and the same object can have incompatible intrinsic
properties by having them at different times, it seems incoherent to suppose that a
single time (or the events at that time) can have incompatible intrinsic properties
at that very time. Posing the problem in this way suggests a solution: why not say
that a single time can possess incompatible properties in just the same way as an
enduring object; that is, by possessing them at different times? For this to be the
case there must exist an additional dimension of time, meta- time, which is such
that ordinary moments of time endure along this extra dimension. There is now
no longer any difficulty in supposing that the same (ordinary) times can change
their properties, as is clear from Figure 2.3.
The bold arrow indicates the location of the present, which moves from t
1
to t
2

and then to t
3
. This movement is rendered possible by the fact that our universe
as a whole (and so ordinary time) is a persisting entity: it endures through
Figure 2.1 Ordinary change. The same
object possessing different (and incompatible)
properties at different times.
Figure 2.2 McTaggart-style change. The same
object possessing different (and incompatible)
properties at one and the same time.
t
1
t
2
t
3
t
1
t
2
t
3
t
1
t
2
t
3
t
1
t
2
t
3
t
1
t
2
t
3
t
1
t
2
t
3
22 TIME AND SPACE
meta- time, as indicated by the vertical line on the left. The present is at t
1
(in
ordinary time) at m- t
1
(in meta- time); at m- t
2
, t
1
is no longer present, but t
2
is,
and at m- t
3
, t
2
is no longer present, but t
3
is. We are now supposing that times
(and the events that occur at them) are enduring entities; they endure through the
additional dimension of meta- time, and one and the same (ordinary) time or event
can have different intrinsic properties at different meta- times, just as an ordinary
material object can have different properties at different (ordinary) times.
But if one problem is solved, others remain. From an ontological perspective,
introducing meta- time is an extravagance: an additional dimension of reality is
being posited. Isnt this a blatant case of multiplying entities beyond necessity? To
this obvious objection the A- theorist might reply:
On the contrary. It is perfectly true that if we want to record when an event
occurred, a one- dimensional time- line suffices, but what we do not usually
recognize is that this one- dimensional representation is drastically incom-
plete: it provides no information at all about the movement of the present.
We know from our ordinary experience that there is a privileged moving
present. A one- dimensional representation of time is thus inadequate; we
need meta- time.
Introducing unobservable entities in order to explain observational data is a per-
fectly respectable scientific strategy. But even if we grant the A- theorist the rel-
evant data that our ordinary experience supports the doctrine that there is a
unique and moving present it is not clear that the proposed meta- time does the
required job, as the following objection makes clear:
Although we are supposing that different ordinary times possess present-
ness at different meta- times, since all meta- times are equally real, doesnt it
Figure 2.3 The two-dimensionalist solution to the overdetermination problem. Ordinary times
(such as t
1
, t
2
and t
3
) persist through a meta-time, and the property of presentness is possessed by
different ordinary times at different moments of meta-time. The claim that one and the same time (or event)
can possess different and incompatible properties is thus rendered intelligible, but the cost is high. The
universe acquires an additional dimension, filled with innumerable copies of the same sequence of events;
all that distinguishes these repeated histories is the location of the present, which is different in each.
M
e
t
a
-
t
i
m
e
Time
t
1
t
2
t
3
t
1
t
2
t
3
t
1
t
2
t
3
m-t
1
m-t
2
m-t
3
P
P
P
McTAGGART ON TIMES UNREALITY 23
remain the case that every moment of ordinary time possesses presentness?
In which case, the uniqueness of the now is lost. Moreover, this model of
time is not in the least way dynamic: the posited two- dimensional system
is entirely static. The proposed second temporal dimension fails to explain
what it is claimed to explain, and so is entirely without justification.
This objection has considerable force. The appeal of the meta- time model lies in
its purported ability to explain how presentness can be gained and lost; but in fact
the model yields the result that every moment permanently possesses presentness,
albeit at different moments of meta- time. Since the theory fails to explain what it
sets out to explain, there is no reason to accept it.
To meet this objection the defender of meta- time might say:
Ah, but you are making the mistake of forgetting that meta- time is itself a
temporal dimension, and as such it has its own unique present: the meta- now.
The ordinary now is that time which possesses presentness in the meta- now.
Since there is just one meta- now, and it is always moving, there is a single
privileged ordinary now, and this too is always moving.
However, this clearly solves nothing. To make sense of a unique moving meta-
now we would have to introduce a further dimension of time, meta- meta- time,
so that different meta- times could be present at different meta- meta- times. But
we are now launched on a regress that is as absurd as it is infinite: to secure a
moving present in meta- meta- time an additional temporal dimension is required;
this meta- meta- meta- time then requires its own moving now, and hence yet a
further temporal dimension, and so on, endlessly.
Second response: other worlds
Instead of positing additional dimensions of time, the A- theorist might seek to
overcome the overdetermination problem by positing additional worlds. Bigelow
has proposed a model of this sort, observing:
Not only can contrary properties attach to the same object at different times,
but also in different possible worlds. Something may be small at one time
and large in another; but similarly, it may be small at one possible world
and large at another. (Which is just to say that, although it may actually be
small, it could have been large.) Possible worlds can be used to keep contrary
properties out of each others hair. (1991: 5)
If temporal passage is real, then statements such as What is now past was present
are true. As Bigelow notes, if something was present, then it is clearly possible for
this thing to be present, which means that it is present in some possible world.
Which possible world? Not the actual world, where it is past, so some other pos-
sible world.
More generally, we can interpret the claim What is present was future, and will
be past in the following modal way: What is actually present could have been
future and could have been past. Or, in terms of possible worlds, something that
has the property of presentness in the actual world also exists in other possible
24 TIME AND SPACE
worlds where it lacks this property; in some of these other worlds it possesses
futurity, in others pastness. Since the thing has these contrary properties in differ-
ent worlds, there is no problem with overdetermination.
Figure 2.4 illustrates Bigelow s model. Each of W(orld)
D
W(orld)
H
represents
the entire history of a possible world, each of which is exactly the same in all
non- temporal respects; a, b, etc., refer to the entire state of a world at a given
moment of time (same letter, same sort of state); pastness is represented by bold
type, futurity by italics and presentness by capitals. Although each of these worlds
contains the same sorts of object, organized in the same ways, the worlds are none-
theless distinct, since the temporal properties (presentness , etc.) are distributed
over these objects in different ways in each world.
World
H
a b c d e f g H i j k l
World
G
a b c d e f G h i j k l
World
F
a b c d e F g h i j k I
World
E
a b c d E f g h i j k l
World
D
a b c D e f g h i j k l
Figure 2.4 Bigelows modal solution
to the overdetermination problem. These
possible worlds (possible universe is a less
misleading term) are exactly the same in all
respects save one: the properties of pastness,
PRESENTNESS and futurity are differently
distributed in each.
These various worlds can be ordered with respect to one another in terms of
pastness and futurity. A world such as W
G
is in the past of a world such as W
H
,
since everything that is past or present in W
G
is past in W
H
; a world such as W
E
is
in the future of a world such as W
D
, since everything that is present or future
in W
E
is future in W
D
. We can now say the following: something was true in a
world W when it is true in some world in Ws past; something will be true when
it is true in some world in Ws future. The claim that what is present was future
is true in a world W just when everything that is present in W is future in some
world in Ws past. The claim what is past was present is true in a world W just
when everything that is past in W is present in some world in Ws past. Similarly
for other formulations that express the passage of time. Bigelow goes on to define
a general earlier/later relation in terms of pastness, presentness and futurity, and
explains how his basic scheme can be modified to accommodate different con-
ceptions of time (e.g. a branching future, a non- existent past). These and other
refinements need not concern us here.
Before we can assess this modal model we need to be clear about its precise
content. Bigelow concludes his discussion thus: Time is real; or at least, time is
no less real than possible worlds (1991: 18), a remark that points to a question
for which he does not venture an answer: what is the status of these other possible
worlds? There are two main positions on this question: modal realism and modal
actualism. According to the modal realist, all possible worlds are equally real, all
exist in exactly the same way; although the inhabitants of each world regard their
own world as actual and the others as merely possible, each claim to actuality is
equally valid. The modal actualist, by contrast, maintains that only one world is
actual, only one world exists, the other possible worlds are merely possible. These
different views of possible worlds, when combined with Bigelows modal model of
passage, yield very different conceptions of time. Unfortu nately, neither concep-
tion is in the least appealing, or dynamic.
McTAGGART ON TIMES UNREALITY 25
If realism is true, there are as many B- series as there are moments of time, and
in each of these series a different moment is present. In this form, Bigelows theory
delivers a model that is, in effect, indistinguishable from the two- dimensional
meta- time that we considered earlier, and is problematic for the same reasons.
Our lives extend through an extra dimension: we are not confined to one world,
or one space time, we are scattered beings whose lives are spread across an infinite
number of four- dimensional planes, but the world we are spread across is a wholly
static world, a world free from temporal dynamism. As Bigelow himself notes the
theory [is] one which can be described in ways which make it seem like a static
structure, and hence in ways which make it seem to miss the point of the passage
of time as a dynamic process (1991: 4).
Modal realism is an extreme and highly controversial view of modality. Let us sup-
pose, as many do, that modal actualism is true. Is the situation improved? Far from
it. If only one possible world is actual, there is just one B- series and just one time that
is present. This is an extremely odd position, which has odd conse quences, however
it is interpreted. The A- theorist could say Although there is just one present, since
your current experience is clearly present, you can be sure that you are enjoying this
privileged moment. But since this entails that we only ever enjoy a single moment
of experience, it is not a position that holds great appeal. But if the A- theorist adopts
the alternative position, and says that the unique present may be located at any point
in history, their account is doubly problematic. Not only is there no passage (since the
present is stuck at a single moment), but presentness becomes an utter irrelevance:
most people (most likely all) will never live through the present moment, but their
lives are none the worse for it. Either way, the modal theory does not provide us
with a truly dynamic account of time, nor does it explain why time seems dynamic,
so there is no reason to accept it.
2.7 Consequences
Since none of the responses to the overdetermination problem hold the slightest
appeal, it seems that we must reject those conceptions of time that are vulnerable to
this problem. Versions of the tensed theory (concerning how time should be charac-
terized) that take tensed predicates to refer to intrinsic tensed properties should be
rejected. So, too, must dynamic conceptions of time such as the Moving-Spotlight
view, which posit intrinsic and transitory A- properties. Does this mean that all
dynamic conceptions of time have been refuted? Not in the least, for there are
dynamic theories that do not posit transitory tensed intrinsics: there is the Growing
Block view, according to which an event is present not by virtue of possessing a
distinctive (but transitory) intrinsic property, but by virtue of being the most recent
addition to the universe; there is the doctrine that only the present exists, a view
sometimes combined with the claim that there is a succession of presents. We will
be considering the intelligibility or otherwise of models such as these in Chapter 6.
It is interesting to note that McTaggart himself realized that there are models of
time that are clearly dynamic, but also quite different from the particular form of
dynamic model he believed (rightly, albeit for the wrong reason) to be paradoxical.
At an early stage during the exposition of his argument he writes:
26 TIME AND SPACE
Could we say that, in a time which formed a B series but not an A series,
the change consisted in the fact that an event ceased to be an event, while
another event began to be an event? If this were the case, we should certainly
have got a change.
But this is impossible. An event can never cease to be an event. It can never
get out of any time series in which it once is
Since, therefore, what occurs in time never begins or ceases to be, or to
be in itself, and since, again, if there is to be change it must be change of
what occurs in time (for the timeless never changes), I submit that only one
alternative remains. Changes must happen to the events of such a nature
that the occurrence of these changes does not hinder the events from being
events, and the same events, both before and after the change.
(1908: 45960)
He goes on to claim that the only properties an event can come to possess and then
lose while remaining the same event are A- properties such as pastness, presentness
and futurity.
What is interesting here is that McTaggart clearly appreciated that someone
might argue that the universe is dynamic by virtue of the fact that events come
into existence or depart from existence. Yet he rejects this view out of hand with
little or nothing in the way of supporting argument. Why? Because he took it for
a given that our ordinary conception of time comprises both the A- series and the
B- series, and the B- series consists of a sequence of events that are all equally real
and permanent.
But this is moving far too quickly. While there is no denying that history books
and memories provide us with information about what occurred in the past, and
calendars and diaries mention future times, facts such as these do not in themselves
entail that the past and future are just as real as the present, nor that most people
believe as much. Indeed, most people believe future events have no existence at
all, and many believe that past events cease to exist when they cease to be present.
These beliefs may turn out to be incoherent, but they are beliefs that many hold.
By simply assuming that the B- series is real, and hence that times and events
cannot be created or annihilated, McTaggart is guilty of begging the question
against those dynamic conceptions of time that maintain the opposite. In effect,
his starting premise is that the world is fundamentally Parmenidean. He then
considers whether such a world can also possess a dynamic Heraclitean character,
while retaining its essentially timeless Parmenidean nature, and, not surprisingly,
he concludes that it cannot. Strangely, or perhaps not so strangely, contemporary
Parmenideans, such as Mellor, who argue that McTaggarts A- paradox deals a
lethal blow to all dynamic theories, do not draw attention to this point.
3 The Block universe
3.1 Time without passage
Many contemporary philosophers are convinced that McTaggart was essentially cor-
rect: our world is a block-like four- dimensional ensemble, lacking a moving present,
wherein all times and events are equally real. But few (if any) of these philosophers
follow McTaggart by holding that time is unreal. According to these contemporary
Block- or B- theorists, what we call time is simply that dimension which exists
in addition to the three of space and in virtue of which the properties of things
can change; a thing changes by having different properties at different locations
in this additional dimension. Since not everything happens at once and things do
change, there clearly is such a dimension, and so time is undeniably real. McTaggart
concluded that time was unreal because he believed that time essentially involves
a moving present, and that no such thing could exist. Modern B- theorists take a
different tack, and while agreeing with McTaggart that there cannot be a moving
present, they argue that this amounts to a discovery about the real nature of time.
We are accustomed to advances in knowledge overturning commonly held beliefs
about how things are. The Earth does not seem to move, but it turns out that it
does; time seems to pass, but it turns out that it does not. Where is the difference?
This chapter and the next are given over to an examination of the Block view of
time. I am not assuming that this theory is true. As we saw in the preceding chapter,
while there are powerful arguments against the view that A- properties are transi-
tory intrinsics, there are also dynamic models of time that take a different form,
and we cannot conclude that the B- theory is true unless and until these alternative
accounts all prove untenable. However, there are good reasons for taking a closer
look at the B- theory before considering these wider issues.
Of all the models of time that we will be considering, the B- theory is currently
the most popular among philosophers and physicists, but in many respects it is
also the strangest and most counterintuitive.
1
Most people believe that time passes;
most people believe that there is an objective difference between past, present
and future. B- theorists maintain that these beliefs are false. Like anybody seeking
to defend a view that runs counter to common sense, the B- theorist is faced with
two tasks, one positive and one negative. The negative task is to provide reasons
for thinking that the common- sense beliefs in question are false. McTaggarts
A- paradox provides one such reason, but we shall see that other B- theorists are
more inclined to lean on scientific considerations. The positive task is more com-
plex. B- theorists must provide us with a clear account of the nature of time and
28 TIME AND SPACE
change, but they must also convince us that this account is one that could be
true. Most people believe that time passes because it seems to: the belief is not
groundless; it is not sustained merely by custom or tradition; there are features of
our experience, features of the world itself, that provide seemingly good reasons
for holding the belief. Hence the B- theorist must explain why it is that so many
of us have false beliefs about the nature of time, and show that, surprising as it
may seem, nothing in our everyday experience requires us to hold that there is an
objective difference between past, present and future. This is no mean task, and
although it will occupy us for this chapter and the next, we will be returning to
it in Chapter 7.
Just as different B- theorists take different approaches to the negative task, they
also approach the more complex positive task in different ways. In several parts of
this chapter I will be drawing on Mellor (1998), a revised edition of his classic Real
Time (1981). But although Mellors writings on this topic have been influential,
not all B- theorists agree with all aspects of his position, and on occasion I will
indicate some of these divergences.
3.2 Passage and experience
Common sense tells us something like this: only the present is fully real; the future
is wholly non- existent and as yet unfixed (in the sense that it is not yet determined
what will happen), whereas the past is fixed but not real in the same way as the
present; the present is constantly advancing. This picture of time is imprecise and
on reflection puzzling (what, precisely, is the status of the past?), but it is also very
appealing. Call it the natural view.
It is not surprising that we find this view appealing, for it seems firmly grounded
in the character of our everyday experience. Someone trying to describe the most
general temporal features of everyday experience might well come up with some-
thing like the following:
We have a good deal of highly detailed knowledge of what happened in the
past, but little or none about what will happen in the future (except for the
case of highly stable and regular systems whose behaviour can be accurately
predicted). We can anticipate experiences that we will have in the future, and
we can remember experiences that we had in the past, but only present expe-
riences are fully and properly real, only sensations that are occurring now
are actually felt. The passage of time is something we can directly observe
in the world: we can see one event happening after another (e.g. we can
observe a car moving along the street passing one house, then another, then
another). Deliberation and choice are oriented towards the future. I am now
(as always) in the position of having to decide what I will do, never what I
was doing. We cannot change or affect the past, but by choosing to do one
thing rather than another we can help determine what happens in the future.
If this is how we are inclined to describe our experience, the appeal of the
natural view of time is readily understandable: the two fit together perfectly. The
idea that only present events are fully real is supported by the confinement of
THE BLOCK UNIVERSE 29
experience to the now. The fact that we can perceive things changing, and find
the experienced present to be continually advancing into the future, supports the
idea that the present is continually advancing into the future. That we have epis-
tem ic access to the past of a sort that we do not have to the future supports the
fixed pastunfixed future doctrine.
That everyday experience makes the appeal of the natural view readily under-
standable is not in question. What is in question is whether there is anything here
that is truly incompatible with the competing B- conception of time. B- theorists
argue that there is not. I will start by considering what the B- theorist can say about
two aspects of the natural view that concern experience.
The presence of experience
Consider first the idea that we can only experience what is present. If we take
this to mean we can only perceive what is present then the claim is simply false.
Light is fast, but not infinitely fast. When you look at the Sun you are seeing it as
it was eight minutes or so ago; when you look at a star you may well be seeing it
as it was millions of years ago. As Mellor notes, whether a thing is past, present
or future doesnt make any difference to how it looks: We cannot for example
refute someone who claims to see the future in a crystal ball by pointing to the
visible pastness of the image: there is no such thing. Our reasons for thinking we
cannot see the future rest not on observation but on theory (1998: 16). This truth
(obvious when our attention is drawn to it) points the way to a more promising
formulation of this confinement doctrine: the objects we perceive may not be
present but the sensory experiences we have when we perceive them surely are.
Conscious experiences are confined to the present.
But are they? Let us suppose (with the B- theorist) that all events are equally
real. It follows that all experiences are equally real: your pain five days ago is just
as real as your pain today, and your pain next week. From this perspective, it is
clearly not true that experience is confined to the present. What is true is this
straight forward tautology: an experience happens when it happens. Every experi-
ence occurs at some time, and is restricted to the time at which it occurs; but this
is true of everything that is located in time. It doesnt follow from this tautology
that the only experiences are the experiences occurring at the time in question.
This would only follow if earlier and later events are unreal. So it seems that the
doctrine that experience is confined to the present doesnt support or confirm
the doctrine that past and future are unreal; it presupposes this doctrine. Even
so, there is something that needs explaining. Why do we feel that only present
experience is really felt, and hence that experience occurs only in the present?
Imagine waking up to find yourself in a strange place. You are sitting in a field
of grass, next to a lamp that illuminates the surrounding area. There is complete
silence. As you look around, you can see nothing whatsoever. Apart from the
small patch of grass illuminated by the lamp there is darkness everywhere. Not
surprisingly, you conclude that you are alone. You could not be more wrong. A
few yards to your right there is another lamp, and another person waking up to
find themselves surrounded by total darkness; likewise to your left. In fact, you are
in a line of people stretching for many miles in either direction, all of whom are
30 TIME AND SPACE
sitting in their own small pools of light, all of whom are alarmed to find themselves
alone in a strange place.
Why is it that nobody can see anyone else? The answer lies with the strange
form of light emitted by the lamps, which only extends a few feet before dying
away. According to the B- theorist, we find ourselves in an analogous position in
time. What stretches only a short distance is not light through space but conscious-
ness over time: the temporal span of direct awareness is very brief. And as in the
analogous spatial case, the fact that at any given time we are not aware of experi-
ences occurring at other times does not mean that these experiences are not there.
Figure 3.1 Illusory spatial solitude. Since it is dark, and light emitted by the lamps fades away after
travelling only a couple of feet, no one can see their neighbours, so everyone concludes they are alone
in the field. But they are wrong.
Im alone!
Nobody here
but me . . .
Where is
everybody
else?
Feeling
lonely

Figure 3.2 Illusory temporal solitude. According to the Block theorist, the impression we have that
reality is confined to a very brief present is (in part) due to the fact that the span of our direct awareness
(or specious present) is very brief. The lived present of a being with a wider span of awareness (such
as the alien depicted here) would include times we would regard as past and future.
Only this
now is
real!
Only this
now is
real!
Only THIS
now is
real!
Only THIS
now is
real!
Time
3.15p.m. 3.20p.m. 3.25p.m. 3.30p.m.
Experiencing passage
How can the B- theorist account for the experience we have of time passing if
time doesnt really pass? According to many B- theorists, Mellor included (1998:
122), the solution lies in the ways our memories accumulate. Suppose you are in
a dentists waiting room. Every few minutes you take an anxious glance at the
clock. Although the minute hand of this clock looks stationary, it is in a different
position every time you look at it. Your first glance shows it to be 10.15. When you
THE BLOCK UNIVERSE 31
next look you see it is 10.25, a while later you see that it is 10.29. Is this all you
experience? Very probably not: in addition to what you perceive, you also have
various memories of what you have recently been doing. So, for example, when
you see that it is 10.25 you also remember seeing the clock showing it to be 10.15;
when you see the clock showing 10.29, you remember observing it showing 10.25
while also remembering having seen it showing 10.15. According to the B- theorist,
our sense that time is passing is due to the fact that our perceptions and memories
combine in this sort of way, not only over the span of a few seconds or minutes,
but over days, months and years.
On the face of it this account seems quite plausible. Vary the scenario, and
suppose you are suffering from Korsakoff syndrome: you forget everything that
happens to you within a minute or so. When you see the clock showing 10.25
not only do you have no memory of having seen it showing 10.15, but you have
no recollection of what you were doing two minutes earlier. Likewise when you
see the clock showing 10.29. In this memory- impoverished condition, what sense
would you have that time is passing?
Memory also has a part to play in explaining why we feel that our lives unfold
in the direction of the future. Even if our future experiences are just as real as our
past ones, as the B- theorist maintains, at any given point during our lives we have
detailed knowledge only of the earlier parts of our life; indeed, for all we know
we might be about to drop dead, and so not have a future at all. It is thus easy to
believe that our lives dont extend beyond the present, even if in fact they do. This
belief leads to another: that any continuation of our lives involves the addition of
something new to the already completed past. Since we remember having these
same beliefs at past stages of our lives, and anticipate having them in the future,
we always have the impression of living at the outermost limit of an ongoing
future- directed process.
Whether this memory- based account of experiencing passage is entirely ade-
quate especially over the very short term is something we will be looking into in
more detail in Chapter 7. It is enough for now to recognize that the B- theorist has
at least the beginnings of a plausible account, one that does not require a moving
present or the non- existence of times other than the present. All that it requires is
for people to have different memories at different stages of their lives.
3.3 A- truth in a B- world
There is a further reason why it is natural to believe that there must be an objective
difference between past, present and future, one that we have not yet considered.
Quite generally, when a claim about the world is true, there is something about
the world, a fact that obtains, that makes the claim true. This fact is the claims
truthmaker. If I say Canada is larger than France, the relevant truthmaker
is the fact that Canada extends over a greater area of the Earths surface than
France does. Suppose I say The battle of Hastings is in the past. I am making an
A- statement, and this statement is clearly true, but what is its truthmaker? The
obvious candidate is the tensed A- fact that Hastings took place in the past. Like-
wise for other A- statements. Not all A- statements are true, but many are. If
32 TIME AND SPACE
those that are true have tensed A- facts as truthmakers, it seems that we live in an
A- world after all.
How can the B- theorist respond to this? Simply denying that there are any true
A- statements is bold, heroic even, but hopelessly implausible. One response, which
dates back to Russell (1915), is to hold that A- statements do not require A- facts as
truthmakers because they are equivalent in meaning to B- statements. If this were
the case, the absence of any A- facts would not matter, since there are B- facts, and
these could serve as truthmakers for both sorts of statement.
This approach is nowadays widely known as the old tenseless theory, and is
generally agreed to fail: A- statements cannot be translated into B- statements with-
out loss of meaning. Even though this is generally agreed to be true, it is worth
pausing to consider some of the difficulties involved in such translations. It seems
clear that two statements cannot have the same meaning if one can be true when
the other one is false. Consider:
S
1
: Event E is past.
S
2
: Event E is earlier than t.
S
1
is a typical A- statement; S
2
is a B- statement. It is easy to see that these can have
different truth- values: if S
2
is true it is true at all times, whereas S
1
is only true when
E is in the past. A- statements have time- varying truth- values, they are true at some
times but not at others, whereas B- statements have time- invariant truth- values.
This alone seems to show that A- and B- statements differ in meaning. But there
are further difficulties. Suppose the following is uttered at 3.25p.m.:
S
3
: Bill is happy now.
We might try to translate S
3
by the B- sentence:
S
4
: Bill is happy at 3.25p.m.
There are two reasons for rejecting the claim that S
3
and S
4
have the same meanings.
First, the two sentences employ different concepts. Since someone might under-
stand now but have no grasp at all of what occurs at 3.25p.m. means, how could
S
3
and S
4
have the same meaning? Secondly, someone could understand both S
3
and
S
4
, but nonetheless believe S
3
to be true while believing S
4
to be false. How is this
possible? Simple. The person in question sees that Bill has a broad grin on his face,
but (mistaken ly) believes that it is 12 noon.
Advocates of the new tenseless theory of time, such as Smart (1980), Mel-
lor (1981, 1998) and many others, agree that A- claims cannot be translated into
B- claims, but argue that this doesnt matter: A- claims can nonetheless have tense-
less B- truthmakers (or truth- conditions). Here is how Mellor has recently stated
the case.
Mellor distinguishes between beliefs, statements of beliefs, the sentences that
are used in stating beliefs and the contents of beliefs, which he calls propositions,
and takes to be what sentences expressing beliefs mean. Thus when Jack says Bill
is happy and Jacques says Bill est heureux, they are using different sentences
to make the same claim or statement; that is they are using different sentences to
express the same proposition, and have beliefs with the same contents. On the face
THE BLOCK UNIVERSE 33
of it, it is not easy to see how an A- belief could have a B- truthmaker. A- beliefs
are true at some times but not at others, whereas B- beliefs always have the same
truth- value. For example:
B
1
= Jim races tomorrow.
B
2
= Jim races on 2 June.
B
2
is always true but, assuming that Jim does indeed race on 2 June, B
1
is only true
on 1 June. Clearly, since the truth- value of any given B- belief is constant across
time, whereas an A- belief such as Jim races tomorrow is true at some times
and false at others, no single B- fact can make such an A- belief true. How many
B- beliefs does it take? The answer, it turns out, is a great many: (roughly speaking)
a different B- fact is required for each different time at which a given A- belief can
be true or false. How can this be? What are the relevant B- facts? How can a single
A- belief have many different B- truthmakers?
The solution relies on the type- token distinction. A token of a type is a particu-
lar instance, specimen or exemplar of it. For example, Mary is a token of the type
humanity; the and the are two tokens of the same type of word, the. Just as
the type- token distinction can be applied to words, it can be applied to beliefs,
sentences and statements.
Our problem was that a given A- belief type can have different truth- values at
different times, so no A- belief type can have a single B- truthmaker. However,
when we switch our attention to tokens of A- beliefs a way forward opens up. A
token of an A- belief is a particular person holding a particular belief at a particular
time. The fact that a person holds a given belief at a certain time is a B- fact, and
hence holds at all times. Consequently, all we need is a way of stating the condi-
tions under which a token A- belief is true in tenseless B- terms. There are two
general strategies available, which Mellor sketches in these terms.
The token- reflexive account: For every A- proposition P about any event E, any
token of P is true if and only if it is as much earlier or later than E as P says
the present is than E.
The date account: For every A- proposition P about any event E, any token of
P is made true at any t by ts being as much earlier or later than E as P says the
present is than E.
Both accounts seem to give the right results. Consider the A- proposition E
occurred yesterday. My saying E occurred yesterday is a token of this proposi-
tion. The first account tells us that this A- statement is true if and only if E does
indeed occur a day earlier than the token itself. (It is because this account makes
reference to the token itself that it is token- reflexive.) The second account says
that, if I make this statement on 2 June, then what I am saying is true if and only
if E occurs on 1 June. Analogous considerations apply for the A- belief E is occur-
ring now. According to the token- reflexive account, a token of this belief is true
if and only if it occurs when E occurs (i.e. the token itself is neither earlier nor
later than E). The date account yields the result that, if the belief is tokened on
2 June, it is true if and only if E also occurs on 2 June. Underlying these differ-
ences is a crucial similarity: both accounts state the conditions under which tokens
34 TIME AND SPACE
of A- propositions (and hence A- beliefs, A- statements, etc.) are true in entirely
B- terms.
In Real Time Mellor preferred the token- reflexive account; in Real Time II, he
prefers the date theory to handle certain tricky cases that critics have developed
in the intervening years. We dont need to go into these details here. The important
point is that the B- theorist has a plausible strategy for explaining how A- statements
and beliefs can be true even if there are no A- facts to make them true.
3.4 Another A- paradox
Mellor takes a further step in arguing that, if we try to supply A- sentences (state-
ments, beliefs, etc.) with tensed truthmakers, we end up in a paradoxical situation,
and consequently we have no option but to accept that A- statements have tenseless
truthmakers. More specifically, Mellor argues that, if tensed sentences have tensed
truth- conditions, the statements that we make using these sentences will undergo
changes in their truth- value, changes that do not in fact occur.
By way of an example, suppose that on Monday I see a man with short hair and
I mistakenly believe that this man is John, who has worn his hair long for many
years. I exclaim, So, John has short hair now. Call this claim C. In actual fact, at
this time John has long hair, so C is false. Suppose that it is now a month later and
John has just taken the plunge, visited the barber and emerged with short hair. Is
C now true? Of course not. The claim that I made a month ago remains false. If
I were now to claim John has short hair now, my claim would be true, but the
claim I made a month ago, that very claim, was false then and is false now. Mellors
allegation is that, if tensed statements have tensed truthmakers, then statements
such as C would change their truth- value. The argument runs thus.
Suppose a tensed sentence S = E is happening now has a tensed truthmaker,
the obvious candidate being the fact that E is now happening. The tensed theorist
will thus say:
Any token of S = E is happening now is true if and only if E is happening
now.
In other words, all tokens of the tensed sentence S, irrespective of when they
occur, are made true by this tensed fact, and hence, if this fact does obtain, then
all tokens of S are true, irrespective of when they occur. Suppose that on Monday
I say It is now Tuesday. Call this statement S
T
. Applying the tensed account,
this statement is true if and only if it is now Tuesday. Since it is now Monday, the
statement is false. However, a day later, on Tuesday, the truthmaker for S
T
now
obtains, so the statement in question is now true: S
T
has undergone a change of
truth- value!
Mellor is certainly right in maintaining that we do not normally suppose that
truth- value changes of this sort occur. Moreover they dont occur if the truth-
makers of tensed statements are of the tenseless sort considered above: for exam-
ple, if a token of It is now Tuesday is true if and only if the token in question
is uttered on a Tuesday, then a token of this sentence uttered on Monday remains
false even on Tuesday. So the idea that tensed claims are made true by tensed facts
THE BLOCK UNIVERSE 35
does look to be in difficulty. Whether this difficulty is insuperable is another mat-
ter. In 5.1 we will encounter a proposal for a cure.
3.5 The indispensability of the A- framework
If there are no A- facts why do we constantly think in A- terms? Mellor argues that
the A- framework, in particular A- beliefs, are of vital practical importance even
though there are no A- facts.
Beliefs contribute to our actions; what we do is a product of what we desire
or want and what we believe. If I want an ice- cream, and believe that there is ice-
cream in the fridge, then I will go to the fridge to get some ice- cream; my belief
helps direct my action in the way appropriate to satisfy my wants. Of course, my
wants will only be satisfied if there really is ice cream in the fridge: successful
action requires true belief.
Now, it is very often the case that the success of an action depends on when it is
carried out. If I want to buy some food from the shops, then my visit to the shops
will only be successful if I go when they are open. I have to act at the right time.
Suppose that I believe (truly) that the shops are open between 8a.m. and 8p.m.
Since this B- belief is always true, it cannot successfully guide my actions: success
depends on my going to the shops at some times but not others. Clearly, the belief
that will successfully guide me must be true at some times but not others; it must be
an A- belief. An example of the right sort of belief would be It is now 3p.m. The
belief If it were 3p.m. then the shops would be open, a B- belief, will not do, for
although this belief is true, it is always true, and so it cannot guide timely action.
To act at the appropriate times we must have A- beliefs that are constantly
updated to reflect the different times at which they are held. I should believe The
shops are open now between 8a.m. and 8p.m. but not at other times. I should
believe The banks are closed today only on days when the banks are indeed
closed. How do we update our A- beliefs? Usually by using our senses. To put it
crudely, we believe what we see, and (cases of known time- lag apart) we believe
that what we see is happening as we see it; happening now. If I look at my watch,
and it says 4p.m. then I believe It is now 4p.m. and act accordingly.
Again, there is a good deal more to be said about these topics, but the important
point is that we can understand why the A- framework is indispensable for practical
purposes without assuming that there are any A- facts. The account just sketched
appeals only to B- facts and B- relationships. A- beliefs play an ineliminable role in
successful timely actions, but, as we have seen, B- facts can serve as truthmakers for
A- beliefs.
3.6 Questions of attitude
Experiences that we are currently undergoing are of more concern to us while we
are undergoing them than experiences we have yet to have, and experiences we
have already had. This is understandable: if I am now in pain, this present pain is
real to me (now) in a way that my past and future pains are not. But we also have
36 TIME AND SPACE
differing attitudes to past and future experiences. Although I suffered horribly at
the dentist last week, this fact doesnt much bother me at all. If someone were
to say Give me 10 and I will change the past, and make it so that pain never
occurred I would not be inclined to pay, even if I believed that the past could
be changed in this way. But if someone were to say Give me 100 and I will
ensure that you feel no pain when you visit the dentist next week, I might well
be tempted by the offer, especially if this were the only way of avoiding the pain
in question. Quite generally, our future experiences are of far greater concern to
us than our past experiences (especially where the experiences in question are of
the painful and pleasurable variety). Perhaps the most dramatic example of our
bias towards the future is Lucretiuss observation that while death (i.e. our own
future non- existence) is something we deeply dread, the vast periods of personal
non- existence preceding our births is of no concern to us whatsoever.
Some A- theorists have held that this asymmetry in our attitudes is problematic
for the B- theory. The best- known of these arguments originated with Prior (1959),
and runs thus. After several long hours of torment, a fierce headache at last fades
away, and you think to yourself Thank goodness thats over! What is it that you
are expressing relief about? Could it be the tenseless fact that this token thought
occurs later than the headache in question? Although according to the B- theorist
it is this tenseless fact that makes the thought true, this answer seems plainly false:
the trouble is that this was as much a fact before and during the headache
as it is now the headache is over. It always was and always will be a fact that
this particular token of Thats over occurs later than the headache it refers
to. What is more, that fact could have been recognized as such in advance. In
particular, you could have decided in advance to say Thats over after the
headache and known about the fact in that way. So if that were the fact that
you were thanking goodness for, you could just have well thanked goodness
for it before or during the headache as afterwards. Which, of course, is non-
sense. So it seems that you must be thanking goodness for some other fact,
something that was not a fact at all until the headache ceased. The tensed
conclusion appears irres istible; the pastness of the headache, for which you
are thanking goodness, must be an extra fact over and above the tenseless
fact that makes Thats over true. (Mellor 1994: 296)
A straightforward response to this objection is to agree with Prior that the fact
that you are thanking goodness for is indeed a tensed rather than a tenseless fact
(Mellor himself recognizes that there are tensed facts, if only in the trivial sense
that every true statement states a fact), but also to hold that the relevant tensed
fact is identical to (or refers to the same state of affairs as) a tenseless fact. As
for why we thank goodness for the tensed fact but not the tenseless fact, there is
a simple explanation. Lois Lane believes that Superman is a superhero, but she
does not believe that Clark Kent is a superhero, yet Superman and Clark Kent are
one and the same person. Our beliefs are attitudes to objects (or states of affairs)
that are represented in a certain way, and the same object (or state of affairs) can
be represented differently. The same holds, it seems reasonable to suppose, for
attitudes such as dread and relief. Trapped in a burning building, Lois Lane would
THE BLOCK UNIVERSE 37
feel relieved if she thought Superman were about to arrive; the thought that Clark
Kent were about to arrive would leave her indifferent.
2
However, recognizing that the same object (or state of affairs) can be repre-
sented differently only solves part of the problem. It is obvious why Lois Lane has
the attitudes she does: she believes Superman to be a superhero, and Clark Kent
to be a mild- mannered reporter. It is not obvious why we tend to care more about
states of affairs when they are characterized in tensed rather than tenseless terms.
Why is it that we have such different attitudes to the past and future?
Do we care more about our future experiences because we can still do something
about them, whereas the past is beyond our control? No. We dread future pains
that we believe are inevitable in just the same way as we dread future pains that we
believe are avoidable (imagine yourself in prison, facing inescapable torture this
afternoon). As Parfit says, Our ground for concern about such [inevitable] future
pains is not that, unlike past pains, we can affect them. We know that we cannot
affect them. We are concerned about these future pains simply because they are
not yet in the past (1984: 168, original emphasis).
That we care more about future pains just because they are future may seem to
pose a problem for the Block theorist. If there is no ontological difference between
past and future if all experiences are equally real what could justify our bias
towards the future? However, while it is far from obvious how this bias could be
justified in the context of the Block view, this is only a problem for the B- theorist if
the bias in question is justified. Perhaps it is rooted in nothing more than irrational
instincts implanted in us by natural selection. It is easy to sketch the outlines of
a suitable evolutionary story: no doubt natural selection has ensured that we are
motivated to act in ways that will enhance our survival prospects; since we can
only act in ways that influence our futures, we have an overriding concern with
the future compared with the past.
But there is a second point. Irrespective of how successful the B- theorists
attempts to explain the bias are, it is by no means clear that adopting a tensed or
dynamic conception of time changes the situation. Why should we care less about
pains that were present than pains that will become present? What is it about past-
ness that makes experiences less worthy of concern?
B- theorists can also take encouragement from the fact that closer exami nation
reveals our bias towards the future to be rather less straightforward than it
initially appears. As Gallois points out (1994), our bias is not towards the future
per se. Were you to travel back in time to the last ice age, the thought of enduring
the terrible cold might well fill you with dread; but the experiences that you are
concerned about in this case lie in the past and not the future; or rather, they lie
in the worlds past but your personal future, a reference frame determined by the
ordering of the stages of your life (see 8.2 for more on the distinction between
personal and world times that time travel generates).
Parfits case of the Past or future suffering of those we love reveals an ambigu-
ity of a different sort:
I am an exile from some country, where I have left my widowed mother.
Though I am deeply concerned about her, I very seldom get news. I have
known for some time that she is fatally ill, and cannot live long. I am now
38 TIME AND SPACE
told something new. My mothers illness has become very painful, in a way
that drugs cannot relieve. For the next few months, before she dies, she faces
a terrible ordeal. That she will soon die, I already knew. But I am deeply
distressed to learn of the suffering that she must endure.
A day later I am told that I had been partly misinformed. The facts were
right, but not the timing. My mother did have many months of suffering, but
she is now dead. (Parfit 1984: 181)
If past pains are of no significance I should be relieved. But am I? Obviously not.
The tragedy for my mother is just as great; she suffered just as much from her
illness, and I am just as upset on her account. So it seems that where other people
are concerned, especially other people whom we love, the timing of any pains they
suffer is of little significance.
The existence of this striking selfother asymmetry poses a problem for anyone
seeking to justify the bias towards the future in the first- person case. Might it
not be that the pattern of concern that we display towards other people is more
rational than our self- oriented concern? More rational because less influenced by
the instinctive drive for self- preservation instilled by natural selection? If so, it
may well be unreasonable for us to care greatly about our own future experiences
while being largely indifferent to our past experiences: the rational pattern of self-
concern (if such there be) may well be entirely in tune with the Block view of time.
3
3.7 B- theories of change
In its crudest form, the B- theory of change is easy to state: an object changes by
virtue of having different properties at different times. McTaggart objected that
a pokers being hot at t
1
and cold at t
2
doesnt amount to change because B- facts
such as these always obtain. In response, the B- theorist can ask, Why cant some
changeless facts be changes? If the poker is hot at t
1
and cold at t
2
, why cant these
facts not constitute a change in the poker? Does it matter that these facts are and
always will be facts? McTaggarts response was to deny that the B- theorist has
any right to talk of times at all in the absence of A- change. Since this is precisely
the assumption that B- theorists reject, this response does not trouble them at all.
However, while agreeing on these fundamentals, B- theorists find much else to
disagree on when it comes to making sense of change.
Objects (trees, people, houses, etc.) are often thought to be different from
events (concert performances, conversations, battles, etc.) because of the kind of
parts that each has. Since a battle is typically spread over a fairly large area it has a
good many spatial parts (i.e. what is going on at different places at a given time),
but it also has temporal parts. To put it crudely, it has a beginning, a middle and
an end, and these different phases are as much parts of the battle as any of its spa-
tial components. Ordinary material objects seem to be different. While a typical
person such as Alice has various spatial parts (her hands, feet, bones, organs, etc.),
it seems wrong to think she has temporal parts in the same way as a battle. Alice
exists now, and she existed yesterday, but it seems that Alice in her entirety exists
at both these times. If you met Alice you didnt meet just part of her, you met the
THE BLOCK UNIVERSE 39
whole thing, Alice herself. Alice- yesterday and Alice- today are not two parts of a
person, they are each a whole person in their own right; moreover, they are the
very same person: Alice- yesterday and Alice- today are numerically identical (the
very same object) even though they are very likely qualitatively different (they
dont have exactly the same properties).
This distinction between objects (or substances, continuants) and events comes
very naturally to us, and some B- theorists, such as Mellor, not only subscribe to
it, but they deploy it in defending themselves against McTaggarts charge that the
B- theory reduces change to mere spatial variation. Mellor points out that change
is something that a single persisting thing undergoes. Events cannot change in this
way. Suppose a performance of a symphony is loud at t
1
and quiet at t
2
. What we
have here are two distinct entities, two temporal parts of the symphony, possess-
ing different properties. Likewise, Jims being happy yesterday and Alices being
sad today doesnt amount to a change; it is a case of two distinct objects having
different properties. Alices being happy yesterday and sad today is a change.
Entities such as Alice can change precisely because they do not possess temporal
parts. Now consider a typical case of spatial variation. If a stick is red at one end
and white at the other, this isnt change; rather than one and the same entity (the
whole stick) possessing different properties, we have two distinct parts of the stick
possessing different properties that is, two numerically different entities. If we
were to say that the stick as a whole is red and white we would be saying something
false, since there are parts of the stick that arent red and parts that arent white.
Hence we have a clear difference between change and spatial variation: the latter
involves numerically distinct entities possessing different properties, whereas the
former involves numerically identical entities possessing different proper ties.
Mellor takes a further step. He argues that persistence involves causation. Alices
present condition is (in part) a causal consequence of her preceding condition. More
generally, if O at t
2
is numerically identical with O at t
1
, then O at t
2
must be causally
dependent upon O at t
1
, directly or indirectly. Since causation operates over time,
but not space (Mellor argues against the possibility of causes being simultaneous
with their effects), we have a further respect in which genuine change differs from
spatial variation. Change requires persistence, persistence requires causation; since
time alone is the dimension of causation, time alone is the dimension of change, and
hence the allegation that the B- theory obliterates the distinction between time and
space (or change and mere variation) is successfully rebutted.
Mellors account of change aims to be non- revisionary. In effect he is saying
something along these lines: The B- theory of time may strike you as very odd
when you first encounter it, but it is not that odd, because it turns out that you
can adopt it without having to undertake a radical overhaul of your current con-
ceptual scheme. Other B- theorists take a quite different line. They start off with
the assumption that if our ordinary ways of thinking about time are mistaken it
may well be that other parts of our ordinary conceptual scheme are mistaken too.
They then argue that if the B- theory is true some natural assumptions about what
persistence involves turn out to be false, and as a result a revisionary account of
change starts to look very appealing. Their reasoning runs as follows.
Think of a spatially extended physical object, such as a worm. Suppose that this
worm has ten segments; some of these segments may be very similar to each other,
40 TIME AND SPACE
but nonetheless each is a distinct object (a worm- segment) in its own right. If we
think of a persisting object from the four- dimensional perspective of the B- theory
a similar picture emerges. The spacetime diagram in Figure 3.3 represents Alice
over a period of time.
The diagram depicts Alice as an extended four- dimensional object. This object
is composed of parts. Some of these parts are of the familiar spatial sort: at t
1
Alice
can be divided into a head, two arms, two legs and a torso. But isnt it clear that
Alice also has other parts too? To be sure, Alice exists at t
1
; but since she also exists
at t
2
(and other times too) it is surely wrong to say that Alice- in- her- entirety exists
at t
1
. If you see Alice at t
1
you arent seeing the whole Alice, you are simply seeing
the part of her that exists at that time (and place). She has plenty of other parts,
existing at other times, and these you dont see. The common- sense assumption
that when you see a person on a given occasion you are seeing the whole person,
and hence that the person is wholly and entirely present before you, would be true
if only the present time were real, and many of us do commonly assume this. But
if we adopt the B- theory, this assumption is false. Since all times are equally real,
people and other persisting things do not exist only in the present; they extend
into the past and future too.
If persisting objects do have temporal parts, the difference between change and
spatial variation that Mellor stresses evaporates. If Alice- at- t
1
and Alice- at- t
2
have
different properties, these properties are possessed not by one and the same object,
but by two numerically distinct objects (person- stages). These objects are indeed
parts of a single thing (the whole temporally extended Alice), but then so, too, are
the red and white ends of a stick, or the front- end and the rear- end of a worm.
This reasoning can be resisted, but I wont enter further into the twists and
turns of this particular debate, which is primarily of importance in the metaphys-
ics of identity and object- hood; for more on these topics see Sider (2001). I want
t
1
t
2
LATER
DOWN
EARLIER
LEFT RIGHT
UP
Figure 3.3 The four-dimensionalist view of persistence. A brief phase of Alices life is shown.
On the traditional view of persistence, Alice-at-t
1
and Alice-at-t
2
are numerically identical. The four-
dimensionalist views matters differently: like all persisting objects, Alice has temporal parts, and these parts
are numerically distinct objects; Alice is identical with the sum of these parts. Consequently, if you were to
meet Alice at t
1
, you would not be meeting Alice-as-a-whole, but merely one of her parts.
THE BLOCK UNIVERSE 41
to mention instead a different (and arguably more profound) way in which the
B- theory can be revisionary to greater or lesser extents.
3.8 Emergent time
If we choose to think of ordinary objects as four- dimensional entities possessing
temporal parts, change becomes more akin to spatial variation, but does it become
spatial variation? Not if the temporal dimension of the four- dimensional block-
universe is different in some way from the three spatial dimensions, different in
a way that renders the distinction between time and space well founded. Many
B- theorists assume that there is such a difference. Some (like Mellor) believe that
causality runs only along the temporal axis. Others, sceptical of causality as a
time- distinguishing feature (see 4.5 and 4.6), look instead to physics. Although
relativity theory posits a four- dimensional manifold (or continuum of points), this
manifold is not a four- dimensional space, and the distinction between space and
time is maintained (albeit in an altered form).
4
But this may change. Relativistic gravitational theory and quantum theory have
yet to be combined in an entirely adequate way, and the search for a satisfactory
theory of quantum gravity continues. Some physicists believe that the distinc-
tion between space and time will not survive in quantum gravity theory, for the
simple reason that this theory will not mention time at all.
5
Rovelli writes, In
quantum gravity, I see no reason to expect a fundamental notion of time to play
any role The time along which things happen is a notion which makes sense
only for describing a limited regime of reality At the fundamental level, we
should, simply, forget time (1999: 13). This is only well- informed speculation
there are plenty of physicists who take a different view but suppose that Rovelli
turns out to be right. Belot and Earman suggest the metaphysical consequences
would be considerable:
there is no change [in such a world] since no physically real quantity evolves
in time change is illusory This is a very radical thesis it is, for instance,
much stronger than the doctrine that there is no room for temporal becoming
in a relativistic world, since proponents of the tenseless theory of time are
confident that they can account for the existence of change. (1999: 178)
If change is illusory, then so, too, is time, or so McTaggart would argue, and it
would be hard to disagree.
But does it follow from the as yet only hypothetical disappearance of time as
an independent variable from the equations of fundamental physics that change
is unreal? Only if we take fundamental physics to be our sole guide in deciding
which (if any) dimension of the universe can legitimately be called time. Not all
B- theorists would want to take this step, as this passage from D. C. Williamss
well- known exposition of the Block theory clearly displays:
it is plain that things, persons and events, as a matter of natural fact, are
strung along with respect to the time axis in patterns notably different from
those in which they are deployed in space. The very concept of things or
42 TIME AND SPACE
individual substances derives from a peculiar kind of coherence and elon-
gation of clumps of events in the time direction. Living bodies in particular
have a special organized trend timewise, a conatus sese conservandi, which
nothing has in spatial section. Characteristic themes of causation run in the
same direction, and paralleling all of these, and accounting for their impor-
tance and obviousness to us, is the pattern of mental events, the stream of
consciousness, with its mnemic cumulation and that sad anxiety to keep going
futureward which contrasts strangely with our comparative indifference to
our spatial girth. An easy interpretation would be that the world content is
uniquely organized in the time direction because the time direction itself is
aboriginally unique. Modern philosophical wisdom, however, consists mostly
of trying the cart before the horse, and I find myself more than half convinced
by the oddly repellent hypothesis that the peculiarity of the time dimension is
not thus primitive but is wholly a resultant of those differences in the mere de
facto run and order of the worlds filling. It is then conceivable, though doubt-
less physically impossible, that one four- dimensional area of the manifold be
slewed around at right angles to the rest, so that the time order of that area,
as composed by its interior lines of strain and structure, runs parallel with a
spatial order in its environment. It is conceivable, indeed, that a single whole
human life should lie thwartwise of the manifold, with its belly plump in time,
its birth at the east and its death in the west, and its conscious stream perhaps
running alongside somebodys garden path. (1951: 468, original emphasis)
The speculation that ends this passage is striking, and for sheer oddness it has few
rivals in the literature of philosophy.
6
And this is why it serves Williamss purposes
so well. The observable (and introspectible) contents of the world are not spread
through the four- dimensional continuum in an entirely haphazard fashion. There
are distinctive patterns of development and organization that are only found along
the temporal axis, so much so that when asked to envisage one of these patterns
a whole human life, from birth to death being distributed across space in the
way that it is usually distributed across time, we simply dont know where to start.
There may be universes where this does not hold where it is common to find
peoples streams of conscious ness running alongside roads and garden paths and
in such universes there might well be no rationale for maintaining or imposing a
general (as opposed to a person- or object- specific) distinction between time and
space, but our universe is not like this. Given the asymmetrical ways in which the
contents of our universe are distributed, the distinction between the temporal and
spatial dimensions is manifestly well founded, even if the distinction between time
and space is not primitive or aboriginal (and so not registered by fundamental
physics), but depends entirely on the de facto run and order of the worlds filling.
In saying that the distinction between space and time is well founded I mean
that it is grounded in objective facts concerning the asymmetrical patterning and
distribution of events (we will be taking a close look at these asymmetries in the
Chapter 4). This is undeniably so at least at the macroscopic (and introspect-
ible) levels in this region of our universe and future developments in physics are
simply irrelevant to the distinction drawn in this way. To this extent time is real
rather than illusory, and the same applies to change.
THE BLOCK UNIVERSE 43
This said, if it does turn out that time is absent from the equations of quantum
gravity theory (or whatever theory in physics turns out to be fundamental), this
will be an interesting and significant result. There will be a clear sense in which
time is not a fundamental feature of the universe, especially if it turns out that
the basic ingredients of the physical world are quite capable of existing without
giving rise to the asymmetrical formations that ground the spacetime distinc-
tion. If this should prove to be the case, time will have the status of an emergent
phenomenon, in somewhat the same way as macroscopic material things such as
planets and animals (which emerge only when elementary particles are arranged
in certain ways). But even if the status of time is thus diminished, its existence is
scarcely threatened, for at the macroscopic level the distinction between spatial
and temporal dimensions remains perfectly real. Cats are emergent entities; are
they illusory?
According to Belot and Earman, the doctrine that time is not physically funda-
mental is stronger and more radical than the doctrine that temporal becoming is
unreal. This is questionable. As far as departures from our ordinary ways of think-
ing are concerned, it is passage (or becoming) that matters: the (metaphysical)
thesis that time does not pass, that all moments are equally real, is truly astonishing
when first encountered; almost impossible to believe. The idea that the spacetime
distinction, although real enough at the macro- level, is emergent relative to the
most fundamental level of physical reality, is far less radical in this respect. If time
can survive in the metaphysical framework of the Block universe, it can certainly
survive discoveries from the frontiers of physics.
4 Asymmetries within time
4.1 The direction of time
A French nursery rhyme runs Leau est comme le temps: il coule, il coule, il
coule. If time really did flow, then, like a flowing river, it would also have a
direction. But if the B- theory is true, time is not going anywhere: there is no mov-
ing now, no slippage into the past, no crystallization from mere possibility into
present actuality. Every time and every event is equally real. But why does time
seem to have a direction if it doesnt? The B- theorists explanation will take this
general form: time does not flow or pass, but there are nonetheless asymmetries to
be found among the contents of time, the material and mental processes we know
to exist, and it is these asymmetries that we mistake for temporal flow and that
provide time with its apparent direction (cf. the passage from Williams in 3.8).
Whether or not these content- asymmetries suffice to give time itself a direc-
tion depends on how we view time. If the substantivalist is correct, and time is
an entity in its own right over and above any contents it might have, then there
is a sense in which time might be perfectly symmetrical in itself, and so lack any
directedness, even if its contents do not. By way of an analogy, imagine looking
at a glass cylinder that gradual ly changes in colour from blue to red; you assume
that the glass itself has an in- built colour asymmetry, but you are wrong: the glass
is perfectly clear throughout (and so entirely symmetrical), but embedded within,
a short way beneath the surface, is an asymmetrically coloured sheet of paper.
This option is not available if we follow the relationist in identifying times with
collections of simultaneous events: asymmetries in content now amount to asym-
metries within time itself.
But there are two further complications to bear in mind. Time as a whole can be
asymmetrical even if its contents are not. For example, rather than being infinite in
both directions, it could be infinite in only one; or it could be infinite in both direc-
tions, but be linear in one and branching in another; or it could branch in both
directions, but in different ways. Secondly, if we ignore these large- scale questions
and assume a linear time, there could be content- asymmetries that are only local,
and so not of a kind that can provide a direction to time as a whole. To return to
our earlier analogy, suppose that our cylinder is blue at both ends, turning to red
in the middle, with exactly similar gradations in shading in both directions. If the
content- asymmetries within time were of this overall form, then although most
people would find time in their neighbourhood to be directed, time as a whole
would lack a direction. We could call one end of this (finite) universe the start
ASYMMETRIES WITHIN TIME 45
and the other the end, but since the universe is perfectly symmetrical about its
centre, the opposite choice would be equally justified.
We will be encountering a universe of this last sort later in the chapter but,
this case aside, we will not be concerned with the more esoteric possibilities just
mentioned. Our focus will be on the observable content- asymmetries to be found in
everyday life. Thus far we have looked only very briefly at just one family of content-
asymmetries: those that contribute to what is sometimes called the psychological
arrow. We do have the impression that we live our lives in the present, and that
the present is constantly advancing into the future. As we have seen, B- theorists
can appeal to the factors such as changing A- beliefs and patterns of memory-
accumulation to explain why it is that we have this impression. How ever, there are
other aspects of the psychological arrow, and further asymmetries within time that
we have not yet tried to explain.
If all conscious beings were to vanish tomorrow there would be nobody left
around to have the subjective impression that time is passing, but there are good
reasons for believing that plenty of asymmetries within time would remain, even
if no one were around to notice them. Using time- lapse photography, the events of
weeks or years can be squeezed on to a film lasting only a few minutes. Suppose that
someone has made a time- lapse film of what has been going on in your garden over
the past few months. Would you have any difficulty in knowing if the film were run-
ning forwards or backwards? Of course not. If the film were running backwards you
would see large plants shrinking rapidly down to small seedlings, smudges of slush
concentrating themselves into ripe plums that suddenly leap upwards and attach
themselves to the branch of a tree, paw- prints gradually appearing in mud prior to
the arrival of a cat walking backwards, rings of water converging into the centre of
the pond from which a ball rises and flies over the fence into the next garden, and
so forth. A film running backwards shows things occurring in the wrong order; later
events are shown as occurring before earlier events.
Given that we can so easily tell whether a film registering everyday sequences
of events is running backwards or not, there must be certain sequences of events
that are always (or typically) ordered in the same way with respect to earlier and
later. This is why, for a world such as our own, it seems reasonable to think that
there is an objective difference between earlier times and later times, a difference
that does not depend on the psychological peculiarities of conscious agents such as
ourselves. In this chapter we will be examining what B- theorists have to say about
asymmetries of this sort. While all B- theorists agree that there are some objective
asymmetries, there is a significant difference of opinion as to precisely what these
are. As we shall see, some B- theorists defend the provocative view that many of
the content- asymmetries we are inclined to think of as objective are not: they are
products of the distinctive perspectives that we bring to bear on a world that is
more symmetrical, in temporal respects, than commonly supposed.
4.2 Content- asymmetries: a fuller picture
Before starting to consider what the B- theorist can say about the various content-
asymmetries, it will help to have more of them clearly in view. Some of the
46 TIME AND SPACE
asymmetries listed below have already been mentioned; others we have yet to
encounter. The brief characterizations provided are only intended to give a rough
idea of what each asymmetry involves. They should not be taken to be definitive
characterizations (nor should the list be taken to be complete).
1
Entropic asymmetry: the second law of thermodynamics states that entropy (or
disorder) increases over time.
Causal asymmetry: some events are causes of others; causes usually occur ear-
lier than their effects.
Fork asymmetry: it is very common to find many later events that are cor-
related with (or caused by) a single earlier event; for example, everyone in a
coach party who had fish at lunch falls ill a few hours later. The inverse fork
(one later event correlated with or causing many earlier events) is much rarer.
Explanatory asymmetry: we tend to explain later events by reference to earlier
events (Jim died because he drank poison, he didnt drink poison because he
died).
Knowledge asymmetry: although we can often make accurate predictions
about what will occur, we have detailed and reliable knowledge of the past of
a kind that is not available for the future.
Action asymmetry: our deliberations are oriented towards the future, not the
past (no one spends time wondering about what to do yesterday); what we do
can influence the future, but not the past.
Experience asymmetry: we experience our lives unfolding in a present that
inexorably moves forwards in the direction of the future.
As will immediately be clear, these asymmetries are not wholly independent; at
least some (perhaps all) are interrelated. But quite how they are related is not so
obvious, and we will see that different B- theorists have different views not only
about the precise pattern of interrelationships, but about which are fundamental
and which depend on others. Before proceeding, it is worth pausing. What exactly
is it that the B- theorist has to explain? A B- theorist might say:
I dont have to explain any of this. Perhaps it is just a brute fact that the con-
tents of time are distributed and related in asymmetrical ways. In any case,
what does the A- theorist have to offer by way of an explanation of these asym-
metries? If temporal passage could exist, it could exist in worlds organized very
differently from our own, in worlds where there are no content- asymmetries
at all, or where there are asymmetries, but they are differently oriented with
respect to past and future. Positing an objective difference between past,
present and future doesnt explain anything; nor does temporal passage.
This response is by no means absurd it may well be true that there are some basic
facts that cannot be explained in terms of other facts but the B- theorist should
only adopt it as a last resort. First, if the B- theorist could explain some or all of the
content- asymmetries, the B- theory itself would be more believable. Secondly, and
more importantly, A- theorists can provide, at least in rough outline, explanations
of at least some important asym metries.
Finding the result of last weeks lottery is easy: just pick up a newspaper. Find-
ing the result of next weeks lottery is not so easy. Past events leave their marks on
ASYMMETRIES WITHIN TIME 47
subsequent presents, but future events do not leave their marks on earlier presents;
this is why our knowledge of the past is far more detailed than our knowledge of
the future. If, as some A- theorists maintain, the future is unreal, this asymmetry
is easily explained: since future events do not exist, they clearly cannot causally
influence or leave traces on anything; past events can and do. This is why we do
not find instances of backward causation, or inverse forks. It is why our actions can
influence the future, which is not yet fixed, but not the past, which is. It is why we
explain later events by reference to earlier events: when an event is present, there
are no future events to influence it.
Can the B- theorist offer anything remotely as plausible by way of explanations
of these phenomena?
4.3 Entropy
It has often been claimed that a crucial key in understanding the direction of time
lies in thermodynamics, and more specifically with the second law of thermo-
dynamics, which dictates that entropy tends to increase over time. Some go as far
as to claim that the entropic asymmetry is the ultimate key: it alone is responsible
for all the other content- asymmetries. Atkins writes:
We have looked through the window onto the world provided by the Second
Law, and have seen the naked purposelessness of nature. The deep structure
of change is decay; the spring of change in all its forms is the corruption of
the quality of energy as it spreads chaotically, irreversibly, and purposelessly
in time. All change, and times arrow, point in the direction of corruption.
The experience of time is the gearing of the electrochemical processes in our
brains to the purposeless drift into chaos as we sink into equilibrium and the
grave. (1986: 98)
To be in a position to evaluate this claim we need to take a closer look at what the
second law of thermodynamics actually amounts to. Since the complexities here
are considerable, I have been obliged to simplify somewhat.
Entropy is a technical concept that can be defined in different ways, but the
general idea is easily grasped. Roughly speaking, entropy is a measure of the extent
to which the energy in a system has spread out in a disorderly fashion through the
space available to it, and hence of how close a system is to its equilibrium state. If
a systems energy has largely dissipated (i.e. it is more or less evenly spread out)
it is in a high entro py state; if, by contrast, its energy is concentrated in just a few
places, it is in a low entro py state. The second law of thermodynamics thus states
that in any closed system (i.e. no outside influences and no leakage) the energy
within that system will tend to spread out, until finally it is evenly distributed
through the available space, and having reached this condition of equilibrium it
will remain there. In our universe, energy is concentrated in stars separated by
vast extents of (very cold) space, so the universe is currently in a low- entropy
condition. But since the stars are pouring their energy into the surrounding space
at an impressive rate, and thus spreading it around, the entropy of the universe is
increasing, in accordance with the second law of thermodynamics.
48 TIME AND SPACE
The entropic account of the earlierlater distinction is in many respects analo-
gous to the gravitational account of the updown distinction:
For Aristotle, the down direction is a primitive notion. He probably believed
that at all points of space a downward direction existed, and that all of
these downs were in the same spatial direction. But now we realize that
down is just pointing in the direction in which the local gravitational force
is pointing. We now understand that there are regions of the universe in
which no direction is down and none is up, and we accept without difficulty
that the downward direction for someone in Australia is not parallel to that
for someone in New York. That is how it is with the futurepast distinction,
Boltzmann claims. Where there is no local entropic asymmetry, there is no
futurepast of time. (Sklar 1992: 149)
As Sklar goes on to note, the gravitational account of down is plausible because
it explains all the relevant facts about the downward direction: if you drop an egg
it falls down; gas filled balloons float upwards (due to air pressure created by grav-
ity); we can feel the difference (due to gravitational forces acting on mechanisms
within our inner ears), and so on. Does the entropic account of earlier and later
similarly explain the relevant facts?
It can certainly explain why some of the processes that allow us to tell whether
a film is running backwards or not unfold as they do. If you see a film showing
leaves blowing about in the wind and ending up in a neat pile in the middle of the
lawn, you can be reasonably certain that the film is running backwards: since this
course of events is highly improbable, it is much more likely that the leaves were
initially piled up by a human gardener, as human gardeners commonly do such
things. What is far less clear is whether any of the other asymmetries listed in 4.2
can be explained in the same sort of way. Would a reversal in entropy also reverse
the direction of causation and explana tion? Would machines (such as clocks) run
in the opposite direction? Would we have detailed knowledge of (what are now)
future happenings? Would the various components of the psychological arrow
reverse their direction? It is certainly not obvious that they would; far from it.
The first point to note is that thermodynamic processes are not in fact irre-
versible. In the course of pioneering work in the nineteenth century, Boltzmann
attempted to reconcile the time- asymmetric second law of thermodynamics with
the time- symmetric laws of classical mechanics. He showed that the tendency for
entropy to rise could be explained in terms of the statistics governing particle
interactions. However, as critics were quick to point out, his explanation also
applies in reverse: even if it is over whelmingly probable that the entropy of a
typical system of particles will always increase, does it not follow that there is
some probability, no matter how small, that the entropy of such a system will at
certain times decrease? This objection first came to Boltzmanns attention in the
1870s, and was reinforced in 1889 by Poincars recur rence theorem, according
to which the particles in any energetically isolated state will, given enough time,
return to a state that is arbitrarily similar to their initial state, and do so infinitely
often. Simplifying a complex story, Boltzmann took the reversibility objection
on board and accepted that so- called irreversible processes are not truly irrevers-
ible after all. Energy in a system does tend to spread out (this is by far the most
ASYMMETRIES WITHIN TIME 49
probable way for a low- entropy system to evolve), but on occasion (very rarely)
concentra tions of matter and energy will occur, and entropy will decrease. But
only for a short while; these improbable concentrations will soon break down,
and equilib rium will be restored. More generally, although there is a tendency
for disorder to increase, there will be exceptions; very rarely, bizarre things will
happen (e.g. milk stirred into tea will separate out again, leaves will be blown by
the wind into neat piles). More generally still, any system that at a given time is in
a high (and highly improbable) state of order is likely to be in a more disordered
(and more probable) condi tion at later times and earlier times.
These considerations do not preclude our defining the direction of time in terms
of entropy, but they do complicate matters. Boltzmann took the view that what
we call the universe was quite likely to be only a tiny part of a far larger whole,
most of which is at thermodynamic equilibrium, or very close to it. From the
vantage point of the universe as a whole, the observable stars amount to nothing
more than a tiny and localized eruption of massively improbable (but statistically
inevitable) order in a vast sea of disorder. As the stars fling their energy into space,
our region of the universe is in the process of returning to equilibrium. Hence,
over our region taken as a whole, entropy is increasing, despite occasional and
improbable small- scale reversals. Boltzmann suggested that this regional entropic
increase provides time with its direction (earlier = lower entropy, later = higher
entropy). Since the universe as a whole is already at equilibrium point (or very
close to it), its average entropy is neither increasing nor decreasing, and time is
without a direction; it is only at the regional scale (e.g. clusters of galaxies), where
significant entropy gradients are to be found, that time has a direction. In recent
years, with the advent (and widespread acceptance) of the big bang cosmology,
most scientists take the view that our universe started off in a low- entropy condi-
tion, and that, with the exception of the occasional small- scale local reversal, its
entropy has been increasing ever since.
2
Leaving these complications to one side, the entropic account of the direction
of time faces another serious problem. What exactly is the connection between
the tendency for entropy to rise and the other observable content- asymmetries?
What is the link between the laws of statistical thermo dynamics (concerning par-
ticle interactions and energy distributions) and asymmetries involving causation,
knowledge and explanation?
As Earman points out (1974), applying the thermodynamic concept to systems
as diverse as decks of playing cards, human beings, printed records, and so on,
involves a considerable extension of the concept beyond its established range of
legitimacy, and there is no guarantee whatsoever that the extension is legitimate.
Moreover, if all the other content- asymmetries were the direct result of local
entropy increase, then we would expect a local decrease in entropy (which is
perfectly possible dont forget) to reverse the other asymmetries. But there is no
reason at all to believe that this would occur. There are many ways in which the
entropy of a region can decrease, and most of these ways seem very unlikely to
have any significant effects on any macroscopic structures (such as people and
mountains) located within the region in question.
For example, suppose that (due to a freakish sequence of molecular motions)
the water in the eastern half of the Atlantic Ocean were to become, on average,
50 TIME AND SPACE
20C warmer than the western half, a process that takes two weeks. Throughout
this period the entropy in the Atlantic would be decreasing. What effect would
there be on its occupants? Well, no doubt a great deal of plant and animal life
would be significantly affected by the dramatic and unseasonal changes in tem-
perature; there would be serious meteorological consequences too. But there is
no reason to believe that any thing else of interest or significance would happen.
Life on board ship might become more difficult due to extreme weather condi-
tions, but it would not start to run in reverse: radio and video transmissions from
the shore would seem perfectly normal (rather than running backwards), clocks
would continue to turn in the same direction, and so on. The same considera-
tions seem to apply quite generally. One way for entropy in a particular region
of outer space to increase is for the particles that were previously distributed in
a uniform fashion (e.g. one atom per cubic metre) to become concen trated in
just one half of it, leaving the other half completely empty. As the latter region
gradually empties, entropy within it is decreasing. If you were to fly a spaceship
into this region, is there any reason to suppose you would find yourself living in a
time- reversed fashion? Again, there is no reason to think so. Since you are enter-
ing a region of space that is almost devoid of matter and energy, and you may well
not interact with anything at all, why should anything on your ship change in any
way whatsoever?
Or consider a more specific case: the knowledge asymmetry. Since a footprint
on a sandy shore will invariably belong to an earlier rather than a later walker, we
reasonably take it as a reliable indicator of what has happened in the recent past.
Can we explain this in terms of the tendency of entropy to increase? Reichenbach
argued that we could. A high entropy beach (i.e. nearly flat, no distinct imprints)
is much more probable than a low entropy beach (one featuring some distinctive
patterns, such as a footprint). So if we encounter a footprint in the sand, it is
reasonable to conclude that this is not due to a random fluctuation from the high
entropy norm, but rather the product of a previous interaction with some external
influence, such as a human foot. This sounds reasonable enough, but why assume
(as Reichenbach does) that the external influence occurred in the past rather than
in the future? This is a perfectly reasonable assumption, since experience has
taught us to expect that any footprints we happen to observe are the product of
earlier rather than later strollings along beaches. But if we are seeking to explain
the direction of time in terms of entropy increase, we cannot simply assume that it
is more probable that the external influence occurred earlier rather than later. The
entropic account must explain why there is this asym metry. But it seems unable
to do so.
3
These considerations suggest at least this much: that there is no direct or obvi-
ous connection between entropy increase and many of the asymmetries that we
normally associate with the direction of time. Of course, if a defender of the
entropic account could provide a detailed and plausible account of how, despite
appearances to the contrary, local entropy increase does in fact explain the other
asymmetries, the situation would be quite different. Some B- theorists are opti-
mistic about the prospects for developing an account along these lines (we will
be encountering a variant of this approach shortly). Others do not share this
optimism, and adopt a very different approach.
ASYMMETRIES WITHIN TIME 51
4.4 The causal route
Of the various ways of explaining the content- asymmetries available to the
B- theorist, there is one that stands out by virtue of its straightforward simplicity:
the causal approach. Consider again the footstep we see on the beach. Why do we
take this as evidence of an earlier rather than a later occurrence? An appealing line
of reasoning runs thus. We know that the footprint is unlikely to have occurred
spontaneously, which means that it must have been created by something else.
To put it another way, the footprint is the effect of some other event, which was
its cause. Since we know that causes precede their effects, we conclude that the
event responsible for the footprint occurred at an earlier (rather than a later) time.
Likewise for all the traces and records that provide us with detailed knowledge
of past occurrences: these are all effects, and since effects always occur later than
their causes, there can be no records or traces of future events. By appealing to
causation we can easily explain an asymmetry that seems highly problematic for
the entropic account.
We can explain much else too. Why do we tend to explain later events in terms
of earlier events? Because explaining an event often consists in citing its cause,
and causes come earlier than their effects. We can perceive past occurrences but
not future ones because to perceive something is to be causally affected by it, and
causes occur earlier than their effects. Memories too can plausibly be viewed as
causal imprints, so the fact that causes precede their effects explains why we have
memories of the past but not the future. Actions have consequences; effective
actions have desired consequences; if actions are the causes of their consequences,
then since an actions effects must occur later than the action itself there is no
mystery in why our actions can influence the future but not the past, and why our
deliberations are future- directed rather than the reverse.
So far so good: it seems that most (perhaps all) of the content- asymmetries
can easily be explained in causal terms. We can now take an additional step and
explicitly define the direction of time in causal terms. Roughly speaking, we can
hold that if E
1
causes E
2
then E
2
occurs later than E
1
precisely because it is an effect
of E
1
. We cannot simply say for any two events, E
x
precedes E
y
if and only if E
x

causes E
y
, since many events are not effects of the events that precede them. All
events are ordered in time but relatively few are causally related to one another.
Mellor, who himself defends a causal account of the direction of time, suggests a
solution along these lines. Take two times, t
1
and t
2
. A lot is going on at these two
times, and many events at t
1
are not causally related to many events at t
2
, yet many
events at t
1
are causally related to events at t
2
, so we can say that t
1
is earlier than t
2

provided that some events at t
1
are causes of events at t
2
. Given the complexity of
the universe, this will always be the case. In short, the temporal order of all events
can follow from the causal order of only some.
4
4.5 Causation in question
This is all wonderfully simple and straightforward, so much so that it may come
as a surprise to discover that many B- theorists (perhaps a majority) find the causal
52 TIME AND SPACE
account untenable. To appreciate why, we need to probe more deeply into what
defenders of the causal account need to establish for their account to be viable.
The claim that for E
1
to be earlier than E
2
it is sufficient that E
1
causes E
2
would
be undermined if it turned out that backward causation were a possibility.
5
In
cases of backward causation, the effect occurs earlier than the cause. Hence, if we
also say that a cause occurs earlier than its effects, we have a situation in which
one event is both earlier and later than another. At the very least, admitting the
possibility of backward causation considerably complicates the causal theorists
task. Whether backward causation is possible (or actual) is a controversial issue,
but many B- theorists are unwill ing to dismiss it (not least because the most obvi-
ous objection to it future events do not exist, and so cannot cause anything is
unavailable to them).
A second problem is no less serious. Unless causal theorists can distinguish
between the events that are causes and those that are effects without appealing to
the earlierlater distinction, their account of the direction of time will be circular,
and so trivial. It will not do to say something along these lines:
Causes and effects are events of kinds whose instances regularly occur in
succes sion, and do so by virtue of the laws of nature rather than mere coinci-
dence; of these pairings, the cause is the earlier of the two, the effect the later
of the two.
According to Hume, causes precede their effects only because there is a linguistic
convention of using the word cause to refer to the earlier instances of event-
types that are known to be regularly paired in the appropriate ways. To avoid the
charge of triviality, the causal theorist must reject the Humean view, and provide
a substantive account of causal priority (i.e. how causes differ from their effects)
that does not appeal to temporal priority (i.e. it will not do to say that causes differ
from their effects by occurring earlier than them).
It might seem that there is no real difficulty here. Surely causes make their
effects happen; effects dont make their causes happen. Hence effects depend on
their causes, in a way that causes dont depend on their effects. Since this notion of
dependency is asymmetrical and (seemingly) non- temporal, we have all we need.
However, on closer examination it turns out to be far from obvious that an
asymmetrical dependency relationship of the required sort really exists. As noted
earlier, it may well be that the laws of physics are time- symmetrical; if so, any type
of event that can be a cause can also be an effect, and vice versa. Moreover, in any
given concrete situation where a cause c produces an effect e, if we say that in the
given circumstances, c is both necessary and sufficient for e, then since the relation-
ship necessary and sufficient is symmetrical, we are clearly committed to holding
that e is necessary and sufficient for c. For example, suppose a spark ignites a fire.
In the circumstances (oxygen present, water absent, no other flames, etc.), the fire
would not have occurred if the spark had been absent, so the spark was necessary
for the fire. It is also true that, in the circumstances, the spark was sufficient for the
fire (nothing else was needed). Now look at the situation in another way. In these
circumstances, for the fire to ignite the spark had to occur, so the fact the fire did
ignite guarantees that the spark occurred: the fire is sufficient for the spark. Also,
since the occurrence of the spark guarantees the fire will ignite, the spark couldnt
ASYMMETRIES WITHIN TIME 53
occur without the fire, so the fire is necessary for the spark. The two events, the
cause and the effect, are thus related to one another in a perfectly symmetrical
way. If this applies generally, what could possibly distinguish any cause c from an
effect e other than the fact that c occurs earlier than e?
Causation is a complex topic; it may well come in different forms, and there
are many competing accounts of it. See Dainton (2009) for an overview of the
main contenders. The only thing that is clear and relatively uncontroversial is that
finding an objective and non- temporal difference between cause an d effect is a far
harder task than one might have imagined, and so a good many philosophers have
concluded that there is no such difference. In which case, the causal asymmetry is
not something that exists in reality, but something our modes of thought project
on to reality, in somewhat the same way as (many would argue) we project colour
on to the physical things we perceive, things that in themselves do not possess
colour- as- we- perceive- it.
If the causal relation is inherently symmetrical, causal accounts of the direction
of time (of the earlierlater relation) cannot succeed; or rather, they are at best
circular and so trivial. However, as we saw earlier, the causal theorist can provide
simple and appealing explanations of the other content- asymmetries (e.g. knowl-
edge). What are we to make of these explanations? As we shall now see, even if
we assume that causation is an inherently asymmetrical relation, its explanatory
potential is suspect.
4.6 Time in reverse
The universe is currently expanding, and it may continue to do so for ever, but
equally it may not. If there is enough mass in the universe to halt the current
expansion (recent evidence suggests not, but the issue remains open), at some
point in the future gravitational forces will cause the universe to fall back on itself.
The big bang will be followed by a big crunch.
Big bangbig crunch models fall into two classes. There are models where
entropy is low at the big bang and keeps on increasing right up to the big crunch
(which is thus far less ordered than the big bang), but there are also models where
entropy increases only until contraction kicks in, at which point it gradually
decreases, and the universe becomes more orderly as it contracts. It is quite pos-
sible for a universe of the latter sort to be highly symmetrical: at either end is a
singularity (a dimensionless point of infinite density), and the sequences of events
running from each singularity to the mid- point are very similar. Highly symmetri-
cal universes of the second type are sometimes called Gold universes, after the
physicist Thomas Gold, who raised the profile of such models in the 1960s.
6
Let us imagine that we live in a Gold universe. If we could observe the post-
contraction phase it would be like looking at a film running backwards; we
would see light and energy pouring into stars rather than out of them, people
would die before they were born, and trees would start off as rotting logs, rise
up to towering heights and then dwindle into seedlings. This might seem utterly
impossible: how could the laws of nature permit such bizarre happenings? Very
easily. As noted earlier, many believe that the laws of funda mental physics are
54 TIME AND SPACE
time- reversible, in which case any physical process can run in reverse.
7
But in
addition we must take care not to assume that our way of looking at things is in
any way privileged. The inhabitants of the other side will regard their half of
universe as expanding and ours as contracting; what we call the end of the uni-
verse, they would call the beginning; what we regard as inflationary expansion
they would regard as gravita tional implosion. If they were to peer over into our
side, it would seem to them that they were looking at a film running backwards:
from their perspective it would seem that time in our half of the universe is run-
ning in reverse. They, too, would wonder how such a thing could be possible,
and their perspective is just as valid as our own.
This scenario opens up some intriguing topics. Could we causally interact with
our (seemingly) time- reversed neighbours? Could we communicate with them? It
might seem not. Suppose we transmitted a sequence of messages M
1
, M
2
, M
3
and
M
4
. From the perspective of our neighbours, M
4
would be the earliest message
and M
1
the latest; consequently, if they respond to M
4
, depending on the mode
of communication adopted, we may well receive the response before we have
transmitted the message itself! However, as MacBeath (1983) shows, with suf-
ficient ingenuity, these hurdles can be overcome (I wont go into details but will
provide a clue: the solution involves programming a computer to transmit ones
initial messages at later times). In a later paper MacBeath suggests that if we say
that a space has n dimensions if there are n respects in which its occupants can,
qua occupants of that space, vary continuously but independently (ibid.: 186),
then time- reversal of the sort we have been considering actually involves time itself
possessing two dimensions rather than one. But interesting as they are, these issues
cannot be pursued here.
Suppose that the causal account of the direction of time were true. Would the
time in the other half of a Gold universe have a direction opposite to what it has
in ours? It would if the arrow of causation pointed in a different direction in the
other half, but there is no reason to believe that this would be the case: if (as we
are currently assuming) causation is an intrinsically asymmetrical relation between
events, why should its direction be reversed just because the universe starts to
contract? So let us assume that causation runs in just one direction from one end
of the universe to the other: from our big bang to the big crunch. Something rather
Figure 4.1 A time-symmetrical Gold universe. A universe that is bounded at both ends, and that is
broadly (not necessarily exactly) symmetrical about its temporal mid-point. The inhabitants of the left-hand
side regard A (their big bang) as lying in their past, and B (their big crunch) as lying in their future; the
inhabitants of the right-hand side regard B (their big bang) as lying in their past, and A (their big crunch)
as lying in their future.
ASYMMETRIES WITHIN TIME 55
disturbing now emerges. It is possible for the apparent direction of time to become
completely disconnected from its real direction. Not only are the inhabitants of
the other half of our universe very similar to us in all respects, but so, too, is
the world they inhabit. The various content- asymmetries (with the exception of
causation) all run in the same direction: away from their big bang, in the direction
of increasing entropy.
I will call the apparent direction of time the world- arrow. Our time- reversed
counterparts exist in a situation where their world- arrow points in the opposite
direction to the causal- arrow. Of course, they would be completely oblivious to
this fact, since, as is now very obvious, the direction of the causal- arrow has no
observable consequences whatsoever. More disturbing still, for all we know our
world- arrow might be pointing in the opposite direction to the causal- arrow! If it
were we would be none the wiser, nor any the worse off.
This thought- experiment does not refute the causal theory; it does not prove
that events are not in fact related by an asymmetrical dependency relationship.
But it does a good deal to undermine whatever intuitive support the causal theory
in this form has, for it brings into sharp focus how little the theory explains: the
apparent asymmetries within time, the asymmetries that determine the direction
of our world- time, may well be entirely independent of the causal- arrow. If so, we
must look elsewhere for an explanation of the temporal asymmetries that we find
in the observable world.
We can draw a further lesson from the thought- experiment. The causal theo-
rists explanations of the content- asymmetries may be suspect, but the fact that
these expla na tions are plausible (at least until examined more closely) suggests that
there is indeed an intimate connection between our ordinary concept of causa-
tion and these other asymmetries: the causal- arrow and the world- arrow point
in the same directions, and it is hard to believe that this alignment is contingent.
Wouldnt our time- reversed counterparts be perfectly justified in thinking that the
direction of causation in their part of the universe runs in the direction they call
earlier- to- later? If this is right, if the causal- and the world- arrow must point in
the same direction, we can ask why? One answer would be that the causal- arrow
determines the direction of the world- arrow, but since this answer is suspect, we
must consider the other option: perhaps it is the world- arrow, or some ingredient
of it, that determines the causal- arrow.
8
4.7 Fundamental forks
In his Asymmetries in Time (1987), Paul Horwich embarks on the ambitious task
of providing an account of the relationships between all the familiar temporal
asym metries within the context of a block-universe. What is striking about the
account he defends is the very lowly status causation occupies within it: see Figure
4.2 (a slightly modified version of Horwich 1987: 201). Rather than being the
asym metry that explains all others, causation comes towards the very end of the
hierarchy of explanatory priority.
For Horwich, the knowledge asymmetry is the key to understanding all those
other asymmetries depicted further to the right; he argues that the knowledge
56 TIME AND SPACE
asymmetry itself is (largely) a product of the fork asymmetry, whereas the latter
exists because of the special conditions that obtained in the early universe.
To bring this chapter to a close I will provide an outline of some of the main
ele ments in Horwichs explanatory scheme. The account, as Horwich himself
concedes, is not entirely adequate as it stands. In addition to certain speculative
components (inevi table, given the current state of physics), a good deal of detail
requires filling in, and some parts are more persuasive than others. Nonethe-
less, by virtue of its treatment of causation, Horwichs approach to the content-
asymmetries is quite different from those that we have considered thus far, and
this alone makes it worth examining.
9
Forks
Since the fork asymmetry plays such an important role in Horwichs scheme we
had best start here. He describes the asymmetry thus: Highly correlated event
types are invariably preceded by some unified common cause, but they need not
Figure 4.2 Horwichs hierarchy of explanatory priority.
Entropy
Initial conditions
(cosmological)
Fork asymmetry Knowledge
Concept of time-order
Explanation
Causation
Counterfactuals
Decision
Value
Freedom

Figure 4.3 The
fork asymmetry.
Time
X Y
Z
W
X Y
frequent infrequent
ASYMMETRIES WITHIN TIME 57
have joint effects. We never find a pattern as depicted [on the right of Figure 4.3]
in which the correlated events are linked only by a joint effect (1987: 73).
If X and Y are types of event whose occurrences are linked, that is they fre-
quently occur together (and so are highly correlated), we will usually find that
there is a certain type of prior event Z that causes both. The reverse of this two
highly correlated events having a typical subsequent effect but no typical prior
cause never occurs (or only rarely). So, for example, thunder and lightning are
strongly correlated types of event (they often occur together), and they have a
common prior cause (high atmospheric static charge), but there is no single type
of effect that invariably follows a lightning discharge (e.g. sometimes a tree is
destroyed, sometimes a house is hit, sometimes a field, etc.). While this all sounds
plausible enough, we cannot characterize this fork asymmetry in terms of causes
and effects if we hope to use this asymmetry to explain causation. Horwich thus
provides an alternative cause- free formulation along these lines: it is common
to find strongly correlated events at t statistically associated with a single earlier
event; it is rare to find events at t similarly associated with a single later event.
It is worth pointing out that there is a more general kind of fork asymmetry,
which is not restricted to strongly correlated events. Detonating a given type of
bomb will have very different consequences in different environments (outer space,
within buildings, under water); the effects of detonations are thus not strongly cor-
related, in Horwichs sense. Yet in each particular case the effects can be traced
back to a single cause. In such cases there is an asymmetry of over determination.
Owens characterizes it thus:
For any event, the number of events at any given point in the future for
which it is necessary is greater than the number of events at any given point
in the past for which it is necessary. (Or equivalently: For any given event,
the number of events at any given point in the future which are sufficient
for it will be greater than the number of events at any given point in the past
which are sufficient for it.) (1992: 89)
Any given detonation will typically have a single sufficient prior condition, such
as the pushing of a switch. The detonation is thus necessary for just one immedi-
ately preceding event. But it is necessary for each of its many subsequent effects,
which would not have occurred in its absence, and hence each of these effects is,
in the circumstances, sufficient for the detonation. In such cases, which are very
common, events are massively overdetermined (in Owenss sense) by later events
but not by earlier events. These overdetermining later events may be strongly
correlated, but they need not be. Bearing this point in mind, let us move on.
Causation
Horwich agrees with Hume that causation is in itself a symmetrical relationship,
and that causes precede effects only because of a convention about how these
words should be used. However, he also maintains that the relevant convention
is not arbitrary. The stipulation that causes come before their effects meshes with,
and is supported by, a cluster of other beliefs about causation that we have, namely:
58 TIME AND SPACE
1. Causes explain their effects.
2. Causes are (in some sense) ontologically more basic than their effects.
3. Correlated events are causally connected.
4. Causes are the means by which we bring about the effects that are our ends.
Given these beliefs about what a cause is, and the temporal asymmetry of knowl-
edge, explanation, and so on, the idea that causes occur earlier than their effects
is entirely natural. Or so Horwich goes on to argue.
Both 2 and 4 are linked to the knowledge asymmetry. The vague idea that
causes are more real than their effects is rooted in the widespread belief that the
past is more real than the future, and this belief is in turn rooted in the fact that
we have detailed knowledge of the past but not of the future. As for 4, Horwich
suggests that causation is (in part) defined by its association with our experience
of deliberation and control: we define causation as that general relation between
events that is exemplified when an event is deliberately brought about by free
choice (1987: 143). If, after due deliberation, I decide to put poison in Jims
drink, and Jim subsequently dies, I have caused Jims death. More generally, we
would all agree that if C is used by someone to produce a desired end E, then C
causes E. As for why our deliberation and intentions are future- oriented, consider
this: could you seriously deliberate about performing a certain action A if you
know for certain that you will not do A? Of course not. We can only deliberate
and form genuine intentions to perform actions that we (believe) we are free to
perform. Horwich argues that:
the feeling of freedom is associated with our ignorance, prior to deliberation,
of what we are going to do. If we knew as much about the future as we do
about the past (and with the same certainty that is provided by memory) this
feeling could not be sustained. We can then explain, in terms of the sense of
free choice, why rational decisions are not made for the sake of past events
and why we care about the future more than the past. (1987: 204)
The knowledge asymmetry also plays a role in the temporal orientation of
explana tion. As we have just seen, the belief that the past is real and fixed, whereas
the future is not, is (according to Horwich) grounded in the knowledge asymmetry,
and this same belief helps sustain the practice of explaining later events in terms
of earlier ones, since for anyone who holds this belief it will seem natural (even
if it is not really the case) to suppose that later events depend on earlier events in
a way that earlier events cannot depend on later events. But the contribution of
the fork asymmetries to the asymmetry of explanation is even more marked. As
Owens stresses, a genuine coincidence consists of a conjunction of events that has
no explanation (1992: ch. 1). Such a conjunction has no explanation if its compo-
nent events each have necessary and/or sufficient conditions that are independent
of one another. If, on the other hand, there is some single event that is necessary
and/or sufficient for each component of the conjunction, we have an explanation,
and the conjunction in question is no coin cidence. Ten people on a coach party
falling ill is a coincidence if they all ate different meals; it isnt if they all ate the fish
for lunch. In virtue of the fact that future- directed forks predominate, and given
ASYMMETRIES WITHIN TIME 59
the asymmetry of overdetermination, non- coincidental conjunctions of events are
nearly always explained by reference to earlier events. It is common to find a
single earlier event that is both necessary and sufficient for several subsequent
events, and that thus reveals the later conjunction to be non- coincidental, but it
is rare to find the same in reverse. Generally speaking, when (as often happens) a
given earlier event is necessary for several subsequent events, none of the latter is
individually necessary for the earlier event, since each is individually sufficient for
it. Since the direction of explanation runs from earlier to later, and since causes
are events which explain their effects, so, too, does the direction of causation.
Knowledge
Horwich argues that the knowledge asymmetry can be explained in terms of the
fork asymmetry, a conclusion that emerges from a highly abstract discussion of
the nature of recording devices (1987: 8490). I will try to convey the gist of the
basic idea in a more informal way.
An expanse of wet concrete constitutes an efficient recording system: if some-
one steps in it, an accurate imprint of their footwear (or foot) will be made, and is
likely to endure for a good while, since once concrete is set it is physically robust.
Anyone who subsequently sees the imprint in the concrete will have a good chance
of being right if they believe that someone with just that shape and size of foot
stepped in the concrete a short while before it hardened. Any foot- shaped imprint
in concrete is a particular macro- state consisting of trillions of particles existing
in a particular configuration. As is clear, these same particles could exist in tril-
lions of different configurations, and since concrete slabs generally start off in a
randomly mixed condition, it is highly unlikely that any given foot- shaped con-
figuration would develop through chance alone. It is even more unlikely that all
the very many (but slightly different) foot- shaped imprints to be found in different
concrete surfaces would arise by chance alone. Viewed from the micro- level, the
creations of these imprints are strongly correlated events: particles in many differ-
ent concrete surfaces all coming to be arranged in highly improbable foot- shaped
configurations. These correlated events require an antecedent event of a highly
specific sort (e.g. a foot pressing downwards). We can be reasonably certain that
the event that explains the correlation occurred earlier than the correlation itself
because we know that, whereas future- directed forks of this kind are common,
inverse forks are rare.
Records come in diverse forms; for example, fossils, newspaper reports and
memories. But despite this diversity there is also similarity:
in each case we have a macroscopic system whose constituent parts have a
ten dency to enter into highly specific (and improbable) configurations;
these configurations are strongly correlated with some other type of event (e.g.
foot- shaped imprints are correlated with downwards pressing feet);
this other type of event almost always occurs earlier than the configuration in
question.
This last fact obtains by virtue of the fork asymmetry. If instances of regularly
occurring improbable conjunctions of event- types were commonly associated
60 TIME AND SPACE
(statistically) with a single type of subsequent event, there might be many reliable
records of future events; but since they arent, there isnt.
The source of the fork asymmetry
Since all the other asymmetries lead back to the fork asymmetry, directly or indi-
rectly, we are left with one outstanding question: why do normal forks predomi-
nate over inverse forks? Why does the fork asymmetry exist at all in a universe
governed by time- symmetrical laws?
This is a crucial question, and Horwich can offer no more than a speculative
cosmological conjecture. He makes the assumption that the early universe was in
a state that combined order with disorder. The disorder obtained at the micro-
level: elementary particles existed in a state of random flux. Within this micro-
chaos there was a limited degree of macro- order: mass- energy was not distributed
uniformly through space, but concentrated in the structures that eventually gave
birth to galaxies, stars and planets. These assumptions are compatible with many
current theories of the early universe. Now, suppose events of types X and Y are
strongly correlated with one another they co- occur at many times and places
(e.g. heat and smoke). It is unlikely that the sequences of events leading up to these
frequent conjunctions of X and Y will consist of two independent processes, for
this would require a vast and repeated conspiracy at the micro- scale, a conspiracy
that is ruled out by the assumption of initial micro- chaos. Hence we would expect
to find that there is some single type of event Z (e.g. fire), the occurrence of which
is strongly correlated with (or nomologically sufficient for) the subsequent occur-
rence of X- and Y- type events. This, of course, is precisely what we do tend to find:
it is the normal fork pattern. The question that needs an answer is why do inverse
forks not occur? Why are strongly correlated simul tane ous events not regularly
followed by a single later event of a specific type? The answer, suggests Horwich,
lies with the implicit microscopic order that exists at later times.
Let us suppose that the universe does not collapse under its own gravitation, but
its expansion gradually slows. From the second law of thermodynamics, we know
that the energy that is currently concentrated in stars will eventually disperse, and
so the later stages of the universe will not contain any recognizable order: there
will be a thin homogeneous spread of low- energy particles and fields. Now, if we
consider the evolution of the universe from the direction of earlier- to- later, we
see a progression from an initial state that combines macro- order and micro- chaos
to later states that consist only of micro- chaos, an asymmetric progression that at
the micro- scale is governed by time- symmetric laws. But now imagine viewing this
same process unfold in the other direction, from later- to- earlier. What would you
see? On the large- scale, you would see the cold thin soup of particles gradually
coming together to form dim stars, which gradually grow brighter as more and
more energy flows into them; occasionally, you would see vast numbers of parti-
cles converging to a single star (a super nova in reverse). On the much smaller scale
of our Earth, you would see huge numbers of astonishing coincidences, such as
wind blowing leaves into neat heaps in the middle of lawns, dirty clothes scattered
around a room ending up washed and neatly folded in cupboards and drawers, and
ASYMMETRIES WITHIN TIME 61
so forth. Adopting the time- reversed view makes one thing very clear: although
the thin cold particle soup that exists at the later stages of the universe might look
like a random distribution, it isnt. Given the distributions of particles that exist at
these times, and the time- symmetrical laws governing them, they will evolve into
the highly organized systems that exist at the present time. The apparently random
distribution is in fact highly ordered; this order is all but impossible to discern, it
is implicit, but it is there nonetheless.
The presence of this implicit order explains why inverse forks are rare. Even if
the scenario sketched above is implausible, and the implicit order dissipates over
long periods of time, it certainly exists over the short term (short by cosmologi-
cal standards). So given the hypothesis of initial micro- chaos, strongly correlated
events will generally only occur when there is a single earlier event that is nomo-
logically sufficient for them all. The implicit order in the events that occur later
than strongly correlated occurrences (even a short while later), means that there
is no need for correlated events to be followed by some single type of event that is
nomologically sufficient for them all: even if the subsequent events are many, var-
ied and apparently random, in each specific case and context, these later events are
nomologically sufficient for the earlier events. And this is what we find. Thunder
and lightning are always preceded by a build up of atmospheric static; there is no
one succeeding condition of remotely comparable significance.
If we explain the content- asymmetries along the general lines sketched out by
Horwich, we cannot rule out the possibility of time- reversed universes. Since the
initial conditions at each end of a Gold universe are very similar both exhibiting
micro- chaos it is to be expected that each half will develop in much the same
way. In which case, not only will both halves of the universe be characterized by
fork asymmetries, but the forks will point in different directions. If, as Horwich
maintains, all the other content- asymmetries depend on the fork asymmetry, these,
too, will be differently oriented, causation included. In which case, the expanding
and contracting phases of the universe will be time- reversed with respect to one
another. Times arrow may thus be less securely fixed than we tend to think. Or
at least, this would be the case if we lived in a B- world, a world where temporal
passage is illusory, where time itself possesses no inherent direction. Whether this
is so or not has yet to be established.
As Horwich himself stresses, his account of why the fork asymmetry exists rests
on largely speculative premises (e.g. a low- entropy big bang), and certainly needs
further development before it can be considered fully adequate. He points out that
all that matters for his account of the other content- asymmetries is the existence of
the fork asymmetry itself (1987: 76), but even here there are questions of detail to
be addressed, and there are those who doubt whether this asymmetry is up to the
job that Horwich assigns to it.
10
But for our purposes what matters is this: even if
the B- theorist cannot appeal to causation to explain the asymmetries that we find
in time, there are alternative strategies that are available, strategies that may well
prove viable. If Horwichs proposal proves problematic, there may well be variants
of it that will prove more successful.
However, in the absence of any fully worked out proposal along these lines, and
given the difficulties encountered by the entropic and causal accounts, B- theorists
cannot yet be said to have provided a fully satisfactory account of the various
62 TIME AND SPACE
content- asymmetries that we have considered. If it proves impossible to formulate
such an account, B- theorists would have no option but to regard the asymmetries
(or at least, those that remain unexplained) as a matter of brute fact: it just so
happens that the contents of our world are distributed through a four- dimensional
continuum in certain asymmetrical patterns. This is not ideal, but neither is it nec-
essarily fatal, since in any theoretical system there is always something that is not
explained in other terms. Of course, adopting this stance would prove damaging
if there were viable dynamic conceptions of time possessing greater explanatory
resources. Whether there are such alternatives to the eternalist or Block concep-
tion is the issue we shall be examining next.
5 Tensed time
5.1 Tense versus dynamism
In 1.6 I noted that we should not assume that the eternaldynamic dispute maps
neatly onto the tenselesstensed dispute. The latter revolves around the sorts of
concepts that are needed to describe the world in a metaphysically adequate way,
whereas the former concerns the reality or otherwise of temporal passage. The
B- theory we have just been considering combines two claims:
The world is a four-dimensional ensemble, and all times and events are equally
real and coexist.
The world can be fully described in tenseless terms.
In Chapter 6 I will be looking at ways of taking issue with the first claim; in this
chapter I will consider the implications of rejecting the second.
In a sequence of papers spread over some years, E. J. Lowe has sought to defend
the A- framework against its critics (such as McTaggart and Mellor), and thus
establish the viability of the tensed conception of time.
1
Since many B- theorists
have argued that the tensed conception is incoherent, his arguments on this
point are of interest in their own right. In Chapter 2 I introduced McTaggarts
A- paradox, but since it proved peripheral to the main point that McTaggart was
trying to make, I did not consider it in any detail; a remedy was suggested but not
examined closely. Similarly, in Chapter 3 we encoun tered Mellors argument that
tensed statements have tenseless truth- conditions, but we have not yet considered
counter arguments. But there is another reason why Lowes position is worth look-
ing at. Although he is a staunch defender of the tensed conception, Lowe does
not engage in grandiose metaphysical speculations of the sort that we will be
considering in Chapter 6. Does this mean that such speculations are pointless and
unnecessary? Can someone who finds the B- theory implausible and/or repugnant
find all the consolation they need in the tensed conception of time? Or is some-
thing more needed? Let us find out.
5.2 Taking tense seriously
Lowe distinguishes the tensed and tenseless views thus: the tensed view takes
McTaggarts A- series terminology to be indispensable for the metaphysics not
just the epistemology and semantics of time, whereas the tenseless view takes
64 TIME AND SPACE
B- series terminology alone to be indispensable for the metaphysics of time
(1998: 43). A mark of Lowes commitment to the A- framework is his firm stance
on the question of the proper use of tenseless sentences. He rejects the idea that
properties can be tenselessly predicated of concrete historical events. The only
objects that can tense lessly possess properties are abstract entities (such as num-
bers) that do not exist in space or time; anything that does exist in time possesses
properties in a tensed way: What it is to exist in time is to be a potential subject
of tensed predications, in my view. That is, for a thing to exist in time is for some
property to be now exemplified by that thing, or to have been exemplified by it,
or to be going to be exemplified by it (1998: 44, original emphasis).
An advocate of the tenseless view would take a sentence such as There is rain
in Durham on 12 March 2099 to be a tenseless sentence akin to 2 + 2 = 4, but
for Lowe it can only be properly understood in a tensed way. Hence it is equivalent
to the sentence There is now rain in Durham on 12 March 2099 (which is false
since it is not now 2099 or 12 March), or it is a grammatically improper way of
trying to say There will be rain in Durham on 12 March 2099, or, finally, it is
an abbreviated way of saying There was, is now, or will be rain in Durham on
12 March 2099.
This point may seem rather pedantic, but it turns out to have far- reaching
implications. For a start, if Lowe is right, all facts about concrete reality are tensed,
and so the idea that a tenseless description of the world could be complete is a
non- starter. But what of the B- theorists claim that tensed statements have tense-
less truthmakers? Lowe has a response to this too.
Mellor argues that A- theorists face an impossible dilemma: they must accept
that either (i) tensed sentences are made true by tensed facts that are non- token-
reflexive, or (ii) tensed sentences are made true by token- reflexive facts that are
tenseless. Adopting (ii) is tantamount to rejecting tensed facts and thus the meta-
physical significance of tense: surely if there were tensed facts one of the things
they would do is make tensed sentences true! But (i) is clearly incoherent. The
tensed sentence World War Two is now over is true if tokened after 1945, and
false if tokened earlier; but how can this be if the sentence is made true by a tensed
fact that is independent of when the sentence is tokened?
Lowe accepts that the truth- conditions of tensed sentences must be token-
reflexive, but rejects Mellors claim that such truth- conditions must be tenseless.
He suggests that the truth- conditions of (tokens) of a sentence such as It is now
raining in Durham can be stated in either of the following ways:
A token u of the sentence- type It is now raining in Durham [is] true if and
only if u [is] uttered at a time t such that it [is] raining in Durham at t.
A token u of the sentence- type It is now raining in Durham [is] true if and
only if u [is] uttered simultaneously with an occurrence of rain in Durham.
Why are the occurrences of is in brackets? Because Lowe wants to make it abso-
lutely clear that these statements of truth- conditions are tensed. Since sentence-
types are abstract entities they can have properties predicated of them tenselessly.
But the formulations above concern sentence- tokens, and concrete time- bound
entities such as these can only be the subject of tensed predication, or so Lowe
TENSED TIME 65
maintains. Accordingly, when he says that a sentence- token is true if and only if
such- and- such is the case he means either that it is now true, or else what he is
saying is a potentially misleading abbreviation of the disjunctive tensed predication
was, is now, or will be true. It is the latter he has in mind in the above statements
of truth- conditions (to make them fully general rather than present- tensed). Hence
[is] true means was, is now, or will be true.
5.3 McTaggart revisited
This approach provides Lowe with a response to McTaggarts A- paradox. Like
C. D. Broad before him, Lowe believes that McTaggart is guilty of a simple mis-
take: the alleged paradox doesnt even get off the ground. As we have already seen
in Chapter 2, McTaggart argued that if we use A- terminology we are committed
to saying that every event is past, present and future, despite the fact that these
are incompatible attributes. Lowe responds (quite plausibly) by maintaining that at
most we are committed to saying something along these lines: For any event e, (i)
it either was, is now, or will be true to say that e has happened, and (ii) it either
was, is now, or will be true to say that e is happening now, and (iii) it either was,
is now, or will be true to say that e will happen (1998: 46). As is clear, there
is nothing paradoxical here at all. There was a time when tokens of the sentence
Napoleon is now alive and well were true, but tokens of this sentence were not
true before Napoleon was born, and they are not true now, and (barring miracles)
they wont be true in the future either. Likewise, tokens of The year 2010 is in
the past, are not true now (or in the past), but they will be one day. And so on.
As Lowe concedes, careful formulations such as the one above can be some-
thing of a mouthful, so we might easily find ourselves falling into misleading
ways of talking. The mistake to avoid is to suppose there is no alternative to the
misleading forms of words. So, for example, we might be tempted to express the
fact that there was a time when tokens of the sentence Napoleons death lies in
the future were true by saying that Napoleons death was future, or even has the
property of being future in the past. McTaggart himself employed these compound
tenses when expounding the A- paradox (as does Smiths disjunctive formulation
quoted in 2.4). But as Lowe points out, interpreted literally, formulations such
as these are simply incoherent: to say that an event E is future in the past is like
saying that an event E is happening here over there. If someone insists But E is
happening here over there, you know what I mean! what they really mean is
something like this: an utterance over there of the sentence E is happening here
is true (Lowe 1987).
5.4 Is tense enough?
These metalinguistic constructions provide Lowe with a plausible way of refuting
the charge that the A- framework is inherently contradictory. Provided we take
the necessary care, we can talk in tensed terms without fear of talking nonsense.
The tensed theory of time cannot be refuted on the grounds that it is internally
66 TIME AND SPACE
incoherent. But what does Lowes tensed theory of time actually amount to? What,
on his view, is the metaphysical difference between past, present and future?
As noted earlier, he says that the tensed view of time is distinguished by the
claim that the notions of past, present and future are indispensable ingredients in
our metaphysical understanding of what time is. If he is right in claiming that for a
thing to exist in time is for some property to be now exemplified by it, or to have
been exemplified by it, or to be going to be exemplified by it, he may well be right
that the notions of past, present and future are indispensable ingredients in our
metaphysical understanding of time. But what does talking in tensed terms really
signify? Wellingtons victory at Waterloo lies in the past. How does this event differ
from an event that lies in the present or future? Do present events have a distinctive
property, presentness, which past and future events lack? Do past events possess a
distinctive property of pastness, which present and future events lack? Lowe rejects
this idea, citing overdetermination difficulties of the sort we encountered in 2.6:
There is one very good reason why I use this sort of metalinguistic construc-
tion, and this is that I do not want to suggest that presentness is some sort
of property of events [on the grounds that] if presentness (and likewise
pastness and futurity) is conceived as being a property of events, then it is
difficult to see how it can either be a property which an event has always had
and always will have, or be a property which an event now has, has had, or
will have only temporarily and no other option seems available.
(1998: 47, original emphasis)
This is a very plausible position, but if the difference between being past, present
and future is not a matter of possessing different properties, what does the differ-
ence consist in?
One compelling answer is that whereas present events actually exist, past and
future events do not, and Lowe would go along with this, up to a point. He is
happy to concede that events that lie in the past do not exist now, and likewise for
events that lie in the future. Of course while events such as the Battle of Waterloo
do not now exist, they did exist, and future events will exist. Is there anything
more to be said? Someone might insist:
What you are saying sounds reasonable, but I want to know more. Are you
saying that past and future events are just as real as present events, that
all events are on the same ontological plane, so to speak, as the B- theorist
maintains? Or are you saying that future and/or past events do not form parts
of the sum total of reality?
Lowe does not address quite this question, although he does address one some-
what similar to it, so the following reply goes some way beyond what Lowe actu-
ally says, but not by much.
I do not say that there is no ontological distinction between past, present and
future. I am happy to concede that some future- tensed statements may be
neither true nor false (e.g. it is now neither true nor false to say It will rain in
Durham on 12 March 2099) (Lowe 1998: 50). But neither do I say that past
and future events are unreal, or that they do not form parts of the sum total
TENSED TIME 67
of reality. What distinguishes past events is precisely that they are past, not
unreal; likewise for future and present events. Moreover, the sum total of
reality is not restricted to the present, since some events are (now) past and
others (now) future (even though we do not as yet know what these will be).
While there is no denying that some events are present whereas others are past,
it is not clear what this difference amounts to on Lowes account, given that he
rejects the idea that pastness and presentness are properties possessed by events.
His response is to say that tensed concepts such as past are so fundamental that
they are semantically irreducible Do we not take the same view of an operator
such as negation? It is not the case that cannot be given a non- circular semantic
explanation, yet of course we grasp its meaning as clearly as we grasp the meaning
of anything (1998: 49, original emphasis). Again, this is a perfectly respectable
reply: some concepts are so basic that they cant be explained in terms of other
(independent) concepts. Perhaps past, present and future fall into this category. But
this reply, although hard to refute, does not entirely satisfy.
Lowe does not explain why some future- tensed statements can lack a truth-
value when (presumably) their past counterparts (e.g. It rained in Durham on 12
March 2099BC) do not. He might well say that this is simply an aspect of what
is involved in being past and future. But is it plausible to suppose that this is all
there is to be said? Suppose someone says,
The future is unreal, not just in the sense of not existing now, but in the
sense of not existing at all. If we could compile a complete inventory of
what exists, not just now but at all times, past and present events would be
included, but not future events.
This claim, if intelligible, would explain why statements about the future contin-
gen cies lack truth- values. By eschewing claim such as these, Lowe is unable to
provide an explanation of comparable power.
Lowes position amounts to a deflationary dynamism.
2
On one level the account
offers opponents of the four-dimensional Block view what they want. There is a
difference between past, present and future; time does pass, at least in the sense
that an event that at one time can truly be characterized as being present can also,
at different times, be truly characterized as past and future. But on another level
the account leaves something to be desired. It turns out that the only difference
between being past, present and future is that events can legitimately be called
past after they have occurred, present when they are occurring, and future
before they occur. Since tensed predicates do not denote genuine properties, it
is hard not to feel that the difference between past, present and future is merely
verbal. Is this all this difference boils down to? It is hard not to feel that there is
something more to be said. The models of time we will be looking at next try to
say more.
6 Dynamic time
6.1 The Growing Block
In Scientific Thought, after a clear statement of the problems posed by suppos-
ing that there are transitory tensed properties, Broad proposes a very different
dynamic theory of time. Focusing on his current state, he says that this:
is just the last thin slice that has joined up to my life- history. When it ceases
to be present and becomes past this does not mean that it has changed its
relations to anything to which it was related when it was present. When
an event, which was present, becomes past, it does not change or lose any of
the relations which it had before; it simply acquires in addition new relations
which it could not have before, because the terms to which it now has these
relations were then simply non- entities.
It will be observed that such a theory as this accepts the reality of the
present and the past, but holds that the future is simply nothing at all.
Nothing has happened to the present by becoming past except that fresh
slices of existence have been added to the total history of the world. The
past is thus as real as the present. On the other hand, the essence of a
present event is, not that it precedes future events, but that there is quite
literally nothing to which it has the relation of precedence. The sum total
of existence is always increasing, and it is this which gives the time- series a
sense as well as an order. A moment t is later than a moment t* if the sum
total of existence at t includes the sum total of existence at t* together with
something more.
The relation between existence and becoming is very intimate. What-
ever is has become, and the sum total of existence is continually augmented
by becoming. There is no such thing as ceasing to exist; what has become
exists henceforth for ever. When we say that something has ceased to exist
we only mean that it has ceased to be present; and this only means that the
sum total of existence has increased since any part of the history of the thing
became. (Broad 1923: 879)
The process that Broad subsequently calls absolute becoming is the coming-
into- being of events. Time passes, the universe is dynamic, because new events
are continually being created. When an event comes into existence it remains
in existence. Past events are perfectly real; just as real as present events. Future
events have no existence at all. As is very clear, Broad has no truck with transitory
DYNAMIC TIME 69
tensed intrinsics: present events do not possess a distinctive property of present-
ness, which past events lack. An event is present by virtue of belonging to the
most recent reality- slice, that is all. It does not change its intrinsic properties as it
sinks into the past; that is, as additional slices of reality are added to the universe.
1

Broads model of time is exactly the same as the B- theorists Block view with one
crucial difference: if Broad is right, the four-dimensional block is not static but
growing.
By what increments? How large are the new slices of reality? In principle,
there seems little reason why the increments could not be huge hunks of time a
universe of ten billion years total duration could come into being in just two
stages but for obvious reasons this idea lacks appeal. If we define later than
in Broads way, these billions of years would contain only two successive times!
More importantly, the appeal of the Growing Block model (as we can call it)
rests on its conformity with the common- sense idea that the present is a brief (or
momentary) interface between a real past and a non- existent future; this is lost
if we take the present to be the most recent addition to reality, and allow these
additions to be of sizeable duration. Hence Growing Block theorists have two
plausible options: they can say that each reality slice is instan taneous, or they can
say that each reality slice is of some finite but minimal duration. Anyone who is
sceptical of the idea that even an infinite number of strictly durationless slices
can ever add up to anything of finite physical duration will prefer the second
option; we shall be taking a closer look at this issue in Chapters 16 and 17.

Figure 6.1 A Growing Block universe.
As new presents are created, the sum total
of reality increases. The growth in question is
temporal. Although our universe (apparently)
originated in a big bang singularity and has been
expanding (spatially) ever since, other possible
universes may start large and shrink in size over
time in such cases each new present will have
a smaller spatial extent than its predecessor.
present
universe (sum total of reality)
This model of time has considerable appeal to anyone reluctant to abandon the
natural view. The past is fixed and determinate, the future is unreal and the present
is both constantly changing and distinctive, even if only by virtue of being real-
itys most recent accretion. Moreover, the model provides an intuitively appealing
explanation for some important content- asymmetries. We have trace evidence
(and hence detailed knowledge) of the past but not the future because the former
is real and the latter is not, and only what is real can leave traces of itself. The
absence of (easily discernible) backward causation can be explained in the same
way: non- existent events cannot cause anything; real events can.
But is the model truly coherent? The closer examination that follows will yield
this tentative verdict: there are versions of the model that may well be coherent,
but in certain respects the model is not the friend of the natural view that it initially
seems.
70 TIME AND SPACE
6.2 Overdetermination
By not positing intrinsic tensed properties, the Growing Block theorist avoids the
overdetermination problem in one form, but the same problem in other guises lies
in wait unless appropriate measures are taken. Consider the following versions of
the model:
A Past, present and future moments of time are equally real and permanent, but
may be empty. As events are created, time is gradually filled with concrete
events.
B Moments of time come into being with things and events; there are present
and past moments, but no future moments.
Version A succumbs to the overdetermination problem that we encountered in
2.6. Does it make sense to suppose that a given moment of time could be both
empty of events and not empty of events? Could the universe at some point in
2010 be both entirely devoid of matter and also at the very same time contain
billions of galaxies? Of course not: just as an object cannot be in two different and
incompatible conditions at a single time, the universe cannot be in two different
and incompatible conditions at a single time. To avoid this problem the Growing
Block theorist must subscribe to version B. If events and the times at which they
occur are created together, times are not overdetermined.
Figure 6.2 The wrong way for a universe to
grow. If we envisage the material universe growing
through pre-existing substantival time, we encounter
the overdetermination problem: there will be times
which possess the incompatible attributes of being
empty of events, and not being empty of events.
time
But has the overdetermination problem really gone away? Consider the uni-
verse at an earlier time t
1
and at a later time t
2
. Let us call the sum totals of reality
that exist at these times S
1
and S
2
. S
1
includes everything that happens at t
1
along
with everything that has happened earlier than t
1
. S
2
includes everything that
happens at t
2
, along with everything that has happened earlier, and so includes
S
1
. Suppose that E is an event that happens at t
1
. It seems that we can say the fol-
lowing about E:
1. The sum total of reality to which E at t
1
belongs consists of S
1
.
2. The sum total of reality to which E at t
1
belongs consists of S
2
.
These two claims are inconsistent; they cannot both be true. A rock cannot, at a
given time, be in a pond containing two different total volumes of water. How
can one and the same event be a part of two different sum totals of reality at the
time it occurs?
There is a solution to this problem. We are familiar with the distinction
between states of affairs that are real, or that exist, or that are actual, and those
DYNAMIC TIME 71
that are not. It is plausible to suppose that the notions of existence and actuality
are primitive. As Tooley plausibly argues at the outset of his impressively wide-
ranging defence of a Growing Block model (see Tooley 1997), in a dynamic
world the actualnon- actual distinction is insufficient, since the states of affairs
that are actual may be different at different times. What we need, accordingly, is
the concept of existing or of being actual as of a given time. What is actual as of
1066 is quite a bit less than what is actual as of 1966. Tooley proposes that we
take this notion of actual as of a particular time as a primitive. Corresponding
to this ontological notion there is the semantic notion of truth as of a time. Since
the totality of facts can vary from time to time, the concept of truth appropriate
to a growing reality is temporally relativized: as reality grows, so, too, does the
totality of truths. Hence when talking about what is true as of a time t, we are
talking about what reality (present and past) contains on the occasion t when its
contents come into being. To put it another way, we are talking about what is the
case relative to the true or objective present, construed as the constantly advanc-
ing frontier of reality.
Thinking in these terms has two consequences: the same proposition can have
different truth- values as of different times, and there are three truth- values: true,
false and neither- true- nor- false. Consider:
3. There are (tenselessly) no dinosaurs.
4. There are (tenselessly) unicorns in the year 2050.
If we suppose that it is now 2010, 3 is not true: the Jurassic Period is actual as
of 2010, and there are dinosaurs in the Jurassic Period. As of 2010, 4 is not true
either, but for a different reason: the year 2050 is not actual as of this time, so
the statement is neither true nor false. As of 2050 it will become true if there are
unicorns then, and false if not. So the truth- value of 4 will change from indeter-
minate to either true or false.
The notion of truth as of a time provides the solution to the problem of the
same event belonging to different sum totals of reality. As of t
1
, 1 is true and 2 is
indeter minate (since some parts of the sum total of reality S
2
are not actual as of
t
1
). As of t
2
, 1 is false and 2 is true.
2
I suggested earlier that a Growing Block theorist should hold that a universe
grows by durationless or very brief stages. A natural follow- up to the question
By what increments does a universe grow? is How fast does a universe grow?
However, if times and their occupants come into being together, this question
has no real sense. The question How much time passes between the addition of
one reality- slice and the next? would be perfectly meaningful if a block- universe
existed in an external time dimension (for example, if a meta- time axis ran verti-
cally past the horizontal blocks in Fig. 6.2). But since the Block theorist recognizes
(or should recognize) no such external time, all we can say is that successive
phases of reality come into being at (and along with) the times when they occur.
That a specific speed cannot be produced in response to the How fast? ques-
tion is sometimes taken to be disastrous (or at best a serious embarrassment) for
the Growing Block theory (cf. Nahin 2001: 199); but since the question in this
context has no sense, the error is with those who press the question expecting an
answer of this sort.
72 TIME AND SPACE
6.3 Dynamism without tense
As I pointed out in 1.6, the issue of whether time is tensed or tenseless is distinct
from the issue of whether or not time is dynamic; consequently, it may be possible
to combine a dynamic view of time with a tenseless view. Tooley develops his
account of the Growing Block model along these lines. A proponent of the tense-
less view subscribes to the claim that the world can be fully described in tenseless
terms. Tooley not only subscribes to this claim, but he rejects the idea that there are
genuine tensed properties, and provides an exhaustive analysis of tensed concepts
in tenseless terms. Space precludes a full exposition of Tooleys account, but I
provide some examples of his approach, which in some respects is novel.
Tooley takes a tensed sentence to be any sentence that employs tensed concepts,
such as present or past. Most tensed sentences are indexical: a sentence such
as It is hot now specifies which time is present not by explicitly stating a date,
but by having a meaning such that the time at which the utterance occurs indicates
the time of the present. Are there tensed sentences that are non- indexical? Tooley
claims that there are:
Event E lies (tenselessly) in the present at time t.
This is tensed (since it mentions the present), but it specifies the time that is present
in a non- indexical way, by mentioning a date. Tooley offers a tenseless analysis of
this sentence:
E is an instantaneous state of affairs, E is actual as of time t, and no state of
affairs that is later than E is actual as of time t.
What of non- indexical sentences about the past and future? Tooley suggests the
following analyses (1997: 196):
E lies in the past at time t: Event E is earlier than time t, and t is an instan-
taneous state of affairs, t is actual as of time t, and no state of affairs that is
later than t is actual as of t.
E lies in the future at time t: Event E is later than time t, and t is an instanta-
neous state of affairs, t is actual of time t, and no state of affairs that is later
than t is actual as of t.
As for indexical sentences, drawing on the above, Tooley suggests analyses such
as these:
Any utterance or inscription, at time t*, of the sentence Event E is now
occurring is true (false, indeterminate) at time t if and only if it is true (false,
indeterminate) at time t that E lies in the present at t*.
Any utterance or inscription, at time t*, of the sentence Event E has
occurred is true (false, indeterminate) at time t if and only if it is true (false,
indeterminate) at time t that E lies in the past at t*.
I will not pursue Tooleys analyses any further. I introduce them solely because they
provide a concrete illustration of how a dynamic view of time can be combined
with the tenseless view. There are no doubt other ways in which the combination
DYNAMIC TIME 73
can be worked. To cite one instance, Tooley defends a causal account of the ear-
lierlater relation, and (cf. Horwich and Price) not all tenseless theorists would
go along with this.
6.4 The thinning tree
Before moving on to assess the Growing Block model, there is another dynamic
model in some respects similar to the Growing Block that deserves a mention.
The account in question has been developed and elaborated over a number of
years by Storrs McCall.
3
For reasons that will shortly be obvious, I will call it the
thinning tree.
The Growing Block model conforms with the widely held view that whereas
the past is real and fixed and the present is real and transitory, the future is (as yet)
unreal, and so consists at most of an open realm of possibilities. In Figure 1.4 these
possibilities are represented schematically, as a fan of dotted straight lines; a more
accurate representa tion would be a collection of spreading and dividing branches.
Tomorrow I may stay at home or go to work, I havent yet decided. If I stay at
home, I will either do some gardening or read a novel; if I go to work I will either
spend the day in my office catching up on paperwork, or attend a lengthy meeting.
The same pattern applies quite generally: each of the initial possible courses of
events at any given time for a given object or system (of which there are usually
many more than just two) is associated with numerous further possibilities, and
each of these sub- branches itself divides into further possible courses of events (if
I spend the day in the garden, I will either repair the lawn, dig the border, prune
the apple tree and so on). With the passage of time (the creation of new reality-
slices) a few of these possibilities are realized. The remainder the vast bulk are
systematically eliminated.
This process of branch attrition is depicted in Figure 6.3. As the present
approaches the first node (n
1
) the possible futures represented by b
1
and b
2
(each
with their attendant sub- branches) still exist; but as the present moves past n
1
,
branch b
1
in its entirety is eliminated (I decided to go to work), leaving b
2
and
its attendant sub- branches (the first two of which are b
3
and b
4
). As the present
passes n
2
, b
3
is actualized and b
4
is entirely eliminated; at this point b
5
and b
6
both
remain possible, but as the present passes n
3
this will change: one branch will be
b
b
b
b
b
b
b
b
b
n
n
n
1
1 2
2
2
2
3
3
3
4
5
6
Figure 6.3 The thinning tree.
74 TIME AND SPACE
actualized, and the other eliminated. And so it continues. In reality, of course,
there are many more branches (perhaps infinitely many) than are depicted here.
Quite how many depends on what we decide to include as possibilities. There
are more logically possible futures than there are physically possible futures (i.e.
ways things might develop that are compatible with the laws of nature). Indeed, in
a universe where nomological determinism holds, there is only a single physically
possible future.
So we see that a Growing Block universe is also a thinning tree universe:
every gain for actuality (every creation of a new reality- slice) is accompanied by a
loss of possibilities; as the present advances, the branching tree of possible futures
is progressively thinned.
McCalls dynamic conception of time is essentially the model just described
the branches of his universe- tree are restricted to physically possible futures
but with just one crucial difference. McCall subscribes to a restricted version of
modal realism (see 2.6), the doctrine that possible states of affairs are just as
real as actual states of affairs. Consequently, he holds that all the future branches
on the universe- tree are just as real as the trunk the portion representing the
past and present:
At each stage, as the present (the first branch- point) moves up the tree, it is
a purely random matter which branch survives to become part of the trunk.
There is no preferred branch, no branch which is singled out ahead of time
as the one which will become actual. Instead, all branches are on a par. All
are equally real and, together with the trunk, constitute the highly ramified
entity I shall call the universe. Since it is never exactly the same at any two
times the universe is a dynamic not a static or eternal thing. Nevertheless
it is the same universe throughout, just as a child can look very different at
different times and yet remain the same person the whole of its life.
(McCall 1994: 4)
For McCall, actual does not mean real, since every branch of the universe-
tree is equally real. An event (or time) becomes actual when, due to the process
of progressive branch- attrition, it stands alone and without competitors. Whereas
the Growing Block model posits an increase in the sum total of reality, in McCalls
model there is a progressive decrease: with every advance of the present untold
billions of perfectly real stars, planets and people wink out of existence.
4
Intriguing though it is, I will not discuss McCalls conception further. If the
Growing Block model is metaphysically coherent, then so, too, I suspect, is
McCalls version of the thinning tree. If a linear block-universe can be of different
sizes as of different times, it is hard to see why a branching block-universe cannot
be of different sizes as of different times in the way that McCall suggests by hav-
ing fewer branches rather than by having different temporal extents. But even if
it is a coherent conception, McCalls model is problematic on several counts. As a
conception of time it is radically counter intuitive; indeed, with its multiplicity of
real futures it is far more counterintuitive than the orthodox Block universe. The
same applies to McCalls version of modal realism, which abolishes the distinction
(as usually understood) between possibility and actuality. But in addition, and per-
haps most seriously, the model is ontologically profligate, and so falls to Occams
DYNAMIC TIME 75
razor. Accepting the enormous expansion in the sum total of reality (as of certain
times if not all) that McCalls model requires would be warranted if there were
no other way to make sense of the world as we find it. But it is far from clear that
this is the case.
5
6.5 How can a block grow?
An important aspect of the Growing Blocks appeal lies in the clear and unambigu-
ous way that the model provides time with a direction. Thus far I have assumed, as
it is natural to do, that there is a genuine sense in which a universe whose contents
as of one time are greater than as of another time is growing. Further reflection
suggests that the situation is not so clear- cut.
In Figure 6.4, S, M and L (standing for small, medium and large, respectively)
refer to three distinct times. At first view, the growing world seems very different
from the shrinking world; whereas the former starts small and becomes larger,
the latter starts big and becomes smaller.
S
M
L
a growing world a shrinking world
Figure 6.4 Growth v. shrinkage.
How does a world that grows via
absolute becoming differ from a world
that shrinks via absolute annihilation?
L
M
S
However, if we focus solely on the facts represented in these diagrams, the
difference begins to seem less obvious. The following descriptions capture all the
facts on display, and yet apply equally to both worlds:
The universe as of S is smaller than it is as of M.
The universe as of M is smaller than it is as of L.
Where is the inherent directedness here? All we have is a dynamic block, a
universe whose temporal extent is different as of different times.
Of course, I have made no mention of becoming, and it does seem natural to
suppose that a universe in which facts come into being is growing rather than
shrinking, but consider another simple universe of finite temporal extent. The
universe in question consists of three slices, S
1
, S
2
and S
3
, occurring at three
times t
1
, t
2
and t
3
. Each slice contains the same rubber ball. In S
1
the ball is blue,
in S
2
it is yellow and in S
3
it is green. Now consider these two ways of describing
this universe.
1. As of t
1
the sum total of reality consists only of S
1
. S
2
and t
2
come into exist-
ence by the process of absolute becoming, and the sum total of reality as of t
2

consists of S
1
and S
2
. S
3
and t
3
are created, and as of t
3
the sum total of reality
consists of S
1
, S
2
and S
3
.
2. As of t
3
, the sum total of reality consists of S
3
, S
2
and S
1
. By the process of
absolute annihilation, S
3
and t
3
depart from existence, leaving t
2
, as of which
76 TIME AND SPACE
time the sum total of reality consists of S
2
and S
1
. S
2
and t
2
undergo annihila-
tion, leaving t
1
, as of which time the sum total of reality consists of S
1
.
It is natural to interpret 1 as a description of a growing universe, and 2 as a
description of a shrinking universe. The scenarios are described sequentially, and
it is easy to adopt the tacit assumption that the sequence is temporal, and so
unfolds in the earlier- to- later direction, where the latter terms have all the usual
associations (we can remember earlier events but not later events, etc.). Inter-
preted in this way, 1 and 2 describe two different universes: one that starts small
and increases in size as time passes, and another that starts large and decreases
in size as time passes. But if we read 1 and 2 without making any assumptions
about how the sequences temporally unfold, the differences between them all but
vanish. As in our initial case, the objective facts about this universe appear to be
these:
the sum total of reality is not constant, there are reality- slices that exist as of
some times but not as of others;
the sum total is smallest as of t
1
, largest as of t
3
and intermediate as of t
2
.
In saying this I am assuming that absolute becoming and absolute annihila-
tion are alternative ways of describing the fact that reality- slices exist as of some
times but not others. Unless this is wrong, and there is some reason for preferring
one term over the other, it seems that 2 is just as accurate a description of our
simple universe as 1. Given that this is so, isnt it just as accurate to describe this
universe as shrinking by incremental annihilation as it is to describe it as growing
by incremental creations?
Not if we define our terms in the way Broad does. Since Broad stipulates that
times as of when the block is larger are later than times as of when it is smaller, it
is true by definition that the universe is growing; that is, that the block is larger at
later times. But there is an alternative to Broads definition. Consider:
D
1
t
2
is later than t
1
if and only if the sum total of reality is greater as of t
2
than
it is as of t
1
.
D
2
t
2
is later than t
1
if and only if the sum total of reality is smaller as of t
2
than
it is as of t
1
.
Broad opts for D
1
, but why is Broads definition of later preferable to D
2
? Any-
one who opts for the latter will regard a universe of the sort depicted in Figure
6.1 to be shrinking rather than growing, and by their lights, this is the correct
description.
The question that we need to consider is whether Broads preference for D
1

over D
2
is warranted. What the discussion so far has shown is that the preference
cannot be justified by appealing to differences in block- size, but this does not
mean that it cannot be justified at all. Let us call the direction of time provided by
D
1
the block- arrow. The world- arrow is the apparent direction of time, the direc-
tion provided by the physical and psychological asymmetries that we examined
in Chapter 4 (e.g. the tendency for entropy to increase, the apparent absence of
back ward causation and trace- evidence of the future, our apparent experience
of temporal passage). Our ordinary under standing of the difference between
DYNAMIC TIME 77
earlier and later is rooted in these content- asymmetries. Consequently, if it
could be shown that the world- arrow and the block- arrow point in the same
direction, either necessarily or contingently, Broad s preference for D
1
would be
fully justified: the claim that times at which the universe is larger are also later
would have a clear rationale.
There are grounds for thinking that the block- arrow and the world- arrow do
point in the same direction. We have trace- evidence of the past but not the future,
and Broads hypothesis provides an explanation of why this is so: the times we call
future are not real as of the times we call present, whereas the times we call
past are. However, for now- familiar reasons, it does not follow from this that
the world- arrow and the block- arrow are in fact pointing in the same direction,
let alone that they must do so.
Suppose that we lived in a content- symmetrical Gold universe, of the sort dis-
cussed in 4.6. If we further suppose that our universe is a Growing Block, we
naturally assume that world- time and block- time coincide, and hence that the
block was smaller in the (world) past. Let us suppose that we are right about this.
Now contemplate the fate of the yet- to- exist inhabitants of the other half of our
universe, whose world- time runs in the opposite direction to our own. These
unfor tun ates will also naturally assume that block- time and (their) world- time
are running in the same direction, but they will be mistaken. Their world will be
constructed in reverse: the block is expanding into their past. Events they will
regard (quite properly, by the criteria of their world- time) to be in the future will
have occurred at earlier block- times, and events they regard as having occurred in
the past will in fact be unreal (they have yet to occur in block- time). Since there is
absolutely nothing in the experience of these people to suggest that this is the way
things are, for all we know we might be in a precisely similar condition. Perhaps
it is we who have futures set in stone and unreal pasts.
As this example clearly illustrates, the apparent direction of time is grounded
in the content- asymmetries that underpin the world- arrow, but there is no logical
guarantee that the world- arrow and the block- arrow point in the same direction,
and if the laws of nature in our universe are time- reversible, there is no nomologi-
cal guarantee either. And this opens the way for the perturbing possibility that the
large- scale structure of our universe might be very different from how it appears:
we might have unreal pasts but real futures!
The version of the Growing Block theory developed by Tooley rules out this
possibility. Tooley believes that there are powerful positive reasons for believing
the Growing Block theory to be true; reasons that go beyond intuitive appeal.
His central argument (in rough outline) runs thus:
1. Events in our world are causally related.
2. The causal relation is inherently asymmetrical. Effects depend on their causes
in a way that causes do not depend on their effects.
3. This asymmetry is only possible if a causes effects are not real as of the time
of their cause.
4. Causes occur before their effects: X is earlier than Y means (roughly) that
some event simultane ous with X causes some event simultaneous with Y.
5. Our universe must, therefore, be a Growing Block.
6
78 TIME AND SPACE
Tooley can thus deny the equivalence between the growing and shrinking descrip-
tions of a dynamic block-universe. Descriptions such as 1 and 2 (p. 75), which
mention only the fact that the sum total of reality is different as of different times,
are not complete, since they make no mention of the pervasive causal asymmetry,
which runs right through and along the times and events that comprise our uni-
verse. It is this causal asymmetry that provides time with its direction, and ensures
that the universe really is growing rather than shrinking.
I cannot attempt a proper evaluation of Tooleys complex arguments here, so I
will restrict myself to a couple of brief remarks. Even assuming that Tooleys claims
about the causal asymmetry are correct, is there any guarantee that the directions
of block- time and world- time will coincide? It might seem so, since the directions
of block- time (as defined by increasing size) and causation are locked together. But
in fact, the issue is far from decided. If the physical laws governing our world are
time- symmetrical (as they may well be), then given the appropriate initial condi-
tions (e.g. those obtaining at the central region of a symmetrical Gold universe)
these laws could generate a time- reversed version of our world. Or rather, they
could generate a universe whose world- time runs in the opposite direction to
causal- time in Tooleys sense. Tooleys causal relation may provide time with a
direction, but if all physical processes are time- reversible, this causal- arrow is
of an invisible (or noumenal) variety, since it is wholly independent of observable
patterns of change and variation, and hence of causation as we actually employ
the notion. If our universe is of the Gold type, our time- reversed counterparts will
take causation to run in the opposite direction from us: from our point of view,
what they call causes occur later than their effects. According to Tooley, the real
direction of causation may be as we describe it, or as they describe it, but one of
us is wrong. Tooleys view of the causal relation may not be incoherent, but it is
certainly suspect.
In a more speculative vein, proponents of Growing Block models should con-
sider this question: is there any reason to rule out the possibility of a block that is
expanding in both directions at once? If new slices of reality can be created, why
suppose that they can only be created at one end of a dynamic block? Is there any
a priori reason to suppose that a universe cannot grow outward in two directions
from a common core? In a doubly dynamic world of this sort there is no reason
to suppose that causation, even as conceived by Tooley, runs in only one direc-
tion. But if such a universe is possible, we can consider it from the other temporal
perspective, according to which the universe is shrinking from both ends. The
continual annihilations of the shrinking universe are, we can suppose, governed
by a principle of this form: let us say that adjacent events are D- related; unless an
event is D- related to other events at adjacent times in both temporal directions it
undergoes spontaneous annihilation. Since, by hypothesis, the outermost events
in either direction are not D- related on one side, they undergo annihilation, thus
exposing the next innermost events, which in turn undergo annihilation, and so
on. Tooley envisages a universe that is positively unstable, a universe that spon-
taneously increases by a process of continual creation. I am propos ing a universe
that is negatively unstable, a universe that spontaneously decreases by a process
of continual annihilation. What reason is there for favouring one vision over the
other?
DYNAMIC TIME 79
6.6 The eternal past
Quite independent of the considerations just advanced, there is an objection to
Growing Block models of a very obvious nature. If we assume that the world- and
block- arrows point in the same direction, to a substantial degree the Growing
Block model agrees with the natural view of time: the future is unreal, the present
is distinctive, the past is fixed. But someone might argue thus:
When an event becomes past it is gone, it has ceased to exist. Napoleon
doesnt just not exist at the present time, Napoleon has ceased to exist com-
pletely. But if the Growing Block model is true, everything that existed in
the past is just as real as what exists now. This is plainly false.
The problem is the status of the past. Most of us believe that there is a definite
and unalterable fact of the matter as to what happened at every point in the past.
Many of us also believe that the past has utterly ceased to exist. The Growing
Block theorist denies this.
The permanence of the past might also provoke concerns of this sort:
From our perspective, there are thousands of years of human history; these
past times are just as real as our time. The people living at these past times
believe that their time is present, but we know they are wrong; they are defi-
nitely in the past. But, by all appearances, there is nothing to distinguish our
time from past times: there is no distinguishing property of presentness that
past events lack. So given that lots of people in the past have experiences just
like ours and are wrong about living in the present, isnt it possible that we
are wrong when we believe we are living in the present? Indeed, wouldnt it
be an astonishing coincidence if the present were located now? Isnt it more
likely that we are somewhere in the middle of the Growing Block? Worse
still, might it not be that our block has finished growing? In which case even
now our world is indistinguishable from a passage-free B- world. Either way,
we are eternally embedded in a static past.
Parts of this may be exaggerated, or even misleading, but the main thrust exposes
what looks to be a very real problem for the Growing Block theorist. The appeal
of dynamic models of time largely rests on their being able to accommodate the
sense we have that we are living out our lives in a privileged and steadily advancing
present. If the reasoning sketched out above is correct, the Growing Block model
fails to satisfy this most basic desideratum: for all we can know, we are not living
in the present, for the present in the form of the leading or cutting edge of real-
ity may have moved on. Or so the objection runs.
There is a real difficulty here, but it is not what it may at first seem to be. First
a clarification. If our universe is a Growing Block with a finite future, it would
be a mistake to suppose that an infinite amount of time will pass after it ceases to
grow, for the Growing Block theorist can argue that time will no longer be pass-
ing when new slices of reality are no longer being added to the block. In which
case, our predicament is not obviously made any worse by our universes having
finishing growing. As for the main point, in one sense at any rate, it is a mistake
to think we can be wrong when we believe we are living in the present. Even if
80 TIME AND SPACE
we do happen to be living in a Growing Block universe, if I say I am living in the
present what I am saying is true if present is construed in an indexical way, as
referring to the time at which it is thought or uttered: clearly, I do exist at the time
at which I utter this sentence. And since what goes for me goes for everyone else,
there is obviously a sense in which everyone in a Block universe can be confident
that they are living in the present.
But, of course, this is not the end of the story. Whereas the ordinary (non-
dynamic) Block theorist holds that the indexical present is the only present, the
Growing Block theorist does not. The latter also recognizes an additional, meta-
physically weighty, concept of the present: things are PRESENT (as we might put it,
clearly to mark the difference) if and only if they happen to be located in the most
recently created reality-slice, when they lie on the interface between existence
and non-existence. Although every momentary event enjoys the privilege of being
PRESENT, the privilege is transitory: as new slices of reality come into being, previ-
ously PRESENT events cease to be PRESENT as they are pushed ever further into the
past. So the worry remains. From a purely subjective perspective, the experiences
of people in the past are no more real- or vivid-seeming than experiences that take
place in the PRESENT. No doubt Napoleon, fully engaged in the Battle of Waterloo
back in 1815, believes his thoughts, experiences and actions are all located at the
leading edge of reality, but from our vantage point we know he is wrong: he is
clearly a part of the past. But since Napoleons experiences in the past are just as
real and vivid (for him, then) as our own are (for us, now), can we be confident
that we are located at the cutting edge of reality?
The Growing Block theorist can at least partially alleviate this concern, by
arguing as follows. Suppose at a time t you think to yourself This very moment
is occurring in the present; call this thought T and, to simplify, let us assume the
thought takes only the briefest of instants to think. If present is interpreted in a
purely indexical way, you can be confident that T is true. But let us further suppose
that you are a firm subscriber to the Growing Block model; as such, you assume
that the very moment you are referring to in T is not only happening at the time
at which it happens, but that it is also PRESENT: it is taking place at the leading edge
of reality. Is your confidence justified? Arguably, it is. The thinking (or tokening)
of T is not something that is repeated over and over, it happens just once; more
specifically, it happens when the reality-slice that contains it comes into being. Let
us use the term occurs* to refer to these episodes of coming into being, or abso-
lute becoming, which lie at the heart of the Growing Block model. On the unique
occasion of its occurrence*, your thinking of T evidently does lie at the leading
edge of reality, and so it is PRESENT. And the same goes for all analogous thoughts
and utterances: when Napoleon pauses for a moment to ponder what is presently
taking place around him during the Battle of Waterloo, the events in question are
themselves PRESENT as those events occur*. In this manner we can accept that what
Napoleon believed to be PRESENT back in 1815 was PRESENT, relative to what was
then occurring*, even though the PRESENT has now moved on.
This way of registering what is PRESENT is not available for the non-dynamic
Block theorist, since for them the sum total of reality is constant, and events do not
occur*. But those who subscribe to the Growing Block model have an additional
resource at their disposal.
7
DYNAMIC TIME 81
There is more to be said, but it may be that Growing Block theorists can be justi-
fied in taking their thoughts and utterances to be not only present (in the indexical
sense) but also PRESENT as they occur. But there is a worry that will not go away.
Although it is true that the future is unreal when our present is PRESENT, it is also
true that the events that happen at this time will remain in existence just as they
are, intrinsically speaking if and when the future becomes real. From an ontologi-
cal perspective, the contents of the present (or PRESENT) are thus no different from
any past time (or indeed, any future time in a non-dynamic Block universe). In this
respect at least, the Growing Block theorists present (or PRESENT) is in no way a
privileged time.
8
Anyone who finds the prospect of being permanently (or time-
lessly) entombed in the past distasteful (or even terrifying) will be attracted to the
Presentist conception of time, according to which both past and future are unreal.
6.7 The varieties of Presentism
Presentism is the doctrine that nothing exists that is not present. Presentism
amounts to a dynamic view of time if this claim is combined with another: reality
consists of a succession of presents.
Although there has been recent upsurge of interest in the doctrine, for most
of the past century or so it has not featured prominently in recent philosophical
discussions.
9
The self- professed Presentist John Bigelow puts this down to the
impact on the modern mind of the four- dimensional view:
Presentism was assumed by everyone everywhere, until a new conception
of time began to trickle out of the high Newtonianism of the nineteenth
century. The Christians Holy Bible says that there is no new thing under
the sun but this is not true The so- called fourdimensionalist theory of
time was something genuinely new when it gradually came into being last
century. (1996: 35)
Once one becomes aware of the Block view, and the scientific considerations
(especially Einsteins special theory of relativity) that can be mustered in its favour,
denying reality to everything but the present can easily come to seem absurd.
Bigelow offers the following observation by way of support for his claim:
the exponential explosion of timetravel stories in the popular media, begin-
ning late in the nineteenth century, is an indication that a very new concep-
tion of time is brewing in the Zeitgeist. The utter absence of any timetravel
stories whatsoever prior to the nineteenth century is a profoundly puzzling
fact I suggest that this is at least partly explained by the utter universality
of presentism prior to the nineteenth century and by the utter absence of any
rivals to presentism. (Ibid.: 356)
If you believe that the past is wholly unreal, it is likely that you would be inclined
to push a button that would send you backwards in time only if you were of a
suicidal frame of mind. It may well be that we are no longer astonished by the very
idea of backward time travel in the way that our predecessors once were. Even if
many of us also have doubts about whether it really is possible to travel in this way,
82 TIME AND SPACE
the fact that time travel is a commonplace in contemporary science fiction may
have produced a change in attitudes towards the past, even if only to the extent of
making Presentism seem less obviously true than it once did.
In any event, in what follows we will be focusing on the metaphysical issues
surrounding Presentism. We shall see that the doctrine comes in different forms,
some a good deal more believable than others. Like most of the other dynamic
temporal models, Presentism at least in its most plausible and palatable guises
posits a universe-wide now, or plane of simultaneity; since this sits uneasily
with Einsteins relativity theories (see Saunders 2002; Balashov & Janssen 2003),
the Presentist must find a plausible explanation as to why this conflict is not as
problematic as it appears to be. Various options are available, and we will be tak-
ing a brief look at some of these in 19.3 and 19.4, after some of the scientific
background has been filled in. But we will be concentrating here on the more
general metaphysical difficulties confronting Presentism: problems that do not
derive from any particular scientific theory.
Some of these problems are specific to particular versions of Presentism, but
two are sufficiently general as to be worth mentioning at the outset. The first
is familiar: does the present have any duration? If so, how much? As noted in
connection with the Growing Block model, the least problematic answer is that
the present is strictly momentary, or of finite but very brief duration. The second
problem is new. A prime motivation for Present ism is the desire to stay close to
common sense, but common sense tells us that, even if the past isnt real, there
are definite facts about what happened in the past. How can this be if the past
is unreal? If the past doesnt exist, how can there be true statements about it?
The problem is especially acute if we subscribe to the truthmaker principle: any
contingent statement about the world that is true is true by virtue of something
in the world that makes it true its truthmaker. If the past doesnt exist, then the
truthmakers for statements about the past dont exist, so it seems that all state-
ments about the past must be false, or at least that none can be true. Needless to
say, this is radically counterintuitive.
Faced with this problem, the Presentist could simply bite the bullet and accept
that there are no true statements about what happened in the past, but this is
very hard to accept. Another option is to hold that statements about the past
do have truthmakers, but these truthmakers all exist within the present. If I say
Dinosaurs once walked on the Earth, what makes this sentence true is not the
existence of dinosaurs in the past, but rather the presently available evidence for
thinking that there used to be dinosaurs, such as fossils. This reductionism about
the past has two unpalatable consequences. First, the past becomes inconstant. If
we suppose (as most Presentists do) that there is more than one present, if different
presents include different traces of the past as they surely would these presents
have different pasts. Secondly, since facts about the past are constituted by what
exists now, the truth about the past is restricted by whatever relevant evidence
now exists. If we take evidence here to mean traces that would enable human
investi gators to reconstruct accurately a past occurrence, the conse quences of this
restriction are dramatic: the bulk of the past would vanish. The claim that only a
minuscule percentage of dinosaurs left fossilized remains would be false, since the
only dinosaurs that existed would be those that left fossilized remains!
DYNAMIC TIME 83
But we need not read the restriction in this way. We might reasonably hold
that the past includes everything that a being with god- like powers of observation
and calculation could infer about it, given a complete description of the present
state of the world down to the sub- atomic level. The truths about the past are not
confined to what we can discover or infer. From this standpoint, if our universe
is deterministic, the totality of facts about the past available to the realist and the
reductionist may well coincide. But if, as currently seems more likely, our universe
is not deterministic, even a being with god- like powers might be unable to infer
very much about what happened even a short while ago. The same will apply (at
least if we downgrade our imaginary beings powers somewhat) if chaos theor-
ists are right in their contention that the occurrence of macroscopic phenomena
can depend on minuscule variations in initial conditions; the so- called butterfly
effect.
The idea that the past is dependent on what exists in the present may seem
absurd and objectionable, but some Presentists have looked on the bright side:
Facts whose effects have disappeared altogether, and which even an omnis-
cient mind could not infer from those now occurring, belong to the realm
of possibility. One cannot say about them that they took place, but only that
they were possible. It is well that it should be so. There are hard moments of
suffering and still harder ones of guilt in everyones life. We should be glad to
be able to erase them not only from our memory but also from existence. We
may believe that when all the effects of these fateful moments are exhausted,
even should that happen only after our death, then their causes too will be
effaced from the world of actuality and pass into the realm of possibility.
Time calms our cares and brings forgiveness.
(Lukasiewicz 1967, quoted in Tooley 1997: 2356)
This last thought is consoling, but is it one that we can believe? We shall be
encounter ing other Presentist responses to the problem of the past in due course.
The claim that nothing exists that is not present is a slippery one, for it can be
interpreted in different ways; Presentism comes in different forms. Not all of these
Presentisms constitute dynamic models of time, but by taking a brief look at these
non- dynamic forms we not only improve our appreciation of what the Presentist
standpoint involves, we can also see more clearly the problems that face dynamic
forms of Presentism.
6.8 Solipsistic Presentism
The most radical version of Presentism is the doctrine that nothing exists that is
not present and that only one present ever exists; this one. This solipsism of the
present moment, as it is sometimes called, is invulnerable to decisive refutation
but impossible to take seriously.
If only this present ever exists we are left with the problem of explaining why
it seems that we have a past. Why do we have memories of a recent past? Why
are there so many signs of a past if the past doesnt exist, and never has? There is
a general principle of reasoning, argument from best explanation, according to
84 TIME AND SPACE
which it is reasonable to believe the best explanation of a phenomenon. Our world
contains all the appearances of having had a long past; isnt the best explanation
of these appearances that there was a past?
More generally, if only this thin reality- slice exists, why is it complex and organ-
ized? The big bang theory of the origin of the universe, coupled with the theory
of natural selection, explains why the world is as it is: how the current complexity
developed from initial simplicity. There is still a lot that is mysterious we do not
know why the big bang occurred, why the initial energy existed, why the laws of
nature are as they are but there is a good deal that is explained. By contrast,
the doctrine that the universe consists of a single brief slice of highly organized
reality, which simply popped into being from nothing, explains nothing and leaves
everything mysterious.
These points provide good reasons for rejecting Solipsistic Presentism. Bearing
them in mind, lets move on.
6.9 Many- Worlds Presentism
The Presentist says that nothing exists that is not present, but the Presentist is not
obliged to say that only one present exists. One option for pluralist Presentists
is to say that, although only what is present exists, there exists a succession of
presents, with each present departing from existence to make way for its successor.
I will examine this view in 6.10 and 6.11. A second option is to say that reality
as a whole includes many momentary presents that are not temporally related
to each other, and so do not succeed each other in any way. Since each of these
presents is a very brief self- contained world in its own right, I call this doctrine
Many-Worlds Presentism. (It should be noted that this version of Presentism is
quite distinct from the many- worlds interpretation of quantum theory, although
there may be a connection, as we shall see shortly.) For the rest of this section I
will call the sum total of these many worlds REALITY.
Since, on this view, REALITY consists of a timeless ensemble of momentary
worlds, it makes no sense to suppose that new worlds can come into being in
time; REALITY encompasses all the worlds that there will ever be, and so cannot
grow. This kind of many- worlds view is thus not dynamic (in the manner of the
growingshrinking block), but it is also distinct from the position of the orthodox
Block theorist. The latter also holds that there are many presents, at least to the
extent that every event in the block-universe is present at the time when it occurs.
But the Block theorist also holds that all these presents are temporally related to
one another, which is denied by the Many- Worlds Presentist. Furthermore, if we
suppose that things belong to the same world if they are spatially and temporally
related, then anyone in an extended block-universe who says My world consists
of only what is happening now is saying something false. This claim is true if
uttered in one of the short- lived worlds posited by the Many- Worlds Presentist.
Despite these differences, there are similarities between the two positions.
10
Block theorists hold that past, present and future are equally real. So, too, does the
Many- Worlds Presentist, after a fashion. If the Solipsist Presentist wants to admit
truths about past events, reductionism is the only option. The pluralist Presentist
DYNAMIC TIME 85
is not so restricted; since other presents exist, there are potential truthmakers for
statements about other times. Depending on the precise contents of REALITY, there
may even be truthmakers for all those statements about the past (or future) that
common sense tells us either are true or could be true.
The easiest way to appreciate how this might be so is as follows. Start by imagin-
ing that we do have a past, and a future, and that all coexist within a four-dimen-
sional block-universe. Now imagine this block carved into many brief slices, each
corresponding to the condition of the world at a particular time, and transport
these slices into timeless REALITY; to capture the fact that there are no spatial or
temporal relations among the slices, imagine that they are randomly shuffled like a
deck of cards. Next, consider all the ways that the slices could be brought together
and stacked; each of these imaginary decks corresponds to a different possible
history. We know (or believe) that, in our world, for the most part change is gradual
and orderly: we dont find whole cities vanishing or suddenly appearing, furniture
stays where it is put, cats dont change into frogs. Consequently, any ordering of
the slices that fails to capture this continuity will not be an accurate representation
of our history. Indeed, it is perfectly plausible to suppose that if we start from our
present, and find those slices that resemble it most closely, and place them in the
order that minimizes overall differences between immediate neighbours, we would
find ourselves with a unique ordering of slices, an ordering that would correspond
with what we believe about our past and (although this is less certain) anticipate
about our future. If we extend this best match method, and try to order all the
slices, we might well be able to reconstruct the block we started with in its entirety.
Reductionism is thus avoidable. A statement about the past (or future) is true if
the following conditions are met:
REALITY contains one or more slices that the statement accurately describes.
The time at which the events are said to occur corresponds to the location of
the slice(s) in the ordering generated by the best match method.
Of course, while it is possible that REALITY includes slices that contain the events
we believe to have occurred, there is no guarantee that it does; the past may be
more impoverished than we believe. But it may also be far richer; REALITY may
contain many more presents than are required to build our history (and future).
It may contain enough presents to build billions of world- stories, some similar to
our own, some very different. Since the best- match procedure might well situate
your current present in more than one world- story, you might well have several
different pasts, and several different futures. Indeed, if REALITY is rich enough,
all your dreams will come true. But this boon comes at a price: so, too, will your
worst nightmares.
The ability of Many- Worlds Presentism to accommodate (albeit in its own dis-
tinctive way) our common- sense beliefs about the past means that it is far easier to
accept than Solipsistic Presentism. But in another respect it is drastically counter-
intuitive. I for one find it hard to believe that I am living a life that is distributed
across many different worlds, and that I spend only the briefest of times in any
one world. The Presentist could, of course, respond by saying that all parts of
my life coexist in REALITY, and it is just that these parts are not related (spatially,
temporally, causally) in the ways usually assumed.
86 TIME AND SPACE
It may seem that the many- worlds view is vulnerable to the objection levelled at
the solipsist. If it is mystery- mongering to suppose that REALITY consists of just one
complex reality- slice containing all the appearances of having a long and complex
past, isnt it even more problematic to suppose that REALITY consists of many
such slices, all independent of one another? But the Presentist could reasonably
respond thus: Is the existence of these many unconnected presents any more
mysterious than the existence of the many connected presents that constitute a
block-universe? This point aside, pluralist Presentists have a further card up their
sleeve: developments in contemporary physics.
In The End of Time (1999), Barbour tries to persuade us that time does not exist.
What he really means is that time is not like we commonly think, for the positive
conception of reality he defends (so far as I understand it) is very close to that of
Many- Worlds Presentism. What is relevant to the objection that we are consider-
ing is that Barbour provides an explanation for why REALITY (he calls it Platonia)
should contain complex world- slices of the sort that we are acquainted with. Given
any physical system, the basic quantum laws enshrined in Schrdingers wave
equation tell us how that system will evolve over time. In fact, the equation deliv-
ers many different possible ways that the system might evolve, and provides no
way of distinguishing which of these possible evolutions will be actual. Adherents
of the many worlds interpretation of quantum theory take the view that it is a
mistake to suppose that just one of these solutions to the equation becomes actual;
they maintain that they are all equally real, and that as a consequence reality has a
branching structure, and is vastly richer than we normally assume. Barbour himself
adopts this many- worlds interpreta tion applied to a Schrdinger equation speci-
fying all the (physically) possible ways the universe could have developed from an
initial big bang and then takes a further step.
We know that standard quantum theory cannot be complete since it is not
integrated with Einsteins general theory of relativity, which is currently our best
theory of gravity. There have been many attempts to bring relativity and quan-
tum theory together, but most attempts have failed. One of the few noteworthy
(although limited) successes is the Wheeler- DeWitt equation.
what is this equation, and what does it tell us about the nature of time?
The most direct and nave interpretation is that it is a stationary Schrdinger
equation for one fixed value (zero) of the energy of the universe. This, if true,
is remarkable, for the Wheeler- DeWitt equation must, by its nature, be the
fundamental equation of the universe [it] is telling us, in its most direct
interpretation, that the universe in its entirety is like some huge molecule in
a stationary state. (Barbour 1999: 247)
The universe, thus represented, is still: it does not evolve for it does not change
over time (since time is absent), and so change itself is rendered illusory (or at best
emergent). This timeless universe is nonetheless richly structured, for it com-
prises an infinite number of momentary three- dimensional worlds, one for each
point of the phase space generated by the Wheeler- DeWitt equation.
11
Barbour
argues that we have the impression that we live in an evolving universe because
there are momentary fragments of this timeless whole (those we inhabit) that
contain apparent traces of earlier and later occurrences. But in reality there is no
DYNAMIC TIME 87
earlier and later, since these three- dimensional world- fragments are tempor-
ally unrelated to one another.
As Barbour concedes, the Wheeler- DeWitt equation is surrounded by contro-
versy. Nonetheless, the canonical quantum gravity programme, which delivers
this equation, is one of the leading contemporary approaches to quantum gravity.
12
Barbour is certainly not alone in believing that this programme poses a threat to
time, for as we saw in 3.8, it is widely accepted that it does; Kuchar writes of the
quantum calamity, known as the problem of frozen time (1999: 178).
What is controversial is Barbours response to this time problem; other theor-
ists are optimistic that they will find time hidden or buried in the Wheeler-
DeWitt equation.
13
But it is clear that from the perspective of physics (if not
common sense), his view is by no means absurd. If something akin to the Wheeler-
DeWitt equation provides the most accurate and fundamental representation
of our world, then perhaps, as Barbour urges, our world does consist of a vast
ensemble of temporally unrelated moments, each of which consists of a three-
dimensional configuration of physical bodies.
Needless to say, there is much that is speculative and controversial here, but the
conclusion is nonetheless surprising: Many- Worlds Presentism is a doctrine that at
least merits serious consideration.
6.10 Dynamic Presentism
The version of Presentism that is easiest to believe at least from the standpoint
of common sense is Presentism of the dynamic variety: the doctrine that reality
takes the form of a succession of instantaneous (or near-instantaneous) presents;
no sooner has one present come into existence than it will depart from it as it is
replaced by another present, and the latter will in turn be immediately annihilated
as a new present comes into existence. Although this version of Presentism remains
committed to the core claim that only what is present is real, by virtue of holding
that presents succeed one another, it is capable of acknowledging that there were
other times and events, and (with luck) there will be others in the future.
14
It may be pleasing to the eyes and ears of common sense, but Dynamic Pre-
sentism nonetheless faces a number of significant hurdles. Since the mid-1990s,
however, interest in the doctrine has been gradually increasing, and a good deal
of effort and ingenuity has been put into developing solutions to the various prob-
lems confronting this form of Presentism. Some of the more important recent
developments are summarized below.
A distinctive doctrine?
Perhaps the least obvious of the problems is the question of how precisely Present-
ism in its dynamic guise should be formulated. Looking at diagrams such as Figure
1.6, where the Presentist model is contrasted with the main alternatives, it may
seem utterly and entirely obvious that we are dealing here with entirely different
conceptions of the large-scale structure of the universe; on both the dynamic and
non-dynamic Block views, reality takes the form of a four-dimensional expanse,
88 TIME AND SPACE
extended (as of now) through billions of years in the temporal direction, whereas
for the Presentist reality consists of nothing but a single three-dimensional slice,
which is instantaneous (or near-instantaneous). What could be more different?
Indeed, in the case of Solipsistic Presentism, which posits a single present P, and
equates the sum total of reality with P, the difference is hard to deny. But by
acknowledging a succession of presents, P
1
, P
2
, P
3
, P
4
, the Dynamic Presentist
muddies the waters a good deal.
Consider this question: given this multiplicity of presents, where precisely does
the Dynamic Presentists conception of the universe differ from that of the ortho-
dox Block theorist? It may seem that there is no difference. After all, for any time
t, past, present or future, is it not the case that both camps can agree that if X
happens or exists at t, then at (or as of) t, X is present? Block theorists will certainly
have no quibble with this: they are perfectly willing to accept that the predicate
is present can truly be ascribed to events at the times at which they occur. But so
too can Presentists, at least of the dynamic persuasion, since they acknowledge that
times other than the current present did exist, or will exist. As of now, Napoleon
does not form part of the sum total of what exists, or so the Presentist will insist,
but Napoleon certainly existed at noon on 18 June 1815, and at this time he was
present or so the Presentist might also insist. So is there a real difference between
these supposedly different and competing models of time?
15
One way of meeting this objection is to hold that the differences between Pre-
sentists and the Block theory clearly and cleanly emerge when we distinguish
between restricted and unrestricted claims about existence. Many ordinary claims
about what exists are intended to apply to restricted domains, even if this is not
explicitly stated. If your house has been flooded by a recent downpour, and you
say Everything is wet!, you probably mean Everything in my house is now wet,
not Everything in the entire universe, past, present and future, is wet; in this
case, and the many others like it, the existence-claim (or its associated domain
of quantification) is tacitly restricted in a context-driven way. But there are also
plenty of instances where claims about existence (or intended quantificational
domains) are not restricted in any obvious way; if someone says Numbers are
abstract entities they are most probably making a claim about all numbers, not
some particular subclass of numbers, or numbers at some times or places but not
others and similarly for Vertebrates have backbones, or Water is wet. Let us
stipulate that, for present purposes, the ordinary existential quantifier, - (or -x),
is maximally unrestricted: it is always as unrestricted as it can be. Accordingly, a
complete response to the question What exists?, where exists is to be taken as
meaning exists in the most general, unrestricted sense will take the form of an
existentially quantified statement (-x)(x
1
, x
2
, x
3
, ), where the xs are an exhaus-
tive list of all the things, events or entities that exist, with no restriction to time,
place or kind. For the Block theorist, the answer to the What exists? question
will be the same, no matter when or where the question is asked: the list (x
1
, x
2
,
x
3
, ) will include the complete contents of the Block universe, past, present and
future. For the Presentist, if the question answered at a time t, the list (x
1
, x
2
, x
3
,
) will include only what is present at t.
This may seem to supply what we are looking for, but there remains a problem.
Proponents of the most plausible form of Dynamic Presentism do not merely
DYNAMIC TIME 89
claim that presents occur successively, they also recognize that statements about
the past Socrates taught Plato, Napoleon was defeated in 1815 at Waterloo,
and the like can be true, even though the past is unreal. Quite how such claims
can be true, if the past does not exist, is a controversial issue to which we shall be
returning shortly, but many Presentists believe that their doctrine can be rendered
compatible with most (perhaps all) of our ordinary beliefs about the character and
content of the past. And, of course, since our current present will soon give way
to another (and then another, and another), there is no reason for Presentists to
reject all talk of what will happen as meaningless. Given this, Presentists who wish
to minimize the revisionary consequences of their view of time will be reluctant to
subscribe to the doctrine that the most general and unrestricted sense of exists
is confined to the present. They will accept that when we ordinarily make very
general claims such as Green vegetables are nutritious, There have been forty-
four presidents of the United States, we are tacitly using a notion of existence, or
at least allowing our quantifiers in this case universal and existential to range
over objects that existed in the past (and in some cases, perhaps the future) as well
as the present. Even if strictly speaking only what is present exists, we nonetheless
need to allow ourselves to talk about past and future objects.
To accommodate this, Presentists can distinguish between the ordinary existen-
tial quantifier, one that they will accept is not confined to presently existing things,
and a more basic quantifier, call it -S, which is confined to presently existing
things. It is this latter quantifier that we use when we want to talk about existence
of the most fundamental or primitive sort, existence simpliciter, as we might call
it. Needless to say, the only things that the Presentist will accept as existing sim-
pliciter are presently existing things. As for the ordinary existential quantifier, for
the Presentist at least, if this ranges over the past and future as well as the present,
it cannot be fundamental. Hence a claim of this sort:
(-x)(x is green)
Or There exists something which is green will be construed as making a more
complex claim along these lines:
(-Sx)(x is green) or WAS(x is green) or WILL BE(x is green)
Translated into ordinary language, this amounts to the following: either something
that exists simpliciter is now green, or something was green, or something will be
green. Crucially, for claims made using the tensed operators it was the case that
P and it will be the case that P (symbolized above by WAS and WILL BE) to
be true, it is not necessary that the objects referred to by the proposition P exist
simpliciter, or so the Presentist will insist. (In this regard they are analogous to the
modal operator it is possible that P; there do not have to be any chess-playing
birds for the statement It is possible that there are chess-playing birds to be
true.) For those who subscribe to the Block theory, the situation is very different.
On this view, the entirety of the four-dimensional universe exists in the same
basic way: hence all objects exist simpliciter, no matter where or when they are
to be found. Accordingly, the Block theorist regards the ordinary (unrestricted)
existential quantifier - as well suited to talking about existence of the most funda-
mental kind. Rather than being a complex composed of conceptually more basic
90 TIME AND SPACE
constituents, it is - that is primitive. As for the Presentists privileged quantifier,
-S, Block theorists will regard this as derivative; from their perspective we arrive
at -S by imposing temporal restrictions on the more fundamental -.
There may be more to be said, but it looks as though Dynamic Presentists have
a response to the charge that they are not offering anything that is ontologically
distinctive vis--vis the Block theory; for more on this theme see Sider (1999,
2006), Crisp (2004), Frances (in progress). As we shall now see, the differences
between the Presentist and Block theorist are by no means confined to the inter-
pretation of exists.
The problem of the past
Issues concerning formulation aside, the Dynamic Presentist is confronted with the
awkward, and potentially very damaging, problem of explaining how there can be
truths about the past if the past is unreal. Presentists who believe that statements
about contingent events lying in future can also have definite truth-values as of the
present are confronted by an analogous difficulty, since for Presentists (of all per-
suasions) the future is as unreal as the past. However, as just noted, most Dynamic
Presentists wish their position to be compatible with as many of our ordinary,
pre-theoretical, beliefs about time as is possible; those who incline to this view
usually do so in large part because they believe that no other conception of time is
more compatible with our ordinary beliefs. Consequently, since the common-sense
position or so it commonly supposed is that the future is unreal (or, at the very
least, indeterminate), many Dynamic Presentists are quite happy to leave it that
way. Hence the more pressing problem is posed by the past.
We encountered the essentials of the problem in 6.7. The truthmaker prin-
ciple has a good deal of plausibility: contingent statements are true because some-
thing in the world makes them true their truthmaker. Since for orthodox and
Growing Block theorists the past is just as real as the present, there is no shortage
of truthmakers for statements about the past; since Presentists hold the past to be
unreal, it seems there are no available truthmakers for statements about the past.
So how can such statements be true? What distinguishes claims about the past from
the claims found in fiction or fantasy?
This is the problem posed by the past in its most general form, but there are
further and related puzzles of a more specific kind. It is natural to think that much
of what we find in the present is a causal consequence of what has happened in the
past, directly or indirectly. Causation is often taken to be a relation between earlier
and later events, and relations between concrete particulars (such as events) are
often taken to be existence-entailing, that is, for the relation to hold both relata
must exist. Since a cause c will typically precede its effect e, if only what is present
is real, the cause will not exist as of the time of the effect. If Presentists want their
model of time to be compatible with the reality of causation, they will need to find
an alternative way of thinking of the latter.
A further problem is posed by referring expressions such as Socrates or
Napoleon or Saturn. In the eyes of many contemporary philosophers, the
meaning of expressions such as these is essentially linked to the particular indi-
vidual they refer to, with the result that Socrates would not have the meaning it
DYNAMIC TIME 91
does have if Socrates had not existed. The problem this poses for the Presentist is
plain: if the past is wholly unreal, then so too is Socrates; if Socrates is unreal, what
sort of meaning (if any) can the name Socrates have? It certainly cannot have the
same sort of meaning as Saturn if the Russellian conception of singular terms
is correct; on this view, the proposition Saturn has rings contains the planet
Saturn as one its constituents. If Socrates does not exist, he presumably cannot be
a constituent of any propositions, and so on the Russellian view, at least there
can be no propositions about Socrates himself.
Let us focus first on the general problem of how there can be truths about the
past at all. One option, also discussed in 6.7, is reductionism: perhaps past-tensed
statements are made true by presently available evidence (fossilized dinosaur foot-
prints, newspaper reports, historical records, and the like). This is certainly one
way of distinguishing factual statements about the past from historical fiction or
fantasy but, as we saw, it is not without its drawbacks. Since vast tracts of the past
(or what we would ordinarily think of as such) have departed without leaving
any readily detectable traces in the present, the reductionist approach leaves the
past greatly diminished. This problem is itself diminished, although by no means
eliminated, if we modify the reductionist approach by holding that the past is
constituted by what a being with god-like powers of observation and ratiocina-
tion could deduce from the totality of presently available evidence. But if, as
seems quite likely, our world is indeterministic, appealing to god-like powers will
still leave the reductionist with an incomplete version of the past. In the light of
these disadvantages, it would not be surprising if Presentists particularly of the
dynamic persuasion sought alternatives. And alternatives there are.
The Lucretian option
One alternative course is to seek to remedy or diminish the deficiencies of reduc-
tionism by continuing to hold that facts about the past are determined by fea-
tures of the present, but introduce a novel kind of feature, one not recognized by
ordinary reductionism, one that is better suited to underpinning the full realm of
facts about the past that are recognized by common sense. The Stoic philosopher
Lucretius had an economical view of reality: he seems to have held that it consists
of nothing but presently existing atoms and the void. This might seem to make
for a problem. If reality is restricted to what presently exists, how can there be
facts about events in the distant past, for example that Helen of Troy was ravished,
that the Trojans themselves were conquered? To accommodate such facts, are we
not obliged to recognize that there is more to reality than what currently exists?
Lucretius argued not: we could put it that whatever has taken place is an accident
of a particular tract of Earth or of the space it occupied (1951: 41). Bigelow
discerns in this doctrine of the Epicureans the outlines of a general solution to
the problem of the past: the truthmakers for claims about the past are properties
of a distinctive kind, which are possessed by presently existing things. The state-
ment that the Trojans were conquered is made true by a property, for example the
Trojans were conquered here, that is presently possessed by a particular portion
of land, or perhaps more plausibly given the scattering of matter particles over
time a region of space. And the same holds more generally:
92 TIME AND SPACE
What exists in the present is a tract of land which has the accident, of being
the earth on which a spark of love kindled a dazzling blaze of pitiless war, of
being the space within which a Wooden Horse set the towers of Illium aflame
through the midnight issue of Greeks from its womb. All that is present are
matter, vacuity, and properties and accidents of these; and among these prop-
erties and accidents are some which are expressed in language by using the
past or present tense. We do not need to suppose the existence of any past or
future things, however, only the possession by present things of properties and
accidents expressed using the past or future tense. (Bigelow 1996: 46)
In principle there is no reason why there could not be properties of this sort
Lucretian properties, as we might call them to act as truthmakers for every
true past-tensed statement. Rather than taking them to be possessed by particular
present objects or regions, Bigelow suggests there are advantages in viewing them
as possessed by the world as a whole.
16
On the face of it, Lucretian properties provide Dynamic Presentists with much
of what they need. There are, needless to say, further issues to be resolved. So far
as causation is concerned, Bigelow points out that one of the options available to
the Presentist is to hold that causation is a relation between presently true proposi-
tions, rather than earlier and later objects or events. Suppose a man receives a knife
wound to the heart; he will soon die because of it. Rather than taking the causal
relation here to involve an earlier thing (a wound) and a later thing (a death), we
take it to hold between the presently true proposition that the man in question
has a serious heart wound, and the true proposition that he will die (ibid.: 42).
17

What of singular referring terms, such as Helen of Troy? The Presentist has
no option but to reject the Russellian view. One alternative is to opt for a purely
descriptive account of such terms that is, we take Helen to be semantically
equivalent to a general descriptive expression, there was an x who was a woman,
was abducted by Paris, had a face that launched a thousand ships, an analysis
that some (athough by no means all) philosophers favour for independent reasons.
Another option is to hold that reference to individuals is secured by means of a
property of a special sort: an individual essence, or haecceity. The haecceity of
Helen is the property of being identical with Helen. If we hold that all individuals
have individual essences of this kind, and that individual essences continue to exist
after their subjects pass away, we have at least potentially a way of referring
to individuals (or at least, their unique individual essences) when they no longer
exist. Properties of this sort are highly controversial, but they offer an alternative
to Presentists who are reluctant to embrace the descriptivist option. For further
discussion and options, see Adams (1986), Sider (1999), Crisp (2003: 3), Keller
(2004: 3.2) and Markosian (2004: 3).
Taking a step back, Lucretian properties may supply Presentists with much of
what they need, but when scrutinized more closely they soon begin to look rather
peculiar. The presently instantiated Lucretian property was conquered by the Tro-
jans does not look to be a relational property, since the Trojans are not presently
part of the sum total of reality. If it is not relational, then presumably it will be an
intrinsic property. But most intrinsic properties make an obvious difference to the
character or nature of the objects that happen to possess them. Shape is an intrinsic
DYNAMIC TIME 93
property, and if a piece of clay is shaped into a cube at a time t, it is clearly very
different from how it would be if it were spherical at t. But the difference Lucre-
tian properties make to what possesses them is a good deal more elusive. Even if
our world has the past we normally assume it to have, there is a logically possible
world W
1
that is an exact replica of our world as it now is at least with regard
to all normal, non-Lucretian properties but which was created only a couple of
seconds ago. Since W
1
lacks a past (worth speaking of), it lacks all the Lucretian
properties that our world possesses. Yet if you were somehow to find yourself in
W
1
, you would be none the wiser: the things in it would appear indistinguishable
from their counterparts in our world indeed, they are indistinguishable, save
for the absence of the decidedly mysterious past-oriented Lucretian properties.
It could be argued, and has been, that we are deluding ourselves if we think
anything as peculiar as Lucretian properties are really properties at all: see Sider
(2001: 3541, 2003: 185), Crisp (2003: 3, 2007: 3) Merricks (2007: 1356)
and Cameron (2010: 2) for more on this.
Ersatzer Presentism
Dynamic Presentists who are suspicious of Lucretian properties do have an alter-
native: they can hold that the truthmakers for past-tensed statements are to be
found in a (so-called) ersatz B-series. By so doing, or so it is claimed, the Presen-
tist can talk about past times and past events in much the same way as the Block
theorist, but without the same ontological commitments. Variants of this proposal
have been developed by Markosian (2004: 3.7, 3.10) and Crisp (2007: 4), and
in most detail by Bourne (2006a,b).
Variations aside, the core idea is reasonably straightforward. To start with, let
us assume the past is as rich and determinate as the Block theorist takes it to be. As
things stand, we know very little about what was occurring in the neighbourhoods
of distant stars at any given occasion in the past in fact, we have little detailed
knowledge about was taking place over large areas of our own planet but we
ordinarily suppose there is nonetheless a determinate fact of the matter as to what
was occurring at every location in space, at every moment in history. If this realist
assumption is correct, then for each and every moment in the past there will be a
complete, maximally detailed, description of what was going on in the entire uni-
verse at that instant. This description can be put in the form of a single proposition;
we can call these complete instantaneous state (CIS) propositions. We can further
suppose, in the orthodox way, that propositions are abstract entities, and that this
applies to CIS propositions too.
Now, in response to the question On your view, what is a time?, a Block theorist
will find it natural to say The totality of material contents of the four-dimensional
universe that exist at an instant, or perhaps Those collections of spacetime points
that are simultaneous with one another. Since Presentists believe that only pres-
ently existing points or objects are real, if they answer What is a time? in the same
way, they will only be able to recognize the existence of just one time, the present.
Hence the appeal of a different answer: they can take times to be CIS-propositions.
Since CIS-propositions are abstract entities, they exist in the same sempiternal way
as numbers or sets: they are unperturbed by the annihilations and creations that
94 TIME AND SPACE
occur in the material universe. Just as times (or momentary states) in the familiar
block-universe form a series ordered by the earlier than relation, so too can ersatz
times in the form of CIS-propositions: the latter can be E-related. Since earlier than
holds among concrete events and real times, and the E-relation holds between time-
less abstract entities, the E-relation must be distinct from the more familiar earlier
than, but it can have the same logical properties: both relations are asymmetrical,
irreflexive and transitive (or so we can suppose). The ersatz B-series is simply the set
of CIS-propositions that is ordered by the E-relation. In taking times to be nothing
more than abstract CIS-propositions, ersatzer Presentists are committed to holding
that the present time is also an abstract entity. However, they will insist that the
present time is distinctive, for it is the only time whose constituent propositions are
all true: it is the only time that is concretely realized. Of course this will change: the
times in the year 2030 have not yet been concretely realized, but they will be. On
this view, the passage of time consists in the CIS-propositions in the abstract ersatz
B-series being concretely realized, one by one.
18
With the ersatz B-series in play, it is possible to explicate talk of earlier and
later times without relying on tensed properties, such as being past or being future.
A time t is past if and only if it is E-related to the present time, and t is future if
and only if the present time is E-related to t; and of course a time t is present if
and only if that times CIS-proposition is true, or concretely realized. The ersatz
B-series also allows us to provide truthmakes for past-tensed statements without
employing Lucretian properties. So, for example, Dinosaurs roamed the earth
is true only if Dinosaurs are roaming the earth is true at one or more past times,
and Aliens will walk the earth is true only if Aliens are walking the earth is true
at one or more future times. Of course, Presentists who opt for an open future will
deny there are truths about future contingencies: Bourne (2006a: ch. 2) develops
an ersatz B-series with a branching future.
By allowing us to talk about past (and perhaps future) times and events, the
ersatz B-series puts Presentists in a stronger position when it comes to handling
causation. According to regularity theorists, facts about what causes what super-
vene on regular patterns among event-types: so it is true to say that the striking of
a particular match causes it to ignite if past (and perhaps future) strikings of similar
matches were followed by ignitions. Since the ersatzer Presentist has no trouble
accommodating talk of past (or future) regularities of this sort, causation in this
form can also be accommodated without difficulty. See Bourne (ibid.: ch. 4) for
a Presentist treatment of other accounts of causation (e.g. Lewiss counterfactual
theory). As for singular reference, although an account along descriptivist lines
remains an option, so too is a weakened version of the causal theory of reference;
see Bourne (ibid.: 99108) for more detail.
The ersatzist version of Dynamic Presentism has its advantages, but it too is
vulnerable on several fronts. It cannot accommodate singular reference to past
individuals in the simple and robust manner of the Block theory; since Socra-
tes is not a part of reality, he (the concrete individual) cannot be a constituent
of propositions that make reference to Socrates. And while causation in some
forms can be handled, the ersatzist must still reject the widely held view that
causation is an existence-entailing relation between concrete events or objects.
There is a further difficulty. The ersatz B-series is composed of a particular set
DYNAMIC TIME 95
of CIS-propositions, those that are E-related, and which also correspond with the
actual history (and perhaps future) of our universe. However, taking a broader
view, there is an infinite number of CIS-propositions and so times that are not
part of the ersatz B-series, namely, those describing possible ways the universe
might be that are never actualized, ways that do not feature anywhere in our past,
or our future for example CIS-propositions describing possible but non-actual
worlds where Napoleon did not lose at Waterloo, or where penguins and pigs
are frequently found flying, and so forth. Why are some times included in the
ersatz B-theory and not others? What underpins the selection? Bourne (2006a:
65) and Crisp (2007: 4.2) concur that nothing does: it is simply a brute fact that
some times are E-related in the way they actually are. This may seem somewhat
unsatisfactory, they also argue that they are not alone in being in this predicament,
for Block theorists face an analogous problem. Why are the concrete objects and
events in the four-dimensional manifold organized as they in fact are? Is there any
informative answer to be given to the question: why does the Block universe have
the character and contents that it does? It is not obvious that there is.
19
6.11 Compound Presentism
To close I want to take a brief look at a rather different way of developing a Presen-
tist model of time. It could be argued, with some justification, that the conception I
have in mind is not strictly a version of Presentism at all, and hence that it is at best
a neo-Presentist position. The terminology is largely irrelevant: the model in ques-
tion is undeniably very much closer to Presentism than any of the other dynamic
models, and so it is appropriate to consider it in this context especially since it
is less problematic in certain respects than the stricter versions of the doctrine.
William James once characterized our immediate experience of temporality in
this way:
Objects fade out of consciousness slowly. If the present thought is of A B C
D E F G, the next one will be of B C D E F G H, and the one after that of
C D E F G H I the lingerings of the past dropping successively away, and
the incomings of the new, are the germs of memory and expectation, the
retrospective and the prospective sense of time. They give that continuity to
consciousness without which it could not be called a stream. (1952: 397)
Irrespective of the merits of this as a description of our consciousness of time,
might it not be that time itself has a structure of this sort? In Jamess example,
seven distinct elements coexist in the form of a succession, but such extravagance
is not needed; the required overlap requires only two successive elements, and
these elements could be very brief indeed. An approximate description of a world
of this sort might run thus:
We start off by assuming that the sum total of reality consists of two coex-
isting but non-simultaneous very brief reality- slices (each spatially three-
dimensional). Suppose A and B are two such, and that A exists at t
1
and B at
t
2
. One of these slices, A, is annihilated and a new slice of reality, C, comes
96 TIME AND SPACE
into existence, and with it a new time, t
3
. Slice B is annihilated and D is cre-
ated, along with t
4
; and so it goes on.
A scenario along these lines is depicted in Figure 6.5, where S
1
S
4
refer to dis-
tinct sum totals of reality, and t
1
t
5
to distinct times. In one respect this diagram
may mislead. It generates the appearance that all these times and events coexist,
when ex hypothesi they do not: temporal dynamism cannot be fully captured in a
static diagram. Two questions need to be kept distinct:
1. What events occur at t
2
?
2. What events are real as of t
2
?
The answer to 1 is obvious: The events that occur in reality-slice B. The answer
to 2 is slightly less obvious. Let us suppose, as seems reasonable, that X is real
as of Y is true if and only if X and Y are times or events that coexist, either sim-
ultaneously or in succession. (And, as usual in the context of dynamic temporal
models, we are taking coexist to be a primitive notion.) With this stipulation in
place we are in a position to answer question 2. Since both A and C coexist with
B, the answer to What events are real as of t
2
? is The events that occur in A,
B and C. This may seem odd: how can the events in A, B and C be real as of t
2

if there is no sum total of reality that includes each of A, B and C? But as we are
using the phrase here, a sum total of reality consists of a collection of times and
events that (timelessly) coexist, and A and C do not coexist in this fashion, despite
the fact that there is a sum total comprising AB, and another comprising BC.
There are a couple of further points to note. Looking at Figure 6.5, it is easy to
fall into thinking that two happenings occur simultaneously at time t
3
: (i) C comes
into being, and (ii) A is annihilated. This way of thinking may be a natural one, but
it is not entirely accurate. With regard to (i), it would be a mistake to think that
t
3
is already there, waiting for the arrival of C (or some other content); to avoid
problems with overdetermination we must hold that t
3
and C come into being
together. With regard to (ii), it would be wrong to suppose that As annihilation
occurs at t
3
: all that happens at t
3
are the events which jointly compose C. Strictly
speaking, the annihilation of A does not happen at any time at all, least of all t
1

(which is annihilated along with A, again avoiding overdetermination). Episodes
of absolute annihilation and creation are not things that occur in time, in the man-
ner of ordinary events; they are better thought of as alterations to time, involving,
as they do, variations in the sum total of what exists.
I shall call the model of temporal passage just outlined Compound Present-
ism. Since neither A nor B (nor any other reality-slice save for the first and last,
Figure 6.5 Compound Presentism. A doubly dynamic
reality: the sum total of reality is not limited to a single present,
and absolute becoming is matched by absolute annihilation.
A B
B C
C D
D E
DYNAMIC TIME 97
if there are such) exists all by itself, neither is a present in the strict and unique
sense of the Dynamic (or Solipsist) Presentist. Nonetheless, since according to this
view the vast bulk of (what we would ordinarily take to be) the past and future do
not exist, this position is much closer to that of the Dynamic Presentist than the
Block theorist, so the label is not without its merits. To avoid confusion, I shall
refer to sum totals such as [AB] and [CD] as extended presents.
Compound Presentism has several significant advantage over its more isola-
tionist kin. First, it does not face any problem with trans-temporal relations. If
two successive reality-slices A and B coexist, there is no problem in supposing
that existence-entailing relations hold between them. Consequently, if causation
properly so-called can hold only between events that are both real, then successive
events can be causally related. Indeed, so far as causal relations between tempor-
ally adjacent events are concerned, Compound Presentists are at no disadvantage
whatsoever with regard to Block theorists. This means that Compound Presentists
can explain how the past can leave traces in the present in a perfectly straightfor-
ward way, without having to turn to the ingenious but questionable stratagems
that other Presentists are obliged to rely on. More generally, the successive brief
phases of a Compound Presentist universe clearly and unambiguously belong to
one and the same universe. This is not something that can be said of the solitary
momentary phases to which other versions of Presentism are confined. Nor is it
clear that the situation in this regard is improved in the slightest by the Lucretian
properties or ersatz B-series to which the orthodox forms of Dynamic Presentism
appeal in their attempts to avoid reductionism about the past.
There is a second advantage. According to orthodox Presentists, the present is
either strictly instantaneous, and so lacking any extension whatsoever, or it has
a minimal extension, that is it possesses some finite duration, but is too brief to
encompass any sort of change, even the most fleeting (if it could encompass change,
then it would contain earlier and later phases that are non-simultaneous, and a
strictly conceived present does not contain phases that are non- simultaneous).
Arguably, both of these options are problematic. With regard to the first option,
there have always been those who have maintained, with some considerable plau-
sibility, that it is a serious mistake to suppose that reality could be entirely com-
posed of zero-sized parts, even if these parts are infinite in number; for more on
this see the discussion of Zenos paradox of plurality in 16.4, 16.6 and 17.5.
Granting the present a finite but minimal extent avoids this problem, but another
is lying in wait. The finite-but-minimal present is too brief to encompass even the
most fleeting of changes, but on the face of it change is something we directly
perceive. Can we not hear one note following on from another in a melody? Do we
not see peoples lips move when they talk, or the runner dashing towards the fin-
ishing line? The doctrine that changes can indeed be perceived has been endorsed
by many eminent figures including James, Bergson, Russell, Broad and Husserl
and the Compound Presentist has no difficulty whatsoever accommodating this.
It is by no means clear that the orthodox Presentist can do likewise.
20
Although these gains look decidedly non-trivial, no doubt there will be those
who find what is being offered by the Compound Presentist metaphysically prob-
lematic; Le Poidevin (2004: 488) is one such. Looking again at Figure 6.5, it could
be maintained that to make proper sense of what is shown there the extended
98 TIME AND SPACE
present [AB] giving way to [BC], which in turn makes way for [CD], and
then [DE] we need to introduce an additional temporal dimension, running
orthogonally to that of ordinary time. In which case, the vertical column listing
the succession of extended presents, S
1
S
4
is incomplete as it stands, for it should
read: [S
1
obtains at meta-time t
1
], [S
2
obtains at meta-time t
2
], and so forth. In
this fashion, McTaggarts paradox returns to the scene once more. Some may find
an objection along these lines plausible, for introducing an additional temporal
dimension probably does make it easier to understand what is going on: we are,
after all, accustomed to plotting changes or variations against time. (Although, of
course, understanding how reality itself could possess two orthogonal temporal
dimensions is not so easy.)
Compound Presentists have a response to this objection: might it not be that
McTaggart-inspired objections of this sort are simply begging the question against
their position? They can argue, with some plausibility, that the demand for a
second temporal dimension derives from an underlying assumption, to the effect
that all the contents of our universe have a location in a single unified dimension,
the contents of which are all equally real. The Compound Presentist, in common
with the Growing Block theorist, is proposing a different picture, a model of an
essentially fragmented reality (as Fine calls it), a model in which the sum total of
what is real, of what exists simpliciter, is not constant over time.
21
Again return-
ing to Figure 6.5, reality-slices A, B and C are real as of t
2
, whereas C, D and E
are real as of t
4
. The story being told of varying sum totals is entirely grounded in
coexistence relations: A and B coexist with one another, as do B and C, but B and
E do not, nor do A and C or A and D. If this pattern of coexistence is intelligible,
why think we also need a meta-time to help us make sense of the situation? Figure
6.6 suggests there is no such need: the required pattern is easily accommodated in
a single temporal dimension.
A B C E D
Figure 6.6 Compound Presentism: a succession
of extended presents forming a single temporal
dimension. The pattern of coexistence relationships (here
indicated by double-headed arrows) characteristic of
Compound Presentism do not require a meta-time. time
The key issue may well be whether coexistence can behave in the manner the
Compound Presentist needs. Some may have their doubts. It is natural to suppose
that for any three items, X, Y and Z, if X coexists with Y, and Y coexists with Z,
then X coexists with Z; that is, that the relationship of coexistence is transitive. If
coexistence is transitive, Compound Presentism is incoherent; if B coexists with
both A and C, then A and C must coexist with one another, and the same would
apply to A and D, and also to A and E. Far from being a dynamic model, Com-
pound Presentism collapses back into the familiar Block model of time.
This problem is potentially a serious one, evidently, but must we assume that
coexistence is transitive in all contexts? The Compound Presentist can argue thus:
DYNAMIC TIME 99
Coexistence is certainly symmetrical, but it neednt be transitive; to suppose
otherwise is simply to deny the dynamic nature of time, which involves pre-
cisely the coming- into- being and departing- from- being of times and events.
Coexistence is transitive in space (at a given time), but to impose transitivity
on time amounts to an unjustifiable spatialization of the latter.
This reply should be treated with some respect. The only obvious way that coexist-
ence could fail to be transitive is if events can cease to exist and come into being.
The failure of transitivity is inevitable if the passage of time involves annihilation
and creation, a doctrine long ago advocated by Heraclitus: This world neither any
god nor man made, but it always was and is and will be, an ever- living fire, kindling
in measures and being extinguished in measures (quoted in Barnes 1982: 61).
Let us move on. Thus far we have been considering a simple case in which the
extended present has two very brief parts, but there are many other possibilities.
Two are depicted in Figure 6.7. The worlds shown here are indistinguishable in
all respects save one: the extent of their extended presents. The events within
these extended presents are of precisely the same durations in both worlds, but
the periods of time relative to which a given moment of time is real are different.
C C
C C
C C E E
E
F
A
B D
D D
D
D B
B
B A
Figure 6.7 Extended presents of
different durations. The duration of the
extended present provides the Compound
Presentist with a measure of temporal
dynamism that does not require an external
meta-time.
World 2 World 3
Since the longer a worlds extended present is, the closer that world is to a
Block universe, there is a sense in which worlds with shorter extended presents
are more dynamic than worlds with longer extended presents. One might even go
so far as to say that the length of the extended present provides a measure of the
rate of times passage. But since this rate is determined solely by relationships
of coexistence within a world there is no need to posit an external meta- time to
render it intelligible.
The status of the past in Compound Presentism is ambiguous, since in this frame-
work the past comes in two forms: the real and the unreal. In our original case, if
we assume that t
1
occurs earlier than t
2
, which occurs earlier than t
3
, then A belongs
to the real past of B, but the unreal past of C. More generally, the real past of a time t
consists of those earlier times and events that coexist with t within a single extended
present, whereas the unreal past of t consists of those earlier times that do not coexist
with t, but are linked to t by what we can call indirect coexistence. Two times are
linked by indirect coexistence if they are at either end of a succession of times, each
of which is related to its neighbours by direct (genuine) coexistence (i.e. they are
linked by a succession of partially overlapping extended presents).
The status of the real past is unproblematic; it is precisely the same as in the
four-dimensional Block model.
22
Of course, there is the additional complication
that the extent of the real past can be different in different worlds; moreover, in
worlds where the extent of the extended present is inconstant, different times
100 TIME AND SPACE
within the same world will have real pasts of different lengths. But these complica-
tions are irrelevant to the ontological status of the real past. The real past is just
as real as any part of the extended present, and hence just as real as the narrow
present, as we might call it: the part of the extended present that has most recently
come into existence.
But what of the unreal past? How can statements about the unreal past be
true, given that the relevant truthmakers no longer exist? Can reductionism be
avoided? Is the Compound Presentist obliged to exploit the questionable devices
of the orthodox Dynamic Presentist Lucretian properties, ersatz B-series and so
forth? Not necessarily, for there is an alternative: as we have just seen, Compound
Presentism allows events in the unreal past to coexist with events in the (narrow)
present, and this opens up new possibilities. With reference to our original case,
consider some event E, which occurs in A at t
1
, and let P be some (true) proposition
about E. Since t
2
coexists with t
1
, P is unproblematically true at t
2
. But what of t
3
?
As of t
3
both t
1
and the events that occurred at it are no longer real, so P lacks a
truthmaker. However, t
2
falls within the real past of t
3
, and this makes a difference.
Suppose that we say the following:
What is true (or factual) as of a time t is also true (or factual) as of all other
times that coexist with t.
Call this the factual inheritance principle. If we subscribe to the factual inherit-
ance principle then since P is true as of t
2
, and t
2
coexists with t
3
in a single extended
present, then P must also be true as of t
3
, even though the relevant truthmakers
are no longer real as of this time. The factual inheritance principle allows truths
(or facts) about earlier times to accumulate at later times, and thus be transmitted
over time. Everything that is true (or factual) as of t
3
is also true (or factual) as of
t
4,
and the same applies to all later times. The reality of the past is thus secured.
This argument, if successful, shows that there can be facts about times and
events that are no longer real. It doesnt follow that all these facts are known or
knowable. Indeed, this is one of the main advantages of this form of Presentism:
reductionism and anti- realism about the past are avoidable. The key assumption is
the factual inheritance principle; it is thanks to this principle that we are entitled to
maintain that facts about the immediate past do not cease to obtain when the past
they concern departs from reality. Is this principle true? It is hard to see how to
argue for it, but there is no denying that it is plausible, for it corresponds with com-
mon sense, although, as we shall now see, it is not as innocuous as it may appear.
Compound Presentism seems to offer dynamists everything they could want
but, as with the Growing Block model, on closer examination it turns out to be
a good deal more hostile to common sense than first appearances suggest. I have
taken the direction of time for granted. I assumed that A occurs earlier than B, that
C comes into existence later than A, and so forth. But this assumption is entirely
unjustified. There is nothing in the simple universe described above that supports
the assumption. Suppose we follow Broad and say:
If event X precedes event Y, then there is a time at which the sum total of
existence includes X but not Y, and a time at which the sum total of existence
includes both X and Y.
DYNAMIC TIME 101
Reading Figure 6.5 from left to right and top to bottom, this yields the result that
A occurs earlier than B. But reading from right to left, bottom to top, Broads
definition yields the result that E occurs earlier than D, since there is a time at
which the sum total of existence includes E but not D, and a time at which the sum
total of existence includes both E and D. If there is a temporal direction in this
universe, it is not due to the pattern of annihilation and creation which constitutes
the dynamic of time itself. This pattern is perfectly symmetrical, as is clear from the
even simpler case shown in Figure 6.8. This universe consists of four times, whose
contents are: a triangle; a triangle and a square; a square and a circle; and a circle.
Nothing in this pattern of occurrences justifies thinking that these different states
of affairs unfold in one temporal direction rather than the other.
If Compound Presentism is true, the sum total of reality is different as of differ-
ent times, and the sum totals at adjacent times partially overlap, but this pattern of
overlap is symmetrical in both temporal directions. If you see two adjacent lights
flashing on and off quickly [on, off], [off, on], [on, off], [off, on] , after a short
while you have no sense of the bulb on the right being on before the bulb on the
left; there is just an impression of rapid alternation. The dynamic model of real-
ity offered by the Com pound Presentist is akin to this. There is an alternation of
times, but no passage of time, at least if passage is taken to be a process that unfolds
in a particular direction. Consequently, any apparent direc tion within time must
be a product of the asym metries that exist among the contents of time, contents
that are themselves sustained by a perfectly symmetrical process of creation and
annihilation.
The symmetry inherent in the Compound Presentists model may not seem dis-
turbing in itself, but it has an implication that is. The argument sketched out above
for realism about the past can be run in the other direction, to reach the conclusion
that the factual domain of any time t includes everything that occurs later than t:
the factual inheritance principle transmits factuality in both directions. If state-
ments about the past have determinate truth- values, so, too, do statements about
the future. It is a surprising irony that this most dynamic model of time may have
precisely the same implications for future truth as the Block view.
To this interesting result we can add another, the principal conclusion of this
chapter: there may well be alternatives to the block-universe model of time.
Many- Worlds Presentism is the least dynamic of these alternatives. Instead of a
single connected block-universe we are presented with a fragmented block-universe,
Figure 6.8 Dynamism without direction. In a
Compound Presentist world, the sum total of reality is
different as of different times, but in cases where the
pattern of becoming and annihilation is symmetrical, the
resulting dynamism lacks an inherent direction. Nothing
in the pattern shown here favours the hypothesis that this
(very simple) universe is developing in the up-direction
rather than the down-direction, and vice versa.
102 TIME AND SPACE
consisting of unconnected worldlets. The Many- Worlds Presentist could be con-
strued as offering a dynamic conception of individuals, since as a continuant per-
sists through time it (in effect) jumps from world to world. But since the worlds in
question constitute a timeless ensemble, the jumps in question are of a singularly
undynamic nature. It would be more accurate to say that persisting individuals
are scattered entities, scattered across spatiotemporally distinct worlds. However,
some genuinely dynamic conceptions of time have emerged: the Growing block,
Dynamic Presentism and Compound Presentism, with its shimmering alternations
of times. If these models are metaphysically intelligible, to discover the nature of
time in our universe we will have to look beyond metaphysics. Perhaps there are
empirical considerations that will decide the issue (we will be looking at one such
in Chapter 7, and taking a closer look at the relativity theories in Chapters 1821).
There are two further conclusions to note. First, a dynamic model of time does
not require transitory tensed properties; no such properties are to be found in
the Growing Block model or Compound Presentism (the situation with regard to
Dynamic Presentism is less clear-cut). Secondly, the questions Is time dynamic?
and Does time have a direction or arrow? have to be kept distinct. At first
view, time in a Growing Block universe seems to possess an inherent and directed
dynamic of its own, a dynamic generated by the creation of times and events, but
on closer scrutiny this inherent directedness evaporates. The same applies in the
context of Compound Presentism: the dynamism generated by the creation and
annihilation of times and events does not run in one way rather than another.
Consequently, if time has a direction, in any significant sense of the term, this is
due to the contents within time and their modes of connection, not time itself.
7 Time and consciousness
7.1 The micro- phenomenology of time
In this chapter I redeem a pledge made in 3.2 and take a closer look at our expe-
rience of time. I will not be concerned with what we believe and feel about time
and its passing at different stages of our lives; my focus will be on the character
of our experience from moment to moment within our streams of consciousness.
My approach will be phenomenological: the task is the tightly circumscribed one
of trying to formulate a description of our short- term experience of time that is
clear, accurate and intelligible. Other interesting problems, such as how our brains
manage to perform the impressive feat of integrating the inputs from our different
senses to produce a real- time representation of our usually changing surroundings
(the typical time- lag between initial stimulation and experience is under half a
second) will be ignored.
The micro- phenomenological topic is of interest in its own right arriving at a
clear account of the temporal features of conscious experience is a non- trivial task
but it may also have wider implications, for it may well be that our experience
of time reveals something (if not everything) about the nature of time itself. Of
particular interest here is the B- theory. Many have voiced the suspicion that time
as manifest in consciousness possesses dynamic features that cannot be explained
with the resources available to the Block theorist:
Should we simply shrug the human experience of time aside as a matter
solely for psychologists? Does the time of an altered state of consciousness
have no relevance at all to the time of Newton or Einstein? Does our impres-
sion of the flow of time, or the division of time into past, present and future,
tell us nothing at all about how time is as opposed to how it merely appears
to us muddle- headed humans?
As a physicist, I am well aware how much intuition can lead us astray
Yet as a human being, I find it impossible to relinquish the sensation of a
flowing time and a moving present moment. It is something so basic to my
experience of the world that I am repelled by the claim that it is only an
illusion or misperception. It seems to me that there is an aspect of time of
great significance that we have so far overlooked in our description of the
physical universe.
1
(Davies 1995: 275, original emphasis)
Before we are in a position to judge whether this is so or not we need a clear idea
of what these features are.
104 TIME AND SPACE
Let us start by gathering some basic phenomenological data. Our typical experi-
ence is a combination of persistence and change. You are, let us suppose, lying in
a deckchair gazing at the sky, where you can see a few white clouds moving very
slowly against the unvarying blue backdrop. You see the occasional bird overhead.
Sometimes they circle around for a while, and often they fly right on by. When
you do see a bird in flight you are directly aware of its motion: the flapping of
wings; the sometimes fast, sometimes slow changes of position. You are as fully and
immediately aware of these movements and changes as you are of the unchanging
blue of the sky. The same holds quite generally for other modes of consciousness.
We see change, but we also hear it (a rapid sequence of notes played on a piano),
feel it (slowly move your finger across the palm of your hand) and imagine it
(recreate in your imagination the final shot in a tennis tournament a ball goes
out, a player leaps with joy, the crowd roars). In addition, there can be changes in
what we smell and taste, and in our emotional state (a surge of anger). And then
there is the inner soliloquy, the conversation we have with ourselves in thought,
conducted sometimes in words, sometimes in images, sometimes in neither, which
rarely ceases during our waking hours.
All this is obvious: consciousness is alive with change and variation. More elu-
sive is the subtle but distinctive sort of dynamism that is characteristic of unchang-
ing sensations. Return to the deckchair scenario. For some moments now you have
been staring at an empty region of blue sky and nothing has changed. Your inner
monologue has (if only briefly) ground to a halt, you have seen no movement, your
visual field is filled with an unvarying expanse of blue. But is your consciousness
entirely still or frozen? Have you come to a complete stop? No. Throughout this
period you remain conscious, and conscious of the blue presence continuing on;
you have a (dim, background, passive) awareness of the blue constantly being
renewed from moment to moment. This passive awareness of continuation and
renewal is perhaps more vivid in the case of auditory experience. Imagine hearing
a sustained but unwavering note played on a cello: you hear a continuous and
continuing flow of sound. This feature call it immanent phenomenal flow is
possessed by all forms of experience (think of the burning sensation on the tongue
caused by biting on a chilli pepper), and is a dynamic feature of experience that is
independent of changes of the ordinary qualita tive sort (the chilli- induced burning
is felt as continuing on even when its intensity and qualitative character remains
constant). Phenomenal flow is especially important in the case of easily overlooked
but ever- present bodily experiences: the feeling of ones limbs being disposed
one way rather than another, the feeling of being vertical rather than horizontal
or upside down, the feelings of pressure on ones skin caused by clothes, seats,
shoes and so on. When we are not moving about, our bodily experience is largely
unvarying in quality, and so goes largely unnoticed, but it remains a constant and
constantly flowing presence within our consciousness.
This short survey of the phenomenological data suggests that any adequate
account of the temporal features of experience must accommodate two facts:
our experience of change is as direct and immediate as our experience of
colour or shape;
our experiences possess the feature of phenomenal flow.
TIME AND CONSCIOUSNESS 105
In what follows I will consider a variety of attempts to account for these features
or rather the first, since the second is not always recognized and, by locating
their deficiencies, move towards a clearer understanding of what they involve.
7.2 Memory based accounts
There is no denying the crucial importance of memory to the macro- phenomenology
of time but, as we saw in 3.2, some philosophers (e.g. Mellor) go further and
argue that memory is also entirely responsible for the character of phenomenal
temporality over the short term. As soon as it is seriously considered this claim
starts to look very implausible.
Suppose you hear a fast sequence of notes played on a piano: CDE. You hear
each note as possessing a short but discernible duration, but you also hear the
succession: C- being- followed- by- D, and D- being- followed- by- E. How could the
memory theorist account for this? One option would be to say When you hear
D, you remember hearing C; when you hear E, you remember hearing D, but
this wont do. Hearing D while remembering having heard C is compatible with
C having occurred at any time, perhaps hours or years ago. To solve this problem,
the memory theorist might forward this proposal:
When I hear D, I remember hearing C, while believing that I have just heard
C; when I hear E, I remember hearing C and D, and believe that C occurred
before D, and that D occurred just before E.
But this wont do either. The problem is the invocation of belief. It is true that we
can form accurate beliefs about the order and duration of events without using
a clock. However, what sort of belief is the memory theorist relying on here?
There are beliefs that are merely latent (e.g. your belief that Berlin is the German
capital, which you possess even when asleep), and there are beliefs whose content
is conscious (e.g. you hear someone ask Is Bonn still the German capital? and
you think No, its Berlin!). Since we are trying to explain our consciousness
of change, the relevant beliefs must be conscious. But this is implausible. The
conscious beliefs (or thoughts) just arent to be found; not always, not in every
experience of change. If they were, then given that we are continually experiencing
change in bodily sensations, auditory and visual perceptions, thoughts and mental
images, our consciousness would be flooded, horribly clogged, with thousands of
boring beliefs about what happened before what. But it isnt.
Faced with this objection the memory theorist might try a different tack, and say:
First I hear C. I then hear D, the experience of which is automatically accom-
panied by a short- term memory- image corresponding to my hearing C. I then
hear E, and as I do I have a short- term memory of C- being- followed- by- D.
This is better, since the account is only using experience in trying to account for our
experience of change, rather than thought or belief. But the account is nonetheless
problematic, for two reasons.
First of all, the memory theorist is positing a short- term memory of an expe-
rience of succession: C- being- followed- by- D. The sort of experience that the
106 TIME AND SPACE
memory- account is meant to explain away is in fact being presupposed: we can-
not remember what we have not already experienced. If the memory theorist is
prepared to admit that we are directly aware of succession when we remember
and imagine, why not admit that we are directly aware of succession in ordinary
experience?
The second difficulty runs deeper. We can distinguish two claims, one weak
and one strong. The weak claim is that experiential memory plays a central and
indispensable role in temporal awareness. The strong claim is that temporal aware-
ness is wholly the product of experiential memory. The various proposals that we
have considered thus far are all versions of the weaker claim. Why? Because of the
example we have been working with: the sequence of tones CDE. Each of these
individual tones has (it was stipulated) a short but noticeable duration a duration
that is directly experienced. If phenomenal temporality is wholly the product of
memory, there can be no direct experience (or memory) of duration or change
whatsoever. This means that our experience of even a single brief tone must be
explained in terms of involuntary short- term memories. But memories of what?
The answer must be a succession of strictly durationless experiences. My experi-
ence of the tone C consists of a large (infinite?) number of momentary experiences,
each (except the first) being accompanied by a large number of nested involuntary
short- term memories of other momentary experiences. And what holds of the
single tone C also holds of the experience of C- being- followed- by- D.
This proposal suffers from a very severe plausibility problem. On the one hand
it is hard to believe that we are not immediately aware of some duration in expe-
rience. Is a strictly durationless auditory experience even possible? On the other
hand we are being asked to believe that our experience of duration depends on
vast numbers of nested momentary memory- images (for it should not be forgotten
that the short- term memories must themselves be durationless). This, too, is very
hard to believe. On hearing the succession of tones CDE, are we aware of vast
numbers of constantly changing momentary memories? I think not. The weak
claim has some plausibility; the strong claim has none whatsoever.
7.3 The pulse theory
Since memory based accounts cannot explain our experience of change or phe-
nomenal flow we must look elsewhere. One option runs thus:
Our streams of consciousness consist of short pulses of experience, each of
some finite duration (perhaps half a second). These pulses correspond to
what is often called the specious present specious because the present of
experience is not the durationless present of mathematical physics. During
a single pulse change is directly experienced: we are conscious of all parts
of the pulse or, to put it another way, all parts of the pulse are co- conscious.
There are no direct experiential relations between successive pulses: these
are linked only by memory.
2
So, for example, if I hear CDEF, my experience could consist of two pulses,
P
1
and P
2
thus:
TIME AND CONSCIOUSNESS 107
P
1
P
2
[CD] [EF]
It might seem that if P
2
comes straight after P
1
, and during P
2
I remember hearing
P
1
, then my experience would seem continuous. I would experience C- followed-
by- D- followed- by- E- followed- by- F. But closer consideration suggests otherwise.
If the pulse theory were correct we should be able to discern two distinct forms
of transition within our consciousness: change within pulses and change between
pulses (e.g. CD, DF). After all, while C and D are experienced together they
are fully co- conscious D and E are not. Surely this should be a noticeable differ-
ence? But in fact we notice no such thing; every transition between any two parts
of the scale is experienced, and experienced in just the same way.
In fact, there is no reason to think that it would even seem to us that P
2
is experi-
enced as following on from P
1
. Given that there are no experiential links between
P
1
and P
2
, the fact that P
2
occurs just after P
1
is irrelevant to the experiential char-
acter of the supposed perceived succession; the only link between P
2
and P
1
is the
memory that P
2
contains of P
1
. Hence, even if P
2
happened a hundred years after
P
1
, the pulse theorist is committed to saying that these two pulses would constitute
an experience of succession, which seems plainly absurd.
The same point can be made differently. Suppose that you and I are simultane-
ously hearing the same sequence of notes. Our streams of consciousness have this
form:
You: [CD] [EF]
Me: [CD] [EF]
Since there are no direct links between the first and second pulses, there are two
other streams of consciousness that, according to the pulse theorist, are just as real:
S
1
[CD] [EF]
S
2
[CD] [EF]
If the pulse theory is true, our streams of consciousness are intermingled. Or
rather, there is no purely experiential fact of the matter as to which experiences
belong to which streams. Again, this seems absurd.
3
The pulse theory may be flawed, but, considering that, it has nonetheless served
a useful purpose since a general feature of our experience has come to light. Each
phase of a typical stream of consciousness is phenomenally bonded to its immedi-
ate predecessor and successor: there is a direct experiential link between adjacent
co- streamal phases. An adequate account of temporal experience must accommo-
date phenomenal bonding.
7.4 Awareness and overlap
An alternative approach takes as its starting point an important distinction: it
is one thing to have a succession of experiences, another to have an experience
of succession. If you were to hear C and then D and then E, you would have
108 TIME AND SPACE
undergone a succession of experiences; if you were to hear C- being- followed- by- D
and D- being- followed- by- E you would have had an experience of succession. The
distinction is plainly valid. Moreover, it is clear that we have experience of suc-
cession, rather than mere successions of experiences. What account can we give
of this?
One solution runs thus. When we are conscious of a sound, we are aware of
it: consciousness consists of an awareness of phenomenal contents, items such as
sounds, colours, bodily sensations, thoughts, and so on. Hence there is a distinc-
tion between phenomenal contents and acts of awareness. Clearly, if we are to be
aware of succession, then a single act of awareness must embrace more than one
momentary content: it must embrace a succession of contents. One way for this
to occur is illustrated in Figure 7.1.
Figure 7.1 Multiple contents falling
under single acts of awareness.
A A
C D E F
A
1 2 3
Figure 7.2 The awareness-overlap model of our
experience of time. Phenomenal bonding is secured by
distinct acts of awareness apprehending (numerically) the
same items of content.
We are to suppose that the experience of CDEF involves two (momentary)
acts of awareness; one of these acts takes in CD, the other EF. However, there is
an obvious difficulty here. Since there is no experienced connection between D and
E, the account is simply the pulse theory in another guise, and we have already seen
why this theory is inadequate. But there is an obvious remedy for this difficulty.
What if the scope (and so the contents) of the two acts overlapped in the manner
indicated in Figure 7.2?
A model of this sort was advocated by Broad (1923). I have shown just three
acts of awareness (apprehending four auditory contents), but this was merely to
present a simple picture. In principle there could be many further acts between A
1

and A
2
(and so likewise between any two acts separated by a noticeable temporal
interval) and Broad suggested that this was the case.
By virtue of the fact that successive acts of awareness apprehend numerically
the same contents, this awareness- overlap model is no longer open to the objection
that it is just the pulse theory in another guise. But it may still be vulnerable to
the objection that it provides an unrealistically fragmented account of our streams
of consciousness. While successive acts provide awareness of the same contents,
A A
C D E F
1 2
TIME AND CONSCIOUSNESS 109
these acts are themselves wholly distinct episodes of experiencing. There is unity
(or rather, partial identity) at the level of content, but there is no unity or continu-
ity whatsoever at the level of awareness. However, there is no need to press this
objection, for the model suffers from a more obvious and devastating problem.
Consider note D in Figure 7.2. This content falls within the region of overlap,
and so is apprehended as a whole by both A
1
and A
2
. As a consequence, the sound
will be heard twice over, when in actual fact it is heard just once. This repeated
contents problem is even worse if there are additional acts occurring between A
1

and A
2
, for the same sound will be heard over and over again! Since the same will
apply for any brief contents that are experienced as wholes by different acts of
awareness, the awareness- overlap model is clearly inadequate.
7.5 The two- dimensional model
In later years Broad adopted a very different account of phenomenal temporality,
an account similar to one advocated by Husserl at one stage.
4
Again, the underly-
ing idea is quite simple. The repeated contents problem arises because one and the
same content is apprehended by successive acts of awareness. To solve the problem
we deny that numerically the same content is apprehended by successive acts.
Instead, we say that each momentary act consists of an awareness of a different
momentary content together with representations of the preceding contents. These
representations are not full- blooded re- presentations of the original contents. If
they were, then when hearing tone E together with representa tions of the preced-
ing C and E we would hear three simultaneous tones a chord rather than a
succession. This doesnt occur, the argument runs, because past contents are not
presented as they initially occurred, but under a distinctive tempor al mode of
presentation. Thus Broad introduces the notion of presented ness. When a content
is initially experienced it possesses this quality to the highest degree, as it is re-
presented with subsequent contents it possesses less and less of this quality, until
in the end it drops out of awareness altogether.
To get an idea of how this works in practice it will be useful to look at an
artificially simple example. A subject hears a succession of brief sounds, which
we label CDEFGHIJK. These are apprehended by nine different acts
of awareness. Let us further suppose that there are six degrees of presentedness,
Pr
1
, Pr
2
, , Pr
6
(ordered by decreasing intensity). In Figure 7.3, presentedness is
indicated by font size (the larger the font, the higher the presentedness).
The hearing of C, let us suppose, was preceded by a period of silence. A
1
is a
direct awareness of C, which is experienced as possessing maximum presented-
ness, Pr
1
. A
2
is a direct awareness of D, along with an awareness of C possessing
presentedness of level Pr
2
as a consequence, C is apprehended as being just
past. A
3
presents E with maximum presentedness, D with Pr
2
and C with Pr
3
as
a consequence, D is experienced under the mode just past, and C a little more
past. And so it continues. C features in A
3
A
6
with ever diminishing presented-
ness, and by the time A
7
occurs this content has vanished from the subjects imme-
diate awareness (or specious present), and from this point on is accessible only
through memory. A
6
A
9
are typical specious presents: the subjects span of direct
110 TIME AND SPACE
awareness is filled with contents of varying degrees of presentedness. As can be
seen, each content enjoys its brief moment of maximum presentedness, and then
sinks gradually into the past. This example is unrealistically simple. In reality there
would be further acts occurring, for example, between A
1
and A
2
, and additional
degrees of presentedness (so as to enable contents to slide smoothly into the past).
Broad and Husserl fill out the details in different ways (e.g. Husserl suggests
that each momentary awareness includes retentions of previous contents along
with protentions of anticipated future contents), but the basics remain the same.
The dynamic character of experience is explained by positing what is, in effect,
a two- dimensional time: there is the earlierlater ordering of acts of awareness
(A
1
A
9
), and the earlierlater ordering of contents presented to these acts (e.g.
in A
7
this runs from D (Pr
6
) to I (Pr
1
)). This is by now a familiar manoeuvre
recall the meta- time required to sustain a moving present but since the posited
additional dimension is located within consciousness rather than the world itself,
it is not open to the objection that we have no reason to believe that such a
thing exists. Indeed, the claim will be that we do have reason to believe that this
extra dimension exists, since this is what is required to explain what we find in
our experience. Nor is the model vulnerable to the objection that it generates
an infinite regress. A regress arises if we posit a present that moves along the
time line; the model we are considering posits no such thing. According to the
two- dimensionalist, the specious present consists of a momentary awareness with
a complex content that seems temporally extended. The appearance of future-
directed motion is a consequence of the ways the complex contents of different
acts are interrelated; for example, tone C is represented in the contents of suc-
cessive acts of awareness, but seems increasingly more past in each. This slippage
of contents into the phenomenal past creates the impression that our immediate
experience is advancing into the future. Since nothing moves along the time- line,
and the impression of passage is not a product of contents being created or anni-
hilated, the model is entirely compatible with the B- theorists static or passage-
free conception of time.
As well as providing an explanation of how we can have an immediate experi-
ence of change, the model provides an account of phenomenal flow, the impres-
sion we have that a qualitatively unvarying sensation is continuously refreshed and
renewed. Figure 7.4 illustrates the experiential structures that result from hearing
an extended C- tone: in each of the successive momentary awarenesses A
1
A
6
a
new C- content is present ed, and these are retained, albeit in modified form, in
subsequent awarenesses. The subjects consciousness soon becomes flooded with
C D E F
C D E F G
C D E F G H
C D E F G H I
C D E F G H I J
C D E F G H I J K
A
1
A
2
A
3
A
4
A
5
A
6
A
7
A
8
A
9
Figure 7.3 The two- dimensional
(BroadHusserl) model of
temporal experience. The same
experiential contents (in this case
brief tones) are apprehended as
possessing different degrees of
presented ness in successive acts of
awareness.
TIME AND CONSCIOUSNESS 111
nothing but the sound of this tone, which is experienced as enduring in the present
whilst simultaneously sinking into the past.
The two- dimensional model is ingenious and possesses explanatory potential,
but it is also seriously problematic. The difficulties fall under three headings:
phenomeno logical inaccuracy; problematic presentedness; and consciousness
fragmented.
Phenomenological inaccuracy
If the two- dimensional model were true, we could expect that:
experiences would never end abruptly, they would always linger on for a short
while as they are represented in successive awarenesses as possessing diminish-
ing presentedness;
at each moment we would be aware of our current experience together with
a constantly shifting complex of representations of recent experiences, and so
our consciousness would be choked with the residues of recent experiences.
But this is wrong on both counts. As far as I can tell, my consciousness is not
clogged up with fading residues of prior experiences, and experiences do not
always linger; they sometimes do end abruptly. It could be objected, Ah, but the
residues are very brief. Thats why you dont notice them. Likewise, for the shift-
ing constellations of rep resentations. This reply is unconvincing. If we cannot
discern the posited representa tions, there is no reason to suppose that they have
any phenomenological reality, and hence no reason to suppose that they contribute
to our actual experience of change and persistence.
Problematic presentedness
For the two- dimensional model to be viable we have to be able to make sense of
the property that creates the extra dimension: presentedness. This is (presumably)
an intrinsic property, the intensity of which can vary, and which is responsible for
an experiences seeming to be actual, or just past, or slightly more past, and so on.
As we have already seen in connection with those accounts of world- time (rather
than phenomenal- time) that posit transitory intrinsics, it is far from obvious that
such a property exists, or could exist; if anything the problem is more severe when
C C C C C C
C C C C C C
C C C C C C
C C C C C C
C C C C C C
C C C C C C
A
1
A
2
A
3
A
4
A
5
A
6
A
7
A
8
A
9
A
10
A
11
Figure 7.4 A two-dimensional account of phenomenal flow.
112 TIME AND SPACE
the bearers of the supposed properties are components of experience. What qual-
ity, added to a pain sensation, would make that sensation seem to be in the past
rather than in the present? If I am now experiencing some sensation, wont the
experience seem to be occurring now irrespective of any peculiar qualities it might
possess? Someone might say, Presentedness is not an additional property that
experiences have. It is akin to Humes force and vivacity. It is simply a measure
of the intensity of an experience; as experiences lose presentedness they become
progressively fainter, until they fade away altogether. But this wont work. Do
some parts of the image in Figure 7.5 appear more past than others?
Consciousness fragmented
Perhaps the most serious objection to the two- dimensional model is the simplest:
the account is profoundly atomistic. Each momentary awareness constitutes a
distinct episode of experiencing in its own right, and so, from a purely experien-
tial view, each awareness is entirely isolated from its immediate neighbours. The
awareness- overlap account atomizes awareness, but does not reduce our streams
of consciousness to entirely distinct moments of experiencing, since successive
acts are held to apprehend numerically the same contents. But according to the
two- dimensionalist, successive acts apprehend numerically distinct contents. The
model thus suffers from the same problem as the pulse account: it fails utterly
to accommodate the reality of phenomenal binding. Since it is obvious that our
experience is not confined to isolated momentary capsules, we must look else-
where for a realistic account of time- consciousness.
7.6 The overlap theory
Recall the repeated contents objection to the awareness- overlap model: since dif-
ferent acts of awareness apprehend the same contents, we would be aware of
every content many times over but we are not. The two- dimensional theorists
try to avoid this problem by keeping the momentary act temporally extended
content schema while multiplying contents: successive momentary acts are not
aware of exactly the same contents, but only similar contents (and these contents
have an apparent, rather than real, temporal extension). We have seen that this
approach is flawed. But as Foster has noted on several occasions (1982, 1985,
1991), there is a far simpler solution to the repeated contents problem.
Figure 7.5 A problem for the notion of presentedness.
TIME AND CONSCIOUSNESS 113
Suppose that I hear the succession CDE. Further suppose that I am directly
aware of CD and also DE, but I am not directly aware of the entire sequence of
notes (although we are directly aware of change, the span of our direct awareness
is brief). There are two ways to account for this:
1. I have two completely separate acts of awareness: A
1
with content CD and
A
2
with content DE.
But this entails that I hear D twice, which I dont. To avoid this problem we can
say the following:
2. I have an awareness A
1
with content CD and an awareness A
2
with content
DE, but these acts of awareness are not distinct: A
1
and A
2
partially overlap,
and the part of A
1
that apprehends D is numerically identical to the part of A
2

that apprehends D.
The problem of repeating contents is now solved. Since A
1
and A
2
overlap, there
is only one experience of D. We have two temporally extended experiences that
share a common part CD, DE but that remain distinct because they do not
completely overlap.
D C E
A A
1 2
Figure 7.6 The key to phenomenal continuity:
temporally extended, partially overlapping experiences.
Of course, if A
1
and A
2
are to partially overlap, these acts of awareness cannot be
momentary, they must have some temporal extension. We must, therefore, aban-
don the doctrine that an awareness of change (or succession) consists of a temporal
spread of content being presented to a single momentary awareness. Instead, we
say that an awareness of change or succession is itself temporally extended. Since
any two acts will overlap to precisely the same degree as their common contents,
it is natural to suppose that acts have precisely the same duration as their contents,
in the manner indicated above.
The idea that the succession CDE involves only two episodes of awareness
is an unrealistic simplification. It is more plausible to suppose that there will be a
sequence of experiencings whose contents differ in just noticeable ways, and so
that overlap almost completely; see Figure 7.7, which represents the experience
of watching a light rapidly growing dimmer. In this case, very short contents (such
as the phase indicated by an arrow) will figure in a larger number of different
experiences but again, since the parts of these experiences that apprehend this
content are numerically identical, the content is only experienced once.
Thus far I have continued to talk as though there is a distinction between phe-
nomenal contents and the acts of awareness that apprehend them. This distinction
is unavoidable if (in the manner of the awareness- overlap account) we hold that
at any given moment we are aware of a temporal spread of content. But since we
114 TIME AND SPACE
are now supposing that contents and awareness run concurrently, the distinction
is doing no work, and unless there is some other reason for maintaining it (and
I cannot see that there is), we are free to dispense with its services. In place of a
two- level view of experience we can adopt a one- level or simple view, according
to which phenomenal contents (such as sounds, colours, pains) are intrinsically
conscious items: they do not need to be apprehended by a separate awareness
to be experienced, the contents themselves are experiences in their own right.
Instead of supposing that different contents (such as a sound and a flash of light)
are experienced together by virtue of falling under a single awareness, we say that
these contents are co- conscious; that is, they are joined by the basic relationship
of being experienced together. This simple overlap theory which makes no
distinction between awareness and contents entails that co- consciousness not
only connects simultaneous experiences, it also ranges a short way over time.
An interesting result now emerges. Consider again the brief stream of con-
sciousness consisting of CDE. Let us suppose once more that the subject expe-
riences C- being- followed- by- D, and D- being- followed- by- E. In saying this I am
assuming that the subject is directly aware of successions involving only two tones:
by the time E occurs, C has dropped out of the subjects specious present. We
thus see that C is co- conscious with D, and D is co- conscious with E, but C and E
are not co- conscious. Since the specious present is of limited duration (probably
less than a second), the relationship of co- consciousness over time (diachronic
co- consciousness) cannot be transitive. If it were transitive, then every part of a
continuous stream of consciousness would be co- conscious with every other part.
Since this is clearly not the case, transitivity is restricted to the brief confines of
the specious present. This may seem a surprising result, but on reflection it seems
both natural and inevitable.
The overlap theory accommodates our direct experience of change, and respects
the phenomenal bonding constraint: experiences occurring at different times (such
as C and D) are linked by precisely the same relationship as experiences occur-
ring at the same time: direct co- consciousness. But there remains a problem. Co-
consciousness may not be transitive, but it is symmetrical: if C is co- conscious with
D, then (necessarily) D is co- conscious with C. If it is co- consciousness that binds
together the successive phases of our streams of consciousness, and this relation-
ship is symmetrical, why does our consciousness flow in a particular direction?
Figure 7.7 Observing a light fade. Many
overlapping experiences are involved in even a brief
phase of a typical stream of consciousness. 1 second
TIME AND CONSCIOUSNESS 115
Why do we hear C- being- followed- by- D rather than D- being- followed- by- C?
Why does a persisting tone seem to continue on in a particular direction?
There seems to be only one solution: the contents that are symmetrically joined
by co- consciousness must themselves possess an inherent directional dynamism.
The C- tone is not a static auditory quality, but a flowing quality, likewise for D
and E. This immanent flow is an essential ingredient of any auditory content, just
as essential as timbre, pitch or volume. It is because these contents possess this
inherent and directed dynamic character that we hear C as enduring for a brief
while and then seamlessly running on into D, which in turn runs on into E. The
same applies to bodily sensations, such as pains or tickles, as well as to olfactory
and gustatory contents. It applies to the visual case also. Suppose that I see a bird
fly overhead. It crosses my field of vision in a second or so. I do not merely see
the bird occupying a succession of different locations at different times. I see the
bird moving. This perceived movement is itself a dynamic and directed intrinsic
feature of my visual experience, and just as essential to it as colour, size or shape.
At the outset of this chapter I suggested that an adequate account of our short-
term experience of time had to be able to accommodate both our direct awareness
of change, and our direct awareness of endurance, or phenomenal passage, a
feature of experience that is independent of qualitative variation. If phenomenal
contents quite generally possess the characteristic of immanent flow, we have a
complete account of phenomenal temporality:
We are directly aware of change because co- consciousness extends over time.
The specious present is of short duration because co- consciousness extends
only a short way over time and is not transitive (at least in the diachronic case).
Consciousness flows in a particular direction because phenomenal contents
possess inherently dynamic and directed contents.
It might be objected that to posit immanent flow as a characteristic of experience
over and above their other phenomenal features reeks of expediency: isnt this
precisely what is required to save the overlap model from phenomenological inad-
equacy? It is, but since the feature in question unquestionably exists, not to rec-
ognize it would guarantee phenomeno logical inadequacy. Moreover, it is wrong
to think that imma nent flow is a separate and additional ingredient that different
kinds of experi ence possess. I use the term merely to indicate that different kinds
of experience all have something in common: they all have a directed and dynamic
temporal character. As soon as ones attention is drawn to this fact, it is obvious
that this feature is built into experience in all its forms.
5
Putting all this together, we now have sufficient resources to state in a general
way the conditions under which experiences are co- streamal, that is, belong to
the same stream of consciousness: experiences that are directly co- conscious are
co- streamal; so, too, are experiences that are indirectly co- conscious, that is, form
part of an overlapping chain of directly co- conscious experiences. Since experi-
ences possess the temporally asymmetrical feature of immanent flow, we can take a
further step and provide co- streamal experiences with a temporal ordering. In the
case of an experienced succession such as CDE, not only does each individual
tone possess immanent flow, but these flows combine to constitute a continuous
flow that runs the length of the succession: C is experienced as flowing into D,
116 TIME AND SPACE
and D as flowing into E. Generalizing, let us say that for any two total states of
consciousness E
x
and E
y
that are linked by direct diachronic co- consciousness, E
x
is
phenomenally prior to E
y
if and only if some part of E
x
flows into some part of E
y

but not vice versa. If E
x
and E
y
are not directly co- conscious, but are co- streamal,
then E
x
occurs earlier than E
y
if and only if E
x
and E
y
occur at opposite ends of a
chain of total states of consciousness,
E
x
E
n
E
n+1
E
n+2
E
y
the adjacent members of which are related by phenomenal priority. We can further
say that experiences belonging to total states of consciousness related by phenom-
enal priority belong to the same phenomenal time.
7.7 The phenomenal arrow
The arrow of phenomenal time has considerable autonomy in relation to the others
that we have encountered. If you were to fly a spaceship into the time- reversed half
of a Gold universe and visit an inhabited Earth- like planet, you would perceive
all manner of bizarre goings on, but you would continue to feel your experience
flowing on in the usual manner, forwards, even though the world around you
seems to be running in reverse. Even if Gold universes are not physically possible,
they are certainly clearly and easily imaginable, and this fact alone establishes the
autonomy of phenomenal flow with respect to the ordinary patterning of events.
6
For an equally bewildering scenario we do not need to posit such widespread
reversals of the familiar content- asymmetries. It suffices to consider a case where
just one is reversed: memory. You wake up one day and are puzzled to discover that
where as you can venture only the haziest speculations as to what you were doing
yesterday, or at any points in your past, you find yourself quite certain about exactly
what you will be doing for the rest of the day, and for much of the remainder of
your life. You still have what seem to be memories about your own experiences
but, as you soon realize, the events you now remember have yet to occur. As
the days and weeks pass, you learn that these future- oriented memories are just
as reliable as their past- oriented counterparts: what you remember happening
does happen. Despite this difference, the character of your experience remains
essentially the same: the direction of immanent flow remains future- directed; you
find your phenomenal present sliding forwards towards future events that you
know will occur.
7
Since phenomenal flow is an intrinsic feature of experience,
its direction is independent of that of memory. It also seems independent of the
knowledge- asymmetry, since it is hard to see how possessing detailed and accurate
records (as well as personal memories) of future events could impinge on the
direction of phenomenal flow.
Newton- Smith (1980: 207) considers a variant of this scenario: you wake up
and find that you can remember past and future experiences equally well. A
person in this predicament, Newton- Smith suggests, would find time to be entirely
lacking in direc tion. He is mistaken. Provided this persons experiences continue
to possess immanent flow, time would still seem to have a direction; the same
direction it always had.
TIME AND CONSCIOUSNESS 117
A world such as our own, a world inhabited by conscious beings, is a world
some of whose contents experiences possess an inherent temporal direction,
a feature that may well be unique to experience. There is nothing to prevent our
defining the earlierlater relation for the world as a whole in terms of phenomenal
priority, in the manner sketched above. Indeed, given the overriding importance
of immanent flow in our experi ence of temporal passage, this is in many ways
a natural move to make. Of course, if we do make this move, we should not be
overconfident about the universality of its appeal; if we live in a Gold universe
there may well be others who have a different sense of where time is going.
It has often been said that in thinking about time we must guard against our
tendency to think in spatial terms (a tendency that may well be rooted in the
dominant role that vision plays in our cognition), and no doubt there is some truth
in this, but I suspect that there is a further and very different reason why thinking
clearly about time can be difficult. Immanent flow is such a pervasive feature of
our consciousness that it is hard to think of anything that does not possess this
feature, time included, for not only do our thoughts possess it as we think them,
but so do any mental images that we call up. To take just one example, in trying to
think about Compound Presentism we naturally think in terms of a series, [P
1
P
2
],
[P
2
P
3
], ,which in the medium of our conscious thought becomes P
1
and P
2

exist, then P
2
and P
3
exist. And already the damage is done: an unthinking pro-
jection of the immanent flow of consciousness on to the world has occurred, for
this alternation in the contents of reality is being apprehended as possessing an
inherent directionality of its own. Even if we realize that it is possible to view this
sequence from another perspective we are liable to think to ourselves Ah, so it
could go [P
3
P
2
] then [P
2
P
1
] and again the immanent directed ness of conscious-
ness is transferred where it doesnt belong, and we will find ourselves thinking
Well, it must surely go one way or the other! The thought that it doesnt go one
way or the other, that all that is really going on is that the sum total of reality is
different as of different times, and that these times do not run in any particular
direction, does not come naturally at all.
7.8 Further consequences
If the overlap account of phenomenal time is correct, at least in the essentials, we
may not be able to draw any conclusions about time in general, but given certain
minimal assumptions there are conclusions that we can draw about the time in
which we live.
Two versions of Presentism are clearly incompatible with the fundamental
structures of our streams of consciousness uncovered in 7.6. Orthodox Presen-
tists invariably assume that the present is of short duration; either strictly momen-
tary, or so brief as to be incapable of including change. Let us assume that this is
the case. According to the Solipsistic Presentist, the sum total of reality consists
of just one present. If this were so, then we would be unable to directly experi-
ence change and persistence for if the overlap theory is true, such experience
unfolds within specious presents that are themselves temporally extended. Since
we do directly experience persistence, we can conclude that we do not inhabit a
118 TIME AND SPACE
single momentary present. The same line of reasoning undermines Many-Worlds
Presentism. According to the latter doctrine, reality consists of a vast number of
momentary presents, all of which are entirely isolated from each other. The fact
that we directly experience successions, such as CD, shows that we do not inhabit
a reality of this kind. Suppose that C occurs in a present P
1
and D in a present P
2
. If
C is co- conscious with D it is clearly wrong to say that P
1
and P
2
are related at most
by similarity. Diachronic co- conscious ness is a perfectly real mode of connection;
it seems quite absurd to suppose that experiences related in this way could belong
to entirely separate and unrelated worlds. More relevantly, it seems impossible
to deny that experiences related by diachronic co- consciousness are temporally
related, and so exist within a single temporal system. This will be so irrespective
of the precise relationship between phenomenal time and ordinary (world) time.
The situation with regard to Dynamic Presentism is not quite so clear-cut, at
least on the surface. Dynamic Presentists hold that reality consists of a succession
of momentary phases, but they also hold that there are true propositions about
past (and perhaps future) events. Is this enough to render this form of Presentism
compatible with the overlap theory? It might seem so. Let us assume that I have
been listening to a sequence of tones CDE, and am currently experiencing the
E-tone. The C- and D-tones no longer exist, and nor do my experiences of them,
but there are true propositions to the effect that I heard C flowing into D, and
D flowing into E. It looks very much as though the Dynamic Presentist is able
to tell a convincing story regarding the character and contents of my stream of
consciousness over the past few minutes a story that is fully compatible with the
overlap theory.
However, in assessing the plausibility of what is being offered here, we must
not lose sight of the model of time with which we are dealing. The overlap theory
tells us that for me to experience C flowing into D, or D flowing into E, my
experiencing of C must be co-conscious with my experiencing of D, and my expe-
riencing of D must similarly be co-conscious with my experiencing of E. When
two experiences are (diachronically) co-conscious, they are experienced together,
in a unified temporally extended episode of experiencing. It is difficult to see how
C could be related to D in this way if C is unreal when D occurs and similarly
for D and E. In order for C and D to be part of one unified experience, C and D
must surely coexist. But according to the Dynamic Presentist, since C and D occur
in succession, they do not coexist.
In reply the Presentist could insist that none of this matters: C and D may
not coexist, but the proposition C is experienced as diachronically co-conscious
with D is nonetheless true, and isnt this all we need? It would be all we needed
if the truthmaker for this proposition were two successive experiences that are
experienced together as a succession, and which thus coexist (albeit in succession).
But we saw in 6.10 what the (non-reductionist) Dynamic Presentist can offer by
way of truthmakers for past-tensed propositions: presently existing Lucretian
properties, or the maximal collections of consistent propositions that compose
an ersatz B-series. Since neither Lucretian properties nor the abstract entities
that feature in ersatz B-series are themselves experiences, they cannot form parts
of the temporally extended episodes of consciousness (or specious presents) within
which our immediate experiences of change and persistence occur. So, given the
TIME AND CONSCIOUSNESS 119
way the Dynamic Presentist ultimately construes propositions such as C is experi-
enced as diachronically co-conscious with D, they are not offering a credible
account of the diachronic structure of our streams of consciousness.
Of course, all this is assuming that the overlap theory of phenomenal temporal-
ity is essentially correct. The rival two- dimensionalist theory is perfectly compat-
ible with both Dynamic Presentism and Many-Worlds Presentism, since it denies
that there are any genuine phenomenal connections between successive phases
of a stream of consciousness. But as we have seen, there are good reasons for
rejecting this account of our immediate experience of temporality. I am also assum-
ing that there are temporal relations (of the ordinary, non- phenomenal variety)
between experiences and other events. Consequently, if experiences E
1
and E
2
are
co- conscious, and hence temporally related, so, too, are all the events (experiential
and non- experiential) that are simul taneous with E
1
and E
2
. But this is hardly a
controversial assumption (ignoring relativistic complications).
8
What of the other models of time that we have considered and found metaphys-
ically intelligible? For better or worse, the overlap theory seems entirely neutral
with respect to these.
Consider the Block view. The overlap theory tells us that co- consciousness
extends over short periods to create temporally extended phenomenal presents,
that successive phenomenal presents overlap, and that the contents of these
phenomenal presents have an inherent dynamic character. That our experience
is structured in this way guarantees that we will have the impression that our
consciousness is confined to a forward- moving present even if, in reality, past
and future experiences are just as real as present ones. From the four- dimensional
perspective, a stream of consciousness is akin to a glowing filament embedded in
a long glass block; the filament is aglow throughout its entire length, but this is
not discernible to the subject of these experi ence, who is only ever aware of those
tiny stretches of experience joined by co- consciousness.
As for Dynamic Block models, the overlap theory is quite compatible with
streams of consciousness growing along with the rest of reality in an incremen-
tal fashion. It also seems quite compatible with streams of consciousness gradually
shrinking. This may seem surprising: wouldnt a person notice if their experiences
started being deleted? There is no reason to think that they would. From the stand-
point of any given present we have no direct awareness of any future experiences
that we might have, and so could hardly be affected by their disappearance; like-
wise, the annihilation of an experience isnt itself something that is experienced,
it is simply the ceasing- to- be of something that was an experience. This said, if it
could be shown that immanent phenom enal flow depended on new experiences
being created and added on to a pre- existing stream of consciousness, then we
could rule out both the Block model, and the version of the Dynamic Block model
that fails to recognize a distinction between growth (via becoming) and shrinkage
(via annihilation). The Growing Block model in its original form would be vindi-
cated on empirical grounds. But if the current analysis is correct, and phenomenal
flow is an inherent feature of experiences themselves and entirely independent of
experiences coming- to- be or ceasing- to- be, this option is not available. Whether
this analysis of phenomenal flow is correct is another issue, and one to which
those attracted to the Growing Block might want to devote attention. It is certainly
120 TIME AND SPACE
tempting to think that the inherent directedness we find within our streams of
consciousness is linked to new phases of experience coming- into- being. But since
it also seems that this same directedness is an intrinsic feature of our experience,
it is hard to see how this could in fact be the case. For more on this see Dainton
(forthcoming).
Dynamic Presentism of the compound variety is not only compatible with the
overlap model, but in some ways it is a precise analogue of it. Whereas the overlap
theorist holds that a stream of consciousness consists of extended phenomenal
presents that partially overlap, the Compound Presentist holds that the world
consists of extended presents that partially overlap. And whereas to make sense
of this the overlap theorist is obliged to deny the transitivity of co- consciousness,
the Compound Presentist is obliged to deny the transitivity of co- existence. But
the similarities end here. The time of the Compound Presentist consists of an
alternation of moments and events that lacks an inherent direction: we could
as easily regard the sequence [P
1
P
2
], [P
2
P
3
] as running forwards or backwards.
By virtue of the dynamic patterning imman ent in experience, a typical stream of
consciousness does have a definite direction.
So by appealing to empirical (rather than metaphysical) considerations we are
in a position to rule out two of our five contenders. It is often assumed that the
only sort of empirical evidence that can further our understanding of time is of the
scientific variety. As should by now be plain, this is a mistake. Scientific theories
attempt to explain and predict the observed behaviour of material things, often
by reference to invisible forces and entities; there is no denying that such theories
can have important implications for how we think of matter, space and time. But
we can also learn something about time by examining experience itself, rather than
what experience reveals about the non- experiential parts of our world.
You may be inclined to think Fine, but what has this foray into the phenomeno-
logical really delivered? The models of time that we have eliminated were never
serious contenders anyway, since of the five they are by far the least plausible. If
so, there are two points to bear in mind. There is no denying that the Solipsistic
and Many-Worlds versions of Presentism are highly implausible, but might this not
be because they both conflict so dramatically with what we find in our immediate
experience? In which case, by working our way towards a clearer picture of how
experiences are interrelated in streams of consciousness, we have put the intui-
tive case against these doctrines on a firmer foundation. (Dynamic Presentism is
rather different; since it tries to accommodate succession within a single world, it
is not so obviously implausible.) Secondly, as we saw in 6.8, the time- problem in
quantum gravity theory has led some scientists to take Many- Worlds Presentism
very seriously indeed; it could even be that the only empirical data that can refute
this doctrine is of the phenomenological variety.
As for the question of what science can tell us about (our) time, we shall be
considering this again in chapters to come.
8 Time travel
8.1 Questions of possibility and paradox
Space offers something that time does not: unconstrained freedom of movement.
Up, down, north, south, east, west, and all points in between, not only are we able
to move in all spatial directions, but we are free to move back and forth as we
wish, and at different speeds. Time is miserly in comparison. As it is by nature one
dimensional, it would be churlish to condemn it for offering us only two direc-
tions, but the additional constraints it places on us can seem unnecessarily severe.
Although we might be said to travel into the future simply by persisting, we are
condemned to the monotonous rate of a second per second, with no option of
skipping over times that we might prefer to avoid; and as for the past, it is utterly
out of bounds. There is a sense in which we are captives of time, but not of space;
there is no temporal counterpart of the car.
This most basic of differences between the dimensions naturally gives rise to
some questions. Is travel through time impossible because of the nature of time?
If so, why? One answer is implicit in Bigelows observation (see 6.7) concern-
ing the rarity of time travel stories prior to the late nineteenth century. Since it
is hard to see how one could travel to the past or future unless these times exist,
Presentists will be inclined to regard time travel as impossible (just as Growing
Block theorists will dismiss leaps into a non- existent future), but it would be mov-
ing far too quickly to assume that our inability to move as we wish through time
entails the falsity of the Block model. It is certainly true that anyone who believes
that times have the same ontological status as places has some explaining to do;
if the difference between now and then is akin to that of here and there,
there is clearly a prima facie problem since we can move more or less as we like
between here and there, but not between now and then. But in fact there are
several responses to the problem available to the Block theorist.
Time travel involves insuperable paradoxes of a broadly logical sort, and so is
metaphysically impossible.
Time travel is not metaphysically impossible, but is forbidden in our world by
contingent physical laws.
Time travel is neither metaphysically impossible nor physically (or nomologi-
cally) impossible: it either does or could occur in our universe.
We will spend most of this chapter exploring these claims. Until further notice I
will be assuming the four-dimensional Block model; we will be in a better position
122 TIME AND SPACE
to assess where dynamic models stand vis- - vis the possibility or otherwise of time
travel when we have a clearer idea of just what the obstacles are.
The time machines that we are familiar with from science fiction fall into sev-
eral categories. Some give access to the past but not the future, others do the
opposite and others do both. Some can travel any temporal distance and others
are limited; for example, the wormhole time machines being seriously explored
by contemporary physicists permit backward travel, but only as far as the earlier
mouth of the hole.
1
Some machines cater for round- trips. Others cater only for
the outward leg, but travelling to a distant time by this route is a risky business,
for unless you have a second machine available for the return journey you will be
stranded. Some machines work by time- slides and others by time- jumps. If you
embark on a time- slide between 1900 and 2000 you will pass through all interven-
ing times, but at a faster- than- normal rate: your trip only takes (say) a week instead
of a century. A time- jump, on the other hand, would take you directly from 1900
to 2000, and the twentieth century would pass in your absence.
Interestingly, the possibility of one form of time travel is not seriously dis-
puted, and that is time- slides into the future. Such trips seem neither physically
nor logically problematic. On the logical side, the problem of self- defeating loops
(What if you killed your grandfather before he sired your mother?) that plague
backward travel do not arise, since your presence in the future cannot, in itself,
causally undermine what has already happened. On the physical side, there are
several ways to achieve the desired effect that seem compatible with known natural
law. Putting someone in a deep- freeze that effectively halts all biological processes
and then successfully thawing them out at a later date is perhaps the most obvious
way to slide someone forwards into the future. When the subject awakes no time
has passed for them but fifty or a thousand years might have passed for the rest
of the world. Relativity theory provides several methods for futureward slides.
Anyone who embarks on a lengthy spaceship voyage of (say) ten years duration
at a sizeable fraction of the speed of light would find on their return that hundreds
or thousands of years had passed on Earth. Fast travel is not the only way to slow
down time. Spending a while orbiting a very strong gravitational field (such as
produced in the vicinity of a black hole) has equally dramatic consequences. (We
will be exploring the theoretical background of such effects in Chapters 1821.)
The fact that slides into the future are comparatively easy increases the pressure
on the four- dimensionalist. Why, if there is no ontological difference between past,
present and future, are slides into the past not equally possible? And if they are
possible, why are they comparatively more difficult? To make any progress with
these questions we will have to untangle a number of physical and metaphysical
issues.
8.2 Misconceptions and multidimensions
One objection to time travel, in either direction, can be speedily dealt with. I step
into the time machine, and two minutes later I am three thousand years in the
past. Isnt this a direct contradiction? Before I step into the machine it is true to
say that two minutes from now (in the future) I shall be thousands of years in the
TIME TRAVEL 123
past. How can one and the same event my stepping out of the machine occur
later than now and earlier than now?
Rather than revealing time travel to be impossible, this objection merely reveals
what time travel, should it occur, necessarily involves. In any time travel story we
have to distinguish two frames of temporal reference. One frame is based on the
time travellers themselves; time as measured by the watches they wear, or by their
mental and biological processes. This is commonly called personal or proper
time. There is one such time for each individual time traveller. The other frame is
that of Earth (or our starting point, or the universe as a whole); call this normal
or external time. In ordinary life our personal time coincides with external time,
but if we embark on time travel the two systems of time reckoning will diverge.
Unless we have independent reasons for supposing that this divergence cannot
occur, we cannot use it to argue that time travel cannot occur (see Dainton (1992)
for more on the ways personal and external times can diverge). In fact, we have
good reason to believe that it can occur. Relativity theory predicts that less time
will pass for the returning space traveller than will pass on Earth (a prediction that
has been empirically confirmed); more generally, objects moving relative to one
another all have their own (diverging) proper times. We can apply precisely the
same distinction to the time travel case: when I step out of my time machine three
thousand years in the past, I am stepping out three thousand years in the past as
measured by external time, and a couple of minutes in the future as measured by
my personal time. Admittedly, backward time travel generates deviations between
personal and external time that are more radical than those permitted by special
relativity we shall see in Chapters 18 and 19 that special relativity limits the
frame- dependence of temporal ordering (to space- like separated events). But
the underlying principle is the same: those in different states of motion have their
own temporal frames of reference.
Another common misconception, a more serious one, is embodied in arguments
such as the following:
But couldnt we travel into the past and witness for ourselves the Egyptians
building pyramids? There is a problem, however. Those centuries have
passed; the events have happened; theyre over. And they all happened with-
out you you werent even born until the twentieth century. The ancient
civilizations came and went, and you werent anywhere in that picture. How
then can you say that you could now go back there and be among the people
then living? It all happened without you, and now you want to say that you
could go back and participate in it, that is, that it did not happen without
you. It did happen without you and it didnt surely, thats a contradiction.
The past is what has happened; not even God could make what has
happened not have happened. Lets say the pyramids were built by 100,000
laborers. Now if you go into the past and join these laborers, there would
be 100,001 laborers. If it was 100,000, it wasnt 100,001; theres a contra-
diction in saying it was both. (Hospers 1997: 121)
There are two key assumptions made here: the past cannot be changed; and
backward time travel involves changing the past. Are these assumptions true? It
depends on the nature of time. There are possible worlds in which the past cannot
124 TIME AND SPACE
be changed, but in such worlds time travel does not involve changing the past;
there are also worlds where the past can be changed, and time travel is a way of
bringing this about.
The idea that the past could be changed is highly counterintuitive. Not only
does it entail two conflicting but true accounts of what occurred at a given time,
but it is quite likely that changes in the past produced by a time travellers arrival
will have wider ramifications, and require the wholesale replacement of the subse-
quent timeline. Consider a scenario familiar from the Terminator films. A nuclear
war in the near future is provoked by out- of- control Strategic Defense Computers;
humanity is reduced to a few ragged bands of survivors; the struggle commences
and human rebels discover the secret of time travel and travel back to prevent the
construction of the computers. They succeed, and history is immediately changed:
the nuclear war does not take place; the billions that perished dont perish, but
happily live out their lives; one entire stretch of human history simply vanishes and
another stretch is immediately created to take its place. Is this scenario absurd? I
am inclined to think that it is.
But dont forget, we are currently assuming the Block view to be true, and in
this context it is simply a mistake to suppose that backward time travel would bring
about changes in the past of the problematic sort just envisaged. If the universe
consists of a single block of times and events, what occurs at a given time is fixed
and unchangeable. This does not mean that you cannot go back to 3000BC and
assist with the building of the pyramids; what it means is that if you do travel
back and assist, that you do so is as true now as it was then. It is true at all times,
including those that occur before you set off. You may have been born in 1975,
but unbeknownst to you and your parents this was not your first appearance in
the worlds affairs: you first entered history as a thirty year old, assisting with the
building of the pyramids several thousand years previously. In short, everything
that you will ever do as a time traveller is built into the past before your first jour-
ney. Prior to your departure you do not remember being in Egypt all those years
ago, but this is because the time you spend there lies in your (personal) future. Of
course, if you left any traces of your visit perhaps you carved your initials on a
sarcophagus these may well be discovered prior to your departure. It is one thing
to affect the past to contribute to what occurred at the times in question quite
another to change it. You certainly affected the past your initials are a testament
to that, and there are Egyptian slaves who were glad of your help but you did
not change it. The building of the pyramids only occurred once, and you were
there at the time.
So the argument fails. But the mistake on which it is founded is easily made. In
imagining how things were in ancient Egypt we envisage ourselves being present
there; since we naturally assume that times later than the present are unreal, it
seems absurd to think someone from the twentieth century could put in an appear-
ance; the future from which they would supposedly emerge is nonexistent. Conse-
quently, if someone from the twentieth century were to travel back, it seems that
they will be emerging into a history that originally unfolded in their absence. All
this shows, however, is that backward time travel is problematic in the context
of certain dynamic models of time. Since we are currently working within the
confines of the Block view, this is an irrelevant result.
TIME TRAVEL 125
In fact, as I suggested above, there are models of time in which the past can be
changed, and some of these models are dynamic. Presentists (and others, for that
matter) who take an anti- realist view of the past reject the idea of a static unalter-
able history. If facts about what occurred at earlier times are constituted by the
evidence (traces, records, etc.) that exists in the present, then if the evidence that
exists at t
2
is different from that which exists at t
1
, the latter time will have a dif-
ferent past from the former. However, since time travel and Presentism are uneasy
bedfellows, this loophole offers little assistance to the would- be time traveller, and
none whatsoever to those who wish to visit the past in order to change it in some
particular (known) respect. More congenial to those in the latter category is the
two- dimensional model depicted in Figure 8.1.
Figure 8.1 Meilands two-dimensional model
of time travel. Each moment has its own distinct
past, indicated by the horizontal lines. If you were
to travel back in time, you would arrive in the past
of the time from which you departed, rather than
a time which preceded your departure on the
primary timeline (indicated by the diagonal).
C
D
E
t
t
t
t
t
t
t
t
Pt Pt Pt Pt Pt Pt Pt Pt
1
1
2
2
3
3
4
4
5
5
6
6
7
7
8
8
In Figure 8.1, t
1
t
8
are present moments, each a year apart. The line P
1
t
1
repre-
sents the past when t
1
is present, and similarly P
2
t
2
represents the past with respect
to t
2
. The point where the line Pt
1
t
1
intersects with P
2
t
2
represents the location
of t
1
in the past of t
2
. As can be seen, each present has a past of its own, and these
pasts are potentially different. Suppose that at t
4
a newly born child, Sam, is sent
back three years. Where does Sam arrive? At t
1
? Yes and no. Sam does not arrive at
t
1
(i.e. at the terminus of the line P
1
t
1
) there were no time travellers present at
t
1
; Sam arrives three years earlier in his own past, at point C. He remains there for
two years: a year later he is at D, three years earlier than t
5
, and yet another year
later he is at E, three years earlier than t
6
. Sam is not present in the past of t
2
, nor
that of t
3
; he is, however, present at the corresponding times in the pasts of t
5
and t
6
.
This two- dimensional model of time is due to Meiland (1974), who explores
its ramifications in some detail but I leave these to the interested reader to
discover for themselves (also see van Inwagen 2010). It can be interpreted in two
ways: as either a static (or eternal) two- dimensional configur ation, or a dynamic
one. In the latter case, as of t
3
no later times are real, and the sum total of reality
includes all those presents that occurred prior to t
3
and their associated pasts,
but no subsequent presents (or pasts); reality grows not just by an accretion of
presents, but by presents and pasts entire timelines come into being with each
new moment.
P
8
P
7
P
6
P
5
P
4
P
3
P
2
P
1
126 TIME AND SPACE
Our world could be like this, but in the absence of any positive reason for think-
ing it is, the hypothesis of a one- dimensional time is rightly favoured on grounds of
ontological economy. The model is of interest, however, for it provides an example
of how backward time travel could involve changing the past without generating
paradoxical consequences. In a world where each present has its own past, there
is no necessity for the past of one present to be the same as the past of another.
Moreover, in such model it is not easy for a time traveller to do something in the
past that threatens his own existence. Suppose baby Toms presence at D results
in the couple who discover him deciding not to have a child of their own; instead
they adopt the mysterious baby they found in their garden a year earlier. This
baby is, of course, none other than Tom himself, the baby they have just decided
not to have. However, their decision does not threaten the existence of Tom, who
was conceived and born on the P
4
t
4
timeline; their decision is made on the P
5
t
5

timeline. Contrary to what Meiland suggests (1974: 173), however, there may
in fact be ways in which a traveller could undermine their own existence in this
model. What if Toms time machine were to land on his own mother at point C,
killing her three years before his own birth?
There are models of reality in which even this sort of eventuality is unproblem-
atic: multiverses. In these realities there are many parallel dimensions, each
containing a distinct spatiotemporal system, some of which resemble one another.
If a way of moving between dimensions were to be discovered, a traveller could
arrive at what seems to be a period in their own worlds past.
In Figure 8.2 each column depicts a portion of the history of a different uni-
verse. Tom, our would- be time traveller, wants to prevent the outbreak of a dev-
astating war, and so travels back into what he believes is his own past. He sets off
from the A- universe and arrives at the B- universe, at a time that strongly resem-
bles an earlier time in the A- universe. Indeed, until the time of Toms arrival,
the A- and B- histories are exactly the same, but diverge thereafter due to Toms
arrival: no (apparent) time traveller arrives at the corresponding time in the
A- universe. Alas, Tom fails to stop the war, and so decides to make a further
attempt, by travelling still further back in time. He arrives at the indicated point
in the C- universe, which, until this time, exactly resembles the A- and B- universes.
Tom succeeds in preventing the war, and lives in the C- universe for the remainder
of his days. This success comes at a cost. Toms activities do not just prevent the
outbreak of war but they result in his parents never meeting, and so Tom (or his
C- counterpart) is not born. But since our Tom originated in a different universe there
Figure 8.2 Interdimensional travel. A trip which takes you to
another universe might seem like a voyage into your own past,
provided you arrive in the other universe at a time when it resembles
your own universe at a time prior to your departure. A, B and C are
distinct universes, the arrows indicate your interdimensional travels.
B and C closely resemble A until the time of your arrival thereafter
what happens is very different.
TIME TRAVEL 127
is no threat of paradox, and no threat to (this) Toms existence. Close variants of this
model of backward time travel can be realized in a number of ways. Instead of travel-
lers arriving in parallel universes they might land in different limbs of a branching
universe. It may well be that their arrival produces the limb in question. The version
of Presentism that takes reality to consist of innumerable momen tary reality- slices
that are related only by resemblance (cf. 6.9) can also be viewed as a Many- Worlds
model, although the worlds in question are of a very distinctive kind: highly frag-
mented and over lapping.
There are writers who believe that time travel, if it is possible at all, must take
this form. Deutsch and Lockwood (1994) argue that backward time travel within a
single universe leads to irresolvable paradoxes, and so can only occur between uni-
verses; as firm adherents to the many- worlds interpretation of quantum mechan-
ics, they are confident that such universes exist (Lockwood 1989; Deutsch 1997).
Not everyone shares their confidence there are less extravagant inter preta tions of
quantum mechanics that posit only one world but we can agree with them on one
point. If we do live in a multiverse that permits movement between worlds, then
the sort of time travel they envisage may be possible depending on the availability
of suitably similar worlds and if so, it would not be threatened by paradoxes that
afflict time travel within a single world. However, whether the envisaged dimen-
sion hopping really amounts to time travel is open to question. It certainly doesnt
involve visiting ones own past.
2
Irrespective of the termino logical proprieties,
it is time travel within a world that has greatest philosophical interest, precisely
because of its potentiality for paradox which leads us to our next topic.
8.3 Self- defeating loops
The most serious threat to backward intra- world time travel is familiar from
science fiction: the self- defeating loop. Someone who travels into their own past
could do something to ensure that they themselves will not exist at the time of
their departure, such as kill a parent, grandparent, or even their own younger self.
Successfully carrying out this action makes it impossible for the action itself to
occur, which is a paradoxical state of affairs if ever there was one.
There are ways of telling the tale that can tempt us into thinking that the impos-
sible is possible. The 50- year- old time traveller accidentally kills his 10- year- old
self, and then as the realization dawns, he sits down, resigned to his inevitable fate:
he knows the past four decades of his life are steadily being wiped from history,
his twenties and thirties are probably already gone, he has only a few moments left
before annihilation. But while we may be able to visualize such a process at least
seeing a film that tells a story along these lines it could never occur. Someone
who dies aged 10 does not live to see fifty. Someone who succeeds in travelling
back in time could not possibly kill their younger self, or do anything else that
would make it impossible for them to travel back. And yet they surely could. If
you were to travel back, what could prevent your buying a gun and blowing out
the brains of your earlier self, whether on purpose or only accidentally? Hence
the real paradox seems to be this: there are actions that a time traveller both can
and cannot perform.
128 TIME AND SPACE
Bearing in mind what has already been said about the fixed character of a
block-universe, a four- dimensionalist wishing to defend time travel may elect to
meet the challenge head- on:
You exist now, and have existed for the past thirty years. Even if you enter
your time machine tomorrow, and go back in time and try to kill yourself
as an infant, we can be certain that you will not succeed simply because you
didnt. Since in actual fact you survived your childhood, it is an unalterable
fact as of now that any attempts made on your life during this period fail. So
there is no paradox. Since it is logically impossible for you to commit auto-
infanticide, it is not true that you both can and cannot kill your earlier self.
While endorsing this general line, Lewis (1976) takes a further step and argues
that the paradox- mongers are guilty of equivocation. To say that someone both can
and cant do X is to make two different claims, and these claims are only contra-
dictory if can has the same meaning in both. Lewis suggests that it doesnt. It is
not logically possible for a time traveller to kill his earlier self, since he didnt, so
in this sense he cant do it. But it is not practically impossible he has the strength
and know- how, he is at the right time and place so there is a perfectly legitimate
sense in which he can perform the act in question, it is just that, for whatever
reason, he doesnt (or didnt).
To say that something can happen means that its happening is compossible
with certain facts. Which facts? That is determined, but sometimes not deter-
mined well enough, by context. An ape cant speak a human language say
Finnish but I can. Facts about the anatomy and operation of the apes
larynx and nervous system are not compossible with his speaking Finnish.
The corresponding facts about my larynx and nervous system are compos-
sible with my speaking Finnish. But dont take me along to Helsinki as your
interpreter: I cant speak Finnish. My speaking Finnish is compos sible with
the facts considered so far, but not with further facts about my lack of train-
ing. What I can do, relative to one set of facts, I cannot do, relative to
another, more inclusive set. Whenever the context leaves it open which facts
are to count as relevant, it is possible to equivocate about whether I can speak
Finnish. (1976: 143, original emphasis)
Likewise for whether you can kill your earlier self. In an ordinary case of attempted
assassination, in judging whether the would- be killer is able to pull off the job, we
take the range of relevant facts to include the power of his weapon, his distance
from the target, his marksmanship, resolve, concentration we would not dream
of including facts about the future in our assessment. By this delineation of the
relevant facts, you are perfectly able to kill your younger self. You dont succeed,
of course, but there is nothing paradoxical about that: we often fail to do things
that we are capable of doing. Luck has a part to play in any success, and without
luck all the preparation in the world will come to naught. However, if we consider
the same question from the perspective of a more inclusive set of facts, a set that
includes facts about the future which, unusually, in this case are known then it
is clear that you cant succeed. Your younger self lives to do a great many things,
not the least of which is take a trip into the past.
TIME TRAVEL 129
8.4 Global consistency constraints
It is hard to fault the logic of this reply, but it is also hard to believe that there is
nothing more to be said. Consider an example that avoids the psychological and
moral complications of self- murder.
You are a committed Block theorist and, although you have no doubts about
the unalterable character of the past, you decide to put it to the test using a time
machine that has recently fallen into your hands. There is a tree in your back garden
on which you carved your initials (and the date) some twenty years ago. You decide
to travel back fifteen years and try to cut it down. Since the tree evidently survives,
you have no doubt that you will fail, but you want to find out what happens when
you make the attempt. So you load a chainsaw, an axe, some dynamite and various
other destructive implements into the time machine, enter the appropriate coordi-
nates, and moments later find yourself in your back garden, as it was fifteen years
ago. Picking up the tools you walk up to the tree, verify that it is the right one yes,
the date and initials are clearly to be seen switch on the chainsaw and set to work.
What happens next? Do the laws of logic prevent the saw from cutting into the
tree? It doesnt seem possible; only material forces and obstacles could do this.
So something else must happen. Some irrational fear stays your hand at the last
moment. Or the saw breaks down, and when you pick up the axe you pull a muscle
and find yourself unable to wield it. No matter what you try, no matter how often
you try, all your attempts fail. Now, assuming that there is no conspiracy here no
one has found out about your plans and has gone to extravagant lengths to thwart
your attempts your long succession of failures is purely coincidental. There is no
unifying explanation; each failure has its own independent cause, and the cause is
always local the fact that the tree exists tomorrow (and for many years to come)
is not what caused your chain saw to fail (it was a faulty electrical connection) or
made your dynamite damp (that was a passing dog).
Hence the problem. Isnt this series of coincidences very peculiar, to say the
least? Is it reasonable to think such a sequence could occur, not only in this instance,
but in every instance in which a time traveller attempts to undo the future? Nahin
writes he must fail because he did fail. To demand an accounting for the specific
why of failure before accepting the failure is as misguided as a stranded motorist
refusing to believe his car wont start until he knows why (1993: 196). But there
is bad luck and there is bad luck. Confronted with many immensely improb-
able sequences of coincidences (one for each sustained attempt to create a self-
defeating loop), the idea that bad luck is the only factor at work seems ridiculous.
Unless some more plausible explanation is forthcoming, the rational response
would surely be to dismiss the possibility of backward time travel.
Coincidences do happen, and from time to time very big coincidences happen,
or at least they could; so this objection does not demonstrate that time travel is
logically impossible. But it does merit a response. Can we at least render it intel-
ligible why block-universes that contain lots of time travellers intent on creating
self- defeating loops must also be universes with lots of highly improbable coinci-
dences? Perhaps.
Imagine that I am a God- like being who has decided to design and then create
a logically consistent universe with laws of nature similar to those that obtain
130 TIME AND SPACE
in our universe, laws that cannot be broken (so miracles will never occur). On
a perverse whim, I have also decided to include within it time travellers of the
troublesome sort. Since the universe will be of the block- variety I will have to
create it as a whole: the beginning, middle and end will come into being together.
Now, given that I will be introducing troublesome time travellers, all of whom
will make many attempts to bring about self- defeating loops, to ensure that their
attempts fail I have to incorporate a good many coincidences into my design.
There is no other option, since I am con strained by both logical consistency and
the no miracles decision. As a consequence, the existence of these coincidences
is not coincidental: they must be there if the universe is to exist at all, given that
the universe is going to contain troublesome time travellers. So we see that, in this
context at least, coincidences are unsurprising; we have an explanation of why
they have to exist.
But what if our universe was not designed in this way, but does include trouble-
some time travellers and the associated coincidences? Do we still have an explana-
tion for why these coincidences exist? Well, assuming that our universe is of the
four-dimensional block variety, even if it never came into being, it nonetheless
exists (timelessly) as a coherent whole, contain ing a globally consistent spread
of events. At the weakest level, consistency here simply means that the laws
of logic are obeyed, but in the case of universes like our own, where there are
universe- wide laws of nature, the consistency constraint is stronger: everything
that happens is in accord with the laws of nature. In saying that the consistency
is global I mean that the different parts of the universe all have to fit smoothly
together, rather like the pieces of a well- made mosaic or jigsaw puzzle. A trouble-
some time traveller can be thought as a particularly complicated and awkward
piece (or collection of pieces) of the mosaic, with parts oddly spread through time.
The pieces that surround the parts of the travellers life when he is trying to create
self- defeating loops will not be unusual in themselves they will be events that are
lawfully related to their surroundings but they will also constitute an improbable
series of coincidences. Or rather, they will when viewed from a local perspective;
from a global perspective, given the need to accommodate all stages of the travel-
lers life including those that occur later than the events in question it will be
obvious that events of that sort had to occur. A universe that conforms to our laws,
is a globally coherent whole and contains awkward time travellers will necessar-
ily include the problematic coincidences. But since, given these constraints, they
had to occur, the coincidences are not in fact problematic: we have a reasonable
explanation for why and how they exist.
3
If this strikes us as odd it is because we are unused to thinking of the universe
as a vast spatiotemporal mosaic, but if the universe is a vast spatiotemporal mosaic
then, given the reality of the future, the future determines the past as much as
the past determines the future. The constraints that later events place on earlier
ones are not always causal (although it can be in the special case of backward time
travellers). It is more typically a matter of coordination: the future events exist in
the same universe as the earlier events, in a coherent, smooth- fitting, law- abiding
whole. So given that the future is in a certain way, there are only so many ways
the past can be.
TIME TRAVEL 131
8.5 Bilking
Backward time travel (as usually conceived) involves backward causation: for
example, the time travellers arrival in the past is the causal consequence of an
event in the future the pressing of the button that activates the machine.
4
But
backward causation comes in other, less dramatic guises; for it involves nothing
more (or less) than an earlier event being causally dependent on some later event.
The earlier event could be the flashing of a light bulb and the later event the press-
ing of a switch; there is no need for objects to be translated through time. Can we
make sense of backward causation? The so- called bilking argument suggests not
(cf. Flew 1954, 1956, 1957), and is worth exploring for the additional light that
it casts on the issues that we have just been considering.
In Dummetts classic discussion (1964), a tribal chief performs a ritual dance
every day during the six- day period in which the young men of the village are
away hunting, even though it takes a good two days to reach and so return
from the hunting grounds. Asked why, he says So as to ensure the young men
perform bravely on the hunt. The chief, it seems, is convinced that his dancing
significantly affects the chances of the young men acting bravely. The question is
put But why do you continue to dance for the final two days, when you know
the hunt is over and they are on their way back? Surely dancing on the first four
days would be enough? The chief s response is that experience has shown other-
wise. Another chief in a neighbouring village tried to cut corners in this way and
the hunters behaved ignobly; the same happened on the two occasions when he
himself fell ill and was unable to dance for the final two days. On the other hand,
whenever the dance is performed on all six days, the hunters usually turn out to
have acted bravely.
Finding these claims suspect we decide to subject them to the rigours of the
experimental method, and we set about the task of designing a series of tests that
will reveal whether the chief s dancing does exert a backward influence on the
behaviour of the hunting party. The tests we come up with are both simple and
effective; so much so that it seems that we neednt bother actually carrying them
out, for, no matter what the result, the backward causation hypothesis is refuted.
Consider the following scenario. We repeatedly send someone out with the
hunt equipped with a radio, which allows us to know whether the hunters have
behaved bravely before they return to the village. On occasions when the hunters
have behaved in a cowardly way, we pass the news on to the chief, and persuade
him to dance anyway in a spirit of cooperation he agrees. Now consider the pos-
sible results. If the chief dances successfully, and does so on numerous occasions
when the men have not been brave, his claim that his dancing produces bravery
is undermined. At the very least, his dancing is not sufficient for the bravery. But
suppose instead that he tries to dance, but repeatedly fails for no apparent reason
(his limbs keep locking up). This result does not refute the hypothesis that bravery
and dancing are causally correlated, but it serves instead to cast doubt on the direc-
tion of the causal influence, for it now seems plausible that it wasnt his dancing
that was causing the bravery, but rather the young mens bravery that was making
it possible for him to dance. The chief s dancing was not, as he thought, an action
that it was in his power to perform or not perform as he saw fit; the hunters acting
132 TIME AND SPACE
bravely was a necessary causal precondition for his dancing. To strengthen our case
we could carry out another experiment. We could try to prevent the chief from
dancing on occasions when (thanks to our advance knowledge) we know that the
hunters have acted bravely. If we succeed on repeated occasions, the hypothesis
that dancing causes bravery is undermined; the dancing has been shown to be
irrelevant to how the hunters act. But suppose that, despite our best efforts, the
chief finds himself compelled to dance and we just cant stop him. The hypothesis
that the dancing and the bravery are causally correlated is not refuted, but we
now have reason to believe that the earlier acts of bravery are causally sufficient
for the dancing.
The general policy adopted in this case is known as bilking. Confronted
with an alleged case of backward causation, we repeatedly observe the presence
(or absence) of the alleged earlier effect (E), and then try to prevent (or produce)
the alleged later cause (L). It seems that, no matter what the result, the hypothesis
that backward causation occurs is falsified. Either there is no correlation between
the two at all L occurs without E, and E without L or else the bilking attempts
fail, in ways that suggest that the later event is in fact causally dependent on the
earlier event.
Is the bilking argument decisive? Not at all, but the reasons why not are less
than entirely straightforward.
As Horwich argues (1987: 939), in cases where the bilking policy is success-
fully implemented (either the chief dances when the young men didnt act bravely,
or he doesnt dance when they were brave), a response is available to the chief.
He could reasonably insist that although his dancing increases the probability of
bravery, and there is lots of prior evidence for this, it is not infallible it is not
nomologically sufficient for bravery in all conditions and that since the condi-
tions in which his dancing is effective are very delicate, no doubt the establishment
of the new channels of communication disrupted them. However, while this reply
does show that the backward causation hypothesis is not automatically refuted
by the successful implementation of the bilking procedure, it does leave us with
some puzzling coincidences. There are many ways that the bilking policy could be
implemented (using radios, carrier pigeons, different stories fed to the chief, etc.),
yet it turns out that every implementation changes the background conditions in
such a way that the chief s dancing proves either ineffective or unnecessary, as the
circumstances require. Coincidences such as these are not impossible, but they are
definitely suspicious.
But suppose that the bilking policy doesnt succeed, and we cannot produce L
when E doesnt occur, and when E does occur, we cannot prevent L from occur-
ring. There are two ways for this to come about.
First, it could be that the types of event in question are nomologically con-
nected. The proponent of the bilking argument construes this connection in a
particular way: the bravery causes the chief to dance; the absence of bravery results
in his losing his ability to dance. But why assume that the no- braverychief- cant-
dance and braverychief- compelled- to- dance combi na tions establish the nomo-
logi cal dependency of the later event on the earlier event? If E is necessary for
L, then L is sufficient for E; if E is sufficient for L, then L is necessary for E.
Regarded in a neutral light, these results are just as compatible with the hypothesis
TIME TRAVEL 133
that the bravery depends on the dancing as they are with the hypothesis that the
dancing depends on the bravery. To maintain that in such cases forwards cau-
sation is more likely than backwards, on the grounds that causa tion normally
acts forwards rather than backwards is clearly begging the ques tion. However, as
Horwich points out, the unorthodox interpretation is in one respect problematic.
The earlier effects the acts of bravery all have their own independent causal
antecedents, such as the young mens upbringing and the growing food shortage,
which are in themselves sufficient to ensure that they would act bravely on this
occasion irrespective of what occurs subsequently. So, assuming the efficacy of the
dancing, we have a case of causal over determination. We are also presented with
more puzzling coincidences. The chief s dancing may make the men act bravely,
but it isnt responsible for the circumstances leading up to the hunt, which are also
responsible for their bravery, and yet the two are correlated: whenever the chief
dances, conditions prior to the hunt are such that the men will act bravely anyway.
The second reason why bilking attempts might fail is familiar from our earlier
discussion of self- defeating loops strangely convenient coincidences such as
when the young men dont behave as bravely as they might, the chief twists his
ankle and cant dance, or the message from the hunting grounds is garbled in
transmission and we end up believing that the young men were brave and so dont
ask the chief to dance; there are many possibilities. The same applies to the cases
where the young men act bravely: through some mishap or other, the chief always
ends up performing his dance.
Horwich suggest that these results, taken together, show that the bilking argu-
ment fails: it turns out that there arent any simple, easy- to- perform experiments
that can decisively refute a backward causation hypothesis. It may well be that
particular instances of backward causation are surrounded by various unexplained
coincidences, but in the light of our discussion of self- defeating loops, this is to be
expected. If backward causation does obtain in a given case, anyone who attempts
to implement a bilking policy will be attempting to create a self- defeating loop
and, since such loops cannot exist, their attempts will invariably fail, even though
there will usually be no single unified explanation for these failures. The cases
where it is hardest to believe this could come about are those in which a person is
compelled to act in one way rather than another. A time traveller from the future
tells you that on a certain date you will order such and such a dish in a particular
restaurant: he has seen you do it. To thwart fate you decide to bilk. No matter what
the cost, you are determined that you will be doing something else on the date in
question. Perhaps you succeed; if so, your informant was not a genuine time travel-
ler, or else was lying. But if he wasnt, if he really did see you ordering that dish
in that restaurant on that day, then, no matter what you do, events will conspire
to ensure that you end up doing precisely this. As for why events so conspire (or
seem to), there is only one explanation: global consistency constraints.
8.6 Quantum retroaction
Anyone who remains unconvinced by the defensive strategies just outlined should
not conclude that backward time travel (or causation) is an absolute impossibility:
134 TIME AND SPACE
there may well be universes where such travel is possible, but only in such a way
that it is impossible to generate self- defeating loops. For example, the physical laws
that make backward travel possible are such that objects always arrive billions of
light- years away from where they set off a distance that makes it impossible for
the travellers to causally interact with their earlier selves. Time travel in this guise
undeniably has less dramatic potential, but it is also far less problematic. And the
same applies to instances of backward causation, where the earlier effect is so
distant from the later cause that bilking is not an option.
5
However, there is a further possibility that we have not yet considered, a way
in which backward causation (and travel) could occur locally without the threat of
self- defeating causal chains arising. The loophole was discovered by Dummett in
the course of his exploration of the dancing chief case, but I will focus on Prices
recent discussions of it, which broaden both the scope and the interest of the basic
idea.
Unwilling to believe that the highly improbable coincidences required to protect
backward causation from successful bilking could occur in our universe, Price
believes that it is reasonable to conclude that physical law does not permit instances
of backward causation that could be bilked.
6
However, from this it only follows
that backward causation is impossible if bilking is always possible, and Price sees
no compelling reason to believe that it is. To enact a bilking policy you must first
discover whether the alleged earlier effect E has in fact occurred (only then can you
decide to prevent or encourage the subsequent cause L). But suppose E is such that
either (i) it is nomologically impossible to discover whether it has occurred before L
occurs, or (ii) it is possible to discover whether E has occurred, but only by carrying
out procedures that are themselves causally relevant to whether E occurs or not. In
such circumstances the bilking strategy cannot be implemented successfully.
For most ordinary past events neither (i) nor (ii) obtains. It is possible, at least
in principle, to discover whether the event in question obtained, and by means
that do not influence the event in the relevant respects; that is, the event would
have happened (or not) irrespective of our attempts at detection. It is, for example,
plausible to suppose that the behaviour of the hunting party would not be affected
by a concealed observer whom they did not realize was present. But there are
conceivable circumstances in which this is not so. The case that follows combines
(apparent) backward causation and inaccessibility of the relevant sort.
Imagine a gambling machine that drops a ball into one of two closed boxes, X
and Y. Players bet on which box contains the ball, and choose which box is opened.
The machine is so constructed that the boxes cannot be opened simultaneously,
and the only (physically possible) way of finding out which box contains the ball
is by opening the box; X- rays along with other penetrative probes either dont
work or cause the device to self- destruct. As it happens, it is always the chosen box
that contains the ball, and gamblers soon realize that they can make a lot of money
by opening the box on which they have placed a bet. As Price notes, It would
seem natural for players to speak of choosing to open box X, in order to ensure
that the ball had dropped into it (1984: 310, original emphasis), thus invoking
backward causation.
We are not obliged to adopt this view as there are two alternative explanations.
The first is the discontinuity option: if the ball has dropped into box Y, it moves
TIME TRAVEL 135
instantaneously to box X when the latter is opened. The second is the indetermi-
nacy option: until box X is opened the ball is in an indeterminate condition it
is located in neither box and opening X instantly causes the ball to acquire a
determinate location. If we are prepared to countenance instantaneous travel, or
objects with indetermin ate properties, we might adopt either of these accounts.
Or, alternatively, we could accept the reality of backward causation: the ball falls
into box X because this is the box that the gambler will open. It would be wrong to
reject this account on the grounds that backward causation could lead to paradox
or (nomologically) impossible coincidences, for in this case they cant. Since there
is no way of finding out where the ball has fallen until the box is opened, bilking
is not an option.
This example is of more than curiosity value, for in key respects it is analogous
to situations that are commonplace in quantum physics. According to quantum
orthodoxy, the state of a quantum system is indeterminate in certain respects until
a measurement is conducted; the act of measurement results in a collapse (of
the wave- function) yielding determinate properties. It is very odd to think that a
particle can fail to have a definite location or momentum (and just as odd to think
that it can have a combination of different locations and momenta) but, in fact,
experimental findings make this a plausible view to hold.
Price cites an example due to Putnam (1979b). Suppose that S is a system con-
taining a large number of atoms, R and T are two incompatible properties and A
and B are two different states about which the following are true.
1. When S is in state A, 100 per cent of the atoms have property R.
2. When S is in state B, 100 per cent of the atoms have property T.
Quantum mechanics predicts that there is a state C of the system a special kind
of combination of A and B of which the following are true.
3. When S is in state C, 60 per cent of the atoms have property R.
4. When S is in state C, 60 per cent of the atoms have property T.
And these predictions are confirmed by experiment. Needless to say, this is a very
perplexing result. If ten marbles are placed in a box, and you know that they are
red or green, and if you discover that six are red, then you would reasonably con-
clude that four must be green; how could you count out six red ones and also six
green ones? We are saved from an outright contradiction because it turns out that
it is impossible to test both 3 and 4 for a given system S. If we test 3 the system is
disturbed in such a way that 4 cannot be tested; likewise mutatis mutandis if we
first test for 4.
Putnam suggested that we can only make sense of such a situation by rejecting
The Principle of No Disturbance (ND), according to which performing a meas-
urement on a given system does not disturb what is being measured, so the value
of the relevant quantity is very nearly the same an instant before the measurement
is made as when the measurement is actually made. This is an understandable reac-
tion. Since it is impossible for 60 per cent of the atoms in S to have both R and T
prior to the measurement the properties are incompatible it seems reasonable
136 TIME AND SPACE
to suppose they have neither until the measurement is conducted, in which case,
the act of measurement must itself be causing the system to enter a determinate
state. However, Price argues, in concluding that we must reject ND, Putnam is
ignoring an alternative explanation, one that relies on backward causation. The
procedure for measuring R is different from the procedure for measuring T, so it
could be that if we decide to put 3 to the test, the fact that an R- measurement will
be carried out causes the system to enter the predicted state 60 per cent of the
atoms having R right at the start. The atoms thus have determinate properties
before they are measured, but only because they are causally influenced by a future
event. And the same would apply if a T- measurement were to be conducted. The
situation is a probabilistic analogue of the gambling machine: the ball falls into X
because X will be opened; if the gambler had bet on and opened Y, the ball would
have fallen into Y instead.
Backward causation can be deployed in other quantum mechanical contexts.
So- called EPR- correlated particles (which we shall be encountering again) are
typically pairs of particles that are produced by a single source event, which then
speed away from one another while remaining intimately related: if at some later
time a measurement is conducted on one of the particles, the other is instantane-
ously affected in a particular way. Or at least, so runs the orthodox interpretation.
This result is par ticu larly problematic since instantaneous interactions seem to be
ruled out by Einsteins special theory of relativity.
7
But there is no need to postulate
instantaneous action- at- a- distance if we introduce backward causation into the
picture, and stipulate that the future measurement retroactively determines the
states of both particles at the moment of their creation.
8
This radical proposal is highly controversial, but so, too, is every other current
interpretation of quantum theory, and it does have the merit of replacing several
bizarre phenomena (indeterminacy, non- local influence) with just one: backward
causation. There is also the striking fact that, since the rules governing quantum
systems render bilking impossible, the principal objections to backward causation
can find no grip. Is this merely a coincidence?
8.7 The inexplicable
Backward time travel opens up the possibility of loops of yet another kind; not
self- defeating (or at least not obviously or directly), but nonetheless peculiar and
intriguing in their own right. Since they also give rise to new issues, they are
worthy of a brief mention.
The Marriage Decision Mary is torn between two suitors. She cant decide
whether to marry Tom or Jack. So she travels to the future and finds out that
she is happily married to Tom. She then travels back, and marries Tom for this
reason.
The Discovery You are looking for a lost sock one day, in the depths of a
drawer, and come across a book that you havent seen before. It seems to
contain details for a time machine. You dismiss it as a prank, but years later,
now an eminent scientist, you realize that the design in the book is a serious
TIME TRAVEL 137
piece of physics. You build the machine, and find that it works. After years of
travelling back and forth and many adventures, you realize that you still dont
know how the book with the design for a time machine got into your drawer.
So you travel back to your home on the day you originally found the book to
see who plants it. The minutes pass, no one comes, and you remember that
your original self will return in about five minutes. Suddenly you realize what
has to happen. With a feeling of unease, you place the book in the drawer
(happily you had it on you) and quietly withdraw from the scene.
The Big Loop Many years from now, a transgalactic civilization has discovered
time travel. A deep- thinking temporal engineer wonders what would happen
if a time machine were sent back to the singularity from which the big bang
emerged. His calculations yield an interesting result: the singularity would be
destabilized, producing an explosion resembling the big bang. Needless to say,
a time machine was quickly sent on its way.
There are many similar tales.
9
Are they coherent? Although there are no glaring
inconsistencies, closer scrutiny reveals that some loops are more dubious than others.
A useful starting point is a distinction between loops in which objects come full
circle, and those that involve only information. Some object- loops are certainly
impossible. In The Discovery a young man finds a book in his drawer, reads it,
carries it about for a number of years, and then replaces it in the drawer for his
younger self to discover. Unless the book is made of a miraculous material that is
entirely immune to the ravages of ordinary wear and tear, the book that is replaced
will be in a different physical condition from the book that is discovered; since
they are supposed to be one and the same, this is impossible, so the envisaged sce-
nario could not take place. This may seem odd: what is to prevent your handing the
book over to your younger self? But this is just the problem of the self- defeating
loop in another guise: since you cant do the impossible you didnt hand the book
over, and since you didnt, you cant. The Big Loop also involves an object the
time machine itself but since the machine does not coincide with itself in two
different conditions, the paradox of the book does not arise. The loop here is of
the causal rather than object- involving variety.
Now consider a variant of The Discovery. One day, in your youth, you receive
a mysterious email. It seems to provide the details of how to build a time machine.
Years later you send the mysterious email to your earlier self. Here there is no
object paradox, but there is still a puzzle. Where did the information about how
to build a time machine come from? It just seems to be there, present throughout a
closed loop. Opinions differ as to how problematic this is. Lewis sees no difficulty:
But where did the information come from in the first place? Why did the
whole affair happen? There is simply no answer. The parts of the loop are
explicable, the whole is not. Strange! But not impossible, and not too dif-
ferent from the inexplicabilities we are already inured to. Almost everyone
agrees that God, or the Big Bang, or the entire infinite past of the Universe,
or the decay of a tritium atom, is uncaused and inexplicable. Then if these
are possible, why not also the inexplicable causal loops that arise in time
travel? (1976: 140, original emphasis)
138 TIME AND SPACE
In the light of The Big Loop the big bang might not be entirely inexplicable, but
this quibble aside, Lewis seems to have a point. Deutsch and Lockwood (1994)
disagree, arguing that informational loops violate a metaphysical constraint that
they attribute to Popper: knowledge comes into existence only by evolutionary,
rational processes; solutions to problems do not spring fully formed into the uni-
verse. But is this no free lunch constraint on information really metaphysical?
According to quantum theory it is nomologically possible (although extremely
improbable) for complex objects to appear from nowhere, random distillations
from the quantum foam; classical thermodynamics carries a similar implication.
An edition of the complete works of Shakespeare is an object like any other. Does
the fact that it carries information mean that it is an object of a special kind, and
distinguished as such by physical law? It seems unlikely. And there are other ways
in which Poppers constraint could be violated. Set enough monkeys randomly
tapping on typewriters long enough, and sooner or later a book containing any
information you like would be produced. If information can spring into being in
these ways, why not in the time travel cases too?
8.8 Voyaging in dynamic time
Time travel may be possible in a Block universe, but what of dynamic worlds?
On the face of it the prospects look dim. The seriousness of the no destination
problem as Grey (1999: 56) aptly calls it is most obvious in the case of Present-
ism. Even if we follow the Dynamic (or Compound) Presentists, and suppose that
one present succeeds another, it is hard to see how travel to another time period
is possible, since at (or, more precisely, as of) the moment of departure there are
no times other than the actual present where one might arrive. Since time travel
involves taking a trip to other times, if there are no other times it seems one can-
not make the trip.
But perhaps it would be a mistake to draw this pessimistic (for would-be travel-
lers through time) conclusion so swiftly. Keller and Nelson (2001) have argued
that the no destination problem does not in fact pose an insuperable barrier to
Presentist time travel. At least in the case of Dynamic Presentism, the world isnt
confined to a single present; rather, it consists of a succession of presents. Accord-
ingly, on the assumption that I will persist through at least the next few minutes,
although I now exist in this present, I will shortly exist in the present which follows
this one, and then the next and so on. Since the model of temporal passage to
which Dynamic Presentists are committed clearly allows an individual to exist in
more than one present, it looks as though the no destination problem cannot
be the insuperable problem it is often taken to be, at least on the assumption
(currently in play) that the Presentists conception of passage is metaphysically
coherent.
Keller and Nelson go on to make a further point. Although Dynamic Presentists
hold that only the present time is real or actual, they also hold that there are true
propositions concerning what occurred in the past, and perhaps the future. If at
some future time I travel back to 1850, there will be propositions such as In
1850, BD steps out of a time machine, and goes looking for a hot dinner that are
TIME TRAVEL 139
true. Since the Presentist can accept that such propositions are true at least when
suitably formulated in terms of temporal operators such as It was the case that
it looks very much as though they can accommodate time travel, quite easily.
And, of course, similar remarks can be made about time travel into the future, for
those Presentists who believe that such travel is possible.
While all this is true, serious doubts remain. Construed thus, what does time
travel really amount to? In the case of the four-dimensional continuum of the
Block universe, everything is clear: if I travel back to 1850, I continue my exist-
ence there, back in the past, in the form of a solid, three-dimensional, live human
being an entity that is just as real as any presently existing thing. In contrast,
since for the Presentist no past times exist, neither myself nor my time machine can
arrive at any past time. There are now true propositions concerning my activities
in 1850, but what makes those propositions true are not the doings of a flesh and
blood human being who is living out a part of his life in 1850. When it comes to
truthmakers for claims about the past, as we saw in 6.10, there are three main
options open to the orthodox Presentist: (i) reductionism; (ii) presently existing
Lucretian properties; and (iii) the collections of mutually consistent propositions
(or world-stories) that constitute ersatz B-series. It is not difficult to see that each
of these options is problematic, albeit in different ways.
Reductionism poses an obvious threat to anyone contemplating backwards time
travel. For unless your trip to the past results in some traces that are discernible
in the present, propositions about your past doings will lack a truth-value, and
so you risk either annihilation, or the limbo of indeterminate existence. There is
a second, and arguably more serious problem. If the relevant truthmakers all lie
entirely in the present (in the form of trace evidence, newspaper reports, etc.), then
a question arises: have you truly succeed in leaving the present at all? It is not clear
that you have: a case could be made for holding that you have been reduced to a
collection of presently existing trace evidence. Essentially the same problem arises
if we suppose that the propositions about my doings in 1850 are made true by
presently instantiated Lucretian properties. These may be enough to make various
propositions about your adventures in the past true, but even so, if I believed that
activating my time machine would reduce me to nothing more than a collection
of presently instantiated Lucretian propositions, I would only be tempted throw
the switch if I were in a suicidal frame of mind. Much the same applies if claims
about the past are made true by a collection of abstract descriptions in an ersatz
B-series. From an ontological point of view, in this case it is not at all clear that my
adventures in the (ersatz) past would be different in kind from the escapades of a
character in fiction. No doubt many would-be time travellers hope their exploits
will one day become the stuff of legends, but they also hope to have their adven-
tures before this happens. If we adopt the ersatzer view of the past, it seems time
travellers will skip the actual adventures and pass into legend directly.
10
It may be that the Compound Presentism, as outlined in 6.11, has the resources
to provide a more solid foundation for claims about the past, and hence of travel
to other times. Since on this view successive times coexist, there is scope for truths
about the past to be transmitted through chains of partially overlapping extended
presents. In which case, it may well be that the (Compound) Presentist can make
acceptable sense of a traveller arriving in the past, and having adventures there.
140 TIME AND SPACE
However, there remains the question of how such a trip would be possible at all.
Magic remains an option: with a wave of the magicians wand you find yourself
instantly transported to another time, with no other technology deployed. This
may be possible, but time travel loses some of its interest if magical methods are the
only methods that can bring it about. But as things currently stand, the methods
of time travel into the past that have a chance of being compatible with our own
physical laws for example wormholes, closed timelike curves (briefly surveyed
in 8.9) all require earlier and later phases of the universe to coexist, and hence
presuppose the four-dimensional Block conception of time. Even if Compound
Presentism renders travel to distant times intelligible, it may not be a nomological
possibility in a universe such as our own.
The situation with regard to the Growing Block model is interestingly differ-
ent. Having constructed (what I believe to be) a working time machine here in
the year 2010, I might think Since the future is not real as of now, I cant travel
there. Fine. But the past is real as of now, so there is nothing to prevent me from
paying it a visit. Ill start with the sixteenth century!, which sounds reasonable
enough. However, the situation looks very different from the perspective of the
selected time in the sixteenth century: as of this time the future is not yet real, so
nobody could arrive from it; since nobody could arrive from it they didnt; and so
my time machine cant possibly materialize in the sixteenth century. Or at least, it
could only arrive by changing the past, which for Growing Block theorists is
simply not possible. This no time of departure problem is the counterpart of the
Presentists no destination problem.
Is there any way around this difficulty? Perhaps, but some fairly radical meas-
ures are required. The Growing Block theorist could try arguing that although the
future is unreal, it is not the case that all claims about future contingencies lack
a determinate truth-value: propositions relating to the future doings of would-be
backward time travellers can have a determinate truth-value, as of the times at
which these travellers arrive in the past. So as of now, the future is unreal, and
many claims about what it contains may be neither true nor false, but if travellers
from the future arrive in our present, we can be confident that however the future
turns out, it will at least feature these same travellers and their time machines. The
idea that the future might be partially determined, but not wholly so, is undeni-
ably strange, but it may not be logically incoherent. See Slater (2005) for further
discussion of this issue.
Thus far we have been considering trips through time of a fairly conservative
sort: in effect, we have been considering whether it is possible to jump across inter-
vals of time within a single dynamic world. Even if this sort of travel is impossible,
there may be more radical modes of time travel that are unaffected by temporal
dynamism. To mention just one, let us suppose our universe is of the Growing
Block variety, and let us further suppose that it is possible to build a machine that
is able to leave this universe (and so our time) and re-enter it at some different
location. Were you to depart from the twenty-first century, spend some (personal
or proper) time in this atemporal limbo and then re-enter our universe in the
sixteenth century, it is not clear that the no time of departure objection has
any force: for you are no longer arriving directly from a non-existent time. If it
is conceivable that someone originating outside this universe could materialize at
TIME TRAVEL 141
some location in the sixteenth century (for the sake of the argument, grant that it
is), why couldnt you do likewise? Why should it matter that you spent some time
in this universe before leaving it?
It is probably fruitless to spend time trying to work out whether scenarios as
far- fetched as this really make sense, especially when there is a slightly less far-
fetched possibility to consider. Even if jumps through dynamic time are impos-
sible, the same may not apply to slides.
Recall the Gold universes introduced in 4.6. I noted in 6.5 that, if we lived
in a finite and symmetrical universe, we might one day encounter or at least
find signals originating from inhabitants from the other side, who, from our
perspective, are living their lives backwards. If we assume that our universe is a
dynamic block, we naturally suppose that it is growing in the direction of our
future, and so the lives of the time- reversed beings are being built up, layer by
layer as it were, in the wrong order. But of course, from their perspective, it is
our lives that are being miraculously put together in the wrong order. Unless there
is some underlying invisible asymmetry in such a world, there will be no fact of
the matter as to which of these views is correct. Suppose that a spaceship from
the other side penetrates our space, and manages to shield itself from dangerous
interactions with our matter (which might well cause explosions of the matter/
anti- matter sort). Those on board could observe life on Earth from afar, and by
the judicious use of suspended animation, selectively visit any times in our his-
tory that they wished to see. They would, in effect, be travelling into our past.
Now for the speculative leap. Aware of the existence of these watchers, we
decide to make contact of a physical sort; we devise a way of changing our matter
into its time- reversed counterpart (perhaps by generating a field that modifies the
appropriate quantum signatures a ploy not unknown in science fiction). Being
an unusually adventurous historian, you volunteer for the process, which in due
course is carried out successfully, and join the time- reversed ship. With your new
companions, you embark on a voyage back through Earths history a history to
which you have always been a witness.
I am not claiming that a scenario of this sort is a genuine physical possibility.
But it does illustrate that backward time travel, of a sort, is at least conceivable in
worlds where time is dynamic. The same method could be employed in Presentist
worlds of the dynamic variety.
8.9 Real time
Once again our results are inconclusive. Had we found an incurable logical contra-
diction in the very idea of backward time travel, we would be in a position to rule
that any physical theory that permits such travel must be false on a priori grounds,
but since no such contradiction was found, we are not in this position. From
a purely conceptual point of view, forward time travel seems entirely unprob-
lematic, backward travel likewise, provided it is restricted in ways that prevent
the establishment of self- defeating loops; even unrestricted backward travel can
occur in universes where coinci dences are improbably common. And this is just
for a one- dimensional block-type time. In multiverses, and worlds where time is
142 TIME AND SPACE
two- dimensional, the past is changeable (after a fashion) and backward travel is
not especially problematic.
Does any of this help us in pinning down what our time is like? The logical
possibility of time travel saves the Block theorist from potential embarrassment,
but from the mere fact that time travel is possible we cannot conclude that we live
in a Block world: we have seen that certain forms of time travel are available in
dynamic universes, and it could be that time travel in our universe is nomologically
impossible. So there are two questions we need to consider. What forms of time
travel are nomologically possible in our universe? What models of time are these
modes of time travel compatible with?
In addressing the first of these questions we have no option but to take current
physics as our guide. We know that the relevant theories are not fully adequate,
but we have no clear idea as to what will replace them, and the implications of
potential successor theories for time travel are even more obscure. Moreover,
unless our current theories are completely misguided, they at least provide some
indication of what can and cannot occur in our universe. Although many physicists
doubt that our laws of nature do permit time travel and/or backward causation,
not all share this view, and there is a fair amount of consensus on how the laws
might allow these phenomena to occur. There are two main routes.
11
We have already seen how some theorists appeal to backward causation to
explain the bizarre behaviour of some quantum systems. Since the causal influence
is transmitted via quantum fields, we are dealing here with perturbations (whether
wave- like or particle- like) that travel backwards through time. The Wheeler
Feynman absorber theory of electromagnetic radiation is another example of the
same sort of phenomenon: the advanced wave that this theory postulates travels
backwards in time; likewise tachyons, hypothetical particles condemned always
to travel faster than light, and backwards in time. Although there is no evidence
that such particles exist, it is at least arguable that their existence is compatible
with current theories. For the sake of the argument, let us suppose that there are
waves or particles that travel backwards through time. Can we conclude from this
that we live in a block-universe? No. If particles could jump instantaneously from
one time to another, it is hard to see how we could resist the conclusion that the
times in question coexist. If such a particle were to emerge now, having jumped
from a future time, we would have to accept that this future time is as real as the
present. But since the phenomena in question travel continuously through time,
their existence at one time does not entail the existence of any later time. The
Growing Block theorist will regard the history of an alleged backward travelling
particle as being built up, slice by slice, in just the same way as an ordinary particle.
The physical properties of such a particle might be odd, but in itself this does not
mean that the future times to which they point are actual.
There is a second way in which backward time travel might come about, one
that is potentially more damaging from the point of view of dynamism. According
to Einsteins general theory of relativity (which we shall be looking at in more
detail in Chapters 20 and 21), the shape of the spacetime fabric is affected by the
presence of mass (or energy). There are certain spacetime structures created by
very special distributions of mass that permit backward time travel (the structure
of both space and time is mass- dependent). Some of these structures arise naturally
TIME TRAVEL 143
in universes very different from our own, for example spacetimes entirely filled
with rotating matter- fields. But there may be ways to create the conditions required
for backward time travel in universes such as our own. One of these hypothetical
mechanisms is potentially lethal to the dynamist: wormhole time machines.
A wormhole is a tunnel that makes shortcuts through space possible. Think
of two ways of getting round a mountain: the long winding road that skirts the
mountain, and the tunnel that cuts directly though it. The mountain tunnel is a
hole through a three dimensional object; it is hard to visualize a tunnel through
three dimensional space: where would the tunnel go? Tunnels between different
regions of space certainly are possible (at least logically and mathematically), but
to visualize them we would need to imagine a three- dimensional space curved back
on itself, which is something we cant do. So focus on a two- dimensional space;
think of a piece of paper folded over so that the two opposite ends nearly touch
one another. A two- dimensional being confined to the paper wouldnt realize that
one side of his world is very close to the other, and this proximity would be of no
use, since it only exists in the third dimension, to which our being has no access
(indeed, we can suppose that the envisaged two- dimensional space is not embed-
ded in a three- dimensional space). But given the shape this two- dimensional space
now has, we could easily create a shortcut by extending the space, for example
by bridging the gap with a small flap or tube of paper. Figure 8.3 illustrates a
wormhole bridge between Earth and Epsilon Eridani, with one spatial dimension
suppressed; the long way round is a distance of 10.5 light- years (or roughly
60 000 billion miles) but the wormhole route is only 4 miles.
Figure 8.3 A wormhole shortcut between two distant planets.
Wormholes of this sort can cut down travel times, but they are not time machines.
To turn a wormhole into a time machine, a temporal discrepancy between the two
mouths has to be created, so that passing through the wormhole amounts to tak-
ing a shortcut between times. We know from Einsteins special theory of relativity
that this is easily done: it would suffice to take one of the wormhole mouths on a
high- speed round- trip (cf. Thorne 1994: ch. 14). If a temporal wormhole a hun-
dred metres long connected London in 1960 with London in 2010, the fifty- year
interval could be crossed in a matter of minutes, or even seconds. By establishing a
network of such wormholes, it would be possible to travel back and forth between
distant times, and times in between. Figure 8.4 is somewhat misleading, in that the
wormholes appear no shorter than the spacetime regions that they connect; this is
an unavoidable artifact of trying to represent a variably curved four- dimensional
spacetime on a two- dimensional surface.
144 TIME AND SPACE
Since a structure of the sort depicted in Figure 8.4 could only exist in a block-
universe, should we ever find ourselves able to take this sort of shortcut we can
be certain that the times in question all exist. This form of time travel thus opens
the way to an empirical way of determining whether the past and future are as
real as the present. Unfortunately, since the jury is still out on whether the laws
of physics will permit such wormholes to exist, and even if they do there is no
guarantee that we will ever be able to create them ourselves (or stumble across
them on our travels), it seems unlikely that we will be carrying out the experiment
in the near future. If we want answers to the great conundrums posed by time, we
have no option but to employ more traditional methods of inquiry, with all their
limitations for the time being.
12
Figure 8.4 A network of wormhole
time machines.
CE 3000
CE 2000
CE 1000
9 Conceptions of void
As I mentioned at the outset, there are various metaphysical issues connected with
space: structure at the very small and very large scales, dimensionality, uniqueness,
topology and geometry. But, as with time, I am going to concentrate on a single
central topic: is the substantival view of space correct, or is the relational view
correct? Is space an object in its own right, in addition to the material bodies that
occupy it, or does it consist of a network of relations between material bodies?
This question is quite unlike that of the passage of time. We could make a good
deal of progress on the latter without considering what science has to say; purely
meta physical considerations ruled out some models of time and empirical facts
concerning experience ruled out others. The existence of space is different; purely
metaphysical considerations dont get one very far, and little that is relevant can be
gleaned from considering the structure of our experience. Scientific considerations
will thus be entering the scene very quickly. In this preliminary survey I will set the
stage by distinguishing some very general ways of thinking of space, a discussion
that will yield some general constraints on a satisfactory account of it. I will then
try to clarify the substantivalist and relationist views.
9.1 Space as void
For many of us these days, a conception of space that has considerable intuitive
appeal runs thus: space is an infinite expanse of featureless emptiness within which
physical bodies are located and move. It is not surprising that we are drawn to this
idea, for we are brought up to think of the Earth as a planet revolving around the
Sun, which is just one star out of billions, strewn through the vastness of galactic
and intergalactic space. The so- called outer space that separates the planets,
stars and galaxies is, we are told, a hard vacuum, empty save for the odd molecule,
perhaps one per cubic metre. An initial characterization of this view of space might
run along these lines:
The Void Conception: Space in itself is nothing at all; it has no intrinsic
features of its own, it is mere absence. Objects can be separated by different
spatial distances London is closer to Paris than it is to New York and
we know this because of the different amounts of time it takes to travel or
transmit signals between them; we cannot directly measure magnitudes of
space, since space is itself featureless void.
146 TIME AND SPACE
Although for most of us this naive picture is later qualified when we learn of
the vast quantities of electromagnetic radiation being pumped out by the stars,
the underlying idea of space itself as empty nothingness, devoid of intrinsic fea-
tures, not only remains in place, but retains a good deal of appeal. After all, if
we removed all particles and fields from a region of space, what would be left?
Absolutely nothing.
If we do assume that space is featureless emptiness, several further ideas can
seem compelling. Since there is always more nothingness beyond any alleged limit
to the universe, space cannot have an outermost limit, and so space surely must
be infinite in all directions. If space is infinite, there is surely just one space, for if
there were two distinct spaces, where could the other one be located? Not within
our space, clearly, but since our space has no boundary, it is hard to see that this
other space could exist outside our space. The question What structure does
space have? seems empty: if space is just nothing, how can it have a structure?
Now, although this view of space may seem quite natural, it was not always
so. It is certainly true that the equation of space with pure nothingness resembles
some of the (early) Greek atomists conception of the void within which atoms
are located. According to Torretti, the atomists void, though boundless, is not
an expanse that contains everything, but rather the nothingness outside all things
(1998, original emphasis).
1
But the void soon fell out of favour. According to the
cosmologies of Aristotle and Ptolemy, which dominated scientific thought until the
late Middle Ages, the universe is divided into two regions: the heavenly and the
sub- lunary. The heavens consist of a nested series of rotating shells, carrying the
Moon, planets and stars; the Earth is at the centre of the universe, surrounded by
the lunar sphere; the sub- lunary realm is filled throughout with matter, water, fire
and air. Since Aristotle believed the notion of a vacuum to be incoherent, he argued
that it made no sense to suppose that space continues on past the outermost sphere
of the heavens. In fact, Aristotle often talks not of space but of place, which he
defined as the inner surface of the first motionless containing body.
2
Descartes also
rejected the void, arguing that inter planetary space was filled with a subtle fluid.
Although Newton, the most influential modern thinker on these questions until
Einstein, rejected Descartess material plenum in favour of an infinite absolute
space, he was also firmly committed to a substan tival view of this space. As he
said in De Gravitatione et Aequipondio Fluidorum: space is eternal in duration
and immutable in nature Although space may be empty of body, nevertheless
it is not itself a void; and something is there, because spaces are there, although
nothing more than that (Huggett 1999: 113).
3
In rejecting the void these various thinkers were often motivated by different
considerations, which I will not enter into here (although we will be looking at
Newtons position in some detail later on). I will restrict myself to pointing out
two respects in which the void conception is problematic. The exercise is well
worth the trouble, for isolating these deficiencies will provide us with two criteria
for a satis factory account of the nature of space.
CONCEPTIONS OF VOID 147
9.2 The unseen constrainer
Suppose we could create a Flatland: a very thin two- dimensional plane inhab-
ited by two- dimensional creatures.
4
Those who like their imaginary cases fleshed
out with some harmless science- fictional detail can take it that Flatland is an
intense tightly confined electromagnetic field; the occupants of Flatland both
the sentient Flat landers themselves, and the objects in their world consist of
intricate self- organizing but potentially mobile patterns of perturbation within this
electromagnetic membrane. Current forms of so- called artificial life only exist
as computerized simulations, but it may not always be so.
The Flatlanders are, we can stipulate, ignorant of the fact that their world is sus-
pended a couple of feet above the floor of a laboratory between the field- generating
devices. They are perfectly free to move around in their flat planar world, but they
cannot leave it: they can move north, south, east and west, but not up or down.
Not only are the Flatlanders unable to move in the third dimension, but they are
wholly unaware of its existence; indeed, their language does not contain terms
for the directions that we call up and down. Their perceptual systems are
geared to their two- dimensional world (their visual experience takes the form of
a thin coloured line), and their cognitive systems are attuned to their perceptual
experience. As a consequence they literally cannot imagine spaces or objects that
have more than two dimensions; although they are fully acquainted with objects
such as squares and circles, they can no more visualize cubes or spheres than we
can envisage hypercubes or hyperspheres, the four- dimensional counter parts
of three- dimensional cubes and spheres.
We may not be able to visualize four- dimensional objects but, logically and geo-
metrically speaking, a fourth spatial dimension is perfectly possible, and mathema-
ticians are familiar with the geometrical properties of objects such as hypercubes
and hyperspheres.
5
There is thus a very real sense in which we find ourselves in
the same predicament as the Flatlanders. If our three- dimensional space existed
as a plane within a four- dimensional space, someone enclosed in a locked room
could escape without opening the door, or burrowing through the walls: they
could (in principle) be lifted upwards into the fourth dimension and then set
Figure 9.1 Two conceptions of Flatland. The Flatlanders on the left are free to rove where they like
on their plane. The Flatlanders on the right also inhabit a plane, but are further confined to moving back
and forth along their worlds surface (the thick line). Consequently, these Flatlanders would have to climb
over the tree to get past it (likewise for each other). Note also the way their internal organs are clearly
visible from the vantage point of the third dimension.
148 TIME AND SPACE
down again somewhere else in three- dimensional space, just as we could rescue a
Flatlander trapped within a square. Inhabitants of four- dimensional space could
peer into our three- dimensional plane and see our internal organs laid out before
them, just as easily as we can see the insides of the Flatlanders (see Figure 9.1),
a perturbing thought to say the least. And an utterly baffling one. The idea that
there could be another spatial axis at right angles to each of the familiar three,
and hence another two directions in which to move, is not one that we can make
any intuitive sense of; there just doesnt seem enough room in space for any other
directions. Since we cannot visualize this additional spatial dimension we can only
grasp it by analogy. Flatlanders find the idea of a third dimension hard to grasp, but
we know perfectly well that a third dimension is possible and, indeed, Flatlanders
could discover the same for themselves by considering the plight of the inhabitants
of one- dimensional Lineland and reasoning analogically.
We will be taking a closer look at Flatland later on, but this brief introduction
is all that is needed for the simple point I have in mind. If you were to look down
on the Flatlanders going about their business, you might find yourself thinking
Ignorance is bliss. How astonished and disturbed they would be if they ever dis-
covered that they were surrounded by an extra spatial dimension, that in addition
to the directions they know about north, south, east and west there is also an
up direction and a down direction! The same applies to us. But assuming that our
universe is the real thing (rather than a simulation run by four- dimensional beings
experimenting with artificial life), this is not something that we have to worry
about. There clearly are only three spatial directions, at least at the macroscopic
scale.
6
We do not encounter objects appearing as if from nowhere; objects do
not vanish into thin air. There are only three spatial axes (backwardsforwards,
leftright and updown) within which we can move.
But now suppose that the void conception of space were true. If space is simply
nothing why should our possibilities for movement be so tightly constrained? If
space is pure void, not only would we be free to wander in the additional direc-
tions made available by a fourth dimension, but there would be nothing to prevent
our wandering in the limitless different directions made available by 5- , 6- , 12- ,
101- , and n- dimensional spaces, which are just as possible, mathematically speak-
ing, as 1- , 2- and 3- dimensional spaces. The big bang would not have created a
sphere of energy expanding in just three spatial dimensions: it would have been an
explosion into n- dimensions, and as energy condensed into matter, the resulting
objects would themselves have anything up to n- dimensions. Clearly, our universe
is nothing like this. Every macroscopic object in the known universe not only
seems three- dimensional, but its possibilities for movement are similarly restricted.
But if space is simply nothingness, this would not be the case, or at least, there is
no obvious reason why it should be the case.
Of course, the fact of our confinement to three dimensions is not at all evident
to us; we naturally suppose that the familiar three dimensions are the only pos-
sible dimensions, and so we naturally suppose that nothingness itself can have
only three dimensions. But once our eyes are opened to the logical possibility of
four- , five- and higher dimensional spaces, the fact of our confinement has to be
faced, and any satisfactory theory of space has to explain these unseen constraints
in some way.
CONCEPTIONS OF VOID 149
9.3 Connection in question
You are in a dark room. You know that you are not alone, and you want to
locate the other person without making any noise. One way to do it would be to
wander about with arms outstretched until physical contact is achieved, but you
dont want to run the risk of stumbling over pieces of furniture. Happily, you are
equipped with a torch. You shine the torch in different directions, and soon locate
your companion. Facts of this sort that objects can move and collide with other
objects, that objects are connected by paths through space along which light can
travel are so familiar that we find them utterly unpuzzling; it is just the way
things are. However, this sort of fact would be puzzling if the void conception
were true. For one thing, as we have just seen, there would be many more possible
directions in which objects could be located than we usually suppose, so it might
take you a long time to point your torch in the right direction. But the problem
runs deeper: if space is mere absence, the very possibility of spatial connection
becomes problematic, and with it the notion of spatial distance, since the only
way we have of measuring expanses of void is in terms of the time taken to travel
between spatially separated objects. To bring this point clearly to the fore, I will
again make use of a thought- experiment.
We find it natural to suppose that we live in a single all- embracing spatial
framework: at any given time, any object, of any kind, no matter where it is
located, is some definite spatial distance away from every other object. The idea
that space is necessarily unified in this way can be undermined by describing an
imaginary scenario in which it would be reasonable to believe that one was living
in a two- space universe; that is a universe consisting of two collections of spatial
locations, S
1
and S
2
, such that every member of S
1
lies at some distance from every
other member of S
1
, and likewise for the members of S
2
, yet no member of S
1
lies
at any distance from any member of S
2
, and vice versa. Probably the best- known
scenario of this type is Quintons Lakeside village story (1962). What follows is
based on a variant devised by Newton- Smith (1980: ch. 4).
7
The world Pleasantville is the sole planet orbiting a solitary star located in
other wise empty space. Pleasantville is a near- perfect sphere. It is always warm
and sunny, but otherwise it is much like the present- day Earth. The inhabitants
of Pleasantville discover that eating a certain root causes them to vanish, only to
reappear some time later. When a root- eater reappears, they invariably bring back
tales of another world, Harshland, a cube- shaped planet orbiting a solitary star
located in otherwise empty space, which is always cold and inhospitable. On van-
ishing from Pleasantville, root- eaters finds themselves on the surface of Harshland,
where they amuse themselves as best they can until their return. People who eat the
root together are transported to Harshland as a group. People who eat the root
at different places on Pleasantville end up at different places on Harshland. The
various descriptions of the features of Harshland and the events that take place
there brought back by different root- eaters invariably agree.
Different interpretations of this scenario are possible. It could be argued that the
root induces intermittent existence accompanied by collective delusions, but given
the intersubjective agreement on the character of Harshland, the hypothesis that
the latter is as real and objective a world as Pleasantville seems more reasonable
150 TIME AND SPACE
(and certainly just as reasonable). Another option would be to maintain that
Harshland occupies some remote region of Pleasantville space, and that the root
induces near- instantaneous spatial relocation. But there are no stars to be seen
in the night skies of either planet. The Pleasantville folk once travelled far and
wide through their universe, but never detected any trace of any star other than
their own; they gave up looking when they discovered that light in their universe
travels instantaneously, so if there were any other stars in their universe, they
would be visible. Since the Harshland star is nowhere to be seen, the hypothesis
that Pleasant ville and Harshland each exist within a separate space is certainly a
reasonable one.
In response, it could be maintained that the two planets exist in different three-
dimensional spaces embedded in a common four- dimensional space (analogous to
the way that any number of two- dimensional spaces can be stacked like parallel
sheets of paper in a three- dimensional space). But, as Newton- Smith points out,
unless we are provided with some reason for thinking this to be the case, the
hypothesis has nothing to recommend it. If Pleasantville scientists possessed a well
confirmed physical theory that explained why it was impossible to travel from
one of these three- dimensional spaces to the other, and could provide data on the
relative locations of these spaces in the encompassing four- dimensional space, then
matters would be different; but, as it happens (we can suppose), no such theory
is forthcoming. Yet another option would be to claim that, on disappearing from
Pleasantville, the root- eaters remained within the same space, but travelled either
forwards or backwards in time, to the remote temporal location of Harshland.
But the facts count against this: the scientists of Pleasantville are able to roughly
estimate the age of both their planet and their universe. When the age of Harsh-
land is measured by the same procedures, a similar result is found. The idea that
Harshland is located within the Pleasantville universe, but at some distant point
in the past or future, can thus be ruled out; the time travel hypothesis seems quite
ad hoc. The one- time/two- space hypothesis is the simplest available.
This story shows that our experience could be such as to warrant the postula-
tion of two distinct spaces, which is an interesting result in its own right, but I
introduced the scenario simply to undermine the complacent assumption that all
objects are neces sarily spatially related. There could be two objects that are not
connected by any spatial path. Bearing this point in mind, let us return to the void
conception.
To simplify, let us suppose that extended material objects are intuitive solids;
that is, their matter completely fills their spatial volume they are not composed
of particles separated by space. Now consider this proposal:
The idea that the universe could consist of material objects dispersed through
a void at different distances from one another is simply incoherent, or at
least it is a seriously misleading way of describing a state of affairs that is
better charac terized in terms of many worlds. For consider, each solid object
constitutes a three- dimensional space in its own right. Take any two points in
an object O
1
. There is a continuous spatial path through the matter in O
1
that
connects these points. Likewise for any two points in a distant and distinct
object O
2
. But is there any spatial path between any point in O
1
and any point
CONCEPTIONS OF VOID 151
in O
2
? There is not: since (by hypothesis) these objects are separated from
one another by an expanse of void, there is absolutely nothing between them,
or that links them, so the objects are entirely unconnected. Any two material
objects are properly viewed as distinct and independent spatial worlds in
their own right. Since there are no spatial paths or distances between two
distinct objects, it is impossible for one object ever to collide with another.
If an object breaks into two parts, these parts would not drift gradually away
from one another, they would each immediately come to constitute isolated
worlds in their own right.
Once we accept the possibility of multiple unconnected spaces, this proposal is by
no means absurd. Anyone who says that the only connection between two spatially
separated objects is an expanse of void does seem vulnerable to the objection that a
connection of this sort is no connection at all, and hence that there is no reason at
all to suppose that the objects in question inhabit the same space; and if they dont
inhabit the same space, each object constitutes a space (or universe) of its own.
Once again, we know perfectly well that objects in our world do inhabit a com-
mon space: objects can collide with one another, light and other signals can pass
from one object to another, and so on. What should now be clear is that this is
by no means a trivial matter. There could easily be a collection of objects that are
neither connected nor connectable in these ways.
So we have a further requirement that a satisfactory account of space must
meet. In addition to providing an explanation of the constraints on spatially
related things, such an account must provide some explanation of the distinctive
ways that such things are connected. This is something the void conception signally
fails to deliver.
9.4 Substantivalism: a closer look
The substantivalist solves the problems of constraints and connections in a direct
and intuitively appealing way. Consider:
The Fluid World: an infinite and infinitely divisible three- dimensional liquid
plenum, a great ocean. The intelligent inhab it ants of this world are jelly-
fish; they are mostly liquid themselves, they float and swim about, feeding
on jelly- plants and they live in tunnels in jelly- planets. Since they are fully
permeated by Fluid, the jelly- fish dont notice or feel the Fluid as they move
through it. The presence of the Fluid is completely undetectable.
There is reason to suppose that in this world the Fluid is space: it is, after all, that
within which other objects are located, but which is not itself located within anything
else. No less importantly, the Fluid constrains the movements of physical objects:
the various jelly- things move by floating through the Fluid, and they are confined
to moving in three dimensions because the Fluid is itself three- dimensional. If in
some regions the Fluid is four- dimensional, the jelly- fish will have an extra spatial
dimension within which to move; if in other regions the Fluid is two- dimensional,
they will (if they can enter such a region) have only two dimensions within which
152 TIME AND SPACE
to move. If there are any empty regions to be found in the Fluid, these cavities
should not be thought of as regions of space, but rather as holes in space, and the
jelly- fish would only be able move around them, not through them. Similarly, if
the Fluid were finite in volume, rather than infinite, then space itself would be
finite and have edges beyond which no traveller could pass. As is also clear, there is
no connection problem in such a world. Objects are not separated by expanses of
nothingness. They are separated by expanses of Fluid, and so there is no difficulty
in understanding how they can move closer together or further apart. Should the
Fluid undergo fission, and become divided into two unconnected expanses, space
itself will have divided into two unconnected regions, and the occupants of each
of these regions would be cut- off from one another: there would no longer be any
spatial relations, or distances, between the two.
You might be inclined to object: You are misdescribing this world. The Fluid
cant be a space, its the wrong sort of thing: it is merely an object like any other
its simply too substantial to be a space! This objection is natural, but misguided.
The idea that space could be substantival or thing- like initially strikes many people
as very odd. Isnt saying that space is substance- like akin to saying that something
can be nothing, or nothing can be something? But this thought is rooted in the void
conception, which as we have already seen is problematic. On the assumption that
space is substantival, why does the Fluid seem too substantial to be a space? Surely,
an entity that was entirely featureless, entirely lacking in properties, wouldnt exist
at all. In which case, if space is substantival it must have a nature of some kind. If it
has a nature of some kind, why cant the Fluid be a space? I am not suggesting that
our space is akin to this hypothetical Fluid it may well be that substantival spaces
in different worlds have different intrinsic characteristics but a substantival space
must have some nature to call its own.
8
Once this point is fully taken on board, there seems no need whatsoever for a
sub stantival space to be transparent to its inhabitants, or, more generally, for it to
be bereft of detectable physical features. Imagine a variant of the jelly- world where
moving through empty space is accompanied by a resistance or drag that affects
different sorts of object to different degrees; or where motion creates slight ripples in
space that result in objects themselves undergoing small but observable movements;
or a variant where the Fluid is coloured, and is seen as such by the inhabitants.
There is a general question that any substantivalist must answer that we have
not yet considered. When a substantival space contains an object, by virtue of what
is the object located within that space? What is the relationship between a space
and an object within it? It may seem as though nothing interesting can be said here.
Isnt it obvious that objects are located at places in space, what more is there to
say? That there is something more to be said is obvious as soon as the following
possibilities are distinguished.
Relational substantivalism: space and material objects are equally basic types
of entity; there is a primitive relationship of spatial locatedness that holds
between objects and places within space.
Container substantivalism: space and material objects are equally basic types
of entity; material things are enclosed by or embedded within substantival
space, but there is space only outside and between material things.
CONCEPTIONS OF VOID 153
Super- substantivalism: space is the basic material object; other material objects
consist of this space taking on various properties. Objects are thus adjectival
on space; space is the only basic physical entity.
The space of the Fluid World as described above doesnt fall neatly into any of
these categories, but is probably closest to the third, since the jelly- things are
entirely permeated by Fluid and couldnt exist without it. If we modify the
scenario, and suppose that the only occupants of the Fluid are regions of colour,
and that these regions consist of the points within the Fluid possessing monadic
colour- properties, we have a clearer instance of super- substantivalism. By varying
the scenario again, and supposing that the occupants of the Fluid are solid steel
spheres, we get an instance of container substantivalism: the Fluid surrounds its
occupants, but does not permeate them. If, as in this case, the occupants of sub-
stantival space are themselves spatially extended (rather than punctiform), there
are really two modes of spatiality in such worlds; there is the space within which
material things exist, and there is the space constituted by material things them-
selves. Illustrations of relational substantivalism are harder to envisage clearly,
but easy enough to describe: we simply state that certain material objects bear
the location relation to certain points in space. Since these objects are not in their
space in an intuitively obvious manner, this doctrine might be thought to leave
something to be desired, but of course an advocate of this position will maintain
that it is our intuitive conception of containment that is flawed.
By way of summing up, the substantivalists position includes the following
claims:
the physical world does not consist solely of material objects such as stars and
atoms, it consists of objects of this sort together with an additional entity:
space;
this space possesses an intrinsic nature of some kind;
it possesses certain topological or geometrical properties; for example, if our
space is Euclidean it is composed of a collection (or manifold) of points that
conform to the axioms of Euclidean geometry;
material objects are related to space in one of the ways mentioned above;
material objects possess some (and probably all in the case of super-
substantivalism) of their spatial properties by virtue of being related to space;
for example, if two point- like objects are separated by a distance d it is because
they occupy points in space that are separated by d;
the ways material objects can move are constrained by space;
since objects which are located in the same space are parts of the same physical
world, the extent of our space determines the extent of our world.
9.5 Relationism: a closer look
There is no denying that a substantival space constrains and connects, but some
find the price excessively high.
The existence of the various substantival spaces that we encountered in 9.4
may have seemed unproblematic, but if so this was in large part due to the way
154 TIME AND SPACE
they were introduced into the discussion. I not only stipulated that these entities
existed, I also stipulated that they possess material features of a familiar sort and,
in doing so, I assumed an all- knowing God- like vantage point on the worlds being
described. When it comes to our world we are not able to adopt a perspective of
this kind, and there is no obvious evidence that we are immersed in a space of this
sort: hence the appeal of the void conception. Even those who believe that our
space is substantival space concede that it is not composed of the same sort of stuff
as other material things, and is not directly observable in the way other things are.
Given these facts, and in light of Occams razor, it is obviously worth investigating
whether we can get by without positing such a thing. This is precisely the task that
relationists set themselves.
Different relationists go about the task in different ways. As Earman notes
there are almost as many versions of relationism as there are relationists (1989:
12). While this is certainly true when it comes to the details, most relationists
would agree with the following:
In claiming that objects inherit their spatial properties from the regions
of space that they occupy, the substantivalist is inserting an invisible and
redundant intermediary between objects. We cannot observe space itself,
but we can observe objects at various distances from one another. The most
economical way of making sense of this is simply to say that objects are
directly related to one another by spatial relations. Instead of appealing to
spaceobjects relations, we can appeal to objectobject relations, where the
relations in question are of a spatial sort. These spatial relations should not
be thought of as material objects in their own right, but as distinctive proper-
ties, of a relational sort, that material objects can possess.
On the face of it, the relationist programme has considerable appeal. In effect, the
relationist is proposing a modified void conception: space is nothing at all, mere
absence, the world consists of nothing but material objects, but these objects have
relational properties that perform the same role as the vast entity posited by the
substantivalist. Here is how Leibniz characterizes the doctrine:
I will here show, how Men come to form to themselves the Notion of Space.
They consider that many things exist at once, and they observe in them a
certain Order of Coexistence, according to which the relation of one thing
to another is more or less simple. This Order is their Situation or Distance
Those which have such a Relation to those fixed Existents, as Others had
to them before, have now the same Place which those others had. And That
which comprehends all their Places, is called Space.
(Alexander 1956: 6970)
But there are clearly questions that need answering. What are these relations?
How exactly should we think of them? What is it, exactly, that they relate?
The obvious answer to the last question is material objects, but there are
some potential problems here. Container substantivalism is an awkward doctrine
because it posits two modes of spatiality: that of space itself, and that of the solid
material things that are immersed in space but not pervaded by it. Relationists
who want to avoid this predicament will have to extend the account they offer
CONCEPTIONS OF VOID 155
of the external spatial relations that hold between objects to the interior regions
of these objects. This can be done if extended objects are regarded as fusions of
point- particles, which themselves are spatially related. Most relationists adopt this
view, which conforms with the standard (although no longer universal) account of
elementary particles in contemporary physics.
What of the relations themselves? I will have more to say on this topic in the
next section, so for now I will restrict myself to the basics.
Generally speaking, relations and relational properties go together, in that, if
we say that X bears some relation R to Y, then X and Y can each be said to pos-
sess a relational property, either singly, as in X is R- related to Y, or jointly, as
in X and Y are R- related. There are many different relations (e.g. is larger
than , is redder than , is to the right of , is the sister of
), and different relations can have different properties. Rather than attempt
an exhaustive categorization, I will focus on the specific characteristics of spatial
relations, conceived as distances separating objects. In itself this is a considerable
simplification, since as we shall see in due course it should not be assumed that
any two spatially related items are necessarily at some determinate distance from
one another, but this complication need not concern us at this preliminary stage.
The formal properties of a spatial relation are not difficult to specify. Distance is
a concrete relation one that holds between concrete rather than abstract entities
and it is symmetrical, transitive and reflexive (an object is zero distance from itself).
Some relations are internal and others are external: roughly speaking, if X is inter-
nally related to Y, then X cannot exist without Y, whereas if X is externally related
to Y, it can. Most would agree that spatial relations are external rather than inter nal.
I may be spatially related to you, but if you were to vanish into thin air, I wouldnt
necessarily follow, and nor would my intrinsic properties undergo any change (or if
they do, it is not because I am no longer spatially related to you, but because of the
changes your disappearance makes to gravitational and other fields that affect me).
We can further distinguish between supervenient and non- supervenient rela-
tions. A relation such as is taller than is supervenient in the intended sense since
the truth of Jim is taller than Bob depends entirely on some other property
of Jim and Bob, namely their height. To express the point in a picturesque way,
in creating a universe in which Jim is taller than Bob all God had to do was to
make Jim 1.9m tall, and Bob 1.6m tall; no further property needed to be created
to make it true that Jim is taller than Bob. Supervenient properties of this sort
are sometimes called ontological free lunches, for obvious reasons. Given our
earlier discussion of the connection problem, it is clear that spatial relations are
not properties of this kind.
Suppose O
1
and O
2
are two objects, each of a certain size, mass, material con-
stitutions, and so forth, that happen to be a certain distance apart. Since it seems
perfectly conceivable for O
1
and O
2
to exist on their own as not spatially related
to one another (or anything else for that matter) it seems clear that spatial relat-
edness is not an ontological free lunch. The spatial relation that actually exists
between O
1
and O
2
required an additional creative act on Gods part, and since this
applies quite generally, spatial relations are non- supervenient, As a consequence,
relationism is by no means a cost- free doctrine, ontologically speaking: we save on
substantival space but pay extra in the coin of spatial relations.
156 TIME AND SPACE
Spatial relations are not only ingredients of reality in their own right; they
perform distinctive functions. Bearing in mind the connection problem that
we encountered in 9.3, they amount to genuine links or bonds between the
objects they relate. Not all relational properties amount to genuine connections
in this sense: has the same number of sides as is a relational property
that objects in a different universes can possess. Spatial relations are potent, since
they constrain or limit the nomologically possible movements of what they relate.
Objects in Flatland can only move in two dimensions because all available spatial
relations inhabit the same plane; we live in a three- rather than two- dimensional
world because the spatial relations in our world have an extra degree of freedom
the laws governing these relations allow them to be oriented with respect to one
another in additional directions.
A key difference between relationism and substantivalism lies in the different
stances they adopt towards empty space. The substantivalists account entails the
possibility of empty places: space is constituted of points, and since these do not
depend on material bodies for their existence they can exist when empty of matter.
Since relationists only recognize the existence of material bodies and the spatial
rela tions between material bodies, they are committed to the non- existence of
unoccupied places and regions. But while the rejection of unoccupied locations
amounts to a considerable ontological economy, it is also problematic, for it is the
substantivalists position that conforms with our usual ways of talking and think-
ing. Suppose we are in empty space; I fire a harpoon at you; I know it will hit you
because I have shot it straight at you and there is nothing between us; I know there
is a path through empty space that the harpoon can follow if there werent such a
path, how could it reach you? Or to take another example, suppose that we agree
to rendezvous next week at the mid- point between Earth and Jupiter. This point
is unoccupied. Surely the point exists. How else could we agree to meet there?
Most relationist are prepared to concede the legitimacy of these ways of talking,
but they deny that they have the ontological import that the substantivalist grants
them. The general relationist line runs something like this:
In practice we all work with more or less precise maps of space. Such a
map will typically include references to regions of empty space. But these
parts of the map do not refer to anything that exists or that is actual. A map
reflects a representation of reality that we have created; it embodies our
beliefs (theories) about the ways objects could be spatially related. Our the-
ories about the ways objects can move in relation to one another tell us that
objects could occupy a place mid- way between Jupiter and Earth, so we put
this place on our map, but there is no such place unless an object is actu-
ally at these distances from Earth and Jupiter. As for distances themselves, it
may be natural to think of these as quantities of space, but this is a mistake.
Since empty space isnt an ingredient of reality, there cannot be quantities
of it. We can, of course, measure distances between objects, for instance
by taking a ruler and moving it around, or sending signals (such as light,
sound, carrier pigeons) back and forth, and measuring the time taken. The
success of these procedures can be explained in wholly relational terms: no
reference to empty regions of space is required. Not even the substantivalist
CONCEPTIONS OF VOID 157
says that we can measure our space directly, since our space is invisible and
intangible.
This position is not obviously incoherent and is certainly worthy of serious
consideration. All that exists are objects and their actual spatial relations; since
objects can change their spatial relations to one another in certain restricted
ways, we can draw up (or imagine) maps (or representa tions) that reflect these
possibilities, but it remains the case that at any given time only objects and their
actual distance relations exist. All remaining points on the map do not corre-
spond to anything real.
9
There is more to be said about these relations, but before moving on to this
topic there is one further point that is worth clarifying. I noted at the outset
that although relationists reject the idea that space is an entity in its own right
they do not claim that all propositions about the sizes of objects or the distances
between them are false. The key relationist claim is that spatial facts of this sort
do not require or involve a spatial substance. The position of the relationist thus
construed can usefully be contrasted with that of the radical spatial anti- realist. I
have in mind idealists who deny the reality of the physical world: if reality consists
only of immaterial minds (as Berkeley held), then clearly there are no material
objects and no spatial relations between them. Nonetheless, there are idealists
who maintain that there are spatial facts (even though there is no space) just as
there are facts about physical objects (even though there are no such objects or
so they believe). How is the trick worked? Well, the idealists in question want to
leave our ordinary ways of talking and describing the world intact. They do not
advocate adopting a completely new vocabulary. So if I say that theres a teacup
on my table (true), they would allow that this proposition is true, but maintain
that what makes it true is not a fact about material objects (cups, tables) but rather
patterns in sensory experience. They would adopt a similar stance towards proposi-
tions about spatial relations.
We clearly need to distinguish two sorts of relationism. Relationists of both
sorts accept that there are true statements about spatial relations, and hence spatial
facts. The difference is this:
Realist relationists: true statements about space are made true by facts about
material bodies and the way they are related.
Idealist relationists: true statements about space are made true by facts about
human minds and/or patterns of sensory experience.
In what follows, unless I state otherwise I will be concerned with realist relationists
people who assume that the physical world is real. The distinction between the
two forms of relationism is, however, worth noting at a preliminary stage because
of one central figure in the debate: Leibniz, the founding father of relationism.
Leibnizs various writings on space can be confusing. Although he was himself an
idealist, and some of his pronouncements concerning space are clearly made from
an idealist standpoint, many others are not. When arguing against the Newtonian
conception of (substantival) space, he often seems to be assuming a realist position,
at least for the sake of argument. We will be concerned only with the Leibnizian
arguments of the latter sort.
158 TIME AND SPACE
9.6 Two concepts of distance
The relationists rejection of the reality of unoccupied locations does bring into
focus an important aspect of the way they conceive spatial relations: these rela-
tions operate, as it were, across space (or void) but not through it. Imagine building
a model of the solar system using plastic balls of different sizes and lengths of stiff
wire. Your planets are all connected to the Sun at the centre of the system
by thin wires extending through empty space. Relationists do not (usually) take
spatial relations to be entities akin to these wires (not surprisingly, since there is
no empirical evidence for the existence of anything resembling them). The envis-
aged relations are held to connect spatially separated objects directly, without
passing (or extending) through the medium of intervening empty space (which, of
course, relationists reject). These relations are more akin to action- at- a- distance
forces than the material connections with which we are familiar in everyday life.
However, the force analogy should not be stretched too far; spatial relations do
not attract or repel, and standard scientific theories do not posit anything akin to
a spatial force. So far as most relationists are concerned, the connecting relations
they posit are among the basic ingredients of reality, ingredients we are obliged
to recognize in order to explain the world as we find it in the most economical
fashion.
Connections of this sort are certainly not easy to imagine, but does this mean
that they cannot exist? Nerlich suggests as much:
One reason for taking space as a real thing is the strongly intuitive belief
that there can be no basic, simple, binary spatial relations Consider the
familiar (though not quite basic) relation x is at a distance from y. There is
a strong and familiar intuition that this can be satisfied by a pair of objects
only if they are connected by a path. Equivalently, if one thing is at a distance
from another then there is somewhere half way between them. Distances are
infinitely divisible, whether the intervening distance is physically occupied
or whether the space is empty. (1994a: 19)
If Nerlich is right, relationism simply doesnt make sense. But is he right? We
clearly need to take a closer look at the notion in question: spatial distance.
Useful light is shed on this whole issue by a distinction explored in Bricker
(1993). We start off with a familiar type of space, a Euclidean plane, which we
construe in a substantival way: the space is full of points. Two of these points, P
and Q, are 20m apart (see Figure 9.2a). We now perform some spatial surgery. In
the first case, a circular region of 10m diameter is removed (Figure 9.2b); in the
second case, an infinitely long segment 10m wide is cut out of the space (the points
within the region are removed) (Figure 9.2c).
Q P Q P Q P
Figure 9.2 Spatial surgery. Deleting points leaves holes in space but is the distance between P and
Q altered as a result? The answer depends on the conception of distance we adopt.
(a) (b) (c)
CONCEPTIONS OF VOID 159
The question that can now be posed for each post- surgery case is how far apart
are P and Q? Someone might answer, In (c) they no longer stand in any distance
relation; in (b) they are more than twenty metres apart. Someone else might
answer, In both cases P and Q are still exactly twenty metres apart. Each answer
reflects a different conception of distance.
The first answer would be given by anyone who thinks that distance is shortest
path length, where the path is a continuous route through the points in a space.
Imagine a Flatland geometer painstakingly measuring off distances between points
using a long piece of string and a ruler. He aligns his string along every continuous
path between the relevant points, measures its length, and declares the shortest
path to be the distance between them (in the case of a tie, the distance is that
of the shortest paths). The seventeenth- century mathematician Gauss envisaged
a procedure of this sort in his work on curved surfaces, which is why Bricker
calls this the Gaussian conception of distance. Since in (c) there is no continuous
path between P and Q the deleted region extends infinitely in either direction
there is no (Gaussian) distance between the points. In (b), the shortest path is
shown in Figure 9.3, and it clearly adds up to more than 20m. The exact distance
is 10\3 + 5t/3

m.
P Q
Figure 9.3 The intrinsic distance between P and
Q is the length of the straight line (passing through
the centre of the hole) connecting them; the
Gaussian distance, indicated by the single-headed
arrows, is rather longer.
Gaussian paths are themselves parts of space, consisting as they do of collections
of points. On this conception, paths are the primary bearers of distance; it is only
after lengths have been assigned to the paths that connect them that distances
can be assigned to points. This conception of distance lives on: it is the basis of
differential geometry, the formalism used in contemporary spacetime theories.
The person who answers Its still twenty metres for both (b) and (c) clearly
isnt working with the Gaussian conception of distance. They are working with
what Bricker calls the intrinsic conception of distance: the distance between P and
Q (and any other two points) depends solely on the properties of the points in
question. For P and Q to remain 20m apart despite the removal of parts of the
intervening space entails that the distance between these points depends on noth-
ing but the points themselves and how they are directly related to one another.
This direct intrinsic distance is independent of the space within which P and Q
are embedded: all the points surrounding P and Q could be removed and they
would still be 20m apart. Like its Gaussian counterpart, the intrinsic conception
also lives on:
The mathematical embodiment of the intrinsic conception is the abstract
structure of a metric space. A metric space consists of a non- empty universe
of points together with a family of distance relations (or a single distance
160 TIME AND SPACE
function it matters not) satisfying the axioms of distance The distance
relations are taken as primitive, and other features of space e.g. topologi-
cal are defined in terms of the distance relation. (Bricker 1993: 278)
The fact that the intrinsic conception has a home in mathematics does not,
in itself, mean that it has a physical application. A proponent of the Gaussian
conception, such as Nerlich evidently is, might argue: But it makes no sense to
suppose that P and Q are still twenty metres apart, especially in (c), where the
radical spatial surgery left two disconnected spaces. How can P and Q be at a
distance from one another if they no longer inhabit the same space? But while
this objection has some intuitive force, it is by no means overwhelming. Someone
might reply, Yes, there is no continuous path linking P and Q, but it remains the
case that they are twenty metres apart. This is a fact about P and Q themselves.
Why should it be affected by what happens to other parts of space? Not only
does this response also have some intuitive force I think we can make sense of
it at a gut- level, for what thats worth but it can be reinforced by imaginary
tales. For instance, it is easy to envisage a communications system that continues
to function despite the spatial deletion (it doesnt work by sending signals along
paths through space), and so not only can messages still be sent from P to Q,
but the time delay is unchanged it remains consistent with P and Q being 20m
apart a fact that suggests that the relevant laws of nature are sensitive to intrinsic
rather than Gaussian distance.
Bricker starts by assuming that points and distances are real, and then poses
the question as to how these points are interwoven to form the fabric of space:
Are there direct ties only between neighbouring points, so that points at a dis-
tance are connected only indirectly through series of such direct ties? Or are
there also direct ties between distant points, so that the fabric is reinforced, as
it were, by irreducibly global spatial relations? (1993: 271). If the latter, then
space retains its overall shape even when sliced into two dis connected pieces,
as in Figure 9.2c. This point- realism amounts to spatial substanti val ism, for, as
we have seen, the relationist denies the reality of unoccupied points. However,
although the concept of intrinsic distance can be applied to substantival spaces, is
there any reason why it cannot be applied to relational spaces? There is no obvi-
ous reason why not, and indeed, the notion is just what the relationist requires.
From the Gaussian perspective, the individual material bodies scattered through
the void are each island universes; since they are not linked by continuous paths
through space (unoccupied points do not exist), they are not spatially related,
and so at no distance from one another. By introducing intrinsic distances into
the picture a unified cosmos is restored. Nerlich might object that the notion of
intrinsic distance simply doesnt make sense in a relational context, but it is hard
to see why this should be so. If we can make sense of points in a substantival
space being linked by intrinsic distances and there is reason to think that we
can given that these distances are unaffected by the presence or absence of
intervening points, we can surely make sense of points (or objects) in a relational
space being similarly linked.
CONCEPTIONS OF VOID 161
9.7 Two conceptions of motion
I suggested earlier that in deciding between the relationist and substantivalist
accounts, empirical or scientific issues play a crucial role, and we can now see why.
Although relationism and substantivalism amount to very different conceptions of
reality, both seem metaphysically coherent, at least when considered on their own
terms. It may well be that there are logically possible spatial worlds where space
is substantival and distances Gaussian, and others where space is relational and
distances intrinsic. If so, then in trying to decide whether our space is substantival
or not we cannot confine ourselves to a purely metaphysical inquiry. We must
consider this question: is there some observable feature of our world that requires,
or is best explained in terms of, a substantival space?
One crucial scientific issue, which will loom large over the next few chapters,
concerns the dynamics of moving bodies (the science of forced motion). Without
entering into any details it is important to be clear right at the outset that the
different accounts of the nature of space entail very different accounts of what
motion itself involves.
The substantivalist says: a body moves if it occupies different locations in
substantival space at different times.
The relationist says: a body moves if its distance relations with other material
bodies change.
For the substantivalist, at least some motions are absolute in this sense: the bodies
in question occupy different locations in substantival space at different times. For
the relation ist, there is no such thing as absolute motion in this sense, since the
supposed substantival space does not exist. Consequently, all motion is relative to
other material bodies.
Although standard, this terminology is potentially confusing. You might think,
The substantivalist is also claiming that motion is relative: a body moves when it
changes its location relative to substantival space. You would be perfectly correct.
For under standable historical reasons, a tradition has developed of calling motion
with respect to substantival space absolute. It is also true that the substantivalist
recognizes that bodies move relative to one another (e.g. the distance between
bodies A and B increases/decreases). The key difference is that the substantival-
ist recognizes a sort of motion that the relationist denies: movement relative to
substantival space.
The implications of this difference are far- reaching. For now note just one
point. Suppose that there are two rocket ships S and T close to one another but
millions of light- years from the nearest star; the two ships are moving apart at
10mph. Since no stars are visible, it is impossible to tell how fast either ship is
moving relative to other bodies. The captain on S might say, We are stationary
and T is moving away from us at 10mph. The captain on T might say, We are
stationary and S is moving away from us at 10mph. Who is right? No test will
reveal the right answer. Indeed, there are plenty of other possible answers; both
ships could be moving apart at 5mph. The different accounts of the nature of
space have different implications about what is really happening in situations of
this sort.
162 TIME AND SPACE
Substantivalists will say, There is a fact of the matter about the real motions of
the two ships, namely how they are moving (if at all) relative to substantival space,
even if we dont know what these real motions are. What can relationists say?
One thing they cant say is that there is a fact about how fast either ship is moving
relative to substantival space. So they must say that the only facts in this situation
are the speed at which the ships are moving apart relative to one another, and the
speed at which the ships are moving towards/away from other material objects in
the universe. If we now suppose that all the other objects in the universe vanish from
existence, leaving just our two ships, what can the relationist say? Clearly, there is
now just one fact: the ships are moving apart from one another at 10mph. For the
substantivalist the presence or absence of other objects is irrelevant, since each ship
has an absolute velocity determined by its motion relative to substantival space.
The relationists conception of motion can seem odd. It is easy to feel that there
must be a right answer to the question How fast is ship S really moving? But
then again, it can also seem quite plausible. Who, when travelling in a fast car or
plane, hasnt had the feeling Its hard to believe that I am actually moving! If we
think of things as the relationist does then, when we are in motion relative to the
Earth, it is just as true to say that the Earth is moving at several hundred miles per
hour and we are stationary as it is to say the reverse, which is just how it seems.
9.8 Matters terminological
In On the Existence of Time, Space and Space- time (1978), Horwich notes that
terms such as absolute, relative and relational are used in connection with
space (and time) to mean very different things. Horwich is certainly right about
different writers using the same terms to refer to different doctrines and combina-
tions of doctrines, and this can be confusing. It helps to have some idea of what
the various doctrines are. Horwich lists the following relativist theses, each of
which is denied by a corresponding absolutist thesis.
1. Space and time depend for their existence upon the existence of objects and
events.
2. Space and time are not substances.
3. The spatial and temporal location of an object or of an event is to be analysed
in terms of spatial and temporal relations between those things and other
objects and events.
4. All motion is relative motion: an object moves only in virtue of changes in its
distance from other objects.
5. Certain features of space and time, such as geometrical properties and congru-
ence relations between intervals, must be defined in terms of the behaviour of
objects and events. (This is the thesis of the intrinsic metrical amorphousness
of space, maintained by Adolf Grnbaum.)
6. All spatial and temporal facts are to be analysed in terms of non- spatial and
non- temporal concepts. (For example, according to the so- called Causal
theory of time, the notion of temporal betweeness may be reduced to vari-
ous causal relations between events.)
CONCEPTIONS OF VOID 163
7. Space and time are relative in the sense that certain magnitudes such as dura-
tion and distance vary from one frame of reference to another. (This is implied
by Einsteins special theory of relativity.)
Many of the claims made in this list have cropped up in earlier chapters, even if the
labels used were different. In this chapter we have mostly been concerned with 2,
and I have eschewed absolutism and relativism in favour of substantivalism
and relationism (except when discussing 4), but beware: the latter pair of terms,
too, have sometimes been used in connection with most of the other doctrines
listed above.
10 Space: the
classical debate
10.1 The last of the magicians
Over the next three chapters we will be examining the debates concerning motion
and space in the context of classical Newtonian mechanics. Newton himself will
be playing the role of arch- substantivalist and Leibniz the arch- relationist, with
Galileo and Descartes fitting uneasily between the two. As far as our space is
concerned, this debate has in some respects been rendered redundant by subse-
quent developments, since Newtons physics has been replaced by Einsteins, but
the earlier debate is well worth studying. Not only does it provide useful prepara-
tion recent developments can only be fully understood against the backdrop of
the positions from which they emerged but it is also of interest in its own right,
for in considering the different ways that the substantivalist and relationist can
account for a basic physical phenomenon such as motion, we learn a good deal
about both the character and explanatory resources of the two competing frame-
works, and what we learn may well be relevant outside the narrow confines of the
classical Newtonian worldview.
Newtons Principia was first published in 1687 (Philosophiae Naturalis Prin-
cipia Mathematica), and in the opinion of many it is the single most important
work in the history of science. Within half a century most scientists were Newto-
nians, and most were to remain so for the next two hundred years.
1
Building on
the various develop ments in the understanding of motion and forces made during
the previous centuries, Newton not only developed a comprehensive theory of
mechanics (one that is still in use today), but he also devised his theory of universal
gravitation and the mathematics of the calculus, which allowed him to predict the
movements of the heavenly bodies with great accuracy, explain why planetary
orbits were elliptical rather than circular, and much else besides. As Barbour puts
it, So comprehensive was his genius, it appeared to open all doors into nature, to
leave nothing really major to discover. Life after Newton seemed a mere walking
through the garden into which his genius had directed us (1989: 629).
Although the importance of the work was immediately recognized, it also gave
rise to considerable controversy. In particular, contemporaries such as Huygens
and Leibniz were highly critical of the cornerstone of the Newtonian edifice:
absolute substantival space. In the famous Scholium following definition eight, at
the very start of the Principia, Newton lays bare his commitments on this score:
2
Hitherto I have laid down the definitions of such words as are less known,
and explained the sense in which I would have them understood in the
SPACE: THE CLASSICAL DEBATE 165
following discourse. I do not define time, space, place and motion, as being
well known to all. Only I must observe, that the common people conceive
these quantities under no other notions but from the relation they bear to
sensible objects. And thence arise certain prejudices, for the removing of
which it will be convenient to distinguish them into absolute and relative,
true and apparent, mathematical and common.
Absolute, true and mathematical time, of itself, and from its own nature,
flows equably without relation to anything external
Absolute space, in its own nature, without relation to anything exter-
nal, remains always similar and immovable. Relative space is some movable
dimen sion or measure of the absolute spaces; which our senses determine by
its position to bodies, and which is commonly taken for immovable space;
such is the dimension of a subterraneous, an ariel, or celestial space, deter-
mined by its position relative to the Earth. Absolute and relative space are
the same in figure and magnitude; but they do not remain always numerically
the same. For if the Earth, for instance, moves, a space of our air, which
relatively and in respect of the Earth, remains always the same, will at one
time be one part of the absolute space into which the air passes; at another
time it will be another part of the same, and so, absolutely understood, it
will be continually changed.
Absolute motion is the translation of a body from one absolute place into
another. Thus in a ship under sail, the relative place of a body is that part of
the ship which the body possesses; of that part of the cavity which the body
fills, and which therefore moves together with the ship: and relative rest is
the continu ance of the body in the same part of the ship, or of its cavity. But
real, absolute rest, is the continuance of the body in the same part of that
immovable space, in which the ship itself, its cavity, and all that it contains
is moved. (Huggett 1999: 11819)
While conceding that we are perfectly familiar with space, time and motion,
Newton points out that for the most part in ordinary life we deal only with rela-
tive locations and motions; that is, with how material bodies are spatially related
to one another. When on land, we typically assume that the Earth is motionless
and so place it at the centre of our frame of reference; we generally take a things
speed to be how fast it is moving with respect to the Earths surface. (Imagine the
reaction to this line of defence from someone caught speeding: But your Honour,
I was only doing 50 with respect to the Moon!) When on board a ship far from
land, we might well take the ship to provide our stationary point of reference, and
so find ourselves saying of a lazy crewman, He hasnt moved an inch all day,
even if, relative to the Earth- based frame of reference, he has moved many miles.
This choice of object- based reference frame is entirely a matter of convenience:
people living on other worlds might well take their planet (or star) to be stationary,
and their choice would be just as well founded as ours.
Newton does not criticize these practices, but he does warn against the preju-
dice of assuming that all frames of reference are on an equal footing, and hence
that all motion is relative. He insists that there is a privileged frame of reference,
one that is not arbitrary, one that provides a measure of real motion the frame
166 TIME AND SPACE
constituted by absolute space. As Newton makes clear elsewhere, he takes absolute
space to be infinite, perfectly uniform and Euclidean in geometrical structure.
Although invisible, it is perfectly real, and an entity in its own right, as this extract
from the earlier De Gravitatione makes clear:
[H]ence it is not an accident. And much less may it be said to be nothing,
since it is rather something, than an accident, and approaches more nearly
to the nature of substance. There is no idea of nothing, nor has nothing any
properties, but we have an exceptionally clear idea of extension, abstract-
ing the dispositions and properties of a body so that there remains only the
uniform and unlimited stretching out of space in length, breadth and depth.
And furthermore, many of its properties are associated with this idea In
all directions, space can be distinguished into parts whose common limits
we usually call surfaces; and these surfaces can be distinguished in all direc-
tions into parts whose common limits we usually call lines; and again these
lines can be distinguished in all directions into parts which we call points.
Furthermore spaces are everywhere contiguous to spaces, and extension is
everywhere placed next to extension And hence there are everywhere all
kinds of figures, everywhere spheres, cubes, triangles, straight lines, every-
where circular, elliptical, parabolical and all other kinds of figures, and those
of all shapes and sizes, even though they are not disclosed to sight. For the
material delineation of any figure is not a new production of that figure with
respect to space, but only a corporeal representation of it, so that what was
formerly insensible in space now appears to the senses to exist. In the same
way we see no material shapes in clear water, yet there are many in it which
merely introducing some colour into its parts will cause to appear in many
ways. However, if the colour were introduced, it would not constitute mate-
rial shapes but only cause them to be visible. (Huggett 1999: 11112)
It was this doctrine that Leibniz took as his target:
These gentlemen maintain, therefore, that space is a real absolute being
As for my own opinion, I have said more than once, that I hold space to be
something merely relative, as time is; that I hold it to be an order of coexist-
ences, as time is an order of successions I have many demonstrations, to
confute the fancy of those who take space to be a substance, or at least an
absolute being. (Alexander 1956: 256)
The debate between relationists and substantivalists was thus joined.
Newtons Principia may have been the defining moment in the development
of modern science, but Newton himself was fully a man of his time, and as such
was motivated by concerns alien to those that we now tend to associate with the
scientist. Far from devoting himself entirely to physics, he was also deeply religious
and as a committed (although secret) Unitarian, he laboured long on the project
of demon strating that the doctrine of the trinity was an invention of the early
church fathers. Combining work in physics and theology was not uncommon at
the time Leibniz did the same but Newtons interests were particularly broad,
extending as they did far into the arena of the esoteric: he was a devotee of both
alchemy and numerology. Not for nothing did Keynes describe him as the last of
SPACE: THE CLASSICAL DEBATE 167
the magicians. These broader interests almost certainly exerted an influence on
Newtons thinking about space; his work in physics was, in part at least, inspired
by a mystical- cum- theological vision of absolute space as a divine attribute. As
Barbour aptly says, Newton is explicitly seeking to demonstrate, through phe-
nomena, the transcendent basis of all motion. But he aims even higher; the boy
from Grantham has set his eye on the anatomy of God (1989: 628).
3
This may well have been Newtons ultimate goal, but there is little in the first
edition of the Principia to suggest as much: Newtons many arguments including
those concerning absolute space are entirely scientific in character. Only in one
remarkable sentence does Newton explicitly reveal the profound importance that
the doctrine of absolute space had for him. The final paragraph of the Scholium
terminates thus:
It is indeed a matter of great difficulty to discover, and effectually to dis-
tinguish, the true motions of particular bodies from the apparent; because
the parts of immovable space, in which these motions are performed, do by
no means come under the observation of our senses. But how are we to
obtain the true motions from their causes, effects, and apparent differences,
and the converse, shall be explained more at large in the following treatise.
For to this end it was that I composed it. (Huggett 1999: 1234)
The Principia as a whole was intended to show that the distinction between real
(absolute) and relative motions is indispensable in an adequate account of the
observed behaviour of material bodies. If discerning the anatomy of God was
Newtons ultimate goal he set about reaching it in an entirely modern way, with
no appeal to wizardry or revelation; he firmly believed the existence of absolute
space could be established on empirical grounds alone.
4
Evaluating the strength of Newtons case is by no means an easy matter, and
I will approach the topic from different directions. In this chapter I set the stage
by taking a brief look at the relevant doctrines of Galileo and Descartes (the lat-
ter was almost certainly Newtons target in the Scholium), and then move on to
Leibnizs attacks on substantivalism. In Chapter 11 we encounter the considera-
tions that Newton himself believed were decisive, and survey the main relationalist
responses to Newton. In Chapter 12, I consider in some detail the impact on the
classical debate of the contem porary concept of spacetime. While anachronistic,
this broadening of scope could also be regarded as perfectly justifiable in an inves-
tigation of a broadly philosophical sort, since to arrive at the four- dimensional
spacetime way of conceiving the material universe requires only a conceptual
revision, which Newton or Leibniz could easily have arrived at by thinking alone
no empirical discoveries are needed. Unfortunately, while the four- dimensional
perspective sheds useful new light on the whole debate, it also brings problems
and difficulties of its own, as we shall see.
10.2 Galileo
The pivotal role that Galileo played in the development of the science of motion is
well known and well described elsewhere, as are his troubles with the Inquisition,
168 TIME AND SPACE
so I will dwell only on those of his doctrines that are especially relevant to our
concerns. In his Dialogue Concerning the Two Chief World Systems, Ptolemaic and
Copernican (1632) Galileo set out to prove that the Earth moves; specifically, that
it rotates on its own axis once per day, and rotates around the Sun once per year.
This hypothesis was by no means novel it had advocates in ancient Greek times
and although interest in it had increased since the publication of Copernicus s
De Revolutionibus in 1543, there were few who believed the hypothesis could be
true, and fewer still prepared to risk ridicule (and the condemnation of the church
authorities) by defending it in public. Religious considerations aside, Galileo had
to overcome objections from two sources: common sense and Aristotelian physics,
which to some extent acted to reinforce one another.
The objection from common sense is simple: the Earth must be stationary
because if it were moving there would be all manner of dramatic observable con-
sequences, none of which are actually observed. Take some mud, squeeze it into
a firm lump and set it down on a table: the lump remains intact. But if you attach
the same lump to a piece of string and whirl it around your head, the lump would
quickly fly apart. Wouldnt the same happen to the Earth if it were flying around
the Sun? If you put a cup on top of a book and snatch the book away the cup is left
behind, stranded in mid- air, just as we would be if the Earth were whirling through
space at thousands of miles per hour. We could also expect a powerful wind (move
your hand through the air). Birds would not be seen flying in all directions with
equal ease; they would have to fly extra hard to catch up with the moving Earth.
Objects dropped from towers wouldnt land directly below, at the base, but many
miles away. There is no need to continue: the case seems secured.
These powerful intuitions are confirmed and enshrined in Aristotles physics,
where every kind of thing has its own natural place, its own natural motion,
and things that are removed from their natural place strive to return there. Celestial
objects (e.g. the heavenly spheres carrying the planets) are not subject to change or
decay, and their natural motion is circular, which explains why they are observed
to move around the Earth at constant speeds; they dont need anything to keep
them going since by their very nature they move in circles. The essential natures of
the four corruptible non- heavenly elements earth, water, air, fire are defined
in terms of their tendencies to move towards their natural places. The natural
motion of heavy things (earth, water) is downwards, towards the centre of the uni-
verse; that of light things (air, fire) is upwards, away from the centre. The heaviest
things are those that sink below those bodies whose natural motion is downwards,
whereas the lightest things are those that rise above those bodies whose natural
motion is upwards. It immediately follows that if the Earth is already at the centre
of the universe it would not move away from it; and since objects when dropped
are always observed to move straight down, even at different locations on the
Earths surface, we can be sure that the Earth is already where it belongs. Aristotle
did, of course, recognize that material objects can move in directions other than
downwards stones can be thrown upwards, lumps of wood can be set in motion
across ponds but in all such cases there is a force applied, and objects will return
to their natural state of rest, in their natural place (or as close to it as they can
reach), unless a force is continually applied. Why does an arrow keep moving
forwards when it leaves the bow? Shouldnt it fall straight down as soon as it loses
SPACE: THE CLASSICAL DEBATE 169
contact with the string? Aristotle was aware of the problem and suggested some
possible solutions. Either the bow string causes a mobile agitation in the air, which
keeps pushing the arrow forwards, or the arrows movement causes air to loop
towards the rear and push it forwards. These Aristotelian arguments may not fully
convince, but the underlying principle that moving objects naturally return to a
state of rest in the absence of a continually applied force does conform to what
we observe in daily life.
In arguing that the Earth goes round the Sun, Galileo did not find it necessary to
abandon every element of Aristotelian physics and cosmology. Far from rejecting
the idea of natural circular motions, he extended it. He followed Copernicus in
arguing that since the planets (of which the Earth is one) are all spherical bodies,
they each possess their own natural circular motions around their own axis and
around the Sun and moreover, that each planet determines its own up and
down away from and towards its own centre. Consequently, a ball dropped on
Earth naturally moves towards the centre of the Earth, whereas a ball dropped on
Mars would naturally move towards the centre of Mars, rather than towards the
centre of the universe, which is occupied by the Sun. As for the alleged observable
effects of Earthly motion, he deployed numerous brilliantly contrived examples,
such as the following:
For a final indication of the nullity of the experiments brought forth, this
seems to me the place to show you a way to test them all very easily. Shut
yourself up with some friend in the main cabin below decks on some large
ship, and have with you there some flies, butterflies, and other small flying
animals. Have a large bowl of water with some fish in it; hang up a bottle that
empties, drop by drop into a wide vessel beneath it. With the ship standing
still, observe carefully how the little animals fly with equal speed to all sides
of the cabin. The fish swim indifferently in all directions; the drops fall into
the vessel beneath; and, in throwing something to your friend, you need
throw it no more strongly in one direction than another, the distances being
equal; jumping with your feet together, you pass equal spaces in every direc-
tion. When you observe all these things carefully (though there is no doubt
that when the ship is standing still everything must happen in this way),
have the ship proceed with any speed that you like, so long as the motion is
uniform and not fluctuating this way and that. You will discover not the least
change in all the effects named, nor could you tell from any of them whether
the ship was moving or standing still. In jumping, you will pass on the floor
the same spaces as before, nor will you make larger jumps toward the stern
than toward the prow even though the ship is moving quite rapidly, despite
the fact that during the time that you are in the air the floor under you will
be going in a direction opposite to your jump. The droplets will fall as
before into the vessel beneath without dropping toward the stern, although
while the drops are in the air the ship runs many spans. The fish in the water
will swim towards the front of their bowl with no more effort than toward
the back the butterflies and flies will continue their flight indifferently
toward every side, nor will it ever happen that they are concen trated toward
the stern, as if tired out from keeping up with the course of the ship, from
170 TIME AND SPACE
which they have been separated during long intervals by keeping themselves
in the air. And if smoke is made by burning some incense, it will be seen going
up in the form of a little cloud, remaining still and moving no more toward
one side than the other. The cause of all these correspondences of effects is
the fact that the ships motion is common to all the things contained in it,
and to the air also. (Barbour 1989: 3956)
There is nothing in the least surprising here for anyone who has spent hours in a
commercial airliner and watched normal life unfold around them while moving at
many hundreds of miles per hour, but of course no one in Galileos day had the ben-
efit of such experiences, and the obvious had to be painstakingly pointed out and
argued for.
5
When it is, the message sings through loud and clear. No observations
that we can make within a ships cabin will enable us to determine whether the ship
is stationary, or moving at a constant speed in a straight line (on a smooth sea). If
the ship is moving, everything in the cabin falling drops of water, jumping people,
floating smoke, flying insects, particles of air will have the same forward motion
as the ship itself, and so nothing is left behind as the ship moves smoothly on. Since
both the ship and its contents share in the same forward movement, in considering
the movements of objects within the ship this movement can be subtracted and
ignored. What applies to the moving ship applies to the moving Earth.
Galileos case did not rest solely on analogies of this sort, persuasive though
they are. Rather, he used the analogies to illustrate and confirm basic principles of
motion for which he provided independent arguments. The ships cabin example
illustrates what we now call the principle of Galilean invariance, which for-
mulated crudely says that the laws of motion do not distinguish between (and so
remain the same within) frames of reference that are moving at uniform speeds in
straight lines with respect to one another (we will be returning to this notion). Of
Galileos various discoveries, two are of particular importance: the principle of
inertia (albeit in a restricted form) and his account of projectile motion.
According to Aristotle an arrow will keep moving only as long as it is being
pushed; stop pushing and it will fall straight to the ground as dictated by its natu-
ral motion and the same applies to all material things. Galileo rejected this, and
made the startling claim which we now know to be true that once an object is
set in motion it will continue moving for ever, without slowing down and without
the need for additional pushes, unless acted upon by external influences, such
as gravity and friction. That it resisted discovery for so long is powerful testimony
to the unobvious nature of this principle. Since everything we ever observe going
up soon comes down, it is very hard to accept that it is just as natural for a ball to
move upwards as downwards. Once the principle has been stated and absorbed,
the ships cabin argument seems almost obvious: the objects that share the ships
motion those that are moving in the same direction and the same speed will
continue to do so unless something impedes them. Newton made the principle of
inertia the first of his three laws of motion:
Law I Every body continues in its state of rest, or in uniform motion in a straight
line, unless it is acted upon by an external force.
Law II The change of motion (acceleration) is proportional to the motive force
impressed; and is made in the direction of the force.
SPACE: THE CLASSICAL DEBATE 171
Law III To every action there is an equal and opposite reaction; or, the mutual
actions of two bodies upon each other are always equal, and directed to
contrary parts.
What is striking to the modern (post- Newtonian) mind is the fact is that although
Galileo had clearly grasped the essential point namely that motion is as natural
a condition for material objects as rest he did not state the principle in its New-
tonian form. Rather, he endorsed a principle of circular inertia: all Earthly objects
have a natural tendency to move in circles, just as the Earth does.
Galileos reasoning ran like this. Imagine a long, smooth plane that first slopes
gradually down, then gradually up, and a ball rolling down it, without any impedi-
ment from air or surface friction. As the ball rolls down it first gathers speed, but
then, as the plane inclines upwards, it gradually slows. Now imagine a variant.
The ball has a certain speed, but the plane does not rise, it remains horizontal;
why should the ball ever come to a stop, given the lack of impediments? Wont it
go on rolling at the same speed for ever? Galileo thus drew the conclusion that a
ball rolling along a horizontal plane would continue moving at the same speed for
ever, unless impeded by some external force, such as friction. He didnt, however,
take the horizontal plane to be a straight line, but rather the spherical surface
of the Earth. For his purposes this was all that was required: it is because a ball
dropped from a tower possesses a natural tendency for circular motion that it fol-
lows the Earths movement, and falls at the base of the tower. In effect, Galileos
principle of circular inertia is simply a generalization of the Aristotelian principle
of natural circular motions, and no doubt this made it easier for him and his
readers to believe it.
6
Prior to Galileo it was commonly believed that projectiles such as arrows and
cannon balls moved in straight lines. According to official doctrine, a cannon
ball first moves in a straight line in the direction of the barrel, since its natural
downward motion is completely overwhelmed by the impetus generated by the
explosion. Then, some time later, when the impetus from the explosion is greatly
weakened, the natural downwards motion wins the battle and the ball drops ver-
tically downwards like a stone. This account seemed compelling because it was
assumed that different kinds of motion natural and forced could not coexist.
Galileo knew otherwise. Careful experimental work had led him to a correct
formulation of the principle of free- fall: the distance travelled by a freely falling
object increases as the square of the time. He then relied on his principle of the
persistence of horizontal motion, and argued again correctly that a cannon ball
fired in a horizontal direction (e.g. from a high castle wall) would keep on travel-
ling at precisely the same speed in the horizontal direction, while simultaneously
falling in a vertical direction in accord with the law of free- fall. The resulting
trajectory, as Galileo demonstrated, is a parabola (Figure 10.1).
This influential piece of analysis played a vital role in the defence of Coper-
nicanism. Galileo argued that a ball dropped from a tower will hit the ground at
the towers base irrespective of whether the Earth is moving or not, and will take
the same time to do so. For this to be possible, a ball dropped on a moving Earth
must have a complex motion with a vertical component that is exactly the same
as that found in the ball falling on a stationary Earth, and for this to be possible
172 TIME AND SPACE
the various components of a complex motion must not interfere with one another,
which is just the result delivered by Galileos analysis of projectiles. Precisely the
same sort of compounding of independent motions is required in the case of the
ships cabin. The motions of rising smoke particles in stationary and moving cabins
are exactly alike in all respects save one: in the latter case there is an additional
horizontal component, but since this component is the same for all the particles,
there is no observable difference between the two cases.
Perhaps Galileos contribution is best summed up in his own words: Motion,
in so far as it acts as motion and among things which all share equally in any
motion, it does not act, and is as if it did not exist. It is a perturbing result, as
Galileo himself appreciated: you are not the first to feel a great repugnance
toward recognizing this inoperative quality of motion among the things which
share it in common (Barbour 1989: 391).
10.3 Descartes
Although Newton worked all Galileos significant discoveries into the synthesis
of the Principia indeed, he credited him (not entirely accurately) with having
discovered the first and second laws of motion when he carried out his first
important work in the 1660s, the dominant physics, which he was both influenced
by and reacted against, was that of Descartes. Since Newtons arguments about
absolute and relative space in the Scholium and elsewhere often take Descartes
as their target, it will prove helpful to know something of what Descartes himself
had to say about these matters.
Descartes was working on The World (Le Monde) completed around 1633
and including the essentials of Cartesian physics at around the same time as
Galileo was composing the Dialogue, but whereas Galileo was content merely to
modify the Aristotelian worldview, Descartes was intent on demolishing it entirely,
and largely succeeded. He was inspired by the dream of explaining everything in
entirely mechanical terms: by universal natural laws governing the movements
and interactions of material bodies in space. This mechanical outlook is taken
for granted by most contemporary scientists, but it was far from widespread in
Descartess day, and it was a perspective that Newton found appealing.
More specifically, in his construction of the cosmos Descartes allowed himself
just two basic ingredients, matter and motion, and three funda mental principles,
inertia, the conservation of matter, and the conservation of motion. According to
the latter pair, there is a fixed quantity of matter and motion in the universe that
h h h h
Figure 10.1 Projectile
motion. Galileo discovered that
objects such as cannonballs, fired
into the air near the surface of the
Earth (at an angle) follow parabolic
trajectories. The horizontal speed
remains constant, only the vertical
speed varies.
SPACE: THE CLASSICAL DEBATE 173
can neither diminish nor increase. Descartes here anticipates the contemporary
principle of energy conservation. As for inertia, here, too, Descartes subscribed
to the principle in its modern (or at least Newtonian) form: objects tend naturally
to move in straight lines at constant speeds, and will continue to do so for ever
unless acted on by some external force or influence. Whereas previously people
had wondered why a stone released from a throwers arm should continue to
move, Descartes turned the question around, and wondered why it should ever
stop. He went further and pointed out that the forces needed to move and stop
objects are equal: the action needed to move a boat which is at rest in still water is
no greater than that needed to stop it suddenly when it is moving (The Principles
of Philosophy II, 26; Huggett 1999: 96). The lingering Aristotelian influence that
led Galileo to subscribe to circular inertia has been utterly expelled.
7
The details of Descartess scheme, although interesting in their own right, need
not detain us long. He took the universe to be a vast (possibly infinite) plenum;
that is, there is matter everywhere, and no empty space. Matter is mobile and
comes in portions of different sizes. The smallest particles are very fast- moving
and are associ ated with fire; the medium- sized pieces move less quickly and are
spherical (having been ground into this shape by eons of constant motion), the
largest pieces move yet more slowly and compose large bodies such as the Earth.
Descartes assumed that the space vacated by a moving body would immediately be
filled by other matter (much as water immediately rushes in to occupy the places
vacated by a moving fish), and since the easiest way for this to happen is for matter
to move in continuous circular streams, he argued that after a long period of time
matter would tend to congregate in a number of spherical vortices. At the centre of
each vortex was a star; our Sun is just one of many. Since Descartes believed that
the planets were transported around the Sun by rotating matter- fields, the vortices
were huge at least the size of the solar system. In accordance with the principle of
rectilinear inertia, the fast- moving particles at the centre of a vortex are constantly
trying to move in a straight line (just like the stone propelled by a sling), but they
are prevented from doing so by the surrounding larger, slower- moving pieces of
matter, which constrain them to move in circles.
Several key ingredients of Newtonian physics are now in place: the principle
of inertia; the conservation principles; and, no less importantly, the fundamental
idea that the universe as a whole can be explained solely in mechanical terms. But
what of absolute space? This is where things get interesting. In The World it seems
that Descartes simply assumed that space had a Euclidean structure and existed
independently of matter much in the way that Newton would subsequently do.
But The World was never published in Descartess lifetime. Hearing of Galileos
troubles with the Inquisition in 1633, troubles linked to the issue of whether the
Earth moves, Descartes preferred to suppress it. Instead, Principles was published
posthumously in 1644, and contains all the essential ingredients of Cartesian phys-
ics that were to be found in The World, but with one significant difference: superim-
posed on this familiar backdrop is an entirely new (and confusing) account of space.
He starts off with a relationalrelativist account of both position and motion:
The terms place and space, then, do not signify anything different from
the body which is said to be in a place; they merely refer to its size, shape
174 TIME AND SPACE
and position relative to other bodies. To determine the position, we have to
look at various other bodies which we regard as immobile; and in relation
to different bodies we may say that the same thing is both changing and not
changing its place at the same time. For example, when a ship is under way,
a man sitting on the stern remains in one place relative to the other parts of
the ship with respect to which his position is unchanged; but he is constantly
changing his place relative to the neighbouring shores Then again, if we
believe the Earth moves, and suppose that it advances the same distance from
west to east as the ship travels from east to west in the corresponding period
of time, we shall again say that the man sitting on the stern is not changing
his place, for we are now determining the place by means of certain fixed
points in the heavens. Finally, if we suppose that there are no such genuinely
fixed points to be found in the universe (a supposition which will be shown
below to be probable) we shall conclude that nothing has a permanent place,
except as determined by our thought
(Principles II, 13; Huggett 1999: 934)
There is here no mention of motion with respect to space itself; Descartes is now
assuming that all movement consists in bodies changing their distance relations
with one another.
The rejection of space as the ultimate frame of reference does not, in itself,
entail that all motion is relative. Aristotle did not believe in the existence of space,
but nor did he believe that motion is relative: the stationary Earth and constantly
revolving heavens (whose distance from the Earth is constant) provided secure
points of reference. But in rejecting both space and the closed Aristotelian cosmos,
Descartes has cut himself adrift. In a cosmos consisting of nothing but pieces of
matter in constant motion, where even the stars are continually jostled from their
positions, there are no fixed bearings. The only fixed points are those that we
choose to regard as fixed.
This position is radical and disturbing, but in the context of the perpetual flux
of the Cartesian cosmos it is also quite intelligible (although as Newton was later
to point out, it is scarcely consistent with Descartess principles of motion). But no
sooner has he advanced this relativist position than he takes a step back:
If, on the other hand, we consider what should be understood by motion,
not in common usage but in accordance with the truth of the matter, and
if our aim is to assign a determinate nature to it, we may say that motion is
the transfer of one piece of matter, or one body, from the vicinity of the other
bodies which are in immediate contact with it, and which are regarded as
being at rest, to the vicinity of other bodies.
(Principles II, 25, original emphasis; Huggett 1999: 934)
Descartes now seems to be telling us a different story. The relativist position
conforms with common usage (as Newton also maintained in the Scholium), but
there is a truth of the matter: a body really moves if it changes its position with
respect to a motionless body with which it is in immediate contact. This to some
extent agrees with Aristotle, who, it will be recalled, said that we should define
an objects place in terms of the nearest surrounding motionless body. But as
SPACE: THE CLASSICAL DEBATE 175
Descartes himself surely realized, relativism has scarcely been vanquished: in the
context of the Aristotelian system motionless could mean really motionless,
but not so in the Cartesian system. This, of course, explains why we find Descartes
defining true motion in terms of bodies regarded as being at rest; there is no fact
of the matter as to which bodies are really moving, only convention. Descartes
himself goes on to stress just this point: transfer is itself a reciprocal process: we
cannot understand that a body AB is transferred from the vicinity of a body CD
without simultaneously understanding that CD is transferred from the vicinity of
AB (Principles II, 29).
Confusing it might be, but this doctrine served Descartes very well vis- - vis the
Inquisition. Galileo found himself in trouble for suggesting that the Earth might
really move. Descartes has immunized himself against any such accusation. Since
he held that the material in the vortex that is in direct contact with the Earth is
motionless relative to the Earth, given his definition of true motion, it follows
that the Earth does not truly move, since it is stationary with respect to its immedi-
ate surroundings.
8
It may well be that Descartes did not himself believe the account of motion he
develops in the Principles; it is hard to tell. But as Barbour says, What is not in
doubt is the most remarkable of the paradoxes we meet in Descartes: that he was
simultaneously the effective founder of the two diametrically opposed concepts of
motion the absolute and the relative (1989: 437).
10.4 Leibniz
We now jump forward in time to November 1715, some eighty years after Des-
cartes and thirty years after the publication of Newtons Principia, when Leibniz
sent a letter to Caroline, Princess of Wales, warning her of the theologically perni-
cious character of Newtons various doctrines, not least the claim that space is
the sensorium of God. The eminent scientist and theologian Samuel Clarke
responded on Newtons behalf, and the result was the LeibnizClarke Correspond-
ence (Alexander 1956), which was published in 1717, shortly after Leibnizs death.
At various points Clarke seems to have consulted Newton when drawing up his
replies.
9
As the Correspondence developed, the focus fell increasingly on the exist-
ence of absolute space, which, of course, Leibniz resolutely denied. The arguments
that Leibniz deployed against absolute space are, as we shall see, a rather mixed
bag, but he did succeed in bringing to light a crucial difficulty in Newtons position.
Whether the difficulty is surmountable will be the topic of Chapters 11 and 12.
Leibniz pointed out two implications of Newtons physics. Each involves a way
the actual world could be differently related to absolute substantival space; each
posits a sort of shift that cannot be detected.
Static shift: Newton claims that the material universe has a particular loca-
tion within absolute space. If the material universe were situated somewhere
else in absolute space, then, provided all the distances and relative motions
between bodies remained the same, everything would appear just as it does.
For example, if all the bodies in the universe were located ten trillion miles to
176 TIME AND SPACE
the south of where they actually are, there would be no empirically discernible
consequences.
Kinematic shift: According to Newton, everything in the material universe
has a definite but undetectable state of motion relative to absolute space. If
everything in the entire universe had a different state of motion relative to
absolute space, provided the distances and relative motions between bodies
remain the same, there would be no observable differences. For example, if
all the bodies in the universe were boosted to a velocity of a thousand mph in
a north- easterly direction, there would be no empirically discernible conse-
quences after the boost.
These shifts generate what we can call Leibniz alternatives, possible situations
with the following characteristics:
all the material bodies in the universe are differently related to absolute space;
all the spatiotemporal relations between bodies are the same;
the situations in question are exactly the same in all observable respects.
The kinematic shift is illustrated in Figure 10.2 for a simple universe containing
just three objects in absolute space. The scene on the left illustrates how things
actually are; the three objects are moving away from one another in the directions
indicated by arrows. On the right are the same three objects, with the same rela-
tive motions with respect to one another, but in addition, the whole system has an
additional velocity component (with respect to absolute space) in a north- easterly
direction. Since absolute velocity is undetectable, so is this boost.
Before proceeding it is important to note that in claiming that absolute velocity
is without empirical consequences Leibniz was not guilty of imputing to Newton
a doctrine to which the latter did not subscribe; far from it. The kinematic shift
is simply Galileos ships cabin example writ large, and is a further illustration
of the principle that Galileo was so keen to establish: the inoperative quality of
motion among the things which share it in common. Newton himself recognized
the point in the Principia: The motions of bodies included in a given space are
the same among themselves, whether that space is at rest, or moves uniformly in a
right line without any circular motion (Corollary V to the laws of motion; Thayer
1974: 29). The import of this statement is far- reaching.
t
1
t
2
t
1
t
2
Figure 10.2 A universe in motion: the kinematic shift.
SPACE: THE CLASSICAL DEBATE 177
By a given space Newton means a relative space. The easiest way of defining a
space of this sort is by taking one body as our origin our reference body and
establishing three orthogonal axes (i.e. at right angles to each other). We then plot
the movements of other bodies by reference to these coordinates. Corollary V is
concerned with relative spaces of a particular sort: those established on reference
bodies that are either at rest, or moving (but not rotating) with respect to one
another at constant speeds and in straight lines. We can call relative spaces that
meet these conditions Galilean frames. Now, if like Newton we assume that
absolute space exists, any Galilean frame has some state of absolute motion, or
rest, depending on whether the relevant reference body is at absolute rest or not.
On the face of it, the laws governing moving bodies could easily be such that bod-
ies behave differently depending on their state of motion with respect to absolute
space. For example, the force of gravity could be weaker for bodies in absolute
motion. If this were the case, it would be possible at least in principle to dis-
cover whether or not a given Galilean frame is at absolute rest or not. Suppose that
two stars, Alpha and Beta, are moving apart from one another at uniform speed
in a straight line. Each star is at the centre of its own Galilean frame, Frame- Alpha
and Frame- Beta, and both stars have planetary systems. By collecting data about
the masses and orbits of the various planets, we could quickly ascertain which of
these two frames has the greatest absolute velocity.
But, it turns out, no such test is possible. Newtons law of universal gravitation,
for the simplest case, states that for two bodies of mass m
1
and m
2
separated by a
distance d, each exerts on the other a force that is proportional to the product of the
two masses and inversely proportional to the square of the distance between them,
and acts along the straight line joining them (hence the equation F = km
1
m
2
/d
2
,
where k is the gravitational constant). There is no mention here of absolute quanti-
ties, the strength and direction of the force is determined by the relative configura-
tions of bodies, and so is entirely independent of their position and velocity with
respect to absolute space. The same applies to Newtons other laws: the forces
between interacting bodies depend solely on their relative configurations and veloc-
ities. As a consequence, there is no way of determining the absolute state of motion
of any Galilean frame. The bodies in Frame- Alpha and Frame- Beta will behave in
exactly the same ways irrespective of any uniform absolute motion of these frames.
This property is called Galilean invariance.
As we can now see, Newton is plainly committed to the undetectability of both
the static and kinematic shifts. Leibniz pounced on this, and with a series of argu-
ments sought to establish that this commitment leads to absurd or objectionable
conse quences. Since these various difficulties all flow from the assumption that there
is a distinction between absolute and relative motion, he argued that we must reject
this assumption, and with it the notion that there is such a thing as absolute space.
10.5 The argument from indiscernibility
Leibniz was committed to the principle of the identity of indiscernibles (PII). When
applied to objects, PII amounts to the following. Let S and T be names of objects.
If S and T have exactly the same genuine properties then S and T are one and the
178 TIME AND SPACE
same thing. Applied to possible worlds, PII entails that there cannot be two distinct
worlds W
1
and W
2
that are exactly similar with respect to all genuine properties. If
they are indiscernible in this way, W
1
and W
2
are one and the same world.
10
Here
is how Leibniz deploys the principle:
To suppose two things indiscernible, is to suppose the same thing under two
names. And therefore to suppose that the universe could have had at first
another position of time and place, than that which it actually had; and yet
that all the parts of the universe should have had the same situation among
themselves, as that which they actually had; such a supposition, I say, is an
impossible fiction. (Alexander 1956: 37)
To say that God can cause the whole universe to move forward in a right line,
or in any other line, without making otherwise any alteration in it; is another
chimerical supposition. For, two states indiscernible from each other, are the
same state; and consequently, tis a change without a change.
(Alexander 1956: 38)
The first passage is concerned with a static shift, the second a kinematic shift.
Newton is committed to saying that both sorts of shift produce different states
of affairs. Leibniz asserts that since in both sorts of case the pre- shift universe is
indiscernible from the post- shift universe, given PII, the two supposedly distinct
states of affairs are in fact one and the same. Newtons theory entails a result that
is either absurd or false. If, on the other hand, we adopt the relational theory, and
so hold that the only spatial relations that bodies have are relations with other
bodies, then we get the right result: since in both cases the relative spatial relations
are the same, the alleged shifts cannot have taken place.
The argument from PII is uncompelling. The principle is itself very controver-
sial, and it is by no means the case that everyone accepts it.
11
But we need not
enter into this dispute, for Leibnizs argument begs the question against Newton
in a quite blatant fashion. The argument starts from the premise that the envis-
aged Leibniz alternatives are indiscernible with regard to all genuine properties.
Since Newtonians hold that relations between material bodies and absolute space
are real features of the world, they will simply deny the premise: the pre- and
post- shift worlds are different, since the relations between objects and space are
different in the two cases. It is true that these differences are not detectable, but
the PII does not say Two objects (or worlds) which do not differ in any detectable
ways are numerically identical; it says Two objects (or worlds) which do not dif-
fer with respect to any genuine properties are numerically identical. Since Leibniz
has not yet established that only empirically detectable differences in properties
are real differences, his argument as it stands fails. The fact that indiscernible
can be read as meaning not detectably different can easily obscure this point.
10.6 The argument from sufficient reason
Leibniz was also firmly committed to the principle of sufficient reason (PSR):
nothing happens without a sufficient reason why it should be so, rather than
otherwise (Alexander 1956: 25). This leads him to argue as follows:
SPACE: THE CLASSICAL DEBATE 179
Space is something absolutely uniform; and without the things placed in it,
one point of space does not absolutely differ in any respect from another
point in space. Now from hence it follows, (supposing space to be something
in itself, besides the order of bodies among themselves,) that tis impossible
there should be a reason why God, preserving the same situations of bod-
ies among themselves, should have placed them in space after one certain
particular manner, and not otherwise; why every thing was not placed the
quite contrary way, for instance, by changing East into West. But if space is
nothing else, but that in order or relation; and is nothing at all without bod-
ies, but the possibility of placing them; then those two states, the one such
as it is now, the other supposed to be the quite contrary way, would not at
all differ from one another. Their difference therefore is only to be found in
our chimerical supposition of the reality of space in itself.
(Alexander 1956: 26)
Leibniz seems here to be envisioning static shifts in the context of an initial act
of creation: God had to plant the material bodies somewhere in space but, given
the uniformity of space, what reason could he have for planting them in one col-
lection of locations rather than another? If we assume that God does nothing for
which he lacks a good reason, then we can be certain that God would not create
a substantival space, for if he were to create such a space, he would be putting
himself in a position where reasonless choices would have to be made, which he
would never do. By not creating a Newtonian space, by creating only material bod-
ies that are spatially related in certain ways to one another, the problem is avoided.
Do the same considerations apply to the kinematic shift? Suppose that someone
were to argue thus:
Just as God could have created the world elsewhere than he did, he could
have given it any number of states of motion relative to space itself. Just as
there is no reason for creating the world in one place rather than another,
there is no reason to give it one state of motion relative to space rather than
another.
While there is no obvious reason why God should choose to locate the world
in one part of absolute space rather than another, since by hypothesis this space
is infinite and homogeneous, this same homogeneity suggests a solution to the
motion problem. Since there is nothing to choose between any of the available
directions, once God had decided where to place the world there would be no
reason for him to set it in motion in one direction rather than another, so he would
have good reason not to set it in motion: he would, therefore, create a cosmos
whose overall centre of mass is at absolute rest.
Of course this leaves the initial location problem unsolved. However, what-
ever force this objection might have against theistic Newtonians, it has virtually
none against the atheistic Newtonian. If we leave God out of the picture, the
problematic choice never arises, and the issue of whether or not there could be
rational grounds for making it is simply irrelevant.
Even if we work within Leibnizs framework, and assume that a rational God
did create the universe, it is not at all clear that the argument from PSR is successful.
180 TIME AND SPACE
The original source of the argument suggests that not all theistically minded phi-
losophers would agree with Leibniz: it was Clarke who first introduced the shifts
into the discussion, arguing that relationism is refuted by the obvious fact that
God could have chosen to place the whole material universe somewhere else in
absolute space while retaining all spatiotemporal relations between bodies.
12
As for
finding a reason for putting the universe in one place rather than another, to sug-
gest that the task is beyond God is to suggest that God is no wiser than Buridans
ass. Or as Clarke puts the point:
when two ways of acting are equally and alike good to affirm in such cases
that God cannot act at all, or that tis no perfection in him to be able to act,
because he can have no external reason to move him to act one way rather
than the other, seems to be denying God to have in himself any original
principle or power of beginning to act, but that he must needs (as it were
mechanically) be always determined by things extrinsic.
(Alexander 1956: 323)
It should be pointed out that PSR can be taken in a non- theistic way. The claim
that nothing happens without a sufficient reason why it is so can be interpreted
to mean everything has a sufficient cause. If we read PSR in this way, how can
the substantivalist respond to the question Why is the world as a whole located
where it is? Sklar suggests that the substantivalist can answer thus:
Because yesterday it was, as a whole, in this position and no forces arose to
move it. So there is a sufficient reason for the present place of the material
world in substantival space its previous position and the forces that have
acted in the meantime. After all, why is Jupiter where it is today relative to the
Earth? Answer: Because of the positions and velocities they had yesterday and
the forces acting upon them in the intermediate period. (1977: 180)
Explaining the way things are at present in terms of causal laws and the way
things were earlier is what counts as providing a sufficient reason in contem-
porary science. Leibniz would have refused to accept this sort of answer as provid-
ing a genuinely sufficient reason, since it explains one contingent state of affairs
in terms of another contingent state of affairs, and he believed that ultimately
nothing was contingent; everything is as it is because it had to be this way, as a
matter of necessity. His argument thus rests on a particular version of PSR that,
as Sklar notes, most present- day philosophers and scientists would reject out of
hand (1977: 181).
10.7 The methodological argument
There is a further argument to be found in Leibniz, one that he did not greatly
emphasize himself, but that has proved by far the most influential subsequently.
In his Fifth Paper he writes:
The author replies now, that the reality of motion does not depend upon
being observed; and that a ship may go forward, and yet a man, who is in
SPACE: THE CLASSICAL DEBATE 181
the ship, may not perceive it. I answer, motion does not indeed depend upon
being observed; but it does depend upon being possible to be observed.
There is no motion, when there is no change that can be observed. And when
there is no change that can be observed, there is no change at all.
(Alexander 1956: 74)
I noted earlier, in connection with the argument from PII, that Leibniz had failed
to defend the claim that only empirically detectable differences are real differ-
ences. He is now claiming that although real changes can occur without being
observed, any real change is such that it must be possible for someone to observe it.
Is this a plausible position? There are those who have found it so. The logical
positivists subscribed to the verification principle: for a claim about the world
to be meaningful it must be empirically verifiable. This principle yields the result
that claims about uniform absolute motion are meaningless, since none can be
empirically verified. It is not at all clear that this is what Leibniz had in mind, and
since few now are convinced that the verification principle is true, perhaps this is
just as well. Scientists these days are quite prepared to posit unobservable entities,
and thus unobservable changes, provided these unobservable goings on are among
the postulates of a theory that explains and predicts what can be observed.
This last point takes us to the heart of the matter. Many of the entities postu-
lated by contemporary scientists may not themselves be directly observable, but
they do have effects that are observable. Sub- atomic particles are not visible to the
naked eye, but not only are they the building blocks of the macroscopic objects
that we can perceive, their movements can be (briefly) tracked by the detectors
in particle accelerators. No scientist would be happy in accepting a theory that
postulated entities that have no discernible empirical consequences. By way of an
example, suppose that two theories of particle physics are on offer, P and Q. In
many ways theory P is just like theory Q: both explain the behaviour of all known
particles and forces in equally elegant ways, but there is one difference. Theory P
posits an additional type of particle the squillion which are swarming every-
where all the time. Theory Q does not posit squillions. More over, according to P,
squillions are in principle unobservable, directly or indirectly: they never interact
with any other particles. Clearly, we would have no reason to believe theory P,
and so would reject the claim that squillions exist. According to Leibniz, Newtons
absolute space is akin to the squillion. Not only is it unobservable, directly or
indirectly, but we have an equally powerful theory of space, relationism, which
is just as good as Newtons, but that does not posit any unobservable entities, or
empirically indiscernible states of affairs.
This is a potentially compelling argument. I say potentially because Newton (as
we shall see shortly) disputes Leibnizs claim that motion with respect to absolute
space has no empirical consequences. Newton claims that dynamic (forced, accel-
erated) shifts are detectable.
11 Absolute motion
11.1 Inertial motion
Newton deploys two basic arguments against relationism. The best known is the
argument from inertial effects: the notorious bucket argument. The second is the
argument for real inertial motion. Both are connected with Newtons first law of
motion, namely that any object will continue to move at a constant velocity (or
remain at rest) unless acted on by a force. Before proceeding it will help to review
some of the relevant terminology.
As we have already seen, Newtons first law is often called the law of inertia.
Inertial motion is motion in the absence of impinging forces: a moving body will
continue to move at the same speed in a straight line for ever if no forces act on it.
Inertial forces are the forces experienced by a body that is undergoing acceleration.
We already know that in the context of Newtons physics a body is at absolute
rest if it is stationary with respect to absolute space, and in absolute motion if is
moving relative to absolute space. An object possesses relative velocity if it is mov-
ing at a certain speed and direction with respect to another object. Accelera tion is
rate of change of velocity. So an object is undergoing absolute acceleration when it
is changing its velocity with respect to absolute space, whereas relative acceleration
consists in a change in velocity with respect to some other bodies.
We can establish a frame of reference by centring a coordinate system on a par-
ticular body that we assume to be at rest. We can specify the motions of all other
bodies relative to the reference body; these other bodies are at rest or in constant
motion or accelerating, relative to this frame. So-called inertial frames of refer-
ence are centred on bodies that are not undergoing absolute acceleration; these
are the Galilean frames encountered in 10.4.
11.2 The argument for real inertial motions
We start with De Gravitatione, a work that remained unpublished until recently,
but that contains important arguments not to be found in the Scholium. Newtons
explicit target here is Cartesian physics, as formulated in Descartess Principles,
and he leaves rubble in his wake.
In 10.3 we saw that Descartes in Principles if not The World combined a
commit ment to the law of inertia with a relational conception of motion. As New-
ton is quick to point out, the two are incompatible. The law of inertia states that a
ABSOLUTE MOTION 183
body will continue to move in a straight line at constant speed unless acted on by
a force. This entails that a body that is not moving in a straight line is being acted
on by a force. Since the existence of forces is not a conventional matter, the law
of inertia presupposes a real and objective difference between straight and curved
paths. However, if like Descartes we hold that all motion is relative to move is
to have different distance relations to other bodies at different times there is no
such difference. Each body can legitimately be regarded as stationary, and so can
be used to define a reference frame in terms of which the motions of other bodies
can be determined. In Figure 11.1, A and B are two spaceships. In the situation
depicted on the left, A takes itself to be moving in a straight line, and regards B
as flying along a curved path. B does likewise, and regards A as flying along a
curved path. For the relationist, neither perspective is privileged; both are equally
valid. The relationist is thus committed to the view that every body can rightly be
considered to have many different motions, in which case there is no objective fact
of the matter as to what a bodys real motion is, and hence there is no objective
fact of the matter as to whether a body is moving in a straight line or not.
Consider another example involving competing reference frames. According to
frame of reference F
1
, the Earth is in orbit around the stationary Sun; according to
frame of reference F
2
, the Sun is in orbit around the stationary Earth. If we now
suppose that the whole solar system is moving uniformly in a straight line, these
two reference frames ascribe very different motions to the Earth. According to F
1
,
the Earth is moving in a spiral (since it rotates around the Sun, which is moving
in a straight line), whereas according to F
2
, the Earth is moving in a straight line,
and the Sun is moving in a spiral. Bearing in mind that from a relationist point of
view these two frames are equally valid, we now bring the law of inertia into the
picture. According to F
1
, the Earth is moving in a spiral, and so a force must be
acting on it, dragging it away from its inertial straight- line path; by contrast, since
the Sun is moving in a straight line, no forces are acting on it. But according to
F
2
, it is the Sun that is rotating around the Earth, whereas the Earth is following
its straight- line inertial path, and so while there are no forces acting on the Earth,
a force must be acting on the Sun. Clearly these two accounts are inconsistent.
They ascribe different forces to the same bodies in the same situation; both cannot
be true, for the existence of a real force cannot depend on an arbitrary choice of
reference frame. For a consistent physics we need a notion of real inertial motion:
there must be an objective fact of the matter as to which motions really are inertial,
and which are not. Newtons concept of absolute velocity fits the bill.
B
A
A
B
Figure 11.1 Ships in flight
but along which paths?
If facts about motion are
determined solely by the distances
between objects at different times,
the distinction between straight
and curved paths becomes a
matter of perspective.
184 TIME AND SPACE
Newton follows up with a further assault on relationism. If an objects loca-
tion consists solely in its distances from other objects at a given time, then if (as
Descartes believed, and as seems largely true) bodies are continually changing
their positions relative to other bodies, there are no facts about past locations,
and this leads to disastrous consequences, as Newton argues in this passage from
De Gravitatione:
For since, according to Descartes, place is nothing but the surface of surround-
ing bodies, or position among some other more distant bodies, it is impossible
(according to his doctrine) that it should exist in nature any longer than those
bodies maintain the same positions from which he takes the individual desig-
nation. And so, reasoning as in the question of Jupiters position a year ago,
it is clear that if one follows Cartesian doctrine, not even God himself could
define the past position of any moving body accurately and geometrically now
that a fresh state of things prevails, since in fact, due to the changed position
of the bodies, the place does not exist in nature any longer.
Now as it is impossible to pick out the place in which a motion began
(that is, the beginning of the space passed over), for this place no longer
exists after the motion is completed, so the space passed over, having no
beginning, can have no length; and hence, since velocity depends upon the
distance passed over in a given time, it follows that the moving body can
have no velocity, just as I wished to prove at first. Moreover, what was said
of the beginning of the space passed over should be applied to all intermedi-
ate points too; and thus as the space has no beginning nor intermediate parts
it follows that there was no spaced passed over and thus no determinate
motion, which was my second point. It follows indubitably that Cartesian
motion is not motion, for it has no velocity, no definition, and there is no
space or distance traversed by it. So it is necessary that the definition of
places, and hence of local motion, be referred to some motionless thing such
as extension alone or space in so far as it is seen to be truly distinct from
bodies. (Huggett 1999: 10910)
Newton argues that, for there to be an objective fact of the matter as to how
fast a body is moving at a given moment, and in which direction, there have to be
objective facts concerning where it was located at previous times. If we define spa-
tial locations by reference to material bodies, and if material bodies are constantly
changing their spatial relations to one another, the required facts are simply not
available. It would be different if we could take just one body as our standard: if
(say) we knew the Sun to be motionless, we could plot the movements of Jupiter
by measuring the distance between the two at different times. But this move is
unavailable to the relationist. The situation is very different if we recognize abso-
lute space, and hence absolute motion: there is now an objective fact as to the
directions objects are moving in and the distances between them.
This argument is intriguing, but not decisive. There may be ways for relation-
ists to meet Newtons challenge. But Newton is certainly right about there being
a difficulty here, a difficulty to which his notion of absolute space supplies a
solution.
1
ABSOLUTE MOTION 185
11.3 The argument from inertial effects
In the light of the arguments considered thus far, we can provisionally grant New-
ton this much: the law of inertia does indeed presuppose a real difference between
bodies that are in uniform straight- line motion and bodies that have some different
form of motion (e.g. circular), a difference that the relationist who appeals only to
distance relations between bodies cannot recognize. However, what is to prevent
the relationist from rejecting the distinction in question? Is there anything to pre-
vent Leibniz from holding that claims such as the Sun is moving in a straight line
and the Earth is rotating about it and the Earth is moving in a straight line and
the Sun is rotating about it are equally true? It is here that Newtons arguments
in the Scholium come into play.
2
While Newton fully accepted that absolute velocity has no detectable conse-
quences there are no experiments that will reveal how fast you are moving with
respect to absolute space he took pains to point out that the same does not apply
for changes in velocity; that is, accelerations. We all know that forces are associated
with acceleration. If you are in a quickly accelerating car you feel yourself being
pulled back into your seat. This is a typical instance of so- called inertial forces
making themselves felt: these are the forces that are experienced by a body that is
not following its inertial path through space as a result of some applied force, such
as that supplied by a cars engine. The trouble is, just as velocity is frame- relative
for the relationist, so is acceleration. If A and B are accelerating apart, then for the
relationist there is no fact of the matter as to who is really accelerating: each of A and
B can regard themselves as being at rest and the other as accelerating with respect to
them. If I accelerate away from your car, which remains stationary at the pavement,
then, relatively speaking, it is as true to say that I am accelerating away from you as
it is to say that you are accelerating away from me. But it is clear that if we describe
the situation in terms of relative accelerations we are omitting a crucial factor: it is
only me who feels inertial forces, you do not. I am the one pulled back into my seat,
not you. The obvious explanation of this is that only one of us is really accelerating.
Hence it seems that an adequate dynamics requires a distinc tion between real and
relative acceleration, for without it we cannot account for the presence/absence of
inertial forces (i.e. those experienced by the accelerating body). Since the relationist
has no means of grounding this distinction, his account of (accelerated) motion is
inadequate. If we assume, as most do, that acceleration is a change of motion with
respect to something, then what alternative is there to Newtons view that real
acceleration is a change of motion relative to absolute space?
Once the existence of absolute space is established in this manner, we can
deduce that material objects that are not accelerating must be either in uniform
motion or at rest with respect to it, even though there is no observational evidence
that will tell us which state a particular body is in.
In making these points in the Scholium Newton employed two thought-
experiments. Neither involves linear accelerations of the sort that we have just
been considering; rather they feature rotations. Newton claims: The effects which
distinguish absolute from relative motion are, the forces of receding from the axis
of circular motion. For there are no such forces in a circular motion purely rela-
tive, but in a true and absolute circular motion they are greater or less, according
186 TIME AND SPACE
to the quantity of the motion (Huggett 1999: 122). And of course, rotation is
itself a form of acceleration. Since acceleration is change in velocity, and velocity
involves moving in a particular direction, a rotating object is continually changing
its direction of movement, and hence is continually accelerating. In the case of a
stone fixed to a string and whirled around ones head, at any instant the velocity
of the stone is in a direction that is tangential to the circle it is moving around,
the direction it would fly off in if released. The string has to exert a continual
force on the stone to prevent it from pursuing this natural inertial motion. In the
case of a body that follows a curved (but not circular) path, such as rocket ship B
on the left in Figure 11.1, forces supplied, let us suppose by small rocket motors,
must be applied to push the body off its inertial path. Since these motors apply
force to the body of the ship, not to its occupants, it is left for the ship itself to
force its occupants off their inertial path: unless strapped down, those on board
would find themselves being pushed by the walls of the ship, just as people in a
car turning a corner find themselves being pushed against the car doors. Similar
effects, although in one case rather more complex, underpin Newtons examples.
The bucket
Suppose that there is a bucket full of water, suspended by a rope (or elastic cord)
from the ceiling. Twist the bucket and let the rope wind up, then release the bucket.
The following sequence of events will unfold.
1. The bucket and water are initially at rest with respect to one another and the
surface of the water is flat.
2. After a short while, as the twisted rope starts to unwind, the bucket begins to
rotate, but the water remains stationary and flat.
3. Then, a while later, as the rotation of the bucket is gradually communicated to
the water via friction, the water starts to turn, and soon is rotating at exactly
the same rate as the bucket; the water is no longer flat but concave, as it rises
up the sides.
We are all familiar with the phenomenon: fluids rotating in containers rise up
the sides. Who hasnt spilt their tea when stirring in milk? Newton talks of a ten-
dency to recede from the axis of rotation, but this is not entirely accurate. The
particles in a rotating fluid have a natural tendency to move at a tangent to the
curvilinear path they are following. This is how they would move if unconstrained,
just like a stone released from a sling. But since their natural (inertial) movement
is constrained by the other particles in the fluid, and ultimately the walls of the
container their actual movement takes the form of an outwardly expanding
spiral. Since all the particles in the fluid move thus, there soon occurs a build- up
around the sides of the container. (This build- up generates a marked increase in
(downwards) fluid- pressure, which often prevents the liquid from overflowing,
depending on the rate of rotation.)
The behaviour of the water in this very basic and familiar sort of case poses a
severe problem for the relationist. Newton has an explanation for why the water
in the third stage curves: it is moving with respect to absolute space. The relationist
ABSOLUTE MOTION 187
cannot adopt this explanation, and so has to say that the curvature of the water is
due to its moving relative to some other material thing. But what? Descartes, it will
be recalled, said that real motion takes place when a body moves in relation to
its immediate surround ings. If this were right, then we should expect to find the
curvature at a maximum when the relative motion of the water with respect to the
bucket is at a maximum. But at this point, stage 2, the water is flat! The curvature
is at a maximum at the stage 3, when there is no relative motion between the two,
when both bucket and water are rotating at the same rate.
bucket at rest bucket rotating bucket rotating
water rotating water at rest water at rest
Figure 11.2
Newtons buckets.
If we cannot explain the inertial forces by appealing to the relative motion of
the water with respect to the bucket, can we say that the water is rotating with
respect to itself? No, for the same effect would be observed if none of the particles
that make up the water rotated with respect to one another. Could the curvature
be the product of motion with respect to we who are observing the bucket? Once
again the answer is no. The curvature certainly occurs when the water is rotating
relative to the onlooking observer, but it is clearly not the product of this relative
motion. If the observer were to run around the bucket very quickly the water
would remain perfectly flat (ignoring vibrations). The same would apply if we
were to put the surrounding buildings on a giant turntable and whirl them around
the motionless bucket. The experiment would be expensive to conduct but there
is no reason to suppose that the water would be in the slightest way perturbed.
Newtons explanation, that the inertial effect occurs when the water is rotating
relative to absolute space, certainly has merit.
The globes
Newton does not explicitly consider whether the curvature of the water in the
bucket case could be due to relative motion to some more distant bodies, such
as the stars. Instead he provides a further example, one that is surely (in part)
intended to refute the suggestion:
if two globes, kept at a given distance one from the other by means of a cord
that connects them, were revolved about their common centre of gravity,
we might, from the tension of the cord, discover the endeavor of the globes
to recede from the axis of motion, and from thence we might compute the
quantity of their circular motions. (Huggett 1999: 123)
If the cord contained a device for measuring tension (such as an appropriately
attached spring- balance), we could ascertain whether the globes are rotating or
188 TIME AND SPACE
not simply by looking at it: if it shows a tension the globes are rotating. So far,
so familiar. Newton now takes us a stage further. He points out that precisely the
same would apply if the globes were not surrounded by stars and planets, but
located in an immense vacuum, where there was nothing external or sensible
with which the globes could be compared (Huggett 1999: 123). So imagine two
possible worlds, W
1
and W
2
. Each world contains a pair of globes connected by
a cord, with a built- in tension meter, and nothing else. In W
1
the meter registers
zero pressure and in W
2
the meter registers a significant pull. Newton can explain
the difference: in W
2
the globes are rotating in absolute space, and in W
1
they are
not. What can the relationist offer? Since there are no objects other than the globes
and cord in either world, and the distance between the globes remains constant,
from a relational standpoint there is no relative motion in either world, and so the
globes in both W
1
and W
2
are stationary. It seems that the only difference between
these worlds is the existence of the tension in the cord in W
2
. The relationist is thus
pushed into the uncomfortable position of having to view the tension in W
2
as an
entirely inexplicable occurrence. Given the availability of an alternative explana-
tion Newtons this position seems quite untenable.
Taken together, the bucket and globes amount to a formidable challenge to
relation ism. In both cases, accelerations give rise to observable forces; these accel-
erations are not, it seems, accelerations relative to ordinary physical bodies; but if
acceleration is a change of velocity relative to something, what can this something
be if not absolute space?
3
As Earman suggests (1989: 64), Newtons argument
is best construed as (in part at least) an inference to the best explanation of the
mechanical phenomena:
P1 The best explanation of mechanical phenomena in general (and rotating
phenom ena in particular) utilizes absolute acceleration (and absolute rotation
in particular).
P2 Absolute acceleration (and absolute rotation in particular) must be under-
stood as acceleration (and rotation) relative to absolute space.
If we accept the Newtonian account we have an explanation for why the con-
nection between acceleration and force is so ubiquitous; why it exists at all places
and all times. If all acceleration is a change in velocity with respect to absolute
space, and absolute space is the same everywhere, it is to be expected that similar
accelerations give rise to the same forces everywhere and everywhen, which as far
as we know, they do. For as we have already seen, and as Newton was to go on to
demonstrate in the rest of the Principia, inertial forces are ubiquitous, and are by
no means confined to rotating bodies: every time you feel yourself pushing into
the wall of a turning bus or car you are experiencing them.
11.4 Stalemate?
Although Newtons bucket amounts to a weighty argument, relationists did not
simply surrender. This is not surprising, for although the argument from inertial
effects is powerful, the relationists stance is far from absurd. Indeed, the positions
might seem perfectly balanced:
ABSOLUTE MOTION 189
Leibniz argued (and Newton accepted) that in Newtonian physics uniform
absolute velocity has no empirical consequences. If you have an absolute
velocity, it is impossible to tell what it is. If we assume that uniform absolute
velocity is a suspect notion, since it has no empirical significance, then how
can absolute acceleration have any empirical significance? After all, absolute
acceleration is simply variation in absolute velocity.
Newton argued in the other direction. Having good grounds for believing that
absolute acceleration has significant empirical consequences, he concluded
that there must be such a thing as absolute velocity, even though the latter is
without physical consequences.
From this we see that Leibnizs kinematic shift continues to pose a problem
for the Newtonian: the empirical insignificance of uniform motion with respect
to absolute space renders the relevance of acceleration (relative to absolute space)
questionable. This said, if relationists are to reject Newtons theory they must
still come up with an alternative explanation of inertial effects. If these are not
produced by absolute acceleration, construed as change in velocity with respect
to absolute space, what does produce them? In the remainder of this chapter we
will consider three lines of response.
11.5 The Leibnizian response
Although Leibniz was well aware of the bucket argument, and the general prob-
lem posed by acceleration for the relationist, he never got round to publishing a
detailed response to it. However he did make some suggestive remarks. In his fifth
letter to Clarke he wrote:
53. I find nothing in the Eighth Definition of the Mathematical Principles
of Nature, nor in the Scholium belonging to it, that proves, or can prove,
the reality of space in itself. However, I grant there is a difference between
an absolute true motion of a body, and a mere relative change of situation
with respect to another body. For when the immediate cause of the change
is in the body, that body is truly in motion; and then the situation of other
bodies, with respect to it, will be changed consequently, though the cause of
the change is not in them. Thus I have left nothing unanswered, of what
has been alleged for the absolute reality of space. (Alexander 1956: 74)
Unfortunately, Leibniz died before he could spell out exactly what he had in
mind, but it may have been something along these lines. Since Newtons first law
of motion tells us that absolute accelerations are correlated with forces, cant we
say that inertial effects are produced not by change of motion with respect to abso-
lute space, but rather the forces that Newton rightly recognizes are inseparably
bound up with real as opposed to merely relative accelerations? Real accelerations
are those in which the immediate cause of the change is in the body.
This idea works well enough for some cases; for example, if two rockets in outer
space are in relative acceleration it is easy to apply Leibnizs criterion to determine
which is really accelerating it is the one whose rocket motor is firing. However,
190 TIME AND SPACE
not all cases of real acceleration are like this. We might extend Leibnizs argument
to collisions, and say that (in a manner of speaking) when contact between moving
bodies occurs, the cause of the change is in the object. But there remain many
instances of real accelerations that are not produced by external forces in this sort
of way. First, there are the accelerations produced by forces that act at a distance,
such as gravita tional and magnetic attraction (at least in Newtonian physics). Sec-
ondly, as Friedman stresses:
uniform rotation is not correlated with external forces in Newtonian theory.
Even if we could replace the reference to absolute acceleration [in some
cases] with a reference to force- generating causal processes, Newtonian
theory does not allow us to replace the reference to absolute rotation in a
similar manner. (1983: 229)
A rotating solid disc will continue to rotate at a constant angular velocity, without
any need for any external force or torque to be applied: it is maintained in motion by
the preservation of angular momentum (a basic principle of Newtonian physics, and
confirmed by, for example, the Earths daily rotation). And yet, if the disc rotates
fast enough, the centrifugal forces generated by its rotation will tear it apart.
In other (earlier) Leibnizian writings a different approach can be discerned. In
the Specimen Dynamicum he claims that all motion is in straight lines or com-
pounded of straight lines, and then says:
From these considerations it can be understood why I cannot support some
of the philosophical opinions of certain great mathematicians on this
matter, who admit empty space and seem not to shrink from the theory of
attraction, but also hold motion to be an absolute thing and claim to prove
this from rotation and the centrifugal force arising from it. But since rotation
arises only from a compo sition of rectilinear motions, it follows that if the
equipollence of hypotheses is saved in rectilinear motions, however they are
assumed, it will also be saved in curvilinear motions.
(Loemker 1970: 44950)
His idea may have been that although objects look as though they are rotating,
in reality they are composed of microscopic particles that are moving in straight
lines, and are zig- zagging back and forth. This is an intriguing speculation, but,
as Earman points out (1989: 72), Leibnizs analysis is based on the assumption
that there is a fact of the matter as to which trajectories are straight and which are
not, and as we have already seen (10.2), in a relational space a trajectory that is
straight relative to one reference body will be curved relative to another, and each
perspective is equally valid.
11.6 The Machian response
The nineteenth- century physicist Ernst Mach criticized Newton for positing
absolute space on empiricist grounds. Mach was deeply suspicious of unobserv-
able entities (such as absolute space), and believed that science should concern
itself only with relation ships between observable things and events. This general
ABSOLUTE MOTION 191
approach, which exerted a profound influence on Einstein, naturally leads to a
relationist approach to space.
Like Berkeley, Mach criticized Newton for (as he saw it) going beyond the
observa tional data. The globes argument is particularly vulnerable on this score.
Newton claims that globes in an otherwise empty universe would behave in just the
same way as globes in our universe; a tension would be created in the joining cord
when the globes are rotating. Mach complains that this is an entirely groundless
speculation. On the one hand, since the experiment has never been carried out
we have no way of knowing how a pair of rotating globes would react if the rest
of the matter in the universe were to disappear. On the other hand, the experi-
ment is impossible to carry out. If the globes are the only objects in the universe,
there couldnt possibly be anyone present to observe the results of their rotation.
Newton is guilty of assuming the result he purports to be arguing for. Yes, given
his theory, the globes would behave as he describes, but his theory is the only
reason we have for believing that the globes would behave as he describes, hence
the thought- experiment doesnt prove his theory true.
But what of the bucket? What of inertial effects that are observable in the actual
universe? Since Mach is convinced that talk of absolute space is nonsense, and
that the only motion it makes sense to talk about is relative motion, he argues that
inertial effects must be the product of change relative to other material bodies in
the universe.
How does this work in practice? Which other bodies? Well, for most practical
purposes, from the standpoint of the Earth, the stars can be regarded as provid-
ing a fixed frame of reference. So instead of saying that the water in the rotating
bucket rises because it is accelerating relative to absolute space, we say that it rises
because it is rotating relative to the fixed stars. It is important to note that all the
inertial effects we observe on the Earth at least do in fact arise when objects
accelerate relative to the stars. Since we cannot observe motion relative to space,
this relative acceleration is all we can directly detect. Mach is on firm ground here.
That there are predictive differences between Machs account and Newtons is
readily grasped. For Mach, since it is the relative motion between water and stars
which causes the water to rise, it doesnt matter, or even make sense to say, which
of the two is really rotating: if there is relative motion between water and stars
the water will curve. For Newton, however, it does make sense to suppose that
the water is really stationary, and if it is stationary (i.e. at rest relative to absolute
space) it will remain perfectly flat no matter how fast the rest of the universe is
rotating about it. Similarly, if there is no relative motion between water and stars,
then according to Mach the water will remain flat. Not so for Newton. If the water
and the stars are rotating at the same rate relative to absolute space, the water will
rise, even though there is no relative motion between the two.
So far so good; now for the hard part. Mach has correctly pointed out that for
most practical purposes on Earth we can regard inertial effects as arising when
objects move relative to the stars, which we can regard as fixed. But this is only
an approximation: in fact the stars are moving relative to one another, it is just
that this motion is comparatively slow (at least when observed from Earth). So
what precise theory of dynamics is Mach offering? If the stars are not fixed, and
inertial effects are caused by relative motions, we need to specify what the relevant
192 TIME AND SPACE
relative motions are. What must an object be accelerating with respect to in order
to undergo inertial effects? What are the relevant laws of nature? What is the
mechanism?
It is here that Mach is rather vague. The general character of the needed theory
is plain. The distinction between inertial and non- inertial (acceler ated) motion
must be grounded in the overall arrangement of matter in the universe. It is
because matter is configured in a certain way that certain paths are inertial and
others are not. His one concrete proposal is that inertial forces are produced when
a body accelerates with respect to the centre of gravity of the entire universe. By
what mechanism, though, and why?
Presumably, there must be something akin to a force that links all bodies in
the universe, and that is responsible for their moving as they do. This is a bold
conjecture, one that can seem implausible: Think about it for a moment. Would
you seriously entertain the possibility that the reason you get seasick is due to your
motion relative to the stars? Is this notion any more plausible than astrology?
(Earman 1989: 211). Newton himself recognized the possibility of explaining the
bucket in this sort of way but was sceptical: But who will imagine that the parts
of the [water] endeavor to recede from its centre on account of a force impressed
only on the heavens? (De Gravitatione; Huggett 1999: 108). It is not clear that
this scorn is entirely justified: Newtons own theory of gravity, after all, posits
a force of attraction acting at a distance that operates instantaneously and links
every material body to every other material body! This is just the sort of force that
Mach needs, but there is a key difference: Newton had a precise quantitative the-
ory that matches known empirical data and whose predictions are well confirmed.
Mach can offer only a sketch of a theory of comparable power. In the absence of
a complete physical theory, the Machian approach cannot be judged successful.
4
11.7 The Sklar response
In responding to Newton, the Leibnizian and the Machian both make the same
assumption: they agree with Newton that inertial effects are due to acceleration,
and that acceleration is change of velocity with respect to something. It is this last
assumption, that change of motion has to be relative to something, that Sklar ques-
tions (1977: 22932). Sklar pointed out that the relationist can simply abandon
this assumption.
What follows if we do? Well, recall the problem. If we adopt a relationist view of
acceleration, how can it be that when two bodies are undergoing acceleration rela-
tive to one another only one body experiences inertial effects? What account can
the relationist give of this asymmetry? Sklars suggestion is that there is a primitive
monadic property that bodies moving on some trajectories possess, which comes in
different quantities, and inertial effects are associated with the possession of this
property. If we like we can call this property absolute acceleration, but this is
only a label; objects possessing this property should not be thought of as undergo-
ing changes in motion relative to Newtonian absolute space, or Machs centre of
the universe. The property in question is primitive, and thus quite independent of
absolute space, which we now say does not exist.
ABSOLUTE MOTION 193
While Sklars proposal on behalf of the relationist is interesting, in that it
reveals a previously unsuspected region of logical space, in other respects it is
not very appeal ing. The Sklarian relationist posits just what is needed for the
relationist to solve the problem of inertial effects in a way that is immune to the
problems facing the Leibniz ian and the Machian, but the posited property has no
independent motivation. In order to deliver quantitative predictions, the Sklar-
ian will presumably employ all the normal machinery of Newtonian mechanics,
and then re interpret the results in his own terms; re describing trajectories that
Newtonians view as absolute accelerations in terms of varying quantities of the
monadic property. The resulting picture is rather odd: there are bodies moving
along various paths (construed in a relational way); only some of these objects
possess the monadic property, and only those that do suffer inertial effects. Now,
there is nothing in the objects or the paths they follow that distinguishes the objects
that possess the monadic property from those that do not. The property seems to
be distributed in a quite accidental or unprincipled fashion.
Of course the Sklarian can respond: Not at all! The property is correlated with
inertial effects. There is nothing unprincipled about its distribution at all. This
is true, but does it help? The problem facing relationists is to explain how and
why inertial effects arise as they do. Sklar is solving the problem by introducing a
primitive property; the possession of this property is the proposed explanation of
inertial effects. But, and this is the real problem, all we are told about this property
is that it is that property which is found wherever inertial effects are found. By
comparison, the Newtonian offers a richer account: there is a difference between
objects that suffer inertial effects and those that dont, namely, inertial effects are
suffered by objects that are undergoing absolute acceleration. If we buy into the
idea that some objects are moving uniformly with respect to absolute space and
some are not, we have a principled explanation of what distinguishes those objects
that suffer inertial effects from those that do not. The Sklarian cannot offer as
much.
5
So at this stage of the game the Newtonian is clearly ahead on points, but the
game is not yet over. We will next look at ways to improve the Newtonian account,
but we shall also see that this revision gives rise to new problems and difficulties.
12 Motion in spacetime
12.1 Newtonian spacetime
Newton may be ahead on points, but his account is still blighted by a serious
problem. To account for inertial effects Newton recognized absolute acceleration;
having recognized absolute acceleration he also had to recognize absolute velocity
(change of which constitutes absolute accelera tion), despite the fact that absolute
velocity has no physical consequences. In positing absolute velocities, Newtonians
expose them selves to the claim that their theory gives rise to empirically indistin-
guishable states of affairs. Which of the infinite number of possible universes that
can be generated by Leibnizs kinematic shift do we inhabit? Which corresponds
to the actual world? If Newtons theory is correct, there is no empirical test that
could be performed that would reveal the answer.
Since the work of Cartan and others in the 1920s and 1930s, Newtonians have
come to see a better way to formulate Newtonian theory, a way that eradicates
absolute velocities but retains (in a way) absolute accelerations. The improved
formulation involves two steps: first, Newtons theory is formulated in terms
of spacetime; then this Newtonian spacetime is discarded in favour of neo-
Newtonian (or Galilean) spacetime.
1
Before we can appreciate the difference between these two, we need to know
what a spacetime is. We can then consider some of the ways that spacetimes can
differ. Before proceeding there are some misconceptions to dispel. On hearing the
word spacetime one naturally thinks of Einstein and relativity theory. But while
Einsteins special and general theories of relativity are indeed formulated in terms
of spacetime, they employ a particular form of spacetime (one in which spatial
and temporal intervals are relativized to reference frames), but spacetime comes in
many forms, and there are classical spacetimes that do not share this feature (see
Earman (1989: ch. 2) for a survey of classical spacetimes). A second misconcep tion
concerns the four- dimensional Block view of the universe. It is certainly true that
spacetimes are often, and understandably, construed in this way, but it would be
a mistake to suppose that they must be. As we shall see in due course, there are
alternatives. But in the meantime, to avoid overcomplicating, we can overlook this
point. In what follows I will be assuming that all parts of a spacetime coexist in a
four- dimensional ensemble.
There are two ways of approaching a spacetime: top- down and bottom- up.
The top- down approach is to start with a complicated spacetime Newtonian,
for example and then to subtract various elements of structure. The bottom- up
MOTION IN SPACETIME 195
approach is to start with the elementary components of a spacetime a collection
of distinct individual points and then to add different types of structure. We will
be following the top- down route.
Newton viewed space as an infinite and immutable three- dimensional Euclidean
continuum of points, all of which endure through time (which is also infinite).
To convert this space + time model into a spacetime we simply replace Newtons
enduring spatial points with a succession of momentary and numerically distinct
spacetime points. The switch in perspective is indicated in Figure 12.1. On the left
are four spatial points persisting through time; on the right, these four persisting
points have been replaced with an array of spacetime points. Newtonian space has
been chopped into bits, in this case into forty numerically distinct points, although
of course in a real spacetime there would be an infinity of points. Taken as a whole,
the collection of spacetime points constitute a four- dimensional continuum: a
spacetime.
Time
enduring
spatial points
momentary
spatial points
Figure 12.1 From space to spacetime.
A spacetime continuum (or manifold) is composed
of numerically distinct points-at-times, rather
than the enduring points of classical space.
While spacetime and the Block view are certainly
compatible, it is perfectly possible to combine the
spacetime view with a dynamic conception of time.
A Growing Block theorist, for example, will hold
that spacetime increases in (temporal) size as layers
of spacetime points come into being.
Just as a three- dimensional volume (think of a brick) can be thought of as
being composed of a collection of two- dimensional planes, so a four- dimensional
volume can be regarded as a collection of three- dimensional volumes, or hyper-
planes. In Figure 12.2 the two cubes represent a Newtonian space at two moments
of time (so each cube represents an infinite three- dimensional volume of space). A
succession of these cubes composes a four- dimensional spacetime. By suppressing
a spatial dimension, we can represent a four- dimensional spacetime by a succes-
sion of flat surfaces, as on the right. Each of these surfaces represents a three-
dimensional hyperplane.
In Newtonian spacetime every point is at a determinate spatial and temporal
distance from every other point. Vertical lines running down through the planes
Figure 12.2
Four-dimensional
spacetime as a sum (or
stack) of hyperplanes.
196 TIME AND SPACE
indicate points at different times that are at zero distance from one another (and
so, in effect, the same place) (Figure 12.3).
Distinguishing absolute velocity and absolute acceleration in Newtonian space-
time is easy. Persisting material objects are represented in spacetime diagrams by
worldlines (or, in the case of objects larger than points, world- tubes, but I will ignore
this complication for now). Each of these worldlines represents an object persist-
ing through a succession of different spacetime points; the worldline in its entirety
represents the complete history of the object in question. Objects that are at absolute
rest have vertical worldlines; objects in uniform motion have straight worldlines
that cut through the layer of grids at an angle; the greater the degree of deviation
from the vertical, the faster the velocity. All curved lines passing through successions
of hyperplanes represent the worldlines of bodies undergoing absolute acceleration;
the steeper the curve, the greater the acceleration. In Figure 12.4 the worldlines of
nine objects are depicted. Three of these objects are at absolute rest, three are in
inertial (uniform straight line) motion, and three are accelerating, at different rates
and in different directions. The objects in inertial motion are not moving relative
to one another: the distance between their worldlines remains constant over time.
Figure 12.3 A Newtonian spacetime. Points in different
hyperplanes (i.e. at different times) are spatially related. The
diagram employs the convention that points directly above and
below each other are at no spatial distance from one another;
hence straight vertical lines represent the same place over time.
Figure 12.4 The states of rest, uniform motion
and acceleration in Newtonian spacetime.
Looking at spacetime in a more general way, we can distinguish inertial
and non- inertial trajectories within spacetime itself. These are continuous suc-
cessions of spacetime points, occupied or unoccupied; straight paths are inertial,
curved paths non- inertial. The straight paths are those that bodies will follow if they
are either at rest or moving with a uniform velocity. All the curved paths represent
accelerated motions. Again, the vertical lines represent absolute rest (each such line
represents a different enduring point in three- dimensional space). Taken together,
the collection of straight lines vertical and non- vertical is said to represent the
inertial structure of the spacetime. This is quite natural: taken together, these lines
represent all possible inertial trajectories that bodies could take.
MOTION IN SPACETIME 197
12.2 Neo- Newtonian spacetime
To turn a Newtonian spacetime into a neo- Newtonian (or Galilean) spacetime is
not difficult. In Newtonian spacetimes there are distance relations between points
in different simultaneity hyperplanes; if we eliminate these distance relations, leav-
ing everything else unchanged, we are left with a neo- Newtonian spacetime.
Once the distance relations between hyperplanes are abolished, it no longer
makes sense to talk of how much distance separates a point in one hyperplane
from a point in a different (earlier or later) hyperplane. Moreover, the concept
same place cannot be applied over time, since points at different times are at
no distance from one another; only points that are simultaneous are located at a
spatial distance from one another.
Since all other aspects of Newtonian spacetime are preserved, the temporal
relations between hyperplanes remain intact, and the points within a given hyper-
plane are all simultaneous. Crucially, the distinction between straight and curved
paths is retained, despite the elimination of transtemporal spatial distance. Formally
speaking, this is achieved by first establishing coordinates for each three- dimensional
space (or hyperplane); this allows numbers to be assigned to points, and distances to
be given a numerical value. The next step is to stipulate which sequences of points
through successive three- dimensional spaces constitute continuous sequences. With
these lines distinguished, we then specify which ones are curved and which are
straight. In technical terms, the latter procedure is known as specifying the affine
structure of the spacetime; the specification of straight- line paths is known as sup-
plying the connection. Since the straight lines are intended to represent the paths
through spacetime that bodies would follow if no forces were acting on them, they
represent the inertial paths, and the totality of straight lines represents the inertial
(or affine) structure of the spacetime as a whole. The distinction between inertial
and non- inertial motion is thus preserved: curved paths are followed by accelerating
bodies, straight paths by non- accelerating bodies.
2
To appreciate fully the difference between Newtonian and neo- Newtonian
spacetime it will help to have a simple example on hand. Consider just two neo-
Newtonian inertial frames, P and Q. As in the Newtonian case, P can be repre-
sented by a collection of parallel worldlines running through the simultaneity
planes, and Q likewise. We start by assuming that the P- frame is at rest, and so
depict its worldlines as vertical, and those in the Q- frame as slanting. In Figure
12.5 the rectangles represent the collections of lines corresponding to these two
frames, but for the sake of clarity only a small proportion are shown. The depicted
transformation is a realignment of the hyperplanes, which renders the lines of the
Q- frame vertical (and hence motionless). A transformation of this kind is called
an inertial transformation, since it preserves inertial structure.
In neo- Newtonian spacetime every inertial frame can be turned into the zero-
velocity frame by an inertial transformation and every such transformation
preserves the distinction between straight and curved lines, and hence between
uniform and accelerated motion (see Figure 12.6). Given the stipulation that there
are no distance relations between distinct hyper planes, there is no right way
of aligning them, and the selection of one frame as privileged or motionless is
entirely arbitrary.
198 TIME AND SPACE
In Euclidean plane geometry, transformations that leave the structure of the
space unchanged are called symmetries. The symmetries of the plane are rota-
tion about a point (by any angle), reflection about any axis, and translation (or
displacement) in any direction by any distance. As shown in Figure 12.7, the
geometry of a shape remains the same under these transformations. For the same
reason, inertial transformations are symmetries within a neo- Newtonian space-
time, for they leave spacetime structures unchanged. As we shall see later, different
spacetimes have different symmetries (the Minkowski spacetime of special relativ-
ity (see 18.5) has Lorentz transformations as symmetries).
Velocity is distance travelled divided by time taken; since the notion of dis-
tance over time is undefined in neo- Newtonian spacetimes, there are no absolute
velocities in such spacetimes. However, there are absolute accelerations, and these
can easily be measured empirically. Suppose that we have an object that we have
reason to believe is accelerating (e.g. it is being boosted by a rocket motor). Select
some part of its trajectory as the starting point, and select an inertial frame of
reference for which the object is at rest at this time. Now select a later part of its
Q
P
Q
P
Figure 12.5 An inertial transformation. P and Q represent two distinct inertial frames of reference;
each frame consists of a family of (straight) parallel paths. On the left is a representation of spacetime that
shows the P-frame to be the zero-velocity state. On the right is an alternative representation of the same
spacetime created by altering the distance relations between hyperplanes in which the Q-frame is the
zero-velocity state.
o
Figure 12.6 Neo-Newtonian spacetime.
Every inertial frame can be transformed
into the zero-velocity frame by an
inertial transformation (or realignment of
hyperplanes). Since there are no distance
relations between hyperplanes, there is
no fact of the matter as to which of these
representations is correct. In a Newtonian
spacetime there are distance relations
between hyperplanes, and so one of these
representations will reflect reality, the
remainder will not.
Translation Reection Rotation
Figure 12.7
Symmetries of the
Euclidean plane.
MOTION IN SPACETIME 199
trajectory as the end point, and another inertial frame with respect to which
the object is now at rest. All we need do now is ascertain the relative velocity of
the second inertial frame with respect to the first. This gives the absolute velocity
change, i.e. a change of velocity that observers in all inertial frames can agree on.
If we now divide this absolute velocity change by the time separating the starting
and end points, we derive the average absolute acceleration of the object over this
part of its trajectory. The instantaneous absolute acceleration can be worked out
using the usual limiting process (Sklar 1977: 205).
The following points may firm up your understanding of neo- Newtonian space-
time:
3
This particle is moving at constant velocity makes sense in neo- Newtonian
spacetime: it simply means that the worldline is straight.
This particle has a speed of 10

cm

s
1
does not make sense. A particular speed
requires a definite time between two events and a definite spatial distance. In
neo- Newtonian spacetime the former is well defined but the latter is not. People
in different reference frames will disagree about what speeds objects have, and
there is no fact of the matter as to which groups are right or wrong.
This particle is at rest does not make sense.
This particle collided with that one makes sense: two particles collide if
their worldlines intersect.
This rope is straight makes sense: the claim is true if every temporal stage
of the rope is straight in ever y three- dimensional space (or hyper plane).
This particle has travelled 10

cm doesnt make sense: talk of spatial distances
over time makes no sense.
Particle A is moving faster than particle B does not make sense: in some
frames A will be at rest and B moving, in others B will be at rest and A will be
moving, and all the frames are equally legitimate.
Particle A is moving at 10

cm

s
1
relative to particle B makes sense.
12.3 The only reasonable view?
Neo- Newtonian spacetime seems very odd initially, since we are certainly accus-
tomed to assuming that where we are now is spatially related to where we were a
little while ago, but is it really all that strange? Some writers suggest not. Geroch
says: The Galilean [neo- Newtonian] view, although it takes a little getting used
to, is really a simple and remarkably natural attitude about how space and time
operate. It would be easy to be lulled into the position that it represents the only
reasonable attitude one could take (1978: 52). In a similar vein, Huggett writes:
Again, it might seem strange that some points have a spatial distance between
them when others do not, but points dont have to have every property that
we can think of. For example, we could define some point to be the centre
of the Euclidean plane, so that every point would be a definite distance from
the centre, but we dont. Similarly, we can define a spacetime in which there
is no distance between some points. Indeed, it seems plausible that spacetime
really is this way: How far in space are you from where you were in space a
200 TIME AND SPACE
minute ago? 1m? 1km? 1,000km? It is reasonable to think that such a ques-
tion has no answer. (1999: 194)
If you dwell for a few moments on the question How far in space am I now
from where I was a minute ago?, there is one thought that might occur to you
that would seriously undermine any lingering conviction you might have that this
question has an answer. The thought in question is this: Suppose it is the case,
as I sometimes think, that the past is unreal, completely and utterly gone from
existence If this is so, then of course it makes no sense to think that I am now
some distance away from where I was. There is an amusing twist here, for you
might well think, There is surely no metaphysic more hostile to this view of the
past than the four- dimensionalist way of thinking about space and time! How-
ever, while it may be natural to think like this, it is not necessarily accurate: a
spacetime of the neo- Newtonian sort, equipped as it is with a well- defined simul-
taneity relation, is not in itself incompatible with the view that only the present
is real.
It is interesting to note that the Minkowski spacetime of special relativity theory
also has no place for the notion of absolute velocity, although it, too, retains the
notion of absolute acceleration. But the reason is quite different. Whereas the
neo- Newtonian keeps a rigid framework of temporal relations (simultaneity and
succession) but abandons spatial distance over time, in Minkowski spacetime the
same effect (abolition of absolute velocity) is achieved by a different means: aban-
doning absolute temporal intervals. Since velocity is the measure of how much
space you are traversing in a given interval of time, the notion can be undermined
from two directions, so to speak. This fact leads Geroch to say:
we observe that the transition from the Aristotelian [Newtonian] view to the
Galilean view results in fewer things making sense. Although the evidence on
this is perhaps a bit scanty, it seems to be the case that physics, at least in its
fundamental aspects, always moves in this one direction. For example, rela-
tivity theory consists essentially of the realization that certain of the notions
in the Galilean view dont make sense either In quantum mechanics, to
take another example, such notions as the position of a particle or the
speed of a particle do not make sense. It may not be a bad rule of thumb to
judge the importance of a new set of ideas in physics by the criterion of how
many of the notions and relations that one feels to be necessary one is forced
to give up. (1978: 52)
The claim that advances in physics result in fewer things making sense can be
read in two ways:
as physics advances it produces worldviews that are increasingly utterly unin-
telligible; and
as physics advances, we find we can still make sense of the world despite the
fact that some of our most basic assumptions about how things work have
been refuted.
I leave it to the reader to judge which of these verdicts applies to neo- Newtonian
spacetime.
MOTION IN SPACETIME 201
12.4 A threat vanquished
Assuming that all this is at least reasonably clear (we shall be returning to the ques-
tion of just how neo- Newtonian space should be interpreted), we can move on.
As should by now be plain, the move to neo- Newtonian spacetime solves the
prob lem posed by the undetectablity of absolute velocity. Velocity is the measure
of how much distance is traversed in a given time; since points within successive
hyperplanes are not at any distance from one another, it no longer makes sense to
talk of absolute velocity, of how fast a body is moving relative to absolute space.
Of course, in practice this is not what happens. In practice, we can set up a
reference frame of bodies that are at rest with respect to one another, and from
the point of view of this reference frame we can measure the velocity of other
bodies. However, as we have just seen, there are other reference frames, compris-
ing bodies that are not moving relative to one another, from the perspective of
which our reference frame is moving. Given the existence of these other frames,
our frame has indefinitely many velocities. Since there is no longer any such thing
as absolute motion or rest, none of these inertial frames is privileged, and so there
is no frame- independent fact of the matter about what our velocity really is. The
only facts about velocities are facts about relative velocities.
So by situating Newtonian dynamics in neo- Newtonian spacetime the most
serious Leibnizian objection is overcome. Since absolute velocities are no longer
meaningful, the Newtonian is no longer guilty of positing innumerable physi-
cal states of affairs that are empirically undetectable and so indistinguishable.
The move to neo- Newtonian spacetime does not solve the problem of the static
shift. The Leibnizian could argue that, provided inter- object spatial relations are
retained, the whole system of material bodies could, in each hyperplane, be located
at a different and distant collection of spacetime locations. However, as we saw
earlier, the static shift argument only has bite for those in thrall to the principle of
sufficient reason, and these days few are.
However, we should pause before drawing the conclusion that Newton has been
fully vindicated by the move from Newtonian to neo- Newtonian spacetime. The
debate between the substantivalist (or absolutist) and relationist takes on a somewhat
different complexion in the neo- Newtonian context, as we shall now see.
12.5 The charge of explanatory impotence
Teller (1991) questions the extent to which the substantivalist neo- Newtonian
actually explains inertial effects. Teller suggests that in the context of standard
Newtonian space the substantivalists complaint against the relationist is best con-
strued thus:
We know that there are systematic connections between acceleration and
inertial effects. With only relative accelerations to which to appeal, relation-
ists have no way of distinguishing between the relative accelerations which
do and those which do not experience inertial effects. Thus they have no way
of giving a systematic description of the relation between acceleration and
202 TIME AND SPACE
inertial effects or of giving an explanation which such a systematic descrip-
tion might constitute. (1991: 370)
In response, the relationist can make Sklars move and bite the bullet by main-
taining that there is a primitive unexplained property (which we can call absolute
accelera tion if we like), which is systematically associated with inertial effects.
This meets the challenge of the substantivalist. The relationist can say that the
difference between relative accelerations that are, and are not, associated with
inertial effects is that the former possess the primitive property while the latter do
not. But this seems an unappealing move. The substantivalist will argue:
Sklars move is ad hoc. To say that inertial effects are always accompanied by
a primitive property, a property that is characterized solely as the property
that goes with inertial effects, amounts to nothing more than saying that iner-
tial effects are found where they are found. The substantivalist isnt restricted
to this truism, for the substantivalist can say that absolute accelera tion can
be characterized independently of inertial effects, in terms of motion with
respect to space (or spacetime). It is this independently characterized abso-
lute acceleration that is connected with inertial effects.
Teller now states his claim. This plausible sketch of an argument relies on a
slide from characterizing absolute acceleration relative to space, to characteriz-
ing absolute acceleration relative to spacetime. The argument is compelling in
the context of absolute Newtonian space, but not in the context of spacetime,
particularly neo- Newtonian spacetime. Since there is good reason for rejecting
Newtonian space in favour of neo- Newtonian spacetime, the substantivalists
argument against Sklar has no force, for it turns out that the neo- Newtonian sub-
stantivalist is in a position that is precisely analogous to that of Sklars primitive
property theorist.
The justification for this claim runs as follows. In Newtonian space the dis-
tinction between absolute and merely relative acceleration is real and clear: if
an object is located at the same point as time passes, it is at absolute rest; if
an object is located at different points at different times, such that equal times
are accompanied by equal changes in distance, and the changes in distance are
always in the same direction, then the object is in uniform absolute motion; and
all other motions constitute absolute accelerations. In neo- Newtonian spacetime
the situation is quite different: absolute velocity is not well defined and absolute
acceleration consists in having a curved worldline, rather than in change in abso-
lute velocity (that is, change of rate of motion through persisting substantival
spatial locations). So the notion of worldline curvature is key. But what does
this involve? Let us recall the manner in which neo- Newtonians construct their
spacetime.
We begin with a collection of spacetime points, organized into distinct Eucli-
dean three- dimensional spaces, one for each moment of time, and impose addi-
tional structure:
We specify which points are very close together; these count as next to one
another.
MOTION IN SPACETIME 203
We further specify what are to count as continuous sequences of spacetime
points: these are the paths or trajectories through spacetime continuous
paths that objects could follow.
We next specify the connection on the space; that is, we stipulate which
trajectories are straight, and thus which are curved.
We can now identify straight trajectories with inertial paths. Once the connection
is specified, we are provided with a distinction between absolute or true accel-
eration and merely relative or apparent acceleration. The problem now emerges.
Teller suggests that the substantivalist:
has an embarrassment analogous to Sklars. For let us ask, Which trajectories
are the ones to be called straight? That is, Which represent unaccelerated
motion? Or again, Which, if followed by an object, would describe the object
as unaccelerated? There is nothing in the collection of bare space- time points
and their next to relations to answer this question. The facts presented so
far do not enable a substantivalist to say more than that the trajectories to be
labeled straight are just the ones which will exhibit no inertial effects (ones
which, if followed by an object, describe motion free of inertial effects.) Such
a substantivalist has no more of an account of inertial effects than Sklars
relationist. (1991: 373)
This seems a sound point, for as Teller presents the situation, the only guide the
neo- Newtonian has as to which trajectories are to be construed (or made) straight,
and which curved, are which trajectories are associated with the production of
inertial effects. So, just as the Sklarist relationist can do no more than tell us that
the trajectories associated with inertial effects are the trajectories associated with
inertial effects, so too for the substantivalist. Consequently, it seems plain that the
neo- Newtonian substantivalist provides only a trivial explanation of the origin of
inertial effects.
12.6 A rebuttal
The underlying worry here, I suspect, is that the move to neo- Newtonian space-
time seems to have fundamentally altered what absolute acceleration amounts to;
since absolute velocity is no more, it can no longer consist in a change of rate of
motion relative to an underlying substructure. As a consequence, our intuitive
grasp on what absolute (as opposed to merely relative) acceleration consists in has
all but vanished. It is not easy to see what role substantival spacetime is playing
in the new scheme of things. However, while there is certainly some truth in this,
Tellers argument is suspect.
As Teller presents the situation, the substantivalist constructs a spacetime to fit
the bill (the observed interactions between bodies), and so in deciding which point-
sequences to count as straight he looks to the inertial forces. He has, it seems, no
other way of distinguishing spacetime point sequences that are straight or curved.
But this way of looking at things is potentially misleading: isnt Teller overlooking
the fact that for the substantivalist spacetime is a real entity, one possessing its own
204 TIME AND SPACE
distinctive structure? In constructing a neo- Newtonian spacetime the physicist is
constructing a mathematical model that is intended to represent this indepen-
dently existing entity. The mathematical model doesnt constitute space time; it is
merely a useful way of picturing it. From the standpoint of the substantivalist, the
physicist doesnt introduce inertial structure into spacetime, rather this structure
is already there, as it were, as a real feature of an autonomous physical entity. The
substantivalist can thus claim:
There is a real distinction in spacetime between straight and curved tra-
jectories, and inertial effects are encountered when objects follow curved
trajectories. So it is not the case that the substantival ist is in the same boat as
Sklar. Sklar can do no more than say that inertial effects are always accom-
panied by a primitive property, a property that is characterized solely as the
property that goes with inertial effects, which amounts to nothing more than
saying that inertial effects are found where they are found. The substantival-
ist isnt restricted to this truism: absolute acceleration can be characterized
independently of inertial effects, in terms of curved trajectories through neo-
Newtonian spacetime.
While this reply may go some way towards dispelling the doubts raised by Teller,
it doesnt go as far as it might. The question still remains, what exactly is the
substantivalist offering by way of an explanation of inertial effects? One answer
would run thus:
An object that suffers inertial effects possesses a distinctive physical property;
there is a law of nature that links the occurrence (and magnitude) of this
physical property with curved spatiotemporal paths.
Now, there is nothing in principle wrong with an answer along these lines, in
that it is certainly conceivable that there should be laws of this sort, laws that bring
it about that objects following certain types of trajectory through spacetime come
to possess intrinsic properties of a certain sort. However, I suspect that it would
be a mistake to view matters thus: there is no reason to think that inertial effects
involve intrinsic properties of a distinctive kind. Once this point is appreciated, it
becomes clearer just what the neo- Newtonian is offering by way of an explanation
of inertial forces.
What do inertial effects consist of? Typically, they consist of internal stresses
between the component parts of a body. Objects follow inertial paths unless
acted on by a force (which is created by gravity, magnetism, or an expenditure of
energy). When objects are forced off their inertial path, this is typically achieved
by applying force to one part of the object only, which sets up tensions within the
object: that is, some parts of the object start exerting forces on other parts. In the
rotating globes case, the tension is registered in the cord, which exerts a force on
the globes (whose inertial motions are tangential to their circular motion). When
a jet plane deviates from its inertial motion it pushes those parts of your body that
are in contact with it off their inertial paths and you feel this as pressure on your
skin. Since a similar motion is not imparted to the other parts of your body, a ten-
sion is created as your rear- parts push against your middle- parts (which is often
accompanied by a sensation of queasiness). All this is explicable without positing
MOTION IN SPACETIME 205
any special properties. The neo- Newtonians explanation of inertial effects is thus
quite simple: the internal stresses arise when absolute velocity differences arise. As
we saw above, to make sense of absolute velocity differences we have no need to
posit absolute velocities. The absolute differences are those that observers in all
inertial reference frames will agree on despite differing in their views as to who is
moving and who is stationary.
The notion of inertial paths lies at the heart of this account. What does the
neo- Newtonian have to say about these? Something like this:
certain paths through spacetime are privileged by the natural laws which gov-
ern the relationship between material bodies and spacetime;
the relevant natural law is that bodies continue to move along these paths
unless acted on by a force exerted by another moving body, or a locally gener-
ated impulse (e.g. jet engine), or a distant attractive/repulsive impulse (gravita-
tion, magnetism); and
these nomologically privileged paths are the inertial paths.
This explanation seems non- trivial, and eminently reasonable.
12.7 Newtonian spacetime relationism
I want to conclude by looking at another way in which the debate between rela-
tionists and substantivalists may be affected by adopting the spacetime perspec-
tive. In a recent article Maudlin has defended the following interesting claims: if
we adopt the spacetime perspective, in the context of Newtonian spacetime the
relationist is in a very strong position, since the argument from inertial effects is
completely without force; however, this changes once we move to neo- Newtonian
spacetime, which is an extraordinarily hostile environment for the relationist
(1993: 196).
Returning to Newtons buckets and globes and the argument from inertial
effects, Maudlin poses this question: exactly which sorts of relationism does this
argument work against? Well, suppose a relationist were to stipulate that the only
spatiotemporal relations are simultaneity, temporal succession and spatial distance
relations between simultaneous events. For such a relationist a Leibnizian rela-
tionist (as Maudlin puts it) the situations that the substantivalist describes as
involving different rotations are all exactly alike physically. They must be. Since
there is no spatial distance relation between states of affairs at different times,
there is no way of distinguishing a pair of globes at rest from a pair of globes that
are rotating. So, in a case in which the substantivalist explains different observed
tensions by reference to different rates of rotation, the relationist can say nothing:
there are simply the different tensions, which are just found to occur.
Introducing Newtonian absolute space solves the problem. Since absolute
space consists of points that endure through time, it provides a reference frame
that allows us to define distances between simultaneous and non- simultaneous
events. And this is the key point. Newtons ability to account for inertial effects
doesnt depend immediately on the positing of absolute space, but rather on the
206 TIME AND SPACE
extension of the domain of the (spatial) distance relation from simultaneous to
non- simultaneous occurrences. This sug gests a new form of relationism, which
Maudlin calls (for obvious reasons): Newton ian (spacetime) relationism.
The advocate of this doctrine rejects absolute space, and maintains that all
spatiotemporal facts are facts about the relations between material bodies, but
these relations now include a distance relation between non- simultaneous events.
Material objects are now conceived in the four- dimensional spacetime way: an
object consists of a succession of temporal parts (see 3.7). This change of view
gives the relationist invaluable additional resources. The Newtonian relationist
will say that a body is at absolute rest if the spatial distance between its successive
temporal stages at different times is zero, and an object is in absolute uniform
motion if its successive temporal stages are separated by the same spatial distance
per unit time. The stages of an object undergoing linear acceleration are separated
by different distances per unit time (think of a heap of placemats skewed to one
side). Rotation poses no problem: the particles in a rotating sphere have different
distance relations to one another at different times, even though they retain the
same distance relations to one another at any given time.
With these augmentations to his armoury the Newtonian relationist has no
trouble accommodating the bucket. In the stationary case, the successive bucket-
stages are aligned on top of one another in spacetime. In the rotating case they are
not: each stage is rotated to some degree with respect to its immediate predecessor.
Once all distance and direction relations between material objects are specified,
the entire system can be regarded as situated or embedded in a full Newtonian
space time, a four- dimensional continuum of points, some of which are occupied,
and many not. The relationist will not regard this spacetime as a real entity, but as
a fictional representation that is useful for purposes of calculation and prediction:
for, once the relational system is so embedded, the entire mathematical apparatus
that Newtonians have devised can be deployed by the relationist. The only real
spacetime locations that the relationist will recognize are those occupied by mater-
ial things and likewise for real spatial relations. The substantivalist, on the other
hand, will regard the material bodies and their spatial relations as constitut ing just
one part (or sub- model) of the totality of real spatiotemporal facts. This restriction
provides the relationist with a solution to the static shift (the problem of initial
location). For any given Newtonian spacetime S, regarded as a particular collec-
tion of (fictional) locations forming an infinite four- dimensional continuum, there
will be indefinitely many different ways to embed a given system M of material
bodies into S that preserve all spatiotemporal relations between the bodies in M.
For a start, M could be embedded at different locations in S. Alternatively, M can
be set at rest relative to S or given a uniform velocity, of any magni tude.
4
However,
since the embedding space is itself fictional, the relationist will not regard these
embeddings as representing physically distinct states of affairs: each embed ding is
simply a different way of representing one and the same physical state.
5
As Maudlin plausibly suggests, this option has not been noted by historical
figures, because it requires a four- dimensional perspective and an event ontology,
but it is a distinct position in its own right, and it shows that Newtons bucket
argument is not effective against relationism per se. Maudlin concludes, Relations
that are sufficiently rich can provide the means for explaining inertial effects.
MOTION IN SPACETIME 207
12.8 Neo- Newtonian spacetime relationism
Can the same trick be worked in the context of neo- Newtonian spacetime? Can
the relationist devise a satisfactory theory of dynamics if all that exists are material
objects whose successive temporal parts are linked by the spatiotemporal relations
recognized by the neo- Newtonian? Maudlin suggests not:
Neo- Newtonian spacetime does not have the complete four- dimensional
metric provided by absolute space. It results rather from the addition to
Leibniz ian spacetime of an affine connection which (roughly) specifies
which spacetime trajectories are straight, and hence inertial. So a reasonable
version of neo- Newtonian relationism would add to the Leibnizian relations
a three- place predicate col(x,y,z) which has as its extension all triples of non- -
simultaneous collinear events. That is, col(x,y,z) iff x, y and z all lie along some
inertial trajectory. Specifying col(x,y,z) completely in a spacetime would fix
the affine connection therein.
The relationist, however, is given as data only the restriction of col(x,y,z) to
the set of occupied points. This will, in general, be information too meagre to
specify an embedding of the occupied trajectories into the full spacetime up to
the freedom associated with choice of reference frame. For example, consider
two particles rotating about their common centre of mass. Until the first rota-
tion is complete, no triple of occupied event locations are collinear. Even after
any number of rotations, the collinearity relations among occupied points
will be consistent with any periodic rotation, uniform or nonuniform. So
while the Newtonian relationist could make use of the full power of Newto-
nian absolute space, starting only with the relations between occupied points,
embedding them in a fictional space, and deriving via Newtonian mechanics
exhaustive predictions about the future relations between occupied points,
the neo- Newtonian cannot perform the same trick. Knowing, for example,
that two particles remain at a constant distance through some period of time
and that no three points in their world- lines are collinear, the neo- Newtonian
relationist could not predict when, if ever, a triple of points would be col-
linear. Nor could inertial effects be predicted or explained, since the absolute
acceleration cannot be inferred from the data. (1993: 1934)
To simplify, suppose that we are dealing with two rotating particles in an other-
wise empty universe. Imagine the two particles as black dots on a thin sheet of
glass. In describing the evolution of this system the relationist has to do two things:
put the sheets into a particular order and provide them with an orientation with
respect to one another. The temporal ordering is unproblematic (it is a given in
neo- Newtonian contexts), but the orientation is not. If we draw an imaginary line
between the particles, the relationist can align the successive sheets in any number
of ways. If, say, the lines in successive sheets coincide vertically, the particles are
at rest; if just 1 degree separates successive sheets, they are rotating but slowly; if
the separation is 30 degrees, they are rotating more quickly, and so on. Now, since
there are no distance relations between non- simultaneous spacetime locations, it
seems that the relationist not only has no means of specifying what the correct
orientations are, but he cannot even recognize that the different orientations are
208 TIME AND SPACE
physically distinct states of affairs. The only fact about the relative orientations of
the slices available to the relationist is collinearity: the relation that holds between
particles that lie on the same inertial path. Given this, the relationist can tell when
a rotation is complete, but there are no facts about how fast the particles have
been moving in the intervening period. The Newtonian relationist is in a very
different position. Since there are distance relations between particles at different
times, there are facts about how fast the particles are moving from moment to
moment, and the behaviour of the system through the relevant period of time is
well defined. This is clearly not the case for the neo- Newtonian relationist.
The significance of this underdetermination is clearer for slightly more com-
plex situations. When astronomers observe the behaviour of triple star systems
they see the stars moving about in seemingly bizarre ways: the orbits are (seem-
ingly) highly irregular, with stars quickly moving together and shooting off again.
(Some executive toys approximate this behaviour, using magnetism rather than
gravity.) Although the behaviour of the three bodies is seemingly random, it is in
fact determined, and so (in principle, if not practice) predictable using Newto-
nian theory, provided one is given certain initial information, and this is when it
becomes interesting.
As Poincar showed, if we are given two snapshots of the system that show
the relative distances between the bodies, then, provided we know both the time
interval between the shots and the relative orientations of the snapshots, the future
behaviour of the system is determined. However, if we are not given the information
about the orientations, the future evolution of the system is not determined; see
Barbour (1999: ch. 5). Figure 12.8 shows just a few of the ways that such a system
could develop; the different evolutions show the effects of different orientations.
These different orientations correspond to different initial absolute accelerations.
As these diagrams make plain, facts about absolute acceleration play a crucial role
in the way a system of three mutually gravitating bodies behaves over time. The
Newtonian relationist can view such systems in just the same way as the Newtonian
substantival ist, since all the relevant distinctions are available to both. Not so for the
neo- Newtonian relationist: in the absence of spatial distance relations between suc-
cessive object- stages the absolute accelerations of objects is drastically underdefined.
In response to this problem the relationist could adopt an instrumentalist
strategy. Is there anything to prevent the relationist from using the substantival
Time Time Time
x
y y y
x x
Figure 12.8 A system
of three gravitationally
interacting bodies.
Different initial
accelerations lead to
dramatic differences later
on. This is a depiction
inspired by Barbour
(1999: 84).
MOTION IN SPACETIME 209
neo- Newtonian apparatus for predicting the behaviour of moving bodies, but
regarding this as nothing more than a fictional device useful for predictive pur-
poses? Suppose that, when mapping his spacetime, the relationist pretends to
be a substantivalist, and gets into the habit of using maps that display unoccupied
locations. When the map is complete (or as complete as it can be given the observ-
able facts), what is to prevent the relationists deleting from his map all the unoccu-
pied points and trajectories recognized by the substantivalist? What remains after
the deletion are worldlines; that is, straight and curved four- dimensional material
objects. At any given time, the spatial distances between the relevant temporal-
stages of these objects are well defined. So, too, are the intrinsic curvatures: either
zero, for straight worldlines, or non- zero for curved worldlines, with greater or
lesser intrinsic curvatures for greater or lesser absolute acceler ations. In trying to
measure absolute accelerations the neo- Newtonian relationist has exactly the same
observational data as the neo- Newtonian substantival ist: inertial effects found in
accelerating bodies along with at- a- time distance relations. Moreover, the relation-
ist can explain alterations in curvature in just the same ways as the substantivalist:
in terms of collisions (worldlines that meet) and action- at- a- distance forces (e.g.
gravity induces worldlines to curve towards one another). Hence it is not at all
clear that the unoccupied positions and paths that the substantivalist (or spacetime
plenist) posits are doing any real work. Look again at Figure 12.8. If we suppose
that the relationists world (a very simple one) is accurately represented by one of
these spacetime diagrams, given the data he has, isnt it highly plausible that he
could tell which one he lives in? More generally, it seems quite plausible to think
that, given equal amounts of data, the relationist and substantivalist could work
out equally accurate maps of their world.
This instrumentalist manoeuvre may seem innocuous enough, but closer
scrutiny reveals it to be dubious. The substantivalist provides a precise explana-
tion of the behaviour of the system in terms of Newtonian dynamics; according
to this theory, every particle, at every moment of time, has a definite absolute
accelera tion, a feature that plays an indispensable role in determining how the
dynamic system to which the particle belongs will evolve. Since the relationist
does not recognize the same range of dynamic features as the substantivalist (in
particular, the absolute accelerations of particles is often undefined), he cannot
regard the Newtonian account as true. But since the relationist has nothing to
offer by way of an alternative theory of remotely equivalent empirical power,
he has failed to show that the dynamical features he rejects are irrelevant to the
physics of moving bodies. Consequently, although the relationist is offering an
ontologically more economical account, we have no reason at all to accept his
account as true, and no reason to suppose that the discarded dynamical features
do not exist.
12.9 Relationism redux
Is this the end of the story? Not necessarily, for as Maudlin himself notes in con-
nection with Newtonian relationism To defeat one tribe of relationists is not to
vanquish the whole nation (1993: 187) for the simple reason that the predictive
210 TIME AND SPACE
and explanatory power of a relationist theory depends both on the nature of the
spatio temporal relations admitted and on the domain over which the relations are
defined (ibid.: 194). Newtonian relationism emerged as a serious rival to New-
tonian substan tiv al ism when the domain of the distance relation was extended to
include temporal parts of the same objects at different times. Might it not be that
the objections to neo- Newtonian relationism can be overcome if the relationist
embraces a richer collection of spatiotemporal relations?
Consider again the system consisting of two rotating particles in an otherwise
empty universe. It is quite true that in the absence of across- time spatial distances
(we are working in a neo- Newtonian context) the relationist cannot distinguish
between periodic rotations that are uniform and non- uniform, and so is incapable
of discrimi nating states of affairs that are physically distinct. However, the rela-
tionist can solve the problem by introducing distance relations among simultane-
ous objects that have a specific direction as well as a size. There are two options: we
can suppose either that distance relations have a relative orientation with respect
to one another over time, or that distance relations have an absolute orientation.
Since the latter eschews across- time comparisons it remains more faithful to the
basic tenets of neo- Newtonianism, and for this reason it is this option that I will
consider here.
For an intuitive picture of what is involved, start by imagining a motionless
sphere in Newtonian spacetime. We now stipulate that every straight line connect-
ing a point on the surface of the sphere to its centre both possesses and defines
a different intrinsic direction. Focus on just one of these lines, and imagine many
more with the same alignment, all originating at the centre of the sphere, but
extending outwards for different distances, up to infinity; this collection of lines
corresponds to all the potential spatial relations with the same intrinsic direction.
A similar operation on the other radii generates the remaining spatial relations.
Building a world is a matter of choosing different sorts of particles, plucking
distance relations from the sphere, and fitting them all together. When a distance
relation is removed from the sphere it retains its original direction and resists
reorientation. Since we are operating in a spacetime framework, strictly speaking
you are not connecting particles but rather particle- stages (momentary or very
short- lived), and each configuration of particle- stages constitutes a temporal slice
of your world.
Equipped with these newfangled distance relations we can reconsider the
two- particle system. If the particles are not rotating, their neighbouring (but not
simultane ous) temporal parts will be connected by a distance relation of the same
size and the same intrinsic direction. If they are rotating, neighbouring temporal
stages of the particles will be connected by distance relations of different intrinsic
directions. Should the rate of rotation be uniform, the angles between the distance
relations will be the same for unit intervals of time; in cases of non- uniform rota-
tions, the angular variation will be different for unit times.
Intrinsically directed distance relations solve some problems but not all. In Figure
12.9, A and B are two bodies in an otherwise empty universe moving closer together
in response to gravitational attraction; the curves represent the worldlines of these
bodies, and B is more massive than A. R
1
and R
2
depict two different ways these
worldlines could converge. Since in actual fact A is the less massive of the two, it is
MOTION IN SPACETIME 211
accelerated more than B, and so its worldline will be curved to a higher degree. R
1
is
therefore the accurate representation; R
2
misrepresents the situation, since it gives B
the greater acceleration. However, while the difference between these two scenarios
is both real and apparent on the page (which we can take to represent a Newtonian
substantival space), the difference cannot be captured in a neo- Newtonian relational
spacetime where only the following inter- object relations exist:
inertial worldlines the collinearity relation between particle- stages on iner-
tial trajectories; and
intrinsically directed distance relations between simultaneous space time particle-
stages.
Neither of these objects is on an inertial trajectory, and at any time the dis-
tance relations between the object- stages is exactly the same (cf. the dotted lines).
To capture the difference between R
1
and R
2
we need to introduce an additional
property: different degrees of instantaneous intrinsic curvature.
6
As soon as
the relationists augments his armoury with this property, the difference between
states of affairs such as those depicted above is readily captured: the worldlines of
A in R
1
and R
2
have different intrinsic curvatures at each moment of time.
I may have said enough to make this conjecture plausible: the neo- Newtonian
relationist who embraces both intrinsic direction and different degrees of intrinsic
worldline curvature has the resources to accommodate all the dynamically relevant
features to be found in both relational and substantival Newtonian worlds. Of
course, there is no need for the relation ist to recognize every embedding of a given
system into full Newtonian substantival space as physically distinct. For exam-
ple, the normal and upside- down embedding of our world into Newtonian
spacetime may well be physically indiscernible. Also, it may be that, with intrinsic
curvature on board, the relationist can dispense with the services of intrinsically
directed distance relations. Since a pair of particles rotating at a uniform rate have
differently curved worldlines than a pair at rest, or a pair in non- uniform rota-
tion, we only need directed distance relations if we want to be able to distinguish
rotations of the same sort in different directions. We may not wish to regard such
a difference as physically real and, even if we do, we may be able to capture it in
a more economical fashion by positing inherently directed curvatures. But I will
not attempt to explore these matters further. I introduce them only to illustrate
T
i
m
e
Distance
R
1
R
2
A B A B
Figure 12.9 Distance and intrinsic
direction are not enough.
212 TIME AND SPACE
that Maudlin may have underestimated the force of his own point: a relationist
who is prepared to countenance sufficiently exotic species of spatial properties can
reasonably hope to have available all the dynamical distinctions that are necessary
for Newtonian theory.
Another lesson has emerged from the discussion thus far. Shifting to the stand-
point of spacetime improves the prospects for the relationist; perhaps the geo-
metrical way of conceiving motion from this standpoint makes it easier to develop
viable forms of relationism. In any event, bearing in mind the resources available in
the spacetime framework, one might be tempted to say that, for systems of material
bodies conform ing to Newtonian laws of motion, purely dynamical considerations
do not weigh decisively in either direction: substan tivalists and relationists may
both be able to provide viable accounts of forced and unforced motions and their
associated inertial effects.
7
If this is so, then if the issue between the relationist and
substantival ist is to be resolved, we must look further afield. Perhaps recent scien-
tific discoveries funda mental ly change the picture; even if the two conceptions of
space (or spacetime) are equally viable in the context of Newtonian physics, the sit-
uation may well be very different when (say) electro magnetism or relativity theory
enters the picture. Or then again, perhaps only purely metaphysical arguments can
decide the issue, or at least tilt the balance. I have in mind questions such as these:
What are these distance relations that the neo- Newtonian (and Newtonian)
relation ist simply assumes?
How can these distance relations conspire to control the behaviour of bodies
in inertial and non- inertial motions?
Is a theory that posits vast numbers of relations any simpler than one that
posits a substantival space or spacetime?
We will be turning to some of these questions shortly. However, lest it be thought
that the interest of the issues we have been dealing with is entirely historical, the
following point is worth bearing in mind. I noted earlier that the Minkowski space-
time of relativity theory is at once similar to and different from neo- Newtonian
spacetime. It is similar because the concept of absolute velocity has no role, and
different because of the reason for this: in neo- Newtonian spacetime there are
no spatial distances between non- simultaneous points, and in Minkowski space-
time there is no absolute simul taneity. Recent writers have drawn attention to the
fundamental incompatibilities between relativity and quantum theory. One such
incompatibility is that quantum theory may be committed to absolute simultaneity,
whereas relativity certainly is not. If, as many believe, the theory that synthesizes
relativity and quantum theory will retain the basic character of a quantum theory,
it may be that absolute simultaneity is about to make comeback.
8
Since there is
no reason to believe that absolute velocity will do the same, might it not be that
neo- Newtonian spacetime better reflects the actual character of our spacetime
than any of its current competitors?
13 Curved space
13.1 New angles on old problems
We may have pursued the classical debate on space and motion a good way beyond
its original boundaries, but there are further modifications to the classical world-
view of a still more fundamental kind that we have yet to explore. Viewing motion
as a process unfolding in spacetime does not require or involve the abandonment
of the classical conception of space as a three- dimensional Euclidean structure,
and the various spacetimes we have considered thus far retained this conception
of space: there are no spatial distances over time in neo- Newtonian worlds, but
the spaces- at- times that remain are entirely Euclidean. The spacetime perspective
came into its own when Minkowski gave a spacetime interpretation of Einsteins
special theory of relativity in 1908, but significant advances in the understanding
of space advances that play a crucial role in Einsteins general theory of relativ-
ity had already occurred. By the middle of the nineteenth century it had become
clear to mathematicians that space can take different forms and that the structure
of Euclidean space is just one spatial structure among many.
This discovery gave rise to new questions: what sort of space do we live in and
how can we find out? It also transforms the debate between substantivalists and
relationists, as we shall see in Chapter 14. However, before entering these debates
we need to know something about these strange non- classical, non- Euclidean
curved spaces, and this chapter is devoted to this end (and so can safely be skipped
by those already familiar with the basics).
13.2 Flat and curved spaces
In 9.2 we encountered Flatland, a two- dimensional plane inhabited by two-
dimensional (but nonetheless intelligent) creatures. This imaginary world was
used to undermine a deep- seated assumption that we naturally make and never
normally have reason to question: that space, even nothingness itself, is necessarily
three- dimensional. This same imaginary world is equally useful in undermining
another basic assumption, namely that space is necessarily Euclidean. The Flat-
land space of 9.2 was truly flat: not only smooth and two- dimensional, but also
lacking curvature. If by a Flatland space we mean any two- dimensional plane
surface, not all Flatland spaces are flat in the latter sense. A tablecloth is a two-
dimensional surface, and is flat when spread out over a table, but once picked up
214 TIME AND SPACE
and rumpled there are any number of different and differently curved shapes that
it can adopt. A Flatland space (or part of one) could have any of these more exotic
non- flat shapes.
To make matters more concrete let us consider in more detail just one instance
of a curved two- dimensional space, a Flatland shaped like the surface of a sphere.
As previously, the Flatlanders live out their lives within their two- dimensional
space and are entirely unaware of the third dimension, which they cannot move
into (it is as if the interior of the sphere does not exist). Would the Flatlanders be
aware that their space is curved? Not necessarily. Our Earth is curved, but because
it is very large (in comparison with us) this isnt immediately obvious. Viewed
from the surface it looks more or less flat. As ships sail away they do eventually
disappear over the horizon, but it is easy to put this down to the fact that things
look smaller as they get further away. However, this points to a potential giveaway.
A Flatland explorer intent on discovering the edge of the world, who sets out in
a particular direction and sticks to it, moving always in a straight line never
deviating to the right or to the left would eventually find himself back where he
started, having gone right around his space. After a number of such excursions,
from different points of departure and in different directions, the Flatlanders
would be in no doubt as to the shape of their world.
This result could be confirmed by other experiments. On a flat surface the
internal angles of a triangle always add up to a total of 180 degrees. In the special
case of a right- angled triangle, one angle is 90 degrees, and the other two are both
of 45 degrees. As can be seen below, this is not the case for triangles inscribed on
the surface of a sphere. If two Flatlanders were to set off from the North Pole trav-
elling at right angles to each other, and continue until they reached the equator,
and then turn another 90 degrees and walk along the equator until they met, they
would trace out a triangle whose internal angles add up to a total of 270 degrees,
since each internal angle is 90 degrees. A triangle of this sort cannot exist in flat
two- dimensional space. The deviation from the 180 degree two- dimensional norm
is not always so large. A triangle drawn within the boundaries of the one depicted
below will have internal angles somewhere between 180 and 270 degrees, and the
smaller the triangle the closer it will be to 180 degrees; a fact that illustrates the
more general truth that the geometry of curved surfaces more closely resembles
that of flat ones over short distances. Small areas of a sphere are flatter than larger
areas. But there are also more extreme deviations. If the Flatlanders had set off
from the North Pole walking in directions separated by 130 degrees, they would
trace out a triangle whose internal angles add up to 130 + 90 + 90 = 310 degrees.
N
Figure 13.1 Sphere geometry. The internal angles of a
triangle on a flat plane always add up to two right-angles (or
180 degrees). It is very different on the surface of a sphere.
CURVED SPACE 215
There are further important differences between the geometry of spherical
and flat two- dimensional spaces. The Flatlanders are, we can suppose, perfectly
familiar with the geometry of circles on flat (or very nearly flat) surfaces: they
know that the circumference of such a circle is its diameter multiplied by pi (t =
3.14159). Suppose that one Flatlander decides to measure the circumference of
Flatland space by travelling right around the equatorial circle depicted in Figure
13.1. After a trip lasting some years he returns to his point of departure. Another
Flatlander decides to measure what is (in effect) the diameter of the equatorial
circle. She starts from the same place as the first explorer, but heads directly north,
crosses the North Pole, and continues in the same direction until she hits the
equator again. She then returns to base, where she makes a curious discovery: on
multiplying the length of the diameter by t, she obtains a result for the length of
the circumference that is greater than its actual length (as measured by the other
Flatlander). On reflection the reason for the discrepancy is obvious: the trip over
the North Pole is considerably longer than the trip through the sphere to the same
destination. Of course, the latter route, which corresponds to the diameter of the
equatorial circle in flat two- dimensional space, does not exist in the curved space
that we are currently considering. Hence the ratio of the diameter of a circle to
its circumference on the surface of a sphere is always less than t, a figure that
only gives the correct result for circles and diameters inscribed on flat surfaces. As
with the internal angles of triangles, the discrepancy from the flat norm is not
constant: it is smaller over smaller areas.
Parallel lines are straight lines that never meet, no matter how far they are
extended. In a flat two- dimensional space, parallel lines not only exist, but there
are plenty of them. For any given line, there is an infinity of others, at different dis-
tances both and above and below it, that are parallel with it. Does the same apply
in our Flatland sphere- space? It might seem so. Are not all the (dotted line) circles
in Figure 13.2a parallel with the equatorial circle? They certainly do not intersect
the latter, and to this extent they share a feature of parallel lines on a plane. But
parallels are straight lines that never meet, and there is a case for supposing that
the dotted rings in Figure 13.2a are not straight lines in Flatland sphere- space.
Someone inclined to literal mindedness might reasonably argue:
The notion of a straight line is simply inapplicable in a curved Flatland: if we
take straight to mean what it does in a flat space then all the lines on the
surface of a sphere are curved to some extent, and so not straight. And since
there are no straight lines, there cannot be any parallels either.
(a) (b)
X
Y
Figure 13.2 Unobvious parallels.
216 TIME AND SPACE
This may be going too far. Although there are no lines on a sphere that are exactly
like their flat kin, might there not be some lines that are analogous to ordinary
non- curved straight lines in important respects, and in virtue of this similarity
deserve to be regarded as the counterparts of straight lines in sphere- space?
There are indeed. A straight line in flat space is the shortest possible distance
between two points. If we apply this criterion to the surface of a sphere we find
the following: for any two points, the shortest line between them is a line that
forms part of a great circle. Each of these great circles is a circumference of
the sphere- space: the equator is one great circle, and so is a circle that joins the
North and South Poles, but there are infinitely many others (see Figure 13.2b).
A simple experiment reveals as much. A reliable way to find the shortest distance
between two points on a sphere, X and Y, is to take a solid globe, pin a short
length of rubber band to a point corresponding to the location of X and stretch
the band to a point corresponding to the location of Y; the band will strive to find
the shortest route between the two. Will it follow the horizontal ring? Well, try
it (at least in your imagination). If you pull the band from X in the direction of Y
horizontally, you will find that it immediately starts slipping upwards, only to stop
when it reaches the North Pole: this great circle (running between the two poles)
is the shortest route between X and Y. (This is why airliners flying between North
America and Europe often take a northerly route rather than flying directly
across the Atlantic the route north is actually shorter.) We can now see that our
literal minded objectors conclusion is true, even if not for the stated reason: there
are indeed no parallel lines in our Flatland sphere- space, but not because there are
no straight lines. There are lines that are analogous to straight lines in this space,
but these are all paths along great circles, and when these paths are extended as
far as possible, they all intersect, as can be seen in Figure 13.2b.
To sum up, a two- dimensional spherical space differs from a flat space in three
key respects:
the internal angles of triangles add up to more than 180 degrees;
the ratio of the diameter of a circle to its circumference is less than t; and
there are no parallel straight lines.
Interestingly, there is another sort of Flatland space where each of these results
obtains in reverse, where:
the internal angles of triangles add up to less than 180 degrees;
the ratio of the diameter of a circle to its circumference is more than t; and
there are many more parallels than there are in flat space.
One instance of a space with these properties is depicted in Figure 13.3. We will
learn more about the characteristics of spaces of this type shortly, but as is intuitively
Figure 13.3 Negative curvature. The internal angles
of a triangle add up to less than 180 degrees.
CURVED SPACE 217
clear, a triangle inscribed on the two- dimensional surface of this object will have
sides that bulge inwards (the opposite of what happens on the sphere- surface, where
they swell outwards), and so its internal angles add up to less than 180 degrees.
In considering these curved Flatland spaces we have an advantage over the
Flatlanders themselves: looking down from the vantage point provided by the
third dimension we can see that the spaces are not flat. Confined as they are to
their two- dimensional space, the Flatlanders cannot do the same. But as we have
also seen, even so confined, there are ways the Flatlanders could learn about the
shape of their space. Provided they know some of the basics of plane geometry,
they could soon discover whether their space is flat or not by measuring distances
and angles. There is nothing to prevent our doing the same, in order to find out if
our space is curved. Once again we have not been considering Flatland for its own
sake, but for the light it sheds on our predicament. Since the Flatlanders visual
field is restricted to a straight line, and their visual imaginations likewise, they
can neither perceive nor imagine a curved two- dimensional space. In analogous
fashion, we cannot visualize a curved three- dimensional space. Just try imagining
a three- dimensional volume that extends the same distance in all directions but
nonetheless curves back in on itself in the same way as the surface of a sphere. But
a space of this sort is mathematically possible. In two- dimensional space there are
circles, in three- dimensional space there are spheres, and in four- dimensional space
there are hyperspheres, and the surface of a hypersphere is a three- dimensional
surface that curves back on itself in just the same way as the two- dimensional
surface of an ordinary sphere. If our space were of this form, a four- dimensional
being could look down on us and see that our space is curved, in just the way that
we look down on the spherical surface that constitutes Flatland.
There is, however, one important respect in which the Flatland cases can mis-
lead. The curved Flatlands we have considered have all been embedded in three-
dimensional space, but while viewing them thus makes their curvature undeniably
apparent, it would be a mistake to suppose that curved spaces of n dimensions
can only exist within a space of n + 1 dimensions. Mathematically speaking this
is not so: the higher- order embedding space need not exist. It may be that a rec-
ognizably physical universe of just two dimensions, without any thickness at
all, is an impossibility, in which case the embedding in three- dimensional space is
not an optional extra. But there is no reason whatsoever to suppose that a three-
dimensional space, whether curved or flat, can only exist when embedded in a
four- dimensional space. This point, that a surface (or space) can be curved without
having to be curved in anything else, is of crucial importance. It means we can make
sense of the idea that our own space might be curved without having to suppose
that our space exists as a layer in some higher- dimensional space. And the same
applies to the notion of a curved spacetime. Although spacetime itself has four
dimensions rather than three (the additional dimension being time), the curved
spacetime of (general) relativity theory is regarded as a self- contained entity, one
that is not embedded in any higher- order spatial (or spatiotemporal) medium. The
sort of curvature that can exist independently of an embedding space is known as
intrinsic; we will be looking at this notion in slightly more detail in 13.4.
Contemporary cosmologists often tell us that the universe is expanding, a claim
that is not particularly puzzling in itself, but that quickly becomes puzzling when
218 TIME AND SPACE
they also tell us that not only is space itself expanding (as opposed to the objects
within it), but also that irrespective of the direction in which astronomers point
their telescopes they find that distant galaxies are invariably moving away from us,
at speeds proportional to their distance (the more distant the galaxy, the faster the
motion). It would be easy to see how this could be the case if we were located at
the centre of the universe the fragments from a bomb blast all move away from
the site of the explosion but the cosmologists also tell us that the Sun is situated
in a nondescript suburb of the Milky Way, which is itself a perfectly ordinary
galaxy whose location is in no way distinctive or distinguished. It is not obvious
how to make sense of these cosmological claims in the context of a space that is
both infinite and flat, but, as should by now be clear, they make perfect sense for a
three- dimensional space that is finite and has a shape analogous to the surface of a
two- dimensional sphere. Imagine a balloon covered in dots (the galaxies) which is
expanding in size: from the vantage point of each dot all the others are receding,
and the more distant the dot the faster the rate. If our universe is of this type, our
three- dimensional space has the form of expanding hyperspherical surface.
13.3 The fifth postulate
These ideas may be mind- boggling when first encountered, but they are not par-
ticularly difficult to grasp, so it comes as something of a surprise to find out that
it was only in the nineteenth century that physicists and mathematicians started
to take seriously the idea that our space could be curved. The main reason for this
lies in the history of geometry, and the spell cast down the centuries by the work
of Euclid.
Many ancient peoples in particular the Egyptians and Babylonians had put
together rules for calculating areas and volumes (on the basis of lengths), as these
were handy for doing lots of practical tasks (such as allocating land and construct-
ing buildings). The rules were generalizations based on practical experi ence. For
example, the Egyptians knew that if you make a triangle out of pieces of wood of
lengths 3, 4 and 5 units then the triangle will contain a right angle. Around 600BC,
Thales of Miletus travelled to Egypt, learned something of their geometry, and did
something quite new with it: he regarded geometric statements not as empirical
generalizations, but as provable propositions, and showed that it was possible to
logically deduce complex geometrical truths from more basic principles. Progress
was swift (at least compared with the pace of later develop ments). Euclids Ele-
ments were put together around 300BC, and contained a masterly synthesis of
geometrical results, many discovered by others, some by Euclid himself. The hun-
dreds of derived results fill thirteen books. What particularly impressed Euclids
contemporaries, and many subsequent genera tions, was the structure of the work:
large swathes of plane (and some solid) geometry were logically deduced from a
select handful of simple and largely self- evident axioms. Euclid himself may not
have presented his work as having direct application to the physical world, but it
was not long before people were working on the assumption that his geometry
describes the structure of physical space. This view was far from unreasonable.
Geometry is useful, in surveying, navigation, engineering and farming, and it
CURVED SPACE 219
provides us with knowledge about how the things in the world are related to one
another by virtue of being located in space.
Viewed as an intellectual accomplishment, whether in science or pure math-
ematics, the Elements was unmatched in sophistication in its day and remained
so long after. There are errors in Euclids work, but these were mostly minor,
and were detected and corrected over the subsequent centuries; the plane geom-
etry taught in schools nowadays is essentially the same as in 300BC.
1
Later work
in astronomy and mechanics such as the Almagest of Ptolemy presupposed
Euclidean geometry; Newton himself compared the mechanics of his Principia
to the Elements, suggesting that the two works were on a par in method, scope
and significance. The influence on philosophy was equally profound. Plato had
already taken geometry as a paradigm for all knowledge, and concluded not only
that we could discover ultimate truths by reasoning alone, but that the objects of
these truths would be perfect and immaterial entities, of which Euclids circles
and triangles can be seen as prime examples (Plato pointed out that a truly perfect
circle is never to be found in the natural world). The rationalist founda tional ism
of Descartes is a more recent example of the same line of thought. Although
Kant denied that we could know anything of reality in itself, questioning Euclid
never occurred to him: he insisted that the principles of Euclidean geometry were
hardwired into our minds (or whatever aspect of noumenal reality is associated
with our capa cities for thought and perception), so that even if space itself is not
necessarily Euclid ean we cannot possibly conceive of such a thing.
The status of the Elements as a cornerstone of Western intellectual life is one
reason why it remained unquestioned for so long, but by no means the only one.
The handful of axioms and postulates from which Euclid derived his geometrical
results are such that it is not easy to see how any sane person could deny them,
and, given the deductive character of the Elements, if the basic principles are true,
so is the rest of the system. Euclid starts off with twenty- three definitions, none
of which is controversial; for example, 1. A point is that which has no part. 2.
A line is breadthless length. 3. The extremities of a line are points . Following
these are five axioms (or common notions), all of which seem self- evidently true:
1. Things equal to the same thing are equal to each other.
2. Equals added to equals yields equals.
3. Equals removed from equals yields equals.
4. Coincident figures are equal to one another in all respects.
5. A whole is greater than any of its parts.
Next come the five postulates:
1. Two points determine a straight line.
2. A straight line may be extended in either direction.
3. About any point a circle of a specified radius exists.
4. All right angles are equal.
5. From a point outside a given line, one and only one line can be drawn in a plane
which does not intersect the given line, no matter how far it is extended.
2
There is nothing obviously suspect here either but, if you grant Euclid this much,
you must grant him the rest.
220 TIME AND SPACE
Of the five postulates the fifth, the so- called parallel postulate, is by far the
most complex. In Figure 13.4 the lower line is the given line, the X the point
outside it, and the line with arrows, being perfectly parallel with the lower line,
will never intersect it, no matter how far either is extended. How could there be
another line running through X that is also parallel with the lower line?
Figure 13.4 Euclids fifth postulate.
However, although no one doubted its truth, the fifth postulate was regarded
with some wariness. Euclid himself only relied on it when he had to. He proved
his first 26 propositions without using it at all. Noting its relative complexity, later
geometers suspected that it might be redundant, and so tried to show that it could
be deduced from the first four. All such attempts failed, and there were many such
attempts.
In 1733 an Italian Jesuit, Girolamo Saccheri, tried to prove the parallel postul-
ate by assuming it to be false, and deducing an absurdity from this assumption and
the other four postulates (and the five axioms). This form of reasoning is familiar
to us from logics reductio ad absurdum: if you can deduce a contradiction from
proposition P, then you can conclude that not- P is true (on the assumption that
your other premises are true). Saccheris proof had two parts. He first assumed
that no parallel lines exist at all and deduced a contradiction; he then assumed
that more than one parallel line exists and deduced a contradiction. He thought
that he had vindicated Euclid and published the proof under the title Euclid Freed
from Every Flaw. In fact his work had the opposite effect.
In the second part of his proof Saccheri demonstrated that a good many propo-
sitions that seemed false can be deduced when the many parallels postulate is
assumed. When other mathematicians carefully examined the proof they found
that, although Saccheris propositions were counter intuitive, they did not in fact
harbour a logical contradiction. Euclid was tottering. By the start of the nine-
teenth century several mathematicians Gauss, Bolyai and Lobachevski came
to the same conclu sion, more or less at the same time: if Euclids fifth postulate
is replaced by the alternative many parallels postulate, and combined with the
other nine basic assumptions, it is possible to develop a consistent geometrical
system. That there was an alternative to Euclidean geometry was a deeply shocking
result, so much so that Gauss, the pre- eminent mathematician of his day, never
published his results, fearing the clamor and cry of the blockheads.
3
But it soon
got worse. Just a few years later Riemann discovered that the assumption that
there are no parallels to a given line also fails to generate a contradiction, and
he developed yet another new geometry on this basis. Logically speaking, these
alternative geometries were on a par with Euclids. It couldnt be proved that any
of the systems was perfectly consistent, so it remained possible that any one of
them contains a concealed contradiction, but it was proved that if any one of the
geometries contains a contradiction, then they all do. So, if a non- Euclidean sys-
tem should yield a contradiction, then so will Euclids. This relative consistency
proof destroyed the logical priority of Euclidean geometry.
CURVED SPACE 221
We have already encountered Riemanns non- Euclidean plane geometry. One
possible realization of it is the Flatland space constituted by the surface of a sphere,
so this geometry is often called spherical or (more often) elliptical. Of the
various theorems, the following are of special interest:
1. Through a given point external to a line no non-intersecting lines can be drawn.
2. The internal angles of a triangle add up to more than 180 degrees; as the area
of a triangle approaches zero its internal angles approach 180 degrees.
3. The ratio of the circumference of a circle to its radius is less than 2t.
4. All straight lines are of the same finite length when fully extended.
The alternative Lobachevskian (or hyperbolic) non- Euclidean geometry yields
very different theorems:
1. Through a given point external to a line an infinite number of non-intersecting
lines can be drawn.
2. The internal angles of a triangle add up to less than 180 degrees; as the area
of a triangle approaches zero its internal angles approach 180 degrees.
3. The ratio of the circumference of a circle to its radius is more than 2t.
4. Straight lines are infinitely extendible.
One possible realization of a hyperbolic geometry is a flat but circular plane in
which objects become smaller as they move further from the centre, shrinking to
infinitesimal size as they approach the outermost boundary. Such a world would
seem infinite to its inhabitants: they can never reach its edge, for as they approach
it distances (as measured by them) get larger and larger. The geometrical proper-
ties of the space will be the same (when measured) by all occupants anywhere
within the space. Straight lines are (in effect) curves that converge on the boundary
at right angles. This boundary should not be thought of as being in the space at
all, and neither should the Euclidean space surrounding the disc; it is only present
as an artifact of our representa tion. (Some of Eschers fish pictures depict worlds
with this geometry.) The surface of the object depicted in Figure 13.5 is another
realization of a space possessing Lobachevskian geometry.
Figure 13.5 A surface of constant
negative curvature: the pseudo-sphere.
222 TIME AND SPACE
13.4 Intrinsic curvature
In two papers of 1825 and 1827 Gauss showed how to measure the intrinsic
curvature of a plane without referring to any external embedding space. Riemann
subsequently extended the method, and laid the basis for our contemporary math-
ematical treatment of n- dimensional manifolds, outlining his main ideas in the
now famous 1854 Gottingen lecture On the hypotheses that underlie geometry.
The informal sketch of the Gaussian approach that follows is primarily intended
to introduce some of the terminology used in the philosophical and scientific
literature.
For any point P on a one- dimensional curved line it can be shown that there
is a unique circle that most closely matches the curvature of the line at P. Circles
with greater or smaller radii match the curvature of the line less well. The letter
k stands for the Gaussian measure of the curvature of the line at P; the value of
k is given by the reciprocal of the radius R of the best- matching circle: k = 1/R.
The same procedure can be extended to curved two- dimensional surfaces. It
can be proved that all the lines running through a point P on such a surface have
tangents that lie on the same plane: the so- called tangent plane. Every point on a
curved surface has a unique tangent plane, and each tangent plane has a unique line
that is perpendicular to it and that passes through the relevant point on the sur-
face: the so- called normal line. Returning to our point P, any plane that includes
Ps normal line will be perpendicular to Ps tangent plane, and slice through, or
section, the surface; this normal- section will be a curved one- dimensional
line, and since we already know how to determine the curvature of such a line,
we can determine the value of k for it. If we now take all the normal- sections (e.g.
by rotating our initial section) through P, and find their curvatures, we will have
completely determined the curvature of the surface at P. This is a complicated busi-
ness, but Gauss showed how it could be simplified. He proved that for any point
there will be two normal- section curves, one with a unique maximum curvature
and the other with a unique minimum curvature, and moreover that the planes
containing these sections will be at right angles to each other. These two principal
curvatures, k
1
and k
2
, completely determine the intrinsic curvature of the surface
at the point in question. The Gaussian curvature, K, at P is given by the product
of the principal curvatures: K = k
1
k
2
. Although particular values for k
1
and k
2

depend on the units of measurement used, the magnitude of their product K does
not; Gaussian curvature is thus an invariant and intrinsic feature of a surface.
We can now assign signs to our curvatures by arbitrarily calling one direction
of a normal line N positive and the other negative. A normal- section that is
concave (or bends towards) the positive direction of N is positively curved, and a
normal- section that does the reverse is negatively curved. If k
1
and k
2
for a given
Figure 13.6 For any point on a curved line, one
circle matches the curvature at that point more
closely than any other circle.
P
CURVED SPACE 223
point both have the same sign, whether positive or negative, their product, K, will
be positive (a negative multiplied by a negative yields a positive), and the surface
close to the point will be curved like a sphere or a bowl. If k
1
and k
2
have different
signs, then K will have a negative value, and the surface close to the point will be
curved like a saddle- back. If either one of k
1
or k
2
is zero, K will also be zero, and
in this case the surface has zero curvature in the neighbourhood of the point in
question, and the region has the same shape as the Euclidean plane.
These concepts allowed us to place the axiomatic geometries that we met earlier
in a new perspective: Euclidean geometry is the geometry of a plane whose cur-
vature is everywhere zero; hyperbolic (LobachevskyBolyai) geometry is that of a
surface of constant negative curvature; elliptical (Riemannian) geometry is that of
a surface of constant positive curvature.
A few examples will help to make all this clearer. Think of a sphere, such as the
Earth, and take a point P on the equator. Since the surface curves equally in all
directions, we can take for our principal curves any great circles running along the
surface that pass through the point, such as the equator itself, and the line running
through P and the North and South poles. Call the radius of these circles R, then
K = 1/R 1/R. Now imagine a sphere of infinite size. Its surface is effectively
flat, and the radius of all principal curves on it will be infinite: K = 1/ 1/
= 0. This is why the Gaussian curvature of a flat Euclidean plane is zero. Now
imagine a horizontal torus (e.g. a doughnut on a tabletop), and think of the surface
inside the hole. There are two relevant curves here: one vertical (curving upward
and downward from the central ring) and one horizontal (going round the inner
surface). In this case the curves bend in opposite directions, and will have differ-
ent signs, one positive, and one negative. K = k
1
+k
2
, and so is itself negative.
Note also that the curvature on a torus is posi tive on its outer surface, since both
circles bend in the same direction.
More intriguing is the case of a piece of paper that is bent into a gentle smooth
curve (even a cylinder). Along one direction the paper is straight, and so the (maxi-
mum) radius of curvature is 0, but in the other direction the radius of curvature,
R, will be finite. So the measure of curvature is 0 R = 0. Hence we see that the
Gaussian curvature of a tubular piece of paper is the same as a flat piece of paper!
A Flatlander confined to this plane who went about measuring the internal angles
of triangles would get exactly the same results as a Flatlander living on a flat plane.
More generally spaces with the same Gaussian curvature can possess the same
geometrical properties, even if intuitively the spaces are very different.
Figure 13.7 Appearances can
mislead. The plane and the
cylinder have surfaces with the
same intrinsic geometry. Although
the global properties of these
spaces are different, this would
not be apparent locally.
224 TIME AND SPACE
In more formal terms, the properties of a space, S, that could (in principle) be
discovered by a geometer working within S, without any knowledge of the sort
of space S is embedded in (if any) are intrinsic features, whereas those features of
S that require knowledge of the embedding space are extrinsic. The case of the
cylinder illustrates the fact that a two- dimensional surface with the same intrinsic
features can be embedded in three- dimensional space in different ways; the dif-
ference is evident from the perspective of the third dimension, but invisible from
within the space itself. As for curvature, while it is easier to grasp what K involves
by considering two- dimensional planes from the perspective of the third dimen-
sion, which is what Gauss himself did, the higher- order embedding space is in
fact inessential: the curvature of a given surface is exactly the same irrespective of
whether the surface inhabits a space of higher dimensions. And although the value
of K can be determined from the vantage point of a higher- order dimension (e.g.
by positioning circles on the surface), it can also be determined from within the
space itself, by measuring distances and angles.
Of course, a Flatlander living in a cylindrical space could discover that his space
had a different overall structure from an infinite Euclidean plane by successfully
travelling all the way around it in a straight line and arriving back where he set
off. The points on a cylinder and a plane are path- wise connected in very differ-
ent ways, and this is an intrinsic difference. However, it is a difference that can
only be discovered by investigating the space as a whole; it cannot be discerned
locally. Hence we need to distinguish between the local and the global properties
of a space. Spaces with different global features can be locally indistinguishable.
The notion of intrinsic curvature was developed further in Riemanns extension
of Gausss methods, work that laid the foundations of modern differential geom-
etry. Riemann provided the means for characterizing path- lengths and continu-
ously varying curvatures in three- , four- and higher- dimensional spaces. Although
he followed Gauss to the extent of supposing that the curvature in the vicinity of
each point could be characterized by means of a Euclidean space of the required
dimensionality, no actual embedding of the curved space in flat space is involved
or presupposed: the claim is simply that the curved space around a specified
point, provided the area involved is sufficiently small, can be approximated to any
desired degree of precision, by a Euclidean space. The intrinsic curvature of more
complex spaces cannot be captured by a single number such as Gausss K. More
sophisticated mathematical devices are required, such as the multi- component
function now called the Riemann metric tensor, g
ik
, and the Riemann curvature
tensor, R
i
jkl
. (A tensor is a mathematical operator, acting on vectors, which
delivers numbers or vectors as output.)
Gauss had shown how to define the lengths of paths relative to an arbitrary
coordinatization on a curved surface via his g- function, which assigns a collection
of three numbers to each point on the surface; the values of these metrical coef-
ficients vary from point to point, depending on the curvature (and so, in effect,
these values register the effect of curvature on distance). Given the coordinates of
two points, the lengths of all the different paths between them can be calculated,
provided the values of g- functions are known for the points in the space. Riemann
generalized this approach, and showed how it could be applied in spaces of higher
dimensions. For a four- dimensional space, ten metrical coefficients are required,
CURVED SPACE 225
represented by the symbols g
11
, g
12
, g
13
, g
14
, g
21
, , g
34
, g
44
(these can be organized
into a matrix of sixteen members, but six of these are redundant since g
21
= g
12
, and
so on). Together these go to form the metric tensor. Twenty functions produced
from certain combinations of the metrical coefficients together determine the cur-
vature tensor, R
i
jkl
, which provides a full description of the geometrical properties
of the space around the point in question. The twenty numbers (or components)
of this tensor are needed to capture the way a four- dimensional space bends and
twists in different directions. Generally speaking, the higher the value of the metric
tensor, the more curved the space at the point in question. In a flat n- dimensional
space all of its components are zero. This vanishing of the metric and curvature
tensors at every point in a space is what distinguishes n- dimensional Euclidean
spaces. For a complete representation of the curvature of an entire space the value
of these tensors for every point in a space must be supplied. (Such collections of
tensors are known as a tensor fields.)
Using the tensor apparatus to represent the properties of a space has a useful
feature: the quantities thus represented (if not the actual numbers employed) are
independent of the system of coordinates used to refer to points in the space.
Coordinate independence (or covariance) is a criterion of a quantity being
physically real as opposed to an artifact of a particular mode of representation.
The mass of a truck does not depend on whether it is measured in pounds or
kilograms, and just as pounds and kilograms are interconvertible, so, too, are
the tensor representations of geometrical quantities based on different systems of
spatial coordinates.
4
It was this property that led Einstein to seek out the tensor
apparatus when formulating the general theory of relativity.
One further piece of terminology will prove useful later on: in n- dimensional
variably curved spaces, the counterparts of straight lines in Euclidean space paths
of shortest distance are called geodesics; these are also the lines of least curvature.
13.5 Topology
As well as developing methods for representing differently curved spaces,
nineteenth- century mathematicians moved on to investigate the properties of spaces
at a still more abstract level, and topology was born. The topological properties of
an object are insensitive to continuous transforma tions. Intuitively, the operations
Figure 13.8 Continuous transformations and their limits. Topologically speaking, the sphere and
the ovoid (rugby ball) are the same shape a continuous transformation leads from one to the other (and
back). But no such transformation (which permits stretching and squeezing but no tearing) can turn either
of these shapes into a torus. The sphere and the torus are therefore topologically distinct.
226 TIME AND SPACE
of bending, stretching and squeezing (e.g. as applied to a piece of soft rubber)
count as continuous transformations, provided no cuts, perfora tions or joining of
edges takes place. Hence a teacup and a doughnut are topologically indistinguish-
able stretching without tearing can take you from one to the other but both are
topologically distinct from a sphere, which is itself topologically indistinguishable
from a rugby ball and a cube. There are various interesting concepts in topology,
but I will mention just one, which will prove relevant later.
5
Manifolds that (like the torus) are holed (or that possess a handle like a
teacup) are said to be multi- connected, whereas manifolds lacking holes (such as
the surface of a sphere, or the Euclidean plane) are said to be simply- connected.
If a surface is simply- connected, every loop on it can be shrunk down to a point;
on multi- connected surfaces, the presence of one or more holes means there will
be some loops that cannot be shrunk down to points. Although it becomes harder
to define, this distinction applies to all spaces, irrespective of their dimensional-
ity. From a purely mathematical perspective multi- connected spaces are far from
uncommon; indeed, they tend to outnumber their simply- connected counterparts.
6
A torus can be made from a rectangular rubber sheet: first take two opposite
edges and curl the sheet into a tube, then take the two ends of the tube and bend
them towards each other until they join, and apply glue. Such operations cannot be
performed on a topological manifold (since no joining of edges is allowed), but we
can achieve much the same result by supposing that the points along the opposing
edges of the rectangle are numerically identical. A Flatlander on the surface of a
torus who moves straight up or down (imagine that the torus is horizontal, like
a doughnut lying flat on a plate) will soon end up where they started, likewise if
they keep moving forwards, thus taking the longer of the two routes around their
curved space. The dotted line connecting X and X* in Figure 13.9 represents a
short route, whereas the line between Y and Y* represents one of the long
routes. Anyone who has played the (now venerable) video game Asteroids will
have enountered a toroidal space of this sort. Objects moving across the screen
behave in the same way as the aircraft shown in Figure 13.9.
The torus created by identifying the opposite edges of a rectangle is geo-
metrically distinct from the more familiar torus depicted in Figure 13.8: unlike
the latter it is entirely flat everywhere (and so is called the flat or normal
torus), and in this respect resembles the surface of a cylinder (which is also locally
A B
C D
X
X*
Y* Y
1
2
3
4
5
6
Figure 13.9 A flight through
toroidal space. Sides AB and CD
are identical, as are AC and BD. To
an observer ignorant of the topology
of this space, the plane would seem to
leap (instantaneously) across space as
it flies through X, before resuming on
its original heading at X*. But in fact,
since X and X* are the same point, no
such leap occurs. Similarly, when the
plane arrives at Y on AC, it appears
immediately at Y* on BD, and continues
on its way. Some video games employ
spaces of this type.
CURVED SPACE 227
Euclidean). The surface of the ordinary torus, by contrast, curves both positively
and negatively. Unlike its curved counterpart, the flat torus cannot exist in three-
dimensional space.
We construct a flat hypertorus by identifying the opposite faces of a cube;
the surface of the hypertorus constitutes a three- dimensional multi- connected
space that is locally Euclidean. The rectangle and cube constitute the fundamen-
tal domains of the (flat) torus and hypertorus. Many other geometrical figures
can serve as fundamental domains for more complicated multi- connected spaces.
In the three- dimensional case, the fundamental domains are polyhedra whose
faces (or facets) are pairwise identified, sometimes after being rotated by specific
amounts relative to one another (see Figure 13.10). Some of these spaces are flat
but most are curved, either positively or negatively (usually the latter). Some are
closed (and so have finite volumes), and others are open. The number of multi-
connected three- dimensional spaces is infinite, and mathematicians have yet to
complete their classification.
Detecting multi- connectedness in a space can be quite easy. If both you and
the contents of your bedroom were miraculously transported to a hypertoroidal
space, whose fundamental domain is a cube with sides a hundred yards long, the
unusual character of your space would be immediately obvious. Looking around,
you would see something remarkable: repeated replicas of yourself and your bed-
room furniture, appearing from different angles, stretching off into the distance
on all sides. It would be rather like being in a room whose walls, ceiling and floor
are all mirrors. These images are due to light making several passes through
your space (in the manner of the aircraft in Figure 13.9) before arriving at your
eyes. If you were to wait long enough, you would eventually be able to see every
object in your room from every angle. More generally, the apparent structure of
Figure 13.10 The fundamental
domains of three-dimensional multi-
connected spaces are polyhedra. The
hypertorus is created by identifying the
opposite faces of a cube.
Figure 13.11 A closed,
multi-connected, hyperbolic space viewed
from within. The tiling of the fundamental
polyhedron a dodecahedron appears
to stretch off to infinity. (Courtesy of the
Geometry Centre, image created by
Charlie Gunn.)
228 TIME AND SPACE
a multi- connected space is a function of the fundamental domain; an observer in
such a space will see a tiling of endlessly repeated images of the relevant poly-
hedron stretching out, seemingly to infinity on all sides. Figure 13.11 shows the
effect for a closed SeifertWeber space.
7
Of course, if the fundamental domain is
very large, these effects will not be so easily discerned; if the fundamental domain
is larger than the visible universe, they will not be detectable at all. But more on
this later (20.7). It is time to consider some of the philosophical implications of
these geometrical discoveries.
13.6 Conventionalism
Most of the early pioneers of non- Euclidean geometry had this thought: Since
there are alternatives to Euclidean geometry, we cannot be certain that our space
is Euclidean, and went on to devise tests that they hoped would reveal the truth.
Gauss used ordinary surveying equipment to measure the internal angles of a size-
able triangle (69

km 85

km 107

km) formed by the Brocken, Hoher Hagen
and Inselberg mountains. Lobachevsky worked on the astronomical scale, and
conducted measure ments on a triangle that had the star Sirius as apex and one of
the diameters of the Earths orbit as base. Both were disappointed to discover that
there were no discernible deviations from what one would expect if space were
Euclidean. However, they also realized that this proved nothing. Very gradual
spatial curvature would only reveal itself over very large scales indeed, of a sort
that were beyond the measuring techniques available to Gauss and Lobachevsky.
The existence of other galaxies was unknown at the time (indeed, their existence
was only firmly established in the 1920s).
At the turn of the twentieth century, shortly before the advent of Einsteinian
relativity theory, Poincar put forward an argument that, in the words of one
commentator, demonstrated once and for all the futility of this controversy and
the fallacy of any attempt to discover by experiment which of the mutually exclu-
sive geometries applies to real space (Jammer 1993: 165). The conventionalist
approach to space that Poincar advocated initiated much subsequent discussion
(see Sklar 1977: 88147 for a thorough overview), and was an important influence
on neo- positivist thinkers such as Reichenbach and Grnbaum.
Poincars starting point is uncontroversial. Since Gaussian methods for
ascertaining intrinsic curvature involve measuring distances and angles, before
we can even start to investigate the geometry of physical space we have to give
physical meaning to basic geometrical concepts; we must settle on some practical
ways of measuring distances, straight lines and angles. This was not usually seen
as problematic. Short distances can be measured by putting rigid rods end to end,
or with tape or string; for longer distances it suffices to measure the time taken
for a fast- moving signal with known speed, such as light or sound, to cross the
inter vening space. Light rays also provide a measure of straightness, at least on the
assumption that light always travels in straight lines through a vacuum.
8
Poincars
basic point was simple. If we conduct measurements that suggest that our space
is non- Euclidean we cannot immediately draw the conclusion that our space is
non- Euclidean, for there is always an alternative hypothesis, namely that our space
CURVED SPACE 229
is in fact Euclidean but (for one reason or another) our measuring equipment is
delivering inaccurate results.
Reichenbach provides a good example of how this might come about (1958:
1.3). Imagine a Flatland inhabited by two- dimensional creatures who move
about making measurements as they attempt a large- scale geographical survey
of their world. They find that most of their space has Euclidean characteristics
(with respect to triangles, circles, etc.), but there is one region where this isnt the
case. As they proceed through this region their measurements suggest that space is
becoming first negatively curved (a two- dimensional saddle- space), then positively
curved (like the surface of a sphere) and then negatively curved again, before flat-
tening out. They conclude (reasonably enough) that their space is non- Euclidean:
although mostly flat it contains a hump- shaped region.
These beings live in world A. Imagine now beings who live in world B. Although
world B is completely flat, the observable behaviour of physical bodies in world
B is exactly the same as in world A: when moved across a certain region of space
they change shape. The relationship between the two is shown in Figure 13.12.
Whereas the rod in the curved region of world A retains the same length but
becomes curved, the corresponding rod in world B remains straight, but contracts
in length. Since all bodies behave in this way, expanding and contracting, the
inhabitants of world B would be utterly unaware of this behaviour, at least until
they conducted detailed measure ments. We further suppose that all processes that
can be used to measure geometrical properties (such as light rays) act in ways
that render the size changes unobvious. When they do conduct detailed surveys
and measurements, the inhabitants of world B soon discover that one region is
different from others: at the centre of this region parallel lines converge, but at
the interchange region many parallels exist; they find similarly odd results for
measurements on triangles and circles.
Figure 13.12
Reichenbachs world.
A
B
Since the results obtained using measuring rods and protractors are just as
in world A, the inhabitants of world B could postulate that their space is non-
Euclidean, and has a hump. But as it happens, they dont. The geometers of world
B argue thus:
There is a region in our world where measuring rods shrink and expand in
various odd ways. These changes in shape are such that someone who didnt
realize that they were occurring could easily be led to conclude that our
space contains a region curved like a hump, for if the measurements were
taken to be accurate, this would be the case. But since the equipment is in
fact distorted, this is not so: our space is flat.
230 TIME AND SPACE
Not only does this hypothesis seem just as reasonable as the one proposed by the
geometers in world A, but if it were put forward by a dissenting geometer in world
A there is no observation that could directly refute it. Even so, a proponent of
curved space might reasonably argue like this:
Our account has the merit of explaining exactly why we get the measure-
ments we do: the medium in which we are conducting the measurements is
curved in ways that we can specify very precisely. By contrast, the postulated
shape changes are brute facts that are entirely unexplained.
This objection can be countered, but the reply reveals an important feature of the
alternative theory.
To explain the changes in shape we can posit a special kind of force- field ema-
nating from the centre of the relevant spatial region, and the measurable altera-
tions in shape can be explained in terms of variations within this field. Since
physicists often postulate force- fields to explain observed effects, this general
strategy is perfectly respectable, but the type of field required in this case is some-
what unusual. Since metal bars expand when heated and contract when cooled,
it might be thought that something like heat (or temperature variation) is causing
the size change of measuring rods in world B. But heat has differential effects, in
that different kinds of substance expand at different rates in response to the same
applied heat. Put a glass rod and an iron rod into the same oven and the latter
will emerge longer than the former. The contraction in world B is a universal
effect, which changes all kinds of body in the same way and to the same degree,
depending only on their location. If the force- field affected some bodies but not
others, its influence could be detected, and world B would differ from world A in
observable ways.
When the flat- space hypothesis is supplemented by a detailed account of the
location and action of the universal force, it offers as powerful an explanation of
the observational data as the curved- space hypothesis. So what should the inhabit-
ants of these worlds conclude about the geometry of their space? Poincar suggests
that since they cannot know the shape of their space, the preference of one hypoth-
esis over the other can amount to nothing more than a choice of convention. The
inhabitants of world A have decided that they prefer to think of their space as
curved, whereas the world B inhabitants prefer the flat- space hypothesis. This is
just one example, but it is easy to devise others (Poincar himself used an imagi-
nary disc- world whose inhabitants shrink as they move from the centre), and the
lesson applies in three dimensions as much as in two. No observational evidence
could decisively refute the hypothesis that our space is Euclidean since the relevant
deviations can always be explained by postulating a universal force that produces
precisely the same observational results as curved space. More generally, since for
any physical universe the observational data will always be equally compatible
with any number of hypotheses concerning the geometry of physical space, the
preference for one hypothesis over the others is not justified by the empirical
evidence. In actual fact, Poincar suggested, we will always favour the simplest
available hypothesis, and since Euclidean geometry is simpler than the alternatives,
we will never adopt a theory that ascribes a non- Euclidean geometry to our space.
There is, however, no reason to think that our preferred theory reflects the real
CURVED SPACE 231
structure of space; what it reflects, rather, is our preference for theories we find
simple and natural.
13.7 Realism versus anti- realism
In one respect Poincar was soon proved wrong: by the 1920s most physicists
had come to accept Einsteins general theory of relativity, according to which
gravitational effects are due to mass- induced curvatures in spacetime. Scientists
do have a preference for simpler theories, and Euclidean geometry is simpler (in
some respects) than non- Euclidean, but what Einstein saw (and Poincar did not)
is that an entire theoretical package of the form {non- Euclidean space + laws +
forces + entities} could provide a simpler account of the observational data than
{Euclidean space + laws + forces + entities}. Spatial geometry is just one aspect of
a physical theory, and the complexity that comes with non- Euclidean geometries
can be more than compensated for by simplifications elsewhere. Einstein showed
that by introducing variable curvature into spacetime we can economize on forces:
we no longer need to posit gravity as an attractive force (more on this later).
However, Poincars failure to appreciate this point does not undermine his
argument as a whole. Before going into the reasons why this is so it is important
to be clear on what the argument purports to establish. The thrust of Poincars
conventional ism is epistemological rather than metaphysical. He is not claiming
that the very idea of a substantival space is incoherent, he is not denying that such
a space exists. The claim, rather, is that, even if a substantival space does exist, we
could never be justified in thinking that we know its true geometry.
The reasoning that Poincar uses to establish this conclusion is a particular
application of a familiar anti- realist strategy in the philosophy of science. The
anti- realism in question is directed against theories that posit unobservable enti-
ties, and the claim is that, even when such a theory provides a plausible explana-
tory account of the observable data, and delivers accurate predictions as to how
observables will behave, we should not believe that the posited unobservable enti-
ties exist. The reason for this is that quite generally scientific theories are under-
determined by observational data. The relevant data does not logically entail the
truth of any one theory, since there will always be alternative theories theories
that posit different unobservable entities that explain the same data equally
well. These alternatives may not strike us as being equally plausible, indeed it will
often be the case that one theory seems to explain things better than the others,
but it would be a mistake to think that this means the theory in question has a
greater chance of being true. Owing to the quirks of human psychology we find
some explanations better or more satisfying than others; similarly, the theo-
ries that we find plausible are those we find easy to believe. But what reason do
we have for thinking that our preferences in these matters are reliable guides to
the structure and functioning of reality? Assuming that we evolved through natu-
ral selection, our minds are far more likely to be geared to solving the problems
encountered by our ancestors struggling to survive in the hostile environments
in which they found themselves, rather than the invisible deep structures and
basic ingredients of physical reality. Given this, we should adopt an attitude of
232 TIME AND SPACE
cautious agnosticism toward even our best theories concerning these structures
and ingredients.
There are those who are convinced by this argument, but, equally, there are
many who are not. Realists point out that the best explanation for why a theory
generates tremendously accurate predictions across a vast range of cases is that
the theory is true, or at least approximately so. As for the charge that the criteria
we use in evaluating theories are subjective (at least at the species- level), realists
can accept that, while there is some truth in this, it would be wrong to think that
our criteria are entirely fixed. As science has progressed (and most anti- realists
accept that progress has been made), our notions about what counts as a plau-
sible theory or good explanation have evolved a good deal the standards of
plausibility in current physics are not Aristotles and it is not unreasonable to sup-
pose that the criteria we now employ better reflect the functioning of nature itself.
This is not the place for a thorough treatment of this important issue. For our
purposes it will suffice to note the implications of the general realistanti- realist
debate in the philosophy of science for the substantivalistrelationist dispute. If
it turns out that our best theory of matter and motion requires a substantival
space, realists will be substantivalists; but if our best theory takes a relational form,
realists will be relationists. Those of anti- realist inclinations will keep an open
mind on the ontological issue. No matter what theoretical simplifications accrue
from positing a substantival (or relational) space, anti- realists will remain uncom-
mitted on the question of whether space really is substantival (or relational).
14 Tangible space
14.1 Manifestations of curvature
Now that we have a reasonable grasp of the distinction between flat and curved
spaces we can turn to our main business, and consider the significance of the
distinction for the reality of space. That the distinction does impact on the debate
between substantivalists and relationists has been forcefully argued by Nerlich, on
various occasions (1979, 1991, 1994b: 1.7). I will start by looking at some of the
ways that spatial curvature can manifest itself in discernible ways, and them move
on to consider the use that Nerlich makes of these manifestations.
Leibnizs various anti- substantivalist arguments all rely on the innocuous ness
of space: there would be no discernible or significant difference if the physical
universe as a whole were shifted with respect to space, either by having a differ-
ent velocity, or by being located elsewhere. A further variant along these general
lines that Leibniz could have employed is nocturnal doubling. If everything were
to double in size overnight, would we be able to tell? You would wake up in the
morning twice your previous size, but since your bed would also have doubled,
this would not be immediately apparent. Similarly, your room would also have
doubled in size, as would your house, your town and everyone in it. Distances
between houses and towns are now twice what they were, but since all measuring
implements have also doubled in size, there is no way that this can be detected.
Your trip to work takes no longer than usual. To make the example pedant- proof
we must further suppose that the laws of nature have been altered to conceal the
doubling: light travels twice as fast; the mass of an object is increased in propor-
tion to its length rather than its volume and so on. Let us simply assume that the
alterations required for the doubling to be indiscernible have all occurred.
If this all seems very plausible it is because a tacit assumption is being made:
that the doubling occurs within a flat Euclidean space. Were the doubling to occur
within a non- Euclidean space it would be very different. This is most easily seen
for a space of constant positive curvature in two dimensions, the sort of space
inhabited by Flatlanders living on a spherical surface. (Precisely analogous conse-
quences obtain in curved three- dimensional spaces of the same type.) One of the
distinguishing proper ties of spaces of this sort is that the internal angles of triangles
add up to more than 180 degrees. Moreover, this divergence from the Euclidean
norm is size- dependent: the smaller a triangle is the more closely it approximates
to the Euclidean, and the larger it is the more it deviates. As we saw in Chapter 13,
triangles that stretch from the pole to the equator can easily have internal angles
234 TIME AND SPACE
greater than 270 degrees. As a conse quence, a doubling in size in such a space
would be easily detectable, since it would result in easily measured changes in the
shapes formed by objects separated from one another by discernible distance.
Nerlich goes on to point out that it is not just triangles that are affected. Since any
polygon can be decomposed into triangles, and any closed curve can be approxi-
mated by some polygon, shapes of all kinds will change on being doubled. (In a
space of constant negative curvature there are changes too, but of the opposite
kind: the internal angles of triangles will decrease with the increase in size.)
There is the potential for more dramatic consequences. As the two- dimensional
spherical surface illustrates, a space of constant positive curvature can be finite,
and as far as doubling is concerned this makes a definite difference. If our Flat-
landers space is more than half full of matter (think of a planet whose continents
cover more than half its surface area), there wont be enough room for them all to
double in size. In such cases, nocturnal doubling is simply impossible. Shifting up
to three dimensions, if our universe were both finite and positively curved, and the
stars are uniformly spread through space, it makes no sense whatsoever to suppose
that the distances between the stars could all double.
As might be expected, in spaces of variable curvature the effects of doubling
are more variable. Again, consider the two- dimensional case. Our Flatland now
undulates, and contains mountains, valleys and plains; although in some places the
curvature is zero, in others it is positive and in others negative. The changes that
would be produced by a doubling in size now depend on where we envisage the
doubling starting, for whether or not an object gets pushed from a valley on to a
plain, or from a mountain top into a saddle- back, depends where we centre the
doubling; different centres yield different results. As Nerlich says:
there are uncountably many ways of performing the thought experiment
and uncountably many distinct thingthing relations resulting from them. So
doubling- in- thought experiments depend on which thingplace relations of
spatial occupancy (position, place) are preserved, if any, and how others are
altered. (1991: 174)
Spaces that vary markedly in curvature also cast new light on Leibnizs static
and kinematic shifts. In forwarding these arguments Leibniz assumed Newtonian
space to be uniform. Whatever force they have in the context of uniform space
does not survive in the context of non- uniform spaces. It may make no discernible
difference where God chooses to locate the material universe in a flat space, but it
does in a variably curved space: the measurable (Gaussian) geometry of space will
be different in differently curved regions of space. The same applies to the kin-
ematic shift. If the material universe were moving through the three- dimensional
analogue of a Flatland mountain range, the discernible geometry would vary over
time in quite dramatic ways, and the patterns of variation would vary, depending
on the speed and direction taken.
In extreme cases, the changes in shapes and alterations in distances wrought by
spatial curvature would be both tangible and visible in ways not yet considered,
as Nerlichs example of the non- Euclidean hole clearly shows (1994a: 38, 1994b:
7). In this imaginary (but logically possible) case there is a region of space in the
room you are now in, about the size of a football, where space is drastically curved,
TANGIBLE SPACE 235
but both positively and negatively; space elsewhere in the room is flat. (The hole
in question is not an absence of space (construed substantivally), but simply a hole-
shaped region where space is curved.) When light rays approach the hole they are
moving in parallel straight (Euclidean) lines; on entering the hole they continue
to follow geodesic paths, but by virtue of this regions intrinsic curvature they
first all converge and then all diverge again before leaving. The fact that light rays
change direction means that objects viewed through the hole will look different,
just as they do when viewed through a flaw in a windowpane, or through a heat
haze. Wandering around the room, observing how different things look distorted
when looked at from different angles, would soon give you a firm impression of
the holes location, shape and size. Indeed, as Nerlich points out, If all this came
about, would we say, not just that we see that there is such a hole, but that we see
the hole itself, just as we see warps in clear glass? I think we well might (1994a:
39, original emphasis). In cases such as this, space itself becomes visible, or at least
regions of it do.
This is not all: these regions would be tangible. Like rays of light, moving
particles follow geodesics unless acted on by a force. Suppose you take a sizeable
cube of soft foam rubber and move it into the hole. What would happen? Would
it pass freely through? There are no material barriers to prevent it, just air and
empty space. In fact, you would very likely feel a resistance. On entering the hole,
the particles in the rubber will try to follow geodesic paths, and so initially they
will converge, and then diverge. But this convergence will be resisted by the inter-
particle bonds: you can imagine these electromagnetic forces as akin to elastic
bands connecting the particles; as the particles try to converge, the bands have to
stretch. Consequently, for the rubber cube to succeed in entering the hole a force
needs to be supplied to stretch the bonds and hence you feel resistance when
pushing the cube forwards. What if you were to walk through the hole? The same
would apply. You would have to push yourself forwards, you would feel resistance,
and since stress tensions would be generated among your middle parts you would
feel a distinctly queasy sensation. You would feel the curvature of space.
1
14.2 The detachment thesis
We will be returning to the tangible manifestations of spatial curvature, but we
have seen enough for the purposes of Nerlichs pro- substantivalist argument.
To start with, recall the general form of Leibnizs case against Newtonian space.
Underpinning the argument is an assumption about what we should regard as
real. Leibniz assumes that an entity or property that can change without making
any observable difference to anything is not a genuine ingredient in or feature of
the world. Given that unobservable entities now play a more prominent role in
scientific theories than they did in Leibnizs day, this criterion can be broadened
to something like this: only entities that are needed to explain and predict the
observable behaviour of objects should be regarded as real.
On the face of it the substantivalists space satisfies this criterion. An objects
size is the quantity of space that it fills, its location is the place in space that it
occupies, the distances between objects are the lengths of the paths through space
236 TIME AND SPACE
that connect their locations, and, quite generally the spatial relations between
things are all mediated by relations of things to space (Nerlich 1991: 171).
Leibnizs anti- substantivalist argument can now be viewed as akin to a reductio:
Lets suppose that this substantival space does exist. The doubling case (or one
of the other shifts) is then introduced, and straight away the initially plausible
assumption concerning the role of space is under mined. Since all spatial relations
between things can remain observably and measurably the same through dramatic
changes in the relations between things and space, it seems that substantival space
does no useful work after all; it is an idle and inconsequential posit. To put it
more succinctly, all thingspace relations can change while all thingthing relations
remain the same (thingthing relations are spatial relations that are empirically
detectable).
Nerlich now suggests that Leibnizs argument relies on the following general
principle:
Detachment thesis: thingthing spatial relations are logically independent of
thingspace relations. (1991: 172)
The viability of relationism presupposes the detachment thesis. Relationists hold
that only material objects and spatial relations exist. Someone could object, But
spatial relations require the existence of space! Things are spatially related when
some path through space connects them. For relationists to avoid this charge of
circularity they must establish that spatial relations can exist independently of
space; hence the need for the detachment thesis.
So the question to be addressed is whether the detachment thesis is true. Ner-
lichs central contention is simple: the detachment thesis is only plausible for
the special case of Euclidean space, it is manifestly false for spaces that are not
Euclidean. We have already seen why. A universal doubling may have no observ-
able consequences in a flat space, but it has in curved space, and likewise for the
kinematic and static shifts. Since in curved spaces varying the thingspace relations
can produce dramatic variations in observable thingthing relations, the claim that
thing space relations are inconse quential is straightforwardly false, likewise the
claim that thingthing relations are logically independent of thingspace relations.
Of course at the time Leibniz formulated his arguments for the detachment
thesis, the existence of non- Euclidean spaces was not even suspected, let alone
established, so he can scarcely be blamed for not noticing the flaw in his case.
But since his arguments do have some force in the context of Euclidean space, a
relationist might argue thus:
Even if I grant for the time being that the detachment thesis is false for certain
non- Euclidean spaces, rather than rejecting the thesis entirely, we should
restrict its scope to spaces of zero curvature. If it then turns out that our space
is flat, the detachment thesis is true for our world and the argument against
substantivalism retains its force.
Not so argues Nerlich:
the dependence of the outcome of doubling on the nature of space applies
with equal force to the special and actually indiscernible case of doubling
TANGIBLE SPACE 237
in Euclidean space. It is not only when doubling will change shapes that
we need to specify in which space we are performing the thought experi-
ment. That the doubling in Euclidean space has no thing thing consequences
depends on specific symmetries unique to the structure of that space. These
symmetries pick out the similarity transformations from the rest. They do
not show that there is no structure and no space. They mean that there are
specific structures in a specific space. (1991: 175)
The various changes in distances and shapes that universal doubling produces
in non- Euclidean spaces are to be explained by the specific features of the space
concerned: different curvatures produce different changes. But equally, the fact
that operations such as doubling, translation and relocation produce no varia-
tions in shapes and distances in Euclidean space is to be explained by the specific
features of this space; it is only because Euclidean space possesses its distinctive
symmetries that these operations leave no discernible traces. Euclidean space is no
less influential than other sorts of space, but by virtue of its distinctive geometry its
influence is uniquely discrete: by virtue of its symmetries, linear translations of any
size and in any direction do not change the distance relations between a systems
component parts, and doubling (or trebling, or halving) operations do not change
the relative sizes or distances between the objects in a system.
As Nerlich goes on to argue, once the active role of Euclidean space is appreci-
ated, the substantivalists position looks stronger than ever. Given the Euclidean
symmetries, the hypothesis that we live in a substantival space whose geometry is
Euclidean explains why it is that the difference between rest and uniform motion,
doubling and constant size cannot be observed: The realist owns the symmetries,
not the relationist. It is the realists realism that tells him (and tells us) why we
cannot care about the difference between one of these states and another. In fact, it
is the relationist who has no clear, coherent, non- circular account of the relativity
of motion (1991: 1812). And this means that a standard objection against the
substantivalist is without force. There is no reason for the substantivalist to be con-
cerned about the fact that various translations through space have no observable
consequences, for this is precisely what his own hypothesis (space is Euclidean)
predicts.
In 9.2 I pointed out that since a pure void has no determinate structure or
dimensionality, the void- conception of space leaves something unexplained why
it is that our movements are constrained in the ways they are and I called space
the unseen constrainer. The point still applies, but, as is now clear, this particu-
lar label is particularly appropriate in the case of Euclidean space.
14.3 The explanatory challenge
Nerlichs argument can usefully be recast in the form of a challenge to the relation-
ist. Objects moving through non- Euclidean spaces especially spaces of extreme
and variable curvature will behave oddly: their shapes will vary in observable
ways; they will move closer together and further apart; internal stresses will
be generated within them. Since no known forces (e.g. gravity, magnetism) can
238 TIME AND SPACE
explain these movements, we are left with a puzzle. The substantivalist has a
simple and economical explanation for the strange behaviour: a substantival space
exists; this space is curved; and it is a law of nature that objects follow geodesics
through this space unless acted on by some external force. For this explanation to
work the substantivalist must be able to fill in the details, by providing at least an
approxi mate mapping of the actual shape of space, and show that this mapping
suffices to explain the observed motion of bodies when combined with data about
the known forces that are in play. Providing these details is a non- trivial task, but
if it can be done the substantivalists explanation certainly deserves to be taken
seriously. What can the relationist offer in its stead?
In some cases it is very hard to see how the relationist could come up with
a remotely plausible explanation. To see this it suffices to imagine a world that
contains non- Euclidean holes of the type introduced above. From the substantival-
ists stand point, the space of this world contains various stationary regions of
tightly curved space, and these regions have discernible effects on moving bodies,
light rays, and so on. How can the relationist explain these various effects?
Consider first a very simple case that the substantivalist would describe in these
terms. The system contains one star, one orbiting planet and one hole. The curva-
ture of the hole is symmetrical around its centre; the planets orbit is elliptical and
regular, and it passes through the centre of the hole at precisely the same time each
year. Since the relationist denies the existence of both substantival space and the
hole, the various discernible effects that the substantivalist puts down to passing
through the hole must be explained solely in terms of the relation ship between the
planet and its star. Can we say that the effects are correlated with the planet being
at a certain distance from the star? No, since the planet is at the same distance from
its star twice per orbit, and the problematic effects only occur once per orbit. But
another simple option is available: correlate the effects with time. Since the orbit
is perfectly regular, the time between the onset of the effects is regular too they
happen exactly once each year so the relationist can explain the effects by posit-
ing a law that states that the various effects occur at yearly intervals. Laws of this
form are odd, but perhaps not impossible; there might be worlds where all green
things change to red for one day at yearly intervals. Even so, there is a sense in
which the relationists explanation is less complete than the substantivalists. The
latter can (in principle at least) explain the precise form the effects take, in terms
of differing degrees of positive and negative curvature within the hole, and why
these effects only occur on the planet but not the Sun. The relationist must take
these to be nomologically basic features of the world.
In this simple case the discrepancy between what the relationist and substanti-
val ist can offer is not dramatic, but this quickly changes if we introduce more
complexity into the system. Suppose that we introduce just one more planet on
a different orbital plane. The two planets are attracted to one another by both
gravity and magnetism in such a way that their orbits are complex and aperiodic,
but still deterministic and predictable. Both planets pass through the hole region
at varying intervals, and along varying trajectories. Since they dont always pass
through the centre of the hole the effects are usually only felt in some regions but
not others. Since the relationist can no longer explain the effects by the simple time
law that worked in the previous example, the only remaining option is to seek an
TANGIBLE SPACE 239
explanation in terms of the relative positions of the two planets and the star. But
given the irregular orbits of the planetary bodies, and the different ways the effects
are felt by the planets on different occasions, there will be no simple correlation
between relative orbital configurations and effects. At best the relationist will be
able to come up with a law of this type: {C
1
E
1
, C
2
E
2
, C
3
E
3
, }, where each C
stands for a different orbital configura tion and each E for a different pattern of
effects. Not only is each distinct pairing a brute nomological fact, but the list may
well be infinitely long. Opposing this hideously complex gerrymandered effort is
the simple and powerful account offered by the substantivalist. There is simply
no competition. And all this holds for a very simple system; if a few more holes
and planets are introduced, if the star moves, the relationists predicament swiftly
becomes utterly hopeless.
However, if the sort of universe just considered is extremely hostile to the
relationist, others are far less so. To take another simple example, consider a world
consisting of just three equally spaced non- interacting point- particles moving away
from each other in straight lines at uniform speeds. In a spacetime diagram of this
world, the particles worldlines resemble an inverted tripod. Needless to say, the
substantival ist can account for this pattern of movement in terms of geodesics
through flat space, but this account is not very economical, featuring as it does
vast (perhaps infinite) tracts of empty substantival space, the vast bulk of which
serves no explana tory purpose. The relationists account is just as simple and far
less profligate. At any instant the universe contains just three objects and three
spatial relations, and the evolution of the universe is controlled by a simple law:
the distance relations between the objects increase at a constant rate over time.
More complex worlds present the relationist with more complex challenges. As
we saw in the course of our earlier explorations of Newtonian dynamics, there are
worlds where the relationist needs to distinguish between inertial and non- inertial
paths. But, as we also saw, there are ways that the relationist can hope to do this.
As for the conclusions that we can draw from all this, I suspect that there is
just one: there are no easy answers. There are, however, three lessons that have
emerged. First, even in worlds where there is no evidence of the sorts of behaviour
that can be explained in terms of spatial curvature, there is still something to be
explained, namely why the movements of material bodies conform to the geome-
try of Euclidean space. Since there are many other possible geometries, this is a fact
that needs explaining, and the explanation advanced by substantivalists is simple
and effective. Unless the relationist can provide an account of comparable merit
we have no reason to reject the account offered by the substantivalist. Secondly,
in worlds where the substantivalist finds it natural to posit a variably curved space
to explain the observed motions of bodies, it is quite likely that the relationist
will be unable to offer an explanation of comparable simplicity. Finally, however,
there are worlds where the relationist may be able to provide accounts of observed
motions that rival or even surpass the substanti val ists accounts in simplicity and
economy. In cases such as these we cannot pronounce a priori on the likely out-
come; all depends on the details of the specific theories that are proposed.
We will be returning to these issues in Chapters 20 and 21, when we consider
Einsteins general theory of relativity; a theory that posits a variably curved space-
time. We shall see that while a substantivalist interpretation of this theory is hard
240 TIME AND SPACE
to avoid, it is also far from unproblematic. But before moving on to these real- life
scientific cases, there is a famous a priori argument for spatial substantivalism that
we have not yet examined.
14.4 A solitary hand
At various times dating back to the 1768 essay Concerning the Ultimate Founda-
tion of the Differentiation of Regions in Space, Kant used incongruent counter-
parts in a novel attempt to refute the relationist conception of space. Sometimes
he used the argument to reject relationism in favour of Newtonian absolutism, and
at others he used it to defend the idea that our concept of space was of a structured
entity, distinct and independent of the things in space. This is all he could say in his
anti- realist phase. Although the argument (and the phenomenon in question) have
generated a consider able literature, I will not be entering into the debate very far,
but since Kants argument does cast light on the main theme of this chapter from
a new angle, this is the place to consider it.
The discussion will move more swiftly if some relevant terminology is intro-
duced at the outset. Objects are enantiomorphs or incongruent counterparts if they
are related to one another in the same way as mirror images. Although in one sense
they are of exactly the same shape, in that their parts are of the same size, distance
and angular separation, they cannot be brought into congruence with each other
(i.e. made to occupy the same part of space) by any amount of sliding and turning.
The letters b and d are two- dimensional examples (at least in certain fonts). If both
letters are confined to the same two- dimensional plane, they cannot be made to
occupy the same space. There are three- dimensional examples too, such as the left
and right gloves in a pair. Of course, if the letters were made of clay, the b could
easily be squished into a d shape, but in the present context, bending, stretching
and turning inside- out are not allowed. The permitted spatial transla tions are
called continuous rigid motions, or crms. Any two objects whose parts are of
the same size and disposition, but that cannot be brought into congruence by a
crm are enantiomorphic; objects are homomorphic if they can be made congruent
by crms.
Kant noticed the peculiarity of handed objects: What can be more similar in
every respect and in every part more alike to my hand and to my ear than their
images in a mirror? And yet I cannot put such a hand as seen in the glass in the
place of its archetype (Kant 1996: 55). Hands and ears are prime examples of
three- dimensional incongruent counterparts. No measurement of size or angle will
reveal a geometrical difference between the left and right ears of someone who
is perfectly symmetrical, and yet the two are clearly different, since they cannot
be made to occupy the same space. Kant saw that this difference could be very
troublesome indeed for the relationist. To bring this out he used a memorable
thought- experiment.
Imagine that the universe is empty except for a single human hand. Is it a right
hand or a left hand? Since there are no intrinsic differences between enantio-
morphic objects, we have no basis for calling the hand left or right. Of course, if
you imagine yourself looking at the hand floating in otherwise empty space, you
TANGIBLE SPACE 241
will see it as a right hand or a left hand, but in imagining this you are putting
yourself into the space of the hand, together with your sense of right- and left-
handedness. If you perform the thought- experiment rigorously, and dont include
yourself in the picture, it seems clear that we would have no more basis for calling
the hand left or right than saying that it is pointing up or down. Now suppose that
a human body yours materializes in space near the hand. You have with you a
pair of gloves. Clearly only one of these gloves will match the floating hand. Let
us suppose that the glove for the left hand fits. Since we can stipulate that your
appearance on the scene does not change the hand in any way, it seems clear that
the hand must already have been a left hand before you appeared in the space.
In which case, there must have been a fact of the matter as to the handedness of
the hand even when it was the sole occupant of the universe. The key question is
whether the relationist can explain this fact, and it seems not. Since by definition
all the spatial relations (distances, angles) among the parts of a pair of incongru-
ent counter parts are identical, it seems that from a purely relational perspective
there can be no difference whatsoever between a right hand and a left hand. Kant
concluded that since there clearly is a difference between them, and this differ-
ence cannot be explained relationally, it must somehow be due to the relationship
between the hand and substantival space.
14.5 Global structures
Unfortunately, Kant was not very explicit as to how introducing substantival space
into the picture explains the difference between oppositely handed objects. Later
substantivalists have tried to fill this lacuna, and their efforts have helped clarify
the phenomenon of handedness. I will summarize their arguments, and then con-
sider the relationist replies.
Some have tried to refute Kant by arguing thus:
Kants argument rests on the assumption that it makes sense to say that a
solitary hand is left- handed or right- handed. In fact this doesnt make sense.
To see why this is so, think of a solitary (and asymmetric) flat hand on the
plane surface of Flatland. Does it make sense to say that this flat hand is right-
handed or left- handed in the absence of any Flatlanders? The answer is no.
From our three- dimensional perspective, the hand is either a right hand or a
left hand, depending on the direction that we view the plane from (e.g. seen
from above it is a left hand, and from below it is a right hand). So it seems
that it doesnt make sense to say that the hand is a right hand or a left hand
when it is all alone, because what we say it is depends on the perspective that
we view it from. One and the same hand is both a right hand and a left hand,
depending on how we view it, and so in itself it is neither.
Suppose that we now introduce a Flatlander into the world. She has two
gloves, one for a right hand and one for a left hand, labelled R and L. Obvi-
ously only one of these gloves will fit the hand. Let us suppose that the left
one does. Does it mean that the hand was a left hand before we introduced
the Flatlander? Not at all, because the glove that fits depends on how we
242 TIME AND SPACE
introduce the Flatlander; place her in Flatland one way, and she finds that
the hand fits the left- hand glove; turn her over and put her down, and she
finds that the hand fits the right- hand glove. In turning the Flatlander over
in three- dimensional space, we dont change anything about her or the hand
in Flatland; all we change is how they are spatially related to each other.
As is clear, precisely the same applies to us in three- dimensional space. We
can imagine materializing next to the floating hand, and finding that our
left- hand glove fits it; but this doesnt mean that the hand was a left hand
before we arrived, because if we had been rotated in four- dimensional space
we would have found that the hand fitted the right- hand glove. Nothing in
the hand itself determines in advance which glove will fit.
The Flatlanders look at their pairs of gloves and wonder how can two things be
so alike and yet so different? We three- dimensional beings can understand. The
Flat landers hands are exactly alike; it is only because they are trapped, physically
and imaginatively, in their two- dimensional space that they think the gloves are
different. Likewise, a pair of three- dimensional gloves are exactly the same shape,
since a four- dimensional being could make them match exactly; they only appear
different because of the way they are implanted in three- dimensional space.
This line of argument may seem to provide strong grounds for supposing that
enantiomorphs are not intrinsically different, and so it might seem to undermine
Kants argument. However, the substantivalist can reply as follows:
Far from refuting Kant, the argument from Flatland strengthens his conclu-
sion. It is true that one and the same two- dimensional hand can be a right
hand or a left hand, but, as Kant foresaw, the handedness it has depends on
how it is inserted into space. Turning a Flatlanders left- hand glove into a
right- hand glove by rotating it in the third dimension amounts to inserting
the glove into space in a different way.
We can now see that the very possibility of enantiomorphism depends on
the properties of space itself. A pair of two- dimensional glove shapes can
only be enantiomorphs in two- dimensional space; in a three- dimensional
space they are homomorphs, since there is always a crm that will make the
hands congruent. Quite generally, an asymmetrical n- dimensional object
is enantiomorphic in a space of n dimensions, but not in a space of n + 1
dimensions. So we can see that the phenomenon of enantiomorphism depends
on two factors: the shape of a particular object, and the dimensionality of the
space in which the object exists.
2
This conclusion can be strengthened, for there is another general topological
property of space that is relevant to enantiomorphism: orientability. The technical
definitions of orientable and non- orientable spaces are quite complex, but
what the distinction amounts to is easily grasped.
Suppose that space is an infinite two- dimensional plane. Take two shapes, such
as b and d, which cannot be smoothly slid so that one coincides perfectly with the
other (contrast with b and q). The b and d are enantiomorphic both locally and
globally: no matter how far you move the two apart (and then bring them together
again) you will be unable to make them congruent. A space like this is orientable.
TANGIBLE SPACE 243
Now consider b and d on a Mbius strip.
3
In the immediate neighbourhood of b
and d it is impossible to get the two letters to perfectly coincide: no crm will render
them congruent. However, if you slide b right around the strip, when it returns it
will look just like d; the trip through this two- dimensional space has transformed
the b into a d. A space where this can happen is non- orientable. (In effect, such a
space has a kinked structure when viewed globally.) In an orientable space, any two
figures that can be brought into congruence by global continuous transforma-
tions can be brought into congruence locally, too; in a non- orientable space,
objects can be locally enantiomorphic with out being globally enantiomorphic. If
we define enantiomorphism in terms of the non- existence of any crms that will
render two asymmetrical objects congruent, then in a non- orientable space of n
dimensions, all n dimensional objects are homomorphic.
The substantivalist seems to have a strong case: whether an object is enantio-
morphic or homomorphic depends on the global properties of space itself, the
relevant properties being dimensionality and orientability. However, in the light
of the discus sion earlier in this chapter, it should be clear how relationists must
respond. They must show that we can make sense of the relevant features of space
in purely relationist terms, mentioning only material bodies, spatial relations, and
their nomologically possible modes of combination. Once again, the prospects for
this seem mixed.
Dimensionality poses no problem; the relationist can explain it in terms of natu-
ral laws governing the angles or directions in which objects can be spatially related.
Orientability, however, is a very different kettle of fish. In a simple world contain-
ing two objects we might perhaps posit a law that states that the objects undergo
a Mbius- like flip each time they move a certain distance apart, and remain in
the same orientation unless they first move closer together and then move further
apart again. But what of a universe that consists of numerous objects all moving
relative to one another that behave as if they exist in a static non- orientable space?
As with the case of multiple non- Euclidean holes, it is very likely that in such cases
any laws that the relationist can devise to explain the observed behaviour of bodies
will be hideously complex and unwieldy.
It seems that the phenomenon of enantiomorphism yields precisely the same
lessons as spatial curvature. When all possible spatially structured worlds are taken
into consideration, substantivalism will often be the only viable option. In the vast
majority of cases the relationist will be unable to provide an explanatory account
of why objects behave as they do that is of comparable power and simplicity. How-
ever, as we have also seen, some spaces are more dispensable than others. In worlds
where the motions of objects are not such as to warrant the postulation of one of
Figure 14.1 Relations with attitude.
A relationist who posits distance relations possessing
intrinsic directions has an account of how objects such as
those depicted here differ from each other, irrespective of
how they can move relative to one another.
244 TIME AND SPACE
the more exotic forms of space, worlds where a substantival space if it existed
would be comparatively boring and bland, the relationists prospects of providing
a satisfactory account of why things move as they do are greatly increased.
Having conceded a good deal to the substantivalist, it should be noted that the
shoe is sometimes to be found on the other foot. While we can explain enantio-
morphism in terms of the global structure of a substantival space, the relationist
who is prepared to recognize inherently directed vector- like spatial relations can
recognize handedness as an intrinsic property, as Figure 14.1 illustrates.
4
15 Spatial anti- realism
15.1 Foster on matter and space
While the conceptions of the large- scale composition of the physical universe
offered by substantivalists and relationists are undeniably very different, there is
also a sense in which they do not differ at all: both camps accept that we inhabit
a spatial world. The disagreement is limited to the question of how space (or
spacetime) should be characterized. Despite their differences, substantivalists and
relationists both count as spatial realists. In this chapter we will be examining a
purely metaphysical argument for spatial anti- realism; the claim that our universe
is not in fact spatial, despite appearances that suggest the contrary. The claim
that space is unreal may well seem as absurd as McTaggarts assertion that time is
unreal, but just as McTaggarts argument (or at least a close relative of it) proved
to be both non- trivial and independent of idle sceptical considerations, so too will
the argument we will be considering here.
In The Case for Idealism (1982) Foster develops an argument intended to
establish that the physical world is not (and could not) be a part of what he calls
ultimate reality, and the key step in reaching this conclusion is an argument
for spatial anti- realism. Unfortunately, despite its interest, Fosters argument is
complex, multi- faceted and often difficult; to do it full justice would require a
lengthy exposition, and this is not the place for it. Consequently I will be selective
in my approach, and confine my attentions to what he has to say about spaces
that are substantival in character and taken to be ontologically basic components
of the physical world. Foster develops similar (and equally potent) arguments
against relationism, but since these depend more heavily on the sometimes exotic
and unfamiliar framework employed in his book I will leave them for interested
readers to explore for themselves. (In his more recent The succinct case for
idealism [1993] Foster himself employs just this strategy; in his still more recent
A World for Us [2008], his approach is somewhat different see Forrest [2009]
for an overview.)
Although the import of Fosters argument is ontological whereas that of Poin-
cars conventionalism is epistemological, in certain respects the routes they take
are similar. Both arguments are concerned with spatial geometry, and there are
similarities in approach: both employ thought- experiments in which natural laws
conspire to conceal the real structure of the underlying (spatial) reality. The simi-
larities do not end there. As the following sketch of an argument shows, Poincars
conventionalism might easily be taken to have ontological implications:
246 TIME AND SPACE
1. Physical space has the geometry specified by our best physical theory.
2. This geometry is a matter of convention.
3. The character of objective reality is not a matter of convention.
4. Hence the geometry of physical space does not belong to objective reality.
5. As the bearer of this geometry, physical space does not belong to objective
reality either.
Whatever its merits, this reasoning will not convince those who adopt a realist
stance towards scientific theories, for anyone falling into this camp will reject
conven tion alism, and hence point 2. Fosters argument poses a threat even for
realists. He begins by assuming that there is a physical world, and that our best
scientific theories provide accurate guides to its laws and structure. Then, having
thus assumed that the scientific picture of the world is broadly correct, he goes on
to show that this assumption is incompatible with adopting a realistic stance with
respect to physical space.
Although I will be concentrating on Fosters spatial anti- realism, rather than
his more general physical anti- realism, a brief comment on the transition from
the latter to the former is in order. It may not be immediately apparent why an
argument that threatens the existence of physical space is a threat to the existence
of the physical world per se. Supposing that such an argument were successful,
shouldnt we conclude not that the physical world doesnt exist, but rather that
it consists solely of material objects? Bearing in mind what was said in Chapter 9
about the constraining and connecting role that space plays, this is not a plausible
option. Even if we could make sense of individual physical objects existing in the
absence of space, it is surely uncon tro versial that for a physical world to exist these
objects must be spatially connected. In the absence of space (whether construed
relationally or substantively) this is simply impossible.
15.2 The intrinsic and the inscrutable
We can start with a doctrine that plays an important role in what is to come. Foster
sometimes calls it the inscrutability of physical content. It can be broken down
into a conjunction of claims, one of which has two components:
1. The content thesis: at least some material things possess their own intrinsic
natures.
2. The inscrutability thesis: (i) it is impossible for us to know anything about
these intrinsic natures; (ii) these intrinsic natures do not fall within the scope
of physical science.
If intrinsic nature is taken to mean intrinsic property, 1 is clearly true, but 2
is equally clearly false. Shape, size, mass, colour, impenetrability and charge are
commonly taken to be intrinsic properties of material bodies, but all are eminently
scrut able. However, in talking of intrinsic natures Foster is not talking of intrin-
sic properties in this familiar sense, as the following passage makes plain:
from our observations, and from the way these support certain kinds of
explanatory theory, we may be able to establish the existence of an external
SPATIAL ANTI-REALISM 247
space with a certain geometrical structure we can never find out what
the space is like in itself Likewise, while, by the same empirical means, we
may be able to establish the existence of external objects located in this space
we can never discover the ultimate nature of their space- filling content
we can never discover, beyond a knowledge of their spatio- temporal proper-
ties, what the simplest objects (the elementary particles) are like in themselves
(Foster 1993: 295)
The idea that there is something that elementary particles are like in them-
selves has to be taken in the right way. The claim is not that there is something
that it is like to be an elementary particle; Foster is not saying that elementary
particles are conscious entities. The claim is simply that the most basic physical
things have some intrinsic characteristics over and above their spatiotemporal and
causal properties, over and above all the properties we ever actually ascribe to
physical things. Put in this way, the content thesis may well seem to lack motiva-
tion (to put it mildly). Why believe that these additional properties exist? In fact,
as quickly becomes apparent, the content thesis has a good deal of plausibility.
Consider the intrinsic properties we do ascribe to ordinary physical objects.
Some are spatiotemporal: size, shape and motion fall into this category shape,
size and mode of arrangement are sometimes called structural properties, for
obvious reasons.
As for the others, it may not be immediately obvious, but they are all causal
powers: that is, capacities to produce certain specific types of change in (or to be
changed by) other objects. This is often obviously the case. Objects that are rigid
have the capacity to retain their shape under pressure. Objects that are brittle have
a tendency to shatter when struck. Electric charge is associated with a variety of
causal properties, such as the capacity to attract or repel other charged objects
and the capacity to interact with magnetic fields. Mass is a causal property, too:
more massive things are harder to stop, and require more pushing to get them
moving. Likewise impenetrability: if we know that two particles are impenetrable
with respect to each other, we know that neither can occupy the space the other
occupies, that each has the capacity to repel the other at its borders, as it were.
Less obviously, physical colour is a causal capacity, too. Or at least, it is if we fol-
low Locke, Foster and many others in rejecting naive realism and subscribe to a
representative theory of perception, according to which, to say that a physical
object has a certain colour amounts to saying that the object is causally disposed
to produce in perceivers a certain sort of visual experience in certain sorts of cir-
cumstances (e.g. normal daytime lighting conditions). The same applies to all the
other secondary qualities, such as warmth, texture, and solidity (as these feature
in our experience).
The question now arises as to whether all non- structural intrinsic proper-
ties could be causal (or dispositional). If so, then the physical world as a whole
might consist of nothing but bare powers or bare potentialities; bare because
the potenti alities in question are not grounded in anything that is not itself a
potentiality. Foster maintains that this conception of physical reality is incoher-
ent: the causal potentialities that physical things possess must be grounded in
(i.e. possessed by) something that is not itself nothing but a potentiality. Hence
248 TIME AND SPACE
physical things must possess some properties that are not mere potentialities. These
non- dispositional (or categorical) properties are the intrinsic natures referred
to in the content thesis.
Foster is certainly right about one thing: the idea of a world consisting of noth-
ing but bare powers is a very odd one indeed. In a very real sense nothing can ever
happen in such a world. Such changes as there are amount only to alterations in
some of the various potentialities that exist; but since everything that ever exists is
a potentiality, any activation of a power only results in an alteration in some other
potentiality, and this potentiality in turn can never result in any actuality. Is a world
without anything that is more than a potentiality, a world in which the activation
of any potentiality can only bring about alterations in other potenti alities, really
possible? I suspect not.
Consider a simple world, consisting of just two atoms moving in space. Atom
A
1
has the power to repel another atom A
2
should it ever approach, and A
2
has the
power to repel A
1
. We can now ask, what is atom A
1
? Given that it is nothing but
a spatially located mobile disposition, we can only answer that A
1
is the capacity
to repel A
2
. But what is A
2
, if not the power to repel A
1
? This seems problematic,
since in order to make sense of what A
1
is, we need to make sense of what A
1
is a
power to do, since it is nothing but a causal power, but all that A
1
can do is repel
something which itself is nothing more than a power to repel A
1
, and of course the
same applies to A
2
, which is nothing more than a power to repel something which
itself is nothing more than a power to repel A
2
. It seems that, no matter where we
begin to consider the situation, we are left with nothing. Before we can take A
1

to be something that exists at all, we need to be able to specify what it is able to
do; but we cannot do this, because ex hypothesi all it can do is repel A
2
, and the
status of A
2
as something that exists is only secure when we have stated what it
can do, but this cannot be done, since ex hypothesi, all it can do is repel A
1
, whose
status as an extant thing we have yet to secure. So it seems that it is impossible for
the universe we are envisaging to contain anything at all, contrary to the initial
stipula tion that it contains two atoms.
The argument above concerned only two atoms and a single causal power. But
it is clear that adding other atoms (consisting of nothing but the power to repel
other atoms) will not alter the situation at all. Moreover, adding further causal
powers will not do any good, unless they are powers to affect something intrinsic
(i.e. non- dispositional), which is something we have ruled out. Going back to our
(allegedly) two- atom universe, let us suppose that each atom has the power to
attract the other, in addition to their powers of repulsion. As soon as we try to
spell out what each atoms attractive power is the power to do, we run into the
same problem as before: A
1
is the power to attract A
2
, which is itself nothing more
than the power to attract A
1
, and so on. Adding to the specification of A
2
that it
possesses the power to repel A
1
achieves nothing, since we already know that the
power of mutual repulsion is insufficient to secure the existence of either atom.
Once again, bringing additional purely dispos itional atoms into the picture does
nothing to improve matters (see Foster 1982: 6770).
The bare power ontology may be incoherent, but does it follow that all
physical things must possess an intrinsic nature in addition to their causal powers?
Probably not. But if some physical things are nothing more than mobile clusters
SPATIAL ANTI-REALISM 249
of causal potentiality, they can only exist because there are other physical things
that do possess intrinsic natures. As for which physical things fall into the latter
category, Foster is adamant on one point: a substantival space must possess an
intrinsic nature of its own. And he is surely right about this, at least on the assump-
tion that a substantival space, as a fundamental ingredient of physical reality, does
not depend for its existence on any other type of physical thing.
Let us turn from the content thesis to the inscrutability thesis. Now that we have
a better idea as to what Foster means by intrinsic nature, it is hard to disagree
with claim (i), which states that we know nothing about them. We know that (some)
physical things have intrinsic natures, but this is all we know: the precise character
of these intrinsic natures is a complete mystery. Science tells us a great deal about the
intrinsic properties that physical things possess, but as we have already seen, since
these properties are entirely causal or structural in character, they tell us nothing
about the intrinsic natures of the things that have these properties.
But Foster goes further: the inscrutability thesis also includes claim (ii), accord-
ing to which intrinsic natures do not even fall within the scope of physical science.
As for why this is so, Foster argues that physical concepts the concepts that it
is legitimate to apply to the physical world in scientific theorizing are without
exception entirely neutral on the question of intrinsic natures. This topic neutral-
ity thesis is a concep tual truth, or so Foster maintains: genuine physical concepts
are restricted to those that can feature in an empirically testable theory, and claims
about intrinsic natures cannot feature in such theories.
An example will help bring these claims into clearer focus. Imagine a world
super ficially very similar to our own, but consisting of a substantival space and
a single type of atom a simple and indivisible particle of small but finite size.
Physicists have a well confirmed and seemingly complete theory of atomic interac-
tions and motions, and this theory call it T is topic neutral. Now, there are any
number of competitor theories that are not topic neutral. These theories are the
same as T in all respects save one: they attribute intrinsic natures to both atoms
and space. Our purposes will be served by considering just two of these competi-
tors. According to T
1
, atoms are spheres of an intrinsic quality Q
1
, whereas space
consists of an infinite expanse of an intrinsic quality Q
2
. According to T
2
, space is
filled with Q
1
, and atoms are pervaded by Q
2
.
Now, Foster does not claim that we can make no sense whatsoever of theories
such as T
1
and T
2
, for we can, especially if we give the posited qualities a positive
interpretation in phenomenal terms (e.g. by thinking of Q
1
as red and Q
2
as blue, as
these colours feature in our experience). But it is one thing for a claim to be intelligi-
ble, and another for it to be an intelligible scientific claim. Real- life physicists do not
produce theories such as T
1
and T
2
, and any physicist who did would not be taken
Q
1
Q
2
Figure 15.1 Two competing
accounts of the distribution of
intrinsic qualities.
250 TIME AND SPACE
seriously; intrinsic natures, in Fosters sense, play no role in actual scientific theo-
rizing. Although the difference between T
1
and T
2
is easy to depict (and imagine),
it would be impossible to detect in practice. Since both T
1
and T
2
ascribe exactly
the same causal powers to atoms and space as T, they make no novel or testable
predictions: objects composed of Q
1
- type atoms look and behave in just the same
ways as objects composed of Q
2
- type atoms. Quite generally, the only properties
that science recognizes (or should recognize) are those that contribute to explaining
and predicting the observable features of the physical world. All such properties are
causal in character or have causal implication in the manner of structural properties;
it is precisely because of their causal character that they can contribute to explaining
the observed behaviour of material things. Since intrinsic natures are properties that
things possess over and above all their causal and structural characteristics, they do
not fall within the proper purview of science. Hence Fosters claim: the concepts
that can legitimately feature in science are topic neutral.
15.3 Modes of deviancy
Substantivalism, in its traditional guise, is the doctrine that space is an entity in
its own right, possessing its own inherent structure. The existence and structure
of such a space is (to some degree at least) independent of the existence, distribu-
tion and behaviour of any material bodies that it may happen to contain. If we
follow Foster, and subscribe to the content thesis, then a substantival space will
also possess an intrinsic qualitative nature of its own. Given the inscrutability
thesis, we cannot hope to discover what the intrinsic nature of our own space is.
Nonetheless, it is easy to imagine how a space could have its own intrinsic nature.
A Flatland space could take the form of a homogeneous circular blue plane. To
imagine a space with the same structure but a different intrinsic nature, simply
picture a homogeneous circular red plane. The surface of a red sphere represents a
Flatland space with the same intrinsic nature as the latter but a different structure.
Three- dimensional analogues of flat two- dimensional spaces are easily imagined
(just picture a colour- pervaded volume); three- dimensional analogues of curved
Flatland spaces are beyond our imagining, but certainly possible. As these exam-
ples make plain, the geometry of such spaces and their intrinsic natures are linked,
but the determination goes only one way: the geometry of a space is determined
by the structure of its qualitative filling, but spaces of different qualitative types
can have the same geometrical structure.
Fosters argument is general in scope, but easier to expound in the context
of simple spaces. So let us suppose that our space (or spacetime) call it S is
Newtonian in character, and that material bodies are composed of a single kind of
atom. To make matters more concrete, we can imagine these atoms as minuscule
mobile spheres of red, and S as an unbounded volume of uniform grey. Thinking
of these items thus is a useful way of reminding ourselves that they possess intrinsic
natures, but it should not be misconstrued: these colours are not visible to us.
S does not impede light, and so is entirely transparent, whereas material objects
reflect light in different ways (and so have different observable colours) depending
on the way their surface atoms are configured.
SPATIAL ANTI-REALISM 251
Since the material things we perceive are composed of atoms moving through S, it
is natural to think that S and physical space call it P are one and the same. Fosters
key claim is that S and P cannot possibly be one and the same: they are, and must be,
numerically distinct entities. The argument for this conclusion relies on the possibility
of certain modes of deviancy, ways in which P and S might differ from one another.
One mode of deviancy features variations in the intrinsic natures of S and its
occupants. For example, S could contain a cube- shaped region, R, which is sys-
tematically linked (through a law of nature) to variations in the intrinsic natures
of atoms: as atoms enter R they change from red to blue, and remain blue
until they leave (see Figure 15.2b). Both sorts of atom have exactly the same causal
powers, so they behave in exactly the same ways. Alternatively, we can suppose
that R itself has a different qualitative nature than the rest of S; for example, it is
a different shade of grey, as depicted in Figure 15.2c. Possibilities such as these
pose no threat to the status of the physical world; there is no obstacle to our hold-
ing that S and P are one and the same space. Nonetheless, they are a useful way of
bringing to the fore the implications of the inscrutability thesis.
(a) (b) (c)
Figure 15.2 Deviant and non-deviant distributions.
Since the qualitative changes exist solely at the level of the intrinsic natures
of the items concerned, given the inscrutability thesis, they are not empirically
detectable. Hence the qualitative variation (or deviancy) in S found in the case
in Figure 15.2c would not register in our physical theorizing: our best physical
theories would posit a homogeneous space, and a single kind of atom. Now, a
natural response to this situation would be to suppose that these theories provide
us with a largely accurate but incomplete account of the physical world, but this is
a mistake. Not only are the changes in intrinsic nature undetectable, but they are
not physical changes at all. If, as the inscrutability thesis states, physical concepts
are without exception topic neutral, then the worlds in Figure 15.2 fall under
precisely the same physical descriptions, and so are physically indiscernible in all
respects. The worlds are different, but not physically different, and the relevant
physical theories are not incomplete.
The mode of deviancy that Foster is most concerned with a mode that poses
more of a threat to the status of physical space is geometrical rather than qualita-
tive. His primary example (he provides plenty of others) consists of a case in which
the laws governing the behaviour of material things conspire to conceal their real
spatial locations. Although the mechanism involved is very simple, its dramatic
consequences may not be immediately apparent.
Once again S is a three- dimensional Newtonian space. Although S itself is
homogeneous, there are two spherical regions within it, R
1
and R
2
, some distance
252 TIME AND SPACE
apart, whose boundaries have unusual effects on any material objects that cross
them, effects due to the relevant laws of nature. Any object that enters R
1
emerges
in R
2
and continues on its way, with the same speed and in the same direction; simi-
larly, an object entering R
2
emerges in R
1
, and continues on its way, with the same
speed and in the same direction. An object that leaves R
1
appears at the boundary
of R
2
, and continues on its way, and in a similar fashion, an object that leaves
R
2
appears at the boundary of R
1
, and continues on its way. All of these transi-
tions occur instantaneously, and all forms of matter and energy (light included)
are affected. These deviant laws have the effect of making it appear as though R
1

is located where R
2
actually is, and vice versa. A concrete example will help make
this clear.
Let us suppose that R
1
is a spherical region of ten miles diameter in Cambridge-
shire, and R
2
is a similarly sized and shaped region in Oxfordshire. We go back
several thousand years, to a time when these areas were just starting to be occupied
by the first human settlers. A particular settler, walking through the region we now
know as Oxfordshire, finds that he wants to be alone, and so decides to build a hut
a few miles beyond his tribes most outlying village, which it so happens is close
to the boundary of R
2
. (Since this boundary is itself invisible, and has no observable
effects, no one realizes it is there.) Our settler walks five miles further on, and gets
to work. He builds a small hut. After a few months of solitude, our settler decides
he wouldnt mind some company after all, and it is not long before a small village
has formed, centred on his original hut. From this modest beginning, the city we
now know as Oxford developed. However, although Oxford appears to be within
the Oxfordshire countryside, in reality it is situated in Cambridgeshire, in region
R
1
. As our founding settler first crossed the boundary of R
2
he was re located inside
R
1
, and since R
1
is in Cambridgeshire, so too is the hut that he built, and the village
that developed around it. When the villagers went in search of suitable stones to use
in their constructions, they frequently crossed the boundary of R
1
, and emerged in
Oxfordshire; the material they brought back to the village after crossing the R
2

boundary all end up inside Cambridgeshire, which is why the resulting village is
constructed of what seem to be local materials (of the sort found in Oxfordshire),
even though it is actually situated in Cambridgeshire. All subsequent movements
of animals, people, matter, light undergo the same instantaneous translations. We
can tell a similar story for a settler walking through the Cambridgeshire fenlands:
on crossing the boundary of R
1
, this settler emerges in Oxfordshire, and founds the
village that develops into the city we know as Cambridge.
So we see that, in this imaginary scenario, the actual locations of Oxford and
Cambridge are the opposite of their apparent locations: Oxford is in Cambridge-
shire, and Cambridge is in Oxfordshire, but given the immediate re locations that
occur when things cross the boundaries of R
1
and R
2
, no one realizes this. Figure
15.3 (cf. Foster 2000: 268) illustrates the journey of someone travelling from
Oxford to Cambridge by car. As can be seen, the journey through S (space as it
really is) and the apparent journey, are very different.
1
Since light is similarly affected, anyone flying high over Oxfordshire who looks
down on to R
2
will see light emanating from R
1
(as depicted in Figure 15.4), and
so will actually perceive Oxford, even though Oxford in fact lies a good many
miles to the east.
SPATIAL ANTI-REALISM 253
Oxfordshire
R
2
Oxford
R
1
Figure 15.4 The consequences of deviancy. Light from a spherical region of S in Cambridgeshire
(where Oxford is situated) emerges in Oxfordshire, leaving the pilot of a plane flying over Oxfordshire with
the impression that Oxford lies directly below.
Oxford Cambridge
Cambridgeshire Oxfordshire
t
1
t
2
t
3
t
4
apparent journey
R
2

corresponding
to Cambridge
R
1

corresponding
to Oxford
t
3
t
4
t
3
t
2
corresponding to
Oxfordshire
corresponding to
Cambridgeshire
the journey through S
Figure 15.3 The OxfordCambridge case.
t
2
t
1
254 TIME AND SPACE
It might be thought that none of this makes any sense: if all physical proc-
esses are nomologically organized in exactly the same manner as they would be if
Oxford were really in Oxfordshire, and Cambridge in Cambridgeshire, what sense
is there in the idea that this is not the case? It is here that the underlying (but invis-
ible) space S plays a crucial role. With respect to this space, which we are assuming
is perfectly real, Oxford really is in Cambridgeshire, and Cambridge really is in
Oxfordshire. As noted earlier, if per impossible, we could see S, it would be glar-
ingly obvious that Oxford is surrounded by the Cambridgeshire countryside, and
Cambridge by the Oxfordshire countryside.
15.4 Intrinsic versus functional geometry
Let us say that the laws governing the behaviour of atoms in space are nomologi-
cally uniform if they put the same constraints on the behaviour of atoms at all
points in space, and, in particular, the laws apply to all the geometrical properties
of atoms in a uniform way throughout space (and time), where by the geometrical
property of an atom we mean its shape and possible spatial trajectories. It is clear
that the laws operative in S are not nomologically uniform: they require atoms to
move continuously in regions outside R
1
and R
2
, but discontinuously across the
boundaries of these two regions, and a spherical atom will remain spherical as
it moves through space except when it crosses a boundary of R
1
or R
2
, at which
point it ceases to be spherical, and for a while is, in effect, divided into two parts.
In addition to not being nomologically uniform, the laws operative in S have a
distinctive feature: they are such that, if the regions R
1
and R
2
were interchanged,
then nomological uniformity would be restored. Since the imagined interchange
would result in R
1
(which contains Oxford) being located in Oxfordshire, and
R
2
(which contains Cambridge) being located in Cambridgeshire, the envisaged
laws no longer force objects to follow discontinuous paths. A car driven through
Oxfordshire along a road that leads to Oxford will pass through Oxford (not
Cambridge): as the car crosses the critical boundary it emerges in R
1
, just as
before, but since R
1
is now in Oxfordshire rather than Cambridgeshire, the cars
path through S is continuous.
Taking nomological uniformity to be the norm as far as the nomological
organization of a world is concerned, Foster calls worlds that are nomologically
non- uniform and also possess this distinctive feature nomologically deviant. A
nomologically deviant physical world is one that is not nomologically uniform,
but in which nomological uniformity would be brought about by some alteration
in the geometry of the world. In the case we have been considering, the required
alteration is comparatively mild the points in two small regions have simply to
be switched around but Foster describes more radical cases where the deviance
is at the level of dimensionality (the laws range over three spatial dimensions, but
the underlying space is only two- dimensional).
Foster next draws a distinction between what he calls functional and intrinsic
geometry. The intrinsic geometry of a space is its real geometry (the colour-
determined geometry in the case of S). The functional geometry is the geometri-
cal structure that empirical theorizing (whether at the level of common sense
SPATIAL ANTI-REALISM 255
or advanced science) will lead us to ascribe to the physical world. Since space
itself is invisible, we have no direct empirical access to its real geometry, and so
functional geometry is ascribed on the basis of the observed behaviour of material
bodies together with the assumption that the laws of nature are uniform (always
the default assumption in the absence of evidence to the contrary). In short, func-
tional geometry is the apparent geometry of a world, the geometry the world in
question would seem to have to beings such as ourselves beings with no direct
empirical access to the real (intrinsic) geometry of the underlying substantival
space. In the OxfordCambridge case, the functional and intrinsic geometries are
clearly different. Let us call the intrinsic geometry in this scenario G
i
and the
functional geometry G
f
. According to G
f
, the points in R
1
(those underlying
the various material processes in and around the city of Oxford) are close to,
and surrounded by, points in Oxfordshire, whereas, according to G
i
, the points
underlying Oxford are close to, and surrounded by, points in Cambridgeshire;
the same applies, mutatis mutandis, for Cambridge and Oxfordshire. G
f
and G
i

are thus distinct, since they assign different metrical and topological properties to
points and regions associated with the same physical objects and processes. The
question of quite how this situation should be interpreted is a difficult one, and
we will be turning to it shortly.
It is worth noting that Fosters distinction between intrinsic and functional
geometry is not the same as the usual distinction between pure and applied (or
physical) geometry. Both functional and intrinsic geometries are applied; the dif-
ference between them lies in what they are applied to. The intrinsic geometry of a
space is the one we would ascribe to it if we could look upon the space as it really
is, from an all- revealing transcendental perspective (Foster sometimes uses the
term transcendentally natural intrinsic (or TNI) geometry). Construing a space as a
single colour- field gives us a concrete way of understanding what this means. The
functional geometry of a space, on the other hand, is the geometry the space would
possess if the laws controlling the behaviour of its occupants were nomologically
uniform. Fosters contention is that the two could differ. Provided we grant his
assumption that a genuine substantival space possesses both an intrinsic nature
and a spatial structure that is independent of the behaviour of its occupants, then
he is surely right.
15.5 The nomological thesis
Foster now goes on to argue that it is logically impossible for any physical space
to be nomologically deviant, since it is a conceptual truth that physical geometry
coincides with functional geometry. He calls this claim the nomological thesis,
since in effect it posits a conceptual connection between the structure of physical
space and the laws governing the behaviour of its occupants, these same laws being
the basis upon which functional geometry is constructed. The nomological thesis
is crucial to the anti- realist argument, and its defence occupies several chapters of
The Case for Idealism.
2
The first point to note is that there could never be any empirical evidence in
favour of the hypothesis that a physical space is nomologically deviant. Empirical
256 TIME AND SPACE
evidence points, however unreliably, towards functional geometry alone. By defini-
tion, the hypothesis that the geometry of physical space is its functional geometry
will be simpler than, and to this extent preferable to, any alternative geometrical
hypothesis (which does not preserve uniformity), provided we take nomological
uniformity to be our criterion of simplicity, which of course we do. Consider,
for example, what we would make of the hypothesis that Oxford is really in
Cambridgeshire, when all the observational evidence suggests that Oxford is in
Oxfordshire. Although someone could construct a theory involving non- uniform
laws that explains why it seems that Oxford is in Oxfordshire when really it is in
Cambridgeshire, the needless complexity of this theory means we would never be
inclined to take it seriously.
But Fosters claim is stronger than this. He holds that not only could we never
have any scientific grounds for accepting the hypothesis that our physical space
is nomologically deviant, but the hypothesis itself is incoherent. Consider again
the Oxford and Cambridge case, in which the intrinsic geometry of the external
colour- field (S) does not correspond with the functional geometry. What can we
say here about the physical world and its geometrical structure? Three positions
are open to us:
1. The physical world does not exist.
2. The physical world exists, and its geometry is the intrinsic geometry of the
underlying space S.
3. The physical world exists, and its geometry is its functional geometry.
Foster suggests that 1 is not a genuine option: It is just obvious that the local
organizational quirk with respect to R
1
and R
2
, and the resulting effects on empiri-
cal appearance, do not suffice to eliminate the physical world altogether (2000:
269). Since this is clearly right, the issue is between 2 and 3.
In regard to 2, Fosters claim is that it simply makes no sense to think that the
physical world could have a structure that is both quite different from its apparent
structure and wholly concealed from us as a consequence of natural law. However,
as Foster acknowledges, it is far from obvious that this is so, and he defends his
claim with a variety of arguments.
Where the deviation between functional and intrinsic geometry is as slight as in
the Oxford and Cambridge case, it might be objected:
Surely in this case it does make sense to hold that the structure of physical
space might be that of the intrinsic geometry, even if we could never discover
what this geometry is. After all, our concep tion of the physical leaves room
for some discrepancy between how the physical world really is, and how our
best scientific theory represents it as being.
The force of this point can be countered by considering cases in which the functional
and intrinsic geometries deviate more wildly. Foster supplies two such scenarios
(in enough detail to show that they are workable). In the first, the underlying
geometry is two- dimensional, whereas the functional geometry of the resulting
physical world (i.e. the world to which we have access) is three- dimensional. In
the second case, a single physical space is sustained by events in two quite sepa-
rate spatial systems, so there is spatial unity at the functional level, but disunity
SPATIAL ANTI-REALISM 257
at the intrinsic level. Since in both cases we can suppose that the empirically
manifest world is exactly like the physical world we are actually acquainted with,
it would be very implausible to claim that the physical world in either scenario
really possesses the structure of the underlying space (a structure to whose exist-
ence we are entirely oblivious). A more plausible position would be that physical
space exists in such cases, as a unitary three- dimensional continu um, but given the
discrepancy between the geometries of this space and the under ly ing reality, we
cannot regard physical space as being metaphysically fundamental.
What Foster says might be correct for cases in which there are radical dis-
crepancies between intrinsic and functional geometries, and we must admit that
such cases are logically possible, but what of those cases in which intrinsic and
functional geometries exactly coincide, or nearly coincide (as in the Oxford and
Cambridge case)? Cant we say that in these sorts of case the physical geometry is
the intrinsic geometry, or at least, that it makes sense to think this? Foster rejects
this. What grounds could there be for adopting two different methodological prin-
ciples for selecting a geometry for physical space? If we admit that the func tional
criterion is appropriate in cases of radical divergence, then it must be appropriate
in all cases, since even when functional and intrinsic geometries coin cide, our
epistemological predicament is exactly the same as when the geometries are radi-
cally different. We never have any evidence that functional and intrinsic geometries
are similar, and so never have any grounds for employing any selection criterion
other than the functional one.
But, an objector might continue:
Isnt the hypothesis that the functional geometry coincides with the intrinsic
geometry the simplest hypothesis? Doesnt this hypothesis yield a simpler
set of physical laws? If we take nomological simplicity to be an explanatory
virtue, doesnt this hypothesis provide the best explanation for the worlds
being as it seems to be? If so, then the hypothesis that physical and intrinsic
geometries coincide at least makes sense, even if it is not true.
The inscrutability thesis comes into play here. Foster does not deny that sim-
plicity considerations have a role to play in the selection of an explanatory hypoth-
esis concern ing the structure and laws of the physical world. But since physical
theory is topic neutral, simplicity considerations can only enter into the empirical
justifi cation of theories that make no claims about the intrinsic nature of physical
reality. The hypothesis that intrinsic and functional geometries coincide clearly
isnt topic neutral, and so is not a hypothesis to which the usual canons of empiri-
cal reasoning can apply.
This point seems decisive, and doesnt rest simply on an appeal to intuitions
about the sorts of reality it is appropriate to call physical. The intrinsic geometry
of the external item underlying the physical world is grounded in the intrinsic
nature of this item; for example, the intrinsic geometry of the colour- field S is
fixed by the trajectories and distance relations between the different regions of this
field. The same applies in other cases, no matter what the intrinsic nature of the
external item is (we cant of course imagine in any way the nature of this item if
we suppose it to be non- phenomenal). Now, if we accept the inscrutability thesis,
it follows that considerations concerning the intrinsic nature of the external item
258 TIME AND SPACE
cannot play any role at all in the considerations that lead to the adoption of a
physical geometry. The only considera tions that do have a role to play are those
concerning the behaviour of those aspects of the external item to which we have
empirical access. The empirical evidence may under determine physical spatial
geometry there may be a family of competing hypoth eses each assigning to the
physical world a different combination of geometries and laws, each of which is
equally compatible with all possible empirical evidence but every member of this
family will be an acceptable hypothesis only because it helps explain the observed
behaviour of physical objects. The intrinsic geometry of the underlying external
reality does not, and cannot, enter into the selection principles we employ in
choosing a physical geometry.
If we grant this much, the case for the nomological thesis seems compelling: it is
a conceptual truth that the physical geometry of a space is its functional geometry.
15.6 Nomological contingency
One further general assumption is relevant to the main argument. Foster assumes
that the physically relevant geometry of S is contingent. I shall refer to this as the
nomological contingency assumption. The nomological contingency assumption
amounts to the claim that the physically relevant laws governing S and its occupants
could be different from how they actually are. Given its importance to his central
arguments, it is somewhat surprising to find that Foster never provides any argu-
ment for the nomological contingency assumption; he simply takes it to be self-
evidently true. However, one does not need to reflect for very long before finding
oneself quite convinced that the assumption in question is eminently reasonable.
Suppose that S is a colour- field, and its occupants are tiny mobile spheres of
(a different) colour moving about within it. It is easy to imagine these spheres
conforming to any number of different laws of motion without the structure of S
changing. The intrinsic geometry of S is no more affected by the behaviour of the
atoms that move through it than is the geometry of a cinema screen by the play of
light across its surface during the projection of a film. Although this sort of exercise
lends the nomological contingency assumption considerable prima facie plausibil-
ity, we are dealing here only with a model. Might it not be that the laws governing
real- life equivalents of S and its occupants are necessary rather than contingent?
If the bare powers metaphysic applied that is, if a purely causaldisposi-
tional account of the constituents of the physical world were viable then it would
be very plausible to maintain that physical things possess their causal powers essen-
tially, since such things are nothing but clusters of causal powers. But given the
content thesis, S and (at least some of) its occupants must possess space- filling
intrinsic natures of some kind, so this hypothesis can be ruled out. Conse quently,
the question we need to focus on is whether there could be necessary connec-
tions between intrinsic natures and their causal powers. Since the intrinsic natures
of S and its occupants are entirely inscrutable, it is hard to see how the claim
that a particular intrinsic nature is necessarily associated with a particular set of
causal characteristics could ever be substantiated. A non- contingent connection
between an intrinsic nature and its causal characteristics cannot be analytic: since
SPATIAL ANTI-REALISM 259
ex hypothesi we have no conception of the intrinsic nature we are considering, no
amount of conceptual analysis will do the job. But nor can the connection be a
posteriori since, given the inscrutability thesis, no amount of empirical investiga-
tion will shed even a glimmer of light on the intrinsic nature of something external
to our minds. The claim that water is necessarily H
2
O may be an a posteriori truth,
but this truth does not reveal any connection between an intrinsic nature and a
set of causal properties, simply because we have no idea of the intrinsic nature of
hydrogen and oxygen atoms (or their constituents).
The nomological contingency assumption is also supported by purely meta-
physical considerations. We are working on the assumption that S is the metaphys-
ically basic item. If we held that S itself could not exist if its occupants obeyed
different laws, we would be saying that S depended for its existence on its occu-
pants behaving in a specific way, a claim that is manifestly incompatible with the
assumption that S is ontologically basic, for an ontologically basic item is one that
is dependent on nothing else for its existence.
15.7 Realism rejected
We can now, finally, turn to the crucial part of Fosters case: the argument that
yields the conclusion that physical space cannot be a part of what is metaphysically
basic, or ultimately real. To simplify, we again assume that the external reality
corresponding to the physical world is a giant colour- field, S. We also assume that
the physical world exists, and that one of its metaphysically basic constituents
is physical space, P. Facts about physical objects supervene upon facts about the
distribution of atoms through S. We have seen already that there is a sense in
which S can possess two geometries: its own intrinsic geometry, and the functional
geometry that applies to its occupants. Since the physically relevant geometry is
the functional geometry, the geometry of P is the functional geometry of S. The
question now arises as to the relationship between these two spatial items, P and
S. Fosters claim is that S and P are, as a matter of logical necessity, numerically
distinct entities.
His argument rests on some widely accepted assumptions concerning the
identity- conditions for spaces and their constituent points. The most general level
at which the geometrical properties of a space can be characterized is topological.
Spaces that are topologically indistinguishable may be different in other geomet-
rical respects (e.g. as to whether or not a Euclidean metric applies). Foster makes
the minimal assumption that the topology of a space is essential to it.
3
So for a
given space S with topology T, in any possible world in which S exists, it pos-
sesses T. We are assuming here, of course, that S is a concrete item, a particular,
not merely a mathematical structure (which will exist in all logically possible
worlds). Foster further assumes that spatial points owe their individuality to the
spaces to which they belong: whatever geometrical features are essential to a space
are also essential to the points that constitute the space.
Now, from this it follows that there are logically possible worlds that outwardly
resemble ours in all respects, but where S and P are numerically distinct. Foster
provides several examples of worlds where S and P are topologically different the
260 TIME AND SPACE
OxfordCambridge scenario is one such and in these cases not only must S and
P be numerically distinct spaces, but none of the constituent points in S exist in
P, and vice versa. But, as Foster is fully prepared to concede, we have no reason
whatso ever to suppose that our world is like this. Suppose that in our world there
is no discrepancy or deviance; the laws are uniform and S and P have the same
topology. In this instance, is there any obstacle to holding that S and P are one
and the same space? Foster maintains that there is: S and P may share their actual
properties, but they still differ in their modal properties.
If P is a genuine space, as we are currently assuming, it must possess its topology
essentially. As it happens, S shares this topology. But does it do so essentially? No.
Given the nomological contingency assumption, there are possible worlds where S
has exactly the same intrinsic geometry as in the actual world, but where the laws
governing its occupants are such that it has a very different functional (physically
relevant) geometry. In some of these worlds the functional topology of S is differ-
ent from that of P. In such worlds P does not exist (in its place is a quite distinct
physical space). Hence there are possible worlds where S exists but P does not.
If we assume that identities are necessary (i.e., that if x = y, then x = y in every
possible world where either exists), then S and P cannot be numerically identical.
The same conclusion can be reached by a slightly different route. Leibnizs prin-
ciple of the identity of indiscernibles may be highly controversial, but the converse
the indiscernibility of identicals is not. According to this principle, if x and y are
numerically identical then x and y must possess the same genuine properties. Need-
less to say, if x and y are merely different names for the same object, it is hard to
see how this principle could be false. Now, S has a physically relevant geometry with
a certain topology; call this topology T. P has precisely the same topology. But there
is a difference: S possesses T only contingently, but P does not. Although in actual
fact both S and P have the same physically relevant topology, it is possible for S to
exist without this topology (if the occupants of S obeyed different laws of motion,
Ss physically relevant topology would be different), but P cannot do likewise (on
the assumption that P is a genuine space in its own right, it possesses its topology T
essentially). Let us give the modal property could exist without T a convenient
label, X. The argument that P and S are numerically distinct is simple:
1. S possesses property X.
2. P does not possess property X.
3. So, by the indiscernibility of identicals, S cannot be numerically identical with
P.
Provided that X is a genuine property, which it certainly seems to be, the argument
is sound.
4
Fosters argument bears a striking resemblance to a somewhat more familiar
argument concerning the relationship between ordinary physical objects, such as
statues and plastic dishpans, and the matter out of which they are composed.
The idea that two physical objects could exactly coincide in space throughout
their entire careers, but nonetheless be numerically distinct may seem odd (just as
odd as Fosters claim that S and P can have the same geometry, but nonetheless
be numerically distinct spaces). But there is a powerful argument that leads to
precisely this conclusion.
SPATIAL ANTI-REALISM 261
To take the best- known example, consider Lumpl and Goliath (Gibbard 1975).
Goliath is a statue, and Lumpl is the lump of clay from which Goliath is made
(we can treat both Lumpl and Goliath as proper names, since we can name
lumps of clay if we so choose). Goliath and Lumpl come into existence at exactly
the same time, and both cease to exist at exactly the same time. (Take a lump
of clay to be a collection of bits of clay that are stuck together. There were first
two lumps of clay. The first is modelled into an upper- body shape, the second into
a lower- body and these two lumps are then stuck together at the waist, bringing
into being both a new lump of clay, Lumpl, and a new statue, Goliath. At a later
date, Lumpl- Goliath is smashed to bits.) Since Lumpl and Goliath are composed
of exactly the same material constituents throughout their respective careers, and
likewise share exactly the same physical properties (at least of the kind recognized
by physics), it could be argued that there is only a single physical object here, not
two. However, there is an obstacle to this identification.
Even though Lumpl and Goliath coincide (occupy the same volume of space
at each moment) throughout their spatiotemporal careers, it is possible that they
might not have done. Shortly after completing Goliath the artist might have become
bored with his work, and squeezed Lumpl into a ball. Provided that no bits of clay
drop off during the squeezing, Lumpl survives; but Goliath clearly does not. This
doesnt happen, but it might have. In which case, the modal property might have
been squeezed into a ball without being destroyed is possessed by Lumpl but
not by Goliath. Hence, by the law of the indiscernibility of identicals, Lumpl and
Goliath cannot be numerically identical. Anyone who accepts this conclusion (and
many do) must also accept the validity of Fosters modal argument. If they wish to
resist his conclusion, they must do so by rejecting his premises.
If we do accept Fosters conclusion, that S and P cannot be numerically identi-
cal, the only remaining task is to settle on the ontological status of physical space.
Since physical space exists only in virtue of the particular contingent laws that
apply to the underlying external reality, and the latter could exist without sustain-
ing any physical space at all, it is clear that physical space has an ontological ly
dependent status, contrary to our initial assumption. Furthermore, if we take the
view that physical space is the metaphysically basic item in the physical world, and
that physical objects depend for their existence on physical space, then given our
conclusion that physical space is necessarily numerically distinct from the space of
the underlying reality (should this reality be spatial at all), the underlying reality
is, necessarily, wholly non- physical. Even if we discard this assumption, and take
the view that material things are not ontologically dependent on physical space,
we are left with the conclusion that a substantival space cannot be an ontologi-
cally basic ingredient of reality. Spatial realism in its most interesting form is thus
undermined.
One last point. Although he argues that we should be anti- realists with respect
to physical space, Foster is not saying that statements such as Paris is closer to
London than to New York cannot be true. His position is analogous to that of the
B- theorist, who recognizes tensed truth while maintaining that reality itself is not
tensed: the truthmaker of a tensed statement such as World War One is past is
the tenseless fact that this token statement occurred at a time later than the end-
ing of the First World War. In a similar vein Foster would argue that statements
262 TIME AND SPACE
about the spatial properties of physical objects have as truthmakers states of affairs
that are entirely non- physical. Foster is perfectly happy to concede that there are
physi cal facts, but only in the trivial sense that there is a fact corresponding to
every true statement. Since statements about the physical world are made true by
non- physical facts, physical facts are not basic; rather they supervene upon, or
are logically sustained by, non- physical facts. What holds for facts also holds for
objects. Not all objects are basic: chairs exist, but in our world at least chairs
are always composed of various particles, and any particular chair supervenes
on, or is the logical creation of, facts concerning these more basic entities. For
Foster a fact is basic if it is not logically sustained by any other fact or facts, and
an entity is basic if it is not the logical creation of any fact or facts. What he calls
ultimate reality is the totality of basic entities and facts. Formulated in these
terms, Fosters anti- realism amounts to the claim that ultimate reality is wholly
non- physical. So, on the assumption that there are physical entities and facts, these
must be logically dependent (sustained or created) by something else, some reality
that is non- physical. Foster suggests that this ultimate reality is mental, hence his
idealism. But whatever its nature, it is this non- physical reality that is responsible
for our sensory experience being as it is.
15.8 Geometrical pluralism
Any argument as complex as Fosters can be challenged in different ways. For
example, the argument presupposes the necessity of identity and geometrical
essentialism the view that the geometry of a space is essential to the identity
of all its points and although both doctrines are widely accepted, they could
certainly be questioned.
5
Indeed, if there were no alternative, many might prefer to
abandon one of these doc trines rather than accept Fosters anti- realist conclusion.
But a better response is available, and emerges as soon as we question another basic
assumption about the nature of space, an assumption that is by no means unique
to Foster, but that his own argument shows to be questionable.
Not only is Foster committed to geometrical essentialism, but he is committed
to what we can call geometrical monism. His anti- realist argument is based on
the unstated assumption that a genuine space can only have a single geometrical
structure. It is taken as obvious that when P and S have different geometries they
must be numerically distinct; even when their geometries coincide the possibility
that they could diverge ensures their distinctness. But if a genuine space is not
restricted to a single geometry if geometrical pluralism obtains we have the
option of regarding the divergent geometries of S and P as belonging to one and
the same space. If a case for pluralism can be sustained, realism can be restored
without rejecting the other essentialist doctrines that Fosters argument requires.
We saw in 9.6 that there are two viable accounts of distance relations within
a space. Intrinsic distance consists of a direct spatial relation between a pair of
points; Gaussian distance is measured by the shortest continuous path through
space. Although the two accounts yield the same results for classical Euclidean
spaces, they can easily diverge; for example, creating a hole in a space leaves the
intrinsic distances between points on either side unaltered, but increases their
SPATIAL ANTI-REALISM 263
Gaussian distances. When presenting these accounts I tacitly assumed that the
points in any given space could be related by only one sort of distance relation, but
there is no obvious reason why this must be so. Consider the case of a corrugated
space: an infinite two- dimensional plane that undulates in a regular way, its crests
and troughs taking the form of alternating half- cylinders (see Figure 15.5). Call
this space C and compare it with the flat Euclidean plane E. When a flat surface
is bent without any tearing or stretching, the lengths of paths confined to the
surface do not change, so from the Gaussian perspective C and E are geometrically
equivalent (both have zero intrinsic curvature). But clearly, if we suppose that
the points in C are connected by intrinsic distance relations indicated by the
arrow- headed lines in Figure 15.5 the spaces are not geometrically equivalent.
Corrugating a space changes the distances between points; for example, the intrin-
sic distances between points on successive mounds (and valleys) are shortened as
they are moved together.
P
Q
R
S
T
U
V
Figure 15.5 Gaussian versus intrinsic distances in a corrugated space. In a flat space, intrinsic
and Gaussian distances diverge when points are deleted from the manifold. In a curved space there are
other ways for divergence to occur.
On the assumption that the intrinsic distance relations are genuine properties
of the points in C it is tempting to suppose that C is really a bent space, that its
true geometry is the one captured by the intrinsic measure of distance. But the
Gaussian conception of distance remains perfectly valid; the facts about the con-
tinuous path- lengths between points within C are not affected by the existence or
non- existence of intrinsic distance relations. The significance of this fact is easily
obscured by what happens when we try to picture C: we will embed it within
three- dimensional space and imagine ourselves looking down on a surface that vis-
ibly undulates. If we can see the bends in C, isnt it obvious that this space is really
bent? However, the obviousness of the ripples in C evaporates when we abolish
the embedding space and the Gaussian paths through three- dimensional space
connecting its peaks and troughs. If we were Flatlanders living in C, no ordinary
measuring devices would reveal the ripples in our space and their existence would
not be obvious at all.
This case could be construed in a Fosterian way. The true geometry of C is its
intrinsic geometry (as determined by intrinsic distances), for a being equipped
with transcendent God- like powers of observation would discern the intrinsic
distance relations between the points of C, and hence the ripples. The Gaussian
geometry of C is clearly its physical geometry, since the Gaussian properties of
a space are those that can be discovered by geometers working within the space
using rods, strings and so forth. Since C is intrinsically corrugated but physically
flat, physical space is numerically distinct from C and, given that C really exists,
264 TIME AND SPACE
physical space itself cannot be ultimately real. But this interpretation of the case is
far from compelling. For, as is clear, the Gaussian geometry is a real feature of C
itself. Gaussian distances comprise the shortest continuous path- lengths through
a space, and so are determined by the presence or absence of points within C: if a
hole in C is created, path- lengths connecting points at opposite sides of this region
are increased. Rather than populating our ontology with additional (and not fully
real) spaces, why not simply accept that C possesses two equally valid geometrical
structures simultaneously, each structure depending on real features of C itself
Gaussian and intrinsic distances.
The discrepancy between physical and intrinsic geometry in C is of a compara-
tively mild form. The sort of deviancy exhibited in Fosters Oxford and Cam-
bridge example is not only more extreme geometrically, but a product of a form
of nomological deviance that is absent from C. What it reveals is different, too:
not the possibility of Gaussian and intrinsic distances coexisting while diverging,
but rather the existence of two distinct forms of Gaussian or path- based distance,
which likewise can coexist without coinciding.
We can distinguish two varieties of shortest path- length, which I will call
intrin sic geodesics and material geodesics. The material geodesics of a space
are the paths of shortest distance as measured by some specified physical process.
A natural (and standard) way of giving this notion substance is to say that inertial
paths (those followed by bodies that are not being acted on by a force) are geodesics.
Other defini tions are perfectly conceivable, but for present purposes the inertial
account will serve. Intrinsic geodesics, on the other hand, are defined relative to the
structure and composition of space itself, and so independently of the behaviour
of the material (or other) contents of a space. On the assumption (which we are
currently making) that space is substantival, and so a real entity in its own right, the
idea that space has intrin sic features that possess a specific geometrical structure
seems eminently reasonable (all the more so given the strength of Fosters case for
the content thesis). There is no denying that an entity shaped like the surface of a
sphere is geometrically distinct from one shaped like a flat piece of paper; likewise,
a flat Euclidean plane with a hole and an otherwise similar plane without a hole are
geometrically different. Quite what geometrical features a given space possesses
will depend on the nature of the spatial fabric concerned, which may come in very
different forms. It is sufficient for present pur poses that we accept the hypothesis
that a substantival space can possess geometrical features that are intrinsic to it, and
this is something most substantivalists will be prepared to accept.
We can now draw a further distinction, between naive and circumspect substan-
tivalism. The naive substantivalist assumes without question that material and
intrinsic geodesics are one and the same; the idea that they could be different
is never considered or taken seriously. The circumspect substantivalist is willing
to countenance at least the possibility that the two might not always coincide.
On finding a region of space where bodies in inertial motion do not move in
(Euclidean) straight lines, and where this deviation cannot be explained in terms
of known forces pulling or pushing the bodies, the naive substantivalist will imme-
diately conclude that a non- Euclidean hole has been discovered (a hole in Ner-
lichs sense, a region of strong local curvature in an otherwise flat space, rather
than an absence of points). While circumspect substantivalists will accept that
SPATIAL ANTI-REALISM 265
this is one possible explanation for the peculiar motions, they will be prepared to
consider an alternative: that the intrinsic geometry of space in the relevant region
is flat, but the nomological organization of the world is such that it seems holed;
bodies behave as if they are moving along geodesics in a holed space, but they are
not. In short, in this region the material geodesics and the intrinsic geodesics fail to
coincide; perhaps when the universe was created occasional lapses in nomological
uniformity occurred. Unless this possibility can be ruled out, nor can the alterna-
tive explanation, or so circumspect subtantivalists will argue.
The naive substantivalist could try arguing thus:
You should reconsider your position. If you are prepared to recognize this
sort of explanation for the deviant behaviour, you are committed to recog-
nizing another possibility: there could be variations in the intrinsic geometry
of space that are concealed by nomologically deviant behaviour. There could
be regions of space that are intrinsically curved, but where the material geo-
desics are Euclidean, thanks to the laws governing spacebody interactions
that apply in these regions. Isnt it absurd to posit physical structures that
are by their very nature empirically undetectable?
But it is by no means clear that this objection should sway the circumspect sub-
stantivalist, who can offer this response:
It would be nice to rule out even the possibility of material and intrinsic
geodesics coming apart, but this is simply not an option. There is no rea-
son to suppose that it is metaphysically necessary for bodies in force- free
inertial motion to follow intrinsic geodesics. If they do follow them, this is
only because of the particular nomological organization that obtains in this
world; other nomological regimes are perfectly conceivable. Once the pos-
sibility of geodesic dislocations is recognized, the possibility of concealed
dislocations must also be recognized. This said, there is no reason whatso-
ever to suppose that such dislocations are commonplace, or even exist at
all: for like you I believe we have good reason to suppose that our world
is governed by basic physical laws that are simple and uniform. Unlike you,
if we did discover a region where (for no apparent reason) the material
geodesics are curved, I am prepared to countenance the possibility that this
deviation from the norm is due to unusual laws rather than variations in the
shape of space itself.
In the absence of any reason for thinking that dislocations between intrinsic and
material geodesics are absolutely impossible, I think we should conclude that sub-
stantivalists should be circumspect rather than naive.
In our earlier case of geometrical conflict (intrinsic distance versus Gaussian dis-
tance) I suggested that the substantivalist should embrace the option of geometri-
cal pluralism, and hold that both geometries are simultaneously possessed by the
space in question. The same applies when material and intrinsic geodesics fail to
coincide. Here, too, pluralism is a compelling option: since material geodesics are
paths through physical space that are privileged by physical law, there is every
reason for the substantivalist to accept them as genuine geometrical features. On
the assumption that intrinsic geodesics also exist, pluralism is unavoidable.
266 TIME AND SPACE
It might be claimed that intrinsic geodesics are more basic from a metaphysical
perspective than their material counterparts. This claim has some justification:
there are certainly counterfactual situations in which a space retains its intrinsic
geodesics but has different material geodesics, thanks to alterations in the relevant
natural laws. But this fact does nothing to impugn the status of material geodes-
ics as geometrical features of a substantival physical space; after all, not all the
features of a physical entity are essential to it. Indeed, if we focus on the physical
world as a whole, rather than its space, it is the material geodesics that have most
prominence: as the paths privileged by the laws connecting material objects with
space, they provide us with the only clues we have as to the structure of our space.
These amount to strong grounds for taking both geodesic structures as essential to
the identity of a space, at least for worlds where they come apart.
Where does this leave us vis- - vis Fosters case for anti- realism? Let us grant
that in worlds where the discrepancy between the real and apparent spatial struc-
ture of the world is large, physical space is as Foster maintains: an indispensable
theoretical construct that supervenes on a non- physical reality. (I have in mind
Fosters more radical examples, where the physical world partly depends on each
of two distinct external spaces, or a two- dimensional external space.) Since, as
Foster himself concedes, we have no reason to believe that our world is anything
like this, the more interesting possibilities are where the discrepancy between real
and apparent spatial structure is comparatively small or non- existent. In these
instances, by rejecting geometrical monism, and accepting that a substantival
physical space can have a multiplicity of different geometrical structures at the
same time, we can accept that the scenarios envisaged by Foster are possible with-
out rejecting realism. In the Oxford and Cambridge example there are extreme
but highly localized divergences between material and intrinsic geodesics, and the
circumspect substantivalist will regard both geometries as real features of the same
spatial entity: physical space. No less importantly, in the case of the actual world,
where, as Foster willingly concedes, we have no reason to suppose that material
and intrinsic geometries diverge, the mere possibility that they could gives us no
reason to reject realism, for the realist can take the view that the material geodesics
are geometrical features of a physical space that are possessed only contingently.
But perhaps the realist does not need to concede even this much. If we choose to
individuate physical spaces by reference to both intrinsic and material geodesics, it
no longer makes sense to suppose it is possible for the two to diverge, if in actual
fact they coincide.
16 Zeno and the
continuum I
16.1 Motion and the continuum
Motion can easily seem the most unproblematic of phenomena. That things do
move seems beyond question: as I watch a car moving along the street, or my hand
waving in front of my face, I am literally seeing objects in motion, or so it seems
natural to say. Even if, as some would hold (see Chapter 7) it is not strictly speaking
correct to say that motion is directly perceived, there is no denying that motion is
a ubiquitous feature of everyday life. From a more theoretical standpoint, at least
in the context of the simpler and more familiar models of space and time (e.g. flat,
Euclidean, Newtonian), motion is equally unproblematic: it is simply a matter of
an object varying its location in space over time. In more exotic spacetimes there
are additional complications, but much remains unchanged. A spacetime is gener-
ally viewed as a structured collection of spacetime points, and whether or not one
is moving in a spacetime S will be determined by the trajectory ones worldline
takes in S, that is, by the points ones worldline passes through. In spacetimes
where motion is relative rather than absolute there is no fact of the matter as to
who is really moving, but motion itself is easily discerned: two objects are steadily
moving towards one another over an interval t
1
t
2
if the spatial distance between
them is steadily decreasing between t
1
t
2
; they are moving apart from one another
if the distance is steadily increasing; and they are at rest relative to one another if
the distance is unchanging.
However, for present purposes, the additional complications that some of the
more complex spacetimes bring can be set to one side. The motion-related prob-
lems that are the main topic of this chapter and the next do extend to the space-
times of the less familiar varieties, but they also apply with full vigour to the more
familiar forms for example Euclidean, Newtonian and they can safely, and
more easily, be explored in the context of the latter. We can also set aside, at least
temporarily, the debate between substantivalists and relationists; the problems in
question apply to both ways of conceiving space, time and spacetime. (However,
we shall see in 17.5 that these differing conceptions are not entirely without
relevance.)
As for the problems themselves, they derive from the eponymous paradoxes
of Zeno of Elea. Of Zeno himself, comparatively little is known. If what Plato
says in his Parmenides is true that Zeno was forty when Socrates was a young
man then Zeno was born around 490 BCE. We also know that, as well as being
the author of one or more books of paradoxes, Zeno was close to Parmenides.
268 TIME AND SPACE
Despite appearances to the contrary, Parmenides believed that reality consisted of
just one changeless entity. Zenos paradoxes were intended to bolster this static
monistic metaphysic by demonstrating that alternative conceptions conceptions
that posit a plurality of moving and changing objects collapse into absurdity.
Unfortunately, none of Zenos own writings have survived, and what little we
know of his doctrines comes from Aristotles brief discussions, and some passages
in Simplicius, who mentions a number of Zenonian paradoxes in his (fifth century
CE) commentaries on Aristotle. Of some forty paradoxes directed against plural-
ism, only a small handful have survived; in addition to these we have only the four
paradoxes of motion discussed by Aristotle.
Some of the latter are very well known indeed, and have been discussed in
many a school playground. The Achilles may well be the most widely known.
In this ill-matched race, the fleet-footed warrior Achilles sets off in pursuit of a
slow-moving tortoise, who we are told has been given a head start of a few
metres. Given his speed, it will not take long for Achilles to overtake the tortoise,
or so one might think. Zeno brings out a difficulty with this natural assumption.
Let us suppose that Achilles has yet to overtake, and the tortoise has just reached
point p
1
. It does not take long for Achilles to cover the intervening distance, and
shortly he too has reached p
1
. But, of course, by now the tortoise who is moving
continuously at a steady if slow speed has moved on, to p
2
; once again it does
not take long before Achilles has reached this point, but by now the tortoise has
moved on, to point p
3
. And so it goes on. By the time Achilles has arrived at p
3
the
tortoise has advanced to p
4
, and by the time Achilles has arrived at p
4
the tortoise
has moved on to p
5
, and so on and on and on, for ever smaller intervals. Hence it
seems Achilles will always lag behind in this manner, and so never succeed in catch-
ing up with the tortoise.
1
This seems absurd: dont we know that faster moving
things can catch up with and overtake slower moving things? Well yes, but where
has Zenos reasoning gone astray?
Writing of the Achilles, the American philosopher C. S. Peirce wrote: this
ridiculous little catch presents no difficulty at all to a mind adequately trained in
mathematics and in logic, but is one of those which is very apt to excite minds of
a certain class to an obstinate determination to believe a given proposition (1935:
vol. IV, 122). Bertrand Russell was more generous, claiming that Zenos arguments
are not mere foolish quibbles: they are serious arguments, raising difficulties
which it has taken two thousand years to answer, and which even now are fatal to
the teachings of most philosophers ([1929] 2001: 47). This divergence of opinion
is frequently to be found in the literature. Some are of the opinion that one (or
more) of the paradoxes involve trivial mistakes; others like Russell think they
raise deep issues, which require the resources of modern mathematics to resolve.
Alfred North Whitehead wrote: To be refuted in every century after you have
written is the acme of triumph No one ever touched Zeno without refuting
him, and every century thinks it worth while to refute him (1947: 114). Sorenson
responds: I think this compliment will not be paid by future centuries all of
Zenos paradoxes were solved by Cantor a hundred years ago. Cantors triumph
shows that some important paradoxes can be solved. We now have answers to
Zenos riddles that satisfy the exacting standard set by modern mathematics
(2003: 567).
ZENO AND THE CONTINUUM I 269
The idea that Georg Cantors work on infinity in the final decades of the nine-
teenth century along with the work of Augustin-Louis Cauchy, Richard Dede-
kind and others on the continuum allows a satisfying and definitive solution to
Zenos more serious and significant paradoxes is widely held, and not without
good reason. But as we shall see, these more recent developments are not entirely
puzzle- or problem-free.
16.2 Numbering the continuum
We shall be looking in more detail at some pertinent features of Cantors concep-
tion of the infinite in due course (see 16.5) but before proceeding further it will
be useful to take a step back. In arguing against the possibility of motion, one
of Zenos main targets is the doctrine that space and time are continuous. This
conception of the small-scale structure of space and time is the standard one today
it is built into the mathematical framework within which our leading physical
theories are formulated hence the continued interest in, and relevance of, Zenos
arguments. The other main view concerning the microstructure of the continuum
is that space and time are discrete, that is, composed of finite atomic units, or
quanta. Zeno also attacked this conception, or so it is generally thought, and we
shall be looking at what he had to say in due course. But let us start by taking a
closer look at the mainstream view.
What exactly does it mean to say that space, time or any other continuum is
continuous? If someone says The engine has been running continuously for three
hours now, they mean it has been running without any interruptions or gaps.
So much is common sense and ordinary language. In basic Euclidean geometry a
simple example of a continuous entity is a straight line. Such a line is obviously
gap-free; otherwise it would be two lines, rather than one. But it also has the prop-
erty of being infinitely or indefinitely divisible. Although the latter feature is not
obviously part of our common-sense understanding of continuous, it is not obvi-
ously not part of it. If you think of a continuous expanse of colour for example,
what you see when looking up at a clear blue sky it does not have an obvious
granular structure, and there is no obvious limit to how finely it could be divided.
Many of the changes or movements that we experience for example, a car we
observe disappearing down the road seem perfectly smooth and continuous.
In any event, that a geometrical line is infinitely divisible has long been assumed.
Such a line is usually regarded as being composed of an ordered collection of
points. Euclid defined a point as that which lacks parts, which mathematicians
subsequently took to mean that points are of zero size. Euclid also defined a line
as that which has no breadth. If a line is composed of extensionless points, there
is no limit to how far it can be divided, and no limit on the number of points it
contains. In more technical terms, a geometrical line has the property of being
dense, which means: between any two points there is always another point.
2

Denseness by itself might seem to ensure that a line is free from gaps, and
hence continuous. How could there be gaps in a continuum that is dense? That
something more than density might be required becomes evident by considering
arithmetic number lines.
270 TIME AND SPACE
Consider first the positive and negative integers:
, n, , 3, 2, 1, 0, 1, 2, 3, 4, , n, n + 1,
This number line is infinite in both directions: however big a positive integer n
one might encounter it always has a successor, namely n + 1, and the same holds in
the case of the negative integers (although now the successor is n 1). Although
there is an infinite number of integers, from a numerical perspective they are not
fully continuous: the integer line undeniably contains all the integers there are, but
it does not contain all the numbers there are. The fractions
1
2,
3
4,
33
4 and
5
7, are all
perfectly respectable numbers, but they are not to be found on our line of integers.
So let us move on to consider the rational number line. The latter includes all
the fractions, positive and negative, and since the integers can also be expressed as
fractions (1 =
1
1, 2 =
4
2) it automatically includes all the integers too. Like the inte-
gers, the rationals are infinite in number. But the rationals also possess a property
the integers lack: denseness. Between any two rational fractions, no matter how
close in size, there is always another to be found.
3
Does the fact that the rationals
are dense mean they are also continuous, in the sense of having no gaps? It does
not, for as the Pythagoreans famously discovered to their horror, there are irra-
tional numbers: numbers that (provably) cannot be expressed in the form of a
rational fraction. One such is the length of the diagonal of a square with sides one
unit long: by Pythagoras theorem this length must be equal to the square root of
1
2

+ 1
2
, or \2, which cannot be (accurately) expressed as a fraction. The irrationals
can be represented by infinite non-repeating decimals: for example 1.4142135
(= \2); 3.14159 (= t); 27.849279; 10001.89383948383; and so forth.
Adding the irrationals to the rationals gives us the so-called real number line,
symbolized as , and often called simply the real line.
+
11.34332 0 1 t 101.3456 \2
The real line is both dense and gap-free, and so from a numerical perspective
it can be regarded as being fully continuous. Not surprisingly, given this, it is often
simply called the continuum. The structure of the real-line continuum, along
with precise and adequate definitions of real number, were only fully worked
out by Dedekind, Cantor and others in the nineteenth century.
Since the real line is fully continuous, it is generally taken as being our reliable
and informative guide to the structure of the geometrical line.
4
More specifically, it
is assumed that the points on any interval of the geometrical line can be put into a
one-to-one order-preserving correspondence with a portion of the real line. More
importantly for our purposes, the real line is also commonly taken to be a guide to
the structure of the physical continua of space and time. (This is one of the reasons
why the term continuum often does double duty, being used to refer both to
mathematical and physical spaces.) Of course, whereas time is one-dimensional,
space has three dimensions, and spacetime four. But can easily be extended to
Figure 16.1 The real number line, .
ZENO AND THE CONTINUUM I 271
x
1
x
2
x
3
(x
3,
y
3
)
(x
2
, y
1
)
(x
1
, y
2
)
y
3
y
2
y
1
Figure 16.2 Extending the
reals to two dimensions.
three, four or more dimensions. In a two-dimensional real space the location
of each point requires two real numbers (r
1
, r
2
); in the three-dimensional case we
need three real numbers for each point, (r
1
, r
2
, r
3
); in the case of four dimensions
we need four numbers (r
1
, r
2
, r
3
, r
4
), and so on. The procedure is illustrated for
a simple case of a small portion of a two-dimensional Euclidean plane in Figure
16.2.
If our own (physical) space can be coordinated in this way with triplets of
real numbers, then our space must have the same small-scale structure as the real
numbers, and so be dense and continuous. Since it is generally assumed that this
is the case, let us call this the standard or orthodox view.
There is a further point to note. Since the days of Newton and Leibniz and
their independent invention of the calculus, the principal equations of physics and
cosmology have been real-valued functions: functions that take as their inputs
real numbers, and which deliver real numbers as their outputs (or answers). The
distinctive and novel feature of the calculus is its ability to determine how much
some quantities are varying with respect to others at a precise point in space and
time. Hence, as Roger Penrose observes, the orthodox view of the small-scale
structure of space and time has played a key role in our most successful physical
theories, such as those of Archimedes (who anticipated some of the principles that
underlie the calculus), Newton, Maxwell, Einstein and Schrdinger:
All successful dynamical theories have required notions of the calculus for
their formulation. Now the conventional approach to calculus requires the
infinitesimal nature of the reals to be what it is. That is to say, on the small
end of the scale, it is the entire range of real numbers that is in principle
being made use of. The ideas of calculus underlie other physical notions,
such as velocity, momentum, and energy. Consequently, the real-number
system enters our successful physical theories in a fundamental way for our
descriptions of these quantities also. (2004: 61)
Hence the importance and contemporary relevance of Zenos arguments.
According to the orthodox view, space and time have the same fine-grained struc-
ture as the real numbers, and so are infinitely divisible. According to Zeno, anyone
who holds that space and time are infinitely divisible cannot avoid disastrous
paradoxes. Let us move on and take a closer look at these.
272 TIME AND SPACE
16.3 The Dichotomy
A runner is on the final lap of the race, heading for the finish line. Can this run-
ner ever cross the line? In the paradox known as the Dichotomy (or sometimes
the Racetrack) Zeno suggests not. Lets suppose the runner is at point Z and
the finishing line is at Z*. To reach the finishing line, the runner must reach the
point (we can call) Z
1
, which is halfway between Z and Z*. Next he must reach
Z
2
, which is halfway between Z
1
and Z*; next he must reach Z
3
, which is halfway
between Z
2
and Z*, and so on (see Figure 16.3). Since the sequence of intervening
Z-locations has no end, the runner has to complete an endless sequence of Z-runs
(as we can call them), and so will never reach the finishing line. Since the same
considerations apply to any moving object, no matter how short its journey, it is
impossible for any object to reach any destination: motion is impossible.
Z Z*
Z
1
Z
2
Z
3
Z
4
1/2 1/4 1/8 1/16
Figure 16.3
Aristotle succinctly encapsulated Zenos reasoning thus: that which is in loco-
motion must first arrive at the halfway stage before it arrives at its goal. The
rationale for the paradoxs name is thus made clear: it involves the progressive
dividing into two of distances.
There is an obvious kinship between the Dichotomy and the Achilles both
involve an object moving successively through an infinite sequence of intervals
but what precisely is the relationship between them? Commentators tend to agree
that the Dichotomy is the more fundamental, and it is not difficult to see why. If
we know the speeds at which Achilles and the tortoise are moving, provided we
also know when and where they started off from, it is perfectly possible given a
little basic physics to work out precisely where Achilles will draw level with the
tortoise. Call this point p. If it is possible for Achilles to reach point p, then Zeno
is undone: Achilles can catch up with and subsequently overtake the tortoise.
However, if the reasoning behind the Dichotomy is sound, it is impossible for
Achilles ever to reach p, and so the conclusion of the Achilles is safe.
As for what we should make of the Dichotomy itself, the answer depends
entirely on what precisely we take the problem to be. In moving from Z to Z*
our runner has to complete the infinite series of Z-runs in order to pass through
the infinite series of intervening Z-locations, and this is deemed to be impossible.
But why, exactly, is this impossible? Aristotle took Zeno to be claiming that it is
impossible to travel through an infinite number of intervals in a finite time. If this
is right, Zeno may well have had an argument along these lines in mind:
1. In moving from Z to Z* the runner has to passes through an infinite series of
finite spatial intervals (e.g. m, m )
2. To move through an infinite number of finite spatial distances requires an
infinite amount of time.
ZENO AND THE CONTINUUM I 273
3. Hence our runner will require an infinite amount of time to move from Z to
Z*, and so will never arrive.
The same key assumption an infinite number of finite quantities adds up to an
infinite quantity generates an analogous absurdity if we leave time out of the
picture entirely, and focus solely on space:
1. In moving from Z to Z* the runner traverses an infinite series of finite spatial
intervals (e.g. m, m )
2. An infinite number of finite lengths added together yields an infinite length.
3. Hence the spatial distance between Z and Z* is infinite.
But of course, the distance between Z and Z* is not infinite. Both arguments evi-
dently generalize to all movements through space, and so if sound would establish
that motion (as we normally conceive it) cannot occur.
With regard to the first of these arguments, Aristotle objected that Zeno had
overlooked the fact that both space and time are infinitely divisible. He writes:
Zenos argument makes a false assumption in asserting that it is impossible
for a thing to pass over or severally come in contact with infinite things in
a finite time. For there are two senses in which length and time and gener-
ally anything continuous are called infinite: they are so called either in
respect of divisibility or in respect of their extremities. So while a thing in
a finite time cannot come into contact with things quantitatively infinite, it
can come in contact with things infinite in respect of divisibility: for in this
sense the time itself is also infinite: and so we find that the time occupied
by the passage over the infinite is not a finite but an infinite time, and the
contact with the infinites is made by means of moments not finite but infinite
in number. (Physics Z9, quoted in Kirk & Raven 1969: 293)
As usually construed, Aristotle is here arguing that the infinite divisibility of
time comes into play in the following way: if our runner let us suppose it is
Zeno himself is travelling at a constant speed of one metre per second, then he
will complete his first half-metre in half a second, the next quarter of a metre in
a quarter of a second, the next eighth of a metre in an eighth of a second, and so
on. Since the runner takes progressively less time to complete each stage of the
journey, it is possible to complete the entire trip in a finite time, in this particular
case precisely one second. In assessing Aristotles reasoning it is useful to distin-
guish the following claims:
(A) The sum of an infinite number of finite lengths (spatial or temporal) of con-
stant size is always infinite.
(B) The sum of an infinite number of finite lengths (spatial or temporal) of ever
increasing size is always infinite.
(C) The sum of an infinite number of finite lengths (spatial or temporal) of ever
decreasing size can be finite.
5
Claim A appears to be true. It is obvious that the sum of 1 + 1 + 1 + 1 +

is
infinite, at least given the assumption that means and so on to infinity,
which is precisely how it is intended to be taken. It is almost equally obvious that
274 TIME AND SPACE
sum of the fractions + + +

also equals infinity. After all, the latter series

can be rewritten as ( + ) + ( + ) +

, which is equivalent to 1 + 1 + 1 +

.
As for claim B, given that an infinite series of terms of constant finite size sums to
infinity, an infinite series of terms that are always increasing in size will surely do
so too. So far as the Dichotomy is concerned, claim C is the one that matters. At
first glance this certainly has some plausibility. If we imagine or try to imagine a
case in which the relevant quantities very rapidly diminish, it is easy to convince
oneself that the overall sum will be finite rather than infinite. To make matters
more concrete (see Benacerraf 1962), we can suppose that as Zeno proceeds with
his run he diminishes in size by a factor of one half each time he passes through
a Z-point (i.e. each time he halves the remaining distance to the finishing line).
Since it is not at all clear that there will be anything left of Zeno by the time he
completes his run, it is easy to believe that in this case at least, infinitely many finite
quantities in this instance, rapidly shrinking phases of Zeno can fit into a finite
extent of space. If this is possible for spatially extended things, why should it be
otherwise in the temporal case?
So the Aristotelian reasoning has some plausibility, but it is still open to ques-
tion: when it comes to estimating infinite sums, can our intuitions be trusted?
Can we really be sure that an infinite series of finite quantities can have a finite
sum? This is a mathematical question, and mathematics in Aristotles day had not
advanced to the stage where it was possible to answer it in the affirmative. Thanks
to subsequent developments particularly work by Fourier, Bolzano, Cauchy,
Dirichlet, Riemann and others in the nineteenth century contemporary math-
ematicians are in a far better position: they are able to assert with full confidence
that there are indeed infinite sequences of finite terms whose sum is also finite.
In contemporary parlance, and simplifying somewhat, each of our Z-points
[] or [ + ] is a partial sum. The partial sums corresponding to Zenos run
incorporate our now-familiar sequence:
S
1
= [], [ + ], [( + ) +
1
8],
As the series proceeds, it is clear that the individual partial sums are becoming
progressively larger [] is smaller than [ + ] = , and so forth but also,
and importantly, these increases in size are becoming progressively smaller with
each step taken. As a result, the partial sums never become indefinitely large: in
this case, it can be demonstrated that there is no partial sum in the series that is
larger than 1, even though as the sequence progresses the partial sums get ever
closer to 1. A sequence of partial sums with this property is said to be convergent.
Needless to say, not all sequences have this property:
S
2
= [1], [1 + 1], [(1 + 1) + 1],
S
3
= [1], [1 + 2], [(1+2) + 3],
In cases such as these the partial sums get bigger and bigger, without end. Such
sequences are said to be divergent.
Focusing on the relevant properties of convergent series, if an unending or infi-
nite sequence of partial sums approaches arbitrarily close to a particular number
L as the sequence proceeds, but never exceeds this number, then L is said to be
ZENO AND THE CONTINUUM I 275
the limit of that sequence. Our Dichotomy sequence, S
1
, can be rigorously dem-
onstrated to have 1 as its limit, whereas S
2
and S
3
do not have finite limits at all.
The next key step is to hold that the sum of an infinite series of numbers s
1
, s
2
, s
3
,
is equal to the limit of the sequence of partial sums [s
1
], [s
1
+ s
2
], [(s
1
+ s
2
) + s
3
],
[(s
1
+ s
2
+ s
3
) + s
4
], , if such a limit exists. This provides us with what we need in
the case of the Dichotomy. We can now safely say that the quantities that form the
Z-series + +
1
8 +

sum to a finite number, 1, rather than an infinite number.

And of course the same applies in the case of the many other infinite series that
can be proved to converge on finite limits.
It is important to note that in allowing ourselves to talk in terms of sums for
infinite sequences and series, we are moving beyond our ordinary notion of add-
ing up a series of numbers, for the latter is limited to finite series of numbers. As
Huggett notes (1999: 43), it would be a mistake to suppose mathematicians have
demonstrated that if one were to sit down for an infinite period of time adding
to to
1
8, and so on, one would discover that 1 is the answer one would reach
after adding the final number in the series. There is no last member of the series,
for no matter how far one advances, there is always another number to come. The
arithmetic of finite and infinite sums are thus quite different. In defining the sum
of an infinite series in terms of its limit, mathematicians are really introducing a
new stipulation as to how the term sum can be used in a new context: that of
infinite rather than finite series of numbers. Nonetheless, it would be also be
a mistake to think that we have made no advance on Aristotle. Whereas the latter
simply claimed that an infinite series of quantities could have a finite sum, we now
have a detailed mathematical theory that shows us in a systematic and consistent
way how it is possible for infinite sums to have finite totals.
Is that all that needs to be said about the Dichotomy (and by extension, the
Achilles)? By no means. We have dealt with one of the conundrums to which
Zenos puzzle gives rise, but arguably it is neither the deepest nor the most baffling.
We now know that it is possible for an infinite series of quantities whether spatial
or temporal to have a finite sum. There is thus no difficulty in supposing that
Zeno takes no more than a single second to traverse all the infinity of Z-locations
and arrive at Z*. The difficulty that does remain is understanding how he can
arrive at Z* at all: to arrive at Z* he must pass through each of the Z-points,
but there is an infinite number of these: Z
1
, Z
2
, Z
3
, , Z
n
. Since there is no last
member of the Z-series no final Z-run it seems, that Zeno can never reach his
destination after all. Yet he does.
This is puzzling indeed. But before addressing this difficulty it will be useful to
take a look at another of Zenos paradoxes.
16.4 The paradox of plurality
Although only fragments of Zenos forty paradoxes directed against plurality have
survived, a line of argument that can be extracted from some passages in Simplicius
is well worth considering; indeed, some view the Plurality as the deepest of all
the Zenonian paradoxes. (It also goes by the name of the Metrical Paradox, for
reasons that will become clear.)
276 TIME AND SPACE
Zeno takes as his target an extended thing. If such a thing can exist, he argues,
it must be divisible into parts. What is more, this division into parts can be
repeated indefinitely; for example, an object that is one metre long has two parts
of m, three parts of
1
3 m, four parts of m, and so on, to infinity. It may not
be physically possible to carry out this division, but this does not matter. For
suppose it turns out that ordinary material objects are composed of elementary
particles; if these particles are spatially extended to the slightest degree they still
possess parts for example their upper and lower halves even if these parts
cannot be separated. We are thus led to the conclusion that any extended object
is composed of an infinite number of parts. So far as the size of these parts is con-
cerned, there are only two options: either they have zero size, or they have some
small but finite non-zero size. If the latter, then our object will be infinitely large,
since an infinite number of non-zero quantities seems itself to be infinite in size.
If, on the other hand, the parts are of zero size, our object will itself have zero
size, since 0 + 0+ 0+ 0+

= 0. It seems that our ordinary notion of an extended
object collapses into absurdity.
Although this argument is not directed at motion per se, it easily extends to that
which motion supposedly takes place within: the continua of space and time. Let
us focus on the spatial case:
1. Any segment of the continuum is composed of an infinite number of homo-
geneous point-like parts. Consider a particular segment S.
2. The point-parts of S are either (a) of zero length, or (b) of some non-zero
length.
3. If they are of non-zero length then the total length of the segment S will be
infinite.
4. If they are of zero length, then the total length of the segment S will be zero.
As already noted, on the standard contemporary conception, space and time do
consist of infinitely divisible continua. Given this, Zenos argument is no straw-
man.
Nor is this argument quickly and easily refuted. A natural first thought might
be Ah, but as we saw in the context of the Dichotomy, it is possible for an infi-
nite sum of finite quantities to have a finite sum, therefore Zenos premise 3
in the above is undermined! But this is to overlook the distinction see AC
in 16.3 between infinite series whose terms are of constant size (or increas-
ing), and those whose terms are progressively decreasing. Whereas the latter can
indeed converge on a finite limit, and thus have a finite sum, in the case we are
currently considering the terms of the series are of a constant finite size thanks
to the homogeneity of the continuum. Accordingly, they will indeed sum to infin-
ity, just as Zeno claims.
So we are left with the problem of explaining how an infinite number of noth-
ings can add up to more than nothing. As with the Dichotomy, proponents of the
standard conception of the continuum claim that modern mathematics has the
resources to meet Zenos challenge. What follows is an attempt to present at least
some of this response to Zeno in as non-technical manner as possible. In the next
section some of the relevant mathematics will be expounded, in an elementary
way; in 16.6 we will see how this can be applied to the Plurality paradox.
ZENO AND THE CONTINUUM I 277
16.5 Cantors continuum
As standardly conceived, a spatial or temporal continuum is an ordered collection
of points that can be put into one-to-one correspondence with the real number
system. The real numbers themselves are infinitely many, and a spatial (or temporal,
or spatiotemporal) continuum is taken to consist of an actually existing infinity of
points. Until comparatively recently many philosophers but also scientists and
mathematicians were suspicious of the idea that an actual infinity of anything
could exist. In part this was due to Aristotles well-known and influential rejection
of the actual infinite in favour of the potential infinite. Continuous entities, Aristotle
believed, could be divided into ever finer sections whether in actuality, or merely
in thought but these smaller parts only come into being when their wholes are
divided; on this long-influential view, wholes are not composed of pre-existing
point-like parts. The Greek word closest in meaning to infinity is apeiron, which
means without bounds, but also out of control; in general, its connotations are
closer to the English word chaos. Zenos paradoxes are but one illustration of the
difficulties that dealing with infinity can bring. Take the seemingly innocuous series
1, 1, 1, 1, 1, 1,
Given that each term cancels out its predecessor to leave 0, there may seem no
doubt whatsoever that this series sums to zero. This conclusion is vindicated if we
make explicit the sum we are dealing with in this manner:
(1 1) + (1 1) + (1 1) +

This series clearly does reduce to 0 + 0 + 0

. However, there is a second way
of formulating the sum:
1 + (1 + 1) + (1 + 1) + (1 + 1) +

This is equivalent to 1 + 0 + 0 +

. Hence our original series seems (impossibly)
to illustrate that 1 = 0. A different, and ultimately more substantive problem, was
discussed by Galileo in the 1600s. In his Dialogues and Mathematical Demonstra-
tions Concerning Two New Sciences Pertaining to Mechanics and Local Motions
he observed that each of the natural numbers (the upper of the two series below,
G1) can be paired off in a one-to-one fashion with its square (as shown in the
lower series, G2).
G1 1 2 3 4 5 91 92 n

G2 1 4 9 16 25 8281 8464 n
2
In a quite general fashion, if we have two collections of objects C
1
and C
2
whose
members can be paired off in a one-to-one way, without remainder, we can con-
clude that these two collections are equinumerous, that is, they have precisely the
same number of members. Galileos two series, G1 and G2, are both infinite and
278 TIME AND SPACE
so we cannot count them off, one by one but it is perfectly clear and obvious that
their members can be paired off with one another, in a one-to-one way, without
leaving any remainder. This suggests that G1 and G2 must have precisely the same
number of members. But while in one respect this is unproblematic each of the
series is infinite, after all when looked at in another way it is decidedly puzzling.
G2 may be infinite, but it looks as though it also contains fewer numbers than G1
does. After all, G1 contains all the numbers between the squares as well as the
squares themselves, whereas G2 contains just the squares; to put it another way,
there are many numbers that are not in G2 that are to be found in G1. This is
puzzling, to put it mildly: how can the two series both be the same size and not be
the same size? How can a proper part (in this case G2) have the same number of
members as the whole of which it is a part (in this case G1)? Euclids fifth axiom
states that the whole is always greater than the part. Is this seemingly self-evident
claim one we need to reject?
Galileos response to this paradox was to suggest that the properties of being
greater than, less than or equal to apply only to finite quantities, not to infinite
ones: or, in other words, that different forms of arithmetic apply to infinite and
non-infinite numbers. But Galileo himself did not develop this idea any further;
this was left to Cantor, in the nineteenth century.
When demarcating finite from infinite collections or sets Cantor took Galileos
paradox as his starting-point. In more formal terms, he held that two sets S
1
and
S
2
have the same cardinal number (intuitively, they are the same size) if and only
if there exists a one-to-one correspondence between the members of S
1
and the
members of S
2
. Cantor then argued that what marks out an infinite set S as infi-
nite is the fact that a one-to-one correspondence exists between the members of
S and one of Ss proper subsets (a proper subset of a set S contains some of the
members of S but not all of them). A set is finite if and only if it is not possible to
establish that a one-to-one correspondence exists between all of its members and
a (proper) subset of its members. David Foster Wallace calls this Cantors first
stroke of incredible, nape-tingling genius, although it may not look like much at
first (2005: 248). It was certainly a bold move, as well as an ingenious one. The
property that had previously been seen as paradoxical as throwing the whole
notion of the infinite into serious question is simply taken by Cantor to be the
defining feature of infinite sets.
Cantor used the one-to-one correspondence method of determining when two
sets have the same size (or cardinality) to establish a remarkable result. It is natural
to think that all infinite quantities must be of the same size. Suppose you and I are
both fortunate enough to have bags containing an infinite number of chocolates; if
you were to claim that your bag contains more chocolates than mine, I might eas-
ily be tempted to say, Thats just silly. Since each of our bags contains an infinite
number of chocolates, they each contain as many chocolates as it is possible to
possess! But Cantor succeeded in showing that not all infinite collections are of
the same size: some are bigger than others.
To introduce some further terminology, let us say, following Cantor, that an
infinite series is denumerable (or countable) if it is possible to set up a one-to-one
correspondence between its members and the positive integers. The series of frac-
tions familiar from the Dichotomy is denumerable, as shown below:
ZENO AND THE CONTINUUM I 279
1 2 3 4 5


1
8
1
16
1
32
What of the rational numbers, the set of all numbers that can be expressed
in terms of ratios of integers? This set includes all the fractions found in the
Dichotomy series, along with many more, for example
1
1000000,
373
12,
1777
8932542, and
so on, without end. There is no smallest rational number there is no limit to the
size we can divide 1 by, to make ever smaller numbers and as we saw in 16.2,
the rationals are dense: between any two, no matter how close together, there is
always a further rational to be found. With these properties, it would be natural
to think that the set of rationals must be larger than the set of positive integers.
But no. Cantor proved that it is possible to set up a one-to-one correspondence
between the rationals and the integers: the rationals are thus denumerable, they
have the cardinality he designated
0
(or aleph-0). What of the set of real
numbers, the set that includes all numbers expressible by infinite decimals? In a
further remarkable proof employing his method of diagonalization Cantor
established that it is not possible to set up a one-to-one correspondence between
the integers and the reals. Hence is non-denumerable or uncountable for it
has a greater cardinality than that of the integers: Cantor called this new cardinal
number c. (For more on this proof see the Appendix to this chapter.) In further
proofs he established that the infinite cardinals go on getting bigger without end:

1
,
2
,
3
, .
6
For our limited purposes we neednt pursue these higher cardinals further:
it is with the properties of c that we need to concern ourselves. How many real
numbers are there between 0 and 1? Since the reals, like the rationals, are dense,
the quick and easy answer is an infinite number. But now we know that infinite
numbers come in different sizes, this answer no longer suffices. Cantor proved
that the interval 01 contains just as many reals as the entire set . Figure 16.4
amounts to a quick and easy demonstration of why: once again we are dealing with
one-to-one correspondences.
0
0 1
1
0.742 0.31 1.31 2.101

+
p
1
p
2
x
Figure 16.4 A graphical demonstration of how all the points on the real line can be put into a
one-to-one correspondence with the points within the interval [0, 1]. It suffices to draw a line from
all the points on to point x; then draw a vertical line upwards from the point at which each of these lines
crosses the semicircle. Note how 0.31 gets mapped onto p
1
, and 1.31 onto p
2
.
280 TIME AND SPACE
There is nothing in the least special about the interval between 0 and 1: the
same reasoning applies to all finite intervals on the real line: every interval, no
matter how small, contains the same number of points c as every other line,
no matter how long, and indeed, the entire real line itself.
The idea that there are as many points in an interval that is only a billion bil-
lionth of a nanometre across as in an interval that is a trillion trillion metres long
can seem very odd indeed, to put it (very) mildly. But Cantor and his followers
will simply point out that finite and infinite collections do have very different
properties, and if the properties of infinite collections strike us as bizarre, it is
only because our intuitions have evolved in response to our everyday dealings,
which for most of us are exclusively with finite quantities and collections. In
his dimension proof Cantor took matters a stage further: he established that
there are enough points in a one-dimensional real line to cover, without any gaps,
the entirety of a two-dimensional continuum. The method, once again, involved
establishing a one-to-one correspondence: in this instance, between all the points
in an interval of the real line and the points on a two-dimensional plane. (Report-
ing this result to Dedekind, Cantor wrote: I see it, but I dont believe it.) This
same method can be extended to prove that the sets of points that compose objects
of three or more dimensions also have the same cardinality as the real line, c.
16.6 Plurality, measure and metric
For our purposes getting to grips with Zenos paradox of plurality it is these
last results that are of prime significance.
Zenos challenge, it will be recalled, is as follows: given that an interval of a
continuum is composed entirely of points that are uniformly of zero size, how
can finite parts of this interval have lengths other than zero? Underlying this chal-
lenge is an assumption: namely, that the length of an interval on a line (to take the
simplest case) is generated by adding up the sizes of its smallest constituents. In the
case of ordinary objects and their finitely sized parts, this assumption is correct.
We can measure the length of a racetrack whether for humans or horses, cars or
bicycles by counting up the number of metre-length parts it possesses. Similarly,
if we know that a pile is ten bricks high, and each brick has a height of 10 cm, it fol-
lows at once that the pile is 10 10 = 100 cm tall; we also know that if there were
double the number of bricks in the pile, it would be twice the height. In cases such
as these ordinary cases, of the sort we encounter in everyday life the lengths of
wholes is determined by, or is a function of, the magnitudes of their smaller parts.
Indeed, even in the cases of infinite series of the Dichotomy variety, containing
denumerably many members, the sum of the series is a function of the sizes of the
parts: the series sums to 1 precisely because of the sizes of its parts. Hence Zenos
claim that if spatial intervals are composed of parts with zero size, in the form of
dimensionless points, no spatial interval could ever have a size larger than zero.
This line of argument certainly poses a serious threat to the standard view of
the continuum, and of all Zenos (known) arguments, the Plurality may well have
the greatest intuitive power. This does not mean, however, that it is correct or
unanswerable. Zenos conclusion would be unavoidable if the points that compose
ZENO AND THE CONTINUUM I 281
the Cantorian continuum behaved in the same way as ordinary finite quantities,
or denumerably infinite quantities. But as we have just seen, this is not case: there
are the same number of points in a millimetre of the continuum as there are in a
segment that is a trillion miles long. (More precisely, the set of points in the first
interval has cardinality c, as does the set of points in the second interval.) Given
this, the length of a line is not a function of the number of points it contains. How
could it be, if spatial intervals (or line segments) of very different lengths contain
the same number of points?
So we see that our ordinary concept of addition the concept that allows
us to calculate the height of a wall by adding up the heights of the bricks that
jointly compose it does not apply to non-denumerable (or uncountable, or super-
denumerable) collections of points. Since (on the orthodox view) the spatial con-
tinuum does contain a non-denumerable infinity of points, it is a mistake to think
that our ordinary concept of addition applies to these points. But of course Zenos
plurality paradox depends on the assumption that it does: he invites us to add up
all the points in a line, in precisely the same way as we add up 1 + 1, 2 + 2, or 0 + 0,
or 0 + 0 + 0. But if the concept of addition is inapplicable to non-denumerable
collections of points, we must decline this invitation. Here is Adolf Grnbaum
summarizing the state of play:
[Zeno] could not anticipate that the addition of a super-denumerable infinity
of numbers, be they zero or positive, presents a problem altogether different
from adding either a finite sequence of numbers such as 3, 4, 7 or a denu-
merable infinity of numbers such as 1, , ,
1
8,
1
16,
1
32,
1
64, . Although
arithmetic has evolved a definition of the sum of a denumerable infinity
of numbers by a consistent generalization of the concept of finitary sum on
the basis of the limit concept, this definition is utterly useless and irrelevant
when the addition of a super-denumerable infinity of numbers is called for.
Thus, while length is denumerably additive in standard mathematics, the
addition of the lengths of all the points composing a unit segment is a mean-
ingless (undefined) operation in analytic geometry and hence cannot be used
to compute a value for the total length of the segment. In particular, the
deduction of Zenos paradoxical result zero for the length of a unit segment
is precluded by this arithmetical fact. ([1955] 2001: 168)
When dealing with the continuum, to avoid paradox or inconsistency we must
restrict addition in certain ways. The standard mathematical solution is to hold that
an interval I can be regarded as the arithmetic summation of its non- overlapping
sub-intervals, i
1
, i
2
, i
3
, , i
n
, only under certain conditions, for example the sub-
intervals are all of finite length and finite in number, or denumerably infinite in
number and converging to some well-defined limit. Provided we adhere to these
rules we dont run into difficulties; needless to say, these rules do not allow us to
generate lengths by summing individual (zero-sized) points.
Grnbaum goes on to point out that contemporary mathematics is equipped
with a more general way of measuring the sizes of sets containing non- denumerably
many members: the measure theory, initially developed by Henri Lebesgue and
others in the 1900s. Let us suppose that an interval I on the real line is the set of
points lying between two endpoints A and B. Does I include the two endpoints, that
282 TIME AND SPACE
is, points A and B themselves, or not? From the point of view of measure theory,
it matters not: the addition of a single point (or two points) to an interval makes
no difference to its measure. This seems right, intuitively, since it is difficult to see
how adding a dimensionless point to an interval can change its length. Ordinary
length is additive, so too are measures: if interval I divides into two non-overlapping
sub-intervals i
1
and i
2
, the measure of I will be identical to the sum of the measures
of i
1
and i
2
. Since the same applies to countably infinite sets of intervals i
1
, i
2
, i
3
, ,
i
n
, measure is said to have the property of being denumerably additive. Since
measure is additive, given that adding a single point to an interval does not increase
its measure, it is natural to conclude that the measure of a single point must be zero,
and this is what measure theory states: more precisely, the measure of the degener-
ate interval where A = B is taken to have measure zero. By virtue of the fact that
measure is denumerably additive, it follows that a countably infinite set of zero-sized
points (or degenerate intervals) also has measure zero. An interval has to contain
a non-denumerably infinite number of points in order to have a positive, non-zero
measure. As standardly applied, measure theory delivers the result that the real-line
interval [0, 1] has measure 1, and more generally, that the measure of an interval [a,
b] is the distance between the end-points of the interval, that is, the length b a.
Viewing matters from the measure-theoretic perspective, Zeno was both correct
and incorrect. An infinite number of dimensionless points when added together
does always yield zero or at least, they will always constitute an interval whose
measure is zero but this applies only if the infinity is denumerable. A non-
denumerably infinite collection of points does add up to something: the points can
constitute an interval whose measure is non-zero. To put it metaphorically, only
collections of points with c-many members possess any substance; to put it less
metaphorically, only collections of points with c-many members possess extension.
If this seems peculiar, then so too is the Cantorian continuum; recall the manner
in which every interval of the real line, no matter how small, has the same size (or
cardinality) as every other interval, no matter how large. Given this, it is not really
that surprising to learn that it takes c -many points to make any difference to the
size (or length, or measure) of anything which inhabits the continuum.
7
So, thanks to developments in mathematics, the proponent of the standard
conception of the continuum has an answer to Zeno. Accepting that a line L is an
aggregate of points that are each of zero length does not oblige one to accept that
L itself is of zero length. Relying as it does on considerations deriving from analysis
and set theory, someone might query whether Grnbaums approach applies to
non-mathematical entities, as we shall see in 17.5, but we shall not pursue this
now. Accepting the mathematics-based solution even if only tentatively gives rise
to a new question: if length is not an intrinsic feature of collections of points, what
does determine the length of an interval? According to the approach standardly
adopted in contemporary geometry and spacetime theories, a collection of points
has to be associated with a metric (or distance function) before it acquires metrical
properties. The basic idea is that we assign coordinates to our collection points
in the collection, in the form of clusters of real numbers; the numbers are chosen
in such a way that the arithmetic relationships between these numbers accurately
reflect the geometrical properties of the space with which we are dealing. In the
case of a simple space such as the Euclidean plane, our coordinates for individual
ZENO AND THE CONTINUUM I 283
points will take the form of pairs of real numbers (x, y), with the values chosen
so as to ensure a Euclidean structure. This can be achieved by ensuring that the
square of length of the hypotenuse of any right-angled triangle equals the sum of
the square of the other two sides, in accord with Pythagorass theorem. For more
complex spaces, more sophisticated treatments are required. In curved spaces,
real number coordinates are assigned to families of non-overlapping curves, and
distances are determined by a metric tensor usually symbolized by g following
work by Gauss, Riemann and others. For details see Sklar (1985: IIB), Newton-
Smith (1980: ch. 7), Moore (2007a) and Norton (1992).
This metricization procedure itself gives rise to new questions. Those who
incline to a strongly realist view will hold that a metric simply reflects real and
intrinsic features of spacetime itself. However, some argue that there is an ineradi-
cable conventional element to metrical spacetime features. Grnbaum himself has
argued that such features are not fully objective, at least in the case of spaces, times
or spacetimes that have the structure of a Cantorian continuum. This is because
such continua are metrically amorphous: they consist of non-denumerably many
points that entirely lack intrinsic metrical features. But this view too is disputed;
others take the view that the metrical features can be as fundamental and intrin-
sic as topological properties, at least in the case of real physical spacetimes. For
discussion and criticism of Grnbaums stance on this issue see Friedman (1972),
Horwich (1975a), Nerlich (1994b: ch. 9).
16.7 The Dichotomy revisited
In 16.3 we left Zeno in a precarious position vis--vis his attempt to reach the fin-
ishing line of his race. To complete the race he has to pass through all the infinity of
Z-points lying between Z and Z*. One argument that this is impossible goes thus:
Zeno will never reach the finishing line, because to do so he would have to
traverse the final metre of the track, and to do that he would have to travel
first m, then m, then
1
8 m, and so on to infinity. Since he is travelling
at only a finite speed, and he needs to traverse an infinite series of finite
distances, it will take him an infinitely long time to complete this journey.
However, as we saw earlier, it is perfectly possible for an infinite series of quanti-
ties whether spatial or temporal to have a finite sum, provided the quantities
in question converge on a limit. There is thus no difficulty in supposing that
Zeno takes no more than a single second to traverse all the infinity of Z-locations
and arrives at Z*. But as we also saw, a further difficulty remains, which can be
formulated thus:
The real problem is understanding how Zeno can arrive at Z* at all. To arrive
at Z* he must pass through each of the Z-points, but there is an infinite
number of these: Z
1
, Z
2
, Z
3
, , Z
n
. Since there is no last member of the
Z-series, it seems, that Zeno can never reach his destination after all.
The difficulty derives from the denseness of the continuum. Between any two
points, p
1
and p
2
, no matter how close together they might happen to be, there is
284 TIME AND SPACE
always a further point p
n
lying between p
1
and p
2
. Since this applies quite generally,
without restriction, it means that there is also a further point p
n+1
lying between p
n
and p
2
, and so on, without end. Consequently, as Zeno closes in on the finishing
line there is always another point lying between his current location and his final
destination: Z*. Given this, how can he ever arrive at Z*? The difficulty is the
more acute in the light of Cantors having established the coherency and consist-
ency at least in a mathematical sense of conceiving of the continuum as an
actual infinity of points. It can easily seem as though there is an insurmountable
gulf separating Zeno from his destination.
8
Some preliminary clarifications are in order. For Zeno to reach Z* there is a
sense in which he must complete an infinite series of tasks, for he must traverse
an infinite series of finite spatial intervals. The very idea of completing an infinite
series of tasks can seem absurd: such a series has no last member, so how can it
possibly be completed? But as Lowe (2002: 290) observes, much depends on how
we interpret the formulation in question. If we take it to mean completing the last
task in the series after completing all of its predecessors, then since such a series
does not have a last member, bringing it to completion is impossible. But another
interpretation is available completing all of the tasks in the series in the order
in which they occur in the series and, as Lowe notes, it is not at all obvious that
this is incoherent or impossible. As for the problem of how our runner (here A)
manages to reach the finishing line, Lowe has this to say:
[E]very one of the infinite series of tasks that A must complete in order to
arrive at Y involves his arrival at a point between X and Y which is distinct
from Y. It may seem, thus, that even having completed all those infinitely
many tasks, there is still something for A to do, namely, arrive at Y itself;
and this, it may seem, is his last task, which succeeds all his infinitely many
previous tasks. But, in fact, it would seem that the proper thing to say is that,
although there is no single task in the infinite series of tasks by completing
which A arrives at Y, none the less, by completing all of these infinitely many
tasks, A arrives at Y there is nothing further that A need do in order to
arrive there than to complete every task in the series. Thus, the whole task
of running from X to Y, the performance of which results in As arrival at Y,
is just the sum of the infinitely many subtasks in the series, even though no
one of those subtasks is such that, by performing it, A arrives at Y.
(Ibid.: 297)
While this is useful something along these lines probably has to be said in
response to the Dichotomy as Lowes final sentence makes clear, something
of a puzzle remains. If none of the individual Z-runs (the sub-tasks in question)
that jointly compose the whole task manages to take Zeno to the finishing line
at Z*, how is it that he arrives there at all? The puzzling character of this may be
ineliminable: motion through a continuous (in the mathematical sense) medium
has some radically unfamiliar features, some of which are highly counterintuitive.
9

Nonetheless, the extent to which such motion is puzzling can be reduced some-
what by taking a closer look at some features of the continuum itself.
First some further terminology. It is standard practice in set theory and topol-
ogy to distinguish between open intervals (a, b), and closed intervals [a, b], of the
ZENO AND THE CONTINUUM I 285
continuum (or real line). The closed interval between points a and b consists of
all the non-denumerably many points lying between them, together with a and
b themselves. The corresponding open interval also contains all the points lying
between a and b, but does not contain the points a and b themselves. To put it
another way, (a, b) and [a, b] are indistinguishable save for a difference of just two
points. Using open to characterize an interval such as (a, b) is apt, for, given the
denseness of the continuum, there is no last point in the interval: as we get closer
and closer (speaking metaphorically) to as location, there is always yet another
point that is closer to, but distinct from, a itself. Extending this terminology, we
can also speak of intervals that are half-open, such as (a, b] and [a, b); in the former
case point b is included but a is omitted, in the latter case a is included, b is not.
From this vantage point, Zeno appears to be guilty of a significant misrepresen-
tation. In effect, we are invited to consider the half-open series
[Z, , , ,
1
8,
1
16, )
and asked whether performing the sequence of tasks (Z-runs) that corresponds
with this series could ever allow Zeno to reach the finishing line by virtue of occu-
pying or passing through Z*. But of course, by virtue of the fact that the series is
half-open, point Z* is excluded, and so necessarily Zeno cannot occupy or pass
through Z*. A more accurate representation of the total journey undertaken by
Zeno would take the form of the following closed series:
[Z, , , ,
1
8,
1
16, Z*]
Since Z* is now included, if Zeno were to occupy successively all the points cor-
responding to the series of movements represented here, his travels would take
in Z*.
This sheds some useful light on our problem, but there remains the issue of how
Zeno can reach Z* given that there is no final member of the series of Z-runs that
takes him to Z*. A further set- (or measure-) theoretic property of open and closed
intervals is of some assistance here, as registered briefly at the end of the 16.6.
Although the closed interval [a, b] contains one point more than the corresponding
half-open interval [a, b) since the latter does not include point b, whereas the
former does when it comes to the lengths of these intervals, there is no difference
at all. Not only is the length of [a, b] exactly the same as that of the half-open
intervals [a, b) and (a, b], but it is also exactly the same as the length (or measure)
of the wholly open (a, b). Hence to reach point Z*, Zeno does not have to travel
a greater distance than he has already travelled in completing the Z-series. In this
metrical sense at least, completing the Z-series is sufficient to complete the race.
10
Appendix
For interested readers who have not encountered this material previously, here are
a couple of brief informal illustrations of some of Cantors easier proofs, drawn
from Wallace (2005: 24753). The first proof concerns the relationship between
the rationals (or fractions) and the positive integers. On the face of it, there are
strong grounds for supposing that the rationals will vastly outnumber the integers.
286 TIME AND SPACE
Not only are there many fractions between any two integers (e.g. the Dichotomy
series lying between 0 and 1, or 1 and 2), but also, by virtue of the fact that the
rationals are dense, but there is an infinite number of rationals between any two
rationals! However, by showing that the rationals and integers can be paired off,
in a one-to-one fashion and without remainders, Cantor established that the set of
the integers and the set of the rationals have the same size, or cardinality.
The first step is to find a suitable way of representing all the rationals; the grid
shown in Figure 16.5 does the trick. All fractions can be expressed in the form of
a ratio p/q, and since the grid below contains all values of p (moving horizontally
to the right) and all values of q (moving vertically down), and also each instance of
p/q, it contains or would contain if it were shown in full every rational number.
1 2 3 4 5 6 7
p
1

1
2
2
2
3
2
4
2
5
2
6
2
7
2
p
2

1
3
2
3
3
3
4
3
5
3
6
3
7
3
p
3

1
4
2
4
3
4
4
4
5
4
6
4
7
4
p
4

1
5
2
5
3
5
4
5
5
5
6
5
7
5
p
5

1
6
2
6
3
6
4
6
5
6
6
6
7
6
p
6

1
7
2
7
3
7
4
7
5
7
6
7
7
7
p
7


1
q
2
q
3
q
4
q
5
q
6
q
7
q
p
q

The next step is to show that the rationals can be paired-off with the integers in
the required way. Figure 16.6 demonstrates that it is possible to draw a line which
connects the integers (along the top row) with the rationals in a systematic man-
ner, which when extended as far as it will go, horizontally and vertically will
eventually take in all the integers and rationals.
1 2 3 4 5 6 7
p
1

1
2
2
2
3
2
4
2
5
2
6
2
7
2
p
2

1
3
2
3
3
3
4
3
5
3
6
3
7
3
p
3

1
4
2
4
3
4
4
4
5
4
6
4
7
4
p
4

1
5
2
5
3
5
4
5
5
5
6
5
7
5
p
5

1
6
2
6
3
6
4
6
5
6
6
6
7
6
p
6

1
7
2
7
3
7
4
7
5
7
6
7
7
7
p
7


1
q
2
q
3
q
4
q
5
q
6
q
7
q
p
q

Figure 16.5
Figure 16.6
ZENO AND THE CONTINUUM I 287
If we take the numbers picked out by this line and set them out in a long row,
preserving their order, we get the following:
1, 2, ,
1
3,
2
2, 3, 4,
3
2,
2
3, ,
1
5,
2
4,
3
3,
4
2, 5, 6,
5
2,
4
3, ,
2
5,
1
6,
1
7,
2
6,
3
5, ,
p
q,
This series contains some redundancy since, for example,
1
3 and
2
6 designate the
same rational. By eliminating all such redundancies we arrive at the more stream-
lined series:
1, 2, ,
1
3, 3, 4,
3
2,
2
3, ,
1
5, 5, 6,
5
2,
4
3, ,
2
5,
1
6,
1
7,
3
5, ,
p
q,
Since this series includes all the rationals, and can obviously be put into a one-
to-one correspondence with the series 1, 2, 3, 4, , we have established that
the rationals are countable, and therefore they have the same cardinality as the
integers, namely
0
.
Although the proof sketched above is sometimes referred to as Cantors diago-
nal proof , for obvious reasons, in fact this is incorrect. His real diagonal proof
is (one of) the procedures he used to demonstrate that the cardinality of the reals,
c, is greater than
0
. This proof works by reductio: we start by assuming that the
real numbers are in fact denumerable and so can be put in a long row, and paired
off in a one-to-one way with the series 1, 2, 3, 4, and then prove that this
assumption is untenable.
Once again, the first step is to find a way of representing all the reals. The most
natural choice is decimal notation. Irrational numbers (such as \2) can be put into
decimal form, but so too can the rational numbers. Some of the rationals have
non-terminating decimal form, for example,
1
3 = 0.333333; in other cases the
decimal form terminates, for example, = 0.5. So as to have a uniform mode
of representation we will opt to represent 0.5 (and all analogous cases) by the
non-terminating 0.499999999999 . (This is not obligatory; the non-terminating
0.50000000 is an equally serviceable alternative.) Now suppose that it is pos-
sible to list, or enumerate, the reals. Any such enumeration can be put into the
form shown in Figure 16.7. As can be seen, the infinite decimals on the right have
been paired off with the positive whole numbers on the left, which can be used to
count off the reals, one by one. In the representations on the right, X
1
, X
2
, X
3
,
correspond to the pre-decimal point parts of the reals, and the a
1
, a
2
, a
3
, a
4
, a
5
, a
6
,
a
7
, correspond to the infinite series of digits that occur after the decimal point.
1 X
1

a
1
a
2
a
3
a
4
a
5
a
6
a
7

2 X
2

b
1
b
2
b
3
b
4
b
5
b
6
b
7

3 X
3

c
1
c
2
c
3
c
4
c
5
c
6
c
7

4 X
4

d
1
d
2
d
3
d
4
d
5
d
6
d
7

5 X
5

e
1
e
2
e
3
e
4
e
5
e
6
e
7

6 X
6

f
1
f
2
f
3
f
4
f
5
f
6
f
7

Figure 16.7
288 TIME AND SPACE
Since this table is supposed to be able to list all the reals, to bring off the desired
reductio we need to prove that this is not the case: that is, show that there is a real
number that this list does not include. Let us designate the hypothetical missing
real number R. The latter will also have the form of a decimal X
n

n
1
n
2
n
3
n
4
n
4
n
5

.
We can let the pre-decimal part X
n
of R be any integer we like: it matters not. What
does matter is that the decimal part of R consists of an infinitely long series of digits
that is different from any that can be found in our table, and Cantor shows us how
to construct the decimal part of R so that it has precisely this feature.
The construction achieves this by ensuring: (i) that the digit in the first decimal
place of R (n
1
) differs from the digit in the first decimal place of the first real on
our list (a
1
); (ii) that the digit in Rs second decimal place (n
2
) differs from the digit
in the second decimal place of the second real on our list (a
2
); and similarly for all
the rest n
3
differs from c
3
, n
4
from d
4
, and so on. As for how this can be achieved,
we simply create a rule that does what is required: we say that the value of n
1
is
one less than the value of a
1
, unless it so happens that a
1
is 0, in which case we let
the value of n
1
be 9. So if a
1
= 4, then n
1
= 3; if a
1
= 9, then n
1
= 8; and if a
1
=
0, then n
1
= 9. We can use exactly the same rule to ensure that n
2
differs from b
2
,
n
3
differs from c
3
, n
4
differs from d
4
, and so on along the diagonal line traversing
our entire infinite table.
We now have what we need. If the reals are denumerable, they will all occur
somewhere in our table. But it is clear that R = X
n

n
1
n
2
n
3
n
4
n
4
n
5
is not included
in the table, since its decimal part (n
1
n
2
n
3
n
4
n
4
n
5
) is not to be found anywhere
in our table. So the assumption that the reals are denumerable turns out to be
false. Since the rationals and integers are denumerable, the cardinality of the reals
is greater than that of the rationals and integers.
17 Zeno and the
continuum II
17.1 The Arrow
Imagine an arrow in flight. At every instant during its flight, the arrow is at some
specific location, occupying a volume of space that corresponds exactly to the
arrows own size. Since instants are durationless, and motion takes time, the arrow
is not in motion at (or during) these instants. Yet the arrows entire flight is com-
posed of these instants, and nothing but these instants. If the arrow is not in motion
at any of the instants that jointly make up its flight, it is not in motion at all.
This is Zenos Arrow paradox.
1
The reasoning obviously generalizes what
holds for the arrow holds for any object in motion and so if the argument is
sound, motion does not and cannot occur. The argument is sometimes taken as
being directed along with Stadium, which we will be looking at shortly
against the doctrine that space and time are discrete, rather than continuous.
While this may well be true for the Stadium, it is not obviously true in the case of
the Arrow. In any event, the Arrow reasoning certainly poses a significant problem
for those who believe space and time are continuous rather than discrete, the
conception of the continuum with which we are currently concerned. If space
and time are composed of dimensionless points, and the arrows flight involves
its occupying a continuous succession of different locations, we are undeniably
confronted with the problem of explaining how it is possible for the arrow to have
moved, given that it is not (seemingly) moving during any of the momentary stages
which constitute its flight.
The at-at theory
Those who subscribe to the orthodox conception of the continuum have a response
to the Arrow challenge that is widely, although by no means universally, accepted
as adequate. The proposal runs as follows. Whether or not an object is in motion at
a given time t does not depend on the objects condition or state at time t. Rather,
it depends on how the object behaves at those temporal instants neighbouring t.
If at these neighbouring times the object is at precisely the same location as it is
at t, then we say that it is not moving at t; but if it is at different locations, then
we say that it is moving at t. Since everyone can agree that an object is moving
if it is changing its location over time, this simple account seems to give us all
we might want. Or as Russell puts it, Motion consists merely in the fact that
bodies are sometimes in one place and sometimes in another, and that they are
290 TIME AND SPACE
at intermediate places at intermediate times (1917: 84); hence the at-at label
for this view.
It will be useful to take a closer look at what proponents of the at-at account
are proposing. Suppose we start with an interval I
1
, running from half a second
before t to half a second after t (see Figure 17.1). We calculate the arrows average
velocity over I
1
by dividing the distance travelled over the course of this interval
by the time taken, which in this case is one second. This gives us an initial, reason-
ably accurate, approximation of the objects velocity at t. To get more accurate
approximation we repeat the procedure for a shorter interval I
2
, which is also
centred on t, but running (we can suppose) from a quarter of a second before it to a
quarter of a second after it. By working out the average velocities over shorter and
shorter t-centred intervals we will generate increasingly accurate estimates of our
objects velocity at t itself. In standard fashion we now introduce the limit-value
of the series at t: it is the average velocity to which the series of ever-decreasing
intervals converges as the end-points of the intervals approach arbitrarily close to
t itself. We can now stipulate that our objects instantaneous velocity at t is simply
this limit-value.
This procedure can be applied quite generally to deliver the instantaneous
velocities of any moving object at any time. Indeed, the procedure just outlined
is essentially that used in the calculus, where velocity is the first derivative of
change of position with respect to time (acceleration, the rate of change of velocity,
is the second derivative). Since the day of Newton and Leibniz, calculus has been
at the heart of physics and applied mathematics, and it is undeniable that much of
the success of the former is due to the latter; without the methods of the calculus
we would not be able to calculate the rates of change of interdependent quantities
in the precise and accurate ways our science and technology require.
17.2 Velocity as intrinsic
For all its advantages, the at-at conception of motion also has its critics. It can seem
very natural to think of movement as a purely intrinsic property of objects. If an
object is moving at a time t, the fact that it is moving is a fact about the objects
intrinsic condition at time t; accordingly, what is taking place at other times and
Time
t
1/2 +1/2 1/4 +1/4 1/8 +1/8
I
1
I
2
I
3
I
n
Figure 17.1 The arrows speed at t
is the limit value of its average speed
over the infinite series of t-centred
intervals I
1
, I
2
, I
3
, , I
n
, , as these
approach arbitrarily close to t.
ZENO AND THE CONTINUUM II 291
places is of no relevance. But if the at-at conception is correct, motion is not
intrinsic in this way. Whether or not an object is in motion at t depends solely and
entirely on whether the object changes its position at other times neighbouring t:
the objects condition or state at t has no bearing on the issue whatsoever. This
can easily seem strange, or even quite wrong.
Suppose a billion-ton hunk of rock call it R1 is moving swiftly towards the
surface of the Moon, where it will soon impact, causing a massively destructive
explosion. Let us further suppose that on the other side of the Moon there is a
rock R2 that is very similar to R1 in size, shape and mass. There is, however, a
difference: whereas R1 has spent the past few hours plummeting towards the
Moons surface, picking up more and more speed and momentum with each pass-
ing second, R2 has been sitting in a motionless state, at rest in a peaceful lunar
dustbowl. Now focus on an instant of time just before R1 slams into the lunar
surface. At this particular instant, if the at-at account is correct, there is no intrinsic
difference between R1 and R2. That is to say, so far as their instantaneous intrinsic
properties are concerned, these momentary phases of R1 and R2 are completely
indistinguishable. But can this be right? Given that R1 is moving at that instant
and moving with a frightening velocity whereas R2 is motionless, there is surely
a significant difference between the two: a difference between how the two are at
that very instant.
Although scenarios along these lines can make it seem intuitively very plausible
to suppose that motion makes a difference to the intrinsic condition of an object,
proponents of the at-at approach will be swift to point out that much of their
power derives from a tacit assumption that there is a real, frame-independent,
distinction between motion and rest. Velocity may be absolute in Newtonian
spacetimes, but as we saw in Chapter 12, there are classical spacetimes (of the
neo-Newtonian) variety where uniform velocity is frame-relative. If a body at a
time t can be described with equal legitimacy as being both in motion and at rest
at t, the notion that the intrinsic condition of an object is affected by its state of
motion is greatly undermined. And of course, it is widely agreed that velocity in
our universe at least is relative rather than absolute.
Recognizing that motion is relative in many possible universes if not all
goes a long way towards making the at-at theory acceptable. However, there are
other challenges lying in wait. From a purely conceptual point of view, if the at-at
account is correct it is literally senseless to suppose that an object that exists only
for a single moment could be in motion. For if being in motion involves nothing
more (nor less) than occupying different spatial locations at neighbouring times,
an object cannot possibly be in motion if it does not exist at any neighbouring
times. But this can seem strange. Cant we imagine an object flashing into exist-
ence only for a moment, before disappearing again, but being in motion for the
brief time it does exist? In a similar vein, Tooley presents a more detailed scenario,
involving an unusual force field that causes objects to flash in and out of exist-
ence an infinite number of times over any spatial interval (1988: 2478). More
specifically, objects (momentarily) exist only at those points whose distance from
the centre of the field is specified by an irrational number, relative to a particular
method of measurement. As an object enters and moves through this field at a
constant speed something we can easily visualize it seems very natural to say
292 TIME AND SPACE
that it is moving at each of the moments at which it exists, even though it does
not exist at the moments neighbouring these moments.
2
In reply defenders of the
at-at theory must acknowledge that on their view Tooleys object is indeed not in
motion, but they will also cast doubt on the reliability of the relevant intuitions:
is it really that clear that the object in question is moving at the instants at which
it exists?
Arntzenius has recently focused on a different difficulty:
Consider a ball moving from right to left through some region in space, and
a qualitatively identical ball, perhaps the same one, some time later, moving
from left to right through that very same region. If the full state at a time of
a ball does not include an instantaneous velocity then the full state of the two
balls is exactly the same when they occupy the same region. So how does the
one ball know to keep moving right and the other ball know to keep
moving to the left? Or, removing the anthropomorphism: why does the one
ball subsequently move to the left and the other ball subsequently move to
the right? Surely it is conservation of velocity, or something like that, which
determines that the one ball will keep moving to the left and the other will
keep moving to the right. But if there is no such thing as instantaneous veloc-
ity, as there is not in the at-at theory, then why do the balls continue their
motions in different directions? (Arntzenius 2000: 2)
The motion of a classical object depends on a combination of dynamic and
kinematic factors: if a force impacts on an object at a time t, the way it reacts
depends on (a) the mass of the body, (b) the size and direction of the force, but
also (c) the objects state of motion at t, that is, its velocity (its speed in a given
direction). The equations of classical physics do treat velocity as a property of
strictly instantaneous states. However, these equations are formulated in the cal-
culus. And in the latter in the form made rigorous by Cauchy and Weierstrass
values for the velocity of an object at a time t are arrived at by calculating series
of ratios of distance to time in the neighbourhood of t, in the manner outlined
above (see Figure 17.1). Hence, strictly speaking, and despite appearances to the
contrary, instantaneous velocities play no role in classical physics; what do play an
important role are temporally extended phases of systems in the case of velocity,
variations in location over finite intervals of time.
Taking these temporally extended segments to be fundamental, from a causal
and explanatory point of view, may turn out to be acceptable, Arntzenius sug-
gests, but there is also a significant cost: the state of a ball at a time does not
determine the direction in which it continues its motion. And this implies that it
is not the case that all influences from the past upon the future are represented
in the present (ibid.: 3). As a result of this, the at-at view is incompatible with
determinism in the form elaborated by Laplace. Writing in 1814, he suggested in
his Essai philosophique sur les probabilits that:
An intellect which at a certain moment would know all forces that set nature in
motion, and all positions of all items of which nature is composed, if this intel-
lect were also vast enough to submit these data to analysis nothing would
be uncertain and the future just like the past would be present before its eyes.
ZENO AND THE CONTINUUM II 293
The key phrase here is at a certain moment. Determinism, in its classical
guise, is the doctrine that it is possible to predict the entire future history of the
universe, and retrodict its entire past history, from a complete description of its
condition at the present moment of time. Supply Laplaces vastly powerful intel-
lect with all the positions and velocities of all the objects that exist at any one
moment in history, and it will be able to work out everything that happens before
and after that moment. But if velocities are not intrinsic to the instantaneous states
of moving objects, this form of determinism is untenable. In the light of quantum
(and chaos) theory, we might well think there are strong grounds for supposing
that Laplacean determinism does not apply in our universe. But it is decidedly odd
to think that the nature of velocity, all by itself, rules out determinism and does
so even in classical universes.
If these consequences of the at-at theory are deemed unacceptable, rather than
merely counterintiuitive, what alternatives are on offer? The obvious route to take
is to hold that the at-at account does not provide a sufficient condition for motion:
something more than change in location over time is involved. But what is this
additional feature? Several proposals are on offer.
Worried by what keeps a projectile (such as an arrow) in motion after the initial
act of propulsion, some medieval thinkers posited an intrinsic property known as
impetus, which serves precisely this function: possession of this property disposes
an object to move at a certain speed in a certain direction. (It was a part of this
view that the impetus of a moving body is gradually used up, which is why objects
slow down and eventually come to rest.) Perhaps surprisingly, a number of con-
temporary authors have sought to defend something along the same lines. Lowe
writes we can perhaps say that an instantaneous velocity of an object at a time t
is a directional tendency which it possesses at t, in virtue of which, if it continues
to possess it at t, it will undergo a subsequent change of position (2002: 302).
Tooley (1988) takes instantaneous velocity to be a functionally defined intrinsic
property; roughly speaking, it is that intrinsic property of an object whatever its
intrinsic nature turns out to be which plays the role of instantaneous velocity in
physics, and does so in virtue of the laws of nature operative in our universe. For
Tooley, along with Bigelow and Pargetter (1990), velocity should be regarded as a
primitive property of material objects, in the manner of mass or charge.
Critics have pointed out that since this new primitive property has precisely
the same empirical consequences as the orthodox (at-at) account of velocity, it is
vulnerable to Occams razor. Marc Lange also finds the primitive property view
objectionable, but for a different reason. The idea that mass or charge have a real
but inscrutable intrinsic nature has some plausibility, as we saw in 15.2 in con-
nection with Fosters spatial anti-realism. But the analogous doctrine in the case
of velocity that we know nothing about what it really is, only what it does is
absurd, suggests Lange: for we surely know that velocity involves motion along a
trajectory (2005: 45051). Since he too has doubt about whether the at-at account
can adequately accommodate the causal and explanatory roles that velocity plays,
he proposes a variant on the impetus theory: I propose that classical instantane-
ous velocity is something like a dispositional property, a tendency, a power or a
propensity. To be moving at t
0
in a given direction at 5 centimetres per second is
to have a potential trajectory that is, to be threatening to proceed along a certain
294 TIME AND SPACE
trajectory (ibid.: 4523). Lange goes on to explicate this dispositional property
in counterfactual terms: (roughly) an object possesses it at t
0
if it would be moving
in a certain way over arbitrarily short intervals after t
0
if the object were to exist
at those times. This account has its advantages instantaneous velocity is being
analyzed in purely kinematic terms but as Lange himself points out, it too has
its counterintuitive consequences: if a body is moving now at 5 cm/s not in virtue
of its current categorical properties, but in virtue of where it would be were it to
continue to exist, then two bodies could be thoroughly alike now, as far as their
categorical properties are concerned, and yet possess different velocities now
(ibid.: 460). Adherents to the more traditional understanding of impetus an
intrinsic property that makes objects apt to move in certain ways will not find
this aspect of Langes account palatable.
17.3 The Stadium
The fourth and last of Zenos motion paradoxes is often called the Stadium, or
sometimes (and more illuminatingly) the paradox of the moving rows. Quite
how Zeno formulated it, and precisely what he intended it to demonstrate, is
unclear. There is, however, something of a consensus as to the most interesting and
intriguing way of construing it. Suppose we have three rows, A, B and C with each
row containing three equally spaced objects, positioned in the way shown in Phase
1 in Figure 17.2. We further suppose that row A is not moving, but rows B and C
are, and at the same constant speed but in opposite directions: row B is moving
to the right, and row C is moving to the left. One second later, the objects have
acquired the new positions shown in Phase 2: A
1
, B
1
, C
1
are now aligned in the up
down direction, and similarly for A
2
, B
2
, C
2
and A
3
, B
3
, C
3
. (As Aristotle relates the
tale, these objects pass one another on a racecourse, hence the Stadium label.)
There is nothing in the least puzzling or problematic here, one might think.
Zeno thought otherwise. Take a look at object C
1
. This starts off aligned with
B
3
and it ends up aligned with B
1
, so it has moved the whole distance separating
B
1
and B
3
in our allotted time of one second. But in relation to row A this same
object has moved only half this distance over this time: C
1
starts off aligned with
A
2
and finishes aligned with A
1
, that is, a distance of one block rather than two.
A
1
A
2
A
3
B
1
B
2
B
3
C
1
C
2
C
3
A
1
A
2
A
3
B
1
B
2
B
3
C
1
C
2
C
3
Phase 1
Phase 2
Figure 17.2 Zenos Stadium paradox. The B-rows and C-rows move in opposite directions, ending
up in alignment under the A-row. The paradox is also known as Moving Rows (or Moving Blocks).
ZENO AND THE CONTINUUM II 295
So C
1
s movement relative to row B is twice that of its movement relative to row
A. As a result, in a single second C
1
moves double the distance with respect to B
1
as it does to A
1
. Zeno found this discrepancy deeply objectionable. But it is by no
means obvious that he was right to do so. If C
1
managed to accomplish this feat by
moving at the same rate relative to each of the other rows, we would certainly be
faced with a problem. But of course it does no such thing: it is moving twice as fast
relative to the Bs as to the As. Since the Bs are moving in the opposite direction
to the Cs, there is nothing at all surprising here. As Aristotle noted: The fallacy
of the reasoning lies in the assumption that a body takes an equal time in moving
with equal velocity past a body that is in motion and a body of equal size that is
at rest (Physics Z9, quoted in Kirk & Raven 1969: 295).
Is that all there is to it? Was Zeno simply oblivious to the existence of relative
velocities, and the fact that an object can be moving at different velocities relative
to each of two other objects that are moving relative to one another?
3
On a more
charitable interpretation, there is rather more to it. Whereas Zeno intended clearly
some of his paradoxes (e.g. the Achilles and the Dichotomy) to undermine the view
that space and time are continuous, he may also have wanted to undermine the
competing position: the doctrine to which some ancient Greeks subscribed that
time and space are granular or discrete. On the latter view, there are, in effect,
atoms of space and time (sometimes called hodons and chronons, respec-
tively). Rather than being infinitely divisible, on this view spatial and temporal
intervals are only finitely divisible: there will be a certain fixed number of spatial
atoms in a metre, and a certain fixed number of temporal atoms in an hour. No
change can occur within a single chronon, and nothing can be divided beyond
the spatial limit set by the dimensions of the hodon. The discrete conception can
be elaborated in different ways, but according to one common approach hodons
in a two-dimensional space are small areas, whereas in three-dimensional spaces
they are small volumes. Cellular automata provide a familiar and useful model for
discrete spacetimes; see Figure 17.3 for a simple example.
4
As for the problem the Stadium poses for the discrete model, let us suppose
that the scenarios depicted in Phase 1 and Phase 2 in Figure 17.2 are separated
by the least possible temporal distance (or one temporal atom). It follows that
nothing can occur between these two times. Now focus once more on C
1
. At the
t
3
t
1
t
2
Figure 17.3 Cellular automata inhabit discrete spaces and times. In this simple two-dimensional
example, each box (or cell) has just two states live (black) or dead (white) and change unfolds in a
series of discrete steps, or ticks. Appropriate patterns of activation can give rise to stable structures that
can move, in the way illustrated here. In Conways well-known game of life, remarkably complex patterns
of behaviour are generated by a few simple rules: (i) any live cell with fewer than two live neighbours dies;
(jj) any live cell with more than three live neighbours dies; (iii) any live cell with two or three live neighbours
lives on to the next generation; (iv) any dead cell with three live neighbours becomes a live cell.
296 TIME AND SPACE
first discrete instant it is (vertically) aligned with B
3
and A
2
; at the next it finds
itself aligned with A
1
and B
1
. Since A
1
is lying adjacent to A
2
there is nothing at
all puzzling here, but it is otherwise with the B-row: what has happened to B
2
?
It seems that C
1
has managed to pass right by B
2
without there being any time at
which it did so (see Figure 17.4). On the assumptions currently in play, there are
no instants between the two depicted. If we find this absurd, we only have one
obvious alternative: we must abandon the discreteness hypothesis, and accept as
on the continuity model that there is a instant between any two instants.
If this is Zenos argument, it is certainly an intriguing one. However, it is by no
means unanswerable, for the proponent of the discrete conception can reply along
these general lines: Zenos scenario doesnt demonstrate that the discrete model is
incoherent; it merely illustrates that it has some unusual, and perhaps counterin-
tuitive, features. So far as C
1
and B
2
are concerned, the proponent of discreteness
will simply hold that they do not pass each other by in the normal sense: that is,
there is no time at which they are vertically aligned. A complete description of
what happens to C
1
over the interval in question is simply this: at one instant C
1

is aligned with B
3
(and A
2
); at the very next instant it is aligned with B
1
(and A
1
).
To assert that there must be a time at which C
1
is aligned with B
2
is begging the
question against the discreteness hypothesis, for it is tantamount to stipulating that
there is always a time between any two times.
It is true that in the scenario being envisaged objects seemingly leap discon-
tinuously from one location to another, but these discontinuous movements are
problematic only if we start from the assumption that space and time are continu-
ous; if they are discrete, then there is simply no other way to move. Indeed, as
Wesley C. Salmon points out (1975: 65), if space is discrete, the sort of pass-
ing without actually passing that Zeno found puzzling would be commonplace.
Imagine there are two arrows moving towards one another through a discrete
space along adjacent parallel paths; the arrows are moving in the simplest and
most straightforward fashion possible in a discrete spacetime, at a rate of a single
grid-square per single unit of time. As can be seen in Figure 17.5, the arrows pass
each other by, but they do so without spending any time adjacent to one another.
B
1
B
2
B
3
C
1
C
2
C
3
one moment
B
1
B
2
B
3
C
1
C
2
C
3
the very next moment
Figure 17.4 Zenos worry. C
1
travels right past B
2
, but at no time is it right next to it.
Figure 17.5 Motion through a discrete
space. The solid outline shows the
current location of the arrows, and the
broken outline shows where they will be
at the very next moment of time.
ZENO AND THE CONTINUUM II 297
More complicated forms of motion can also be envisaged: for example, instead
of moving in a straight line at a rate of one atom of space per one atom of time, an
object might move two (or more) atoms of space per one atom of time. Alterna-
tively, an object might move only at every other atom of time. Again, while these
modes of motion bigger jumps, pauses between jumps might well strike us as
odd, they are not obviously logically incoherent. Moreover, they will be necessary
if the discrete model is going to be able to handle objects moving with a wide range
of different speeds.
Tiling problems
The discrete model faces hurdles other than the one constructed by Zeno. In his
Philosophy of Mathematics and Science, Hermann Weyl writes:
So far the atomistic theory of space has always remained mere speculation
and has never achieved sufficient contact with reality. How should one
understand the metric relations in space on the basis of this idea? If a square
is built up of miniature tiles, then there are as many tiles along the diagonal
as there are along the side; thus the diagonal should be equal in length to the
side. (1949: 43)
What is now generally known as Weyls tile argument is found in this passage.
To grasp what Weyl has in mind, let us suppose that we model a plane in a discrete
space with a 5 5 grid, as shown in Figure 17.6. Now consider the lengths of the
triangle ABC in the diagram on the left: its vertical and horizontal sides are both
five units long, and so is the hypotenuse AC assuming we measure length in the
particular way indicated, along the path of the dotted diagonal, where distance is
determined solely by the number of squares traversed (as with a queens movement
in chess). Hence the problem. In a Euclidean space, the lengths of sides of a right-
angled triangle conform to Pythagorass theorem: a
2
+ b
2
= c
2
. Applying this to
our right-angled triangle, the square of the length of the diagonal (or hypotenuse)
should be equal to the sum of the square of the two others: 25 + 25 = 50. The
square root of 50 is approximately 7.07, which is considerably more than 5. A
natural response to the tile argument is to appeal to the micro-level: perhaps the
A
B C
A
B C
Figure 17.6 Weyls tile problem. The hypotenuse AC of the right-angled triangle ABC in the diagram
on the left is exactly the same length (5 units) as the shorter sides AB and BC. Reducing the size of the
spatial atoms produces no improvement, as can be seen from the figure on the right.
298 TIME AND SPACE
A
B
C
Figure 17.7 In a discrete space, lines will have a
finite width, and so, in effect, be rectangles. Jean
Paul van Bendegem (1987) shows that Weyls
problem can be avoided if we take the length of a
line to be determined by the number of squares it
wholly or partially covers or contains.
discrepancies will be so small as to be negligible if the atomic spatial units are very
small. However, as Salmon points out, this is not the case. If we double the number
of unit-squares as shown in the diagram to the right of Figure 17.6 the lengths
of the sides is now 10, but the length of the diagonal is also 10, and so nothing
has changed. And the same applies in a quite general fashion: irrespective of how
small we make the tiles, the hypotenuse will always have the same length as the
other two sides. Hence Salmons conclusion: This is one case in which transition
to very small atoms does not help at all to produce the needed approximation to
the obvious features of macroscopic space. It shows the danger of assuming that
such approximation will automatically occur as we make the divisions smaller and
smaller (1975: 66).
5
Weyls tile problem is a serious one, but responses are available. One option
is to start from the assumption that lines in a discrete space will always have
some finite thickness, and then define length in terms of the number of spatial
atoms in the resulting rectangles, as shown in Figure 17.7. Simplifying somewhat,
rather than equating lengths with tile-sequences of this kind , which
are quite sparse numerically speaking, we work instead with sequences of this
kind: . See Van Bendegem (1987, 2010) for further details.
A disadvantage of Van Bendegems scheme is that it requires ordinary (continu-
ous) Euclidean geometry to work out the details, and so is not wholly in alignment
with the tenets of the discrete model. Peter Forrest opts for an alternative. He sug-
gests that what Weyls argument shows is that it is a mistake to represent a discrete
space as a tiling of squares, or any other regular shape. Not only does it lead to
metrical problems, but if the tile picture is taken literally, individual points have
extension, and their different parts can be at some distance from one another. This
is an undesirable result: distance in a discrete space is supposed to be defined in
terms of relationships between distinct points (Forrest 1995: 33031). By way of
an alternative, Forrest models discrete spaces in terms of chains of adjacent points;
although at the micro-level distances in this sort of space are non-Euclidean, on
larger scales they can mimic Euclidean geometry to any degree of closeness one
might want (ibid.: 2.3, 343ff.).
The tiled-plane mode of representation has a further disadvantage: it naturally
gives rise to preferred or distinguished directions (e.g. in the case of a chequer-board,
ZENO AND THE CONTINUUM II 299
Figure 17.8 A Penrose tiling.
along the horizontal or vertical axes), whereas Euclidean space is isotropic, that is,
it has the same structure in all directions. As Van Bendegem (2010: 2.5) notes,
one option is to use non-periodic tiling for example, the Penrose tiling shown
in Figure 17.8 to model discrete spaces. By virtue of being aperiodic, there is
no analogue of a classical straight line in a Penrose tiling. In the 1980s physicists
working in the field were surprised to discover, and initially reluctant to accept,
that there are states of matter quasicrystals whose structures are aperiodic in
just this sort of way. If matter can be aperiodic, perhaps a discrete space can be too.
A second option is to accept Forrests proposal, and abandon the idea that spatial
atoms have a determinate shape at all. The spaces Forrest develops are anisotropic
locally, but this anisotropy is a microgeometric phenomenon which becomes
less and less noticeable at the macrogeometric limit (1995: 338). The latter claim
looks quite plausible if, as Forrest speculates, there are in the order of 10
30
spatial
atoms per centimetre: so sized, the irregularities are very remote indeed.
17.4 Could our spacetime be discrete?
So the discrete model, in some form or other, may well turn out to be conceptu-
ally coherent. This gives rise to a question: might it apply to space and time in
our universe?
The discrete model undeniably sits badly with the mathematical techniques
standardly employed in physics, for as we saw in 16.2, most of these rely on
the assumption that space and time are continuous (more on this in a moment),
but against this it also has very definite advantages of its own. Zenos Dichotomy
reasoning obviously depends on space and time being infinitely divisible: runners
will have no difficulty reaching the finishing line if they only ever have to traverse
a finite number of spatial atoms to do so. Similarly for the Plurality paradox:
since on the discrete conception space and time are not composed of durationless
instants, the issue of how even an infinite number of such things could ever add up
to anything possessing a finite extension simply does not arise. There is a further
300 TIME AND SPACE
advantage. We saw in 16.6 that although metrical properties can be assigned
via coordinates, the Cantorian continuum does not possess any intrinsic metrical
properties. Discrete spaces and times do: counting the number of atoms that
compose an interval provides a natural measure of that intervals size, and what
goes for intervals goes for volumes, whether of space or spacetime.
Of course, none of this means the discrete model does apply to our world; it just
means that there would be certain advantageous consequences if it did. The most
obvious objection to the claim that it does is that in our world things mostly seem
to move in continuous ways: in the case of ordinary-sized objects even our most
sensitive measuring equipment doesnt reveal discontinuous and jerky motions of
the kind we have been envisaging. However, the proponent of discrete space and
time has a ready reply: the granularity of the world is real, but exists only at the
micro-level, at distances we are unable to observe; this is why the world seems
smooth, even if it is really discrete. And this reply seems effective, albeit rather
unmotivated, given the empirical inaccessibility of the postulated granularity.
Nonetheless, the discrete theorist can at least lay claim to an account of the world
that appears to be empirically equivalent to the orthodox continuity-based model.
More important, ultimately, is the conception of space and time (or spacetime)
to be found in our best scientific theories. Newtonian space and time are continu-
ous, not discrete, and the same applies to the spacetimes of Einsteins special and
general theories of relativity. By contrast, quantum theory might seem to be more
congenial to the discreteness hypothesis: if physical quantities such as angular
momentum and charge are quantized, might not the same apply to space and time?
Grnbaum warns against this. From the fact that some properties of single-particle
systems are quantized, it would be wrong to conclude that space and time are also
quantized in the theory:
For they are not. Every point of continuous physical space is a potential
sharply defined position of, say, an electron, and, separately, every instant of
a continuous time is potentially the time of a physical event Of course, this
sense of affirming the continuity of space and time fully allows the quantum
theory to repudiate via the Heisenberg Uncertainty Relations the exist-
ence of well-defined particle trajectories in the sense of classical mechanics.
(1968: 116)
However, the fact that our best current theories are formulated in terms of
continuous space and time does not mean that our future best theories will be.
Quantum theory and general relativity our best current guides to the realms of
the very small and the very large, respectively, are incompatible. Quite what form
a viable quantum theory of gravity will take is currently unknown; much work on
the topic has been done, and some headway has been made. Interestingly, some
of the more promising lines of development reject the continuous spacetime of
general relativity in favour of a quantized conception. Here is Lee Smolin describ-
ing one such: According to loop quantum gravity, space is made of discrete atoms
each of which carries a tiny unit of volume. In contrast to ordinary geometry, a
given region cannot have a volume which is arbitrarily big or small instead,
the volume must be one of a finite set of numbers (2000a: 106). Since these
minimum volumes are exceedingly small 10
99
would fit into a thimble it is not
ZENO AND THE CONTINUUM II 301
surprising that the macroscopic world seems to be entirely smooth (we will be
encountering some further features of this approach in 22.7).
We should not race to any conclusions other approaches to quantum grav-
ity, such as some versions of string theory, are formulated in terms of orthodox
continuous spacetimes. But nor should we rush to rule out the discrete conception
simply because history (thus far) has favoured the smooth alternative.
6
17.5 The standard continuum: basic concerns and further puzzles
As is by now familiar, on the standard view, space and time are composed of
dimensionless points, and the latter form a non-denumerable continuum, in the
manner of their counterparts on the real number line. For all its consilience with
the mathematical methods deployed with such spectacular success by physicists
and engineers over the past few centuries, there has always been opposition to
this way of conceiving of the micro-structure of space and time. Some of this
opposition derived from paradoxes and puzzles of the genre associated with Zeno.
As we have seen over the course of this chapter and it predecessor, proponents of
the orthodox view have responses to these, responses that many find satisfactory.
Even so, opposition to the standard view has continued, with some of it coming
from mathematicians and physicists with a deep understanding of the mathematics
underlying the standard view. To round things off we shall take a brief look at what
motivates this long-running opposition, and some of the developments to which it
has led. We have already encountered one alternative, in the form of the doctrine
that space and time are quantized or discrete, rather than continuous. As we shall
see, this is not the only alternative.
Density and continuity
The defining characteristic of a continuum is continuity. Can a collection of dis-
crete dimensionless points ever amount to something truly continuous? Poincar
was fully versant with Cantors achievements, but he nonetheless found cause for
complaint:
The continuum thus conceived is no more than a collection of individuals
arranged in a certain order, infinite in number, it is true, but external the
one to the other. This is not the ordinary conception, in which it is supposed
that between the elements of the continuum exists an intimate connection
making of it one whole, in which the point has no existence previous to the
line, but the line does exist previous to the point. Multiplicity alone subsists,
unity has disappeared the continuum is unity in multiplicity, according
to the celebrated formula. the real mathematical continuum is quite dif-
ferent from that of the physicists and from that of the metaphysicians.
(1952: 18, translation modified)
Gdel was of a similar view: According to this intuitive concept, summing up
all the points, we still do not get to the line; rather the points form some form of
scaffold on the line (quoted in Wang 1974: 86). In a similar vein, Weyl spoke of
302 TIME AND SPACE
the orthodox model shattering the continuum into isolated points (1998: 115),
and sought an alternative, as did Brentano and Peirce.
7

In one respect at least, it is not difficult to see what these critics are worried
about. Because the orthodox continuum is dense, between any two points there is
always a further point, and as a consequence it is impossible for two points to be
right next to one another. Points in a dense continuum can never be in immediate
contact with one another, and for this reason such a continuum might be said not
to be maximally or truly continuous. Defenders of the orthodox view will insist
No, this objection is utterly mistaken. The continuum is truly continuous by
virtue of the fact that it contains no gaps whatsoever: all the gaps between any
two points are always filled, exhaustively, with more points. And, of course, this
is perfectly correct.
Even so, since none of these further points will ever be in immediate contact
with one another, there is still room for critics to claim we have not yet achieved
and never will achieve a true or maximal continuity. In our ordinary perceptual
experience for example, of smooth spatially extended expanses of colour, or
temporally extended sounds or tones we find plenty of examples of continua
that are (seemingly) wholly gap-free and seamless, where the adjacent parts of
phases are welded together, as it were, to form a unified plenum. By virtue of its
point-structure and density, the orthodox continuum does not and cannot possess
anything resembling this sort of unity, and for this reason at least in the eyes of
Bergson, Brentano, Poincar, Weyl, Whitehead and others it falls short of pos-
sessing authentic continuity, hence the search for an alternative.
Plurality and substance
The problem posed by Zenos Plurality challenge runs thus: how can an extended
line or volume be composed of ingredients which are dimensionless? Even an
infinite number of 0s adds up to 0. We saw in 16.6 that there are responses
available to proponents of the orthodox view. As Grnbaum noted in his influen-
tial discussions of the problem, the notion that the size of a whole is an additive
function of its constituent parts is certainly viable for ordinary finite things, and
can be extended to things with denumerably many parts, but it has no applica-
tion whatsoever for things possessing non-denumerable numbers of parts and of
course if the continuum is isomorphic with the real number system, then it does
possess a non-denumerable infinity of parts. When dealing with non-denumerable
infinite collections, our ordinary ways of assigning and combining magnitudes will
not work; new methods are needed. Contemporary measure theory fits the bill.
It allows us to assign a positive non-zero measure (or size) to continua that are
entirely composed of dimensionless points.
Although some find this resolution satisfactory for example, Graham Oppy
has recently praised Grnbaums analysis as exemplary (2006: 104), and see
Zimmerman (1996a: 1) it is vulnerable to an objection along the following lines:
The considerations that Grnbaum brings into play are very relevant to
various mathematical issues relating to the real-valued continuum, but they
contribute little or nothing to resolving some of the key metaphysical issues.
ZENO AND THE CONTINUUM II 303
For consider: can we really make sense of the claim that a concrete thing a
material object, say is composed entirely of zero-sized parts? Ontologically
speaking, brick, trees and planets are very unlike abstract entities such as
sets or numbers. This is a difference that matters. Mathematicians may have
worked out fruitful ways of assigning magnitudes to infinite collections of
abstract entities, but this is of limited relevance to the real problem posed by
Zeno. For it remains very difficult to comprehend how a concrete material
thing could be entirely composed of dimensionless parts. A piece of matter
that is truly of zero size would be indistinguishable from nothing at all; a
million nothings is nothing, a countably infinite number of nothings is still
nothing, and the same applies to an uncountably infinite number of nothings.
You cant construct a brick, a tree, or any other substantial material thing
from ingredients that singly and collectively amount to absolutely nothing.
This sort of objection has some considerable intuitive force in the case of ordi-
nary material things, or rather, ordinary material things as ordinarily conceived. The
pre-scientific (or naive) conception of a material object is of a three- dimensional
impenetrable solid, filled with coloured stuff of some sort, in the manner of a hunk
of modelling clay. It is certainly not easy to comprehend how substantial solids
of this sort could be entirely composed of zero-dimensional parts. But of course
modern physics undermines the intuitive conception of the material world; if
current theories are broadly correct, bricks and trees are composed of elementary
particles that are themselves patterns of activity in various space-filling quantum
fields. It is by no means so obvious, intuitively, that these continuous fields could
not be composed of zero-sized parts.
However, our concern here is not with matter, but with space (or spacetime).
Substantivalists regard space as a concrete material thing in its own right, albeit of
a distinctive kind. Further, by virtue of its continuity and homogeneity, substanti-
val space has at least some of the features of the ordinary pre-scientific conception
of material objects; it consists of extended volumes, filled everywhere with (spatial
or spatiotemporal) stuff. Accordingly, if the substantival conception of space is the
correct conception of space, the objection to Grnbaum outlined above still has
considerable force. It is not easy to see how a substantival space could be entirely
composed of zero-sized parts, for a zero-sized portion of substantival space just
does not exist, or so it might reasonably be claimed.
The situation looks very different, however, on the opposing relationist con-
ception of space. For relationists, the spatial (or spatiotemporal) continuum
does not exist as a physical ingredient in its own right; all that exist are mate-
rial objects and their distance relations. On this view the extended space-filling
continuum is at best a useful fiction. Since this fiction standardly takes the form
of a mathematical (or set-theoretical) representation, Grnbaums treatment of
Zenos Plurality puzzle looks eminently appropriate: if the continuum is no more
than a mathematical abstraction, there is no obvious objection to applying a con-
sistent set-theoretic account of magnitude or measure to it. But since one of the
main lessons of earlier chapters is that the substantival view of space and space-
time has a great deal to be said for it, this result does not have the significance it
might otherwise possess.
8
304 TIME AND SPACE
Division and contact
The standard view of the continuum is afflicted by a variety of puzzles and problems,
some of which we have already encountered, explicitly or implicitly; others are new.
To start with one of the simplest, take the real number interval [0, 1], and imagine
dividing it without remainder into two portions of exactly equal size, P
1
and P
2
, by
inserting an (infinitely sharp) knife at the point p that lies exactly halfway between
0 and 1. At first view this may seem unproblematic. Locating p is easy: the point
at 0.5 lies halfway between the points 0 and 1, so it suffices to insert our (mental)
cleaver right there. But matters are not quite so straightforward. What do we do
with p itself? Since our division is to leave no remainder, p must end up in P
1
or P
2
.
If it ends up in P
1
, the latter becomes a closed interval and P
2
is left half-open (in
the senses noted in 16.7), whereas if p ends up in P
2
the latter is now closed, and
P
1
is open. One peculiarity of this situation is that nothing in the geometrical or
arithmetical facts seems to determine where our division leaves p: the stipulation
that it is to be found in P
1
is as well motivated as the stipulation that it has ended
up in P
2
. There is a further peculiarity: whichever of these options we choose, the
result will be that P
1
and P
2
will have a point the other doesnt get, and so will in
this respect be larger, even though they are of the same length from a metrical (or
measure-theoretic) perspective. Gdel once remarked that it is natural to think that
dividing a geometrical line at a point will produce a perfectly symmetrical result, that
is, that the two resulting lines will be exactly alike, or mirror images of one another
(see Putnam 1994: 3). The standard model is incapable of yielding this result.
There are further oddities if we apply the standard model to physical reality.
Take a cube-shaped region of (substantival) space and divide it into two equally
sized portions with the aid of a two-dimensional plane P. The result will be one
region that is open in the neighbourhood of P and one region that is closed by
virtue of having P as its outermost surface. Intuitively, it is very difficult to make
sense of this: imagining a closed region of space poses no difficulties (it has well-
defined boundaries on all sides), but what is an open region like? It would be
wrong to lay too much emphasis on this; after all, when dealing with the infinitely
divisible, what we can or cannot easily imagine may well be a poor guide to what
is really possible. But the envisaged state of affairs is potentially problematic for
a further reason. The two regions created by our division have exactly the same
measurable size, but they are nonetheless different: one possesses a (single) layer of
points that the other lacks. Are we to suppose that there can be regions of physical
space that differ by only a single layer of points? Do dimensionless points have any
physical significance at all? The standard model may be guilty of positing a surplus
of structure. This is the conclusion Michael Dummett draws:
The classical model [of the continuum] is to be rejected because it fails to
provide any explanation of why what appears to intuition to be impossible
should be impossible. It allows as possibilities what reason rules out, and
leaves it to the contingent laws of physics to rule out what a good model of
physical reality would not even be able to describe. (2000: 505)
The density of the continuum can also be a source of puzzles. If material objects
exactly occupy closed regions of space as seems most plausible, intuitively at
ZENO AND THE CONTINUUM II 305
least their outermost layer will consist of a two-dimensional skin that is just
one point thick. This has the surprising consequence that material objects cannot
possibly come into direct contact with one another: for no matter how close to one
another the points in their outermost layers manage to get, thanks to the density
of space there will always be a further point separating them, and so they will not
actually be touching one another. We might suppose that when they get very close
to one another they instantaneously cross the remaining points that separate them,
and achieve contact by coming to share a boundary, in the form of a surface layer
of points that is common to both objects. But this is interpenetration or overlap,
and not contact as we normally conceive it. Is it really impossible for material
objects to just touch each other?
As noted above, modern physics teaches us that macroscopic objects are not
solids, in the intuitive sense being presupposed here, and are kept apart by the
repulsive negative charges in the electrons in their surfaces. Nonetheless, it is sur-
prising to discover that the standard view of the continuum seems to be incompat-
ible with contact between objects as we intuitively conceive of them. If truly solid
objects were to exist by virtue of some fluke or miracle the nature of space is
such that they would not be able to touch each other. For more on this see Zim-
merman (1996a,b), Hudson (2005: chs 2 & 3).
Measure-related puzzles
Last but not least, there are the so-called paradoxes of measure, the best-known
of which is the TarskiBanach paradox. The latter is the astonishing proof that
a spherical region of space (or object) S can be divided into five (or more) pieces,
which when rearranged form two new spheres, each of which is precisely the
same size as S was originally. This is undeniably bizarre: when we divide a sphere
into parts, the parts will occupy the same volume as the whole sphere (assuming
nothing is lost in the cutting-up), so how can simply rearranging the parts lead to
a doubling of volume?
The answer lies with the peculiarities of the Cantorian continuum. As we saw
in 16.5, every interval on the real number line contains as many points as every
other interval, or indeed, as many points as are contained in a two-dimensional
plane, or a three- or n-dimensional volume. Since the original sphere contains as
many points as the two it divides into, it is not wholly surprising that Tarski and
Banach could prove what they did. Nonetheless, there is still a difference: the
rearrangement employed in the TarskiBanach proof consists in the manipula-
tion of only a finite number of regions, not an indenumerably infinite number
Figure 17.9 A puzzling division. The asymmetrical
result of slicing a three-dimensional object here depicted
only schematically into two halves along the line L. One
of the resulting halves is fully closed and contains the
points in L, and the other is left partially open.
L
L
306 TIME AND SPACE
of distinct points. So how is the trick worked? The details of the proof depend
on each of the finitely sized parts being infinitely complex in form, in a way that
means they are unmeasurable, that is, have no meaningful size or volume. It is
this indeterminacy that allows them to combine in the seemingly remarkable way
they do. Since it is not easy to believe that real space can behave in this manner,
this measure-related result provides those hostile to the standard view of the con-
tinuum with yet a further weapon.
9
So far as the tenability of the orthodox conception of the continuum is con-
cerned, it may be that none of the difficulties just outlined is decisive, but they
do have a cumulative effect: an alternative model with fewer (or perhaps just
different) problematic features would certainly be welcome. As for alternatives,
there has been no shortage of proposals, but most fall into one or other of two
main categories: some theorists have been led to posit more points, whereas others
have been driven to denying the existence of points altogether. Let us start with
the first option.
17.6 Are more points the answer?
Although Peirce was of the view that the continuum is metaphysically prior to the
existence of any points it may contain, he also believed unlike Aristotle that
we could legitimately view the continuum as possessing such points. But whereas
the orthodox view is that there are only as many points in the continuum as there
are real numbers, Peirce believed that this did not go nearly far enough. As Can-
tor demonstrated, there are higher-order infinities, an infinite number of them.
Since there are bigger infinities than c, if we hold that the continuum has only c
points then it doesnt have as many points as it is possible to have. For Peirce this
was a deficiency: the true continuum will have the maximum possible number of
points. This absolute infinity, as it is sometimes called, includes all the infinite
sets in Cantors hierarchy of ever larger infinities; as such it is larger than any one
set, and so not itself a set (in contemporary terms it is what is known as a proper
class). Somewhat obscurely, Peirce held that when points are packed together in
this hyper-dense fashion, they lose their individuality and fuse together to form
a true, cohesive, continuum. This doctrine allows Peirce to develop an intuitively
more plausible conception of geometrical division.
On the orthodox view, slicing a rectangle into two along a line L will produce
one open and one closed object, as shown in Figure 17.9. Peirce argues for a differ-
ent result. On his view, the points that jointly constitute the line have a rich inner
structure; in effect, each individual point on the real line contains many further
points, at least as many as the power set of c. When the rectangle is divided along
line L, the points in L themselves undergo fission; and the newly created points
go to form two whole and complete lines, L* and L**. As a result we end up with
two fully closed objects, rather than the odd combination of one closed and one
partially open object, as shown in Figure 17.10.
10
In a more measured but perhaps not entirely dissimilar vein, the various forms
of non-standard analysis developed by Robinson and others since 1960, have
successfully rehabilitated the infinitesimal. Mockingly dubbed ghosts of departed
ZENO AND THE CONTINUUM II 307
quantities by Berkeley, infinitesimals quantities with the dubious-seeming prop-
erty of being larger than zero but less than any ordinary finite quantity were for a
long time indispensable tools in calculus, but were rendered redundant by Cauchy
and Weierstrasss work on limits in the nineteenth century. Thanks to the work
of Robinson and his followers, mathematicians are now perfectly happy to use
infinitesimals in various branches of their subject. Figure 17.11 shows the hyper-
real number line of non-standard analysis. The entirety of the real line is included
in the finite numbers; to the left are the negative infinite numbers (i.e. numbers
smaller than any real number) and to the right are the infinite positive numbers
(i.e. numbers bigger than any real number). Above the horizontal is a blown-up
(massively magnified) view of the region in the immediate neighbourhood of 0.
As can be seen, 0 is surrounded by a cluster (or monad) of infinitesimals the
latter are sufficiently small to fit between any of the ordinary reals.
Figure 17.10 Peirces continuum.
Cutting a geometrical object in half along
a line L creates two fully closed objects.
infinite positive
numbers
infinite negative
numbers
0
The region around 0, highly magnified
( =infinitesimal quantity)
0 2 2
Figure 17.11 The hyperreal
(non-standard) line.
L
L* L**
nite numbers
(including real line)
McLaughlin and Miller (1992; see also the more accessible McLaughlin 1994)
have argued that non-standard analysis, in the particular form of Nelsons inter-
nal set theory (IST), can be used to resolve some of Zenos paradoxes in a more
intuitively compelling way than would otherwise be possible. Consider first the
Dichotomy. To reach the finishing line, Zeno has to complete all the Z-runs, but
there is an infinite number of these; how is it possible to complete an infinite
sequence of distinct tasks? In rough outline, non-standard analysis of the IST
variety is supposed to help as follows. Call the parts of Zenos journey that have
coordinates that can be expressed in real numbers the R*-points, and those that
require coordinates expressed in non-standard (or infinitesimal) terms the I*-
points. Despite this additional complexity, from the point of view afforded by IST,
308 TIME AND SPACE
Zenos journey taken as a whole can be viewed as consisting of a finite sequence of
steps. McLaughlin and Miller further argue that although Zeno passes through all
the I*-points, these portions of his journey have no empirical significance: our best
measuring techniques will only ever be able to discern his position on R*-points.
So as Zeno approaches ever closer to the finishing line, he will eventually move
beyond the last R*-point and into the infinitesimal realm, and it is from here that
he will actually arrive at, and cross on to the finishing line itself. As for the way
Zeno moves through this infinitesimal region, there are many possibilities:
Might the process of motion inside one of these intervals be a uniform
advance across the interval or an instantaneous jump from one end to the
other? Or could motion comprise a series of intermediate steps, or else a
process outside of time and space altogether? The possibilities are infinite,
and none can be verified or ruled out since an infinitesimal interval can never
be measured. (McLaughlin 1994: 18)
The Arrow is dealt with in a similar manner. It is true that Zenos flying arrow
is motionless or frozen at all the R*-points, and these are the only points at
which we can ever hope to observe this (or any other) moving object. However,
the arrow also passes through the I*-points along its trajectory. Consequently, we
are at liberty to regard the arrow as being in motion over the infinitesimal regions
surrounding each of its R*-points, even though this motion cannot be empirically
detected, and its precise character cannot be determined.
These resolutions are intriguing, but they also give rise to plenty of questions.
Joseph Alper and Mark Bridger point out that although McLaughlin and Millers
use of finite does correspond with Nelsons use in IST, in the present context it is
misleading: a finite set (in the sense they intend) includes all the standard reals plus
much else (1997: 1514)! This aside, have we really been provided with an intui-
tively clear picture of how Zeno arrives at his finishing line? This is certainly ques-
tionable: since McLaughlin and Millers solution leaves parts of Zenos journey
indeterminate, arguably one mystery has merely been traded for another; we shall
be returning to this shortly. A further (and related) concern with what is being pro-
posed is their overt reliance on a form of verificationism: The tactic of appealing
to infinite sets to argue against the possibility of motion has been denied to Zeno
by showing that such sets always contain references to nonverifiable occurrences
(McLaughlin & Miller 1992: 383). Is it really true, as McLaughlin asserts that we
are not responsible for explaining situations we cannot observe (2003: 20)? This
is not obviously true; after all, particle physicists and cosmologists spend much of
their time trying to uncover the secrets of regions that are not directly observable.
In fact, it is not entirely clear what brand of verificationism McLaughlin and
Miller are endorsing. There are two main options:
(A) The movements of objects through the infinitesimal regions of space are in
principle empirically inscrutable, and for this reason there is no fact of the
matter as to what these movements are.
(B) The movements of objects through the infinitesimal regions of space are in
principle empirically inscrutable, but there is nonetheless a fact of the matter
as to what these movements are.
ZENO AND THE CONTINUUM II 309
Each of these options is problematic, albeit for different reasons. Option A
introduces a genuine ontological indeterminacy into the world, a step that should
not be taken lightly. Quantum theory also introduces indeterminacy into the
micro-realm, but this indeterminacy was accepted as a reality only after exhaustive
investigations failed to reveal any alternative. The indeterminacy McLaughlin and
Miller want to introduce does not fall into this category; it arises only if we opt
to follow their recommendation to reject the standard model of the continuum in
favour of their non-standard alternative. It is by no means clear, to put it mildly,
that the proposed resolution of Zenos motion paradoxes is sufficiently superior
to the orthodox alternatives to merit taking this dramatic step. Option B does not
introduce ontological indeterminacy, and for this reason it is less objectionable
than A. However, anyone opting for B is vulnerable to the charge that Zenos
worries are merely being swept under the carpet. For we are now to suppose that
moving objects pursue a quite definite trajectory through the non-standard points
that are to be found in the infinitesimal regions of space, even though we cannot
find out what these trajectories are. Since some of these trajectories (presumably)
involve traversing infinite series of non-standard points, we have not circumvented
the Dichotomy puzzle: how is it possible to complete an infinite series of move-
ments? Since the trajectory of a moving object will consist in its being located, in
static fashion, on a series of non-standard points, the Arrow is also alive and well.
In this context, McLaughlins suggestion that we are under no obligation to try
to explain what we cannot observe is surely problematic; we can observe motion,
even if only along paths formed from R*-points. Under option B the motion we
can observe is possible only because objects also move along perfectly determi-
nate although unobservable paths through the infinitesimal regions of space.
If observed motion depends in this way on unobserved motion, are we not under
an obligation to understand how the latter is possible?
11
17.7 Extension as fundamental
If we want a model of the continuum that best accords with our intuitive concep-
tion of what a continuum is truly like a conception no doubt rooted in part in the
smooth homogeneities we find in our sensory experience the more promising (at
least from a metaphysical point of view) are the attempts to go pointless, that is,
devise a physical geometry that makes no appeal to points whatsoever, at least at
the fundamental level. According to Peirce: A continuum (such as time and space
actually are) is defined as something any part of which however small itself has
parts of the same kind. Every part of a surface is a surface, and every part of a line
is a line (quoted in Moore 2007b: 429).
12
This characterization entails that space
or time (or any other true continuum) is infinitely divisible in a distinctive way:
take any part you like, however tiny, and it will always be composed of further
and smaller extended parts, and so on and on, without end. Since the orthodox
continuum modelled on the real line is also infinitely divisible, it may not be
immediately clear where the difference lies, but a difference there is. The orthodox
continuum (as standardly conceived) possesses parts that do not themselves pos-
sess parts, and parts that are not themselves extended; these are none other than
310 TIME AND SPACE
the dimensionless points that have been the cause of so much concern. The sort
of continuum implicit in Peirces characterization is very different. Since all of its
parts, without exception, can be divided into ever-smaller extended parts, this sort
of continuum does not include any points, as traditionally conceived, just smaller
and smaller finite segments. Or, as Weyl puts it: that it has parts, is a fundamental
property of the continuum it is inherent in the nature of the continuum that
every part of it can be further divided without limitation (1998: 115).
From a metaphysical perspective, the move we are now considering consists
in taking extended entities as fundamental; in no sense is an extended thing to be
regarded as composed of more basic parts that are themselves entirely lacking in
extension. In the case of a two-dimensional line, the basic entity is an extended
line interval or segment; in the case of a three-dimensional space it is an extended
three-dimensional spatial volume. Within this framework we can still talk of points
and lines, but only as more or less useful abstractions (or fictions) that we are
imposing on a fundamentally extended reality; the latter does not possess parts
that correspond to either points or lines. Like its more familiar Cantorian counter-
part, the essentially extended continuum possesses an infinite degree of complex-
ity: since every part of it contains a smaller part, without end, each of its parts
(no matter how small they might be) is infinitely divisible. Indeed, by virtue of the
fact that the essentially extended continuum lacks a bottom-level of indivisible
point-parts, it might well be regarded as being more thoroughly divisible than its
orthodox counterpart.
Conceiving of the continuum in this sort of way brings one significant advantage
directly: it successfully circumvents Zenos Plurality paradox. If space and time are
not composed of dimensionless points, the issue of how any number of such points
could combine to compose a finite extension simply does not arise. It also gives rise
to a conception of the structure of substantival space that is metaphysically very
simple, and arguably superior in some respects to both the orthodox point-
filled continuum and the discrete or quantized alternative. Proponents of the latter
approaches are confronted with a potentially awkward question. For a collection
of points (or spatial atoms) to constitute a single substantival space, it is not enough
for them simply to exist they might exist in separate worlds, at no distance from
one another they must be unified in some way, unified so as to constitute a single
spatial substance. To put it another way, there must be some unifying connection
between the basic spatial ingredients, a connection that serves and suffices to unite
them into a single extended spatial entity. Perhaps there is nothing objectionable
in positing a sui generis relation that would do the job required, but so doing is
at the very least ontologically expensive. On the essentially extended conception,
there is no need for the additional relation. There is no need for a relation to bind
zero-dimensional points into single substances, since (by hypothesis) there are no
such points. There are, however, small extended regions, as on the discrete model,
but these regions are not indivisible or exclusive in the manner of spatial atoms.
Hence, as shown in Figure 17.12, regions of essentially extended space can: (i)
contain other regions nested within them; (ii) be wholly contained within larger
spatial (or spatio-temporal) regions; and (iii) be connected to one another by one
or more series of partially overlapping regions. In effect, a space (or spacetime) of
this kind is self-unifying.
ZENO AND THE CONTINUUM II 311
In recent parlance, an entity structured in the way now being envisaged is said
to be composed of atomless gunk, or just plain gunk. This usage derives from Lewis
(1991), who used the term to characterize infinitely divisible but atomless matter.
However, the term is now used in connection with space, time and spacetime; in
the literature it is not uncommon to find the pointy conception of space being
contrasted with the gunky view of space (cf. Arntzenius 2008). The terminol-
ogy may be newly minted, but the view itself has a long history: arguably Aristotle
subscribed to a view along these line; Descartes may well have viewed space (and
hence matter) as gunk-like; likewise Kant, and also Peirce (on some occasions).
In the twentieth century the essentially extended gunk-like view was championed
most notably by Whitehead (1919, 1920, and 1929, where a different approach is
taken). Whitehead devised his method of extensive abstraction to define planes,
lines and points in terms of extended volumes. For example, a point can be defined
in terms of an infinitely converging set of nested spherical regions (see Figure
17.13). Broad was immediately convinced of the merits and promise of White-
heads point-free approach, and described how it could be applied to space (1923:
ch. 1); Tarski (1956) went on to develop it axiomatically. Although Whiteheads
systems suffered from a number of flaws, mathematically speaking, subsequent
work has succeeded in ironing them out, albeit in different ways: see Grzegorczyk
(1960), Clarke (1985), Gerla (1990), Forrest (1996a), Roeper (1997), Biacino &
Gerla (1996).

?
? ?
? ?
?
?
Discrete space
Figure 17.12 What unifies a space? In discrete and point-constituted spaces the question arises as
to what unifies the basic spatial parts so as to form a single unified space. This question is easily and
economically answered in the essentially extended framework; in this kind of space (or spacetime) regions
can be nested within one another, or partially overlapping, in the manner shown on the right. More distant
regions in a single space are always connected by a series of partially overlapping regions, and always
contained within the maximal region that consists of the sum of all sub-regions.

Essenally extended space
Figure 17.13 Points and lines
defined. In essentially extended
gunky space, points and lines
do not exist, but can be defined in
terms of nested series of areas and
volumes (from Broad 1923: 47).
312 TIME AND SPACE
As interest in this point-free conception of continua has grown, new problems
have been uncovered. Intriguingly, it turns out that gunk also gives rise to measure-
theoretic paradoxes (see Forrest 1996b; Hawthorne & Weatherson 2004; Arn-
tzenius 2008); for a survey of escape routes see Sanford Russell (2008). Although
all forms of gunk are such that any of their parts has an infinite number of smaller
parts, it may be that some forms of gunk are more finely divisible than others:
Nolan (2004) suggests hypergunk may be possible.
13
There is also the issue of
how to reconcile gunk-like space and time with the orthodox point-based math-
ematics used in the major theories of physics. One option is to follow Whitehead
in defining points and lines in terms of extended volumes, and then deploy these
in much the usual way in physical theorizing. Although adopting this option has
many advantages, it is also vulnerable to the objection that we havent really suc-
ceeded in eliminating points from our ontology. If points continue to play such an
important role in our most fundamental theories, isnt there a case for holding that
they actually exist? For a survey of alternative and more radical approaches to
this issue see Arntzenius & Hawthorne (2005).
So far as Zenos paradoxes are concerned, for the reasons already mentioned,
if space and time are gunk-like, the paradox of Plurality is no more. The problem
posed by the Arrow is also much diminished. In the absence of points, space and
time always come in finite intervals, however small. In this context it no longer
makes sense to conceive of the moving arrow being located at a particular point
at a particular instant of time; rather, the arrow can only be thought of as existing
over finite intervals however brief over the course of which it is continuously
changing its spatial location. The stroboscopic character of motion through
the orthodox pointy continuum is thus avoided. The situation with regard to
the Dichotomy appears to be more mixed. Since Zenos route to the finishing
line will consist of a series of overlapping intervals of finite duration, there is
obviously no difficulty in understanding how he completes the race, no matter
how short these overlapping intervals happen to be. But since gunk-like space
is also infinitely divisible, Zeno will still have to complete an infinite number of
sub-journeys to reach the terminus of any one of these sub-intervals. For those
who find this unacceptable, one alternative is the discrete model we encountered
in 17.3. However, this conception of the continuum also faces difficulties, not
least of which is the unity problem we have just encountered. Another option
would be to remain within the confines of the essentially extended framework,
but to find a way of introducing into it a discrete or quantized element, in such a
way that Zeno can reach his destination after only a finite number of steps. The
fuzzy realist model of the temporal continuum outlined in Dummett (2000) is
one way forward; no doubt there are other possibilities to be explored.
14
Also, as
we saw in 17.4, when it comes to discreteness, developments in physics may have
an important role to play.
As of now only one thing is clear. It may well be that we will never arrive at
a conception of the continuum that offers all that we might want and expect.
Nonetheless, since there are several avenues worthy of further exploration, and
no doubt others that have yet to be discovered, there is every reason to persevere
with the search.
18 Special relativity
18.1 Time, space and Einstein
Our investigations up to this point have not been in vain they have certainly led
us to a deeper understanding of the complexities of the issues we are concerned
with but they have also been inconclusive. Very different accounts of the large-
scale structure of time seem both metaphysically viable and compatible with the
character of our everyday experience. Space has proved an equally stubborn topic.
Substantivalism may be ahead on points in certain respects, but there have been
few clear- cut victories. But for those looking for definite answers all is not yet
lost. We have yet to consider the impact of more recent scientific developments,
and some believe that answers to the questions we have been considering can be
found here. Of course, these answers will only concern the space and time of our
world, rather than space and time in general, but being so restricted would scarcely
reduce their interest.
However, although contemporary science has a great deal to teach us, anyone
looking for definitive answers to our questions will be disappointed. Our current
best theory of the very small quantum theory and our current best theory of the
very large Einsteins general theory of relativity (GTR) have yet to be reconciled,
and there is no consensus among the physicists working in these areas as to what
the overall character of theory that synthesizes the two, quantum gravity, will
be. Some believe that the successor theory will retain the most distinctive features
of both relativity and quantum theory; others believe that at least some of these
features will disappear. The waters are further muddied by the fact that there
are continuing disputes over the correct interpretation of relativity and quantum
theory themselves. As a consequence, much of what we can learn about our space
and time from current physics can only have a provisional status. This said, the
results that can be gleaned from physics are vitally important, for, no matter how
unreliable they may prove to be, there is at present no better guide to the real
nature of our own time and space. They are also fascinating in their own right. The
advances in physics made in the past century are such that one thing seems almost
certain: time and space are stranger than Newton ever imagined.
That we can know this is in large part due to Einsteins special theory of rela-
tivity (STR), which we will be looking at in this chapter and the next. Among the
counterintuitive consequences of STR, according to its standard interpretation at
least, are the following:
314 TIME AND SPACE
1. Common sense says that the rate at which time passes, as measured by clocks,
is the same for everyone and everything, irrespective of whether they are in
motion relative to one another. Consider two twins, Alice and Bob; if Alice
embarks on a long journey and returns, she will still be precisely the same age
as Bob. According to STR this is wrong: Alice will be younger than stay- at-
home Bob in effect, time has passed at a slower rate for her than for Bob.
2. Common sense says that temporal and spatial intervals are independent of
one another. If two events are separated by a spatial distance d and a temporal
distance t, then this is just a fact about how things are. According to STR this
is wrong: these events will be separated by different spatial and temporal
intervals in different frames of reference. Time and space thus lose their inde-
pendence from each other, and we are left with a notion of Einsteinian (or
Minkowski) spacetime, a unitary entity that decomposes into different spatial
and temporal intervals for different reference frames.
3. Common sense says that there is a fact of the matter about which events
happen at the same time as other events, irrespective of their spatial separa-
tion, and hence that simultaneity is an objective feature of the universe. This
belief also falls victim to STR: as a consequence of 2, whether or not two
events are simultaneous depends on the reference frame from which they are
considered; observers in different reference frames will find different events
simultaneous, and there is no sense in the idea that one observer is right and
the other wrong.
There is no denying the strangeness of these ideas, so it is important to under-
stand the grounds for accepting the theory that yields them, and I will provide
some of the necessary background. But I will also be considering the metaphysical
implications of the theory. The debate between substantivalists and relationists
takes a new turn within the context of STR, but since I will be focusing on this
debate when discussing GTR (in Chapters 20 and 21) I will postpone dealing with
it until then. More intriguing is the impact of STR on the debate about the nature
of time, which is commonly thought to be dramatic: influential philosophers have
argued that STRs relativization of simultaneity (which is preserved in GTR) estab-
lishes beyond any reasonable doubt the Block view of time. If they are right, STR
provides the solution to one of our two fundamental problems; I will be examining
this issue in Chapter 19. But before entering these murky waters, the basics of STR
itself must be laid bare.
18.2 Lightspeed
Einsteins special theory of relativity may have many bizarre consequences, but
they all flow from just two simple principles:
The Relativity Postulate: the laws of nature are the same in all inertial frames
of reference, so there is no experiment that can reveal whether one is at rest
or in uniform motion.
The Light Postulate: the speed of light (in a vacuum) is a constant, c, which is
independent of its source and the same in all inertial frames of reference.
SPECIAL RELATIVITY 315
Quite why these seemingly innocuous principles have such momentous conse-
quen ces may not be immediately obvious, but the reasoning involved is not par-
ticularly complex, as we shall see. The difficult step, which Einstein alone took,
was to take seriously the possibility that these principles might be true, for at the
time (STR first appeared in print in 1905), this was far from obvious.
The Relativity Postulate is a strong formulation of our old friend the principle
of Galilean equivalence. As Leibniz enjoyed pointing out, although Newtonian
physics postulates an absolute space, there is no empirical method of detecting
absolute velocity, and this fact had continued to trouble physicists over the inter-
vening period. But as the nineteenth century drew to a close it seemed that this
problem might finally be solved: a way of measuring absolute velocity had finally
emerged. The key lay in the behaviour of light.
In the 1860s James Clerk Maxwell published a work on electrical and mag-
netic phenomena that synthesized the sizeable quantity of experimental work
done over the previous centuries (and especially that done by Faraday earlier in
the same century). His theory is embodied in four equations, which accounted
for all the (then) observable behaviour of electrical and magnetic phenomena,
and showed how electrical and magnetic forces are intimately interrelated. In
particular, he demonstrated (theo retically) how electrical and magnetic fields
could combine to form self- propagating electromagnetic waves. He could predict
the velocity of these waves from his equa tions, which turned out to be the speed
of light (which had already been discovered experimentally). It seemed improb-
able that this was just a coincidence, so he proposed that light itself was an elec-
tromagnetic wave, which duly turned out to be the case. Since waves can vary
in frequency and wavelength, Maxwell also predicted that there would be other
kinds of electromagnetic wave, and was again proved right: at the low frequency
end of the spectrum are radio waves, whereas at the higher end X- rays and
gamma radiation are to be found.
Physicists had been studying waves prior to the 1860s, and since all known
waves were perturbations on or in something water, air, the surface of drums,
and so on everyone assumed that light waves also consisted of perturbations in
a medium. But there was a problem. Light can pass through a jar from which all
trace of gas has been removed, so the required medium was clearly of a very special
sort: it completely permeates seemingly empty space even the space between
the stars. Moreover, since light also passes through the solid glass of the jar itself,
its medium extends through solid objects, as well as seemingly empty space. This
invisible but all- pervasive medium was given a name: the aether.
Having posited this remarkable medium, questions soon arose about its proper-
ties: aether- physics was born. Since planets and other bodies are not noticeably
retarded by anything when they move in a vacuum, the aether doesnt create
any frictional drag; since light seems to travel in straight lines, even in the close
proximity of large moving bodies, these bodies cannot produce any distortion in
the aether, which must therefore be very rigid. One question was of particular
significance: what is the aethers state of motion? After finding reasons for ruling
out various possibilities it was generally agreed that the aether must be in a state of
absolute rest (some took a further step and argued that the aether and Newtonian
absolute space were one and the same entity). This consensus was to prove fruitful,
316 TIME AND SPACE
for, as was soon realized, an experimental test for distinguishing absolute motion
from absolute rest was now possible.
At this point some basic properties of waves (as understood in Newtonian
physics) become relevant. The velocity of a projectile depends on the velocity of
the object that expels it. Suppose that shells fired from a jet fighters wing cannon
move at 2000 km

h
1
when the plane is stationary. If the same shells were to be fired
when the plane is travelling at 1000 km

h
1
, their velocity would be 3000 km

h
1
.
Waves do not behave like this; their velocities vary depending on the medium (e.g.
light is slowed by glass), but they are not affected by how fast their sources are
moving. The sound waves created by a plane that flies overhead move at the same
speed of 1200 km

h
1
in all directions; they do not move faster in the direction
the plane is flying in. So if the plane is flying at 1000 km

h
1
, and the pilot has an
onboard device that detects and displays the sound waves radiating from the plane,
he will see the waves directly in front of him moving forward at only 200 km

h
1
,
whereas in the opposite direction the waves are speeding away at 2200 km

h
1
.
Someone with a similar device on the Earth, directly below the pilot, will see the
waves travelling at the same speed in each direction.
In a precisely similar fashion, Maxwells equations predict that light rays
will travel from their point of origin in all directions at the same speed:
c = 300

000

km

s
1
. If we set about measuring the speed of the light waves emitted by
a particular source, we will only measure them as travelling at that speed if we are at
rest relative to the aether when we conduct the tests; if we are moving through the
aether, we will find that the waves moving in the same direction as ourselves have a
slower speed than those moving in the opposite direction. In 1881 a young Ameri-
can physicist, Albert Michel son, set about testing this prediction using very precise
equipment that he had invented himself. Assuming that the Earth moves through
the aether as it orbits the Sun, it must have different states of motion relative to the
aether at different times of the year. Michelsons equipment measured the time it
took for light to travel along two paths of equal distance in a laboratory; one of these
paths was aligned in the direction of the Earths motion around the Sun, the other
at right angles to this direction. If, when the experiment is conducted, the Earth is
moving relative to the aether, the time taken for the two trips would be different, or
so theory predicted (just as the pilot in the plane gets a different speed reading for
sound waves moving in the same direction as him than for waves moving upwards,
downwards or rearwards). Michelson repeated the experiment at different times of
the year but found no evidence whatsoever for any variations in light speed. Light,
it seemed, travels at exactly the same speed in all directions, irrespective of the
direction of the Earths motion through the aether. This was an astonishing result,
and a worrying one since it threatened accepted theory, so in 1887 Michelson, now
working with Edward Morley, repeated the experiments to a much higher degree
of precision. Again, no variations were detected, even though the equipment was
sensitive enough to detect variations a hundred times smaller than those predicted
to occur. All subsequent tests have confirmed the MichelsonMorley result.
The finding that light always travels at the same speed irrespective of who
or what is observing it, and how fast they are moving relative to its source, is as
remarkable now as it was unexpected then, for it conflicts with some of our most
basic intuitions about motion. If a car is moving towards you on collision course,
SPECIAL RELATIVITY 317
you can reduce the speed at which it will hit you by running away from it, or
increase the speed by running towards it. The Michelson Morley experiments
show that light doesnt conform to this principle.
18.3 Compensation or revolution?
This result could not be explained by classical physics, but suggestions as to how
to accommodate it were quickly forthcoming, the most notable of which was the
compensatory theory proposed by Fitzgerald and developed in detail by Lorentz.
The general idea is that when you are moving in the same direction as a ray of light,
the light actually does overtake you at a slower speed than it would if you were
motionless or moving in the opposite direction, but the decrease is not detected
because of changes caused by your motion through the aether: your measuring
instruments contract in length and your clocks slow down. Lorentz provided equa-
tions that show just how much contraction and time dilation (slowing) is needed
to conceal the changes in the speed of light in different circumstances.
The solution that Einstein proposed also involves length contractions and time
dilation indeed, he made use of Lorentzs own equations but in one crucial
respect it was far simpler. The compensatory theory says that a given ray of light
has different speeds for different observers (in motion relative to it), but nature
conspires to conceal these differences. In a bold stroke Einstein did away with the
need for a conspiracy. He reasoned that, if nature makes it impossible for us to
discover our state of motion with respect to the aether, perhaps we are wrong in
supposing that the aether exists. Why not take nature to be as it presents itself? In
rejecting the aether he also rejected the notions of absolute motion and rest, and
took all inertial frames to be equivalent. He then accepted the consequence: a ray
of light really does travel at the same speed relative to all observers, irrespective of
their state of motion relative to it.
To get some impression of what this can mean in practice, suppose that Alice
and Bob are in outer space. Alice points her torch in Bobs direction and switches
it on. She sees the light ray moving at 300

000

km

s
1
towards Bob. Now Bob, as
it happens, is moving away from Alice at the very high speed of 150

000

km

s
1
,
but despite this, as he measures the speed at which the light passes him using a
measuring rod and a clock, he finds that it is moving at 300

000

km

s
1
. How can
this possibly be? Einsteins explanation runs thus. When Alice observes Bob she
will see that his measuring rod has contracted in length and his clock is running
slow (compared with her clock). Since speed is distance travelled divided by time
taken, Bobs readings are now comprehensible: given his deviant equipment, it is
to be expected that his readings will differ from Alices. But this isnt quite the
end of the story. Given the Relativity Postulate and the abandonment of absolute
motion, Bob is perfectly entitled to regard himself as at rest and Alice to be mov-
ing at a speed of 150

000

km

s
1
. From Bobs standpoint it is his readings that are
correct and Alices that are deviant, since when he observers Alice he will see that
her measuring rod has contracted and her clock is running slow. Since (according
to Einstein) both perspectives are equally valid, there is no objective fact of the
matter as to who is right. And of course, there are many other perspectives on the
318 TIME AND SPACE
same facts those of potential observers in different inertial reference frames, all
of whom are entitled to regard themselves as at rest, and all of whom will find
Alice and Bob to have different states of motion relative to themselves.
The distinctive innovation of STR is the replacement of one absolute for another.
Newtons absolute space and time are both rejected, and a new absolute or invariant
quantity is introduced; the velocity of light. Since velocity involves both distance
and time, the only way for a given light ray to have the same velocity in frames of
reference that are in motion relative to one another is for distances and times to
vary in systematic ways in the relevant frames, and the Lorentz equations allow us to
calculate the precise differences. Hence the recognition of the new invariant required
the rejection of the old: according to Newton the distances and times between
objects are both absolute quantities, and so invariant across frames of reference.
Time dilation and length contraction initially strike us as utterly bizarre phe-
nomena because they are not observed to occur in ordinary life. When travel ling
on a moving train, we can make the phone call we agreed to make at 2 p.m.
without using the Lorentz equations to work out how much train- time differs
from land- time. A fast moving train does not look contracted or scrunched
up to passers- by, but this is because the relativistic effects only become significant
at enormously high speeds. The time dilation and length contraction of a train
moving at 150

km

h
1
will be minute about one part per hundred trillion. Even a
jet travelling at twice the speed of sound will only contract by less than the width
of an atom. Compared with the speed of light these speeds are snail- like. If a jet
could be photographed moving at 85 per cent of light speed, it would have half its
normal length, and time on board (as measured by change-registering systems such
as clocks or people) would pass at half speed compared with time on the ground.
Beings who normally travelled at near- light speeds would be all too familiar with
the relativistic effects we find bizarre.
The fact that we live in slow motion (comparatively speaking) does not mean
that it is impossible for us to find confirmation of the predicted relativistic effects.
Cosmic radiation striking our atmosphere produces showers of muons a species
of sub- atomic particle which speed towards the Earths surface. Even though the
muons are travelling very quickly (at about 95 per cent of c), they are so short- lived
(typically a millionth of a second) that they should perish long before completing
their journey to the surface, but they dont. In virtue of their great speed, the
length of their journey is shorter and their internal clock, as it were, runs slower,
so the trip only occupies a small portion of their lifespan. On a more mundane
level, very accurate atomic clocks can measure the time dilation produced by trans-
atlantic flights. The experiment has been carried out repeatedly, and the results
match the predictions of STR.
1
When Einstein worked out some of the implications of STR for other physical
quantities he discovered that as the velocity of an object increases so does its iner-
tial mass, with the consequence that as the velocity of an object approaches that of
light the amount of energy needed to move it tends to infinity, which means that
no moving material body can ever be accelerated up to the speed of light.
Surprisingly, the implications of this for long- distance space travel are not as
disastrous as one might think: it does not mean that stars hundreds of light- years
away cannot be reached within the lifetime of a normal (unfrozen) human being.
SPECIAL RELATIVITY 319
For a ship moving at near- light speed, spatial distances would shrink and (speaking
loosely) time would slow to such an extent that a trip of a million light- years could
take only a decade or so of ship- time.
You might think: If it is possible for a spaceship to traverse a distance it would
take light a million years to cross in only ten years, doesnt this mean the ship must
be travelling faster than light, which STR tells us is impossible? But dont forget
that there are two perspectives on the lights voyage. From our point of view on
Earth, the light crosses vast tracts of space in a million years, but if per impossible
the light ray could carry a clock and a ruler, it would be found that from its point
of view the trip takes no time whatsoever, and involves crossing no distance what-
soever; for a photon the entire universe shrinks to zero size and time stops. Hence,
from the perspectives of the ship and the light ray, the ship takes longer to make
the trip than the light. From the perspective of the Earths frame of reference, if
the light takes a million years to complete the journey, the ship will take millions
of years more, and the return journey will be just as long.
Unfortunately, while STR does open up the possibility of travelling to even the
most distant parts of the universe within a single human lifetime (as measured
by ship- time), significant obstacles of a practical sort remain. Perhaps the most
intractable of these is the problem of attaining speeds approaching c. This requires
acceleration, and acceleration produces inertial forces; the acceleration needed
to boost a spaceship up to a significant proportion of c within a human lifespan
would be far in excess of those that are survivable.
18.4 Simultaneity
So far I have said little about the aspect of STR that has the greatest (potential) sig-
nificance for our understanding of time: the relativity of simultaneity. Because this
topic is of especial importance I will deal with it more carefully, and rehearse the
argument Einstein himself used to undermine the Newtonian (and neo- Newtonian)
assumption that there is an absolute fact of the matter as to which events occur at
the same time.
2
Einstein begins by asking how we measure the time interval between events.
For events that happen together at the same time at the same place, there is no
problem. But what about events that are separated by some spatial distance, say
at two points X and Y several miles apart? To make matters more concrete, let us
suppose that X and Y are each struck by a bolt of lightning; how do we go about
finding out if they are hit at the same time?
If we are somewhere between X and Y, we will have to rely on signals coming
from X and Y, for example the observable flashes produced when the bolts hit the
ground. Provided we know the speed of light (which we do), and have previously
measured out the distance between ourselves, X and Y, we can easily calculate
when the bolts struck. For example, if we are situated at the mid- point between X
and Y, and the light reaches us simultaneously, then we know that the bolts struck
at the same time. This procedure may seem rather involved and roundabout, but
Einstein points out that this method (or a variant of it) is unavoidable. Since we do
not have direct instantaneous access to what is going on in other parts of space, we
320 TIME AND SPACE
have no option but to use signals that take some finite time to cross the intervening
space, but as soon as we do, if we want to know exactly when some distant event
occurred, calculations are unavoidable.
To see this it suffices to consider an alternative scenario: suppose that we have
stationed observers at X and Y, and equipped them with walkie- talkie radios and
clocks. We wait until after the bolts have struck and give them a call. Both observ-
ers say the same: the bolts struck at exactly 12 noon. Does this mean the bolts hit
simultaneously? Only if the observers clocks are synchronized. How can we go
about checking this? One obvious solution is to call them both up on their radios,
and say something along the lines of Set your clocks at 12.00 now. But of course
this is our original scenario all over again: given that radio waves travel at the same
speed in all directions, they will only receive the message at the same time if we
send it from midway between them.
Einsteins method does work, and can be used to establish a common time
system for a group of people who are stationary with respect to one another; eve-
ryone in this group will agree on which events are simultaneous and which are not.
But as Einstein then went on to show, the situation is otherwise for people who
are moving with respect to one another: it is impossible for such people to agree
on a common time system. He used an example with trains to make the point.
We are to consider a train running along a straight track of indefinite length
at very high speed, say 0.6c, or 60 per cent of the speed of light. You are by the
side of the track, and just as the mid- point of the train passes you lightning strikes
both ends of the train, producing two flashes, which you see simultaneously. You
reason as follows:
I was standing opposite the mid- point of the train when I saw the flashes. The
light from each end of the train had the same distance to travel. The speed
of light is absolute it travels at the same speed in all directions no matter
what your state of motion so the lightning must have struck both ends of
the train at the same time.
However, an observer, O, sitting on the roof at the centre of the train witnesses a
quite different sequence of events. Since O is moving towards the light travelling
from the front of the train, he decreases the distance it has to travel, and so this
light ray reaches him before the ray travelling from the rear of the train, which has
a longer distance to cover. O reasons thus:
I saw two flashes occur one after the other, the earlier at the front of the
train, the later at the rear. These flashes could not have occurred simultane-
ously. Since the speed of a light ray is the same irrespective of how you are
moving, and the light from both ends had the same distance to cover (since
I was sitting at the mid- point of the train), if the flashes had occurred at the
same time I would have seen them at the same time, but I didnt. Clearly, the
lightning hit the front of the train before it hit the rear.
Os reasoning seems impeccable, and so (from his perspective) he is quite right to
conclude that the two bolts did not strike simultaneously.
Now suppose you see a third lightning bolt strike the front of the train shortly
after the first two strikes. Again, O sees something quite different: the light from
SPECIAL RELATIVITY 321
the second bolt to strike the front of the train reaches him at the same time as the
light from the bolt that struck the rear of the train. So he reaches the conclusion
that these two bolts are simultaneous.
What we have here are three events, 1, 2 and 3, whose time ordering is different
from different perspectives. Employing Einsteins method for determining simul-
taneity, you find that 1 and 2 are simultaneous, whereas 3 occurs later; applying
precisely the same method, O finds that 1 occurs first, and is then followed by 2
and 3, which are simul taneous. (It is important to note that the presence of human
observers is irrelevant; the same results would be obtained if there were no people
involved, just automatic measuring devices.) Given the Relativity Postulate the
two perspectives are equally valid; you might take yourself to be motionless, but
so might O (who would thus take the ground to be moving beneath the station-
ary train), and there is no fact of the matter as to who is right. As a consequence,
there is no fact of the matter as to whether the events in question occurred at the
same time or not.
In appreciating the symmetrical character of the situation it is useful to con-
sider it from the perspective of the person on the train; thus far we have been
describing things from the ground- based point of view. So imagine you are on the
train, and are conduct ing a simple experiment. You are sitting at the mid- point of
your carriage, holding a switch in your hand. Connected to the switch (by equal
lengths of wire) are two lamps at either end of the carriage. You flick the switch
and see light arrive from each lamp at the same time. You naturally conclude that
the lights came on simultaneously: after all, the light from each had the same
distance to travel, and the speed of light is absolute. Now suppose that at the
moment you press the switch you are directly opposite O who is sitting at the
trackside. What do you think O would observe and conclude? It is easy to work
it out. At the moment you flicked the switch O was racing away from you at 0.6c
in a rearwards direction, and so O was racing towards the light emitted by the
rear bulb and away from the light coming from the forward end of the carriage.
So he would see the rear light first, then the forward light at a later time. Since O
knows that the distance between you and the two bulbs is exactly the same, and
given the absoluteness of c, O must conclude that the bulbs didnt come on at the
same time.
The role of the Light Postulate is no less important; even if we assume in
line with the Relativity Postulate that both perspectives are equally valid, the
reasoning that leads to the conclusion that simultaneity is relative relies on the
assumption that light rays travel at the same speed with respect to all observers,
no matter how fast they are moving relative to its source. Return to the original
case. O is justified in thinking that the front bolt hits before the rear bolt because
(i) he is situated right in the middle of the train, so the light has equal distances
to travel, (ii) the light coming from the front and rear of the train is travelling at
the same speed. If O found that the light coming from the front of the train was
travelling faster than the light coming from the rear, then he wouldnt be justified
in thinking the front bolt hit first just because the light coming from it arrived
first: this is to be expected if the light coming from the front is moving faster. But
this isnt the case. If O measures the speed of the light when it reaches him, he
will find it has the same speed in both directions.
322 TIME AND SPACE
We have been looking at just one example, but the lesson generalizes. In the
context of STR, simultaneity is not absolute but frame- relative. Spatially separated
events that are simultaneous from the perspective of one inertial frame are not
simultaneous from the perspective of other inertial frames, and since the perspec-
tives of all inertial frames are equally valid, there is no sense in the idea that the
events in question are really simultaneous or not.
3
18.5 Minkowski spacetime
The new conception of space and time was clarified by Minkowski in 1908, in a
lecture whose opening words were to become famous:
The views of space and time which I wish to lay before you have sprung
from the soil of experimental physics, and therein lies their strength. The are
radical. Henceforth, space by itself, and time by itself, are doomed to fade
away into mere shadows, and only a kind of union of the two will preserve
an independent reality.
Minkowski showed that although STR renders some quantities relative and frame-
dependent, there are other quantities that are as absolute as any in Newtonian
dynam ics, it is just that these quantities are neither spatial nor temporal as we
ordinarily understand the terms. The new clarity resulted from formulating STR
as a spacetime theory. We have already encountered the general idea of spacetime,
and two different types of spacetime structure: Newtonian and neo- Newtonian.
The spacetime structure of STR is interestingly different from both.
Newtonian spacetime is rich in structure. Any two points, no matter when or
where they are located, are separated by a definite spatial and temporal interval,
and these intervals are invariant: they are the same when viewed from any inertial
frame. Newtonian spacetime also has an affine structure: there is a distinction
between curved and straight paths through the spacetime, which is also invariant.
The difference between absolute rest and absolute motion is well defined; like-
wise the distinction between uniform motion and acceleration. In neo- Newtonian
spacetime some of this structure is absent. Affine structure is retained the dis-
tinction between straight and curved paths remains intact and hence there is
a real difference between accelerated and non- accelerated motion. There is an
invariant spatial distance between points that occur at the same time, and non-
simultaneous points are separated by an invariant temporal interval. What is lost
is spatial distance between points at different times. There is consequently no
distinction between absolute rest and absolute motion, and the notion of absolute
velocity has no sense. It is, of course, for this reason that Newtonian dynamics
(in which absolute velocity plays no role) is best formulated in terms of neo-
Newtonian spacetime. However, although in STR neither spatial nor temporal
distances are invariant, the speed of light is. Since there are no invariant velocities
in neo- Newtonian spacetime, it is clear that we need a different sort of spacetime
to make sense of STR.
Minkowskis solution was to build a spacetime around the paths that light rays
can take in a vacuum. The idea that luminal trajectories are built into the very
SPECIAL RELATIVITY 323
fabric of spacetime is a natural response to the abolition of the aether, for if light
is not a disturbance within any material medium, it seems there is only one candi-
date for the structure that constrains or determines the paths light rays can take:
spacetime itself. Since all known space is flooded with electromagnetic radiation,
the luminal structure of spacetime extends through the otherwise empty regions
between stars and galaxies. For reasons that I will come to shortly, this invariant
feature of Minkowski spacetime is often called the light- cone structure.
The speed of light is not the only invariant quantity in Minkowski spacetime;
the so- called interval between points and events is the same in all inertial frames.
The spacetime interval is neither spatial nor temporal, but a mixture of both.
A purely spatial analogy is useful in grasping the general principle involved.
By Pythagoras theorem, the square of the hypoteneuse (the longest side) of a
right- angled triangle is equal to the sum of the squares of the other two sides:
z
2
= x
2
+ y
2
. As Figure 18.1 illustrates, there are many ways of drawing a right-
angled triangle with a hypoteneuse of a given length L.
In a similar fashion, a given spacetime interval is related in a systematic way to
two other distances, but here the similarities end: first, only one of these distances
is spatial, the other is temporal; secondly, to calculate the interval we do not add
these distances, we subtract them. The relevant formula for interval I is:
I
2
= d
2
c
2
t
2
where d is the spatial distance between events (given by x
2
+ y
2
+ z
2
), t is the time
separation and c (as usual) stands for the speed of light. Observers in different
inertial frames who use this formula to calculate the interval between a given pair
of events will all arrive at the same answer, even though they will disagree about
the spatial and temporal distances between the two. Since all inertial frames are
equivalent, there is no fact of the matter as to the correct decomposition of the
interval into spatial and temporal components, and so the only objective (frame
independent) fact about the events is the magnitude of the spacetime interval that
separates them.
L
m
n
L
m
n
L
m
n
interval
time
space
interval
time
space
interval
time
space
Figure 18.1 Spacetime interval. Although observers in different frames of reference will disagree
about the spatial and temporal distances between events, they will all agree on the spacetime interval
between them interval is thus an invariant quantity in Minkowski spacetime. Calculating interval is
similar to calculating the length of the hypotenuse of a right-angled triangle using Pythagoras theorem:
the square of the interval between two events is systematically related to the squares of the spatial and
temporal distances between them.
324 TIME AND SPACE
I will say a little more about the standard way of construing the luminal struc-
ture of Minkowski spacetime, a device that provides a useful aid in arriving at
a proper understanding of the feature of STR that has the greatest philosophical
significance the relativity of simultaneity. The key notion is that of a light- cone,
a choice of terminology that is apt in one sense but misleading in another. Imagine
a light source O in outer space, which is turned on for a brief time. Light spreads
from the source in all directions, and so forms a spherical surface or wave- front.
As the seconds tick away, the spherical surface expands. Depicting the expansion
of this spherical front in a single two- dimensional diagram is impossible (it can be
approximated with a sequence of pictures, each showing a slightly larger sphere).
However, if we use only two dimensions to represent space, the expanding sphere
becomes a succession of increasingly large circles, which together form the surface
of a cone. We can represent all the light arriving at the source O by the surface of
a second cone, which expands backwards into time. When this is combined with
the future- directed cone (represent ing light emitted by the source) they together
yield the characteristic egg- timer shape shown in Figure 18.2.
Note the following.
The points on the surfaces of the cones are those that can be connected by
light rays travelling in a vacuum, and are said to be light- like connected.
The spacetime interval between such events is zero. Zero interval is not like
spatial distance, since events separated by very large spatial distances can be
connected by an interval of zero.
The region of spacetime that lies outside the cones is known as absolute else-
where; the spatiotemporal separation between points in this region is such
that only a signal travelling faster than light could connect them; such points
are said to be space- like separated. Events that are very close together in
time but spatially at a considerable distance fall into this category. The interval
between space- like separated points is positive.
The points lying inside the cones constitute the regions known as absolute past
and absolute future; their spatiotemporal separation is such that they can be
connected by signals travelling slower than light, and they are said to be time-
like connected. The interval between time- like separated points is negative.
Figure 18.2 The past and future light-cones of
spacetime point O. A four-dimensional structure,
shown here with one spatial dimension suppressed.
Region F is Os absolute future, region P is Os
absolute past. The points on the surfaces of the cones
are light-like connected to O; the regions within the
cones can be connected to O by processes travelling at
sub-light speeds. For a signal originating at any point
outside the cones to reach O it would have to travel
faster than light this region is known as absolute
elsewhere.
SPECIAL RELATIVITY 325
For any space- like separated events there is an inertial frame in which they are
simultaneous, and so differ only in their spatial coordinates.
For any time- like separated events there is an inertial frame in which they
occur at the same spatial coordinates, and so differ only in their time of occur-
rence.
Assuming that no physical influence can travel faster than light, events that are
time- like connected can be causally related, but no causal relation can hold
between space- like events. It is for this reason that Minkowski spacetime is
sometimes said to embody the causal structure of the universe.
As shown in Figure 18.3, it is possible for two planets X and Y to causally interact
because they are linked by paths through spacetime that are always time- like; X
and Z, on the other hand, cannot causally interact, because all the spacetime paths
between them are space- like at least in part. For obvious reasons, causal influence
is sometimes said to thread the light- cones.
For our purposes, the most significant facts to draw from the above is that
although STR relativizes simultaneity, it does so within very strict limits: the rela-
tivization applies only to space- like separated events. Events that are time- like
separated are non- simultaneous in all frames of reference. Figure 18.4 shows the
upshot of this for a single spacetime point O (which lies where the tips of the two
cones meet). The temporal ordering of all the events lying within the cones is the
same in all frames of reference, although the temporal distances between them
are not. Three simultaneity hyperplanes are shown slicing through the region
of absolute elsewhere; each hyperplane is an extended three- dimensional region
of four- dimensional spacetime, and consists of a collection of space- like separated
events (or spacetime locations) that are simultaneous in different frames of refer-
ence centred on O. The middle (horizontal) hyperplane represents the events that
are simultaneous in Os frame of reference; the other two hyperplanes represent
events that are simultaneous in the frames of reference that are moving relative to
O for example that belong to two spaceships close to O that have different states
Figure 18.3 Threading light-cones.
The solid lines represent time-like
trajectories the sort that objects
travelling slower than light have.
The dotted line shows a space-like
trajectory only something travelling
faster than light can have one of these.
Since no time-like path links X and Z,
these two planets are causally isolated
from each other.
326 TIME AND SPACE
Figure 18.4 Simultaneity hyperplanes
falling within the space-like connected
region of absolute elsewhere.
of motion. As can be seen, each of these alternative hyperplanes contains events
that are not simultaneous in Os frame of reference. Indeed, they both contain
events that, from Os perspective, lie in the past and the future (but not Os abso-
lute past or future, i.e. the regions lying within the upper and lower cones). Only
a few of the many possible alternative simultaneity hyperplanes that are centred
on O are shown: all the events that are space- like related with O all the events
in Os absolute elsewhere are simultaneous with O from some inertial frame.
A similar state of affairs obtains at each spacetime point.
Looking at pictures like these it easy to fall into thinking that from the perspec-
tive of each spacetime point the temporal ordering of half the entire universe is
frame- relative according to STR, but this is not necessarily so: it depends on the
age, shape and size of the spacetime in question, and the location of the spacetime
point. If current estimates for the age, size and rate of expansion of our universe
are approximately correct, the bulk of spacetime may well lie in our absolute past
and future. Light- cone diagrams conceal these complications because they con-
form to the useful (but potentially misleading) convention of indicating light- like
connections with lines at forty- five degrees from the horizontal (given its role as
an absolute standard, light is taken to move one unit through space per one unit
of time).
In the context of the comparatively small distances we are mostly concerned
with in everyday life, STR does not altogether abolish the present, it merely renders
it slightly fuzzy. For events occurring in an ordinary sized room the region of
indeterminacy is minuscule a few millionths of a second. For events a couple of
hundred miles apart, it is slightly larger, of the order of a thousandth of a second. It
should also be borne in mind that, although the temporal ordering of events within
these regions is frame- relative, there is no question of this relativity extending to
causal ordering: the relativity of simultaneity applies only to events that are so
nearly simultaneous (by everyday standards) that nothing travelling slower than
light and this includes subluminal causal influences can connect them.
Maudlin suggests that A test of how well one has internalized the spirit of
Relativity is how one is tempted to answer the question: what would it be like
to travel at 99.99% the speed of light? (1994: 57). I would be flattened like a
pancake and feel time passing very slowly is a bad answer. Only slightly better is
Although I would be flattened and slowed down, I wouldnt notice because my
brain processes would be slowed down too, and since my rulers and measuring
tapes would be contracted to the same extent as my body, nothing would measur-
ably change in size. But, as Maudlin points out, the truly relativistic answer is:
SPECIAL RELATIVITY 327
right now you are travelling at 99.99% the speed of light in some perfectly
legitimate inertial frame. It is no more correct to say that you are now at
rest than that you are infinitesimally close to moving at light speed. Indeed,
except for things travelling at light speed, it makes no sense to attribute any
absolute velocity to any object. Not being invariant, sub- light velocities are
not real properties of objects. There can be no dynamical effect of travelling
near the speed of light because there is no such objective states as travelling
near the speed of light. Things dont shrink or slow down. Rather, there are
always infinitely many ways of expressing space- time intervals in terms of
distance in space and elapsed time. None of these ways, represented by the
various inertial frames, is any more valid, or less valid, than any other.
(1994: 57)
Reading this you may be tempted to think So, the curious effects predicted by
STR are nothing but illusions, of the sort to be found in a hall of distorting mir-
rors! and to an extent you would be right. Length contraction and time dilation
are certainly linked to the behaviour of light, and there is a sense in which they are
always in the eye of the beholder. But the analogy extends no further. The image
seen in a curved mirror is truly a distortion, in that someone standing next to the
mirror and looking in your direction would not see you flattened or stretched.
However, if you and I pass one another travelling at near- light speed, and I look
across and see you flattened and slowed, I am not seeing a distorted image of you.
I am seeing you as you truly are, just as someone who shares your velocity and who
doesnt see you flattened and slowed also sees you as you truly are. We naturally
think of shape and duration as invariant intrinsic properties; STR renders them
perspectival; objects and processes still have determinate shapes and durations,
but these are now frame- relative properties (only interval is frame- invariant). We
can no longer speak of the shape of an object. We can speak only of its shape in
such and such inertial frame, and likewise for events and their durations. Moreo-
ver, it would also be a mistake to suppose that the frame- relativity of shape is
nothing more than an optical effect, a matter of how a thing looks, for in this
instance appearances do not deceive: an objects apparent, frame- relative shape
is an accurate guide to the shape of a hole that the object can successfully pass
through (and the shape of the hole is itself frame- relative). Similarly, an external
clock on a fast- travelling spaceship that is seen to run slow when viewed through
telescopes on Earth is an accurate guide to the rate at which processes within the
ship are unfolding; on their return the astronauts will have aged less than those
who remained behind. All this may take some getting used to, but it is by no means
obviously incoherent.
19 Relativity and reality
19.1 Reality unconfined
We turn now from the physics of STR to its metaphysical interpretation. I want to
focus on some influential lines of argument that attempt to draw very significant
conclusions about the nature of time from the distinctive spacetime geometry of
STR. These arguments takes as their target all those dynamic models of time accord-
ing to which there is an ontological asymmetry between past, present and future.
As we have seen, there are several such models. Presentists of all persuasions are
agreed that the past and the future are unreal; the Growing Block theory of Broad
and Tooley is more liberal, and withholds reality only from the ever- shrinking
future. In both cases, the present has special ontological significance, constituting
as it does the frontier between what is real and what is not.
1
In outlining these
theories I tacitly assumed that the present extends through out the universe and is
absolute (i.e. non- perspectival, non- frame- relative). Let us call models of time that
posit an absolute and universal present, and hence an absolute and universal tide
of becoming (or annihilation) classical dynamic models, or CDMs. Given what
we have seen so far, it is obvious that there are certain tensions, to say the least,
between CDMs and STR. One of the questions we will be addressing in this chapter
is whether these tensions are fatal to CDMs. We will also be exploring the options
for non- classical dynamic models of time in the context of STR.
The mere fact that STR is a spacetime theory does not in itself mean it is a
hostile environment for CDMs. Minkowski spacetime diagrams depict worldlines
representing all stages of an objects history stretching across the page, and future
light- cones are represented in precisely the same way as past light- cones. But
while this may well give the impression that a Minkowskian world is an eternal
four-dimensional block of events, this interpretation is by no means obligatory.
Newtonian physics can be formulated as a spacetime theory. Although there is no
distinction drawn between past, present and future in Newtonian spacetime dia-
grams, it does not follow that there is no such distinction in a Newtonian world;
spacetime diagrams of the standard form may not include complete information
about what they depict. More threatening is what STR has to say about the present
itself. The most obvious challenge posed by STR to CDMs runs thus:
For all CDMs the division between what is present and non- present has onto-
logical significance. Under STR, however, the distinction between what is
present and non- present is dependent on an arbitrary choice of reference
RELATIVITY AND REALITY 329
frame. Isnt it absurd to suppose that the division between what is real and
unreal can be similarly dependent?
Consider an event E on the recently discovered planet that orbits Epsilon Eri-
dani (10.5 light- years from Earth). According to the Earths frame of reference E
lies in the present, but according to Jupiters frame of reference (let us suppose)
it lies in the future. If like CDM theorists we hold that whether an event is real
or not depends on where it stands in relation to the present, what is real relative
to one state of motion is not real relative to another. Relative to the Earth, E is
real. Relative to Jupiter it lies in the future, and so is unreal. Do we really want to
relativize reality in this way? If not, then the only obvious option is to subscribe
to the block-universe conception, and hold that all events are real, irrespective of
their spacetime location.
The suspicion that CDM theories and STR are irreconcilable can be put on a
firmer footing. Rietdijk (1966) and Putnam (1967) argue that, provided we make
certain plausible assumptions, it can be proved that the Block view is the only
model of time compatible with STR.
Let R be the relation of being real with respect to, so Rxy means that x is real
with respect to y. Let us further suppose that R is reflexive (a thing is real with
respect to itself), symmetrical (if Rxy then Ryx) and transitive (if Rxy and Ryz, then
Rxz). As already noted, in a Newtonian world it is natural to suppose that each
successive present is a three- dimensional universe- wide reality- slice. Or, in the
jargon of spacetime theories, the present is a single three- dimensional simultaneity
hyperplane of momentary (or very brief) duration.
Now consider Figure 19.1. This may look a little crowded, but to start with we
need concern ourselves with just three elements: E
1
, E
2
and E
3
, each of which is a
spacetime location. E
1
is the spacetime point at which two spaceships S
1
and S
2
that
are in motion relative to one another make a fleeting contact; two of their radio
antennae brush against one another as they pass. The pointed arrows represent
E
1
s light- cones, and P
1
is the simultaneity hyperplane of S
1
.
Now look at E
2
and E
3
, each of which is space- like related to E
1
. Since E
2
lies
on P
1
, it is simultaneous with E
1
relative to S
1
, whereas E
3
lies in the future with
P
1
P
2
E
4
E
5
E
1
E
2
E
3
Figure 19.1 STR and the status of distant times. Two spaceships, S
1
and S
2
, make fleeting contact.
P
1
is the simultaneity hyperplane of S
1
and P
2
is the simultaneity hyperplane of S
2
. An event (E
3
), which lies
in S
1
s future, is present for S
2
; an event (E
2
), which is present for S
1
, is past for S
2
.
330 TIME AND SPACE
respect to the latter. If we suppose that the Growing Block model applies, then E
1

is R- related to E
2
but not E
3
. More generally, everything up to and including P
1
is
real, but nothing later than P
1
is yet real.
But this is far from the end of the story. P
2
is the simultaneity hyperplane of
S
2
, the other ship at E
1
, and since the ships are moving relative to one another
the contents of this hyperplane (the events that those on board S
2
can justifiably
regard as being simultaneous at the time in question) will be different from those
in P
1
recall Einsteins train. In fact, as can easily be shown, P
2
will be tilted
with respect to P
1
in the way depicted in Figure 19.1. As can be seen, from the
perspective of P
2
it is E
3
that is simultaneous with E
1
, whereas E
2
occurs earlier
than either E
1
or E
3
.
This yields a significant result. Although from S
1
s perspective E
3
has yet to
occur, it no longer seems to make sense to suppose that E
3
is unreal. Since S
2
is
real with respect to S
1
(recall that the two ships are touching), and E
3
is real with
respect to S
2
, how can E
3
not be real with respect to S
1
? More formally, given the
transitivity of the is real with respect to relation R, since E
3
is R- related to E
1
(via
P
2
), and E
1
is R- related to E
2
(via P
1
) it follows that E
3
is R- related to E
2
.
We have focused on just two of the alternative simultaneity planes running
through E
1
, but since the same reasoning extends to all other locations that are space-
like related to E
1
, all spacetime locations within this region are real with respect
to the latter. Furthermore, it can be shown that the same considerations extend to
locations that are within E
1
s forward light- cone, and so in its absolute future, such
as E
5
. For any such location, there will be some frame of reference (such as the one
depicted to the right of Figure 19.1) containing a location that is R- related to both
E
1
and E
5
. Given this result, reality cannot be confined; it is a contagion that spreads
to all points future, and the Growing Block model is thus shown to be false.
Presentism is similarly afflicted, and not just because reality inexorably extends
forwards in time; it also extends backwards, and for exactly the same reasons. Sup-
pose we say that only what is present is real, and take this to mean that only what
is simultaneous with E
1
exists. From the perspective of S
1
the location E
4
lies in
the past, but E
4
also lies in the simultaneity plane of S
2
, and so is real relative to S
2
.
Since S
2
is real relative to S
1
(they are in physical contact at E
1
), then E
4
must be too.
If this reasoning is valid, dynamic models of time are in serious trouble. Putnam
goes further:
I conclude that the problem of the reality and determinateness of future
events is now solved. Moreover, it is solved by physics and not philosophy. We
have learned that we live in a four- dimensional and not a three- dimensional
world, and that space and time or, better, space- like separations and time-
like separations are just two aspects of a single four- dimensional continuum
with a peculiar metric Indeed, I do not believe that there are any longer
any philosophical problems about Time; there is only the physical problem of
determining the exact physical geometry of the four- dimensional continuum
that we inhabit. (1967: 247, original emphasis)
This is a momentous conclusion, but is the reasoning that supports it valid? And
even if it is valid, are there ways in which it can be resisted? There are, and they
fall into two camps. Some dynamists respond to the Rietdijk/Putnam argument by
RELATIVITY AND REALITY 331
rejecting STR, at least in its standard form; I will be looking at this strategy in 19.4.
Others adopt a compatibilist approach, and try to find a way of accommodating
dynamism without rejecting any of the key principles of STR. We shall start by
looking at the options open to compatibilists.
19.2 Compatibilism
A very simple and direct way to bring about the reconciliation is noted by Sklar.
The dynamist faces the problem that under STR different events are simultaneous
in different frames of reference, so if we suppose that (i) simultaneous events are
R- related to one another (where Rxy means x is real with respect to y), and (ii)
R is a transitive relation, then we cannot avoid the result that all events are equally
real. But, Sklar asks, why assume that R is transitive? Since anyone who embraces
STR will accept that simultaneity is both non- transitive and frame-relative, what
is to prevent them from taking a step further and holding that R is non- transitive
and frame-relative as well? It could even be argued that anyone who refuses to
take this extra step is failing fully to enter the relativistic spirit. To be sure, it is
odd to suppose that if x and y are real with respect to one another, and likewise
y and z, it is possible that x and z are not real relative to one another, but so is
the idea that simultaneity can fail to be transitive. Is relativizing existence any less
odd than relativizing simul taneity? Those who dismiss any connection between
temporality and existence may be inclined to think it is, but exponents of dynamic
models of time do not share this view; anyone who believes that time and existence
are intimately connected might be very tempted to suppose that if simultaneity is
frame- relative, existence must be too. Sklar concludes, The science can change
the philosophy and put the dispute in a new perspective, but it cannot resolve the
dispute in any ultimate sense (1977: 275).
While this relativization shows that STR does not in itself entail the Block con-
ception, but does so only with the benefit of a questionable metaphysical assump-
tion, it does nothing to assist us in choosing between dynamisms: it seems equally
compatible with the Growing Block model and the different versions of Present-
ism. In a later essay (1981) Sklar argues that some Presentists may find themselves
in an uncomfortable position. The Presentists in question are those who reject the
reality of both past and future on the grounds of epistemic remoteness: unlike
what is happening here and now, our knowledge of the past and future is based on
more or less unreliable inferences of one kind or another. Since the here and now
is causally isolated from the events to which it is space- like related, the epistemi-
cally motivated Presentist should surely place these events in the same category as
past and future events, and hold them all to be unreal. Presentism thus collapses
into solipsism of the most extreme sort: other than this spacetime point nothing
is real. However, while these considera tions may lead epistemically motivated
Presentists to reconsider their position, those who reject the past and future on
purely metaphysical grounds for example the conviction that when an event is
past it is utterly gone will not feel the same inclination.
Sklar suggests that relativizing the relationship of determinate reality is by
no means inconsistent or patently absurd (1977: 275), but since relativizing
332 TIME AND SPACE
here means rendering non-transitive, some may disagree. Of course, Compound
Present ists willing ly embrace the non-transitivity of coexistence, for they see no
incoher ence in the idea that events may undergo absolute annihilation as well as
absolute becoming, and I suspect they are right. But any dynamist who takes a dif-
ferent view on this issue, and who also inclines towards compatibilism, will have
to look elsewhere. The version of compatibilism proposed by Stein (1968, 1991)
may well fit the bill.
19.3 Time fragmented
Stein suggests that anyone who wishes to reconcile STR with real becoming must
answer two questions:
What is the spatiotemporal nature of the stages in which reality comes into
being?
What is the spatiotemporal criterion that, at any given stage, distinguishes the
definite from the not yet settled?
He cites the following as uncontroversial general principles to which answers to
these questions should comply:
(i) The fundamental entity, relative to which the distinction of the
already definite from the still unsettled is to be made, is the here
and now; that is, the space- time point
(ii) If the state at point b is already definite as of point a, then whatever
is definite as of b is (a fortiori) definite as of a. (Thus the notion is
already definite as of is to be a transitive relation between points.)
(iii) The state at any point a is already definite as of a itself.
(iv) For any point a, there are points whose state is unsettled as of a.
(1991: 148)
He further assumes that there is a definite time- orientation throughout Minkowski
spacetime, an assumption needed to make sense of the claim that what lies in the
future (rather than the past) is not yet definite or settled. With these assumptions
in place he goes on to establish, in rigorous fashion, that for any given point x,
the only points that are R- related to x (and so those that are already definite as of
x) are those that lie in xs causal past: the points on or in the past light- cone of x,
and, of course, x itself. Points that are outside xs past light- cone, those that are
space- like related to x, have the same status as the points in xs future light- cone;
they are not real or definite with respect to x. No other assignment of a domain
for R is compatible with the stated assumptions.
2
The relativization of becoming to individual points has the effect of fragment-
ing reality. No longer is there a universe- wide division between what is real and
determinate and what is not; the division is different for every point, since every
point has a different past light- cone, and on Steins model, what is real and definite
for a given point does not extend beyond the past light- cone. As a consequence,
no two observers will fully agree on what is real and definite unless they are
coincident, for it is only then that their past light- cones will entirely coincide. If
RELATIVITY AND REALITY 333
these observers are in motion relative to one another, they will still agree about
what is real and definite, but they will disagree about what is present. But since
for Stein everything in the absolute- elsewhere has not become and so is not yet
determinate this disagreement does not concern anything that is factual as of
the relevant point.
Stein does not take himself as having proved that objective becoming actually
occurs in Minkowski spacetimes; he remains neutral on the issue. His aim is simply
to show that the Rietdijk/Putnam considerations do not rule out the possibility that
becoming occurs, and many commentators agree that he has succeeded.
3
Steins
contention that assumptions (i)(iv) are uncontroversial is debatable as he notes
himself, not all writers on this topic share them but the important point is that
they are both reasonable and defensible. Rietdijk and Putnams objection succeeds
against CDMs, models of time that posit a single universe- wide ontological division
between what is real (or definite) and what is not, but, as Stein notes, this choice
of target seems singularly misconceived. The notions of temporal becoming
and spatially extended now seem to be conceptually independent, so there is
no obvious or compelling reason to think that the former requires the latter, and
this is especially true in the context of the special theory. STR explicitly denies
the objectivity (or frame- independence) of a spatially extended present, so given
the assumption that real becoming is an objective process, it is clearly unfair to
inflict on the dynamist a spatially extended division between what is as of now
determinate and what is not: to insist (without supporting argument) upon a
notion of present [spatially distant] actualities, in assessing special relativity, is
simply to beg the question, since it is fundamental to that theory that it rejects
any such notion (Stein 1991: 152). Sklar is guilty of the same error, for he too
assumes that the dynamist will want to transplant a concept of extended becom-
ing that, while perfectly appropriate in a Newtonian setting, is quite alien to the
environ ment of STR.
Sklars approach also requires the dynamist to adopt a non-transitive notion
of real or determinate with respect to. Steins approach does not. Provided we
stipulate, as Stein does, that the R- relation can only hold between a pair of points if
one is in the causal (or absolute) past of the other, R can be transitive without risk
of becoming universal (i.e. spreading from the past to the future). If any point can
be R- related only to points in its causal past, although it follows from Rab and Rbc
that Rac, it also follows that c is in the absolute past with respect to a, and the same
applies generally. The key factor is that while R is transitive and reflexive, it is not
symmetrical: if b is definite as of a, by virtue of lying in as causal past, it does not
follow that a is definite as of b. If it did, then we would again be in the situation in
which all points are equally definite with respect to one another. The relation
is real as of in the BroadTooley version of the Growing Block theory is also
asymmetrical: 1900 is real as of 2000, but as of 1900, the year 2000 is not real.
Steins compatibilism may be coherent, but in one respect it is ambiguous. The
avowed task is to reconcile a metaphysic of becoming with STR, but when it
comes to specifying what this metaphysic actually involves, Stein himself is less
than fully clear. He talks of a notion of temporal evolution as (in some sense) a
becoming real, or becoming determinate, of what is not yet real or determinate
(1968: 14, original emphasis), and of a distinction between a part of the worlds
334 TIME AND SPACE
history that had already become or is ontologically fixed and definite and a
part that is not yet settled (1991: 148). Is the not yet settled part of reality
wholly non- existent? Or does it exist, but only in a sketchy, partially complete
form? It may not matter. Perhaps the notion of a real but not fully determinate
event can be made coherent; if so there are dynamic models of time other than
those we explored previously. What does seem clear is that Steins account of
becoming can be applied to models of time that posit a clear distinction between
what is real and what is not, where becoming is absolute becoming, the ex nihilo
creation of time and events or rather, spacetime points and events.
An understandable reaction to what has been said so far would be:
I think I have grasped what Steins account of becoming amounts to; we
cannot ask what is real as of a given moment of time, if by this is meant a
temporal cross- section of the universe as a whole, but we can ask what is real
as of a given place at a time, a spacetime point. And what is real as of one
point is different from what is real as of every other point. What I cant do
is form an intuitively clear picture of what this point- relativized dynamism
amounts to.
A universe that becomes la Stein is hard to visualize, but since this can be put
down to contingent (and understandable) limitations on our imaginative abilities
we can only easily picture planes of becoming that are spatial this is not neces-
sarily a problem for Steins model. Dorato provides some suggestive descriptions:
the universal tide of becoming of the CDMs is replaced by water flowing through
an uncorrelated, non- denumerable set of narrow creeks (1995: 184) and shat-
tered into small disconnected fragments, that is, convex infinitesimally small
regions in the future lobe of the light- cone at any point p within [the Minkowski
manifold] M (1995: 185).
Perhaps the easiest way of putting some pictorial flesh on the model is to consider
things from the perspective of individual worldlines, Doratos narrow creeks of
becoming. Consider a collection of point- particles scattered through a region of
space; from one moment to the next the worldlines of these particles grow, and
as they do so they drag their past light- cones with them. To generate an image,
imagine a large cloth sheet lying on the ground that is gradually pushed into a
shape resembling a mountain range by several rigid rods slowly rising up from the
ground beneath. Looking down from above, not only do we see the entire cloth,
but we see the peaks associated with each growing worldline all rising together, in
synchrony. But these are artifacts of the mode of representation to which nothing
in the envisaged reality corresponds. There is no one privileged perspective within
this world to which the world as a whole is revealed, or to which the world as a
whole is factual; there are just the different pers pectives of each point, and what
is real for each point is different (although of course, for neighbouring points the
difference is infinitesimally small). Consequently, there are no temporal relations
between the different peaks. All that is real from the perspective of each peak is
what lies on and beneath the downward pointing part of the sheet (i.e. its causal
past). In Figure 19.2 the arrows point towards what is not real as of each peak.
There is one further respect in which pictures like Figure 19.2 might mislead:
the undulations are not purely spatial, they are spatiotemporal. Figure 19.3 depicts
RELATIVITY AND REALITY 335
the past light- cones at two neighbouring locations on the worldline of a particle,
P
1
and P
2
; the smaller triangle exists in the causal past of the larger, and the dif-
ference between the two represents the regions of spacetime that are real relative
to P
2
but not P
1
. The regions of spacetime indicated by the arrows at the bottom
of Figure 19.3 are very remote in the causal pasts of the points at the apex of the
light- cones: billions of light- years away. Consequently, some of the events that are
real for P
2
but not for P
1
are not merely a long way off in space, they are a long
way off in time: billions of years in the past. It is not just worldlines that grow in
Steins model, it is light- cones and all they contain; so as our worldlines grow, so
too do the regions of the very distant past that are real relative to us. Indeed, as
your worldlines grow, by far the greater proportion of what becomes real for you
lies in the spatially and temporally remote past, whereas what becomes present
is always strictly confined to the here and now. Light, in Steins model, does not
merely reveal; it makes real.
Stein provides a way of retaining a form of the Growing Block metaphysic in
the context of STR, but his model does not provide a congenial home for Present-
ism. However, a simple modification does the trick. We retain the idea that what
is real is always relative to a single spacetime point; but rather than regarding the
whole past light- cone of this point as real and determinate, we exclude the interior
region, and hold that only the surface of the cone is real; the final step is to regard
the entire past light- cone of a point as the present of that point. For understandable
reasons, Hinchliff calls this position Cone Presentism.
4
From the standpoint of common sense, Cone Presentism has a highly appealing
feature. You may recall how surprised you were when you were taught that every-
thing you see is actually in the past, due to the time- lag generated by the finite speed
of light; astronomers observing distant stars and galaxies see them as they were mil-
lions or billions of years ago. If Cone Presentism is true, the naive view is vindicated:
Figure 19.2 Steins model: becoming relativized to spacetime points. As worldlines grow, their
past light-cones expand. Despite appearances, the portions of the cones indicated by the arrows are not
real relative to one another.
P
2
P
1
Figure 19.3 The coming-into-being of the distant past.
336 TIME AND SPACE
everything you see is in the present (or at most only a fraction of a second in the past
the time it takes the brain to process the retinal image). The doctrine has a second
appealing feature. By restricting the present to the here and now (a single point),
Steins model has the consequence that (strictly speaking) we are always alone: no
one else shares our present. This is no longer true under Cone Presentism, since
everyone and everything on our past light- cone shares our present.
As might be expected, Cone Presentism also has some counterintuitive features.
Symmetry and transitivity must be denied. If event E is in my present (by virtue
of being on the surface of my past light- cone), I am not in Es present (since I am
not on the surface of Es past light- cone, but rather on the surface of Es future
light- cone). If an event D lying within my past light- cone occurs on the surface of
Es past light- cone, then although D is present for E, and E is present for me, D is
not present for me (it is past). But no doubt the most counterintuitive feature of all
is the way in which temporally distant events are rendered present, for example, a
supernova that occurred a billion years ago, but is currently being observed by an
astronomer. While this does conform with the phenomenology of perception as
we have just noted it also seems very odd, to say the least. In saying that an event
that occurred a billion years in the past can also be present, are we not stretching
the notion of the present beyond breaking point?
However, given the way STR renders temporal distances frame- dependent, this
objection does not have the force it otherwise would (e.g. in Newtonian space-
times). Since the temporal (and spatial) distances between two distinct events can
vary enormously from one reference frame to another, there is no objective fact as
to the true spatiotemporal distance between the events in question. Consequently,
the Cone-Presentist does not equate the present with a collection of events (and
points) that occur simultaneously, where simultaneously means at the same
time, or zero temporal distance apart; rather, the present is composed of events
(and points) that are at zero spacetime interval from one another. Unlike its pre-
relativistic counterpart, Cone Presentism is not a doctrine about time at all, it is a
doctrine about spacetime, of the Minkowski variety, and in this context present
takes on a new meaning.
5
19.4 Absolute simultaneity: the quantum connection
Dynamism can be rendered compatible with STR, but the cost to our common- sense
understanding of reality is more than some dynamists are prepared to pay. For
those falling into this camp an approach that may appeal is the refusnik strategy:
There is a single, universe- wide division between what is real and what is
not, a division that coincides with the present. Since STR denies this, STR must
be false, and the same applies to theories such as GTR that also relativize the
present.
This stance has the merit of exemplifying a healthy scepticism towards current
scientific theorizing, for it may well be that Einsteins physics will one day meet the
same fate as Newtons, and be displaced by a quite different theory with superior
explanatory and predictive power. However, the strategy suffers from two defects.
RELATIVITY AND REALITY 337
First, if there is an absolute and universe- wide present there must be an absolute
simultaneity relation. Since no such relation is required or possible in STR, an
opponent could reasonably argue thus:
I agree that we could have found evidence for absolute simultaneity. For
example, we might have discovered that some signals or influences travel
instantaneously they are received at the same moment they are sent,
irrespective of the distance they have to travel; in which case we could have
defined an absolute simultaneity relationship via these signals. But no such
evidence has been found. The fact that your pet metaphysical model of time
requires such a relation does not, in itself, give us any reason to believe that
your model applies to our world.
And this has some force. In the absence of any reason for believing that events in
our world are absolutely simultaneous, compatibilist dynamisms seem less costly
than the CDMs, for at least they do not require us to posit facts for which there is
no empirical evidence.
Secondly, STR may indeed one day be replaced by another theory, but since the
relativistic empirical effects such as measurable time dilation and length contrac-
tion are well confirmed, the successor theory will have to explain them in its own
terms. And so must anyone seeking to defend CDMs.
Tooley is one dynamist who is alert to these problems and also committed to
absolute simultaneity and a universe- wide division between what is real and what
is not real. Rejecting the refusnik strategy he sets himself the task of devising an
alternative to STR that:
is closely related to STR, and compatible with all the observational data that
are usually taken to confirm it;
entails that events stand in relations of absolute simultaneity; and
is superior to STR in other respects, and so preferable independently of its
conforming to the requirements of CDMs.
I will conclude by briefly summarizing Tooleys position (elaborated in Tooley
1997: ch. 11), which although bold is not without its rationale.
6
Tooley starts by showing that absolute simultaneity can in fact be defined within
the framework of STR provided we make certain metaphysical assumptions about
space time. The assumptions Tooley proposes are these:
Spacetime points are substantival entities, and likewise spacetime itself is the
sum of these entities.
Spacetime points can be causally related to one another.
Every spacetime point causes at least one other spacetime point to exist, and
was itself caused to exist by a spacetime point.
Causally related spacetime points occur at the same place (i.e. on the same line
through spacetime).
If P, Q, R and S are four spacetime points, such that the existence of P causes
the existence of Q, and the existence of R causes the existence of S, then the
lines defined by PQ and RS are parallel.
338 TIME AND SPACE
The picture, then, is of an absolute substantival spacetime that is uniform and
causally self- propagating over time. We now assume that spacetime (rather than
any aether) is the medium through which light propagates, and that the speed
of light is constant relative to absolute space (which is what we would expect if
natural laws are uniform). As for simultaneity, Tooley accepts Einsteins opera-
tional definition: two events E
1
and E
2
occur simultaneously if light from each
would arrive at an object O which is both equidistant and at rest relative to them
(i.e. in the same inertial frame) at the same time. Starting with this definition we
only reach the conclusion that simultaneity is relative if, like Einstein, we assume
that all inertial frames are on an equal footing, and this Tooley rejects. If absolute
space exists, then some inertial frame is motionless with respect to it, and so we
can say that Two events E and F are absolutely simultaneous means the same as
E and F are simultaneous relative to some frame of reference that is at rest with
respect to absolute space (1997: 344).
According to STR in its standard form, light has the same speed in all directions
in all inertial frames. In Tooley s modified theory, light has different speeds in
different inertial frames. To focus on the simplest case, a light ray will have a
speed of (c v) in a frame moving at velocity v in the same direction as the ray,
and a speed of (c + v) in a frame moving at v in the opposite direction to the ray.
Since empirical evidence supports the same speed in all inertial frames predic-
tion, doesnt Tooleys theory fall at the first hurdle: compatibility with known
empirical data? Not in the least. What the empirical data actually supports is the
weaker hypothesis that, for any two locations L
1
and L
2
, observers in all inertial
frames will agree on the round- trip speed of a light ray travelling from L
1
to L
2

and back again the speed that results from dividing the total distance travelled
by the time taken. But as Einstein himself recognized in his original 1905 paper,
there is no empirical evidence that the one- way speed of light is constant in all
inertial frames; he simply regarded it as a reasonable assumption, and made it true
by definition. There have been many subsequent attempts to devise experiments
that would demonstrate that the one- way speed of light is constant, but thus far
none has been successful. In the absence of any countervailing empirical evidence,
Tooley feels free to abandon Einsteins assumption, and hold that light travels at
different speeds in inertial frames that are in motion relative to absolute space.
As for why this is not detected, he relies on a Lorentz- style compensatory theory:
physical processes in moving frames are affected in ways that systematically con-
ceal the variations in light speed that actually obtain. Since the original Lorentz
transformations presuppose that the one- way speed of light is the same in all
inertial frames, they will not perform the required job, but there is an alternative
that will: the so- called c- Lorentz transformations (the terminology is due to
Reichenbach), which specify the necessary compensa tions corresponding to dif-
ferent assumptions concerning the one- way speed of light. Winnie (1970) showed
that a suitably modified version of STR entails the c- Lorentz transformations, and
that the modified theory is inconsistent only if STR is itself inconsistent.
That there should be an empirically equivalent competitor to STR is not surpris-
ing as we saw during our discussion of conventionalism, there is always room
for such a competitor but is there any reason for preferring Tooleys version of
STR over the standard form? Discounting any purely metaphysical preferences we
RELATIVITY AND REALITY 339
might have, in the absence of any evidence for an absolute simultaneity relation,
defenders of the status quo (standard STR) will object that theories such as Tooleys
are guilty of positing a real relationship absolute simultaneity which we have
no reason to believe exists. It is at this point that Tooley turns to quantum theory.
Defending the possibility that absolute simultaneity could exist in our universe,
Lucas remarks:
if some superluminal velocity of transmission of causal influence were
dis covered, we should be able to distinguish frames of reference, and say
which were at rest absolutely and which were moving. If for instance, we
were able to communicate telepathi cally with extraterrestrial beings in some
distant galaxy, or if God were to tell us what was going on in Betelgeuse now,
then we should have no hesitation in restricting the principle of relativity to
electromagnetism only: hence it cannot be an absolute principle foreclosing
absolutely any possibility of absolute time. (1999: 910)
And in this he is surely right. Were we to discover that a truly instantaneous
connection exists between objects at different places in space, then assuming the
connection has some detectable effects, not only would the notion of absolute
simultaneity have a real application, but we would have a way of determining
which frames of reference are at absolute rest and which are not: it is only with
respect to frames truly at rest that the relevant changes would occur at precisely the
same time. The idea that there are contra Einstein instantaneous connections
in nature has gained favour in recent years as the significance of a now famous
theorem due to John S. Bell has gradually sunk in.
In 1935 Einstein, Podolsky and Rosen (EPR) came up with an ingenious argu-
ment designed to show that quantum mechanics, at least in its orthodox form,
cannot provide a complete description of nature. According to quantum ortho-
doxy, particles only acquire definite properties when these same properties are
measured: a photon only acquires a definite polarization when its polarization is
measured; an electron only acquires a definite momentum when its momentum is
measured; and so on. EPR devised a thought- experiment that (as they thought) cast
doubt on this doctrine. Quantum theory predicts that it is possible (in principle
at least) to create pairs of particles that are correlated with one another, in the
sense that there is some property that both particles possess, which is such that
if the property has a certain value for one of the particles it will have a certain
value for the other, irrespective of the spatial separation of the particles. To take
an artificially simple example, suppose that the property is P, and the value of P
for both particles must add up to 1. Then if particle A is measured to have a P of
0.25, particle B will have a P of 0.75, whereas if P for A is 0.5, it will be 0.5 for
B, and so on. (Actual instances of P include quantum properties such as spin and
polarization, whose numerical values take different forms.) Suppose that we create
a pair of correlated particles that speed off in opposite directions, and measure the
value of P for A at time t, and find that it is 0.6. We know at least this much: from t
onwards the value of P for B is 0.4. But when precisely did B acquire this property?
One possibility is that neither A nor B possesses a determinate value of P until
time t; at the very same moment that A is measured as possessing a P of 0.6, B
acquires a P of 0.4. It is necessary that the particles acquire determinate values
340 TIME AND SPACE
absolutely simultane ously. If this does not occur, if the P- value of B remains truly
indeterminate for some period after t (for some frame of reference), there is noth-
ing to prevent our measuring the value of P for B and finding a value other than
0.4, which we know is nomologically impossible. But this can only be nomologi-
cally impossible if the value of P for B becomes fixed at precisely the same time
as P for A is fixed. However, we are now confronted with a situation in which
what is done to A instantaneously affects B, irrespective of how far the two par-
ticles are spatially separated from one another: it is only because we conduct a
P- measurement on A at t that B acquires a determinate P- value at t. How can B
know what is being done to A this quickly?
If Bs state is instantaneously affected by the measurement performed on A, one
thing seems clear: STR cannot provide a complete account of the spatiotemporal
relations that actually exist, since superluminal causal connections between spa-
tially separated material things are explicitly ruled out by the theory. For obvious
reasons, this is Tooleys preferred interpretation.
But there is a second possibility, which for understandable reasons Einstein
found more palatable, namely that both particles always had determinate values
for P: at the moment they were created A had a P- value of 0.6 and B a value of
0.4. The measurement of As P- value at t merely reveals a property that A always
possessed, and this is why if we were to measure the P- value of B at t we would find
it to be 0.4 this was the value B had all along. Hostile to the idea of superluminal
spooky action- at- a- distance, this is the option that EPR preferred. And hence
their conclusion: quantum mechanics cannot provide a complete description of
nature. Since quantum theory does not ascribe a P- value to A or B until a P meas-
urement is conducted on one of the particles, if we suppose that their P- values are
fixed when they are created, there are properties that exist but quantum theory
does not recognize.
The situation changed radically in 1964 when Bell showed that EPRs preferred
interpretation was not in fact an option. Bell proved that the quantitative predic-
tions made by quantum mechanics are incompatible with the possibility that A
and B possess determinate P- values before any measurements are made. As Tooley
puts it: the thrust of the EinsteinPodolsky Rosen thought- experiment is no
longer merely that the Special Theory of Relativity, or else quantum mechanics, is
incomplete. It is rather that either the Special Theory of Relativity is incomplete,
or quantum mechanics is false (1997: 359). The Aspect experiments on quantum-
correlated particles, carried out in the early 1980s (and successfully replicated on
many subsequent occasions) revealed that the quantitative predictions of quantum
theory for this type of situation are in fact correct, and so, given Bells result, the
EPR interpretation is falsified: the particles cannot possess determinate P- values
prior to measurements being conducted. If Bs P- value becomes fixed only when
As P- value is measured, it seems that some events at least occur absolutely simul-
taneously.
Or so Tooley concludes. And he is not alone. Although Tooley does not mention
the fact, there is a well- known and respected interpretation of quantum mechan-
ics due to David Bohm which is highly non- local and requires (and posits) a
preferred reference frame with respect to which non- local interactions are instan-
taneous.
7
RELATIVITY AND REALITY 341
But these matters are complex; the correct interpretation of quantum theory
remains controversial, likewise the relationship between quantum theory and
relativ ity. Nonetheless, the possibility of physics recognizing a preferred reference
frame (and so abandoning Lorentz invariance) remains real. After a thorough
investigation of rec ent debates Maudlin (1994: 240) concludes that the space- like
connection established by the Aspect experiments:
does not require superluminal matter or energy transport;
does not entail the possibility of superluminal signalling;
does require superluminal causal connections; and
can be accomplished only if there is superluminal information transmission.
There are ways of accommodating the quantum results within STR but, as Maudlin
notes, the cost is high:
Embedding quantum theory into the Minkowski space- time is not an impos-
sible task, but all the available options demand some rather severe sacrifices.
Bohms theory and orthodox collapse theories seem to require the postula-
tion of some preferred set of hyper planes, ruining Lorentz invariance and
jettisoning the fundamental ontological claims of the Special Theory. Lorentz
invariant theories might be developed using explicit backwards causation, so
that the birth of every pair of correlated particles is already affected by the
sorts of measurements to which they will be subject, perhaps hundreds of
years later One can achieve Lorentz invariance in a theory with instantane-
ous collapses by embracing a hyperplane dependent formalism, but the exact
ontological implications of such a view are rather hard to grasp Or finally,
one can avoid collapses and retain locality by embracing the Many Minds
ontology, exacting a rather high price from common sense.
the common thread that runs through all these proposals is that no
results are to be had at a low price. Indeed, the costs exacted by those theo-
ries which retain Lorenz invariance is so high that one might rationally prefer
to reject Relativity as the ultimate account of space- time structure.
(1994: 220)
It is with considerations such as these in mind that Lucas writes: [although] the
Special theory made time seem spacelike quantum mechanics is kind to time
[it] redresses the balance. We no longer need feel obliged to construe time in
a non- temporal way in order to be truly scientific and philo sophically respect-
able (1999: 11). But while quantum mechanics alone may indeed be kind to
our ordinary understanding of time, we have already seen in 6.9 that quantum
mechanics may yield a very different result when integrated with GTR. Attempts to
introduce gravity into quantum theory give rise to the quantum problem of time:
if time exists in the equations of quantum gravity theory, it is not obvious where.
What can we conclude? Given the manner in which STR relativizes simultaneity,
it is easy to suppose that Minkowski spacetimes are of the Block variety, and hence
agree with Putnams claim that the problem of the reality and determinateness of
the future is now solved, and solved by physics and not philosophy. But this
is wrong, for there are compatibilist and non- compatibilist ways of reconciling
dynamic models of time with the empirical evidence that supports STR. Steins
342 TIME AND SPACE
point- relativized dynamism preserves the spirit of STR while retaining an ontologi-
cal division between what is as yet real and what is not; following on from this
lead, Cone Presentism goes a step further, and restricts reality to what is present
(relativistically construed). Tooleys modified version of STR goes still further, and
by rejecting the one- way constancy of the speed of light, retains a universe- wide
plane of becoming, without abandoning empirical equivalence with orthodox STR.
Indeed, we have seen that there may well be support from physics for an absolute
simultaneity relation. But as this support is rooted in the still murky and much dis-
puted arena where quantum theory and relativity come together, it is too early to
know precisely what impact physics will eventually have on our understanding of
time in our universe. However, even if STR has not provided the definitive solution
to the problem of whether or not time is dynamic, it has certainly cast new light
on the whole issue, and forced us to entertain previously unforeseen possibilities.
The debate has moved on, even if where it will lead as yet remains unclear.
8
20 General relativity
20.1 The limits of STR
In this chapter several threads of our inquiry thus far come together. Our topic
is Einsteins general theory of relativity (GTR), the product of ten years arduous
labour, which followed the completion of STR in 1905. During this intervening
period Einstein wrote never in my life have I tormented myself anything like this
Compared to this problem the original relativity theory [STR] is childs play.
1

The metaphysical implications of this elegant and intriguing theory are of interest
in their own right, for it remains our best account of our universes large- scale
spatiotemporal structure. But it is of interest for two further reasons.
Since GTR posits a variably curved spacetime manifold, it constitutes a challenge
to Poincars conventionalism: the claim that, although non- Euclidean spaces are
mathematically intelligible, we could never, in practice, have reason to accept a
theory that ascribes a non- Euclidean geometry to our world. We will be looking
into the considerations that led Einstein (and most other physicists) to accept a
theory that does precisely this. GTR is also of interest in the light of the discussion
in 14.23, where we arrived at the conclusion that in worlds where material
objects behave as if they inhabit a variably curved space, there is a strong case for
supposing that the space in question is substantival rather than relational. We shall
see that, although there are strong grounds for construing GTR in a substantivalist
manner, there are also difficulties of a hitherto unsuspected variety.
As for why Einstein found himself obliged to develop GTR in the first place, it
was not because he discovered a flaw in STR, but rather because the latter is lim-
ited in its aspirations: it was designed to explain some basic physical phenomena,
but not all (hence the label special theory of relativity as it only applies to cer-
tain special cases). STR provides an account of the difference between accelerated
and inertial motions (Min kowski spacetime has a well- defined affine structure),
and a harmonious integration of the dynamics of moving bodies and Maxwells
electromagnetic theory, but it was silent on one crucial phenomenon: gravita-
tion, the force that keeps the planets in orbit around their stars, that prevents us
from floating off into outer space, and which provided Newton with his greatest
triumph.
There are two main reasons why Newtons theory of gravity cannot be smoothly
slotted into the framework of STR. First, Newtons theory of gravity presupposed
that time and space were separate and invariable quantities; this presupposition
was abandoned by Einstein with STR, so he needed to develop a theory of gravity
344 TIME AND SPACE
that was compatible with length and time dilation and the relativity of simultane-
ity. It is easy to see how relativistic effects lead to anomalies when combined with
Newtonian gravitation. The strength of Newtons gravita tional force between a
pair of bodies depends on their distance apart. The fact that under STR this dis-
tance will be different in the inertial frames of the relevant bodies (if they are in
relative motion) means that observers on the two bodies will arrive at different
results when they calculate the strength of the gravitational attraction between
them. Secondly, Newtonian gravity is a force that acts between all material bod-
ies instantaneously, irrespective of their distance. Since STR puts a finite limit on
causal propagation, Einstein had to find an alternative explanation of gravitational
effects.
The explanation he came up with was remarkable. Whereas Newton regarded
gravity as a force, which acted at a distance to attract material bodies towards each
other, Einstein explained gravitational effects without appealing to such a force,
and did so in a way that dispensed with the need for instantaneous action- at- a-
distance. Einstein replaced Newtons attractive force with mass- induced space-
time curvature. Quite why this suffices to keep us rooted to the Earths surface
will emerge in due course.
20.2 Equivalence
Einstein started serious work on the problem of gravity in 1907, and his first
breakthrough occurred in November of that year: I was sitting in a chair in the
patent office at Bern, when all of a sudden a thought occurred to me: if a person
falls freely, he will not feel his own weight. He was later to call this the happiest
thought of my life.
The fact Einstein grasped is no doubt a good deal more obvious these days
than it was in 1907. Suppose you are in a lift, going up, and the cable snaps.
The lift cabin begins to fall freely. What happens? Does the roof of the cabin
rush down and bang into your head? No. As Galileo discovered, all objects,
irrespective of their size, weight and composition, fall at exactly the same rate
when released gravity is not a discriminating force and so the cabin and all its
occupants will fall at the same speed. In fact, you will find yourself weightlessly
drifting around the cabin just as you would if you were in outer space. We have
all seen films of astronauts floating around in weightless conditions, and this is
just how it would be for you in the lift cabin. The analogy is closer still since
astronauts in orbit around the Earth are in effect falling towards the Earth all
the time (they remain in orbit only because they also have a considerable forward
directed velocity). In short, if you go with the flow near a gravitating body, you
dont feel any force; to avoid feeling the pull of gravitational forces, you need
only to be allowed to continue with your free- fall path. You only feel a force
inside a gravitational field if something prevents you from continuing along your
free- fall trajectory, if something else applies a force to you, such as a floor, or the
Earths surface.
Thanks to this insight, Einstein realized that the relativity principle he used in
STR could be extended. Roughly speaking, the relativity principle of STR says that
GENERAL RELATIVITY 345
the laws of physics are the same in all inertial frames, that is, those that are at rest
or moving in straight lines (and not under the influence of external forces). But
if the effects of gravity disappear in free- fall, then we can also say that the laws
of physics take the same form in frames that are freely falling in gravitational
fields as they do in inertial frames. Suppose you are in the cabin of a windowless
spaceship, enjoying weightless con ditions. You wonder about your state of motion.
Are you drifting aimlessly in interstellar space, or are you accelerating towards a
nearby planet under the influence of its powerful gravitational field? Surprisingly,
alarmingly, there is nothing you can do to find out. All experiments that you can
perform within the cabin will yield exactly the same results if you are accelerating
under gravity as they would if you are not. Nature does not distinguish between
inertial frames and frames that are freely falling in gravitational fields. This became
known as the equivalence principle.
Einstein recognized another respect in which the presence of gravity is unde-
tectable. Suppose you wake up in your windowless space cabin to find that you
are no longer weightless; in fact, it feels just as it would if the cabin were lying
motionless on the Earths surface. Nothing is floating in mid- air, you can walk
about as normal, you can feel the difference between up and down (a disturbing
absence in weightless conditions), objects fall when dropped, and so forth. Can
you conclude that you are in fact on Earth (or at least on some massive body)?
After a few moments thinking back to the (Newtonian) physics you learned in
school, you realize that the answer is no. Newtonian physics can explain all the
relevant phenomena in two different ways:
you are being held motionless near a planet or star, and feeling the effects of
gravity; or
you are not subject to any gravitational forces, but are being uniformly acceler-
ated (by a silent rocket motor, say); we know that acceleration gives rise to
forces (when a plane takes off you are pulled back into your seat).
Once again, no test you could carry out inside your windowless laboratory would
tell you if you were inside a gravitational field, or being accelerated by a rocket.
(Of course, both could be happening, but to keep things simple we will focus on
the simple either/or case.) This was a further key point that Einstein grasped: so
far as their physical effects on a body are concerned, the pulls generated by
gravitational attraction and acceleration are indistinguishable.
The equivalence principle in this form is particularly significant. Why? Because
whereas acceleration is a well understood and purely local phenomenon, grav-
ity at least as construed by Newton is a mysterious force linking all bodies in
the universe, which acts instantaneously. Here was a promising hint as to how the
problem posed by gravity might be solved: given that there was no way of tell-
ing, locally (over small distances), between gravitational effects and acceleration
effects, perhaps there is no difference between the two; perhaps gravity and accel-
eration are essentially the same phenom enon. It was several more years, which
involved numerous detours and false starts, before Einstein finally worked out
how this might be.
346 TIME AND SPACE
20.3 Spacetime curvature
Starting from the equivalence principle, there are several lines of reasoning that
lead to the idea that gravity is not a force acting in or over spacetime, but rather a
modification of the very structure of spacetime itself.
The equivalence principle yielded a novel prediction: light should be affected
by gravity (this was generally denied at the time, since it was widely believed that
light was massless). Einstein reasoned thus. If you were in a sealed laboratory being
accelerated by a rocket motor and aimed a light ray across the room, it would hit
the opposite wall at a lower position than if the laboratory were not moving the
velocity (and so direction) of light is independent of the source, and the laboratory
changes position during the time the light takes to cross the room. If accelerated
motion and gravitational attraction are equivalent, then if the laboratory is held
stationary in a gravitational field, the light ray should also hit the wall lower down.
The diagrams in Figure 20.1 depict the apparent path of the light ray from the
perspective of someone within the laboratory.
As we saw in the previous chapter, Minkowski spacetime is structured around
the trajectories of light rays. Not only do they define the absolute past and future
of every point, but they are the paths of shortest possible distance: the interval
between light- like connected points is zero (and so there is, from the perspective
of light, no spatial or temporal distance between them at all). Consequently, if
gravity affects the path light takes, and one is already assuming that we live in
a Minkowski spacetime, it is not a huge leap to the idea that gravity affects the
structure of spacetime itself.
There were other considerations, of a more concrete sort, that convinced Ein-
stein that acceleration would produce warpings of both time and space. Extending
Figure 20.1 The equivalence principle and light. Light travelling across the cabin of an accelerating
spacecraft follows a curved path. If gravity has the same (local) effects as acceleration, light crossing the
cabin of a stationary spacecraft within a gravitational field should follow a similar trajectory.
GENERAL RELATIVITY 347
the reasoning he employed in STR, he imagined a group of people on a rotating
disc who perform various measurements, of both distance and time, and worked
out exactly what they would find. Suppose you have the role of external observer,
and have measured the circumference and radius of the disc before it starts moving,
using a string and ruler. Once the disc is in motion, you pass the ruler to someone
on the disc, Sam, who uses it to make similar measurements. As he sets about
measuring the circumference, you observe that Sams ruler is shortened (due to the
Lorentz contrac tion), and realize that the number he will arrive at for the circum-
ference will be larger than yours. When he measures the radius, on the other hand,
the ruler is not pointing in the direction of rotation, but at ninety degrees to it, so it
will not be shortened (length contraction occurs in the direction of motion), and his
measurement will agree with yours. When Sam puts the results of his measurements
together, he finds that the ratio of the circumference to the radius is greater than
2t, which as we saw earlier is a result that is indicative of a non- Euclidean space. If
we now imagine that there are some people at the edge of the disc who, at regular
intervals, compare their watches with that of someone at the centre, we know that
these watches will not keep the same time. Since the people at the edge are moving
faster than the person in the middle, and STR predicts that the faster you move the
slower time passes for you (speaking loosely of course) relative to others who dont
share your motion, time will pass more slowly at the edge of the disc.
What is the relevance of these disc- based results to gravity? Once again the
equivalence principle comes into play. Rotation is a form of acceleration, and if
accel era tion is associated with time dilations and changes in spatial geometry in
relativistic terms, warpings of the chronogeometry of spacetime then, given the
equivalence principle, gravity must be too.
What Einstein actually concluded was this: gravity is nothing but the warping of
space and time. Material bodies do not exert a gravitational pull on one another,
rather material bodies warp space and time, and these warpings produce the effects
we associate with gravity the effects Newton explained in terms of an attractive
force. It was mysterious how anything could act- at- a- distance in the required way.
Einstein does away with action- at- a- distance, since gravity for him doesnt exist as
a force operating between bodies. Instead, matter bends the spacetime in its vicin-
ity; the more matter, the greater the distortion, and the matter- induced curvature
is transmitted through spacetime at the speed of light, not instantaneously.
By way of an analogy, think of walking on the icy surface of a frozen lake. As
you get to the middle, the ice suddenly cracks beneath you. The cracks spread
across the surface, from the centre outwards, at a finite speed. Matter- induced
spacetime curva ture works rather like this. A large concentration of matter induces
strong curvature just where it is, and the spacetime in the immediate vicinity is very
curved; the space time in the next region isnt directly affected in any way by the
matter, but is affected by the adjacent spacetime, so the curvature is transmitted
through spacetime, from region to region, like cracks spreading through ice. But it
gradually weakens. As you get further away from the mass, the curvature of each
successive spherical region of spacetime diminishes. The important point is that all
gravitationally generated causal relations are local in GTR, they are directly trans-
mitted from one region of spacetime to the neighbouring regions of spacetime,
they dont work magically across the distances separating material bodies. Since all
348 TIME AND SPACE
the material bodies in the universe induce curvature in spacetime, the overall shape
of spacetime is the product of their combined influence, and as the bodies move
about the overall shape of spacetime alters; spacetime structure is thus dynamic
rather than static. There is a standard two- dimensional analogue. Think of a big
flat sheet of rubber stretched out like a trampoline. Now drop some heavy metal
balls on to it. The rubber will be bent, the deflection from flat being greatest near
each ball. Now move the balls about. The overall shape of the surface will change;
it will be differently curved.
At this point the concepts introduced during our earlier discussion of curved
space become relevant. In Newtonian (and STR) physics the natural path for a body
to take when it isnt subject to any forces is a straight line, and a body moving
along its natural path will continue to do so for ever, without slowing or speeding
up. These inertial paths are those that spacetime forces on a body, and a body only
deviates from an inertial path if some external force acts on it. We have seen that
when a body falls under the influence of gravity, it does not experience any forces
provided it goes with the flow and continues along its natural free- fall trajectory,
and this suggests that free- fall paths in gravitational fields are themselves inertial
paths. The path of a spaceship that quickly flies past a planet will be deflected; it
will be as if the planet is pulling at the ship, causing it to move in a curve rather
than a straight line. But if the spaceship is not being accelerated by its engines,
but simply continuing on its natural free- fall path, as it passes the planet it wont
feel any force acting on it (just as you didnt feel any force pulling on you in the
falling lift). The ships path bends inwards because the planets mass is altering the
inertial path of the spaceship: the planet is not pulling the ship towards it like a
magnet, it is altering the shape of spacetime. In a curved spacetime the equivalents
of Euclidean straight lines are the geodesics of the spacetime (the paths of shortest
length), and mass affects the curvature of these geodesics: the larger the mass, the
more the local geodesics deviate from Euclidean straight lines.
Figure 20.2 Mass-induced
spacetime curvature. The
greater the concentration of
mass, the greater the curvature.
Inertial paths cease to be
Euclidean straight lines in the
vicinity of mass-concentrations.
When the concentrations are
large, this becomes noticeable.
GENERAL RELATIVITY 349
20.4 Feeling the grip of spacetime
To see how all this applies to some simple cases of motion and explains the gravita-
tional effects we are familiar with in everyday life, we will start by ignoring time,
and focus on purely spatial curvature.
On the face of it, Newtons account of gravity as an attractive force has a signifi-
cant advantage over Einsteins: it explains why objects fall towards the Earth when
dropped, and why astronauts who have spent months in space and have become
unaccustomed to gravity feel the Earth dragging them down when they return.
How can these seemingly real instances of attraction be explained by altera tions
in the geometry of space? As it happens, quite easily. In response to the question
Why do we feel pulled downwards? Einstein answers, We dont, we feel pushed
up. We are prevented from following our natural free- fall trajectory (which is
determined by our prior motion and the matter- controlled shape of space) by the
surface of the Earth, and so we feel a force on us, as the Earth keeps preventing us
from continuing on our free- fall path. Why is our free- fall motion downwards? It
isnt. It is a greatly elongated corkscrew or spiral motion, since the Earth is moving
forwards (around the Sun) and rotating; given the distortion in space produced
by the Earths mass, our free- fall slightly inclines us into the Earths centre, and
the Earths surface pushes against us, preventing us from following our natural
path. So the only force applied to us is due to the Earths surface preventing us
from following our inertial path. There is no gravitational force pulling us, just an
alteration in the curvature of our local space.
If we accept the Einsteinian picture, it is not difficult to see why Newton
thought that gravity was a force. Suppose that Bob and Sue set off to roller- skate
around an artificial metal planet, using magnetic roller skates, which stick to the
planets smooth metal surface.
2
They set off along parallel paths, four metres
north and south of the equator, each of which circumscribes a circumference of
the planet, and they never deviate from these paths. Let us suppose that there is
no gravitational pull exerted by the planet on Bob and Sue; they just coast along,
on its perfectly smooth surface.
Bob and Sue dont realize that the planet they are on is spherical; they think it
is a great big flat plane, which is just how it looks viewed from the surface. Con-
sequently, they assume that, since they set off along parallel paths, the distance
Equator
Bobs path
Sues path
Figure 20.3 Roller-skating at the equator.
Bob and Sue, viewed from above, just prior to
setting off on their around-the-world trip.
350 TIME AND SPACE
separating them will remain constant, unless some force pushes them off their
original paths. Hence their surprise when they notice that they are slowly drifting
nearer to each other. In an attempt to stave off collision they each take hold of the
opposite ends of one of Bobs ski- poles, which they keep extended between them.
But while this prevents their drifting closer together, they now feel a constant
inward pressure. In effect, they have to be pushed apart to keep from coming
together. Not surprisingly they conclude that some force is pulling them together:
a mysterious force of attraction.
But if we look at them from above, from an orbiting spacecraft, we see that
there is no force involved; nothing is pulling them together. Given the way they
started off, they are merely following the great circle paths shown in Figure
20.4, and these paths gradually converge and intersect. These are their natural
(inertial) paths given the intrinsic geometry of the Earths surface, and the pressure
on the stick is due to the force needed to deflect them from these paths. When they
hold the stick between them, they are forcing themselves to follow paths that are
not inertial given the Earths geometry, and this is why they feel a force.
Figure 20.4 A meeting of ways. Bob and Sue
continue to move straight ahead, without deviating to the
right or the left; but since each is moving along a different
great circle, their paths converge, and eventually cross.
Suppose now that a fat uncle follows them. Sue finds that he set off just behind
her, and follows exactly her path. Since she is attributing her movement towards
the equator to the action of a force, and since the fat uncle moves like her, she
concludes that the force acts more strongly on greater masses, since it manages
to make her uncle move just like she does. Finally, when they find that all bodies
move in the same way, no matter what their masses, they conclude that the attrac-
tive force is proportional to the mass of each object: the more massive, the more
force, and the same resulting path.
So Sue and Bob are led to postulate a force that looks like Newtonian gravita-
tion: a universal force that is proportional to mass. But they are making a mistake.
There is no mysterious force of attraction, only a non- Euclidean intrinsic geom-
etry. Force- free objects are doing just what they should do: following straight- line
paths unless acted on by a deflecting force.
The examples we have looked at so far involved space, and might seem compat-
ible with a broadly Newtonian view of the universe. But of course, what is really
being curved, according to Einstein, is the spacetime manifold. This means that
there are time- dilation effects produced by gravity, as well as space- bending effects:
clocks tick more slowly in the vicinity of large material objects; the stronger the
spacetime curvature, the slower the clocks tick. Could Newton have explained
GENERAL RELATIVITY 351
gravity in space- bending terms, if he had given up the absolutist conception of
space as something whose shape is fixed and unalterable, and independent of
the bodies in space? At first sight it might seem so, but in fact the empirical data
wouldnt have permitted it.
If gravity is a curvature in space alone, then we would expect all bodies start-
ing from the same place and time to move along the same curved path under the
influence of gravity. For example, suppose you are at the other side of the room
from me, and I throw a ball towards you at the same time as I fire a bullet at you,
aiming in the same upwards direction (so the bullet should miss your head). Both
ball and bullet set off in the same direction, but one is moving much faster than
the other. They wont travel in a straight line: due to the influence of gravity their
paths will curve downwards. But will the curvature of both paths be the same?
No, clearly not. The slow- moving balls path will be strongly curved and the bullet
will move in almost a straight line. This difference means we cannot explain the
effects of gravity is terms of the curvature of space. It is quite different if we move
to considering the event in spacetime, rather than space alone.
Time
Space
Space
bullet
ball
Figure 20.5 Projectile motion in
spacetime. When we plot the paths
of the bullet and the ball in two spatial
dimensions and one time dimension, we
see that their curvatures are the same
they are different in space alone.
When time is explicitly introduced, we can see that, although both ball and bul-
let travel the same distance, the ball is in flight for longer, and when this is taken
into account, the curvature of its trajectory (in spacetime) is precisely the same as
that of the bullet, which is what we would expect if both are in inertial motion
through a medium with the same intrinsic curvature.
Hence we see why it must be spacetime, not space alone, that is the medium for
the motion of masses, and so Newton could not have explained gravity in terms
of purely spatial curvature.
3
20.5 Evidence
It is one thing to show, in a qualitative way, by appealing to images and analogies,
that gravitational effects can be explained by spacetime curvature, but another to
show that this account is superior to alternative accounts. It is here that empirical
evidence and precise quantitative predictions enter the fray. Despite the enormous
difference in basic concepts between GTR and (orthodox) Newtonian theory, when
it comes to predicting how material bodies should behave there is much com-
mon ground. For almost all the situations we can create in terrestrial laboratories
352 TIME AND SPACE
the two theories generate virtually identical predictions, and for the most part
the same applies to the behaviour of the planets. When planning interplanetary
probes, for example, scientists still use Newtons equations rather than Einsteins:
they are easier to work with, and the extent to which their predictions differ from
Einsteins are too small to have any significant effect. But there are differences.
In the 1900s there were only two known discrepancies between Newtons law
of gravity and the observed motions of the planets: one involved the Moon (and
eventually proved to be due to erroneous observations) and the other Mercury.
The orbit of Mercury is not a perfect ellipse. It completes each trip around the
Sun at a slightly different position from where it set off. Newtonian theory could
account for much of this perihelion shift but not quite all of it. There was a resid-
ual advance of 43 arc- seconds per century; a small difference, but a significant one,
since astronomers at the time reckoned that their measurements were accurate to
within 0.01 arc- seconds (equivalent to a hairs breadth at a distance of 10 km).
The observed perihelion shift is precisely predicted by Einsteins equations. I was
beside myself with ecstasy for days, he wrote after making this discovery in 1916.
An even more dramatic confirmation occurred in 1919. Einstein worked out that
starlight passing very close to the Sun on its way to the Earth should be deflected
by 1.75 arc- seconds. Dyson and Eddington (Director of the Greenwich observa-
tory and Secretary of the Royal Astronomical Society respectively) mounted an
expedition to Principe (off the coast of West Africa), where a total eclipse was due
(only then are the stars adjacent to the Sun visible). The confirmation of Einsteins
prediction made headlines around the world. In more recent years, using the more
sensitive measuring equipment that has become available, there have been many
further confirmations.
GTR does more than improve on the accuracy of Newtons theory. It has ena-
bled physicists to postulate, and in some cases confirm the existence of, entirely
new phenomena, such as spinning black holes, gravitational lensing and gravity
waves. One of the theorys earliest and most striking predictions that given
certain plausible assumptions the universe as a whole must be either contracting
or expanding was confirmed by Hubbles discovery of the red- shift in 1928. We
now know the universe to be expanding; what we dont know is whether it will
contract again or not. More on this shortly.
These successes do not mean that GTR is true; Newton had many successes to
his credit as well. But anyone who favours a generally realist stance with regard
to scientific theories certainly has reason to take the theory seriously, especially in
the absence of serious alternatives. Even if the theory in its current form will one
day be supplanted, its core insight that gravitational effects are manifestations of
spacetime geometry may well be retained.
4
20.6 Equations
Not surprisingly, given that they have to accommodate all the different ways a
four- dimensional continuum can bend and twist, the equations Einstein used to
make these predictions are complex, and can only be fully understood by those
with a thorough grounding in differential geometry; Einstein himself found the
GENERAL RELATIVITY 353
mathematics far from easy.
5
They consist of a set of ten coupled, non- linear, partial
differential tensor equations. The tensor formulation (see 13.4) means the various
quantities represented are independent of any particular coordinate system; the
tensor formalism also allows the equations to be expressed in very concise ways.
One such is:
G = 8tT
On the left- hand side of the equation, the Einstein tensor, G, describes the
curvature of spacetime in a metrical form. On the right the stress- energy tensor,
T, describes the distribution, density and pressure of mass- energy and momentum
at the same region of spacetime.
6
We know from STR that matter and energy are
interconvertible, and various non- gravitational fields (such as the electromagnetic)
also contribute to the energy density but to keep things simple I will often over-
look these complications and refer simply to matter distributions. So in slightly
less schematic form we have:
[Spacetime curvature] = 8t [Mass- energy and momentum density]
The equations apply to every spacetime point they are field equations and tells
us that, for any point P, the curvature in Ps immediate neighbourhood is related
to the mass- energy and momentum density around P. In effect, the equations estab-
lish a law- like connection between the intrinsic geometry of spacetime and the
distribution of mass- energy through spacetime. Or, as Wheeler succinctly puts it,
matter tells space- time how to bend, and space- time tells matter how to move.
Needless to say, given their complexity (concealed by the schematic forms used
here) finding exact solutions for the equations is a formidable task. In practice,
physicists usually make drastic simplifying assumptions (e.g. that mass is uniformly
distributed) in order to render them manageable.
7
Wheelers formulation could mislead. It should not be taken to mean that mass
causes spacetime to curve. Einsteins equations do no more than posit a system-
atic relationship between spacetime curvature and distributions of matter: certain
distributions of matter can only exist with certain spacetime curvatures. There is
no one causing the other here (or at least, the theory is agnostic on this issue);
there is simply a constraint on the way physically possible worlds that conform
to GTR might be.
20.7 Relativistic cosmology
When it comes to the actual geometry of the universe, Einsteins equations tell us
less than we would like to know, for there is no oneone correspondence between
global (or cosmic) matter- distributions and spacetime structures. The same distri-
bution of matter can exist in spacetimes of different curvatures and topologies. To
solve the equations at all, we need to assume certain boundary conditions (such
as whether the universe is finite); the equations themselves do not tell us which
boundary conditions apply in the real world, and different boundary conditions
yield different geometries for the same matter distributions. More generally, in
exploring the ramifications of GTR it is sometimes useful to consider universes
354 TIME AND SPACE
very different from our own (e.g. entirely filled with a uniform dust and rapidly
rotating); any possible universe whose matter distribution and spacetime geometry
conforms to Einsteins equations is said to be a model of GTR. There are an infinite
number of these cosmological models, and at present we do not know which
model best corresponds to our universe. Although recent astronomical results have
allowed us to rule out certain classes of models, many possibilities remain open.
Black holes
However, before turning to the implications of GTR for the general shape and
structure of the universe, mention should be made of the theorys most notorious
by- product. In 1916 Schwarzschild used an exact solution of Einsteins equations
to model the warping of space and time around a spherical star. The Schwarzschild
solution clearly implied that if a concentration of mass exceeds a certain critical
value, the spacetime distortion will be such that nothing, not even light, will
ever be able to escape from it.
8
Anyone unfortunate enough to cross the event
horizon surrounding one of these concentra tions would never be able to return.
It was a while before the physics community settled on a name for objects in this
condition compressed stars, dark stars and frozen stars were among the
early candidates but the term black hole, coined by John Wheeler in 1967, was
the one that stuck. Stephen Hawking has called them possibly the most mysteri-
ous objects in the universe (Thorne 1994: 11), despite the fact that a good deal
has subsequently been discovered about their properties, with Hawking himself
being a notable contributor. Interest in the phenomenon started to increase after
1930, when Chandrasekhar showed that stars with a mass above what is now
called the Chandrasekhar limit should collapse into black holes when they run
out of nuclear fuel to burn. Black holes were no longer a merely theoretical pos-
sibility. Evidence for their existence has accumulated rapidly in recent years, and
it may well be that super- massive black holes with billions of times the mass
of the average star can be found at the heart of most galaxies.
Of all the conundrums that black holes pose, perhaps the most interesting is
the fate of the matter inside them. Does it reach a state, finite in volume, beyond
which it cannot be compressed any further, or does it keep on collapsing until a
singularity a dimensionless point of infinite density is formed? Since current
physical laws break down (do not apply) to circumstances where infinite ener-
gies come into play, most physicists would have preferred the former answer, but
it was not to be: in 1964, Penrose published a singularity theorem to the effect
that all black holes must contain singularities. Subsequent work over the next
decade convinced most physicists that this is indeed the case. Since a singularity
is not a point in spacetime (at least in the usual sense), matter falling into a black
hole vanishes from the universe altogether.
9
The shape of space
Let us move on to the more general cosmological implications of GTR. The most
significant assumption made in relativistic cosmology is that the universe is
homogene ous and isotropic.
10
This so- called cosmo logical principle a latter- day
GENERAL RELATIVITY 355
formulation of the Copernican principle that our location is in no way special
initially strikes one as highly unrealistic. How can a universe in which matter is
concentrated into stars, galaxies and clusters of galaxies (not to mention black
holes), all separated by huge tracts of empty space, be considered homogeneous?
Size is the solution: at scales measured in hundreds or thousands of light- years, the
distribution of matter is statistically uniform. And although a typical star contains
a considerable quantity of matter compared with a typical human body, given the
scale of cosmological distances (driving to the nearest star at normal motorway
speeds would take millions of years) individual stellar matter concentrations have
insignificant effects on the large- scale curvature of spacetime. (Think of the effect
of a grain of sand on a taut trampoline.) The Cosmic Background Explorer (COBE)
satellite data on the microwave radiation that pervades the universe usually
supposed to be a remnant of the big bang supports the isotropy assumption; the
temperature of the universe seems very much the same (to one part in a hundred
thousand) in every direction.
The cosmological principle has an important geometrical consequence. If the
distribution of matter is uniform, then spacetime (and so the universe) must have
a constant average curvature. This constant curvature can be positive, negative or
zero. No other options exist. While this result simplifies things considerably, many
possibilities remain open. Quite how many possibilities was not at all obvious
when Einstein formulated GTR much of the relevant mathematics had not been
done and is still a topic of some controversy today.
When Einstein was formulating his field equations it was widely assumed the
universe was static. The available astronomical data supported this assumption:
the relative motions of the stars appeared to be small (the existence of other
galaxies had still to be established). But a static universe posed a problem for
Einstein. His equations told him that a universe would not remain static for long,
since gravitational effects would lead to an overall contraction; portions of mat-
ter would move towards one another, due to the curvature in spacetime that
they themselves create. Determined to save appearances, Einstein introduced an
additional factor into his equations, the cosmological constant, A (or sometimes
the lower-case ), which was used to represent a repulsive influence whose value
is fixed so as to precisely counterbalance the attractive effects of gravity. The
resulting theory of the overall shape of the universe which appeared in Einsteins
1917 paper Cosmological Considerations on the General Theory of Relativity
had an appealing simplicity. The universe Einstein described was of uniform
positive curvature and closed: finite in volume but without boundaries or edges.
Recall the example of Flatlanders confined to the surface of a sphere; the spatial
component of Einsteins universe is analogous: it is the three- dimensional surface
of a four- dimensional hyper sphere. (Remember, too, that, as in the Flatland case,
there is no interior to the sphere.) Despite its appeal not the least of which was
the solution to the long- standing problem of how space could be finite without
possessing a boundary that someone could easily cross this cosmological model
was not entirely satisfactory, as Einstein was well aware. Not only does the cosmo-
logical constant have the air of an ad hoc face- saving device, but it has a peculiar
feature in its own right. Other known forces (such as magnetism) become weaker
with distance, but the A-factor becomes stronger with distance.
356 TIME AND SPACE
For a brief time Einstein hoped that GTR, when combined with the homogeneity
and isotropy assumptions, would have just one cosmological solution. These hopes
were soon dashed. Other solutions all compatible with the constraints deriv-
ing from the cosmological principle soon appeared. In the 1920s the Belgian
cleric- mathematician Georges Lematre, and the Russian geo- physicist Aleksandr
Friedmann, discovered a variety of non- static solutions to the GTR equations; some
of these models were of expanding universes, others of contracting universes,
some employed a cosmological constant, others did not. Moreover, a serious flaw
in Einsteins initial model was discovered by Lematre in 1927. It turned out that
Einsteins static model, using a A- factor that precisely cancels gravitational attrac-
tion, was radically unstable: the slightest irregularity in matter distribution would
result in expansion or contraction. So even if Einsteins model did apply, our
universe would be unlikely to have remained static for long; it would be expanding
or contracting. But the most influential discovery came from astronomy. In 1929
Hubble published convincing evidence, based on the red- shift of distant galaxies
also observed, but not effectively publicized, by Slipher some fifteen years earlier
that the universe is actually expanding. The rate of expansion is fast (at least by
terrestrial standards), and increases with distance; the more distant a galaxy, the
faster it is receding. The discovery by Slipher and Hubble led to renewed interest
in the non- static models of Friedmann and Lematre, not to mention the latters
deduction that an expanding universe must have originated from a far smaller
cosmic egg: what we now call the big bang singularity. Einstein embraced the
expanding model of universe, and repudiated the cosmological constant calling it
his greatest mistake not surprisingly regretting having missed the opportunity
of going down as discoverer of the dynamic cosmos.
11
By the end of the 1930s relativistic cosmology had stabilized somewhat. Most
physicists (with the notable exception of Lematre) followed Einstein in rejecting
time-axis
big bang
15 billion years ago
8 billion years ago
A
B
the universe today
Figure 20.6 The visible universe. The teardrop shape shows the region of the expanding universe
that is visible from present day Earth (located at the intersection of the two spatial axes). Light emitted by
objects outside this region has not had time to reach us yet. Light from galaxy A has taken 8 billion years
to reach us, so we see it as it was 8 billion years ago. Galaxy B will not be visible from Earth for another
8 billion years or so.
GENERAL RELATIVITY 357
the cosmological constant; there were still those who resisted the idea that the
universe originated in a big bang, but it was widely assumed that it did, and so
attention turned to other questions.
12
How old is the universe? What is its rate of
expansion? Is its curvature positive, negative or zero? Is it open or closed? The
answers to these questions depend on a number of interrelated factors, not least
of which is the average density of mass-energy in the universe, which it is not easy
to measure with any precision. If the density is above the critical value, O (or
W), the universe will eventually slow to a halt and then collapse back in on itself
to a big crunch; if the density is less than O it will keep on expanding forever.
13
Estimates of the mean density of luminous matter in the universe based on the
measured numbers and locations of the stars and galaxies in those parts of the
universe that we can observe suggest that all the luminous matter taken together
only adds up to around a half of 1 per cent of the critical density O. If these esti-
mates are even approximately accurate, then either O is much less than 1 and the
universe will go on expanding forever, or the vast bulk of the mass (or energy) in
the universe does not take the form of luminous stars or gas clouds. More on the
possibility of so-called dark matter shortly.
Figure 20.7 shows some of the more influential cosmological models. A and B
are the closed and open Friedmann (or FriedmannRobertsonWalker) models,
of positive and negative curvature respectively. C depicts the 1932 EinsteinDe
Sitter model, which was a popular model for some sixty years but currently out of
favour (it cannot be reconciled with current estimates of the age of the universe).
This model posits a flat and open spacetime, with a steadily decelerating rate of
expansion. D shows Lematres (1931) hesitating universe model: after an initial
fast expansion, the universe (whose curvature is positive) slows for a while before
accelerating again. Variations on this model which alone of the four includes a
cosmological constant have recently become more popular in response to discov-
eries concerning the rate at which the universe is expanding. But these discoveries
properly fall into the next part of this brief history of relativistic cosmology.
S
i
z
e
Time
big bang now big crunch
A
B
C
D
Figure 20.7 The standard big bang models. The different models have different implications for both
the distant future and the distant past (Luminet 2001: 64).
358 TIME AND SPACE
Dark matter
Recent decades have seen significant advances in our understanding of various
cosmological issues, but the new discoveries that led to these advances have also
revealed new puzzles, problems and unknowns. So far as the overall or global
curvature of the universe is concerned, cosmologists are confident that they now
know the answer. It is neither positively curved nor negatively curved, but flat; or,
at least, very close to it it is estimated that O = 1, to an accuracy of within 1 per
cent. This unambiguous answer emerged from detailed scrutiny of the precise data
gathered from the Nasas Wilkinson Microwave Anisotropy Probe (WMAP) in the
years since its launch in 2001. This satellite has allowed small irregularities in the
microwave background radiation to be studied with unprecedented precision. The
zero curvature result first emerged in 2003, and was confirmed by further studies
announced in 2006.
The result that O = 1 was welcomed in some circles, since it agrees with the
predictions of the widely accepted inflationary model. By the 1970s the big bang
hypothesis was seen to be problematic on a number of fronts. In particular, it was
by now clear that energy is distributed throughout the universe in a very smooth,
homogeneous manner. Globally speaking, ignoring small local concentrations (i.e.
the sparsely distributed galaxies), it is precisely the same in every direction. This
is a surprise. If the universe began in a vast, uncontrolled explosion, one might
expect mass and energy to be distributed in a very varied, almost random, manner.
As Guth and others pointed out in the early 1980s (see Guth 1997), this problem,
along with several others, is solved if we suppose that the presently observable
universe derives from a very small (and so homogeneous) region of space that
underwent very rapid expansion shortly after the big bang. The time scales in the
various inflationary scenarios are short, at least by ordinary human standards: the
period of rapid expansion may have started as little as 10
40
seconds after the big
bang, and come to a halt before the universe was one second old; during this brief
time the universe may have doubled in size a hundred times over. Importantly,
over the course of this rapid expansion in size, any initial curvature in spacetime
is all but eliminated.
The results from the microwave background surveys that confirmed that the
universe is flat, or very nearly, were thus welcome, but they also gave rise to a
problem. For as noted earlier, estimations of the energy density of the universe
based on the observed numbers and sizes of stars, galaxies, gas clouds and the like,
all suggest a value for O that is as little as 0.005, or around a half of 1 per cent of
1. So the universe may be flat, but it is not clear how it can be.
The only obvious solution assuming our current theories and observational
data are correct is that there are other sources of mass or energy, sources that
we have been unaware of thus far. Since this missing mass and energy exists
over and above that of all the luminous sources we are able to observe directly, it
is unsurprisingly classed as dark.
One of the first firm indications that dark matter might exist in considerable
quantities emerged from the studies of rotating galaxies carried out by Vera Rubin,
and announced in 1975. In our own solar system, the more distant planets for
example Jupiter, Uranus, Neptune orbit the sun at a slower rate than those that
GENERAL RELATIVITY 359
are closer to the sun, such as Mercury and Mars. The reason for this was discov-
ered long ago, by Kepler. For a small object orbiting a far larger one, the orbital
speed of the object is proportional to two factors: the mass of the large central
body, and the radius of the orbit. The more massive the central body, the faster
the orbital speed; the larger the radius of the orbit, the slower the orbital velocity.
(More specifically, orbital speed is inversely proportional to the square root of the
orbital radius.) The stars and dust clouds in the outlying regions of disc-shaped
or spiral galaxies such as our own Milky Way are also in orbit: they trace out
curved, near-circular paths around their heavily populated galactic centres, which
are filled with dense concentrations of stars (and sometimes super-massive black
holes). Theory predicts that stars that are in orbit around these bulges of densely
packed nuclear stars will conform to Keplerian dynamics, and hence that their
orbital speeds will decrease the further away from the galactic centre (or nucleus)
they happen to be. Surprisingly, this does not in fact occur: the stars in the more
peripheral regions of a galaxy have much the same orbital speed as stars closer to
the centre. When orbital speed is plotted against orbital radius, rather than the
speed dropping off as radial distance increases, it remains flat (see Figure 20.8).
The most economical explanation for this divergence from predicted behaviour
is that the visible stars in a galaxy are surrounded by a spherical layer, or halo,
of invisible dark matter, lying some distance away from the galactic cores. Obser-
vations of the orbital behaviour of the stars in our own Milky Way, and evidence
from a variety other sources, combine to suggest that as much as 90 per cent of
the galaxys total mass exists in the form of dark matter, and the latter is assumed
to lie in a halo situated up to 500,000 light-years from the galactic centre. (To
put this in some sort of perspective: the sun lies some 27,000 light-years from the
centre, and it takes it around 225 million years to complete its near-circular orbit
at a speed of approximately 230 kilometres per second; all the luminous mass in
the Milky Way adds up to around 120 billion solar masses the estimated dark
matter is of the order of a 1000 to 1500 billion solar masses.) Observations suggest
that other spiral galaxies possess a broadly similar ration of ordinary-to-dark mat-
ter. The situation is less clear-cut in the case of the so-called elliptical galaxies:
since these lack a simple, well-defined pattern of rotation, arriving at an estimate
of the dark:luminous ratio is more difficult, but there is evidence that the larger
ellipticals also possess significant quantities of dark matter in their composition.
Figure 20.8 The problem of missing
mass in rotating galaxies. Rather
than dropping away with distance, as
predicted by theory, the orbital velocity of
stars in outlying regions remains constant
or flat: the stars that are further away
from the galactic core are rotating at the
same speed as those closer to the core.
distance from galacc centre
orbital
velocity
bulge periphery
actual
predicted
360 TIME AND SPACE
Evidence for dark matter deriving from the rotation patterns of stars within
individual galaxies may only have emerged in the 1970s, but evidence pointing
in the same direction deriving from studies of entire galactic clusters was first
obtained some years earlier: from studies on the Coma cluster (of some 2000
galaxies, some 300 million light-years away) carried out by Fritz Zwicky in 1933.
After measuring the Doppler shifts of individual galaxies within the cluster, which
allowed him to estimate their velocities, his calculations showed that the clusters
must contain a good deal of mass over and above that possessed by the visible stars
they contain: something like fifty times as much! This led Zwicky to propose that
the Coma cluster contains a vast amount of very dim, or entirely dark, matter.
Subsequent studies of galactic clusters have generally confirmed Zwickys result:
it is now estimated that the ordinary mass in a typical cluster mainly made up of
the stars in luminous galaxies along with inter-stellar and inter-galactic dust only
contributes around 10 per cent of the total mass.
The discovery of dark matter gives rise to at least two pressing questions. First
and most obviously, what is dark matter? What is its nature and composition?
Secondly, is there is enough dark matter to account for the geometry of the uni-
verse? For as we have seen, the total amount of ordinary matter (and energy) in the
universe falls far short of what would be required to achieve the critical density,
at which O = 1; if it is indeed the case that hence O = 1, is there enough dark
matter in the universe to make up the shortfall?
The simplest and most straightforward answer to the first question is that dark
matter is nothing more than ordinary baryonic matter (i.e. matter that is mainly
composed of protons and neutrons) in a form that is non-luminescent, and so dif-
ficult or even impossible to observe. Astronomers had little difficulty coming up
with suitable candidates: for example, low luminosity red and brown dwarf stars,
large dark planet-like objects, and so forth. In dark matter circles, these hard-to-see
objects became known as Massive Compact Halo Objects (MACHOs). But while
MACHOs do exist, it soon became clear that they can only form a comparatively
small part of the total amount of dark matter. Evidence from studies of lensing
effects suggested that only 20 per cent or so of the dark matter in galactic halos can
be composed of MACHOs. More serious still, the leading theoretical model of big
bang nucleosynthesis the process by which hydrogen gets converted into heavier
elements shortly after the big bang entails that baryonic matter only contributes
45 per cent to the total amount of mass-energy in the universe. Analyses of
recent WMAP data point in the same direction: they suggest that baryonic matter
contributes only around 4 per cent of the critical density. If ordinary baryonic
matter is ruled out, what fills the gap? Theoreticians currently favour cold dark
matter (CDM), in the form of weakly interacting massive particles (WIMPs); these
were forged in the early universe, and do not interact to a significant extent with
baryonic particles. As for the precise identity of the non-baryonic ingredients,
there is as yet little consensus, but some of the particles predicted by extensions to
the Standard model of particle physics for example axions, neutralinos would
seem to fit the bill.
GENERAL RELATIVITY 361
Dark energy
What of our second question: is the combination of dark and baryonic matter
enough to explain the flatness of the universe? Given the comparative paucity of
ordinary matter, is there enough CDM to render O equal to the critical density?
Although in the 1980s and early 1990s cosmologists hoped that this was the case,
these hopes were dashed by the WMAP data: the precise pattern of irregularities
in the microwave background is compatible with CDM making up only 22 per
cent of the critical density. With ordinary matter contributing only 4 per cent, we
have a significant and potentially alarming shortfall: some seventy-four per cent
of the mass-energy of the universe is missing! If the total mass component of the
critical density cannot exceed around twenty-six per cent, the remainder must,
presumably, exist in the form of energy. But what kind of energy? Well, since it
is not visible in the usual way, we can conclude that it is dark energy. While apt,
this label does not take us any further, for we are no wiser as to what dark energy
really is.
Assistance with this problem was forthcoming from another direction. As we
have seen, the various cosmological models on offer yield different predictions
as to the ways in which the contents of the universe behave after the big bang.
In the early 1990s, it was widely believed that the energy density of the universe
was such as to either (i) stop the expansion and cause a contraction and eventual
re-collapse, or (ii) be insufficient to bring the expansion to a stop, but nonetheless
be enough to slow the expansion down. Arriving at ever more accurate estimations
of the energy density is one way of shedding light on this issue, but there are other
avenues that can be explored.
Thanks to the finite speed of light, when we observe distant galaxies, we are in
effect looking into the past; when we observe very distant galaxies, we are seeing
the universe as it was billions of years ago. Further, the Doppler shift in the light
we receive from these galaxies allows us to judge how fast they are receding from
us, which is itself a measure of the extent to which space has stretched or expanded
since the light was emitted. In virtue of this, careful astronomical observations
of galaxies at a range of distances should allow us to gauge the rate at which the
universe has been expanding, and whether the expansion rate has been constant
or variable, over the past few billion years. Unfortunately, there is an obstacle: for
the observations to be used in this way, we need to know how far away from us
the galaxies in question are, and arriving at reliable and accurate measures of such
distances is not easy. We could measure the distance separating us from a distant
galaxy if we had what astronomers call a standard candle, that is, if there were
one or more objects in the galaxy that are bright enough to be seen from Earth,
and we also know, in absolute terms, how much light they are emitting. In the
1930s, Fritz Zwicky suggested that supernovae (exploding stars that can briefly
outshine entire galaxies) might be able to serve as standard candles. But since
further study revealed that the power outputs and spectral emissions of novae can
vary a great deal, nothing came of the suggestion. As more was learned, however,
the situation changed: it turns out that a particular kind of supernova, Type IA, are
always of the same brightness, because they always involve the burning of the same
amount of matter in the same sort of way; they are exploding white dwarf stars,
362 TIME AND SPACE
which collapse catastrophically on achieving 1.4 solar masses. The discovery of
a standard candle was a significant advance, but problems remained: supernovae
are spectacular events, but very rare ones, which means large numbers of galax-
ies large areas of the sky have to be observed over long periods if we are going
to detect significant numbers of them. By the mid-1990s these difficulties had
been overcome, and several teams set about detecting Type IA events. When the
first clear results arrived in 1998 they were a shock: the IA-type novae in distant
galaxies were systematically fainter than was expected that is, the light being
received from them was fainter than it would have been if the universe had been
expanding at a steady rate, or decelerating in the expected way. Once alternative
explanations had been ruled out for example, dust clouds that absorb some of
the light only one viable explanation remained: the light was up to 20 per cent
fainter than expected because it had travelled further than expected. And the only
obvious explanation for this: the expansion of the universe was accelerating, not
decelerating. Further studies, on more and different supernovae, have confirmed
these early results (for further detail see Kirshner 2002).
If the expansion rate of the universe is increasing, there must be something driv-
ing or fuelling this acceleration: some repulsive or anti-gravitational influence. To
do the job required this repulsive influence requires some special features. It must
extend through all of space in a homogeneous manner; it must also be able to exert
an influence on the global structure of spacetime without interacting with local
concentrations of matter (as found in our solar system, or galaxies); furthermore,
its power must not diminish as the space that contains it expands. Cosmologists
were very familiar with just such a driving agent: the cosmological constant, A, that
Einstein had included in his early version of GTR to counteract gravitational attrac-
tion has precisely these features. It was starting to look as though the inclusion of
the constant was not the great blunder Einstein feared, but an act of prescience.
The benefits of restoring A to the equations of GTR do not end there: we now
have a plausible candidate for the mysterious dark energy that (seemingly) makes
up 74 per cent of the universe. It so happens that the cosmological constant is
itself associated with a distinctive form of energy. From Einsteins STR we know
that matter and energy are inter-convertible, and according to GTR all forms of
matter and energy influence gravity, by constraining the geometry of spacetime.
According to GTR, pressure is one of the factors along with mass, radiation,
and so forth that contribute to the overall energy density of a system. GTR also
allows for pressure to come in two forms. Ordinary pressure, of the outward-
pressing from (e.g. the force exerted by compressed air in a can), is known as
positive pressure, and its effect on gravity is similar to ordinary matter; in effect,
it generates gravitational attraction. There is also what is known as negative pres-
sure, which takes the form of an inward-directed tension, as found in a stretched
rubber band, or spring. The gravitational effect of negative pressure is the oppo-
site of that of positive pressure; it is repulsive rather than attractive. Given this,
the cosmological constant can be regarded as being akin to a smoothly distrib-
uted tension in the fabric of spacetime, and hence as a form of negative pressure
whose gravitational influence is repulsive. Crucially, as a form of pressure, it can
contribute to the total mass-energy of the universe. Calculations show that the
energy associated with the cosmological constant is sufficiently massive that it
GENERAL RELATIVITY 363
can easily constitute the required 74 per cent of the critical density, O. Thus
construed, dark energy can account for both the flatness of the universe and the
accelerating rate of expansion.
14
There are, of course, ingredients in O in addition to dark energy: around 4 per
cent of ordinary baryonic matter (a small portion in the form of stars, more in the
form of dust and gas), and around 22 per cent of dark matter of the CDM variety (if
current mainstream thinking is correct). The hypothesis that the universe is largely
made up of cold dark matter and dark energy, in the proportions just mentioned, is
sometimes referred to as the ACDM model, the assumption being that dark energy
is to be explicated in terms of the cosmological constant, A. So accurately does
ACDM accommodate the recent astronomical observations that it is also known as
the concordance model of big bang cosmology. It is the new standard cosmologi-
cal model. Needless to say, this situation may well change; relativistic cosmology
has seen so many dramatic developments over the past couple of decades that it
would be foolhardy to rule out the possibility of further revelations or discover-
ies, and these could occur sooner rather than later (more on this in a moment).
Nonetheless, as things currently stand, there is no denying that ACDM does suc-
ceed in accounting for the otherwise problematic empirical findings very well:
hence we certainly cannot be confident that the universe is not largely composed
of dark energy and dark matter. If this turns out to be the case, the Copernican
revolution and the concomitant de-centering of humankind will have been car-
ried a stage further. We have known for some time that our spatial location in the
grand scheme of things is nothing special: our planet is in orbit around an average
star in a typical galaxy. If current thinking is correct, we may have to get used to
the idea that 95 per cent of the universe is composed of matter and energy that
are quite unlike the forms with which we are familiar; the forms we have always
(and quite naturally) assumed constitute everything else that is to be found in the
physical universe.
The horizon is not entirely bleak for anyone who finds this an unwelcome
result. Much of the data ACDM accommodates so well the recent findings regard-
ing the precise relative proportions of dark and ordinary matter, and (crucially)
the accelerating rate of expansion are in many cases founded on theories and
assumptions that may yet prove to be false or unfounded, and delicate observations
that may prove to be unreliable or misleading. This general point aside, there is a
more specific worry: reconciling dark energy in the guise required by cosmology
with quantum theory has proved difficult. On the face of it the prospects of a suc-
cessful marriage look promising: we have seen that A can be viewed as energy pos-
sessed by spacetime itself, and quantum theory also features a vacuum energy,
Figure 20.9 The composition of the
universe according to the currently
popular ACDM (or concordance)
cosmological model. The smallest
(unlabelled) segment of the chart
represents the 4 per cent of the
universe that is composed of
ordinary matter.
74% dark
energy
22% dark
matter
364 TIME AND SPACE
owing in large part to the short-lived virtual particles that are a consequence of
Heisenbergs uncertainty principle. Unfortunately, the estimated strength of the
quantum mechanical vacuum energy is some 10
120
times greater than that required
to explain the observed expansion of the universe! This is a serious discrepancy,
to put it mildly, and one that continues to baffle physicists.
15
There are also a number of alternative explanations of the data that are actively
being explored. When a serious conflict arises between a successful and long-
established theory and new observational data, it is always tempting at least for
those who have grown so used to the theory they find it difficult to contemplate
any alternative to find a way of hanging on to the theory while accommodating
the new data, even if so doing means positing the existence of as yet unobserved
objects. The strategy is sometimes successful. Urbain Le Verrier explained the dis-
crepancy between Newtonian theory and the observed orbit of Uranus by predict-
ing the existence of an additional planet. He turned out to be correct: the planet
in question Neptune was soon found, just where Le Verrier had predicted. But
the tactic does not always work. Le Verrier posited yet another new planet, Vulcan,
to explain the discrepancies in Mercurys orbit. But no such planet was found:
Einstein went on to show that it was Newtons theory of gravity that was at fault.
Bearing this in mind, rather than positing vast amounts of unobservable dark
matter to explain the divergence from predicted results for example in the
rotation of galaxies we should be open to the possibility that it is our current
theory of gravity that should be rejected or modified. Perhaps the best-known and
most promising proposal along these lines is the Modified Newtonian Dynamics,
(MOND) first elaborated by Mordehai Milgrom in 1983, in response to the puzzling
rotation curves for spiral galaxies (see Figure 20.8). The idea that we can avoid the
need to posit dark matter by modifying our standard theory of gravity was by no
means novel, even in 1983. An early suggestion was that the strength of gravity
declines less than Newtons inverse square law predicts, albeit over very large (i.e.
galactic) distances. However, this proposal entails that the gravitational behaviour
of very large galaxies should deviate from what Newtonian theory predicts to a
greater extent than smaller galaxies, and observations do not back this up: there
are large numbers of small galaxies which also behave as though they include a
lot of missing mass. Milgroms proposal circumvents this problem. For according
to MOND, it is not distance that is the crucial factor, but rather acceleration. If an
object is accelerating at a rate that is greater than a so-called transition factor,
then gravity behaves as Newton decreed: its force is inversely proportional to the
square of the distance between bodies. But if the rate of an objects acceleration
is less than the transition factor, then gravity is inversely proportional only to the
distance, not the distance squared. In this case, the force of gravity is halved when
the distance is doubled, rather than quartered. More relevantly, whereas in the
Newtonian case the speeds of orbiting bodies fall rapidly away with distance (at
a rate inversely proportional to the square root of the distance), in systems where
MOND-type gravity governs, the speed of such bodies remains flat (or constant),
irrespective of their distance from the central mass.
16
If gravity were as MOND describes, there would be no need to introduce dark
matter: a considerable saving, ontologically speaking. Since the orbiting objects in
our solar system all have accelerations many times greater than the transition value,
GENERAL RELATIVITY 365
MOND predicts that the gravitational behaviour of these objects will conform to
Newtons laws, which is just what we find. Interest in MOND increased significantly
in 2004, when Jacob Beckenstein developed a relativistic version: Bekensteins
Tensor-vector-scalar gravity theory (TeVeS). This has the added advantage of
applying to photons, and can accommodate the lensing effects created by galactic
clusters as easily as dark matter models; some theorists believe that it may be
possible to account for dark energy within the TeVeS framework. There are other
promising alternatives, for example John Moffats Modified Gravity (MOG) (also
known as scalar-tensor-vector gravity, or STVG), which can accommodate the
galactic rotation curves and the supernovae observations without recourse to dark
matter or dark energy in the form of A (see Moffat & Toth 2007).
Although the modified gravity programme is developing quickly, and showing a
good deal of promise, a great deal of further work needs be done before we will be
in a position to conclude that the Einsteinian account of gravity requires revising
in a fundamental way: MOND, TeVeS and MOG (or one of their kin) may be able to
accommodate some astronomical data, but it is too soon to say whether they will
be able to account for the wide range of phenomena that can be explained within
the framework of orthodox, unmodified GTR. And there are other intriguing pos-
sibilities. According to one recent suggestion (Clifton et al. 2008), the type IA
supernovae results can be fully explained if we suppose that we are located close
to the centre of a cosmic void, a large region of space of the order of size of the
observable universe where the local matter density is atypically low. If we make
this supposition by no means idle, since recent large-scale galaxy surveys (e.g.
the 6dFGS) has found evidence of huge voids the observational evidence can be
accommodated without positing an accelerating rate of expansion, and so there is
no need to posit a dark energy to fuel this expansion.
If the void hypothesis proves viable, we will have an economical explanation
for the otherwise puzzling type IA supernova data. However, this hypothesis too
has profound implications, for embracing it means means rejecting a cornerstone
of cosmology for the past three hundred years: the Copernican doctrine that our
location in the grand scheme of things is entirely typical. If the void hypothesis is
correct, our location is quite untypical. This may be consoling in some respects,
but it has an awkward and potentially disturbing consequence: we will no longer
be able to take local measurements to be reliable guide to how things are elsewhere
in the universe.
17
Topological issues
Despite their various differences, the cosmological models we have considered
thus far share a common feature: all are topologically timid. The spatial com-
ponents of these spacetimes are all simply- connected (in the manner of a plane);
none are multi- connected (in the manner of a torus).
18
Friedmann pointed out the
compatibility of GTR with multi- connected topological manifolds in 1924, but
the suggestion fell by the wayside, despite the fact that Einsteins field equations
are entirely neutral with respect to this issue. However, the prejudice in favour
of simply- connected models lifted in the 1980s and 1990s, and the possibility
that our universe is multi- connected began to be taken seriously.
19
The number
366 TIME AND SPACE
of possible multi- connected spaces of constant curvature (flat, positive and nega-
tive) is vast, and many are highly complex. But even the simpler cases (such as the
hypertorus) have the potential for creating mischief, in the form of topologically
generated ghost images of nearby galaxies.
The shaded circle in Figure 20.10 represents the region of the universe that is
observable; objects outside this region are so distant that the light they emit has
yet to reach us, so we cannot detect them (see Figure 20.6). The central square
with a bold boundary represents the entirety of the actual universe (only a few
stars and a single galaxy are depicted); the remainder of the squares are filled with
ghost images produced by light from the same few stars and galaxies weaving
endlessly back and forth through the multi- connected spacetime manifold before
being detected by telescopes on Earth.
In one respect this depiction is misleading. Since a galaxy emits light in all direc-
tions, although we would always be observing the same astronomical objects, we
would be observing them from many different angles: we would see each galaxy
from behind, below, the right, the left (relative to the Earth), and all angles in
between. Also, since light from a given object can make many passes through a
multi- connected space before arriving in our neighbourhood, the heavens will be
filled with images of the same galaxies at different stages of their career, and in
different configurations. In principle, this opens up the possibility of seeing our
own sun and solar system as they were at different points in the past. All we need
do is detect a ghost image of our own region of the galaxy, which originated in
the desired historical period, and point a sufficiently powerful telescope in the
appropriate direction. Unfortunately, if Gott is correct (2001: 80) in his calcula-
tions (based on assuming a toroidal topology), we can expect the closest image of
our galaxy to be 5 billion light- years away, and hence 5 billion years old a billion
years before the Earth itself formed.
In combination, these factors make the ghosts very difficult to detect. It
would be far from obvious to Earth- bound astronomers that they are observing
Figure 20.10 Topological ghost
images on a cosmic scale, creating
an illusion of infinity. The night sky of
this small toroidal universe is filled with
repeated images of the same few stars
and galaxies (Luminet 2001: 116).
GENERAL RELATIVITY 367
repeated images of the same galaxies. Nonetheless, in principle it should be pos-
sible to ascertain whether this is in fact the case. Various methods of detecting the
distinctive signatures imposed by multi- connected topologies on galactic distribu-
tions have been proposed, and the search for these patterns is being pursued by
a number of cosmologists. Although these enquiries are still in their preliminary
stages, there is no decisive evidence so far that our universe is not multi- connected
and some models (in particular, one that is closed, positively curved and dodecahe-
dral) can explain otherwise puzzling features of the WMAP data-releases (Luminet
2001, 2008).
Interestingly, the possibility of a multi- connected universe was noted by
Schwarz schild (who, as we have seen, would later make significant contributions
to GTR) as early as 1900.
One could imagine that as a result of an enormously extended astronomical
experience, the entire universe consists of countless identical copies of our
Milky Way, that the infinite space can be partitioned into cubes each contain-
ing an exactly identical copy of our Milky Way. Would we really cling on
to the assumption of infinitely many repetitions of the same world? We
would be much happier with the view that these repetitions are illusory, that
in reality space has peculiar connection properties so that if we leave any
cube through a side, then we immediately reenter it through the opposite
side. (Luminet & Roukema 1999: 23)
Schwarzschild is certainly correct about the way we would be inclined to interpret
such findings; the strange geometry hypothesis would have much the greater
appeal. So we have here a further kind of case that undermines Poincars claim
that we would always opt for a cosmological hypothesis that posits the simplest
(i.e. Euclidean) geometry.
A
B
A
B
A
B
Figure 20.11 Winding ways. The length of a path linking two points on the surface of a cylinder
depends on how many times it winds around en route. The same holds in multi-connected spaces; in many
instances there is an additional complication: there may be more than one route between any two points
(in a toroidal space, for example, the windings can go in two opposite directions). Since the speed of
light is finite, ghost images will often be considerably out of date.
21 Spacetime metaphysics
21.1 Substantival spacetime
Newtonian space is both substantival and absolute: able to exist in the absence of
matter, possessed of an immutable Euclidean geometry, it is entirely unaffected
by the presence and distribution of matter within it. The spacetime of GTR is very
different; its geometry is variable, and affected by the presence and activities of
material things. GTRs spacetime is certainly not absolute, but is it substantival?
Einstein certainly came to think it was. In his 1920 Leyden lecture Ether and the
theory of relativity he remarked that the theory has, I think, finally disposed of
the view that space is physically empty (1922: 18). It is not hard to see why he
adopted this view.
We saw in 14.2 and 14.3 that in variably curved spaces for example, worlds
where non- Euclidean holes occur at irregular intervals the relationist faces a
near- impossible task in seeking to explain the behaviour of bodies solely in terms
of objectobject forces and relations.
1
Since the internal tensions suffered by a
body moving through such a hole cannot be explained by alterations in the shape
of space, they must be explained in terms of relations between the body in ques-
tion and other bodies; since the other bodies may be in any number of different
configurations, the laws involved if it proves possible to formulate any will be
horrendously complex.
Since, according to GTR, variations in the geometry of spacetime are related to
the movements of matter, areas of strong localized curvature sitting motionless in
empty tracts of spacetime are unlikely to be encountered. However, GTR does posit
dynamic equivalents of the non- Euclidean hole, which prove just as problematic
for the relationist, the most dramatic example of which is the phenomenon of
gravity waves. Large rotating concentrations of mass (e.g. a pair of black holes)
generate outwards spiralling ripples in the fabric of spacetime, and these mobile
geometrical disturbances of varying curvature affect the shape and trajectories of
any material objects they encounter (although the waves themselves are unaffected
by any matter they pass through). A gravity wave traversing the Earth stretches
and squeezes it in directions at right angles to its own. Moreover, the precise way
a gravity wave affects a material body will depend on the geometry of the local
spacetime, which varies from place to place depending on local matter distribu-
tions. Is it possible to explain the subtle and complex motions in material systems
that would be produced by a gravitational wave (or worse, many such waves act-
ing simultaneously) solely in terms of forces and laws that relate material bodies
SPACETIME METAPHYSICS 369
themselves? Perhaps, although it would not be easy. But even if it could be done,
the resulting theory would bear no resemblance to GTR. It would be an altogether
different type of theory.
The way in which GTR makes use of variations in the geometry of empty space-
time regions in explaining the motions of bodies renders relationist reinterpreta-
tion highly problematic. The spacetime manifold is not just a con venient heuristic
device; it is an agent in its own right.
2
We would be led to the same conclusion
if cosmologists were to discover evidence that our universe is multi- connected:
what manner of laws governing objects and their spatial relations could explain
the behaviour of objects in such worlds? It is easier and more plausible by far to
find in favour of a strange but substantival spatiotemporal manifold.
There is a further fact, specific to the spacetime of GTR, that seems decisively to
confirm its substantival character: gravitational energy. The following description
of the fate of a binary system comprising two black holes provides an illustration of
the way energy can be transmitted back and forth between objects and spacetime:
As they depart for outer space, the gravitational waves push back on the
[black] holes in much the same way as a bullet kicks back on the gun that fires
it. The waves push drives the holes closer together and up to higher speeds;
that is, it makes them slowly spiral inward towards each other. The inspiral
gradually releases gravitational energy, with half of the released energy going
into the waves and the other half into increasing the holes orbital speeds
the closer the holes draw to each other, the faster they move, the more
strongly they radiate their ripples of curvature, and the more rapidly they
lose energy and spiral inward. (Thorne 1994: 3589)
Here we see objects transferring part of their energy into spacetime, which then
transmits it at light speed through empty space in the form of gravitational waves.
How can something that can behave like this be anything other than substantival
in nature?
21.2 Machs Principle
The question of whether the spacetime of GTR is substantival or not is sometimes
addressed by focusing on a different but related issue: does GTR confirm Machs
Principle? It will be recalled that, in the context of Newtonian dynamics, Mach
asserted that there was no need to posit an absolute space to explain inertial
phenomena, for the latter could all be explained solely in terms of relationships
between material things (e.g. the water rises in Newtons rotating bucket because
it is moving relative to the centre of gravity of the universe, not absolute space).
Transferred into the context of GTR, Machs Principle can be restated in a weak
(WMP) or strong (SMP) form, depending on whether or not we include only inertial
structure (which Mach himself was primarily concerned with), or all geometrical
features.
WMP: The inertial structure of spacetime is uniquely determined by the
distribu tion of mass- energy through all of spacetime.
370 TIME AND SPACE
SMP: All geometrical features of spacetime are uniquely determined by the
distri bu tion of mass- energy through all of spacetime.
As the stronger principle, SMP entails WMP.
3
It may well be that SMP must be satisfied
if relationism is to be viable, but the converse does not follow. Even if GTR were
Machian in this strong way, the properties it ascribes to spacetime are such that it
is hard to see how it could fail to be substantival. The fact that an entitys structure
is entirely dependent on how some other entities are distributed within it does not
in itself entail that only the latter entities exist. As it happens, and despite Einsteins
own hopes, GTR does not satisfy Machs Principle, in either form, for, as already
noted, the same distributions of matter are compatible with very different spacetime
structures. It is worth exploring this point a little further, for the light it casts on the
extent to which spacetime structure and mass distribution are interwoven in GTR.
4
The environment of STR is extremely hostile to the Machian: the distinction
between inertial and non- inertial paths is built into the fabric of Minkowski space-
time, and is entirely independent of the presence or absence of material bodies.
In a universe consisting of just one particle, the distinction between inertial and
non- inertial (straight versus curved) paths and worldlines remains valid. It is not
hard to see why the innovations introduced by GTR raised hopes among Machians.
The theory states that the geometry of spacetime is dependent on the distribution
of material bodies; furthermore, the theory predicts that a stationary object that is
surrounded by a large quantity of rotating matter will experience inertial effects
similar to those that would be produced if the object itself were rotating which
supports Machs contention that inertial effects are entirely due to the relative
motions of bodies. But deeper explorations soon reveal respects in which GTR is
profoundly non- Machian.
The strong form of Machs Principle fails because the same mass distributions
can occur in spacetimes of different topologies. The spacetime that houses a finite
matter system may itself be finite (closed) or infinite (open) the matter distribu-
tion itself does not determine which. So far as we can tell from observations, our
universe is expanding and (at least approximately) isotropic (similar in all direc-
tions), and homogeneous (similar at all points). Ellis (1978) has shown that there
is a model of GTR that is compatible with these observa tions but where spacetime
is inhomogeneous and anisotropic. The matter in this universe pours out of one
singularity and is sucked into another; globally this universe is not expanding, but
it seems to be in the region around the source singularity.
If WMP were true, two consequences would follow. First, universes that con-
tained no matter would be devoid of inertial structure; if inertial structure is deter-
mined by matter distribution, then no matter implies no structure. Secondly, it
would be absurd to suppose that all the matter in the universe could be rotating; or
at least, if the matter- field were rotating it would have no empirical consequences
(since the relations between material things remain the same as in the non- rotating
case), and so the distinction between a stationary and rotating matter- field would
have no physical content. Unfortunately for the Machian, there are models of GTR
that refute both predictions.
In most open GTR universes, regions of spacetime that are a long way from
any matter have the structure of flat Minkowski spacetimes, so it is reasonable
SPACETIME METAPHYSICS 371
to suppose that in an entirely matter- free GTR universe the spacetime would take
the same form. But since in Minkowski spacetime the distinction between inertial
and non- inertial worldlines is well defined in the absence of matter, it would be
here too. And as Sklar notes, GTR allows for stranger possibilities: Spacetime
curvature has its own gravitational self- energy. So it is possible to have non- zero
curvature in an empty universe, or for there to be regions of curved spacetime
whose deviation from flatness is supported by no matter at all but simply by the
self- energy of the curved spacetime region (1992: 78). In empty worlds such as
these the inertial structure is quite definite, but quite different from that of flat
Minkowski spacetime.
There are several solutions to GTRs field equations that posit global rotation.
The Oszvth and Schcking solution characterizes a globally rotating, spatially
closed, homogeneous dust- filled universe; Kerrs solution is for an isolated rotat-
ing mass; in Gdels solution matter is represented as a perfect fluid present
throughout spacetime, which is rotationally symmetric about all points. Although
all the matter in a Gdel universe is rotating, there are experiments that would
reveal that for each spacetime point there is a certain distinctive plane, and light
(or any other particle) sent out along this plane will trace out a spiral path (just
like a marble propelled from the centre of a rotating turntable). Consequently, it
would seem to all observers that they themselves are at the centre of the universal
rotation. This is a bizarre state of affairs, but, as we shall see, the Gdel universe
has still stranger properties.
There are, then, models of GTR that are radically non- Machian in character, in
both the strong and weak sense, and so GTR is not a Machian theory. But there
are also models that are Machian. Closed Friedmann models, which characterize
finite isotropic homogeneous dust- filled universes, are not globally rotating, and
their inertial and metrical structures are uniquely determined by their matter dis-
tributions. They thus satisfy WMP. (Since matter- fields of the same type may also
occur in open spacetimes, they do not uniquely determine topology, and so fail to
conform to SMP.) Since Friedmann models provide good approximations for our
universe they are the standard way of representing big bang cosmologies the
Machian cause is not entirely lost. But it should not be forgotten that, like all GTR
models, these are only approximations. The stress- energy tensor specifies only
the gross, large- scale patterns within the matter- field; local variations are simply
not registered. So given the non- Machian character of GTR itself, there may well
be regions within the observable parts of our universe where WMP fails. And, of
course, it could be that the parts of the universe we can observe are not representa-
tive of the larger picture; if so the models of GTR that correspond most closely to
the global structure of our spacetime may be very different from the Friedmann
models that satisfy the strictures of WMP.
21.3 The hole argument
While it is generally conceded that GTR renders substantivalism (of some descrip-
tion) hard to avoid, the mathematical formalism in which the theory is couched
renders this commitment problematic. In the 1980s the significance of the hole
372 TIME AND SPACE
argument a version of which dates back to Einstein (in about 1913) was rec-
ognized by Stachel, and deployed against orthodox substantivalist interpretations
of GTR by Earman and Norton. Reading the already considerable literature on the
hole argument brings some clarity to how the issue of substantivalism fares in GTR,
as well as new puzzles and perplexities.
As we have already seen, modern spacetime theories start by positing a manifold
of spacetime points and then proceed to add additional structures. In the case
of GTR, the manifold, M, is a collection of points possessing certain topological
properties it is four- dimensional and continuous and the additional structures
consist of a metric field and a matter- field, defined by the metric tensor, g, which
specifies the distances and geometrical relations between points, and the stress-
energy tensor T, which provides an approximate representation of the distribution
of matter and energy. Given that all models of GTR have the three components [M,
g, T] the substantivalist is faced with a question: if spacetime is a substance, which
of these components constitute this substance? Is it the manifold, or manifold +
metric, or is it manifold + metric + matter- field? Since the last answer can be ruled
out it obliterates the distinction between spacetime and matter the real issue is
whether we take spacetime to be M alone or [M + g]. Earman and Norton argue
that the substantivalist should opt for M alone:
The advent of general relativity has made most compelling the identification
of the bare manifold with spacetime. For in that theory geometric structures,
such as the metric tensor, are clearly physical fields in spacetime. The metric
tensor now incorporates the gravitational field and thus, like other physical
fields, carries energy and momentum
If we do not classify such energy bearing structures as the [gravitational]
wave as contained within spacetime, then we do not see how we can consist-
ently divide between container and contained. (1987: 51819)
Traditionally substantivalists have viewed space as a sort of container: it is that
extensive medium that contains the material world. Earman and Norton argue
that if we wish to retain this fundamental idea in the spacetime framework of
GTR we must identify the container (space) with the bare manifold M, for if
we include the metric we no longer have a sharp distinction between spacetime
and matter, between container and contained. According to GTR the metric field
contains energy, and hence has a material aspect since matter and energy are
inter convertible.
So, let us suppose that the substantivalist adopts this view: spacetime is identical
with the manifold of points. A problem now arises.
GTR is a generally covariant theory, which (roughly) means that the mathemati-
cal form of the laws of physics are the same no matter which system of coordi-
nates are used to map spacetime points. A coordinate system is a way of assigning
numbers to the manifolds points; GTR requires that we can smoothly assign four
numbers to each point in the manifold thus guaranteeing that spacetime is four-
dimensional and continuous (there are no gaps between points), and hence differ-
entiable. There are any number of different coordinate systems that one can adopt
(different ways of assigning numbers). General covariance is a desirable property
for a spacetime theory to have. Something would be wrong if the mathematical
SPACETIME METAPHYSICS 373
form of basic laws depended on an arbitrary choice of coordinates, so all spacetime
theories (not just GTR) are nowadays presented in a generally covariant form. But
while desirable, general covariance also brings a potentially problematic freedom.
If we take one model of GTR [M, g, T] we can apply a diffeomorphism a per-
mutation (or switching about) of points in M satisfying certain restrictions; for
example, the deformation of the original must be smooth (no topology- destroying
breaks or tears). This transformation call it h yields a different model [M,
h*g, h*T]. Although this new model describes a universe that is observationally
equivalent to the original, there is a significant divergence: it spreads the matter
and metrical fields over the points of the manifold in a different way, and so the
same physical processes are differently located in spacetime.
5
Since there are any
number of diffeormophisms that can be applied to the original model, there are
any number of ways the material contents of the universe can be located with
respect to the manifold.
E
F
N
O
E
F
N
O
M, g, T M, h*g, h*T
Figure 21.1 The hole diffeomorphism. The hole transformation affects what happens where. The
diagrams show the same four galaxies in free-fall motion, in an expanding universe. In the original GTR-
model, the galaxy that traverses the hole passes over a sequence of points that includes E and F; in the
second GTR-model, E and F are unoccupied, and the galaxy passes over a sequence of points that includes
N and O (which are themselves unoccupied in the first model). The locations of processes outside the
hole-region are unaltered.
To illustrate the problem this freedom poses for the substantivalist Earman and
Norton focus on a particular kind of transformation: a hole diffeomorphism. The
hole is a particular region of spacetime; it is not a hole in spacetime (i.e. an
absence of points) nor need it be empty of matter. The diffeomorphism leaves all
assignments of matter to points outside the hole region unchanged, but alters the
assignments within the hole; hence the transformation assigns different spatiotem-
poral locations to the material events occurring within the hole. So, for example,
the original model might say that the worldline bundle of a given galaxy passes
through a spacetime region E, whereas the new model (after the hole transforma-
tion) says that the galaxy doesnt pass through E: it occupies a new collection of
spacetime locations. This is problematic because it can be viewed as a failure of
determinism. As Norton writes:
The physical theory of relativistic cosmology is unable to pick between the
two cases. This is manifested as an indeterminism of the theory. We can
specify the distribution of metric and matter fields throughout the manifold
374 TIME AND SPACE
of events, excepting within the region designated as The Hole. Both the
original and the transformed distribution are legitimate extensions of the
metric and matter fields outside The Hole into The Hole, since each satis-
fies all the laws of the theory of relativistic cosmology. The theory has no
resources which allow us to insist that only one is admissible. (2008b: 6)
With reference to our earlier example, given complete data about what occurs
outside the hole, GTR cannot predict whether or not the galaxy in question will
pass through region E. More generally, since we can select any region as the hole,
GTR cannot predict which events will happen where, even given maximal data
concerning what happens outside the chosen hole. As Hoefer puts it, If A is a
spacetime point that will (in fact) exist around here sometime tomorrow, the past
plus the field equations do not determine whether A will underlie me, or you, or
some part of a star in a distant galaxy (1996: 9). The hole transformations do
not render anything that is observable indeterminate; they leave all predictions
concerning the relative distances between objects intact. The indeterminism only
extends to what happens where: the spread of material processes over the mani-
fold of points. An indeterminism of this sort will not be a problem for a relationist
who regards spacetime as a fiction, but substantivalists cannot dismiss it lightly.
For any model of GTR that accurately represents the observable universe, there are
many other models, observationally equivalent to the first, that situate any given
material process on a different collection of space time points.
Earman and Norton recognize that we cannot reject a theory just because it is
indeterministic in some respect, or else we would have to reject quantum theory
and much classical physics. However, they argue that the way in which determin-
ism fails in the case of GTR is deeply suspect: it doesnt fail for any reason of phys-
ics, but rather because of a commitment to a particular metaphysic.
The argument does not rest on the assumption that determinism is true,
much less on the assumption that it is true a priori, but only on the assump-
tion that it be given a fighting chance. To put the matter slightly differ-
ently, the demand is that if determinism fails, it should fail for a reason of
physics. (Earman 1989: 180)
As for why the failure of determinism is due to reasons of metaphysics rather than
physics, recall Leibnizs argument against Newton. We called worlds where all
inter- body relations are the same Leibniz equivalents. Leibniz is committed to
the doctrine of Leibniz equivalence: the claim that there is no physical difference
between universes that do not differ in any observable way, but do differ in the way
the bodies are related to a hypothetical underlying space. Since Newton is commit-
ted to the view that the material bodies could be differently located with respect
to absolute space, and that this difference is a genuine physical difference, Newton
rejects the principle of Leibniz equivalence. Earman and Norton argue that in the
context of GTR there is only a failure of determinism if we reject Leibniz equiva-
lence: for if we accept Leibniz equivalence then there is no longer any physical
difference between the models of GTR that are generated by hole transformations.
It is only if we insist that the spacetime points in M are real physical individuals
that the transformations represent physically different states of affairs. If we reject
SPACETIME METAPHYSICS 375
this metaphysical commitment to substantivalism, we are free to accept Leibniz
equivalence, and determinacy is restored. As Norton puts it:
If we deny manifold substantivalism and accept Leibniz equivalence, then
the indeterminism induced by a hole transformation is eradicated. While
there are uncountably many mathematically distinct developments of the
fields into the hole, under Leibniz Equivalence, they are all physically the
same. That is, there is a unique development of the physical fields into the
hole after all. Thus the indeterminism is a direct product of the substantival-
ist viewpoint. Similarly, if we accept Leibniz equivalence, then we are no
longer troubled that the two distributions cannot be distinguished by any
possible observation. They are merely different mathematical descriptions
of the same physical reality and so should agree on all observables the
manifold substantivalist advocates an unwarranted bloating of our physical
ontology and the doctrine should be discarded. (2008b: 6)
21.4 Metrical essentialism
Since the hole argument (at least in its original form) applies to any spacetime
theory that is formulated in a generally covariant way, not just GTR, those inclined
(for general metaphysical reasons) towards substantivalism may well be tempted
to dismiss it as no more than an artifact of a particular mode of mathematical
representation. But while there is something to be said for this response after
all, we have already seen that there are very strong grounds indeed for supposing
that the spacetime of GTR is substantival it carries little weight unless accompa-
nied by a viable alternative formalism, one that renders GTR immune to the hole
argument. Needless to say, such alternatives are not easy to come by. But there
are less radical responses available, responses that meet the threat posed by the
hole diffeo morphism while retaining GTR in its standard form. Indeed, there are
several such responses, and a full survey cannot be supplied here.
6
But to provide
some indication of the way explicitly metaphysical considerations lie at the heart
of the debate I will explore the rationale of an approach that several writers have
found appealing.
Earman and Norton argue that if one is to embrace both GTR and substanti-
valism, then one should endorse manifold substantivalism; that is, of the three
components of GTR models [M, g, T] we take M alone to represent substantival
spacetime. If we do, then it makes sense to think of the very same spacetime
points being differently related to g and T, and so the very same spacetime points
can have different metrical properties and be differently related to the spread of
mass- energy. This particular conception of substantivalism is crucial to the success
of the hole argument. For the hole transforma tions to be possible we must be
able to make sense of the very same spacetime points being differently related to
the metrical fields and matter- fields. But suppose we adopt a different version of
spacetime substantival ism: metrical essentialism. On this view, we take substan-
tival spacetime to be identical with a particular metrical field; that is, spacetime
points that are structured by a metric. The distance relations between points (and
376 TIME AND SPACE
their geometrical properties) are now taken to be essential to them: it no longer
makes sense to suppose that a given point could exist in different metrical relations
to other points.
If the substantivalist adopts this position, the hole argument cannot get off the
ground: the hole alternative involves numerically the same points bearing different
metrical relations to one another, so if we adopt metrical essentialism, the hole
alternative is metaphysically impossible. Consequent ly, the empirically equivalent
alternatives generated by the hole transforma tions do not exist, and the associated
failure of determinism is averted.
An example may make things clearer. Suppose our original model of GTR situ-
ates the physical events in France at a given time on a collection of points situated
where we normally take France to be between Germany and the UK. Call this
collection of points P(F), and call the collection of points that underlie the events
in Germany at this time P(G). The proponent of the hole argument maintains that
there is an observationally equivalent model of GTR in which P(G) underlies the
goings- on in France, and P(F) underlies the goings- on in Germany, so the very
same points that in the original model were where we normally take Germany to
be are now where we normally take France to be, and the corresponding French
points have been shifted eastwards. Since the equations of GTR together with
occurrences outside the FrancoGerman region do not allow us to choose between
these two models, we have a failure of determinism. The metrical essentialist will
argue thus:
The competing model represents the same spacetime points existing in dif-
ferent relations to both material processes and other spacetime points. This
makes perfect sense if we hold that spacetime is a manifold of points that
can be rearranged in this sort of way without losing their identities. But this
conception of point identity is not obligatory. If we adopt the view that the
identity of a point is fixed by its spatiotemporal relationships with other
points, the competing model cannot exist. A spacetime is a collection of
points with a specific spatiotemporal organization.
Since GTR invites a substantivalist interpretation, the fact that metrical essential-
ism immunizes it against the hole argument which threatens such an interpreta-
tion is one reason for adopting the doctrine. Another is the suspect character of
the main alternative: manifold substantivalism.
Recall what M consists of: a collection of points that form a smooth four-
dimensional continuum. In the absence of a metric there are no spatial or tem-
poral distances, no affine structure; the light- cone structure is absent and so past
and future are not distinguished. Our spacetime is four- dimensional (at least on
the macro- scale), but it seems very plausible to suppose that something more is
needed, beyond mere four- dimensionality, for a collection of points to be a space-
time. How could a collection of points for which the distinction between space and
time does not exist constitute a spacetime? Yet this distinction does not exist prior
to the imposition of the metrical field on M. The manifold substantivalist might
object: Look, in moving beyond Newton and adopting relativity, youve had to
take on board some pretty strange ideas about space and time; common sense has
largely gone out of the window. Might it not be that the truth about space and
SPACETIME METAPHYSICS 377
time is stranger still? This might seem to be a reasonable response, but it is still
problem atic. In urging us to recognize that spacetime is a real entity in its own
right, the substantivalist points to the useful work that spacetime does in unifying
and explaining physical phenomena such as acceleration, inertial motion and light
propagation. But it is the metrical field of a GTR spacetime which brings with
it inertial structure, spatiotemporal distances and light- cone structure that does
this work, not the bare manifold, which in itself explains very little (cf. Hoefer
1996: 12).
It seems, then, that substantivalists have a good reason for rejecting manifold
substantivalism in favour of metrical substantivalism, of some form or other. What
of the charge that only manifold substantivalism saves the distinction between
container and contained from obliteration? Earman and Norton argue that,
since the metric field carries energy and momentum (as shown by the phenomena
of gravity waves), if we allow spacetime to include the metric field, we lose the
sharp distinction between spacetime and its contents. But the metrical essentialist
can ask:
It is true that Newton assumed space was devoid of material characteristics
such as energy and momentum, but why must the contemporary substantiv-
al ist seek to retain the distinction between container and contained in its
traditional form? In accepting GTR we have accepted that Newton was wrong
about a good deal, why not accept that he was wrong about this too? Since
the spacetime of the metrical substantivalist contains energy and momentum,
the case for recognizing spacetime as a substance is all the stronger: spacetime
is now a physical entity in its own right, one which has causal powers just
like any other physical entity spacetime becomes all the more substantival!
On the face of it, this seems a viable position (we shall be considering it from a
different vantage point in the next section).
Earman and Norton have a further argument for excluding the metric from
spacetime, an argument that appeals to the very nature of GTR. According to GTR
the metric is dynamic: the geometry of spacetime is (in part) a function of the
distribution of matter g and T are interdependent so different matter- fields
yield different metrics. But surely, they continue, an essential feature of relativistic
spacetime must be a feature that is invariant, a feature that occurs in all models of
GTR. Since metrical structure is dynamic, varying from model to model, it is surely
right to classify it as an accidental, rather than essential, feature of spacetime.
Hence we are led to the view that spacetime essentially consists of the manifold
(viewed as a four- dimensional continuum of points) and nothing but.
It is hard to see that this argument has any real force. For one thing, if we were
to follow the logic of Earman and Nortons argument all the way, and hold that
the essential features of spacetime must be invariants, then topological properties
become non- essential too, since there are models of GTR with different topologies.
We would thus be led to the simple point substantivalism: spacetime is a collection
of points that need not be related to one another spatiotemporally at all. Needless
to say, this is a singularly unappealing doctrine. But more importantly, there is no
compelling rationale for the doctrine that essential properties must be invariant
across models. If we were still working in a Newtonian framework, where space
378 TIME AND SPACE
endures through time, then a dynamic metric would clearly be incoherent: since
the very same spatial points would have different metrical properties at different
times, we could scarcely claim that metrical properties are essential to points. But
in the four- dimensional framework of GTR we have abandoned enduring points;
a spacetime is composed of momentary points, ephemeral entities that do not
endure. There is surely no obstacle to maintaining that the metrical properties of
these entities are essential to them.
Metrical essentialism comes in different forms. Hoefers preferred version leads
him not only to reject primitive identities for points, but to hold that points are to
be individuated by reference to both their relations to other points and material
processes. Consequently, he maintains, substantivalists should embrace Leibniz
equivalence and hold that all qualitatively (observationally) identical worlds are
physically identical. If we take this step, it no longer makes sense to suppose that
all the material bodies in the universe could have existed three feet to the east of
where they actually are.
While this position may have a certain appeal from a purely scientific point
of view observationally equivalent worlds are physically identical it is less
appealing at the metaphysical level. If spacetime is a genuine substance, in worlds
where physical law makes spacetime structure independent of material bodies,
why couldnt God (or chance) locate the material system elsewhere? As Nerlich
stresses, given the symmetries of Newtonian spacetime, the substantivalist has a
perfectly respectable explanation of why Leibniz equivalents are observationally
equivalent. But they are not qualitatively identical if we take (as a substantival-
ist should) the domain of real properties to include spacetime location: worlds in
which the same objects are differently located are not qualitatively identical from
the standpoint of a full- blooded substantivalism. Hoefer construes the doctrine of
primitive identity as entailing that two actual space time points could have all their
properties systematically interchanged, and (seemingly) assumes that if we reject
this doctrine we should hold that all of a points actual properties are essential to
it. But the metrical essentialist need not go so far. The option of holding that points
have their metrical relations to other points essentially is all that is required, and
this limited essentialism allows the substantivalist to accept the logical possibility
of the same spacetime existing without material objects, or with material objects
differently distributed (as in Newtonian worlds). Indeed, it would be quite in
the spirit of substantivalism to maintain that the identities of material objects are
essentially linked to their locations in spacetime, a doctrine that precludes the pos-
sibility of there being two distinct worlds in which numerically different objects
occupy the same locations in the same spacetime.
But these are issues of comparative detail. The important point is that metrical
essentialism, however the fine print is formulated, seems a clear and well- motivated
form of substantivalism. There is a cost. Spacetime points lose the ontological
autono my they have in manifold substantivalism and, to this extent, the substan tiv-
al ists position might be thought weakened: an attribute of traditional substances
is their ability to exist independently of all other entities. It is not clear, however,
that the loss is to be regretted. The doctrine that spacetime is a real entity does
not in itself entail the view that this entity has component parts that are capable
of independent existence. It may be that not all real substances can be broken into
SPACETIME METAPHYSICS 379
parts in the way of macroscopic material things. Last but not least, it is worth not-
ing that Newton himself seemed to subscribe to a form of metrical essentialism. To
quote once again from De Gravitati one: The parts of duration and space are only
understood to be the same as they really are because of their mutual order and posi-
tion, nor do they have any hint of individu ality apart from that order and position
which consequently cannot be altered (Huggett 1999: 112).
7
If GTR makes metrical essentialism an appealing position for the substantivalist,
it is also a position that is vulnerable to Fosters argument. Foster argues that the
same underlying spacetime could (metaphysically speaking) sustain a number of
different physical geometries. This possibility poses no problem for anyone who
takes spacetime points to have primitive identities and who takes a particular space-
time to consist of a particular collection of points. Put together, these doctrines
permit the same spacetime to have different intrinsic and physical geometries.
But adopting metrical essentialism rules out this response to Fosters scenarios. If
the underlying space and physical space have different geom etries, they must
be composed of different points, and hence be numerically distinct. Of course,
as we have seen, this fact only yields an anti- realist conclusion if we subscribe to
geometrical monism. By accepting that a single spacetime can possess two distinct
geometries at the same time, one physical and one intrinsic, anti- realism is averted.
Hence the substantivalist who adopts metrical essentialism should also embrace
geometrical pluralism.
21.5 An outmoded debate?
Earman and Norton argue that unless substantivalists equate spacetime with M the
sharp distinction between container and contained is lost; construing space-
time as [M + g] or [M + g + T] infects spacetime with material attributes, since
the metric field g possesses paradigmatic physical properties such as energy and
momentum. I suggested above that the substantivalist might welcome this result,
since it confirms the status of spacetime as a real entity. However, in a recent
paper Rynasiewicz has argued that a quite different conclusion should be drawn:
spacetime does indeed take on distinctively physical attributes in GTR, but, rather
than vindicating the substantivalist position, this shows that the issue between
substantivalists and relationists no longer has any clear sense; it is a debate that has
been rendered pointless and irrelevant by developments in science.
The fundamental fallacy here is the supposition that the alleged issue can be
stably formulated in terms that transcend any particular historical or concep-
tual context The distinction between space and matter, or alterna-
tively, between container and contents, necessary for the debate, though
immediate to common sense and sustainable to the microlevel for the earlier
participants of the debate, ultimately evaporates in the course of develop-
ment of the physics of the aether. Relativity, instead of restoring the original
contrast, has made it only more problematic. The current attempts to con-
tinue the debate do so by misguided attempts to legislate how that distinction
should be projected onto a framework that has left it behind. (1996: 306)
380 TIME AND SPACE
Summarizing rather brutally, Rynasiewiczs argument involves the following
claims:
Lorentzs aether can equally well be viewed as a successor to Newtons absolute
space or Descartess subtle fluid. Unlike Newtonian space it is the bearer of
physical properties, since at every point within it the electromagnetic field has
a definite value. But since the aether is at rest, it is natural to view it as playing
the same role as Newtons space, and hence take the distinction between inertial
and non- inertial motion to be grounded in the aethers structures. Is the aether
spatial or material in character, or something neither spatial not material? The
concepts of space and matter are not themselves sufficiently determinate to
settle the issue. The question is not one of fact (1996: 288).
The spacetime of GTR is simply the Lorentzian aether in relativistic guise. It can
no longer be regarded as being at absolute rest, but it continues to ground the
distinction between inertial and non- inertial movement and possesses physi-
cal properties. As such, it is neither clearly spatial nor material in character,
and there is no way of projecting on to it the distinction between space and
matter that has lost the clear rationale it possessed in the days of Newton and
Descartes.
It must be conceded that the concepts of space and matter have evolved consid-
erably, and so transposing the classical substantivalistrelationist controversy into
the contemporary context does involve extending concepts beyond their original
sphere of application. But as Hoefer forcefully argues, sometimes indeed quite
often there is one natural and compelling way to make such extensions, all other
decisions looking forced and awkward at best. To identify Lorentzs ether with
absolute space is natural; to equate it with a Cartesian subtle matter, forced at best
(1998: 457). Unlike the aether, Descartess fluids could not be at rest; indeed, they
were meant to explain inertia and gravity, as these apply to solid bodies, through
their motions. Furthermore, unlike both absolute space and Lorentzs aether, they
do not pervade the interiors of solid material things. As for GTR, we have already
seen that there is a strong case for identifying substantival space with the metrical
field. Since the latter possesses its own energy, the question arises as to whether it is
legitimate to regard this field in particular, rather than the others now recognized
by physicists, as the latter- day counterpart of Newtons absolute space. Einstein
thought it was. In his Leyden lecture he noted a remarkable difference between
the gravitational and electromagnetic fields. Whereas it is quite conceivable that a
part of space could exist without its electromagnetic field, there can be no space,
nor any part of space, without a gravitational (i.e. metric) field, for these confer
upon space its metrical qualities, without which it cannot be imagined at all
(Einstein 1922: 21). Another writer makes the point thus:
Granting certain similarities of the metric field to other physical fields, still
the metric plays a unique role in any relativistic theory. For the metric is
indispensable. There is a theory of what the universe would be like without
electromagnetism or without strong gauge fields. One can describe universes
that are complete vacua, with stress- energy tensors identically zero. But
there is no theory of space- time with electromagnetism but with no metrical
SPACETIME METAPHYSICS 381
structure. Without affine structure and light- cone structure and covariant
differentiation the laws governing other fields could not be written. It is
the physics itself, not just philosophical fancy, that singles out the metric as
peculiarly intertwined with space- time. (Maudlin 1989: 87)
There is a further and quite different reason for supposing that the terms of the
classical debate remain relevant. GTR and quantum theory remain unreconciled.
Some of the more radical proposals for reconciliation hope to show that the mac-
roscopic spatiotemporal features of the universe are the product of inter relations
between pre- spatial entities. Were this approach to prove successful, there is a
clear sense in which classical relationism would stand vindicated: the universe
would consist of objects that, far from requiring spacetime to exist, generate space-
time or rather, a world that manifests itself to us in spatiotemporal form by
their modes of interrelation. Of course it may be that the successor to quantum
mechanics and GTR will take a different form entirely, one that truly does render
the substantivalistrelationist debate redundant. But equally, it may not.
8
21.6 GTR and time
What of the implication of GTR for the nature of time? In this regard, the impact of
GTR is potentially enormous, but also intriguingly ambiguous. The field equations
do not decisively point one way or the other, but there are particular models of
the theory that do, and others that might.
Nothing in GTR is obviously incompatible with the Block conception. As in
STR, simultaneity when defined by Einsteins light- signalling method is relative
to inertial frames, albeit with a difference. The inertial frames of STR are valid
throughout the whole of Minkowski spacetime; given the possibility of variable
curvature there are no global inertial frames in typical GTR worlds (even though
the distinction between inertial and non- inertial motion is at each place well
defined). The fact that simultaneity is localized as well as relativized in GTR might
lead one to expect that the latter theory is even more hostile to dynamic models
of time than its predecessor. But this is not so.
A certain class of GTR models are surprisingly friendly to those versions of dyna-
mism that posit a universe- wide tide of becoming. The manifolds in the models
in question can be exhaustively partitioned into foliations of non- intersecting
global (three- dimensional) hyperplanes (or leaves) that are orthogonal to the
time- like geodesics. By regarding all the points on each of these hyperplanes as
simultaneous, we can assign a global or cosmic time to all events, despite the fact
that the events in question are not causally connectable by a light signal, and so
fall into the region of space- like connected events that evade Einsteins operational
definition of simul taneity. The models in question only apply to universes that
are isotropic and homogeneous, but, as we saw in 20.7, the Friedmann models
used by cosmologists for modelling our universe fall into this category (as do the
Schwarzschild models used for modelling the solar system), so it is at least possible
that they apply to our world, even if the hyperplanes in question will be somewhat
uneven due to local variations in matter densities.
9
382 TIME AND SPACE
If our universe is of this form, it may be possible for the dynamist to embrace
relativity without resorting to Steins stratagem and relativizing becoming to
spacetime points. Dynamic (or Compound) Presentists will regard a universe pos-
sessing a cosmic time as a succession of thin hyperplanes, created and anni hi lated
in turn. Rather than a hollow shell, Growing Block theorists will view the universe
as akin to an expanding solid sphere (assuming the big bang), growing as hyper-
plane is added to hyperplane;
10
hence Lucas:
In many of the models that cosmologists use there is a worldwide cosmic
time that flows, if not evenly and uniformly, at least generally and universally.
There are thus also preferred hyperplanes (not necessarily flat) of simultane-
ity constituting a worldwide present and separating a real unalterable past
from a possible future not yet actualized. (1999: 10)
But further assumptions are at work here, which require additional justifica-
tion. Even if our universe allows the postulation of worldwide simultaneity planes,
it does not necessarily follow that there is an objective difference between past and
future of the sort Lucas here describes. Since the cosmic time of GTR is perfectly
compatible with the Block model, the dynamist cannot simply assume that the
events we call future have a different status to those we call past. Lucas himself
appeals to quantum considerations: he construes the collapse of the wave- packet
as a real physical occurrence in which a definite actuality emerges from a spread
of mere probabilities, but he concedes that this interpretation is contro versial and
rejected by many. Others will attempt to locate a cosmic asymmetry in causal or
entropic considerations; others may be content with a directionless dynamism.
However, as Torretti has noted this remarkable revival of absolute time is
more than compensated by the complete destruction of universal time order in
other conceivable G[T]R worlds (1999a: 77). Among the GTR worlds Torretti
has in mind are those corresponding to the Gdel solutions mentioned earlier.
In these worlds there are no global time slices, and no globally consistent tempo-
ral ordering. But these spacetimes have a more remarkable property: they con-
tain time- like paths (one through each point), which, when followed in a given
direction, eventually lead back to nearly the same point. These paths are known
as closed time- like curves or CTCs. If you were to travel along one of these
Figure 21.2 Cosmic time.
The FriedmannRobertsonWalker models
of GTR assume an even distribution of
mass-energy throughout the universe; the
resulting spacetimes can be regarded as
consisting of nested non-intersecting global
hyperplanes, shown here in cross-section,
with one spatial dimension suppressed. Each
shell represents the entirety of the universe
at a given moment of time. This distinctive
foliation generates a consistent universe-wide
time-ordering for events. However, there
is a sense in which cosmic time is just one
reference frame among others: the effects of
relative motion on measurements of temporal
intervals (and simultaneity) remain.
SPACETIME METAPHYSICS 383
paths, although your watch would always be moving forwards you would end up
approaching the place and time from which you originally started. In this sense,
a trip along a CTC is a journey into the past. A worldline that forms a CTC is no
different locally than an ordinary worldline which is why the hands on your
watch would continually advance but viewed as a whole the events along a CTC
cannot be temporally ordered in the usual way, since each of these events occurs
both earlier and later than all the others. As can be seen in Figure 21.3, your
approach (A) to your original point of departure (D) occurs both later and earlier
than the latter.
A
D
Time
Figure 21.3 A closed time-like curve. Anyone who
completes a trip around this spacetime loop finds themselves
back where they started, in time as well as space. The time of
departure (D) occurs earlier and later than the time of arrival
(A) as do all the other times on the curve.
Although CTCs have been found in many other solutions to the GTR equations,
the relevant models represent matter distributions quite unlike those that obtain in
our universe. Can we conclude that our universe is free of CTCs? Not necessarily.
There may well be no naturally occurring CTCs in our universe, but this does not
rule out the possibility of creating them by artificial means. Tipler (1974) showed
that a quickly rotating cylinder of very dense matter would make backward time
travel possible; by flying your spaceship around the cylinder a few times you
could return to Earth before you set off. Since it is not clear that a cylinder of
finite length would generate the kind of spacetime warping required for this, it
may well be that a Tipler time machine could never be constructed. More recently
Thorne and others have worked on the possibility of wormhole time machines of
the sort mentioned in 8.9.
11
Again, it is as yet far from clear that constructing
such a device is possible. Leaving aside the formidable technological difficulties
for example, the creation of sufficient negative energy to keep the wormholes
open the question of whether GTR and quantum theory will allow such things
to exist is unresolved.
Gdel himself inclined to the view that whether CTCs actually exist is irrelevant;
the mere fact that they are physically possible (assuming GTR is true) suffices to
establish the Block view. This is the upshot of his modal argument:
The mere compatibility with the laws of nature of worlds in which there is
no distinguished absolute time, and [in which], therefore, no objective lapse
of time can exist, throws some light on the meaning of time also in those
worlds in which an absolute time can be defined. For, if someone asserts that
this absolute time is lapsing, he accepts as a consequence that whether or not
an objective lapse of time exists (i.e., whether nor not time in the ordinary
384 TIME AND SPACE
sense of the word exists) depends on the particular way in which matter and
motion are arranged in the world. This is not a straightforward contra dic-
tion; nevertheless, a philosophical view leading to such consequences can
hardly be considered satisfactory. (1949: 5612)
In talking of lapses of time, Gdels is talking about absolute becoming: The
existence of an objective lapse of time [means] that reality consists of an infinity
of layers of now which come into existence successively (1949: 558). It is not
clear whether Gdel was committed to the Growing Block view or Presentism,
but his contention that real time, time as we ordinarily understand it, is dynamic
may well be correct. The more intriguing and more controversial suggestion is
that time in GTR worlds cannot be dynamic because in some physically possible
GTR worlds time is of the block variety. Dynamists might be tempted simply to
dismiss Gdels claim:
Fine, GTR worlds with CTCs are Block worlds where time is non-dynamic, but
mass- energy in our world is differently organized, and the actual universe
may well possess a global time function whose simultaneity planes are perfect
candidates for becoming; the fact that some GTR worlds are passage-free does
not mean that all are.
But this reply, as it stands, is insufficient to counter Gdels point.
Suppose it were the case that by simply rearranging mass- energy in our world
wormholes allowing fast and unhindered access to past and future times could be
opened up. Producing the required conditions might not be practically possible for
civiliza tions with technological capabilities similar to our own, but this limitation
is of no consequence. If in our universe CTCs are physically possible, if they could
be produced simply by rearranging matter, then the past and future times to which
these CTCs would provide access must be real. Unless they are real, it would not
be physically possible to travel there, but we are assuming that it is.
It is clear what dynamists must do to counter Gdels argument: they must deny
that the relevant kind of CTCs are physically possible in our universe. This might
seem problematic, given the fact that there are solutions to the GTR equations that
allow CTCs, but as we have seen on previous occasions, the GTR equations need
supplying with boundary conditions, certain assumptions concerning the large-
scale structure of the universe. The boundary conditions applied by the dynamist
are of a distinctive sort; the Presentist will maintain that times earlier and later
than the present do not exist; the Growing Block theorist will maintain that times
later than the present do not exist. With these stipulations in place, the potentially
problematic paths through spacetime are not physically possible in our universe.
These stipulations, it is true, do not derive from GTR itself, but then few dynamists
claim that the laws responsible for the systematic creations and/or annihilations
that render time dynamic can be derived from current scientific theories. If future
developments in physics provide an underpinning for dynamic time, so be it; if
not, then dynamists will have to rely on other considerations to support their view.
In the same paper Gdel makes another suggestion aimed at undermining the
position of dynamists. He is talking about beings living in the Gdel world cor-
responding to his own solution to the GTR equations:
SPACETIME METAPHYSICS 385
the decisive point is this: that for every possible definition of a world time
one could travel into regions of the universe which are past according to
that definition. This again shows that to assume an objective lapse of time
would lose every justification in these worlds. For in whatever way one may
assume time to be lapsing, there will always exist possible observers to whose
experienced lapse of time no objective lapse corresponds (in particular also
possible observers whose whole existence objectively would be simultane-
ous). But, if the experience of the lapse of time can exist without an objective
lapse of time, no reason can be given why an objective lapse of time should
be assumed at all. (1949: 561, original emphasis)
Remarkably, in the Gdel world, scenarios similar to the one described by Wil-
liams (see 3.8), where an entire life is lived at a single time, are a real possibility.
12

But more important in the current context is Gdels final point. Savitt (1994:
468) expands it into an argument, which (in slightly abbreviated form) runs like
this:
1. It is physically possible that some galaxy in a physically possible universe is
inhabited by creatures just like us. Let us suppose this universe is the Gdel
world.
2. The direct experience of time of the creatures just like us in the Gdel world
will be just like our own direct experience of time.
3. It is possible to have direct experience of time just like ours in a universe in
which there is no objective lapse of time.
4. Our direct experience of time provides no reason to suppose that there is an
objective lapse of time in our universe.
5. Our direct experience of time provides the only reason to suppose that there
is an objective lapse of time in our universe.
6. Since there is no objective lapse of time in the Gdel world, there is no reason
to suppose that there is an objective lapse of time in our universe either.
As we have seen, there are dynamists who would not accept 5. Some Growing
Block theorists appeal to content- asymmetries (such as the fact that we have no
detailed knowledge of the future), and Tooley rests his case on causal consid-
erations. While proponents of the Block view reject these arguments, we saw in
Chapter 4 that, although a good deal of progress has been made on the task of
explaining content- asymmetries without appealing to temporal dynamism, work
remains to be done. But what of the point Gdel himself is making? Does the fact
that his GTR worlds can be just like ours (locally) prove that it is possible to have
direct experience of time just like ours in a block-universe?
No. Once again, a question is being begged. A dynamic theorist who holds that
the character of our direct experience of time (in particular the immanent flow
discerned in 7.6) depends on temporal becoming would be justified in maintain-
ing that the creatures in the Gdel world could not possibly have temporal experi-
ence exactly resembling our own for the simple reason that in their world there
is no temporal becoming. This said, as noted in 7.8, it is far from clear that the
flowing character of our experience does depend on times being dynamic. If
this is the case, then it may well be that Gdelians could have conscious lives
386 TIME AND SPACE
that are phenomenologically indistinguishable from our own. But of course, if a
more penetrating phenomeno logical analysis were to reveal that experience such
as our own could only exist in a temporally dynamic universe, this verdict would
have to be revised.
13
22 Strings
We have now encountered a number of ways in which contemporary physics threat-
ens common- sense views of space and time. With STR comes the inter mingling of
space and time and the abandonment of absolute simultaneity; with GTR comes
curved spacetime; there are interpretations of quantum theory that posit backward
causation, multiple universes and instantaneous action- at- a- distance. Mention has
also briefly been made of an attempt to integrate these two theories. We saw in
3.8 that canonical quantum gravity theory has significant and potentially disturb-
ing implications for time, even if one interpretation of what these are is empirically
suspect (see 6.9 and 7.8). There is another theoretical programme aimed at inte-
grating (or replacing) GTR and quantum theory that more than merits a mention:
string theory. This too has radical implications, but they concern space rather than
time. If current versions of string theory are to be believed, we live in a universe of
at least nine but possibly ten (or more) spatial dimensions, not just three.
String theory is still in the early stages of development. New versions are con-
tinually being developed, and there is as yet no empirical evidence that strings
themselves exist. Given their size (see below), direct detection may never be pos-
sible. But although the theory is far from universally accepted, it is also taken seri-
ously, and in the opinion of many it is our best prospect for a theory of everything.
And as we shall see, the theory not only expands the dimensions of space, but it
provides space with new roles.
22.1 Higher dimensions
The idea that we live in a nine- or ten- dimensional space might seem quite absurd.
Why is our freedom of movement so restricted if these extra dimensions exist?
Why dont objects disappear into them? Before considering how string theorists
reply to these questions, it is worth considering how evidence of concealed dimen-
sions could emerge in a simple context provided by Flatland.
Imagine a universe containing a Flatland in which certain (non- organic) physical
objects are three- dimensional. These objects are homogeneous, stationary, immov-
able and indestructible solids of fixed but differing orientations with respect to the
two- dimensional plane in which they are embedded.
If we suppose that the Flatlanders, and their senses, are wholly confined to
their two- dimensional plane, the three- dimensional nature of these objects would
be wholly concealed from them. All they can detect using their senses are their
388 TIME AND SPACE
two- dimensional cross- sections. If the thought were to ever to cross their minds
(there is absolutely no reason why it should) that these immobile solids were
three- dimensional, the available evidence would not allow them to determine
their shapes. The way a cube manifests itself in Flatland depends on its orientation:
it might appear as a square, but if it is tilted with respect to the plane, it will be a
quadrilateral with sides of unequal lengths; if it is not fully embedded in the plane,
it will appear as a (not necessarily equilateral) triangle, or even a straight line.
What would have to change in this scenario for the Flatlanders to begin to
suspect the truth? Suppose the objects are as before, but the Flatlanders can also
push them about. They would find that objects of the same (two- dimensional)
size and appearance require more effort to shift than others. But this fact could be
explained in terms of the objects in question possessing different material constitu-
tions or densities. A mathematically sophisticated Flatlander might propose an
alternative explanation of these mass differences, pointing out that these differen-
ces would be what you would expect if material objects of the same constitution
were in reality three- dimensional solids. But in the absence of any independent
evidence of a third dimension of space, it is unlikely that anyone would take this
explanation seriously. Imagine a medieval alchemist explaining the difference in
weight between equal volumes of water and gold in an equivalent manner, in terms
of invisible extrusions into a fourth dimension of space.
However, the three- dimensional hypothesis would become much more plausible
if it were discovered that objects could be moved without actually touching them.
For example, by pushing a square towards a rectangle from a certain direction, the
rectangle starts to move before the square comes into contact with it (in the Flatland
plane), and at this same moment it becomes corres pon dingly harder to push the
square forwards. This would arise if objects were embedded in the plane at differ-
ent angles, so that they could come into contact above or below the plane (see the
two small blocks tilting towards one another on the right and to the rear in Figure
22.1). If a systematic dynamics explaining such phenomena could be worked out
by positing a hidden third dimension, it might well be accepted on methodological
grounds (not least because action- at- a- distance would be eliminated).
This example shows that beings whose own movement (and perception) is
confined to n dimensions could come to believe that certain objects are of n + 1
Figure 22.1 Hidden depths: static three-dimensional
objects embedded in a plane. Flatlanders would be oblivious
to the true geometrical character of these objects if they could
move them this could change.
STRINGS 389
dimensions. There is no reason why higher dimensional dynamics need stop here.
It could turn out that to explain the observed interactions our Flatland physicists
are led to hypothesize that they are dealing with four- dimensional objects: neither
we nor they can visualize such things, but their mathematical properties can be
worked out, and likewise for objects of even higher dimensionality. But what of
space itself? How could there be more spatial dimensions than there appear to be?
One possibility is enforced confinement. A Flatland physicists could argue thus:
Since we know that there are three- dimensional objects in our universe,
we must be living in a three- dimensional space, but for some reason we
are restricted to moving in a two- dimensional plane. We are in the grip of
a force- field that prevents us from enjoying the full spatial possibilities our
universe offers.
This is not an entirely unreasonable hypothesis, but it is vulnerable to Occams
razor. In the absence of any reason to posit the force- field, the [two- dimensional
space + three- dimensional object] hypothesis is more economical; there is, after
all, no obvious metaphysical reason for supposing the three- dimensional objects
must be completely immersed in a surrounding spatial medium.
Considerations of a quite different kind could suggest the existence of addi-
tional spatial dimensions. Perhaps movement in the additional dimension is not
only possible, but happening all the time, but this dimension is so small that the
movements through it are miniscule, and so not easily detected. Suppose that our
Flatland physicists have developed very precise measuring equipment. After con-
ducting a series of experiments on colliding bodies they discover that the numbers
dont add up; objects have more energy than their current theories predict they
can have. Numerous accounts of why this should be are proposed, but only one
proves viable. The proposal runs thus. The additional energy is due to the fact that
objects are vibrating in a previously unsuspected way. Flatland space is extended
a tiny way into the third dimension the space has a slight thickness and all
objects are constantly vibrating along this extra axis, but the distances they move
are so small they cannot be observed directly. The phenomenon is illustrated (in
much exaggerated fashion) in Figure 22.2, which shows Flatland space edge- on,
and, as can be seen, it has a slight thickness. The object depicted is exploiting this
additional degree of freedom by moving up and down as it slides along. In reality,
the updown movements are so small they cannot be directly detected.
The reasoning that leads string theorists to believe there are additional spatial
dimensions is far more complex and subtle than anything to be found in this simple
fictional scenario, but the underlying reasoning is comparable.
Figure 22.2 Hidden dimensions: Flatland with added depth. If the extrusion into the third
dimension were sufficiently small, Flatlanders would be oblivious to its existence any wobbles would
be undetectable.
390 TIME AND SPACE
22.2 KaluzaKlein theory
A mechanism not dissimilar to the one just described was first proposed as a piece
of serious physics by Kaluza in 1919, and developed further by Klein in 1926.
Although Einstein formulated GTR in terms of a four- dimensional (3 + 1) space-
time, the equations he used could be extended to universes with additional spatial
dimensions. Kaluza derived the modified equations for a (4 + 1) universe and
made an astonishing discovery: the five- dimensional equations included Maxwells
equations for electro magnetism. A theory that was designed to explain only gravity
also explains electro magnetism when an additional spatial dimension is introduced.
This unexpected unification at the mathematical level suggests that a single physical
mechanism might be responsible for two very different phenomena: perhaps both
gravity and electro mag netism are nothing but spatial perturbations, with gravity
consisting of disturbances in the familiar three dimensions of space, while electro-
magnetism consists of disturbances in the additional fourth spatial dimension.
This is an exciting idea, but for it to be anything but a mathematical curiosity
a plausible story of how the additional dimension is related to the four of GTR
is needed. Kaluza supplied this too. He proposed that the extra dimension was
curled into a small circle attached to each point of the (3 + 1) spacetime manifold.
(Hence the Maxwell connection: mathematically speaking, electromagnetism is a
U(1) gauge theory, and U(1) is the group of rotations around a circle.) Figure
22.3 shows a simplified KaluzaKlein space. Only a few points are represented;
in reality there would be a circle at every point.
Figure 22.3 KaluzaKlein and beyond. The two uppermost images are alternative ways of depicting
KaluzaKlein space. Ordinary three-dimensional space is shown in Flatland form, with many points omitted:
only a grid-like structure remains. The fourth spatial dimension takes the form of a tiny Planck-scale circle,
one for each point in three-dimensional space. In the 1970s, physicists explored the idea that the problems
with the original version of KaluzaKlein theory could be solved by positing still more spatial dimensions.
The two lower images depict two variants of five-dimensional space, with the two curled up dimensions
having the topology of the sphere and the torus. But these proposals soon ran into difficulties (cf. Greene
1999: 186201).
STRINGS 391
An electron moving through spacetime is free to occupy different positions in
the extra dimension. In effect, it changes its location on the circles as it moves.
This motion is not detectable because the circles are so small: according to Kleins
calculations, their size is of the order of the Planck length, 10
33
cm. Since modern
particle accelerators can only probe down to distances of around 10
16
cm, it is
not surprising that the wobbles of particles moving in the extra dimension go
undetected.
For all its initial promise the KaluzaKlein theory soon ran into difficulties. The
theory predicted mass- charge ratios for the electron that differed greatly from the
measured values. Quantum theory burst on to the scene, and its successes were
such that rival approaches such as KaluzaKlein theory were soon obscured.
Before long many new particles had been discovered, providing new challenges:
discovering the precise properties of the new particles was a difficult task, and
finding a theoretical scheme that explained these properties proved harder still.
The idea that additional curled up (or compactified) dimensions might exist
and have an important explanatory role was not entirely forgotten, but it was
fifty years before its potential significance re- emerged.
22.3 The standard model
By the mid to late 1970s all the main planks of the standard model of particle
physics were in place. This theory could account for all the experimental results in
particle physics that could be carried out at the time with great accuracy, and was
consistent with both quantum theory and special relativity. The general framework
in which it was formulated is known as quantum field theory. In more recent
decades, as more powerful accelerators have been built, it has become possible to
test further predictions of the model, and these too have been vindicated. Evidence
for the elusive top quark was discovered only in 1995.
The standard model was put together with remarkable speed. As late as the
1960s physicists were baffled on two fronts: particles and interactions. Hundreds
of different types of particle had been discovered, and the existence of many more
types was suspected. There was evidence that all the strongly interacting particles
were composed of just three more basic particles quarks but there was much that
remained mysterious. Four different and seemingly fundamental interactions had
been discovered: the electromagnetic, the weak, the strong and gravity. Unlike grav-
ity and electro magnetism, the weak and strong interactions work over very short
distances, typically within atomic nuclei.
1
Physicists only possessed a fully worked
out quantum theory for one of these interactions: electro magnetism. The weak and
strong forces were not completely understood, and the best account of gravity GTR
was not itself a quantum theory. All this was soon to change. By the early 1970s
the WeinbergSalam electroweak theory (proposed in 1968), which unified the
weak force and electro magnetism, was firmly established, and in 1973 a promising
account of the strong interaction quantum chromodynamics was proposed, and
confirmed in 1975 with the discovery of the predicted charmed quark.
Thereafter things rapidly fell into place. The electroweak theory and quantum
chromodynamics were combined to provide a complete account of three of the
392 TIME AND SPACE
four known interactions, and the standard model was born. Three basic types of
particle were recognized: quarks (which combine to form protons, neutrons and
other more massive particles); leptons (which include electrons and muons); and
force carriers, particles which mediate the various interactions (gluons for the
strong interaction, photons for electromagnetism, and weak gauge bosons for
weak interactions). Before long there was talk of a grand unified theory (GUT)
first proposed by Georgi and Glashow in 1974 which would show the electro-
magnetic, weak and strong forces to be modes of a single more basic kind of
interaction. Support for this idea came with calculations that suggested that, while
the three interactions are of very different strengths in (comparatively) low- energy
contexts, this changes at (extremely) high- energy levels, where all three have the
same strength. Why should the interaction strengths converge in this way unless
they are all related? GUTs provide an answer: there is only one underlying mode
of interaction.
Work on quantum field theory and GUTs continued, with mixed success, but
even if success had been forthcoming a problem would have remained: the grand
unifica tion like the standard model itself deals with only three of the four inter-
actions. No mention is made of gravity. This omission did not affect the standard
models ability to provide highly accurate predictions of particle interactions
gravity is by far the weakest of the four interactions, and has negligible effects on
such minuscule and isolated masses but it was clear that the omission would have
to be rectified if a truly complete physical theory was ever to be devised. Unfortu-
nately, all attempts to incorporate gravity into the quantum field theory framework
of the standard model failed. Some of the general properties of the particle that
would mediate gravitational interactions the graviton could be worked out,
but when the normal rules of quantum field theory were applied they generated
nonsensical results. When two gravitons approach one another the theory pre-
dicted that the force between them becomes infinite (which is impossible); other
quantum equations generate infinite quantities, but none of the techniques used to
eradicate them can be extended to the gravitational case. Although the spacetime
of general relativity can vary in curvature, spacetime itself is always smooth (and
effectively flat over very short distances). Quantum field theory delivers something
quite different. The uncertainty principle, which lies at the heart of quantum
theory, entails that all quantum properties will be subject to significant fluctuations
over small spatial and temporal intervals, and the smaller the intervals, the larger
the fluctuations. When the uncertainty principle is applied to the vacuum, the
Energy (GeV)
strong
weak
electromagnetic
Figure 22.4 Intimations of
unity. If the strong, weak and
electromagnetic forces have
the same strength at very high
energies, perhaps they are merely
different modes of a single, more
basic, type of interaction.
STRINGS 393
energistic eruptions it permits affect the geometry of spacetime. Over very short
Planck- scale distances the fluctuations are of such violent intensity that spacetime
comes to the boil. The result is quantum foam, a spacetime of dramatically vary-
ing geometry and topology, where tiny wormholes are continually being created
and annihilated. The idea that spacetime resembles a roiling frothy liquid at the
very small scale is not in itself problematic the disruptions even out over larger
areas, and in any case are far too small to have observable consequences but it
does make the incorporation of GTR into quantum field theory highly problematic.
22.4 Strings
For reasons that will shortly become clear, string theory is a descendant of the
KaluzaKlein theory, and so could be viewed as dating back to 1919. However,
the theory in its current form originated in work done during the 1970s and
early 1980s, work largely ignored at the time due to the successes of the standard
model, with a landmark paper by Scherk and Schwarz appearing in 1984. For a
detailed account of the origins of the theory see Davies and Brown (1988) and
Greene (1999).
From the mathematical perspective string theory is considerably more daunt-
ing than either quantum theory or relativity, but some of its essential features are
easily grasped. All the various particles of the standard model are considered to
be structure less points, these point- particles have various properties velocity,
position, mass, electric charge, spin, colour (a charge- like property of quarks)
and different types of particle have different combinations of these properties.
As we have already seen, there are hundreds of different particle types. String
theory does away with this particle zoo and posits just one fundamental entity:
a string. As its name suggests, a string is not point- like. It is one- dimensional and
so line- like; it has length but no breadth. But not much length: strings are roughly
10
33
cm long the now- familiar Planck- length and so resemble point- particles
when viewed from afar (i.e. at all distances that can be probed by current technol-
ogy). Strings can be closed, like loops, or open, like short lengths of bendable wire.
As a point- particle moves through space it traces out a worldline. As strings move
through space they trace out world- tubes (closed) or world- sheets (open). Strings
interact by combining and splitting apart. Some simple closed string interactions
are depicted in Figure 22.6.
open string closed string
Figure 22.5 A world-sheet and a world-tube.
String-theorys substitutes for the world lines of classical
point-particle physics.
394 TIME AND SPACE
As well as being able to join and split apart, strings can also vibrate. Most of
us have probably experimented with the vibrational properties of cords in the
school playground. If two people each hold one end of a (longish) skipping rope
and move their arms up and down in random fashion, the rope starts to jiggle
chaotically; but by timing the movements in the right way, orderly oscillations
emerge. Slow rhythmic movement will produce the pattern in Figure 22.7a; speed
up, and the pattern in Figure 22.7b emerges; speed up some more and the pattern
in Figure 22.7c will develop. Each of these vibrations corresponds to a resonant
frequency of the rope: the peaks and troughs are evenly spaced, and an exact
whole number of waves can fit between the two endpoints. Elementary strings
can vibrate or pulse in a precisely analogous fashion. Three possible modes
are shown in Figure 22.8.
Figure 22.6 Simple string
interactions: world-tubes
undergoing fusion and
fission.
Figure 22.7 Resonant oscillations.
(a) (b) (c)
Figure 22.8 Resonant modes: some simple cases. The physical properties of a string depend on the
way it vibrates. Different resonance patterns generate the properties associated with the various types of
particle recognized by the standard model (e.g. electron, up-quark, photon), whereas different amplitudes
and wavelengths are associated with different energies and masses.
Different vibration patterns in a guitar string produce different sounds. The
different vibrational modes of an elementary string generate different physical
properties. One pattern of pulses causes a string to behave like an electron, another
like a photon, another like a neutrino, and another like a weak gauge boson. The
way this works is most obvious in the case of mass. Quite generally, the energy car-
ried by a wave is a product of its amplitude (the greater the distance between peaks
and troughs, the greater the amplitude) and wavelength (the distance between
one peak and the next). Higher amplitudes and shorter wavelengths correspond
STRINGS 395
to more frenetic vibrations that possess greater energy; think of the way a gen-
tle pluck on a guitar string creates a gentle vibration, whereas a more forceful
pluck (which transfers more energy) produces a more energetic vibration. Since
we know from STR that mass and energy are interconvertible, it follows that more
energetic strings are also more massive. The same applies, although less obviously,
to all the other properties particles can possess, such as spin, colour (in the quan-
tum chromo dynamics sense) and electric charge. As Greene puts it: What appear
to be different elementary particles are actually different notes on a fundamental
string. The universe being composed of an enormous number of these vibrating
strings is akin to a cosmic symphony (1999: 146).
The standard model can handle a vast variety of different particle types, and
even generate predictions about the properties of particles as yet unobserved, but
it doesnt explain why the particles have their various properties; that they do
is taken as a brute fact. String theory goes deeper. It is no longer a mystery why
certain combina tions of mass, spin, charge, and so on are found: the relevant
property combinations correspond with different vibrational modes of a single
fundamental type of entity.
I have yet to mention the most significant advance achieved by string theory.
One of the resonance patterns generates precisely the properties the standard
model ascribes to the graviton, but whereas the standard model could not be
extended so as to incorporate this particle in a consistent way, its existence is guar-
anteed by string theory. It is there right at the outset, as a basic mode of string
oscillation. Moreover, it is possible to make sense of interactions between gravi-
tons in string theory. There are none of the problematic and ineradicable infinities
that arise in quantum field theory. In large part this is due to the extended nature
of strings themselves. The equations of the standard model deliver problematic
results because the interacting particles are zero- dimensional points that can
become arbitrarily close to one another; in string theory interactions are not con-
centrated into single points, but are always spread over small regions of space (as in
the world- tubes depicted in Figure 22.5). The extended character of string interac-
tions prevents unwanted and physically unrealistic infinities from occur ring.
2
Strings may be able to explain a lot, but their remarkable properties do not stop
here. For one thing, being approximately of Planck length, they are very small.
The figure of 10
33
cm (a millionth of a billionth of a billionth of a billionth of a
centimeter) is easy to write down or say, but hard to grasp. Atomic nuclei are very
small billions could fit into a pinhead but strings are a hundred billion billion
times (10
20
) smaller still. The known universe is very big. The star nearest to us,
Alpha Centauri, is a hundred million times further from us than the Moon (a
mere 240

000 miles); the diameter of our galaxy, the Milky Way, is about 20

000
times greater than the distance between here and Alpha Centauri; there are about
a hundred billion other galaxies in the universe. After pondering on this for a
moment, imagine a single atom boosted in size so it fills the known universe. How
wide would the individual strings in this atom be after this boost? Not much wider
than an average house.
The minuscule size of strings means we may never be able to detect their exist-
ence directly; we would need an accelerator a million billion times more powerful
than any now available to smash matter into individual strings. The explanation
396 TIME AND SPACE
for the minuscule scale of strings is as remarkable as the scale itself: internal string
tension. Calculations suggest that this is of the order of 10
39
tons: the so- called
Planck tension. It is because strings are so taut that they find themselves pulled
into so compact a size.
The fact that strings (assuming they exist at all) are orders of magnitude smaller
than atoms, which are themselves incredibly small by ordinary standards, makes
for a problem. As things stand, it is not possible to subject the predictions string
theory makes about the character of the micro-realm to direct experimental tests.
It may well be that we will never develop probes that are able to delve down so far.
For some of the more radical of string theorys critics, this means the theory is not
properly part of physics at all: it is mere metaphysics, or post-modern science.
Although conceding that the lack of empirically testable predictions is a real prob-
lem, string theorists themselves are less pessimistic. The approach is comparatively
new, and still under development; perhaps currently unforeseen ways of testing
it empirically will emerge when the theory itself is better understood. Also, the
theory does yield some predictions notably, that ordinary particles will turn out
to have supersymmetric partners which may soon be testable directly. They can
also point out that accepting the story the theory tells about the micro-realm can
be abductively supported. Provided the theory provides a better explanation of the
observable parts of the universe than any of the alternatives, we would be justified
in taking seriously the claims it makes concerning the properties of elementary
particles, even if these claims can never be directly tested empirically. But while
abduction (or inference to the best explanation) is by no means unknown in
science, there are those who question its legitimacy (see van Fraassen 1980). And
even if this mode of inference is legitimate, it may be a long time before we know
whether a version of string theory really is the best explanation for the observable
phenomena in our universe; some physicists believe other approaches are more
promising. For a more detailed survey of the problems facing string theory, and a
discussion of some of the alternatives, see Smolin (2006) and Woit (2006).
But even if string theory is more problematic than its more enthusiastic support-
ers like to admit, it does have successes to its name, and many physicists (although
not all) believe it is currently the most promising path to a quantum theory of
gravity.
3
What makes it of particular relevance here are its distinctive and dramatic
implications for one of primary concerns: the nature of space. Over the next two
sections we shall take a brief look at these.
22.5 CalabiYau space
Investigations into the earliest versions of the theory revealed a problem. Although
the infinities associated with point- interactions did not arise when particles were
taken to be one dimensional rather than zero dimensional, the equations gener-
ated predictions that featured negative probabilities; another reliable sign that
something has gone awry. Further work suggested a surprising solution. It was
discovered that the negative probabilities were connected with the number of
different directions in which strings can vibrate. If string vibrations are confined
to the directions available in three- dimensional space the negative probabilities
STRINGS 397
cannot be avoided; moving to four or five dimensions doesnt help. But move to
nine spatial dimensions (plus one for time) and the problem disappears completely.
Does this mean that string theory must be false? No, for as we saw earlier, Kaluza
and Klein discovered a way in which our universe could have more spatial dimen-
sions than it seems to have: if the additional dimensions are curled up enough they
would be effectively invisible. Since strings are themselves very small, they do not
need much room in which to vibrate: the six extra dimensions need only be of the
Planck magnitude.
String theorists were thus led to resuscitate the discarded KaluzaKlein theory,
but in a more radical guise. Kaluza and Klein introduced one extra dimension
to unify relativity and electromagnetism (and the attempt failed); string theory
introduces six extra dimensions, but in so doing provides a unified framework that
encompasses all known particles and all known interactions.
The extra six dimensions can be compactified in a multitude of different ways,
each compactification corresponding to a different six- dimensional manifold. We
saw earlier how in KaluzaKlein theory the extra dimension takes the form of a
circle attached to each point of the familiar four- dimensional spacetime manifold.
This strategy can be extended: instead of attaching one circle we attach six in
effect a six- dimensional torus to each point in ordinary space. But while this
simple solution is often adopted (it makes calculations easier) it turns out not to
deliver quite the right results. It has been demonstrated that to get the right results
the extra dimensions must compactify on to a particular sort of six- dimensional
space: a CalabiYau manifold. Unfortunately, CalabiYau manifolds themselves
come in many thousands of forms, and it is not yet clear which is required for
string theory (recent work has suggested that different CalabiYau manifolds are
related through conifold transitions, which may simplify things). Being six-
dimensional, there is no way to visualize a CalabiYau manifold in all its complex-
ity, so Figure 22.9 is not realistic.
4
It is of course rather odd to think that with
every move we make each and every part of us is continually shifting through
these contorted six- dimensional manifolds, but dont forget their size: the result-
ing wobbles are exceedingly slight.
This multiplication of dimensions is only one of the ways in which string the-
ory requires us to revise our preconceptions concerning space. Another concerns
Figure 22.9 Concealed corridors of space.
A single point in four-dimensional spacetime,
with an impressionistic rendering of the
additional six dimensions at the Planck scale.
398 TIME AND SPACE
topology. As was noted earlier, quantum theory predicts that at very small scales
space resembles a turbulent frothing liquid, with the spatial fabric being repeat-
edly torn apart and stitched back together again, with the consequence that space
does not have a stable unvarying topology. General relativity allows space to bend
and flex in response to variations in mass- energy density, but it does not permit
topology- destroying tears in the manifold (which is one reason why black hole
singularities are thought by many to be a disastrous consequence of GTR). String
theory does. Work in the 1990s proved that a certain kind of topology trans-
formation (a flop) could occur within a CalabiYau manifold without drastic
physical consequences. Indeed, it now seems that such processes are common-
place, assuming that current versions of string theory are pointing us in the right
direction.
5
More intriguing still is the way in which string theory provides space with an
entirely new role. Newton ascribed two functions to space. By virtue of being
the medium in which all physical things exist, it constrains (or makes possible)
the ways material bodies can move, and hence the ways in which they can be
spatially related; it also grounds the distinction between inertial and non- inertial
motion. With the advent of GTR, space (or spacetime) continues to fulfil these
functions, but it loses its absolute status: not only is its structure no longer
fixed, but its structure is responsive to movements of matter. In string theory all
this is retained space still determines possible movements but it does some-
thing more: it directly controls the properties that material things possess and can
possess.
Why does an oboe sound different from a saxaphone? To a large extent it is
because of the way air can vibrate in the interior spaces of the instrument; differ-
ent shapes permit different patterns of vibration, which results in different tonal
qualities. As we saw earlier, elementary strings take on different physical attributes
depending on how they vibrate, and as with the air in a musical instrument, the
way they can vibrate depends on the shape of the space they inhabit. As a string
moves through space, pulsing as it goes, it travels through the twisting byways of
the compactified CalabiYau manifolds, the complex geometry of which signifi-
cantly affects the vibrational patterns available to the string, and hence its physical
properties:
Since tiny strings vibrate through all of the spatial dimensions, the precise
way in which the extra dimensions are twisted up and curled back on each
other strongly influences and tightly constrains the possible resonant vibra-
tional patterns. These patterns, largely determined by the extradimensional
geometry, constitute the array of possible particle properties observed in
the familiar extended dimensions. This means that extradimensional geom-
etry determines fundamental physical attributes like particle masses and
charges that we observe in the usual three large space dimensions of common
experience. (Greene 1999: 206, original emphasis)
The italics are appropriate; it is not every day that a new function for space is
discovered.
STRINGS 399
22.6 Branes and bulk
Other developments in string theory also have interesting implications for the spa-
tial character of our universe. By the early 1990s, physicists had begun to realize
that most versions of string theory permit (or require) objects that are not confined
to just one line-like spatial dimension in the manner of strings themselves. Among
the simplest are a two-dimensional membrane whose energy when stretched is
proportional to its area not just its length, and a three-dimensional membrane
whose energy is proportional to its volume rather than its area and so on, mutatis
mutandis, for higher-dimensional cases. The term brane was introduced to refer
to objects of this kind: two-dimensional membranes are known as two-branes,
three-dimensional blob-like objects are known as three-branes, and so on up
the dimensions.
In 1995, Edward Witten provided reasons for supposing that the five version
of string theory might well all be manifestations of a single more basic theory,
labelled M-theory. The unification comes at a cost: M-theory posits a total
of ten spatial dimensions, rather than nine, to make up an eleven-dimensional
spacetime. Moreover, in the case of a model studied by Hor ava and Witten, the
additional tenth spatial dimension is extended, rather than compactified. This
extended tenth dimension sometimes called the bulk can house and be
bounded by lesser branes. Of particular interest are models that include branes of
nine dimensions, six of which are compactified; since branes such as these possess
just three extended spatial dimensions, they have the potential to house (or be)
worlds that resemble our own. These braneworlds open up many new possibilities,
which particle physicists and cosmologists have been exploring with enthusiasm.
I shall mention only a few of the more intriguing.
First of all, braneworld models have proved useful to those trying to account for
the weakness of the gravitational force. The idea that gravity is weak may come as
a surprise to anyone who has broken a limb after falling from a tree or ladder, or
even watched a science fiction movie featuring an asteroid smashing into a planet,
with devastating effect, under the influence of gravity alone. Clearly, gravity can and
does exert a powerful grip on material things, but it is nonetheless weak in compari-
son with the other basic forces known to science: those associated with the weak
force (which governs interactions among quarks), the strong force (which holds
atomic nuclei together) and electromagnetism (which holds atoms and molecules
Figure 22.10 Branes and bulk. Each of the solids
represents an entire (possibly infinite) three-dimensional
space. Provided all the familiar particles and forces (with
the sole exception of gravity) are confined to the three-
dimensional surface of our brane, we would be oblivious
to both the existence of the surrounding bulk and the other
brane even if the distance separating us from it is very
small indeed, for example a millimetre or even less.
BRANE
(ours)
BRANE
BULK
400 TIME AND SPACE
together). More specifically, gravity is roughly 10
39
times weaker than the strong
force, which in turn is roughly a hundred times stronger than the electrical and mag-
netic forces. Ordinary experience provides evidence of this enormous divergence.
If we try, most of us can jump a few inches off the ground; proof that our leg mus-
cles alone are enough to allow us to (briefly) overcome the gravitational influence
of the entire Earth, which weighs some 6,000,000,000,000,000,000,000,000 kg
(6 10
24
kg). For purposes of comparison, think of how difficult it can be to pull
apart two small magnets, each weighing just a few hundred grams. The vast differ-
ence in strength between gravity and the other currently recognized basic forces
may not be a problem at the level of logical possibility, but it does pose significant
problems for those seeking to unify all the forces within a single theoretical frame-
work: their task would be a good deal easier if the disparity between gravity and the
other forces could be diminished.
Why might braneworld models play a useful role in this regard? Because of the
distinctive way in which gravity is treated in them. All ordinary particles pho-
tons, electrons, quarks are construed as open or two-ended strings, and since
they are firmly bound to the surfaces of branes they are not free to move into the
surrounding bulk. It is for precisely this reason, of course, that the inhabitants of
a brane would take themselves to be living in a three-dimensional space: they are
unable to see or touch anything beyond their brane. By contrast, gravitons, the
hypothetical particle associated with the gravitational influence, are closed strings,
and these are free to move into the bulk, unlike the other members of the standard
models particle zoo. These additional degrees of freedom are suggestive: might
the comparative weakness of gravity be linked to its ability to extend through
the higher dimensions? In 1999, Lisa Randall and Raman Sundrum published a
detailed specification of a braneworld scenario with precisely this feature. In their
RS-1 model the universe consists of two three-dimensional branes separated by
a tiny gap in a five-dimensional spacetime. On one of these branes the gravity
brane gravity is very heavily concentrated, and is similar in strength to the
other basic forces. We live on the second brane; since comparatively few gravitons
manage to cross the bulk to our brane, we experience gravity as a comparatively
weak force. Other scenarios, some even more exotic at least by common-sense
standards have been developed; for informal introductions to these see Randall
(2005).
Assistance with gravity aside, as Brian Greene observes, these braneworld
models have implications for the long-running dispute between substantival and
relational conceptions of space:
If we are living within a three-brane if our four-dimensional spacetime is
nothing but the history of a three-brane through time then the venerable
question of whether spacetime is a something would be cast in a brilliant new
light. Familiar four-dimensional spacetime would arise from a real physical
entity in string/M-theory, a three-brane, not from some vague or abstract
idea. In this approach, the reality of our four-dimensional spacetime would
be in a par with the reality of an electron or a quark. (2004: 391)
In effect, if we live in a three-brane, we are occupying the interior of a very
large elementary particle, something with the same material and ontological
STRINGS 401
characteristics of any of the other basic constituents of matter. The reality of the
surrounding ten-dimensional bulk space may still be in question assuming that it
is not itself a brane but the substantival character of our space would not longer
be in any doubt.
There is a further interesting exploitation of the braneworld scenario, perhaps
the most striking (so far) of them all: the cyclic model of the cosmos, developed
by Paul Steinhardt and Neil Turok as an alternative to the standard big bang cos-
mological model. According to the latter (in its contemporary form) our universe
came into existence some thirteen or fourteen billion years ago, in the form of
a massive explosion from an initial singularity, a dimensionless point of infinite
pressure and density; soon after this birth-event, the universe underwent a brief
period of rapid inflationary expansion, a process that effectively smoothed and
flattened our space, providing it with the characteristics we find today. The infla-
tionary big bang model explains a great deal, and it has been developed in great
detail over the past few decades, but there are nonetheless questions it does not
answer. How and where did the mass-energy in the initial singularity originate?
What, if anything, came before the big bang? The ability of physics (as opposed to
metaphysics) to furnish answers to such questions is seriously compromised by the
defining feature of the model itself: since all our main physical theories cease to
apply in the infinitely dense singularity, it is impossible to use these laws to probe
any further back in time.
The starting point of the cyclic model is a hypothesis as to the structure of the
cosmos. We are to suppose it consists of two infinite three-dimensional branes,
separated by a small distance, rather like two (infinitely big) panes of glass in a
double-glazed window, but with one key difference: unlike the panes of glass,
these branes are not fixed but are able to move back and forth through the bulk.
We are to further suppose that there is an attractive force that pulls the branes
together, a force that is very weak when the branes are at their maximum dis-
tance apart, but that gains in strength rapidly when they approach. The history of
this (model) universe takes the form of a series of collisions. Every trillion years,
or so, the two branes accelerate towards one another, and then collide, and the
collision generates vast amounts of mass-energy, resulting in the branes moving
rapidly away from one another again. After another trillion years, or so, the whole
process repeats. Each of these collisions corresponds to a distinct big bang, and
all the matter and energy we find in the universe today is a product of the most
recent collision. After each collision, the space within each of the branes expands
under the influence of dark energy, in conformity with current observational data.
Whereas the orthodox big bang is spatially localized at a single point, the big bangs
in the cyclic model are not: when the branes collide, the whole of infinite space
is filled with energy, and quickly reaches a temperature around 10
15
times greater
than found on the surface of the sun, which is more than sufficient to obliterate all
structures (black holes excepted) from the previous cycle. Since these bangs are
spatially extended, there is no concentration of vast quantities of mass-energy into
point-sized regions, and so no singularities. As a consequence, our current theories
can extend right through each phase of the process (or so it is hoped work on
this issue continues). All of space may be filled with a creative energy, as it were,
but this energy is not distributed in a perfectly homogeneous manner. Thanks to
402 TIME AND SPACE
quantum uncertainty, small irregularities occur in the surfaces of the branes as they
come together, and these lead to a slightly irregular distribution of energy after
each collision, of a kind that permits matter to cluster under the influence of grav-
ity, a process that eventually leads to galaxies forming. Importantly, the structure
of these irregularities corresponds closely with the scale-invariant pattern in the
early universe detected by the COBE and WMAP satellites. Although new matter and
energy are created at the start of each cosmic cycle, fresh energy is also drawn from
the (inexhaustible) gravitational field, so the dynamics of the process are such that
there is no reason why it cannot continue forever, and no reason why it ever had
to have a beginning.
Needless to say, this cyclic cosmological model is comparatively new at least in
its brane-theoretic form and it is still very much a work-in-progress (for further
details see Steinhardt & Turok 2007). But it is obviously and interestingly different
from the more established inflationary big bang model, not least because of the
new role allocated to space. Far from being a mere passive container, space (in
brane-form) is now cast in the role of a maker of worlds or so it might be argued.
Since we cannot reach into the bulk, we cannot directly test for the existence or
non-existence of a neighbouring brane, but the cyclic model is nonetheless vulner-
able to empirical refutation. The orthodox inflationary process is so violent that
it would create ripples in the spacetime fabric in the form of large-scale cosmic
gravitational waves; the brane collisions of the cyclic model are also cataclysmic,
but they are not so violent as to produce the same sort of large-scale gravitational
waves. Hence experiments due to be carried out over the next decade, aimed at
detecting cosmic gravitational waves, have an added interest: as well as confirming
a prediction of Einsteins GTR, they may allow us to rule out at least one of these
competing cosmological models.
22.7 Shards
Current formulations of string theory unashamedly presuppose a spacetime mani-
fold. Indeed, as we saw in 22.5, the physical properties of strings are controlled
by the manifolds structure. Might this change? Can the spacetime manifold itself
be explained in terms of something else? In a speculative vein, Greene suggests
that it might:
we can ask ourselves whether there is a raw precursor to the fabric of space-
time a configuration of strings of the cosmic fabric in which they have
not yet coalesced into the organized form that we recognize as spacetime.
Notice that it is somewhat inaccurate to picture this state as a jumbled mass
of individual vibrating strings that have yet to stitch themselves together into
an ordered whole because, in our usual way of thinking, this presupposes a
notion of both space and time the space in which a string vibrates and the
progression of time that allows us to follow its changes in shape from one
moment to the next. But in the raw state, before the strings that makeup
the cosmic fabric engage in the orderly, coherent, vibrational dance we are
discussing, there is no realization of space or time. Even our language is too
STRINGS 403
coarse to handle these ideas, for, in fact, there is no notion of before. In a
sense, its as if individual strings are shards of space and time, and only
when they appropriately undergo sympathetic vibrations do the conven-
tional notions of space and time emerge just as we should allow our artist
to work from a blank canvas, we should allow string theory to create its own
spacetime arena by starting in a spaceless and timeless configuration.
(1999: 3789)
He goes on to note that a number of leading string theorists are working along
these lines.
6
In the light of the many hard- to- believe- but- possibly- true scientific theories we
have encountered thus far it would clearly be a mistake to prejudge this issue. The
idea that ordinary material things are constructed from more basic building blocks
has proved fruitful; perhaps the same will apply to what underlies material things.
Perhaps space is a construct from more basic ingredients, and perhaps these are
strings themselves, or related entities such as the zero- branes, currently viewed
as perhaps the most basic ingredient in M- theory, or alternatively, the constituent
parts of the tangled spin networks of loop quantum gravity theory, one of string
theorys current rivals. However, there is an obstacle that all such approaches must
overcome.
When discussing the connection problem in 9.3 I suggested that any theory
that eliminates space (or spacetime) as a fundamental entity has to provide an
account of how the entities that remain can connect with one another. If spacetime
is the product of a cooperative string enterprise, the strings doing the work must be
able to interact with one another, and a story must be told of how this is possible
in the absence of space and time. More generally, no matter what the nature of the
pre- spatial shards that are jointly responsible for the physical world as we know it,
there must be a connection of some sort between them, for in the absence of such
a link these entities would in effect consist of entirely distinct island universes
in their own right. A substantival manifold is one mode of connection, but it is not
the only mode. Perhaps there are pre- spatial and pre- temporal shards linked by
relations of some sort if so the relations must themselves be real ingredients of
reality in their own right or perhaps the shards are themselves directly connected
to one another (we shall be considering such a case shortly). The mistake to avoid
is supposing that these ingredients could start off isolated in the void and then
come together to constitute an extended spatial network. Nothing can come
together in the void; it is not that sort of place. The component parts of a space
must come into being together, as a connected (even if growing) whole. Space may
not be indispensable, but it is not easily disposed of.
A paradigmatic relational world comprises material objects separated by void
(nothingness) but connected by spatial relations of the intrinsic variety (the precise
nature of which is usually left unspecified); the paradigmatic substantival world,
by contrast, consists of a continuous spatial manifold that may, or may not, contain
some material objects. There are possible universes that are spatial, but whose
space falls between two stools, being neither relational nor substantival in the
usual senses of the terms. One such possibility is the mesh world depicted in
Figure 22.11 in Flatland form. This space is akin to a grid, with the fibres or
404 TIME AND SPACE
threads of space being separated by void (in accord with relationism), and where
the gaps in the grid are smaller than the smallest particles. Since these particles
extend beyond the gridlines they fill a small region of void (in line with classical
relationalism), but their motions are constrained by the components of the mesh
(in accord with classical substantivalism).
We can further suppose that elementary particles in this world do not move
in a smooth continuous fashion, rather they jump instantaneously from cell to
cell (there are four such jumps between t
1
and t
2
, and between t
2
and t
3
). Provided
the mesh is fine enough (e.g. sub- Planck length) these discontinuities would not
be discernible empirically; space would appear continuous, even though it is not.
Although there is certainly a relationalist aspect to this world the presence of
surrounding (or permeating) void given the role the mesh plays in constraining
motion, there is clearly a case for regarding it as a substantival space, albeit of an
unusual sort. The growing mesh depicted in Figure 22.12 shows how such a
space can expand into the non- existent unstructured void, creating new locations
and new possibilities for movement as it does so, in just the manner one would
expect of a substantival space. This model has more than mere curiosity value:
mesh space bears some resemblance to the conception of space (if not objects and
Figure 22.11 Part-void,
part-substance: mesh space.
An atom is moving through mesh
space, constrained by the gridlines
and extending through void.
t
1
t
2
t
3
t
1
t
2
Figure 22.12 A growing mesh space.
motion) to be found in loop quantum gravity theory, the competitor to string
theory mentioned above.
7
In one of its forms this theory incorporates a variant of the spin network
approach originated by Roger Penrose; instead of a spacetime manifold it posits
an evolving network of looping intersecting flux- lines. Whereas the spacetime
of GTR is continuous, that of the spin network is not, so space and time come in
discrete, quantized units. These units are incredibly small. The smallest areas are
of the order of 10
66
cm
2
, and the smallest volumes about 10
99
cm
3
. The area of a
given surface is determined by the number of edges of the network it crosses (not
STRINGS 405
the expanse of interior void), and the volume it encloses is given by the number of
nodes (or intersections) it contains. Hence circle A in Figure 22.13 encompasses a
larger area and volume of the fragment of space shown than the (seemingly larger)
circle B. The way these loops and lines intersect and knot together determines
the geometry of space at the Planck- scale (the knotted aspect of the network
would be more apparent in a three- dimensional depiction). Since the network is
constantly changing its structure, so, too, is space, albeit at this very (very) small
scale. Given their size, these geometrical fluctuations together with the discrete
character of space are invisible for all practical purposes.
Intriguingly (if also optimistically) Smolin suggests that rather than remaining
competitors, loop quantum gravity and string theory may one day come together:
space may be woven from a network of loops just as a piece of cloth is
woven from a network of threads. The analogy is fairly precise. The proper-
ties of the cloth are explicable in terms of the kind of weave, which is to
say in terms of how the threads are knotted and linked with one another.
Similarly, the geometry of the space is determined only by how the loops link
and intersect one another.
We may then imagine a string as a large loop which makes a kind of
embroidery of the weave. From a microscopic point of view, the string can
be described by how it knots the loops in the weave. But on a larger scale
we would see only the loop making up the string. If we cannot see the fine
weave that makes up space, the string will appear against a background of
some apparently smooth space. This is how the picture of strings against a
background space emerges from loop quantum gravity.
If this is right, then string theory will turn out to be an approximation to
a more fundamental theory described in terms of spin networks.
(2000a: 186)
Be this as it may, for our purposes what matters is the type of world being pro-
posed in these speculations. On various occasions Smolin suggests that not only
was the loop quantum gravity approach inspired by Leibnizian- style relationism,
but it also vindicates the latter.
8
In one sense it does: what is being proposed is
undeniably very different from Newtons absolute space. The structure of the
spin network is dynamic, its geometry is continually evolving, and the way one
part evolves depends entirely on its relations with other parts; moreover, this
evolution does not take place against the backdrop of a separate and independ-
ent spatiotemporal framework the spin network constitutes the spatiotemporal
frame work. But this is only part of the story. Leibniz was also rejecting a substan-
tival conception of space and time, and it is far from clear that space and time as
conceived in loop quantum gravity are not substantival. Indeed, the loop quantum
conception seems as substantival as any conception we have encoun tered thus far.
A
B
Figure 22.13 An evolving spin network.
406 TIME AND SPACE
It could even be viewed as vindicating super- substantivalism: rather than providing
the framework that contains and constrains material objects, space, time and the
properties they possess are now being viewed as the sole constituents of physical
reality. What could be less Leibnizian?
However, we must not too get carried away by all this. The string and loop
theories have fascinating implications, likewise the canonical approach and some
of the other more radical approaches being considered, but it is too early to tell
whether a viable quantum gravity theory assuming that physicists succeed in
developing one will resemble anything so far conceived. Nonetheless, since these
theories are the best guides we have as to what the future may bring, it may very
well be that the next revolution in physics will be as remarkable as any of its pred-
ecessors, and its metaphysical implications just as difficult to discern.
Notes
1. Preliminaries
1. In the first edition I often used the term static conception, or the static Block view when
talking about the conception of time that in this edition I usually refer to simply as the Block
view. The use of static had a clear rationale, since proponents of this conception deny that
passage exists, and hence deny that there is anything resembling a privileged moving present;
in this respect, the Block model is a static conception of time. However, talking of static time
also has its disadvantages, for it can also suggest a conception of time in which nothing at all ever
moves or changes, ordinary objects such as cars, trains, clouds (and human beings) included, and
this is certainly not what Block theorists maintain: objects in a block-universe move by varying
their spatial locations and, more generally, undergo change by virtue of having different properties
at different temporal locations. In his review (Nerlich 2002), and subsequent correspondence,
Graham Nerlich succeeded in persuading me that the disadvantages of static in this context
significantly outweigh the advantages; my thanks to him for his sustained effort in this regard. And
Nerlich was not alone: Nathan Oaklander (2003) made a similar complaint. Hence the change in
terminology in this second edition.
2. Parmenides B8, Barnes (1982: 178).
3. The temporal determinism entailed by the Block view should not be confused with nomological
determinism. In a world where the laws of nature are deterministic, how things are at any one time
fixes how things will turn out at later times (as well as how things were at earlier times). One can be a
nomological determinist without being a temporal determinist (by holding that although the present
completely determines how the future will unfold, as of the present time future events are unreal).
And one can be a temporal determinist without being a nomological determinist (by holding that
although future events are just as real as present events, they are not nomologically deter mined by
them). Since many contemporary physicists believe that our world is (nomologi cally) indeterministic,
the greater threat (if threat it be) may well be posed by determinism of the temporal variety.
4. McTaggart (1908). It has been claimed that the distinction is also to be found in the writings of
Iamblichus (died c. 325 CE), see Sorabji (1983: ch. 3).
2. McTaggart on times unreality
1. Shoemaker (1969) describes a world (quite unlike ours in some respects there are regular local
freezes) where it would be reasonable to posit the occurrence of occasional universe- wide tempo-
ral vacua.
2. It is important to note that the A- theorist cannot seek to avoid the paradox by interpreting succes-
sively in a B- series way. Suppose we say something along these lines: No event is past, present and
future at the same time. An event E at t is present at t, past at times earlier than t and future at times
later than t. This is tantamount to denying the reality of A- series change: all events are equally and
timelessly present; an event doesnt change its A- attributes it is always true that E is present at t,
likewise it is always true that before t, E lies in the future and after t, E lies in the past.
3. I do not know of any contemporary moving spotlight theorists who have defended the doctrine
in precisely the moderate manner outlined here, but some have certainly embraced similar (if
not stronger, and arguably stranger) positions. Dean Zimmerman summarizes thus: Todays
408 NOTES
spotlighters generally agree that when an object ceases to be present, it loses a great many of its
most interesting properties. Quentin Smith [2002] and Timothy Williamson [1999], for example,
believe that the entire block of past, present, and future individuals and events exists; but that,
when objects and events pass from being future to present to past, they change in much more
than just their A-properties i.e., their presentness, or their degree of pastness or futurity. For
example, they think that future things and events are not located anywhere in space until they are
present; and, when past, they again become non-spatial though of course it remains true that
they once occupied space. These spotlight theorists strip past and future objects and events of all
their interesting intrinsic properties, too. A table yet-to-be-made has no shape or mass or color;
and when it is destroyed, it will lack these properties as well (2008: 215). For a more detailed
survey of the options available to the spotlighter, see Zimmerman (2005). For a rather different
defence of the moving-now view, see Skow (forthcoming).
3. The Block universe
1. Competing accounts have their adherents, but none of these is as popular as the B- theory.
2. See Garrett (1988); for other responses to Priors argument see Oaklander & Smith (1994: part
III) and Gallois (1994).
3. Derek Parfit argues for this view by considering Timeless, a being who lacks the bias towards the
future: Our bias towards the future is bad for us. It would be better for us if we were like Timeless.
We would lose in certain ways. Thus we should not be relieved when bad things were in the past.
But we should also gain. We should not be sad when good things were in the past There would
be other, greater gains. One would be in our attitude to ageing and to death We do not regret
our past non- existence. Since this is so, why should we regret our future non- existence? If we
regard one with equanimity, should we not extend this attitude to the other? As our life passes,
we should have less and less to look forward to, but more and more to look backward to In
giving us this [future] bias, Evolution denies us the best attitude to death (1984: 1737).
Einstein was of a similar view. Shortly before his own death, he wrote to the children of one of
his oldest and closest friends, Michele Besso, who had just died: And now he has preceded me
briefly in bidding farewell to this strange world. This signifies nothing. For us believing physicists,
the distinction between past, present and future is only an illusion, even if a stubborn one (quoted
in Hoffmann 1972: 2578). Many of us regard death as a misfortune, as something to be feared
and dreaded, for we equate it with the onset of complete non- existence. In a Block universe this
fear is misplaced, for if all times and events are equally real, death does not bring absolute annihila-
tion, lives do not cease to exist, they simply have beginning and ends. If it were not for our deep-
seated (stubborn) bias towards the future, this idea would be more comforting than it actually is.
4. In a four- dimensional space, the distance between two points is given by an extended version of
Pythagoras theorem: d
2
= w
2
+ x
2
+ y
2
+ z
2
; the spatio temporal distance (or interval) between
two spacetime points in special relativity is given by an equation that treats temporal intervals
differently from spatial intervals. See 18.5.
5. We will encounter this hypothesis again, in a rather different context, in 6.8.
6. If you find it hard to grasp what having your belly plump in time amounts to, it may help to
imagine the time axis in Figure 3.3 rotated counter- clockwise so it coincides with (and displaces)
the leftright axis. Alices life now unfolds through space, rather than time (or rather, the dimen-
sion that the rest of us call time). In a footnote Williams adds, I should expect the impact of the
environment on such a being to be so wildly queer and out of step with the way he is put together,
that his mental life must be a dragged- out monstrous delirium. In this he is surely right, although
if he had been writing this passage now he may well have wanted to reconsider the remark about
our comparative indifference to our spatial girth.
4. Asymmetries within time
1. To keep the discussion to a manageable length I have simplified somewhat. A fuller discussion
would include the radiation asymmetry (we often encounter ripples (e.g. of water or light)
moving away from a common source- point, but rarely do we encounter converging ripples, even
though they are physically possible), and the counterfactual asymmetry (true counterfactuals
NOTES 409
generally concern future rather than past possibilities e.g. If I had struck the match, it would
have lit versus If the match had lit, it would have been struck).
2. In fact, it is far from obvious why a big bang should be a low- entropy event; a universe that is very
small in spatial size and extremely hot throughout (which is how our universe is envisaged as being
just after the initial explosion) would be in a state verging on maximum thermodynamic disorder,
and hence close to maximum entropy. The inflationary big bang model provides a solution; a
period of very rapid spatial expansion in the very early universe has the effect of creating a sudden
decrease in entropy, hence the subsequent tendency for entropy to increase. Although they are no
longer in favour, Boltzmanns cosmological speculations are remarkable on several counts. As well
as providing a statistical account of the origin of complex structures such as stars, he also predicted
that time- reversed regions of the universe exist. Think of the equilibrium condition of the universe
as a whole as a straight baseline and the occasional eruptions of order as hills or humps along this
line; each summit represents a state of maximum local order (minimum entropy). It follows (given
the entropic account of the direction of time) that we inhabit a slope of one of these humps, and that
the summit of our hump is in our past; ahead of us (in our future) lies disorder and equilibrium. But
what lies behind, on the far side of the hump? Boltzmanns answer: a region of the universe where
entropy is also increasing, and hence where the local arrow of time points in the opposite direction
to ours. Hence, for Boltzmann, the arrow of time is only a regional phenomenon, and time- reversed
regions of the universe are possible. Of course, this is all based on the assumption that the various
observable temporal asymmetries depend on entropic increase, which is precisely what is at issue.
3. Consider the following reasoning: The processes that are involved in the production of records
(such as human memories, or video films, or books) invariably lead to entropy increase, since
energy is expended and spread when making the record. Thus the tendency for entropy to increase
explains why it is that we have records of the past but not the future. The fact that producing
records is an entropy-increasing process doesnt in itself explain why the records produced provide
information about the past but not the future. For more on Boltzmann and Boltzmanns views, see
Lindley (2001), Price (1996, 2002a,b), Carroll (2010) and North (forthcoming).
4. Cf. Mellor (1998: 10.4). Since Mellor takes causation to relate facts rather than events, his
formulation differs from that given here.
5. Simultaneous causation poses a similar problem. As noted earlier, Mellor argues that causes cannot
be simultaneous with their effects, but many disagree.
6. This was not the first appearance of this notion in science. As noted earlier, Boltzmann deduced
the possibility of time- reversed regions of the universe from his own premises.
7. There are, however, certain complications. Time-reversibility can mean different things in the
context of different physical theories. See Carroll (2010: ch. 7) and North (2008, forthcoming)
for further discussion.
8. Are content-reversal scenarios of this sort really compatible with the physical laws that operate in
our universe? Some have their doubts; see Maudlin (2007: 3). Others believe that these doubts
are misplaced; see Price (forthcoming: esp. 3.10, Maudlin v. Boltzmann on backward brains).
9. Horwich is far from alone in his view of causation, so his approach is just one implementation of
a more general strategy. But few other implementations are so ambitiously wide- ranging.
10. See Price (1996: ch. 6) for a critical assessment of the strategy of grounding the causal asymmetry
on the fork asymmetry. (Augmenting Horwichs fork asymmetry with the more general overdeter-
mination asymmetry cf. Owens goes some way towards reducing the force of Prices argument.)
Although critical of the fork proposal, Prices view of causation is similar (broadly speaking) to
that of Horwich.
5. Tensed time
1. Lowe (1987, 1993). I will draw mainly from Lowe (1998).
2. There are similarities with deflationary theories of truth.
6. Dynamic time
1. There is at least one advocate of the Growing Block who combines the view with transitory intrin-
sic property variations: see Forrest (2004, 2006), who holds that consciousness only exists at the
410 NOTES
leading edge of reality, where being and non-being meet. Transitory intrinsic properties lead to the
overdetermination problem we encountered in 2.6. How can any subject be both conscious and
non-conscious at the same time? For this reason we shall not be considering this hybrid model here.
2. Tooley argues that the dynamic theorist also needs a more general and non- temporal concept of
existence (or actuality) existence (or actuality) simpliciter so as to be able to make sense of talk
of non- temporal entities (such as abstract things), and the concept of a total, dynamic world, as
contrasted with the history of a dynamic world up to some point in time (1997: 40). It is not
clear that this is a necessary move, or a wise one: how can states of affairs that are not actual as of
one time become actual as of another time if the states of affairs in question are actual simpliciter?
I prefer not to saddle the Growing Block model with Tooleys timeless notion of actuality simplic-
iter, and so will ignore it in what follows.
3. The model has evolved over the years. See, among others, McCall (1976, 1984). I rely here on
McCall (1994).
4. Or at least they do in universes where indeterminism holds sway. In deterministic universes there
is only one nomologically possible future, and so McCalls universe- tree is so slender (and branch
free) that it is indistinguishable from a static Block universe.
5. It should be noted that the extended case that McCall mounts for his model in A Model of the
Universe (1994) rests entirely on its explanatory potential (e.g. Chapter 4 provides an interesting
interpretation of quantum theory, Chapter 5 is devoted to probability theory). But while the model
does have explanatory and problem- solving potential, there are competing solutions to the same
problems that bring less surplus baggage.
6. Causal laws, rather than being merely regularities in the history of the world, control the course
of history; they underlie, and account for, any patterns that the world may exhibit over time. But
how is this control to be understood? One way and, I think, the only satisfactory way is if
causal laws, in conjunction with what is actual as of a given time, determine what states of affairs
will ultimately exist, in a tenseless sense (Tooley 1997: 111).
7. Other ways of making sense of the situation are available to Growing Block theorists. Rather than
talking in terms of occurrences*, they can eschew any reference to episodes of absolute becoming,
and talk instead in terms of variations in the sum total of reality. Accordingly, they can hold that
thought T is PRESENT in that sum total of reality in which there is no time later than t, the time at
which T occurs.
8. The treatment of this issue has been slightly expanded (and I hope improved) in this second edition;
although it followed the same general line, my discussion in the first edition has given rise to some
misunderstanding. In their recent discussions of the issue, Braddon-Mitchell reaches the conclusion
that in the framework of the Growing Block theory we cannot be justified in believing that our
own time is the PRESENT, and Bourne (2002, 2006a: ch. 1) reaches the same verdict. Button argues,
in some detail (2006, 2007), that Growing Block theorists do have the resources to overcome the
problem, provided they are prepared to recognize irreducible tensed facts the sort of facts Tooley
(for one) very much wishes to avoid. As noted earlier, Forrest responds in a different and intriguing
way. On his dead past proposal, the inhabitants of the past are zombies; they are entirely lacking in
consciousness because Forrest suggests consciousness arises only at the PRESENT, on the interface
between existence and non-existence, generated by incomplete causal processes (2004, 2006). This
speculative suggestion may solve the how do we know it is now now problem conscious states
of the sort we enjoy are found only in the PRESENT but at a cost of overdetermination: subjects
are both conscious and non-conscious at precisely the same times. For some further discussion of
Forrests proposal see Braddon-Mitchell (2004: 202) and Heathwood (2005).
9. For an online listing of the recent literature see http://philpapers.org/browse/presentism.
10. Any Block theorist who adopts a strict Humean view of causation (and so denies real connec-
tions between events occurring at distinct moments) may well be endorsing a position that is very
similar to that of the Many-Worlds Presentist.
11. A phase space is an abstract mathematical representation of a dynamic material system, often of
infinite dimensions, whose coordinates are given by the set of independent variables characterizing
the state of the system. Each individual point in the phase space corresponds to a possible state of
the entire system (in this case, the entire universe), and paths through it correspond to possible
histories of the system; or at least they do when time is one of the coordinates. If time is absent,
this interpretation is not available.
NOTES 411
12. The programme consists of casting general relativity in Hamiltonian form, that is treating it like
any other quantum theory. Another approach string theory posits basic items that are already
quantized. We will be taking a brief look at strings in Chapter 22.
13. For details see Belot & Earman (1999, 2001), Butterfield & Isham (1999), Kuchar (1999); for a
thorough and illuminating review of Barbour, see Butterfield (2002).
14. If these presents have a very small but nonetheless finite duration, this talk of one present suc-
ceeding another can be taken literally. If each present is durationless, and presents are densely
ordered so that between any two there is always another then it cannot be taken literally: in a
densely ordered series of moments no moment is directly followed (or succeeded) by another. For
more on these matters, see Chapter 16.
15. Crisp, himself a Presentist, develops the triviality charge in a different way (2003, 2004). Sup-
pose we take the Presentists key claim to be: (Pr) Only present things exist. Is the exist here
tensed or tenseless? It can easily be read in either of these ways. Indeed, Crisp argues, there seem
to be two tensed ways of construing the claim. It might be read as: (Pra) Only present things exist
now (i.e. at present). But since this is a trivial truism it is hardly news to be told the only things
existing in the present are those that exist in the present it is reasonable to conclude that the
Presentist has something else in mind. Perhaps by exist they mean something more general and
wide-ranging. Perhaps they have the following in mind: (Prb) Only present things existed, exist,
or will exist. But this is also problematic. Since the Dynamic Presentist wants to accept that lots
of things that dont exist now did exist or will exist, they will want to deny the truth of (Prb). So
perhaps the exists should be construed as tenseless, hence: (Prc) Only present things (tenselessly)
exist. Since this is supposed to be claim about things that exist in time, as opposed to abstract things
that exist atemporally, it is natural to suppose that it is equivalent to Only present things existed,
exist, or will exist. But since this is (Prb), and we have just seen that Presentists will want to deny
that (Prb) is true, it appears that Presentism is either trivial or false.
16. For doubts about whether Lucretius himself was a Presentist, in the contemporary sense, see War-
ren (2006).
17. A further option here is to follow Mellor (2004), and hold that causation (of the most basic kind)
is a relation between facts. A causal fact has the form E because C, where E and C are facts,
and because a sentential connective (rather than a relation between earlier and later concrete
particulars).
18. Presentists typically maintain that this successive realization of distinct presents is immune to
McTaggart-type problems, but a case can be made for thinking otherwise (see Oaklander 2004,
2009).
19. When it comes to making sense of the past, the options available to the Presentist are not restricted
to Lucretian properties and the ersatz B-series. Cameron (2008, 2010) argues that the solution may
lie in temporal distributional properties. These are the temporal analogue of spatial properties
such as x extends over several square inches, and also has a particular distribution of black spots
across it; such properties, suggests Cameron, may be possessed by presently existing things, and
are capable of fixing the history of the world, and so grounding claims about the past. Kierland
& Monton (2007) advocate brute past presentism. On the latter view, although there are no
past things, the past itself exists as an aspect of the presently existing reality: The brute past has
a certain intrinsic nature we like to think of this intrinsic nature in terms of the past having a
certain shape. This shape does not consist in a structure of things having properties and standing
in relation to one another. The past is an aspect of reality, even though no past things are. How
can this be? There is no reductive answer to this question (ibid.: 491). For a discussion of further
problems with Presentism, see Sider (2001: ch. 2).
20. As we shall see in Chapter 7, when we explore the temporal features of consciousness in more
detail, this is a more complex issue than it first appears. There are a number of different accounts
of our immediate experience of change and persistence, not all of which are incompatible with
orthodox Presentism, although the most plausible of these accounts probably is. These complica-
tions aside, the main point stands: a Presentist who grants the present zero or minimal temporal
depth does have very real difficulties in accommodating our experience. For different perspectives
on this issue see McKinnon (2003) and Hestevold (2008). Both authors argue that the temporal
features of our ordinary experience cannot be accommodated by the (orthodox) Presentist, and
this leads Hestevold in the direction of a thick presentism that is broadly similar to Compound
412 NOTES
Presentism. McKinnon also discusses a similar doctrine, Durational Presentism, but argues that it
is problematic. Simplifying somewhat, he argues that if an extended present could encompass two
successive experiences, why not three, or four, or more? If the present is, in principle, indefinitely
extendible, do we have any reason for supposing that in the actual world the present is not arbitrar-
ily long? And in the absence of such a reason, we no longer have a position that is significantly dif-
ferent from the tenseless Block view. This argument has some force, but it is by no means decisive.
It may not be easy for the Compound Presentist to justify a claim that the extended present has
a particular duration (0.5 s, say, rather than 0.25, or 0.3), but this may not be impossible. What
matters more, however, is whether the model of time being proposed by the Compound Presentist
is incoherent from a purely metaphysical perspective, and McKinnons argument does not show
that it is.
21. When outlining his fragmentalist conception of time, Fine does not commit himself to Com-
pound Presentism, but what he writes seems largely consistent with the latter: One naturally
assumes that in a correct account of reality all apparent contradictions will be ironed out. If
something is both hot and cold, it must be because one part is hot and the other cold, or because
it is both hot and cold at different times, or because being hot is somehow compatible with being
cold. But on the present view, this fundamental assumption is given up. It is taken to lie in the
character of reality that certain apparently contradictory aspects of it cannot be explained away.
Reality may be irredeemably incoherent. Under such a view, reality will be fragmentary. Certain of
the facts constituting reality will cohere and some will not. Any fact is plausibly taken to belong to
a fragment or maximally coherent collection of facts; and so reality will divide up into a number
of different but possibly overlapping fragments (Fine 2005: 281).
22. It should be pointed out that not everyone sees the status of the past in such models as unproblem-
atic; see Dummett (1978). Dolev (2000) supplies a succinct introduction to Dummetts position.
7. Time and consciousness
1. In conversation with Carnap, Einstein expressed similar qualms: Once Einstein said that the
problem of the Now worried him seriously. He explained that the experience of the Now means
something special for man, something essentially different from the past and the future, but that
this important difference does not and cannot occur within physics. That this experience cannot
be grasped by science seemed to him a matter of painful but inevitable resignation. I remarked that
all that occurs objectively can be described in science; on the one hand the temporal sequence of
events is described in physics; and, on the other hand, the peculiarities of mans experiences with
respect to time, including his different attitude towards past, present and future, can be described
and (in principle) explained in psychology. But Einstein thought that these scientific descriptions
cannot possibly satisfy our human needs; that there is something essential about the Now which
is just outside the realm of science (Carnap 1963: 378).
2. Advocates of this account include Whitehead: The final facts are all alike, actual entities; and
these actual entities are drops of experience (1929: 25). See also Sprigge (1983: ch. 1).
3. This, of course, is the problem facing Presentism of the non- compound variety, transposed to the
phenomenal level.
4. See Broad (1938) and Husserl (1950, 1991).
5. See Dainton (2000: chs 57; forthcoming b) for more on the topic of phenomenal time.
6. This example does not establish that the direction of phenomenal flow is independent of the
direction of causality; the relationship between the two is unclear; as unclear as the notion of the
direction of causation. A Humean could define a personal or subjective arrow of causation
in terms of an earlierlater distinction grounded in the direction of phenomenal flow; it seems
quite conceivable that this subjective arrow could point in the opposite direction to the objective
arrow of the inhabitants of a (seemingly) time- reversed planet. Non- Humeans (such as Tooley)
might want to say that the phenomenal and causal arrows necessarily point in the same direction,
but it is obscure how this claim could be supported, especially since it is far from clear that imma-
nent phenomenal flow depends on block growth.
7. If this claim seems implausible, consider what it is like to watch televised replays of familiar events:
you remember exactly what will happen over the next few seconds (the player will miss the penalty,
the ball will fly over the crossbar), but is the forward- flow of your experience in any way affected?
NOTES 413
8. See Lee (2007) for a thought-provoking exploration of some of the tensions between relativistic
doctrines and some common assumptions about the nature of consciousness: There are hard
questions about what kind of metaphysics a temporally extended stream might have in a Relativ-
istic world. For example, it is natural to think of perceptual consciousness as a continuous stream
that has a certain qualitative character at each moment that it is being enjoyed. Just as it is natural
to regard a material object as having a continuous series of instantaneous overall physical states
(mass, shape, etc.), so it is natural to regard consciousness as having a continuous series of overall
phenomenal characters: to put it somewhat metaphorically, it is like a densely packed series of
infinitely thin phenomenal slices. This picture comes into obvious tension with Relativity, since
in a Relativistic world, there are no moments of time, or at least, they only exist in a thin frame-
relative sense. One question is how a continuous stream of phenomenology could nonetheless exist
in space-time, in a way that is consistent with this fact (ibid.: 342).
8. Time travel
1. This class of machine undermines the Where are they? refutation of the possibility of backward
time travel popularized (if only semi- seriously) by Stephen Hawking.
2. Some science fiction tales seem to involve a different mode of time travel which isnt really
time travel: a powerful device reconfigures the present- day world so that it resembles (with the
exception of the presence of whatever travellers may be involved) the world as it was at some
past time. (I am grateful to Stephen Clark for drawing my attention to this model.) Obviously, in
this sort of case there is no objection to changing the past, since what is being changed is not
the past at all.
3. See Earman (1995: ch. 6) for further discussion of the relevant consistency constraints, and how
they should be conceived. For more on time travel and potentially problematic coincidences, see
Sider (2002) and Dowe (2003).
4. Or at least it does in worlds where the distinction between earlier and later is sufficiently well
defined. In the Gdel universe that we shall be encountering later, there is no consistent global
time order (moreover, locally speaking, causation along closed time- like curves in this world is
usually regarded as always being future directed).
5. Tooley (1997: 638) describes a world where backward and forward causation coexist, but are
contained within different spatial regions, separated by a barrier that severely restricts interaction
in that light (and nothing else) can pass through one way but not the other.
6. What follows is drawn mainly from Price (1984). For more detail see Price (1996: chs 79).
7. In 19.4 we shall see that the matter is far from clear- cut.
8. In Cramers transactional interpretation of quantum mechanics (1980, 1983, 2000), quantum
states are determined by wave- functions that propagate both forwards in time and backwards in
time, in the manner of the advanced and retarded electromagnetic waves in the Feynman
Wheeler absorber theory. See Price (1996: chs 3 & 9) for further details of such proposals, and
Maudlin (1994: 196201) for criticisms of Cramers theory.
9. See Nahin (1993: 20517) for a full survey; see Heinlein (1970), Harrison (1979) and MacBeath
(1983) for intriguing sexual loops. Some scientists take the Big Loop scenario seriously, albeit
not in this form. The idea that the universe originated in a causal loop is sympathetically explored
in Gott & Li (1998) and Gott (2001), a particularly elegant response to the What caused the first
event? question, since the universe, in effect, creates itself. Of course, the problem of why the
primordial loop exists at all remains unexplained.
10. For further discussion on this topic see Sider (2005). Keller and Nelsons defence of Presentist
time travel is itself nuanced: The presentist believes that there are past- and future-tensed truths,
but does not believe that the past or future exist. Should we then say that the presentist does not
really believe in the possibility of time-travel, but only in the possibility of odd arrangements
of past- and future-tensed truths? Perhaps we should. Perhaps presentist time-travel is not real
time-travel, but only something that parodies time-travel, in the same way that the presentist
arrangement of past- and future-tensed truths is only a parody of the past and future. But when
it comes to telling stories, even this conclusion, we think, is strong enough. The presentist is just
as entitled to tell time-travel stories as to tell any stories about things that happened in the past or
will happen in the future. Even if presentist time-travel is not real time-travel, the presentist can
414 NOTES
be just as committed to the statement There might be time travellers as she is to the statement It
might rain tomorrow( 2001: 345).
11. For a recent survey see Nahin (2001) and Arntzenius & Maudlin (2005).
12. For more advanced surveys of the prospects for time travel in light of the theories of contemporary
physics see Arntzenius & Maudlin (2005), Earman & Smeenk (2009) and Smeenk & Wthrich
(forth coming).
9. Conceptions of void
1. See also Selby (1982).
2. Think of water flowing over a stone on a riverbed. The water contains the stone, but the water in
contact with the stone is constantly changing; consequently, if we were to say that the stones place
is the inner surface of the surrounding water, we would be committing ourselves to the view that
the stone was constantly changing place, which is absurd. So Aristotle suggests that we identify
the stones place with the inner surface of the nearest stationary containing body: the river bed.
This definition sits uneasily within the rest of Aristotles cosmology. The first containing body of
the Earth and the sub- lunary world as a whole is the Moons shell, which is rotating; but so is the
next (planetary) shell, and so is the outermost shell. Consequently, the Earth itself has no place, if
we apply Aristotles own theory rigorously; a fact noted by many commentators.
3. Something akin to Newtons conception of space had been popular in neo- Platonist and mystical
circles for some time. According to Giordano Bruno (15481600) in On the Infinite Universe and
Worlds: There is a single general space, a single vast immensity which we may freely call VOID;
in it are innumerable globes like this one on which we live and grow. This space we declare to
be infinite, since neither reason, convenience, possibility, sense- perception nor nature assign to
it a limit It diffuseth throughout all, penetrateth all and it envelopeth, toucheth and is closely
attached to all, leaving nowhere any vacant space (Jammer 1993: 89).
4. Abbotts Flatland (first published in 1884 and in print ever since) is the classic literary source
(Abbott 1986). See also Dewdney (1984), Rucker (1984) and Stewart (2001).
5. Visit http://casa.colorado.edu/~ajsh/sr/hypercube.html (accessed June 2010) for a simulation of a
rotating hypercube.
6. The additional spatial dimensions postulated by string theorists only make their presence felt at a
very, very small scale. See Chapter 22.
7. See also Hollis (1967).
8. It is worth noting, however, that some contemporary physicists have floated the idea that our
universe is in fact fluid- filled. Grady has devised a model in which the material universe is a giant
crystal growing in a five- dimensional fluid. Matter is simply the fluid in frozen form (think of the
way thin shards of ice form in water that is nearly frozen). See Chown (1999).
9. See Manders (1982) and Mundy (1983) for some detailed relationalist reconstructions.
10. Space: the classical debate
1. The conversion to Newtonianism was not uniform across national and cultural boundaries. The
French in particular were reluctant to abandon Cartesianism, but did so eventually.
2. The Scholium is only a few pages long, and reproduced in full in Thayer (1974), Barbour (1989)
and Earman (1989), not to mention Newton (1729). I have given references for Huggett (1999),
which also contains useful selections from the relevant writings of Descartes and Leibniz, as well
as extracts from Newtons De Gravitatione.
3. The idea that space is an infinite, homogeneous and immovable entity, independent of material
bodies but capable of containing and constraining them, has a long history, as does the doctrine
that space thus conceived is a divine attribute. The latter doctrine dates back at least as far as
Palestinian Judaism during the Alexandrian period, and was transmitted through the Jewish eso-
teric tradition to Italian Renaissance figures such as Mirandola and Campanella, and thence to
figures such as Fludd and Gassendi, and subsequently to More and Barrow, both of whom were
Cambridge contemporaries of Newton and known to have influenced him. More, who wrestled
with the problem of space between 1648 and 1684, sought to synthesize cabalistic doctrines with
the light- metaphysics of Neoplatonism, a doctrine that associated space with light, and light with
NOTES 415
God, and which led Grosseteste to believe that geometrical optics was the key that would unlock
the secrets of the universe. Newtons early renown, it is worth noting, rested on his own works in
optics.
4. There are passages in Newtons writings in which his mystical- cum- theological bent is evident.
In the General Scholium added to the second edition of the Principia (1713) we find: He is not
eternity and infinity, but eternal and infinite; he is not duration or space, but he endures and is
present. He endures for ever, and is everywhere present; and by existing always and everywhere,
he constitutes duration and space. In Query 28 added to the second edition of the Opticks in
1706 he writes, does it not appear from Phaenomena that there is a Being incorporeal, living,
intelligent, omnipresent, who in infinite Space, as it were in his Sensory, sees things themselves
intimately, and thoroughly perceives them, and comprehends them wholly by their immediate
presence to himself . It was this last claim that attracted Leibnizs attentions, and initiated the
Correspondence. Newton only published these clarifications of his theory of absolute space in
response to theologically- inspired criticisms of the doctrine that had started to appear in response
to the first edition of the Principia.
5. Prior to Galileo it was not unknown for anti- Copernicans to argue that since a cannon ball dropped
from the mast of a ship in motion does not fall straight down to the base of the mast, if the Earth
were in motion we could expect the same to apply to objects dropped from towers on land.
6. The situation is rather more complicated: Galileo also subscribed to a principle of straight- line
(rectilinear) inertia for things such as bullets shot from guns. The fact that he did not find it
necessary to discuss the relationship between circular and rectilinear inertia is another sign of his
continued allegiance to Aristotelian doctrines, in this instance the distinction between natural and
violent motions.
7. Pondering the reasons why the true laws of dynamics took so long to discover, Barbour observes
that the law of inertia would only seem obvious when people became aware of the possibility of
free continuation of motion for ever once commenced: [A]nd from what activities could man
have gained such an awareness? Skating perhaps, but that was scarcely a sport practised by the
ancients in the Mediterranean. (One wonders if it is entirely fortuitous that the first reasonably
clear statement of the law of inertia in its modern form was made by a man Descartes who
spent most of his adult working life in Holland, at a time moreover in which skating had become
a great vogue.) (1989: 35).
8. It follows from this that in the strict sense there is no motion occurring in the case of the Earth or
even the other planets, since they are not transferred from the vicinity of those parts of the heaven
with which they are in immediate contact, in so far as those parts are considered as being at rest
(Principles II, 28).
9. Although Leibniz was a long- standing critic of Newton (in 1705 he had argued that the latters
gravitational force was an unwelcome reintroduction of occult qualities into physics), they rarely
entered into direct correspondence; relations between them had been strained beyond breaking
point by an acrimonious priority dispute over the invention of the calculus.
10. I leave the notion of genuine property vague, but there are certain properties for which PII
is certainly not a valid principle. For example, if the property is believed by Lois Lane to be a
superhero is a genuine property, then PII entails that the names Clark Kent and Superman
cannot refer to the same person.
11. There is a well known (but disputed) counterexample to PII (due to Max Black). Couldnt there be
a universe whose sole contents are two iron spheres, of exactly the same size and shape, sitting at
a distance of six feet away from one another? It seems so. But since the two spheres have exactly
the same properties, including the relational property of being six feet away from an iron sphere
of such and such shape, size, etc., PII entails that there arent in fact two spheres at all, only one.
Note also that PII is not the only principle which is sometimes called Leibnizs Law, there is also
the principle of the indiscernibility of identicals: if a and b name the same object, then a and b have
exactly the same properties. This is far less controversial, although it has to be indexed to times to
accommodate the fact that persisting objects can change their properties.
12. As Maudlin notes, Leibniz, in a classic modus ponens/modus tollens reversal, argues that since
only those relations can be observed, the supposed shifts create no real change at all (1993: 188).
416 NOTES
11. Absolute motion
1. As well as arguing that the movement of material objects consists of motion with respect to abso-
lute space, Newton also argued that the parts of absolute space are necessarily immobile. For an
interesting discussion of these arguments, in De Gravitatione and the Scholium, see Nerlich (2005)
and Huggett (2008).
2. For a discerning analysis of just what Newton was trying to achieve in the Scholium see Rynasie-
wicz (1995).
3. The Newtonian account of inertial forces requires absolute time as well as absolute space. Accelerated
motion gives rise to effects that nonaccelerated motion does not, and this acceleration is absolute. It
isnt a matter of moving with respect to other bodies; acceleration is rate of change of velocity, i.e.
how much space is covered during a certain interval of time. Now, the relationist holds that there is
no absolute measure of periods of time, just more or less convenient measures as recorded by differ-
ent clocks. If I select as my clock one that reveals a body to be accelerating, but the body suffers no
force, we can conclude that the body isnt really accelerating, so the clock is wrong. Hence there is
an absolute time scale, and good clocks accurately reflect this time scale, bad clocks dont.
4. In recent years work has been done to improve on Mach. Barbour & Bertotti (1977, 1982) have
formulated precise and rigourous Machian- style theories of gravitation and inertia. Their propos-
als are serious and well regarded, and although far from complete (they have not as yet incorpo-
rated electromagnetism or quantum mechanics into their Machian schemes), do show that it is
possible to go far further in the direction that Mach advocated than many have supposed. Still, it
remains the case that the viability of a Machian approach remain as yet unproven. For evaluations,
see Earman (1989: 926), Belot (2000), Misner et al. (1970: 21.12) and Brown & Pooley (2002).
5. See Skow (2007) for further discussion of Sklars proposal.
12. Motion in spacetime
1. For an early and influential statement of this view, see Howard Steins Newtonian space- time
(1970).
2. The following succinct definition of neo- Newtonian spacetime is due to Huggett: Between any
two events p and q, (i) let there be a definite temporal interval, T(p, q); (ii) if they are simultaneous
(T(p, q) = 0), let there be a definite spatial distance D(p, q); (iii) given any curve c through p, let c
have a definite curvature at p, S(c, p) (i.e. a curve c is straight if S(c, p) = 0 for all points p on c)
(1999: 194).
3. Largely drawn from Robert Geroch (1978: ch.3).
4. If the relationist recognizes absolute directions there is no choice of orienting M in S in different
ways; this additional degree of freedom becomes available if only relative angular differences are
recognized.
5. This way of characterizing the difference between relationists and substantivalists is now standard:
see Friedman (1983: ch. 6) and Mundy (1983).
6. While this notion may be unproblematic in the context of the worldtubes of non- punctiform
objects, defining a notion of curvature for the strictly one- dimensional worldlines of point- particles
is another matter. Perhaps the curvature of a given worldline can be defined as the limit of series
of nested worldtubes, all with the same (or minimally different) curvature along the time- axis, but
diminishing radii (in a manner reminiscent of Whiteheads method of extensive abstraction). In
any event, there is no reason why the relationist need shy away from introducing new geometrical
primitives, even if working out the detailed mathematics is a non- trivial task.
7. See Huggett (2006) for details of a viable-looking form of relationism, one that relies on the best
system account of natural laws.
8. Maudlins Quantum Non- Locality and Relativity (1994) is entirely devoted to exposing the dif-
ficulties of reconciling special relativity and quantum theory.
13. Curved space
1. A perfected Euclidean system was finally achieved at the end of the nineteenth century by Hilbert,
who used a total of twenty axioms.
NOTES 417
2. This is the Playfair version. Euclids version is If a straight line falling across two straight lines
makes the sum of the interior angles on the same side less than two right angles, then the two
straight lines intersect, if sufficiently extended on that side.
3. In a letter to Bessel, quoted in Jammer (1993: 148).
4. See Norton (1992: Pt. II) for a good discussion of this point in a simple context. There is clear
and succinct account of the basics of Riemannian geometry in Reichenbach (1958: 39). See
Egan (1998, II) for a somewhat more technical (but not too advanced) explanation of what
is involved mathematically; Leiber (2008) also provides an excellent and accessible introduc-
tion. As Sklar notes (1977: 45), in characterizing the intrinsic structure of a Riemannian
space it is not necessary to start off with a particular coordinatization and specification of the
g- function; there are coordinate free methods, and much of modern differential geometry is
devoted to the problem of characterizing the intrinsic structure of general curved spaces without
reference to an imposed coordinatization of the space. See Egan (1998) and Baez (1996) for
comparatively accessible routes to the curvature tensor via parallel transport of vectors and con-
nection coefficients. For an idea as to what parallel vector transport involves, imagine walking
around the four sides of a football field holding a walking stick out in front of you, making sure
you keep the stick pointing in the same direction at all times (this may take some doing when you
reach the corners, but you are allowed to change hands). Assuming that the field is flat, when you
complete the trip and arrive back where you started, the stick (which has undergone a parallel
transport) will still be pointing in the same direction. If you conduct a similar round- trip on a
curved surface, the stick will not arrive back pointing in the same direction; the difference in direc-
tion constitutes a measure of the curvature of the relevant surface the greater the curvature, the
greater the direction change after parallel- transport. It is easy to verify this effect of curvature by
spending a few minutes experimenting with an apple and pencil.
5. See Callender & Weingard (1996) for an informal exposition of some of the fundamentals.
6. In the two- dimensional case, five types of space are locally Euclidean: the plane, the cylinder, the
Mbius strip, the torus and the Klein bottle. Of these, only the plane is simply- connected. There
are two spaces of constant positive curvature elliptical (or spherical) space and the projective
plane and the latter is multi- connected, the former simply- connected. As for two- dimensional
spaces of constant negative curvature, the Lobachevskian space that we encountered earlier is
simply- connected, but the number of multi- connected spaces of this type is infinite. The same pat-
tern extends to the three- dimensional case, but direct comparisons become trickier as the numbers
on both sides are infinite.
7. For more on life in topologically unusual spaces see Reichenbach (1958: ch. 1, 12), Nerlich
(1994b: ch. 5).
8. In fact, in travelling between two locations light follows the path that takes the least time which
even in vacuo isnt quite the same as travelling in a Euclidean straight line, as we shall see.
14. Tangible space
1. This is oversimplifying considerably. Although an object entering the hole will acquire stress energy,
and (given the principle of the conservation of energy) this energy must come from somewhere;
in some cases it may be supplied from the outside (a push), but it may also be supplied (partly or
wholly) from within: the moving object will lose some of its kinetic energy, and so change either its
velocity or momentum. But as Nerlich notes: The interaction of the tensor field of stress with the
vector field of momentum is so complex that we can say very little in general about how they will
sum over a volume. Furthermore, in certain cases, kinetic energy may be gained at the expense of
stress energy, and the elastic volume will accelerate without the intervention of an outside force
(1994a: 5, 1789). Even so, having ones middle regions undergo unforced accelerations would
be likely to induce an unusual sensation.
2. See Van Cleve (1998) for a different response to the lesson from Flatland.
3. As most will recall from early school days, you can make one by taking a longish strip of paper,
giving it a half- twist, and then sticking the two ends together.
4. For further useful discussions on this topic, see Hoefer (2000), Huggett (2000), Pooley (2003) and
Lee (2006).
418 NOTES
15. Spatial anti- realism
1. The odd goings- on across the boundaries of R
1
and R
2
in Fosters scenario are due to place- specific
laws of nature, which, in effect, teleport objects between the regions concerned. There may be
another way in which the same effect could be produced. In 13.5 we encountered a space with a
toroidal topology. When a plane exits from the space on the left, it instantly re- emerges on the
right; when it exits on the right, it re- emerges from the left, and so on. These effects were not
due to strange laws of nature, but rather to the fact that the points on opposite sides of the space
are numerically identical. If, in similar fashion, the points on the boundaries of R
1
and R
2
in the
OxfordCambridge case were numerically identical, on entering R
1
from the left you would find
yourself inside R
2
, continuing on in the same direction. Would this re location be detectable?
2. To be specific, the bulk of Chapters 810.
3. Foster points out (1993: 305) that for his purposes he could take any aspect of the geometrical
structure of a space to be essential to it (e.g. metrical or affine structure), since the functional
and intrinsic geometries of S could differ in any of these other ways too. But since topological
essentialism is the weakest version of geometrical essentialism it is also the least contentious.
4. Here is how Foster himself formulates the modal argument: We know that the physically relevant
geometry (topology) of S logically depends upon its nomological organization. And we also know
that S possesses this nomological organization (including those aspects relevant to the physical
geometry) only contingently. From these premisses, it immediately follows that S possesses its
physically relevant geometry (topology) only contingently. But we also know that, as a genuine
space, P possesses its physical geometrical (topological) structure essentially. So, standing as they
do in different modal relations to the physical (or physically relevant) geometry, P and S must be
numerically distinct. Moreover, since the physical geometry (topology) which is essential to P is
essential to the identities of its points, we can conclude, more strongly, that the points of P are
numerically distinct from the points of S. And, points being the smallest components of a space,
this means that there are no divisions of S and P relative to which the components of one can be
identified with those of the other (1993: 306). To simplify, I have replaced Fosters F (in the
original) with S.
5. Rejecting Fosters essentialist doctrine concerning points would permit a response of this sort: If
geometry (or topology) is not essential to the identity of points, spaces with different geometries
can contain the same points; consequently, the fact that S and P can differ geometrically is no
obstacle to holding that they are the same space, i.e. the same collection of points. Among those
who question the doctrine of the necessity of identities, in its traditional guise, are Lewis (1971,
1986), Noonan (1991, 1993) and Gallois (1998).
16. Zeno and the continuum I
1. Aristotle presents Zenos argument in an economical way, thus: The second is the so-called Achil-
les. It consists in this, that the slower will never be overtaken in its course by the quickest. For the
pursuer must always come first to the point from which the pursued has just departed, so that the
slower must necessarily be always still more or less in advance (Physics VI.9 239b).
2. Euclid himself did not make this denseness property explicit; Hilbert did so in his improved
axiomatization of Euclidean geometry. Contemporary mathematicians talk of points as possessing
size zero.
3. It may not be intuitively obvious that the rationals are dense, but it is easy to demonstrate. Take two
rationals expressed as fractions that are successive in the manner of
33
101 and
34
101; now double all
four numbers, giving
66
202 and
68
202 this immediately reveals the fraction lying half-way between
our original two, namely
67
202. Since this procedure can obviously be applied to any successive
fractions, the rationals must be dense.
4. Hilary Putnam refers us to the basic postulate upon which the subject of analytical geometry is
founded (and with it the whole study of space in modern mathematics, including the topological
theory of manifolds). This is the postulate that there is a one-to-one order preserving correspond-
ence between the points on the line and real numbers (1979a: 64). The doctrine that there is an
intimate connection between numbers and geometry can be traced back to Descartes analytic
geometry, within which numbers are conceptualized as points on a line. For a useful critical
overview see Lakoff & Nez (2000: pt. IV).
NOTES 419
5. To speak more strictly, the lengths in question here are always positive, non-zero and non-
infinitesimal.
6. The set of all subsets of a set S is called the power set of S. Cantor proved that the power set of
any infinite set S always has a greater cardinality than S itself. For more on Cantor and infinity see
Rucker (1982), Moore (2001), Salmon (2001: appendix) and Wallace (2005).
7. Note that not all collections of points in c are measurable; such collections are not assigned
a measure. These badly behaved collections of points will play a role in the TarskiBanach
Paradox, which we shall encounter in 17.10. See Salmon (1975: 5262) for a more detailed
exposition of Grnbaums use of measure theory; see Sherry (1988) for a more critical discussion;
Chris Mortensen and Graham Nerlich (1978) are strongly critical of Grnbaums treatment of
intervals as sets, and offer alternatives.
8. There is a related puzzle: since, thanks to the denseness of the continuum, there is no first point
that follows Z (the point of departure) one might wonder how Zeno ever manages to get going
in the first place. This is sometimes called the reverse, regressive or inverted dichotomy,
which Max Black describes thus: To reach any point, he must first reach a nearer one. So in order
to be moving at all, Achilles must already have performed an infinite series of acts must, as it
were have travelled along the series of points from the infinitely distant and open end. This is an
even more astounding feat than the one he accomplishes in winning the race against the tortoise
(1954: 69). For reasons of space we will not be discussing this variant here, but the considera-
tions outlined in 16.3 concerning the addition of infinite series of finite quantities also apply to
the inverted dichotomy, as do the considerations that follow, mutatis mutandis. Since, as Black
observes, the regressive dichotomy can seem more baffling than the standard (or progressive) form,
there are pedagogical advantages in addressing the easier variant first. For further discussion see
Grnbaum (1968: 3A), Lowe (2002: ch. 16).
9. This is as good a place as any to mention a fascinating possibility opened up by relativistic space-
times: so-called bifurcated supertasks. In an accessible exposition of the essentials, John Earman
and John Norton (1996) outline a scenario that is permitted by the MalamentHogarth class of
general relativistic spacetimes. The scenario involves a Slave who runs through the entirety of
an infinitely long series of calculations perhaps putting to the test Goldbachs conjecture that
every even number greater than 2 is the sum of two primes while being continuously observed
by a Master. From the Slaves own temporal perspective (or proper time) this process takes an
infinite amount of time: the calculations are not performed infinitely quickly, but at a normal rate,
subjectively speaking. However, from the Masters perspective the whole operation takes only a
finite time, so it is possible for the latter to discover whether Goldbachs conjecture is true or false
by observing all the Slaves calculations. It is as yet unclear whether implementing this sort of
scenario is physically possible, but as Earman and Norton go on to note, there may nonetheless be
significant ramifications for the philosophy of mathematics. For further discussion and references
see Laraudogoitia (2004) and Oppy (2006: 4.6).
10. See Grnbaum (1968: 745, 1267). It should also be noted that in terms of Cantors theory of
ordinals, the series of sub-journeys leading to Z* is of order type e, that order type of the whole
journey (including Z*) is e + 1. By way of a more concrete illustration, the entire series of positive
integers 1, 2, 3, 4, has order type e; the ordered set that includes the entirety of this series
together with 0 that is, 1, 2, 3, 4, , 0 has the order type e + 1. For further details see Moore
(2001), Salmon (2001: appendix) and Wallace (2005).
17. Zeno and the continuum II
1. Aristotle again states Zenos argument with an enviable succinctness: he says that if everything
when it occupies an equal space is at rest, and if that which is in locomotion is always in a now,
the flying arrow is therefore motionless (Physics VI.9 239b30).
2. Difficulties of this sort are pressed by Tooley (1988), Bigelow & Pargetter (1989, 1990) and Car-
roll (2002); see Butterfield (2006a) for a good overview of recent discussions, and a rebuttal of
some of Tooleys arguments.
3. Aristotle relates Zeno as concluding that half a given time is equal to double that time (quoted
in Kirk & Raven 1969: 295); for attempts to explain precisely what Zeno had in mind see Owen
(1957) and Sorabji (1983: 33033). Sorabji also mentions an intriguing conjecture of David
420 NOTES
Sedleys: rather than being confused about the relativity of motion, Zeno may in fact have discov-
ered that motion is relative, and he took this to mean that motion is less than fully real (ibid.: 332).
4. Discrete spaces are sometimes construed as networks of points, with each point having a finite
number of closest neighbours (in different directions, depending on the dimensions of the space)
for example, Carnap (1994: 90) but this is not the most natural way to go: see Nerlich (1994b:
ch. 9).
5. It has been known since medieval times (at least) that discrete spaces do not precisely conform to
Pythagorass theorem, and so (in this sense) they are non-Euclidean. If space is discrete, all lengths
will consist of a whole number multiple of the elementary distance. But as we saw in 16.2, the length
of the hypotenuse of a right-angled triangle whose shorter sides are both 1 unit is the square root of
2, which is an irrational number 1.414213 . However, this discrepancy diminishes with greater
lengths: if the shorter sides of our triangle are 5 units long, the hypotenuse will be the square root
of 50, or 7.07106, which is not much more than the closest discrete approximation, 7.
6. There are more radical proposals: in his monumental A New Kind of Science, Stephen Wolfram
has suggested that cellular automata along the lines depicted in Figure 17.3 provide the keys
to unlocking the real nature of our universe. Needless to say, Wolfram believes that space and
time (in our universe, at least) are discrete, rather than continuous. Gray (2003) provides a useful
overview and (often sympathetic) criticism. For a collection of reviews and assessments of Wolfram
see http://shell.cas.usf.edu/~eclark/ANKOS_reviews.html (accessed June 2010).
7. And many others: Bergson complained that the orthodox continuum consisted of nothing more
than a possire des instant (dust of instants) (1946: 21920). For a survey of these and other
dissenting voices in the nineteenth and early twentieth centuries for example Du Bois-Reymond,
Veronese and Brouwer see Bell (2005: ch. 5).
8. Galen Strawson has argued along these lines, with some vigour: If one is being metaphysically
straight, the intuition that nothing (concrete, spatio-temporal) can exist at a mathematical point,
because there just isnt any room, is rock solid. Do not be cowed by physicists or philosophers of
physics. If we had been metaphysically sensible we might have got to something like string theory
(M-theory or brane theory) decades sooner. It seems intuitively obvious, by the grace of mathemat-
ics, that to introduce real, concrete entities that are infinitely small and therefore metaphysically
impossible into ones theory will lead to infinite largenesses popping up in protest elsewhere in
ones equations. And so it came to pass (2006: 16).
9. Anyone who finds the TarskiBanach result unpalatable, but who also finds the orthodox (Can-
torian) conception of the continuum appealing, has another option: since the proof relies on the
axiom of choice of standard set theory, rejecting the latter is also an option. However, since
there are serious consequences in mathematics for so doing, it is a step most mathematicians would
prefer not to take. The axiom of choice states (informally) that for any collection C of non-empty
sets, we can select one member from each set in C even if we lack a rule for making the selections.
By way of an informal illustration of an unmeasurable set, take the rational numbers to be a set,
and then consider all the sets that can be formed by shifting or translating the rationals by a fixed
amount along the real line. Using the axiom of choice, select one member from each of these sets,
and combine them to make a set C. This set will be unmeasurable.
10. I rely here on Putnams reconstruction of Peirces views on the continuum, as outlined in the lat-
ters 1898 Cambridge lectures. Putnam suggests that, for Peirce: The reason that the line is a col-
lection of points that lack distinct individuality is that it is a collection of possibilia, and possibilia
are not fully determinate objects for Peirce. What answers to our conception of a continuum is
a possibility of repeated division which can never be exhausted in any possible world, not even in a
possible world in which one can complete abnumerably [i.e. non-denumerably] infinite processes.
That is what I take Peirces daring metaphysical hypothesis to be (1994: 17). For more on Peirces
continuum see Zalamea (2003), Moore (2007b).
11. For further discussion see Lakoff & Nez (2000: ch. 11), Alper & Bridger (1997) and White
(1999). Infinitesimals have been introduced into recent mathematics by a quite different route: in
the synthetic differential geometry or smooth infinitesimal analysis deriving from work done
by Lawvere, Kock, Reyes and others. As Bell describes it: employing the methods of category
theory, smooth infinitesimal analysis provides an image of the world in which the continuous is
an autonomous notion, not explicable in terms of the discrete. It provides a rigorous framework
for mathematic analysis in which every function between spaces is smooth (i.e. differentiable
NOTES 421
arbitrarily many times, and so in particular continuous) and in which the use of limits in defining
the basic notions of the calculus is replaced by nilpotent infinitesimals, that is, quantities so small
(but not actually zero) that some power most usefully the square vanishes (2009: 9). In
Robinsons brand of non-standard analysis all theorems of orthodox analysis are valid; this does
not apply in the case of smooth infinitesimal analysis, which abandons the law of excluded middle
and employs an intuitionist logic. For introductions see Bell (2005, 2009), Hellman (2006); for
applications to Zenos paradoxes see Harrison (1996).
12. This is from his essay The Conception of Time Essential in Logic, from his 1873 Logic. Moore
plausibly suggests that Peirce was influenced by Kants doctrine in the Anticipation of Perception
in the first Critique: The property of magnitudes on account of which no part of them is the
smallest (no part is simple) is called their continuity. Space and time are quanta continua, because
no part of them can be given except as enclosed between boundaries (points and instants), thus
only in such a way that this part is again a space or a time. Space therefore consists only of spaces,
time of times (A170/B212).
13. Perhaps there can be a greater degree of divisibility still. One might think that it would be possible
for there to be a stuff that no matter how many of its parts you accounted for, you would not have
accounted for all of its parts for it would be able to be decomposed into strictly more parts than
the ones you in fact decomposed it into. I call such gunk Hypergunk, where Hypergunk is gunk
such that for any set of its parts, there is a set of strictly greater cardinality of its parts. There would
thus be too many pieces of hypergunk to form a set for any sized set, there would be enough
pieces of hypergunk to make up a larger set Nolan (2004: 5). As Nolan goes on to note, there
are affinities between this view and that of Peirce.
14. Time, on this model, is not composed of durationless instants, densely ordered, but of overlap-
ping intervals having some temporal extension (Dummett 2000: 506). On Dummetts model
all physical quantities are subject to a finite maximum rate of change, which rules out the radical
discontinuities permitted by the standard model of the continuum.
18. Special relativity
1. Paul Davies provides a nice example of a case in which these effects are observable in everyday life:
Most metals have a silvery appearance, but not gold. Its distinctive and attractive glitter can be
traced to the effects of relativity on the motions of the electrons inside the metal that are respon-
sible for reflecting light. So it is no exaggeration to say that this precious metal is precious and
financially valuable partly as a result of time dilation operating within the gold atoms (1995:
69).
2. It should be pointed out that Einsteins theory has implications of a very practical kind. Einstein
published two papers on STR in 1905: the first lays down the basic principles of the theory, the
second (only three pages long) establishes the equivalence of mass and energy via the famous equa-
tion E = mc
2
. Matter and energy are interconvertible, and the amount of energy contained in a
body at rest is huge by ordinary standards (the energy stored in a 1 kg mass would keep around
30 million hundred watt light bulbs going for a year). While the full implications of this did not
emerge at once, it was only another forty years before the first atomic bombs exploded.
3. STR gives rise to several intriguing puzzle cases (often referred to as paradoxes). For more on
these see entries for The Twin Paradox, The Barn and the Pole, The Superluminal Sissors and
(more advanced) Bells Spaceship Paradox in the Usenet Relativity FAQ at http://crib.corepower.
com:8080/~relfaq/relativity.html (accessed May 2010). In a different vein, Galison (2003) devel-
ops and intriguing case in support of the contention regarding the background against which
STR was developed. Galison suggests that the theorys roots are profoundly interwoven with the
practical concerns of both Einstein and Poincar at the turn of the twentieth century; more specifi-
cally, with the accurate timetabling of trains (Einstein), and with mapmaking (Poincar). In both
instances, the synchronization of clocks over long distances was a pressing problem.
19. Relativity and reality
1. To keep things as simple as possible I will restrict my attention to Growing Block theories, but the
same considerations will apply to dynamic block models that do not posit a preferred temporal
direction.
422 NOTES
2. Howard Stein states his theorem thus: If R is a reflexive, transitive relation on a Minkowsi
space (of any number of dimensions of course at least two), invariant under automorphisms
that preserve the time- orientation, and if Rab holds for some pair of points (a, b) such that ab is
a past- pointing (time- like or null) non- zero vector, then for any pair of points (x, y), Rxy holds if
and only if xy is a past- pointing vector (1991: 149).
3. For example, Dorato (1995: ch. 11), Clifton & Hogarth (1995) and Tooley (1997: 337).
4. Hinchliff (1998), but see also Godfrey- Smith (1979), who seems to have been the first to propose
this version of Presentism. Compound Presentists will extend reality to sheaves of neighbouring
light- cones. Another option for Presentists would be to hold that the present of a given point
consists of the space- like separated regions of spacetime (the points absolute elsewhere).
5. The Cone Presentist is confronted with another awkward question: why confine the present only
to the surfaces of past light- cones? Why not include future light- cones too? After all, events on
the latter are also light- like related. Hinchliff rejects this Double-Cone Presentism, appealing to
an asymmetrical causal relation: The difference between the cones is due to the asymmetry built
into the nature of a light ray or signal. And that asymmetry arises from the asymmetric nature of
causation itself, which is a non- arbitrary foundation on which to rest the distinction between cone
and Double- Cone Presentism (1998: 582). Presentists who reject this view of causation may well
be obliged to embrace Double- Cone Presentism, and all that it entails.
6. For a different line of response see Smith (1993, 1998).
7. [We] cannot maintain a Lorentz invariant interpretation of the quantum nonlocal connection of
distant systems there has to be a unique frame in which these nonlocal connections are instan-
taneous Bohm & Hiley (1993: 271). See also Bohm (1952) and Albert (1992: ch. 7). The theory
(also known as the De BroglieBohm Theory since it is a development of a line of thinking
that dates back to De Broglie in the 1930s) posits a form of non- locality over and above the EPR
variety. The global wave function, which according to the theory simultaneously fixes the motions
of all particles, is itself influenced by the condition of individual particles everywhere, so it turns
out that in determining the trajectory of a particle, one may have to take into account the wave
functions of particles in other galaxies! (Chalmers 1996: 344).
8. Needless to say, there are further controversies concerning STR. On the standard interpretation,
the geometry of Minkowski spacetime plays a crucial explanatory role: material objects undergo
relativistic effects, such as length contraction and time dilation, because they inhabit a Minkowksi-
type spacetime. On another view, recently resurrected (see Brown 2005; Brown & Pooley 2006),
these effects, ultimately, are a consequence of the quantum laws governing material things, not the
geometry of spacetime. This geometric structure, purportedly left behind even if per impossible
all the matter fields were removed from the world, is, I shall argue, the space-time analogue of
the Cartesian ghost in the machine, to borrow Gilbert Ryles famous pejorative phrase (Brown
2005: viii). For further discussion of this issue, see Norton (2008a) and Skow (2006).
20. General relativity
1. In a letter to Sommerfeld, quoted in Greene (1999: 62).
2. This example is drawn from Maudlin (1994: 2257).
3. What Newton could have done (had he had the requisite mathematics) was formulate an explana-
tion of gravity in term of mass- induced curvature in Newtonian spacetime and the precise form
such a theory would take has been worked out. See Friedman (1983: 95104) for details. Friedman
comments, In an important sense, then, general relativity is just Newtonian gravitation theory
plus special relativity (1983: 184). But since the empirical predictions of this theory differ from
those of Einsteins theory, and the latters predictions have (so far) proved to be correct, there are
no takers for the modified Newtonian theory.
4. The conventionalism stance is far from dead, especially among scientists who regard theories as
primarily tools for prediction: cf. Thorne (1994: ch. 11).
5. See Leiber (2008) for an unusually accessible elmentary introduction to the relevant mathematics.
6. The twenty components of the Riemann curvature tensor which generate the twenty numbers
needed to completely describe curvature in four dimensions can be divided between the Ricci
tensor and the Weyl tensor. The Ricci tensor describes changes in volume, whereas the Weyl
tensor describes changes in shape (objects can change their shape without changing their volume).
NOTES 423
The Einstein tensor is defined via the Ricci tensor, and since it only has ten components (likewise
the stress- energy tensor) it does not completely determine the geometry of spacetime. The remain-
ing aspects of curvature are characterized by the Weyl tensor. In regions where there is no mass-
energy the stress- energy tensor is zero, and likewise the Einstein tensor and the Ricci curvature.
But the Weyl curvature in such regions need not be zero: there is thus an aspect of curvature that
is independent of mass- energy. Gravitational waves and tidal forces involve variations in the Weyl
curvature. See Baez (2001) and Penrose (1989: 21011). Note also that since GTR is a spacetime
theory, it is formulated in terms of a pseudo- or semi- Riemmanian manifold, not the spatially
four- dimensional Riemmanian manifold we encountered in 13.4.
7. Torretti writes: The [stress- energy] tensor is something of a joke, for we are quite incapable of
accurately representing the distribution of matter and energy in the universe and, if we could do it,
we would be unable to solve the [Einstein Field Equations] EFE for a matter- energy tensor field of
the required complexity. In actual practice, the EFE are solved exactly only for ideally simple fields.
Thus the Schwarzschild solution, used in celestial mechanics, concerns the spherically symmetric
field of a central body, and the Friedmann solutions, used in cosmology, assume a distribution of
matter which is perfectly isotropic about every point in space (1999a: 77).
8. Newtonian gravitational theory leads to the same result. In 1785, John Michell, working on the
assumption that light is corpuscular, showed that mass, sufficiently concentrated, would have a
gravitational pull so that strong light could not escape (assuming that light is affected by gravity in
the way of other particles). Since nothing ruled out the possibility of matter existing in so condensed
a state, he speculated that space might contain a huge number of dark stars, all invisible to us.
9. For more on black holes and their strange properties see Penrose (1989: 33045) and Thorne
(1994). For a discussion (often advanced) of the problems posed by singularities, see Earman
(1995).
10. These properties are distinct: an infinite field of exactly similar (and similarly spaced) blades of
grass, all blown in precisely the same direction is homogeneous (the same everywhere) without
being isotropic (there is a preferred direction).
11. Dynamic in this context should not be confused with dynamic time (of the sort we looked at
in Chapter 6). A universe is dynamic in the current sense if it is of different spatial sizes at different
times; universes in which time is dynamic are of different temporal sizes as of different times. It
is perfectly possible for a Block-type universe to be spatially dynamic, just as it is possible for a
dynamic- time universe to be spatially static (or dynamic).
12. A significant minority of cosmologists remained hostile to the big bang: the steady state theory,
proposed by Bondi, Gold, Hoyle and Narlikar (among others) in the late 1940s, remained a
serious competitor for at least two decades, before falling out of favour due to its incompatibility
with observational data. Underlying the theory is the so- called perfect cosmological principle,
according to which the universe is homogeneous and isotropic in time as well as space. Since
the steady state theorists accepted that the universe is expanding, they had to come up with an
explanation for how the average density of matter could remain constant over time (other things
being equal, it would decrease with the expansion of space). Their solution was the C- field, which
generates a steady flow of new matter (by a process of continuous creation) to compensate for
the drop in density brought about by expansion. Continuous creation has never been observed in
the laboratory, but given the tiny quantities required by the theory only one hydrogen atom per
cubic metre over the entire age of the universe this was not regarded as problematic. The C- field,
along with the notion of continuous creation, could be regarded as a spatial (albeit small- scale)
analogue of the continual diachronic ex nihilo creation that Growing Block models of time require.
See Coles (2001: 33031) for more on the steady state theory.
13. Current estimates place the critical value at around 10
26
kg m
3
, which amounts to an average of
five or six hydrogen atoms per cubic metre over the entire universe which is around one hundred
trillion trillion times less dense that the air at the Earths surface. This estimate of the critical value
depends on current estimates of the Hubble constant, which place it in the vicinity of 70 km Mpc
1
,
so a galaxy at a distance of 1 megaparsec (= 3.26 light years) will be receding from us at a speed
of 70 km s
1
, whereas a galaxy 100 Mpc away will be receding from us at a rate of 7000 km s
1
.
14. The cosmological constant is by no means the only explanation of dark energy: another candidate
is quintessence, a dynamic scalar field that need not be uniform across space and time, unlike
the cosmological constant. For more see Steinhardt & Ostriker (2001).
424 NOTES
15. There is no obstacle to imagining that all of the large and apparently unrelated contributions
[from quantum theory to the cosmological constant] add together, with different signs, to produce
a net cosmological constant consistent with the limit, other than the fact that it seems ridiculous.
We know of no special symmetry which could enforce a vanishing vacuum energy while remaining
consistent with the known laws of physics; this conundrum is the cosmological constant problem
which at this point remains one of the most significant unsolved problems in fundamental
physics (Carroll 2001).
16. For a large collection of MOND-related resources see www.astro.umd.edu/~ssm/mond/ (accessed
June 2010).
17. For a critical assessment of the void approach, see Zibin et al. (2008). See Nicolson (2007) for
a detailed yet accessible (and well-illustrated) treatment of recent developments relating to dark
matter and dark energy.
18. Cf. 13.5.
19. For a full review see Lachize- Rey & Luminet (1995); also Luminet (1998, 2001), and Luminet
et al. (1999). The fundamental domains for (three- dimensional) multi- connected spaces are poly-
hedra; the idea that these geometrical objects could have cosmological significance has definite
Platonic overtones.
21. Spacetime metaphysics
1. These holes, it will be recalled, are not cavities in (or absences of) substantival space, but rather
regions of extreme local curvature in substantival space.
2. It should be noted that gravity waves have yet to be detected under laboratory conditions, but there
is astronomical evidence for their existence, for example the observed energy loss of binary pulsar
systems match the energy that GTR predicts will be transported away by gravitational ripples. See
Will (1988: 2016) and Thorne (1994: ch. 10); for a more idiosyncratic (but interesting) take on
the topic, see Martin (2009).
3. Wheeler combines the two: The geometry of spacetime, and therefore the inertial properties
of every infinitesimal test particle are determined by the distribution of energy and energy- flow
throughout all space (quoted in Ray 1991: 134).
4. For more detail see Sklar (1977: 21044), Raine (1981), Earman (1989: 1059) and Ray (1991:
ch. 7).
5. The diffeomorphism is of the active rather than passive variety; i.e. it involves a switching
about of geometrical objects points rather than merely a relabelling of the same objects via a
different system of coordinates or names.
6. See, for example, Butterfield (1989a,b), Brighouse (1994) and Rynasiewicz (1994). Some might
be tempted simply to reject the hole argument on the grounds that it is foisting upon GTR super-
fluous metaphysical commitments concerning the numerical identities of spacetime points. To
take a simpler case, suppose we have a theory that predicts the forces that exist within systems of
interconnected load- bearing beams at any given time. We would have grounds for complaint if
our theory delivered alternative models of a given beam structure, some of which predict collapse,
some of which dont. But we would scarcely complain if the differences between the alternative
models involved nothing more than the numerical identity of the beams themselves (e.g. there
are transformations that switch the locations of intrinsically similar beams within a structure at
a time). All our theory purports to offer is an account of synchronic forces within structures of
certain types of beam, so its failure to be sensitive to the identities of token beams is not really
a failure at all. Might not precisely the same be said of the alleged failures of determinism in
GTR? Proponents of the hole argument would say not: unlike our imagined theory of beams, the
mathematical formalism used in GTR is sensitive to token- identities, and so is not neutral on the
question of the numerical identities of spacetime points; consequently, the alternative models are
genuinely distinct by the lights of the theory itself.
7. For further discussion of the hole argument see Pooley (2006) and Norton (2008b); for further
interpretation of Newtons remarks on the immobility of space see Nerlich (2005) and Huggett
(2008).
8. I say a little more about these issues in Chapter 22.
9. See also Figure 20.6. The uniform microwave background radiation left over from the big
NOTES 425
bang (measured with great accuracy by COBE and WMAP provides a valuable guide here:
Although the view from Earth is of a slightly skewed cosmic heat bath, there must exist a
motion, a frame of reference, which would make the bath appear exactly the same in every
direction. It would in fact seem perfectly uniform from an imaginary spacecraft travelling at
350 km per second in a direction away from Leo (towards Pisces, as it happens) We can
use this special clock to define a cosmic time Fortunately, the Earth is moving at only
350 km per second relative to this hypothetical special clock. This is about 0.1 per cent of the
speed of light, and the time- dilation factor is only about one part in a million. Thus to an excellent
approximation, Earths historical time coincides with cosmic time, so we can recount the history
of the universe contemporaneously with the history of the Earth, in spite of the relativity of time
(Davies 1995: 1289). See also Dorato (1995: ch. 13).
10. This analogy should not be extended too far. If our spacetime is, say, negatively curved and open,
what grows is already spatially infinite.
11. Thorne (1994: ch. 14); see Al-Khalili (1999: 22539) for a more recent survey.
12. Savitt explains it thus: [A]ny way we extend the Gdelians local time function to the whole
spacetime will have an odd consequence. Pick any spacetime point x somewhat distant from the
Gdelians but simultaneous (according to this extended time function) with some point of their
history. A possible observer, Kurt, can be born at x, live an entire lifetime experiencing time just
as they (and we) do, yet no part of his life is assigned a time different from that assigned to x.
Thus Kurts whole existence objectively would be simultaneous if one tried to identify objectively
lapsing time with the Gdelians time (1994: 4678).
13. For further discussion of Gdels argument see Yourgrau (1991), Earman (1995: ch. 6, appendix)
and Bourne (2006: ch. 8).
22. Strings
1. Strong interactions bind protons and neutrons (and their component quarks) together in atomic
nuclei, which is roughly the range of the strong force (approximately 10
13
cm). The weak force
only operates over distances a hundred times smaller than this, but nonetheless plays an important
role in particle physics: neutrinos feel the weak force but not the strong, and are produced in bulk
by the nuclear reactions that power the sun.
2. Needless to say, there is rather more to it. See Greene (1999: 15265) for a fuller account of the
way string theory overcomes the problems associated with point- interactions.
3. Empirical testability aside, it looks as though there may be a very large number (perhaps as many
as 10
500
) of string landscapes, that is, models permitted by the theory that involve different
elementary particles, physical constants and physical laws. Although some see this as a serious
problem, for others it is merely a profound lesson about the nature of our universe: it may be
larger and more varied than we suppose. See Susskind (2006) for more on this.
4. It is possible to produce two- dimensional representations of what three- dimensional sections
of such a manifold would look like. There is a beautiful example in Greene (1999: 207), which
resembles a cross between a tangle of worms and an impossibly intricate piece of origami.
5. See Greene (1999: ch. 13) for more on string theory and black holes. In the mid- 1990s the five
main competing versions of string theory were brought into a single unified framework known as
M- theory. Once again, the unification was brought about by the discovery of another spatial
dimension: in M- theory there are ten spatial dimensions, and one of time. Thanks to the extra
dimension, strings can possess length and depth. M- theory posits not just one- dimensional strings,
but two- dimensional membranes (two- branes), along with pulsating blobs (three- branes) and other
exotic Planck- scale entities. See Greene (1999: ch. 12; 2004: ch. 13). More on this shortly.
6. In his Reflections on the Fate of Spacetime (1996), Witten suggests spacetime seems destined
to turn out to be only an approximate, derived notion, much as classical concepts such as the
position and velocity of a particle are understood as approximate concepts in the light of quantum
mechanics (Callender & Huggett 2001: 134).
7. Or at least it does when one variant of the theory is construed both literally and realistically, rather
than being viewed as merely a useful mathematical model.
8. This theme runs throughout Smolins book (e.g. 2000a: 11920). The theoretical underpinning
is diffeomorphism invariance, a feature of GTR that is often taken to have Leibnizian implications
426 NOTES
(see 21.3 & 21.4), and which is preserved in loop quantum theory; see Gaul & Rovelli (1999).
Whereas in the context of classical GTR it is questionable whether the classical point- manifold
really is dispensable, there is no reason for such misgivings in the loop quantum case, since a
substitute for the manifold (the spin lattice) is supplied. For more on loop quantum gravity see
http://cgpg.gravity.psu.edu/people/Ashtekar/articles/rovelli03.pdf and http://relativity.livingre-
views.org/Articles/lrr-2008-5/ (accessed June 2010).
Glossary
Many of the technical terms used in the main text can be found below, sometimes explained in less
detail, occasionally more. I have taken this opportunity to introduce some standard philosophical
terminology that is used but not explained in the main text.
absolute: quantities or relationships that are independent of frames of reference and coordinate sys-
tems.
absolute becoming: the coming- into- existence of objects, events and times posited by dynamic theories
of time, such as the Growing Block model and Presentism in some of its guises. Absolute annihila-
tion is the ceasing- to- exist of objects, times and events.
absolute motion: motion with respect to substantival space. See substantivalism, relationism.
absolute simultaneity: two events are simultaneous if the temporal interval between them is zero; if the
temporal interval between them is zero in all frames of reference, they are absolutely simultaneous.
Since absolute simultaneity is an equivalence relation, it partitions points in space and time into
mutually exclusive and exhaustive planes of simultaneity. The notion is abandoned in Einsteins
special theory of relativity, a fact that has led many to suppose that the notion of a single universe-
wide now cannot be physically real.
absolute space: the sort of space Newton believed in: substantival and with a structure that is independ-
ent of, and unaffected by, any material bodies that occupy it. Subsequent developments (such as
Einsteins general theory of relativity) have made it clear that the structure of a substantival space
need not be independent of its occupants.
acceleration: rate of change of velocity. Absolute acceleration is rate of change of velocity with
respect to a substantival space.
action- at- a- distance: a force or influence that acts between two objects directly, without passing through
the intervening space. Prime example: Newtonian gravity. In recent time, theories that posit action-
at- a- distance have been regarded as suspect, or even spooky (see field), but developments in
quantum theory have brought the notion to the fore once again (see Bells inequality, non- local).
affine geometry: the study of properties that remain constant under affine transformations, i.e. those
that preserve parallelism and collinearity (points are collinear if they lie on the same straight line),
but not angles or distances. The slogan In affine geometry, all triangles are the same is a useful
reminder of what is involved.
analyticsynthetic: another way of categorizing truths and falsehoods, also the subject of dispute.
Analytic truths are true in virtue of the meanings of the words used to express them (Bachelors are
unmarried), whereas synthetic truths are not (Jupiter is the largest planet in the solar system).
anti- realism: the term is used in a variety of ways in contemporary philosophy, but, generally speaking,
anti- realism is concerned with the rejection (or downgrading) of existence and/or truth. An anti-
realist about unicorns (say) maintains that although we talk about unicorns in much the same way
as we talk about cats or oranges, in reality unicorns do not exist, and so statements about unicorns
are either false, or neither true nor false; a realist about unicorns maintains that unicorns exist, and
428 GLOSSARY
that statements about them can be true. So far as scientific theories are concerned, some anti- realists
hold that we have no reason to believe that any of our theories are true, others say that we have no
reason to believe that the unobservable entities posited by certain theories exist (these claims can,
of course, be combined). Some anti- realists (or reductionists) about the past hold that, although
statements about the past can be true or false, what makes these statements true are not past states
of affairs, but presently available evidence, and, consequently, that statements about the past that
cannot be verified or falsified by presently existing evidence are neither true nor false a position
sympathetically explored by Michael Dummett.
a prioria posteriori: a statement or claim is a priori if it is possible to determine whether it is true
(or false) without consulting the world via observation or experiment. To verify (or falsify) an a
posteriori claim one has to consult the world. The application of the distinction is much disputed,
but most would agree that 1 + 1 = 2 and widows are women are a priori, whereas some
cats are black all over, others are not and radiation causes cancer are a posteriori. See also
analyticsynthetic, necessary, contingent.
A- properties/B- properties: A- properties include present, past, future; B- properties include earlier than,
simultaneous with, later than. The A- properties of an event are different as of different times,
B- properties are unchanging. A- properties are also called tensed and B- properties tenseless. X
happened three days ago (i.e., in the past) is an A- sentence, and X happened three days earlier
than Y is a B- sentence.
arrow of time: used in connection with the idea that time has a direction, in a way that space does
not. Temporal passage, if it were real, would provide time with an arrow, but so too might asym-
metries among the contents of time. There are numerous apparent content- asymmetries: the causal,
electromagnetic and thermodynamic, the psychological (intention, belief), the phenomenal
(involving experience), together with the knowledge- asymmetry (we know more about the past
than the future). How these various arrows are related to one another is one of the central problems
in the philosophy of time. See also time- reversibility, entropy.
A- theory/B- theory: prior to the 1980s, an A- theorist was usually someone who believed that B- sentences
could be translated (without change of meaning) by A- sentences, and a B- theorist someone who
believed the opposite. As it became clear that such translations are impossible, usage diversified.
The new (tenseless) theory of time is often called the B- theory, but the term is also used simply
to refer to the Block view, irrespective of other commitments. Consequently, an A- theorist might
be someone who either (i) believes that there are genuine A- properties and facts, and that it is
these that make A- statements true, and/or (ii) believes that an adequate account of time cannot be
formulated in entirely tenseless terms, and/or (iii) rejects the Block view, and holds that time is (in
some way) dynamic.
atoms: the tiny constituents of complex material objects and substances. A classical (or metaphysical)
atom is simple (has no parts) and indivisible. The atom of contemporary science consists of a
nucleus (made up of protons and neutrons), and a surrounding swarm of electrons. The atoms of
different elements have different numbers of protons, neutrons and electrons.
backward causation: normally, a cause c occurs before its effect e; in backward causation, e occurs
before c.
Bells inequality (or theorem): if the inequality holds, quantum mechanics is non- local. Experiments
have confirmed that the inequality holds, but the precise manner in which quantum mechanics is
non- local remains a matter of controversy.
big bang model: the hypothesis that the universe began somewhere between 10 and 15 billion years
ago in a tiny primeval fireball of extreme density, and has been expanding and cooling ever since.
Among the key pieces of evidence for the model are the observed red- shift of light from distant
galaxies (which means the distance between us and them is increasing) and microwave background
radiation.
Block view: the doctrine that past, present and future are all equally real (sometimes known as eter-
nalism). Although Block theorists subscribe to a static conception of time since they reject
GLOSSARY 429
temporal passage and the moving present they would reject the allegation that a Block universe is
static in the sense of containing no change. For the Block theorist, a universe (or object) changes
if it has different properties at different times, and our universe does. One occasionally (not here)
finds the block-universe being used to refer to universes where determinism (of the nomological)
variety obtains.
calculus: the branch of mathematics concerned with the study of the rates of change of continuously
varying quantities; used in most branches of science. Differential calculus provides the means for
finding the slope of a tangent to a curve at a given point, and the maximum and minimum values
of functions; it can be used to work out the precise rate at which one variable quantity is changing
relative to another (e.g. velocity is rate of change of distance with respect to time, acceleration is
rate of change of velocity with respect to time). Integral calculus is used to calculate the lengths
of curves, the areas of flat and curved surfaces, and hence magnitudes such as the total force acting
on a body.
causal theory of time: a programme of analysing temporal concepts in causal terms.
centre of gravity: the point at which the total weight of a body can be considered to act.
classical physics: Newtonian and post- Newtonian physics, prior to the development of relativity and
quantum mechanics in the early years of the twentieth century.
congruence: two objects or shapes are congruent if they have exactly the same size and shape.
connection: any sort of link or relation, but in spacetime theories a specification of points which fall
on straight (or straightest) lines (see geodesic).
consciousness: your consciousness as you read this sentence consists of everything you are experienc-
ing: your visual, auditory and tactile experiences, your thoughts, mental images, bodily sensations,
and so forth. We are not conscious of (in the sense of thinking about) most of our consciousness
for most of the time. Consciousness (in this semi- technical sense) does not require wakefulness: we
are conscious (i.e. have conscious states) when dreaming.
contingent: something that is not necessary. A true statement is contingent if it could have been false, an
entity is contingent if it might not have existed, an objects condition is contingent if it could have
been different. The laws of nature are usually taken to be contingent since it seems reasonable to
suppose there are logically possible worlds where the laws of nature are different.
continuant: a persisting object (or substance), specifically an object that lacks temporal parts (see
event).
continuum: in discussions of space and time continuous usually means gap-free and infinitely
divisible, and this is usually (but not always or inevitably) taken to mean that there are an infinite
number of points between any two points, no matter how close together they are. A continuum
is thus a space, time or spacetime which is free from gaps and infinitely divisible. In other contexts,
the real number line is often referred to as the continuum. This is because the real numbers are
free from the gaps that are a feature of the rational number line. It is commonly (although, again,
not always or inevitably) assumed that physical space and time have the structure of the real line.
coordinate system: a systematic assignment of numbers to spatial, temporal (or spatiotemporal) points,
such that each point receives a different number. For three- dimensional spaces, the Cartesian coor-
dinate system (familiar from school geometry, and invented by Descartes) employs three axes, x, y,
z, centred on a given point O, all at right angles to each other, and pointing in fixed directions; by
convention, the x- and y- axes are represented as horizontal, the z- axis as vertical; each point can
then be identified by three numbers, one for each axis. In a two- dimensional space, two numbers
suffice to identify each point; in a four- dimensional spacetime, four numbers are required, and so
on up hence the connection between coordinates and dimension. Other methods of coordinatiza-
tion are available: for example polar coordinates use combinations of distance and angle. Points in
curved spaces can also be assigned coordinates, but somewhat different procedures are required,
and in the case of variably curved spaces it will often be the case that several different coordinate
systems (or charts) are required to cover the whole space.
430 GLOSSARY
cosmic time: if we assume that the universe is effectively homogeneous and isotropic it is possible to
define a single timescale that is valid across the entire universe; observers who move with the cosmic
expansion can synchronize their clocks by reference to the local matter- density. In homogeneous
universes equipped with a cosmic time coordinate, it is possible (mathematically) to ignore the
intermixing of space and time that relativity theory usually brings, and treat space as an independent
three- dimensional continuum.
counterfactual: a statement that makes a claim that is contrary to fact, describing a possible but non-
actual state of affairs. A counterfactual conditional is an ifthen claim with a counterfactual
antecedent (If it were sunny then I would be happier.)
covariance: the property possessed by quantities that do not depend on their method of representation
in a given theory (e.g. the unit of measurement, or system of coordinates); only features that are
not artifacts of particular methods of representation are candidates for being physically real. A field
theory is said to be generally covariant if it is insensitive to the points underlying the field being
swapped around in certain specified ways (the relevant transformations are called diffeomorph-
isms).
curved space: spaces that conform to Euclidean geometry are said to be flat, and consequently spaces
that are non- Euclidean are said to be curved.
determinism: the nomological (or causal) determinist believes that the laws of nature are such that how
things are at one moment of time precisely fixes how things will be at all subsequent times, and
how things were at all earlier times. The temporal determinist believes that all times and events are
equally real, and hence that we can no more change the future than we can change the past (which
is not to say that what we do in the present will not contribute to the future being as it is). The two
determinisms are independent: one can be a nomological determinist without subscribing to the
Block view (and hence temporal determinism), and the laws of nature in a Block universe need not
be (nomologically) deterministic. See also fatalism.
differential geometry: the branch of mathematics that involves the application of calculus to curved
surfaces and spaces, of any number of dimensions, hence the study of differential (or differenti-
able) manifolds.
dimension: when used loosely (as in I believe in other dimensions) the term simply means universe,
in the narrow sense. In geometry, a point is said to be zero dimensional, a line one- dimensional, a
plane two- dimensional, and a cube (or sphere) three- dimensional, likewise the space in which we
appear to live. Although we cannot imagine them clearly, spaces of higher dimensions are logically
possible, and their properties have been explored by mathematicians, and exploited by physicists:
the spacetime of relativity theory is four- dimensional, and the idea that there are additional (but
concealed) spatial dimensions has proved useful in particle physics in recent years. Formulating a
precise mathematical definition of dimension is a non- trivial task (most current definitions appeal
to concepts in topology). The intuitively appealing idea that a spaces dimensionality is given
by counting up how many numbers are needed to provide a unique address for each point in
the space using a Cartesian- style coordinate system was undermined by the nineteenth- century
discovery that a single continuous line could completely fill (and hence be used to coordinatize) a
two- dimensional square.
dynamic models of time: dynamic theorists accord metaphysical significance to the passage of time, and
hence reject the Block view. There are different accounts of what passage involves: some dynamists
equate it with the acquisition and loss of the A- properties pastness, presentness and futurity (which
are construed as making a real difference to what possesses them), other dynamists posit variations
in the sum total of reality.
dynamics: the branch of physics concerned with bodies moving under the influence of forces (cf.
kinematics)
electromagnetic spectrum: the range of frequencies at which electromagnetic waves (also known as
radiation) can exist. It is customary to divide the spectrum into several broad bands, consisting
of radio waves (starting at about 3 10
4
Hz), microwaves, infrared, visible light (a narrow band
GLOSSARY 431
between 10
14
and 10
15
Hz), ultraviolet, X- rays and gamma rays (starting at about 3 10
21
Hz).
Lower frequency waves have longer wavelengths: radio waves can have wavelengths of several
hundred metres, whereas gamma waves are at the pico- metre scale (10
12
metres); the wavelength
of visible light ranges between 700 nanometres (red) and 420 nanometres (violet).
electromagnetic wave: a wave that consists of an electrical field and a magnetic field, oscillating at the
same rate at right- angles to each other in short, a ripple in the electromagnetic field. Electro-
magnetic waves travel at the speed of light in a vacuum, and, in effect, are self- perpetuating and
self- propelling, and so unlike other waves (such as sound), do not require a medium in which to
propagate. The existence (and speed) of such waves was predicted by Maxwell in the nineteenth
century; he also drew the (correct) conclusion that light was itself a form of electromagnetic radia-
tion. Maxwells theory of electromagnetism is a classical theory; the quantum theory of electro-
magnetism developed by Feynman quantum electrodynamics (QED) is regarded by some as being
the most successful current theory of matter.
empirical: concerned with what can be discovered using observation and experiment, as opposed to
what can be discovered using reason alone (see a priori).
entropy: a measure of the internal disorder (or randomness) of a closed system; the greater the disor-
der, the greater the entropy. The second law of thermodynamics says that the entropy of a closed
system can never decrease, and so remains constant or increases. The second law is now seen as
probabilistic rather than absolute, and neutral with respect to time: if a system exists in a state of
highly improbable order, it is likely to exist in a condition of greater (and more probable) disorder
at later and earlier times. This temporal symmetry leaves us with a puzzle: why does our universe
have a low- entropy past? See also time- reversibility.
epistemology: concerned with knowledge and what can be known.
essential v. non- essential: an entitys essential properties are those that it cannot exist without. A
triangle is essentially three- sided; I am not essentially two- legged (I could lose one and still survive).
event: a happening or process, typically extended over both time and space; standard examples include
battles, concert performances and arguments; events such as these have both spatial and temporal
parts (e.g. the first- half and second- half of a performance). Some maintain an event requires a
change of some sort to occur; others, drawing a distinction between boring and interesting
events, disagree. My usage in the main text reflects the latter view: I often use event to refer
to what happens at a single moment of time events of this duration cannot involve change. In
spacetime theories this practice is taken a step further: individual spacetime points, occupied or
unoccupied, are called events. According to traditional ontology, although entities such as cats and
planets are spatially extended and endure through time, they do not have temporal parts, and so
fall into an ontological category of their own: they are objects or things rather than events (see
substance, continuant). The distinction between object and event has come under pressure since
the advent of the four- dimensional (spacetime) way of thinking; in this context, everyday objects
are often regarded as extended events possessing both spatial and temporal parts.
fatalism: the doctrine that what will be will be. In the philosophical literature the term is often used to
refer to the doctrine that statements about contingent future happenings (Jones will win the lottery
next week.) have a definite truth-value as of the present. Thus construed, fatalism is entailed by
the Block view of time, at least in its standard guise. The doctrine of temporal determinism is also
called fatalism by some writers.
FeynmanWheeler absorber theory: an elegant but controversial account of electromagnetic radiation.
Electromagnetic waves that propagate in the earlier- to- later direction are called retarded they
arrive after their departure. These are the only sort of electromagnetic waves we are familiar with:
we never receive television broadcasts from the future, we receive them only from the past (even
live broadcasts suffer a slight time- delay, since electromagnetic radiation travels at the speed of
light, i.e. about one foot per nanosecond). However, Maxwells equations for electromagnetism
also predict advanced waves, which travel backwards in time and arrive before they were emit-
ted. This was generally regarded as an insignificant mathematical oddity, but in the 1940s Feynman
and Wheeler developed an account of electromagnetic radiation which involved both retarded and
432 GLOSSARY
advanced waves. According to this account, when energy is converted into electromagnetic radia-
tion, only half of the resulting radiation takes the form of retarded waves, the remainder is sent back
into the past, in the form of advanced waves. The retarded waves that do get emitted are absorbed
by electrons in the future, which in turn emit both advanced and retarded waves; these advanced
waves travel back in time to stimulate the original emitter so doubling its output, in accord with
what we observe as well as cancelling out all the advanced waves it emits which explains why
we never seem to receive emissions from the future. One thing the theory does not explain is why
the overall earlierlater asymmetry exists in the first place.
field: in physics, not a patch of grass but a force or influence that extends through space. A particular
field is represented by an assignment of numbers (or, if the relevant quantities have direction, vec-
tors) to each point of space at each moment of time. The field- concept came to the fore in physics
with Maxwells theory of electromagnetism in the nineteenth century. Field theories are local, in
that physical (or causal) influence is always transmitted from one small region of space to another,
hence there is no action- at- a- distance. Einsteins theory of gravity is a field theory, Newtons is not.
frame of reference: a coordinate system centred on a particular point O in space or spacetime, and
assumed to be at rest. In classical (Newtonian) physics, the standard reference frame is Cartesian
(three axes at right angles to each other). Different frames of reference can be created by moving
the original axes to a different point in space, retaining their directions (a translation), or by
rotating the axes around O, or by combining translation and rotation.
function: hair- dryers have a function to dry hair. The term is used differently in mathematics: a
function is akin to an input- output device, where what comes out depends upon the input and the
function. A simple example is the function + = : insert 2 either side of the + and 4 is
delivered as output, insert a couple of 3s and 6 is delivered.
Galilean relativity (or Galilean invariance or Galilean equivalence): physical laws have this prop-
erty if they remain valid in reference frames in uniform motion relative to one another. Newtons
laws of motion are Galilean invariant. Suppose we start with a frame of reference centred on O,
which we assume to be at rest in absolute space, and a system of bodies S; we now introduce a
second frame, M, moving uniformly relative to O: the axes of O and M remain parallel to one
another, but the distance between O and M is increasing at a steady rate. The bodies in S which
are moving inertially relative to O will also be moving inertially relative to M. Any non- inertial
motions (accelerations) will have the same magnitudes in both frames, and if Newtons laws are
used to calculate the forces acting on these accelerating bodies, precisely the same results will be
obtained in both frames. As a consequence, the fact ( if we assume it is one) that O is at rest and
M is moving has no experimental or observational consequences in the circumstances envisaged.
geodesic: the path of least distance connecting two points in a flat or curved space. Geodesics are also
paths of least curvature: although in curved regions of a space geodesics are themselves curves,
their curvature is limited to what the space itself imposes; they have no additional curvature, as
it were. (These notions can be made mathematically precise.)
Gold universe: a universe of finite duration that is (largely if not completely) symmetrical about its mid-
point, at least in respect of discernible physical processes. If the inhabitants of one side of a Gold
universe could travel to the other half , it would seem as if everything were running backwards.
gravity: what keeps us anchored to the Earths surface and the planets in their orbits. According to
Newtons universal law of gravitation, each material body exerts an attractive force on every other
body in the universe, a force that is proportional to its mass, and that decreases with the square of
the intervening distance. According to Einstein, gravitational effects are due not to a force, but to
spacetime curvature. See also inertial mass vs. gravitational mass.
Growing Block model: the universe is a spacetime block that is gradually increasing in size; past and
present are real, but the future is not.
hole (in space): there are different types of spatial hole referred to in the literature on space and
spacetime. A hole in a substantival space is a region within a manifold that contains no spatial
points, where the spatial substance, so to speak, is absent. A non- Euclidean hole is a region of
GLOSSARY 433
strong local spatial curvature. In the hole argument, the hole is a region in representations of a
spacetime where the points are switched about via a mathematical transformation.
idealism: the doctrine that our universe is entirely mental (i.e. composed of minds and mental states).
identity: two senses must be distinguished: numerical and qualitative. To say that x is identical (or the
same as) y could mean x and y are one and the same object this is numerical identity or it
could mean x and y are two distinct objects which are exactly alike this is qualitative identity.
identity- conditions: an objects synchronic (or at- a- time) identity conditions are the way an object must
be in order to qualify as an object of a particular kind. Spelling out precisely what these are is not
easy, but it is clear that the conditions an object must meet (at a given moment) to be a brick are
quite different from those it must meet to be a horse. Specifying an objects diachronic (across- time)
identity conditions involves spelling out the sorts of changes an object of a given kind can (and
cannot) undergo while remaining in existence. Different sorts of object have different diachronic
identity conditions.
identity of indiscernibles: the (dubious) claim that, for any objects x and y, if x and y are exactly alike
then they are also numerically identical (see identity). Not to be confused with the indiscernibility
of identicals, the far less contentious claim that if x and y are numerically identical (i.e. x and y
are simply different names of the same object), then x and y are exactly alike.
iff: an abbreviation for if and only if .
indexicals: words such as here, now, I, her, whose reference (e.g. a particular place, time,
person) depends upon the context of utterance, i.e. where a particular token of the indexical is
used, when, by whom, etc.
inertia: the tendency to resist acceleration; bodies with greater mass have greater inertia, and so require
greater force to accelerate them.
inertial forces: the forces which are suffered by a body when due to the impact of forces upon it it
moves non- inertially (i.e. when it accelerates or decelerates).
inertial frame (of reference): any frame of reference in which Newtons laws of motion hold; more
generally, a non- accelerating frame.
inertial mass v. gravitational mass In Newtonian physics, an objects inertial mass is the property
responsible for its resistance to acceleration; the greater an objects inertial mass, the greater the
force needed to accelerate that object by a given amount. Gravitational mass is the property
which features in the universal law of gravity: the greater an objects gravitational mass, the more it
attracts (and is attracted by) other objects. Inertial and gravitational mass are exactly equal which
explains the observed fact that bodies of different masses starting from the same point undergo the
same gravity- induced acceleration when dropped. However, there is no obvious reason why such
distinct properties should be exactly equal. (Electrical charge, which like gravitational mass
generates a force, is not equivalent to inertial mass.) In developing his alternative account of gravity,
Einstein sought an explanation for this mysterious coincidence; his solution involved the abolition
of gravity as a force operating between bodies.
inertial motion: the motion of a body which is not subject to any forces; sometimes called free- fall
motion. In Newtonian physics, bodies in inertial motion move in straight lines at constant speeds;
in general relativity straight line is replaced by geodesic.
infinite: sometimes taken to mean without boundary or limit. Space is infinite in this sense if, for any
distance one has travelled in a given direction, it is always possible to go further. Space is infinitely
divisible in this sense if, for any interval that has already been divided, it is always possible for it
to be divided again. This sort of infinity is the potential infinite introduced by Aristotle. As its
name suggests, an actual infinite involves more than mere potentiality. There is an actual infinity
of points in space if the total number of points that exists is greater than any positive finite number
(i.e. greater than any number in the series 1, 2, 3, 4, ). Although Aristotle believed that actual
(or completed) infinities are not to be found in nature, this is no longer generally assumed to tbe
the case. Many (although not all) contemporary theorists hold that space and time contain infinite
434 GLOSSARY
numbers of points. In the nineteenth century, Georg Cantor established that some infinite numbers
are larger than others.
inflationary universe: not a universe- wide tendency for prices to rise, but a universe that undergoes
a period of very rapid growth shortly after the big bang. The big bang models of the 1960s and
early 1970s faced a number of problems. One worry was the flatness problem (why the universe
at the present time doesnt have a higher degree of curvature, whether positive or negative, than
observations suggest it actually possesses), another was the smoothness problem (why the average
density of matter is so close to being homogeneous), another was the source of the primordial
density fluctuations responsible for galaxy formation. It turned out that all these problems could
be solved by positing a short period of exponential expansion: inflation, as it become known.
The inflationary period starts when the universe is about 10
43
seconds old, and comes to an end
shortly afterwards, well before the first second has ended. This short span of time is sufficient for
very considerable growth. One version of the inflationary theory has it that the universe doubles
in size every 10
34
seconds, a rate that amounts to 100 doublings in 10
32
seconds. Several different
quantum- theoretical explanations of where the energy that fuels inflation comes from have been
developed.
intentional: as well as using the term to talk about actions that are performed on purpose, philosophers
also use it in connection with representation. Mental states (such as thoughts, beliefs, desires), along
with signs, pictures and sentences (whether written or spoken) can be about things other than them-
selves; such items are said to possess the property of intentionality. There are also intentional
objects the intentional object of Texas is big is Texas. Since we can think (and have beliefs)
about entities such as unicorns and elves, intentional objects need not exist.
interval: in special relativity, the spatial and temporal distances between events are not absolute, they
vary between reference frames. However, one quantity is absolute: the square of the separation in
space minus the square of the separation in time. This quantity known as the spacetime interval
does not vary between reference frames. Events whose spatial separation is greater than their
temporal separation have a positive interval, and are said to be space- like connected; events whose
temporal separation is greater than their spatial separation have a negative interval, and are said to
be time- like connected; events whose interval is zero are said to be light- like connected.
intrinsic curvature: the curvature of a space (or surface) which is determined solely by features of the
space itself, without reference to any higher- order embedding space. Since the universe as a whole
cannot be embedded in a higher- dimensional space the idea that the space of our universe could
be curved only became a serious option when mathematicians (starting with Gauss and Riemann)
worked out ways of characterizing curvature intrinsically.
intrinsic nature: the way an object is in itself. Foster uses the expression in a more restricted way, to
refer to the qualitative characteristics a thing has over and above its causal and structural proper-
ties (intrinsic properties such as size and shape).
invariants: laws or quantities which remain unchanged by certain specified transformations (usually
with respect to position or coordinates). Shape and size are invariants of translations, rotations
and reflections in Euclidean space; Newtons laws of motion are invariant under translations and
rotations (sudden, once and for all) of reference frames. The speed of light is an invariant in special
relativity.
is: often used with different meanings, and hence a potential source of confusion. John is washing
up features the so- called tensed is; the utterance means that John is now doing the washing up.
Paris is the capital of France features the so- called tenseless is: there is no reference to any spe-
cific time; the same holds for the present tensed sentence Water boils at 100 degrees centigrade.
Superman is Clark Kent features the is of identity, whereas Superman is strong employs the
is of predication.
isotropic: a universe in which no particular direction is distinguished by the laws of nature, or the
behaviour and distribution of the contents of the space. A universe where this is not the case is
anisotropic.
GLOSSARY 435
kinematics: the branch of physics concerned with unforced motion.
logical possibility: the only things that are not logically possible are those which involve an internal
incoherence. Five- headed flying pigs are logically possible, round squares are not. See possible
worlds.
manifold: used loosely, another term for a spatial, temporal or spatiotemporal dimension. The term has
a more technical usage: in differential geometry the term refers to a continuous space (or collection
of points); a manifold in this sense has no spatiotemporal properties until one or more spacetime
structures are imposed (e.g. affine structure, a metric).
measurement problem: see quantum mechanics
metaphysics: the investigation into the most basic and/or general features of either reality itself, or the
conceptual schemes with which we operate. P. F. Strawson called the latter descriptive metaphys-
ics and the former revisionary metaphysics the rationale being that revisionary metaphysicians
are usually open to the possibility that reality may not conform to our ordinary ways of thinking.
So- called metaphysical systems (many of the best- known philosophers have produced one) are
full- scale attempts to bring all aspects of reality under a unified conceptual system. Although physics
is also concerned with the most basic and general aspects of reality, physicists confine themselves to
trying to explain the observable features of the universe (albeit often by positing unobservable enti-
ties and forces); for better or for worse, metaphysics (of the revisionary sort) is not so constrained.
metric: a measure of the distance between points in a space or spacetime, the numerical value of
which will depend on the coordinate system employed. In a flat space equipped with Cartesian
coordinates, the distance between two points is given by Pythagoras Theorem: d
2
= x
2
+ y
2
+ z
2
,
where x, y and z are the distances separating the points along the x- , y- and z- axes respectively. In
a curved space, the axes of any coordinate system will themselves be curved, and a more complex
formula is required, one employing correction factors (or curvature coefficients), the precise
numerical value of which will depend upon the coordinates chosen and the curvature of the space
in the relevant regions. In discussions of general relativity, the metric often refers to the metric
field tensor g (or g
ik
) which determines the inertial structure of spacetime (the affine connec-
tion), and thus which motions are accelerated and which are not, as well as the space- like/time- like
distinction, and the distances between points along all paths connecting them.
microwave background radiation: the slowly fading microwave glow left by the big bang, measured
to be around 2.7 degrees above absolute zero, a result which fits the predictions made by most
versions of the big bang model.
Minkowski spacetime: the flat spacetime of special relativity theory; see interval.
modal: concerning necessity and possibility. An objects modal properties concern the sorts of change
it is possible for it to undergo while remaining in existence (see identity conditions).
monadic: a non- relational property or predicate, such as is running or is sad; not to be con-
fused with monads: the spatially non- extended non- physical minds which feature prominently
in Leibnizs metaphysical system.
multiverse: sometimes used to refer to all the myriad branches of the universe posited by many- worlds
interpretations of quantum mechanics; more generally, any universe (in the broad sense) which
contains more than one spatiotemporal system can be said to constitute a multiverse.
necessary: a necessary truth is one which could not be otherwise than true, a necessary entity is one
which could not fail to exist. See also contingent and possible worlds.
necessary condition: see sufficient condition
new (tenseless) theory of time: the central claim is that although tensed sentences cannot be translated
by tenseless sentences, the conditions under which tensed sentences (as used on particular occa-
sions) are true can be specified in tenseless terms. There can thus be tensed truth in a world
lacking genuine tensed properties.
436 GLOSSARY
nomological: concerning the laws of nature. Something is nomologically necessary (or physically
necessary) if it is required by the laws of nature; something is nomologically possible if it is
permitted by the laws of nature. In nomologically possible worlds the laws of nature are the same
as in the actual universe, but history unfolds differently, often very differently.
non- local: theories which posit direct interactions between objects that are independent of distance
(see action- at- a- distance) are said to be non- local, whereas local theories only recognize inter-
actions which are transmitted in a continuous fashion across space, usually through fields.
noumenal: Kant distinguished the phenomenal world the realm of appearances, the world we
can know something about from the noumenal world the world as it really is, about which
(Kant claimed) we can know nothing at all. More generally, something is noumenal if it has no
empirical consequences.
Occams razor: the methodological precept of not multiplying entities beyond necessity; entities that
are not needed for explanatory purposes should be expelled from ones ontology i.e. not taken
to exist.
ontology: the branch of metaphysics concerned specifically with what exists. The ontology of a
theory (or its ontological commitments) are the entities (objects, properties, etc.) that you will
have to believe exist if you accept the theory as true.
overdetermination: an event is causally overdetermined if it has two (or more) causes, each of which
would have sufficed to bring the event in question about. Overdetermination of a different sort
occurs when an object or event is ascribed combinations of properties which cannot be possessed
together.
passage of time: the future- directed movement of the present, which results in present times becom-
ing past, past times becoming more past, and future times becoming less future, until they finally
become present and then past. The dispute between dynamic and non-dynamic views of time is
centred on the question of whether time really does pass.
persistence: entities that endure through time are said to persist; there are competing accounts of
what persistence involves. See also substance, event.
phenomenal property: those properties that we are directly aware of in our experience, for example
the smell and colour of a rose, the sound of a car- horn, the taste of nutmeg, pains, aches, feelings
of joy or nausea, and so forth. See secondary property, consciousness.
phenomenology: the enterprise of describing the character of consciousness, in all its aspects, in as clear
and systematic way as is possible. Phenomenology is concerned only with describing how things
seem to the experiencing subject, and ventures no claims about the causes of experience, or what
lies outside or beyond experience.
photon: a particle (or basic quantum- unit) of light.
positivism: see verificationism
possible worlds: a (logically) possible world is a way the universe as a whole might have been, but
isnt. There is a possible world which consists of nothing but a single elementary particle; there is a
possible world which is exactly like ours save for a single trivial detail; there are untold numbers of
possible worlds which are richer and more complex than the actual universe, and possible worlds
where the laws of nature are very different. A possible world is a consistent ensemble of logically
possible (non- contradictory) states of affairs; every consistent set of sentences describes a possible
world but no doubt there are possible worlds which cannot be described by any human language.
The concept of a possible world has proved useful in modal logic, and provides a convenient way of
talking about possibility and necessity; for example, a statement that expresses a necessary truth is
true in every possible world, whereas a contingent truth is true in at least one possible world. Most
philosophers are of the opinion that only one possible world is real or actual this one. But a minor-
ity modal realists take the view that all possible worlds are equally real. Also see nomological.
predicate: any linguistic expression used to ascribe a property to something.
GLOSSARY 437
Presentism: the doctrine that only what is present is real. Despite the simplicity of the basic claim, there
are very different variants of Presentism.
proper time: time as measured by a single observer (or any other system which changes in a regular
way); in special relativity, the time between two events as measured by an observer whose worldline
passes through both events.
property: a feature or characteristic, usually of a thing though some believe ordinary things are
themselves nothing but collections of interrelated properties. An objects intrinsic properties
depend on how the object is in itself ( is square, has a hole), whereas an objects relational
properties depend on how the object is related to other objects ( is bigger than , is
between and ). Since is a round- square seems a perfectly respectable predicate, it should
not be assumed that every predicate refers to a genuine property, i.e. something which could be a
real feature of a world. The question of whether A- predicates such as is past refer to genuine
properties is much disputed (see A- theory/B- theory).
proposition: often used to refer to the content of an utterance or belief. (See also sentence v. statement.)
quantum mechanics: our current best theory of the realm of the very small, which underlies a good deal
of our modern technology (e.g. lasers, computer chips). Quantum mechanics originated with the
discovery that electromagnetic radiation is packaged into discrete units (or quanta); in the 1920s
and 1930s detailed mathematical formalisms were developed (by Schrdinger, Heisenberg and
Dirac, among others) which proved highly successful in delivering accurate predictions; later devel-
opments rendered quantum theory compatible with special (but not general) relativity. Despite
its empirical power, quantum theory has several highly puzzling features, and debate continues
to rage over how the quantum formalisms should be interpreted. The measurement problem
is particularly significant in this respect. The fundamental dynamical law of quantum theory is
Schrdingers wave equation, which assigns specific probabilities to all the different ways a given
system of particles might evolve, but does not specify the way the system will in fact evolve this
can only be discovered by performing a test or observation. When a test is conducted, the wave
equation is said to collapse, since once a system is observed as being in a given state, the prob-
ability of finding it in all the other possible states reduces to zero. Prior to an observation being
made, the system is said to exist in a superposition of states, though quite what this means for
the state in question is unclear. According to the Copenhagen interpretation, from which the
preceding terminology derives, the question is simply illegitimate; a system only acquires definite
properties as and when these properties are observed or measured. However, this response leaves
many questions unanswered, and many theorists unsatisfied. Why should conducting a measure-
ment have such profound consequences? What happens to a system if no measurement is ever
conducted, perhaps because no conscious observers exist at the time? Various alternatives to the
Copenhagen interpretation have been proposed. Hidden variables theorists argue that the system
in question is in a definite state, and since quantum theory does not tell us what that state is, the
theory is incomplete. Many worlds theorists argue that each of the possibilities which feature
in the Schrdinger equation are realized: the wave equation never collapses; rather, when certain
interactions occur (including measurements) the universe splits into separate branches, one for
each element of the superposition. Other theorists argue that the wave equation quickly collapses
of its own accord due to interactions with other systems at the micro- level, which is why we never
encounter strange superposed objects in everyday life. Quantum theory has other notable and
surprising features, including wave- particle duality and the Heisenberg uncertainty principle,
according to which the more we know about one quantum quantity (e.g. position) the less we can
know about certain others (e.g. momentum). See also Bells inequality.
realism: see anti- realism
reductionism: see anti- realism
relations: objects can stand in many different relations to one another, and different relations them-
selves have different properties. I will only mention a few of the most important. Let x, y and z
refer to objects, and is R- related to mean bears the relation R to . If a relation is
symmetrical, then if x is R- related to y, then y will be R- related to x, for example is next to .
438 GLOSSARY
If a relation is transitive, then if x is R- related to y, and y is R- related to z, then x is R- related to z,
for example is larger than . A relation is reflexive if anything which possesses it bears the
relation to itself, for example is the same size as (we are all the same size as ourselves). A relation
which is reflexive, symmetrical and transitive is known as an equivalence relation. Equivalence
relations partition all their relata into distinct mutually exclusive groups, in such a way that each
member of a group bears the relation in question to every other member of the group. Having the
same age is an equivalence relation: it divides a collection of people into non- overlapping groups,
and each person belongs to only one group. Not all relations involve genuine connections between
objects, though some do: compare is tied to with is the same shape as . Objects with
the relational property same shape need not be in the same universe, but objects which are tied
together (by a piece of string, say) are linked by a real physical connection.
relationism: the view that space, time or spacetime do not exist as entities in themselves, over and above
material objects and their spatiotemporal relations.
relative motion: motion relative to other objects, rather than space itself.
relativity theory: a shorthand way of referring to Einsteins special theory of relativity, his general
theory of relativity, or both. The special theory posits a flat spacetime, and is based on the idea that
the laws of physics (and the speed of light) are the same in all inertial frames; it has several coun-
terintuitive consequences, such as length contraction, time dilation and the abolition of absolute
simultaneity. The general theory extends the special theory by explaining gravitational effects in
terms of spacetime curvature.
secondary properties (or qualities): a race of intelligent aliens with very different perceptual systems
from our own might well have no idea what we mean when we say that a certain flower is red
and tastes bitter, or than an object makes a whining sound, but they would have no trouble with
properties (and concepts) such as square, mass and size. The latter sort of property are called
primary, the former secondary. More generally, secondary properties are perceived charac-
teristics of objects that depend heavily upon the peculiarities of the perceptual system of whoever
is doing the perceiving; by contrast, primary properties are independent of the peculiarities of
perceptual systems. According to one influential account (Lockes), although secondary properties
such as colour seem to be intrinsic features of objects, in just the same way as the primary proper-
ties, this is an illusion; in reality, secondary qualities are causal properties: a red object is one that
causes certain types of experience (red- as- we- experience- it) in certain sorts of perceiver (typical
human) in certain sorts of circumstance (ordinary lighting conditions). As a consequence, there is
no reason to suppose red- as- we- experience it is an intrinsic feature of red objects themselves; if so,
physical reality, as it is in itself, is very different from how it appears to us.
sentence/statement: a sentence is a grammatical sequence of words (or more generally, meaningful
signs), a statement is what a particular sentence expresses on a particular occasion of use. Different
sentences can express the same statement (Its snowing, Il neige), and the same sentence can
express different statements (if you say Im hungry and I say Im hungry, we are saying dif-
ferent things using the same words). The term content is often used to refer to what statements
and beliefs (and other mental states) have in common when they involve the same assignment of
properties to the same objects; for example I say Mary is unhappy, you believe that Mary is
unhappy, and someone else hopes that Mary is unhappy. See proposition.
singularity: roughly speaking, a point or region of spacetime where curvature (or some other quantity)
becomes infinite, spacetime structure breaks down, and the standard laws of physics cannot be
applied. There are different types of singularity.
solipsism: the doctrine that only one thing exists, me. Solipsism of the present moment is the doc-
trine that only one moment of time exists, this one.
space- like (and time- like) separation: see interval
spacetime: often used to refer to the four- dimensional spatiotemporal frameworks employed in relativ-
ity theory, according to which space and time are no longer independent dimensions, but inter-
mingled in a distinctive way: the spatial and temporal distances between the same two events can
GLOSSARY 439
vary from one frame of reference to another, and so are no longer absolute. However, the concept
has wider applicability, and is used to refer to any collection of individual points which jointly con-
stitute a single spatiotemporal system. Contemporary geometrical treatments of spacetime begin
with a manifold of points and add one or more geometrical structures (such as a metric).
substance: in ordinary language the term can be used to refer to just about anything (Whats that
strange green sticky substance on your shirt?), but in metaphysics the term is generally used
exclusively to refer to enduring objects (dogs, bicycles) rather than stuffs (patches of green sticky
material, a litre of beer), or events. Another term sometimes used for substance in this sense is
continuant. Further restrictions are sometimes imposed (e.g. only an object which does not depend
on any other object for its existence is a substance), but not here.
substantivalism: the doctrine that space, time or spacetime are entities in their own right; denied by
relationism.
sufficient condition: if X is a sufficient condition of Y, then if X obtains, so too does Y. If X is a neces-
sary condition of Y, then Y cannot exist in the absence of X. If X is necessary and sufficient for Y,
then Y obtains iff X does, and vice versa.
supervenience: an asymmetrical relationship of dependency, which can come in different forms, depend-
ing on the precise mode of dependency, and the items so related. A (comparatively) uncontroversial
example of a supervenience claim: The large- scale shape of physical objects supervenes upon their
micro- structure. At the very least, this means that two objects with similar micro- constituents
arranged in similar ways will have similar macro- shapes, and that changes in macro- shape depend
on, and are brought about by, changes at the micro- level.
symmetry: in physics, used in connection with properties of a system which are unaffected by specified
transformations.
tangent: a line which touches a curve at only one point.
temporal part: see event
temporal vacuum: the temporal counterpart of the hole in space, which like the latter, can come
in different forms. A mild temporal vacuum consists of a period of time during which nothing
moves or changes the universe, as it were, undergoes a temporary total freeze. A stronger form
of temporal vacuum consists of a period of time during which nothing exists other than time and
space the universe is completely empty (of matter, minds) for a while. Relationists will gener-
ally reject the possibility of (at least) the stronger form of temporal vacuum, whereas space time
substantivalists will be inclined to accept the possibility of both.
tensor: a mathematical object, a generalization of the vector- concept, specified with respect to a given
coordinate system and able to undergo transformation to other coordinate systems; there are vari-
ous types (or ranks) of tensor (a vector is a first- rank tensor); generally employed to deal with
vectors in complex dependence relationships representing quantities that are not parallel. A tensor
in a spacetime is a function at a given point which takes vectors (or functions of vectors) as inputs
and delivers as output a number or vector, where the output depends on the inputs in a linear way.
A tensor field is an assignment of tensors to each point in a spacetime. Einsteins general theory of
relativity is formulated in terms of tensor- equations.
thermodynamics: the branch of physics concerned with the transformation of energy by heat and work.
tidal forces: forces suffered by extended bodies in a gravitational field. If you were to fall feet- first
into a black hole your body would be torn apart by tidal forces before you reached the singularity,
since a stronger pull would be exerted on your feet than on the upper parts of your body (your
feet being nearer to the source of the gravitational field).
time- reversibility: if the laws of nature are fully time- reversible (or time- symmetric), any physical
process would be able to unfold in either temporal direction. The laws of classical physics are
time- reversible, likewise relativity and the basic equations of quantum mechanics (or so it is gener-
ally held). However, some interpretations of quantum theory do seem committed to a deep asym-
metry between past and future: those which both posit and attribute ontological significance to
440 GLOSSARY
the collapse of the wave- equation. On this view, the past and present are fixed and determinate,
whereas the future is unfixed and indeterminate. Whether this interpretation of quantum theory is
correct is another matter. One piece of evidence from particle physics suggests the laws of nature
can in fact distinguish between the two directions of time: the decay of the neutral K- meson (or
kaon) is not perfectly time- symmetrical. The significance of this rare and subtle effect is not yet
known, but there is no obvious connection between it and the various macroscopic asymmetries.
See arrow of time, entropy.
tokens v. types: a token is a particular instance of a general kind, or type. In dddd there are four
tokens of a single type (in this case, a letter of the alphabet).
topic- neutral: a theory is topic- neutral if it restricts itself to making claims about causal and/or struc-
tural features, and says nothing about the intrinsic natures of the items which possess these features.
topology: the branch of geometry concerned with the properties of spaces (and spacetimes) that are
independent of metric: i.e. with how the different parts of a space are connected, rather than how
far apart they are. Squeezing and stretching a space (or shape) do not affect its topological proper-
ties, provided the space isnt torn and the local betweeness relation between nearby points is
not altered.
transcendental: pertaining to reality as it really is.
universe: in the broad sense of the term, the universe is all that exists; in its narrow sense, a universe is
a single spatiotemporal system and all that it contains. It may well be that the universe (in the broad
sense) consists of more than one spatiotemporal system. If so, then these different systems will not
be spatially or temporally related to one another, so to call them parallel universes is misleading.
vector: a quantity which has both a magnitude and a direction (e.g. velocity, acceleration); scalar
quantities have magnitude but lack direction (e.g. weight).
velocity: speed in a given direction.
verificationism: the doctrine (associated with logical positivism) that unless a (non- analytic) statement
is empirically verifiable or falsifiable it is meaningless. Hence claims about the world which go
beyond what is scientifically verifiable (or falsifiable) are literally meaningless (or cognitively
empty); many traditional metaphysical claims fall into this category or would do, if verification-
ism were true. In the opinion of many, the doctrine applies to itself and so is self- refuting.
void: nothingness.
wave function: see quantum mechanics
worlds: in philosophical writings, worlds are often entire universes, rather than individual planets;
see possible worlds.
worldline: the career of an object in spacetime, from beginning to end, represented as an extended
four- dimensional entity. Strictly speaking, only points have worldlines; spatially extended objects
such as your body occupy four- dimensional volumes. In spacetime diagrams, a curved worldline
indicates acceleration, a straight world line indicates uniform velocity, and an object at rest has a
worldline that is straight and vertical.
Web resources
The web changes quickly, and links come and go. The links below were all available in March 2010. For
a regularly updated listing of materials relevant to the issues discussed in the book, see the dedicated
site:
http://web.me.com/barrydainton1/timeandspace/contents.html
Listed below are a few of the more useful science- oriented sites.
1. The Los Alamos e- print archive is a fully automated electronic archive and distribution server
for research papers in physics, astronomy and other sciences. The search engine is efficient, and
downloads are available in multiple formats.
http://arXiv.org
2. Relativity on the World Wide Web has a wealth of information, including links to the physics and
relativity frequently asked questions (FAQs), guides and tutorials on relativity pitched at beginner,
intermediate and advanced levels.
http://webplaza.pt.lu/public/fklaess/html/relativity_on_www.html
3. Foundations, by Greg Egan, has useful introductions to STR, GTR and quantum theory.
www.netspace.net.au/~gregegan/FOUNDATIONS/index.html
4. This Weeks Finds in Mathematical Physics, by John Baez.
http://math.ucr.edu/home/baez/TWF.html
5. Jeff Weekss topology games (do crossword puzzles in multi- connected spaces).
www.geometrygames.org/
6. Andrew Hamiltons homepage has videos of what you would see as you fall into a black hole
singularity. Also pay a visit to the STR tour, which features a Guide to relativistic flight simula-
tors (web- based); the simulations of four- dimensional geometrical objects is also worth a visit.
http://casa.colorado.edu/~ajsh/home.html
7. Nasa has a page on dark matter and dark energy, with links to other useful pages.
http://nasascience.nasa.gov/astrophysics/what-is-dark-energy
8. For introductions to quantum theory, a useful starting point is the Thinkquest site Quantum
mechanics made simple:
http://library.thinkquest.org/C005775/frameset.html
442 WEB RESOURCES
9. A number of useful resources are available at the Perimeter Institute.
www.perimeterinstitute.ca/index.php?lang=en
10. Steve Prestons SciFi Science site has useful information and links.
http://freespace.virgin.net/steve.preston/Time.html
See also www.pbs.org/wgbh/nova/time/
11. Striking Hubble Space Telescope images of the far reaches of space and time.
http://oposite.stsci.edu/pubinfo/pictures.html
12. A time travel website (John W. Carroll and others):
http://www4.ncsu.edu/~carroll/time_travel/index.html
13. The Stanford Internet Encyclopedia has many useful entries on relevant topics in physics, some
quite technical, but many not. For example:
Quantum theory:
http://plato.stanford.edu/entries/qm/
Early Philosophical Interpretations of General Relativity
http://plato.stanford.edu/entries/genrel-early/
The Hole Argument
http://plato.stanford.edu/entries/spacetime-holearg/
Quantum Gravity
http://plato.stanford.edu/entries/quantum-gravity/
Holism and Nonserability in Physics
http://plato.stanford.edu/entries/physics-holism/
Absolute and Relational Theories of Space and Motion
http://plato.stanford.edu/entries/spacetime-theories/
Bibliography
Abbot, E. A. 1986. Flatland: A Romance of Many Dimensions by a Square. New York: Penguin.
Adams, R. M. 1986. Time and Thisness. In Midwest Studies in Philosophy, Vol. XI, P. A. French,
T. E. Uehling & H. K. Wettstein (eds), 31529. Minneapolis, MN: University of Minnesota Press.
Albert, D. Z. 1992. Quantum Mechanics and Experience. Cambridge, MA: Harvard University
Press.
Alexander, H. G. 1956. The LeibnizClarke Correspondence. Manchester: Manchester University
Press.
Al- Khalili, J. 1999. Black Holes, Wormholes and Time Machines. Bristol: Institute of Physics.
Alper, J. S. & M. Bridger 1997. Mathematics, Models and Zenos Paradoxes. Synthese 110(1):
14366.
Arntzenius, F. 2000. Are There Really Instantaneous Velocities? Monist 83(2): 187208.
Arntzenius, F. 2003. An Arbitrarily Short Reply to Sheldon Smith on Instantaneous Velocities. Stud-
ies in the History and Philosophy of Modern Physics 34(2): 2812.
Arntzenius, F. 2008. Gunk, Topology, Measure. In Oxford Studies in Metaphysics, D. Zimmerman
(ed.), vol. 4, 22547. Oxford: Oxford University Press.
Arntzenius, F. & J. Hawthorne 2005. Gunk and Continuous Variation. Monist 88(4): 44165.
Arntzenius, F. & T. Maudlin 2005. Time Travel and Modern Physics. Stanford Encyclopedia of
Philosophy, http://plato.stanford.edu/entries/time-travel-phys/ (accessed March 2010).
Arthur, R. 1994. Space and Relativity in Newton and Leibniz. British Journal for the Philosophy of
Science 45: 21940.
Atkins, P. W. 1986. Time and Dispersal: The Second Law. See Flood & Lockwood (1986), 8098.
Baez, J. C. 1996. General Relativity: Long Course Outline. http://math.ucr.edu/home/baez/gr/out-
line2.html (accessed March 2010).
Baez, J. C. 2001. The Meaning of Einsteins Equation. www.math.ucr.edu/home/baez/einstein/
(accessed March 2010).
Balashov, Y. & M. Janssen 2003. Presentism and Relativity. British Journal for the Philosophy of
Science 54: 32746.
Barbour, J. 1982. Relational Conceptions of Space and Time. See Butterfield et al. (1996), 14164.
Barbour, J. 1989. Absolute and Relative Motion. Cambridge: Cambridge University Press.
Barbour, J. 1999. The End of Time. London: Weidenfeld & Nicholson.
Barbour, J. & B. Bertotti 1977. Gravity and Inertia in a Machian Framework. Nuovo Cimento 38B:
127.
Barbour, J. & B. Bertotti 1982. Machs Principle and the Structure of Dynamical Theories. Proceed-
ings of the Royal Society of London 382: 295306.
Barnes, J. 1982. The Presocratic Philosophers. London: Routledge.
Bell, J. L. 2005. The Continuous and the Infinitesimal in Mathematics. Milan: Polimetrica.
Bell, J. L. 2009. Continuity and Infinitesimals. Stanford Encyclopedia of Philosophy. http://plato.
stanford.edu/entries/continuity/ (accessed March 2010).
Belot, G. 2000. Geometry and Motion. British Journal for the Philosophy of Science 51: 56195.
Belot, G. & J. Earman 1999. From Physics to Metaphysics. In From Physics to Philosophy, J. But-
terfield & C. Pagonis (eds), 16686. Cambridge: Cambridge University Press.
Belot, G. & J. Earman 2001. Pre- Socratic Quantum Gravity. See Callender & Huggett (2001),
21355.
444 BIBLIOGRAPHY
Benacerraf, P. 1962. Tasks, Supertasks and the Modern Eleatics. Journal of Philosophy 59: 76584.
Bergson, H. 1946 The Perception of Change. In The Creative Mind, M. L. Andison (trans.). New
York: Philosophical Library.
Biacino, L. & G. Gerla 1996. Connection Structures: Grzegorczyks and Whiteheads Definitions of
Point. Notre Dame Journal of Formal Logic 37(3): 4319.
Bigelow, J. 1991. Worlds Enough for Time. Nos 25, 119.
Bigelow, J. 1996. Presentism and Properties. Philosophical Perspectives 10: 3552.
Bigelow, J. & F. Pargetter 1989. Forces. Philosophy of Science 55: 61430.
Bigelow, J. & F. Pargetter 1990. Science and Necessity. Cambridge: Cambridge University Press.
Black, M. 1954. Problems of Analysis. London: Routledge & Kegan Paul.
Bohm, D. 1952. A Suggested Interpretation of Quantum Theory in Terms of Hidden Variables, Parts
I and II. Physical Review 85: 16693.
Bohm, D. & B. J. Hiley 1993. The Undivided Universe. London: Routledge.
Bourne, C. 2002. When am I? A Tense Time for some Tense Theorists? Australasian Journal of
Philosophy 80: 35971.
Bourne, C. 2006a. A Future for Presentism. Oxford: Oxford University Press.
Bourne, C. 2006b. A Theory of Presentism. Canadian Journal of Philosophy 36(1): 124.
Braddon-Mitchell, D. 2004. How do we Know it is Now Now? Analysis 64(3): 199203.
Bricker, P. 1993. The Fabric of Space. In Midwest Studies in Philosophy, Vol. XVIII, P. A. French,
T. E. Uehling & H. K. Wettstein (eds). Notre Dame, IN: Notre Dame University Press.
Brighouse, C. 1994. Spacetime and Holes. In Proceedings of the 1994 Biennial Meeting of the Phi-
losophy of Science Association, D. Hull, M. Forbes & R. M. Burian (eds), vol. 1, 11725. Lansing,
MI: Philosophy of Science Association.
Broad, C. D. 1923. Scientific Thought. London: Kegan Paul.
Broad, C. D. 1938. An Examination of McTaggarts Philosophy. Cambridge: Cambridge University
Press.
Broad C. D. 1998. McTaggarts Arguments Against the Reality of Time: An Excerpt from Examina-
tion of McTaggarts Philosophy. See Van Inwagen & Zimmerman (1998), 749.
Brown, H. R. 2005. Physical Relativity: Space-Time Structure from a Dynamical Perspective. Oxford:
Clarendon.
Brown, H. R. & O. Pooley 2002. Relationalism Rehabilitated? I: Classical Mechanics. British Journal
for the Philosophy of Science 53: 183204.
Brown, H. R. & O. Pooley 2006. Minkowski Space-Time: A Glorious Non-Entity. In The Ontology
of Spacetime, D. Dieks (ed.), 6789. Amsterdam: Elsevier.
Butterfield, J. 1989a. The Hole Truth. British Journal for the Philosophy of Science 40: 128.
Butterfield, J. 1989b. Albert Einstein meets David Lewis. In Proceedings of the 1988 Biennial Meeting
of the Philosophy of Science Assocation, A. Fine & J. Leplin (eds), vol. 2. East Lansing, MI: Philosophy
of Science Association. Reprinted in Butterfield et al. (1996), 295311.
Butterfield, J. 1999. The Arguments of Time. Oxford: British Academy.
Butterfield, J. 2002. The End of Time. British Journal for the Philosophy of Science 53: 389330. A
longer version available at http://arxiv.org/abs/gr-qc/0103055v1 (accessed June 2010).
Butterfield, J. 2006a. Against Pointillisme about Mechanics. British Journal for the Philosophy of
Science 57: 70954. http://arXiv.org/abs/physics/0512064 (accessed March 2010).
Butterfield, J. 2006b. Against Pointillisme about Geometry. In Proceedings of the 28th Ludwig Witt-
genstein Symposium, F. Stadler & M. Stoeltzner (eds). Longer version available at: http://philsci-
archive.pitt.edu/archive/00002552/ (accessed March 2010).
Butterfield J. & C. Isham 1999. On the Emergence of Time in Quantum Gravity. See Butterfield
(1999), 11168.
Butterfield, J. & C. Pagonis (eds) 1999. From Physics to Philosophy. Cambridge: Cambridge University
Press.
Butterfield, J., M. Hogarth & G. Belot (eds) 1996. Spacetime. Aldershot: Dartmouth.
Button, T. 2006. Theres no Time like the Present. Analysis 66(290): 13035.
Button, T. 2007. Every Now and Then, No-futurism Faces no Sceptical Problems. Analysis 67(296):
32532.
Callender, C. 1998a. The View from Nowhen. British Journal for the Philosophy of Science 49:
13559.
BIBLIOGRPAHY 445
Callender, C. 1998b. Shedding Light on Time. Proceedings of the 1998 Biennial Meeting of the
Philosophy of Science Association 67(3), part II, supplement.
Callender, C. 2000. Is Time Handed in a Quantum World? Proceedings of the Aristotelian Society
100(3): 24769.
Callender, C. (ed.) 2002. Time, Reality and Experience. Cambridge: Cambridge University Press.
Callender, C. (ed.) (forthcoming). The Oxford Handbook of Time. Oxford: Oxford University Press.
Callender, C. & N. Huggett (eds) 2001. Physics Meets Philosophy at the Planck Scale. Cambridge:
Cambridge University Press.
Callender, C. & R. Weingard 1996. An Introduction to Topology. Monist 79: 2133.
Cameron, R. 2008. Comments on Merrickss Truth and Ontology. Philosophical Books 49(4):
292301.
Cameron, R. 2010. Truthmaking for Presentists. In Oxford Studies in Metaphysics, D. Zimmerman
(ed.), vol. 6. Oxford: Oxford University Press.
Carnap, R. 1963. Carnaps Intellectual Biography. In The Philosophy of Rudolf Carnap, P. A. Schilpp
(ed.). La Salle, IL: Open Court.
Carnap, R. 1994. An Introduction to the Philosophy of Science, M. Gardner (ed.). New York: Dover.
Carroll, J. W. 2002. Instantaneous Motion. Philosophical Studies 110(1): 4967.
Carroll, S. 2001. The Cosmological Constant. Living Reviews in Relativity. http://relativity.livingre-
views.org/Articles/lrr-2001-1/ (accessed March 2010).
Carroll, S. 2010. From Eternity to Here. London: Dutton.
Cau, T. C. 1998. Conceptual Developments of 20th Century Field Theories. Cambridge: Cambridge
University Press.
Chalmers, D. 1996. The Conscious Mind. Oxford: Oxford University Press.
Chown, M. 1999. Cosmic Crystal. New Scientist 161(2173) (13 February): 42.
Chown, M. 2000. Backwards to the Future. New Scientist 165(2224) (6 February): 26.
Clarke, B. 1981. A Calculus of Individuals Based on Connection. Notre Dame Journal of Formal
Logic 22(3): 20418.
Clarke, B. 1985. Individuals and Points. Notre Dame Journal of Formal Logic 26(1): 6175.
Clifton, R. & M. Hogarth 1995. The Definability of Objective Becoming in Minkowski Spacetime.
Synthese 103: 35587.
Clifton, T., P. Ferreira & K. Land 2008. Living in a Void: Testing the Copernican Principle with
Distant Supernovae. http://arXiv.org/abs/0807.1443 (accessed June 2010).
Coles, P. (ed.) 2001. The New Cosmology. London: Routledge.
Craig, W. L. 2000. The Tensed Theory of Time. Dordrecht: Kluwer.
Cramer, J. G. 1980. Generalized Absorber Theory and the EinsteinPodolskyRosen Paradox. Physi-
cal Review D, 22: 17789.
Cramer, J. G. 1983. The Arrow of Electromagnetic Time and the Generalized Absorber Theory.
Foundations of Physics 13: 887902.
Cramer J. G. 1986. The Transactional Interpretation of Quantum Mechanics. International Journal
of Theoretical Physics 27: 22736.
Cramer, J. G. 2000. The Plane of the Present and the New Transactional Paradigm of Time. In Time
and the Instant, R. Durie (ed.), 17789. Manchester: Clinamen Press.
Crisp, T. M. 2003. Presentism. In The Oxford Handbook of Metaphysics, M. J. Loux & D. W. Zim-
merman (eds), 21145. Oxford: Oxford University Press.
Crisp, T. M. 2004. On Presentism and Triviality. See Zimmerman (2004), 1520.
Crisp, T. M. 2007. Presentism and the Grounding Objection. Nos 41(1): 90109.
Dainton, B. 1992. Time and Division. Ratio 5: 10228.
Dainton, B. 2000. Stream of Consciousness. London: Routledge.
Dainton, B. 2009. Causation and Laws of Nature. In Central Issues of Philosophy, John Shand (ed.),
193207. Oxford: Blackwell.
Dainton, B. (forthcoming a). Time, Passage and Immediate Experience. See Callender (forthcoming).
Dainton, B. (forthcoming b). Temporal Consciousness. Stanford Encyclopedia of Philosophy.
Davies, P. 1986. Time Asymmetry and Quantum Mechanics. See Flood & Lockwood (1986), 98124.
Davies, P. 1995. About Time: Einsteins Unfinished Revolution. Harmondsworth: Penguin.
Davies, P. C. W. & J. Brown (eds) 1988. Superstrings: A Theory of Everything? Cambridge: Cambridge
University Press.
446 BIBLIOGRAPHY
Deutsch, D. 1997. The Fabric of Reality. Harmondsworth: Penguin.
Deutsch, D. & M. Lockwood 1994. The Quantum Physics of Time Travel. Scientific American
270(3) (March): 5056.
Dewdney, A. K. 1984. The Planiverse. London: Picador.
Diekemper, J. 2005. Presentism and Ontological Symmetry. Australasian Journal of Philosophy
83(2): 22340.
Dolev, Y. 2000. Dummetts Antirealism and Time. European Journal of Philosophy 8: 25376.
Dorato, M. 1995. Time and Reality: Spacetime Physics and the Objectivity of Temporal Becoming.
Bologna: CLUEB.
Dorato, M. 1998. Becoming and the Arrow of Causation. Proceedings of the 1998 Biennial Meeting
of the Philosophy of Science Association 67(3), part II, supplement.
Dowe, P. 2000. The Case for Time Travel. Philosophy 75(3): 44151.
Dowe, P. 2003. The Coincidences of Time Travel. Philosophy of Science 70: 57489.
Dummett, M. 1964. Bringing About the Past. Philosophical Review 73: 33859. Reprinted in Le
Poidevin & MacBeath (1993), 11733.
Dummett, M. 1978. On the Reality of the Past. In his Truth and Other Enigmas. London: Duckworth.
Dummett, M. 1986. Causal Loops. See Flood & Lockwood (1986), 13569.
Dummett, M. 2000. Is Time a Continuum of Instants? Philosophy 75: 497515.
Dummett, M. & A. Flew 1954. Can an Effect Precede its Cause? Aristotelian Society Supplement
28: 2762.
Durie, R. (ed.) 2000. Time and the Instant. Manchester: Clinamen Press.
Earman, J. 1971. Kant, Incongruous Counterparts and the Nature of Space and Space- Time. Ratio
13: 118.
Earman, J. 1974. An Attempt to Add a Little Direction to The Problem of the Direction of Time.
Philosophy of Science 41: 1547.
Earman, J. 1989. World Enough and Spacetime. Cambridge, MA: MIT Press.
Earman, J. 1995. Bangs, Crunches, Whimpers and Shrieks. Oxford: Oxford University Press.
Earman, J. & J. Norton 1987. What Price Space- Time Substantivalism? The Hole Story. British
Journal for the Philosophy of Science 38: 51525.
Earman, J. & J. D. Norton 1996. Infinite Pains: The Trouble with Supertasks. In Benacerraf and His
Critics, A. Morton & S. Stich (eds), 23161. Oxford: Blackwell.
Earman, J., A. J. Janis, J. Massey & N. Rescher (eds) 1993. Philosophical Problems of the Internal and
External Worlds. Pittsburgh, PA: University of Pittsburgh Press.
Earman, J. & C. W. Smeenk 2009. Do the Laws of Physics Forbid the Operation of Time Machines?
Synthese 169: 91124.
Egan, G. 1995. The Hundred Light Year Diary. In his Axiomatic, 2139. London: Orion/Millenium.
Egan, G. 1997. Diaspora. London: Orion/Millenium.
Egan, G. 1998. Foundations 14. www.netspace.net.au/~gregegan (accessed March 2010).
Einstein, A. 1922. Sidelights on Relativity. London: Methuen.
Einstein, A. 1961. Relativity: The Special and the General Theory. New York: Three Rivers Press.
Ellis, G. F. R. 1978. Is the Universe Expanding?. General Relativity and Gravitation 8: 8794.
Fine, K. 2005. Tense and Reality. In his Modality and Tense: Philosophical Papers, 261320. Oxford:
Oxford University Press.
Fitzgerald, P. 1985. Four Kinds of Temporal Becoming. Philosophical Topics 13: 14577.
Flew, A. 1954. Can an Effect Precede its Cause? Proceedings of the Aristotelian Society, Supplemen-
tary Volume 28: 4562.
Flew, A. 1956. Effects Before their Causes? Addenda and Corrigenda. Analysis 16: 10410.
Flew, A. 1957. Causal Disorder Again. Analysis 17: 816.
Flood, R. & M. Lockwood (eds) 1986. The Nature of Time. Oxford: Blackwell.
Forrest, P. 1995. Is Space-Time Discrete or Continuous? An Empirical Question. Synthese 103:
32754.
Forrest, P. 1996a. From Ontology to Topology in the Theory of Regions. Monist 79(1): 3450.
Forrest, P. 1996b. How Innocent is Mereology? Analysis 56(3): 12731.
Forrest, P. 2004. The Real but Dead Past: A Reply to Braddon-Mitchell. Analysis 64(3): 35862.
Forrest, P. 2006. Uniform Grounding of Truth and the Growing Block Theory: A Reply to Heath-
wood. Analysis 66(2): 1613.
BIBLIOGRPAHY 447
Forrest, P. 2009. A World for Us: The Case for Phenomenalistic Idealism John Foster. Philosophical
Quarterly 59(237): 74043.
Foster, J. 1979. In Self- Defence. In Perception and Identity, G. F. Macdonald (ed.). London: Macmillan.
Foster, J. 1982. The Case for Idealism. London: Routlege & Kegan Paul.
Foster, J. 1985. Ayer. London: Routledge & Kegan Paul.
Foster, J. 1991. The Immaterial Self. London: Routledge.
Foster, J. 1993. The Succinct Case for Idealism. In Objections to Physicalism, H. Robinson (ed.).
Oxford: Clarendon.
Foster, J. 2000. The Nature of Perception. Oxford: Oxford University Press.
Foster, J. 2008. A World for Us. Oxford, Oxford University Press.
Frances, B. (in progress). The Relation of Existence to Time, www.fordham.edu/philosophy/frances/
(accessed March 2010).
Friedman, M. 1972. Grnbaum on the Conventionality of Geometry. Synthese 24(12): 21935.
Friedman, M. 1983. Foundations of Space- Time Theories. Princeton, NJ: Princeton University Press.
Galison, P. 2003. Einsteins Clocks, Poincars Maps. London: Hodder & Stoughton.
Gallois, A. 1994. Asymmetry in Attitudes and the Nature of Time. Philosophical Studies 76: 5169.
Gallois, A. 1998. Occasions of Identity. Oxford: Clarendon.
Garrett, B. J. 1988. Thank Goodness Thats Over Revisited. Philosophical Quarterly 39: 2015.
Reprinted in The New Theory of Time, L. N. Oaklander & Q. Smith (eds), 31621 (New York: St
Martins Press, 1994).
Gaul, M. & C. Rovelli 1999. Loop Quantum Gravity and the Meaning of Diffeomorphism Invari-
ance. http://arXiv.org/abs/gr- qc/9910079 (accessed March 2010).
Gerla, G. 1990. Pointless Metric Spaces. Journal of Symbolic Logic 55: 20719.
Geroch, R. 1978. General Relativity from A to B. Chicago, IL: Chicago University Press.
Gibbard, A. 1975. Contingent Identity. Journal of Philosophical Logic 4: 187222.
Goddu, G. C. 2003. Time Travel and Changing the Past: (Or How to Kill Yourself and Live to Tell
the Tale). Ratio 16: 1732.
Gdel, K. 1949. A Remark about the Relationship Between Relativity Theory and Idealistic Philosophy.
In Albert Einstein: Philosopher- Scientist, P. A. Schilpp (ed.). La Salle, IL: Open Court.
Godfrey- Smith, W. 1979. Special Relativity and the Present. Philosophical Studies 36: 23344.
Gold, T. 1962. The Arrow of Time. American Journal of Physics 30: 40310.
Gott, R. J. 2001. Time Travel in Einsteins Universe. London: Orion.
Gott, R. J. & L.-X. Li 1998. Can the Universe Create Itself? Physical Review D 58: 023501.
Gray, J. 2003. A Mathematician Looks at Wolframs New Kind of Science. Notices of the American
Mathematical Society 50(2): 200211.
Greene, B. 1999. The Elegant Universe. London: Jonathan Cape.
Greene, B. 2004. The Fabric of the Cosmos. London: Penguin.
Grey, W. 1999. Troubles with Time Travel. Philosophy 74: 5570.
Grnbaum, A. 1952. A Consistent Conception of the Extended Linear Continuum as an Aggregate of
Unextended Elements. Philosophy of Science 19(4): 288306.
Grnbaum, A. 1955. Modern Science and the Refutation of the Paradoxes of Zeno. Scientific
Monthly 81: 234-239. Reprinted with revisions in Salmon (2001), 16475.
Grnbaum, A. 1968. Modern Science and Zenos Paradoxes. London: George Allen & Unwin.
Grzegorczyk, A. 1960. Axiomatizability of Geometry without Points. Synthese 12: 22835.
Guth, A. H. 1997. The Inflationary Universe. London: Jonathan Cape.
Harrison, C. 1996. The Three Arrows of Zeno: Cantorian and Non-Cantorian Concepts of the
Continuum and of Motion. Synthese 107(2): 27192.
Harrison, J. 1979. Jocastas Crime. Analysis 39: 65.
Harrison, J. 1980. Report on Analysis Problem no. 18. Analysis 40: 659.
Hawthorne, J. & B. Weatherson 2004. Chopping Up Gunk. Monist 87(3): 33950.
Heathwood, C. 2005. The Real Price of the Dead Past: A Reply to Forrest and to Braddon-Mitchell.
Analysis 65(3): 24951.
Heinlein, R. A. 1970. All You Zombies. In The Mirror of Infinity, R. Silverberg (ed.). San Francisco,
CA: Canfield.
Hellman, G. 2006. Mathematical Pluralism: The Case of Smooth Infinitesimal Analysis. Journal of
Philosophical Logic 35(6): 62151.
448 BIBLIOGRAPHY
Hestevold, H. S. 2008. Presentism Through Thick and Thin. Pacific Philosophical Quarterly 89:
32547.
Hinchliff, M. 1998. A Defense of Presentism in a Relativistic Setting. Proceedings of the 1998 Biennial
Meeting of the Philosophy of Science Association 67(3), part II, supplement.
Hoefer, C. 1996. The Metaphysics of Spacetime Substantivalism. Journal of Philosophy 93: 527.
Hoefer, C. 1998. Absolute Versus Relational Spacetime: For Better or Worse, the Debate Goes On.
British Journal for the Philosophy of Science 49: 45167.
Hoefer, C. 2000. Kants Hands and Earmans Pions: Chirality Arguments for Substantival Space.
International Studies in the Philosophy of Science 14: 23756.
Hoefer, C. & N. Cartwright 1993. Substantivalism and the Hole Argument. In Philosophical Prob-
lems of the Internal and External Worlds, J. Earman, A. J. Janis, J. Massey & N. Rescher (eds), 2343.
Pittsburgh, PA: University of Pittsburgh Press.
Hoffman, B. 1972. Albert Einstein: Creator & Rebel. New York: New American Library.
Hollis, M. 1967. Time and Spaces. Mind 76: 52436.
Horwich, P. 1975a. Grnbaum on the Metric of Space and Time. British Journal for the Philosophy
of Science 26(3): 199211.
Horwich, P. 1975b. On Some Alleged Paradoxes of Time Travel. Journal of Philosophy 72: 43244.
Horwich, P. 1978. On the Existence of Time, Space and Space- time. Nos 12: 396419.
Horwich, P. 1987. Asymmetries in Time. Cambridge, MA: MIT Press.
Hospers, J. 1997. An Introduction to Philosphical Analysis, 4th edn. London: Routledge.
Hudson, H. 2005. The Metaphysics of Hyperspace. Oxford: Oxford University Press.
Huggett, N. 1999. Space from Zeno to Einstein. Cambridge, MA: MIT Press.
Huggett, N. 2000. Reflections on Parity Non-Conservation. Philosophy of Science 67: 21941.
Huggett, N. 2006. The Regularity Account of Relational Spacetime. Mind 115(456): 4173.
Huggett, N. 2008. Why the Parts of Absolute Space are Immobile. British Journal for the Philosophy
of Science 59(3): 391407.
Husserl, E. 1950. The Phenomenology of Internal Time- Consciousness, J. S. Churchill (trans.). Bloom-
ington, IN: Indiana University Press.
Husserl, E. 1991. On the Phenomenology of the Consciousness of Internal Time (18931917), J. B.
Brough (ed. & trans.). Dordrecht: Kluwer.
James, W. 1952. The Principles of Psychology. Chicago, IL: Encyclopedia Britannica.
Jammer, M. 1993. Concepts of Space, 3rd edn. London: Dover.
Kant, I. 1996. Prolegomena to Any Future Metaphysics, B. Logan (ed.). London: Routledge.
Keller, S. & M. Nelson 2001. Presentists Should Believe in Time Travel. Australasian Journal of
Philosophy 79(3): 33345.
Keller, S. 2004. Presentism and Truthmaking. See Zimmerman (2004): 83104.
Kierland, B. & B. Monton 2007. Presentism and the Objection from Being-Supervenience. Australa-
sian Journal of Philosophy 85(3): 48597.
Kirk, G. S. & J. E. Raven 1969. The Presocratic Philosophers. Cambridge: Cambridge University
Press.
Kirshner, R. P. 2002. The Extravagant Universe: Exploring Stars, Dark Energy, and the Accelerating
Cosmos. Princeton, NJ: Princeton University Press.
Kuchar , K. 1999. The Problem of Time in Quantum Geometrodynamics. See Butterfield (1999),
16995.
Lachize- Rey, M. & J.-P. Luminet 1995. Cosmic Topology. Physics Reports 254: 136. Also available
at http://arXiv.org/abs/gr- qc/9605010 (accessed March 2010).
Lakoff, G. & R. Nez 2000. Where Mathematics Comes From. New York: Basic Books.
Lange, M. 2005. How can Instantaneous Velocity Fulfil its Causal Role? Philosophical Review
114(4): 43368.
Laraudogoitia, P. 2004. Supertasks. Stanford Encyclopedia of Philosophy. http://plato.stanford.edu/
entries/spacetime-supertasks/ (accessed March 2010).
Le Poidevin, R. 1991. Change, Cause and Contradiction. Basingstoke: Macmillan.
Le Poidevin, R. (ed.) 1998a. Questions of Time and Tense. Oxford: Oxford University Press.
Le Poidevin, R. 1998b. The Past, Present and Future of the Debate about Tense. See Le Poidevin
(1998a), 1342.
Le Poidevin, R. 2004. Review of Time and Space by Barry Dainton. Philosophy 79(3): 48690.
BIBLIOGRPAHY 449
Le Poidevin, R. & M. MacBeath (eds) 1993. The Philosophy of Time. Oxford: Oxford University
Press.
Lee, G. 2006. The Experience of Left and Right. In Perceptual Experience, J. Hawthorne, & T. Szabo
Gendler (eds), 291315. Oxford: Oxford University Press.
Lee, G. 2007. Consciousness in a Space-Time World. Philosophical Perspectives 21(1): 34174.
Leiber, L. 2008. The Einstein Theory of Relativity. Philadelphia, PA: Paul Dry Books.
Lewis, D. 1971. Counterparts of Persons and their Bodies. Journal of Philosophy 68: 20311.
Lewis, D. 1976. The Paradoxes of Time Travel. American Philosophical Quarterly 13: 14552.
Reprinted in Le Poidevin & MacBeath (1993), 13446.
Lewis, D. 1986. On the Plurality of Worlds. Oxford: Blackwell.
Lewis, D. 1991. Parts of Classes. Oxford: Oxford University Press.
Lindley, D. 2001. Boltzmanns Atom. New York: Free Press.
Lockwood, M. 1989. Mind, Brain and the Quantum: The Compound I. Oxford: Blackwell.
Loemker, L. E. (ed.) 1970. Leibniz: Philosophical Papers and Letters. Dordrecht: Reidel.
Lowe, E. J. 1987. The Indexical Fallacy in McTaggarts Proof of the Unreality of Time. Mind 96:
6270.
Lowe, E. J. 1993. McTaggarts Paradox Revisited. Mind 101: 3236.
Lowe, E. J. 1998. Tense and Persistence. See Le Poidevin (1998a), 4360.
Lowe, E. J. 2002. A Survey of Metaphysics. Oxford: Oxford University Press.
Lucas, J. R. 1984. Space, Time and Causality. Oxford: Clarendon.
Lucas, J. R. 1986. The Open Future. See Flood & Lockwood (1986), 12534.
Lucas, J. R. 1999. A Century of Time. See Butterfield (1999), 120.
Lucretius 1951. On the Nature of the Universe, R. E. Latham (trans.). Harmondsworth: Penguin.
Ludlow, P. 2004. Presentism, Triviality, and the Varieties of Tensism. See Zimmerman (2004): 2136.
Lukasiewicz, J. 1967. Determinism. In Polish Logic 19201939, S. McCall (ed.). Oxford: Oxford
University Press.
Luminet, J.- P. 1998. Past and Future of Cosmic Topology. http://arXiv.org/abs/gr-qc/9804006
(accessed March 2010).
Luminet, J.- P. 2001. LUnivers Chiffonn. Paris: Fayard.
Luminet, J.-P. 2008. The Shape and Topology of the Universe. http://arXiv.org/abs/0802.2236
(accessed March 2010).
Luminet, J.- P. & B. F. Roukema 1999. Topology of the Universe: Theory and Observation. http://
arXiv.org/abs/astro- ph/9901364 (accessed March 2010).
Luminet, J.- P., G. D. Starkman & J. R. Weeks 1999. Is Space Finite? Scientific American Online
(April), www.scientificamerican.com/article.cfm?id=is-space-finite (subscription only; accessed
March 2010).
MacBeath, M. 1982. Who was Dr Whos Father? Synthese 51: 397430.
MacBeath, M. 1983. Communication and Time Reversal. Synthese 56: 2646.
Manders, K. L. 1982. On the Space- Time Ontology of Physical Theories. Philosophy of Science 49:
57590.
Markosian, N. 1993. How Fast Does Time Pass? Philosophy and Phenomenological Research 53:
82944.
Markosian, N. 2004. A Defense of Presentism. See Zimmerman (2004): 4781.
Martin, A. 2009. Beware Invisible Cows. London: Simon & Schuster.
Maudlin, T. 1989. The Essence of Spacetime. Proceedings of the Philosophy of Science Association
1988, vol. 2, 8299. Reprinted in Butterfield et al. (1996), 31322.
Maudlin, T. 1993. Buckets of Water and Waves of Space: Why Spacetime is Probably a Substance.
Philosophy of Science 60: 183203. Reprinted in Butterfield et al. (1996), 26383.
Maudlin, T. 1994. Quantum Non- Locality and Relativity. Oxford: Blackwell.
Maudlin, T. 2007. On the Passing of Time. In his The Metaphysics Within Physics, ch. 4. New York:
Oxford University Press.
McCall, S. 1976. Objective Time Flow. Philosophy of Science 43: 33762.
McCall, S. 1984. A Dynamic Model of Temporal Becoming. Analysis 44: 1726.
McCall, S. 1994. A Model of the Universe. Oxford: Oxford University Press.
McKinnon, N. 2003. Presentism and Consciousness. Australasian Journal of Philosophy 81(3):
30523.
450 BIBLIOGRAPHY
McLauglin, W. I. [1994] 2003. Resolving Zenos Paradoxes. Mathematical American (December
2003): 1620; www.sciamdigital.com/index.cfm?fa=Products.ViewIssuePreview&ARTICLEID_
CHAR=61564081-2B35-221B-63C168F9993F293D [via subscription].
McLaughlin, W. I. & S. L. Miller 1992. An Epistemological Use of Nonstandard Analysis to Answer
Zenos Objections Against Motion. Synthese 92: 37184.
McTaggart, J. E. 1908. The Unreality of Time. Mind 17: 45774.
McTaggart, J. E. 1927. The Nature of Existence, vol. 2. Cambridge: Cambridge University Press.
Meiland, J. W. 1974. A Two- Dimensional Passage Model of Time for Time Travel. Philosophical
Studies 26: 15373.
Mellor, D. H. 1981. Real Time. Cambridge: Cambridge University Press.
Mellor, D. H. 1994. Thank Goodness Thats Over. In The New Theory of Time, L. N. Oaklander &
Q. Smith (eds), 293304. New York: St Martins Press.
Mellor, D. H. 1998. Real Time II. London: Routledge.
Mellor, D. H. 2004. For Facts as Causes and Effects. In Causation and Counterfactuals, J. Collins,
N. Hall & L. A. Paul (eds), 30923. Cambridge, MA: MIT Press.
Merricks, T. 2007. Truth and Ontology. Oxford: Oxford University Press.
Meyer, U. 2003. The Metaphysics of Velocity. Philosophical Studies 112(1): 93102.
Minkowski, H. [1908] 1964. Space and Time. In Problems of Space and Time, J. J. C. Smart (ed.),
297312. London: Macmillan.
Misner, C. W., K. S. Thorne & A. Wheeler 1970. Gravitation. San Francisco, CA: Freeman.
Moffat J. W. & V. T. Toth 2007. Modified Gravity: Cosmology without Dark Matter of Einsteins
Cosmological Constant. http://arXiv.org/abs/0710.0364 (accessed March 2010).
Moore, A. W. 2001. The Infinite, 2nd edn. London: Routledge.
Moore, M. 2007a. On the Completeness of the Real Line. Critica 39(117): 6186.
Moore, M. 2007b. The Genesis of the Peircean Continuum. Transactions of the Charles S. Peirce
Society 43(3): 42569.
Mortensen, C. & G. Nerlich 1978. Physical Topology. Journal of Philosophical Logic 7: 20933.
Mundy, B. 1983. Relational Theories of Euclidean Space and Minkowski Space- Time. British Journal
for the Philosophy of Science 50: 20526.
Nahin, P. J. 1993. Time Machines. New York: American Institute of Physics.
Nahin, P. J. 2001. Time Machines, 2nd edn. New York: Springer- Verlag.
Nerlich, G. 1979. What can Geometry Explain? British Journal for the Philosophy of Science 30: 6983.
Nerlich, G. 1991. How Euclidean Geometry has Misled Metaphysics. Journal of Philosophy 88: 16989.
Nerlich, G. 1994a. What Spacetime Explains. Cambridge: Cambridge University Press.
Nerlich, G. 1994b. The Shape of Space. Cambridge: Cambridge University Press.
Nerlich, G. 2002. Review of Time and Space, Barry Dainton. Australasian Journal of Philosophy
80(4): 53031.
Nerlich, G. 2005. Can Parts of Space Move? On Paragraph Six of Newtons Scholium. Erkenntnis
62: 11935.
Newton, I. [1729] 1962. Mathematical Principles of Natural Philosophy, A. Motte & F. Cajori (trans.).
Berkeley, CA: University of California Press.
Newton- Smith, W. H. 1980. The Structure of Time. London: Routledge & Kegan Paul.
Newton- Smith, W. H. 1986. Space, Time and Space- Time: A Philosophers View. See Flood &
Lockwood (1986), 2235.
Nicolson, I. 2007. Dark Side of the Universe: Dark Matter, Dark Energy, and the Fate of the Cosmos.
Baltimore, MD: Johns Hopkins University Press.
Nicolson, I. 2008. Dark Side of the Universe. Bristol: Canopus Publishing.
Nolan, D. 2004. Classes, Worlds and Hypergunk. The Monist 87(3): 321.
Noonan, H. 1991. Indeterminate Identity, Contingent Identity and Abelardian Predicates. Philo-
sophical Quarterly 41: 18393.
Noonan, H. 1993. Constitution is Identity. Mind 102: 13346.
North, J. 2008. Two Views of Time Reversal. Philosophy of Science 75: 20133.
North, J. (forthcoming). Time in Thermodynamics. See Callender (forthcoming).
Norton, J. 1989. The Hole Argument. In Proceedings of the 1988 Biennial Meeting of the Philosophy
of Science Assocation, A. Fine & J. Leplin (eds), vol. 2. East Lansing, MI: Philosophy of Science
Association. Reprinted in Butterfield et al. (1996), 28593.
BIBLIOGRPAHY 451
Norton, J. 1992. Philosophy of Space and Time. In Introduction to the Philosophy of Science, M.
Salmon (ed.), 179232. New Jersey: Prentice Hall. Reprinted in Butterfield et al. (1996), 356.
Norton, J. 2008a. Why Constructive Relativity Fails. British Journal for the Philosophy of Science
59: 82134.
Norton, J. 2008b. The Hole Argument. Stanford Encyclopedia of Philosophy. http://plato.stanford.
edu/entries/spacetime-holearg/ (accessed March 2010).
Oaklander, L. N. 2003. Review: Time and Space. Mind 112: 50913.
Oaklander, L. N. 2004. Presentism, Ontology and Temporal Experience. In his The Ontology of
Time, 83101. Amherst, NY: Prometheus Books.
Oaklander, L. N. 2009. McTaggarts Paradox and Crisps Presentism. Philosophia (11 September).
Oaklander, L. N. & Q. Smith (eds) 1994. The New Theory of Time. New York: St Martins Press.
Oppy, G. 2006. Philosophical Perspectives on Infinity. Oxford: Oxford University Press.
Owens, D. 1992. Causes and Coincidences. Cambridge: Cambridge University Press.
Owen, G. E. L. 1957. Zeno and the Mathematicians. Proceedings of the Aristotelian Society 58:
199222.
Parfit, D. 1984. Reasons and Persons. Oxford: Oxford University Press.
Peirce, C. 1935. The Collected Papers of Charles Sanders Peirce. Cambridge, MA: Harvard University
Press.
Penrose, R. 1989. The Emperors New Mind. Oxford: Oxford University Press.
Penrose, R. 2004. The Road to Reality. London: Jonathan Cape.
Poincar, H. 1952. Science and Hypothesis. New York: Dover.
Pooley, O. 2003. Handedness, Parity Violation, and the Reality of Space. In Symmetries in Phys-
ics: Philosophical Reflections, K. Brading & E. Castellani (eds), 25080. Cambridge: Cambridge
University Press.
Pooley, O. 2006. Points, Particles and Structural Realism. In The Structural Foundations of Quantum
Gravity, D. Rickles, F. French & J. Saatsi (eds). Oxford: Oxford University Press.
Price, H. 1984. The Philosophy and Physics of Affecting the Past. Synthese 16: 299323.
Price, H. 1996. Times Arrow and Archimedes Point. Oxford: Oxford University Press.
Price, H. 2002a. Boltzmanns Time Bomb. British Journal for the Philosophy of Science 53: 83119.
Price, H. 2002b. Burburys Last Case: The Mystery of the Entropic Arrow. See Callender (2002),
1956.
Price, H. forthcoming. The Flow of Time. See Callender (forthcoming). Preprint available at http://
philsci-archive.pitt.edu/archive/00004829/ (accessed June 2010).
Prior, A. N. 1959. Thank Goodness Thats Over. Philosophy 34: 1217.
Putnam, H. 1967. Time and Physical Geometry. Journal of Philosophy 64: 24047.
Putnam, H. 1979a. What is Mathematical Truth?. In his Philosophical Papers Vol. 1: Mathematics,
Matter and Method. Cambridge: Cambridge University Press.
Putnam, H. 1979b. A Philosopher Looks at Quantum Mechanics. In his Philosophical
Papers Vol. 1: Mathematics, Matter and Method, 2nd edn. Cambridge: Cambridge University
Press.
Putnam, H. 1994. Peirces Continuum. In Peirce and Contemporary Thought: Philosophical Inquiries,
K. Ketner (ed.). New York: Fordham University Press.
Quinton, A. 1962. Spaces and Times. Philosophy 37: 13047. Reprinted in Le Poidevin & MacBeath
(1993), 20320.
Raine, D. J. 1981. Machs Principle and Space- Time Structure. Reports on Progress in Physics 44:
115195.
Randall, L. 2005. Warped Passages. London: Allen Lane.
Ray, C. 1991. Time, Space and Philosophy. London: Routledge.
Reichenbach, H. 1956. The Direction of Time. Berkeley, CA: University of California Press.
Reichenbach, H. 1958. The Philosophy of Space and Time. London: Dover.
Rickles, D. 2008. Symmetry, Structure and Spacetime. Elsevier: Amsterdam.
Rietdijk, C. W. 1966. A Rigorous Proof of Determinism Derived from the Special Theory of Relativ-
ity. Philosophy of Science 33: 3414.
Roeper, P. 1997. Region-Based Topology. Journal of Philosophical Logic 26(3): 251309.
Rovelli, C. 1999. Quantum Spacetime: What do we Know? http://arXiv.org/abs/gr-qc/9903045
(accessed March 2010). Reprinted in Callender & Huggett (2001), 10122.
452 BIBLIOGRAPHY
Rucker, R. 1982. Infinity and the Mind: The Science and Philosophy of the Infinite. Sussex: Harvester.
Rucker, R. 1984. The Fourth Dimension. Boston, MA: Houghton Mifflin.
Russell, B. 1915. On the Experience of Time. Monist 25: 21233.
Russell, B. 1917. Mathematics and Metaphysicians. In his Mysticism and Logic. London: George
Allen & Unwin.
Russell, B. [1929] 2001. The Problem of Infinity Considered Historically. See Salmon (2001). Origi-
nally published in Our Knowledge of the External World, 18298 (New York: Norton, 1929).
Rynasiewicz, R. 1994. The Lessons of the Hole Argument. British Journal for the Philosophy of
Science 45: 40736.
Rynasiewicz, R. 1995. By their Properties, Causes and Effects: Newtons Scholium on Time, Space,
Place and Motion. Studies in History and Philosophy of Science 26:13353, 295321.
Rynasiewicz, R. 1996. Absolute Versus Relational Space- Time: An Outmoded Debate? Journal of
Philosophy 93: 279306.
Rynasiewicz, R. 2000. On the Distinction between Absolute and Relative Motion. Philosophy of
Science 67: 7193.
Sainsbury, R. M. 2009. Paradoxes. Cambridge: Cambridge University Press.
Salmon, W. C. 1975. Space, Time and Motion. Encino, CA: Dickinson Publishing.
Salmon, W. C. 2001. Zenos Paradoxes. Indianapolis, IN: Hackett.
Sanford Russell, J. 2008. The Structure of Gunk: Adventures in the Ontology of Space. In Oxford
Studies in Metaphysics, D. Zimmerman (ed.), vol. 4, 24874. Oxford: Oxford University Press.
Saunders, S. 2002. How Relativity Contradicts Presentism. See Callender (2002), 27792.
Savitt, S. F. 1994. The Replacement of Time. Australasian Journal of Philosophy 72: 46374.
Savitt, S. F. (ed.) 1995. Times Arrow Today. Cambridge: Cambridge University Press.
Savitt, S. F. 1996. The Direction of Time. British Journal for the Philosophy of Science 47: 34770.
Savitt, S. F. 1998. Theres No Time like the Present (in Minkowski Spacetime). In Proceedings of the
1998 Biennial Meeting of the Philosophy of Science Association, 67(3), part II, supplement.
Savitt, S. F. 2006a. Being and Becoming in Modern Physics. Stanford Encyclopedia of Philosophy,
http://plato.stanford.edu/entries/spacetime-bebecome/ (accessed March 2010).
Savitt, S. F. 2006b. Presentism and Eternalism in Perspective. In The Ontology of Spacetime, D. Dieks
(ed.), 11127. Amsterdam: Elsevier.
Schilpp, P. A. (ed.) 1959. Philosophy of C. D. Broad. New York: Tudor Publishing Company.
Schlesinger, G. N. 1991. E pur si muove. The Philosophical Quarterly 41: 42741.
Selby, D. 1982. Two Conceptions of Vacuum. Phronesis 27(2): 17593.
Sherry, D. M. 1988. Zenos Metrical Paradox Revisited. Philosophy of Science 55: 5873.
Shoemaker, S. 1969. Time without Change. Journal of Philosophy 66: 36381. Reprinted in Le
Poidevin & MacBeath (1993), 6379.
Sider, T. 1999. Presentism and Ontological Commitment. Journal of Philosophy 96(7): 32547.
Sider, T. 2001. Four-Dimensionalism. Oxford: Oxford University Press.
Sider, T. 2002. Time Travel, Coincidences and Counterfactuals. Phlosophical Studies 110: 11338.
Sider, T. 2003. Reductive Theories of Modality. In The Oxford Handbook of Metaphysics, M. J. Loux
& D. W. Zimmerman (eds), 180208. Oxford: Oxford University Press.
Sider, T. 2005. Travelling in A- and B-time. The Monist 88(3): 32935.
Sider, T. 2006. Quantifiers and Temporal Ontology. Mind 115(457): 7597.
Sklar, L. 1974. Incongruous Counterparts, Intrinsic Features and the Substantivality of Space. Jour-
nal of Philosophy 71: 27790.
Sklar, L. 1977. Space, Time and Spacetime. Berkeley, CA: University of California Press.
Sklar, L. 1981. Up and Down, Left and Right, Past and Future. Nos 15: 11129. Reprinted in Le
Poidevin & MacBeath (1993), 99116.
Sklar, L. 1985. Philosophy and Spacetime Physics. Berkeley, CA: University of California Press.
Sklar, L. 1992. Philosophy of Physics. Oxford: Oxford University Press.
Skow, B. 2006. Review of Brown, Physical Relativity: Space-time Structure from a Dynamical
Perspective. Notre Dame Philosophical Reviews, 5 November 2006, http://ndpr.nd.edu/review.
cfm?id=6603 (accessed March 2010).
Skow, B. 2007. Sklars Maneuver. British Journal for the Philosophy of Science 58: 77786.
Skow, B. (forthcoming) Why Does Time Pass?. http://web.mit.edu/bskow/www/research/whydoes-
timepass.pdf (accessed March 2010). Forthcoming in Nos.
BIBLIOGRPAHY 453
Slater, M. 2005. The Necessity of Time Travel (On Pain of Indeterminacy). The Monist 88(3): 3629.
Smart, J. C. C. 1980. Time and Becoming. In Time and Cause, P. van Inwagen (ed.), 315. Dordrecht:
Reidel.
Smart, J. J. C. 1964. Problems of Space and Time. London: Macmillan.
Smeenk, C. & C. Wthrich (forthcoming) Time Travel and Time Machines. See Callender (forth-
coming).
Smith, Q. 1993. Language and Time. New York: Oxford University Press.
Smith, Q. 1998. Absolute Simultaneity and the Infinity of Time. See Le Poidevin (1998a), 13584.
Smith, Q. 2002. Time and Degrees of Existence: A Theory of Degree Presentism. See Callender
(2002), 11936.
Smolin, L. 2000a. Three Roads to Quantum Gravity. London: Weidenfeld & Nicolson.
Smolin, L. 2000b. The Present Moment in Quantum Cosmology: Challenges to the Arguments for the
Elimination of Time. In Time and the Instant, R. Durie (ed.), 11243. Manchester: Clinamen Press.
Smolin, L. 2006. The Trouble with Physics. London: Allen Lane.
Sorabji, R. 1983. Time, Creation and the Continuum. London: Duckworth.
Sorenson, R. 2003. A Brief History of the Paradox. Oxford: Oxford University Press.
Sprigge, T. 1983. The Vindication of Absolute Idealism. Edinburgh: Edinburgh University Press.
Stachel, J. 1993. The Meaning of General Covariance. In Philosophical Problems of the Internal
and External Worlds, J. Earman, A. J. Janis, J. Massey & N. Rescher (eds), 12960. Pittsburgh, PA:
University of Pittsburgh Press.
Stein, H. 1968. On EinsteinMinkowski Space- Time. Journal of Philosophy 65: 523.
Stein, H. 1970. Newtonian Spacetime. Texas Quarterly 10: 174200. Reprinted in Butterfield et
al. (1996), 79105.
Stein, H. 1991. On Relativity Theory and the Openness of the Future. Philosophy of Science 58:
14767. Reprinted in Butterfield et al. (1996), 23959.
Steinhardt, P. & J. P. Ostriker 2001. The Quintessential Universe. Scientific American 284(1): 4653.
Steinhardt, P. & N. Turok 2007. Endless Universe. London: Orion.
Stewart, I. 2001. Flatterland. London: Macmillan.
Strawson, G. 2006. Realistic Monism. Journal of Consciousness Studies 13(1112): 331.
Susskind, L. 2006. The Cosmic Landscape: String Theory and the Illusion of Intelligent Design. New
York: Little, Brown and Company.
Tarski, A. 1956. Foundations of the Geometry of Solids. In Logic, Semantics, and Metamathematics,
J. H. Woodger (ed.), 249. Oxford: Clarendon Press.
Teller, P. 1991. Substance, Relations, and Arguments about the Nature of Space- Time. The Philo-
sophical Review 100: 35397.
Thayer, H. S. (ed.) 1974. Newtons Philosophy of Nature: Selections from his Writings. New York:
Macmillan, Hafner Press.
Thorne, K. 1994. Black Holes and Time Warps: Einsteins Outrageous Legacy. London: Picador.
Tiles, M. 1989. The Philosophy of Set Theory. Oxford: Blackwell.
Tipler, F. 1974. Rotating Cylinders and the Possibility of Global Causality Violation. Physical Review
D9: 22036.
Tooley, M. 1988. In Defense of the Existence of States of Motion. Philosophical Topics 16: 22554.
Tooley, M. 1997. Time, Tense and Causation. Oxford: Oxford University Press.
Torretti R. 1998. Space. In Routledge Encylopedia of Philosophy, CD- ROM. London: Routledge.
Torretti, R. 1999a. On Relativity, Time Reckoning and the Topology of Time Series. See Butterfield
(1999), 6582.
Torretti, R. 1999b. The Philosophy of Physics. Cambridge: Cambridge University Press.
Van Bendegem, J. P. 1987. Zenos Paradoxes and the Tile Argument. Philosophy of Science 54(2):
295302.
Van Bendegem, J. P. 2010. Finitism in Geometry. Stanford Encyclopedia of Philosophy. http://plato.
stanford.edu/entries/geometry-finitism/ (accessed March 2010).
Van Cleve, J. 1998. Incongruent Counterparts and Higher Dimensions. See Van Inwagen & Zim-
merman (1998), 11120.
Van Cleve, J. & R. Frederick (eds) 1991. The Philosophy of Left and Right. Dordrecht: Kluwer.
Van Fraassen, B. C. 1969. On Masseys Explication of Grnbaums Conception of Metric. Philoso-
phy of Science 36(4): 34653.
454 BIBLIOGRAPHY
Van Fraassen, B. 1980. The Scientific Image. Oxford: Oxford University Press.
Van Inwagen, P. 2010. Changing the Past. In Oxford Studies in Metaphysics, Volume 5, Dean Zim-
merman (ed.), 328. Oxford: Oxford University Press.
Van Inwagen, P. & D. W. Zimmerman (eds) 1998. Metaphysics: The Big Questions. Oxford: Blackwell.
Vlastos, G. 1967. Zeno of Elea. In The Encyclopedia of Philosophy, P. Edwards (ed.), vol. 8, 36979.
New York: Macmillan.
Wallace, D. F. 2005. Everything and More: A Compact History of . London: Phoenix.
Wang, H. 1974. From Mathematics to Philosophy. New York: Humanities Press.
Warren, J. 2006. Epicureans and the Present Past. Phronesis 51(4): 36287.
Weyl, H. 1949. Philosophy of Mathematics and Natural Science. Princeton, NJ: Princeton University
Press.
Weyl, H. 1998. On the New Foundational Crisis in Mathematics. In From Brouwer to Hilbert: The
Debate on the Foundations of Mathematics in the 1920s, P. Mancosu (ed.), 3779. Oxford: Oxford
University Press.
White, M. 1999. Incommensurables and Incomparables: On the Conceptual Status and Philosophical
Use of Hyperreal Numbers. Notre Dame Journal of Formal Logic 40(3): 42046.
Whitehead, A. N. 1919. An Enquiry Concerning the Principles of Natural Knowledge. Cambridge:
Cambridge University Press.
Whitehead, A. N. 1920. The Concept of Nature. Cambridge: Cambridge University Press.
Whitehead, A. N. 1925. Process and Reality. Cambridge: Cambridge University Press.
Whitehead, A. N. 1929. Process and Reality. New York: Macmillan.
Whitehead, A. N. 1947. Essays in Science and Philosophy. New York: Philosophical Library.
Will, C. M. 1988. Was Einstein Right? Oxford: Oxford University Press.
Williams, D. C. 1951. The Myth of Passage. Journal of Philosophy 48: 45772.
Williamson, T. 1999. Existence and Contingency. Proceedings of the Aristotelian Society, Sup. 73:
181203.
Winnie, J. A. 1970. Special Relativity Without One- way Velocity Assumption. Philosophy of Science
37: 8199, 22338.
Witten, E. 1996. Reflections on the Fate of Spacetime. Physics Today 96(4), 2430. Reprinted in
Callender & Huggett (2001), 12537.
Woit, P. 2006. Not Even Wrong. London: Jonathan Cape.
Wolfram, S. 2002. A New Kind of Science. Champaign, IL: Wolfram Media.
Yourgrau, P. 1991. The Disappearance of Time: Kurt Gdel and the Idealistic Tradition in Philosophy.
Cambridge: Cambridge University Press.
Zalamea, F. 2003. Peirces Logic of Continuity: Existential Graphs and Non-Cantorean Continuum.
The Review of Modern Logic 9: 29, 115162.
Zangari, M. 1994. Zeno, Zero and Indeterminate Forms: Instants in the Logic of Motion. Australa-
sian Journal of Philosophy 72(2): 187204.
Zelicovici, D. 1989. Temporal Becoming Minus the Moving- Now. Nos 23: 50524.
Zibin, J. P., A. Moss & D. Scott 2008. Can We Avoid Dark Energy? Physical Review Letters 101(25).
Available at http://arXiv4.library.cornell.edu/abs/0809.3761 (accessed March 2010).
Zimmerman, D. W. 1996a. Could Extended Objects Be Made Out of Simple Parts? An Argument for
Atomless Gunk. Philosophy and Phenomenological Research 56(1): 129.
Zimmerman, D. W. 1996b. Indivisible Parts and Extended Objects: Some Philosophical Episodes from
Topologys Prehistory. Monist 79(1): 14880.
Zimmerman, D. W. 1998. Presentism and Temporary Intrinsics. See Van Inwagen & Zimmerman
(1998), 20619.
Zimmerman, D. W. 2004. Oxford Studies in Metaphysics, vol. I. Oxford: Oxford University Press.
Zimmerman, D. W. 2005. The A-Theory of Time, the B-Theory of Time, and Taking Tense Seri-
ously. Dialectica 59(4): 40157.
Zimmerman, D. W. 2008. The Privileged Present : Defending an A-Theory of Time. In Contempo-
rary Debates in Metaphysics, T. Sider, J. Hawthorne & D. W. Zimmerman (eds), 21125. Oxford:
Blackwell.
Index
A- and B-statements
problems with translations between 32
a priori/a posteriori 428
Abbott, E. A. 414
A-beliefs, practical importance of 35
absolute acceleration 182, 18896, 199,
200204, 207209, 427
absolute annihilation 76, 332
absolute becoming 68, 757, 3324, 384, 427
narrow creeks of 334
absolute motion 161, 177, 1812, 184, 2012,
31617, 322, 427
absolute past v. absolute future v. absolute
elsewhere (in STR) 324
absolute rest 165, 177, 179, 182, 196, 202,
206, 315, 322, 339, 380
absolute simultaneity 200, 212, 337, 339, 342,
387, 428
absolute velocity 162, 1767, 1835, 189,
1946, 200205, 212, 315, 322, 327
change 199
acceleration
as primitive monadic property 192, 202
real v. apparent 185
in spacetime 196
Achilles 268, 272, 275, 295, 418, 419
action-at-a-distance 158, 209, 340, 344, 347,
388, 427
actuality simpliciter 410
aether 31517, 323, 338, 379, 380
A-facts 11, 14, 312, 345
affine connection 207
affine geometry 427
affine structure 197, 322, 343, 376, 381, 418
A-framework 11, 35, 635
Albert, D. Z. 422
Alexander, H. G. 154, 166
Alper, Joseph 420
analyticsynthetic 427
angular momentum 190
anisotropic 4, 370
anti-realism 231, 246, 262, 266, 379, 427
A-properties 19
as intrinsic and transitory 20
v. B-properties 428
Archimedes 271
Aristotle 6, 48, 146, 168, 170, 174, 232, 268,
2727, 2945, 306, 311, 414, 41819
outline of his physics 168
Arntzenius, Frank 292, 31112, 413
arrow of time 6, 428
B-theory of 44
the causal account 51
content-asymmetries 46
and entropy 47
and the Growing Block model 75
Horwichs account 56
McCalls account 73
phenomenal v. memory/knowledge v. causal v.
world 116
Arrow paradox 289
A-series 1011, 1314, 1517, 26, 63, 407
A-statements 11, 16, 314
asymmetry of overdetermination 57, 59
at-at (theory of motion) 290
A-theorists/B-theorists 12, 428
A-theory/B-theory 428
Atkins, Peter 47
A-truths, token-reflexive and date accounts 33
autoinfanticide 128
awareness-content model 108
rejection of 114
awareness-overlap model 108, 112
repeated-contents problem 109
axiom of choice 420
backward causation 52, 69, 76, 1316, 142,
428
Dummetts loophole 134
Baez, John 417, 423
Balashov, Yuri 82
Barbour, Julian 867, 164, 167, 170, 175,
41516
bare powers 2478, 258
Barrow, Isaac 414
basic phenomenological data 104
Beckenstein, Jacob 365
Bell, John L. 42021
Bell, John S. 340
Bells inequality 42930
Belot, Gordon 41, 43, 411, 416
Benacerraf, P. 274
Bergson, Henri 420
Berkeley, G. 157, 191, 307
Besso, M. 408
B-facts 11, 14, 325, 38
456 INDEX
B-framework 11
bias towards the future 368, 408
bifurcated supertasks 419
big bang 534, 61, 84, 86, 137, 148, 3557,
371, 382, 401, 424
low entropy event 409
standard models of 357
big crunch 534, 357, 401
Bigelow, John 235, 81, 912, 121, 293, 419
bilking 131-6
black holes 4, 122, 3525, 3689, 398, 423,
425
Black, Max 415, 419
Block universe 41, 43, 55, 74, 78, 128, 142,
144, 385, 410
Block view 79, 1112, 27, 378, 41, 69, 81,
88, 124, 130, 194, 314, 3835, 407, 428,
430, 4323
block-arrow 767
Bohm, David 34041, 422
Boltzmann, Ludwig 48, 409
cosmological speculation and time-reversal
409
Bolyai, J. 220, 223
Bolzano, B. 274
boundary conditions 353, 384
Bourne, Craig 935, 410
Braddon-Mitchell, David 410
branching-time model 73
branes 399401, 403, 425
braneworld (cosmological) models 399
Bricker, Phillip 15860
Bridger, Mark 420
Brighouse, Carolyn 424
Broad, C. D. 17, 18, 65, 689, 767, 10810,
311, 328, 412
Brown, Harvey 416
Bruno, Giordano 414
B-series 11, 1314, 1921, 256, 64, 407
B-statements 32
B-theory, in contemporary guise 27
bucket argument 186
bulk (space) 399
Butterfield, Jeremy 411, 419, 424
Button, Tim 410
CalabiYau spaces 3968
calculus 164, 271, 290, 292, 307, 415, 421,
429, 430
Callender, Craig 417, 425
Cameron, Ross 411
Campanella, T. 414
canonical quantum gravity programme 87,
387, 406
Cantor, G. 26870, 27780, 2848, 301, 306,
419
cardinal numbers 279
Carnap, Rudolf 420
Carroll, John W. 419
Cartan, E. 194
Cauchy, A.-L. 269, 274, 292, 307
causal asymmetry, the projectivist view 53
causal propagation 344
causal-arrow 55, 78
causation
asymmetrical relationship 51
backward and forward coexisting 413
necessary and sufficient conditions 52
CDMs (classical dynamic models of time) 328,
3334, 337
cellular automata 295
centre of gravity 429
centre of mass 192
centrifugal force 190
C-field 423
Chalmers, David 422
Chandrasekhar limit 354
change
B-theories of 38
and causation 39
as requiring the A-series 14
as unreal (quantum gravity theory) 41
chronogeometry 347
chronon 295
CIS-propositions 93
Clark, Stephen 413
Clarke, B. L. 311
Clarke, Samuel 175, 180
ClarkeLeibniz correspondence 175
classical physics 317, 374, 429
Clifton, R. 422
Clifton, Timothy 365
clogging of consciousness 105
closed time-like curves 3823, 413
COBE 355, 424
co-consciousness 1067, 11419
over time, failure of transitivity 114
coexistence 98100
coincidences
defined 58
and time travel 129
cold dark matter 360
collinearity 2078, 211
communication and time-reversal 54
compactification 391, 397
compatibilism 3323
compensatory theory 317, 338
Compound Presentism 95102, 120, 139, 412
compound temporal predicates 16
concordance model 363
Cone Presentism 335, 422
confinement doctrine 29
congruence 430
connection (geometry) 197, 203
connection problem 152, 155, 403
consciousness 429
of time 95
content thesis 247
content-asymmetries
causal account 51
need for explanation 46
a partial list of 45
contingent 429
continuants 39, 431
INDEX 457
continuous rigid motions (crms) 240
continuous transformations 225, 243
continuum 68, 412, 195, 206, 257, 26970,
2767, 28089, 300306, 30912, 330,
376, 377, 41921, 429
and denseness 301
conventionalism 228, 231, 2456, 338, 343,
422
convergent v. divergent series 274
Conway, John 295
coordinate independence 225
coordinate system 182, 353, 372, 429
Copernican principle 355
Copernican revolution 363
Copernicus, Nicolaus 1689
cosmic time 3812, 424, 430
cosmic void 365
cosmological constant 3557, 3623, 4234
cosmology 4, 169, 354, 356, 373, 414, 423
co-streamality, defined 115
counterfactual 430
asymmetry 408
covariance 225, 3723, 381, 430
Cramer, John 413
Crisp, Thomas 90, 923, 411
critical value (O) 357
C-series 17
curvature tensor 2245, 417, 422
curved space 5, 213, 217, 225, 233, 236, 368,
417, 430
in the absence of an embedding space 217
made manifest 235
straight lines in 215
tests for 214
curved spacetime in the absence of matter 371
cyclic model of the cosmos 401
dark energy 3615, 401, 4234
dark matter 35860, 3634, 424
Davies, Paul 103, 393, 421, 425
De Broglie, Louis 422
De Sitter, Wihelm 357
death 16, 36, 42, 58, 65, 83, 175, 408
Dedekind, R. 26970, 280
deflationary dynamism 67
denseness (as a property of continua) 269
Descartes, Ren 13, 146, 164, 167, 1725,
182, 184, 187, 219, 380, 415
the Cartesian cosmos 172
detachment thesis 236
determinism 374, 376, 407, 424, 430
and the at-at theory of motion
and the hole argument 373
temporal v. nomological 407
Deutsch, David 127, 138, 413
deviant laws 252
Dewdney, A. K. 414
diagonal proof 287
diagonalization 279
Dichotomy paradox 272
diffeomorphism 3735, 425
active v. passive 424
differential effects 230
differential geometry 159, 224, 352, 417, 430
dimension (concept of) 430
direct awareness, its temporal span 30
direction of time
and content-asymmetries 44
defined in causal terms 51
Dirichlet, L. 274
discrete space (and spacetime) 295301
disjunctive tensed predication 65
distance, Gaussian v. intrinsic 159, 262, 265
distance relations 157, 1835, 197, 21012,
239, 257, 2623
intrinsically directed 211
Dolev, Yuval 412
Dorato, Mauro 334, 422, 425
doubly dynamic world 78
Dowe, Phil 413
Dummett, Michael 131, 134, 304, 312, 412,
421
Dynamic Presentism 8794, 97, 118
dynamic time 25, 26, 62, 430
and Presentism 84
initial characterization 7
dynamics 161, 185, 191, 201, 2079, 239,
322, 343, 369, 3889, 415, 430
E = mc
2
421
Earman, John 41, 43, 49, 154, 188, 190, 192,
194, 3737, 379, 41114, 416, 419, 4235
Eddington, A. S. 352
Egan, Greg 417
Eiffel Tower 10
Einstein, Albert 4, 81, 86, 103, 136, 1423,
146, 1634, 191, 194, 213, 225, 231, 239,
313, 31521, 330, 336, 33840, 34372,
38081, 390, 4213
on the problem of the now 412
train argument 321
electromagnetic spectrum 430
electromagnetism 212, 315, 339, 343, 380,
39092, 397, 416, 43031
electroweak theory 391
emergent entities 43
empirical 431
enantiomorphs 24042
entropy 6, 4650, 53, 55, 61, 76, 409, 431
epistemology 431
EPR (EinsteinPodolskyRosen) 136, 33940,
422
Epsilon Eridani 143, 329
equilibrium state 47
equivalence principle 3457
equivalence relation, 428, 441
ersatz B-series 934, 97, 100, 118, 139, 411
Escher, M. C. 221
essential v. non-essential 431
essentially extended continuum 310
eternalism 7
Euclid 21820, 278, 417
axioms and postulates 219
Elements 21819
458 INDEX
Euclidean geometry 153, 21920, 223, 228,
23031, 269, 343, 368
Euclidean plane 271
active nature of 237
symmetries 198
event
broad sense of the term 20
defined 431
existence as frame-relative 331
existence claims, restricted and unrestricted
88
existence simpliciter 89
experience of passage, B-theory of 30
experience of time or temporality 47, 95,
103, 115, 385
the BroadHusserl theory 109
memory- and belief-based accounts 105
pulse theory 107
extended presents 120
extensive abstraction 311
factual inheritance principle 100
Faraday, Michael 315
fatalism 431
FeynmanWheeler absorber theory 142, 413,
431
field, concept (in physics) 432
Fine, Kit 98, 412
Flatland 1478, 159, 21317, 221, 229, 234,
2412, 250, 355, 3879, 403, 414, 417
and concealed higher dimensions 387
flat v. curved 213
flatness problem (in big bang cosmology) 434
Flew, Anthony 131
Fludd, R. 414
Fluid World 151
foliations 381
fork asymmetry 567, 5961, 409
Forrest, Peter 2989, 31112, 40910
Foster, John 112, 24566, 293, 379, 412, 418
Fourier, J. 274
fourth spatial dimension 147, 388, 390
frame of reference 182, 432
Frances, Bryan 90
free-fall 171, 3445, 3489
Friedman, Michael 190, 283, 416, 422
Friedmann, Aleksandr 356, 365, 371, 381,
423
FriedmannRobertsonWalker models (of GTR)
357
frozen time, problem of 87, 341
function 432
fundamental domain 227
Galilean frames 177, 182
Galilean invariance 170, 177
Galilean relativity 432
Galileo, G. 164, 167, 16976, 277, 344, 415
on projectile motion 171
Galison, Peter 421
Gallois, Andr 37, 408, 418
game of life 295
Gassendi, P. 414
Gaul, M. 426
Gauss, C. F. 159, 220, 222, 224, 228, 283
Gaussian curvature 2223
Gaussian distance 16061, 265
general covariance 372, 375
general theory of relativity (GTR), 4, 86, 142,
213, 225, 231, 239, 31314, 336, 341, 343,
347, 3516, 36593, 398, 404, 411, 4225
Gdels solution 371, 382
Kerrs solution 371
Oszvth and Schcking solution 371
geodesic 225, 235, 2389, 264, 2656, 348,
381
influenced by mass 348
intrinsic v. material 264
geometrical concepts, physical meaning of 228
geometrical essentialism 262, 418
geometrical line 270
geometrical monism 262, 266, 379
geometrical pluralism 262, 265, 379
geometry
intrinsic v. functional 254
pure v. applied 255
Georgi, H. 392
Gerla, G. 311
Geroch, Robert 199200, 416
g-function 224, 417
ghost images 366
Glashow, S. L. 392
global wave function 422
gluons 392
Gdel, Kurt 301, 304
Gdel universe 371, 413
Gdels modal argument 3845, 425
Godfrey-Smith, W. 422
Gold universe 534, 61, 778, 11617, 141,
432
Gott, R. 366
grand unification 392
grand unified theory (GUT) 392
gravitational energy 369, 371
gravitational field 122, 3446, 372
gravitational waves 3689, 423
graviton 392, 395
gravity 41, 48, 867, 170, 177, 187, 204, 208,
209, 231, 238, 34152, 355, 377, 380,
39092, 4224, 432
Gray, Lawrence 420
Greene, Brian 393, 395, 398, 400, 402, 422,
425
Grey, W. 138
Grosseteste, R. 415
Growing Block model 25, 716, 79, 82, 330,
331, 410, 432
and Gold universes 77
growth v. shrinkage 75
positive v. negative instability 78
the problem of the past 79, 410
size of increments 69
Tooleys causal account 77
Grnbaum, A. 162, 228, 2813, 300303, 419
INDEX 459
gunk 31112, 421
Guth, Alan 358
haecceity 92
Hamiltonian 411
Harrison, Craig 421
Harrison, J. 413
Harshland 14950
Hawking, Stephen 354, 413
Hawthorne, John 312
Heinlein, Robert 413
Hellman, G. 421
Heraclitus 8, 26
Hestevold, H. Scott 411
Hilbert, David 416, 418
Hiley, B. J. 422
Hinchliff, Mark 335, 410, 422
hodon 295
Hoefer, Carl, 374, 3778, 380
Hogarth, M. 422
hole argument 372, 3756
hole (in space) 152, 158
Hollis, M. 414
homomorphic 240, 243
Horwich, Paul 17, 55, 5661, 73, 1323, 162,
283, 409
Hospers, John 123
Hoyle, F. 423
Hubble, Edwin 352, 356
Hudson, Hud 305
Huggett, Nick 184, 199, 275, 416, 425
Hume, David 52, 57, 112
Huygens, Christiaan 164
hypercube 147
hypergunk 421
hyperplane 1957, 201, 341, 381, 382
hypersphere 147, 217
hypertorus 227
Iamblichus 407
ice skating 415
idealism 262, 433
identity 433
identity of indiscernibles 177, 260, 433
identity-conditions 433
iff 433
immanent phenomenal flow 104, 11517, 385,
412
implicit order 61
incongruent counterparts 24041
indexicals 433
inertia 433
inertial effects 204
inertial forces 182, 185, 1878, 192, 2034,
319, 416, 433
inertial frames of reference 182, 314, 433
inertial mass 318
inertial mass v. gravitational mass 433
inertial motion 1823, 186, 1967, 204,
2645, 351, 377, 38081, 398, 433
inertial paths 205
inertial structure 1967, 204, 36971, 377
inertial transformation 197
infinite
actual v. potential 277
denumerable v. indenumerable 278
infinitesimal 306, 308
inflationary model 358
inflationary universe 54, 409, 434
inscrutability thesis 246, 24951, 257, 259
instantaneous action-at-a-distance 136, 387
instantaneous connection 339
instantaneous velocity 29093
intentional 434
interval (spacetime) 323, 408, 434
intervals, closed v. open 285
intrinsic curvature 211, 222, 224, 228, 235,
263, 351, 434
intrinsic direction 21011
intrinsic nature 20, 153, 24651, 255, 2578,
434
intrinsic tensed properties 25, 70
is 434
isotropic 3, 354, 37071, 381, 423, 434
James, William 95
Jammer, Max 228
Janssen, Michel 82
KaluzaKlein theory 39093, 397
Kant, Immanuel 219, 24042, 421
glove/hand argument 240
Keller, Simon 92, 138, 413
Keynes, John Maynard 166
Kierland, Brian 411
kinematic shift 176, 1789, 189, 194, 234
kinematics 432, 435
Kirshner, Robert P. 362
Klein bottle 417
knowledge asymmetry 50, 55, 589
Kock, A. 420
Kuar, K. 87
ACDM model 363
Lachize-Rey, M. 424
Lange, Marc 293
Laplace, Pierre-Simon 292
Laraudogoitia, J. P. 419
law of inertia 182, 185
law of universal gravitation 177
Lawvere, F. W. 420
Le Poidevin, Robin 17, 97, 410
Le Verrier, Urbain 364
Lebesgue, H. 281
Lee, Geoffrey 413, 417
Leiber, Lillian 417, 422
Leibniz, G. W. 154, 157, 1647, 17581, 185,
18990, 194, 2336, 260, 271, 290, 315,
3745, 378, 405, 415
Leibniz alternatives 1768
Leibniz equivalence 3745
Lematre, Georges 3567
length contraction 318, 337, 347
Lewis, David 128, 1378, 311, 418
460 INDEX
light postulate 314, 321
light-cone 3234, 33036, 376, 381
light-like v. space-like v. time-like connections
324
lightspeed 318, 338, 369
one-way v. round-trip 338
Lineland 148
Lobachevsky, N. 220, 228
Locke, J. 247, 438
Lockwood, Michael 127, 138, 413
logical possibility 435
loop quantum gravity theory, 300, 4046, 426
loops, object v. informational 137
Lorentz, H. A. 198, 31718, 338, 341, 347,
380, 422
Lorenz invariance 341
Lowe, E. J. 637, 284, 293, 409, 419
Lucas, J. R. 339, 341, 382
Lucretian properties 100, 118, 139, 411
Lucretius 36
Lukasiewicz, Jan 83
Luminet, Jean-Pierre 3667, 424
Lumpl and Goliath 261
MacBeath, Murray 54, 413
Mach, Ernst 19092, 369, 370, 416
Machs Principle, strong and weak forms 369
MACHOs 360
MalamentHogarth spacetimes 419
Manders, K. 414
manifold substantivalism 3758
manifolds, multi-connected v. simply-connected
226
many-worlds 847, 118, 11920, 127, 410
Many-Worlds Presentism, empirical refutation
of 118
Markosian, Ned 923
mass-energy 60, 353, 36970, 375, 384, 398,
423
Maudlin, Tim 2057, 209, 212, 326, 341, 381,
413, 41516, 422
Maxwell, J. C. 271, 31516, 343, 390
McCall, Storrs 735, 410
McKinnon, Neil 411
McTaggart, J. M. E. 10, 1319, 257, 389,
41, 63, 65, 245, 407
McTaggarts paradox 17, 98
Lowes solution 65
measure theory 281, 285, 302, 304, 419
measurement problem 435, 437
Meiland, J. W. 126, 452
Mellor, Hugh 17, 2636, 3941, 51, 634,
105, 409, 411
A-paradox 34
memories and the experience of passage 31,
105
Merricks, Trenton 93
mesh world 403
metaphysics 435
indispensability of 5
meta-time 213, 25, 110
metric 435
metric field 372, 377, 379, 380
metric tensor 2245, 372
metrical amorphousness 283
metrical coefficients 224
metrical essentialism 3756, 378, 379
Michell, John 423
Michelson, Albert 31617
MichelsonMorley result 316
micro-chaos 60
microwave background radiation 424, 428,
435
Milgrom, Mordehai 364
Miller, S. L. 307
Minkowski spacetime 198, 200, 21213, 314,
3225, 328, 3326, 341, 343, 346, 370,
381, 435
Mirandola, G. P. 414
Mbius strip 243, 417
modal actualism 245
modal realism 245, 74
modality 435
modified gravity theories 365
Moffat, John 365
momentum density 353
monadic property 153, 1923, 435
MOND 3645, 424
monstrous delirium 408
Monton, Bradley 411
Moore, Adrian 419
Moore, Matthew 283, 42021
More, Thomas 414
Mortensen, Chris 419
motion
absolute 161
inoperative quality of 172, 176
relative 162
for substantivalists and relationists 161
motion as intrinsic (or not) 291
moving present 22, 27, 31, 103, 110, 119
Moving Spotlight model (of time) 18, 407
M-theory 399400, 403, 420, 425
multi-connected space 2268, 3659, 417,
424
multiple unconnected spaces 151
multiverse 127, 435
Mundy, B. 414, 416, 452
muons 318, 392
Nahin, Paul 129, 413
natural place (and natural motion) 168
n-dimensional space 148, 225
necessary (truth) 435
negative energy 383
Nelson, Michael 138, 413
neo-Newtonian spacetime 197, 200207,
212, 322
relationism 207
Neoplatonism 414
Nerlich, Graham 158, 160, 2337, 283, 378,
417, 420
neutral K-meson (or kaon) 443
neutrinos 1
new tenseless theory 32
INDEX 461
Newton, Isaac 6, 103, 116, 146, 1647, 170,
1728, 18195, 201, 2056, 219, 271, 290,
313, 318, 336, 34352, 369, 374, 37680,
398, 405, 41416, 422
the bucket 186
criticism of Descartes 184
De Gravitatone 146, 166, 182, 192, 379
the globes 187
law of universal gravitation 177, 192
Principia 164, 166, 167, 172, 1756, 188,
219, 415
on relative and absolute space and motion
165
Scholium 164, 167, 172, 174, 182, 185, 189,
41516
three laws of motion 170
univeral law of gravitation and STR 343
wider interests 166
Newtonian (spacetime) relationism 206
Newtonian gravity 344
Newton-Smith, W. H. 116, 149, 283, 414
Nicolson, Iain 424
nocturnal doubling 2334
Nolan, Daniel 312
nomological contingency assumption (NCA)
258
nomological determinism 74, 407
nomological deviance 254
nomological thesis 255
nomological uniformity 2546, 265
nomologicality 436
non-Euclidean geometry 5, 231
elliptical/spherical 221
hyperbolic 221
positive v. negative curvature 222
non-Euclidean hole 234, 238, 243, 264, 368
non-Euclidean space 231, 233, 347
non-indexical tensed sentences 72
non-local connections 136, 422, 436
non-standard analysis 3067, 421
normal line 222
North, Jill 409
Norton, John 283, 3725, 377, 379, 417, 419
nothingness 1, 146, 148, 152, 213, 403
noumenality 439
Oaklander, Nathan 17, 4078, 411
Occams razor 75, 154, 389, 436
ontology 436
open regions 304
Oppy, Graham 302, 419
Ostriker, J. P. 423
overdetermination 21, 23, 25, 70, 436
overlap theory 118
Owen, G. E. L. 419
Owens, David 578, 409
OxfordCambridge case 256, 266
t 215
parallel lines in sphere-space 215
parallel postulate 220
parallel-transport 417
Parfit, Derek 378, 408
Pargetter, Robert 293, 419
Parmenides 8, 26, 267, 407
partial sum 274
passage of time 17, 19, 245, 28, 73, 145, 436
past and future pains (and other experiences),
differing attitudes towards 35
Peirce, C. S. 268, 302, 306, 30911, 42021
Penrose, Roger 271, 354, 404, 423
Penrose tiling 299
perihelion shift 352
persistence 436
and causation 39
four-dimensionalist conception 40
personal (proper) time 37, 123, 437
phase space 86, 410
phenomenal bonding 107, 114
phenomenal connections 119
phenomenal flow 104, 106, 110, 119, 412
and absolute becoming/annihilation 11920
phenomenal property 436
phenomenal time 11618, 412
defined 116
phenomenology 105, 336, 386, 436
philosophy of science, realism v. anti-realism
231
physical objects as spacetime worms 39
Planck length 391, 395, 404
Planck scale 393, 405
Planck tension 396
Plato 219, 424
Platonia 86
Pleasantville 149
plenum 146, 151, 173, 212
Poincar, Henri 8, 48, 208, 22831, 245,
3012, 343, 367
point-free space 310
Pooley, Oliver 41617
Popper, Karl 138
possible worlds 45, 234, 178, 188, 25960,
353, 436
predicate 437
presence of experience 29
present 80
presentedness 109, 111
Presentism 3, 817, 95, 11720, 125, 127,
328, 33031, 3356, 342, 384, 422, 427,
437
collapse into pure solipsism 331
Many-Worlds 84
Pluralistic 84
Solipsistic 83, 85
and time travel 81, 121
presentness 7, 11, 1826, 669, 79
attempt at characterization 19
as relational 19
pre-spatial shards 403, 414
Price, Huw 73, 1346, 409, 413
principle of circular inertia 171
principle of inertia 170, 173
Prior, Arthur 36
properties 437
462 INDEX
causal and structural 20
intrinsic v. relational 18
proposition 437
protentions and retentions 110
psychological arrow 45
Ptolemy 146, 219
Putnam, Hilary 1356, 32930, 333, 341,
420
Pythagoras theorem 270, 323, 408
quantum chromodynamics (QCD) 391
quantum entanglement 339
quantum field theory (QFT) 3915
quantum foam 138, 393
quantum gravity 41, 43, 120, 313, 341,
4045, 426
quantum theory 46, 41, 846, 127, 1368,
200, 212, 313, 33942, 374, 3813, 387,
3913, 398, 41011, 416, 437
many-worlds interpretation 86
transactional interpretation 413
quark 391
Quinton, Anthony 149, 414
Racetrack (paradox of) 272
radiation asymmetry 408
Raine, D. J. 424
Randall, Lisa 400
Ray, C. 424
real v. rational (number lines) 270
realism 441
reality of past and future, an empirical test
144
real-valued functions 271
recording devices, Horwichs account 59
rectilinear inertia 173, 415
reductionism 824, 437
about the past 82, 85
refusnik strategy 336
Reichenbach, Hans 50, 2289, 338, 417
relationism 23, 10, 1558, 1613, 18084,
188, 2059, 212, 236, 240, 245, 370, 381,
4045, 438
characterized 154
and empty space 156
Newtonian (spacetime) 206
problem of inertial motions 183
realist v. idealist 157
relations, varieties of 155, 437
relative consistency proof 220
relativity postulate 31417, 321
relativity theory 441
relativization
of existence (Sklars suggestion) 331
of what is real and definite 332
Reyes, G. E. 420
Riemann, Bernhard 22024, 274, 422
Riemannian geometry 417
Rietdijk, C. W. 32930, 333
Robinson, Abraham 3067, 421
Roeper, P. 311
rotation 205, 210
as form of acceleration 186
Rovelli, Carlo 41, 426
Rucker, Rudy 414, 419
Russell, Bertrand 32, 268, 289
Rynasiewicz, R. 37980, 416, 424
Saccheri, G. G. 220
Salmon, Wesley C. 296, 298, 419
Sanford Russell, J. 312
Saunders, Simon 82
Savitt, Steven 385, 425
Scherk, J. 393
Schlesinger, George 17
Schrdinger, E. 86, 271
wave equation 440
Schwarz, J. A. 393
Schwarzschild, K. 354, 367
science fiction 4, 82, 122, 127, 413
second law of thermodynamics 6, 60
secondary properties 438
Sedley, David 420
SeifertWeber space 228
Selby, D. 414
selfother asymmetry 38
sentence/statement 438
Shakespeare, William (works as object) 138
Shoemaker, local freezes 407
Sider, Ted 40, 90, 923, 413
simple overlap theory 114
simple point substantivalism 377
Simplicius 275
simultaneity, 314
absolute 339
Einsteins method 321
relativization of (in STR) 325, 329
singular terms 91
singularity 53, 137, 3546, 370, 438
singularity theorem (Penrose) 354
Sklar, Lawrence 48, 180, 1923, 199, 2024,
228, 283, 331, 333, 371, 417, 424
Skow, Brad 408, 416, 422
Slater, Mathew H. 140
Slipher, V. M. 356
Smart, J. C. C. 32, 4523
Smeenk, Chris 414
Smith, Quentin 17, 65, 408, 422
Smolin, Lee 300, 396, 405, 425
smooth infinitesimal analysis 420
solipsism 442
Solipsistic Presentism 88
empirical refutation of 117
Sorabji, Richard 419
Sorenson, R. 268
space
absolute 146, 1657, 17393, 201, 2057,
315, 318, 338, 369, 374, 380, 405, 415,
416, 428
continuous v. discrete 269
corrugated 263
as divine attribute 167, 414
expansion of 218
finite or infinite 4
INDEX 463
flat v. curved 216
its innocuousness explained 237
intrinsic metrical amorphousness 162
intrinsic v. extrinsic geometrical features 224
local v. global properties 224
as maker of worlds 402
new function for 398
non-classical 213
orientable v. non-orientable 2423
radical anti-realism 157
strange geometry v. unusual forces 229
topology-destroying tears 398
toroidal 226
as void 145, 148
woven from loops 405
spacetime 3, 1423, 159, 167, 194213, 231,
239, 245, 250, 314, 32298, 4024, 408,
4235
absolute and substantival (Tooley) 338
curved 143
as fictional representation 206
Galilean 197, 200
inertial v. non-inertial paths 196
ingredients of 195
neo-Newtonian 194, 322
construction of 202
Newtonian 194, 197, 202, 2056, 322, 328,
378, 422
not just a mathematical model 204
and relativity theory 194
semi-Riemmanian 423
simply- v. multi-connected 365
spacetime curvature 351
matter induced 347
spacetime interval 323
spatial anti-realism 2456
spatial girth, indifference to 408
spatial location, non-unified 149
spatial relations, formal properties 155
spatial substantivalism, three variants 152
spatial surgery 158
special theory of relativity (STR) 81, 123, 136,
143, 198, 213, 31348, 353, 370, 381, 387,
391, 395, 408, 416, 421, 422
puzzle cases 421
Winnies modified form 338
specious present 106, 10910, 11415
spin networks 4035
squillion 181
standard candle 361
standard model 3915
Starkman, G. D. 424
static shift 175, 178, 201, 206
time as eternal or static 78, 110
steady state theory 423
Stein, Howard 3326, 341, 382, 416, 422
Steinhardt, Paul 4012, 423
Stewart, Ian 414
Strawson, Galen 420
stream of consciousness 103
string theory 387, 393405, 411, 414, 425
substance 439
substantial change 21
substantivalism 25, 8, 10, 44, 145, 1516,
16067, 180, 20112, 23244, 2646, 343,
37181, 404, 406, 439
argument from spatial curvature 236
naive v. circumspect 264
neo-Newtonian 202
relational v. container v. super 152
and Zenos Plurality Paradox 303
sufficient v. necessary condition 439
sum total of reality 9, 66, 67, 7071, 748, 95,
117, 125
Sundrum, Raman 400
Superman 36, 415
superposition 437
superstrings 4
supervenience 442
Susskind, Leonard 425
symmetry 439
failure of 333
and substantivalism 237
tachyons 142
tangent 439
plane 222
Tarski, A. 305, 311, 419, 420
TarskiBanach paradox 305, 41920
Teller, P. 2014
temporal determinism 9
temporal parts 3841, 2067, 210
temporal passage, characterized 7
temporal vacua 14, 407, 439
tensed properties, as basic 67
tensed theory 12, 17, 25, 66
tensed token-reflexive truth-conditions 64
tenseless theory 12
tensor 2245, 353, 3712, 417, 422, 423, 439
Ricci v. Weyl v. Einstein 422
stress-energy 380, 423
tensor field 423
tensorvectorscalar gravity theory 365
Terminator films 124
Thales 218
Thank goodness thats over! 36
The Principle of No Disturbance (ND) 135
thermodynamics 47, 49, 138, 433, 443
thinning tree (model of time) 734
Thorne, Kip 143, 354, 369, 383, 416, 4224,
425
tidal forces 423, 443
time
absolute 339, 382-3, 416
branching 4
the causal theory 162, 429
as the dimension of change 14
dynamic models 18, 25, 73
dynamic view 12
as emergent 41
Lowes version of the tensed conception, 63
modal model 24
the natural (or common sense) view 28, 69
its ontological status 3
464 INDEX
tensed and tenseless conceptions 11
as two-dimensional 23
at the phenomenal level 110
as unreal 13
what we can learn from our immediate
experience 120
time and space
as absolute 4, 314
as containers 2
differences between 6
existence of 1
fundamentally the same 41
structural features 3
time dilation 31718, 327, 337, 344, 421
time travel 37, 81, 12144, 383, 413
and the Block view 124
and changing the past 123
consistency constraints 130
no destination problem 138
in Growing Block universes 140
Meilands (two-dimensional) model 125
in a multi-verse 126
and presentism 138
problem of coincidences 129
relativity and futureward slides 122
self-defeating loops 122, 12930, 1334,
141
sexual loops 413
slides v. jumps 122
through two-dimensional time 125
timeless 408
time-reversible 48, 54, 78, 409, 439
times and events, creation and annihilation 26
token-reflexive account 33-4
tokens v. types 440
Tooley, Michael 713, 778, 83, 291, 293,
328, 33742, 385, 410, 41213, 419, 422
topic-neutral 25051, 257, 440
topology 145, 2256, 25960, 371, 373, 393,
398, 418, 440
Torretti, R. 146, 382, 414, 423
torus, flat 226
transcendental 440
transitory tensed property (model of time) 18,
409
transitory tensed intrinsics 25, 69
transtemporal spatial distance, elimination of
197
truthmaker 314, 64, 82, 85, 912, 100,
11718, 139, 2612
truthmaker principle 82, 90
Turok, Neil 4012
Twin Paradox 421
typetoken distinction 33, 440
ultimate reality 245, 262
underdetermination of theory by data 231
universe
broad v. narrow sense 440
closed 355
critical value (density) 357
different senses of dynamic 423
expanding 217, 356
as giant crystal 414
as mechanical 173
universe-tree 74, 410
unmeasurable set 420
vacuum 145-6, 188, 228, 31415, 322, 324
and quantum field theory 393
vacuum energy 363
Van Bendegem, J. P. 2989
Van Cleve, J. 417
van Fraassen, Bas 396
variably curved space 234, 239, 343
vectors 224, 417, 440
verification principle 181, 440
void 403
void conception (of space) 146, 14854
Vulcan 364
Wallace, David Foster 285, 419
Warren, James 411
waves, basic properties 316
weak force 391, 425
Weatherson, Brian 312
Weeks, Jeffrey 424
Weierstrass, K. 292, 307
Weyl, Hermann 2978, 3012, 310, 422, 423
Weyls tile argument 297
Wheeler, J. A. 353
WheelerDeWitt equation 867
White, M. J. 420
Whitehead, A. N. 268, 302, 31112, 412, 416
Wilkinson Microwave Anisotropy Probe (WMAP)
358
Williams, Donald C. 412, 44, 385, 408
Williamson, Timothy 408
WIMPs 360
Winnie, J. A. 338
Witten, Edward 399, 425, 454
Woit, Peter 396
Wolfram, Stephen 420
world-arrow 55, 767
worldline 1969, 20911, 239, 328, 37071,
416, 440
growing 334
worlds (different uses of the term) 440
worlds-in-reverse 53, 77
Boltzmanns hypothesis 409
wormhole 122, 1434, 3834, 393
time machines 143
Wthrich, Christian 414
Yourgrau, P. 425
Zalamea, F. 420
Zeno 7, 26789, 2949, 3013, 30712,
41819, 421
Zimmerman, Dean 302, 305, 407
zombies 410
Zwicky, Fritz 36061