PHILIP KEAREY & FREDERICK J.

VINE

Global
Tectonics
SECOND
EDITION
GLOBAL TECTONICS
Global Tectonics

PHILIP KEAREY
Department of Geology
University of Bristol

FREDERICK J. VINE
School of Environmental Sciences
University of East Anglia, Norwich

SECOND EDITION

b
Blackwell
Science
© 1990,1996 by Blackwell Science Ltd,
a Blackwell Publishing company

BLACKWELL PUBLISHING

350 Main Street, Maiden, MA 02148-5020, USA

9600 Garsington Road,Oxford OX4 2DQ,UK
550 Swanston Street, Carlton, Victoria 3053, Australia

All rights reserved. No part of this publication may be reproduced, stored in a retrieval
system, or transmitted, in any form or by any means,electronic, mechanical, photocopying,
recording or otherwise, except as permitted by the UK Copyright, Designs,and Patents Act
1988, without the prior permission of the publisher.

First edition published 1990

Second edition published 1996

9 2005

Library of Congress Cataloging-in-Publication Data

Kearey,P.
Global tectonics / Philip Kearey, Frederick J. Vine. — 2nd ed.
p. cm.
Includes bibliographical references(p. -)and index.
ISBN 0-86542-924-3
1. Plate tectonics. I. Vine, F. J. II. Title.
QE511.4.K43 1996
551.1^36—dc20 95-49623
CIP

ISBN-13: 978-0-86542-924-6

A catalogue record for this title is available from the British Library.

Set by Semantic Graphics, Singapore

Printed and bound in the United Kingdom
by TJ International, Padstow, Cornwall

The publisher's policy is to use permanent paper from mills that operate a sustainable
forestry policy, and which has been manufactured from pulp processed using acid-free and
elementary chlorine-free practices. Furthermore, the publisher ensures that the text paper
and cover board used have met acceptable environmental accreditation standards.

For further information on

Blackwell Publishing, visit our website:
www.blackwellpublishing.com
Contents
Preface, viii
Acknowledgements, x
1 Historical perspective,1
1.1 Continental drift, 1
1.2 Sea floor spreading and the birth of plate
tectonics, 5
1.3 Geosynclinal theory,6
1.4 Impact of plate tectonics, 7
2 internal structure of the Earth,8
2.1 Earthquake seismology, 8
2.1.1 Introduction, 8
2.1.2 Earthquake descriptors, 8
2.1.3 Seismic waves,8
2.1.4 Earthquake location,9
2.1.5 Mechanism ofearthquakes, 10
2.1.6 Focal mechanism solutions of
earthquakes, 10
3 Continental drift,46
2.1.7 Ambiguity in focal mechanism
solutions, 12
2.1.8 Seismic tomography, 15
2.2 Velocity structure ofthe Earth, 17
2.3 Composition of the Earth, 19
2.4 The crust, 19
2.4.1 The continental crust, 19
2.4.2 Upper continental crust, 20
2.4.3 Lower continental crust, 20
2.4.4 The oceanic crust, 21
2.4.5 Oceanic layer 1, 21
2.4.6 Oceanic layer 2, 22
2.4.7 Oceanic layer 3, 23
2.5 Ophiolites, 24
2.6 Metamorphism of oceanic crust and upper
mantle, 25
2.7 Differences between continental and
oceanic crust, 26
2.8 The mantle, 26
2.8.1 Introduction, 26
2.8.2 Seismic structure ofthe mantle, 27
2.8.3 Mantle composition, 27
2.8.4 The mantle low velocity zone, 27
2.8.5 The mantle transition zone, 28
2.9 The core, 28
2.10 Deformation in the crust and mantle, 30
2.10.1 Introduction, 30
2.10.2 Brittle deformation, 30
2.10.3 Ductile deformation, 31
2.10.4 Creep in the mantle, 33
2.11 Isostasy, 34
2.11.1 Introduction, 34
2.11.2 Airy's hypothesis, 35
2.11.3 Pratt's hypothesis, 35
2.11.4 Flexure ofthe lithosphere, 36
2.11.5 Isostatic rebound, 37
2.11.6 Tests ofisostasy, 38
2.12 Lithosphere and asthenosphere,40
2.13 Terrestrial heat flow,42
3.1 Introduction, 46
3.2 Continental reconstructions, 46
3.2.1 Euler's theorem,46
3.2.2 Geometrical reconstructions of
continents,46
3.2.3 The reconstruction ofcontinents
around the Atlantic, 47
3.2.4 The reconstruction of Gondwana,48
3.3 Geological evidence for continental
drift, 49
3.4 Palaeoclimatology, 51
3.5 Palaeontological evidence for continental
drift, 52
3.6 Palaeomagnetism, 55
3.6.1 Introduction, 55
3.6.2 Rock magnetism,55
3.6.3 Natural remanent magnetization, 55
3.6.4 The past and present geomagnetic
field, 56
3.6.5 Apparent polar wandering
curves, 57
VI CONTENTS

3.6.6 Continental reconstructions based on 7.3 Oceanic fracture zones, 125

palaeomagnetism, 59 7.4 Continental strike-slip faults, 128
7.4.1 Introduction, 128
7.4.2 Transpression and transtension, 130
4 Sea floor spreading,62
7.4.3 Pull-apart basins, 131
4.1 Introduction,62 7.4.4 Fault wedge basins, 132
4.2 Marine magnetic anomalies,62 7.4.5 Deep structure oftransform
4.3 Sea floor spreading,64 zones, 133
4.4 Geomagnetic reversals, 64 7.5 Sheared continental margins, 135
4.5 The Vine-Matthews hypothesis,67
4.6 Magnetostratigraphy,69
4.7 Deep sea drilling results, 74 8 Subduction zones,138
4.8 Introduction to transform faults, 75 8.1 General morphology of oceanic
subduction zones, 138
8.2 Gravity anomalies ofsubduction
5 Theframework of plate
zones, 139
tectonics,76
8.3 Structure ofisland arc systems from
5.1 Plates and plate margins, 76 earthquakes, 139
5.2 Distribution of earthquakes, 76 8.4 Thermal structure of the downgoing
5.3 Relative plate motions, 78 slab, 148
5.4 Hotspots and absolute plate motions,81 8.5 Metamorphism at convergent
5.5 True polar wander,83 margins, 150
5.6 Superplumes,88 8.6 Ocean trenches, 155
5.7 Direct measurement of relative plate 8.7 Accretionary prism, 155
motions,90 8.8 Volcanic and plutonic activity, 160
5.8 Instantaneous and finite rotations, 92 8.9 Marginal seas, 163
5.9 Triple junctions,95 8.10 Example: the Lesser Antilles subduction
zone, 170
6 Ocean ridges, 103
6.1 Ocean ridge topography, 103 9 Mountain ranges,174
6.2 Structure ofthe axial zone, 103 9.1 General, 174
6.3 Broad structure ofthe lithosphere below 9.2 Andean-type mountain ranges, 174
ridges, 104 9.2.1 Introduction, 174
6.4 Origin ofanomalous upper mantle 9.2.2 General geological structure of the
beneath ridges, 107 Peruvian Andes, 174
6.5 Shallow structure ofthe axial region, 108 9.2.3 Seismicity ofthe Andes, 175
6.6 Heat flow and hydrothermal 9.2.4 Plate tectonic model of the
circulation, 112 Andes, 177
6.7 Petrology ofocean ridges, 114 9.3 Collisional mountain ranges, 180
6.8 Differences between fast and slow 9.3.1 Introduction, 180
spreading ridges, 115 9.3.2 Models of continental
6.9 Depth-age relationship of oceanic collision, 180
lithosphere, 116 9.3.3 Himalayan geology, 182
6.10 Origin of the oceanic crust, 116 9.3.4 Deep structure ofthe
6.11 Propagating rifts, 118 Himalaya, 184
9.3.5 Evolution of the Himalaya, 185
7 Transform and transcurrent 9.3.6 Indentation tectonics, 187
9.3.7 The mechanism ofcontinental
faults,122
collision, 194
7.1 Introduction, 122 9.4 Arc-continent collision, 195
7.2 Origin oftransform faults, 123 9.5 Suspect terrains, 198

10 Continental rifts, continental
margins and sedimentary
basins,209
10.1 Continental rifts, 209
10.1.1 Introduction, 209
10.1.2 Characteristics of rifts, 209
10.1.3 Classification of rifts, 209
10.1.4 Petrology and petrogenesis of rift
rocks, 210
10.1.5 Structure ofcontinental rifts, 211
10.1.6 Origin of rifts, 212
10.2 Aulacogens, 216
10.3 Continental splitting, 218
10.3.1 Previous theories, 218
12 Implications of piate
10.3.2 Requirements for continental
splitting, 221
10.3.3 The origin of regional continental
tension, 223
10.3.4 Evidence from continental
margins, 225
10.3.5 Models of continental
splitting, 226
10.3.6 Volcanic margins, 231
10.4 The Wilsonian cycle, 234
10.5 Vertical movements associated with plate
tectonics, 235
10.5.1 Introduction, 235
10.5.2 Sedimentary basin formation, 236
10.5.3 Foreland basins, 245
10.5.4 Inversion, 246
11 The mechanism of plate
tectonics,248
11.1 Introduction, 248
11.2 Contracting Earth hypothesis, 248
11.3 Expanding Earth hypothesis, 248
11.3.1 Calculation of the ancient moment
ofinertia ofthe Earth, 249
11.3.2 Calculation of the ancient radius
ofthe Earth, 250
CONTENTS VII
11.4 Implications of heat flow, 250
11.5 Convection in the mantle, 252
11.5.1 The convection process, 252
11.5.2 Feasibility of mantle
convection, 253
11.5.3 The vertical extent of
convection, 255
11.6 The forces acting on plates, 258
11.7 Driving mechanism of plate tectonics, 260
11.7.1 Mantle drag mechanism, 260
11.7.2 Edge-force mechanism, 261
11.8 Areal distribution ofconvection cells, 262
tectonics,269
12.1 Palaeozoic plate tectonics, 269
12.2 Precambrian plate tectonics, 269
12.2.1 Introduction, 269
12.2.2 Heat flow in the Precambrian, 271
12.2.3 Archaean tectonics, 272
12.2.4 Precambrian palaeomagnetic
studies, 275
12.2.5 Proterozoic tectonics, 279
12.2.6 Proterozoic plate tectonics, 282
12.3 Plate tectonics and economic
geology, 285
12.3.1 Introduction, 285
12.3.2 Autochthonous and allochthonous
deposits, 285
12.3.3 Deposits ofsedimentary
basins, 292
12.3.4 Deposits related to climate, 294
12.3.5 Geothermal power, 295
Review questions, 297
Appendix: The geological timescale and
stratigraphic column, 301
References, 303
Index, 323
Acknowledgements
We thank Professors H. Robert Burger, Smith Col
lege; Stephan P. Phipps, University of Pennsylvania;
and Eli A. Silver, University of California at Santa
Cruz, for their most helpful reviews of the original
manuscript. Professor Tony Hallam provided useful
comments on palaeontological matters and Pro
fessor W. Jason Morgan advice on plate kinematics.
We also acknowledge the intellectual stimulation
provided by our colleagues and students in the Uni
versities of Bristol and East Anglia, and at Princeton
University where FJV spent a sabbatical term during
the preparation of this book.
1 / Historical perspective
1.1 Continental drift
Although the theory of the new global tectonics, or
plate tectonics, has largely been developed since
1967,the history ofideas concerning a mobilist view
of the Earth extends back considerably longer
(Rupke, 1970; Tarling & Tarling, 1971; Hallam,
1973a; Vine, 1977; Frankel, 1988). Ever since man
first charted the coastlines of the continents around
the Atlantic Ocean in the 16th century, he has been
intrigued by the similarity of the coastlines of the
Americas and ofEurope and Africa. Possibly the first
to note the similarity and suggest an ancient separa
tion was Abraham Ortelius in 1596(Rom,1994). In
1620, Sir Francis Bacon, in his Novum Or£fanum^
commented on the similar form ofthe west coasts of
Africa and South America: that is, the Atlantic coast
of Africa and the Pacific coast ofSouth America. He
also noted the similar configurations ofthe New and
Old World,'both of which are broad and extended
towards the north, narrow and pointed towards the
south.' Perhaps because of these observations, for
there appear to be no others. Bacon is often errone
ously credited with having been first to notice the
similarity or 'fit' of the Atlantic coastlines of South
America and Africa and even with having suggested
that they were once together and had drifted apart.In
1666, Fran9ois Placet, a French prior, related the
separation of the Americas to the Flood of Noah.
Noting from the Bible that prior to the flood the
Earth was one and undivided, he postulated that the
Americas were formed by the conjunction offloating
islands or separated from Europe and Africa by the
destruction of an intervening landmass, 'Atlantis'.
One must remember,ofcourse,that during the 17th
and 18th centuries geology, like most sciences, was
carried out by clerics and theologians who felt that
their observations, such as the occurrence of marine
fossils and water-lain sediments on high land, were
explicable in terms of the Flood and other biblical
catastrophies.
Another person to note the fit of the Atlantic
coastlines of South America and Africa and to sug
gest that they might once have been side by side was
Theodor Christoph Lilienthal, Professor of Theo
logy at K6nigsberg in Germany. In a work dated
1756 he too related their separation to biblical cata-
strophism,drawing on the text,'in the days ofPeleg,
the earth was divided.' In papers dated 1801 and
1845,the German explorer Alexander von Humbolt
noted the geometric and geological similarities ofthe
opposing shores of the Atlantic, but he too specu
lated that the Atlantic was formed by a catastrophic
event, this time 'a flow ofeddying waters... directed
first towards the north-east, then towards the north
west, and back again to the north-east... What we
call the Atlantic Ocean is nothing else than a valley
scooped out by the sea.' In 1858 an American,
Antonio Snider-Pellegrini, made the same observa
tions but postulated 'drift' and related it to 'multiple
catastrophism'—the Flood being the last major cata
strophe. Thus Pellegrini suggested drift sensu stficto^
and he even went so far as to suggest a pre-drift
reconstruction (Fig. 1.1).
The 19th century saw the gradual replacement of
the concept ofcatastrophism by that of'uniformitar-
ianism' or 'actualism' as propounded by the British
geologists James Hutton and Sir Charles Lyell. Hut-
ton wrote'No powers are to be employed that are not
natural to the globe, no action to be admitted of
except those of which we know the principle, and no
extraordinary events to be alleged in order to explain
a common appearance.' This is usually stated in Sir
Archibald Geikie's paraphase ofHutton's words,'the
present is the key to the past,' that is, the slow
processes going on at and beneath the Earth's surface
today have been going on throughout geological
time and have shaped the surface record. Despite this
change in the basis of geological thought, the pro
ponents of continental drift still resorted to cata
strophic events to explain the separation of the
continents. Thus, in 1882 Oswald Fisher associated
 CHAPTER 1
Europe
Africa
Australia
America
Fig.1.1 Snider-Pellegrini's reconstruction of 1858.
drift with the origin of the Moon out of the Pacific,
and this idea persisted well into the 20th century.
A uniformitarian concept of drift was first sug
gested by F.B. Taylor, an American physicist, in
1910,and Alfred Wegener,a German meteorologist,
in 1912. For the first time it was considered that drift
is taking place today and has taken place at least
throughout the past 100-200 Ma of Earth history.
In this way drift was invoked to account for the
geometric and geological similarities of the trailing
edges of the continents around the Atlantic and
Indian oceans and the formation of the young fold
mountain systems at their leading edges. Taylor, in
particular,invoked drift to explain the distribution of
the young fold mountain belts and 'the origin ofthe
Earth's plan'(Taylor, 1910)(Fig. 1.2).
The pioneer of the theory of continental drift is
generally recognized as Alfred Wegener, who as well
as being a meteorologist was an astronomer, geo-
physicist and amateur balloonist (Hallam, 1975),
and he devoted much of his life to its development.
Wegener detailed much of the older, pre-drift,
geological data and maintained that the continuity of
the older structures,formations and fossil faunas and
floras across present continental shorelines was more
readily understood on a pre-drift reconstruction.
Even today,these points are the major features ofthe
geological record from the continents which favour
the hypothesis ofcontinental drift. New information
which Wegener brought to his thesis was the
presence of a widespread glaciation in Permo-
Carboniferous times which had aflfected most of the
southern continents while northern Europe and
Greenland had experienced tropical conditions.
Wegener postulated that at this time the continents
were joined into a single landmass, with the present
southern continents centred on the pole and the
northern continents straddling the equator.
Wegener termed this continental assembly Pangaea
(literally 'all the Earth') (Fig. 1.3), although we
currently prefer to think in terms of A. du Toit's idea
of its being made up of two supercontinents (du
Toit,1937)(Fig. 3.5). The more northerly ofthese is
termed Laurasia(from a combination ofLaurentia, a
region of Canada, and Asia), and consists of North
America,Greenland,Europe and Asia. The southerly
supercontinent is termed Gondwana (literally 'land
of the Gonds' after an ancient tribe of northern
India), and consisted of South America, Antarctica,
Africa, Madagascar,India and Australasia. Separating
the two supercontinents was a former 'mediterra
nean' sea termed the palaeo-Tethys Ocean (after the
Greek goddess of the sea), while surrounding
Pangaea was the proto-Pacific Ocean or Panthalassa
(literally 'all-ocean').
Wegener propounded his new thesis in a book Die
Entstehun£f der Kontinente und Ozeam {The Origin of
Continents and Oceans)^ of which four editions ap
peared in the period 1915-28. Much ofthe ensuing
academic discussion was based on the English trans
lation of the 1922 edition which appeared in 1924,
consideration ofthe earlier work having been delayed
by World War I. Many Earth scientists of this time
found his new ideas difficult to encompass as accep
tance of his work necessitated a rejection of the
existing scientific orthodoxy, which was based on a
static Earth model. Wegener based his theory on data
drawn from several different disciplines, in many of
which he was not an expert. The majority of Earth
scientists found fault in detail and so tended to reject
his work in toto. Perhaps Wegener did himself a
disservice in the eclecticism of his approach. Several
of his arguments were incorrect: for example, his
estimate of the rate of drift between Europe and
Greenland using geodetic techniques was in error by
an order of magnitude. Most important, from the
point of view of his critics, was the lack of a rea
sonable mechanism for continental movements.
Wegener had suggested that continental drift oc-

i
i
Fig. 1.2 Taylor's mechanism for the formation of Tertiary
mountain belts by continental drift (after Taylor, 1910).
curred in response to the centripetal force experi
enced by the high-standing continents because ofthe
Earth's rotation. Simple calculations showed the
forces exerted by this mechanism to be much too
small. Although in the later editions of his book this
approach was dropped,the objections ofthe majority
of the scientific community had become established.
Du Toit, however, recognized the good geological
arguments for the joining ofthe southern continents
and A. Holmes,in the period 1927-29, developed a
new theory of the mechanism of continental move
ment(Holmes, 1928). He proposed that continents
were moved by convection currents powered by the
heat of radioactive decay (Fig. 1.4). Although dif
fering considerably from the present concept of con
vection, Holmes laid the foundation from which
modern ideas developed.
Between the world wars two schools of thought
developed—the drifters and the non-drifters, the
latter far outnumbering the former. Each ridiculed
the other's ideas. The non-drifters emphasized the
lack of a plausible mechanism, as we have already
HISTORICAL PERSPECTIVE
noted, both convection and Earth expansion being
considered unlikely. It is interesting to note that
historically many southern hemisphere geologists,
such as du Toit, Lester King and S.W. Carey, have
been advocates of drift, perhaps because the geologi
cal record from the southern continents and India
favours their assembly into a single supercontinent
(Gondwana) prior to 200 Ma ago. The non-drifters
had difficulty in explaining the present separation of
faunal provinces, for example, which could be much
more readily explained if the continents were for
merly together, and their attempts to explain these
apparent faunal links or migrations also came in for
some ridicule. They had to invoke various improb
able means such as island stepping-stones, isthmian
links or rafting.
Very little was written about continental drift
between the initial criticisms of Wegener's book and
about 1960. In the 1950s, employing methodology
suggested by Lord Blackett, the palaeomagnetic
method was developed and S.K. Runcorn and his
co-workers demonstrated that relative movements
had occurred between North America and Europe.
The work was extended by K.M. Greet into South
America and by E. Irving into Australia. Palaeomag
netic results became more widely accepted when the
 4 CHAPTER 1
Panqaea
Panthalassa
pper Carboniferous
-ower Quaternary
technique of magnetic cleaning was developed in
which primary magnetization could be isolated.
Coupled with dating by faunal or newly developed
radiometric methods, the palaeomagnetic data for
Mesozoic to Recent times showed that there had
been significant differences, beyond the scope of
error, in the motions between various continents.
Panthaiassa
Fig. 1.3 Wegener's reconstruction
of the positions ofthe continents
from Carboniferous to Quaternary
An important consideration in the development
of ideas relating to continental drift was that prior
to World War II geologists had, necessarily, only
studied the land areas. Their findings had revealed
that the continental crust preserves a whole spectrum
of Earth history, ranging back to nearly 4000 Ma
before the present, and probably to within a few
 Borderland
Deep I Geosyncllne
ocean
» X X X X
\ X X X
V'X X X X J
^ X X X'
Fig. 1.4 The concept ofconvection as suggested by
Holmes(1928), when it was believed that the oceanic
crust was a thick continuation ofthe continental 'basaltic
layer',(a) Currents ascending at A spread laterally, place a
continent under tension and split it, providing the
obstruction ofthe old ocean floor can be overcome. This
is accomplished by the formation of eclogite at B and C,
where sub-continental currents meet sub-oceanic currents
and turn downwards. The high density of the eclogite
causes it to sink and make room for the continents to
advance,(b) The foundering ofeclogite at B and C
hundred million years ofthe age ofthe Earth and the
solar system itself. Their studies also revealed the
importance ofvertical movements of the continental
crust in that the record was one ofrepeated uplift and
erosion, subsidence and sedimentation. But as J.
Tuzo Wilson, a Canadian geophysicist, has said, this
is like looking at the deck of a ship to see if it is
moving.
1.2 Sea floor spreading and the
birth of plate tectonics
Ifthere is a possibility that the continental areas have
been rifted and drifted apart and together, then
presumably there should be some record of this
within the ocean basins. However, it is only since
World War II and notably since I960 that sufficient
data have been obtained from the 60% ofthe Earth's
surface covered by deep water for some understand
Borderland
GeosyncllnejDeep
S\^ll ocean
X X X X
X X X X X/
\ X X X X /
^ X X X f
contributes to the main convective circulation. The
eclogite melts at depth to form basaltic magma, which
rises in ascending currents at A, heals the gaps in the
disrupted continent and forms new ocean floor. Local
swells, such as Iceland, would be formed from old sial left
behind. Smaller current systems, initiated by the buoyancy
of the basaltic magma,ascend beneath the continents and
feed flood basalts or, beneath 'old'(Pacific) ocean floor,
feed the outpourings responsible for volcanic islands and
seamounts (redrawn from Holmes, 1928).
ing of the origin and history of the ocean basins to
have emerged. It transpires that, in contrast to the
continents, the oceanic areas are very young geo
logically (probably no greater than 200 Ma in age)
and that horizontal, or lateral, movements have been
all-important during their history offormation.
In I96I, following intensive surveying of the sea
floor during post-war years, R.S. Dietz proposed the
mechanism of'sea floor spreading' to explain contin
ental drift, a concept originally conceived by H.H.
Hess of Princeton University, whereby the contin
ents moved in response to the growth ofocean basins
between them. It was suggested that oceanic crust is
created from the Earth's mantle at the crest of the
mid-ocean ridge system, a volcanic submarine swell
or rise which occupies a median position in many
of the world's oceans (Fig. 1.5). Oceanic crust is
much thinner than continental crust, having a mean
thickness of about 7km, compared with the average
 CHAPTER 1
Continent
Mid-ocean
Trench
ridge
Mantle Warmer Nj
Cooler
1000 km
continental thickness of about 35 km; is chemically
different; and is structurally far less complex. The
lateral motion ofthe oceanic crust was believed to be
driven by convection currents in the upper mantle in
the fashion of a conveyer belt. In order to keep the
surface area of the Earth constant, it was further
proposed that the oceanic crust is thrust back down
into the mantle and resorbed at oceanic trenches.
These are vast bathymetric depressions, situated at
certain ocean margins and associated with intense
volcanic and earthquake activity. Within this frame
work the continents are quite passive elements—rafts
of less dense material which are drifted apart and
together by ephemeral ocean floors. The continents
themselves are a scum of generally much older
material that was derived or separated from the
Earth's interior either at a very early stage in its
history or, at least in part, steadily throughout geo
logical time. Instead of blocks ofcrust, we now think
in terms of 'plates' of comparatively rigid upper
mantle and crust, perhaps 50-100 km thick and
which we term lithosphere (a term originally coined
by R.A. Daly many years ago and meaning 'rock
layer'). Lithospheric plates can have both continental
and oceanic crust embedded in them.
The theory ofsea floor spreading was confirmed in
the period 1963-66 following the suggestion of
F.J. Vine and D.H. Matthews that the magnetic
lineations of the sea floor might be explained in
terms of sea floor spreading and palaeomagnetism
(see Chapter 4). On this model the conveyor belt
of oceanic crust is viewed as a tape recorder which
registers the reversal history of the Earth's magnetic
field.
A further precursor to the development of the
theory of plate tectonics came with the recognition.
Fig.1.5 The concept ofsea floor spreading (after
Hess, 1962).
by J.T. Wilson in 1965, of a new class of faults
termed transform faults, which connect linear belts
of tectonic activity (see Chapter 7). The Earth was
then viewed as a mosaic of six major and several
smaller plates in relative motion. The theory was
put on a stringent geometrical basis by the work of
D.P. McKenzie,R.L. Parker and W.J. Morgan in the
period 1967-68 (see Chapter 5), and confirmed by
earthquake seismology through the work of B.
Isacks, J. Oliver and L.R. Sykes (see Chapter 8).
The theory has been considerably amplified by
intensive studies of the geological and geophysical
processes affecting plate margins. Probably the aspect
about which there is currently the most contention is
the nature of the mechanism which causes plate
motions (see Chapter II).
Although the basic theory of plate tectonics is well
established, by no means is understanding complete,
and its implications will fully occupy Earth scientists
for many decades to come.
1.3 Geosynclinal theory
Prior to the acceptance of plate tectonics, the static
model of the Earth encompassed the formation of
tectonically active belts, which formed essentially by
vertical movements, on the site of geosynclines. A
review of the development of the geosyncline
hypothesis and its explanation in terms of plate
tectonics is provided by Mitchell & Reading(1986).
Geosynclinal theory envisaged elongate, geo
graphically fixed belts of deep subsidence and thick
sediments as the precursors of mountain ranges in
which the strata were exposed by folding and uplift of
the geosynclinal sediments (Dickinson, 1971). A
plethora ofspecific nomenclature evolved to describe

the lithological associations of the sedimentary fill
and the relative locations ofthe geosynclines.
The greatest failing ofgeosynclinal theory was that
tectonic features were classified without there being
an understanding of their origin. Geosynclinal no
menclature consequently represented an impedi
ment to the recognition of a common causal
mechanism. The relation of sedimentation to the
mobilistic mechanism of plate tectonics (Mitchell &
Reading, 1969) allowed the recognition of two
specific environments in which geosynclines formed,
namely passive continental margins and subduction
zones. Although some workers retain geosynclinal
terminology to describe sedimentary associations
(e.g. the terms eugeosyncline and miogeosyncline
for sediments with and without volcanic members,
respectively), this usage is not recommended, and
the term geosyncline must be recognized as no
longer relevant to plate tectonic processes.
1.4 Impact of plate tectonics
Plate tectonics is of very great significance as it
represents the first theory which provides a unified
explanation of the Earth's major surface features. As
such it has enabled an unprecedented linking of
many different aspects of geology, which had previ
ously been considered independent and unrelated. A
deeper understanding of geology has ensued from
the intepretation of many branches ofgeology within
the basic framework provided by plate tectonics.
Thus,for example, explanations can be provided for
the past distributions of flora and fauna, the spatial
relationships of volcanic rock suites at plate margins.
HISTORICAL PERSPECTIVE
the distribution in space and time of the conditions
of different metamorphic facies, the scheme of
deformation in mountain belts and the association of
different types ofeconomic deposit.
The fact that plate tectonics is so successful in
unifying so many aspects of geology should not be
taken to indicate that it is completely understood.
Indeed,it is the critical testing ofthe implications of
plate tectonic theory which has led to modifications
and extrapolations,for example in the consideration
of the relevance of plate tectonic processes in con
tinental areas and the more distant geological past. It
is to be hoped that plate tectonic theory will be
employed cautiously and critically.
Further reading
Frankel, H.(1988) From continental drift to plate tecton
ics. Nature 335,127-30.
Hallam, A. (1973a) A Revolution in the Earth Sciences.
Clarendon Press, Oxford.
Hallam, A.(1975) Alfred Wegener and the hypothesis of
continental drift. Sci. Am. 232(2), 88-97.
Mitchell, A.H.G. & Reading, H.G.(1986) Sedimentation
and tectonics. In: Reading, H.G. (ed.) Sedimentary
Environments and Facies., pp.471-519. Blackwell
Scientific Publications, Oxford.
Rupke,N.A.(1970)Continental drift before 1900. Nature
227, 349-50.
Stewart, J.A. (1990) Drifting Continents and Colliding
Raradi£fms: Perspectives on the Geoscience Revolution. Indi
ana University Press, Bloomington.
Vine,F.J.(1977)The continental drift debate. Nature 266,
19-22.
2/ Internal structure of the Earth
2.1 Earthquake seismology
2.1.1 Introduction
Much of our knowledge of the internal constitution
ofthe Earth has come from the study of the seismic
waves generated by earthquakes. These waves follow
various paths through the interior of the Earth, and
by measuring their travel times to different locations
around the globe it is possible to determine its
large-scale layering. It is also possible to make infer
ences about the physical properties of these layers
from a consideration of the velocities with which
they transmit the seismic waves.
2.1.2 Earthquake descriptors
Earthquakes are normally assumed to originate from
a single point known as the focus or hypocentre
(Fig. 2.1), which is generally within about 700 km of
the surface. In reality, however, most earthquakes are
generated by movement along a fault plane so that
the focal region may extend for several kilometres.
The point on the Earth's surface vertically above the
focus is the epicentre. The angle subtended at the
centre ofthe Earth by the epicentre and the point at
which the seismic waves are detected is known as the
Epicentre Seismograph
Centre of
Earth
Fig.2.1 Illustration of epicentral angle, A.
8
epkentral an£(le A.The magnitude ofan earthquake is
a measure ofits energy release on a logarithmic scale;
a change in magnitude of one on the Richter scale
implying a 30-fold increase in energy release (Bullen
&Bolt, 1985).
2.1.3 Seismic waves
The strain energy released by an earthquake is trans
mitted through the Earth by several types of seismic
wave (Fig. 2.2), which propagate by elastic deform
ation of the rock through which they travel. Waves
penetrating the interior of the Earth are known as
body waves,,and consist oftwo types corresponding to
the two possible ways of deforming a solid medium.
P waves,, also known as lon£fitudinal or compressional
waves, correspond to elastic deformation by com
pression/dilation. They cause the particles of the
transmitting rock to oscillate in the direction of
travel ofthe wave so that the disturbance proceeds as
a series of compressions and rarefactions. The velo
city of a P wave Fp is given by:
where k is the bulk modulus, p the shear modulus
(rigidity) and p the density of the transmitting
medium. 5 waves,, also known as shear or transverse
waves, correspond to elastic deformation of the
transmitting medium by shearing and cause the par
ticles of the rock to oscillate at right angles to the
direction of propagation. The velocity of an S wave
Fs is given by:
Because the rigidity ofa fluid is zero,S waves cannot
be transmitted by such a medium.
A consequence ofthe velocity equations for P and
S waves is that the P velocity is about 1.7 times

Rayleigh wave
Love wave ground motion
ground motion___
Epicentre
ocus
S wave motion P wave motion
Fig.2.2 Focus and epicentre of an earthquake and the
seismic waves originating from it (after Davies, 1968, with
permission from Iliffe Industrial Publications Ltd).
greater than the S velocity in the same medium.
Consequently, for an identical travel path, P waves
arrive before S waves. This was recognized early in
the history of seismology, and is reflected in the
names of the body waves (P is derived from primus
and S from secundus). The passage of body waves
through the Earth conforms to the laws of geomet
rical optics in that they can be both refracted and
reflected at velocity discontinuities.
Seismic waves whose travel paths are restricted to
the vicinity of a free surface, such as the Earth's
surface, are known as surface waves, Raylei^fh waves
cause the particles of the transmitting medium to
describe an ellipse in a vertical plane containing the
direction of propagation. They can be transmitted in
the surface of a uniform half space or a medium in
which velocity changes with depth. Love waves are
transmitted whenever the S wave velocity of the
surface layer is lower than that of the underlying
layer. Love waves are essentially horizontally polar
ized shear waves, and propagate by multiple reflec
tion within this low velocity layer, which acts as a
wave guide.
Surface waves travel at lower velocities than body
waves in the same medium. Unlike body waves,
surface waves are dispersive, that is, their diflferent
wavelength components travel at different velocities.
Dispersion arises because ofthe velocity stratification
of the Earth's interior, longer wavelengths penetrat
ing to greater depths and hence sampling higher
velocities. As a result, surface wave dispersion studies
INTERNAL STRUCTURE OF THE EARTH 9
provide an important method of determining the
velocity structure and seismic attenuation character
istics ofthe upper 600 km ofthe Earth.
2.1.4 Earthquake location
Earthquakes are detected by seismographs, instru
ments which respond to very small ground displace
ments, velocities or accelerations associated with the
passage of seismic waves. There exists a global net
work of seismographs used in the monitoring of
earthquake activity known as the World-Wide Stan-
darized Seismograph Network(WWSSN),which was
set up in 1962. WWSSN observatories employ at
least two sets of three component seismographs in
order to monitor ground motions ofshort and long
periodicity in the vertical and two, orthogonal, hori
zontal directions. In the 1980s a new generation of
seismographs evolved based on digital recording;
they make up the Federation of Digital Seismic
Networks.
Earthquakes occurring at large, or teleseismk,, dis
tances from the seismograph are located by the iden
tification of wmoMSphasesor seismic arrivals, on the
seismograph records. Since,for example,the direct P
and S waves travel at different velocities, the time
separation between the arrival ofthe P phase and the
S phase becomes progressively longer as the length of
the travel path increases. By making use ofa standard
model for the velocity stratification ofthe Earth, and
employing many seismic phases corresponding to
diflferent travel paths along which the seismic waves
are refracted or reflected at velocity discontinuities, it
is possible to translate the differences in their travel
times into the distance of the earthquake from the
observatory. Triangulation using distances com
puted in this way from many observatories then
allows the location ofthe epicentre to be determined.
The focal depths of teleseismic events are deter
mined by measuring the arrival time difference be
tween the direct phase P and the phase pP (Bath,
1979). The pP phase is a short path multiple event
which follows a similar path to P after first under
going a reflection at the surface of the Earth above
the focus, and so the P-pP time difference is a
measure offocal depth. This method is least accurate
for foci at depths of less than 100 km as the P-pP
time separation becomes very small. The focal depths
of local earthquakes can be determined if a net
work of seismographs exists in the vicinity of the
10 CHAPTER 2
epicentre. In this case the focal depth is determined
by triangulation in the vertical plane, using the P-S
time difference to calculate the distance to the focus.
2.1.5 Mechanism of earthquakes
Most earthquakes are believed to occur according to
the elastic rebound theory^ which was developed after
the San Francisco earthquake of 1906. In this theory
an earthquake represents a sudden release of strain
energy which has built up over a period oftime.
In Fig. 2.3a a block of rock traversed by a pre
existing fracture (or fault) is being strained in such a
way as eventually to cause relative motion along the
plane ofthe fault. The line AB is a marker indicating
the state of strain ofthe system, and the broken line
the location ofthe fault. Relatively small amounts of
strain can be accommodated by the rock (Fig. 2.3b).
Eventually, however, the strain reaches the level
at which it exceeds the frictional and cementing
forces opposing movement along the fault plane
(Fig. 2.3c). At this point fault movement occurs
instantaneously(Fig. 2.3d). The 1906 San Francisco
earthquake resulted from a displacement of 6.8 m
along the San Andreas Fault. In this model,faulting
(a) (b)
tr
Tt-
0 i-Compression®?
(0 (d)
Fault
Fig.2.3 Elastic rebound mechanism of earthquake
generation.
reduces the strain in the system virtually to zero, but
ifthe shearing forces persist,strain would again build
up to the point at which fault movement occurs. The
elastic rebound theory consequently implies that
earthquake activity represents a stepwise response to
persistent strain.
2.1.6 Focal mechanism solutions of
earthquakes
The seismic waves generated by earthquakes, when
recorded at seismograph stations around the world,
can be used to determine the nature of the faulting
associated with the earthquake, to infer the orienta
tion ofthe fault plane and to gain information on the
state of stress of the lithosphere. The results ofsuch
an analysis are referred to as afocal mechanism solution
or fault plane solution. The technique represents a
very powerful method of analysing movements of
the lithosphere, in particular those associated with
plate tectonics. Information is available on a global
scale as most earthquakes with a magnitude in excess
of5.5 can provide solutions,and it is not necessary to
have recorders in the immediate vicinity ofthe earth
quake, so that data are provided from regions which
may be inaccessible for direct study.
Accordi^ig to the elastic rebound theory, the strain
energy released by an earthquake is transmitted by
the seismic waves which radiate from the focus.
Consider the fault plane shown in Fig. 2.4 and the
plane orthogonal to it, the auxiliary plane. The first
Dilation
Auxiliary
Focus
Dilation |Compression|
Fault plane
S
Fig.2.4 Quadrantal distribution of compressional and
dilational P wave first motions about an earthquake.

seismic waves to arrive at recorders around the
earthquake are P waves, which cause compression/
dilation of the rocks through which they travel. The
shaded quadrants defined by the fault and auxiliary
planes are compressed by movement along the fault,
and so the first motion ofthe P wave arriving in these
quadrants corresponds to a compression. This is
recognized by an initial upward displacement at the
Earth's surface and recorded on the vertical seismo
graph. Conversely, the unshaded quadrants are
stretched or dilated by the fault movement. The first
motion of the P waves in these quadrants is thus
dilational, and recognized by an initial downward
displacement at the seismographs. The region
around the earthquake is therefore divided into four
quadrants on the basis of the P wave first motions,
defined by the fault plane and the auxiliary plane. No
P waves propagate along these planes as movement
of the fault imparts only shearing motions in their
directions; they are consequently known as nodal
planes.
Simplistically, then, a focal mechanism solution
could be obtained by recording an earthquake at a
number of seismographs distributed around its epi
centre, determining the nature ofthe first motions of
the P waves, and then selecting the two orthogonal-
planes which best divide compressional from dila
tional first arrivals, that is, the nodal planes. In
practice, however, the technique is complicated by
the spheroidal shape of the Earth and the progressive
increase ofseismic velocity with depth, which causes
the seismic waves to follow curved travel paths be
tween the focus and recorders. Consider Fig. 2.5.
The dotted line represents the continuation of the
fault plane, and its intersection with the Earth's
surface would represent the line separating compres-
Compression
Nodal plane
Fault planp^:Focal sphei
Fig. 2.5 Distribution of compressional and dilational first
arrivals from an earthquake on the surface of a spherical
Earth in which seismic velocity increases with depth.
INTERNAL STRUCTURE OF THE EARTH 11
sional and dilational first motions ifthe waves gener
ated by the earthquake followed straight line paths.
The actual travel paths, however, are curved and the
surface intersection of the dashed line, correspond
ing to the path that would have been followed by a
wave leaving the focus in the direction of the fault
plane, represents the actual nodal plane.
It is clear, then, that simple mapping of compres
sional and dilational first motions on the Earth's
surface cannot readily provide the focal mechanism
solution. However, the complications can be over
come by considering the directions in which the
seismic waves left the focal region, as it is apparent
that compressions and dilations are restricted to
certain angular ranges.
A focal mechanism solution is obtained firstly by
determining the location ofthe focus by the method
outlined in Section 2.1.4. Then, for each seismo
graph recording the earthquake,a standard model for
the velocity structure ofthe Earth is used to compute
the travel path of the seismic wave from focus to
seismograph, and hence to calculate the direction in
which the wave left the focal region. These directions
are then plotted, using an appropriate symbol for
compressional or dilational first motion,on an equal
area projection of the lower half of the focal sphere.,
that is, an imaginary sphere of small but arbitrary
radius centred on the focus (Fig. 2.5). An equal area
net, which facilitates such a plot, is illustrated in
Fig. 2.6. The scale around the circumference ofsuch
a net refers to the azimuth,or horizontal component
ofdirection, while dips are plotted on the radial scale
from 0° at the perimeter to 90° at the centre. Planes
through the focus are represented on such plots by
great circles with a curvature appropriate to their dip;
hence a diameter represents a vertical plane.
Let us assume that,for a particular earthquake,the
fault motion is strike-slip along a near vertical fault
plane. This plane and the auxiliary plane plot as
orthogonal great circles on the projection ofthe focal
sphere, as shown on Fig. 2.7. The lineation defined
by the intersection of these planes is almost vertical,
so it is apparent that the direction of movement
along the fault is orthogonal to this intersection,that
is, near horizontal. The two shaded and two un
shaded regions ofthe projection defined by the nodal
planes now correspond to the directions in which
compressional and dilational first motions, respec
tively, left the focal region. A focal mechanism solu
tion is thus obtained by plotting all the observational
12 CHAPTER 2

270°

180° Fig.2.6 Lambert equal area net.

data on the projection of the focal sphere and then

fitting a pair of orthogonal planes which best divide
the area of the projection into zones of compres
sional and dilational first motions. The more seismo
graphs recording the earthquake, the more closely
defined will be the nodal planes.

2.1.7 Ambiguity in focal mechanism

solutions

It is apparent from Fig. 2.7 that the same distribution

Fig.2.7 Ambiguity in the focal mechanism solution of a
strike-slip fault. Regions ofcompressional first motions are
shaded.
of compressional and dilational quadrants would be
obtained if either nodal plane represented the actual
fault plane. Thus, the same pattern of first motions
would be obtained for sinistral motion along a
north-south plane as for dextral motion along an
east-west plane.
In Fig. 2.8a an earthquake has occurred because of
faulting along a westerly dipping thrust plane f^. fj
and its associated auxiliary plane ap^ divide the re
gion around the focus into quadrants which experi-
 INTERNAL STRUCTURE OF THE EARTH 13

Fig.2.8 Ambiguity in the focal

mechanism solution of a thrust
fault. Shaded areas represent
regions ofcompressional first
motions(C), unshaded areas
represent regions of dilational first
motions(D).f refers to a fault
plane, ap to an auxiliary plane.
Changing the nature ofthe nodal
planes as in (a) and (c) does not
alter the pattern of first motions
shown in (b), the projection ofthe
lower hemisphere of the focal
sphere.

ence either compression or dilation as a result ofthe
fault movement. The directions in which compres
sional first motions and C2 and dilational first
motions and D2leave the focus are shown and C2
and D2 are plotted on the projection of the focal
sphere in Fig. 2.8b, on which the two nodal planes
are also shown. Because Fig. 2.8a is a vertical section,
the first motions indicated plot along an east-west
azimuth. Arrivals at seismographs at other azimuths
would occupy other locations within the projection
space. Consider now Fig. 2.8c, in which plane api
becomes the fault plane f2 and f^ the auxiliary plane
ap2. By considering the movement along the thrust
plane it is obvious that the same regions around the
fault are compressed or dilated, so that an identical
focal sphere projection is obtained. Similar results are
obtained when the faulting is normal (Fig. 2.9). In
theory the fault plane can be distinguished by making
use of Anderson's simple theory offaulting (Section
2.10.2) which predicts that normal faults have dips
of more than 45® and thrusts less than 45®. Thus,fj
is the fault plane in Fig. 2.8 and f2 the fault plane in
Fig. 2.9.
It is apparent that the diflferent types of faulting
can be identified in a, focal mechanism solution by
the distinctive pattern of compressional and dila
tional regions on the resulting focal sphere. Indeed,
it is also possible to differentiate earthquakes which
14 CHAPTER 2
(a)
have originated by a combination offault types,such
as dip-slip accompanied by some strike-slip move
ment. The precision with which the directions ofthe
nodal planes can be determined is dependent upon
the number and distribution ofseismographs record
ing arrivals from the event. It is not possible, how
ever, to distinguish the fault and auxiliary planes.
At one time it was believed that distinction be
tween the nodal planes could be made on the basis of
the pattern ofS wave arrivals. P waves radiate into all
four quadrants of the source region as shown in
^ Fig.2.9 Ambiguity in the focal
mechanism solution of a normal
fault. Legend as for Fig. 2.8.
Fig. 2.10a. However,for this simple model, which is
known as a type I, or single-couple, source,S waves,
whose corresponding ground motion is shearing,
should be restricted to the region of the auxiliary
plane(Fig. 2.10b),Recording ofthe S wave radiation
pattern should then make it possible to determine
the actual fault plane. It was found, however, that
instead of this simple pattern, most earthquakes
produce S wave radiation along the direction ofboth
nodal planes (Fig. 2.10c). This observation initially
cast into doubt the validity of the elastic rebound

(a) (b)
max
Fig.2.10 (a)P wave radiation pattern for a type I and
type II earthquake source mechanism;(b)S wave radiation
pattern from a type I source (single couple);(c)S wave
radiation pattern from a type II source (double couple).
theory. It is now realized, however, that faulting
occurs at an angle, typically rather less than 45®, to
the maximum compressive stress, and the bi
sectors of the dilational and compressional quad
rants, termed P and T, respectively, approximate to
the directions of maximum and minimum principal
compressive stress, thus giving an indication of the
stress field giving rise to the earthquake (Fig. 2.10c).
This type II, or double-couple, source mechanism
gives rise to a four-lobed S wave radiation pattern
(Fig. 2.10c) which cannot be used to resolve the
ambiguity ofa focal mechanism solution. Generally,
the only constraint on the identity of the fault plane
comes from a consideration of the local geology in
the region ofthe earthquake.
2.1.8 Seismic tomography
Tomography is a technique whereby three-
dimensional images are derived from the processing
ofthe integrated properties of the medium that rays
encounter along their paths through it. Tomography
is perhaps best known in its medical applications, in
which images of specific plane sections of the body
INTERNAL STRUCTURE OF THE EARTH 15
are obtained using X-rays. Seismic tomography refers
to the derivation of the three-dimensional velocity
structure of the Earth from seismic waves. It is
considerably more complex than medical tomo
graphy in that the natural sources of seismic waves
(earthquakes) are of uncertain location, the propaga
tion paths of the waves are unknown and the re
ceivers (seismographs) are of restricted distribution.
These difficulties can be overcome, however, and
since the late 1970s seismic tomography has pro
vided important new information on Earth struc
ture. The method was first described by Aki et aL
(1977) and has been reviewed by Dziewonski &
Anderson (1984)and Thurber & Aki(1987).
Seismic tomography makes use of the accurately
recorded travel times of seismic waves from geo
graphically distributed earthquakes at a distributed
suite of seismographs. The many different travel
paths from earthquakes to receivers cross each other
many times. If there are any regions of anomalous
seismic velocity in the space traversed by the rays, the
travel times of the waves crossing this region are
aflfected. The simultaneous interpretation of travel
time anomalies for the many criss-crossing paths
then allows the anomalous regions to be delineated,
providing a three-dimensional model of the velocity
space.
Both body waves and surface waves(Section 2.1.3)
can be used in tomographic analysis. With body
waves, the actual travel times of P or S phases are
utilized. The procedure with surface waves is more
complex,however,as they are dispersive; that is, their
velocity depends upon their wavelength. The depth
of penetration of surface waves is also wavelength-
dependent, with the longer wavelengths reaching
greater depths. Since seismic velocity generally in
creases with depth, the longer wavelengths travel
more rapidly. Thus, when surface waves are utilized,
it is necessary to measure the phase or group velo
cities of their different component wavelengths.
Because of their low frequency, surface waves pro
vide less resolution than body waves. However, they
sample the Earth in a diflTerent fashion and, since
either Rayleigh or Love waves(Section 2.1.3) may be
used, additional constraints on shear velocity and its
anisotropy are provided.
The normal procedure in seismic tomography is to
assume an initial model ofthe velocity space in which
it is assumed to be radially symmetrical. The travel
time ofa body wave from earthquake to seismograph
16 CHAPTER 2

is then equal to the sum of the travel times through Sensors

I I
the individual elements of the model. Any lateral
velocity variations within the model are then re
flected in variations in arrival times with respect to
CO
D
I Vi

the mean arrival time of undisturbed events. Simi

U
/7
f
M V2

larly, the dispersion of surface waves across a hetero

geneous model differs from the mean dispersion
Initial
model
0)

c / : \
V3

through a radially symmetrical model. The method

makes use of a simplifying assumption based on
Fermat's Principle, which assumes that the ray paths
for a radially symmetrical and laterally variable veloc
ity model are identical ifthe heterogeneities are small
and that the diflferences in travel times are caused
solely by heterogeneity in the velocity structure of
the travel path. This obviates the necessity of com
puting the new travel path implied by refractions at
the velocity perturbations.
There are two main approaches to seismic tomo
graphy depending upon how the velocity heterogen
eity of the model is represented. Local methods make
use of body waves and subdivide the model space
into a series of discrete elements so that it has the
form of a three-dimensional ensemble of blocks. A
set of linear equations is then derived which link the
anomalies in arrival times to velocity variations over
the different travel paths. A solution ofthe equations
can then be obtained, commonly using matrix inver
sion techniques, to obtain the velocity anomaly in
each block. Global tnethods express the velocity vari
ations of the model in terms ofsome linear combin
ation ofcontinuous basic functions,such as spherical
harmonic functions.
LxDcal methods can make use of either teleseismic
or local events. In the teleseismic method (Fig. 2.11)
a large set of distant seismic events is recorded at a
network ofseismographs over the volume ofinterest.
Because of their long travel path, the incident wave-
fronts can be considered planar so that deviations
from expected arrival times are caused by velocity
variations beneath the network. In practice, devi
ations from the mean travel times are computed to
compensate for any extraneous eflfects experienced by
the waves outside the volume of interest. Inversion
of the series of equations of relative travel time
through the volume then provides the relative velo
city perturbations in each block of the model. The
method can be extended by the use of a world-wide
distribution of recorded teleseismic events to model
the whole mantle.
Local events can be used, with the seismic sources
E
0 Vi/ L..
\
\
V4
a
/
Q.
D / .V5
Iv-
Standard earth
Fig.2.11 Geometry ofthe teleseismic inversion method.
Velocity anomalies within the compartments are derived
from relative arrival time anomalies of teleseismic events
(redrawn from Aki et al.^ 1977, with permission from the
American Geophysical Union).
located within the volume of interest (Fig. 2.12). In
this case the location and time of the earthquakes
must be accurately known, and ray-tracing methods
used to construct the travel paths of the rays. The
inversion procedure is then similar to that for
teleseisms.
Global methods commonly make use of both
surface and body waves with long travel paths. If the
Earth were spherically symmetrical, these surface
waves would follow great circle routes. However,
again making use ofFermat's Principle, it is assumed
that ray paths in a heterogeneous Earth are similarly
great circles, with anomalous travel times resulting
from the heterogeneity. In the single-station con
figuration, the surface wave dispersion is measured
for the rays travelling directly from earthquake to
receiver. Information from only moderate-size
events can be utilized, but the source parameters
have to be well known. The great circle method uses
multiple circuit waves, that is, waves which have
Sensor
/'
Vi
V2
V3
*
Earthqijake V4
Fig.2.12 Geometry ofthe local inversion method.

Fig. 2.13 Great circle paths from two earthquakes (stars)
to recording stations (dots)(after Thurber & Aki, 1987).
travelled directly from source to receiver and have
then circumnavigated the Earth to be recorded again
(Fig. 2.13). Here the differential dispersion between
the first and second passes is measured, eliminating
any undesirable source effects. This method is appro
priate to global modelling, but can only use those
large magnitude events which give observable mul
tiple circuits.
2.2 Velocity structure of the Earth
Knowledge of the internal layering of the Earth has
been largely derived using the techniques of earth
quake seismology. The shallower layers have been
studied using local arrays of recorders, while the
deeper layers have been investigated using global
Fig. 2.14 Reduced time-distance
relationship for direct waves (Pg)
and waves critically refracted at the
Moho(P„)from an earthquake
INTERNAL STRUCTURE OF THE EARTH 17
networks to detect seismic signals that have traversed
the interior of the Earth.
The continental crust was discovered by Mohorov-
icic from studies of the seismic waves generated by
the Croatia earthquake of 1909(Fig. 2.14). Within a
range of about 200 km from the epicentre, the first
seismic arrivals were P waves which travelled directly
from the focus to the recorders with a velocity of
5.6 km s"^. This seismic phase was termed Pg. At
greater ranges, however, P waves with the much
higher velocity of7.9 km s"^ became the first arrivals,
termed the P^ phase. These data were interpreted by
the standard techniques of refraction seismology,
with P„ representing seismic waves which had been
critically refracted at a velocity discontinuity at a
depth of some 54 km. This discontinuity was sub
sequently named the Mohorovicic discontinuity, or
Moho,and it marks the boundary between the crust
and mantle. Subsequent work has demonstrated that
the Moho is universally present beneath continents
and marks an abrupt increase in seismic velocity to
about 8kms"^. Continental crust is, on average,
some 35 km thick, but thins to less than 20 km
beneath some tectonically active areas and thickens
to up to 80 km beneath young fold mountain belts.
A discontinuity within the continental crust
was discovered by Conrad in 1925 using similar
200
X, Distance(km)
■'^5.6 km s ^ '54 km
sSsMohoss
fis km s
18 CHAPTER 2
200 300
X, Distance(km)
methods. As well as the phases Pg and P^ he noted
the presence of an additional phase P* (Fig. 2.15)
which he interpreted as the critically refracted arrival
from an interface where the velocity increased from
about 5.6 to 6.3 km s"^. This interface was subse
quently named the Conrad discontinuity. Conrad's
model was readily adopted by early petrologists who
believed that two layers were necessarily present in
the continental crust. The upper layer, rich in silicon
and aluminium, was called the SIAL and was be
lieved to be the source ofgranitic magmas, while the
lower, silicon- and magnesium-rich layer or SIMA
was believed to be the source of basaltic magmas.It is
now known, however, that the upper crust has a
composition more basic than granite, and that the
majority of basaltic magmas originate in the upper
mantle. Consequently,the petrological necessity ofa
two-layered crust no longer exists and, where applic
able, it is preferable to use the terms upper and lower
crustal layer. Unlike the Moho,the Conrad disconti
nuity is not always present within the continental
crust, although the seismic velocity generally in
creases with depth. There are some regions, however,
where it is believed that a low velocity layer may be
present at the base of the upper crust (Mueller &
Landisman, 1966).
The oceanic crust has principally been studied by
explosion seismology. The Moho is always present
and the thickness of much of the oceanic crust is
remarkably constant at about 7km irrespective ofthe
Fig.2.15 Reduced time-
distance relationship for direct
500
waves (Pg), waves critically
refracted at the Conrad
^ 5.6 km 8"^ discontinuity(P*)and waves
Conrad critically refracted at the Moho
/-''"/Xi-; 6.3 km s"^
(P„)from an earthquake
source.
depth of water above it. The internal layering of
oceanic crust and its constancy over very wide areas
will be discussed later (Section 2.4.4).
In studying the deeper layering of the Earth, seis
mic waves with much longer travel paths are em
ployed. The velocity structure has been built up by
recording the travel times ofbody waves over the full
range of possible epicentral angles. By assuming that
the Earth is radially symmetrical, it is possible to
invert the travel time data to provide a model of the
velocity structure. A modern determination of the
velocity-depth curve (Hart et 1977) for both P
and S waves is shown in Fig. 2.16.
Velocities increase abruptly at the Moho in both
continental and oceanic environments. A low velo
city zone (LVZ) is present between about 80 and
300 km depth. This appears to be universally present
for S waves, but may be absent in certain regions for
P waves, especially beneath ancient shield areas.
Between 400 and about 670 km velocity increases
rapidly in a stepwise fashion within the mantle trans
ition zone, which separates the upper mantle from
the lower mantle. Each velocity increment probably
corresponds to a mineral phase change to a denser
form at depth (Section 2.8.5). Both P and S velocities
increase progressively in the lower mantle.
The Gutenberg discontinuity marks the core-
mantle boundary at a depth of2891 km,at which the
velocity of P waves decreases abruptly. S waves are
not transmitted through the outer core, which is

Crust
Upper rriaritle- pllll"J"
Transition zone
Fig.2.16 Variation ofP and S
wave velocities with depth in the
Earth (based on the data of Hart et
«/., 1977) and their correlation
with compositional and rheological
layering.
consequently believed to be in a fluid state. The
geomagnetic field (Section 3.6.4) is believed to
originate by the circulation of a good electrical con
ductor in this region. At a depth of 5150 km the P
velocity increases abruptly and S waves are once again
transmitted. This inner core is thus believed to be
solid as a result of the enormous confining pressure.
There appears to be no transition zone between inner
and outer core, as was originally believed.
2.3 Composition of the Earth
All bodies in the solar system are believed to have
been formed by the condensation and accretion of
the primitive interstellar material which made up the
solar nebula. The composition ofthe Sun is the same
as the average composition of this material. Gravita
tional energy was released during accretion, and
together with the radioactive decay of short-lived
radioactive nuclides eventually led to heating of the
proto-Earth so that it differentiated into a radially
symmetrical body made up of a series ofshells whose
density increased towards its centre. The differentia
tion prevents any estimate being made ofthe overall
composition of the Earth by direct sampling. How
ever, it is believed that meteorites are representatives
of material within the solar nebula and that esti
mates of the Earth's composition can be made from
them. The presence of metallic and silicate phases in
meteorites is taken to indicate that the Earth consists
ofan iron/nickel core surrounded by a lower density
silicate mantle and crust.
INTERNAL STRUCTURE OF THE EARTH 19
Velocity(km s
1.1 ithngnhprp
iiiiif'-vz
I -Astheno-
sphere
j: MesoH;
Sspherei;
Depth (km)
Seismic data, combined with knowledge of the
mass and moment of inertia of the Earth, have
revealed that the mean atomic weight ofthe Earth is
about 27, with a contribution of 22.4 from the
mantle and crust and 47.0 from the core. No single
type of meteorite possesses an atomic weight of 27,
the various types ofchondrite being somewhat lower
and iron meteorites considerably higher. However,it
is possible to mix the proportions of different met
eorite compositions in such a way as to give both the
correct atomic weight and core/mantle ratio. Three
such models are given in Table 2.1.
It is apparent that at least 90% ofthe Earth is made
up ofiron, silicon, magnesium and oxygen, with the
bulk of the remainder comprising calcium, alumin
ium, nickel, sodium and possibly sulphur.
2.4 The crust
2.4.1 The continental crust
Only the uppermost part of the crust is available for
direct sampling at the surface or from boreholes. At
greater depths within the crust virtually all informa
tion about its composition is indirect. Much of our
information about the crust has been derived from a
knowledge ofthe variation of seismic velocities with
depth and how these correspond to experimental
determinations of velocities measured over ranges of
temperature and pressure consistent with crustal
conditions. Pressure increases with depth at a rate of
about 30MPa km"\ mainly due to the lithostatic
20 CHAPTER 2

Table 2.1 Estimates ofthe bulk composition ofthe Earth

2.4.2 Upper continental crust
and Moon (in weight percent)(from Condie, 1982 with
permission). Past theories of crustal construction suggested that
the upper continental crust was made up of rocks of
Earth Moon
granitic composition. That this is not the case is
1 2 3 4 evident from the widespread occurrence of large
negative gravity anomalies over granite plutons,
Fe 34.6 29.3 29.9 9.3
30.7 30.9 42.0
which demonstrate that their density(about 2.67 Mg
O 29.5
Si 15.2 14.7 17.4 19.6
m"^) is some 0.10-0.15 Mg m"^ lower than the
Mg 12.7 15.8 15.9 18.7 average value ofthe upper crust. The mean composi
Ca 1.1 1.5 1.9 4.3 tion of the upper crust can be estimated, albeit with
Al 1.1 1.3 1.4 4.2 some uncertainty due to biasing, by determining the
Ni 2.4 1.7 1.7 0.6 mean composition of a large number of samples
Na 0.6 0.3 0.9 0.07 collected world-wide and from analyses of sedimen
S 1.9 4.7 -
0.3 tary rocks which have sampled the crust naturally by
the process of erosion (Taylor & Scott, 1985). This
1: 32.4% iron meteorite (with 5.3% FeS) and 67.6%
oxide portion of bronzite chondrites. composition corresponds to a rock type between
2:40% type I carbonaceous chondrite, 50% ordinary granodiorite and diorite.
chondrite, and 10% iron meteorite (containing 15%
sulphur).
2.4.3 Lower continental crust
3: Non-volatile portion oftype I carbonaceous chondrites
with FeO/FeO + MgO of0.12 and sufficient Si02 The velocity range of the lower crust from 6.5 to
reduced to Si to yield a metal/silicate ratio of 32/68. 7.6 km s~^ cannot be explained by a simple increase
4: Based on Ca, Al, Ti = 5 x type I carbonaceous of seismic velocity with depth. Consequently, either
chondrites, FeO = 12% to accommodate lunar density,
the chemical composition must be more basic, or
and Si/Mg = chondritic ratio.
denser,high pressure phases are present. Early crustal
models suggested a basaltic composition for the
confining pressure ofthe overlying rocks, but also,in lower crust, as it was believed that basaltic magmas
some regions, with a contribution from tectonic were derived from this level. Although this would be
forces. Temperature increases at an average rate of in accord with experimentally derived P wave veloci
about 25 km~^, but decreases to about half this ties for rocks of basaltic composition, investigations
value at the Moho because of the presence of radio of the mineral assemblages present under the
active heat sources within the crust (Section 2.13). physical conditions of the lower crust reveal that
Seismic velocities vary with depth because of a basalt, as such, cannot be present. Green & Ring-
number of factors. The increase of pressure with wood (1967) showed that as pressures increased to
depth causes a rapid increase in incompressibility, 2100 MPa at a temperature of 1100®C under dry
rigidity and density over the topmost 5 km as pores conditions, the amount of garnet increases and the
and fractures are closed. Thereafter the increase of plagioclase content decreases so that basalt trans
these parameters with pressure is balanced by the forms to eclogite via the intermediate stage ofgarnet
decrease resulting from thermal expansion with in granulite. The velocity of P waves in eclogite is
creasing temperature so that there is little further 8 km s~^, considerably higher than that observed in
change in velocity with depth. Velocities change with the lower crust.
chemical composition, and also with changes in If the lower crust is dry, its composition must be
mineralogy resulting from phase changes. Abrupt more acidic than basalt and corresponds to high
velocity discontinuities are usually caused by changes pressure forms ofrocks ranging in composition from
in chemical composition, while more gradational granodiorite to diorite (Christensen & Fountain,
velocity boundaries are normally associated with 1975; Smithson & Brown, 1977). An alternative
phase changes which occur over a discrete vertical suggested by Green (1970) is that the lower crust is
interval. ofgabbroic anorthosite composition, which is also of

the correct velocity range. Dewey & Burke (1973)
have used this idea to explain the development of
anorthositic lower crust in terms of a continent-
continent collision (Section 9.3.2) wherein the crust
is thickened, heat flow increases and granitic melts
migrate upwards, leaving a refractory lower crust.
The Conrad discontinuity is envisaged as the top of
this layer, and it is interesting to speculate that,
according to this model, the Conrad discontinuity
may only be developed in terrains which have suf
fered orogenesis of this type.
Ifthe lower crust is wet, basaltic rocks would occur
in the form ofamphibolite. If mixed with more silicic
material, this would have a seismic velocity in the
correct range.
Information on the lower crust has also been
provided by estimates of Poisson's ratio, which can
be measured from the P and S wave velocities (Zandt
&Ammon,1995).The mean value of-^0.30 strongly
suggests a dry lower crust of mafic composition.
Undoubtedly, the lower crust is compositionally
more complex than suggested by these simple
models, and studies of deep crustal xenoliths and
crustal contaminated magmas (Kay & Kay, 1981)
indicate that there must be significant regional varia
tions in its composition, age and thermal history.
Deep seismic reflection investigations (Oliver, 1982)
have shown that this compositional complexity is
matched by complex structure.
2.4.4 The oceanic crust
The oceanic crust (Francheteau, 1983)is in isostatic
equilibrium with the continental crust according to
the Airy mechanism (Section 2.11.2), and is con
sequently much thinner. Seismic refraction studies
have confirmed this and show that the oceanic crust
is 6-7 km thick beneath an average water depth of
4.5 km.
The earliest refraction surveys produced time-
distance data of relatively low accuracy which, on
simple inversion using plane-layered models, indic
ated the presence ofthree principal layers which were
found ubiquitously in the ocean basins. The veloci
ties and thicknesses of these layers are shown in
Table 2.2. More recent refraction studies, employing
much more sophisticated equipment and interpreta-
tional procedures (Kennett, 1977), have shown that
further subdivision of the main layers is possible
INTERNAL STRUCTURE OF THE EARTH 21
Table 2.2 Oceanic crustal structure (after Bott, 1982a
with permission from Edward Arnold).
P velocity Average thickness
(km s"^) (km)
Water 1.5 4.5
Layer 1 1.6-2.5 0.4
Layer 2 3.4-6.2 1.4
Layer 3 6.4-7.0 5.0
Moho
Upper mantle 7.4-8.6
(Harrison & Bonatti, 1981) and that, rather than a
structure in which velocities increase downwards in
discrete jumps, there appears to be a progressive
velocity increase with depth (Kennett & Orcutt,
1976; Spudich & Orcutt, 1980). Figure 2.17 com
pares the velocity structure of the oceanic crust as
determined by early and more recent investigations.
2.4.5 Oceanic layer 1
Layer I has been extensively sampled by coring and
drilling. Sea bed surface materials comprise uncon-
solidated deposits including terrigenous sediments
carried into the deep oceans by turbidity currents,
and pelagic deposits such as brown zeolite clays,
calcareous and silicic oozes and manganese nodules.
These deep sea sediments are frequently redistrib
uted by bottom currents or contour currents, which
are largely controlled by thermal and haline anoma
lies within the oceans. The dense, cold saline water
produced at the poles sinks and underflows towards
equatorial regions, and is deflected by the Coriolis
force. The resulting currents give rise to sedimentary
deposits which are termed contourites(Stow & Lovell,
1979).
Layer I is on average 0.4 km thick. It progressively
thickens away from the ocean ridges, where it is thin
or absent. There is, however, a systematic difference
in the sediment thicknesses of the Pacific and
Atlantic/Indian oceans, as the former is rimmed by
trenches which trap sediments ofcontinental origin,
and the latter are not, allowing greater terrestrial
input. The interface between layer I and layer 2 is
considerably more rugged than the sea bed, because
22 CHAPTER 2
Velocity (km s~^)
4 6 8 I—--—izTd
I 1 1 1
Depth
below
basement
(km) 4 (L3) 6.7
n 11
Anomalous mantle
(Close to ridge axis)
Fig.2.17 P and S wave velocity structure ofthe oceanic
crust and its interpretation in terms oflayered models
proposed in 1965 and 1978. Numbers refer to velocities
in km s"^. Dashed curve refers to gradational increase in
velocity with depth deduced from more sophisticated
inversion techniques (after Spudich & Orcutt, 1980 and
Harrison & Bonatti, 1981).
of the volcanic nature of layer 2. Within layer 1 are
a number of horizons which show up as prominent
reflectors on seismic reflection records. Edgar(1974)
has described the acoustic stratigraphy in the North
Atlantic, where up to four suprabasement reflectors
are found (Fig. 2.18). Horizon A corresponds to an
Eocene chert, although deep sea drilling indicates
that it maintains its reflective character even when
little or no chert is present. In such locations it may
correspond to an early Tertiary hiatus beneath the
chert. Horizon A* occurs beneath A,and represents
the interface between Late Cretaceous/Early Ter
tiary metal-rich clays and underlying euxinic black
clays. Horizon P represents the base of the black
clays, where they overlie a Late Jurassic/Lower Cre
taceous limestone. Horizon B may represent a sedi
mentary horizon,although it has also been identified
as basalt similar to that at the top oflayer 2.
Reflectors similar to A and B have been identified
in the Pacific and Caribbean, where they are termed
A', B' and A",B", respectively.
1965
Sediment
(L2A) 3. Fractured basalt
Massive basalt
(L2) 5.0 (L2B) 5.2 with dykes
Dykes with
(L2C) 6.1 massive basalt
Metagabbros and gabbros
(L3A) 6.8 with pockets of plagiogranite
and protrusions of serpentinite
Gabbros and metagabbros
(L3B) 7.3 with serpentinite protrusions
and pockets of cumulate
ultramafic
Mantle
8.1 Serpentinized ultramafics
Harzburgite and Iherzolite
2.4.6 Oceanic layer 2
Layer 2 is variable in its thickness, in the range
1.0-2.5 km. Its seismic velocity is similarly variable
in the range 3.4-6.2 km s~^. This range is attribut
able to either consolidated sediments or extrusive
igneous material. Direct sampling and dredging of
the sediment-free crests of ocean ridges, and the
necessity of a highly magnetic lithology at this level
(Section 4.2), overwhelmingly prove an igneous
origin. The basalts recovered are olivine tholeiites
containing calcic plagioclase, and are poor in potas
sium,sodium and the incompatible elements(Sun et
aL^ 1979). They exhibit very little areal variation in
composition,with the exception oflocations close to
oceanic islands (Section 5.4).
Three subdivisions of layer 2 have been recog
nized. Sublayer 2A is only present on ocean ridges
near eruptive centres in areas affected by hydro-
thermal circulation ofsea water, and ranges in thick
ness from zero to 1 km.Its porous, rubbly nature, as
indicated by a P wave velocity of 3.6 km s"^, permits
such circulation. The very low velocities(2.1 km s"^)
of the top of very young layer 2 located on the
Mid-Atlantic Ridge (Purdy, 1987) probably indicate
a porosity of 30-50%, and the much higher veloci
ties of older layer 2 imply that the porosity must be
reduced quite rapidly after its formation.Sublayer 2B
forms the normal acoustic basement oflayer 1 when

(a)
North America Atlantic Ocean
North American Basin
Continental margin
(b) Pleistocene sand and clay Site 105
Horizon A.
Horizon A*
Horizon0
Basement
Fig.2.18 (a) Major seismic reflectors in the western
Atlantic Ocean,(b) Corresponding lithoiogies determined
by deep sea drilling (after Edgar, 1974, with permission
from Springer-Verlag).
sublayer 2A is not developed. Its higher velocity of
4.8-5.5 km s~^ suggests a louver porosity. With time
layer 2A may be converted to layer 2B by the infilling
of pores by secondary minerals such as calcite, quartz
and zeolites.
Sublayer 2C is about 1 km thick, where detected,
and its velocity range of5.8-6.2 km s~^ may indicate
a high proportion ofintrusive, basic rocks. This layer
grades downwards into layer 3.
2.4.7 Oceanic layer 3
Layer 3 is the main component of the oceanic crust
and represents its plutonic foundation (Fox &
Stroup, 1981). Some workers have subdivided it
into sublayer 3A, with a velocity range of 6.5-
6.8 km s"^, and a higher velocity lower sublayer 3B
(7.0-7.7 km s"^) (Christensen & Salisbury, 1972),
although the majority of seismic data can be ex
plained in terms of a layer with a slight positive
velocity gradient (Spudich & Orcutt, 1980).
Hess (1962) suggested that layer 3 was formed
from upper mantle material whose olivine had
reacted with water to varying degrees to produce
serpentinized peridotite, and, indeed, 20-60% ser-
pentinization can explain the observed range of P
INTERNAL STRUCTURE OF THE EARTH 23
Mid-Atlantic Ridge
Acoustic Basement
/Tertiary hemipelagic mud
— — ^^Upper Cretaceous-
Lower Tertiary multicoloured clay
Cretaceous black clay
Late Jurassic and Neocomian limestone
Basalt
wave velocities. This notion can now be discounted,
however, as the value ofPoisson's ratio for layer 3A,
which can be estimated directly from a knowledge of
both P and S wave velocities, is much lower than
would be expected for serpentinized peridotite. In
fact, Poisson's ratio for layer 3A is more in accord
with a gabbroic composition, which also provides
seismic velocities in the observed range, and gabbros
have been encountered in this level in one DSDP
borehole. It is possible, however, that all or at least
part of layer 3B, where recognized, consists of ser
pentinized ultramafic material.
The concept ofa predominantly gabbroic layer 3 is
in accord with models suggested for the origin of
oceanic lithosphere (Section 6.10). These propose
that layer 3 forms by the crystallization of a magma
chamber, with an upper layer, possibly correspond
ing to sublayer 3A, of isotropic gabbro and a lower
layer, possibly corresponding to 3B, consisting of
cumulate gabbro and ultrabasic rocks formed by
crystal settling. This layering has been confirmed by
direct observation and sampling by submersible on
the Vema Fracture Zone in the North Atlantic
(Auzende et al., 1989).
Recent in situ evidence from an ODP hole on the
Costa Rica rift in the eastern Pacific has shown, at
least for this location, that the layer 2/3 seismic
boundary lies within the basalt and is associated with
gradual changes in porosity and alteration (Detrick^^
uL 1994).
The origin of the relatively small amounts of
24 CHAPTER 2

oceanic crust which do not conform to the standard
layering is discussed in Section 7.3.
2.5 Ophiolites
The study of oceanic lithosphere has been aided by
investigations of characteristic rock sequences on
land known as ophiolites (literally'snake rock', refer
ring to the similarity of the colour and texture to
snakeskin; see Nicolas, 1989, for a full treatment of
this topic). Ophiolites usually occur in collisional
mountain belts, and their association of deep sea
sediments, basalts, gabbros and ultrabasic rocks sug
gests that they originated as oceanic lithosphere and
were subsequently thrust up into their continental
setting by a process known as obduction (Dewey,
1976; Ben-Avraham et 1982; see also Section
9.5). The complete ophiolite sequence (Gass, 1980)
is shown in Table 2.3. The analogy ofophiolites with
oceanic lithosphere is supported by the gross simil
arity in chemistry (although there is considerable
difference in detail), metamorphic grades corre
sponding to temperature gradients existing under
spreading centres, the presence of similar ore miner
als and the observation that the sediments were
formed in deep water (Moores, 1982). Salisbury &
Christensen (1976) have compared the velocity
structure of the oceanic lithosphere with seismic
velocities measured in samples from the Bay of
Islands ophiolite complex in Newfoundland, and
concluded that the determined velocity stratigra
phies are identical. Figure 2.19 shows the correlation
between the oceanic lithosphere and three well-
studied ophiolite bodies.
At one time it seemed that investigations of the
petrology and structure of the oceanic lithosphere
could conveniently be accomplished by the study of
ophiolite sequences on land. However, in recent
years this simple analogy has been challenged, and it
has been suggested that ophiolites do not represent
typical oceanic lithosphere, and were not emplaced
exclusively during continental collision (Mason,
1985).
Dating ofevents indicates that obduction of many
ophiolites occurred very soon after their creation.
Continental collision, however, normally occurs a
long time after the formation of a mid-ocean ridge.

Table 2.3 Correlation ofophiolite stratigraphy with the

oceanic lithosphere (after Gass, 1980, with permission
Oceanic Semail Bay of Islands Troodos
from the Ministry of Agriculture and Natural Resources
km lithosphere ophiolite ophiolite ophiolite
Cyprus). 0-

Complete ophiolite sequence Oceanic correlation 2- Sea

Sediments Layer 1
4-
rLayer h
Basic volcanics, commonly| Layer
___

6-
pillowed, merging into ^ Layer 2
Basic sheeted dyke complex J 8-
Layer 3
High level intrusives 10-
S ..
Trondhjemites 1 Layer 3
Gabbros / 12-
Upper
mantle
14-
Layered cumulates
Olivine gabbros| — Moho — I Pillow Sheeted
Pyroxenitcs > 1 Sediment
' lava dykes
Peridotites J I Ultrabasic
Upper mantle
B Gabbro I ,. \yi Peridotite
I cumulates '—'
S-seismological Moho P- petrological Moho
Harzburgite, commonly
serpentinized ±Iherzolite, Fig.2.19 Comparison ofoceanic crustal structure with
dunite, chromititc ophiolite complexes (after Mason, 1985,with permission
from Blackwell Scientific Publications).

so that the age of the sea floor obducted should be
considerably greater than that of the collisional oro
geny. Ophiolites consequently represent lithosphere
that was obducted while young and hot. Geochem-
ical evidence (Pearce, 1980; Elthon, 1991) has sug-
geted that the original sites of ophiolites were
marginal seas (Section 8.9; see also Cawood 8c Suhr,
1992), or Red Sea-type oceans.
There have been many different mechanisms pro
posed for ophiolite obduction, none of which can
satisfactorily explain all cases. It must thus be recog
nized that there may be several operative mechanisms
and that, although certainly formed by some type of
accretionary process, ophiolite sequences may differ
significantly from lithosphere created in the major
ocean basins.
2.6 Metamorphism of oceanic
crust and upper mantle
Many of the rocks sampled from the ocean basins
show evidence of metamorphism, including abun
dant greenschist facies assemblages and alkali meta
somatism: In close proximity to such rocks, however,
are found completely unaltered species.
It is probable that this metamorphism is accom
plished by the hydrothermal circulation of sea water
within the oceanic crust. There is much evidence for
the existence ofsuch circulation, such as the presence
of metalliferous deposits which probably formed by
the leaching and concentration of minerals by sea
water, observations of active hydrothermal vents
Fig. 2.20 Schematic model for
hydrothermal metamorphism ofthe
Magma
oceanic crust at a spreading centre chamber
(redrawn from Elthon, 1981, with
permission from John Wiley 8c
Sons Ltd).
INTERNAL STRUCTURE OF THE EARTH 25
on ocean ridges, and the observed metamorphism
within ophiolite sequences.
Hydrothermal circulation takes place by convec-
tive flow, probably through the whole of the oceanic
crust (Fyfe 8c Lonsdale, 1981), and is of great signi
ficance. It influences models of heat production, as
it has been estimated that approximately 25% of the
heat escaping from the Earth's surface is vented at
the mid-ocean ridges. The circulation must modify
the chemistry of the ocean crust, and consequently
will affect the chemical relationship of lithosphere
and asthenosphere over geological time because of
the recycling oflithosphere that occurs at subduction
zones. It is also responsible for the formation of
certain economically important ore deposits, partic
ularly massive sulphides.
These hydrothermal processes are most conven
iently studied in the metamorphic assemblages of
ophiolite complexes, and the model described below
has been derived by Elthon (1981).
Hydrothermal metamorphism of pillow lavas and
other extrusives gives rise to low temperature
(<230''C) and greenschist facies assemblages
(Fig. 2.20). The distribution of alteration is highly
irregular, and is controlled by the localized fissuring
ofthe extrusive rocks. Higher temperature metamor
phism is widespread within the sheeted dyke com
plex, producing assemblages typical ofthe actinolite
facies, although pockets of unaltered rock do occur.
The highest metamorphic temperatures are achieved
at the base of the sheeted dyke complex and the
upper part ofthe gabbroic section. Rarely, retrograde
Sediments Facies
Pi low lavas
Greenschist
retrograded to
Zeolite
Greenschist
Sheeted dikes
Actinolite
retrograded to
Greenschist
Actinolite
26 CHAPTER 2
rocks of the greenschist facies occur at this level.
Alteration decreases to only about 10% within the
top kilometre of the gabbroic section and thereafter
metamorphism is restricted to the locality of fissures
and dykes, although metamorphism does not com
pletely terminate at depth. According to this model,
sea water circulation occurs extensively in the upper
3 km of the crust, producing the metamorphic as
semblages and cooling the crust. High temperature
metamorphism only occurs near the spreading
centre. At depth the circulation becomes diminished
as secondary minerals are deposited within the flow
channels.
As the ridge spreads continuously, oceanic litho-
sphere is moved laterally from the heat source and
undergoes retrograde metamorphism. This depends
upon an adequate water supply, as water distribution
is the major control of metamorphic grade. The
absence ofsufficient water allows the preservation of
relict high temperature assemblages. The hetero
geneous nature of the distribution of metamorphic
facies is consequently explained by a similarly hetero
geneous distribution ofcirculating fluids rather than
extreme temperature variations.
2.7 Differences between
continental and oceanic crust
On the basis of information presented in this and
following chapters, the major differences between
continental and oceanic crust can be summarized as
follows:
1 Layering. The large-scale layering ofthe continen
tal crust is ill defined and highly variable, reflecting a
complex geological history. In places there is a broad
subdivision by the Conrad discontinuity, but this is
not globally developed. By contrast, the layering of
the oceanic crust is well defined into three distinct
layers which are found in all the ocean basins. How
ever, the nature of these layers, in particular layers 2
and 3, may change quite markedly with depth. In the
case of layers 2 and 3 this reflects the complex
development oftheir parental magma chamber.
2 Thickness, The thickness ofcontinental crust aver
ages about 35-40 km but is quite variable, thinning
to only a few kilometres beneath rifts and thickening
to up to 80 km beneath young mountain belts. The
oceanic crust maintains a relatively constant thick
ness of about 7km, although layer 1 increases in
thickness towards the ocean margins.
3 A£fe. Continental crust may be as old as 3960 Ma,
the age ofthe oldest rocks yet discovered. On a very
broad scale the oldest crust consists of Precambrian
cratons or shield areas which are surrounded by
younger orogenic belts, both active and inactive.
Oceanic crust, however, is nowhere older than
180 Ma, and progressively increases in age outwards
from oceanic ridges (Section 4.3). Oceans are conse
quently viewed as essentially transient features of
the Earth's surface. About 50% ofthe surface area of
the oceans has been created during the last 65 Ma,
implying that 30% of the Earth's surface has been
created during the most recent 1.5% of geological
time.
4 Tectonic activity. Continental crust may be exten
sively folded and faulted and contains evidence of
being subjected to multiple tectonic events. Oceanic
crust, however, appears to be very stable and has
suffered little deformation except at plate margins.
5 Igneous activity. Very little igneous activity occurs
on the great majority of the continental crust. The
only major locations ofactivity are mountain belts of
Andean type (Section 9.2). The activity within the
oceans is very much greater. Ocean ridges and island
arcs are the location of the Earth's most active areas
of volcanic and plutonic activity.
2.8 The mantle
2.8.1 Introduction
The mantle constitutes the largest internal subdiv
ision of the Earth by both mass and volume, and
extends from the Moho at a mean depth of about
21 km to the core-mantle boundary at a depth of
2885 km. It is generally believed to be chemically
homogeneous, apart from the abundances of minor
and trace elements, and formed of silicate minerals.
The mineralogy and structure of the silicates change
with depth and give rise to a transition zone between
400 and 670 km depth, which separates the upper
and lower mantle.
Mantle materials are only rarely brought to the
surface, in ophiolite complexes (Section 2.5), in
kimberlite pipes and as xenoliths in alkali basalts.
Consequently, most of our information about the
mantle is indirect and based on the variation of
seismic velocities with depth combined with studies
of mineral behaviour at high temperatures and pres
sure, and in shock-wave experiments. Geochemical

studies of meteorites and uitramafic rocks are also
utilized in making predictions about the mantle.
2.8.2 Seismic structure of the mantle
The uppermost part of the mantle constitutes a
high velocity lid abut 80 km thick in which seismic
velocities remain constant at a figure in excess of
7.9 km s~^ or increase slightly with depth. This part
of the mantle makes up the lower portion of the
lithosphere (Section 2.12), and its behaviour is inti
mately related to that ofits overlying crust. That the
crust and lithospheric mantle are not decoupled is
indicated by the strong correlation of upper mantle
velocity with overlying crustal type (Condie, 1982).
Beneath the lithosphere lies a low velocity zone
extending from a depth of approximately 100 to
300 km (Fig. 2.16). This appears to be present
beneath most regions ofthe Earth with the exception
of the mantle beneath cratonic areas. From the base
ofthis zone seismic velocities increase slowly until a
major discontinuity is reached at a depth of about
400 km, marking the upper region of the transition
zone. There is a further velocity discontinuity at a
depth of 670 km, the base of the transition zone.
Within the lower mantle velocities increase slowly
with depth until the basal 200 km where gradients
decrease or low velocities are present.
2.8.3 Mantle composition
The fact that much ofthe oceanic crust is made up of
material of a basaltic composition derived from the
upper mantle suggests that the upper mantle is com
posed of either peridotite or eclogite (Harrison &
Bonatti, 1981). The main difference between these
two rock types is that peridotite contains abundant
olivine and less than 15% garnet, whereas eclogite
contains little or no olivine and at least 30% garnet.
Both possess a seismic velocity which corresponds to
the observed upper mantle value of about 8 km s"^.
Several lines ofevidence now suggest very strongly
that the upper mantle is peridotitic. Beneath the
ocean basins the velocity is frequently anisotropic,
with velocities over 15% higher perpendicular to
ocean ridges. This can be explained by the preferred
orientation of olivine crystals, whose long[100]axes
are believed to lie in this direction. None of the
common minerals of eclogite exhibit the necessary
crystal elongation. A peridotitic composition is also
INTERNAL STRUCTURE OF THE EARTH 27
indicated by estimates ofPoisson's ratio from P and S
velocities, and the presence of peridotites in the basal
sections of ophiolite sequences and as nodules in
alkali basalts. The density of eclogites is also rather
too high to explain the Moho topography of iso-
statically compensated crustal structures.
The bulk composition of the mantle can be esti
mated in several ways: by using the compositions of
various uitramafic rock types, from geochemical
computations, from various meteorite mixtures and
by using data from experimental studies. It is neces
sary to distinguish between undepleted mantle and
depleted mantle which has undergone partial melt
ing so that many ofthe elements which do not easily
substitute within mantle minerals have been
removed and combined into the crust. Ringwood
(1975) has suggested a number of possible combin
ations of compositions, the simplest of which con
sists of three parts dunite (a monomineralic rock
containing olivine), which represents fully depleted
mantle, and one part basalt, corresponding to the
partial melt. This hypothetical rock type has been
termed pyrolite, and further work has led to pyrolite
compositions containing more complex combina
tions ofingredients.
Pyrolite can exist in four different forms depend
ing upon the ambient temperature, pressure and
water vapour pressure. The normal form, existing to
a depth of 70 km is pyroxene pyrolite, which trans
forms into garnet pyrolite at depths of 80-350 km.
The other forms can only exist in environments very
rarely encountered within the Earth.
Although estimates of mantle composition vary in
detail, it is generally agreed that at least 90% of the
mantle by mass can be represented in terms of the
oxides FeO, MgO and SiOj. A further 5-10% is
made up ofCaO,Ai203 and Na20,and at least 98%
ofthe mantle comprises these six oxides.
2.8.4 The mantle low velocity zone
The low velocity zone (Fig. 2.16) is characterized by
low seismic velocities, high seismic attenuation and
also, probably, a high electrical conductivity. The
seismic effects are more pronounced for S waves than
for P waves. The low seismic velocities could arise
from a number of different mechanisms, including
an anomalously high temperature, a phase change, a
compositional change,the presence ofopen cracks or
fissures and partial melting. All but the latter appear
28 CHAPTER 2
to be unlikely, and it is generally accepted that the
lower seismic velocities arise because ofthe presence
of molten material. That melting is likely to occur in
this region is supported by the fact that it is at this
level that mantle material most closely approaches its
melting point (Section 2.12, Fig. 2.36).
Only a very small amount of melt is required to
lower the seismic velocity of the mantle to the ob
served values and to provide the observed attenu
ation properties. A liquid fraction of less than 1%
would, if distributed along a network of fissures at
grain boundaries, produce these effects (O'Connell
& Budiansky, 1977). The melt may also be respons
ible for the high electrical conductivity of this zone.
For the partial melting to occur, it is probable that a
small quantity of water is required to lower the
silicate melting point, and this is probably supplied
from the breakdown of hydrous mantle phases. The
base of the low velocity zone is probably controlled
by the availability of water.
The mantle low velocity zone is of major import
ance to plate tectonics as it represents a low viscosity
layer along which relative movements of the litho-
sphere and asthenosphere can be accommodated.
2.8.5 The mantle transition zone
There are two major velocity discontinuities in the
mantle at depths of400 km and 670 km.The former
marks the top ofthe transition zone and the latter its
base. The discontinuities are rarely sharp and occur
over a finite range in depth,so it is generally believed
that they represent phase changes rather than
changes in chemistry. Some workers, however, pro
pose that the transition zone represents a change in
composition between an upper mantle dominated
by Mg2Si04 in the form of olivine, and a lower
mantle more rich in silica composed of MgSiOa with
perovskite structure. It is also possible that large scale
lateral shearing associated with mantle convection
contributes to the seismic signature of the disconti
nuities (Jeanloz, 1993).
High pressure studies have shown that the olivine
in the mantle, which is 89% forsterite ((Mgo.89
^^^0.11)2^104), when subjected to a pressure corres
ponding to a depth of400 km and a temperature of
1600®C, transforms into a distorted form of spinel
structure known as the P-phase, in which the atoms
are more closely packed(Ringwood & Major, 1970).
It is believed that olivine constitutes about 57% by
mass of the mantle at this level and so this phase
transition is probably responsible for the 400 km
seismic discontinuity. A further contribution is the
phase change ofaluminous orthopyroxenes to garnet,
which occurs over the depth range of350-400 km.
The P-phase Mg2Si04 transforms into a spinel
structure over the depth range of 500-600 km, and
at a depth of 670 km the spinel structure breaks
down to a mixture of MgSiOs with perovskite struc
ture and periclase (MgO). This involves an increase
in density(measured at zero pressure) ofabout 11%.
At about the same depth pyrope garnet transforms to
an ilmenite structure and thence to a perovskite
structure at a depth ofabout 800 km.The spinel and
garnet transformations are thus probably responsible
for the 670 km discontinuity. The mineral assem
blages of the upper mantle and transition zone are
summarized in Table 2.4.
A weaker discontinuity at 520 km depth is also of
global extent (Shearer, 1990). It probably represents
the transformation from modified spinel to spinel
structure.
The recognition of the transition zone as repre
senting a series of phase changes rather than com
positional changes has important ramifications for
the mechanism of plate tectonics. It means that
convection currents in the mantle are capable of
crossing the transition zone (Section 11.5.3) if con
vection is to occur across the full vertical extent of
the mantle.
The lower mantle is probably relatively homo
geneous in its mineralogy, having an essentially
perovskite structure. The lowest 200-300 km of the
mantle, however, are characterized by a decrease in
seismic velocity, which is probably related to an
increased temperature gradient at the mantle-core
boundary. The lowest 200-300 km ofthe mantle are
known as the D" layer (Knittle & Jeanloz, 1991).
Laboratory experiments suggest that the liquid iron
of the core reacts with mantle silicates in the D"
layer, with the production of metallic alloys and
non-metallic silicates from perovskite. Earthquake
seismology has shown that these reactions create a
very heterogeneous layer. It is probable that mantle
plumes (Sections 5.4, 5.6) originate in the D"layer.
2.9 The core
The core, a spheroid with a mean radius of 3486 km,
occurs at a depth of2891 km and occupies the centre
 INTERNAL STRUCTURE OF THE EARTH 29

Table 2.4 Possible mineral assemblages ofthe upper mantle and transition zone (after Brown & Mussett, 1981).

Depth in the Zero-pressure

Increasing upper density
depth mantle(km) (Mgm~^)
Olivine Pyroxene Garnet 3.38

c. 400

Spinel (str) Garnet 3.66

r. 670

\ 1 / \
Strontium llmenite (str) Perovskite (str)
plumbate (str) + + 3.99
simple oxides simple oxides

c. 1050

Perovskite (str)+ calcium ferrite (str)

and/or corundum (ilmenite (str))+ 4.27
stishovite + periclase + wiistite (max)

(str), structure; dashed lines indicate alternative routes.

ofthe Earth. The core-mantle boundary(Gutenberg
discontinuity) generates strong seismic reflections
and thus probably represents a compositional inter
face.
The outer core, at a depth of2891-5150 km,does
not transmit S waves and so must be fluid. This is
confirmed by the generation ofthe geomagnetic field
in this region (Section 3.6.4) by dynamic processes
and by the long period variations observed in the
geomagnetic field. The convective motions respons
ible for the geomagnetic field involve velocities of
'^lO'^ma"^, five orders of magnitude greater than
convection in the mantle. A fluid state is also indi
cated by the response ofthe Earth to the gravitational
attraction of the Sun and Moon.
The boundary between the outer core and inner
core at 5150 km depth is sharp, and not represented
by any form of transition zone. The inner core is
believed to be solid for several reasons. Certain oscil
lations of the Earth produced by very large earth
quakes can only be explained by a solid inner core. A
seismic phase has been recognized which travels to
and from the inner core as a P wave, but traverses the
inner core as an S wave. The amplitude of a phase
reflected off the inner core also suggests that it must
have a finite rigidity and thus be a solid.
Shock wave experiments have shown that the ma
jor constituents of both the inner and outer core
must comprise elements of an atomic number
greater than 23, such as iron, nickel, vanadium or
cobalt. Of these elements, only iron is present in
sufficient abundance in the solar system to form the
major part of the core. Again, by considering solar
system abundances, it appears that the core should
30 CHAPTER 2
contain about 4% nickel. This iron-nickel mixture
provides a composition for the outer core which is
8-15% too dense and it must therefore contain a
small quantity of some lighter element. The inner
core, however, has a seismic velocity and density
consistent with a composition of pure iron.
There are several candidates for the light element
present in the outer core, which include silicon,
sulphur, oxygen and potassium (Brett, 1976). Sili
con requires an over-complex model for the forma
tion of the Earth and sulphur conflicts with the idea
that the interior of the Earth is highly depleted in
volatile elements. Oxygen appears to be the most
likely light element as FeO is probably sufficiently
soluble in iron. The presence of potassium is specu
lative, but is interesting in that it would provide a
heat source in the core which would be active over
the whole ofthe Earth's history. It would also help to
explain an apparent potassium deficiency in the Earth
compared to meteorites.
2.10 Deformation in the crust
and mantle
2.10.1 Introduction
.When a solid is stressed, it becomes strained. The
relationship between stress and strain consequently
governs the form of the deformation experienced by
the solid. A comprehensive account of deformation
processes is given by Ranalli (1987).
Stress (a) is defined as the force exerted per unit
area of the rock surface, and is thus a pressure
measured in pascals (Pa). The confining pressure
experienced by rocks with increasing depth in
the crust and mantle increases at a rate of about
33MPa km~^ (0.33 kbarkm"^) depending on the
density ofthe overlying rocks. Any stress acting upon
a surface can be expressed in terms ofa normal stress
perpendicular to the surface and two components of
shear stress in the plane of the surface. The state of
stress within a medium is conveniently specified by
the magnitudes and directions ofthe three principal
pressures which act on the three particular planes in
the medium along which the shear stress is zero.
These are orthogonal and are termed and
<^min5 that is, the maximum, intermediate and mini
mum principal stresses, respectively. In geology,
compressive stresses are normally treated as positive
and tensional stresses negative.
Strain (e) is defined as the fractional change in
dimension of a material when it is subjected to a
stress, for example, the change in volume per unit
volume. Materials act in an elastic manner according
to Hooke's law, in which strain is proportional to
stress, until a critical stress, known as the elastic limit,
is reached. Beyond the elastic limit, brittle materials
deform by fracturing and ductile materials deform
by flow. The elastic limit is a function oftemperature
and pressure, so the behaviour of materials is
governed by these variables.
In considering the deformation of Earth materials
it is important to recognize the importance of the
length of time for which stress is applied (Park,
1983). The behaviour of rocks in the short term is
significantly different from that of the same material
stressed over geologically realistic durations of
months or years. This difference in behaviour arises
from different strain rates. Rocks exhibit higher
strength at high strain rates than at low strain rates.
This is typified by the behaviour of pitch. When
struck with a hammer,that is, subjected to 'instanta
neous'strain, pitch shatters. When left for a period of
months, however, a block of pitch deforms slowly by
flowing. The long-term behaviour ofstrained materi
als is known as creep.
2.10.2 Brittle deformation
Brittle fracture is believed to be caused by progressive
failure along a network of micro- and mesocracks
by the amplification of stress within those cracks
orientated in the most suitable directions. When the
magnitude of an applied tensional stress exceeds the
tensile strength ofthe material, cracks orthogonal to
the tension fail first and extensional fracture occurs.
Below a depth of a few hundred metres all prin
cipal pressures are usually compressional. Any cracks
progressively close with increasing overburden pres
sure, and are probably completely closed when a
depth ofabout 5 km is reached. This implies that the
compressive strength of a material at low confining
pressures is much greater than the tensile strength, as
failure is facilitated by pre-existing lines of weakness.
For example, the compressive strength of granite at
atmospheric pressure is 140 MPa, and its tensile
strength only about 4 MPa. When all cracks are
closed,fracture is controlled by the inherent strength
ofthe material,the normal stress on a plane offailure
and the coefficient of friction on that plane, which
 INTERNAL STRUCTURE OF THE EARTH 31

<5 max
icr mIn
l^int

min • cr max «^max

(b) (c)

Fig.2.21 Three classes offault determined by the nature ation with depth might cause the lithosphere to have
ofthe vertical principal stress:(a) normal fault;(b)thrust a 'sandwich' type theological layering (Ranalli &
fault;(c) strike-slip fault. Murphy, 1987), with perhaps several ductile layers
separating brittle layers. For example. Brace & Kohl-
resists relative motion across it (the Coulomb- stedt (1980) have investigated the limits of litho
Navier hypothesis). Under this closed crack regime, sphere strength based on measurements on quartz
the type of faulting which results, according to the and olivine, which are typical ofthe continental crust
simple theory of Anderson, depends upon which of and upper mantle, respectively. Their findings are
the principal pressures is vertical (Fig. 2.21). Nor illustrated in Fig. 2.23 (Molnar, 1988). Within the
mal,thrust and strike-slip faults occur depending on oceanic lithosphere the upper brittle crust gives way
whether a„ Omin or Oinj, respectively, is vertical. to a region of high strength at a depth of 20-60 km
Faulting at depth, which would otherwise appear to depending on the temperature gradient. The
require very high shear stresses because of the high strength thence gradually decreases and grades into
norrnal stresses, is probably facilitated by the reduc
tion of the frictional coefficient and the normal
stresses across the fault by the presence ofpore waters Oi + P
and hence a pore fluid pressure which counteracts the
normal stress.
The strength of brittle rocks increases with con
fining pressure but decreases with temperature. In
I .1. I
the uppermost 10 km the former effect is dominant
and the strength increases. Below this depth the
effect of temperature takes over, and rocks progres p:Rollmg;|
sively weaken downwards.
The mechanism of deformation of brittle solids
may be that of cataclasis (Fig. 2.22)(Ashby & Ver-
rall, 1977). This results from repeated shear fractur
ing, which acts to reduce the grain size of the rock,
and by the sliding or rolling ofgrains over each other.

2.10.3 Ductile deformation

The transition from brittle to ductile behaviour takes

place at a depth of 10-50 km in continental areas,
the actual value depending upon the local geo-
thermal gradient. The brittle-ductile transition
O^l + P
marks an important level below which rocks are
mechanically weaker. Fig.2.22 Deformation ofa brittle solid by cataclastic
The strength of rock depends upon temperature, flow (redrawn from Ashby & Verrall, 1977, with
pressure and composition. Thus compositional vari permission from the Royal Society).
32 CHAPTER 2

Strength Strength
Wa^r
Crust' Britt e-
Moho
ductile Crust
transition

Brittle
ductile Moho
Brittle-
transition
ductile
^ 50- ► Mantle transition

Mantle
Fig. 2.23 Schematic strength
Oceanic Continental
lithosphere
profiles through oceanic and
lithosphere
continental lithosphere (after
Molnar, 1988, using information
100-* 100^ from Brace & Kohlstedt, 1980).

that of the asthenosphere. Continental crust, how strain rate, that is, lower the apparent viscosity, while
ever, is much thicker than oceanic, and at the temper increased pressure produces a more sluggish flow.
atures of 400-700 ®C experienced in its lower parts Plastic flow (Fig. 2.24) occurs when the yield
the minerals are much weaker than the olivine found strength of the material is exceeded. Movement takes
at this depth in the oceanic lithosphere, so, whereas place in a manner similar to cataclasis, in that defor
the oceanic lithosphere behaves as a single rigid plate mation occurs by gliding motions along intersecting
because of its strong centre, the continental litho planes within crystals and movement along crystal
sphere is characterized by a layer of weakness at lower
crustal depths. This theological layering may explain
how the continental crust may detach from the Oi + p
mantle part of the lithosphere on continental colli
sion and create a Himalayan-type mountain range by
the stacking of thrust slices of crust (Section 9.3).
The mechanisms of flow in crystalline solids have XWVXXYWT"
Dislocation glide
been largely deduced from studies of the behaviour on intersecting ^
of metals, which have the advantage that they deform planes
in a ductile manner at lower temperatures and pres
sures than silicates. Suflicient is known of silicate
deformation, however, to validate deducing their
flow behaviour at high temperatures and pressures
from that of metals. It is now possible, using modern
apparatus, to determine directly the temperature and
pressure dependence of creep in mantle materials
(Borch & Green, 1987).
When the temperature is low, up to about half the
melting temperature {T^ in kelvin), materials react
to low stresses by transient creep, which is a very slow
flow in the solid state. Only a small amount of
deformation results and large permanent strains are
not generated.
There are three types of ductile flow which are
thought to occur beneath the brittle-ductile transi
tion (Ashby & Verrall, 1977). All are dependent Fig. 2.24 Deformation by plastic flow (redrawn from
upon the ambient temperature and, less markedly, Ashby & Verrall, 1977, with permission from the Royal
pressure. Increased temperature acts to increase the Society).

CTi + p
Cell Boundary
formation sliding
Tj_ Diffusion_x
Dynamic recrystallization
Oi + p
Fig.2.25 Deformation by power-law creep (redrawn
from Ashby & Verrall, 1977, with permission from the
Royal Society).
boundaries. The yield strength of materials deform
ing in this way is thus controlled by the magnitude of
the stresses required to overcome the resistance of
the crystal framework to this form of dislocation.
The strain produced tends to be limited by strain- or
work-hardening.
Power-law creep (also known as hot creep, recovery
creep or Weertman creep) (Fig. 2.25) takes place at
temperatures in excess of 0.55T^. In this form of
creep the strain rate is proportional to the nth. power
ofthe stress, where » ^ 3. Power-law creep is similar
to plastic flow, with deformation taking place along
crystal glide planes, but diflfers in that diffusion of
atoms is permitted by the higher temperatures so that
any barriers to movement are removed immediately
as they form; hence, work-hardening does not occur
and steady state creep is facilitated. The dislocations
take up a polygonal form whose size increases with
increased stress. Sliding along crystal boundaries
and recrystallization may also occur. Because of the
higher temperature the yield strength is lower than
for plastic flow,and strain results from lower stresses.
INTERNAL STRUCTURE OF THE EARTH 33
+P
Boundary
diffusion
Diffusive
fluxes
\ / Lattice
diffusion
Qi + P
Fig.2.26 Deformation by diffusion creep (redrawn from
Ashby & Verrall, 1977, with permission from the Royal
Society).
Power-law creep is believed to be the most important
form of deformation occurring in the mantle, and
governs convective flow within it(Weertman,1978).
Diffusion creep becomes dominant as temperatures
exceed O.SST^^ and results from the migration
of atoms in a stress gradient (Fig. 2.26). Atomic
migration through the crystal lattices is known as
Nabarro-Herring creep, while migration along crys
tal boundaries is known as Coble creep. In this form
of creep the strain rate (k) is proportional to the
applied stress (a) so that 0= 11^ the constant of
proportionality being the dynamic viscosity (ri). ^
increases as the square of the grain radius. Diffusion
creep is believed to occur within the asthenospheric
part of the mantle (Section 2.12).
2.10.4 Creep in the mantle
In applying the creep laws for metals to behaviour
in the mantle, it should be recognized that experi
mental data cannot duplicate the very slow strain
rates experienced naturally. In addition, the effect of
34 CHAPTER 2
pressure on creep rates is poorly known, rock
strength is significantly reduced by the presence of
water, and mantle material is probably not mono-
mineralic. Consequently, the strain rates calculated
for the mantle may be much in error.
Low temperature plastic flow is likely to be opera
tive immediately beneath the brittle-ductile transi
tion (Carter & Ave'lallemant, 1970; Goetze, 1978),
and because of the reasonably high yield strength
stress differences of about ICQ MPa may be present.
As the temperature increases to 0.55 power-law
creep is expected to become dominant. Because of
the differing compositions this would be at about
830 K in the subroceanic mantle, but at only 530 K
in the continental crust. The presence of structural
evidence in mantle-derived olivine nodules suggests
that power-law creep exists to a depth of at least
200 km.
The strain rate at which power-law creep occurs
should increase drastically on nearing the mantle low
velocity zone, where temperatures most closely ap
proach the melting temperature. At greater depths,
diffusion creep may become increasingly prominent
as the stress diflferences decrease and the temperature
rises. The rate ofdiffusion is believed to be controlled
by oxygen, which is the slowest ofthe atomic species
present in olivine. However,computations based on
both Nabarro-Herring and Coble mechanisms seem
to suggest that power-law creep is dominant, except
for the slowest strain rates.
2.11 Isostasy
2.11.1 Introduction
The phenomenon of isostasy concerns the response
ofthe outer shell of the Earth to the imposition and
removal oflarge loads. This layer, although relatively
strong, is unable to support the large stresses gener
ated by, for example, the positive weight of a moun
tain range or the relative lack of weight of an ocean
basin. For such features to exist on the Earth's
surface, some form of compensating mechanism is
required to avoid the large stresses that would other
wise be generated.
Isostasy was first recognized in the 18th century
when a party of French geodesists were measuring
the length of a degree of latitude in Peru in an
attempt to determine if the shape of the Earth cor
responds to an oblate or a prolate ellipsoid. Plumb
lines were used as a vertical reference in the surveying
and it was recognized that a correction would have
to be applied for the horizontal deflection caused by
the gravitational attraction of the Andes. When this
correction, based on the mass of the Andes above
sea level, was applied, however, it was found that
the actual vertical deflection was less than predicted
(Fig. 2.27). This phenomenon was attributed to the
existence of a negative mass anomaly beneath the
Andes which compensates, that is to say, supports,
the positive mass of the mountains. In the 19th
century similar observations were made in the vicin
ity of the Himalaya and it was recognized that the
compensation of surface loading at depth is a wide
spread phenomenon.
The presence of subsurface compensation is con
firmed by the variation in the Earth's gravitational
field over broad regions. Bouguer anomalies (Kearey
& Brooks, 1991)are generally negative over elevated
continental areas and positive over ocean basins
(Fig. 2.28). These observations confirm that the
positive topography ofcontinents and negative topo
graphy ofoceans is compensated by regions at depth
with density contrasts which are, respectively, nega
tive and positive and whose mass anomaly approxi
mates that of the surface features.
The principle of isostasy is that beneath a certain
depth, known as the depth of compensation, the
pressures generated by all overlying materials are
everywhere equal; that is, the weights of vertical
columns of unit cross-section, although internally
Andes
Compensating
mass
deficiency
Fig.2.27 Horizontal gravitational attraction ofthe mass
ofthe Andes above sea level would cause the deflection (c)
ofa plumb bob from the vertical (a). The observed
deflection (b)is smaller, indicating the presence of a
compensating mass deficiency beneath the Andes(angles
of deflection and mass distribution are schematic only).
 INTERNAL STRUCTURE OF THE EARTH 35

Mountain range Ocean

.'-'-Root':-

Dep^ of
compensation

^Continent^ Sea level

Ocean

swa

Lrr N^piPc
Fig.2.28 Inverse correlation of Bouguer anomalies with
topography indicating its isostatic compensation.

variable, are identical at the depth ofcompensation if Depth of

the region is in isostatic equilibrium.
Two hypotheses regarding the geometric form of
local isostatic compensation were proposed in 1855
by Airy and Pratt.
2.11.2 Airy's hypothesis
Airy's hypothesis assumes that the outermost shell of
the Earth is ofa constant density and overlies a higher
density layer. Surface topography is compensated by
varying the thickness of the outer shell in such a way
that its buoyancy balances the surface load. A simple
analogy would be blocks of ice of varying thickness
floating in water, with the thickest showing the
greatest elevation above the surface. Thus mountain
ranges would be underlain by a thick root, and
ocean basins by a thinned outer layer or antiroot
(Fig. 2.29a). The base of the outer shell is conse
quently an exaggerated mirror image of the surface
topography. Consider the columns of unit cross-
section beneath a mountain range and a region of
zero elevation shown in Fig. 2.29a. Equating their
weights gives:
+ TaPc + rPc + £>aPJ
=.5tTAPc +''Pm+i>APm]
where^ is the acceleration due to gravity.
Rearranging this equation gives the condition for
compensation
Fig.2.29 (a) Airy mechanism of isostatic compensation.
height of mountain above sea level; 2, depth of water of
density p,^,; normal thickness of crust of density p^; r,
thickness of root; thickness of antiroot; depth of
compensation below root; p^,density of mantle,(b)The
Pratt mechanism of isostatic compensation. Legend as for
(a)except Tp, normal thickness ofcrust; p^, density of
crust beneath mountain; p^, density ofcrust beneath
ocean;Dp,depth ofcompensation below Tp.
isostatic equilibrium:
hpc
^"(Pm-Pc)
A similar computation provides the condition for
compensation of an ocean basin:
g(Pc - Pw)
a =-
(Pm-Pc)
2.11.3 Pratt's hypothesis
Pratt's hypothesis assumes a constant depth to the
base of the outermost shell of the Earth, whose
density varies according to the surface topography.
Thus, mountain ranges would be underlain by
36 CHAPTER 2
relatively low density material and ocean basins by
relatively high density material (Fig. 2.29b). Equat
ing the weights of columns of unit cross-section
beneath a mountain range and a region of zero
elevation gives:
which on rearrangement provides the condition for
isostatic equilibrium ofthe mountain range:
^pPc
Ph =
(Tp+Z;)
A similar computation for an ocean basin gives:
(Tppc-Zpw)
Pz =
(Tp-z)
In these early models of isostasy it was assumed that
the outer shell of the Earth whose topography is
compensated corresponded to the crust. Certainly
the large density contrast existing across the Moho
plays a major part in the compensation. It is now
believed, however, that the compensated layer is
rather thicker and includes part ofthe upper mantle.
This strong outer layer of the Earth is known as the
lithosphere (Section 2.12). The lithosphere is under
lain by a much weaker layer known as the astheno-
sphere which deforms by flow, and which can thus
be displaced by vertical movements of the litho
sphere. The density contrast across the lithosphere-
asthenosphere boundary is, however, very small. A
further complication in assessing the level at which
isostatic compensation takes place arises because the
continental crust can be theologically stratified into
brittle and ductile layers (Section 2.10.3). Isostatic
compensation can thus also take place at the brittle-
ductile transition. Indeed, some structures, such as
continental rift zones (Section 10.1.5), may be fully
compensated at this level.
Both the Airy and Pratt hypotheses are essentially
applications of Archimedes' Principle whereby
adjacent blocks attain isostatic equilibrium through
their buoyancy in the fluid substratum. They assume
that adjacent blocks are decoupled by fault planes
and achieve equilibrium by rising or subsiding in
dependently. These models of local compensation,
although widely used, in fact imply unreasonable
mechanical properties for the crust and upper mantle
(Banks et aL^ 1977), because they predict that in
dependent movement would take place even for very
small loads. The lithosphere is demonstrably not as
weak as this implies, as large gravity anomalies exist
over igneous intrusions with ages in excess of
100 Ma. The lithosphere must therefore be able to
support stress differences of up to 20-30 MPa for
considerable periods oftime without the necessity of
local compensation.
2.11.4 Flexure of the lithosphere
More realistic models of isostasy involve regional
compensation. A common approach is to make the
analogy between the lithosphere and the behaviour
of an elastic sheet under load. Figure 2.30 illustrates
the elastic response to loading; the region beneath
the load subsides over a relatively wide area by
displacing asthenospheric material, and is comple
mented by the development of peripheral bulges.
Over long periods of time, however, the lithosphere
may act in a viscoelastic manner and undergo some
permanent deformation by creep (Section 2.10.4).
For example, the vertical displacement z of the
oceanic lithosphere under loading can be calculated
by modelling it as an elastic sheet by solving the
fourth order differential equation:
d^z
^^+(Pm-pw)2^ = P(^)
where P{x) is the load as a function of horizontal
distance x^£f the acceleration due to gravity and pj„,
the densities of asthenosphere and sea water,
respectively. D is a parameter termed the flexural
rigidity, which is defined by:
D = ET^/\2{\-<5^)
where E is Young's modulus,a Poisson's ratio and T
the thickness of the lithosphere.
The specific relationship between the displacement
z and load for the two-dimensional sediment load of
half-width height h and density p^ shown in
Fig. 2.30 is:
2max = ^(Ps"Pw)(l"C"*-" COS A^)/(Pm"Ps)
where
^=^(Pm-Pwk/4I>
and p^, Pn, the densities of water and the mantle,
respectively.
Note that as the elastic layer becomes more rigid,
D approaches infinity, X approaches zero and the

"L-rrv.x? «?J!|P§j Lithosphere
Fig.2.30 Flexural downbending ofthe lithosphere as a
result of a two-dimensional load of half-width a, height h
and density Pg.
depression due to loading becomes small. Con
versely, as the layer becomes weaker, D approaches
zero, X approaches infinity and the depression ap
proaches h{p^ - pj/(p^ - Ps)(Watts & Ryan, 1976).
This is equivalent to Airy-type isostatic equilibrium
and indicates that for this mechanism to operate the
elastic layer and fluid substrate'must both be very
weak.
It can be shown that loads with a half-width ofless
than about 50 km are supported by the finite strength
of the lithosphere. Loads with half-widths in excess
ofabout 500 km are in approximate isostatic equilib
rium. Figure 2.31 illustrates the equilibrium attained
by the oceanic lithosphere when loaded by a sea-
mount(Watts etaL^ 1975).
- 3000 - — Observed
Computed D = 2 x 10^^ Nm
E 2000
o D = 6x1022Nm
D =4xlO"Nm
^ 1000
100 200 300
Distance(km)
Fig.2.31 Interpretation ofthe free air anomaly ofthe
Great Meteor Seamount, northeast Atlantic Ocean,in
terms offlexural downbending ofthe crust. A model with
the flexural rigidity(D)of6 x 10^^ N m appears best to
simulate the observed anomaly. Densities in Mg m"^.
Arrow marks the position 30"N,28 ®W (redrawn from
Watts^a/., 1975, with permission from the American
Geophysical Union).
INTERNAL STRUCTURE OF THE EARTH 37
2.11.5 Isostatic rebound
The equilibrium flexural response of the lithosphere
to loading is independent of the precise mechanical
properties ofthe underlying asthenosphere as long as
it facilitates flow. However,the reattainment ofequi
librium after removal of the load, a phenomenon
known as isostatic rebound^ is controlled by the viscos
ity ofthe asthenosphere. Measurement ofthe rates of
isostatic rebound provides a means ofestimating the
viscosity of the upper mantle. Fennoscandia repre
sents an example of this type of study as precise
levelling surveys undertaken since the late 19th cen
tury have shown that this region is undergoing uplift
following the melting of the Pleistocene ice sheet
(Fig. 2.32). The maximum uplift rates occur around
Lithosphere
Asthenosphere
Fig.2.32 Theory ofisostatic rebound,(a) The load of an
icecap on the lithosphere causes downbending
accompanied by the elevation ofthe peripheral lithosphere
and lateral flow in the asthenosphere (b). When the icecap
melts (c), isostatic equilibrium is regained by reversed flow
in the asthenosphere, sinking ofthe peripheral bulges and
elevation ofthe central region (d).
38 CHAPTER 2
the Gulfof Bothnia, where the land is rising at a rate
of over 10 mm a"^. Twenty thousand years ago the
land surface was covered by an ice sheet about 2.5 km
thick (Fig. 2.32a). The lithosphere accommodated
this load by flexing (Fig. 2.32b), resulting in a sub
sidence of600-700 m and a lateral displacement of
asthenospheric material. This stage currently pertains
in Greenland and Antarctica where, in Greenland,
the land surface is depressed by as much as 250 m
below sea level by the weight ofice. Melting ofthe ice
was complete about 10000 years ago (Fig. 2.32c),
and since this time the lithosphere has been return
ing to its original position and the land rising in
order to regain isostatic equilibrium. A similar situa
tion pertains in northern Canada where the land
surface around Hudson Bay is rising subsequent to
the removal of an icecap. The rate of isostatic re
bound provides an estimate for the viscosity of the
upper mantle of 10^^Pas (Pascal seconds), and
measurements based on world-wide modelling of
post-glacial recovery and its associated oceanic
loading suggest that this figure generally applies
throughout the upper mantle as a whole (Peltier &
Andrews, 1976). Compared to the viscosity of water
(10"^ Pa s) or a lava flow (4x10^ Pa s), the viscosity
of the sub-lithospheric mantle is extremely high and
its fluid behaviour is only apparent in processes with
a large time constant. Knowledge of the viscosity of
the mantle, however, provides an important control
on the nature of mantle convection, as will be
discussed in Chapter 11.
2.11.6 Tests of isostasy
The state ofisostatic compensation ofa region can be
assessed by making use of gravity anomalies. The
isostatic anomaly^ lA^ is defined as the Bouguer anom
aly minus the gravity anomaly of the subsurface
compensation. Consider a broad, flat plateau of
elevation h compensated by a root ofthickness r. The
terrain correction of such a feature is small in the
central part of the plateau so that here the Bouguer
anomaly,BA,is related to the free air anomaly,FAA
by the relationship:
BA^FAA-BC
where BC is the Bouguer correction, equal to
2nGp^h^ where is the density of the compensated
layer. For such an Airy compensation:
where A^oot is the gravity anomaly of the compen
sating root. Since the root is broad compared to its
thickness, its anomaly may be approximated by that
of an infinite slab, that is 2kG{p^ - p^)^, where p^, is
the density of the substrate. Combining the above
two equations:
lA = FAA - 2nGp^h - 27cG(pc - Pn,)r
From the Airy criterion for isostatic equilibrium:
f =/;p,/(p„-pc)
Substitution of this condition into the equation
reveals that the isostatic anomaly is equal to the free
air anomaly over a broad flat feature, and this repre
sents a simple method for assessing the state of
isostatic equilibrium.
More rigorous tests of isostasy over topographic
features of irregular form are made by accurate com
putation of isostatic anomalies. This procedure in
volves calculating the shape of the compensation
required by a given hypothesis ofisostasy,computing
its gravity anomaly and then subtracting this from
the observed Bouguer anomaly to provide the iso
static anomaly. Figure 2.33 shows free air, Bouguer
and isostatic anomalies over a feature with varying
degrees ofcompensation.
Gravity anomalies can thus be used to determine if
a surface feature is isostatically compensated at
depth. They cannot, however, reveal the form of
compensation and indicate which type of mechanism
is in operation. This is because the compensation
occurs at a relatively deep level and the differences in
the anomalies produced by a root/antiroot (accord
ing to the Airy hypothesis) or by different density
units (according to the Pratt hypothesis) would be
very small. Moreover, the gravity anomalies over
most regions contain short wavelength components
resulting from localized, uncompensated geological
structures which obscure the differences in the re
gional field arising from the different forms of
compensation.
Information on the geometric form of isostatic
compensation can be gained by a combined analysis
of gravity and seismic,refraction data, as the latter
technique can provide a reasonably detailed picture
of the subsurface structure of the region under con
sideration. Such studies have demonstrated that the
broad isostatic equilibrium ofcontinents and oceans
is mainly accomplished by variations in crustal thick
ness according to the Airy hypothesis. Figure 2.34
 INTERNAL STRUCTURE OF THE EARTH 39

o 500 Airy isostatic

Airy isostatic

Bouguer
Bouguer
'0% compensation
70% compensation

Q
/-\ 50-^
Rn - : ■•
>,... •• 3 30 I
..V .. "i *50

(b) 0 1000 km

Airy isostatic
Airy isostatic

o 1500-

Bouguer
Bouguer
30% compensation 0% compensation

I 40-

Fig. 2.33 Free air, Bouguer and Airy isostatic anomalies
over an idealized mountain range (a) in perfect isostatic
equilibrium, (b) with 70% isostatic compensation,
(c) with 30% isostatic compensation, (d) uncompensated.
Densities in Mg m"^.
shows seismic velocity sections from the western
USA in which surface topography is largely compen
sated by Moho topography, although in several loca
tions density variations in the upper mantle must be
invoked to explain the isostatic compensation. A
cross-section across the western USA (Fig. 2.35) re
veals, however, that crustal thickness is not necessar
ily related to topographic elevation as the Great
Plains, which reach a mean height of 1 km, are
underlain by crust 40-50 km thick and the Basin and
Range province, at 2 km elevation, is underlain by
25-30 km thick crust. Clearly, the Basin and Range
province must be partially compensated by a Pratt-
type mechanism resulting from the presence of low
density material in the upper mantle. Similarly,
ocean ridges (Chapter 6) owe their elevation to a
region of low density material in the upper mantle
rather than to a thickened crust.
There are regions of the Earth's surface which do
not conform to the concepts of isostasy discussed
here. The lithosphere has sufficient internal strength
to support relatively small loads without the
40 CHAPTER 2

o-
O •

8 8 •
E -1000-
o

• Isostatic anomaly
o Bouguer anomaly
3.6 4.64 4.8 3.8

5.68.v. 5.6 ® °
>.11 6.6
6.07

•1—1 7,58 j__b#^i7"34n \

:'7?7^|8.16 8.18^^811;:
7 95
u1
7.87;
J

(D
C
CO
c c
CO CO
c
8.07
CD
c
D3
C
(/)
CO
CO
c
hUta CO
T3
■a
c
<0
'c
CO
"c (0
CD
CO
'c
c
o
CO CO (/) CO k.
CO CO CO o _CD O)
E c N > 0 o o o
c c c c c CO CD U c
o C
o o O o o
< Z lo "co CO "cD crt
Fig. 2.34 Bouguer and isostatic
5 2 2 < CO u U 'c CJ CO

5 si 2 lU
"c
(/3 C/j o "co s:
gravity anomalies and their relation
z CO
to seismic velocity sections from the
western USA. Velocities in km .s"'
(redrawn from Garland, 1979),

Colorado

I .c >- c I
slillill Basin and Range Colorado
Il1|l Great

siMii
u

^West f
Province Plateau I CO tr 2 I Plains
,1 Jr.'.Cs', East Fig. 2.35 Section from San
E 2o4 Francisco, California to Lamar,
Colorado based on seismic

^id velocity refraction data (redrawn from

Pakiser, 1963, Journal of Geoplr/sical
Research 68, with permission from
the American Geophysical Union).

necessity of subsurface compensation. Such feamres

include small topographic irregularities and varia
tions in crustal density such as small granitic and
basic bodies. As will be seen in Section -10.1.5, rift
valleys conform to a modified mechanism of com
pensation. The hypotheses discussed above all as
sume that the support of surface features is achieved
by their attaining hydrostatic equilibrium with the
substrate. In certain areas, however, in particular
convergent plate margins, surface features are sup
ported dynamically by horizontal stresses. Such fea
tures provide the largest isostatic anomalies observed
on the Earth's surface.
2.12 Lithosphere and
asthenosphere
It has long been recognized that for large scale
structures to attain isostatic equilibrium, the outer
most shell of the Earth must be underlain by a weak
layer that deforms by flow. This concept has assu med
fundamental importance since it was realized that the
subdivisions of the Earth controlling plate tectonic
movements must be based on theology, rather than
composition.
The lithosphere is defined as the strong, outermost
layer of the Earth that deforms in an essentially

elastic manner. It is made up ofthe crust and upper
most mantle. The lithosphere is underlain by the
asthenosphere, which is a much weaker layer and
reacts to stress in a fluid manner. The lithosphere is
divided into plates, of which the crustal component
can be oceanic and/or continental, and the relative
movements of plates take place upon the astheno
sphere.
However, having made these relatively simple
definitions, examination of the several properties
which might be expected to characterize these layers
reveals that they lead to different ideas of their
thickness. The properties considered are thermal,
seismic, elastic and temporal.
Temperature is believed to be the main pheno
menon which controls the strength of subsurface
material. Hydrostatic pressure increases with depth
in an almost linear manner,and so the melting point
ofrocks also increases with depth. Melting will occur
when the temperature curve intersects the melting
curve (solidus) for the material present at depth
(Fig. 2.36). The asthenosphere is believed to repre
sent the location in the mantle where the melting
point is most closely approached. This layer is cer
tainly not completely molten,as it transmits S waves,
but it is possible that a small amount of melt is
present. The depth at which the asthenosphere
occurs depends upon the geothermal gradient and
2000
1600-
a
o
(U 1200-
L-
D 30-
to
800-
A Continent
B Precambrian shield
C Oceanic ridge
400-
D Ocean basin
100 200 300 400 500
Depth (km)
Fig. 2.36 Variation of temperature with depth beneath
continental and oceanic regions (redrawn from Condie,
1982,Plate Tectonics and Crustal Evolution^ with
permission from Pergamon Press Ltd).
INTERNAL STRUCTURE OF THE EARTH 41
the melting temperature of the mantle materials(Le
Pichon et al.^ 1973). Beneath ocean ridges, where
temperature gradients are high, the asthenosphere
must occur at shallow depth. Indeed, since it is
actually created in the crestal region (Section 6.10),
the lithosphere there is particularly thin. The gradi
ent decreases towards the deep ocean basins, and the
lithosphere thickens in this direction, the increase
correlating with the depth ofwater as the lithosphere
subsides as a result of contraction on cooling
(Section 6.9). The mean lithosphere thickness on
this basis beneath oceans is probably 60-70 km.
Beneath continents a substantial portion of the ob
served heat flow is produced within the crust(Section
2.13), so the temperature gradient in the sub-crustal
lithosphere must be considerably lower than in
oceanic areas. It is probable that the mantle solidus is
not approached until a significantly greater depth,so
that the continental lithosphere has a thickness of
110-150 km, being at a maximum beneath cratonic
areas.
The depth of the LVZ for seismic waves agrees
quite well with the temperature model oflithosphere
and asthenosphere. The LVZ progressively thickens
away from the crests of mid-ocean ridges (Forsyth,
1975)(Fig. 2.37), reaching an average thickness for
old oceanic lithosphere of 70-80 km. The LVZ is
rather deeper beneath continents, and occurs at a
depth of 110-150 km. Within the LVZ attenuation
of seismic energy, particularly shear wave energy, is
very high. Both the low seismic velocities and high
Age(Ma)
10 20 30 40 50
60-
E
90-
.c
Q.
0)
Zone of partial melting 4.1 km s ^
Q 120-
150-
180-
Fig.2.37 Shear wave model ofthe thickening of oceanic
lithosphere with age. Velocities in km s~^ (redrawn from
Forsyth, 1975, with permission from the Royal
Astronomical Society). The 150 km transition may be
somewhat deeper.
42 CHAPTER 2
attenuation are consistent with the presence of a
relatively weak layer at this level. As would be ex
pected for a temperature-controlled boundary, the
lithosphere-asthenosphere interface is not sharply
defined,and occupies a zone several kilometres thick.
When the Earth's surface is loaded, the lithosphere
is believed to react by downward flexure. This is
frequently observed on continental areas in response
to loading by ice sheets, large sedimentary basins and
mountain ranges, but is less common in the oceans
where only the loading by seamounts occurs with any
frequency. The amount of flexure depends upon the
magnitude of the load, the flexural rigidity of the
lithosphere and the lithospheric thickness. The mag
nitude of the load can be calculated and the flexural
rigidity of the lithosphere is now known with reason
able accuracy. Consequently, flexure studies can be
used to determine the elastic thickness of the litho
sphere. The elastic thickness of the oceanic litho
sphere deduced in this way is 40 km or less, and
increases in thickness with distance from ocean
ridges (Watts et al., 1980)(Fig. 2.38). Beneath con
tinental shield areas the elastic thickness can vary
from 30 to 110 km. Under tectonically active areas.
Age of oceanic lithosphere(Ma)
40 80 120 160
20- m
"2^ 35(
Ela
Elastic
thickness
Seismic
thickness
Fig.2.38 Comparison of short term 'seismic' thickness
and long term 'elastic' thickness for oceanic lithosphere of
different ages (redrawn from Watts et«/., 1980, with
permission from the American Geophysical Union).
such as the Basin and Range Province, the elastic
thickness may be as small as 4 km, which is less than
the crustal thickness (Bechtel etal., 1990). Such very
thin elastic thicknesses are undoubtedly due to very
high geothermal gradients.
The concept of the lithosphere as a layer of uni
formly high strength is seen to be over-simplistic
when the theological layering is considered. The
upper 20-40 km of the lithosphere are brittle and
respond to stress below the yield point by elastic
deformation accompanied by transient creep. Be
neath the brittle zone is a layer which deforms by
plastic flow above a yield point of about 100 MPa.
The lowest part, which is continuous with the as-
thenosphere, deforms by power-law creep and is de
fined as the region where the temperature increases
with depth from 0.55 to 0.85 T^.
The type of deformation experienced depends
upon the duration ofthe applied loads. Over pei iods
of a few thousand years, most of the region exhibit
ing power-law creep does not deform significantly
and consequently is included within the elastic litho
sphere. Long term loading, however, occurring over
periods of a few million years, permits power-law
deformation to occur so that this region then be
longs to the asthenosphere.
It can thus be appreciated that the lithosphen; can
be defined in a number of different ways w hich
provide different estimates of its thickness. This
must be borne in mind throughout any consider
ation of plate tectonic processes.
The asthenosphere is believed to extend to a depth
of about 700 km. The properties of the underlying
region are only poorly known. Seismic waves w hich
cross this region do not suffer great attenuation
(Section 8.3), and so it is generally accepted that this
is a layer of high strength, termed the mesosphere. The
compositional and theological layering of the Earth
are compared in Fig. 2.39.
2.13 Terrestrial heatflow
The study of thermal processes within the Earth is
somewhat speculative because the interpretation of
the distribution of heat sources and the mechanisms
of heat transfer are based on measurements made at
or near the surface. Such a study is important, how
ever, as the process of heat escape from the Earth's
interior is the direct or indirect cause of most tec
tonic and igneous activity.

Compositional layering
Mohorovicic\jJ^
discontinuity \ ^
400 krtr^
670 knA
Gutenberg —
2891 km^
discontinuity
Fig. 2.39 Comparison of the
compositional and rheological
layering of the Earth.
The vast majority of the heat affecting the Earth's
surface comes from the Sun, which accounts for
some 99.98% of the Earth's surface energy budget.
Most of this thermal energy, however, is reradiated
into space, while the rest penetrates only a few
hundred metres below the surface. Solar energy
consequently has a negligible effect on thermal pro
cesses occurring in the interior of the Earth. The
geothermal energy loss from heat sources within the
Earth constitutes about 0.022% of its surface energy
budget. Other sources of energy include the energy
generated by the gradual deceleration of the Earth's
rotation and the energy released by earthquakes, but
these make up only about 0.002% of the energy
budget. It is thus apparent that geothermal energy is
the" major source of the energy which drives the
Earth's internal processes.
It is believed that the geothermal energy is derived
in part from the energy given off during the radio
active decay oflong-lived isotopes, in particular
U23S, and Th^®^, and also from the heat re
leased during the early stages of the formation of the
Earth. These isotopes would account for the present
geothermal loss if present in proportions similar to
those of chondritic meteorites. Radioactive decay is
INTERNAL STRUCTURE OF THE EARTH 43
Rheological layering
6 km Continent Inn
""" U35h5phe;^--2*{n
!Lower mantle;; ;;Mesosphere;;
5150 km^
exponential, so that during the early history of the
Earth the concentration of radioactive isotopes
would have been significantly higher than at present
and the thermal energy available to power its internal
processes would have been much greater (Section
12.2.2). Currently accepted models for the forma
tion of the Earth require an early phase of melting
and differentiation of its originally homogeneous
structure. This melting is believed to have been
powered in part by thermal energy provided by the
decay ofshort-lived radioactive isotopes such as AP®,
Ee®" and CP^.The differentiation ofthe Earth would
also have contributed energy to the Earth arising
from the loss in gravitational potential energy as the
dense iron-nickel core segregated to a lower energy
state at the centre of the Earth.
The heat flow through a unit area of the Earth's
surface, H,is given by:
H.fK oz
where 6T/5z is the thermal gradient perpendicular
to the surface and K the thermal conductivity of the
medium through which the heat is flowing. The units
ofJT are mW m"^.
44 CHAPTER 2
On land, heat flow measurements are normally
made in boreholes. Mercury maximum thermomet
ers or thermistor probes are used to determine the
vertical temperature gradient. Thermal conductivity
is measured on samples ofthe core using a technique
similar to the Lee's disc method. Although appearing
relatively simple, accurate heat flow measurements
on land are difficult to accomplish. The drilling of a
borehole necessitates the use of fluid lubricants
which disturb the thermal regime of the borehole so
that it has to be left for several months to allow the
disturbance to dissipate. Porous strata have to be
avoided as pore water acts as a heat sink and distorts
the normal thermal gradients. Consequently, it is
rarely possible to utilize boreholes sunk for the
purposes ofhydrocarbon or hydrogeological explora
tion. In many areas readings may only be undertaken
at depths below about 200 m so as to avoid the
transient thermal effects of glaciations.
Heat flow measurements are considerably easier to
accomplish at sea. The bottom temperatures in the
oceans remain essentially constant and so no compli
cations arise because of transient thermal perturba
tions. A temperature probe is dropped into the upper
soft sediment layer of the sea bed and, after a few
minutes' stabilization, the temperature gradient is
measured by a series of thermistor probes. A corer
associated with the probe collects a sediment sample
for thermal conductivity measurements; alterna
tively, the role of one of the thermistors can be
changed to provide a source of heat. The change in
the temperature of this probe with time depends on
the rate at which heat is conducted away from it, and
this enables a direct, in situ measurement of the
thermal conductivity ofthe sediment to be made.
A large proportion of geothermal energy escapes
from the surface by conduction through the solid
Earth. In the region of the oceanic ridge system,
however, the circulation of sea water plays a major
role in transporting heat to the surface and about
25% of the geothermal energy flux at the Earth's
surface is lost in this way.
The pattern of heat flow provinces on the Earth's
surface broadly correlates with major physiographic
and geological subdivisions. On continents the
magnitude of heat flow generally decreases from the
time of the last major tectonic event (Sclater et»/.,
1980). Heat flow values are thus low over the
Precambrian shields and much higher over regions
affected by Tertiary orogenesis. Within the oceans
the heat flow decreases with the age of the litho-
sphere (Section 6.9), with high values over the
oceanic ridge system and active marginal seas and
low values over the deep ocean basins and inactive
marginal seas.
The average heat flux in continental areas is
57 mW m"^, and in oceanic areas lOI mW m~^, of
which about 30% is contributed by hydrothermal
activity at the mid-oceanic ridge system (Davies,
1980). The continental value is about 60% of the
oceanic; about 75% of the geothermal energy is
lost through the oceans, and 25% through the
continents.
Further reading
Anderson, D.L. (1989) Theory of the Earth. Blackwell
Scientific Publications, Oxford.
Ashby, M.F. & Verrail, R.A.(1977) Micro-mechanisms of
flow and fracture, and their relevance to the rheology of
the upper mantle. Phil. Trans. Roy. Soc. Land. 288A,
59-95.
Bott, M.H.P.(1982)The Interior ofthe Earth. Its Structure,
Constitution and Evolution (2nd edn). Edward Arnold,
London.
Brett, R.(1976) The current status of speculations on the
composition ofthe core ofthe Earth. Rev. Geophys. Space
Phys. 14, 375-83.
Brown, G.C. & Mussett, A.E. (1981) The Inaccessible
Earth. Unwin Hyman,London.
Condie, K.C.(1982)Plate Tectonics and Crustal Evolution.
Pergamon Press, Oxford.
Edgar, N.T. (1974) Acoustic stratigraphy in the deep
oceans. In: Burk, C.A. & Drake, C.L.,(eds) The Geolo^
of Continental Marpfins, pp. 243-6. Springer-Verlag,
Berlin.
Elthon, D. (1981) Metamorphism in oceanic spreading
centres. In-. Emiliani, C. (ed.) The Oceanic Lithosphere,
The Sea 7, pp. 285-303. Wiley, New York.
Fowler,C.M.R.(1990)The Solid Earth. Cambridge Univer
sity Press, Cambridge.
Francheteau,J.(1983)The oceanic crust. Sci. Am. 249(3),
68-84.
Green, D.H. & Ringwood, A.E.(1967) An experimental
investigation of the gabbro to eclogite transition and its
petrological applications. Geochim. Cosmochim. Acta. 31,
767-833.
Jacobs,J.A.(1991)The Deep Interior ofthe Earth. Chapman
& Hall, London.
Kay, R.W. & Kay, S.M. (1981) The nature of the lower
continental crust: inferences from geophysics, surface
geology and crustal xenoliths. Rev. Geophys. Space Phys.
20, 735-60.

Mason,R.(1985) Ophiolito. Geokjgy Today 1,136-40.
Moores, E.M.(1982) Origin and emplacement of ophio-
lites. Rev. Geophys. Space Rhys. 20,735-60.
Nicolas, A.(1989) Structure of Ophiolites and Dynamics of
Oceanic Lithosphere. Kluwer Academic Publishers,
Dordrecht.
Park, R.G.(1988) Geolopfical Structures and Moving Plates.
Blackie, Glasgow.
INTERNAL STRUCTURE OF THE EARTH 45
Parson, L.M., Murton, B.J. & Browning, P. (eds)(1992)
Ophiolites and Their Modem Oceanic Analogues. Spec.
Pub. 60, Geological Society ofLondon.
Ringwood, A.E. (1975) Composition and Petrolopy of the
Earth^s Mantle. McGraw-Hill, New York.
3/ Continental drift
3.1 Introduction
In the 18th century it had been noted that the
western and eastern coastlines of the Atlantic Ocean
appeared to fit together like the pieces of a jigsaw
puzzle (Section 1.1). The significance of this obser
vation was not fully realized, however, until the 19th
century, when the geometrical fit of continental
outlines was invoked as a major item of evidence in
constructing the hypothesis of continental drift. The
case for the hypothesis was further strengthened by
the correspondence of geological features across the
juxtaposed coastlines. Application of the technique
of palaeomagnetism in the 1950s and 1960s pro
vided the first quantitative evidence that continents
had moved at least in a north-south direction during
geological time. Moreover, it was demonstrated that
the continents had undergone relative motions, and
xis of rotation
ole of rotation
ngle of rotation
Great circle or
equator of rotation
Small circles or
latitudes of rotation
Fig. 3.1 Euler's theorem. Diagram illustrating how the
motion of a continent on the Earth can be described by an
angle ofrotation about a pole of rotation.
this confirmed that continental drift had actually
occurred.
3.2 Continental reconstructions
3.2.1 Euler's theorem
In order to perform accurate continental reconstruc
tions across closed oceans it is necessary to be able to
describe mathematically the operation involved in
making the geometrical fit. This is accomplished
according to a theorem ofEuler, which states that the
movement ofa portion ofa sphere across its surface is
uniquely defined by a single angular rotation about a
pole of rotation (Fig. 3.1). The pole of rotation and
its antipodal point on the opposite diameter of the
sphere are the only two points which remain in a
fixed position relative to the moving portion. Conse
quently, the movement of a continent across the
surface of the Earth to its pre-drift position can be
described by its pole and angle of rotation.
3.2.2 Geometrical reconstructions of
continents
Although approximate reconstructions can be per
formed manually by moving models of continents
across an accurately constructed globe, the most
rigorous reconstructions are performed mathemat
ically by computer, as in this way it is possible to
minimize the degree of misfit between the juxta
posed continental margins.
The technique generally adopted in computer-
based continental fitting is to assume a series of poles
of rotation for each pair of continents arranged in a
grid of latitude and longitude positions. For each
pole position the angle of rotation is determined
which brings the continental margins together with
the smallest proportion of gaps and overlaps. The fit
is not made on the coastlines, as continental crust
extends beneath the surrounding shelfseas out to the
 CONTINENTAL DRIFT 47

continental slope. Consequently, the true junction
between continental and oceanic lithosphere is taken
to be at some isobath marking the midpoint of the
continental slope, for example the 1000 m contour.
Having determined the angle of rotation, the good
ness offit is quantified by some criterion based on the
degree of mismatch. This goodness offit is generally
known as the objective function. Values ofthe objec
tive function are entered on the grid ofpole positions
and contoured. The location ofthe minimum objec
tive function revealed by this procedure then pro
vides the pole of rotation for which the continental
edges fit most exactly.
3.2.3 The reconstruction of
continents around the Atlantic
The first mathematical reassembly of continents
based solely on geometrical criteria was performed by
Bullard et al. (1965), who fitted together the conti
nents on either side of the Atlantic (Fig. 3.2). This
was accomplished by sequentially fitting pairs of

Fig. 3.2 Fit ofthe continents Overlap

around the Atlantic Ocean,
obtained by matching the 500
fathom (927 m)isobath
(redrawn from Bullard etaL^
1965, with permission from the
Royal Society).
48 CHAPTER 3
continents after determining their best fitting poles
of rotation by the procedure described in Section
3.2.2. The only rotation involving parts ofthe same
iandmass is that of the Iberian peninsular with re
spect to the rest ofEurope. This is justified because of
the known presence ofoceanic lithosphere in the Bay
ofBiscay which is closed by this rotation. Geological
evidence (Section 3.3) and information provided by
magnetic lineations in the Atlantic (Section 4.5)
indicate that the reconstruction represents the con
tinental configuration during early Jurassic times
some 200 Ma ago.
Examination of Fig. 3.2 reveals a number of
overlaps of geological significance, some of which
may be related to the process ofcontinental splitting
(Section 10.3). Iceland is absent because it is of
Tertiary to Recent age and its construction during
the opening ofthe Atlantic postdates the reconstruc
tion. The Bahama Platform appears to overlap the
African continental margin and mainland. It is prob
able, however, that the platform represents an accu
mulation ofsediment capped by coral which formed
after the Americas separated (Dietz & Holden,
1970). Similarly,the Niger Delta ofAfrica appears to
form an overlap when in fact it also developed after
rifting.
A major criticism of the reconstruction is the
overlap of Central America on to South America and
absence of the Caribbean Sea. This must be viewed,
however,in the light ofour knowledge ofthe history
of the opening of the Atlantic based, for the most
part, on its pattern of magnetic lineations. Palaeo-
magnetic evidence (Section 3.6) suggests that the
Palaeozoic crustal blocks which underlie Central
America were originally situated within the region
now occupied by the GulfofMexico,an area existing
within the reconstruction (White, 1980). The North
Atlantic started opening about 180 Ma ago, and the
South Atlantic somewhat later, about 135 Ma ago.
The poles of rotation ofthe North and South Atlan
tic were sufficiently different that the opening created
the space between North and South America now
occupied by the Caribbean. This also allowed a clock
wise rotation of the Central American blocks out of
the Gulf of Mexico to their present locations. About
80 Ma ago the poles of rotation of the North and
South Atlantic changed to an almost identical loca
tion in the region of the present north pole so that
from this time the whole Atlantic Ocean effectively
opened as a single unit.
3.2.4 The reconstruction of Gondwana
Geometrical evidence alone has also been used in the
reconstruction of the southern continents which
make up Gondwana. The first such reconstruction
was performed by Smith & Hallam (1970) and is
illustrated in Fig. 3.3. The shapes of the continental
edges of the east coast of Africa, Madagascar, India,
Australia and Antarctica are not quite so well suited
to fitting as the circum-Atlantic continents, and so
the reconstruction is more controversial. It does.
Fig. 3.3 Fit of the southern
continents and India (redrawn
from Smith & Hallam, 1970,
with permission from Nature
225,139-44. Copyright ©
1970 Macmillan Magazines
Ltd).

however, represent the best least-squares fit with a
minimum of overlap and gaps. The main uncertain
ties are the locations of Madagascar and India within
the reconstructions, and Fig. 3.4, from Powell et al.
(1980), illustrates some of the alternative arrange
ments which have been suggested. Differentiation
Fig.3.4 Possible alternative reconstructions of
Gondwana (after Powell et 1980, with permission
from Elsevier). Note the different positions of Madagascar
and India.
CONTINENTAL DRIFT 49
between these must now be based on geological
(Section 3.3), palaeomagnetic (Section 3.6) and sea
floor spreading (Section 4.6) criteria.
3.3 Geological evidence for
continentai drift
The continental reconstructions discussed in Sec
tions 3.2.3 and 3.2.4 are based solely on the geomet
rical fit of continental shelf edges. If they represent
the true ancient configurations of continents it
should be possible to trace continuous geological
features from one continent to another across the
fits. The matching of features requires the rifting of
the supercontinent across the general trend of geo
logical features. This does not always occur as the
location ofthe rift is often controlled by the geology
of the supercontinent, and takes place along lines of
weakness which may run parallel to the geological
grain. However, there remain many geological fea
tures which can be correlated across juxtaposed con
tinental margins, some of which are listed below.
1 Fold belts. The continuity of Palaeozoic fold belts
across the assembled Laurasia and Gondwanaland
supercontinents is illustrated in Fig. 3.5. An espe
cially well-studied case is the continuation of the
Appalachian fold belt ofwestern North America with
the Caledonian fold belt ofnorthern Europe. Within
the sedimentary deposits associated with fold belts
there is often further evidence for continental drift.
The grain size and composition ofthe sediments can
be used to determine the nature and direction of
their source. The source of sediments in the Cale-
donides of northern Europe lies to the west in a
location now occupied by the Atlantic, indicating
that, in the past, this location must have been occu
pied by continental crust. The source ofthe 1 Ga old
Roraima Formation sediments of Guyana lies to the
east and appears to be situated in northwest Africa.
Diamonds at the base of this formation increase in
size to the east, and in West Africa diamonds in
similar sediments also increase in size eastwards to
their source in the Sudan (Tarling & Tarling, 1971).
2 A£fe provinces. The correlation of the patterns of
ages across the southern Atlantic is shown in
Fig. 3.6, which illustrates the matching of both Pre-
cambrian cratons and rocks of Palaeozoic age (Hal-
lam, 1975).
3 Ipfneous provinces. Distinctive igneous rocks can be
traced between continents as shown in Fig. 3.7. This
50 CHAPTERS

Mountain belts
Younger
formed earl ier
than 260 m llion
years ago

^ ( Ural ^
/ Mountains

Greenland

^Europe]
Caledonian Mountains

Appalachian Mountains
tP a
■HpF South Pole Tethys Ocean
440 Ma ^>
South ^ /T
America 1) Africa ^ Andia ■
Australia,
Antarctica

Fig. 3.5 Alignment of mountain

belts on a reconstruction of Pangaea
(redrawn from Dewey, 1972).

s
[yl Mesozoic dolerite
[Hm Cratons Limit of Permo - Carboniferous glaclation
1: :. j Younger mobile belts • Precambrian anorthosite

Fig. 3.6 Correlation of cratons and younger mobile belts Fig. 3.7 Correlation of Permo-Carboniferous glacial
across the closed southern Atlantic Ocean (redrawn from deposits, Mesozoic dolerites and Precambrian anorthosites
Hallam, 1975). between the reconstructed continents of Gondwana.
 CONTINENTAL DRIFT 51

O flj

j*: O
< B
(D "O c 2 0) 2
O)

<
^
— £
to 3 -C
o ^
CO <
O
X <
C
a<
|l
5 < LU <

Jurassic

Middle-
Upper
Triassic

Lower
Triassic

Upper
Triassic
Permian

Lower Permian Permian

Basal Permian

Carboniferous

Devonian

ro

■1000 m

Tillite Coal Glossopteris Gangamopteris

Fig. 3.8 Correlation of stratigraphy between Gondwana

continents (redrawn from Hurley, 1968).
applies both to extrusive and intrusive rocks, such as
the belt of Mesozoic dolerite which extends through
southern Africa, Antarctica and Tasmania, and the
approximately linear trend of Precambrian an-
orthosites through Africa, Madagascar and India
(Smith & Hallam, 1970).
4 Strati£fmphic sections. Distinctive stratigraphic se
quences can also be correlated between adjacent
continents. Figure 3.8 shows stratigraphic sections
of the Gondwana succession, a terrestrial sequence of
sediments of Late Palaeozoic age (Hurley, 1968).
Marker beds of tillite and coal, and sediments con
taining Glossopteris and Gangamopteris flora (Section
3.5) can be correlated through South America, South
Africa, Antarctica, India and Australia.
5 Metallo£fenic provinces. Regions containing manga
nese, iron ore, gold and tin can be matched across
adjacent coastlines on such reconstructions (Evans,
1987).
3.4 Palaeoclimatology
The distribution of climatic regions on the Earth
is controlled by a complex interaction of many
phenomena, including insolation (i.e., latitude),
wind directions, ocean currents, elevation and topo
graphic barriers. The majority of these phenomena
are only poorly known in the geological record. On a
broad scale, however, latitude is the major control
ling factor of climate and, ignoring small micro-
climatic regions dependent on rare combinations of
other phenomena, it appears likely that the study of
climatic indicators in ancient rocks can be used to
infer, in a general sense, their ancient latitude. Con
sequently, palaeoclimatology, the study of past
climates (Frakes, 1979), may be used to demonstrate
that continents have drifted at least in a north-south
sense. It must be realized, however, that the Earth is
presently in an interglacial period, and so parallels
between modern and ancient climates may not be
completely justified. The important climatic indica
tors are listed below.
52 CHAPTER 3
1 Carbonates and reef deposits. These deposits are
restricted to warm water and occur within 30® ofthe
equator at the present day where temperatures fall in
the narrow range 25-30 ®C.
2 Evaporites. Evaporites are formed under hot arid
conditions in regions where evaporation exceeds sea
water influx and/or precipitation, and are usually
found in basins bordering a sea with limited or
intermittent connection to the ocean proper (Sec
tion 12.3.4). At the present day they do not form
near the equator, but rather in the arid subtropical
high pressure zones between about 10® and 40®
where the required conditions prevail, and it is be
lieved that fossil evaporites formed in a similar latitu
dinal range (Windley, 1984).
3 Red-beds. These include arkoses, sandstones,
shales and conglomerates which contain haematite.
They form under oxidizing conditions where there is
an adequate supply ofiron. A hot climate is required
for the dehydration oflimonite into haematite,and at
presentthey are restricted to latitudes ofless than 30 ®.
4 Coal and oil. These deposits are formed by the
accumulation and degradation of organic remains,
which require a warm, humid climate (Section
12.3.3). Both are predominantly formed in latitudes
ofless than 30®.
5 Phosphorites. At the present day phosphorites form
within 45 ® ofthe equator along the western margins
of continents where upwellings of cold, nutrient-
rich, deep water occur, or in arid zones at low
latitudes along east-west seaways.
6 Bauxite and laterite. These aluminium and iron
oxides only form in a strongly oxidizing environ
ment. It is believed that they only originate under the
conditions of tropical or subtropical weathering.
7 Desert deposits. Care must be employed in using
any of these deposits because desert conditions can
prevail in both warm and cold environments. How
ever, the dune bedding of desert sandstones can be
used to infer the ancient direction of the prevailing
winds. Comparison ofthese with the direction ofthe
modern wind systems found at their present latitude
can indicate if the continent has undergone any
rotation.
8 Glacial deposits. Glaciers and icecaps, excluding
those oflimited size formed in mountain ranges, are
limited to regions within about 30® of the poles at
the present day.
In addition to these sedimentological indicators
of climate, it is possible to make direct estimates
of palaeotemperatures based on oxygen isotope
measurements. In certain marine organisms which
secrete calcium carbonate under equilibrium condi
tions with the sea water, the isotopic composition of
oxygen in the calcite or aragonite depends upon
temperature (Hoefs, 1980). Consequently, by mak
ing assumptions about the isotopic composition of
the sea water, measurement of the ratio
provides an estimate of water temperature. This
method is not directly applicable to terrestrial depos
its, as the composition of the meteoric water from
which calcium carbonate is precipitated is highly
variable. However, information is available from
studies of isotopic compositions of oxygen and car
bon in snail shells and in cave deposits of calcium
carbonate (Faure, 1986).
The results of applying these palaeoclimatic tech
niques strongly indicate that continents have
changed their latitudinal position throughout geo
logical time. For example, during the Permian and
Carboniferous the Gondwana continents were
experiencing an extensive glaciation (Martin, 1981)
and must have been situated near the south pole
(Fig. 3.9). At the same time in Europe and eastern
USA,coal and extensive reef deposits were forming,
which subsequently gave way to hot deserts with
evaporite deposits. The northern continents were
thus experiencing a tropical climate in equatorial
latitudes.
3.5 Palaeontological evidence for
continental drift
Continental drift has affected the distribution of
ancient animals and plants (Briggs, 1987) by creat
ing barriers to their dispersal (Hallam, 1972). An
obvious example of this would be the growth of an
ocean between two fragments of a supercontinent
which prevented migration between them by terres
trial life-forms. The past distribution of tetrapods
implies that there must have been easy communica
tion between all parts of Gondwana and Laurasia.
Remains of the early Permian reptile Mesosaurus are
found in Brazil and southern Africa. Although
adapted to swimming, it is believed that Mesosaurus
was incapable of travelling large distances and could
not have crossed the 5000 km of ocean now present
between these two localities.
Oceans can also represent dispersal barriers to
certain animals which are adapted to live in relatively
 CONTINENTAL DRIFT 53

Fig. 3.9 Use of palaeoclimatic data to
control and confirm continental
reconstructions (redrawn from Tarling &
Tarling, 1971, in Continental Drifts with
permission from Unwin Hyman Ltd).
Areas of tropical coal forests at 300 Ma which
m some 50 Ma later became vast hot deserts
Areas of glaciation between 250 and 300 Ma which
with arrows indicating known directions of ice movement

shallow marine environments. The widespread dis
persal of marine invertebrates can only occur in their
larval stages when they form part of the plankton
(Hallam, 1973b). For most species the larval stage is
too short-lived to exist for the duration of the cross
ing of a large ocean. Consequently, ancient faunai
province boundaries frequently correlate with su
tures, which represent the join lines between ancient
continents brought into juxtaposition by the con
sumption of an intervening ocean. The distribution
of Cambrian trilobites strongly suggests that in
Lower Palaeozoic times there existed several conti
nents separated by major ocean basins. The similarity
between ammonite species now found in India,
Madagascar and Africa indicates that only shallow
seas could have existed between these regions in
Jurassic times.
Palaeobotany similarly reveals the pattern ofconti
nental fragmentation. Before break-up, all the Gond-
wana continents supported, in Carboniferous times,
the distinctive Glossopteris and Gangamopteris floras
(Hurley, 1968; Plumstead, 1973), which are be
lieved to be cold climate forms. At the same time a
varied tropical flora existed in Laurasia (Fig. 3.10).
After fragmentation, however, the flora of the indi
vidual continents diversified and followed separate
paths of evolution.
A less obvious form of dispersal barrier is climate,
as the latitudinal motions of continents can create
climatic conditions unsuitable for certain organisms.
Indeed, relative continental movements can modify
the pattern of ocean currents, mean annual temper
ature, the nature of seasonal fluctuations and many
other factors (Valentine & Moores, 1972). Also,
plate tectonic processes can give rise to changes in
topography which modify the habitats available for
colonization.
The diversity ofspecies is also controlled by conti
nental drift. Diversity increases towards the equator
so that the diversity at the equator is about ten times
that at the poles. Consequently, drifting in a north-
south direction would be expected to control the
54 CHAPTERS

i ! i mMi:
\ : i-i

/
^ ,-r
/ ..

■ flora
TropicalwithLauraslan
many
LV\^
Vvw
Polar Gondwanan
flora with few species
Tethys marine
foramlnlfera
species and areas of eurydesma fauna
of identical reef-forming
corals followed later by
tropical coal forests

Fig. 3.10 Present distributions of Pangaean flora and

fauna (redrawn from Tariing & Tarling, 1971,in
Continental Drifts with permission from Unwin Hyman
Ltd).

diversity on a continent. Diversity also increases

with continental fragmentation (Kurten, 1969). For
example, 20 orders of reptiles existed in Palaeozoic
times on Pangaea, but with its fragmentation in
Mesozoic times 30 orders of mammals developed on
the various continents. Each continental fragment
becomes a nucleus for the adaptive radiation of
the species as a result of genetic isolation and the
morphological divergence of separate faunas. Con Fig. 3.11 Correlation of invertebrate diversity with time
sequently, more species evolve as different types and continental distribution. A, earlier Pangaea;
occupy similar ecological niches. Figure 3.11, from B,fragmentation of earlier Pangaea producing oceans
Valentine & Moores(1972), compares the variation preceding Caledonian (1), Appalachian (2), Variscan (3)
and Uralian (4)orogenies; C,suturing during Caledonian
in the number of fossil invertebrate families existing
and Acadian orogenies; D,suturing during Appalachian
in the Phanerozoic with the degree of continental
and Variscan orogenies; E,suturing of Urals and
fragmentation as represented by topological models. reassembly of Pangaea; F, opening of Tethys Ocean;
The correlation between number ofspecies and frag G,fragmentation of Pangaea. a, Gondwana;
mentation is readily apparent. An example of such b, Laurasia; c. North America; d. South America;
divergence is the evolution of anteating mammals. e, Eurasia; f, Africa; g, Antarctica; h, India; i, Australia
As the result of evolutionary divergence this special (after Valentine & Moores, 1972, with permission from
ized mode of behaviour is followed by different the University of Chicago Press).

orders on separated continents: the antbears (Eden
tata) ofSouth America, the pangolins (Pholidota) of
northeast Africa and southeast Asia, the aardvarks
(Tubulidentata) of central and southern Africa and
the spiny,anteaters(Monotremata) of Australia.
Continental suturing leads to the homogenization
offaunas by cross-migration (Hallam, 1972)and the
extinction ofany less well-adapted groups which face
stronger competition. Conversely, continental split
ting leads to the isolation of faunas which then
follow their own distinct evolutionary development.
For example, marsupial mammals probably reached
Australia from South America in the Upper Creta
ceous along an Antarctic migration route (Hallam,
1981) before the Late Cretaceous marine transgres
sion flooded west Antarctica and closed the route for
the later evolving placental mammals. Sea floor
spreading then ensured the isolation of Australia
when the sea level dropped, and the marsupials
evolved unchallenged until the Late Tertiary when
the collision of Asia and New Guinea allowed the
colonization of placental mammals from Asia.
3.6 Palaeomagnetism
3.6.1 Introduction
The science of palaeomagnetism is concerned with
studies of the fossil magnetism that is retained in
certain rocks. If this magnetism originated at the
time the rock was formed, measurement of its direc
tion can be used to determine the latitude at which
the rock was created. If this latitude differs from the
present latitude at which the rock is found, very
strong evidence has been furnished that it has moved
over the surface of the Earth. Moreover, if it can be
shown that the pattern of movement differs from
that ofrocks ofthe same age on a different continent,
relative movement must have occurred between
them. In this way, palaeomagnetic measurements
demonstrated that continental drift has taken place,
and provided the first quantitative estimates of rela
tive continental movements. For fuller accounts of
the palaeomagnetic method,see Tarling (1983) and
Piper(1987).
3.6.2 Rock magnetism
Palaeomagnetic techniques make use ofthe phenom
enon that certain minerals are capable of retaining a
CONTINENTAL DRIFT 55
record of the past direction of the Earth's magnetic
field. These minerals are all pamm(i£fnetic^ that is,
they contain atoms which possess an odd number of
electrons. Magnetic fields are generated by the spin
and orbital motions of the electrons. In shells with
paired electrons, their magnetic fields essentially
cancel each other. The unpaired electrons present in
paramagnetic substances cause the atoms to act as
small magnets or dipoles.
When a paramagnetic substance is placed in a weak
external magnetic field, such as the Earth's field, the
atomic dipoles rotate so as to become parallel to the
external field direction. This induced ma£fnetization is
lost when the substance is removed from the field as
the dipoles return to their original orientations.
Certain paramagnetic substances which contain a
large number of unpaired electrons are ferro
magnetic. The magnetic structure ofthese substances
tends to devolve into a number of mapfnetic domains.,
within which the atoms are coupled by the interac
tion ofthe magnetic fields ofthe unpaired electrons.
This interaction is only possible at temperatures
below the Curie temperature., as above this temper
ature the energy level is such as to prohibit inter
atomic magnetic bonding and the substance then
behaves in an ordinary paramagnetic manner.
Within each domain the internal alignment of
linked atomic dipoles causes the domain to possess a
net magnetic direction. When placed in a magnetic
field the domains whose magnetic directions are in
the same sense as the external field grow in size at the
expense ofdomains aligned in other directions. After
removal from the external field a preferred direction
resulting from the growth and shrinkage of the
domains is retained so that the substance exhibits
an overall magnetic directionality. This retained
magnetization is known as permanent or remanent
mapfnetism.
3.6.3 Natural remanent magnetization
Rocks can acquire a natural remanent magnetization
(NRM) in several ways. If the NRM forms at the
same time as the rock it is referred to as primary; if
acquired during the subsequent history ofthe rock it
is termed secondary.
The primary remanence ofigneous rocks is known
as thermoremanent magnetization (TRM). It is ac
quired as the rock cools from its molten state to
below the Curie temperature, which is realized after
56 CHAPTER 3
solidification. At this stage its ferromagnetic minerals
pick up a magnetism in the same sense as the geo
magnetic field at that time which is retained during
its subsequent history.
The primary remanence in clastic sedimentary
rocks is known as detrital remanent magnetization
(DRM). As the sedimentary particles settle through
the water column, any ferromagnetic minerals pre
sent align in the direction of the geomagnetic field
(Fig. 3.12). On reaching bottom the particles flatten
out, and if of elongate form preserve the azimuth of
the geomagnetic field but not its inclination. After
burial, when the sediment is in a wet slurry state, the
magnetic particles realign with the geomagnetic field
as a result of microseismic activity, and this orienta
tion is retained as the rock consolidates.
Secondary NRM is acquired during the subsequent
history of the rock according to various possible
mechanisms. Chemical remanent ma£fnetization
(CRM)is acquired when ferromagnetic minerals are
formed as a result of a chemical reaction, such as
oxidation. When ofa sufficient size for the formation
of one or more domains the grains become magnet
ized in the direction of the geomagnetic field at the
time of reaction. Isothermal remanent magnetization
(IRM)occurs in rocks which have been subjected to
strong magnetic fields, as in the case of a lightning
strike. Viscous remanent magnetization (VRM) may
arise when a rock remains in a relatively weak mag
netic field over a long period oftime as the magnetic
Direction of
Magnetic .geomagnetic
particle Y field
Wm
domains relax and acquire the external field direc
tion.
Some CRM may be acquired soon after formation,
for example during diagenesis, or during a metamor-
phic event of known age, and hence preserve useful
palaeomagnetic information.
CRM, TRM and DRM tend to be 'hard', and
remain stable over long periods of time, whereas
certain secondary components of NRM, notably
VRMs,tend to be 'soft' and lost relatively easily. It is
thus possible to destroy the 'soft' components and
isolate the 'hard' components by the technique of
magnetic cleaning. This involves monitoring the ori
entation and strength of the magnetization ofa rock
sample as it is subjected either to an alternating field
of increasing intensity or to increasing temperature.
Having isolated the primary remanent magnetiza
tion, its strength and direction are measured with
either a spinner magnetometer or superconducting
magnetometer. The latter instrument is extremely
sensitive and capable of measuring NRM orienta
tions of rocks with a very low concentration of
ferromagnetic minerals.
3.6.4 The past and present
geomagnetic field
The magnetic field of the Earth approximates the
field that would be expected from a large bar magnet
embedded within it inclined at an angle ofabout 12®
Water
Unconsolidated
sediment
Consolidated
sediment
Fig.3.12 Development of detrital
remanent magnetization.

to the spin axis. The actual cause ofthe geomagnetic
field is certainly not by such a magnetostatic process,
as the magnet would have to possess an unrealisti-
cally large magnetization and would lie in a region
where the temperatures would be greatly in excess of
the Curie temperature.
The geomagnetic field is believed to originate from
a dynamic process, involving the convective circula
tion of electrical charge in the fluid outer core,
known as magnetohydrodynamics. However, it is
convenient to retain the dipole model as simple
calculations can then be made to predict the geo
magnetic field at any point on the Earth.
The geomagnetic field undergoes progressive
changes with time, resulting from variations in the
convective circulation pattern in the core, known as
secular variation. One manifestation ofthis phenom
enon is that the direction of the magnetic field at a
particular geographic location rotates irregularly
about the direction implied by an axial dipole model
with a periodicity of a few thousand years. In a
palaeomagnetic study the effects of secular variation
can be removed by collecting samples from a site
which span a stratigraphic interval of many thou
sands of years. Averaging the data from these speci
mens should then remove secular variation so that
for the purposes of palaeomagnetic analysis the geo
magnetic field in the past may be considered to
originate from a dipole aligned along the Earth's axis
of rotation.
Palaeomagnetic measurements provide the inten
sity, azimuth and inclination of the primary reman-
ent magnetization, which reflect the geomagnetic
parameters at the time and place at which the rock
was formed. By assuming the axial geocentric dipole
model for the geomagnetic field discussed above,the
inclination I can be used to determine the palaeo-
latitude (|) at which the rock formed according to
the relationship tan
| (|) = tan I. With a knowledge of
the palaeolatitude and the azimuth of the primary
remanent magnetization, that is, the ancient north
direction, the apparent location of the palaeopole
can be computed. Such computations, combined
with age determinations ofthe samples by radiomet-
ric or biostratigraphic methods, make possible the
calculation of the apparent location of the north
magnetic pole at a particular time for the continent
from which the samples were collected. Palaeo
magnetic analyses of samples of a wide age range
can then be used to trace how the apparent pole
CONTINENTAL DRIFT 57
position has moved over the Earth's surface.
It is important to recognize that remanent magnet
ization directions cannot provide an estimate of
palaeolongitude, as the assumed dipole field is axi-
symmetric. There is a consequent uncertainty in the
ancient location of any sampling site, which could
have been situated anywhere along a small circle,
defined by the palaeolatitude, centred on the pole
position.
Ifa palaeomagnetic study provides a magnetic pole
position different from the present pole, it implies
either that the magnetic pole has moved throughout
geological time, that is, the magnetic pole has wan
dered relative to the rotational pole, or if the poles
have remained stationary that the sampling site has
moved, that is, continental drift has occurred. It
appears that wandering of the magnetic pole away
from the geographic pole is unlikely because the
Coriolis force is thought to be crucial in determining
the axial symmetry ofthe dipole field. Consequently,
palaeomagnetic studies can be used to provide a
quantitative measure ofcontinental drift.
An early discovery of palaeomagnetic work was
that in any one study about half of the samples
analysed provided a primary remanent magnetiza
tion direction in a sense 180® diflferent from the
remainder. Although the possibility ofself-reversal of
rock magnetism remains, it is believed to be a rare
phenomenon, and so these data are taken to reflect
changes in the polarity ofthe geomagnetic field. The
field can remain normal for perhaps a million years
and then, over an interval of a few thousand years,
the north magnetic pole becomes the south magnetic
pole and a period of reversed polarity obtains. Polar
ity reversals are random, but obviously affect all
regions of the Earth synchronously so that, coupled
with radiometric or palaeontologic dating, it is poss
ible to construct a polarity timescale. This subject
will be considered further in Chapter 4.
3.6.5 Apparent polar wandering curves
Palaeomagnetic data can be displayed in two ways.
One way is to image what is believed to be the true
situation,that is, plot the continent in a succession of
positions according to the ages of the sampling sites
(Fig. 3.13a). This form of display requires the as
sumption of the palaeolongitudes of the sites. The
other way is to regard the continent as remaining at
a fixed position and plot the apparent positions of
58 CHAPTER 3

'Triassic-Jurassic

Present

jf Ifc ;
r7«:S« L Permian
^ / lS*dSl i
V

Pennsvlvanian

Cambrian

Devonian

Fig.3.13 Two methods of displaying paiaeomagnetic Subsequent work has modified the detail ofthe
data:(a) assuming fixed magnetic poles and applying movements shown. Note that the south pole has been
latitudinal shifts to the continent;(b) assuming a fixed plotted (redrawn from Creer, 1965, with permission from
continent and plotting a polar wandering path. the Royal Society).

the poles for various times to provide an apparent

polar wandering (APW) path (Fig. 3.13b). As dis W180°E

cussed above, this representation does not reflect real

events, but it overcomes the lack of control of
palaeolongitude and facilitates the display ofinform
ation from different regions on the same diagram.
The observation that the apparent position of the
pole differed for rocks ofdifferent ages from the same
continent demonstrated that continents had moved
over the surface ofthe Earth. Moreover,the fact that
APW paths were different for different continents
demonstrated unequivocally that relative move
ments of the continents had taken place, that is,
continental drift had occurred. Paiaeomagnetic stud
ies thus confirmed and provided the first quantitative
measurements ofcontinental drift. Figure 3.14 illus
trates the APW paths for North America, Eurasia
(with separate spur for Siberia), Africa, Australia and
India. WO°E
APW paths can be used to interpret motions,
Fig. 3.14 Apparent polar wandering paths for North
collisions and disruptions of continents (Piper, America, Europe, Siberia, Africa, Australia and India.-G,
1987) and are especially useful for pre-Mesozoic Cambrian;S, Silurian; C,Carboniferous; P,Permian;T,
continents whose movements cannot be traced by Triassic; J, Jurassic; K,Cretaceous; E, Eocene (redrawn
the pattern of magnetic lineations in their surround from Condie, 1982,Plate Tectonics and Crustal Evolution.
ing ocean basins (Section 4.5). Figure 3.15 repre- with permission from Pergamon Press Ltd).
 CONTINENTAL DRIFT 59

Ocean closure and

suturing of continents

Rifting and
creation of
ocean

Sutured continent

Single continent

Fig. 3.15 Palaeomagnetic signature of plate divergence method of continental reconstruction whereby the
and convergence (redrawn from Piper, 1987, with pole and angle of rotation necessary to bring the
permission from Open University Press). APW paths of two continents together are deter
mined. Indeed, modern palaeomagnetic measure
sents the full Wilsonian cycle (Section 10.4) of the ments,summarized in Van der Voo(1990),confirm
opening and closure of an ocean basin between two the precision ofthe reconstruction ofthe continents
continents. Before rifting the two segments A and B around the Atlantic of Bullard et al. (1965). As
ofthe initial continent have similar APW paths. They
are unlikely to be identical as it is improbable that the
initial rift and final suture would coincide. After
rifting the two segments describe diverging APW
paths until the hairpin at time 8 signals a change in
direction of motion to one of convergence. After
suturing at time 12 the two segments follow a
common polar track. Figure 3.16 shows part of this
seqiience illustrated by the continents of Pangaea.
Gondwana and Laurasia converged during Silurian
times. From then until the Mesozoic they followed a
common APW path, indicating that they formed the
single supercontinent Pangaea. During the Meso
Antarctica
zoic,Pangaea split into the present continents, which
each followed a separate APW path.
if-S. Arrierica.
1
3.6.6 Continental reconstructions
based on paiaeomagnetism
America

Careful examination of Fig. 3.14 reveals that the Europe

APW paths ofNorth America and Eurasia are similar Fig. 3.16 APW path of the south pole during the
in shape, showing that after a suitable rotation these Phanerozoic (redrawn from McElhinny, 1973,
continents would have followed an identical track Ptilaeoma£fnetism and Plate Tectonics^ with permission
over the Earth's surface. This, then, represents a from Cambridge University Press).
60 CHAPTER 3
Australia
Africa
Africa
South America
Africa—
Fig.3.17 Gondwana reconstructions based on polar
wandering curves. Africa is in its present position and the
other APW curves have been rotated to be coincident with
particular geometrical fits of Gondwana:(a)the fit of
Africa and South America;(b)-(e) eastern Gondwana fits.
stated before, palaeomagnetic data provide no in
formation on palaeolongitude, and so the east-west
separation of the continents is indeterminate. How
ever, by making use of the criterion that the conti
nental margins should be as close together as possible
without overlapping, it is possible to make adequate
reconstructions. More accurate reconstructions can
be made for Mesozoic and Tertiary movements by
making use of oceanic magnetic lineations (Section
4.6). However, for Palaeozoic and Precambrian
times, such palaeomagnetic methods represent the
only quantitative methods of reconstruction avail
able.
Figure 3.17, from Tarling (1983), shows recon
structions of Gondwana accomplished by APW path
matching. There is wide agreement on the fit be
tween South America and Africa (Fig. 3.17a), which
Africa
Australia
Africa
Australia
* Australia
Reconstructions shown as insets are based on the studies
of different workers (redrawn from Tarling, 1983;
Palaeoma£fmtism—Principles and Applications in Geologyy
Geophysics and Archaeology^ with permission from
Chapman & Hall, Ltd).
is confirmed by geometrical arguments. The match
ing of APW paths for the other Gondwana con
tinents, however, is more subjective, and four
possibilities are shown in Fig. 3.17b-e. The history
of continental movements since Devonian times,
based on palaeomagnetic data, has been summarized
by Irving (1979).
Further reading
Bullard,E.G.,Everett, J.E.&Smith,A.G.(1965).The fit of
the continents around the Atlantic. Phil. Trans. Roy. Soc.
Lond. 258A,41-51.
Dewey, J.F. (1972) Plate tectonics. Sci. Am. 226(5), 56-
68.
Frakes, L.A. (1979) Climates Throughout Geologic Time.
Elsevier, New York.

Hallam, A.(1973b) Provinciality, diversity and extinction
of Mesozoic marine invertebrates in relation to plate
movements. In Tarling, D.H. & Runcorn, S.K. (eds)
Implications of Continental Drift to the Earth Sciences^ 1,
pp.287-94. Academic Press, London.
Piper, J,D.A.(1987)Palaeomapfnetism and the Continental
Crust. Open University Press, Milton Keynes.
CONTINENTAL DRIFT 61
Powell, C.McA., Johnson, B.D. & Veevers, J.J. (1980) A
revised fit of east and west Gondwanaland, Tectonophys.
63,13-29.
Tarling, D.H.(1983)Palaeomapfnetism. Chapman & Hall,
London.
Tarling, D.H.& Tarling, M.P.(1971) Continental Drift: a
Study ofthe Earth^s Movin£[ Surface. Bell, London.
4/ Sea floor spreading
4.1 Introduction
By the late 1950s much evidence for continental drift
had been assembled, but the theory was not generally
accepted. Up to this time, work had concentrated
upon determining the predrift configurations of
the continents and assessing their geological conse
quences. The paths by which the continents had
attained their present positions had not been deter
mined. In order to study the kinematics of conti
nental drift it was necessary to study the regions
now separating once juxtaposed continents. Conse
quently, at this time interest moved from the conti
nents to the intervening ocean basins.
Any kind of direct observation of the sea floor,
such as drilling, dredging or submersible operations,
is time consuming, expensive and provides only a
low density of data. Much of the information avail
able over oceanic areas has therefore been provided
by geophysical surveys undertaken from ships or
aircraft. One such method involves measuring vari
ations in the strength of the Earth's magnetic field.
This is accomplished using either fluxgate, proton
precession or optical absorption magnetometers,
which require little in the way of orientation so that
the sensing element can be towed behind the ship or
aircraft at a sufficient distance to minimize their
magnetic effects. In this way total field values are
obtained which are accurate to ± I nanotesla(nT)or
about one part in 50 000. Magnetometers provide a
virtually continuous record of the strength of the
geomagnetic field along their travel paths. These
absolute values are subsequently corrected for the
externally induced magnetic field variations which
give rise to a diurnal effect and the regional magnetic
field arising from that part of the magnetic field
generated in the Earth's core. In theory the resulting
ma£fnetic anomalies should then be due solely to
contrasts in the magnetic properties of the under
lying rocks. The anomalies originate from the gener
ally small proportion of ferromagnetic minerals
62
(Section 3.6.2) contained within the rocks, of which
the most common is magnetite. In general ultrabasic
and basic,rocks contain a high proportion of mag
netite and thus give rise to large magnetic anomalies.
Metamorphic rocks are moderately magnetic and
acid igneous and sedimentary rocks are usually only
weakly magnetic. A full account of the magnetic
surveying method is given in Kearey & Brooks
(I99I).
On land, magnetic anomalies reflect the variable
geology of the upper continental crust. The oceanic
crust, however, is known to be laterally uniform
(Section 2.4.4)and so unless the magnetic properties
are heterogeneous it would be expected that marine
magnetic anomalies would reflect this compositional
uniformity.
4.2 Marine magnetic anomaiies
Magnetic surveying is easily accomplished, and mea-
surements have been carried out from survey vessels
since the mid-1950s both on specific surveys and
routinely on passage to the locations ofother ocean-
ographic investigations.
A most significant magnetic anomaly map
(Fig.4.1) was constructed after detailed surveys off
the western seaboard of North America (Mason &
Raff, 1961;Raff& Mason,1961).The magnetic field
was shown to be anything but uniform, and revealed
an unexpected pattern of stripes defined by steep
gradients separating linear regions ofhigh amplitude
positive and negative anomalies. These magnetic
lineations are remarkably persistent, and can be
traced for many hundreds ofkilometres. Their conti
nuity, however, is interrupted at major oceanic frac
ture zones, where the individual anomalies are offset
laterally by distances of up to 1100 km.
Subsequent surveys have shown that magnetic
lineations are present in virtually all oceanic areas.
They are generally 10-20 km wide and characterized
by a peak-to-peak amplitude of 500-1000 nT. They

150° 145°
iim/fwimm aii
Mendocino
Fig.4.1 Magnetic anomaly
lineations in the northeastern
Pacific Ocean. Positive
anomalies in black; also shown
are the oceanic fracture zones at
which the lineations are offset
(after Menard, 1964, with
permission from the estate of
Late Professor H. William
Menard).
run parallel to the crests of the mid-ocean ridge
system (Chapter 6), and are symmetrical about the
ridge axes (Fig. 4.2).
The source of these linear magnetic anomalies
cannot be oceanic layer I, >vhich is made up of
non-magnetic sediments. They cannot originate at a
depth corresponding to layer 3 as sources solely
within this layer would be incapable of generating
the steep anomaly gradients.The source ofthe anom
alies must therefore be, at least in part, in oceanic
layer 2. This conclusion is consistent with the basal
tic composition of layer 2 determined by dredging
and drilling (Section 2.4.6), since basalt is known to
contain a relatively high proportion of magnetic
minerals. The magnetic lineations therefore confirm
that layer 2is everywhere composed ofthis rock type.
If magnetic lineations are generated by a layer of
homogeneous composition, how do the magnetic
contrasts originate that are responsible for the juxta
position of large positive and negative magnetic
anomalies.^ The shape of a magnetic anomaly is
SEA FLOOR SPREADING 63
140° 135° 130° 125°
Juan de Fuca ridge crest
^Cape:
JVIendocIno
Pioneer
Francisco
determined by both the geometric form ofthe source
and the orientation of its magnetization vector.
Oceanic layer 2 maintains a relatively constant depth
and thickness and any anomalies arising because of
rugged topography on the top of the layer would
attenuate too rapidly to account for the amplitude of
the anomalies observed on the surface 3-7 km above
the sea bed. Consequently, the lineations must arise
because adjacent blocks oflayer 2 are magnetized in
different directions. Figure 4.3 shows an interpreta
tion of magnetic anomalies observed over the Juan
de Fuca Ridge in the northeastern Pacific. Layer 2 has
been divided into a series of blocks running parallel
to the ridge crest which have been assigned magneti
zation vectors which are either in the direction ofthe
ambient geomagnetic field or in the reversed direc
tion. The interpretation shows that the observed
anomalies are simulated by a model in which the
intensities of the magnetization vary, and that rela
tively high values of some 10 A m"^ are required to
produce the necessary contrasts.
64 CHAPTER 4
22° 18°
1000 m
Fig.4.2 Magnetic lineations either side of the mid-
Atlantic ridge south ofIceland. Positive anomalies in
black (after Heirtzler a^/., 1966,in Deep Sea Research 13,
428, with permission from Pergamon Press Ltd).
4.3 Sea floor spreading
In the early 1960s, Dietz (1961) and Hess (1962)
had proposed that continental drift might be accom
plished by a process which Dietz termed sea floor
spreadin£(. It was suggested that new oceanic litho-
sphere is created by the upwelling and partial melting
of material from the asthenosphere at the ocean
ridges. As the ocean gradually grows wider with the
progressive creation of lithosphere, the continents
marginal to the ocean are moved apart. The drift
between North America and Europe, for example,
would have been accomplished by the gradual
growth of the Atlantic Ocean over the past 180 Ma.
Since the Earth is not increasing in surface area by
any'significant amount (Section 11.3), the increase
in size ofthose oceans growing by sea floor spreading
would be balanced by the destruction of lithosphere
at the same rate in another, shrinking, ocean by
subduction at deep sea trenches situated around its
margins.
The driving mechanism of these movements was
believed to be convection currents in the sub-
lithospheric mantle (Fig. 1.5). These were thought
to form cells in which mantle ascended beneath
ocean ridges, bringing hot material to the surface and
giving rise to new lithosphere. The flow then moved
horizontally away from the ridge, driving the litho
sphere laterally in the same direction by viscous drag
on its base, and finally descended back into the deep
mantle at the ocean trenches, assisting the subduc
tion ofthe lithosphere. This possible mechanism will
be discussed more fully in Section 11.7.
4.4 Geomagnetic reversals
The possibility that the geomagnetic field reverses
polarity was first suggested during the early part of
the twentieth century, when it was noted that re
versed magnetizations were present in some rock
samples, and that the low amplitudes of magnetic
anomalies observed over certain volcanic sequences
were explicable in terms of a reversed magnetization
vector.
By the early 1960s the concept of geomagnetic
field reversals was being revived, both because of
the large number of palaeomagnetic measurements
revealing reversed magnetization and the demon
stration that self-reversal, whereby a reversed magne
tization can originate from interaction with normally
magnetized material, was a very rare phenomenon.
By the mid-1960s, following the work of Cox et al.
(1964; 1967) on lava flows erupted within the past
few million years, the concept was widely accepted.
More recently, palaeomagnetic studies of rapidly
deposited sediments and slowly cooling igneous in
trusions have shown that a magnetic reversal occurs
over a time interval of about 2000-3000 years. It is
accompanied by a reduction in field intensity to
about 10% of its normal value which commences
some time before the reversal and continues for some
time afterwards, with a total duration of about
10 000 years.
There is no general theory for the origin of the
geomagnetic field. However,it is recognized that the
main part originates within the Earth, and must be
caused by dynamic processes. A magnetostatic origin
appears impossible as no known material is suffi
ciently magnetic to give rise to the magnitude ofthe

• Calculated
Am ' 0
Fig. 4.3 Interpretation of a magnetic anomaly profile
across the Juan de Fuca ridge, nonheastern Pacific Ocean,
in terms of normal and reversed magnetizations of two-
dimensional rectangular blocks of oceanic layer 2. The
arrow marks the ridge crest(redrawn from Bott, 1967,
with permission from the Royal Astronomical Society).
field observed at the surface, and subsiirface temper
atures would be well in excess of the Curie point,
even given that its dependence upon pressure is
largely unknown. The temporal variation of the
internally generated field would also be inexplicable
with such a model.
The geomagnetic field is believed to originate by
magnetohydrodynamic processes within the fluid
(outer) part of the Earth's core, magnetohydro-
dynamics being that branch of physics concerned
with the interaction offluid motions and electromag
netic fields. Indeed,this process is also believed to be
responsible for the magnetic fields of other planets
and certain stars. The process requires the celestial
body to be rotating and to be partly or completely
composed of a mobile fluid which is a good electrical
conductor. The turbulent or convecting fluid consti
tutes a dynamo, because if it moves in a pre-existing
SEA FLOOR SPREADING 65
magnetic field it generates an electric current which
has a magnetic field associated with it. Thus, once
'excited', the dynamo becomes self-perpetuating as
long as there is a primary energy source to maintain
the convection currents. The process is complex, and
analytic solutions are only available for the very
simplest configurations, which cannot approach the
true configuration in the core. The field is thought
to be maintained by convection in the outer core
which is thermally or gravitationally driven (Jacobs,
1975).
Since mathematical solution ofthe magnetohydro
dynamic dynamo is not yet possible, although nu
merical modelling is now possible (Glatzmeier &
Roberts, 1995),the geomagnetic field has been mod
elled in two different ways. One approach is to derive
a complete mathematical solution for a simple me
chanical analogue of the geomagnetic field. A widely
utilized analogue is a disc dynamo consisting of two
rotating conducting discs interconnected by coils of
wire. Under certain conditions the electric currents
flowing in the wire change direction and reverse the
magnetic field direction. Although this to some ex
tent mimics the behaviour ofthe geomagnetic field, a
major drawback to the analogy is that this type of
66 CHAPTER 4
Reversed
Fig.4.4 Possible mechanism for reversal of the
geomagnetic field via an axial quadrupole transition field.
Shaded areas on the left side ofeach sphere represent
positive radial fields, lines on the right the magnetic field
lines (after Fuller, 1987, with permission from Nature
326,132. Copyright © 1987 Macmillan Magazines Ltd).
dynamo depends for its operation on its asymmetry,
while the core is radially symmetrical.
Another approach is to consider the magnetic
reversals as a random, stochastic process, and to
attribute reversals to large convection cells in the
core. These cells develop, as a result of the Coriolis
force produced by the Earth's rotation, into spirals
similar to atmospheric cyclones (Parker, 1969). The
cyclones remove the radial symmetry from the
system as they rotate in opposite directions in the
northern and southern hemispheres. Links between
the cyclones and magnetic reversals assume that the
former are randomly distributed, and various work
ers have suggested the critical conditions under
which reversals occur. Thus, for example, reversals
might be immediately preceded by the random at
tainment ofa highly symmetrical configuration or by
a concentration of spirals near the geographic poles.
The most recent theories of the mechanisms of
geomagnetic reversals (Fuller, 1987) postulate that
the process begins in a localized region of the core
(Hoffman, 1979). Bloxham &Gubbins(1985) have
produced models of the magnetic field at the core-
mantle boundary for the period 1715-1980 by con
tinuing downward the observed field at the Earth's
surface. Gubbins (1987) has used such models to
suggest that the present decay of the geomagnetic
field is caused by the localized development of re
versed magnetic flux in the southern hemisphere. He
predicts that this may be the precursor to a reversal
via the sequence illustrated in Fig. 4.4, although it is
statistically likely that the field intensity will recover
without reversing, as it appears to have done fre
quently in the past. This work implies that all rever
sals originate in the same low latitude region of the
southern hemisphere, whose location is probably
Normal
controlled by topography or temperatures at the base
of the mantle. These latter effects are controlled in
turn by convection in the mantle.
It is estimated that a geomagnetic reversal takes
about 5000 years, and is accompained by a progres
sive decrease in field strength. However, the mea
surement of changes in magnetization direction
within a suite of thin, rapidly cooled lava flows(Coe
et 1995) has produced results which can be
interpreted to suggest that, at times, the geomag
netic field may have changed direction at a rate of6®
per day over several days. The capture of a geomag
netic reversal by a rock sequence in this way is very
rare, but if this result is true, it will have very
important ramifications for our models of magnetic
field generation.
The periodicity of geomagnetic reversals is highly
variable. The number of reversals per million years
averaged over 10 Ma intervals has not remained con
stant throughout geological time (Fig. 4.5). There
has been a gradual increase in reversal frequency
during the Cenozoic (Livermore et 1984; Sto-
thers, 1986). This increase started in the Late Creta
ceous, after the field had been ofdominantly normal
polarity for 35 Ma. Land-based investigations have
0
1
o
c
0
3
C7
0
40 60 80 100 120 140 160
Age(Ma)
Fig.4.5 Relationship between frequency of magnetic
reversals and age (after Livermore et al.^ 1984, with
permission from the Royal Astronomical Society).

0 4 3 2 1
I 1—pJ—, I
1 Gilbert .Gauss Matu-
I reverse|normal
"reverse
10-60 mm
Oceanic
Lithosphere
Asthenosphere
Fig.4.6 Sea floor spreading and the generation of
magnetic lineations by the Vine-Matthews hypothesis
(redrawn from Bott, 1982a, with permission from Edward
Arnold Ltd).
revealed a similar prolonged period of reversed
polarity in Upper Carboniferous and Permian time.
Stothers (1986) has suggested that there is a period
icity of reversals of about 30 Ma superimposed on
the general trends of Fig. 4.5. There is also strong
evidence that the ratio ofthe mean lengths ofnormal
and reversed polarity intervals has changed through
out geological time (McFadden & Merrill, 1984).
Cox(1975)attributed changes in the frequency of
magnetic reversals to changes in the conditions at the
core-mantle boundary. Modern theories have ex
tended this idea and propose that there exists a
coupling between the core and mantle in such a way
that the magnetic reversal frequency can be linked to
convection in the mantle and hence to the mechan
ism underlying plate tectonics. Courtillot & Besse
(1987)have proposed a model in which the thickness
of the boundary layers on either side of the core-
mantle boundary affect convective processes in both
the outer core and mantle. In the region above the
core-mantle boundary, a thermal boundary layer
releases heat pulses to the mantle at a rate dependent
on its thickness. These thermal pulses ascend along
paths influenced by mantle convection cells, eventu
ally to reach the lithosphere as mantle hotspots.
Below the core-mantle boundary cold thermal
pulses are emitted, again at a rate proportional to its
thickness. These thermals affect the pattern of circu
lation within the outer core and moderate the rever
sal rate. The coupling between core and mantle is
SEA FLOOR SPREADING 67
Age(Ma)
0 1 2 3 4 '
I ■ ■ ' . '. '
Brunhesj Matu- [Gauss Gilbert
Innrmai
I normal !j vama
yarna ^ normal reverse
^reverse
Layer 2
Layer 3
-60 mm a
Moho
Upper
mantle
thus seen to be controlled by the thickness ofthermal
boundary layers either side ofthe Gutenberg discon
tinuity. The model explains well the correlation
between the hotspot activity and reversal frequency
noted by Vogt(1975).
Larson & Olson (1991) have noted the inverse
correlation between magnetic reversal frequency and
mantle plume activity (Section 5.4), as measured by
the volume production rates of plume products. The
mechanism proposed involves the cooling of the
outermost core as plume material rises from the
core-mantle boundary, which results in more rapid
convective overturn in the outer core to replace the
lost heat. Above a certain level,this increased convec
tion might result in a magnetic reversal.
4.5 The Vine-Matthews
hypothesis
It is perhaps surprising to note that magnetic maps of
the oceans showing magnetic lineations (Section
4.2) were available for several years before the true
significance of the lineations was realized. The hy
pothesis of Vine & Matthews (1963) was of elegant
simplicity and combined the notion of sea floor
spreading (Section 4.3) with the phenomenon of
geomagnetic field reversals (Section 4.4).
The Vine-Matthews hypothesis explains the
formation of magnetic lineations in the following
way. New oceanic crust is created by the solidification
of magma injected and extruded at the crest of an
ocean ridge (Fig^4'!6). On further cooling, the tem
perature passes through the Curie point below which
ferromagnetic behaviour becomes possible (Section
3.6.2). The solidified magma then acquires a
68 CHAPTER 4

magnetization with the same orientation as the am

bient geomagnetic field. The process of lithosphere
formation is continuous, and proceeds symmetri
cally as previously formed lithosphere to either side
of the ridge moves aside. But, if the geomagnetic
field reverses polarity as the new lithosphere forms,
the crust on either side ofthe ridge would consist ofa
series of blocks running parallel to the crest which
possess remanent magnetizations which are either
normal or reversed with respect to the geomagnetic
field. The crust can thus be viewed as a twin-headed
tape recorder in which the reversal history of the
Earth's magnetic field is registered (Vine, 1966).
Oceanic basalts are characterized by a high Kon-
igsberger ratio, that is, the intensity of remanent
magnetization is significantly larger than the in
duced. Since the shape of a magnetic anomaly is
governed by the orientation of its total magnetiza
tion vector, that is, the resultant ofthe remanent and
induced components, the shapes of magnetic linea- 50 50 km

tions are effectively controlled by the primary re

manent direction. Consequently, blocks of normally Fig.4.7 Variation ofthe magnetic anomaly pattern with
magnetized crust formed at high northern latitudes geomagnetic latitude. All profiles are north-south. Angles
possess a magnetization vector which dips steeply to refer to magnetic inclination. No vertical exaggeration.
the north, and the vector of reversely magnetized
material is inclined steeply upwards towards the The orientation of the ridge also affects anomaly
south. The magnetic profile observed over this por shape and amplitude, because only that component
tion ofcrust will be characterized by positive anoma ofthe magnetization vector lying in the vertical plane
lies over normally magnetized blocks and negative through the magnetic profile affects the magnetic
anomalies over reversely magnetized blocks. A simi anomaly. This component is at a maximum when the
lar situation pertains in high southern latitudes. ridge is east-west and the profile north-south,and at
Crust magnetized at low latitudes also generates a minimum for ridges oriented north-south. The
positive and negative anomalies in this way, but variation in amplitude and shape of the magnetic
because of the relatively shallow inclination of the anomalies with orientation for a ridge of fixed lati-
magnetization vector the anomaly over any particu
lar block is markedly dipolar, with both positive and
negative components.This obscures the symmetry of
the anomaly about the ridge crest, as individual
blocks are no longer associated with a single positive
or negative anomaly. However,at the magnetic equa
tor, where the field is horizontal, negative anomalies
coincide with normally magnetized blocks and posi
N30°E
tive anomalies with reversely magnetized blocks,
precisely the reverse situation to that at high lati N60°E
tudes. In addition, the amplitude of the anomaly
decreases from the poles to the equator as the geo
magnetic field strength, and hence the magnitude of 50 50 km
the remanence,decreases in this direction. Figure 4.7
illustrates how the shape and amplitude of the mag Fig.4.8 Variation ofthe magnetic anomaly pattern with
netic anomalies over an ocean ridge striking east- the direction ofthe profile at a fixed latitude. Magnetic
west vary with latitude. inclination is 45' in all cases. No vertical exaggeration.

tude is shown in Fig. 4.8. In general, the amplitude
of magnetic anomalies decreases as the latitude de
creases and as the strike of the ridge progresses from
east-west to north-south. The symmetry of the
anomalies is most apparent for ridges at high mag
netic latitudes (e.g. greater than 64®, which is equi
valent to geographic latitudes greater than 45®),
north-south trending ridges at all latitudes and east-
west trending ridges at the magnetic equator.
4.6 Magnetostratigraphy
Once the geomagnetic reversal timescale has been
calibrated, oceanic magnetic anomalies may be used
to date oceanic lithosphere. The method has been
progressively refined so that it is now possible to
deduce ages back to mid-Jurassic times with an accu
racy of a few million years.
The Vine-Matthews hypothesis explains the
sequence of magnetic anomalies away from ocean
ridges in terms of normal and reversed magnetiza
tions of the oceanic crust acquired during polarity
reversals ofthe geomagnetic field. Verification ofthe
hypothesis was provided by the consistency of the
implied reversal sequence with that observed inde
pendently on land. Cox et a>L (1967) had measured
the remanent magnetization of lavas from a series of
land sites. The lavas were dated by a newly refined
potassium-argon method, which allowed the con
struction of a reversal timescale back to 4.5 Ma. The
timescale could not be extended to earlier ages, as the
errors involved in K-Ar dating become too large.
Similarly, polarity events of less than 50 000 years
duration could not be resolved. The timescale, to
5 Ma before present as later refined by Cande & Kent
(1992), is given in Fig.4.9. In magnetostratigraphic
terminology, polarity chrons are defined with dura
tions of 10^-10^ years. Chrons may be dominantly
of reversed or normal polarity, or contain mixed
events.
Further verification of the geomagnetic reversal
timescale was provided by palaeomagnetic investiga
tions ofdeep sea cores(Opdyke etal.^ 1966). Unlike
lava flows, these provide a continuous record, and
permit accurate stratigraphic dating from their
microfauna. This method is most conveniently ap
plied to cores obtained in high magnetic latitudes
where the geomagnetic inclination is high, because
the cores are taken vertically and are not oriented
azimuthally. Excellent correlation was found be
tween these results and those from the lava se
SEA FLOOR SPREADING 69
quences,and confirmed that at least 11 geomagnetic
field reversals had occurred over the last 3.5 Ma.
Subsequent work on other cores extended the rever
sal history back to 20 Ma(Opdyke et«/., 1974).
Pitman & Heirtzler(1966) and Vine (1966) used
the radiometrically dated reversal timescale to com
pute the magnetic profiles that would be expected
close to the crestal regions of mid-ocean ridges. By
varying the spreading rate it was possible to obtain
very close simulations of all observed anomaly se
quences (Fig. 4.10), and consequently to determine
the spreading rates. A modern compilation is shown
in Table 4.1. Extensions of this work show that the
same sequence of magnetic anomalies, representing
nearly 100 Ma of spreading and reversals of the
Earth's magnetic field, can be observed over many
ridge flanks (e.g. Fig. 4.11). Later work has shown
that similar linear magnetic anomalies are developed
over oceanic crust dating back to the Jurassic. Al
though there is no oceanic crust older than this,
palaeomagnetic investigations on land have shown
that geomagnetic reversals have occurred at least
back to 2.1 Ga.
That spreading rates have varied with time is appar
ent from an examination of magnetic profiles from
different oceans. Examples are given in Fig.4.12 in
which the spreading rate in the south Atlantic is
assumed to be constant and the distances to various
magnetic anomalies from ridge crests in other oceans
are plotted against the distance to the same anomaly
in the South Atlantic. Inflection points in the curves
for the other oceans indicate when the spreading
rates changed there if the implicit assumption that
the spreading rate has remained constant in the
South Atlantic is correct.
However, spreading rates may have changed with
time in all oceans; hence, independent dating is
required at points of different age in order to trans
form the distance from the ridge axis into a temporal
scale and, therefore, the anomalies into a geo
magnetic reversal timescale. This calibration was first
performed using dates obtained from the oldest
sediments overlying layer 2 penetrated in Deep Sea
Drilling Project (DSD?) holes, but a more modern
technique, made possible by superconducting mag
netometers, is the detection of magnetic reversals in
thick limestone sequences on land which can be
dated from their fossil content.
The first long-term geomagnetic timescale, which
covered the last 80 Ma, was constructed by Heirtzler
et al. (1968). They made the assumption that
CHAPTER 4

Epoch Numerical Polarity Named Chrons

Chrons and Subchrons

BRUNHES

Jaramillo

MATUYAMA

Oiduvai

Reunion

GAUSS

Kaena

2An2
Mammoth

GILBERT

Cochiti

Nunivak

Sidufjall

Thvera
Fig. 4.9 Geomagnetic polarit)'
timescale for the Plio-Pleistocene
(after Cande & Kent, 1992).
Numerical chrons are based on the
numbered sequence of marine
magnetic anomalies.
 SEA FLOOR SPREADING 71

Reykjanes Ridge South Atlantic NW Indian Ocean

60° N
38°S
Profile'C 5° N
15
E

0-
500 nT
500 nT

Model 15mma~^
Model 15 mm a ^

km 50 0 50
—I 1 i_^S.L km 50 — CT
-3.3 3 L
"-5 km -3.3
500 nT
^3 km

East Pacific Rise

Profile reversed
Juan De Fuca Ridge
Models lOmma Profile reversed 46° N

km 50 '0 50 50 '0 50 p |_
500 nT 500 nT
^4 km

Model 29 mm a Model 44 mm a

km 100 100
S.L
3.3

Fig.4.10 Magnetic anomaly profiles and models of

several spreading centres in terms of the reversal timescale
(redrawn from Vine, 1966,Science 154,1405-15, with
permission from the AAAS. Copyright © 1966 by the
AAAS).

Latitude 28.5 °S Latitude 28.2 °S

Longitude 27.2 °W F.Z Longitude 10.2 °W

I
Observed 250 18 13 12
profile nT I
0'
100 km
h
Model 500
profile nT
100 km
1

1500 1000 500

80 70 60 50 40 30 20 10 0 Age (Ma)

Fig.4.11 Magnetic anomaly profile and model over the

southern Mid-Atlantic Ridge (redrawn from Heirtzler et
»/., 1968).
72 CHAPTER 4

Table 4.1 Spreading rates at mid-ocean ridges (^spreading ity were numbered from 1 to 32 with increasing
rate' is defined as the accretion rate per ridge flank). time.
The Heirtzler timescale was checked using in
Observed Predicted
formation collected during the DSDP (Section 4.7),
rate rate

Ridge Latitude (mm a"^) (mm a"^)

and shown to be essentially correct back to 80 Ma
with a maximum error of about ±5 Ma. It appears
Juan de Fuca 46.0'N 29 t that spreading in the major ocean basins has been
Gulf of California 23.4'N 25 24.7 continuous as all polarity events are present, al
Cocos—
though the rate ofspreading has varied.
Pacific 17.2"N 37 39.4
The discovery ofLarson & Pitman(1972)ofolder
3.rN 67 65.4
Galapagos 2.3'N 22 22.0
magnetic anomalies in three regions of the western
n
3.3'N 34 34.6
Pacific allowed the Heirtzler geomagnetic timescale
Nazca— to be extended back to 160 Ma. Lineations ofsimilar
Pacific 12.6''S 75 74.2 pattern were also found in the Atlantic. The time-
Chile Rise 43.4'S 31 30.2 scale was extended by assuming a constant spreading
Pacific- rate in the Pacific, calibrated by DSDP sites in the
Antarctic 35.6'S 50 49.5 Pacific and Atlantic. The longer periods of reversed
51.0°S 44 44.6 polarity in this sequence are numbered MO to M29
65.3'S 26 29.0 (M representing Mesozoic).
North Atlantic 86.5°N 6 5.7
The version of the reversal timescale published by
60.2''N 9.5 9.2
//
42.7°N 11.5 11.9
Harland et al. (1982) is shown in Fig. 4.13. More
Central Atlantic 35.0'N
recent modifications for the Cenozoic have been
10.5 11.0
23.0'N 12.5 12.6
published by Berggren et al. (1985) and for the
Cayman 18.0'N 7.5 5.9 Cretaceous and Jurassic by Kent & Gradstein (1985).
South Atlantic 38.5''S 18 17.6 The timescale is accurate to within a few million
Antarctic- years, and will undoubtedly increase in accuracy in
South America 55.3°S 10 9.3 the future as new age data become available.
Africa- The use of the geomagnetic timescale to date the
Antarctic 44.2'S 8 7.4 oceanic lithosphere is based on the identification of
Northwest Indian Ocean 4.2''N 14 14.6
characteristic patterns of magnetic anomaly linea
"
12.0'S 18.5 17.9
tions and their relation to the dated reversal chrono
24.5°S 25 24.5
Southeast Indian Ocean 25.8'S 28 28.8
logy. Particularly conspicuous markers which are
"
50.0"'S 38 37.3
widely used are anomalies 5, 12-13, 21-26 and
"
62.4''S 34.5 33.7 31-32. Also of interest is the prolonged period of
Gulfof Aden 12.rN 8 8.6 normal polarity in the Cretaceous. This period corre
//
U.b'N 12 12.1 sponds to magnetic quiet zones within the oceans
Red Sea 18.0"N 10 8.2 where there are no linear magnetic anomalies. In
many instances, however,the recognition of particu
Based on data from DeMets et al.(1990)and Vine
(1966).
lar anomalies is not possible, and the usual approach
t Not available because Farallon plate is omitted from the is to construct the anomaly pattern expected for
model. relevant parts ofthe timescale and to compare it with
the observed sequence.
Once the reversal chronology has been established,
spreading in the South Atlantic had remained con lineations of known age can be identified on mag
stant at the same rate as had been deduced for the last netic maps and transformed into isochrons so that
4 Ma. A model of normal and reversely magnetized the sea floor can be subdivided into age provinces.
blocks was constructed which simulated the observed Lineations of the same age on either side of a mid-
anomaly pattern, and the distance axis transformed ocean ridge can be fitted together by employing
into a geomagnetic timescale of reversals. Prominent techniques similar to those used for continental
anomalies corresponding to periods ofnormal polar margins (Section 3.2.2). In this way reconstructions
 SEA FLOOR SPREADING 73

Anomaly number
-70 14- S Pacific Ocean

Distance from
N. Pacific
ridge X S. Indian
Ocean
10-2(km) Ocean
^ -40 in S. Atlantic

6 8 10 12 14 16 20 22 24 26 28 30

Distance from ridge x 10"^(km)

Fig. 4.12 iJCop) Relationship between the distance to a
given anomaly in the South Atlantic and the distance to the
same anomaly in the South Indian, North Pacific and South
Pacific Oceans. Numbers on the right refer to magnetic
anomaly numbers (redrawn from Heirtzler a/., 1968).
Fig.4.13 {Bottom) Magnetic reversal timescale from mid-
Jurassic times to present(redrawn from Harland et«/.,
1982,A Geolo£fical Time Scale^ with permission from
Cambridge University Press).

OMa- r 55 Ma ■ -113Ma
25
Q Pleistocene -115
26
Aptian
0
Pliocene O)
0 Palaeocene -60 MO
1
0
27
-120
a. 28 Barrcmian
.29 Ml

30
M3 ■ 125
31
0
D)
Miocene Maastrichtian
■70 0 Hauterivian M5 I
O
0 32 I <j M7
> 130
M9
M10
20 33 '75 M10N
Valanginian lynif ■135
w Campanian
•80
MI2I
-25
M13|
-140
o
Berriasian M14j
M15
30 § Santonian •85
Oligocene ■M16 . 145
Coniacian M17
Tithonian
Turonian -90 M18
0 -35
c
0 M19 • 150
O)
13
O £ M20
0
Cenomanian
iS
40
•95 Kimmeridgian [^21 I
0 0
Q.
CO 155
-i i—— |\/|22

-100 M23
45
Oxfordian [^241 160
Eocene
M25 I
-105"
M25A|
50 0

(j
Albian
o M27I
165
(0 n n '
Callovian M29

■110 -7

■o 170
■o
■113
74 CHAPTER 4
Fig.4.14 Relative positions of Europe and Africa with
respect to North America illustrating their separation
during the Mesozoic and Cenozoic. Ages of
reconstruction shown in millions of years (redrawn from
Pitman & Talwani, 1972, with permission from the
Geological Society of America).
of plate configurations can be made for different
times, and the whole history of ocean basins deter
mined (Scotese et al., 1988). Figure 4.14 shows this
method applied to the Mesozoic and Cenozoic his
tory of the North Atlantic, while a much more
complex example ofthe evolution ofthe Greenland-
Iceland-Scotland region is described by Nunns
(1983).
4.7 Deep sea drilling results
The Vine-Matthews hypothesis and geomagnetic
reversal timescale were verified independently during
the third leg ofthe DSDP (later to become ODP,the
Ocean Drilling Program). A series of holes was
drilled in the South Atlantic along a traverse at right
angles to the Mid-Atlantic Ridge (Fig. 4.15a) The
age of the oceanic crust would ideally have been
determined by radiometric dating of the layer 2
basalts which were penetrated in each hole. However
the basalts were too weathered for this to be possible,
and so their ages were determined, albeit slightly
underestimated, by palaeontologic dating of the
basal sediments of layer 1. In Fig. 4.1Sb oldest sedi
ment age is plotted against distance from the ridge
axis, and it is readily apparent that there is a remark
able linear relationship, with crustal age increasing
with distance from the ridge. The predicted ages
agree well with the reversal timescale, with a half
spreading rate in this region of 20 mm a"^.
These results provide independent confirmation of
the predictions of the Vine-Matthews hypothesis,
and indicate the reasonable accuracy of the reversal
timescale to ages of at least 80 Ma.

■E 40-
0 400 800 1200 1600
Distance from ridge crest(km)
Fig.4.15 (a) Location map of drilling sites on Leg 3 of
the DSDP in the South Atlantic,(b) Relationship between
greatest sediment age and distance from the Mid-Atlantic
Ridge crest (after Maxwell el al., 1970, Science 168,
1047-59, with permission from the AAAS. Copyright ©
1970 by the AAAS).
4.8 Introduction to
transform faults
The theory of sea floor spreading proposes that
oceanic lithosphere is created at mid-ocean ridges
and is balanced by the complementary destruction of
oceanic lithosphere at subduction zones. While this
theory neatly explains the geometry of lithospheric
behaviour in two dimensions, a problem arises when
SEA FLOOR SPREADING 75
the third dimension is considered, namely where do
ridges and trenches terminate horizontally? This
problem was addressed by Wilson (1965), who pro
posed that the ends ofthese features were linked by a
new class of faults which he called transform faults.
At these faults there is neither creation nor destruc
tion of lithosphere, but rather the motion is strike-
slip, with adjacent lithosphere in tangential motion.
Transform faults will be considered in greater
detail in Chapter 7,but at this stage it is apparent that
there are three types of boundary between blocks of
lithosphere, and that these boundaries divide the
Earth's outer shell into a series ofplates. This is the
first step in the construction ofthe modern theory of
plate tectonics.^ which is covered in the following
chapter.
Further reading
Courtillot, V. & Besse, J. (1987) Magnetic field reversals,
polar wander and core-mantle coupling. Science 237,
1140-7.
Cox,A.& Hart, RB.(1986)Plate Tectonics. How It Works.
Blackwell Scientific Publications, Oxford.
Cox,A.,Dalrymple,G.B.& Doell,RR (1967)Reversals of
the Earth's magnetic field. Sci. Am. 216,44-54.
Fuller, M.(1987) A mechanism for reversals? Nature 326,
132-3.
Gubbins, D.(1987) Mechanism for geomagnetic polarity
reversals. Nature 326,167-9.
Jacobs, J.A. (1975) The Earth's Core. Academic Press,
London.
Kearey,P.& Brooks, M.(1991) Introduction to Geophys
ical Exploration (2nd edn). Blackwell Scientific Publica
tions, Oxford.
Pitman, W.C. Ill & Heirtzler, J.S.(1966) Magnetic anom
alies over the Pacific-Antarctic ridge. Science 154,1164-
71.
Stothers, RB.(1986) Periodicity of the Earth's magnetic
reversals. Nature 322,444-6.
Vine, F.J. (1966) Spreading of the ocean floor: new evi
dence. Science 154,1405-15.
Vine, F.J. 8c Matthews, D.H.(1963) Magnetic anomalies
over oceanic ridges. Nature 199,947-9.
5 / Theframework of plate tectonics
5.1 Plates and plate margins
The combination of the concept of transform faults
with the hypothesis of sea floor spreading led to the
construction of the theory of plate tectonics. In this
theory the lithosphere is divided into an interlocking
network of blocks termed plates. The boundaries of
plates can take three forms (Isacks et 1968).
1 Ocean rid£fes (accretive or constructive plate mar
gins) mark boundaries where plates are diverging.
Magma and depleted mantle upwell between the
separating plates, giving rise to new pceanic litho
sphere. The divergent motion of the plates is fre
quently perpendicular to the strike of the boundary,
although this is not always the case and is not a
geometrical necessity. In the Pacific it appears to be
an intrinsic characteristic of spreading whenever a
steady direction has been established for some time
(Menard & Atwater, 1968).
2 Trenches (destructive plate margins) mark bound
aries where two plates are converging by the mechan
ism of the oceanic lithosphere of one of the plates
being thrust under the other, eventually to become
resorbed into the sub-lithospheric mantle. Since the
Earth is not expanding significantly (Section 11.3),
the rate of lithospheric destruction at trenches
must be virtually the same as the rate of creation at
ocean ridges. Also included in this category are
Himalayan-type orogens caused by the collison of
two continental plates (Section 9.3), where contin
ued compressional deformation may be occurring.
The direction of motion of the underthrusting plate
need not be at right angles to the trench, that is,
oblique subduction can occur.
3 Transform faults (conservative plate margins) are
marked by tangential motions in which adjacent
plates in relative motion undergo neither destruction
nor construction. The relative motion is usually
parallel to the fault. There are, however, transform
faults which possess a sinuous trace, and on the
bends ofthese faults relatively small regions ofexten-
76
sion and compression are created (Section 7.4.2).
For the time being such structural elements are
ignored.
Within the basic theory of plate tectonics plates are
considered to be internally rigid, and to act as ex
tremely efficient stress guides. A stress applied to one
margin of a plate is transmitted to its opposite
margin with no deformation of the plate interior.
Deformation,then,only takes place at plate margins.
This behaviour is rather surprising when it is appreci
ated that plates are only about 80-150 km thick but
may be many thousands of kilometres in width.
When plate behaviour is examined in more detail,
however, it is recognized that there are some loca
tions where intraplate deformation does occur(Gor
don & Stein, 1992). Mountain belts (Section 9.3)
are characterized by extensive thrust faulting and
movements along large transcurrent fault zones(Sec
tion 9.3.6). Even extensional deformation may be
found in such belts. Within oceanic areas there are
regions of crustal extension and accretion in the
back-arc basins that are located on the landward sides
of many destructive plate margins (Section 8.9).
Plates are mechanically decoupled from each
other, although plate margins are in intimate con
tact. A block diagram illustrating schematically the
different types of plate boundaries is presented in
Fig. 5.1.
5.2 Distribution of earthquakes
Plate tectonic theory predicts that the majority ofthe
Earth's tectonic activity takes place at the margins of
plates. It follows, then, that the location of earth
quake epicentres can be used to define plate bound
aries. Figure 5.2 shows the global distribution of the
epicentres of large magnitude earthquakes for the
period 1961-67(Barazangi & Dorman,1969). For a
more detailed discussion of earthquake distribution
see Young et at, (1988). Although in terms of most
geological processes this represents only a very short

Lithosphere
Fig.5.1 Block diagram summarizing the principal
features of plate tectonics. Arrows on lithosphere
represent relative motions. Arrows in asthenosphere may
represent complementary flow in the mantle (redrawn
from Isacks et 196S^ Journal of Geophysical Research
73, with permission from the American Geophysical
Union).
period of observation, the relatively rapid motions
experienced by plates generate very large numbers of
earthquakes over a short interval oftime.
Earthquakes are classified according to their focal
depths: 0-70 km shallow focus, 70-300 km inter
mediate focus, greater than 300 km deep focus.
An important belt of shallow focus earthquakes
follows the crest ofthe ocean ridge system (Fig. 5.2),
0 20 40 60 100 120 140 160
THE FRAMEWORK OF PLATE TECTONICS 77
Lithosphere
Asthenosphere
Mespsphere
where focal mechanism solutions indicate tensional
events associated with plate accretion and strike-slip
events where the ridges are offset by transform faults.
On land, shallow tensional events are also associated
with the Basin and Range Province of the western
USA,the East African rift system and the Lake Baikal
graben system. The first two examples appear to
originate where an ocean ridge penetrates beneath a
continent, and its effects may extend beneath the
continent beyond the rift system (Wilson, 1988).
Fig.5.2 World-wide distribution of epicentres of large
magnitude earthquakes(mb>4)for the period 1961-67
(after Barazangi & Dorman, 1969, with permission from
the Seismological Society of America).
180 -160 -140 -120 -100 -80 -60 -40 -20 0
78 CHAPTER 5

The majority of intermediate and deep events are
associated with destructive plate margins. The north
ern, eastern and western Pacific Ocean is ringed by a
belt of earthquakes which lie on planes occasionally
offset by transform faults dipping at an angle of
about 45" beneath the neighbouring plates. These
planes of earthquake foci, known as Benioff (or
Benioff-Wadati) zones, are typically associated with
volcanic activity at the surface. The deepest events
recorded lie at a depth of about 680 km. Collisional
mountain belts such as the Alpine-Himalayan chain
are similarly characterized by intermediate and deep
focus earthquakes although,since there is no longer a
Benioff zone present in such regions, the seismic
activity occurs within a relatively broad belt. Careful
examination of epicentre locations has revealed,
however, that some of the shallow events lie on
arcuate transcurrent fault zones(Section 9.3.6) asso
ciated with the orogenic deformation.
The intraplate areas are relatively aseismic on this
timescale, although occasionally large magnitude
earthquakes do occur. Although insignificant in their
release of seismic energy, intraplate earthquakes are
important as they can indicate the nature and direc
tion ofstress within plates (Section 11.7).
determined by constructing great circles at right
angles to the trends to transform faults affecting their
common margin and noting their point of intersec
tion. The most convenient type of plate margin to
which to apply this technique is the accretive type
(Fig. 5.3), as ocean ridges are frequently oflfset later
ally by transform faults (Section 7.2). Because of
inaccuracies involved in mapping oceanic fracture
zones, the great circles rarely intersect at a single
point. Consequently, statistical methods are applied
which are able to predict a circle within which it is
most probable that the relative rotation pole lies.
A second method is based on the variation of
spreading rate with angular distance from the pole of
rotation. Spreading rates are determined from mag
netic lineations (Section 4.6) by identifying anoma
lies ofthe same age (usually number 3 or less so that
the movement represents an instantaneous rotation)
on either side of an ocean ridge and measuring the
distance between them. The velocity of spreading is
at a maximum at the equator corresponding to the
Euler pole and thence decreases according to the

Euler pole
5.3 Relative plate motions
The motion of plates over the Earth's surface can be
described by making use ofEuler's theorem (Section
3.2.1), which says that the relative motion between
two plates is uniquely defined by an angular separa
tion about a pole of relative motion known as an
Euler pole. The pole and its antipole are the two
unique points on the surface of the Earth which do
not move relative to either of the two plates. An
important aspect of relative plate motion is that the
pole of any two plates tends to remain fixed relative
to them for long periods oftime. Plate velocities are
similarly constant for periods ofseveral million years
(Wilson, 1993).
There are three methods by which the pole of
relative motion for two plates can be determined.
The first, and most accurate, is based on the fact that
for true tangential motion to occur during the rela
tive movement of two plates, the transform faults
along their common boundary must follow the
traces of small circles centred upon the pole of Fig.5.3 Determination ofthe Euler pole for a spreading
relative motion(McKenzie & Parker, 1967;Morgan, ridge from its offsetting transform faults which describe
1968). The pole ofrotation oftwo plates can thus be small circles with respect to the pole.

Velocity as a percentage of Vmax
20 40 60 80 100%
VmaxCOSij
Latitude 0 with
respect to pole of 0 estimates of relative plate velocities derived from
rotation
-90
Fig.5.4 Variation ofspreading rate with latitudinal
distance from the Euier pole of rotation.
cosine of the Euler pole's latitude (Fig. 5.4). The
determination ofspreading rate at at least two points
along the ridge then allows the pole of relative
rotation to be found.
The final, and least reliable, method of determin
ing the directions of relative motion between two
plates makes use of focal mechanism solutions of
earthquakes (Section 2.1.6) on their common mar
gins. Ifthe inclination and direction ofslip along the
fault plane are known,then the horizontal compon
ent of motion can be found. The data are less
accurate than the other two methods described above
because, except in very well determined cases, the
nodal plates could be drawn in a range of possible
orientations and the detailed geometry of fault
systems is often more complex than implied here
(Section 7.4.2).
Divergent plate boundaries can be studied using
spreading rates and transform faults. Convergent
boundaries, however, present more of a problem,
and it is often necessary to use indirect means to
determine relative velocities. This is possible by
making use ofinformation from adjoining plates and
treating the instantaneous rotations between pairs as
vectors (Morgan, 1968). Thus, if the relative move
ments between plates A and B and between plates B
and C are known, the relative movement between
THE FRAMEWORK OF PLATE TECTONICS 79
plates A and C can be found by vector algebra.
This approach can be extended so that relative
motions can be determined for any number of inter
locking plates. Indeed,the method can be applied to
the complete mosaic of plates which make up the
Earth's surface, provided that there are sufficient
divergent plate margins to be able to compute rela
tive velocities at convergent margins.
The first study of this type was undertaken by Le
Pichon (1968). He made use of globally distributed
transform faults and spreading rates, but not of
information obtained from focal mechanism solu
tions. Le Pichon used a subdivision of the Earth's
surface based on only six large plates. In spite of this
simplification his model provided estimates of
spreading rates which agree well with those observed.
Subsequently more detailed analyses of global
plate motions were derived by Chase (1978) and
Minster & Jordan (1978). These studies subdivided
the Earth's surface into 12 plates, although Chase
recognized a Philippine plate but not the Caribbean,
while Minster & Jordan took the opposing view on
the existence of these relatively small plates. Both
used large datasets of relative motion vectors derived
from transform faults, spreading rates and focal
mechanism solutions. In fact, so many data were
available that the problem became overdetermined,
and in inverting the data set to provide the global
distribution of plate motions, both studies used a
technique whereby the sum ofthe squares ofresidual
motions was minimized. Errors in spreading rates
were generally ofthe order of1 mm a"^,in transform
fault orientation between 1® and 12®, and in earth
quake slip vector direction normally less than 15®.
Although the two studies used different data sets and
plate configurations, the two models are in generally
good agreement. Chase's model is illustrated in
Fig. 5.5, which also shows the relative motions ofthe
plates with respect to the hotspot reference frame
(Section 5.4). The most recent global analysis of
current plate motions is provided by DeMets et al.
(1990), which employs a dataset three times larger
than previous models. This model suggests plate
velocities some 15% slower than before, and recog
nizes separate Indian and Australian plates.
These global models yield several interesting
points. The assumption behind the models, i.e. the
internal rigidity of plates, breaks down in areas such
as the back-arc basin of the Philippine Sea and the
 CHAPTER 5

■-23 J 60°
\\ N y ^
EURASIAN rx- . >72 i\NORTH AMERICAN
y Nt
y •r^25 I
\ Yellowstone/ .•■-—i
^57
, rCARIBBEAN'S.ae
AFRICAN^

H \ ^ ✓''V
^78 PACIFIC "* l^^^cocbs) '"■••. —
^ •' T" 'I ■■■••••■•. - 0°
_^X5r'^^-rr66 south)
y2 "~
"/Reunion 80 -t ^;l n
MacDonaid /y NAZCA
94-— y AMERICAN \-
INDIAN f
r~'2 -^0
'•^es ^
X--J85 ^
•.... 106 ,y
J
X., ^^^40
. _ono
30
; ;■ J, gg I ^ Martin ;
/ Vaz
■•"i Kerguelen
.../.Tl
' ■•-.166 /
/

ANTARCTIC: ANTARCTIC^

• Ocean ridges Subductionzonesandtransformfaults

Fig. 5.5 Directions and rates (in mm a") of relative
motion of the Earth's major plates shown at selected
points on their boundaries. Circled numbers represent
absolute motions (in mm a"') of plates relative to the
hotspot reference frame (redrawn from Bott, 1982a, with
permission from Edward Arnold, based on data in Chase,
1978).
Alpine-Himalayan mountain chain. In these areas
relative motions also occur by internal deformation,
extensional and compressional, respectively, of the
plates. In the Himalaya, data are consistent with a
convergence between India and Asia in a north-
south direction at a rate of approximately 50 mm a~^.
The convergence between Africa and Europe at the
Mediterranean-Alpine boundary is in the same sense
but somewhat slower at 10 mm a^^.
The greatest rates of subduction are found at the
margins of the Pacific Ocean where the values plotted
fall in the range 65-106 mm a~^. The rate of accre
tion of oceanic lithosphere to each plate at the East
Pacific Rise, however, is typically lower than this at
33-93 mm a"^, so that the ocean is shrinking in this
area. In 300 Ma the Pacific would close completely,
although this is unlikely to happen because events
elsewhere might cause a reversal of this situation
(Section 10.4). Subduction at the Aleutian Trench
becomes more and more oblique westwards. The San
Andreas Fault represents a conservative boundary at
which there is relative dextral movement between
the North American and Pacific plates. The most
recent analysis of global plate motions predicts that
the rate of relative motion is 48 mm a~^ (DeMets et
al., 1990). This is significantly higher than the
35 mm a"^ suggested from geological investigations.
It strongly suggests that a significant fraction of the
movement between these two plates is accommod
ated elsewhere, notably within the Basin and Range
Province and a shear zone in the Mojave Desert.
Plates containing parts of the Atlantic and Indian
oceans appear to be growing, in contrast to the
shrinking Pacific, Nazca and Cocos plates. A corol
lary of this is that the Mid-Atlantic Ridge and Oarls-
berg Ridge of the western Indian Ocean must be
moving apart. This has important implications for
the nature of the driving mechanism of plate tec
tonics discussed in Chapter II. Not all ocean ndges
spread in a direction perpendicular to the strike of
their magnetic lineations. It may be significant that
the major obliquities of this type are found in the
more slowly spreading seas, in particular the North
Atlantic, Gulf of Aden, Red Sea and southwestern
Indian Ocean.

The analyses of Chase and Minster & Jordan
imply that there is a slow compressionai movement
between North and South America about a pole to
the east of the West Indies at approximately 22°N.
Some of this movement is probably taken up in the
compressionai ridge/trench regime of the southern
Caribbean Plate (Kellog & Bonini, 1982). To the
east of the Caribbean, however, there is no obvious
seismic manifestation ofthe plate boundary between
the Americas, and it has been suggested that defor
mation takes place over a relatively wide area so that
it is aseismic.
The analyses are consistent with extension across
the East African.rift system at the very slow rate of
about 6 mm a"^. The significance of this slow exten
sion is discussed further in Section 10.3.
5.4 Hotspotsand
absolute plate motions
The major part of the Earth's volcanic activity takes
place at plate margins. However, a significant frac
tion occurs within the interiors of plates. In oceans
the intraplate volcanic activity gives rise to linear
island and seamount chains such as the Hawaiian-
Emperor and Marquesas in the Pacific (Fig. 5.6).
I ^
0 o
o d.
o <? :
T'jLti/o*
Fig. 5.6 Linear island chains in the 20°
Pacific Ocean on a Mercator AustratW
projection. The chains young to the
THE FRAMEWORK OF PLATE TECTONICS 81
Moreover, many of these island chains appear to be
mutually parallel. Where the volcanic centres are
closely spaced, aseismic ridges are constructed, such
as the Ninety-East Ridge in the Indian Ocean and
the Greenland-Scotland Ridge in the North Atlantic
and the Rio Grande and Walvis ridges in the South
Atlantic. These island chains and ridges are associ
ated with broad crustal swells which currently occupy
about 10% ofthe surface ofthe Earth, making them a
major cause of epeirogeny (Crough, 1979).
The island chains are invariably younger than the
ocean crust on which they stand. The lower parts of
these volcanic edifices are believed to be formed
predominantly of tholeiitic basalt, while the upper
parts are alkali basalts (Karl et al., 1988) enriched in
Na and K and, compared to mid-ocean ridge basalts,
have higher concentrations of Fe, Ti, Ba,Zr and rare
earth elements (REE) (Bonatti et al., 1977). Their
composition is compatible with the mixing of juve
nile mantle material and depleted asthenosphere
(Schilling 1976)(Section 6.7). They are under
lain by a thickened crust but thinned lithosphere, and
represent a type of anomalous feature that will even
tually become welded to a continental margin as a
suspect terrain (Section 9.5).
An example of an oceanic island chain is the
North America
%■
•-V °
O'
:
82 CHAPTER 5

Jan Mayen

Iceland

Bowie ^
Ye bwstone
Azores
Cobb
Bermuda
Canary
Hawaii New
Sooerro England
Cape*
Galapagos
Verde I. Caroline
Marquesas Fernando
Samoa Ascension Comores
Pitcairn
San Felix
Society St Helena
Easter Reunion
Martin Vas Lord
Austral Juan •
Howe
Fernandez
Tristan
Marion
AustraliaA ^ yf
Discovery Kerguelen Tasman ^
Crozet
Louisviile

Shona Bouvet
Balleny

Fig.5.7 World-wide distribution of hotspots (after
Duncan & Richards, 1991).
Hawaiian-Emperor chain in the north-central Pacific
Ocean (Fig. 5.6). This chain is some 4000 km long
and shows a trend from active volcanoes at Hawaii in
the southeast to extinct, subsided guyots in the
northwest. Dating of the various parts of the chain
(Jackson etcd.^ 1972)confirms this trend, and shows
that the volcanic activity has progressed from north
west to southeast at a rate of about 99 mm a"^
(McDougall & Duncan, 1980), although Jackson
(1976) disputes the conclusion that voicanism has
moved regularly along the chain. The chain conse
quently cuts across the older Cretaceous structural
patterns of the sea floor. The change in direction of
the chain correlates with a major change in spreading
direction at about 43 Ma. The Hawaiian-Emperor
chain runs parallel to other chains on the Pacific
Plate, along which voicanism has progressed at a
similar rate.
A possible explanation ofthe origin ofisland chains
was proposed by Wilson(1963).It was suggested that
the islands formed as the lithosphere passed over a
hotspot. The hotspots are now believed to originate
from mantle plumes rising from the lower mantle
which thin the overlying lithosphere(Olson & Nam,
1986). The volcanic rocks are then derived from
pressure-release melting and differentiation within
the plume. Such plumes represent material of low
seismic velocity and can be detected by seismic tomo
graphy (Section 2.1.8; Nataf & VanDecar, 1993).
Although the mantle plume mechanism has been
widely adopted,some workers (e.g., Turcotte & Ox-
burgh, 1978; Pilger, 1982) have questioned the ne
cessity for mantle hotspots and suggest that magmas
simply flow to the surface from the asthenosphere
through fractures in the lithosphere resulting from
intraplate extensional stresses. This mechanism obvi
ates the problem of maintaining a mantle heat source
for long periods. It does not, however, explain why
fractures in the same plate should trend in the same
direction and develop at similar rates(Condie,1982).
A useful summary of knowledge about hotspots is
provided by Duncan & Richards(1991).

Morgan (1971, 1972a) proposed that mantle
plumes remain stationary with respect to each other
and the lower mantle,and are oflong duration. Ifso,
the hotspots represent a fixed frame of reference by
which absolute motions of plates can be determined.
The most recent estimate of the present number of
hotspots is about 40 (Stothers, 1993)(Fig. 5.7).
Morgan (1981) has used this hypothesis to de
scribe plate motions over the mantle for the past
200 Ma (Fig. 5.8). According to this analysis the
hotspots appear to be fixed with respect to each other
or to move at relative velocities of only a few milli
metres per year. Significantly, many of the tracks,
such as that of the Madeira hotspot between Green
land and Labrador, and of the St Helena hotspot
between the south coast of West Africa and the north
coast ofBrazil, appear to be the sites offuture rifting.
It is thus possible that hotspots weaken the litho-
sphere so that at future dates rifting will take place
along their trajectories.
Predicted hotspot traces in the Indian Ocean
(Duncan & Richards, 1991) are shown in Fig. 5.9,
superimposed on volcanic structures on the sea floor
and on land. The correlation between the two is
excellent. For example, the Reunion hotspot began,
and was responsible for the Deccan Traps flood
basalts of western India; India's northwards motion
was then recorded by a north-south ridge in the
western Indian Ocean and the Mascarene Plateau.
The hotspot is currently beneath a seamount 150 km
west ofthe volcanically active island of Reunion.
Continental plates also pass over hotspots al
though the effects are not as pronounced as on
oceanic plates. They give rise to domal uplifts with
associated alkaline igneous activity, and may be re
sponsible for areas of high heat flow and geothermal
activity. Crough et al, (1980) have proposed that
kimberlite intrusions are formed by hotspots, and
Fig. 5.10 shows the relationship of the Trindade
hotspot trajectory to the occurrence of kimberlites
and alluvial diamond deposits in Brazil.
The notion of fixed hotspots is important because
plate movements have destroyed all other fixed co
ordinate systems, with the exception of the geo
magnetic field. The proposition can be tested in
several ways. For a single plate, the pole of rotation
can be determined from a single hotspot and its rate
of rotation found using the measured rate of migra
tion of volcanicity along the chain (McDougall &
Duncan, 1980). The trajectories of other hotspots
THE FRAMEWORK OF PLATE TECTONICS 83
on the plate can then be computed and compared
with the actual tracks. This type of calculation indi
cates that there have been no significant relative
movements during the last 10 Ma, when instantan
eous motions are probably representative.
More rigorously, the consistency ofhotspot trajec
tories on several plates can be tested by reconstructing
the relative positions of the major plates at various
times in the past and locating one ofthe plates over its
hotspots. If the hotspot positions are indeed fixed,
they should then all underlie the chains for which they
are responsible. This test requires precise knowledge
of relative plate positions in the past, and has been
applied with increasing accuracy as models of plate
motion have become more refined. Thus,Burke etal.
(1973)and Molnar&Atwater(1973)concluded that
there has been at least 15 mm a~^ relative movement
between hotspots over the last 120 Ma,and Molnar&
Stock(1987) have determined interhotspot motions
of up to 20 mm a"L
The most recent analyses, summarized in Duncan
& Richards (1991), suggest that the maximum
possible relative motion between hotspots has been
only 2-5 mm a"^ over the last 120 Ma. This is con
siderably slower than typical plate velocities, so that
the concept of a hotspot reference frame is not
seriously degraded.
A different approach to absolute plate motions has
been applied by Kaula (1975). He defined an abso
lute reference frame by minimizing the lateral
motions of plate boundaries, that is, by assuming
that plate margins move as little as possible with
respect to the deep interior of the Earth. His model
agrees quite well with that based on hotspots and
provides the following result: oceanic plates move at
about 50mma~^; continental plates move at about
15 mm a" oceanic-oceanic plate boundaries move
at about 20 mm a" and continent-oceanic plate
boundaries move at about 15mma"^. Kaula con
cludes that the forces exerted by the lithosphere are
significant in controlling the locations of accretive
and destructive plate margins.
5.5 True polar wander
In Section 3.6 it was demonstrated that palaeo-
magnetic techniques can be used to construct appar
ent polar wandering paths which track the motions
of plates with respect to the magnetic north pole and
hence, using an axial geocentric dipole model, the
84 CHAPTER 5

Iceland

""^abrador
0 Eifel

Bermuda
Madeira

Meteor q Ahagga
Canary , Tibesti

0
Verde
•■A
Jebel Marra

o Fernando
Cameroon
Guyana h Ascension
^ St Helena ) Comoros
0

Trindade
Tristan ®
0 Vema

« Discovery
Bouvet

1 <$
i

Fig. 5.8 Reconstruction of plates around the Atlantic reference (redrawn from Morgan, 1983, with permission
from 180 Ma to present in a fixed hotspot frame of from Elsevier).
 THE FRAMEWORK OF PLATE TECTONICS 85

1 1

&
86 CHAPTERS

Indian Plate

Arabian
Plate

African
Plate

Seychelles 'Maldives 1 77^

\ / 63"

Comores
r Chagos /j'j \
\ 1^1-62 Ninety-East
^M^scarene 58~^ /
2 °
Reunion /' \

Madagascar
Broken Ridge
Ridge

Australian Plate
Kergueien

Kergueien
Plateau

Antarctic Plate

Fig.5.9 Hotspot tracks in the Indian Ocean. Solid circles
show present locations of hotspots. Numbers refer to
radiometrically dated locations in millions of years. Dark
stipple marks flood basalt provinces (redrawn from
Duncan & Richards, 1991).
spin axis ofthe Earth. In Section 5.4 it was suggested
that hotspots are nearly stationary in the mantle, and
so their trajectories provide a record of the motions
of plates with respect to the mantle. A combination
of these two methods can be used to test if there has
been any relative movement between the mantle and
the Earth's spin axis. This phenomenon is known as
true polar wander (TPW), and has been studied by
several workers, including Livermore et al. (1984),
Andrews(1985) and Courtillot & Besse (1987).
The method employed to investigate TPW is as
follows. Palaeomagnetic pole positions for the past
200 Ma are complied for a number of continents
which are separated by spreading oceans so that their
relative motions can be reconstructed from magnetic
lineation data (Section 4.6). The pole positions are
then corrected for the rotations relative to a single
continent (usually Africa) experienced as a result of
sea floor spreading since the time for which they
apply. In this way a composite or global apparent
polar wander path is obtained. This is then compared
with the track of an axis of the hotspot reference
 THE FRAMEWORK OF PLATE TECTONICS

1202v,122

/ ♦ *80

Trindade

o Kimberlites ♦ Major alluvial diamond deposits

Fig.5.10 Locations of kimberlites and major alluvial

diamond deposits in Brazil compared to the predicted
track of the Trindade hotspot (ages in Ma)(redrawn from
Crough et al., 1980).

frame as viewed from the fixed continent. The TPW

path is then determined by calculating the angular
rotation that shifts the global mean palaeomagnetic
pole of a certain age to the north pole and then
applying the same rotation to the hotspot pole of
the same age (Courtillot & Besse, 1987).
Three TPW paths constructed by different workers
are shown in Fig. 5.11. They bear some similarity in
shape and in the amplitude of the TPW which
increases with increasing age from around 10° up to
100 Ma to 20° at about 180 Ma. The differences
between them result primarily from the use of very
different criteria for the selection of the palaeo
magnetic pole positions used in the analysis.
Courtillot & Besse (1987) maintain that there
is a correlation between the rate of TPW and the
rate of geomagnetic reversals since Jurassic times

Fig. 5.11 True polar wander paths, that is the motion of

the 'geographic' pole, derived from the hotspot model,
with respect to the magnetic pole derived from
palaeomagnetic studies, the latter being equated with the
Earth's rotational axis. Ages in Ma. Polar equal area
projections (a) after Livermore et al. (1984), with
permission from the Royal Astronomical Society;(b) after
Andrews (1985), with permission from the American
Geophysical Union;(c) after Courtillot & Besse (1987),
Science 237, 1140-7, with permission from the AAAS.
Copyright © 1987 by the AAAS.
88 CHAPTERS
(Fig. 5.12e,f), and also, possibly, since Permo-
Carboniferous times. They noted that times ofslow
ing reversal rate are periods ofslow polar wander and
vice versa, and have suggested that the correlation
between TPW and reversal frequency changes may
indicate a coupling between the mantle and the core.
They also noted that the major change in direction of
the TPW path at about 30-40 Ma appears to corre
late with the collision of Africa, Arabia and India
with Eurasia, and so there may also be coupling
between the mantle and the lithosphere. This in
formation has led Courtillot & Besse to propose a
model in which mantle and core dynamics are linked
via thermal exchange across the core-mantle
boundary.
The relative motion between the rotation axis and
the mantle as illustrated by the TPW path may be
interpreted as a shifting of the whole or part of the
Earth in response to some form of internal mass
redistribution which causes a change in the direction
about which the moment ofinertia ofthe mantle is a
maximum (Andrews, 1985). For example,Anderson
(1982) relates TPW to the development of geoid
highs by the insulation from heat loss of mantle
niaterial beneath supercontinents, which cause a shift
of the rotation axis. It is possible that only the
lithosphere or the mantle or both lithosphere and
mantle together shift during polar wander. It is
highly unlikely that the lithosphere and mantle are
sufficiently decoupled to move independently,
and so it appears probable that shifting oflithosphere
and mantle as a single unit takes place during TPW.
Indeed,ifthere is coupling between core and mantle,
the whole Earth may be affected. Andrews's inter
pretation ofTPW is supported by astronomical data
which shows that during the 20th century the loca
tion ofthe Earth's rotational axis has moved at a rate
similar to that computed from palaeomagnetic data,
namely about l®Ma"^. This suggests that at least
part of the mass redistribution takes place in the
mantle, as the continents do not move this rapidly.
Sabadini & Yuen (1989) have shown that both
viscosity and chemical stratification in the mantle are
important in determining the rate of polar wander.
Another mechanism proposed for driving TPW is the
surface mass redistribution arising from major glaci-
ations and deglaciations (Sabadini et al.^ 1982).
However, mantle flow is required to explain TPW
during periods with no evidence of significant con
tinental glaciation, and, indeed, may be responsible
for the majority of TPW. It has also been suggested
that TPW is excited by the mass redistributions
associated with subduction zones (Spada et «/.,
1992), mountain building and erosion (Vermeersen
& Vlaar, 1993).
TPW clearly has important implications in the
study ofmantle inhomogeneities, mantle convection
and palaeoclimatology. The differences between
TPW paths derived by different authors (Fig. 5.11)
will probably diminish with the future availability of
accurate pole positions and age determinations.
5.6 Superplumes
The hotspots described in Section 5.4 are the surface
manifestation of plumes of hot material ascending
from the deep mantle. These are of moderate size and
can be considered to form part ofthe normal mantle
convecting system. It has been proposed, however,
that at least once during the history ofthe Earth there
has been an episode of much more intense volcanic
activity. The cause has been ascribed to a phenomena
termed superplumes^ large streams of overheated
material rising buoyantly from the D" layer at the
base of the mantle (Section 2.8.5) which derived its
heat from the core. These spread laterally at the base
of the lithosphere to affect an area ten times larger
than more normal plume activity.
Larson (1991a; 1991b; 1995) has proposed that
superplumes were responsible for the widespread
volcanic and intrusive igneous activity that affected
abnormally large amounts of ocean floor during the
mid-Cretaceous. One manifestation of this activity
was the creation of numerous chains of seamounts
and ocean plateaux (Section 9.5) in the western
Pacific (Fig. 5.6) at a rate some five times greater
during this period than at other times. Similarly
there were extrusions ofthick, areally extensive flood
basalts on the continents, such as the Parana Basalts
of Brazil.
Phenomena attributed to the mid-Cretaceous
superplume episode are illustrated in Fig. 5.12. At
120-125 Ma the rate of formation of oceanic crust
doubled over a period of 5 Ma,decreased within the
next 40-50 Ma and returned to previous levels about
80 Ma ago (Fig. 5.12c). The additional production
ofcrust required increased subduction rates, and it is
significant that the Andes ofSouth America and the
Sierra Nevada mountains ofthe western USA formed
at this time by the subduction process (Section 9.2).
 THE FRAMEWORK OF PLATE TECTONICS

Superplume
High-latitude surface '
temperature

Sea level

Oil resources

Oceanic crust
30 - production

Reversal rate

TRW velocity

Fig.5.12 Phenomena associated

with the mid-Cretaceous
superplume (after Larson, 1991a,
100
Age(Ma)
I
Tertiary Cretaceous Jurassic
and Courtillot & Besse, 1987).
90 CHAPTER 5
Coupled to the increased crust production, and
caused by the consequent general rise in the level of
the sea floor, was a world-wide increase in sea level to
an elevation some 250 m higher than at the present
day (Fig. 5.12a). The surface temperature of the
Earth also increased by about 10®C, as shown by
oxygen isotope measurements made on benthic fora-
minifera from the North Pacific. This effect was
probably caused by the release of large amounts of
carbon dioxide during the volcanic eruptions, which
created an enhanced 'greenhouse' effect. During the
superplume episode the rates of carbon and carbon
ate sequestration in organisms increased due to the
shallowing ofthe seas and the increased temperature,
which caused plankton to thrive. This is reflected in
the presence of extensive black shale deposits at this
time (Force, 1984) and in the estimated oil reserves
of this period (Tissot, 1979; Fig. 5.12c), which may
constitute about 50% of the world's supply. Also of
economic significance is the placement of a large
percentage of the world's diamond supply at this
time, probably as a result of the diamonds' having
been translated to the surface by the rising plumes.
During the plume episode the rate of geomagnetic
reversals (Section 4.4) was very low (Fig. 5.12e),
with the field remaining in normal polarity for some
40 Ma. This indicates that activity in the core, where
the geomagnetic field originates (Section 3.6.4), was
low, perhaps related to the transfer of considerable
quantities of heat to the mantle.
The correlation ofTPW with the rate of magnetic
reversals was considered in Section 5.5. Courtillot
& Besse (1987) correlated the period of very slow
TPW from 170 to 110 Ma with continental fragmen
tation and the decrease in reversal rate during the
Late Jurassic and Early Cretaceous (Fig. 5.12f).
Shortly after the start of the mid-Cretaceous mag
netic quiet zone,TPW velocity increased suddenly to
close to its present, relatively high rate. Courtillot &
Besse (1987) related the slow TPW and decreasing
reversal rate to reduced convective activity in the
mantle.
Acceptance of a mid-Cretaceous superplume epi
sode is not universal. For example, Anderson (1994)
suggests that the phenomena of this period were
caused by a general reorganization of plates on a
global scale associated with the break-up of Pangaea
and reorganization of the Pacific plate. The mantle
upwelling in the latter may then have been a passive
reaction to plates^being pulled apart by their attached
slabs. The episode would thus be viewed as a period
when mantle ascended passively as a result ofchang
ing plate motions.
Large mantle plumes are also believed to be re
sponsible for the formation of the volcanic margins
of certain continents. They are described in Section
10.3.6.
5.7 Direct measurement of
relative plate motions
It is now possible to monitor the relative motion
between plates by periodic resurveying of baselines
extending across plate boundaries (Stein, 1987).
Before about 1980 the only methods available for
this type ofinvestigation were the standard terrestrial
geodetic methods of baseline measurement using
optical techniques or laser ranging instruments such
as the geodolite (Thatcher, 1979). These methods
are certainly sufficiently precise to measure relative
plate motions of a few tens of millimetres a year.
However,in some regions the strain between certain
plates is not all dissipated across the boundary, and
may extend into the adjacent plates for distances of
up to about 100 km. In order to study these large
scale problems it is necessary to be able to measure
very large distances to very great accuracy. Terrestrial
methods are extremely time consuming on land, and
impossible to use across major oceans. During the
last decade, however, the measurement of very long
baselines using extraterrestrial methods has become
possible via the application ofspace technology.
Three independent methods of extraterrestrial
surveying are available. These are very long baseline
interferometry, satellite laser ranging and satellite
radio positioning.
The technique ofvery long baseline interferometry
(VLBI) makes use of the radio signals from extra-
galactic radio sources or quasers (Niell et«/., 1979;
Carter & Robertson, 1986; Clark et aL, 1987) as
shown in Fig. 5.13. The signal from a particular
quasar is recorded simultaneously by two or more
radio telescopes at the ends ofbaselines which may be
up to 10 000 km long. Because of their different
locations on the Earth's surface, the signals received
at the telescopes are delayed by different times, the
magnitude of the delays between two stations being
proportional to the distance between them and the
direction from which the signals are coming. Typi
cally, during a 24-hour experiment, 10-15 quasars
 THE FRAMEWORK OF PLATE TECTONICS 91

Extragalactic
random radio Difference in
signals time of arrival

antenna

Baseline
Portable
antenna

Calibrations Magnetic j
tape 1
I
Simultaneous
First order
vector baseline
Fig.5.13 Schematic outline of geodetic
solution
point
a VLBI system (redrawn from {General purpose Delays
positioning Cross-correlation
Niell etai^ 1979, with computer)
signal detection
permission from Elsevier). (minicomputer)

are each observed 5-15 times. This scheme provides
estimates of baseline length which are accurate to
about 20 mm (Lyzenga et 1986). The usefulness
of this system has been greatly enhanced by the
development of portable radiotelescopes which frees
the technique from the necessity ofusing fixed obser
vatory installations.
The technique of satellite laser ranging (SLR)
calculates the distance to an orbiting artificial satellite
(e.g. LAGEOS,the Laser Geodynamics Satellite) or a
reflector on the Moon by measuring the two-way
travel time of a pulse of laser light reflected from the
satellite (Cohen & Smith, 1985). The travel time is
subsequently converted to range using the speed of
light. If two laser systems at different sites simultan
eously track the same satellite, the relative location of
the sites can be computed by using a dynamic model
of satellite motion, and repeated measurements pro
vide an accuracy ofabout 80 mm.Periodic repetition
of the observations can then be used to observe
relative plate motions (Christodoulidis ,1985).
The accuracy of the measurements is determined by
the knowledge of the satellite orbit, which in turn
depends on the forces acting upon the satellite. Of
these the Earth's gravitational field is the most im
portant, and improved knowledge of this field will
lead to improved accuracy in SLR.
The technique of satellite radiopositioning makes
use ofradio interferometry from the Global Position
ing System (GPS) satellites (Dixon, 1991). It is a
three-dimensional method by which the relative
positions of instruments at the ends of baselines are
determined from the signals received at instruments
from several satellites. The simultaneous observation
of multiple satellites makes extremely accurate mea
surements possible with small portable receivers,
with relative positions determined to about 1 part
per million ofthe baseline length. This appears to be
the most efficient method of establishing geodetic
control on both local and regional surveys, and it is
probable that an accuracy of O.lp.p.m. will be
achieved. The full constellation of 21 GPS satellites
has now been implemented.
Stein (1987) has summarized the determinations
of relative plate motions determined by these meth
ods. Generally, plate velocities averaged over a few
years ofobservation agree remarkably well with those
averaged over millions of years (Gordon & Stein,
1992). The methods were first applied to the mea
surement of the rate of movement across the San
Andreas Fault in California. Smith et al, (1985),
using SLR, have reported that a 900 km baseline
which crosses the Fault at an angle of 25 ® has been
shortened at an average rate of 30 mm a"^. Lyzenga
et aL (1986) have used VLBI to measure the length
of several baselines in the southwestern USA and
have found that over a period of4 years movement
on the Fault has been 25±4mma~^. If the SLR
results are not in error, it is probable that the differ
ence in rates provided by the two methods will
92 CHAPTER 5
provide constraints on the regional distribution of
tectonic strain in this area.
Direct measurements of the rate of displacement
across the San Andreas Fault are lower than the
48 mm a"^ predicted from global models of plate
movements(DeMets 1987). However,during
the last half decade of observation, no major earth
quakes have occurred. Over longer time intervals, the
discrete jumps in fault movement associated with
the elastic rebound mechanism of large earthquakes
(Section 2.1.5) would contribute to the total dis
placement and provide a somewhat higher figure for
the average rate of movement. In addition, motion
between the Pacific and North American plates may
be occurring along other major faults which run
parallel to the San Andreas Fault (Section 7.4).
Tapley et al. (1985) have measured changes in
length of four baselines by SLR between Australia
and the North American and Pacific plates, and
have found that the rates differ by no more than
3mma"^ from average rates over the last 2 Ma.
Similarly Christodoulidis et al (1985) and Carter &
Robertson(1986)have measured the relative motion
between pairs of plates and have found a strong
correlation with the kinematic plate model of
Minster & Jordan (1978)(Section 5.3).
Herrings?/>/.(1986)have reported VLBl measure
ments between various telescopes in the USA and
Europe. They have determined that the present rate
of movement across the Atlantic Ocean is 19±
10 mm a"^. The uncertainty in this estimate will un
doubtedly decrease in the future as the model ofthe
atmosphere is improved and multiple telescope obser
vations are made,but even so it agrees with the rate of
23 mm a~^ averaged over the past 1 Ma.
BC
These techniques of direct measurement are
clearly extremely important in that they provide
estimates of relative plate movements that are inde
pendent of plate tectonic models. It is probable that
their accuracy will continue,to improve, and that
observations will become more widely distributed
over the globe. The determination of intraplate de
formation and its relationship to intraplate stress
fields, earthquakes and magmatic activity should also
become possible. Important new findings are antici
pated over the next few decades.
5.8 Instantaneous and
finite rotations
The motions of the plates described in Section 5.3
are termed instantaneous as they refer to movements
taking place at a single instant in time or averaged
over a very short period of geological time. Instant
aneous rotations cannot,therefore, provide informa
tion on the paths followed by the plates in arriving at
the point at which the instantaneous motion is mea
sured. Although it is a basic tenet of plate tectonics
that poles of rotation remain fixed for long periods
of time, consideration of the relationships between
plates forming an interlinked spherical shell reveals
that this cannot be the case for all plates (McKenzie
& Morgan, 1969).
Consider the three plates on a sphere A, B and C
shown in Fig. 5.14a. ^bc ^ac represent
Euler poles for pairs of plates which describe their
instantaneous angular rotation. Let plate A be fixed.
Clearly the poles and P^c can remain fixed
with respect to the relevant pairs of plates. Thus,for
example, any transform faults developing along
BC
BCi
BC^
I Fig.5.14 The three plate problem.
Pac5 Pbc and Pba refer to
instantaneous Euler poles between
plates A and C,B and C,B and A
BA ^ respectively, and a'-c, b'^c and b«a
to their relative rotation vectors. In
(b)P'bc is the present location of
Pbc- See text for explanation.

Fig.5.15 (a) Rotation of
plates A and B about pole Pi
\ 33 it 30 /
produces arcuate fracture A 11
zones with a radius of FZ
curvature of 30°;(b) a jump
of the pole of rotation to P2
causes the fracture zones to
assume a radius of curvature
of90°. P'l represents the
positions of pole Pi after
rotation about pole P2(after
Cox & Hart, 1986, with
permission from Blackwell
Scientific Publications).
common plate margins would follow small circles
centred on the poles. Consider now the relative
movements between plates B and C. It is apparent
that if A,Pba and P^c remained fixed, the rotation
vector ofC relative to B (b"c) acts through P^c and is
given by the sum ofthe vectors and which act
about Pba and Pac-> respectively. Thus, Pgc lies
within the plane ofPgA and Pac and is fixed relative
to A.Such a point, however, does not remain station
ary with respect to B and C. Consequently, relative
motion between B and C must take place about a
pole that constantly changes position relative to B
and C (Fig. 5.14b). Transform faults developed on
the B-C boundary will not then follow simple small
circle routes.
Even when a moving pole is not a geometrical
necessity, it is not uncommon for Euler poles to
jump to a new location (Cox & Hart, 1986). In
Fig. 5.15 the pole of rotation of plates A and B was
initially at Pj,and gave rise to a transform fault with
a small circle of radius 30°. The new pole location is
Pj,60° to the north of Pj, so that the transform fault
is now 90° from P2, that is, on the equator of this
pole. The occurrence of this pole jump is easily
recognizable from the abrupt change in curvature of
the transform fault.
Menard 8c Atwater (1968) have recognized five
different phases of spreading in the northeastern
Pacific. In Fig. 5.16 it is shown that the numerous
large fracture zones of this region appear to lie on
THE FRAMEWORK OF PLATE TECTONICS 93
Pi' N
I-*- A
'T"!
/
Ifz^M
small circles centred on a pole at 79°N,111°E. Ifthe
fracture zone patterns are analysed in more detail,
however,it can be seen that the fracture zones in fact
consist of five different segments with significantly
different orientations which can be correlated be
tween adjacent fracture zones. The apparent gross
small circle form of the fractures only represents the
third phase of movement. It is thus apparent that in
the northeastern Pacific sea floor spreading has taken
place about a pole of rotation that was continually
changing position by small discrete jumps. This pro
gression has been analysed and illustrated in greater
detail by Engebretson ft a/. (1985).
Changes in the direction of relative motions of
plates do not cause large scale deformation of the
plate boundaries but rather result in geometric ad
justments of transform faults and ocean ridge crests.
This may be a consequence ofthe lithosphere's being
thin at accretive margins and consequently ofsmaller
mechanical strength (Le Pichon et al., 1973). That
the adjustments are only minor, however, is appreci
ated from continental reconstructions such as shown
in Fig. 5.17, where the earthquake foci associated
with present day activity are superimposed on the
predrift reconstruction. The coincidence ofshape of
the initial rift and modern plate margins indicates
that there has been little post-drift modification of
the latter.
The past relative positions of plates can be deter
mined by the fitting of lineaments which are known
94 CHAPTERS

160° 180° 160° 140° 120° 100°

^
■
Mendocino —:==^

^ — ' tV \
—V. c\af'°!L-

••"■^ h
r.alap^gos .1 /■ -"
— '^^M^qpesas C rest of Fig. 5.16 Fracture zones in the
J East Pacific Rise northeastern Pacific showing
20° L I — ' 1 1- r / trends corresponding to five
160° 140° 120° ' 1?Qo possible spreading episodes, each
®0° with a new pole of rotation
(redrawn from Menard &
Atwater, 1968, with permission
from Nature 219, 463-7.
Copyright © 1968 Macmillan
Magazines Ltd).

to have been originally juxtaposed. One approach is
to fit former plate margins. Fossil accretive margins
are usually readily identified from their symmetrical
magnetic lineations (Section 4.2), and fossil trans
form faults from the offsets they cause of the linea
tions. Ancient transform faults on continents are
more difficult to identify, as their direction may be
largely controlled by the pre-existing crustal geology.
Their trace, however, normally approximately
follows a small circle route, with any deviations
from this marked by characteristic tectonic activity
(Section 7.4). Ancient destructive margins can be
recognized from their linear belts of calc-alkaline
magmatism, granitic batholiths, paired metamorphic
belts, and, possibly, ophiolite bodies.
The features most commonly used for determin-

Fig. 5.17 Present day

earthquake epicentres
superimposed on a
reconstruction of Australia and
Antarctica (redrawn from
McKenzie 8c Sclater, 1971).

ing earlier continental configurations are continental
margins and oceanic magnetic anomalies. The
former are obviously used to study the form of
pre-drift supercontinents (Section 3.2.2). Because
magnetic anomalies can be reliably dated (Section
4.6), and individual anomalies identified on either
side oftheir parental spreading ridge,the locus ofany
particular anomaly represents an isochron. Fitting
together pairs of isochrons then allows reconstruc
tions to be made of plates at any time during the
history of their drift.
Finally, reconstructions can be accomplished by
the fitting of polar wandering curves (Section 3.6.6),
because if two continents were once juxtaposed they
must have shared the same polar path. This tech
nique is not particularly reliable, as polar wandering
paths are rarely based on anything but a series of
discrete measurements, each of which is subject to a
wide range of uncertainty. Consequently, for the
most recent 200 Ma of Earth history this method is
only used as an independent test of the continental
drift suggested by other methods. Before 200 Ma,
this is the only technique available.
If the relative positions of plates are well con
strained during drift by oceanic magnetic anomalies
and fracture zones,it may be possible to compute the
(a)
Fig.5.18 (a),(b) Evolution of a
trench,(c) Alpine Fault of New
Zealand (redrawn from McKenzie
& Morgan, 1969, with permission
(b)
from Nature 224,125-33.
Copyright © 1969 Macmillan
Magazines Ltd).
THE FRAMEWORK OF PLATE TECTONICS 95
progression of finite rotations between the plates
which led to their present position. Thus, although
any relative movement between two plates can be
described in terms of an angular rotation about a
pole, the plates do not necessarily follow the route
predicted by a single rotation, and may have followed
a much more complex path. Indeed, many long
oceanic fracture zones appear to fit more closely a
route consisting of a series of small circles about
poles which suffered small shifts in position during
their history. Each change in pole position corre
sponds to a change in the direction of plate motion.
5.9 Triple junctions
The stability of the boundaries between plates is
dependent upon their relative velocity vectors. If a
boundary is unstable it will exist only instantan
eously and will immediately devolve into a stable
configuration.
Figure 5.18a shows an unstable boundary between
two plates where plate X is underthrusting plate Y at
be in a northeasterly direction and plate Y is under-
thrusting plate X at ab in a southwesterly direction.
The boundary is unstable because a trench can only
consume in one direction, so to accommodate these
I I I I hi- Trench
=Ridge
— Transform fault
Alpine fault
(c)
96 CHAPTER 5
Fig.5.19 Ridge (R)-trench (T)-transform fault
(F)-triple junction between plates, A,B and C.
movements a dextral transform fault develops at b
(Fig. 5.18b). This sequence ofevents occurred in the
development of the Alpine Fault of New Zealand
(Fig. 5.18c),which is a dextral transform fault linking
the Tonga-Kermadec Trench, beneath which Pacific
lithosphere is underthrusting in a southwesterly
direction, to a trench to the south of New Zealand
where the Tasman Sea is being consumed in a north
easterly direction (McKenzie & Morgan, 1969).
A more complex situation arises when three plates
come into contact at a triple junction. Quadruple
junctions are always unstable, and immediately
devolve into a pair of stable triple junctions, as will
be shown later.
In a similar fashion to a boundary between two
plates, the stability of triple junctions depends upon
the relative directions of the velocity vectors of the
plates in contact. Figure 5.19 shows a triple junction
between a ridge (R), trench (T) and transform fault
(a)
(F). From this figure it can be appreciated that, in
order to be stable,the triple junction must be capable
of migrating up or down the three boundaries be
tween pairs of plates. It is easier to vizualize the
conditions for stability of the triple junction if this
phenomenon is first applied to each boundary in
turn.
Figure 5.20a shows a trench at which plate A is
underthrusting plate B in a northeasterly direction.
Figure 5.20b shows the relative movement between
A and B in velocity space(Cox & Hart,1986),that is,
on a figure in which the velocity ofany single point is
represented by its north and east components, and
lines joining two points represent velocity vectors.
Thus, the direction of line AB represents the direc
tion of relative movement between A and B, and its
length is proportional to the magnitude of their
relative velocity. Line ab must represent the locus ofa
point which travels up and down the trench. This
line, then, is the locus of a stable triple junction. B
must lie on ab because there is no motion of the
overriding plate B with respect to the trench.
Now consider a transform boundary (Fig. 5.21a)
between plates B and C, and its representation in
velocity space (Fig. 5.21b). Again,line BC represents
the relative velocity vector between the plates, but
the locus ofa point travelling up and down the fault,
be, is now in the same sense as vector BC,because the
relative motion direction of B and C is along their
boundary.
Finally, consider a ridge separating two plates A
and C (Fig. 5.22a), and its representation in velocity
space (Fig. 5.22b). The relative velocity vector AC is
now orthogonal to the plate margin, and so the line
ac now represents the locus of a point travelling
along the ridge. The ridge crest must pass through
Fig.5.20 (a) Trench (T) between
plates A and B;(b) its
representation in velocity space
with the velocity line ab
corresponding to its related triple
junction.
 THE FRAMEWORK OF PLATE TECTONICS 97

Fig.5.21 (a) Transform fault(F)

between plates B and C;(b) its
representation in velocity space
with the velocity line be
corresponding to its related triple
junction. (a) (b)

Fig.5.22 (a) Ridge(R)between

plates A and C;(b)its
representation in velocity space
with the velocity line ac ■►E
corresponding to its related triple
junction. (a)

the midpoint of velocity vector CA if the accretion

process is symmetrical with plates A and C each
moving at half the rate of accretion.
ab.bc
By combining the velocity space representations
(Fig. 5.23), the stability of the triple junction can be
determined from the relative positions of the velocity
lines representing the boundaries. If they intersect at
one point, it implies that a stable triple junction
exists because that point has the property of being
able to travel up and down all three plate margins. In
the case of the RTF triple junction, it can be appreci
ated that a stable triple junction exists only if velocity
line ac passes through B, or if ab is the same as be,
that is, the trench and transform fault have the same
trend, as shown here. If the velocity lines do not all
intersect at a single point the triple junction is un ►E
stable. The more general case of an RTF triple junc
ab.bc
tion, which is unstable, is shown in Fig. 5.24.
Figure 5.25 illustrates how an unstable triple Fig. 5.23 Velocity space representation of the plate
junction can evolve into a stable system, and how system shown in Fig. 5.19. Velocity lines ab, be and ac
this evolution can produce a change in direction intersect at the single point J, which thus represents a
of motion. The TTT triple junction shown in stable triple junction.
98 CHAPTERS
(a)
Fig.5.24 (a) Ridge (R)-trench (T)-transform fault(F)
triple junction between plates A,B and C.(b)Its
representation in velocity space. As the velocity lines ab,
be and ac do not intersect at a single point, the triple
junction must be unstable.
Fig. 5.25a is unstable, as the velocity lines represent
ing the trenches do not intersect at a single point
(Fig. 5.25b). In time the system evolves into a stable
configuration (Fig. 5.25c) in which the new triple
junction moves northwards along trench AB. The
dashed lines show where plates B and C would have
been if they had not been subducted. The point X
(Fig. 5.25a,c) undergoes an abrupt change in relative
motion ais the triple junction passes. This apparent
change in underthrusting direction can be distin
guished from a global change as it occurs at different
times and locations along the plate boundary. In
order to be stable, the plate configuration shown in
Fig. 5.25a must be as in Fig. 5.25d. When plotted in
velocity space (Fig. 5.25e) the velocity lines then
intersect at a single point.
McKenzie & Morgan (1969) have determined the
geometry and stability of the 16 possible combina
tions oftrench,ridge and transform fault(Fig. 5.26),
taking into account the two possible polarities of
trenches, but not transform faults. Ofthese, only the
RRR triple junction is stable for any orientation of
the ridges. This comes about because the associated
velocity lines are the perpendicular bisectors of the
triangle ofvelocity vectors, and these always intersect
at a single point (the circumcentre of the triangle).
(b)
The FFF triple junction is never stable, as the velocity
lines coincide with the vector triangle, and, of
course, the sides of a triangle never meet in a single
point. The other possible triple junctions are only
stable for certain particular orientations ofthe juxta
posed plate margins.
Only six types oftriple junction are present during
the current phase of plate tectonics. These are RRR
(e.g., the junction ofEast Pacific Rise and Galapagos
Rift 2k)ne), TTT (central Japan), TTF (junction of
Peru-Chile Trench and West Chile Rise), FFR(pos
sibly at the junction of Owen Fracture Zone and
Carlsberg ridge), FFT(junction ofSan Andreas Fault
and Mendocino Fracture Zone) and RTF(mouth of
Gulfof California).
The evolution of the San Andreas Fault illustrates
the importance of the role of triple junctions. In
Oligocene times (Fig. 5.27a), the East Pacific Rise
separated the Pacific and Farallon plates. The trans
form faults associated with this ridge have been
simplified, and only the Mendocino and Murray
fracture zones are shown. The Farallon Plate was
being underthrust beneath the North American
Plate, and, since the rate of consumption exceeded
the rate ofspreading at the East Pacific Rise,the ridge
system moved towards the trench. The first point of
the ridge to meet the trench was the eastern extrem
ity of the Mendocino Fracture Zone. A quadruple
junction existed momentarily at about 28 Ma, but
this devolved immediately into two triple junctions
(Fig. 5.27b). The more northerly was of FFT type,
the more southerly ofRTF type,and both were stable
 THE FRAMEWORK OF PLATE TECTONICS 99

bc~._ C

(b)

ab

bC^^C

A""-

aC

(d) (e)

Fig.5.25 (a) Triple junction between three trenches consumed are shown as dashed lines,(d)Stable
separating plates A,B and C.(b)Its representation in configuration ofa trench-trench-trench triple junction,
velocity space, illustrating its instability,(c) The positions (e) Its representation in velocity space,((a) and (c)
plates B and C would have reached ifthey had not been redrawn from McKenzie & Morgan, 1969).

(insets on Fig. 5.27b). Because of the geometry of formed in response to the migration of these triple
the system the northern triple junction moved north junctions. The southerly migration of the southern
along the trench and the southern triple junction triple junction ceased as the eastern extremity of
moved south. Thus the dextral San Andreas Fault the Murray Fracture Zone reached the trench
 Type Geometry Velocity triangle Stability Examples Type Geometry Velocity triangle Stability Examples
O
O

Stable if the
All orientations East Pacific o
angles between X
stable Rise and
ab and ac, be >
Galapagos Rift TTR(c) A IB respectively,
Zone -O
are equal,or if
Great Magnetic ac,be form a
Bight straight line X
C lab oi
Stable if ab, ac Stable if ac, be Intersection of
TTT(a) Central form a straight the Peru-Chile
form a straight Japan
line, or if be is line, or if C lies Trench and the
parallel to the on ab Chile Ridge
slip vector OA TTF(a)

Stable if the
complicated Stable if be, ab
TTT(b) general condition form a straight
for ab, bo and line, or if ac
ac to meet at goes through B
TTF(b)
a point is satisfied

Stable if ab, ac
Unstable form a straight
FFF line, or if ab,
TTF(c)
be do so

B
ab must go
RRT through circumcentre
of ABC Stable if C lies Owen Fracture
kC-^^ac on ab,or if Zone and the
FFR A B
ac, be form a Carisberg Ridge
straight line West Chile
Ridge and the
ac! lab Unstable, East Pacific Rise
A I B
RRF Aj-^- evolves to FFR
rr^bj Stable if ab, be San Andreas
1 1 FFT
form a straight Fault and Mendocino
line, or if ac, be Fracture Zone
be I
^"ac do so
Stable if ab
TTR(a) goes through C,
or if ac, be form
Stable if ab goes Mouth of the
>k .S: a straight line
A : B RTF(a) A B ab
lau through C, or Gulf of California
ab if ac. Deform a
straight line
A I
be
be Stable if complicated
general conditions
ar. \ are satisifed -.C
A B Stable if ac, ab
^"A ab B RTF{b)
cross on be

Fig.5.26 Geometry and stability of all possible triple junctions (redrawn from McKenzie & Morgan, 1969).
 THE FRAMEWORK OF PLATE TECTONICS 101

illllii A
M Mendocino iiiiiSj!
iiliy
Pacific Plate North America Plate

iiiiii®S Murray S lllil*

'XvIvi6W«t'IvX'l

iSjIjiiS ^ Plate ixj-jiiiiiiljiiiifc

(a)

Fig.5.27 Evolution ofthe San Andreas

Fault (redrawn from Cox & Hart, 1986). (c)
102 CHAPTERS
(Fig. 5.27c). The triple junction changed to FFT type
and began to move northwards. The Farallon Plate
continued to be subducted to the north and south of
the San Andreas Fault, until the geometry changed
back to that shown in Fig. 5.27b when the East
Pacific Rise to the south ofthe Murray Fracture Zone
reached the trench. The triple junction then reverted
to RTF type and changed to a southerly motion
along the trench. This represents the situation at the
present day at the mouth of the Gulfof California.
Further reading
Bott, M.H.P.(1982) The Interior ofthe Earth, its Structure,
Constitution and Evolution (2nd edn). Edward Arnold,
London.
Burke, K. & Dewey, J.F. (1973) Plume generated triple
junctions: key indicators in applying plate tectonics to
old rocks./. Geol. 81,406-33.
Chase, C.G. (1978) Plate kinematics: the Americas, East
Africa, and the rest of the world. Earth planet. Sci. Lett.
37, 355-68.
Courtillot, V. & Besse, J. (1987) Magnetic field reversals,
polar wander and core-mantle coupling. Science 237,
1140-7.
Cox,A.& Hart,RB.(1986)Plate Tectonics. How It Works.
Biackwell Scientific Publications, Oxford.
Duncan, RA.& Richards, M.A.(1991) Hotspots, mantle
plumes, flood basalts, and true polar wander. Bsv. Geo-
phys. 29, 31-50.
Gordon,RG.&Stein,S.(1992)Global tectonics and space
geodesy. Science 256,333-42.
Isacks, B., Oliver, J. & Sykes, L.R (1968) Seismology
and the new global tectonics./. Res. 73, 5855-
99.
Larson, RL.(1995) The mid-Cretaceous superplume epi
sode. Sci. Am.,Feb.,66-70.
McKenzie,D.P.& Morgan,W.J.(1969)Evolution oftriple
junctions. Nature 224,125-33.
McKenzie, D.P. & Parker, RL.(1967) The North Pacific:
an example oftectonics on a sphere. Nature 216,1276-
80.
Minster, J.B. & Jordan, T.H. (1978) Present-day plate
motions.J.^eophys. Res. 83,5331-54.
Molnar,P.& Stock,J.(1987)Relative motions of hotspots
in the Pacific, Atlantic and Indian oceans since late
Cretaceous time. Nature 327, 587-91.
Morgan, W.J. (1968) Rises, trenches, great faults and
crustal blocks.J.pfeophys. Res. 73,1959-82.
Morgan, W.J. (1971) Convection plumes in the lower
mantle. Nature 230,42-3.
Morgan, W.J.(1972) Deep mantle convection plumes and
plate motions. Bull. Am. Ass. Petrol. Geol. 56, 203-13.
Morgan, W.J. (1981) Hot spot tracks and the opening
of the Atlantic and Indian oceans. In Emiliani, C.(ed.)
The Oceanic Lithosphere. The Sea 7, pp.443-87. Wiley,
London.
Olson,P. & Nam,l.S.(1986)Formation ofseafloor swells
by mantle plumes./.^eophys. Res. 91,7181-91.
Stein, RS.(1987) Contemporary plate motion and crustal
deformation. Rev. Geophys. Space Phys. 25,855-63.
6/ Ocean ridges
6.1 Ocean ridge topography
Ocean ridges mark accretive, or constructive, plate
margins where new oceanic lithosphere originates.
They represent the longest,linear uplifted features of
the Earth's surface, and can be traced by a belt of
shallow focus earthquakes (Fig. 5.2) which follows
the crestal regions and transform faults between
offset ridge crests. The total length of the ridge
system is some 60 000 km and ridges are typically
between 1000 and 4000 km in width. Their crests
are commonly 2-3 km higher than neighbouring
ocean basins, and locally the topography can be quite
rugged and runs parallel to the crests.
6.2 Structure of the axial zone
The gross morphology of ridges appears to be con
trolled by separation rate (MacDonald, 1982)(Sec
tion 6.8). At relatively slow rates of 10-50 mm a"^,
such as the Mid-Atlantic and Atlantic-Indian ridges,
a prominent median rift is developed at the ridge
axis. This is commonly 30-50 km wide and 1500-
3000 m deep, and contains rugged topography.
At intermediate rates of50-90 mm a"^, such as at
the Galapagos spreading centre and the most north
erly East Pacific Rise, the median rift is only 50-
200 m deep and its topography relatively smooth. At
fast rates, greater than 90 mm a~^, such as at the East
Pacific Rise, no median rift is developed and the
topography of the ridge crest is relatively smooth.
These three ridge types are shown in Fig. 6.1, based
on detailed bathymetric data obtained using deeply
towed instrument packages. In all three the axis of
spreading is marked by a narrow zone of volcanic
activity which is flanked by zones of fissuring. Away
from this volcanic zone,the topography is controlled
by vertical tectonics on normal faults. Beyond dis
tances of 10-25 km from the axis, the lithosphere
becomes stable and rigid. These stable regions bound
the area where oceanic lithosphere is generated—an
area known as the 'crustal accretion zone' or 'plate
boundary zone'. An axial topographic high about
8 km in width is often present on fast and intermedi
ate spreading rate ridges (Carbotte & MacDonald,
1994). At fast rates the high arises from the buoyancy
ofhot rock at shallow depth;at intermediate rates the
high may form by volcanic construction.
A detailed study ofa median rift has been made in
the Atlantic Ocean between latitudes 36® 30' and
37®N, a region known as the FAMOUS (Franco-
American Mid-Ocean Undersea Study) area, using
both surface craft and submersibles (Ballard & van
Andel, 1977). The median rift in this area is some
30 km wide, bounded by flanks about 1300 m deep,
and reaches depths between 2500 and 2800 m. In
some areas an inner rift valley is developed which is
1-4 km wide and flanked by a series of fault-
controlled terraces (Fig. 6.2). Elsewhere, howeyer,
the inner floor is wider with very narrow or no
terraces developed. The normal faults which control
the terracing and walls of the inner rift are probably
the locations where crustal blocks are progressively
raised, eventually to become the walls ofthe rift and
thence ocean floor, as they are carried laterally away
from the rift by sea floor spreading. Karson et al,
(1987) have described investigations of the Mid-
Atlantic Ridge at 24®N using a submersible, deep-
towed camera and side-scan sonar. Along a portion
of the ridge some 80 km long they found consider
able changes in the morphology,tectonic activity and
volcanism of the median valley. By incorporating
data supplied by investigations of the Mid-Atlantic
Ridge elsewhere, they concluded that the develop
ment ofthe style ofthe median valley may be a cyclic
process between phases of tectonic extension and
volcanic construction. The possible sequence in the
evolution of the median valley is shown in the flow
chart of Fig. 6.3. The topmost sketch shows the
median valley with an extensively developed neovol-
canic zone in which volcanic construction domin
ates. Thereon the neovolcanic zone progressively
103
104 CHAPTERS
FVVF
vents
F ViV F
FVVF
VE'>'4x
Axis
—I—
20 10 0 10
km
Fig.6.1 High resolution bathymetric profiles ofocean
ridges at fast, intermediate and slow spreading rates. EPR,
East Pacific Rise; MAR,Mid-Atlantic Ridge. Neovolcanic
zone bracketed by Vs,zone offissuring by Fs, extent of
active faulting by Ps(redrawn with permission from
MacDonald, 1982,Annual Review ofEarth and Planetary
Sciences 10. Copyright © 1982 by Annual Reviews Inc.).
decreases in width and tectonic extensional features
become increasingly important until volcanic activity
wanes completely and tectonic extension is the
dominant process affecting the median valley. The
applicability of the flow chart depends upon the
small spreading cells beneath the axial region being
stable in time, and it will be modified as more
information becomes available.
Bicknell et al. (1988) have reported on a detailed
survey ofthe East Pacific Rise at 19®30'S. They have
found that faulting is more prevalent than on slow
spreading ridges, and conclude that faulting ac
counts for the vast majority of the relief. They ob
served both inward and outward facing fault scarps
which give rise to a horst and graben topography.
This differs from slower spreading ridges, where the
topography is formed by back-tilted, inward-facing
normal faults. Active faulting is confined to the
West Western
wall marginal Eastern
..... high Western Central Eastern marginal
500 m trough high trough high
1 0
Vertical and horizontal scale
(EPR3°S)
Intermediate
(ERR 21° N)
Slow
(MAR 37° N)
—1—
20 30
region within 8 km ofthe ridge axis, and is asymmet
rical with the greater intensity on the eastern flank.
The half extension rate due to the faulting is
4.1 mm a"^, compared to 1.6 mm a"^ observed on
the Mid-Atlantic Ridge in the FAMOUS area.
6.3 Broad structure of the
lithosphere below ridges
Gravity measurements have shown that free air
anomalies are broadly zero over ridges (Figs 6.4,
6.5), indicating that they are in a state of isostatic
equilibrium (Section 2.11.6), although small scale
topographic features are uncompensated and cause
positive and negative free air anomalies. The small,
long wavelength, positive and negative free air anom
alies over the crests and flanks, respectively, ofridges
are a consequence of the compensation, with the
positives being caused by the greater elevation ofthe
ridge and the negatives from the compensating mass
deficiency. The gravitational effects ofthe compensa
tion dominate the gravity field away from the ridge
crest, and indicate that the compensation is deep.
Seismic refraction experiments by Talwani et aL
(1965)over the East Pacific Rise have shown that the
crust is slightly thinner than encountered in the main
Fig.6.2 Diagrammatic cross-
section ofthe inner rift valley of
the Mid-Atlantic Ridge at
36°50'N in the FAMOUS area
(redrawn from Ballard & van
Andel, 1977, with permission
Tkm from the Geological Society of
America).
 OCEAN RIDGES 105

20 km
'

Greenstones 8(
* Hydrothermal vents '"Vl metagabbros

Neovolcanic zones Serpent!nites

Fig.6.3 Flow chart for median valley evolution
along the Mid-Atlantic Ridge (redrawn from
Karson et 1987, with permission from Nature
M a| Old Neovolcanic zones \ Faults
328,681-5. Copyright © 1987 Macmillan Bml Older basaltic rocks
Magazines Ltd).
106 CHAPTERS

Heat flow

Free air anomaly

-500

Fig.6.4 Heat flow,free air

gravity anomaly and crustal
Layer 2 structure defined by seismic
refraction across the East Pacific
Layer 3
Rise at 15-17®S. P wave
Moho
velocities in km s"^ (redrawn
Mantle
from Talwani et 1965,
Journal of Geophysical Research
Anomalous
mantle
70, with permission from the
500 500 km American Geophysical Union).

3000-
Bougueranomaly

. 2000-

1000-'
500n
Free air anomaly
Fig.6.5 Gravity anomalies and
0 crustal structure defined by
-500-• seismic refraction across the
W ^Seismic line
0- J=t- '-'hh Mid-Atlantic Ridge at about
31°N. Bouguer anomaly
5- reduction density 2,60 Mg m"^,
Layer 2
^(6.5) 7.3 7 38.2 P wave velocities in km s"^
10- Layer 3 (redrawn from Talwani et al.
Moho l96Sy Journal ofGeophysical
Anomalous mantle
Mantle Research 70, with permission
15
from the American Geophysical
500 500 1000 km Union).

ocean basins, and that the upper mantle velocity
beneath the crestal region is anomalously low
(Fig.6.4). Oceanic layer 1 rocks (Section 2.4.5) are
only present within topographic depressions, but
layers 2 and 3 appear to be continuous across the
ridge except for a narrow magma chamber at the
crest. A similar structure has been determined for
the Mid-Adantic Ridge (Fig. 6.5). The suggestion of
this work that layer 3 is not continuous across the
ridge has now been disproved (Whitmarsh, 1975;
Fowler, 1976).
As the crust does not thicken beneath ridges,
isostatic compensation must occur ^vithin the upper
mantle by a Pratt-type mechanism (Section 2.11.3).
Talwani et al.(1965) proposed that the anomalously
low upper mantle velocities detected beneath ridges
correspond to the tops ofregions oflow density. The
densities were determined by making use of the
Nafe-Drake relationship between P wave velocity
and density (Nafe & Drake, 1963), and a series of
models produced which satisfied both the seismic
and gravity data. One of these is shown in Fig.6.6,
and indicates the presence beneath the ridge of a
body with a density contrast of -0.25 Mgm~^ ex
tending to a depth ofsome 30 km. This large density
contrast is difficult to explain geologically, and it is
 OCEAN RIDGES 107

Observed
Calculated
D 1
L-

0 3000-
3
>
O)
D CO ■

O E
GO o 2000-
c
CO

2.60 2.90
0-
Crust
Anomalous mantle
20-
a Mantle
0
Q
40-

500 500 1000 km

Fig.6.6 Possible model ofthe structure beneath the Mid-
Atlantic Ridge from gravity modelling with seismic
refraction control. Densities in Mg m"^(redrawn from
Talwani et \96S^ Journal ofGeophysical Research 70,
with permission from the American Geophysical Union).
probable that the Nafe-Drake relationship does not
apply to rocks which may contain significant frac
tions of partially molten material. An alternative
interpretation, constructed by Keen & Tramontini
(1970), is shown in Fig.6.7. A much lower, more
realistic density contrast of -0.04 Mgm~^ is em
ployed, and the anomalous body is considerably
larger, extending to a depth of 200 km.
Given the ambiguity inherent in gravity model
ling, the two interpretations shown probably
represent end members ofa suite ofpossible interpre
tations. Indeed, seismic tomography (Section 2.1.8)
suggests the low velocity region beneath ocean ridges
extends to a depth of 100 km (Anderson et al,^
1992). They demonstrate without ambiguity, how
ever, that ridges are underlain by large, low density
bodies in the upper mantle whose upper surfaces
slope away from the ridge crests.

6.4 Origin of anomaious upper

Bathymetry
mantle beneath ridges
x:
a
There are three possible sources of the low density
<D
Q regions which underlie ocean ridges and support
them isostatically (Bott, 1982a).
Free air 1 Thermal expansion of upper mantle material be
1000-, gravity anomaly
ly A Calculated Observed neath the ridge crests, followed by contraction as sea
floor spreading carries it laterally away from the
source of heat. Suppose the average temperature to a
depth of100 km below the Moho is 500"C greater at
the ridge crest than beneath the flanking regions, the
^ 100
average density to this depth is 3.3 Mgm"^ and the
Q.
Anomalous
volume coefficient of thermal expansion is 3 x 10"^
mantle Mantle per degree. In this case the average mantle density to
a depth of 100 km would be 0.05 Mg m"^ less than
that of the flanking ocean basins. If isostatic equilib
500 km rium were attained, this low density region would
support a ridge elevated 2.2 km above the flanking
Fig.6.7 Alternative model of the structure beneath the
areas.
Mid-Atlantic Ridge from gravity modelling. Profile at
46 ®N. Densities in Mgm"^(redrawn from Keen & 2 The presence of molten material within the anom
Tramontini, 1970, with permission from the Royal alous mantle. If the degree of partial melting were
Astronomical Society). 1%, the consequent decrease in density would be
108 CHAPTER 6
about 0.006 Mgm"^. Extended over a depth range
of 100 km this density contrast would support a
relative ridge elevation of0.25 km.It is very difficult
to assess the contribution ofthis source as the extent
of partial melting in the anomalous mantle is un
known. However, theoretical relationships between
seismic velocity and density for solid materials do not
seem to apply to the anomalous mantle and so it is
possible that this discrepancy results from the pres
ence ofa molten fraction.
3 A temperature dependent phase change. The high
temperatures beneath ocean ridge crests might cause
a transition to a mineralogy oflower density. Normal
upper mantle is made up of pyroxene pyrolite (Sec
tion 2.8.3), but at the high temperatures and low
pressures experienced within the top 40 km below a
ridge crest the lower density form, plagioclase pyro
lite, would be more stable. This has a density up to
0.07 Mgm"^ lower than pyroxene pyrolite, and, ex
tending over a depth range of20 km,would support
a ridge elevation of some 0.7 km. A problem with
this hypothesis is that the upper mantle in this area
has been depleted by removal ofthe molten fraction
so that plagioclase may not be able to develop in
great quantities.
The major factor contributing to the uplift of
ridges is thus thermal expansion and contraction,
which can also account for the topography of the
ocean basins (Section 6.9). Partial melting and phase
changes may also contribute to a lesser, unknown
extent. Certainly, the model shown in Fig. 6.6 ap
pears unreasonable because too high a density con
trast is required, and the model shown in Fig. 6.7 is
preferred. However, the model of Keen & Tramon-
tini(1970)can also be criticized in that the densities
employed are rather too high and provide too low a
density contrast, and the depth to the base of the
anomalous mass is too great. A model which em
ploys densities of 3.35 and 3.28 Mgm"^ for normal
and anomalous mantle, respectively, with the anom
alous mass extending to a depth ofICQ km,would be
more in accord with geological and geophysical data.
6.5 Shallow structure
of the axial region
Models for the formation of oceanic lithosphere
normally require a magma chamber beneath the
ridge axis within which differentiation leads to the
formation of at least part of oceanic layer 3(Section
6.10). Evidence for the presence of such a magma
chamber has been provided from detailed seismic
surveys.
On the East Pacific Rise,a seismic refraction survey
using ocean bottom seismometers (Orcutt et al.^
1976) detected a pronounced low velocity zone
about 2km beneath the ridge axis, which is underlain
by upper mantle with the anomalously low velocity
of 7.7 km s"^. Away from the crest, the low velocity
zone is absent beneath crust 2.9 Ma old but the
anomalous upper mantle persists. By the time that
crust 5 Ma old is reached, normal oceanic layering
is encountered and normal mantle velocities are
present.
Herron et al. (1978; 1980), employing multi
channel seismic techniques, detected reflections
from the top of a magma chamber beneath the East
Pacific Rise. This was 1.5-2.0 km beneath the ridge
crest, and its width, varying from 2-8 km, appeared
to correlate with the varying width ofthe ridge crest
(Fig. 6.8). The sides ofthe magma chamber were not
detected, being either too steep to provide reflections
at near vertical incidence or too gradational to pro
vide a reflecting horizon. This work has been ex
tended by Detrick et al. (1987), who similarly
Moho
vC-.^
Mono
Moho
I— —1
10 10
West Ridge crest East
Fig.6.8 Seismic structure ofthe East Pacific Rise between
9"15'N and 9'40'N (redrawn from Herron etal.^ 1980).

detected the top of a low velocity magma chamber
1.2-2.4 km deep and less than 4-6 km wide beneath
the axis of the East Pacific Rise. Moreover, by run
ning a reflection profile along the ridge axis they were
able to demonstrate that the magma chamber is
present as a continuous feature along at least 60% of
the crestal region surveyed. Further work on the
records(Mutter 1988a) has demonstrated that
the magma chamber is often displaced laterally from
the rise axis. Tomographic imaging (Section 2.1.8)
ofthe crestal region ofthe East Pacific Rise.(Burnett
et «/., 1989) has shown that the actual magma
chamber(> 50% melt)is quite small, being no more
than a few hundred metres thick and 4km wide
supplied by a narrow feeder. Surrounding the cham
ber is a zone with a velocity up to 1 km s~^ lower than
normal. This is interpreted as a reservoir of very hot
rock containing little melt, whose low density iso-
statically supports the ridge topography. Kent et al
(1990)have further decreased the estimate ofthe size
of the magma chamber, suggesting a width of 1 km
and a thickness of 100 m. Harding et al. (1989),
however, suggest that magma is present within a
zone at least 6 km wide.
The crest of the Mid-Atlantic Ridge has been less
intensively surveyed and the evidence for the pres
ence ofan axial magma chamber is less clear. Francis
& Porter (1973), from the recording of local earth
quakes with ocean bottom detectors, located a low
velocity region about 3 km beneath the median valley
which they interpreted as a magma chamber. Whit-
marsh (1975), using seismic refraction methods,
similarly inferred the presence of a low velocity zone
2-3 km wide extending through the upper 2.5 km of
crust. Calvert(1995) has isolated seismic reflections
from a survey at 23®17'N which have been inter
preted as arising from the top ofa 4 km-wide magma
chamber at a depth of 1.2 km. However, Fowler
(1976)disputed the presence ofany sizeable magma
chamber as S waves propagate across the axis,indicat
ing the absence ofa fluid phase.
These investigations of the East Pacific Rise and
Mid-Atlantic Ridge appear to indicate the presence
of an axial magma chamber a few kilometres wide at
shallow depth. This width, however, is an order of
magnitude less than implied from some ophiolite
models (Section 2.5) in which magma chambers are
inferred to be up to 30 km wide. The seismic data,
moreover, cannot distinguish between the presence
ofliquid magma and a partially molten crystal mush.
OCEAN RIDGES 109
The most recent studies, summarized by Smith &
Cann (1993), suggest that no steady state magma
chamber exists beneath ocean ridges; rather, discrete
pockets of magma rise to feed small volcanoes in the
inner rift. Contrary to the melt-dominated magma
chambers described in Section 6.3, Sinton & Detrick
(1992) have proposed a model in which magma
chambers comprise narrow, hot, crystal-melt mush
zones. Magma chambers are viewed as composite
structures comprising an outer transition zone made
up of a hot, crystal framework with only small
amounts of interstitial melts and an inner zone of
crystal mush with sufficient melt for it to behave as a
very viscous fluid. Pure melt only develops in fast
spreading ridges where sufficient magma is present to
form a thin melt lens on top of the mush zone
(Fig.6.9a). This lens may be tens of kilometres long
along the ridge crest, but only tens or hundreds of
metres in thickness. Slow spreading ridges have in
sufficient magma supplies for a melt lens to develop
(Fig.6.9b) and eruptions only occur when there are
periodic influxes of magma from the mantle. Such a
model is consistent with seismic reflection data from
ocean ridges and petrological observations which
require magma to have been modified by crystalliza
tion or mixing within the crust, which could not
occur in a large, well mixed chamber. It also explains
why less fractionation occurs in the volcanic rocks of
slow spreading ridges. A problem with this model,
however, is that it is not apparent how the layered
gabbros oflayer 3 might develop.
Many early investigations of ocean ridges were
essentially two-dimensional in that they were based
on quite widely spaced profiles oriented perpendi
cular to their strike. More recently 'swath'-mapping
systems have been employed which provide complete
aerial coverage of oceanic features. These systems
have been used to reveal variations in the structure of
ocean ridges along strike. A full review of these
developments is provided in MacDonald et al.
(1988).
Studies ofthe East Pacific Rise have shown that it is
segmented along its strike by the occurrences of
non-transform ridge axis discontinuities such as
propagating rifts (Section 6.11) and overlapping
spreading centres(OSC) which occur at local depth
maxima,and by smooth variations in the depth ofthe
ridge axis. These features may migrate up or down
the ridge axis with time.
OSCs (MacDonald & Fox, 1983) are non-rigid
 110 CHAPTERS
Transition
(b) 2
QJ e Transition
Q 6
Distance (km)
Fig.6.9 Interpretive models of magma chambers beneath
a fast(a) and slow (b) spreading ridge (after Sinton &
Derrick, 1992).
discontinuities where the spreading centre of a ridge
is offset by a distance of 0.5-10 km, with the two
ridge portions overlapping each other by about three
times the offset. It has been proposed that OSCs
originate on fast spreading ridges where lateral off
sets are less than 15 km,and true transform faults fail
to develop because the lithosphere is too thin and
^Volcanics c
Mush
Rift valley
weak. The OSC geometry is obviously unstable, and
its development has been deduced from the behav
iour of slits in a solid wax film floating on molten
wax, which appears to represent a reasonable ana
logue (Fig. 6.10a). Tension applied orthogonal to
the slits (spreading centres) causes their lateral prop
agation (Fig. 6.10b) until they overlap (Fig. 6.10c),
and the enclosed zone is subjected to shear and
rotational deformation. The OSCs continue to ad
vance until one tip links with the other OSC
(Fig. 6.10d). A single spreading centre then develops
 OCEAN RIDGES 111

/ Two knife cuts in frozen wax Axial depth profile

film,spreading initialized Long-wavelength undulation of the axis
Short-wavelength undulations of the_axis

2500-
il III* _ ^1 ...
I

(a)
^ 3000-
TJ

/ Propagation of spreading
Melt segregation events A - E need not
centres along strike
3500- be synchronous
,/ -50-100 km

/■*- Upwelling asthenosphere

(b) /
cbronous
Spreading centres overlap
and curve towards each
other,encircling a zone Chamber
of shear and rotational
deformation. OSC geometry
established
(0 Ascending partial melt y
from 30 - 60 km depth 9

/ Progressive shear and

rotational deformation
(§J continues until one
OSC links with
the other Fig. 6.11 Schematic model for the segmentation of an
ocean ridge by magmatic processes, (a) Strike section of a
(d) /
ridge crest. Roman numerals refer to discontinuities of
orders one, two, three and four. Capital letters refer to
■Mf 1 -► A continuous spreading melt segregation effects occurring at 30-60 km beneath
centre is established. the axis, (b) Enlarged area of the dashed circle in (a)
^/
{$ abandoned OSC and showing a possible cause of fourth order segmentation
1/
1
overlap zone rafted
away
(redrawn from MacDonald etaL^ 1988, with permission
from the Geological Society of America).
(e) '
still thin and hot) and large offset (3-5 km) OSCs
Fig. 6.10 Possible evolutionary sequence in the which cause axial depth anomalies of hundreds of
development of an overlapping spreading centre (redrawn metres. Third order segmentation at intervals of
from MacDonald & Fox, 1983, with permission from
30-100 km is defined by small offset (0.5-3 km)
Nature 302, 55-8. Copyright © 1983 Macmillan
OSCs, where depth anomalies are only a few tens of
Magazines Ltd).
metres. Finally, fourth order segmentation at inter
vals of 10-50 km is caused by very small lateral
as one OSC becomes inactive and is moved away as offsets (<0.5 km) of the axial rift and small devia
spreading continues. tions from axial linearity of the ridge axis. These are
Ocean ridges are segmented at several different rarely associated with depth anomalies and may be
scales (Fig. 6.11). First order segmentation is defined represented by gaps in th^ volcanic activity within the
by fracture zones (Section 7.3) and propagating central rift or by geochemical variation.
ridges, which divide the ridge at intervals of 300- Seismic reflection investigations (e.g. Fig. 6.8)
500 km by large axial depth anomalies. Second order have revealed a strong correlation between the cross-
segmentation at intervals of 50-300 km is caused by sectional form of the ridge, the presence or absence
non-ri^d transform faults (which affect crust that is of an axial rift and the presence or absence of a
112 CHAPTER 6
shallow axial magma chamber. Magma chambers
usually occur where the ridge is broad and where an
axial rift is present. It has been suggested that this
correlation, and the lateral segmentation of ocean
ridges, arises from the variation in space and time of
the availability of magma beneath the ridge axis.
Increasing magma supplies give rise to shallow
magma chambers and a consequent broadening axial
bulge on which an axial rift is developed. Decreasing
magma supply causes the presence of only a small
magma chamber, or its absence, the axial region to
be narrow, and the absence of an axial rift because
the magma chamber is too small to give rise to a
significant collapse structure.
Segmentation of ocean ridges thus appears to
be controlled by the distribution of partial melts
beneath them (Toomey et aL^ 1990; Gente et al,^
1995), which feed magma chambers at discrete loca
tions along them and create local depth anomalies.
The ridge model of Sinton & Derrick (1992), de
scribed above, precludes extensive mixing within the
small axial magma chamber along the ridge, and
could explain the observed geochemical segmenta
tion. With time the magma may migrate away from
its sources,creating a gradual increase in depth ofthe
axis as the pressure within it gradually wanes. This
phenomenon may explain the non-coincidence of
magma chamber and rise culmination noted by
Mutter et al. (1988a). The brittle shell overlying the
magma stretches and cracks and magma intrudes so
that eruptions follow the path of magma migration.
After eruption the removal of supporting magma
gives rise to the formation ofan axial summit graben.
Evidence for the pulse-like, episodic spreading of
ridges,in which sea floor spreading occurs alternately
by stretching and fracturing of existing crust and
then dyke injection and copious volcanicity has been
provided by studies ofophiolites(Harper,1978)and
seismic reflection profiles across old Atiantic litho-
sphere (McCarthy et aL^ 1988). Ridge axis discon
tinuities consequently occur where the magmatic
pulses finally come to rest. The variable bathymetry
and associated geophysical and geochemical differ
ences imply that adjacent segments of ridge have
distinct, different sources in the mantle. First to third
order segmentation is caused by the variable depth
associated with magma migration, and fourth order
effects are caused by the geochemical differences in
magma supply.
Third and fourth order segmentations appear to be
short-lived, as their effects can only be traced for a
few kilometres in the spreading direction. Second
order segmentations, however, create off-axis scars
on the spreading crust consisting of cuspate ridges
and elongate basins which cause differential relief of
several hundred metres. The scars do not follow
small circle routes about the speading pole, but form
V-shaped wakes at 60-80® to the ridge. This indi
cates that the OSCs responsible for the segmentation
migrate along the ridge at velocities of up to several
hundred millimetres per year. Figure 6.12 summa
rizes the three general cases for the evolution of
ridge-axis discontinuities in terms of the movement
of magma pulses.
The residual structures of second order disconti
nuities leave a surprisingly wide wake 30-50 km
wide and disrupt magnetic lineations over a zone
50-80 km wide. It has been estimated that about
20% ofthe sea floor has been affected by such wakes.
All ofthe ridge-axis discontinuities responsible for
segmentation have their origin in the upwelling and
segregation of molten mantle material beneath the
ridge and the subsequent propagation of pulses of
magma along it. Differences between the types of
discontinuity may arise because oftheir relative size,
duration, structural evolution, frequency and
whether they are related to major changes in direc
tion ofthe spreading ridge.
The fine scale mapping of ocean ridges which is
now possible with modern equipment has clearly
allowed new ideas to evolve about the detailed behav
iour of the ridge crests. The ideas concerning lateral
segmentation were derived on the East Pacific Rise,
but Sempere et al. (1990) and Gente et aL (1995)
have demonstrated that a similar segmentation is
present on the Mid-Atlantic Ridge. There, the seg
mentation is recognized by variation in the ridge-axis
morphology. It will be interesting to see if similiar
segmentation is present on currentiy unexplored
regions ofthe ocean ridge system.
6.6 Heatflow and hydrothermall
circulation
On a large scale it has been found that there is a
relationship between the heat flow from oceanic
lithosphere and the inverse square root of its age,
which leads to the relationship between water depth
and age discussed in Section 6.9. On a fine scale,
however, there are discrepancies between observed
 OCEAN RIDGES 113

Case 1. Meeting Magmatic pulses meet head-on.

or are aligned but stop short of
meeting

Small or no
offset or
depth anomaly.
no discordant zone
on flanks

Case 2. Linking and Decapitation Magma pulses misalign

but eventually link
Decapitated
ridge .
tip Off-axis discordant
zone of
relict ridges 0)
oo\ and whole overlap
basins, average
depth 200 m In this case,
deeper than envelope of scars
adjacent indicates S
sea floor migration of
the offset

Case 3. Self-Decapitation Magma pulses misalign and do not

link, ridge tips self-decapitate

N and S Overlap basin

ridges rafted off
curve on W
inside flank as
offset
migrates S

Fig.6.12 Three possible

general cases for the evolution
of ridge-axis discontinuities. Envelope indicates Off-axis discordant zone
Arrows along axis refer to S migration of relict ridges, curved
fault traces, no relict
direction of propagation of
Off-axis overlap basins unless
magmatic pulses, fj, discordant zone migration of offset
refers to a time sequence. of relict ridges, reverses direction
Cases 2 and 3 apply to second whole or partial
and third order overlap basins,
discontinuities, and case 3 average depth 200 m
deeper than adjacent
may also apply to first order
sea floor
propagating rifts (after
MacDonald^f «/., 1988).
114 CHAPTER 6
and expected heat flows for young lithosphere. In
particular, there is a large scatter in heat flow magni
tude near the crests of ocean ridges. Thermal lows
tend to occur in flat-floored valleys and highs within
areas of rugged topography (Lister, 1980). Blanket
ing by sediment does not appear to be the cause of
the low heat flow as the troughs are the least sedi-
mented areas of the ridge and also the youngest and
therefore hottest. To explain these phenomena it
has been proposed that the pattern of heat flow is
controlled by the circulation ofsea water through the
rocks ofthe oceanic crust.
Although the penetration of water through the
hard rock ofthe sea floor at first seems unlikely,it has
been shown that thermal contraction can induce a
suflficient permeability for efficient convective flow to
exist. The cracks are predicted to advance rapidly and
cool a large volume ofrock in a relatively short time,
so that intense localized sources ofheat are produced
at the surface. Active geothermal systems which are
driven by water coming into contact with near-
molten material are expected to be short-lived, but
the relatively gentle circulation ofcool water, driven
by heat conducted from below, should persist for
some time. The chemistry of hydrothermal springs
on the East Pacific Rise and Mid-Atlantic Ridge is
remarkably similar, in spite of the great diflference
in spreading rates of these two accretive margins
(Campbell 1988).
Further evidence that hydrothermal circulation
occurs comes from the presence of metalliferous
deposits in the crestal area. The metals are those
known to be hydrothermally mobile, and must have
been leached from the oceanic crust by the ingress of
sea water which permitted their extraction in a hot,
acidic, sulphide-rich solution (Rona, 1984). On
coming into contact with cold sea water on or just
below the sea floor the solutions precipitate base
metal sulphide deposits. The presence of such de
posits is corroborated by studies of ophiolites (Sec
tion 12.3.2).
Detailed heat flow surveys on the Galapagos Rift
have shown that the pattern oflarge scale zoning and
wide range of individual values is consistent with
hydrothermal circulation. Small-scale variations are
believed to arise from variations in the near-surface
permeability, while larger scale variations are due to
major convection patterns which exist in a permeable
layer several kilometres thick which is influenced by
topography, local venting and recharge at basement
outcrops. The penetration of this convection is not
known, but it is possible that it is crust-wide.
Hydrothermal circulation ofsea water in the upper
crust beneath ocean ridges transports about 25% of
the global heat loss, and is clearly a major factor in
the Earth's thermal budget.
6.7 Petrology of ocean ridges
Under normal conditions the peridotite ofthe upper
mantle does not melt. However,the high heat flow at
ocean ridges implies that the geothermal gradient
crosses the peridotite solidus at a depth of about
50 km (Wyllie, 1981, 1988), giving rise to the
parental magma of the oceanic crust (Fig. 2.32).
A similarly high geothermal gradient is believed to
exist beneath oceanic islands as oceanic lithosphere
traverses a mantle plume or hotspot(Section 5.4),so
generating basaltic rocks by a similar mechanism.
Ocean ridge basalts have the composition of oliv-
ine tholeiite (Kay et «/., 1970), and exhibit only
minor compositional variation caused by variable
alumina and iron contents. They may contain phe-
nocrysts of olivine or plagioclase or, rarely, clinopy-
roxene (Nisbet & Fowler, 1978). Three alternative
sources have been suggested for the parental magma
of olivine tholeiite: a small degree of partial melting
of mantle peridotite at 75-90 km followed by precip
itation ofolivine as the melt ascends; 20-30% partial
melting at depths in the range 30-60 km; shallow
melting with separation ofthe melt at 15-25 km.
The simplest interpretation of the chemistry of
oceanic basalts, suggested from experimental petrol
ogy, is that separation of the partial melt occurs at a
depth of 15-25 km. However, a wide range of alter
native interpretations exist. The analysis of trace
elements reveals that much of the compositional
variation in the basalts is explicable in terms of high
level fractionation. To explain the most extreme
variations, however, it is necessary to invoke the
mixing of batches of magma. The frequent presence
of xenocrysts of deep-level origin indicates that the
rocks only spend a very short time in a high level
magma chamber.
On a smaller scale, a detailed sampling ofthe East
Pacific Rise by Langmuir et al,(1986) has revealed a
series of basalts which are diverse in their major and
trace element chemistry. This compositional vari
ation has been interpreted in terms of a series of
magmatic injection centres along the crest of the

ridge which correlate with bathymetric highs spaced
about 50-150 km apart. Magma moves outwards
from the injection points along the ridge so that the
temperature of eruption decreases regularly from
maxima at the bathymetric highs, which correspond
to the centres of small spreading cells. Batiza et cd.
(1988) have sampled along the axis of the southern
Mid-Atlantic Ridge, and have shown that there are
regular patterns ofchemical variation along it caused
by differences in the depth and extent of partial
melting and degree of fractionation. They conclude
that these patterns imply the presence of a deep
central magma supply, with no melt migration along
the axis and no large, well mixed magma chamber.
6.8 Differences between fast and
slow spreading ridges
The morphology of ridges, and in particular the
development of a median rift, appears to correlate
with the rate of speading and magma supply (Mor
gan 8c Chen, 1993). Spreading rates and morphol
ogy also appear to correlate with the ridge's being
situated within an expanding or a contracting ocean.
In expanding oceans,such as the Atlantic and Indian
Oceans,spreading rates are low,ridges tend to follow
the centre of the ocean basins and a median rift is
developed. In contracting oceans,such as the Pacific,
spreading rates are high, ridges may be displaced
from the centre and no median rift is developed.
There are several hypotheses concerning the presence
or absence of a median rift(MacDonald, 1982).
The 'hydraulic head loss' model suggests that
viscous forces within a conduit tapping the magma
source are sufficient to decrease the hydraulic pres
sure,causing a bathymetric depression over the ridge
axis. This hydraulic head loss then causes uplift of
the rift valley walls relative to the floor. The loss of
hydraulic pressure is proportional to the upwelling
velocity and inversely proportional to the cube ofthe
conduit diameter. At slow spreading ridges, magma
ascends through narrow conduits formed of cool,
old lithosphere. At fast spreading ridges there is
much less head loss as the conduits are flanked by
hotter lithosphere, and a flat topography develops
instead of a rift.
A mechanism similarly dependent upon magma
viscosity (Cann, 1974) suggests that the ascending
asthenosphere beneath the ridge can exert a viscous
drag on the newly formed lithosphere on either side.
OCEAN RIDGES 115
The drag would then prevent the ascent of the
lithosphere in the crestal region while flanking litho
sphere rises by movement along normal fault planes
(Fig.6.13). This model thus relates the formation of
a median rift to the presence of high viscosity
magma, and its absence if the upwelling materials is
oflow viscosity.
A different hypothesis involves 'steady state neck
ing' of the lithosphere. In this model the median rift
forms by necking or thinning in a ductile layer
beneath the rift in the manner of a plastic material
necking out under tension. The layer does not break
as new material is constantly added from below. At
slow spreading ridges the strength ofthe lithosphere
is considered to be sufficiently high that necking
creates a median rift. Fast spreading ridges are char
acterized by young crust which is believed to be too
hot and weak for necking to be significant.
Davis (1984), by contrast, has attributed the for
mation of a median rift to the superimposition of
anelastic deformation due to dyke intrusion/
fracturing onto the isostatic topography predicted by
plate cooling theory.
Flower (1981) has shown that differences in the
lithology and chemistry of basalts generated at mid-
ocean ridges show a simple correlation with spread
ing rate. The differences are not related to processes
in the upper mande, as the primary melts appear
to be identical. They are believed to reflect the
fractionation environment after partial melting.
Slow spreading systems are characterized by a com
plex magma chamber in which there is a widespread
accumulation of calcic plagioclase, the presence of
phenocryst-liquid reaction morphologies and
pyroxene-dominated fractionation extracts. These
phenomena are consistent with fractionation at
many different pressures in a chamber which appears
to be transient. This conclusion is in accord with the
pattern of rare earth elements in basalts sampled
from the Mid-Atlantic Ridge (Langmuir et al..
Magma
chamber
Fig.6.13 Possible mechanism for the development of a
median rift in terms of a magma body of high viscosity.
116 CHAPTERS
1986). Although a homogeneous mantle source is
suggested, the variations in rare earth chemistry
apparent in samples from adjacent areas indicate a
complex subsequent history of differentiation. Fast
spreading ridges, however, suggest low pressure
basalt fractionation trends to iron-rich compositions
with little plagioclase accumulation or crystal-liquid
interaction. The magma chamber thus appears to be
a stable and steady state feature.
6.9 Depth-age relationship of
oceanic lithosphere
As newly formed oceanic lithosphere moves away
from a mid-ocean ridge, it becomes removed from
underlying heat sources and cools. This cooling
has two effects. First, the lithosphere contracts
and increases in density. Secondly, because the
lithosphere-asthenosphere boundary is controlled
by temperature (Section 2.12), the cooling causes
the lithosphere to increase in thickness away from the
mid-ocean ridge. This latter phenomenon has been
confirmed by lithosphere thickness estimates derived
from surface wave dispersion studies in the Pacific
Ocean, which indicate that the thickness increases
from only a few kilometres at the ridge crest to 35 km
at 15 Ma age and 70 km at 50 Ma.
The cooling and contraction of the lithosphere
cause a progressive increase in the depth to the top of
the lithosphere away from the ridge (Sclater &
Francheteau, 1970), accompanied by a decrease in
heat flow.It follows that the width ofa ridge depends
upon the spreading rate, and so provides an explana
tion for the relative widths of the rapidly spreading
East Pacific Rise and more slowly spreading Mid-
Atlantic Ridge. Parsons & Sclater(1977) have deter
mined the nature of the age-depth relationships of
oceanic hthosphere, and have shown that the depth,
d (metres) is related to age t(Ma)by:
rf = 2500 + 350>/f
It has been found, however, that this relationship
only holds for oceanic lithosphere younger than
80 Ma. For older lithosphere the relationship indi
cates a more gradual increase of depth with age. In
order to explain this. Parsons & McKenzie (1978)
suggested a model in which the cooling layer com
prises two units rather than the single unit implied by
Parsons & Sclater (1977). In this model the upper
unit, through which heat moves by conduction is
Distance for spreading rate of
Age(Ma) 10 mm a"^(km)
200 150 100 50 0 500 1000 1500 2000
Fig.6.14 Theoretical relationship of water depth and age
or distance for a spreading rate of 10 mm a"^.
mechanically rigid, and the lower unit is a viscous
thermal boundary layer. As the lithosphere travels
away from a spreading centre, both units thicken and
provide the linear d-yft relationship described
above. However, the lower unit eventually thickens
to the point at which it becomes stable and starts to
convect. This brings extra heat to the base of the
upper layer and prevents it thickening at the same
rate. The age-depth relationship for oceanic litho
sphere older than 80 Ma is then given by:
d = 6400- 3200exp(-f/62.8)
These age-depth relationships are shown in
Fig.6.14, and appear to explain the observed
bathymetry in most areas (e.g., Renkin & Sclater
1988).
6.10 Origin of the oceanic crust
A widely accepted model of the petrological pro
cesses occurring at ocean ridges has been derived by
Cann (1970; 1974). In this model (Fig.6.15), hot
asthenospheric material ascends buoyantly (Nicolas
et»/., 1994) suflSciently rapidly up a narrow zone to
pass through the basalt melting curve and provides
an interstitial melt of basaltic composition. The mol
ten fraction increases in volume as the asthenosphere
rises, and eventually departs the parental material to
ascend independently and produce a magma cham
ber within the lower part of the oceanic crust at the
level oflayer 3. Part ofthis magma is erupted on the
ocean floor through fissures to produce a sequence of
pillow lavas, and immediately beneath is a zone of
dykes. These collectively make up layer 2. Kidd
(1977) has modelled these processes of extrusion
and intrusion and compared them with observations
 OCEAN RIDGES 117

Spreading axis
Pillow lavas
Layer2
Sheeted dyke
complex
3=4 Magma chamber Gabbro

Seismological • • **Cumulates4/>'
Moho t'Tr
antle Peridotite
Petrological
Moho
Hviagma rises in crystal mush

I 1 1
5 0 5 km
Horizontal and vertical scale

Fig.6.15 Diagrammatic cross-section of the structure of clase also crystallizes and layered olivine gabbros
the crust and upper mantle beneath the crestal region of a form.Much ofthe residual liquid, still volumetrically
mid-ocean ridge. The shape of the magma chamber is quite large, then solidifies over a very small temper
based on the seismic reflection model of Detrick et al. ature range to form an upper,'isotropic' gabbro. A
(1987)(after Bott, 1982a). small volatile-rich residuum of this diflferentiation
process, consisting essentially of plagioclase and
of ophiolite complexes. Layer 2C was found to quartz, is the last fraction to crystallize, sometimes
consist entirely of sheeted dykes, which were in intruding upwards to form veins and small pockets of
truded through zones less than 50 m wide, and 'plagiogranite' within the overlying sheeted dyke
which demonstrate a highly extensional regime. The complex. The abundance of volatiles, notably water,
dykes show so^e 10% more chilled margins on one in the uppermost part ofthe magma chamber may be
side than the -^wing that approximately 10% due, at least in part, to interaction with sea water
of the dyke Xu dykes, such that the percolating downwards and/or stoping of the over
margins o' ^ up on opposite lying, hydrothermally altered dykes into the magma
sides of *« sea floor chamber.
spreadir ause The two gabbro units, isotropic and layered, are
dyke the often correlated with seismic layers 3A and SB,
iSt. It respectively. The ultramafic cumulates, rich in oliv
^ dykes ine and pyroxene, would then account for the sub-
low less Moho seismic velocities. Thus, the Moho occurs
dykes are within the crystallized magma chamber at the base of
tst as they the mafic section. Oflf-axis, however, in a lower tem
a result of perature environment, the uppermost ultramafics
,o metamor- may become partially hydrated (i.e., serpentinized)
ibrate at high and as a result acquire lower seismic velocites more
iter. characteristic of layer SB. The seismic Moho would
from the crystal- then occur at a somewhat greater depth, within the
r(Fig.6.15). The ultramafic section. As a result of this uncertainty in
magma chamber, defining the seismic Moho, petrologists have tended
through the magma to define the base of the crust as the base of the
nite with occasional magma chamber, that is, the dunite/chromitite ho
accumuiationr^Ot-.,^^Jj^ With further cooling rizon. Hence, this level is termed the 'petrological
pyroxene crystallizes ahci-^ lulate peridotitic layers Moho'(see Fig.6.15).
(i.e., of olivine and pyroxene) are produced, giving The magma chamber is continuously replenished
way upwards to pyroxenites as the crystallization of by magma rising from the asthenosphere, which
pyroxene begins to dominate. Ultimately, plagio- limits the degree of fractionation possible and
118 CHAPTER 6
explains the preponderance of basaltic rocks within
the volcanics dredged from the oceans. The highly
fractionated varieties occasionally encountered in the
oceans and ophiolite bodies probably result from the
isolation of magma bodies by lateral injection from
the magma chamber.
Cann (1974) has also examined how oceanic
crustal structure would vary with spreading rate. No
change in crustal thickness is predicted as the amount
ofmagma available depends only on the temperature
of the upwelling asthenosphere which controls the
depth at which partial melting takes place. As this is
thought to be the same for all ridges the same
amount of magma will be produced and crust of
similar thickness should result. However,changes in
the internal structure of the crust are predicted as
speading rate increases. The interval between succes
sive dyke intrusions decreases so that dykes cool less
between injections. This causes a decrease in the
thickness ofthe sheeted dyke unit,reaching the limit,
at very high spreading rates, when lava continuously
extrudes without dyke intrusion. The resulting crust
would then consist of pillow lavas directly overlying
homogeneous dolerite grading down into gabbro.
Also, as each dyke cools less before subsequent intru
sion, there is a greater tendency for dykes to intrude
the line of existing dykes, thus decreasing the width
ofthe zone ofintrusion. Because ofthe persistence of
the magma chamber and the high thermal gradients
immediately off-axis, as spreading rate increases
stresses are concentrated in a thinner layer and frac
turing will occur more frequently so that the widths
of individual dykes become less with increased
spreading rate.
The lower crust is probably less affected by increas
ing spreading rate, although the width ofthe magma
chamber would be expected to increase. Oceanic
crustal structure evolving from processes in the
magma chamber would not be expected to differ
greatly.
There is a critical spreading (half separation) rate
of about 5 mm a~^ below which a spreading centre
cannot exist. This is because lateral heat conduction
away from the axial zone is then sufficient to freeze
dykes as they are intruded. Similarly, permanent
magma chambers could not exist because the rate of
replenishment of heat by replenishment of the
magma chamber is insufficient to compensate for
losses by conduction.
The problem of axial structure can be approached
in another way by thermal modelling of the cooling
magma chamber (Sleep, 1975; Kusznir & Bott,
1976). Even though these models took no account
of cooling of the top 2km by hydrothermal circula
tion of sea water, the calculations confirm that
magma chambers can only exist at spreading rates in
excess of5 mm a"^. In contrast to the observations of
Detrick et aL (1987),they predict that the roofofthe
chamber initially dips steeply beneath the axial zone
and thence the slope of the roof decreases laterally
away from the axis. Crystal settling at the base ofthe
chamber narrows it and flattens the floor so that the
maximum horizontal dimension occurs at the junc
tion between the isotropic and layered gabbros.
There is very good agreement between the petro-
logical and thermal models of oceanic crust forma
tion and the lithological succession within ophiolite
units (Section 2.5). All require the presence of a
magma chamber as the necessary prerequisite of
oceanic crust formation. The only major problem,
however, is that the shape of the magma chamber
predicted by modelling does not seem to conform
with the form determined by the most recent seismic
reflection surveys (Detrick et 1987). Instead of
the teardrop shape required by modelling, a mush
room shape appears to explain most closely the
reflection results. Resolution of this apparent anom
aly must await further investigation.
6.11 Propagating rifts
The direction ofspreading at an ocean ridge does not
always remain constant over long periods of time,
but may undergo several small changes. Menard &
Atwater (1968) proposed that the northeastern
Pacific had changed direction five times on the basis
of changes in the orientation of major transform
faults (Section 5.8) and magnetic anomaly patterns.
Small changes in spreading direction have also
been proposed as an explanation of the anomalous
topography associated with oceanic fracture zones
(Section 7.3).
Menard & Atwater (1968) made the assumption
that the reorientation of a ridge would take place by
smooth, continuous rotations of individual ridge
segments until they became orthogonal to the new
spreading direction (Fig. 6.16a). The ridge would
then lie at an angle to the original magnetic anomaly
pattern. Long portions of ridges affected in this way
might be expected to devolve into shorter lengths.

(a)
New spreading
direction
Old spreading
direction
(b)
Fig.6.16 (a) Ridge rotation model ofspreading centre
adjustment;(b)evolution ofa stepped ridge following
rotation (after Yicyetal.^ Journal of Geophysical
Research 93,with permission from the American
Geophysical Union).
facilitating ridge rotation and creating new trans
form faults (Fig. 6.16b). The change in spreading
direction is thus envisaged as a gradual, continuous
rotation which produces a fan-like pattern of mag
netic anomalies which vary in width according to
position.
An alternative model of changes in spreading
direction envisages the creation of a new spreading
centre and its subsequent growth at the expense of
the old ridge. This mechanism has been termed the
propa£aUn£f rift modsX(Hey,1977;Htyetal,^ 1980).
Thus the old,'doomed', rift is progressively replaced
by a propagating spreading centre orthogonal to the
new spreading direction (Fig. 6.17a). Kleinrock &
Hey (1989) have described the complex processes
OCEAN RIDGES 119
Pseudofault
Propagating
rift^
Failed \ Pseudofault
Doomed rift
New spreading
direction
Old spreading
direction
(b)
Fig.6.17 (a) Ridge adjustment by rift propagation;
(b)evolution of a stepped ridge following propagation
(after Hey et al.^ Journal of Geophysical Research 93,
with permission from the American Geophysical Union).
which take place at the tip of the propagating rift.
The boundaries between lithosphere formed at old
and new ridges are termed pseudofaults. Pseudo-
faults define a characteristic V-shaped wake pointing
in the direction of propagation. Between the propa
gating and failing rifts, lithosphere is progressively
transferred from one plate to the other, giving rise to
a sheared zone with a quite distinctive fabric. There
fore, abrupt changes in both the topographic and
magnetic fabric of the sea floor occur at the pseudo-
faults and failed rift, and the new ridge propagates by
the disruption oflithosphere formed by symmetrical
accretion at the old ridge. Figure 6.17b shows a
possible way in which the propagating model could
120 CHAPTERS

give rise to evenly spaced fracture zones. These new

fracture zones are bounded by pseudofauits and/or
failed rifts, because the fracture zones do not form
until propagation is completed. They thus contrast
with the ridge rotation model (Fig.6.16b) in which
the fracture zones are areas of highly asymmetric sea
floor spreading, and which does not produce failed
rifts. The propagation model predicts abrupt bound
aries between areas of uniform magnetic anomaly
and bathymetric trends of different orientation. The
rotation model predicts a continuous fanlike con
figuration of magnetic anomalies whose direction
changes from the old to new spreading direction.
Consequently, detailed bathymetric and magnetic
surveys should be able to distinguish between the (a)

two models.
4°N +
Hey et aL (1988) have reported the results of a
detailed investigation of the region where the direc
tion of spreading of the Pacific-Farallon boundary
changed direction at about 54 Ma,just north of the
major bend of the Surveyor Fracture Zone, using
side-scan sonar, magnetometry and seismic reflec
tion. They found that the change in direction ofsea
floor fabric revealed by sonar is abrupt,in accord with
the propagating rift model. Similar conclusions were
also reached by Caress (1988). Yi&yetal.(1980)
have described the results of a survey of an area west
of the Galapagos Islands at 96 ®W. They concluded
that here a new ridge is progressively breaking
through the Cocos plate, and the magnetic data in
particular (Fig.6.18) provide convincing evidence
that the ridge propagation mechanism is operative.
The results were confirmed in a later work(Hey et al. (b)
1986). Brozena & White(1990)have reported ridge
propagation from the South Atlantic, so this phe r+ + +
96° W 95° 94°
nomenon appears to be independent of spreading
rate. Fig.6.18 (a) Predicted magnetic lineation pattern
Engeln et aL (1988) have pointed out that the resulting from ridge propagation;(b) observed magnetic
growth of the region of overlap between old and anomalies near 96°W west ofthe Galapagos Islands
propagating rifts must take a finite amount of time (redrawn from Hey et al. 19S0^ Journal ofGeophysical
because propagation is not infinitely fast and the new Research 85, with permission from the American
ridge does not reach its full spreading rate instan Geophysical Union).
taneously. At small scales of a few kilometres, this
can lead to the formation of overlapping spreading Easter and Juan Fernandez microplates on the East
centres, such as have been described by MacDonald Pacific Rise (Searle etal.., 1989).
et al, (1987). At much larger scales, and during The cause of the initiation of ridge propagation is
propagation over millions of years, the mechanism unknown. It is probable that it is quite a common
can give rise to the formation of microplates several mechanism by which spreading centres adjust to
hundred kilometres in diameter between the two changing directions of the forces driving plate
rifts. Examples ofsuch features are believed to be the motions (Hey^f a^/., 1980). An important corollary

of this is that the ridge-push force at spreading
centres (Section 11.6)is not a primary driving mech
anism as it appears to be quite easily overridden
during ridge propagation.
Further reading
Atwater, T.(1981) Propagating rifts in seafloor spreading
patterns. Nature 290,185-6.
Cann, J.R. (1974) A model for oceanic crustal structure
developed. Geophys.J. Roy. astr. Sac. 39,169-87.
Fowler, C.M.R. (1976) Crustal structure of the mid-
Atlantic ridge crest at 37®N. Geophys.J. Roy. astr. Sac. 47,
459-91.
Hey, R-N., Menard, H.W., Atwater, T.M. & Caress, D.W.
(1988) Changes in direction of seafloor spreading re
visited. Ros. 93,2803-11.
Keen, C. & Tramontini, C. (1970) A seismic refraction
survey on the mid-Adantic ridge. Geophys. J. Roy. astr.
Sac. 20,473-91.
Kusznir, N.J. & Bott, M.H.P.(1976) A thermal study of
the formation of oceanic crust. Geophys. J. Roy. astr. Soc.
47,83-95.
Lister, C.R.B.(1980) Heat flow and hydrothermal circula
tion. Rsv. Earth planet. Set. 8,95-117.
MacDonald, KC.(1982) Mid-ocean ridges: fine scale tec
tonic, volcanic and hydrothermal processes within a plate
boundary zone.A»». Rop. Earth planet. Sci. 10,155-90.
OCEAN RIDGES 121
MacDonald, K.C., Sempere, J.-C., Fox, P.J. & Tyce, R.
(1987) Tectonic evolution of ridge-axis discontinuities
by the meeting, linking or self-decapitation of neighbor
ing ridge segments. Geology 15,993-7.
MacDonald, K.C. et al. (1988) A new view of the mid-
ocean ridge from the behaviour ofridge-axis discontinu
ities. Nature 335, 217-25.
Open University Course Team (1989) The Ocean Basins:
their Structure and Evolution. Pergamon Press, Oxford.
Parsons, B. & McKenzie, D.P. (1978) Mantle convection
and the thermal structure ofthe pldXts.].geophys. Res. 83,
4485-96.
Parsons, B. & Sclater, J.G. (1977) An analysis of the
variation of ocean floor bathymetry and heat flow with
2LQC.J.ge<^hys. Res. 82,803-27.
Sclater, J.G. & Francheteau, J.(1970) The implications of
terrestrial heat flow observations on current tectonic and
geochemical models ofthe crust and upper mantle ofthe
Earth. Geophys. J. Roy. astr. Soc. 20, 509-42.
Smith,D.K. & Cann,J.R.(1993)Building the crust at the
Mid-Atlantic Ridge. Nature 365,707-15.
Talwani, M., Le Pichon, X. & Ewing, M.(1965) Crustal
structure of the mid-ocean ridges 2. Computed model
from gravity and seismic refraction data./. Res.
70,341-52.
Wyllie,P.J.(1988) Magma genesis, plate tectonics and the
chemical differentiation of the Earth. Rev. geophys. 26,
370-404.
7/ Transform and transcurrent faults
7.1 Introduction
Transform faults constitute conservative plate
boundaries, where adjacent plates are in tangential
contact and are affected by little or no subduction or
accretion.
The existence of large lateral relative movements
of the lithosphere was first suggested from marine
magnetic anomalies in the northeastern Pacific
(Fig. 4.1), which were found to be offset along frac
ture zones. Combined left lateral offsets along the
Mendocino and Pioneer faults amount to 1450 km,
while the right lateral offset across the Murray Fault
is 600 km in the west and only 150 km in the east
(Vacquier, 1965).
However, in interpreting these fracture zones as
large scale strike-slip faults, a major problem arises in
that there is no obvious way in which the faults
terminate, as it is certain that they do not circumnav
igate the Earth to join up with themselves. An answer
to this question was suggested by Wilson (1965),
who proposed that the faults terminate at the ends of
other plate boundaries^ which they commonly meet
at right angles. Wilson termed this new class offaults
transform faults^ as the lateral displacement across
the fault is taken up by transforming it into either the
formation of new lithosphere at a terminated ocean
ridge segment or lithosphere subduction at a trench.
Figure 7.1 shows the plan view of an ocean ridge
crest which has been displaced by transcurrent and
Dextral Sinistral
(a) Transform fault (b) Transcurrent fault
122
transform faulting. The transcurrent fault (Fig. 7.1b)
causes a sinistral offset along a vertical plane which
must stretch to infinity beyond the ridge crests. The
transform fault (Fig. 7.1a), however, is only active
between the offset ridge crests, and the relative move
ment ofthe lithosphere on either side ofit is dextral.
Transform faults differ from other types of fault in
that they imply, indeed derive from,the fact that the
area ofthe faulted medium,in this case lithosphere is
not conserved at plate boundaries.
Wilson (1965) defined six classes of transform
fault which depend upon the types of plate margins
they join (Fig. 7.2). These may be an ocean ridge,the
overriding plate at a trench or the underthrusting
plate at a trench. Figure 7.2a shows the six possible
kinds of dextral transform fault; a further six based
on sinistral movement are also possible. Figure 7.2b
shows how the transform faults would develop with
time. Cases (i) and (v) will remain unchanged, cases
(ii) and (iv) will grow, and cases (iii) and (vi) will
diminish in length with the passage oftime.
Sykes (1967) determined focal mechanism solu
tions for earthquakes occurring in the vicinity of the
Mid-Atlantic Ridge at equatorial latitudes. Events
along the ridge axis are consistent with normal fault
ing along north-south planes. Events along the frac
ture zones are much more common and the energy
release is about a hundred times greater than along
the ridge crest. Between the offset ridge segments
events are of strike-slip type with one nodal plane
Fig.7.1 Comparison oftransform
and transcurrent faults.
 TRANSFORM AND TRANSCURRENT FAULTS 123

(i) (ii) (iii)

uid
dfu

(iv) (V) (vi)

3'
if
(a)

(i) (ii) (iii)

I , I 11
Jc ..rM"

(iv) (V) (vi)

d/u

f uJT
(b)

Mid-ocean ridge or rift -Transform fault d fu

V mountain arc
d = down; u = up

Fig.7.2 (a) Six possible types of dextral transform fault:
(i) ridge to ridge;(ii) ridge to concave arc: (iii) ridge to
convex arc;(iv) concave arc to concave arc;(v) concave
arc to convex arc;(vi) convex arc to convex arc.
(b) Appearance ofthe dextral transform faults after a
period oftime (redrawn from Wilson, 1965, with
permission from Nature 207,334-47. Copyright(© 1965
Macmillan Magazines Ltd).
consistent with dextral transform motion. Events
along the fracture zone beyond the ridge extremities
are rare, confirming the transform mechanism of
faulting.
7.2 Origin of transform faults
Before the recognition of transform faulting, the
parallel fracture zones which appear to displace the
crest of the Mid-Atlantic Ridge in equatorial lati
tudes between Africa and South America were be
lieved to represent sinistral transcurrent faults which
displaced an originally straight crest. Wilson (1965),
however, interpreted them as dextral transform
faults. The evolution of these faults is intimately
related to the rifting event which preceded the open
ing of the Atlantic Ocean. Figure 7.3a shows an
idealized continent containing two lines of weak
ness. When the continent is subjected to tension
approximately parallel to these lines, an incipient
fracture develops which is largely perpendicular to
the direction of tension, but, in the vicinity of the
lines ofweakness,develops preferentially along these
old lines (Fig. 7.3b). After rifting (Fig. 7.3c) the
mid-ocean ridge crest maintains the shape of the
124 CHAPTER?
"^Lines of old
weakness
(a)
(b)
New rift and seismiczone
(c)
Seismic zone and rift
initial rift, and the portions parallel to the lines of
weakness develop into transform faults which are
only active between the offset ridge crests. These
transform fault locations are consequently inherited
from the initial rift and act in such a way that the
mid-ocean ridge can maintain a median position
within the expanding ocean.
An example of transform faulting is shown in
Fig. 7.4, which shows how the northern part of the
Mid-Atlantic Ridge is terminated. In Early Tertiary
times the Mid-Atlantic Ridge bifurcated to the south
of Greenland. The western branch, which is now
inactive, passed through Baffin Bay and terminated
against the Wegener Fault,an extinct,sinistral ridge-
ridge transform fault. The active eastern branch
passes through Iceland, and terminates southwest of
Spitsbergen at the De Geer Fault. This dextral ridge-
ridge transform fault connects to the northernmost
part of the Mid-Atlantic Ridge in the Arctic. This is
itself transformed into the Verkhoyansk Mountains
Fig.7.3 Three stages in the rifting of a
continent(redrawn from Wilson, 1965, with
permission from Nature 207, 334-47.
Copyright © 1965 Macmillian Magazines
Ltd).
of Siberia by rotation about a fulcrum in the New
Siberian Islands (Fig. 7.4 inset).
The different offsets observed across the Murray
Fracture Zone from magnetic lineations are thought
to be due to a change in location ofthe ridge crest to
the south ofthe fracture zone about 40 Ma ago. The
change in offset ofanomalies ofthe same age implies
a 'ridge jump' of approximately 500 km to the east
(Harrison & Sclater, 1972). Similar but better docu
mented ridge jumps, which also greatly reduce the
offset of crust of the same age on either side of a
fracture zone, occur in the extreme south of the
Atlantic Ocean (Barker, 1979). Here,to the south of
the Falkland-Agulhas Fracture 2k)ne, the ridge crest
has jumped westward on three occasions since the
opening of the South Atlantic, that is at 98,63 and
59 Ma. In so doing it has reduced the original offset
of 1400 km to approximately 200 km. Other ridge
jumps, producing major changes in ridge crest
geometry within the past 10 Ma, have occurred to
 TRANSFORM AND TRANSCURRENT FAULTS
%
•'A'S
Canadian Basin
90°W
Siberian
t: Basin
-
Wegener /w:-
De Geer
Fault
Fig.7.4 Northern termination of the Mid-Atlantic Ridge
(redrawn from Wilson, 1965, with permission from
Nature 207, 334-47. Copyright © 1965 Macmillan
Magazines Ltd).
the north of Iceland (Vogt et 1970) and along
the crest of the East Pacific Rise in the east central
Pacific (Herron, 1972).
The proposal that transform fault locations are
inherited from ancient lines of weakness is strongly
supported by considering the continuation of trans
form faults offsetting the Red Sea Ridge into the
adjacent continents (Garson & Krs, 1976). The
faults appear to correlate well with ancient fractures
present in the flanking Precambrian Arabian and
Nubian shields. The geometry ofthe Red Sea has also
been controlled by northerly trending Precambrian
structures(Dixon At/., 1987).
Transform faults occur in both oceanic and con
tinental lithosphere. In the oceans their trace is well
defined, in the absence of sedimentary cover, by
125
fracture zones, but on land their recognition in detail
is considerably more complex. This difference is
partly a function of the proximity of the fault to
observation, as oceanic faults are commonly studied
from the sea surface several kilometres above the
ocean floor. Direct observations ofa fracture zone on
the Mid-Atlantic Ridge (Choukroune et 1978)
have shown, however, that it consists of a complex
swarm of faults occupying a zone 300-1000 m in
width, and Searle (1983) suggests that these multi-
fault zones are wider and more common on fast
spreading ridges such as the East Pacific Rise. More
importantly, however, their differences reflect the
contrasting natures of the oceanic and continental
crusts in their response to predominantly strike-slip
shear failure.
7.3 Oceanic fracture zones
Transform faults in the oceans are marked by fracture
zones, which are long, linear, bathymetric depres
sions which normally follow arcs of small circles on
126 CHAPTER?
the Earth's surface perpendicular to the offset ridge
(Bonatti & Crane, 1984).
Fracture zones mark both the active transform
segment and its fossilized trace. It has been suggested
(Collette, 1979) that the fractures result from ther
mal contraction in the direction ofthe ridge axis. The
internal stresses caused by contraction are much
larger than the breaking strength ofthe rocks, and it
is possible that fracture zones develop along the
resulting lines of weakness.
Dredging of fracture zones has recovered both
normal oceanic crustal rocks and rocks which show
much greater metamorphism and shearing. Very
commonly large blocks ofserpentinite lie at the bases
ofthe fracture zones.
Bonatti & Honnorez(1976)and Fox etal.(1976)
have examined specimens recovered from the thick
crustal sections exposed in the large equatorial Atlan
tic fracture zones, which were found to consist of
ultramafic, gabbroic and basaltic rock types and their
metamorphosed and tectonized equivalents. Serpen
tinite intrusion appears to be quite common within
fracture zones, accompanied by alkali basalt volcan-
ism,hydrothermal activity and metallogenesis. Inves
tigations ofthe Vema Fraction Zone(Auzende
1989) have indicated a sequence similar to normal
oceanic layering. The islands of St Peter-Paul in the
equatorial Atlantic, which lie on a ridge associated
with the St Paul Fracture 2k)ne, are composed of
mantle peridotite.
In the North Atlantic, fracture zone crust is very
heterogeneous in thickness and internal structure
(Derrick et 1993). It is often thin(< 2km) with
low seismic velocities, and layer 3 is absent. The
crustal thinning may extend several tens of kilo
metres from the fracture zone. Geologically, this
structure may represent a thin, intensely fractured
and hydrothermally altered basaltic layer underlain
by serpentinized ultramafic rocks. The apparent
thickness variations may then reflect different extents
ofserpentinization. The thin mafic crust is possibly a
result of reduced magma supply at ridge oflTsets.
Work by Cannat et til. (1995) has suggested that
serpentinized rocks may be much more common in
the oceanic crust than previously expected, even in
areas distant from fracture zones. They dredged in
the region of the North Atlantic Ridge at 22-24®N
over areas of positive gravity anomalies, which are
probably indicative of relatively thin crust, and over
areas with a normal gravity field. Over the former.
which comprised some 23% of the area surveyed,
they encountered serpentinite with very few of the
basaltic rocks which normally characterize oceanic
layer 2(Section 2.4.6). The hypothesis proposed to
explain this extensive serpentinite follows from the
segmentation of ocean ridges and the occurrence of
magma generation only at certain locations on
slow spreading ridges (Section 6.5). The oceanic
crust formed at magmatic centres would be of the
normal,layered type (Section 2.4.4). Sections distant
from the centres would, however, receive an insuffi
cient magma supply to produce basaltic rocks and
upper mantle-derived peridotitic material would be
brought to the surface, where it is altered by sea water
to serpentinite. As magmatic centres grow, migrate
along the ridge axis and decline, the normal oceanic
crust would similarly migrate and would enclose
those regions of serpentinitic crust which originate
where magma was absent. This work is important
as it implies that serpentinized peridotite is more
common in slow spreading oceans than previously
recognized. There are wide ranging implications.
Peridotite is much more reactive with sea water than
basalt and on weathering would release magnesium,
nickel,chromium and noble metals. Sepentinite also
contains far more water than altered basalt, which
would account for much ofthe water supplied to the
mantle in subduction zones (Section 8.8).
Ocean fracture zones must bring oceanic crust of
diflferent ages into juxtaposition. The depth ofthe sea
floor is dependent upon the square root of the age
(Section 6.9), and so it would be expected that a
scarp would develop across the fracture zone from
the younger, higher crust to the lower, older crust
(Menard & Atwater, 1969; DeLong et til.., 1977)
(Figs 7.5, 7.6b). The rate of subsidence of oceanic
lithosphere is inversely dependent upon the square
root of its age (DeLong et at.., 1977), so the higher,
younger crust subsides more rapidly than the lower,
older side. The combination of contraction in the
vertical plane and horizontally perpendicular to the
direction of the ridge axis would result in a small
component of dip-slip motion along the fracture
zone away from the active transform fault. DeLong et
at.(1977) have suggested that this small amount of
dip-slip motion could give rise to fracture zone
seismicity and deformation of rocks within the floor
and walls.
Transverse ridges are often found in association
with major fracture zones and can provide vertical
 TRANSFORM AND TRANSCURRENT FAULTS 127

Fig. 7.5 Differential topography resulting from transform

faulting of a ridge axis.

reliefofover 6 km. These run parallel to the fractures

(Bonatti, 1978) on one or both margins. They are
frequently anomalous in that their elevation may be
greater than that of the crest of the spreading ridge
(Fig. 7.6c, d). Consequently,the age-depth relation
ship of normal oceanic lithosphere (Section 6.9)
does not apply and depths differ from 'normal' crust
of the same age. The ridges do not originate from
volcanic activity within the fracture zone, nor by
hotspot activity (Section 5.4), but appear to result
from the tectonic uplift of blocks of crust and upper
mantle. Transverse ridges therefore cannot be ex
plained by normal processes oflithosphere accretion.
Bonatti (1978) considers that the most reasonable
mechanism for this uplift is compressional and ten-
sional horizontal stresses across the fracture zone
which originate from small changes in the direction
of spreading, so that transform movement is no Fig. 7.6 Different types of basement morphology across
longer exactly orthogonal to the ridge. Several small fracture zones (redrawn from Bonatti, 1978, with
changes in spreading direction can give rise to epi permission from Elsevier).
sodic compression and extension affecting different
parts of the fracture zone. This has caused, for
example, the emergence of parts of transverse ridges distribution of strong and weak portions of fracture
as islands,such as St Peter-Paul, and their subsequent zones.

subsidence (Bonatti 8c Crane, 1984).
Lowrie et al. (1986) have noted that, in some
fracture zones, the scarp height may be preserved
even after 100 Ma. They accept that some parts
of fracture zones are weak, characterized by active
volcanism, and maintain the theoretical depths pre
dicted for cooling lithosphere. Other parts, however
appear to be welded together and lock in their initial
differential bathymetry. The differential cooling
stresses would then cause flexure of the lithosphere
on both sides of the fracture zone. Future work
will reveal if there is any systematic pattern in the
There are certain oceanic transform faults in which
the direction of the fault plane does not correspond
exactly to the direction ofspreading on either side so
that there is a component of extension across the
fault. When this occurs the fault may adjust its
trajectory so as to become approximately parallel to
the spreading direction by devolving into a series of
fault segments joined by small lengths of spreading
centre (Fig. 7.7a). A fault system in which new crust
originates is termed a leaky transform fault(Thomp
son & Melson, 1972; Taylor et al., 1994). A leaky
transform fault may develop where the location of
128 CHAPTER?
(a)
Fig.7.7 (a) Development of a leaky transform fault
because of a change in the pole ofrotation,(b) Leaky
transform fault in the Gulfof California.
faulting is determined by some existing crustal weak
ness. An example of this type is the Gulf of Califor
nia, where the transform fault system developed is
controlled by the presence of a small tract ofoceanic
crust between the continental crust ofBaja California
and Mexico (Fig. 7.7b). An alternative mechanism
for leaky transform fault development occurs when
there is a small shift in the position of the pole of
rotation about which the fault describes a small
circle. The fault would then adjust to the new small
circle direction by becoming leaky. An example of
this type is the Cayman Trough, a leaky transform
fault which defines the western part of the northern
boundary of the Caribbean Plate with the North
American Plate.
1 1
United States
Mexico
Pacific Ocean
7.4 Continental strike-slip faults
7.4.1 Introduction
In strike-slip or wrench faults the primary motion is
horizontal along a vertical fault plane. Fundamental
faults penetrate the whole lithosphere. Only trans
form faults mark active plate boundaries.
Examples of transform faults which displace con
tinental lithosphere are the dextral San Andreas
Fault of California, the dextral Alpine Fault of New
Zealand and the sinistral Dead Sea fault system which
connects the Red Sea to the Bitlis Mountains of
Turkey. Other fundamental faults which do not
represent present plate boundaries (Reading, 1980)
include the Great Glen Fault ofScotland, the North
Pyrenean Fault Zone and the Insubric Line of the
Alps. An additional category is represented by those
fundamental faults associated with indentation tec
tonics (Section 9.3.6).
Large strike-slip faults frequently produce distinc
tive topographic features. These may be a primary
 TRANSFORM AND TRANSCURRENT FAULTS
fault scarp, a linear trough resulting from differential
erosion of contrasting juxtaposed lithologies or ero
sion of a fault breccia (Allen, 1981). The strike-slip
movement causes horizontal displacements of sur
face features such as streams, canyons and ridges.
In contrast to oceanic fracture zones, which are
characterized by a relatively simple linear trough,
plate boundaries on continents are considerably
more complex. This undoubtedly reflects the differ
ence in strength and thickness between oceanic and
continental lithosphere, and also the inhomogen-
eous nature of the continental crust, which may
contain many ancient lines of weakness along which
a plate would rupture preferentially. It is thus neces
sary to distinguish between the transform fault, the
principal rupture; the transform fault zone, a set of
approximately parallel faults which splay from and
rejoin the main fault within a strip about 10 km in
width; and the transform fault system, a set of high
angle faults approximately parallel to the main fault,
forming a belt some 500 km wide, and which may
include a subset striking at a significant angle to the
main fault.
The complex nature ofcontinental transform fault
systems is illustrated by the San Andreas fault system
in California (Fig. 7.8). The system developed in
Oligocene times and its location within a continent
may have been controlled by an existing fracture
network (Beck, 1986), as otherwise the more logical
place for the fault to develop would be the locus
of the transition from continental to oceanic litho
sphere at the continental margin. There has been
about 1500 km ofdextral movement along the trans
form fault zone since Oligocene times. The San
Andreas Fault is the only continuous fault in this
zone,and extends for some 1200 km,but only about
300 km of horizontal movement appears to have
occurred along it. The remaining 1200 km move
ment must have occurred along other northwest
faults within the transform zone and in folding ofthe
rocks along it. There also exist major sinistral faults,
such as the Garlock and Big Pine, which strike north
easterly. Much of the present day seismicity occurs
along such oblique faults (Nicholson et«/., 1986a),
which in places define small crustal blocks which are
rotating in a clockwise sense as a result ofthe regional
stress field. Much of the deformation caused by the
transform fault system changes from brittle to duc
tile at 15 km depth, below which seismicity is rare.
Zoback et al, (1987) have presented information
129
on the stress state of the region of the San Andreas
fault system which has led to a new model for
continental strike-slip faulting. Various types of in
situ stress measurements and other geological and
geophysical indicators demonstrate that, east of the
Great Central Valley of California (Fig. 7.8), there is
much strike-slip and normal faulting, but to the west
the deformation is wholly compressional,dominated
by folding, strike-slip and reverse faulting parallel to
the San Andreas, so that the direction of maximum
horizontal compression is orthogonal to the fault
zone. Classical faulting theory cannot explain com
pression at such a high angle to a strike-slip fault as
the Pacific and North American plate motions are
only slightly convergent. Moreover, Zoback et al.
point to a paradox concerning major strike-slip faults
in that heat flow observations show no frictionally
generated heat,so that the fault must slip in response
to very low shear stresses which are 10-20% ofthose
predicted by frictional faulting theory. They suggest
a common explanation for these phenomena in
terms of a model in which the San Andreas is an
extremely weak fault which can locally reorient the
regional stresses (Mount & Suppe, 1987). Far from
the fault, shear stresses are high and contained by the
frictional strength of the crust, but shear stresses on
planes parallel to the'weak'faults ofthe San Andreas
system must be quite low. Consequently, the prin
cipal stresses become reoriented so as to minimize
shear stress on planes parallel to the San Andreas.
This requires a rotation such that the direction of
maximum principal stress becomes nearly either per
pendicular or parallel to the fault. The maximum
horizontal compressive stress becomes nearly or
thogonal to the fault ifthe regional compression is at
an angle to the fault in excess of 45®, as occurs at
present. However if this angle is less than 45®, the
maximum horizontal compression becomes almost
parallel to the fault. The 20® clockwise change in
relative plate motions that occurred at 4-5 Ma would
have been preceded by a 90® change in stress dir
ection, and would explain the fault-parallel basin
development and fault-normal extension prevalent at
that time.
Zoback et al, suggest that the weakness of the San
Andreas fault system extends to a depth approxi
mately equivalent to the whole lithosphere. A pos
sible mechanism for decreasing the strength of the
fault is the presence of abnormally high pressure
fluids in the fault zone which lowers the effective
130 CHAPTER?

120°

Cape i 0 50100150
Menaocino

Ft Delgada
40'

Ft Arena

Fort Ross
Hayward F.
./ Mojave Desert

San Francisco Colorado

Desert
Pilarcitos F.
^
Gabiian Range
%
Santa Cruz Mts.

N -

White Wolf F

Nacimlento F.
35° ■•••
Gabriel

Banning F
^ Angeles'Tv Salton
Sea

Inglewood-:^ ^ ^
■ F. r.: ^
Diego
^
50 100 150
' isAexico 115°

Fig. 7.8 Generalized fault map of coastal and southern
California (redrawn from Crowell, 1962, with permission
from the Geological Society of America).
normal stress across it. However, the pressure could
not exceed that of the minimum horizontal principal
stress or natural hydraulic fracturing would occur,
and so fault slip would not occur. Zoback et al. thus
propose that some other mechanism must be active
in reducing fault strength.
This revolutionary new model of very weak conti
nental strike-slip faults clearly explains the conflict
ing geological data in central California, and would
be verified by direct in situ stress measurements at
depth in the fault zone. It may also be applicable to
other major continental faults elsewhere.
7.4.2 Transpression and transtension
Since ancient fractures may control the locations of
fault planes in many transform fault zones, the strike
 TRANSFORM AND TRANSCURRENT FAULTS
_^;:;^Releasing
hpnri
(a)
(b)
Fig.7,9 Structures developed along continental wrench
faults:(a) at a double bend a pull-apart forms at a releasing
bend and deformation and uplift at a restraining bend;(b)
sharp pull-apart;(c) severe deformation at a sharp
restraining bend causing folding and thrusting (redrawn
from Crowell, 1974b, with permission from the Society of
Economic Palaeontologists & Mineralogists).
of the faults may depart from a simple linear trend
following a small circle on the Earth's surface. The
curvature of the faults imposed by existing litho-
spheric weaknesses gives rise to alternate zones of
convergence and divergence, where the fault shape
causes blocks on either side of the fault to be com
pressed or extended (Fig. 7.9a). An example ofsuch
behaviour is illustrated by the Transverse Ranges of
California (Fig. 7.8), which have developed along a
portion ofthe San Andreas Fault which strikes signi
ficantly more westerly than the general northwest
trend of the fault system. A consequence of the
compression experienced between southwest and
northeast California across this part of the fault is
the formation of the Transverse Ranges by crustal
shortening.
131
'^'/--Restraining
honri
Sharp restraining ben^
Folds and thrusts
y-yy-'y/
(c)
The combination of strike-slip motion and exten
sion (divergent strike-slip) is known as transtension^
and the combination of strike-slip motion and
compression (convergent strike-slip) is known as
transpression. Transtensile regions exhibit normal
faulting, basin extension and volcanicity, whereas
transpressive regions exhibit thrust faulting, folding
and uplift. Gentle strike-slip fault curvature can thus
give rise to both extension and compression. Various
forms of basin develop under extension, while the
elevated areas resulting from compression may pro
vide sediment for the basin filling.
7.4.3 Pull-apart basins
Where the curvature of a strike-slip fault is pro
nounced,or where one fault terminates and sidesteps
to an adjacent, parallel fault, the curved zone or area
separating the ends of the faults is thrown into
tension (Fig. 7.9b)or compression (Fig. 7.9c). Com
pression results in the development of a region ele
vated by crustal shortening and typified by folds and
thrust faulting. Tension gives rise to an extensional
trough known as pull-apart basin.
132 CHAPTER?
Well documented examples of pull-apart basins
include the Dead Sea (Quennell, 1958) and the
Salton Trough of southern California (Crowell,
1974a). The Salton Trough is located in the San
Andreas fault zone to the north of the Gulf of
California and is growing due to continued dextral
motion along two parallel traces of the fault. It
contains about 6000 m of young sediments
(Crowell, 1974b). Pull-apart basins are of economic
importance as they represent excellent targets for
hydrocarbon exploration.
Pull-apart basins progressively grow in the same
direction as the fault movements so that the oldest
sediments occupy the margins only (Fig. 7.10). The
strike-slip margins are initially straight and parallel,
but may sag with time so that continued fault move
ment may slice off segments and provide a complex
braided zone. The pull-apart margins can exhibit any
shape. They are initially identical in form, but extend
and sag during development of the basin so that in
time their similarity is lost.
As the pull-apart basement grows, its floor
stretches and becomes attenuated, so that eventually
igneous material may be emplaced in the centre of
the basin. The presence of basic igneous material is
suggested by the presence of positive gravity anoma
lies and the location of geothermal areas within
basins such as the Salton Trough. However, many
basins,ofthe order of50-100 km in length,have not
attenuated to the point of lithospheric rupture, and
contain no igneous rocks. Calculations on several
basins indicate that only about 5-10% extension is
required to produce the observed lithospheric attenu
Time
Intrusives
ation, so only relatively small extensions are needed
to initiate the formation of pull-apart basins. A
sketch of an idealized pull-apart basin is presented in
Fig. 7.11, although it is unlikely that any one basin
would exhibit all the features shown.
7.4.4 Fault wedge basins
Strike-slip faults may diverge and converge to form
an anastomosing pattern. Duplexes (Fig. 7.12) may
form at bends (Woodcock & Fischer, 1986), analo
gous to the duplexes formed on the ramps ofdip-slip
faults but differing in that vertical movements are not
constrained at the upper (ground) surface. In this
environment, pull-apart basins are frequently associ
ated with lens-shaped basins and high-standing
ridges and banks ofsimilar shape. These features owe
their origin to the stress regime resulting from the
confluence of two sub-parallel strike-slip faults. The
individual blocks defined by the strike-slip faults are
compressed and uplifted when the faults converge
and stretched and downthrown where the faults
diverge (Fig. 7.13).
In an anastomosing pattern of sub-parallel strike-
slip faults, those most closely following the direction
of plate movements predominate, grow longer and
assume near vertical dips. Other faults at an angle to
the overall direction of movement may then rotate
farther out of alignment and develop dips signifi
cantly less than vertical, so that the fault involves a
component of dip-slip motion. If the fault's curva
ture carries it to a region of extension, a normal
oblique-slip fault develops;ifto a region ofcompres-
Fig.7.10 Stages in the formation of a pull-
apart basin.
 TRANSFORM AND TRANSCURRENT FAULTS 133

Dextral fault continues to northwest

Irregular basin margin

Detachment faults
Irregular normal-slip faults

Stretched and attenuated

marginal floor
Straight basin
margin
Complex unconformities X/
and overlaps in
Volcanic floor
subsurface

Complex
Pull-apart basin
intersection

Complex corner
Oldest intact basin

Remnants of marginal
rocks within volcanics
Oblique-slip
faults
Talus breccias and rapid facies
changes basinward

Small thrust plates'

Slide blocks

Braided right-slip zone

Fig.7.11 Sketch map of an
Irregular basin margin -folds
idealized pull-apart basin (redrawn result of convergence between
from Crowell, 1974b, with boundary dextral faults
permission from the Society of Dextral fault continues to
Economic Palaeontologists & southeast
Mineralogists).

sion, a reversed oblique-slip fault forms. In braided
fault systems of this type, the faults may converge at
depth to produce a flower structure (see Fig. 7.14)
(Harding, 1985).
7.4.5 Deep structure of transform zones
Important data on the deep structure of continental
strike-slip faults have been provided by teleseismic
studies of the crust and upper mantle in southern
California by Hadley & Kanamori(1977). They have
detected a high velocity body beneath the Transverse
Ranges which extends from a depth of40 km down
to about 100 km. This body extends 150 km east of
the San Andreas Fault before terminating, and does
not appear to be oflfset by the fault. The San Andreas
fault system only appears to affect the crust, which is
30-35 km thick, and about 5-10 km of the upper
mantle. The lower lithosphere is not displaced. One
possible interpretation ofthese data is that the upper
lithosphere is decoupled along a horizontal plane,
and has reached its present position by westerly
shearing. The plate boundary at depths in excess of
40 km is, then, situated farther to the east than at the
surface (Fig. 7.15). This idea is supported by earth
quake data from southern California (Nicholson et
fit/., 1986b), which suggest that much deformation
takes the form oftectonic rotations ofrelatively small
blocks. If this is the case, a detachment surface is
required at depth.
A lack of coherence between the upper crust and
mantle in the region of strike-slip faults has been
catalogued by Molnar (1988), who notes that the
deformation of near-surface rocks can be of such
134 CHAPTER?
Releasing bend Restraining bend
Separation
Straight
Leading extensional imbricate fan
Extensional duplex
Contractional duplex
Leading contractional imbricate fan
Fig.7.12 Map view ofidealized dextrai strike-slip systems
(redrawn from Woodcock & Fischer, 19S6,Journal of
Structural Geolqgy^ with permission from Pergamon Press,
Ltd).
magnitude as to cause the amount of horizontal
displacement to vary along the fault. Ifthis deforma
tion affected the whole crust, variations in crustal
thickness would be expected. These have not been
noted by deep seismic reflection surveys, implying
that,in at least some fault zones, the fault penetrates
no deeper than 8-10 km. The upper crust thus
appears to be detached from the lower continental
iithosphere, and to act as a suite of small blocks
separated by minor faults. This suggests that the
crustal blocks are driven by flow in the underlying
lower crust and mantle. Thus,some portions of the
Divergence
Down
Convergence
Releasing offset Restraining offset
Overlap Straight
Trailing extensional imbricate fan
Trailing contractional imbricate fan
San Andreas Fault do not continue down into the
underlying upper mantle. Elsewhere, however,
Lemiszki & Brown (1988) have shown that the San
Andreas Fault appears to pass directly into the mantle
as a narrow vertical feature. Indeed, Lemiszki &
Brown suggest, from available deep reflection data,
that past or present plate boundary transform faults
in continents are characterized by a near vertical fault
zone penetrating the whole crust, while intraplate
wrench faults sole out at depth along a subhorizontal
detachment, allowing the rotation and deformation
of blocks of upper crust.
The possibility of the occurrence of decoupling
within the continental Iithosphere along horizontal
shear planes is clearly most important as it suggests
that, in addition to transpression and transtension,
the upper brittle Iithosphere can respond to distrib
uted shear in the more ductile Iithosphere at depth.
Fig.7.13 Subsidence patterns of
curved and anastomosing strike-slip
faults (redrawn from Reading,
1980, using data from Quennell,
1958, with permission from the
Geological Society, Kingma, 1958,
with permission from the Science
Information Publishing Centre and
Crowell, 1974a, with permission
from the Society of Economic
Palaeontologists & Mineralogists).
 TRANSFORM AND TRANSCURRENT FAULTS
Fig.7.14 (a) Negative flower structure formed in an
extensional strike-slip duplex,(b) Positive flower structure
formed in a contractional duplex (redrawn from
Woodcock & Fischer, 1986,in Journal ofStructural
Geolqgy, with permission from Pergamon Journals Ltd).
The results ofsimilar studies ofother strike-slip faults
will determine whether this behaviour is typical of
such tectonic environments.
7.5 Sheared continental margins
Where a transform fault develops during continental
rifting, as is shown in Fig. 7.3, the continental mar
gin is defined by the transform fault and is termed a
135
San Andreas Fault
Upper llthosphere
Fig.7.15 Block diagram illustrating the possible
subsurface displacement ofthe San Andreas Fault
(redrawn from Hadley & Kanamori, 1977, with
permission from the Geological Society of America).
sheared or transform continental marj/in. The history
of such a margin, first considered by Scrutton
(1979), is likely to be complex as the margin is
initially in contact with its continental counterpart
on the adjacent plate and subsequently with oceanic
lithosphere and an ocean ridge as the separation
proceeds.
One of the best-studied sheared margins is the
Ivory Coast-Ghana margin in the north of the Gulf
of Guinea, which formed during the opening of the
Atlantic at about 130-125 Ma, during which there
was transform motion in this region (Mascle &
Blarez, 1987). The margin has undergone little sub
sequent modification and so can be considered to
represent a fossil sheared margin. On the basis of
dredge sampling and seismic reflection profiles corre
lated with onshore boreholes, it has been possible to
reconstruct the evolution of this sheared margin.
There appear to have been four main phases in this
evolution, which are diagrammatically illustrated in
Fig. 7.16. In phase 1 (Fig. 7.16a) there is contact
between the thick crust of the two continents. The
strike-slip motion probably results in brittle deforma
tion of the upper crust and ductile deformation at
depth (Section 2.10). The former can give rise to
pull-apart basins (Section 7.4.3) and structurally ro
tated crustal blocks. In phase 2(Fig. 7.16b), as diver
gent margins are formed, the contact is between a
continent and the thinner, stretched crust of a conti
nental margin. This newly-created rifted basin would
experience rapid sedimentation from the adjacent
continent and rapid subsidence associated with the
136 CHAPTER?

Divergence
Transform motion between
oontinental crust
(a) \
Transform motion between
oceanic crust

0 Thick continental crust

Thinned continental crust

Oceanic crust

Ocean ridge axis

Marginal ridge with thinned
crustai blocks and deformed
sediment

Fig. 7.16 Simplified model of the formation of a sheared
continental margin,(a) Continental crust in transform
contact,(b) Continent and continental margin in
transform contact,(c) Progressive drift of the continental
margin past hot ocean crust and the ocean ridge,(d)
Formation of an oceanic fracture zone as separation is
completed (redrawn from Mascle & Blarez, 1987).
crustai thinning (Section 10.3.5). The sediments are
folded and faulted by the transform motion and a
marginal ridge is formed. This tectonism is also
recorded in unconformities in the sedimentary se
quence. In phase 3 (Fig. 7.16c) there is contact
between continental margins and oceanic crust. Dur
ing this phase the sheared margin passes adjacent to
the hot oceanic crust of the spreading centre and the
thermal exchanges it experiences must affect the
sheared margin. This is probably recorded in a de
crease in the subsidence rate of the margin, as shown
by a succession ofsediment onlaps, and results in the
uplift of the marginal ridges. In phase 4(Fig. 7.16d)
the transform is only active between blocks of
oceanic crust and thus appears as a fracture zone
(Section 7.3). The sheared margin is then in contact
with cooling oceanic lithosphere and its subsidence
should evolve in a manner similar to other rifted
margins.
Further reading
Bonatti, E. (1978) Vertical tectonism in oceanic fracture
zones. Earth planet. Set. Lett. 37, 369-79.
Bonatti, E. 8c Crane, K.(1984) Oceanic fracture zones. Sci.
Am. 250(5), 36-47.
Crowell, J.C. (1974) Origin of late Cenozoic basins in
 TRANSFORM AND TRANSCURRENT FAULTS 137

southern California. In Dickinson, W.R. (ed.) Tectonics
and Sedimentation. Soc. econ. Paleont. Mineral. Spec.
Publ. 22,190-78.
Hadley,D.& Kanamori,H.(1977)Seismic structure ofthe
Transverse Ranges, California. Bull. £feol. Soc. Am. 88,
1469-78.
Reading, H.G. (1980) Characteristics and recognition of
strike-slip fault systems. In Ballance, P.P. & Reading,
H.G. (eds) Sedimentation in Oblique Slip Mobile Zones.
Spec. Publ. int. Ass. Sediment. 4,7-26.
Woodcock, N.H. & Fischer, M. (1986) Strike-slip
duplexes./, struct. Geol. 8,725-35.
Zoback, M.D. et al. (1987) New evidence on the state of
stress of the San Andreas fault system. Science 238,
1105-11.
8/ Subduction zones

8.1 General morphology of oceanic

subduction zones
Island arc systems are formed when oceanic iitho-
sphere is subducted beneath oceanic lithosphere.
They are consequently typical of the margins of
shrinking oceans such as the Pacific, where the major
ity of island arcs are located. They also occur in the
western Atlantic, where the Lesser Antilles (Carib
bean) and Scotia arcs are formed at the eastern
margins ofsmall oceanic plates isolated by transform
faults against the general westward trend of move
ment.

All of the components of island arc systems are

usually convex to the underthrusting ocean. Excep
tions to this occur where lithosphere was originally
consumed beneath a continental margin at an
Andean-type mountain range. The shape ofthe arc is
then imposed by the edge of the continent and may
be retained as the arc migrates from the continent.
The usual convexity of island arcs may be a conse
quence of spherical geometry, as suggested by Frank
(1968). If a flexible spherical shell, such as a table
tennis ball, is indented an angle 0 (Fig. 8.1), the
indentation is a spherical surface with the same
radius as the shell(R). The edge ofthe indentation is
a circle whose radius r is given by r = ViRO, where 0 is
in radians. Ifthis theorem is applied to a plate on the
Earth's surface, 0 represents the angle of under-
thrusting of oceanic lithosphere, which averages
about 45°. The radius ofcurvature ofthe trench and
island arc on the Earth's surface is then about
2500 km. This value is in agreement with some, but
not all, island arc systems. The general convexity of
island arc systems is undoubtedly a consequence of
spherical geometry, and deviations result from the Fig.8.1 Geometry ofan indentation in a sphere of
oversimplification of this approach, in particular the inextensible material (redrawn from Bott, 1982a, with
fact that the conservation of surface area is not permission from Edward Arnold).
required by plate tectonics. Thus, for example, the
angle of underthrusting at the Mariana arc is almost The generalized morphology of an island arc
90®, but it has one of the smallest radii of curvature system is shown in Fig. 8.2, although not all com
(Uyeda & Kanamori, 1979). ponents are present in every system (Uyeda & Kan-
138

Remnant arc Volcanic arc
I Marginal basin V /
Fig. 8.2 Generalized schematic morphology of an oceanic
subduction zone.
amori, 1979). Proceeding from the oceanward side
ofthe system, a bulge about 500 m high occurs about
120-150 km from the trench. The forearc region
comprises the trench itself, the subduction complex
(the 'first arc' or accretionary wedge or prism) and
the forearc basin. The subduction complex is con
structed of thrust slices of trench fill (flysch) sedi
ments and also possibly oceanic crust, which have
been scraped off the downgoing slab by the leading
edge ofthe overriding plate. The contact between the
accretionary wedge and forearc basin is often a region
of backthrusting (Silver & Reed, 1988). The forearc
basin is a region of tranquil, flat-bedded sedimenta
tion between the forearc ridge and island arc. The
island arc ('second arc') is made up of an outer
sedimentary arc and an inner volcanic arc. The sedi
mentary arc comprises coralline and volcaniclastic
sediments underlain by volcanic rocks older than
those found in the volcanic arc. This volcanic sub
strate may represent the initial site of volcanism as
the relatively cool oceanic plate began its descent. As
the 'cold' plate extended further into the astheno-
sphere the position of extrusive igneous activity
moved backwards to its steady state location now
represented by the volcanic arc. The island arc and
remnant arc (back-arc ridge or'third arc'), first recog
nized by Vening Meinesz(1951),enclose a marginal
sea (back-arc basin) behind the island arc. Such mar
ginal seas are generally 200-600 km in width. In
SUBDUCTION ZONES 139
Sedimentary arc Subduction complex Bulge
Forearc basin \ Trencti j
some island arc systems there may be up to three
generations of marginal seas developed on the land
ward side of the island arc.
8.2 Gravity anomalies of
subduction zones
Figure 8.3 shows a free air gravity anomaly profile
across the Aleutian arc which is typical of most
subduction zones. The bulge of the downgoing
lithosphere to seaward of the trench is marked by a
positive gravity anomaly of about 500 g.u. (Talwani
& Watts, 1974). The trench and accretionary prism
are typified by a large negative anomaly of some
2000 g.u. amplitude which results from the displace
ment ofcrustal materials by sea water and sediments.
Conversely, the island arc is marked by a large posi
tive anomaly. Isostatic anomalies over the trench and
arc are large and exhibit the same polarity as the free
air anomalies. These large anomalies result from the
dynamic equilibrium imposed on the system by com
pression, so that the trench is forced down and the
arc held up out of isostatic equilibrium by the forces
driving the plates.
8.3 Structure of island arc systems
from earthquakes
Island arc systems exhibit intense seismic activity. A
large number of events occur on a plane which dips
on average at an angle of about 45° away from the
underthrusting oceanic plate. The plane is known as
140 CHAPTERS
South
2000 r
Lithosphere
Fig.8.3 Gravity anomalies of an oceanic subduction zone
(after Grow, 1973, with permission from the Geological
Society of America).
a BeniofF (or Benioff-Wadati) zone, after its dis-
coverer(s), and earthquakes on it extend from the
surface, at the trench, down to a maximum depth of
about 680 km. Figure 8.4 illustrates how shallow,
intermediate and deep focus earthquakes in the
southwestern Pacific occur at progressively greater
distances away from the site of underthrusting at the
Tonga Trench. Figure 8.5 shows a section through
the Tonga-Kermadec island arc system with earth
quake foci projected on a vertical plane parallel to the
direction of underthrusting. Further information on
the nature of the BeniofF zone was obtained from a
study of the body wave amplitudes from deep earth
quakes (Fig. 8.6). Seismic arrivals at the volcanic
islands ofthe arc, such as Tonga, were found to be of
far greater amplitude than those recorded to the
front or rear of the arc at stations such as Raratonga
and Fiji. The differences in amplitude are usually
described quantitatively in terms of the ^-factor, the
inverse of the specific attenuation factor, and in
general the higher the ^.-factor the stronger the rock.
High travel paths give rise to little attenuation, and
vice versa. Seismic waves travelling up the length of
the seismic zone appear to pass through a region of
high 2,(about 1000), while those travelling to lateral
recorders pass through a more normal region of low
Q (about 150). The BeniofF zone thus appears to
define the top of a high Qzone about 100 km thick.
Volcanic arc
Bering Sea
The BeniofFzone had originally been interpreted as a
large thrust fault between different crustal provinces.
The seismic data allowed a new interpretation to be
made in terms of a high belt ofPacific lithosphere
underthrust into the mantle. This interpretation has
been refined by Barazangi St Isacks(1971) by the use
of a local seismometer network (Fig. 8.7) in the
region of the Tonga arc. In addition to the pret ious
results, a zone of very high attenuation (extremely
low of about 50) has been defined in the upper
most mantle above the downgoing slab in a region
about 300 km wide stretching between the active
island arc (Tonga) and back-arc ridge (Lau Ridge).
This implies that the mantle beneath the marginal sea
(Lau Basin) is much weaker than elsewhere or that
the lithosphere is considerably thinner. The data have
important ramifications for the origin of back-arc
basins and will be considered in more detail in
Section 8.9.
Detailed investigations of the region above
the subducting lithosphere have also been carried
out using seismic tomography (Section 2.1.8).
Figure 8.8, from Hasemi et al. (1984), shows a
section through the Japan arc at 39.8°N with con
tours in percentages of the relative velocity variation
in the region ofthe subducting Pacific Plate. The top
ofthe downgoing slab is clearly indicated by a region
of high velocity. Also present is a low velocity anom
aly in the wedge above the subducting plate which
extends from the surface to a depth of about 65 km.
This underlies the volcanic arc and is interpreted as
an upwelling diapir of partially molten material. Also

shown is the double BeniofF zone in this region,
which is discussed later in this section.
The earthquake activity of the downgoing slab
occurs as a result of three distinct processes
(Fig. 8.9). In region 'a' earthquakes are generated in
response to the bending of the lithosphere as it
begins its descent. Downward flexure of the litho
sphere throws the upper surface of the plate into
tension, and the normal faulting associated with this
stress regime gives rise to the observed earthquakes,
which occur to depths of up to 25 km.
Flexural bending of the lithosphere also gives rise
to the topographic bulge present in the subducting
plate on the oceanward side of the island arc. This
regional rise of sea bed topography is located about
120-150 km from the trench axis and has an ampli
tude ofseveral hundred metres. It is associated with a
regional positive gravity anomaly of about 500 g.u.
(Watts & Talwani, 1974). Simple beam theory pre
dicts that the presence ofthis bulge is a consequence
of the downward deflection of the subducting plate
(Fig. 8.10). However, closer investigation of litho-
spheric behaviour in this environment (Chappie &
Forsyth, 1979)indicates that the flexure is not com
pletely elastic and must involve considerable plastic
(permanent)deformation. A two-layer elastic-plastic
lithospheric model would deform in a very similar
way to that observed (Fig. 8.11).
Region 'b' (Fig. 8.9) is characterized by earth
quakes generated from thrust faulting along the con
tact between the overriding and underthrusting
plates. Indeed, the overriding plate suffers compres-
sional deformation for several tens of kilometres to
the landward side of the trench. Focal mechanism
solutions for earthquakes associated with regions 'a'
and 'b' of Fig. 8.9 are shown in Fig. 8.12, which
represents the distribution of earthquake types
around the Aleutian island arc (Stauder, 1968). The
belt ofearthquakes to the south ofthe islands is caused
by normal faulting associated with the flexure of the
top part of the Pacific Plate which is underthrusting
the Bering Sea in a northwesterly direction. The
groups ofearthquakes lying under orjust to the south
of the island chain are indicative of thrust faulting.
The nodal planes dip steeply to the south and gently
to the north. It is probable that the latter planes
represent the faults so that these earthquakes are
generated by the relative movements between the
Pacific and Bering Sea lithosphere. The single focal
mechanism solution indicative of strike-slip move
ment is probably an arc-arc dextral transform fault
SUBDUCTION ZONES 141
occurring on a vertical plane perpendicular to the
island chain.
The earthquakes occurring in the Benioff zone in
zone *c' (Fig. 8.9) at depths greater than the thick
ness of the lithosphere at the surface are not gener
ated by thrusting at the top of the descending plate,
because the asthenosphere in contact with the plate is
too weak to support the stresses necessary for exten
sive faulting. At these depths earthquakes occur as a
result of the internal deformation of the strong
descending slab of lithosphere, so that the majority
of events lie about 30-40 km beneath the top of the
slab. Hasegawa etal.(1978), by making use ofa local
array of seismographs, have identified what appears
to be a double Benioff zone beneath the Japan arc
(Fig. 8.13; see also Fig. 8.8). The arrival times of
different seismic phases indicate that the upper of
these corresponds to the top of the descending slab
and that the lower corresponds to the main Benioff
zone where earthquakes originate by internal deform
ation. It is probable that the upper plane earthquakes
result from the stresses caused by the unbending of
the plate which had suffered a certain amount of
permanent, plastic deformation during its initial de
scent(Isacks & Barazangi, 1977). Fujita & Kanamori
(1981) and Spence (1987) attribute the lower seis
mic zone to tension in the slab resulting from the
deeper sinking plate, and the upper seismic zone to
localized compression of the plate as its motion is
resisted by the surrounding mantle. Wiens et aL
(1993) have suggested that the double Benioffzone
results from transformational faulting (see below) at
the edges of the upper part of the descending slab.
Kawakatsu (1986) and Wiens et al. (1993) have
described a similar double seismic zone beneath the
Tonga arc and Gorbatov et al. (1994) a double
seismic zone beneath Kamehatka. With the imple
mentation of more refined location techniques, it is
probable that more will be discovered in the future.
The presence ofearthquakes at depths in excess of
70 km is paradoxical in that below this level the high
pressure causes materials to flow rather than fracture.
From 70 km to 300 km it appears that faulting
occurs during the rapid dehydration of serpentinite
formed higher in the subduction zone by the hydra-
tion of olivine. Below 300 km the earthquake mech
anism is believed to be the sudden phase change
from olivine to spinel, known as transformational or
anticrack faultin£i., which takes place by rapid shear
ing of the crystal lattice along planes on which
minute spinel crystals have grown (Green, 1994).
 IS)

Samoa Is.
Samoa Is.

15°S —

Tonga Is.

South Fiji Ridge $ fo

Tonga
Tonga Trench
South
Trench
Fiji
Basin

Tonga-fiji Tonga-Fiji
earthquakes
Focal depth earthquakes
o 0-100 km
Kermadec A ! X 200-300 km
A 100-200 km
Trench < 300-400 km
Water depth
contours in km o 400-500 km
Kermadec rnO-6 ■ >6 km
Trench
L_^ fJmfltSJBif
175°W 175°W
 SUBDUCTION ZONES 143

Samoa Is.'

15°S

Fiji Is.

Tonga

20°
± wv

#f-v
Tonga
Trench
V

V
V
•V V
25° - w

Tonga-Fiji
earthquakes
V 500-600 km

Kermadec + > 600 km

Fig.8.4 Epicentres ofTonga-Fiji Trench
earthquakes;(a)shallow, showing isobaths ▼ Volcanoes

in kilometres;(b)intermediate;(c) deep
(redrawn from Sykes, 1966^ Journal of
Geophysical Research 71, with permission 180° 175°W
from the American Geophysical Union). (c)

The change to spinel is probably complete by a depth
of about 700 km,explaining the termination of sub-
duction zone seismicity at this depth. It is also
probable that a similar transformation from enstatite
to ilmenite contributes to subduction zone seismic
ity(Hogrefe et al,^ 1994).
Unlike the shallow events of regions 'a' and 'b' in
Fig. 8.9 where the principal stress directions are hor
izontal or vertical, the deep events of region 'c' are
characterized by principal stress directions which are
either parallel or orthogonal to the dip of the de
scending plate (Isacks et al.^ 1969)(Fig. 8.14).Con
sequently, the nodal planes determined by focal
mechanism solutions do not correspond to the dip of
the Benioff zone or a plane perpendicular to it. The
principal stress directions show that the descending
plate is thrown into either downdip compression or
extension. Isacks & Molnar (1969) have suggested
that the distribution ofstress type in the seismic zone
may result from the degree of resistance experienced
by the plate during its descent, and Spence (1987)
has described this resistance in terms of the nett
eflPect of ridge push and slab pull forces (Section
11.6). In Fig. 8.15a the plate is sinking through the
asthenosphere because of its negative buoyancy and
is thrown into downdip tension as its descent is
unimpeded. In Fig. 8.15b the bottom of the plate
approaches the mesosphere, which resists descent
and throws the leading tip into compression. As the
plate sinks farther (Fig. 8.15c), the mesosphere pre
vents further descent and supports the lower margin
of the plate so that the majority of the seismic zone
144 CHAPTERS

Volcanic arc
Coralline arc
1-
A;vy;.vLN
Water depth - >>v Tonga Trench
(km) 5

+200

Distance from Niumate, Tonga (km)

-400 -200 0 +200

200
500
A 8o
ZO

Depth
(km)
550
400 -1r A

A
600
km

600 A

-350 -300 -250 km

Fig.8.5 Vertical section perpendicular to the Tonga arc

showing earthquake foci during 1965. Circles, foci
projected from within 0-150 km north ofthe section;
triangles, from 0-150 km south. Exaggerated topography
(13:1) above. Inset, enlargement of the region of deep
earthquakes (redrawn from Isacks et al.^ 1969, with
permission from the Geological Society of America).

Tonga Trench Rarotonga

Fig.8.6 Hypothetical section

across the Tonga arc based on the
attenuation ofseismic waves
(redrawn from Oliver & Isacks,
1967,Journal ofGeophysical
Research 72,with permission
from the American Geophysical
Union).
 SUBDUCTION ZONES 145

Tonga Tonga
Fiji Islands Lau Ridge Lau Basin Ridge Trench

Hioh Q
Extremely low O

LowQ LowQ

Depth(km)
400
Fig.8.7 Schematic section across the
Tonga arc showing the zone of very
high seismic attenuation beneath the
Lau marginal basin (redrawn from
Barazangi & Isacks, 1971^Journal of
Geophysical Research 76, with
permission from the American 1000 900 800 700 600 500 400 300 200 100
Geophysical Union). Distance from trench(km)

experiences compression. In Fig. S.lSd a section of

Fig.8.8 Contours offractional velocity variations (in %)
derived by tomographic analysis ofthe region above the
Japanese subduction zone at 39.8°N. Earthquake foci
defining the double Benioffzone are also shown (redrawn
from Hasemi ^a/., 1984, with permission from Elsevier).
the downgoing slab has decoupled so that the upper
portion of the plate is thrown into tension and the
lower portion into compression. However, as noted
above it has also been suggested (Isacks & Barazangi,
1977) that compressional events at depth in the
subducting plate may originate from the unbending
ofthe slab. A global summary ofthe stress directions

Volcanic arc

3 O O O

50 —

100 - 8f%
Q.
(D
o

150 -

O O

200 —Qo-
142°E
146 CHAPTERS
w«a:
'•V«%V.V#V.V»*.W.V.V.V»*«%%*A
v.%v.v.%v
.v.v.w.
•.%•••«•••••
vwlm
{.VAV.
K'MwS
itt r--'?:•:•:%■
%wfSft .%%vK%;
5S5S^g
w.w.ss m v,w>>
'.W.K
>X.X<<<«»555m
determined from focal mechanism solutions (Isacks
& Molnar, 1971) is shown in Fig. 8.16.
The stress distributions shown in Fig. 8.15b,d pro
vide one possible explanation for the seismic gaps
observed along the middle parts of the Benioff zone
at certain trenches, such as the Peru-Chile Trench
(Figs 8.16 and 9.30) (Section 9.5) where it is known
that the slab is continuous (James & Snoke, 1990). A
further type of seismic gap appears to be present in
some island arcs at shallow depths. Figure 8.17
shows sections through the BeniofF zone at the
Aleutian-Alaska arc (Jacob et aL^ 1977). There is a
prominent gap in seismicity between the trench and a
point about half-way towards the volcanic arc which
becomes progressively greater from west to east. The
angle of underthrusting is very shallow in this region.
The probable cause of this seismic gap and shallow
underthrusting is the presence of copious quantities
of terrigenous sediments within the trench which
become increasingly abundant towards that section
Sea level
Fig. 8.10 Pownbending of an elastic or elastic-perfectly
plastic plate at a subduction zone (redrawn from Turcotte
etal,^ 1978, with permission from Elsevier).
Lithosphere
KAM. ^^^A^eriosph^ei
>v.w.;
i«iS55!
Fig. 8.9 Plate model of subduction
zones; a, b and c indicate regions of
distinctive focal mechanism
solutions. See text.
of the trench adjacent to Alaska. The unconsolidated
nature of these sediments probably prevents any
build-up of the strain energy necessary to initiate
earthquakes, and their high positive buoyancy may
force the subducting plate to descend at an anom
alously shallow angle.
Information on the depth reached by the down-
going slab has been provided by seismic tomography
(Section 2.1.8), showing it to be dependent on the
past history of the subduction zone (van der Hilst,
1995). In the western Pacific, where the lithosphere
is descending at a rapid rate, the slab reaches depths
of at least 800 km and passes directly through the
base of the mantle transition zone at 670 km (Sec-
Mariana profile
Elastic
Tonga profile
Elastic
Elastic-plastic
-100 300 km
Fig. 8.11 Observed and theoretical profiles of lithosphere
bending at a trench: (a) Mariana Trench, with an elastic
lithosphere 29 km thick; (b) Tonga Trench, better
modelled by an elastic-perfectly plastic plate 32 km thick
(redrawn from Turcotte et al.^ 1978, with permission
from Elsevier).
 SUBDUCTION ZONES 147

Bering Sea

+ 50°

Pacific Ocean

Fig.8.12 {Top) Focal mechanism solutions of
earthquakes in the Aleutian arc, compressional quadrant
shaded (redrawn from Stauder, 196%^ Journal of
Geophysical Research 73, with permission from the
American Geophysical Union).
Fig.8.13 (Bottom) Distribution ofearthquakes beneath
northeastern Japan arc. Shaded line is probably the top
ofthe descending lithosphere (redrawn from Hasegawa
etal.^ 1978, with permission from the Royal Astronomical
Society).

NE Japan

E 100

Q 150

250
140°E 142°E 144°E
148 CHAPTERS
Trench axis
200 km
Depth(km)
r«ioo
mMMWM
Low strength
Increasing
High strong
Fig.8.15 A model ofstress distributions in the
descending lithosphere. Solid circles, extensional stress
downdip; open circles, compressional stress downdip
(redrawn from Isacks & Molnar, 1969).
tion 2.8.5) into the lower mantle,for example in the
south of the Tonga-Kermadec subduetion zone and
at the Izu-Mariana subduction zone (Fig. 8.18a).
However, in the northern part of the Tonga-
Kermadec subduction zone, there is a horizontal
deflection of the slab at the 670 km discontinuity
before it continues its descent (Fig. 8.18b). This is
because, unlike the southern part of the subduction
zone, in the north there has been up to 1000 km of
back-arc spreading (Section 8.9), which caused the
slab to flatten out. The penetration of the slab into
the lower mantle has important ramifications for the
nature of mantle convection, and this will be dis
cussed in Section 11.5.3.
8.4 Thermal structure of the
downgoing slab
The strength and high negative buoyancy ofsubduct
ing oceanic lithosphere and its capacity for sudden
¥;• Trench axis
Fig.8.14 Schematic focal
mechanism solution distribution on
a section perpendicular to an island
arc. Inset shows alternative
intermediate depth mechanism
(redrawn from Isackseta/., 1969,
with permission from the
Geological Society of America).
failure in the generation of earthquakes are conse
quences ofits relatively low temperature with respect
to normal mantle material. The lithosphere can re
tain its separate thermal and mechanical identity to
considerable depths until sufficient heat has been
transferred to it from the mantle to increase its
temperature to that ofits surroundings.
The variation of temperature within the sinking
slab can be calculated from heat conduction equa
tions provided that its thermal properties and bound
ary states are specified. The factors controlling the
temperature distribution are:
1 the rate ofsubduction: the more rapid the descent
the less time there is for absorption of heat from the
surrounding mantle by conduction;
2 the thickness of the descending slab: the thicker
the slab the greater the time taken for it to equilibrate
thermally with the surrounding asthenosphere;
3 frictional heating of the upper and lower surfaces
ofthe slab as the descent ofthe slab is resisted by the
asthenosphere;
4 the conduction of heat into the slab from the
asthenosphere;
5 the adiabatic heating associated with compression
of the slab as the pressure increases with depth;
6 the heat derived from radioactive decay ofminerals
in the oceanic lithosphere,likely to be small as oceanic
plates are largely barren ofradioactive minerals;
7 the latent heat associated with phase transitions of
minerals to denser crystalline structures with depth:
the principal phase changes experienced by the slab
are the olivine-spinel transition at about 400 km
depth which is exothermic, and the spinel-oxides
 SUBDUCTION ZONES 149

North South
100

500

north central south

700
Marianas Tonga Honshu Kurile Izu - Bonin Kermedec Calabrian

South/
100

300
km

500

700 '± / N. Chile New Hebrides

Peru

Flores
100

300
/ :
km

500

700 New New M.America Ryukyus S. Sandwich Burma Romania

Sunda Mindanao Zealand Britain Solomons Aleutians Sumatra Aegean Hindukush

Fig.8.16 Summary ofthe distribution of downdip neither P- nor T-axis parallel to zone; solid lines,
stresses in Benioffzones. Open circles, events with approximate form ofseismic zone (redrawn from Isacks
compressional axis parallel to dip ofzone; solid circles, & Molnar, 1971, with permission from the American
events with tensional axis parallel to dip ofzone; crosses. Geophysical Union).

transition at about 670 km which may be exo- or
cndothermic.
Different solutions ofthe temperature distribution
have been derived by various workers depending
upon the relative contributions of the above phe
nomena. Two such models derived by Toksoz et al,
(1971) and Schubert et al. (1975) are presented in
Fig. 8.19. Although differing in detail, all such mod
els indicate that the downgoing slab maintains its
thermal identity to great depths. Temperature con
trasts of 700® C are shown to exist between the slab
and mantle at a depth of700 km.
The depths to which the subducting oceanic litho-
sphere maintains its separate thermal identity control
the length of the Benioff zone. Molnar et at.(1979)
have deduced that the downward deflection of iso
therms, and hence the length ofthe seismic zone, is
proportional to both the rate of subduction and the
square of the thickness of the lithosphere. Litho-
sphere thickness is proportional to the square root of
its age (Section 6.9) so that the length of seismic
zones should be proportional to the product of
convergence rate and age. That this is generally so is
illustrated by Fig. 8.20, and although there is con
siderable scatter the data appear to fit the relationship
length (km)= rate (mma"^)xage (Ma)/10. Other
controls of the geometry of the subducted plate are
discussed by Cross & Pilger (1982). An alternative
explanation for the absence ofseismicity below about
700 km is that the phase changes suffered by the slab
150 CHAPTERS

Alaska

170'

Pavlof volcano Trench

. . ....
r ■ i
; A
''
km

~
- 100
f Shumagin Islands
View N59°E
iNW
. . . . 1 '• ■ . ... 1 . .. 1 . .. . 1 . 200

Mt. Redoubt Trench

—i ^7^
km

Cook Inlet 100

•xW
ViewN17°E
WNW BSE
j-._L I . ... I . . . . I . . I . I 200

Mt. Spurr Trench

.. 1 .
.. ..j.... 1 0
:c c';
km
Fig 8.17 Location map and cross-
Skwentna 100 sections across the Aleutian arc
View N17°E showing earthquake foci (redrawn
iWNW ese; from Jacob et 1977,with
. .. 1 . . . . 1 . . . . 1 .. . 1 .. .. 1 200
200 400 600 800 permission from the American
km Geophysical Union).

may produce fine-grained materials which behave in a
superplastic manner and thus cannot generate earth
quakes (Ito & Sato, 1991).
8.5 Metamorphism at convergent
margins
The anomalous thermal and pressure conditions as
sociated with subduction zones give rise to distinctive
suites of metamorphic rocks whose disposition de
pends on the direction ofunderthrusting(Fig. 8.21).
An abnormally low geothermal gradient of about
10®Ckm~^ results from the rapid descent of rela
tively cool oceanic lithosphere at trenches (Section
8.4) to depths of about 30 km. The high pressures
and low temperatures in this environment give rise to
a metamorphic complex characterized by the pres
ence ofglaucophane and jadeite, which are indicative
of the blueschist facies. Blueschist metamorphism
has been observed in the Mariana subduction zone
(Maekawa et 1993) and blueschist terrains are
associated almost everywhere with ophiolitic suites

Fig. 8.18 The vertical extent of subduction revealed by
seismic tomography,(a) With no back-arc spreading, the
slab can directly penetrate to the lower mantle, e.g.
Kermadec arc.(b) Back-arc spreading causes a horizontal
slab deflection at the 670 km discontinuity, e.g. Tonga arc
(after van der Hilst, 1995).
(Ernst, 1973)(Section 2.5). The ascent of magmas
generated by dewatering-induced mantle melting
(Section 8.8) gives rise to anomalously high geother-
mal gradients of more than 25°Ckm"'^ up to about
50°Ckm~^ A second metamorphic complex conse
quently arises associated with surface volcanism,
characterized by minerals such as andalusite, which
are generated at high temperatures and low pressures
(Fig. 8.21). Figure 8.22, from Miyashiro (1973),
shows the minerals expected to be formed at the
temperature and pressure conditions experienced in
these two metamorphic environments.
Subduction zones consequently contain paired
metamorphic belts (Miyashiro, 1972); an outer high
pressure/low temperature type on the oceanward
side and a parallel low pressure/high temperature
belt of similar age associated with the island arc,
which are typically about 100-250 km apart.
Paired metamorphic belts were first recognized in
Japan (Fig. 8.23) where three pairs of different ages
have been identified. At present Pacific lithosphere is
being subducted in a northwesterly direction be
neath the Japan arc, and the metamorphic polarity of
the Sangun/Hida and Ryoke/Sanbagawa paired
belts indicates that they were formed by underthrust-
ing in this same direction.The Sanbagawa/Ryoke
belts however, are much closer together than sug
gested by the plate tectonic model,and so it has been
SUBDUCTION ZONES 151
suggested that their common margin, the Median
Tectonic Line, has been the locus ofsome 400 km of
strike-slip movement responsible for bringing the
Sanbagawa and Ryoke belts into juxtaposition. This
transcurrent movement has been amply confirmed
by detailed mapping (Takagi, 1986). The Hidaka/
Kamuikotu paired belt shows the opposite metamor
phic polarity, and must have been formed during a
different phase of plate movements when the direc
tion of subduction was from the west ofJapan.
Paired metamorphic belts have now been identi
fied from many of the subduction zones, of both
island ate and Andean type, around the margin ofthe
Pacific, as shown in Fig. 8.24. Full descriptions of
many of these belts are given in Frnst (1975). In
other subduction zones, however, such as the Phan-
erozoic orogenic belts of the Atlantic region, un
paired metamorphic belts occur in which one of the
two belts is not developed or the contrast between
belts is unclear. Miyashiro (1973) has noted that a
very low geothermal gradient is required for blue-
schist grade metamorphism, and if this is not
achieved only medium pressure metamorphism will
result (Fig. 8.22). Thereafter, if the rate of plate
descent is slow, or if young, hot oceanic lithosphere
is being subducted, the very low thermal gradient
necessary for blueschist grade metamorphism will
not be achieved and the high pressure belt will be
missing. Investigations in California and Japan have
shown that in past subduction systems, blueschist
metamorphism ceased 30-40 Ma before the under-
thrusting ofan ocean ridge as young, hot lithosphere
started subducting. The rapid plate motion and rate
of underthrusting of the Pacific lithosphere thus
account for the common occurrence of paired meta-
152 CHAPTERS

Volcanic High heat flow

Trench line Anomalous mantle
j -0
400X ^ 400°C
800X Continental

1200°C Oxxv^ ■ 800°C

1200°C -200 km

• 600X

Olivine
1600"C -400 km
Spinel

800°C
— 600 km
Spinel
1700°C 1700°C
Oxides

(a)
- 800 km

300

km

-0
400°C
800°C
1200°C

1600°C
- 200 km

- 400 km

2000°C

- 600 km

2400X

800 km

Fig.8.19 Two models of the thermal structure ofthe

descending lithosphere:(a) redrawn from Schubert et»/.,
1975;(b)redrawn from TOksoz^f a/., 1971, with
permission from the American Geophysical Union.
Shading indicates zones of phase changes.
 SUBDUCTION ZONES 153

o Aleutians * Tonga-Kermadec morphic belts around the Pacific.

• Kurlle i Japan Blueschist terrains may be located well within
□ New Zealand X Central America present day continents, and probably represent the
South America • Lesser Antilles suturing of ancient continental margins following
the consumption of an intervening ocean (Ernst,
1973). Blueschist belts are common in Mesozoic and
Cenozoic subduction complexes, but are rarer in
CD
C 1000 older Phanerozoic belts. This may reflect an in
o
creased presence of the more extreme pressure-
■g 800 temperature conditions with time. The presence of
several generations of paired metamorphic belts in
regions such as Japan argues strongly for the lon
gevity of many subduction zones.
400
tu
Barber (1982) has questioned the validity of the
paired metamorphic belt concept by proposing a new
model for the evolution of Japan. He demonstrated
that the deformation of the members of paired belts
2000 4000 6000 80001000012 000 has occurred at different times. Palaeomagnetic in
Rate X age (km) vestigations have revealed that elements of Japan
Fig. 8.20 (Top) Relationship between length of BeniofF
zone and the product of spreading rate and age. Fig. 8.21 (Bottom) Model for the interpretation of
Approximate uncertainties given by error bars in upper left convergent plate margins as applied to southwestern
corner (redrawn from Molnaref «/., 1979, with Japan. MTL, Median Tectonic Line (redrawn from Barber,
permission from the Royal Astronomical Society). 1982, with permission from the Geologists' Association).

Japan Sea Japanese Islands Shikoku Shelf Nankai Trough Pacific Ocean
Volcanicity
' Alkaline Calc-alkaline Tholeiitic

Marginal sea Shelf Magmatic arc Forearc Forearc Trench Ocean floor
basins ridge /

Low pressure metamorphism

4^ Intermediate pressure
rrt-^, I metamorphism MM '
t —LLi iLi i' I I I' I I I I _
Partial melting of
granitic crust
\j^ pressure
^ metamorphism

Partial melting of
100-1 mantle peridotite
(100-200 km)
r^Zoneof partial
melting of basaltic
Z = zeolite fades magma
PP = pumpellyite-prehnite Asthenosphere
gl = glaucophane-iawsonite
eel = eclogite A
154 CHAPTERS

High pressure
metamorphism
1 Albite + H2O = analcime + quartz
1000
2 Lawsonite +2quartz + HgO = laumontite
3 Aragonite = calclte
800 4 Jadelte + quartz = albite
Pressure 5 Kyanite = andalusite
(MPa) 6 Kyanite = sillimanite
600
7 Sillimanite = andalusite
8 Muscovite + quartz= Kfeldspar + AI2Si05+ H2O
400 9 Onset of granite melting
10 Onset of olivine tholeiite melting

200 Left hand side of equation

represents high pressure
assemblage

200 400 600 800 1000

Temperature(°C)

Fig.8.22 ijop) Pressure-temperature conditions of

regional metamorphism. Dashed lines, geothermal curves
for high, medium and low temperature metamorphism;
solid lines, equilibrium curves for the reactions listed
(redrawn from Miyashiro, 1973,with permission from
Elsevier).
Fig.8.23 {Bottom) Three paired metamorphic belts in
Japan, F-F is the Itoigawa-Shizuoka Line (redrawn from
Miyashiro, \972^ American Journal ofScience 272,629-
56, with permission of American Journal ofScience).

Late Cretaceous Hidaka belt (Low pressure)

Early Tertiary

400 km
I I

ijggilHonshu
Sangun belt (High pressure) I Permian -
Hida belt (LOW pressure)

Shikoku

pressure/high temp) 1 Late Jurassic

edian * '
Median Sanbagawa belt (High pressure/low temp) J early Cretaceous
tectonic line

^/A'Taiwan
/7 Celebes
Fig.8.24 Paired metamorphic belts
in the circum-Pacific region. Dotted
lines, high pressure belts; solid lines,
low pressure belts (redrawn from
Miyashiro, 1973, with permission
from Elsevier).
were originally situated in equatorial latitudes, and
that the Japanese islands have undergone a clockwise
rotation since Palaeogene times, when the major axis
of the islands was parallel to the Asian coast. Barber
suggests that northward subduction at an oblique
angle at this time was accompanied by sinistral strike-
slip movements along the trench. The various Japan
ese high and low pressure belts were accreted and
brought together during these phases ofmovements,
that is, they represent suspect terrains (Section 9.5).
Although the paired metamorphic belt model
appears reasonable, this new synthesis for the type
example in Japan questions its validity (Windley,
1984). It is necessary to await detailed investigations
of other paired belts before a decision can be made
about its general applicability.
8.6 Ocean trenches
Oceanic trenches are the direct manifestation of
underthrusting oceanic lithosphere and are devel
oped on the oceanward side of both island arcs and
Andean-type mountain ranges. They represent the
largest linear depressed features of the Earth's sur
face, and are remarkable for their depth and continu
ity. The Peru-Chile Trench is 4500 km long and
SUBDUCTION ZONES 155
NW Canada^
X\ ^ ^
Kamchatka Washington
^California j
^Chil^
New Zealand
reaches depths of 2-4 km below the surrounding
ocean floor so that its base is 7-8 km below sea level.
Trenches are generally 50-100 km in width and in
section form an asymmetric V-shape with the steep
est slope of 8-20® on the side opposite the under-
thrusting ocean. The sediment fill of trenches can
vary greatly from virtually nothing as in the Tonga-
Kermadec Trench to almost complete as in the Lesser
Antilles Trench. All are associated with very large
negative free air and isostatic gravity anomalies which
result from the downward displacement ofthe litho
sphere so that mantle material is displaced by either
sea water or low density sediments.
In section the morphology of the sub-sedimentary
basement of all trenches is very similar (Fig. 8.25),
and is independent of the amount of sediment
present. Consequently, it is inferred that the sedi
ment loading plays no role in the formation of
trenches. This sediment, however, may be deformed
by the underthrusting to form the accretionary
prism.
8.7 Accretionary prism
The accretionary prism (first arc, subduction com
plex, accretionary wedge)forms on the inner wall of
 (Jl
O)

O
X
>
"0

3J
00

—I 0

10 10

100

km

Vertical exaggeration: 50

Fig.8.25 Stack of 35 projected topographic profiles ofPacific subduction zones(redrawn from Hayes & Ewing, 1970, with permission from John Wiley).

an ocean trench (Moore & Silver, 1987). The inter
nal structure and construction have been deduced
from seismic reflection profiles and sampling at
active subduction zones and by the study of ancient
subduction complexes now exposed on land.
Karig (1974) and Karig & Sharman (1975) have
proposed that accretionary prisms develop when
trench-fill turbidites (flysch), and sometimes also
pellagic sediments and underlying oceanic crust, are
scraped from the descending oceanic plate by the
leading edge of the overriding plate to which they
become accreted. The internal structure ofthe accre
tionary prism is consequently made up of a series of
imbricate listric thrusts dipping towards the arc
which define a series ofwedge-shaped pockets within
which are developed complex folds verging towards
the trench (Dickinson,1977)(Fig. 8.26). As subduc
tion continues, older thrust wedges are gradually
moved upwards and rotated arcwards by the addition
ofnew wedges to the base ofthe prism (Platt, 1986).
The older thrusts consequently become more steeply
dipping with time. This mechanism has been termed
'offscraping', 'plastering', 'underplating' and 'sub-
cretion'and is responsible for the outward growth of
the prism. This model requires that the most intense
deformation occurs at the oceanward base of the
sedimentary pile, although the older thrusts may
remain active during their rotation. During this de
formation a forearc basin may develop between
trench and island arc in a trough resulting from the
inability of sediment sources to keep pace with
forearc development. The forearc basin covers the
oldest members of the accretionary wedge. In addi
tion, small sedimentary pockets may develop on top
ofthe thrust wedges.
This model is in accord with observations made on
ancient subduction complexes (Ernst, 1975) in
which sediment age and metamorphic grade increase
from trench to arc and major thrust faults dip to
wards the arc. The oldest, high grade, rocks are
structurally higher and have been uplifted with re
spect to younger deposits. These older rocks, which
may have formed at depths as great as 30 km, have
been translated upwards and may be juxtaposed with
much younger, lower grade rocks.
The uplift of high grade rocks cannot be fully
explained by continued movements along thrust
faults. It is possible that the high grade rocks rise
buoyantly (Ernst, 1975), but this seems unlikely
as their density would be higher than the younger
SUBDUCTION ZONES 157
rocks encountered at structurally lower levels. Seeley
(1979) and Cowan & Silling (1978) have suggested
that, in addition to underplating, mass flow may
occur within the prism. Material may be transported
downward by the underthrusting and subsequently
flow upwards on reaching the lowest part of the
trench (Fig. 8.27). Cowan & Silling (1978) have
constructed a scale model demonstrating that this
form of uplift is feasible, and have also modelled the
mechanics theoretically in terms of the two-
dimensional flow of a viscous fluid round an acute
corner in response to sliding a rigid plate beneath
another fixed plate. Davis et al.(1983) have likened
the overall mechanics in an accretionary prism to
those of a wedge of soil scraped in front of a bull
dozer. The material in the prism deforms internally
and steepens its slope until a certain critical taper
between the top and base of the wedge is attained,
whereupon it slides stably and continues to grow at a
constant taper as new material is added to its leading
edge. Indeed, with the addition of new sedimentary
material at its edges, the prism will deform internally
until it reaches a stable configuration in which the
internal gravitational forces ofthe wedge balance the
traction on its base from the subducting slab (Platt,
1986).
Karig (1982) has suggested that the trajectory of
materials within the subduction complex is influ
enced by the thickness and mechanical properties of
the sedimentary cover on the subducting plate, the
basement relief on this plate and the convergence
rate. Thick sedimentary cover gives rise to large
subduction complexes, although the proportion ac
tually accreted to the overriding plate varies signifi
cantly and is controlled by rheological and porosity
variations within the sediments. Accretion depends
upon the nature and geometries of the arriving sedi
ment (Fig. 8.28). Relatively thick trench turbidites
overlying a thin pelagic cover,such as in the Cascade
arc, allow thrusting to take place at the base of the
turbiditic sediments (Fig. 8.28a). At the Makran arc,
however, where the sediment carried to the trench
may be in excess of 6 km thick, only the upper
2-2.5 km are accreted (Fig. 8.28b). Where the sedi
ment cover is thin, such as the Marianas arc, the
situation at the trench is less certain. An increased
proportion ofoceanic lithosphere may participate in
the accretion, and this would be enhanced by relief
on the oceanic basement surface (Fig. 8.28c) due to
the normal faulting resulting from downbending of
158 CHAPTERS

Volcanic arc
Trench
\1
Oceanic crust -
!w.v.?.v.v

»»!W555553

•Krw.v.%x»x*:<.:

Subduction complex

»:*K^K?S
Lithosphere

Asthenosphere

Fore arc basin

\

<»M%w5

IwvXvIvI
.%%%%%%" >>>»!•;

WM
5>>>S'
.•.w.v

wtwl

100 km

Trench slope basin

(d) \
20 km

Oceanic crust

Fig.8.26 (a-c) Idealized sequence ofthe evolution ofan
accretionary prism (redrawn from Dickinson, 1977, with
permission from the American Geophysical Union):(a)
incipient stage;(b)forearc basin;(c)full forearc basin.
Solid lines in subduction complex, active thrusts; broken
lines, inactive thrusts,(d)Schematic, vertically
exaggerated section ofan accretionary wedge (redrawn
from Cowan & Silling, 1978, with permission from the
American Geophysical Union).

Fig.8.27 Vertically exaggerated generalized kinematic
model of accretion at subduction zones (redrawn from
Cowan & Silling, 1978, with permission from the
American Geophysical Union).
the lithosphere (Section 8.3). In some cases, how
ever, the horst and graben topography might act
rigidly as a file and erode the base of the overriding
plate. This latter possibility is considered unlikely as
the heterogeneous nature of the top of oceanic layer
2 probably ensures that it is relatively weak.
It appears that the situation illustrated in
Fig. 8.28b is quite common,and the underthrusting
of undeformed sediments beneath accretionary
wedges has been observed in several subduction
zones. For example, Westbrook et al. (1982) have
shown that only about the top third of sediments
covering the Atlantic are added to the leading edge
of the Barbados accretionary wedge of the eastern
Caribbean, the remaining sediments being carried
Oceanic crust
Trench sediments
(a) Cascades
Fig.8.28 Modes of accretion as a
function of the form ofsediment
cover on the descending plate
(redrawn from Karig, 1982, with
permission from Academic Press). (c) Mariana
SUBDUCTION ZONES 159
beneath the wedge. The decollement surface thus
marks the plate boundary, and develops in a weak
sediment layer, typically a low permeability hemi-
pelagic mud underlying more rapidly deposited,
stronger, more permeable trench turbidites (Moore,
1989). The ofF-scraped sediments undergo only
limited burial, whereas the underthrust materials are
deeply buried. They are thus more likely to be pre
served in the geological record, where they are seen
to have undergone high pressure metamorphism to
the blueschist or eclogite facies, indicating depths of
burial of at least 30 km.
The sediments above and below the decollement
zone descend so rapidly that they have no opportu
nity ofdewatering before burial. The evolution ofthe
resulting hydrologic system controls many aspects of
the tectonic development of the accretionary wedge
(Moore, 1989). The small angle of taper of the
wedge, and the low strength of the wedge and
underlying sediment imply that the shear stresses
along the decollement must be very low(Westbrook,
1982). This can only be made possible ifthe pressure
of the lithostatic load is balanced by pore water
pressure of similar magnitude. Consequently, the
decollement only exists because of the overpressur-
ing of pore waters in the sediments above and below
it. However, the decollement zone is also known to
act as a drainage channel for the fluid released by
Basinal pelagic sediments
Basinal clastic sediments
(b) Makran
160 CHAPTERS
sediments at depth, and this requires the decolle-
ment zone to possess a lower fluid pressure than its
surroundings. This apparent conflict can be recon
ciled by a model in which the fluid pressure in the
decollement zone varies cyclically. During periods of
strain accumulation there is an interval in which fluid
flows towards fractures developed in the decollement
zone. Eventually the fractures join up as fluid pres
sure increases, and failure and slip take place on the
decollement zone. As this takes place fluid is expelled
along the decollement zone as the connected frac
tures offer a higher permeability route than the
source sediments. There is then a period of recovery
as deformation of the source sediments increases
their pore pressure to the point at which fractures are
again developed and the cycle recommences. Fluid
expulsion thus takes place by a pumping mechanism
(Moore, 1989), and is manifest in the development
of mud volcanoes and diapirism (Westbrook et al.^
1984).
The outward flow of water from accretionary
prisms can explain many of the phenomena associ
ated with them. Reck (1987) has reported thermal
gradients from DSDP holes in the prism ofnortheast
Japan of24-36°C km"^.Ifextrapolated downwards,
temperatures in excess of400®C would be expected
at 15 km depth, taking into account the increasing
thermal conductivity with depth associated with de
creasing porosity. These temperatures are in conflict
with the temperatures of about 150®C predicted
from conductive thermal models and studies of the
metamorphic rocks in ancient subduction com
plexes. These data can be reconciled by the upward
flow of water in the prism decreasing the thermal
gradients to much lower than those at the surface
below a depth ofa few kilometres. Dewatering ofthe
Barbados prism at the decollement zone creates a
unique chemical environment in which clay minerals
of distinctive character are formed (Schoonmaker,
1986). Dewatering also produces characteristic au-
thigenic carbonates derived from the methane-rich
pore waters ofthe Washington-Oregon accretionary
prism (Ritger et al., 1987). The methane, which is
dominantly of biogenic origin,is carried to the upper
sediments of the prism by fluids and oxidized by
sulphate reducers before being incorporated into a
carbonate cement of magnesian calcite, dolomite
and aragonite at the sites of pore water expulsion.
Direct observation by submersible ofthe accretionary
prism of the Nankai Trough off southern Japan (Le
Pichon et al.^ 1987) has revealed the presence of
unique benthic communities which exist by chemo-
synthetic processes along the active tectonic zone at
the leading edge of the prism and whose shells
contain carbonate derived from methane.
Knowledge of the structure and models of evolu
tion of subduction complexes have allowed new
interpretations to be made of ancient complex sedi
mentary terrains in terms of subduction-related pro
cesses. One such terrain is the Southern Uplands of
Scotland (Fig. 8.29a)(McKerrow et al.^ 1977; Leg-
gett et«/., 1982; Leggett, 1987). This area has been
interpreted in terms of an accretionary wedge devel
oped at a subduction zone. In late Ordovician times,
Laurentia (early Palaeozoic North America, includ
ing Scotland) was separated from Europe by the
lapetus Ocean (Fig. 8.30). lapetus subsequently be
gan to close as the result of underthrusting at the
southeast margin of Laurentia. A subduction com
plex grew here by sediment accretion which was
partly overlain by a foreland basin on its landward
side. Laurentia eventually collided with Europe in
late Silurian times, giving rise to a continent-
continent collision represented in the stratigraphic
record as the Caledonian Orogeny, and welding
Scotland to England. The subduction complex is
now preserved as the Southern Uplands and the
forearc basin as the Midland Valley of Scotland
(Fig. 8.29a). Figure 8.29c shows a reconstructed
section ofthis area in Wenlock times, and its similar
ity to models of mdoern subduction complexes is
readily appreciated.
8.8 Volcanic and piutonic activity
Where subducting oceanic lithosphere reaches a
depth in excess of 80 km, an island arc is created at
the surface by volcanic and piutonic activity 150-
200 km from the trench axis. The majority ofisland
arcs are situated along the western and northern
margins of the Pacific, but are also located in the
western Atlantic at the Caribbean and Scotia arcs. At
the eastern edge ofthe Aleutians an island arc gradu
ally transforms into an Andean-type orogenic belt off
Alaska (Fig. 8.17).
Relatively young island arcs are structurally simple
and are underlain by a crust of less than 20 km
thickness. These include the arcs of Tonga-
Kermadec, the New Hebrides, Aleutians and Lesser
Antilles. Older, mature island arcs are more complex
 SUBDUCTION ZONES 161

Devonian granites

Upper Palaeozoic

□ Ordovician and Silurian

Caradoc

Site of Southern
Upland Fault
NW SE

Sea level

(b)
Wenlock

Site of Midland
Valley inliers Site of Southern Site of
Upland Fault Solway Firth

Cockburnland
Fig. 8.29 (a) Map of the Southern v:-:: Sea level
Uplands of Scotland showing the
locations of the seven sequences of Present
(b) and (c); (b) Southern Uplands level of
accretionary prism in Caradoc erosion
times; (c) Southern Uplands
accretionary prism in Wenlock (c)
times (redrawn from McKerrow et
V.E.X2
al.^ 1977, with permission from 10 km
Nature 267, 237-9. Copyright © Northern belt Central belt Southern belt
1977 Macmillan Magazines Ltd).
162 CHAPTERS
VVV V
vvvvvvvv
1000 km
I
lapetus Ocean
Fig.8.30 Possible configuration ofthe lapetus Ocean
and bordering continents in late Ordovician times
(redrawn from Leggett etal., 1982,in Trench-Forearc
Geolqgy^ pp. 495-520, with permission from the
Geological Society).
as they were constructed upon previous generations
of subducting plate margin products. They are gen
erally underlain by thicker crust of 20-35 km thick
ness, and include the Japanese and Indonesian arcs.
Three series of volcanic rocks occur in the island
arc environment(Baker, 1982):
1 The low potassium tholeiitic series, which is dom
inated by basaltic lavas associated with lesser volumes
ofiron-rich basaltic andesites and andesites.
2 The calc-alkali series, dominated by andesites
(Thorpe, 1982) which are more enriched in potas
sium, other incompatible elements and light rare
earth elements. In Andean-type belts dacites and
rhyolites are more abundant.
3 The alkali series, which includes the subgroups of
alkaline basalts and shoshonitic lavas.
The tholeiitic series is found in young arcs. The
Continent
Fe (max), heavy REE, Y, K/Rb, Na/K, Sr®7/Sr8^ FeO/MgO, SiOj
K, Rb, Sr, Ba, Cs, P; Pb, U, Th, light REE, Th/U, Rb/Sr, La/Yb
Carbonates
-z'zi CJastics
E3 Volcanics
Land
Trench
tholeiites have been interpreted as being derived by
the fractional crystallization of olivine from a pri
mary olivine tholeiite magma originating at relatively
shallow levels of 80-120 km. The calc-alkaline and
alkaline series are encountered in more mature arcs,
and are believed to be derived from magmas gener
ated at increasingly greater depths. Indeed, in some
mature arcs a compositional trend is apparent, with
increasing distance from the trench, of tholeiite-
calc-alkine-alkaline volcanic rocks, which represent
magmas derived from increasingly greater depths
(Fig. 8.21). There is also a systematic variation in
abundance, and in elemental and isotope ratios
(Fig. 8.31), which correlates with the depth to the
Benioflfzone.In mature island arcs plutonic rocks are
exposed which represent the residua of magma cham
bers which have crystallized at depth. They are gener
ally of granodioritic composition and exhibit similar
variations to the volcanic rocks.
In the past there has been controversy over the
origin of the magmas supplying island arc volcapic-
ity. Certainly the generation ofthe magmas must be
linked to the BeniofF zone, as there is a very strong
correlation between its depth and the systematic
Fig.8.31 Systematic variation ofelement
Ocean
abundance, elemental ratios and isotope ratios
across a typical subduction zone. Arrows
indicate direction ofincrease (redrawn from
Windley, 1984, with permission from John
Wiley & Sons Ltd).

variation in volcanic rock composition and elemental
abundances. It was originally believed that the mag
mas were derived from the melting of the top of the
descending oceanic slab. However, the volcanic
suites of island arcs show little variation in spite of
great differences in geographic location and tectonic
environment, so it appears that these suites must
develop from a parent with a ubiquitous, common
composition. This, petrological and mineralogical
evidence (Arculus & Curran, 1972) and helium
isotope ratios (Hilton & Craig, 1989)strongly indi
cate that the parental magmas originate largely by
partial melting of the asthenospheric mantle imme
diately overlying the descending plate. However
Karig & Kay(1981),Davidson(1983)and Hilton &
Craig(1989),among others, have demonstrated that
certain isotopic ratios require a large contribution
from continent-derived sediments. Consequently,
sediments from the trench must be carried down the
subduction zone and incorporated into the asthen
ospheric melt(Plank & Langmuir, 1993).
A major problem is the source ofthe heat required
for melting the asthenosphere above the descending
slab. It was originally believed that this was derived
by shear heating at the top of the slab. This is
unlikely, however, as the viscosity of the astheno
sphere decreases with increasing temperature, and at
the temperatures required for partial fusion the asthe
nosphere would have such a low viscosity that shear
melting could not occur. Ringwood (1974; 1977)
has therefore suggested that partial melting takes
place at a relatively low temperature because of the
high water vapour pressure resulting from the dehyd
ration of various mineral phases. Indeed, the greater
the amount of water present, the more the melting
temperature of the mantle is reduced (Stolper &
Newman, 1994).
At depths shallower than 100 km amphibolite
transforms to eclogite and releases water (Fig.
8.32a). The water rises into the asthenosphere wedge
overlying the Benioffzone causing partial melting of
the pyrolite and the production of tholeiitic basalt
magmas at depths of 80-100 km. These magmas
fractionate as they ascend and give rise to the basaltic
andesite and andesite magmas typical of the tholeit-
tic suite present in young island arcs.
At depths in excess of 100 km water is released by
the dehydration ofserpentinite bodies in the oceanic
crust (Fig. 8.32b). The resulting high water vapour
pressure causes partial melting of the quartz eclogite
SUBDUCTION ZONES 163
at depths of 100-200 km producing acidic magmas.
These react with the overlying pyrolite of the asthe
nosphere forming diapirs which rise and undergo
partial fusion. During their ascent they fractionate to
produce the calc-alkaline series of andesite, dacite
and rhyolite typical of mature island arcs. For a more
detailed account of island arc petrogenesis, see
Wilson (1989).
Ringwood's models thus imply that island arc
magmas originate from both the top ofthe descend
ing plate and the overlying asthenosphere. The resid
ual asthenosphere has been irreversibly differentiated
and will never again be able to participate in basaltic
magma generation at ocean ridges. Ringwood esti
mates that 30-60% of the asthenosphere has been
affected in this way.
8.9 Marginal seas
Marginal seas (back-arc basins) are small ocean basins
lying on the inner, concave, sides of island arcs,
bounded on the side opposite the arc by a back-arc
ridge ('third arc', remnant arc). They are most com
mon behind the island arcs ofthe western Pacific, but
are also found in the Atlantic behind the Caribbean
and Scotia arcs. Tarney & Windley (1981) have
presented evidence for their existence throughout
the geological record from Archaean times.
The crust beneath marginal basins is similar in
composition to that ofthe ocean basins, although in
some cases layer 1 is somewhat thicker as would be
expected in a small, enclosed basin. Several genera
tions of marginal basin may occur behind any given
island arc, and Karig (1971) has classified them
according to their heat flow (Fig. 8.33) in terms of
active marginal basins with high heat flow, inactive
basins with high heat flow and inactive basins with
normal heat flow. Active basins always occur imme
diately adjacent to island arcs. The high heat flow
experienced in these basins (Watanabe et aL^ 1977)
correlates with the region of very low (g,in the upper
mantle (Fig. 8.7) and supports the interpretation of
this region as a zone of partial melting.
Karig (1970) has interpreted marginal basins as
developing in response to tensional tectonics,
whereby an existing island arc was rifted along its
length and the two halves, corresponding to the
volcanic and third arcs, separated to give rise to the
marginal basin. This model was deduced from
the Tonga-Kermadec arc(Fig. 8.34)which is flanked
164 CHAPTERS

Island arc
tholeiite series

crust

-Serpentinlte — Basalt, gabbro, ^ —

,nno™ bodies_ greenschists, amphibolite
Differentiating-— ^
hydrous tholeiiticj'''' "
— magmas ^ ^ Sinking peridotiticlithosphere

\ '\ / / //

Partial \\-\20'.,a
melting of}'///'
pyrolite
^;s ^/
< ^ ^£Sgy -- - -/:=«

Eclogite / ^

Calc-alkaline
volcanics

Peridotitic
lithosphere
/ ^ ^
Serpentinite^
/
— —' Y';/>-SegregatiQn and /
/ / / ^
— — T JZ/VA-differentiation of
— — —^ . calc-alkaline
-magmas — Amphibolite ^
— 7 / / / / /
Diapiric uprise \ / / / / /
and partial

^
melting of .
pyroxenite ^\\
'\
\
/ ^ ^ ^ ^ ^^
/
High PhjO
Wet garnet 1
pyroxenite'

/>

Fig.8.32 (a) Early phase of development of an island arc. London^ 1974, with permission from the Geological
(b)Later phase ofisland arc development(redrawn from Society).
Ringwood, 1974,in Journal ofthe Geological Society of
 SUBDUCTION ZONES 165

- 60°N

Aleutian Basin

f ' •«.
"Kamchatka Basin
Bowers Basin

Okhotsk Basin

Shikoku Basin

Bonin Zone - 30°N

Active marginal basin
(inter-arc basin)
Inactive marginal basin
with high heat flow
West
Parece ^5^ Inactive marginal basin
Philippine Vela with normal heat flow
Basin Basin
South ^\ Trench
} China
:■ Basin Mariana Trough

Sulu Basin

Celebes Basin
Bismarck Basin

Solomon Basin

^Coral Basin
^i. j.
Fiji Plateau

New Hebrides

South Fiji
Basin
- 30°S

7 Lau-Havre
^ Trough
i

Tasman Basirf

120°E 150°E 180°

Fig. 8.33 Distribution of marginal seas in the western

Pacific (redrawn from Karig, 1971, with permission from
the American Geophysical Union).
166 CHAPTERS
South Fiji Ejasin
3 Contour interval
2000 m
170" 175" New Zealand 180'
on its landward side by the Lau-Havre Trough and
Lau-Colville Ridge. Evidence for extension comes
from: the asymmetric cross-section of both arc and
ridge, which are mirrored across the centre of the
marginal basin, and the fact that the basin's topo
graphy is aligned parallel to both; the considerable
sediment thickness present to the seaward side ofthe
arc and landward side ofthe ridge and the absence of
sediment within the basin; and the continuation of
the arc-basin-ridge system to the south which corre
lates with the Taupo volcanic region of the north
island of New Zealand, a zone of active extension.
Further support for this model comes from the
subsidence generally experienced by remnant arcs
after the development of marginal basins, as their
dynamic support is removed.
Pacific
Fig.8.34 Bathymetry of the Tonga-
Kermadec region (after Karig, 1970,
with permission from the American
Geophysical Union).
Tamaki(1985) has noted that the initial rifting of
a back-arc basin takes place at the island arc, as this is
the most ductile part of the system. The dip of the
subduction zone controls the nature ofthe split, with
a single rift forming within the narrow volcanic zone
associated with a steeply dipping Benioffzone, and a
multirift system forming in the wider zone associated
with shallow angle subduction.
The oceanic nature of the crust underlying mar
ginal basins implies that it is generated in a manner
similar to that occurring at the ocean ridges, and
older theories involving the thinning of continental
crust have been abandoned. Indeed, a magma cham
ber beneath the Lau Basin has been identified on
seismic reflection projects (Collier & Sinha, 1992).
However, structures corresponding to an accretive

margin are not commonly present within marginal
seas. Moreover, magnetic lineations are not devel
oped in all marginal basins (Weissel, 1981). Those
lineations that are present can usually be correlated
with the magnetic polarity timescale (Section 4.6),
although they tend to be oflower amplitude and less
clearly defined than oceanic anomalies. These data
support the proposal that marginal crust formation
takes place over a relatively broad, diffuse zone of
volcanic and plutonic activity.
Although some workers have suggested that cer
tain marginal basins have been constructed from
leaky transform fault systems, four possible mecha
nisms (Karig, 1974) have been put forward to ex
plain the evolution of the majority of marginal seas
(Fig. 8.35). Karig (1971) has suggested that mar
ginal sea extension has resulted from the forcible
intrusion of basaltic mantle diapirs generated by
partial melting at the top of the descending slab
(Fig. 8.35a) in the same manner as magmas which
erupt at the volcanic arc (Section 8.8). Other work
ers, such as Packham & Falvey (1971), have sug
gested that this magma upwelling is passive rather
than active (Fig. 8.35b), and occurs in response to
other factors which have caused regional extensional
stresses in the back-arc region. These factors include
the trench suction force (Section 11.6) which some
workers, such as Chase (1978) and Fein & Jurdy
(1986), believe originates from a 'roll-back' of the
trench. Such a roll-back is believed to occur in sub-
duction systems where the 'absolute' direction of
movement of the overriding plate is away from the
trench. Carlson & Melia (1984) have calculated the
rates of roll-back in a hotspot reference frame, and
have found values of 100 mm a"^ at the Tonga arc,
50-70 mm a"^ at the South Sandwich arc,zero at the
New Hebrides arc and negative values, that is, an
advancing trench, at the Izu-Bonim Trench. Hsui &
Toksoz (1981) and Jurdy & Stefanick (1983) have
suggested that marginal sea extension is a result of
secondary convection cells in the mantle wedge over
lying the Benioff zone which have been induced by
the descent of the underthrusting slab (Fig. 8.35c).
Such convection is feasible provided that the upper
mantle viscosity is low and/or the superadiabatic
temperature gradient high (Section 11.5.2). How
ever, there is little geological evidence of the crustal
compression that would be produced either side of
the median line of marginal seas by this mechanism,
and focal mechanism solutions indicative of com-
SUBDUCTION ZONES 167
pressional events are rare. Figure 8.35d shows the
formation of a marginal basin by a step-back in the
location of underthrusting. The cause of this behav
iour is unknown,but its effect is to isolate a fragment
ofunderthrusting oceanic lithosphere and to accrete
it to the leading edge ofthe overriding plate. Dvorkin
etal.(1993)have suggested that back-arc basins only
form behind trenches where narrow slabs are sub
ducted, that is, slabs which are restricted in width in
the direction parallel to the trench. Such slabs allow
the ingress of asthenosphere above the slab from
both the end and sides ofthe slab, so overcoming the
hydrodynamic suction believed to support wide
slabs. An alternative model,unrelated to subduction,
has been proposed by Miyashiro(1986)for marginal
seas in the western Pacific. He suggested that the
spreading of these back-arc basins was related to the
northward migration of a hot region in the mantle
from Australia in the Late Cretaceous to its present
location around north China. In this way Miyashiro
explains the general northerly younging of the mar
ginal seas, their location and short duration of back-
arc spreading. He cannot explain, however, the
presence of up to three generations of basins behind
any given island arc.
The mechanism shown in Fig. 8.35d can be iden
tified as having occurred behind island arcs by dating
the remnant arc or marginal basin crust. If these
features are older than the frontal arc, a step-back in
underthrusting has probably occurred. If, however,
the back-arc structures are contemporaneous with
the arc,some continuous process in this region must
be active. Both the Bering Sea and eastern Caribbean
Sea in all probability formed by a step-back in under
thrusting. The Bering Sea crust exhibits magnetic
lineations typical ofocean basins, but these strike at a
high angle to the trend ofthe Aleutian arc(Cooper et
1977). The lineations can be dated and related to
anomalies in the Pacific to reconstruct the events
preceding the formation of the Bering Sea. This
technique and other geological evidence suggest that
the present plate subduction at the Aleutian Trench
occurs 400-1000 km farther south than in Mesozoic
times, and that the Bering Sea was formed by the
initiation of a new subduction zone in Late Creta
ceous times(Section 9.5). The Caribbean case will be
discussed in Section 8.10.
More commonly, marginal basins appear to have
formed contemporaneously with underthrusting at
their parental island arc, and the correlation of active
168 CHAPTERS

Lithosphere

Mesosphere

-,, JC - V-> ';"'/ 0/.\ '\"i- i^'N'i7,

^r-S-:
'
== w

Remnant arc
A

nvA<'5ir ifr.Jh/ wW^-WV vtM'f w,:f:


Indo-Australian Plate
Fig. 8.36 Schematic model of deformation in the region
between the New Hebrides and Tonga arcs. Heavy arrows,
relative motion between Pacific and Indo-Australian
plates; open arrows,'pull-apart basin'; shading, areas of
marginal basins with underlying regions of very high
attenuation of seismic waves (Section 8.3) provides
support for models which invoke either the active or
passive intrusion of diapirs into the back-arc region.
However, the criticism of such models is that the
zone ofextension apparently remains in the centre of
the marginal basin and migrates away from the volca
nic arc. The cessation of activity in a marginal basin,
and the development of successive generations of
marginal basins, may be related to times when the
whole of the mantle wedge above the Benioff zone
had undergone partial melting and the supply of
materials for the creation ofnew back-arc lithosphere
had been exhausted.
It is probable that certain marginal seas have
evolved in response to specific configurations ofplate
boundaries in their vicinity. Hamburger 8c Isacks
(1988) have compiled earthquake data from the
Fig. 8.35 (Facingpage) Models for the formation of
marginal seas: (a) active diapirism generated by Benioff
zone;(b) passive diapirism from regional tensional
stresses;(c) subsidiary convective circulation;(d) step-back
in underthrusting. No vertical exaggeration (after Karig,
1974,Annual Review ofEarth and Planetary Sciences 2,
Copyright © 1974 by Annual Reviews Inc.).
SUBDUCTION ZONES 169
Pacific Plate
known active deformation; diagonal ornament, stable
areas (redrawn from Hamburger & Isacks, 1988, with
permission from Nature 332, 599-604. Copyright ©
1988 Macmillian Magazines Ltd).
southwestern Pacific in the region behind the island
arcs of the New Hebrides and the Tonga Islands, and
have found that the deformation is diffuse and dom
inated by strike-slip mechanisms. This is very dif
ferent from the linear earthquake belt with normal
faulting mechanisms that would be expected if the
back-arc basin developed in a manner similar to the
ocean basins at accretive plate margins. Volcanism is
similarly diffuse in this region, and the concept of a
single spteading centre is not supported by the sea
floot bathymetry and magnetic anomaly data. Ham
burger 8c Isacks consider that this back-arc region has
developed in response to the opposite directions of
underthrusting at the Tonga and New Hebrides
trenches (Fig. 8.36), in which deformation takes
place by shearing and localized crustal generation at
irregular spreading centres of short duration in the
whole region between the island arcs. The entire
interarc region is viewed as a large, internally de
formed pull-apart basin (Section 7.4.3)formed along
a 700 km long sinistral offset of the transform fault
boundary between the Pacific and Indo-Australian
Plates. It is possible that other back-arc basins which
show no evidence of a discrete spreading centre and
lie in a region of complex plate boundaries, such as
the Tyrrhenian Basin and Sea of Japan, developed by
170 CHAPTERS

diffuse spreading of a similar nature. A major prob East Pacific rise (a)
lem is the reason why marginal seas have developed Pacific West CoastTrench
preferentially in the western Pacific and not in the Plate Farallon Plate \ American Plate
east. The answer may lie in the rates at which the \ \-/' / N-,'.' t

lithosphere moves over the asthenosphere as de

duced from a hotspot frame of reference (Section
5.4). Wilson & Burke (1972) have suggested that
marginal seas only develop behind subduction zones
which flank plates that are stationary or slow moving
with respect to the asthenosphere. Volcanism
\l/ >0/
Although true oceanic marginal basins are not
developed adjacent to trenches along the eastern
edge ofthe Pacific, there are regions ofthe continen
tal lithosphere adjacent to these trenches which ap
pear to have been subjected to a tensional stress
regime. One such region is the Basin and Range
Province of the western USA. The region is charac
terized by a well defined zone of high heat flow, thin San Andreas Fault
crust, anomalously low Q in the upper mantle and
extensive normal faulting. Scholz et al. (1971) have
suggested that the tectonics of the Basin and Range
Province are due to the diapiric rise of partially
molten mantle material generated by the under-
•\i/ \i/ (d)
thrusting ofthe oceanic Farallon Plate from the west
in Early Cenozoic times (Fig. 8.37). The diapir
would have been incapable of penetrating the strong Present
0 200 400
continental lithosphere, and so would have flattened 1 L
and spread at its base. When subduction ceased as a
result ofthe East Pacific Rise reaching the trench,the Fig.8.37 Inferred Cenozoic evolution ofthe Basin and
San Andreas Fault was initiated (Section 5.9), com Range Province (redrawn from Scholz 1971, with
pression of the North American Plate resulting from permission from the Geological Society of America).
the underthrusting would have ceased, and the ex
tension resulting from the diapir would have been
enhanced. of back-arc spreading and stress release across a con
A further example of ensialic extension in this vergent plate boundary. The Rocas Verdes Complex
environment is provided by the Rocas Verdes com has been cited as the modern analogue for the forma
plex of the Chilean Andes. In southern Chile ophio- tion of Archaean greenstone belts (Tarney et al.
lite complexes have been interpreted as part of the 1976;Section 12.2.3 below).
floor ofa Cretaceous marginal basin(de Wit& Stern,
1981). The basin widens from north to south
(Fig. 8.38). In the north basaltic magmas intruded
8.10 Example:the Lesser Antilles
subduction zone
continental crust, causing extensive mobilization
and reconstitution. In the southern, wider, section The Lesser Antilles (east Caribbean) arc (Fig. 8.39)
basaltic magmas were emplaced in a localized exhibits all the features typical of the island arcs. It
oceanic-type spreading centre. Consequently, within forms the eastern margin of the Caribbean Plate at
this region there is a range of back-arc floor from which Atlantic Ocean lithosphere is subducting at a
crust intermediate between continental and oceanic rate of about 20 mm a" The Caribbean Plate is
to typically oceanic. This range was interpreted as currently welded to the South American Plate, and
representing various stages in the evolution of a relative movement between the Caribbean and
back-arc basin which formed during the interaction North American plates is taken up along sinistral

Marginal basin
Island arc
\ Dyk^orientation
Andesitic volcanoclastid'
100 km s '/!
Fig.8.38 Reconstruction ofthe Rocas Verdes complex of
southern Chile in Early Cretaceous times (redrawn from
de Wit & Stern, 1981, with permission from Elsevier).
transform faults along its northern margin and to the
east ofthe Lesser Antilles.
The topographic expression ofthe trench is largely
obscured by a thick sediment fill derived from the
Orinoco River of Venezuela, but its presence is
indicated by a belt of negative free air and isostatic
gravity anomalies (Kearey a /»/., 1975). These sedi
ments have been thrust and deformed into a large
accretionary wedge over 20 km thick known as the
Barbados Ridge(Westbrook,1982;Westbrook et al.,
1984), which emerges at the island of Barbados
(Fig. 8.40). Between this ridge and the island arc lies
a forearc basin of tranquil deposition known as the
Tobago Trough. The island arc stretches from Som
brero in the north to Grenada in the south, and has a
history ofcalc-alkaline volcanism since the Eocene. It
exhibits the very small radius ofcurvature of400 km.
In the north the arc is diflferentiated into an outer
sedimentary arc and an inner volcanic arc. These
SUBDUCTION ZONES 171
Remnant arc
Chile' Patagonia
Rise Soi^th Georgia
Tierra del Fuego
Drake
Passage
South Sandwich
arc
Antarctic Peninsula
merge at the island of Guadeloupe, and to the south
the islands consist of volcanics on the western flank
and sediments on the eastern. In the south the oldest
rocks are ofEocene age, while in the north the oldest
rocks on the volcanic arc are Oligocene and the
sedimentary islands are underlain by rocks ofEocene
age. There is no evidence anywhere ofa substructure
older than Eocene in age.
A back-arc basin, the Grenada Trough, flanks the
inner side of the island arc, and is bounded to the
west by the Aves Ridge (Kearey, 1974), an almost
totally submarine bathymetric prominence. The east
ern edge ofthe Aves Ridge is noticeably arcuate, with
a radius ofcurvature very similar to that ofthe island
arc. The ridge is of volcanic origin, and samples of
igneous rocks recovered by dredging and drilling
provide ages no younger than Cretaceous and Palae-
ocene. The Aves Ridge consequently has many ofthe
characteristics of a remnant island arc. Moreover, in
order to explain the sequential nature of volcanism
from the Aves Ridge to Lesser Antilles, it probably
originated by a step-back in underthrusting
(Fig. 8.3Sd) rather than by any process of back-arc
spreading.
172 CHAPTERS

Volcanic islands
Sombrero
I I Sedimentary islands
18°N
100 km

St Kitts

Guadeloupe

16°N
6
Dominica

Martinique

l!)
14°N
St Lucia

St Vincent Barbados

Grenada 12°N

1000 fathom isobath

64°W 62°W 60°W

Fig.8.39 Lesser Antilles island arc, eastern Caribbean.


Aves Ridge Grenada Trough
(Remnant arc) (IN^arginal sea)
100
I I
km
Fig.8.40 Section through the eastern Caribbean (after
Westbrook, 1982,in Trench-Forearc Geolq0y, pp. 275-90,
with permission from the Geological Society).
Further reading
Baker,P.E.(1982)Evolution and classification oforogenic
volcanic rocks. In Thorpe, R.S. (ed.) Andesites^ pp. 11-
23, Wiley, New York.
Barazangi, M. & Isacks, B. (1971) Lateral variations of
seismic wave attenuation in the upper mantle above the
inclined earthquake zone of the Tonga island arc: deep
anomaly in the upper mzntit.].£feophys. Res. 76,8493-
516.
Barber, A.J.(1982)Interpretations ofthe tectonic environ
ment ofsoutheast Japan. Proc.£ieol. Ass. 93,131-45.
Hasegawa, A., Umino, N. & Takagi, A. (1978) Double-
planed seismic zone and upper mantle structure in the
northeastern Japan arc. Geophys. J. Roy. astr. Sac. 54,
281-96.
SUBDUCTION ZONES 173
TobagoTrough Atlantic Ocean
(Fore-arc basin) (Oceanic crust)
Lesser Antilles j Barbados Ridge
(Island arc) (Subduction complex) E
200
Hasemi,A.H.,Ishii, H.& Takagi, A.(1984)Fine structure
beneath the Tohoku district, northeastern Japan arc, as
derived by the inversion ofP-wave arrival times from local
earthquakes. Tectonophys. 101, 245-65.
Karig, D.E. (1971) Origin and development of marginal
basins in the western V2X\^c.J.£feophys. Res. 76,2542-61.
Karig, D.E.(1974)Evolution of arc systems in the western
Pacific. Ann. Rev. Earth planet Sci. 2, 51-75.
McKerrow, W.S., Leggett, J.K. 8c Eales, M.H. (1977)
Imbricate thrust model of the Southern Uplands of
Scotland. Nature 267, 237-9.
Miyashiro, A. (1973) Paired and unpaired metamorphic
belts. Tectonophys. 17, 241-54.
Molnar,P., Freedman, D.& Shih, J.S.F.(1979)Lengths of
intermediate and deep seismic zones and temperatures in
downgoing slabs oflithosphere. Geophys.J. Roy. astr. Soc.
56,41-54.
Moore, J.C.(1989) Tectonics and hydrogeology of accre-
tionary prisms: role of decollement zone.J. struct. Geol.
11,95-106.
Ringwood, A.E. (1974) The petrological evolution of
island arc systems.J.^eol. Soc. Lond. 130,183-204.
9/ Mountain ranges
9.1 General
The subduction of oceanic lithosphere gives rise to
two different types of orogenic belt depending upon
the nature of the overriding plate (Dewey & Bird,
1970)(Fig. 9.1). Subduction beneath oceanic litho
sphere results in the formation ofan island arc and its
associated tectonic features (Fig. 9.1a)(Section 8.1).
Subduction beneath continental lithosphere gives
rise to a linear mountain belt on the overthrusting
plate margin which runs parallel to the subduction
zone (Fig. 9.1a). Such mountain belts are generally
termed Andean-type mountain ranges, the term
Cordilleran-type no longer being in general use
since the recognition that the Western Cordillera of
North America formed by rather more complex
processes (Section 9.5) than implied by this simple
model. Island arcs and Andean-type mountain
ranges consequently appear to represent the response
of the lithosphere to continued, steady state sub
duction, and are likely to be relatively long-lived
features.
Collisional mountain ranges (Fig. 9.1b) develop
when, as a result of subduction adjacent to a conti
nental margin,lithospheric material is brought to the
trench which has a density sufficiently low that its
buoyancy prevents its descent. This anomalous ma
terial is commonly a continent, but may also be
lithosphere carrying the thickened crust of an island
arc, or a microcontinent making up an integral part
ofthe subducting oceanic lithosphere (Section 9.5).
The resulting collision of buoyant crust causes
the creation of a collisional mountain range by the
stacking ofthrust slices ofcrust(Molnar, 1988),and
brings to a close an episode of sea floor spreading.
Lateral movements in the welded continents may
continue, however, for a considerable time interval
after collision (Section 9.3.6). Collisional mountain
ranges are complex since the geological record of
collision is always preceded by Andean-type oro
genesis.
174
9.2 Andean-type mountain ranges
9.2.1 Introduction
The Andes typify the relatively simple mountain
belts resulting from the underthrusting of oceanic
lithosphere adjacent to a continental margin. There
is scant evidence of accreted terrains (Windley,
1984) (Section 9.5). However, the geology of the
whole Andean chain is not known in detail and there
are significant changes in lithospheric structure and
behaviour along the mountain belt.
9.2.2 General geological structure of the
Peruvian Andes
The Andes in Peru (Cobbing & Pitcher, 1972)com
prise two subparallel fold belts (Fig. 9.2). The West
ern Cordillera is of Mesozoic-Tertiary age and the
Eastern Cordillera of Late Palaeozoic age. In south
ern Peru, where the fold belts diverge, they are
separated by the Altiplano, which consists of a thick
sequence ofTertiary molasse.
In the south the Palaeozoic Eastern Cordillera is
made up ofa thick black shale-quartzite assemblage,
while in the north it comprises greenschist facies
pelites of pre-Ordovician age associated with
gneisses, which may represent reworked Precambrian
basement.
The Mesozoic-Tertiary Western Cordillera is di
vided longitudinally into an eastern sedimentary
trough of folded clastic and carbonate rocks ('mio-
geosyncline') and a relatively undeformed volcano-
sedimentary trough ('eugeosyncline'). From lea to
Trujillo the sediments of the Western Cordillera are
so thick that their base is not seen, while south of
Ica regionally metamorphosed rocks form fault-
bounded inliers within the younger rocks. The Phan-
erozoic rocks are thus underlain by old crystalline
rocks and the Andean mountain range is founded on
old continental crust.

New
Australia TasmanSea Hebrides Pacific Ocean
(a)
Indian Ocean India
(b)
Fig.9.1 (a) Section from Australia through the Pacific to
the Atlantic Ocean illustrating island arc and Andean plate
margins,(b)Section from the Indian Ocean to Tibet
across the Himalayan mountain range (redrawn from
Dewey & Horsfield, 1970, with permission from Nature
225, 521-5. Copyright © 1970 Macmillan Magazines
Ltd).
The volcanosedimentary trough of the Western
Cordillera includes the massive Coastal Batholith.
This is a multiple intrusion of granite, tonalite and
gabbro (Cobbing & Pitcher, 1972) extending for
some 1600 km. It consists of more than a thousand
interlocking plutons intruded into the andesitic vol
canic rocks(Cobbing, 1982) via a persistent series of
deep-seated fractures. Drake et aL (1982) have esti
mated that the Peruvian continental margin has
increased in width by up to 200 km over the last
200 Ma by the emplacement ofsuccessive plutons.
The Andean Cordilleras are flanked to the east by
the sub-Andean zone, which consists of continental
sediments deposited on the Brazilian Shield, and to
the west by the Precambrian Arequipa Massif, which
may make up a large part ofthe Peruvian continental
shelf. The boundaries with the older rocks are prob
ably deep faults originating during periods of exten
sion (Myers, 1975) which were probably caused by
the trench suction force (Section 11.6).
Figure 9.3 illustrates the generalized structure of
the region. Controls on the deep structure come
from the interpretation of seismic refraction and
MOUNTAIN RANGES 175
Ar^s South America Atlantic Ocean
Continental
crust
Oceanic
crust
Subcrustal
lithosphere
Tibet
Asthenosphere
gravity data, which show that the base of the crust is
at 12 km below sea level beneath the ocean, at 76 km
under the Altiplano and probably at 40 km beneath
the Brazilian Shield.
9.2.3 Seismicity of the Andes
The general pattern of seismicity and focal mechan
ism solutions of earthquakes are in accord with the
easterly subduction of the Nazca plate beneath the
Andes(Molnar & Chen, 1982). However,a detailed
assessment of the most accurately located events
(Barazangi & Isacks, 1979) indicates that the dip of
the BeniofFzone is not constant(Fig. 9.4). In central
and northern Peru the seismic zone dips shallowly at
about 10" while in southern Peru the dip is 30". The
transition between these two areas must conse
quently be represented by a tear zone or distorted
downgoing plate which would allow subduction to
take place at different angles. Barazangi & Isacks
(1979) and Jordan et aL (1983) have noted that the
dip of the BeniofF zone appears to control the topo
graphy and geology ofthe overlying areas. Above the
shallow dipping plate there is no significant Quater
nary and Neogene volcanism, there is a steady in
crease in topography from the coast to the crest of
the Andes and shallow seismicity occurs over a broad
region. Above the more steeply dipping plate there is
Neogene and Holocene volcanism, a longitudinal
valley is developed between the Eastern and Western
Cordilleras which contains the deposits of the
176 CHAPTER 9

Eugeosynclinal
volcaniclastics

Miogeosynclinal
elastics

Thin cover of platform Mesozoic-Tertiary

sediments mobile belt

Intrusives

Altiplano molasse

Ordovician-Devonian
Dm flysch trough
Pre-Ordovician
schist belt

Precambrian craton

Fig.9.2 Geology ofthe Peruvian Andes (redrawn from

Pitcher, 1978,in the Journal ofthe Geolo£fical Society of
London^ with permission from the Geological Society).

Western Altiplano
Cordillera
"- Moho
Tertiary volcanic
Tertiary molasse Palaeozoic
-=3 Mesozoic Batholith
Fig.9.3 Schematic section through the Peruvian Andes
(redrawn from Cobbing & Pitcher, 1972, with permission
from Nature Phys. Set. 240, 51-3. Copyright© 1972
Macmillan Magazines Ltd).
Altiplano, and the seismicity is confined to a rela
tively narrow BeniofFzone.
The presence of a shallow subduction zone is
interesting as it implies that the continental litho-
sphere has a thickness in this area not exceeding
130 km.Such a thickness would mean that the top of
the descending oceanic plate occurs at the base ofthe
continental lithosphere. There would then be no
wedge of asthenospheric material above the BeniofF
zone. This wedge may be necessary for the develop
ment of volcanic activity, and its absence in this
region may explain the lack of volcanicity associated
with shallow subduction. However, this situation
may not apply to other parts ofthe Andes(Megard &
Philip, 1976).
The cause of the flat geometry of the subducted
plate is not known. The Nazca Plate is younger than
the oceanic plates in the west Pacific and may thus be
thinner and more buoyant. It is most probable that
underthrusting is affected by the subduction of oce
anic rises, the Nazca Ridge in the north and Carnegie
Ridge in the south whose buoyancy might inhibit the
descent of the Nazca Plate (Section 9.5). Whatever
the cause,this flat geometry gives rise to deformation
in the South American plate extending inland some
700 km from the trench, as demonstrated by the
thrust faulting in the sub-Andean zone (Snares et al.,
1983).
MOUNTAIN RANGES
Eastern Sub-Andean
Cordillera fault zone
;V'| Precambrian basement
n Oceanic crust
9.2.4 Plate tectonic model of the Andes
James(1971) has presented the following model for
the plate tectonic evolution of the Andean mobile
belt (Fig. 9.5), which argues for the presence of a
continuous subduction zone in Phanerozoic times
(Cobbing,in Pitcher et a/., 1985).
During late Palaeozoic times the South American
shelf and slope were covered by sediments up to
15 km thick (Fig. 9.5a). Some of these deposits were
derived from continental crust situated to the west
whose origin and mechanism of subsequent dis
appearance are obscure. Marine deposition in the
Eastern Cordillera ceased before the Permian. Un
derthrusting of the Nazca Plate started in late
Triassic-early Jurassic times and pre-dated the sepa
ration of South America from Africa (Fig. 9.5b). A
volcanic arc developed to the west of the Palaeozoic
continental shelf, with the eruption of basalts and
andesites, and minor granite plutonism. During late
Cretaceous-early Tertiary times a major orogenic
event occurred to the east of the site of Triassic-
Jurassic activity, and the major part of the Andean
batholith was emplaced by the invasion of granitic-
intermediate melts (Fig. 9.5c) derived principally
from mantle sources(Atherton, 1984).The horizon
tal compression resulting from this emplacement
was transmitted to the crust to the east and caused
the folding and uplift of the Eastern Cordillera. The
molasse of the Altiplano, which reaches a thickness
of 7 km, was then deposited. The volcanism and
plutonism that began in Miocene times record the
modern phase of igneous activity. The compressive
178 CHAPTERS

81°W
S60W 13030's
\ 550 ^ ^ _ 1100km _

S 200-

300 4 A

Dominant surface
■*■ compression •*-
+5j
- -sc
(D E o4
VT
^ 51

I 100-

g- 200-
Q

1100 km

E
100

Q.
a>
200
Q

300

Dominant surface
— extension — Fig. 9.4 Cross-sections along profiles
A and B, shown in Fig. 9.2, showing
earthquake foci and interpretative
models. In the models, the dashed lines
— 100-
under the Moho (M) show the
-c 200-
approximate position of the base of the
o. South American Plate (redrawn from
0>
Q 300- Megard & Philip, 1976, with
permission from Elsevier).

stress generated by magma intrusion caused folding
and thrusting of the Altiplano and Eastern Cordillera
(Fig. 9.5d). During Cenozic times the Altiplano re
ceived some 15 km of molasse from Eastern and
Western Cordilleras. During late Quaternary, the
igneous activity diminished, compressive stresses
relaxed and graben formed within the Altiplano,
probably in response to the trench suction force
(Section 11.6).
The crustal rocks of the Andes are thus interpreted
as deriving from the melting of the downgoing slab
and overlying mantle, and the crustal thickening has
resulted from the intrusion of magmatic products
from this region. In addition, Molnar & Chen
(1982) suggest that thickening has also arisen from
crustal shortening by thrusts, and infer that the
present seismic activity indicates that this process is
Fig. 9.5 (Facingpa^e) Schematic sequence of the
evolution of the Peruvian Andes. M, Moho; I, presumed
isotherm at which melting occurs in the lithosphere
(redrawn from James 1971, with permission from the
Geological Society of America).
 MOUNTAIN RANGES 179

Distance (km)
200 400 ^0 800 1000
—I— —I—

Late Palaeozoic 50

Lithosphere
(a) 100

Triassic-
^
,/
+ + + + + + + + + + + + + —W-—
+ + + + + + + + + + + + + + + + + + + + + + + + +
t * * * * * + + + + + + + + + + + + + + + +*
++++++ J -
Jurassic
50

1
100

(b) -"150

^ ✓ N'
\ T'
' + + .
Cretaceous- - +
+
+ + + + + +
+ + + + + + + + + +
+ + + + +
+
+
+
early Tertiary + + + + + + + + + + +
• i + + + + + +. ^
^

M 50

100

(c)
150

Western Eastern
cord

Pliocene-
Pleistocene

- 100

Asthenosphere

Lithosphere -150
2 X Vert,exag
(d)

Pre-Palaeozoic Sedimentary

Gneiss Calc-alkaline rocks

180 CHAPTERS
still continuing. The present rate of subduction is
70-120 mm a"^.
There are objections to James's plate tectonic
model (Helwig, 1973) which Zeil (1979) has exam
ined in detail. These include the absence of blue-
schist facies rocks except in the northern Andes, the
variation in time between compressional and ten-
sional episodes, the presence of seismic gaps and
variable dips in the Benioff zone and the question
able validity ofextrapolating recent seismic data back
in time. James's model consequently represents a
working hypothesis which must be critically tested
against continuing field and laboratory investiga
tions. Certainly, the Andes cannot be considered to
have formed by a continuum process, as there have
been considerable changes in the orogen with time.
The history ofthe Andes cannot be related to uniform
subduction, but rather to a variety of processes which
gave rise to different characteristic features in dif
ferent sections of the orpgen. Ramos (1988), for
example, has interpreted the history of the central
Andes in terms of a model which involves the west
wards migration ofthe trench since Palaeozoic times
and the accretion ofexotic terrains (Section 9.5).
9.3 Collisional mountain ranges
9.3.1 Introduction
When an oceanic plate is subducted beneath conti
nental lithosphere, an Andean mountain range de
velops on the edge of the continent (Section 9.1). If
the subducting plate also contains continental litho
sphere, continued underthrusting causes conver
gence of the continents, and eventually brings them
into juxtaposition (Fig. 9.1b). Whereas oceanic
lithosphere is relatively dense and sinks into the
asthenosphere,the greater sialic content ofcontinen
tal lithosphere causes it to possess a positive buoy
ancy in the asthenosphere which prevents its being
subducted any great distance (McKenzie, 1969).
Consequently, the arrival of continental lithosphere
at a trench results in collision with the overriding
continent; rapid relative motion is halted, and a
collisional mountain range is formed by crustal
shortening (Fig.9.1b). According to classic plate
tectonic theory, no further relative movements be
tween the continents are possible, and the forces
responsible for collision, if they remain active, may
cause the initiation ofan oceanic subduction zone at
the new continental margin. This process has been
documented in the Indian Ocean, where a new plate
boundary is developing in response to the collision
ofIndia and Asia(Van Orman 1995). It will be
shown later (Section 9.3.6), however,that the behav
iour of continents after collision is rather more
complex.
The surface marking the locus ofcollision is known
as a suture. The suture zone may preserve slivers of
the oceanic lithosphere which formerly separated the
continents, known as ophiolites (Section 2.5).
The Himalaya represent the youngest collisional
mountain range, in which continental collision
began in Tertiary times. Older mountain ranges are
represented by the Appalachians, the Caledonides,
the Alps and the Urals. Several of the mobile belts
present within Precambrian shields may also have
originated in this way.
It is probable that mechanisms operating in the
Himalaya today are analogous to those which af
fected these older mobile belts and mountain ranges.
Consequently, this section will concentrate on the
evolution of the Himalaya. It has only recently be
come possible to investigate the geology and geo
physics of this region, as it has only been in the last
few years that it has been opened to Western investi
gators. Before considering the Himalaya in detail,
however, a number of theoretical models of colli
sional mountain building will be examined.
9.3.2 Models of continental collision
Several models ofthe evolution of collisional moun
tain ranges have been proposed, and with the accu
mulation of increasing knowledge of these features,
greater geological and geophysical constraints can be
used to discriminate between them.
A review of models for the evolution ofthe Hima
laya and Tibetan plateau has been given by Powell &
Conaghan (1975), and these are shown in Fig. 9.6.
Figure 9.6b shows a model in which both converg
ing plates are equally affected by folding and thrust
ing,so that the resulting orogenic belt is symmetrical
about the suture. This model is discounted because
the main Himalayan suture does not occupy a central
position within the mountain belt.
Figure 9.6c shows a model first proposed by Ar-
gand (1924), and suggests that the suture between
India and Eurasia developed beneath a cover of
Tethyan sediments. Deformation was thought to
 MOUNTAIN RANGES 181

Lower and
higher
S Himalaya . Tibetan Plateau
(a)
Sub
Himalaya
^"u1o^TangWa_^ChangThangPlatformH- p'^oldB^elt-H- Stable^ri sban^
Basin l^oldBelt
- Tibetan Dzuhgarian
Brahmaputra Himalaya Stable Block
Sea level
js ^ J-
Base of crust

Central
Thrust

ndia

Tethys

.. . T-... o.. Chiangtang

^ Southern Northern Tibetan Plate pigt© Tsaidam
Tibetan Plate ^ \ ^ ^ ^ Plate

Indian Plate

ndia
Aiianl31iBi-r^
f^r'^'atpr Indj

100 200 300 400 500

(km)
No vertical exaggeration

Fig.9.6 Different models suggested for the formation of
the Himalaya. S, postulated suture between India and Asia
(after Powell & Conaghan, 1975, with permission from
the Geological Society of America),(a)Surface geology
and nomenclature;(b) orogenic belt symmetrical about
suture;(c) plastic deformation beneath Tethyan sediments
(after Argand, 1924);(d) basement reactivation model
(after Dewey & Burke, 1973, with permission from the
Geological Society of America);(e) multiple block
collision model (after Chang & Cheng, 1973);
(f) underplating model (after Powell & Conaghan, 1973
and Mattauer, 1986);(g) limited underplating model
(after Molnar, 1984).
182 CHAPTERS
occur by plastic deformation and thrusting of the
upper crust, which resulted from the underthrusting
ofIndia beneath Asia.
Figure 9.6d is based on the basement reactivation
model of Dewey & Burke (1973). They envisage a
thickening of the Asian crust following collision by
thrust faulting in the upper crust and ductile creep
in the lower crust. The increased heat flow of the
thickened crust then causes partial melting and
the buoyant ascent ofa melt rich in potassium and the
lithophile elements. The enriched upper crust would
be underlain by a depleted lower crust ofanorthositic
composition, with the boundary at mid-crustal
depths being represented by a zone of migmatites.
This boundary may well represent the Conrad seis
mic discontinuity (Section 2.4.1), and it is interest
ing to speculate that the Conrad discontinuity is only
developed in continental crust which has been af
fected by collision, thus explaining its absence else
where. This model is inadequate to explain the
evolution of the Himalaya, however, as most of the
crustal thickening has occurred within the Indian
Plate and not beneath Tibet. Also, if the increased
heat flow beneath Tibet is caused by the presence of
subducting oceanic lithosphere, even a very shallow
dip of 10® for the subduction zone would only affect
the southern half ofTibet (Fig. 9.6d).
Figure 9.6e, based on the model of Chang &
Cheng (1973), attempts to explain the presence of
many ophiolite belts within the Himalaya by the
successive collision of several small crustal blocks
after the progressive closure of several small oceans
during the Phanerozoic. The model is based on
radiometric dates which span Phanerozoic time and
which young southwards. This mechanism success
fully explains the thickening of the underthrusting
southern plates, but cannot explain why uplift ofthe
Himalaya did not occur until Tertiary times. This
model is also highly relevant to the accretion of
suspect terrains to mainland Asia (Section 9.5).
Figure 9.6f is based on the models of several
workers (e.g. Powell & Conaghan, 1973; Mattauer,
1986) who believe that the thickened crust beneath
the Himalaya has resulted from the underthrusting
ofIndian crust beneath the crust of Asia rather than
by penetrative deformation ofthe crust. This appears
to be supported by the seismic tomographic analysis
(Section 2.1.8) of Roecker (1982) and the thrusts
and double thickness crust revealed by a seismic
reflection profile (Zhao et 1993). The great
elevation of the Himalaya is then attributed to the
isostatic response ofa thickened,low density crust. A
problem ofthis type of model is that some process of
'mantle peeling' is required beneath the Asian block
so that Indian crust can be emplaced directly beneath
Asian crust rather than the base of the Asian litho
sphere. This problem is obviated, however, by the
model of Molnar (1984) in which only a limited
amount of Indian underthrusting is postulated. No
crust-mantle decollement is required, and crustal
thickening takes place in response to sequential
thrusting along the Main Central Thrust and Main
Boundary Fault (Fig. 9.6g). This model is also con
sistent with the seismic studies ofHirn (1995),
which suggest that no Indian lithosphere is present
at depth north of the Indus-Zangbo Suture Zone
(Section 9.3.3) and that crustal thickening is caused
by imbrication. A major criticism ofall models ofthis
type which involve the underthrusting ofIndia is that
they appear to contravene the assertion of McKenzie
(1969) that continental lithosphere is too buoyant
to subduct. Subduction might be achieved if the
continental crust were greatly attenuated, but this
would probably only affect relatively small areas.
Subduction would be facilitated, however, if the
crust were delaminated from the lithospheric mantle
along the brittle-ductile transition at the level ofthe
Moho predicted by some theories of the rheologic
structure of the continental lithosphere (Section
2.10.3). Butler (1986) has suggested that if crustal
thickening is accomplished by thrusting along a
detachment climbing from the Moho (Fig. 9.7), the
base of the footwall section of the crust would be
depressed below the level at which the granulites of
the lower crust undergo a phase change to eclogite.
The increased density of the basal eclogite would
then decrease the buoyancy of the crust and allow
limited subduction to take place, although long
distance underthrusting does not appear to have
occurred (Change?At/., 1986).
9.3.3 Himalayan geology
The Himalayan mountains are 250-350 km wide
and extend for about 3000 km from Afghanistan to
Burma. They comprise a series of lithologic and
tectonic units which run parallel to the mountain
belt and maintain a constant character for long
distances (Fig. 9.8). Summaries of the known geol
ogy are to be found in Molnar(1984) and Windley

lihmMoho'itV»ulm^ui{Muwilly"' jiir(AUi.Bukiljjjii6lW^^
(a)
Foreland Hinterland
basin basin
^Grar^ljtg^;-yXVo^
(b)
^ root
(c)
Fig.9.7 Thrusting ofthe lithosphere. Stipple, crust;
thin dashed line, hypothetical phase boundary,
(a) Undeformed template;(b) movement on a thrust
system with a crust-mantle detachment (cf. western
Himalaya);(c) development of a thrust which climbs
across the crust-mantle boundary (cf. Alps)(redrawn
from Butler, 1986,in the Journal ofthe Geological Society
ofLondon^ with permission from the Geological Society).
(1985). The geology provides evidence for the previ
ous existence of oceanic lithosphere (preserved as
ophiolite bodies), a passive continental margin, an
island arc and Andean-type batholiths. The various
geological units will be considered in sequence from
south to north. A generalized cross-section is given
in Fig. 9.9.
The Himalaya are undergoing rapid uplift at rates
between 0.5 and 4mma~^. They are consequently
experiencing rapid erosion, with the deposition in
the Sub-Himalaya of a thick terrigenous sequence
dating from the Miocene.These comprise the Siwalik
molasse conglomerates and extend southwards into
the Ganga Basin where they are eventually bounded
by the Main Frontal Thrust.
The Lower(or Lesser) Himalaya occupy elevations
from 1500 to 3000 m and are thrust over the Sub-
Himalaya along the Main Boundary Thrust. The
thrust is still active and focal mechanism solutions
suggest that the fault plane dips northwards at a
MOUNTAIN RANGES 183
shallow angle. The Lower Himalaya consist of
Precambrian-Mesozoic low grade metasediments
which are overthrust by nappes of gneisses of the
Higher Himalaya.
The Higher (or Greater) Himalaya reach altitudes
ofover 8000 m and consist ofa basement ofPrecam-
brian gneiss overlain by Palaeozoic and Mesozoic
sediments of Tethyan origin which were originally
situated on the northern margin ofIndia. The unit is
thrust over the Lower Himalaya along the Main
Central Thrust for a distance exceeding 100 km.The
unit is intruded by many granites of Miocene age
which originated by melting of the lower crust.
Several large ophiolite klippen were thrust over the
sediments in Palaeocene times.
The Indus-Zangbo Suture Zone is the major
boundary separating the Precambrian Indian Plate
from the younger Mesozoic-Cenozoic Trans-
Himalaya to the north. The suture, which is a steeply
dipping thrust zone, contains Tethyan ophiolites,
blueschists and granulites. The ophiolites are not
continuous, and in places are replaced by sediments
typical of the forearc environment. Virtually all the
material ofthe Himalayan range,that is, south ofthe
suture, was once part of the Indian Plate and not
derived from the Asian Plate.
The Trans-Himalaya occur to the north of the
Indus-Zangbo Suture Zone in the central eastern
Himalaya where the Southern Tibetan Plate with a
cover of Palaeozoic and Mesozoic sediments is in
truded by the Cretaceous-Eocene Trans-Himalayan
granite batholith (or Kangdese granites) which
formed along an Andean-type plate margin in re
sponse to northwards underthrusting of the Tethys
Ocean. In the western Himalaya the equivalent
unit is an island arc which formed within the Tethys
in mid-Cretaceous times, ultimately to become
squeezed between the converging Indian and Asian
plates at about 100 Ma. This unit is separated by the
minor Northern Suture (Bangong-Nujiang Suture)
from the Karakorum granite batholith to the north.
A further east-west suture (Kokoxili Suture) sepa
rates the Northern and Southern Tibetan plates.
Focal mechanism solutions of earthquakes in the
Himalaya (Fig. 9.8) reveal the present style of fault
ing (Molnar & Chen, 1982). A single event in the
Indian Shield to the south is indicative of normal
faulting along an east-west plane, and is consistent
with the extension experienced in this area as a result
of the bending of the Indian Shield beneath the
184 CHAPTER 9

Karakorum Batholith
Trans-Himalayan Batholiths
and volcanics

Ophlollte klippe
_ Higher
Tethyan sediments ,
Himalaya
Miocene granites J
Central crystallines ]
Sediment, crystallines I Lower
and granites i Himalaya
Outer crystalline klippe
Siwaliks — Sub-Himalaya

Fig.9.8 Geological map of the Himalaya (redrawn from
VVindley, 1985 with focal mechanism solutions of
earthquakes from Molnar, 1984). IZS,Indus-Zangbo
Suture; MCT, Main Central Thrust; MET, Main
Boundary Thrust; MFT, Main Frontal Thrust.
Himalaya. Thrust faulting occurs beneath the Lower
Himalaya, with the fault planes probably dipping
north at shallow angles. Thrust faulting gives way to
normal faulting on north-south planes in the Higher
and Trans-Himalaya, and probably reflects the in
creased loading of the lithosphere, causing the verti
cal stress to pass from being the least to the most
compressive stress. The overall pattern of fault plane
solutions appears to be compatible with a model in
which the Indian Plate bends downwards in front of
the Himalaya and descends beneath them.
9.3.4 Deep structure of the Himalaya
Bouguer anomalies become increasingly negative
northwards from India. This indicates that the crust
thickens in this direction and that the Himalayan
mountains are at least partly compensated by an Airy
type of compensation (Section 2.11.2).
The most detailed imaging of crustal structure has
come from a 500 km-long seismic traverse across the
Himalaya and Tibet in which emphasis was placed on
the recording of critical reflections from the Moho
(Hirn et a-L, 1984a; 1984b). A line drawing of the
interpretation of these events is shown in Fig. 9.10.
The data show that the crustal thickness ofthe Indian
Shield is some 35 km, and that this increases to
55 km beneath the Himalaya and to 70 km under
Tibet. The crust-mantle boundary, rather than being
a sharp contact, occupies a transition zone over a

+ Sub-Himalaya + Lower Himalaya
Outer Central
Siwaliks crystalline crystallines
klippe
Sediments
Upper crus
Upper mantle
50 km
Fig.9.9 Diagrammatic section across the central
Himalaya. Symbols as for Fig.9.8 (redrawn from Windley,
1983,in the Journal ofthe Geological Society ofLondon^
with permission ofthe Geological Society).
vertical interval of some 12 km. The phase has a
velocity of8.5 km s"\ indicating that normal mantle
material is present with no temperature anomalies so
that no isostatic support originates from hot, low
density material in the upper mantle.
The Moho topography is not smooth (Fig. 9.10),
and exhibits a number of steps, some of which are
vertical and some in which there is an overlapping of
a Moho at different depths. All steps produce a Moho
displacement of less than 20 km. The vertical dis
continuities are associated with the Indus-Zangbo
Suture and the Danqiao Ophiolite, and probably
indicate that these features are now strike-slip faults
associated with indentation tectonics (Section
9.3.6). The stacked Moho segments probably repre
sent the thrust faults which were responsible for the
thickening of the Himalaya-Tibet crust. The thick
ness ofthese thrust slices is less than 20 km,which is
considerably less than the normal crustal thickness.
This implies that crustal thickening does not take
place by simple underthrusting of Indian crust be
neath Eurasia (i.e. underplating) in which the Moho
is smoothly depressed northwards. Rather, it sug
gests that thickening has taken place in response to
MOUNTAIN RANGES 185
Higher Himalaya + Trans-Himalaya
Miocene Ophiollte Indus - Cretaceous
granite klippe Zangbo batholiths in
Suture Palaeoz sed.
Tethyan
Sediments
Melting
zone
intracontinental thrusting affecting both the crust
and upper mantle.
Lyon-Caen & Molnar(1983) have interpreted the
gravity field of the Himalaya by assuming that India
slides beneath Eurasia and that the Indian Plate flexes
elastically (Section 2.11.4) under the weight of the
overthrust Himalayan mountain range (Fig. 9.11).
The only unknowns in this interpretation are the
flexural rigidity of the lithosphere, density contrasts
between sediments,crust and mantle,the location of
the end of the elastic plate and the bending moment
applied to its end. Figure 9.11 shows an interpreta
tion which has optimized the correspondence of
observed and calculated gravity anomalies by varying
these parameters. In the light ofthe seismic reflection
data described above, this model is incorrect in that
India does not descend smoothly and coherently
beneath Eurasia. However, the model successfully
simulates the observed gravity field, and it is possible
that modification of the model might well produce
an interpretation consistent with the seismic reflec
tion results.
9.3.5 Evolution of the Himalaya
The Himalaya formed in response to the collision of
India with Eurasia. This collision was brought about
by the northward migration ofthe Indian Plate as the
intervening Tethyan Ocean was subducted beneath
the southern margin ofthe Eurasian Plate. Magnetic
186 CHAPTER 9
High indus-Zangbo
Himalaya ophioiites
50 km j
70km -i
Fig.9.10 Moho reflections beneath Tibet (after Hirn et
«/., 1984b, with permission from Nature 307, 25-7,
Copyright © Macmillan Magazines Ltd).
anomalies in the Indian Ocean and palaeomagnetic
measurements on the subcontinent confirm the
northerly drift of India and allow the reconstruction
of its path (Fig. 9.12). The initial continental colli
sion took place about 50 Ma ago and marks the
cessation of marine sedimentation and the com
mencement of terrestrial deposition in the suture
zone. All oceanic lithosphere had disappeared by
45 Ma, and at about 36 Ma there was a decrease in
the velocity of India's northward drift from over
ICQ mm a"^ to about 50mma~^. This probably
marked the last stage of true continental collision
and the initiation of a regime of indentation tecton
ics (Section 9.3.6).
Magnetic lineations can only provide a relatively
coarse model of the India-Eurasia convergence, and
Observed
N Greater Ganga
' I Himalaya Basin
g^Load -Sediments
100 km
Namu Hu Danqiao
Xaenza ophioiites ophioiites
g - 50 km
1- 70km
geological observations (Allegre et al., 1984) can
provide more detail to the sequential movements.
These indicate that the main continental conver
gence was preceded by the collision of two smaller
plates (Fig. 9.13). The North Tibetan Plate became
welded to Eurasia by about 140 Ma and was followed
by a step-back in the subduction zone to the south of
North Tibet. The underthrusting brought the South
Tibetan Plate into juxtaposition with North Tibet
and eventually welded North and South Tibet to
gether. A further step-back in underthrusting pre
ceded the main collisional event. India collided with
southern Tibet, which then represented the southern
margin of the Eurasian Plate, 50 Ma ago, with the
formation ofthe Indus-Zangbo Suture.
Continued convergence at a slower rate led to the
progressive formation of the Main Central Thrust
and Main Boundary Thrust, accompanied by some
2000 km of crustal shortening.
It is thus apparent that the Himalaya contain at
Computed
Indian S
Shield
Fig.9.11 Observed and computed
Bouguer anomalies over a flexnral
model of the Himalaya (redrawn
from Lyon-Caen Sc Molnar, 1983,
with permission from the American
Geophysical Union).

40''
30°
20°
10°
10°
20°
30°
w 2.1.6) has revealed the pattern of faulting in the
100°
Fig.9.12 Northward drift ofIndia with respect to Asia
from 71 Ma to the present, determined from magnetic
iineations in the Indian and Atlantic oceans (redrawn
from Molnar & Tapponier, 1975, Science 189,419-26,
with permission from the AAAS. Copyright © 1975 by the
AAAS).
least two microcontinents (North and South Tibet)
which became welded to the Eurasian Plate before
the main collision. North and South Tibet therefore
represent suspect terrains (Section 9.5), and it is to
be expected that most collisional orogens will con
tain units of this type.
9.3.6 Indentation tectonics
After rifting from Gondwana, India drifted north
wards towards the main Asian continent as the inter
vening Tethyan Ocean contracted by subduction
MOUNTAIN RANGES 187
beneath southern Tibet. Collision ofthe two contin
ental masses took place about 45 Ma ago,resulting in
the formation of the Himalayan mountain range.
Classical plate tectonic theory predicts that, after a
period of crustal shortening, relative movement
between the two continents should cease because any
significant underthrusting ofcontinental lithosphere
is prevented by its buoyancy (McKenzie, 1969).
Further north-south movements of the lithosphere
would then be accommodated by the development
of a new trench to the south of India and consump
tion of oceanic lithosphere at a northerly dipping
subduction zone. This prediction is not fully real
ized. The Himalayan region is still seismically active,
is undergoing rapid uplift and there remains an area
of active tectonics within a broad zone stretching
some 3000 km north ofthe mountain chain (Molnar
& Chen-, 1982). It is estimated that India is still
moving northwards at a rate ofsome 45 mm a"^, and
has penetrated at least 2000 km into Asia. Clearly,
such continued convergence requires significant de
formation of the continental lithosphere so that the
interacting plates cannot act in a completely rigid
manner.
A combined interpretation of satellite images and
focal mechanism solutions of earthquakes (Section
Himalayan region shown in Fig. 9.14. Thrust fault
ing is restricted to a relatively narrow belt north of
the Himalayan Frontal Thrust. Strike-slip faulting
dominates a region some 1500 km wide north ofthe
Himalaya and extending east into Indo-China. Far
thest from the mountain chain is a region of crustal
extension and normal faulting extending from the
Baikal region of Siberia to the northern China Sea.
An analogy has been drawn between this pattern of
faulting and the lines offailure developed in a plastic
medium when it is indented by a rigid die(Tapponier
& Molnar, 1976). The theory of indentation of
plastic materials has been developed by mechanical
engineers,and it is possible,for simple shapes ofboth
plastic medium and die, to predict mathematically
the configuration of lines of failure, or slip lines.
When this form of analysis is applied to large scale
geological processes the phenomenon has been
termed indentationy extrusion or escape tectonics. In
this case,India is equated with the rigid indenter and
Asia with the plastic medium, whose initial elastic
response to strain is ignored.
The slip lines developed on indentation are termed
188 CHAPTER 9
50°S 40 30 20 10
S.Tibet
Deccan Traps
Fig.9.13 Possible sequence of events in the evolution of
the Himalaya. KS, Kokoxili Suture; BNS,Bangong
Nujiang Suture; IZS,Indus-Zangbo Suture; MCT,Main
Central Thrust; MBT,Main Boundary Thrust(redrawn
from Allegre etal., 1984, with permission from Nature
307,11-22. Copyright © 1984 Macmillan Magazines
Ltd).
a and P lines, which correspond to dextral and
sinistral strike-slip motion, respectively. The pattern
ofslip lines is controlled by the shape ofthe indenter
and the lateral constraints placed on the plastic
medium. Figure 9.15 shows the patterns developed
when the plastic medium is semi-infinite. A flat
indenter (Fig. 9.15a) produces a symmetrical pattern
of slip lines extending into the plastic medium a
distance approximately equal to the width of the
die. Wedge-shaped indenters (Figs9.15b,c) produce
more complex arrangements of slip lines. Figure
9.15b appears to simulate the structure in the west
ern Himalaya and Pakistan where the Indian Plate's
boundary has a wedge-shaped geometry. The Herat
10 20 30 40 ?in°N
N. Tibet
140 Ma
IZS^x t
mot\x V
MBTx^,
L/'y.i'!: Present
and Altyn Tagh faults run approximately parallel to
the a and P lines, respectively, generated by this form
of indenter. The asymmetry of the indenter, whose
northeastern edge impinges more obliquely than the
northwestern/edge, may explain why deformation
occurs over a very large area to the northeast of
the Himalaya. However, a triangular indenter
(Fig.9.15c) may represent a more appropriate model
to explain this phenomenon.In this case the indenter
has penetrated a significant distance into the plastic
medium. The Kangting Fault, a remarkably curved
feature on satellite images,follows a p line, while the
linear Red River Fault follows an a line. This model
can also explain how underthrusting can simulta
neously take place in an easterly direction at the
Burman arc (Fig. 9,14). The subduction results from
the material shouldered aside to the east by the
northerly motion ofIndia.
A further class of models ofthis type is obtained if
plastic material of finite width is indented, as shown
in Fig. 9.16. In this case the region ofdeformation is
bounded by the pair of slip lines which leave the
 MOUNTAIN RANGES 189

70°E 80°E 100°E 120°E 140°E 150°E 50°N

TWl
Strike-slip fault
[)s Motions with respect
Siberia

Normal fault
1^ Direction of extension
Oceanic crust
Thrust fault in SE Asia

60°N

Siberia
V 150°E
WI*. 30°N

50°N

Mongolia
f
140°E
20°N
40°N
China

30°N

thrust

China•§:
Sea i 130°E

20°N

Borneo
10°N

120°E

80°E 100°E

Fig.9.14 Schematic map illustrating extrusion tectonics

in eastern Asia. Numbers associated with arrows, extrusion
phases(1 ~ 50-20 Ma;2 20-0 Ma)(redrawn from
Tapponier^f 1982, with permission from the
Geological Society of America).
190 CHAPTERS
cx lines
p lines
a lines
(b)
Fig.9.15 Plane indentation ofsemi-infinite rigid-plastic
media by different rigid dies, with names corresponding to
tectonic features:(a) flat rigid die;(b) rigid wedge;(c)flat
triangular indenter and hollowed-out medium (redrawn
from Tapponier & Molnar, 1976, with permission from
Nature 264, 319-24. Copyright © 1976 Macmillan
Magazines Ltd).
corners of the indenter and intersect at the opposite
side of the plastic material. At this intersection the
plastic material is thrown into a state oftensile stress,
and it is possible that the crustal thinning and normal
faulting observed in the Baikal and Shansi regions
result from this process.
The indentation models considered above are un
doubtedly reasonably successful in providing expla
nations ofthe orientation and sense of movements of
major faults and the gross tectonic style ofthe Hima-
Mongolia
Fig.9.16 Plane indentation of a bounded rigid-plastic
medium (redrawn from Tapponier & Molnar, 1976, with
permission from Nature 264, 319-24. Copyright ©
Macmillan Magazines Ltd).
layan orogen. However,the mathematical analysis of
slip line fields is complex, requires many simplifying
assumptions and is limited to relatively simple geo
metries of indenter and plastic medium. Moreover,
in many cases ofindentation, the region of deforma
tion increases with time so that the boundary condi
tions of the problem are similarly time-dependent.
Consequently, without the use of unjustifiably com
plex incremental models,the slip line fields produced
are instantaneous and cannot be used to trace the
evolution offaulting.
These problems can be obviated to a certain extent
ifthe mathematical models are replaced by a mechan
ical analogue. Tapponier et al. (1982) have con
structed a mechanical device in which a block of
laminated Plasticine about 300 x 300 mm is de
formed laterally by a rigid indenter 50 mm wide
advanced at a constant rate. Figure 9.17shows evolu
tionary sequences of the indentation of a block of
Plasticine which is either bilaterally confined at the
two edges parallel to the motion of the indenter
(Fig.9.17a-c) or unilaterally confined at only one of
these edges (Fig. 9.17d-f). The bilaterally confined
case produces a symmetrical pattern of slip lines
ahead of a 'dead triangle' which rapidly welds to the
indenter, and the penetration proceeds by the ere-

20 mm
Fig.9.17 Indentation experiments
on Plasticine. F, major fault
(redrawn from Tapponier et al.^
1982, with permission from the
Geological Society of America).
ation of numerous, short-lived, dextrai and sinistrai
faults near the apex of the triangle. The unilateral
case provides asymmetrical deformation in which
faults allowing displacement towards the free edge
predominate, such as F^. The block translated side
ways rotates about 25 ® clockwise and is followed by
the extrusion of a second block along a further
sinistrai fault, F2, which allows a continued rotation
of the first block by up to about 40°. Numerous
pull-apart troughs (Section 7.4.3) develop along the
sinistrai faults because oftheir irregular geometry. As
these movements progress, a gap grows between the
indenter and extruded Plasticine.
Tapponier et al. (1982) suggest that many of the
MOUNTAIN RANGES 191
1 +2
tectonic features ofthe Himalayan region (Fig.9.14)
can be explained by this experiment,for example:
1 sinistrai oflfsets in central China are much greater
than dextrai;
2 pull-apart structures along the sinistrai faults may
be analogous to the extensional regimes in Shansi,
Mongolia and Baikal;
3 the Altyn Tagh Fault may correlate with the major
dislocation F2, and its bending in the west may be
caused by the effects ofthe indenting Indian Plate;
4 the Red River Fault maycorrelate with Fi,and the
succession of activity between the Altyn Tagh and
Red River faults might explain many ofthe complex
ities in the evolution ofsoutheast Asia.
192 CHAPTERS
The concept of indentation tectonics is clearly
successful in explaining the spatial distribution and
relationship of the various styles of faulting and
deformation in the Himalayan region. In applying
the results of indentation studies to deformation of
the lithosphere, however,it should be borne in mind
that several simplifying assumptions are made which
are not completely true for natural Earth materials.
The deformation resulting from indentation is as
sumed to occur in a homogeneous plastic material.
The lithosphere does not exhibit such a simple theol
ogy,and the upper crust certainly stores strain energy
elastically, as evidenced by the presence of earth
quakes. In mitigation, at greater depths creep is
probably a dominant mechanism of deformation.
A major criticism of the indentation model is that
it does not predict, or take account of, the variable
thickness of the deformed plastic layer, in which
thickening occurs in response to thrusting, and thin
ning in response to extension. England (1982) and
England & McKenzie (1982) have adopted a com
pletely different approach to the India-Asia collision
(a)
■llllll
(C)
in which the deformation of Asia is viewed as the
response of a continuous viscous medium to both
the edge forces of continental collision and the
internal forces generated by differences in crustal
thicknesses. This continuum model takes no account
of faulting, but it is argued that fault-induced de
formation is small and restricted to the upper brittle
layer, so that over long intervals of time the ductile
lower crust is the major control of deformation. The
lithosphere is not regarded as rigid, but simply as a
layer of much greater strength than its substrate.
England (1982) has shown that horizontal forces
are generated by regions of elevated and thickened
crust. He has combined these with the edge forces
arising from continental collision and, in order to
account for the observed deformation of a region
such as central Asia, has determined the required
rheological properties of the viscous sheet that is
believed to represent this region. Solution of the
response of this sheet to the applied edge and inter
nal forces then provides the evolutionary trends in
the distribution ofdifferent styles ofdeformation for
Fig. 9.18. Style of deformation in
a thin viscous sheet caused by the
stresses shown in (a) after (a) 8 Ma;
(b)16Ma; (c) 32 Ma; and (d)
40 Ma. Black areas, regions of
compression; vertical and
diagonally hatched areas, regions of
compression and extension,
respectively, both with significant
strike-slip; dotted areas, regions of
predominantly strike-slip motion
with components of compression
( +) and tension ( -) (redrawn from
England, 1982, with permission
from Academic Press).

Region of major crustal thickening
I' I Region of strike-slip faulting
Region of normal faulting
and crustal thinning
Fig.9.19 Distribution of tectonic styles in Asia (redrawn
from Tapponier & Moinar, 1976, in Nature 264,319-
24, with permission. Copyright © 1976 Macmillan
Magazines Ltd).
a model in which collision has occurred along part of
one edge (Fig. 9.18). The black areas correspond to
regions ofcompression in which thrusting would be
dominant;the vertically and diagonally hatched areas
are regions of compression and extension, respec
tively,in which deformation has a significant compo
nent of strike-slip; the dotted areas are regions of
predominantly strike-slip with components of com
pression(+)and extension(-).
Such a geometrically simple model would not be
expected to simulate exactly the distribution of
deformation styles in such a complex region. The
model can explain in broad terms, however, the
northerly progression from thrust faulting in the
Himalaya through a region ofstrike-slip faulting to a
region of tension in northern China and eastern
Siberia (Fig. 9.19). The region of normal faulting in
Tibet, which has no analogue in the indentation
model and which occurs immediately to the north of
MOUNTAIN RANGES 193
140°
i
the main thrust front, is explained here by the ten
dency of thickened crust to spread laterally.
In contrast to the indentation model, in which a
plastic layer of constant thickness progressively de
forms because of the steady advance of an indenter,
the continuum model places greater emphasis on the
role of forces arising from crustal thickening. The
initial continental convergence results in compres-
sional deformation and thickening ofthe crust in the
region of the suture. Eventually the crust reaches
such a thickness that its buoyancy exerts lateral forces
large enough to deform the hinterland. The thick
ened crust is then thrown into tension and the
collisional forces deform the region outside this area
by predominantly strike-slip movements.
Neither the indentation nor the continuum model
can be completely successful in simulating the pat
tern of deformation styles of a collisional orogen,
although the presence of large strike-slip displace
ments has been proved (Tapponier et al.^ 1990).
Moreover, work by England & Houseman (1986)
on the continuum model predicts that deformation
is restricted to the region about 1000 km north and
northeast of the Himalaya, and argues against the
model of Tapponier et al. (1982) in which Indo-
194 CHAPTERS
Fig.9.20 The deforming zone between India, China and
Eurasia. Arrows show directions and approximate
magnitudes of the velocities with respect to Eurasia
(redrawn from McKenzie, 1990),
China has been laterally extruded by indentation.
Both models are forced to adopt simplifying approxi
mations and cannot fully model the complexities of
the Behaviour of continental lithosphere. However,
both models provide an important insight into the
mechanisms of deformation of continental regions,
and it is probable that a combination of both models
(Houseman & England, 1993) will prove most rea
sonable.
The whole notion of indentation tectonics in this
region has been challenged by England & Molnar
(1^90), who suggest that the major sinistral strike-
slip faults are a consequence ofthe presence ofa wide
deforming zone between India, China and Eurasia
(Fig. 9.20). Fig. 9.21a shows the indentation model
in which the faults do not rotate. Fig. 9.21b, how
ever, shows that if the faults rotate in a clockwise
sense, as suggested by recent palaeomagnetic mea
surements, their sinistral motion arises from their
lying in a broad region undergoing dextral shear in a
north-south direction. This model explains the east
ward motion of Tibet relative to Asia and India,
and its northward motion relative to southeastern
China.
Fig.9.21 Schematic representation of contrasting
interpretations offaulting in eastern Tibet. Left figures
show locations and directions of movement on idealized
faults resulting from the schemes shown on the right
Velocities are shown with respect to the solid circle,
(a) Faults do not rotate as slip occurs and material to the
south ofthe faults moves east-south-east with respet t to
the north,(b) Slip on the faults is a result of north-
trending right-lateral shear and clockwise rotation of the
blocks between the faults (redrawn from England &
Molnar, 1990).
9.3.7 The mechanism of continental
collision
The forces responsible for continental collision are
poorly understood. Clearly, the slab-pull force (Sec
tion 11.6) is only active while the intervening ocean
is shrinking, and would cease when the ocean had
been completely destroyed by subduction. However,
in the case of the Himalaya, oceanic portions of the
Indian Plate are still being underthrust at the Indo
nesian arc (Fig. 9.14). The slab-pull force in this area
may well assist the continued northerly movements
of the Indian continent(Molnar, 1984). Although a
new plate boundary would appear to be developing
within the Indo-Australian Plate to the southeast of
India (Weissel et al., 1980), the latest analysis of
current plate motions (DeMets et al., 1990) divides

(a)
Crustal low
velocity zone'
Foreland Thrust Belt
Foreland basin
Peripheral bulge
Crustal low velocity
Foreland/hinterland crust
Thickened crust of plateau
I I Thrust sheets
Fig.9.22 Schematic sections comparing (a) Alpine-type
and (b) Himalayan-type orogens (redrawn from Dewey et
al., 1986).
it into separate Indian and Australian plates. A fur
ther contributor to the continued northward move
ment of India is the ridge-push force exerted on the
southern margin of the Indian Plate.
The Himalaya are the youngest mountains on
Earth, and must represent our analogue for ancient
continental collisions. However, an examination
of schematic sections (Dewey et«/., 1986) through
the Himalaya and Alps reveals several differences
(Fig.9.22), the most important of which is the lack
of a continental plateau, and correspondingly thick
ened crust equivalent to Tibet, in the hinterland of
the Alps. It is not easy to provide an explanation for
the differences between Alpine- and Himalayan-type
collisions. It is possible that the differences arise
because of the initial presence, between India and
Eurasia, of a large intervening ocean whose subduc-
tion caused extensive relative movements. This large
oceanic area was missing from the northern margin
of Africa prior to the Alpine collision (England,
1982). Butler (1986) has suggested that the differ
ences between the Alps and Himalaya depend upon
the initial geometry of the thrust faulting experi
enced on collision (Fig. 9.7). Low angle thrusting
would have occurred in the Himalaya and much
steeper thrusting in the Alps, so that a larger volume
MOUNTAIN RANGES 195
Plateau Hinterland
of crust was thrust into the mantle and metamor
phosed to eclogite, causing it to create its own
'slab-pull' force. These differences in thrust style
might owe their origin to the forces acting on the
plates immediately prior to collision. The Indian
Plate, because ofits larger northerly flanking compo
nent of oceanic lithosphere, would possess a greater
horizontal stress component which might have led
to lower angles of thrusting.
Such suggestions are tentative, but strongly suggest
that if plate tectonic interpretations are to be made of
ancient collisional mountains, a full study must be
made of the plate regime extant for some consider
able time before collision.
9.4 Arc-continent coliision
Collisional mountain ranges can also develop as the
result of a collision between an island arc and a
continent, although the resulting mountains are
smaller than those resulting from continent-
continent collision, such as the Tertiary fold belt of
northern New Guinea(Dewey & Bird, 1970).
The sequence of events occurring in an arc-
continent collision is shown in Fig. 9.23. In
Fig. 9.23a a continent approaches an island arc by
the consumption of an intervening ocean. When
the continental margin is driven into the trench
(Fig. 9.23b), significant underthrusting of con
tinental lithosphere is inhibited by its buoyancy
(McKenzie, 1969). Thrust slices moving towards the
196 CHAPTERS
Island arc
(a)
Flysch
Molasse
Fig.9.23 Schematic sequence ofsections illustrating the
collision ofan island arc with an Andean-type plate
margin, followed by a change in the direction of
underthrusting (redrawn from Dewey & Bird, 1970, with
permission from the American Geophysical Union).
continent develop while flysch and slices of oceanic
crust are driven on to the continental margin
(Fig. 9.23c). No further lithosphere convergence is
possible, and so if the two plates continue to con
verge a new trench may develop on the oceanward
side ofthe island arc (Fig. 9.23d).
Arc-continent collision zones are relatively rare, as
they usually represent an intermediate step in the
closure of an ocean and so are relatively short-lived.
The most recent site of an arc-continent collision is
represented by the Banda arc (Fig. 9.24) (von der
Borch,1979;Bowin^f/i/., 1980;Johnston & Bowin,
1981). The Banda arc is a double arc enclosing the
oceanic crust ofthe Banda Sea.The outer arc(or first
arc—see Section 8.1)is made up ofa submarine ridge
ofdeformed sediment which emerges at the island of
Continental shelf
oceanic crust
continental crust lithosphere
I I lithospheric
mantle
mantle
iT'r \'w r
Timor. The inner volcanic arc stretches from Java to
the east ofFlores. In the west the arc is active and the
volcanism results from the northwards underthrust
ing ofthe oceanic Wharton Basin at the Java Trench.
In the east, however, north of Timor, volcanic activ
ity ceased in the late Tertiary. Strontium isotope
studies of the volcanic arc in this region are consis
tent with the involvement of continental material in
the magma genesis. Here the outer arc is underlain
by continental crust which is continuous with that of
Australia.
The structural and tectonic units of this region
originated by the underthrusting of oceanic litho
sphere northwards at both the Java Trench and
Timor Trough. About 3 Ma ago this consumption of
oceanic material brought the Australian continental
margin to the Timor Trough (Fig. 9.25). Volcanic
activity in this eastern part of the arc then ceased
when the buoyant continental lithosphere choked
the subduction zone, creating a discontinuity be
tween the Timor Trough and Java Trench. No fur
ther convergence of the Australian shelf and Banda

pFlores thrust
Sumb
Arc discontinuity J
Wharton Basin
500 km
110°E 120°E
Fig.9.24 Structural elements of the Banda Sea region.
Diagonal shading, continental or quasi-continental crust;
S-S-S,sedimentary arc; V-V-V,volcanic arc (redrawn
from von der Borch, 1979, with permission from Elsevier).
Sea is possible along the northward dipping trench at
the Timor Trough. However, it appears that the
forces driving this convergence are still active, and
SiWtretal.(1983) have discovered two large thrusts
in the back-arc region. These Wetar and Flores
thrusts are northerly directed (Figs 9.23 and 9.24),
Timor Trough
{rc Australian Continental crust ^
Fig.9.25 True scale section across
the Banda arc in the location shown
in Fig. 9.24(redrawn from von der
Borch, 1979, with permission from
Elsevier).
MOUNTAIN RANGES
ulnearoiiiiis^
Ceram
Q Australian-New Guinea shelf Ms* 10°S
Australia
15°S
130°E
and may represent the precursors of a new subduc-
tion zone caused by a polarity reversal ofthe arc.
A similar arc-continent collision zone has been
described in Taiwan by Hsu (1990).
Charlton (1991) has noted that extension of the
Banda arc collision zone commenced within 5 Ma of
the collision, and that such extension appears to be
common in the orogenic belts of eastern Indonesia.
This extension may arise following decoupling ofthe
subducted oceanic lithosphere as continental litho-
sphere chokes the subduetion zone.
Accretionary prism Banda Sea Vole arc
50 km
I
198 CHAPTERS
9.5 Suspect terrains
The linear Andean-type mountain belts(Section 9.2)
located on the margins ofcontinental plates adjacent
and parallel to oceanic trenches, were originally
thought to represent the steady state response ofthe
continental lithosphere to the continuous process of
subduction of oceanic lithosphere. If this were the
case, such mountain belts would be expected to
exhibit a consistent structure and geology along their
length, to have a tectonic and geological history
which is intimately related to the effects of subduc
tion and to have been subjected to little or no
orogenic deformation. However, since the mid
1970s, detailed integrated geological and geophys
ical investigations have shown that very large areas of
continental lithosphere situated behind trenches do
not conform to this simple picture. The continental
back-trench region is made up ofa collage or mosaic
of units ranging in size from a few hundreds to
thousands of square kilometres with distinct and
diverse geologies and structures which are not related
to present subduction regimes. The units are known
as suspect, exotic or displaced terrains. They are de
scribed in these terms as they are allochthonous and
have become accreted to the continent over a consid
erable period of time, in many cases after travelling
great distances. The process of amalgamation has
been referred to as accretionary, colla£fe or mosaic
tectonics.
A suspect terrain is a mappable unit characterized
by a geological history different from that ofadjacent
units (Jones et at.., 1983). Terrains are bounded by
normal,reverse or wrench faults, and the boundaries
occasionally preserve thin ophiolites, rocks of blue-
schist metamorphic grade or highly deformed flysch.
The recognition of suspect terrains should ideally
be based on contrasts in detailed stratigraphic his
tory, although in many cases this has been destroyed
by later tectonism and the bounding faults obscured
by later processes. Consequently, in order to deter
mine whether the geological histories of adjacent
terrains are compatible with their present spatial
relationships, extremely detailed and comprehensive
investigations are necessary (e.g., Yorath & Chase,
1981). In practice several criteria are used to distin
guish the identity ofseparate terrains(Shermer et at.,
1984), based on contrasts in:
1 the stratigraphic and sedimentary history;
2 petrogenetic affinity and magmatic history;
3 the nature, history and style ofdeformation;
4 palaeontology and palaeoenvironments (e.g.
McKerrow & Cocks, 1986);
5 palaeopole position and palaeodeclination.
Palaeomagnetic investigations have demonstrated
that many terrains have a north-south component of
motion ofseveral thousand kilometres (Beck 1980),
and have undergone rotations ofup to 60°(Simpson
6Cox, 1977; Cox, 1980).
Jones et at. (1983) have recognized four major
types of suspect terrain on the basis of studies in
western North America:
1 Stratigraphic terrains, which are characterized by
distinct stratigraphies typical of:
(a) continental fragments containing land-derived
sediments and perhaps a crystalline basement;
(b) oceanic fragments, containing lithologies typ
ical of oceanic crust, sometimes with overlying
sedimentary sequences charting travel from deep
sea to continental margin environments;
(c) island arc fragments containing intrusive and
extrusive igneous rocks and volcanigenic sedi
ments.
2 Disrupted terrains, which contain a heterogeneous
assembly of flysch, serpentinite, shallow water lime
stones and graywacke, with occasional exotic blocks
of blueschist metamorphic rocks. Some of these
terrains may have originated as subduction zone
assemblages (Section 8.7).
3 Metamorphic terrains, in which a metamorphic
overprint has destroyed the original stratigraphy.
4 Composite terrains, which comprise two or more
sub-terrains which had amalgamated prior to accre
tion to the continent.
The chronological sequence of accretion of
terrains to the continent can be determined from
geological events which postdate accretion and
link adjacent terrains (Fig.9.26). These include the
deposition of sediments across terrain boundaries
(Fig. 9.26a), the appearance of sediments derived
from an adjacent terrain (Fig.9.26b) and the 'stitch
ing' together of terrains by plutonic activity
(Fig.9.26c).
The most fully investigated region of suspect
terrains is the North American Western Cordillera
(CoTity et at., 1980). Figure 9.27 shows the distribu
tion ofthe major terrains in this region, which form a
zone some 500 km wide and make up about 30% of
continental North America. Many of the known
terrains are too small to be illustrated at the scale of

Successor basins or
overlap sequences
Provenance linking
Pluton stitching
Fig.9.26 Geological relations which help to establish the
tiining ofterrain amalgamation and accretion (redrawn
from Jones et al., 1983).
Fig. 9.27. They range in age from Precambrian to
Jurassic, and have accreted to the continent since
Triassic times when subduction was initiated at the
western margin of North America. Generally, the
more distal and oceanic terrains are thrust upon
the inner terrains or on to the craton itself(Coney,
1989). Debiche etal.(1987) have determined,from
plate motion models, the routes by which the ter
rains were carried across the Pacific basin and along
the western margin of North America. Most of the
terrains are bounded by faults and, both during and
after docking with the continent, have undergone
lateral translations along a complex series of strike-
slip faults (e.g. Cache Creek terrain. Fig. 9.27). This
process is still active, as the San Andreas Fault (Sec
tion 7.4) is currently dismembering terrains in south
west North America.
The extensive strike-slip faults that are responsible
for the dismemberment ofaccreted terrains are ofthe
type termed trench-linked, strike-slip faults by Wood
MOUNTAIN RANGES 199
cock (1986). He ascribes their formation to oblique
underthrusting where the velocity vector between
the overriding and underthrusting plates makes an
angle of less than about 35-50° with the trench
between them. They are responsible for taking up the
horizontal component of motion at the plate bound
ary, and tend to form in the overriding plate parallel
to, and about 100 km from, the trench. Woodcock
(1986) argues that motion oblique to plate margins
should be considered as the rule rather than the
exception. An example ofsuch a fault in a region not
greatly affected by collage tectonics is the Atacama
Fault of South America, which runs parallel to the
Peru-Chile Trench.
A well documented suspect terrain is Wrangellia
(Jones era/., 1977), named after the Wrangell Moun
tains in Alaska (Fig. 9.27). It comprises an upper
Palaeozoic island arc assemblage overlain by thick,
subaerial lava flows and capped by a Triassic carbonate
sequence. This distinctive geology has allowed the
identification of several fragments of Wrangellia
which are now scattered along some 2500 km of the
Cordillera, and occupy a latitudinal spread of almost
24°. Palaeomagnetic data, however, suggest tnat the
original spread was only 4°, implying that a large
amount of post-accretion fragmentation and separa
tion has occurred. The palaeolatitude of the frag
ments is centred around 10°,and is in accord with the
tropical environment suggested by their geology. It
appeats that Wrangellia may have originated in the
western Pacific near the present position of New
Guinea in Triassic times, traversed the Pacific as a
complete entity, was accreted to North America and
was subsequently fragmented and translated to its
present locations by strike-slip faulting.
The evidence from the North American Cordillera
and other areas (e.g., Pigram 8c Davies, 1987),
strongly suggests that suspect terrains originate by
the oceanic lithosphere's bringing material of conti
nental, oceanic and island arc origin into contact
with a continental plate at a subduction zone and the
subsequent thrusting of the material on to the cra
ton. Ben-Avraham et al. (1981) have suggested that
modern analogues of the material giving rise to
suspect terrains are the abundant ridges, rises and
plateaux which make up about 10% ofthe area ofthe
present ocean basins (Fig. 9.28). These topographic
highs are not associated with any currently active
plate margins, and represent extinct island arcs, sub
merged microcontinents, anomalous accumulations
200 CHAPTER 9

Sp I Principal terrains

NS North Slope
Kv Kagvik
En Endicott
R Ruby
Sp Seaward Peninsula

xP mPM
j
/
I
NF
Innoko
Nixon Fork
PM Pingston and McKinley
YT Yukon-Tanana
01 Chuiitna
P Peninsular
W Wrangellia
Cg Chugach and Prince William
TA Tracy Arm
T Taku
Ax Alexander
G Goodnews

Ch Cache Creek
St Stikine
BR Bridge River
E Eastern assemblages
Washington, Oregon and Cali'ornia
Ca Northern Cascades
SJ San Juan
O Olympic
S Si letzia
BL Blue Mountains
Trp Western Triassic and Palaeozoic
of Klamath Mountains
KL Klamath Mountains
Fh Foothills Belt
"■V
San Francisco Ti ,
Si S F Franciscan and Great Valley
C Calaveras
(_RM^ Si Northern Sierra
SG San Gabriel
Mo Mohave
Sa Salinia
Or Orocopia

?5^0r S Sonomia
RM Roberts Mountains
X USA GL Golconda

^MEXICOl
B Baja
V Vizcaino

Fig. 9.27 Generalized map of suspect terrains in western ornament, Wrangellia; diagonal ornament. Cache Creek
North America. Stippled ornament, North American terrain (redrawn from Coney rt a/., 1980, with permission
cratonic basement; barbed line, eastern limit of from Nature 288, 329-33. Copyright © 1980 Macmillan
Cordilleran Mesozoic-Cenozoic deformation; solid Magazines Ltd).
 'Chukchi '
Rise
Voering
Plateau
JS'i.' -Faeroe
' . ^^-/^Plateau
Shirshov
Ridge
Rockall
Bowers
Plateau

Umnak
Porcupine^
Alaska Q Bank
Plateau Seamounts
Flemish ^
Shatsky Cap Madiera'
Rise Hess Rise
§ Marcus
Bermuda Rise ^ Plate^
ratosthenes East ^
Plateau Chagos-Laccadive Mid .. Blake-Bahama Canary
Ridge Pacific Mts £>• Plateau Islands
Medina Aves Ridge
PI . Palau-<s^ ^ (Plateau
Kyushu Ridge Barbados Ridge
Manikiki
ySeychelles Cocos Ridge
Plateau
Bank Ontong-Jav
g^Plateau Tuamotu
\ ^^ Ridge Galapagos
Mozambique Exmouth Rise Sao Paulo
Plateau
Fiji Plateau Walvis
Nazca
Q Plateau
Ridge Ridge
Lau Ridge
Madagascar Tuba! Is Juan
Plateau ^ Broken Lord Ridge Fernandez Ridge Rio Grande
\ Ridge Norfolk Plateau Rise
Cape Agulhas South ^Rise Chatham
\ Falkland
Rise Plateau Tasman
Rise South Georgia
Rise
Campbell
® «p. Plateau ^ North Scotia Ridge

Thirty-East Maud
(
Spur Balleny Island South Scotia Bank
Peter Ridge
Island
1 ♦
20° 40° 60° 80° 100° 120° 140° 160° 180° 160° 140° 120° 100° 80° 60° 40° 20° 0°

Fig.9.28 Distribution of present day oceanic plateaux (redrawn from Ben-Avraham ^a/., 1981,Science 213,47-54, with permission from the AAAS. Copyright © 1981 by the AAAS).
202 CHAPTER 9
of volcanic rock and uplifted oceanic crust. Indeed,
Silver & Smith (1983) have suggested that the island
arc, melange zones and continental fragments found
in the Indo-Pacific region of modern southeast Asia
may resemble the configuration ofsuspect terrains of
the Western Cordillera prior to accretion. Coffin &
Eldholm (1994) have shown that many of the frag
ments are ofigneous origin and must have originated
from a complex mantle circulation coexisting with
the convection associated with plate movements.
Geophysical data and sampling have shown that
some of the oceanic topographic highs have strong
continental affinities, possessing, for example, a gra
nitic substrate. Others are of oceanic origin. Most,
however, lie on a crust two to five times thicker than
normal oceanic values. These oceanic highs situated
within shrinking oceans will ultimately be brought to
a trench. Their thickened crust implies a relatively
high buoyancy, which may be sufficient to prevent
their subduction. If not, they should alter the geo
metry ofthe Benioffzone and consequently affect the
tectonic and magmatic process associated with the
subduction.
At present, no oceanic high of continental affin
ities, that is, too buoyant to be subducted, is in the
neighbourhood of a trench. There are several loca
tions, however, where oceanic rises are in the process
ofsubduction (Nur & Ben-Avraham, 1982), such as
at the Peru-Chile Trench (Fig. 9.29) and the Nankai
Trough (Hirahara, 1981). The subducting rises
strike obliquely to the direction of underthrusting,
so that the locus ofthe point ofdescent progressively
moves along the trench. The cpntinental region
behind the segment of trench swept by this locus is
marked by a gap in volcanic activity. Figure 9.30
shows the vertical distribution of earthquakes across
these zones of anomalous subduction and compares
them with regions of normal subduction and conti
nental collision. The buoyancy ofthe descending rise
causes the Benioff zone to dip at a shallower angle
than normal, and a distinct seismic gap is present.
A conceptual model for the accretion of a suspect
terrain is shown in Fig. 9.31,illustrating the possible
evolution of the Bering Sea. Within this region the
Umnak Plateau and Bowers and Shirshov ridges form
positive topographic features, all of which are under
lain by thickened crust. Bowers Ridge is probably
an extinct island arc. Ben-Avraham et al. (1981)
propose that the present structural configuration
evolved as the proto-Bowers Ridge and proto-Umnak
Plateau moved northwards as a result ofsubduction
at a now extinct trench within the Bering Sea
(Fig. 9.31a). The arrival of these features at the
trench may have locked the subduction zone and
caused a new trench to develop at the presently active
Aleutian arc (Fig. 9.31b). The Shirshov Ridge may
have been emplaced at the same time by motion
along a now extinct transform fault. A fuller account
of the evolution of this region is given by Cooper et
al.(1992).
Similar processes of accretion have been proposed
for other regions of suspect terrains (Shermer et al.^
1984), and such models have prompted Nur &
Ben-Avraham (1982) to construct a world map
showing possible present and future suspect terrains
(Fig. 9.32). While it is accepted that the terrains
shown have accreted, or will accrete, to major conti
nental plates following large horizontal movements,
it is important that the nomenclature does not get
out of hand. Most workers would describe the re
gions of Somalia, Arabia and India as plates or
sub-plates, and it would seem more reasonable to
restrict the term 'suspect terrain' to considerably
smaller entities with a rather more obscure history of
movement.
It is probable that many of the suspect terrains in
the Alpine-Himalayan chain represent fragments
rifted from the northern margin of Gondwana and
transported across the Tethys Ocean before accretion
to the southern margin of Laurasia (Fig. 9.33).
Similarly, the circum-Pacific terrains may have
originated from the eastern margin of Gondwana.
The fragments ultimately travelling to the Western
Cordillera of North America are believed to have
come from a region of Gondwana named Pacifica
(Fig. 9.33)(Kamp, 1980), which was probably situ
ated over a geoid high (Anderson, 1982). Since the
Pacific Ocean occupied two-thirds of the Earth's
surface at the start of the Mesozoic, the Pacifica
terrains would have become widely dispersed during
their long voyage. The mechanism of rifting, how
ever, is unknown. It is possibly related to the hotspot
and volcanic activity associated with the geoid high
(Section 10.4)(Anderson, 1982). Wmketal.(1984)
have demonstrated the weakness ofcontinental litho-
sphere compared to oceanic, so that if rifting occurs,
it will preferably follow a continental path and create
the microplates which will subsequently become
suspect terrains.
The mechanism of accretion of suspect terrains is
 MOUNTAIN RANGES 203

Cocos Ridge

Galapagos
Carnegie R. (KJ

Nazca Ridge^

Juan Fernandez R

Submarine ridges
and islands

Deep sea trenches

Cenozoic voicanism

Cenozoic graben
Fig.9.29 Map showing the faults and
relationship between subducting intra-Andean depressions
oceanic rises, Cenozoic voicanism
and graben in western South
America.

poorly understood. Where the terrain is of a buoy
ancy sufficient to prevent its subduction, the trench
may become locked and a new trench may develop to
the oceanward side of the old one, perhaps entrap
ping a marginal sea. Several of the present marginal
seas for which there is no convincing evidence for
back-arc spreading (Section 8.9), such as the eastern
Caribbean and Bering Sea, may have formed in this
way. In other cases, terrains are accreted directly to
the continent as coherent masses of significant
crustal thickness which do riot, however, behave as
rigid sub-plates and undergo strong internal deform
ation (Burchfiel, 1980). Telescoping and thrusting
are dominant and there is no evidence ofsubduction-
related igneous activity. It is possible that the process
resembles an obduction mechanism similar to that
204 CHAPTER 9

(a) Underthrusting normal (b) Underthrusting with buoyant (c) Collision

fragments
W E W E S N
0 200 400 600 800(km) 0 200 400 600 800(km) 500 1000 1500 (km)
0 TT ^

'E 200 200

200 p
400
^400 400 Gap
0 600
.t
^ 600 Pamir(75''E)
Altiplano(~20°S)
600
Nazcagap(~10°S) A
W E W E S
0 200 400 600 (km) ^200 400 600 800 (km) ^ 0 500 1000 1500(km)
0 ^-n

E 200 200
200
400
B 400 Gap
400
600
Q 600
Izu Trench 600 X
/ '-N
Tibet(85°E)
Juan Fernandez gap(~30°S)
S N
100 200 300 (km)
I . r T"

1100 - /Gap/...
#200
Q.
0
Q Kos(SW-NE)

Fig.9.30 Comparison offocal distributions beneath
subduction zones and Himalayan mountain ranges:
(a) normal subduction-related seismicity;(b) disrupted
subduction zone seismicity;(c) collision-related seismicity
(redrawn from Nur & Ben-Avraham, 1982, with
permission from the American Geophysical Union).
proposed by Oxburgh (1972). In this model an
irregular continental mass approaches a subduction
zone (Fig. 9.34a). When the promontory X reaches
the trench and resists subduction, plate movement
continues in response to the subduction of embay-
ments Y, Z, and the leading edge of the overriding
plate acts as a wedge, splitting the lithosphere hori
zontally (Fig. 9.34b) and obducting the upper part
on to the adjacent continent for distances perhaps in
excess of 100 km. The split propagates progressively
downwards so that the whole of the crust may be
come detached. Confirmation of such a mechanism
must await more detailed investigations. However,it
appears that terrain accretion is a viable mechanism
of ophiolite emplacement, as ophiolites frequently
occur at terrain margins. It is also possible that
terrain accretion was a dominant mechanism in the
growth of the continents during Precambrian times
(Section 12.2).
The post-accretionary fragmentation of suspect
terrains is probably a consequence of plates under
going oblique convergence (Barber, 1985),as such a
regime frequently gives rise to transcurrent faulting
in the associated mobile belt (Section 8.5). Oblique
subduction is common in many present-day
trenches, and it would be expected that such faulting
is a ubiquitous feature ofancient mountain belts and
responsible for the fragmentation of accreted ter
rains.
As the Western Cordillera of North America
formed by accretion tectonics, it seems possible that
the Andes formed in a similar way. Although detailed
analyses are not yet available, there are several pieces
ofevidence which suggest that this might be the case.
The Andes comprise several distinct tectonic/
stratigraphic provinces, often separated by major
faults; many Palaeozoic and early Mesozoic struc
tures strike at an oblique angle to the mountain belt;
 MOUNTAIN RANGES 205

Siberia Alaska

Shirshov
-Ridge umnak
Plateau
/ V .h. Bowers
Ridge

Fig.9.31 Model for the

formation ofthe Bering Sea: ^ Direction of plate motion
(a) Late Mesozoic;(b) Early
Tertiary (redrawn from Ben-
Avraham etal.^ 1981,Science
(b)
f * * Active subduction zone

213,47-54, with permission Former subduction zone

from the AAAS. Copyright ©
1981 by the AAAS). Transform fault

during the Palaeozoic and Mesozoic sediment
sources were located to the west. These geological
observations (e.g. Ramos, 1988) and palaeomag-
netic results (Beck et al.^ 1986)suggest the presence
ofsuspect terrains. Other workers, however, such as
Cobbing,in Pitcher (1985), question the pres
ence of exotic terrains and contend that the Andean
margin has been integral during the Phanerozoic.
If both the Western Cordillera and Andes devel
oped by accretionary tectonics,it could be argued that
all orogenic activity is related to the single process of
collision, the complexity of mountain belts being
explicable in terms of multiple deformational epi
sodes associated with the successive accretions of
 North Slope
w Endicot
o
c
o

Faeroe Kolyma > KoryaR Ruby

Plateau Nixon Fork
Okhotsk
Sikhote Alin
Korea' Kamschatk f Yukon-Tanana Rockall ^
Plateau
Apulia Stikine
Goodnews
Adria Sakhalin
Hokkaido Wrangellia
Katagami
Mess Rise
%
-IT yamat Salinia
- J^Mallorca-[o
Balearic c/3 Shatsky Rise
Thailand Benham
Rise Rise
N. Jibe Paracel Island Mexico
Arabia
S. Tibet S. Halmahera Nicaragua
iivchelles G^ion Buton
Ontong-Java
Plateau
Somalia Kontum
Manihiki Peru
Sumba
Plateau Arequ pa
Nazca Ridge
Broken Norfolk
Ridge CO (u
Plateau
Madagascar Juan Fernandez Ridge
Lord Coast Range
Howe Falkland
Agulhas Chatham
Kerguelen Rise Plateau
Plateau Crozet Rise
Plateau
\Ridge
Campbell
Plateau

20° 0° 20° 40° 60° 80° 100° 120° 140° 160° 1P0° 160° 140° 120° 100° 80° 60° 40°

Fig.9.32 Distribution of past, present and future suspect terrains (redrawn from Nur & Ben-Avraham, 1982, with permission from the American Geophysical Union),

Fig.9.33 Possible original
configuration of many ofthe suspect
terrains in the Alpine and Pacific
Mesozoic-Cenozoic orogenic belts
(redrawn from Nur & Ben-Avraham,
1982, with permission from the
American Geophysical Union).
terrains. True continent-continent collision then
represents but one end of a spectrum of possible
accretionary events.It is not essential to the formation
of a mountain belt, but may have the function of
Fig.9.34 (a) Schematic representation ofan irregular
continental margin approaching an Andean-type
subduction zone,(b)Formation of a crustal flake after
collision (redrawn from Oxburgh, 1972, with permission
from Nature 239, 202-4. Copyright © Macmillan
Magazines Ltd).
Trench
Continent Continent
Moho
i-
(a)
MOUNTAIN RANGES 207
^''^gnients
Seyche//
Antarctica ^^stralia
J^smania
ending orogenic activity. The concept of a unique
suture marking the ancient juxtaposition of con
verged plates may thus be illusory for many orogens.
Palaeogeographic reconstructions have previously
been attempted from the fragments of plates pre
served in mountain belts. The recognition ofsuspect
terrains implies that such attempts cannot be success
ful. Terrains now adjacent may have originated from
completely different areas, and even after accretion
may have been translated large distances along
wrench faults.
(b)
208 CHAPTER 9
Further reading
Ben-Avraham, Z., Nur, A., Jones, D. & Cox, A. (1981)
Continental accretion and orogeny. From oceanic
plateaux to allochthonous terranes. Science 213, 47-
54.
Cobbing, E.J. (1982) The segmented coastal batholith of
Peru: its relationship to volcanicity and metallogenesis.
Earth. Sci. Bsv. 18, 241-51.
Coney,P.J., Jones,D.L. & Monger,J.W.H.(1980)Cordil-
leran suspect terranes. Nature 288, 329-33.
Dewey, J.F. & Bird, J.M. (1970) Mountain belts and the
new global i^Qtomcs.J.^eophys. Res. 75,2625-47.
England, P.C. (1982) Some numerical investigations of
large scale continental deformation. In Hsii, K.G. (ed.)
Mountain Building Processes^ pp. 129-41. Academic
Press, London.
Howell, D.G.(1989) Tectonics of Suspect Terranes: Moun
tain Buildinpf and Continental Growth. Chapman & Hall,
London.
James, D.E.(1971) Plate tectonic model for the evolution
of the central Andes. Bull, pfeol. Soc. Am. 82, 3325-
46.
Jones, D.L., Howell, D.G., Coney, P.J. & Monger,J.W.H.
(1983) Recognition, character and analysis of tectono-
stratigraphic terranes in western North America. In
Hashimoto,M.& Uyeda,S.{tds)Accretion Tectonicsin the
circum-Pacific Repdons^ pp. 21-35. Terra Scientific,
Tokyo.
Jordan, T.E., Isacks, B.L., Allmendinger, RW., Brewer,
J.A.,Ramos,V.A.& Ando,C.J.(1983)Andean tectonics
related to geometry ofsubducted Nazca Plate. Bull.pfeol.
Soc. Am. 94,341-61.
Molnar,P.& Chen,W.-P.(1982)Seismicity and mountain
building.In Hsii, K.G.{cd.)Mountain Building Processes.,
pp.41-57. Academic Press, London.
Powell,C.McA.& Conaghan,P.J.(1975)Tectonic models
of the Tibetan plateau. Geology 20,727-31.
Schaer, J.-P. 8c Rodgers, J. (eds) (1987) The Anatomy of
Mountain Ranges. Princeton University Press, Princeton,
NJ.
Shackleton, R.M., Dewey, J.F. 8c Windley, B.F. (eds)
(1988) Tectonic Evolution of the Himalayas and Tibet.
Royal Society, London.
Tapponier, P. 8c Molnar, P. (1976) Slip-line field theory
and large-scale continental tectonics. Nature 264, 319-
24.
Tapponier, P., Peltzer, G., Le Dain, A.Y., Armjo, R 8c
Cobbing, P.(1982) Propagating extrusion tectonics in
Asia: new insights from simple experiments with plas-
ticene. Geology 10,611-16.
Wilson, M.(1989) Igneous Petrogenesis. A Global Tectonic
Approach. Unwin Hyman,London.
Windley, B.F. (1985) The Himalayas. Geology Today 1,
169-73.
10/ Continental rifts, continental margins
and sedimentary basins
10.1 Continental rifts
10.1.1 Introduction
Rifts are defined as elongate depressions where the
entire thickness of the iithosphere has deformed
under the influence ofextensional forces(Neumann
& Ramberg, 1978; Burchfiel, 1980). The term 'rift'
thus applies to major iithospheric features and does
not encompass smaller scale structures associated
with normal faulting. Rifts are relatively common as
the strength ofthe Iithosphere is least under tension,
and are found on the continents in a variety of
structural settings. Examples of presently active rifts
are the East African Rift, the Baikal Rift (south
central Siberia), the Rhine Graben (northwest
Europe) and the Rio Grande Rift and Basin and
Range Province of western USA.
10.1.2 Characteristics of rifts
Although they may form in different tectonic envi
ronments, rifts exhibit many common characteristics
(Neumann & Ramberg, 1978).
Rifts often contain extensive volcanic rocks, al
though in some cases few or no volcanics are de
veloped. Where present, these rocks are mildly to
strongly alkaline, and the alkalinity decreases with
time and increases with distance from the rift axis.
Rifts are commonly spatially associated with
domal uplifts. They may indicate a uniaxial extension
or lack an obvious axis. They may form an interlinked
network apparently dividing a continental plate, or
be isolated and situated far from any plate margin.
The Iithosphere underlying rifts is generally abnor
mally thin and invaded by low velocity, low density
and high temperature material. In this respect they
are similar to ocean ridges (Section 6.4).
The location of rifts is often controlled by existing
zones of crustal weakness (Sykes, 1978). Thus, the
Baikal Rift lies between the Siberian Platform and the
Sayan-Baikal Fold Belt (Logatchev & Florensov,
1978). These zones, however, must be approxi
mately normal to the regional tensile stress system
(lilies, 1977).
The amount of horizontal extension required to
form present-day rifts is relatively small (Logatchev
& Florensov, 1978), being less than 10 km for the
Baikal Rift, 4.5-5 km for the Upper Rhine Graben
and 8-10 km for the Kenya Rift.
10.1.3 Classification of rifts
Continental rifts fall into three major categories, two
of which are associated with continental break-up
and one with continental collision.
1 On continental splitting, the precursor to the
development of a new ocean is the development ofa
rift system, which marks the locus of rupture. A
modern example is often cited as the East African rift
system whose present activity extends from Neogene
times, when the African Plate appears to have come
to rest with respect to the mantle convection pattern.
It should be noted, however, that some theories of
continental splitting (Section 10.3) do not view the
East African rifts as the site of a future major conti
nental rupture. The initial rifting and splitting of a
continent prior to the onset ofsea floor spreading is
considered further in Section 10.3.
2 If continental splitting occurs by the con
catenation of a series of rift-rift-rift triple junctions
(Section 5.9), the rift not utilized in rupture remains
preserved within the continent as a failed rift or
aulacogen (Section 10.2). The aulacogen's sedimen
tary development continues after splitting, and it
may be reactivated at a later stage when the conti
nental margin finally participates in a continent-
continent collision.
3 Continental collision imposes a large scale com-
pressional regime in the vicinity of the juxtaposed
continents. However, in areas distant from the
suture, tensional regions can develop in response to
209
210 CHAPTER 10
(a)
AAAAAAA
AAAAAAA A A A A A A A A A A A A A A A A 0 AA
=== Continental edge of B
1I M Passive continental margin
-V ^ y ■ Subduction zone
AAA Andean-type margin
Fig. 10.1 Diagrammatic maps showing the development
of a rift following collision oftwo continents A & B. See
text for explanation (redrawn from §eng6r, 1976, with
permission from the Geological Society of America).
indentation tectonics (Section 9.3.6). In this way
relatively small, isolated rift valleys can form such as
the Rhine Graben in the hinterland of the Alpine
Orogeny and the Baikal and Shansi Graben in the
distal region of the Himalaya. Such 'collisional' rifts
can be distinguished from aulacogens, which also
strike into collisional fold belts, as they have no
pre-collisional history in their geological record.
§eng6r (1976) has also proposed a model for
certain collision-related rifts whereby graben develop
in the foreland region ofa Himalayan-type orogen as
a result of the collision of irregular continental
margins (Fig. 10.1). The localized compression re
sulting from the collision of a promontory
(Fig. 10.1a) also causes tensional stresses parallel to
the collision front. These give rise to dilational fea
tures,including graben striking at right angles to the
orogen (Fig. 10.1b). Continued convergence length
ens the suture so that collision occurs along an
extensive front, and the foreland may then be cut by
large scale wrench faults (Fig. 10.1c). The Upper
Rhine Graben may be an example ofa rift ofthis type.
AAAA A A
Basic dykes
Graben
Ophiolite nappe
Basement cored nappe
Such collision-generated graben have been termed
impactqgens.
The distinction between rifts generated by conti
nental rupture or collision is made by some authors
by using the terms'mantle activated'(or 'active') for
the former and 'lithosphere activated' (or 'passive')
for the latter (Burchfiel, 1980). These terms refer to
the role of upwelling asthenosphere in creating the
'mantle activated' rift. Mantle activated rifts are char
acterized by large volumes of extrusive rocks, while
lithosphere activated rifts usually contain a greater
abundance of immature clastic sediments than vol
canic rocks. Passive rifts may evolve into active rifts
by the subsequent upwelling and spreading of the
asthenosphere.
10.1.4 Petrology and petrogenesis of
rift rocks
The volcanic rocks extruded at continental rifts are
highly enriched in alkalis, large ion lithophile ele
ments (LILE) such as K, Ba, Rb and the light rare
earths, and volatiles, in particular CO2 and the halo
gens.
Bailey (1983) has suggested that the distinctive
nature of these rocks arises from their formation by
metasomatic processes in which the volatile content
of a large volume of mantle material is focused and
 CONTINENTAL RIFTS, CONTINENTAL MARGINS AND SEDIMENTARY BASINS
concentrated. Volatiles escaping upward from the
deep mantle probably carry LILE, and these would
cause metasomatism at higher levels where amphi-
bole, biotite and carbonate are stable. Bailey suggests
that the volatiles originate in a large mantle reservoir
and migrate upwards through the narrow rift zone
(Fig. 10.2). LILE build up in the conduits in the
overlying mantle and crust by metasomatism. The
volatiles modify the melting behaviour of their sur
roundings and eventually depress the solidus to the
point at which melting occurs, producing volatile
and LILE-rich magmas.
The alkaline magmas probably originate at depths
of approximately 60-100 km. Tholeiitic and calc-
alkaline volcanics may also be present, which equi
librated at much higher levels. In addition to these
systematic variations, which are caused by depth and
degree of partial melting, there also exist unsystem
atic variations related to mantle inhomogeneities and
complex combinations of processes which affect the
magmas during and after melting (Neumann &
Ramberg, 1978).
Volcanic rocks occurring within continental rifts
and oceanic islands are enriched in incompatible
elements and alkalis. Ocean floor basalts, however,
are relatively depleted in these components. It is not
possible to explain these differences in terms of the
conditions ofmagma genesis and evolution,and so it
appears that the mantle from which the magmas are
derived is heterogeneous. Volcanic rocks in conti
nental rifts and oceanic islands must be derived from
a mantle source enriched in incompatible elements
and radiogenic strontium and lead isotopes. Ocean
ridge basalts must be derived from a source depleted
in these components.
It is generally recognized that the depleted mantle
source is the asthenosphere, but opinions are divided
as to whether the enriched source originates above or
below the asthenosphere, that is, from the deep
Mobile elements
Fig. 10.2 Fissuring ofthe lithosphere to allow the escape
of volatiles from a mantle reservoir (stippled)(after Bailey,
1983, with permission from Elsevicr).
211
mantle or the sub-crustal lithosphere. High level
enrichment may result from the trapping of en
riched, primitive asthenosphere at the base of the
lithosphere before depletion, underplating of the
lithosphere by undepleted mantle diapirs, or diffu
sion ofLILE-rich volatiles from the asthenosphere or
deeper mantle into the lithosphere.
10.1.5 Structure of continental rifts
Earthquake focal mechanism determinations, geo
logical investigations and gravity and seismic reflec
tion surveys have demonstrated that the bounding
faults of rift valleys are of normal type. Rifts con
sequently develop in response to tensional stresses,
and older theories which attributed their formation
to compression, in which a central block was de
pressed beneath bounding reverse faults, have been
discounted. Seismic reflection data (Brown et al.^
1980) have shown that the faults bounding the Rio
Grande Rift do not decrease in dip with depth,and so
are not of listric type. Whether or not this observa
tion applies to all rifts must await similar surveys of
other areas.
Seismic refraction experiments have shown that
the crust underlying many rifts is significantly thin
ner than beneath adjacent regions, and indeed
gravity anomalies indicate that the whole of the
lithosphere is thinned (Searle, 1970). Seismological
data using both natural and artificial sources have
shown that a zone of anomalously low seismic velo
cities occurs in the upper mantle beneath most rifts
(Fig. 10.3) and is commonly two to three times
wider than the rift zone. The extent of this zone in
the region of the Baikal Rift (Puzyrev et al. 1978) is
shown in Fig. 10.4 and Gao et al (1994) have
demonstrated a shallowing of the asthenosphere in
this region. The vertical extent of the zone has been
studied beneath the Gregory Rift in Kenya by Long
& Backhouse (1976). They have shown that it
widens downwards into an ellipsoidal structure at a
depth of 150 km. The regional gravity field of the
Gregory Rift is negative (Baker & Wohlenberg,
1971),indicating that the low velocity mantle mater
ial is also of low density. Heat flow measurements,
the presence of volcanic rocks and geothermal phe
nomena show that this low density results from
thermal expansion. Local positive gravity anomalies
over the rift axis indicate the presence of a high
density igneous body about 20 km wide.
212 CHAPTER 10

100° E 120'^E ^6Q°N

(D
c
DC
0)
C .Q
c
CD
Irkutsk j Lake
CD
O
CD
DC j/SaikaK
■>- c5 T3
< Q) Ulan Bator
C
Baikal Rift {/)
Z
CD
UJ DC
Beijing
56 7 8 Z z

China
LVZ
Crust
Moho
100° El 20
Low velocity
50 \ Hi
upper mantle
Normal mantle
km

100-
I
s Basni Province Low velocity
mantle Irkutsk

150
V(km 8"')
200 km
I
Fig. 10.3 P wave velocity structure of continental rift
zones (redrawn from ^uzyrcv et al., 1978, with y Contours of
depth to mantle (km)
permission from Elsevier).
Limit of
•••*' low velocity region
X Geological boundaries
Gravity surveys have shown that the uplifted re of rift

gions are in isostatic equilibrium. However, negative

Fig. 10.4 Schematic diagram of the upper mantle
Bouguer and isostatic anomalies are locally present structure in the Baikal region (redrawn from Puzyrev et
over the rift itself (Fig. 10.5). The negative anomalies al.^ 1978, with permission from Elsevier).
result from the presence of a low density sediment
infill, which may be as much as 6 km thick. -1400 Computed
The low density material in the upper mantle 1800-
beneath rift valleys isostatically supports doming of Observed
-2200-
the crust in the whole region around the rift. The
-2600
crustal doming in the vicinity of the Rhine Graben
(lilies, 1977) is shown in Fig. 10.6, which also shows
WSW Rift valley
the underlying mantle bulge, centred beneath the 2.0. 2.2
Kaiserstuhl volcano. A similar hot low density region
underlies the Basin and Range province of the
western USA.

10.1.6 Origin of rifts

Any mechanism for the origin of rifts must account
for their apparent large local deviation from isostatic
equilibrium in the presence of a tensional stress
regime.
Heiskanen & Vening Meinesz (1958) proposed a
model in which the whole of the crust was assumed
to be elastic. A pair of inwardly inclined normal
faults were developed which penetrated the whole
crust. They predicted, from elastic beam theory, that
in a crust 35 km thick the faults would dip at about
50 100 km
I
Fig. 10.5 Bouguer anomaly profile and model of the
Gregory Rift. Densities in Mg m"^ (redrawn from Baker &
Wohlenberg, 1971, with permission from Nature 229,
538-42. Copyright © 1971 Macmillan Magazines Ltd).
60-65® and be separated by about 65 km at the
surface. Persistent tension would then lead to the
continued foundering of the keystone and the devel-
 CONTINENTAL RIFTS, CONTINENTAL MARGINS AND SEDIMENTARY BASINS 213

Contours of the
, Dulsburg
/ crust-mantle
boundary
' (depth In km)

Major fault zone

Border of the
Alpine fold belt

Miocene volcano
of the Kalserstuhl

Inner trough of the

Rhine graben
Wurzburg

iLuxembourg, V
Amount of uplift:

Stuttgart
1000-1500

M 1500-2000

2000 - 2500

> 2500 m
Fig. 10.6 Mantle bulge beneath
the Rhine Graben with post
middle Eocene uplift (redrawn
from lilies, 1977, with
permission from Kluwer
Academic Publishers).

opment of rim uplift by elastic upbending of the
crust on the outer, upthrown, sides of the rift
(Fig. 10.7a). There would be a corresponding out
ward flow of ductile material in the upper mantle.
This hypothesis clearly explains the dimensions
and morphology of rifts. Moreover, it accounts for
the presence of negative isostatic anomalies. This is
not because the downfaulted wedge is actually out of
isostatic equilibrium, but rather that it deviates from
the assumptions inherent in theories of isostatic
compensation. The Airy model (Section 2.11.2) as
sumes that equilibrium is attained by differential
movements of adjacent blocks of crust along vertica.1
fault planes until their upthrust is sufficient to bal
ance the mass of the overlying material. In the rift
model described above, however, the subsiding fault
wedge is bounded by inclined rather than vertical
faults so that equilibrium is attained at a somewhat
lower level. This is easily understood ifone considers
that a wedge-shaped block of wood would float
somewhat lower in water than a block of the same
thickness with parallel edges (Fig. 10.7b).
The theory of Heiskanen & Vening Meinesz
(1958) breaks down, however, in that the crust is
usually thinned beneath rifts and there is no evidence
for a fault wedge penetrating into the mantle (Artem-
jev & Artyushkov, 1971). It is readily modified ifit is
assumed that the faulting and subsidence are re
stricted to the upper, brittle, part of the crust and
compensating lateral flow at depth takes place in the
ductile lower crust (Section 2.10), whose viscosity is
decreased by the elevated temperatures experienced
in this region. There is geochemical evidence that the
lithosphere thins beneath rifts (Perry a/. 1988).
A more realistic sequence of events in rift forma
tion (Fig. 10.8) is thus as follows (Bott, 1981):
214 CHAPTER 10
(a)
?^65 km-^
Mantle —i
0$k
Fig. 10.7 (a) Vening Meinesz's hypothesis for the
formation of a rift valley (after Heiskanen & Vening
Meinesz, 1958);(b)isostatic equilibrium of a wedge and
parallel sided block ofthe same thickness.
1 Under tensional stress, a normal fault develops in
the upper crust (Fig. 10.8a).
2 The fault movement warps the crust, whose bend
ing gives rise to maximum tension where the curva
ture is greatest (Fig. 10.8b).
3 A second fault develops in this location and a rift
valley is established (Fig. 10.8c). Continued tension
causes continuous subsidence ofthe keystone, aided
by the weight ofsediments, and the development of
rim uplifts in the adjacent upper crust. These move
ments are accompanied by compensatory flow in the
ductile lower crust.
The model described above assumes that normal
faulting is relatively high angle and limited to the
region between the scarps which define the rift valley.
Wernicke & Burchfiel(1982)and Wernicke (1985),
however, have considered the consequences of large
scale strain in extensional areas. They have concluded
that large, very low angle normal faults (sometimes
called detachment faults) dominate extended ter
rains,and that normal faults ofboth planar and listric
geometries are common components ofthe hanging
walls. This model is considered more fully in Section
10.5.2. Bosworth (1987) has applied this theory to
the Gregory Rift, which forms part of the East Afri
can rift system.
Bosworth (1987) has noted that the large vol
canoes associated with the rift are situated 100-
Ductile
< lower
crust
Moho
Upper
mantle
Sediments
Fig.10.8 Development of a rift valley by faulting in the
upper crust and ductile flow in the lower crust (c)(after
Bott, 1981, with permission from Elsevier).
150 km to the east and west ofthe axis, and appear to
form an alternating pattern on each side (Fig. 10.9).
These volcanic centres correlate with asymmetric
development of the rift valley in which the edge
nearer to the volcano is an abrupt topographic break,
whereas the farther edge is characterized by a more
gradual rise via a series of smaller, parallel fault
scarps. The rift structure is interpreted as a half
graben defined by a set of imbricate listric normal
faults (Fig. 10,10). These faults sole out into a major
low angle planar normal fault, which reaches the base
of the regionally thinned crust beneath the locus of
the bordering volcanoes and thins it further. It is
possible that the regional seismicity (Fig. 10.9) may
be explained by events occurring on this detachment,
which are not restricted to the vicinity of the rift's
bounding faults.
The zones between detachments of opposite dip
 CONTINENTAL RIFTS, CONTINENTAL MARGINS AND SEDIMENTARY BASINS 215

Gregory Geological evidence from many rift areas indicates

rift
that the sequence of events during the development
Crustal ofa rift is firstly volcanism,followed by doming and
-oV Marsabit then rifting. As doming always appears to precede
rifting, it is possible that the doming itselfcreates the
^1 ^
^ Elgeyo necessary stresses for rifting to occur. The volcanism
' «'^"7 and uplift result from the development of a hot
UGANDA'^ detachment region in the upper mantle, which itself probably
originates from upwelling from the deeper mantle.
The thermal expansion produced by this heating
KENYA creates a low density region which, upon attaining
v^Aberdar isostatic equilibrium,supports the observed doming.
Lithospheric
^ N'
»' Moreover,the presence ofhigh temperature material
tachmehif^ thinning
Sif'J N causes the lithosphere to thin (Spohn & Schubert,
TANZANIA 1982),as the lithosphere-asthenosphere boundary is
50 km
-1 N \ / J I I
largely dependent upon the thermal state of the
Nguruman subsurface. The uparching throws the overlying crust
Lithospheric '/i-.C/A'/ detachment into tension, but ifthe whole crust acts elastically the
thinning Kilimanjaro stresses generated would be too small to cause fault
ing. However, Bott & Kusznir (1979) have shown
that if only the upper crust behaves elastically, with
Fig. 10.9 Tectonic interpretation of the region around the lower crust acting in a ductile manner, the
the Gregory Rift in western Kenya. Bold arrows represent tensional stresses become confined within the upper
relative motion of upper crust above each detachment. crust and amplified to magnitudes at which crustal
AZ,accommodation zone; solid circles, epicentres of rupture can take place.
historical earthquakes (redrawn from Bosworth, 1987, The rupture ofthe crust by this mechanism would
with permission from the Geological Society of America). only occur at a single point corresponding to the area
immediately above a hotspot. In order to propagate
laterally, the crust must be in an overall state of
direction have complex histories of wrench and horizontal tension, and the possible causes of this
oblique-slip faulting (Fig. 10.9), and have been stress state will be considered in Section 10.3.3.
termed accommodation zones. These are expected to
extend laterally away from the rift valley until they
merge with the detachment at depth. Where this Fig. 10.10 Schematic cross-section across the Gregory
merging occurs at the base of the thinned crust, the Rift. No vertical exaggeration below sea level (redrawn
wrench structures might facilitate ascent of magma from Bosworth, 1987, with permission from the
and the formation of volcanoes. Geological Society of America).

Mt Elgon Elgeyo escarpment Rift valley Mt Marsabit

Sea level
Aberdare detachment Elgeyo detachment
Lithosphere

Differentiatinq magma bodies Asthenosohere

Kenya Dome hot spot

20 km
216 CHAPTER 10
10.2 Aulacogens
Aulacogens (literally 'born as furrows') are long-
lived, deeply subsiding sedimentary troughs, often
fault bounded, that extend at high angles from the
margins towards the interiors of cratons (Dickinson,
1974). They were first described from the USSR in
the early 1940s, and occur in the geological record
from late Proterozoic times (Hoffman, 1973) on
wards (Milanovsky, 1981).
Aulacogens have a number of properties in com
mon (Burke, 1977), all of which must be explained
by any model for their origin.
1 They are located at re-entrants on continental
plate margins, and their initial formation is contem
poraneous with continental rupture.
2 They are characterized principally by vertical tec
tonics, with an early history as narrow graben and
subsequently as broader sedimentary basins. The
sediment fill is some three times thicker than on the
adjacent craton. This is usually undeformed or only
gently folded,in contrast with the extreme tectonism
experienced by adjacent orogenic belts.
3 They have a long duration, which corresponds to
the span of activity of the associated plate margin.
4 They exhibit the tendency to reactivation by re
newed faulting and subsidence long after termina
tion of activity in associated orogenic belts.
5 Their initiation is frequently accompanied by the
intrusion of alkaline igneous rocks.
Aulacogens provide a complete igneous, sedimen
tary and structural history of events associated with
an orogenic cycle, and are thus very useful in deci
phering sequences ofcomplex plate interaction. The
stages in development of an aulacogen in compari
son to a plate margin are shown in Fig. 10.11.
Aulacogens are believed to form as the result ofthe
evolution of a rift-rift-rift triple junction within a
continental plate (Fig. 10.12a). As plate separation
proceeds, one of the arms of the triple junction
becomes inactive, and remains preserved in the con
tinent as a failed rift or aulacogen (Fig. 10.12b).
More rarely, a similar abandoned rift may develop
from an inherently unstable ridge-ridge-transform
fault triple junction (Dickinson, 1974) as shown in
Fig. 10.12c.
Burke & Dewey(1973) have suggested that conti
nental lithosphere, especially if it is stationary with
respect to the mantle, develops a rift-rift-rift triple
junction in response to the effects of an underlying
mantle plume (Fig. 10.13a). The increased heat flow
causes uplift and the development of three rifts,
typically meeting at 120® (Fig. 10.13b). Separation
across the rifts then occurs by the emplacement of
axial dykes, potentially leading to the generation of a
ridge-ridge-ridge triple junction, and marking the
initiation of a new phase of sea floor spreading.
Frequently, however, one of the arms of the triple
junction becomes inactive, and an aulacogen is
formed. Hoffman et al. (1974) have suggested that,
since the poles of rotation of most of the present
plates are in high latitudes, the east-west rift arms are
most likely to fail. If rift failure occurs before
the intrusion of igneous material, the aulacogen is
floored by continental lithosphere (Fig. 10.13c),and
if after axial dyke emplacement, by oceanic litho
sphere (Fig. 10.13d). The aulacogen represents a
very favourable location for the development of a
river system (Fig. 10.13e), which carries detritus
from the craton. If local conditions are suitable, a
large delta will form which progrades progressively
oceanward. A modern example is the Niger River
which drains the Benue Trough into the Gulf of
Guinea. This situation is maintained if the adjacent
ocean starts to contract (Fig. 10.13f), and on com
plete ocean closure (Fig. 10.13g) a collisional orog
eny ensues (Section 9.3) and the sediment transport
direction reverses. The aulacogen is then preserved,
striking at a high angle to the mountain belt.
The reactivation of aulacogens as depositional
centres is probably explained by the occurrence of
continental splitting preferentially along existing
lines of weakness. The present Atlantic Ocean, for
example, opened along approximately the locus of
opening and closing ofthe proto-Atlantic,indicating
that sea floor spreading may be initiated along
ancient sutures (Vink et al., 1984), so reactivating
the aulacogens associated with the earlier splitting.
Dewey & Burke (1974) have suggested that conti
nental splitting (Section 10.3) occurs in direct re
sponse to the development of a series of rift-rift-rift
triple junctions at hotspots across a continental plate
(Fig. 10.14). At each hotspot two of the rifts extend
to link up with adjacent rifts, while the third becomes
an aulacogen. This results in an irregularly shaped
continental margin, whose jaggedness is a con
sequence of the locus of the triple junctions. The
aulacogens consequently occupy re-entrants in the
continental margin.
Burke (1976; 1977) has applied this theory to the
rupture of Gondwana and Laurasia, and has recog
nized the existence of many aulacogens along the
 CONTINENTAL RIFTS, CONTINENTAL MARGINS AND SEDIMENTARY BASINS 217

Plate margin Aulacogen

Dome with crestal graben filled by continental

sediments and alkali basalt-rhyolite Graben stage

Mantle upwelling Mantle upwelling

Red Sea-type small rift ocean

Oceanic lithosphere produced

Abyssal Plain Rise Continental shelf Downwarp stage

o:;-\

1/1 -'I

I- WN-.- ~\y-yl^. - /

■l*i
Mantle contraction

Arc-trench system
Volcanic arc
Arc-derived Trench
^ sediments
aM;ri 'n:t;,r"XT^

' ^ / y\
/C' ^ 1 f 'T 'w \<

' Oceanic lithosphere subduced

Orogenic belt
Tibetan-type hinterland Nappes Compressional stage
plateau

'K"./"li"* J ^r'.^rV ■>, ''

¥:¥:%¥x*x*x«x*x

Continental collision Medial block founders

Fig. 10.11 Comparison of the evolution of plate margins

and aulacogens (redrawn from Hoffman etal., 1974).
218 CHAPTER 10

eastern margins ofthe Americas and western margins

of Europe and Africa (Fig. 10.15). Many of these
have been investig«. in great detail because oftheir
importance in hydrocaroon exploration (e.g. White-
man 1975).
An example of the evolution of an aulacogen is
provided by the history of the Palaeozoic Southern
Oklahoma Aulacogen (Ham, 1969; Hoffman et«/.,
1974) of the central USA (Fig. 10.16, 10.17). The
aulacogen was initiated as a graben transverse to the
North American Platform, underlain by Precambrian
granitic rocks. Until mid-Cambrian times it was filled
by up to 5000 m of coarse clastic sediments accom
panied by intrusive and extrusive volcanic rocks.
From Late Cambrian to Late Ordovician times the
trough evolved into a broad downwarp which was
filled by up to 3100m of carbonates. The rate of
subsidence then decreased, and the thickness of
Silurian and Devonian sediments was only some
Aulacogen 355 m, being the same as that on the surrounding
platform. In Late Palaeozoic times the aulacogen was
subjected to mild compression and fractured by a
braided system of faults in which movement was
both vertical and transcurrent. This gave rise to a
complex configuration of paired uplifts and fault
bounded basins in which up to 7000 m of Carbonif
erous elastics were deposited. This compressional
episode corresponds to the major deformation ofthe
Appalachian Fold Belt as a proto-Atlantic Ocean
closed.

10.3 Continental splitting

Continental splitting is an inherent part of the
Wilsonian cycle of plate evolution (Section 10.4)
and involves major reorientations of global plate
Aulacogen
motions. This also implies major changes in the
nature and directions of the forces responsible for
plate movement (Section 11.6). The mechanism of
continental break-up is only poorly understood.

10.3.1 Previous theories

Runcorn (1962) suggested that continental splitting

Fig.10.12 (a) Rift-rift-rift triple junction.

(b) Aulacogen forms by failure of one ofthe arms shown
in (a),(c) Aulacogen forms as the failed arm of an
unstable ridge-ridge-transform fault triple junction
((b) and (c) redrawn from Dickinson, 1974).
 CONTINENTAL RIFTS, CONTINENTAL MARGINS AND SEDIMENTARY BASINS
Volcano—
(a)
Sediment transport
Miogeocline
Fig. 10.13 Schematic origin and
evolution of plume-generated
triple junctions (redrawn from
Burke & Dewey, 1973). (f)
occurs as a result of the changing pattern of convec
tion in the mantle associated with the growth of the
core (Fig. 10.18).The growth ofthe core would have
progressively decreased the volume ofthe mantle and
so controlled the extent of the region in which
convection could take place. The width of convec
tion cells is of the same order as their thickness.
Hence, early in the history of the Earth the mantle
would have been considerably thicker than today and
so would have been populated by a few large convec
tion cells. As the core grew and the mantle became
smaller there would have been a series of times at
which the convection cells became smaller and more
numerous. Runcorn suggested that each of these
219
Axial Dyke
rrr junction
Riftyalley crust rrrjunction
Foreland basin
Thrust sheets
(g)
times would represent the initiation of a phase of
splitting of continents sited over the rising limbs of
adjacent convection cells. This mechanism is no
longer valid as it is now known that the core stabi
lized early in the history ofthe Earth. It also suggests
that plate motion is via the mantle drag mechanism
(Section 11.7.1), which is unlikely.
Oxburgh & Turcotte (1974) have proposed that
the lithospheric stresses which give rise to continen
tal splitting are only indirectly related to the forces
which cause plate motion. They suggested that these
stresses originate because of the Earth's oblate ellip
soidal form, whose radius ofcurvature at the poles is
greater than that at the equator. When a plate moves
220 CHAPTER 10
Alkaline
magmatism
wr¥:yx*xW:';":
xwr-x-wy:-:-
'•*,*»**w***»\'»\*»w»\***0*»*»*»***»
"^A~ Continent Ocean il
SSi:::!®:!;
Hotspot
Fig. 10.14 Evolution of a continental rift to continental
margins by the concatenation of a series of rift-rift-rift
triple junctions (redrawn from Dewey 8c Burke, 1974).
longitudinally across the Earth's surface it must ac
commodate to the changing radius ofcurvature, and
this results in intraplate stresses known as membrane
stresses^ which originate as a result of membrane
tectonics (Fig. 10.19). When a plate moves from
equator to pole,compressional stresses are generated
in the centre of the plate as it attempts to adjust to a
greater radius of curvature, and radial graben are
formed at the edges ofthe plate. When a plate moves
towards the equator, however,the stresses generated
in the centre of the plate are tensional and radial
compression ridges form at the edges ofthe plate. It
was proposed that many types of propagating frac
ture could form in this way, such as rift valleys and
intraplate volcanic zones. The latter phenomena are
xvivX::
5%:xSixSxSSxix
V.V.V.V.V.V.V.•••••.V.V.'.V
X-X-XvX*X*X'X'X*X*XvX
:-Xy:;::Xv:::::::x
XWX'X'X-X'SWW^^
X'I'X'X-X •x*x-x*:X'X
River flowing down
failed arm
X'I'X'X'
:'X'x:::x:x>:::::
xwxWx
x«
- Ridge
- :wx<Wx::
more commonly interpreted in terms of the passage
of the lithosphere over nearly stationary hotspots
(Section 5.4), and a membrane tectonic origin does
not appear likely as the linear volcanic zones are,
within a single plate, all elongated in the direction of
motion of that plate rather than forming a radial
pattern. Oxburgh & Turcotte suggested that the East
African rift system was formed as a result of Africa's
northward motion over the past 100 Ma,as revealed
by palaeomagnetic investigations. Tensile stresses
were generated as Africa moved towards the equator
so that the rift propagated southwards with time.
There is no evidence, however, of any other radial
cracks expected to be formed by this mechanism.
Membrane tectonics consequently does not re
quire the active participation of the mantle in the
rifting process, the only contribution being passive
as mantle material wells up into the propagating rift.
It would seem that membrane stresses are neither
 CONTINENTAL RIFTS, CONTINENTAL MARGINS AND SEDIMENTARY BASINS 221

which strike at high angles to the incipient rift, as

Barentz these control the location of transform faults which
ultimately offset the spreading centre in the newly
Nares
formed ocean basin (Section 7.2). There are other
examples, however, in which existing geological fea
tures have not determined the locus of splitting. It
Central North Sea seems, therefore, that the location of continental
North America Europe break-up is most closely controlled by the contempo
Biscay
Nova Scotia rary stress field, and older structures are utilized only
Connecticut Grand Banks
Newark
when they are close to the incipient rift.
The two major requirements for continental split-
ing are, first, the existence of a horizontal deviatoric
Africa
tension sufficient to rupture the lithosphere (devia
Casamance
toric stress represents the departure ofa stress system
Abidjan from symmetry and controls the extent of shape
Maracaitx)
change or distortion in a body, e.g. Park, 1983) and
Takatu ^
Amaz(^ Sao Lui^ second, a sufficient supply of basaltic magma in the
Reconcavo Gabon upper mantle to form a new ocean.
South America For the time being. East Africa will be taken as the
model for continental rifting, although it will be
Luanda
shown later that this view may be erroneous. East
Africa is characterized by an epeirogenic uplift of
some 2km and copious basaltic igneous activity.
Argentina
Geological evidence suggests that rifting followed a
phase of crustal doming, which was, in turn, pre
ceded by volcanism, with magma derived from the
Falkland) mantle (Davidson 8c Rex, 1980). From this one may
Plateau^-'"
infer that the primary cause ofthe initiation ofrifting
is the presence of some sort of thermal anomaly in
Fig. 10.15 Aulacogens developed around the Atlantic the upper mantle (Kusznir 8c Park, 1984), which is a
Ocean as a result of the model shown in Fig. 10.14
common feature of many rifts (Morgan, 1983; Ben-
(redrawn from Burke, 1976).
Avraham 8c von Herzen, 1987;Hooper, 1990). This
is in accord with the recognition ofthe lithosphere-
sufficiently large nor sufficiently persistent to be asthenosphere boundary as being temperature con
responsible for continental rifting. The rate of prop trolled. Any increase in temperature would cause this
agation is so slow that the membrane stresses are boundary to rise and consequently thin the litho
likely to be dissipated by creep. sphere (Spohn 8c Schubert, 1982). The thermal
anomaly most probably originates from hotspot ac
tivity. True convection is rare in the mantle beneath
10.3.2 Requirements for
continents, as much ofthe observed heat flow origin
continental splitting
ates within the crust (Section 11.4). However, there
The location of continental splitting may be con must be some sub-crustal component of heat flow,
trolled to some extent by the presence of ancient and this is released by spasmodic hotspot igneous
structural lineations within the continental plate activity (Crough, 1979). Basaltic magma is conse
(Windley, 1984), such as the trends and tectonic quently supplied by the partial fusion of mantle
grain of Precambrian mobile belts and sutures. For material as it undergoes adiabatic decompression on
example, the opening of the North Atlantic in the ascent.

Mesozoic took place approximately along the suture A deviatoric tension of 100-200 MPa is required
ofthe proto-Atlantic that closed in Palaeozoic times. to rift the continental lithosphere. The horizontal
Wilson (1965) also drew attention to lineaments drag on the base ofthe lithosphere resulting from the
222 CHAPTER 10

Late Proterozoic-Middle Cambrian

Late Cambrian-Early Devonian

Late Devonian-Mississippian

Pennsylvanian-Permian

Quartzite Marine shale Conglomerate

Rhyolite, basalt Marine shale with
hypabyssal sills, Marine carbonates sandstone and
tuffs, sediments conglomerate

Granitic basement

Fig. 10.16 Schematic transverse sections illustrating the

evolution ofthe Southern Oklahoma Aulacogen (after
Ham,1969).
 CONTINENTAL RIFTS, CONTINENTAL MARGINS AND SEDIMENTARY BASINS 223

Fold Belt

North American Platform

Cordilleran
Fold Belt
#7 I \ '9'°

Southern
Oklahoma
\ w Ouachlta
Aulacogen
Fold Belt

V Gulf coastal plain

500 km

Fig. 10.17 Major geological structures ofsouthern North
America.
lateral spreading of material welling up from the
hotspot is unlikely to be large enough to cause such
tension as the material is oflow viscosity and so would
not effectively couple with the overlying lithosphere.
response to the uplift associated with the buoyant
ascent ofthe mantle plume. This would be in accord
with the observation that crustal doming precedes
rifting. Tensional stresses would then be created as
the lithosphere was squeezed between the weight
of the uplift and the upthrust of the low density
material associated with the mantle plume. If the

10.3.3 The origin of regional

continental tension Fig. 10.18 Pattern of mantle-wide convection controlled
by the size ofthe core (after Runcorn, 1962, with
It is possible that the tensile stress is generated in permission from Academic Press).

0
0
224 CHAPTER 10
whole lithosphere behaved in an elastic manner, an
uplift of 2km would produce a tensional stress of
only SOMPa. Bott & Kusznir (1984), however,
recognized that the crustal component of the litho
sphere was subdivided into an upper, brittle layer and
a lower plastic layer. Consequently, only the upper
crustal component ofthe lithosphere would respond
elasticaily to stress. This would have the effect of
concentrating the tensional stress in the upper crust
as the lower crust and lithospheric mantle deformed
by creep(Fig. 10.20).The amplification ofthe tensile
stress by its concentration in the upper crust would
increase the stress to a value ofsome 400 MPa,which
is more than enough to cause rifting.
This mechanism of stress concentration is clearly
effective in initiating the rifting process. The stress
would, however, be confined to the area of plateau
uplift, and no explanation is provided for the gener
t40
<D
D compensation in the upper mantle. The
80 ^100 MPa
400 km
Fig. 10.19 Mechanism of membrane
tectonics,(a) and (b) refer to
continents moving to regions of higher
and lower radius ofcurvature
respectively(redrawn from Oxburgh &
Turcotte, 1974, with permission from
Elsevier).
ation ofregional stress whereby a whole continent is
placed under tension. Also, previous episodes of
rifting were not always associated with uplift, such as
the rifting ofthe North Sea in Mesozoic times.
The notion that hotspots are responsible for conti
nental rupture has been expanded by Burke (1976;
1977). He overcomes the objection that rifting is
confined to the region of uplift associated with a
hotspot by postulating that continental splitting
takes place in response to the tensional forces gener
ated by a series ofhotspots which form a line marking
the location of the initial rift (Fig. 10.14). Each
hotspot would initially give rise to a rift-rift-rift
triple junction (Burke & Dewey, 1973). A modern
example of such evolution is the Afar rift system of
Djibouti (Barberi et 1974), at which sea floor
spreading is taking place at two arms in the Red Sea
and Gulf of Aden and rifting on the third—the
Visco'elastic Fig. 10.20 Deviatoric stresses in the
Mono lithosphere resulting from a plateau
uplift of2km and the upthrust of its
Viscoelastic
subdivision of the crust into an upper
elastic layer 20 km thick and a
viscoelastic lower layer has the effect of
concentrating stress in the upper crust
(redrawn from Bott, 1981, with
permission from Elsevier).
 CONTINENTAL RIFTS, CONTINENTAL MARGINS AND SEDIMENTARY BASINS
, Alpine erogenic belt
if 1
I(I
Fig. 10.21 Simplified plate tectonic framework ofthe
Red Sea—Gulf of Aden area. Dark areas, oceanic crust;
light stippled areas, transitional crust consisting of
stretched, thinned continental crust invaded by basaltic
dykes (redrawn from Bonatti, 1987, with permission from
Nature 330,692, Copyright © 1987 Macmillan
Magazines Ltd).
Ethiopian East African rift (Fig. 10.21). Subse
quently, one arm fails to evolve, giving rise to an
aulacogen (Section 10.2), while the other pairs of
rifts continue to evolve and link up to give rise to a
rift system. This mechanism appears to explain quite
satisfactorily the initial rifting of the Atlantic Ocean.
It does, however, require the fortuitous simulta
neous occurrence of a linear series of hotspots be
neath the same continental plate, a configuration
which seems to have only a low probability.
Sawkins(1976)suggests that the concentration of
hotspots beneath supercontinents is to be expected,
as continental assembly would move all oceanic
spreading centres to areas geographically remote
from the supercontinent. The lack of access to heat
dissipating systems and the thermal insulation of the
continental lithosphere might necessitate the gener
ation of new heat dissipation mechanisms and the
development of hotspots in the mantle. Hoffman
(1989) has suggested that this mechanism was active
in the evolution of Laurentia. This may help to
225
explain the relatively short-lived nature of large
continental assemblages in Phanerozoic times. The
major drawback to the multiple hotspot hypothesis,
as mentioned earlier, is that not all past episodes of
rifting were associated with uplift. Consequently, a
means of placing a continental plate under tension is
required which has a greater likelihood of frequent
occurrence, and which is not necessarily associated
with igneous activity.
A possible mechanism for placing a whole conti
nent under regional tension is provided by the trench
suction force (Section 11.6), which affects the region
landward of the overriding plate at a subduction
zone. This effect would be most pronounced if a
continental plate were being underthrust on oppo
site sides by oceanic lithosphere. The combined
trench suction of both subduction systems would
then throw the whole of the intervening continental
plate into tension. An argument against this mecha
nism, however, is that Africa is surrounded to the
west, south and east by ocean ridges, and only the
northern boundary with the European Plate is a
subduction zone. Africa, therefore, cannot be placed
in tension at right angles to the East African rift
system by trench suction forces and,indeed, none of
the present continental plates is in such a position. In
spite of this, trench suction appears to be the only
realistic means of generating regional tensional
stresses within a continent, and so none of the
present continents would be expected to be in a state
of incipient splitting. The use of East Africa as a
model of continental rupture may be incorrect, and
this will be discussed in more detail later.
10.3.4 Evidence from continental margins
Evidence on the rifting process should be available
from passive continental margins, that is, continental
margins which are now within plate interiors which
were once the locus of continental rupture, such as
the eastern margin of the Americas or the western
margins of Europe and Africa.
Investigation of the northern margin of the Bay of
Biscay during DSDP leg 48 (Montadert et al., 1979;
Whitmarsh^r»/., 1986) has revealed no indication of
a major rift valley or extensive igneous activity. There
is similarly no evidence of regional uplift. Indeed,
splitting appears to have been associated with subsid
ence of about 2 km.
At other passive margins continental volcanism
was widespread during rupture. For example, the
226 CHAPTER 10
Tertiary northeast Atlantic igneous province extends
from west Greenland to Scotland and is character
ized by volcanism which correlates in time with the
initiation of opening of the North Atlantic (White
et al.^ 1987). This region, then, was most probably
underlain by a hotspot which became active some
60 Ma ago and had moved away from it after a period
of some 3 Ma or less. This hotspot now underlies
Iceland, and was responsible for the creation of the
Greenland-Scotland Ridge (Bott, 1983). The differ
ence between volcanic and non-volcanic margins
thus arises because ofthe difference in temperature of
the underlying asthenosphere. Mutter (1988b)
suggest that the copious igneous material emplaced
at volcanic margins arises from convective partial
melting ofthe upper mantle.
The evidence from volcanic margins argues against
the adoption of the East African rift system as a
model of continental rupture. In East Africa, for
example, rifting commenced 25 Ma ago, but separa
tion is still not complete;the North Atlantic splitting
implies that rifting should have been completed
within a timespan ofonly a few million years. Pavoni
(1993) has suggested that the horizontal extensional
regime of Africa originates from a large mantle
plume;it is probable that the tractional forces on the
base of the lithosphere arising in this way are too
small to initiate full continental break-up.
10.3.5 Models of continental splitting
The most reasonable model for continental splitting
seems to require a regional tension to exist, affecting
the whole plate, and on this is superimposed a
localized region of hotspot-related uplift which is
responsible for the initial rupture of the lithosphere.
Once ruptured at a single point, the regional tension
would cause the rupture to propagate rapidly at right
angles to its direction, with the location of the rift
controlled to some extent by regional lines of weak
ness in the plate which trend in the appropriate
directions.
Bott(1982b; 1992)has recognized three principal
stages in the rifting process (Fig. 10.22):
1 The continental lithosphere is thrown into a state
of tension by the trench suction force acting on
opposite edges ofthe plate (Fig. 10.22a).
2 A hotspot develops in the upper mantle by local
ized convection. This thins, and consequently weak
ens, the lithosphere. The resulting uplift and its hot.
low density compensation then increase the tension
about the hotspot, leading to rifting and the intru
sion of dykes into the rupture (Fig. 10.22b).
3 Once commenced,the split propagates away from
the hotspot as the tensional forces are concentrated
at the tips of the rupture (Fig. 10.22c). It eventually
reaches the margins ofthe plate and splits it into two.
The tensile stress then gives over to compression near
the rupture as the ridge push force is initiated.
It is thus apparent that this model predicts the
development of two types of rift: those associated
with doming, extensive volcanism and late stage
rifting, and those characterized by initial graben
formation with little or no volcanism. These have
been termed 'mantle active' and 'mantle passive'
rifts, respectively, and the model predicts that the
former are related to the phase ofinitial rifting under
the influence of hotspot activity and the latter to
propagation of the rift through a continent placed
under tension by plate driving forces. This is not the
only explanation, however, and Dunbar & Sawyer
(1988) have proposed that both styles of rifting can
occur in response to regional horizontal stresses. The
different rift types are proposed to be related to
different kinds of pre-existing weaknesses in the con
tinental lithosphere, with volcanic rifts formed over
areas of mantle weakness and non-volcanic rifts over
weakness in the crust. Storey(1995)has summarized
the information available about the role of mantle
plumes in the break-up of Gondwana and has
demonstrated that not all episodes of continental
splitting are associated with plume activity.
A criticism of the model of continental splitting
described above is that it assumes the rifting process
to be infinitely rapid, so that sea floor spreading
commences simultaneously at every point along the
rift. This cannot be the case, and it appears likely that
the rift would propagate through the continent over
a finite period oftime. The process would be akin to
that proposed for propagating ocean ridges (Section
6.11), but differs in that the rift would propagate
independently through a continent, whereas in the
propagating ridge model one ridge propagates at the
expense of a nearby ridge receding at the same
velocity.
Courtillot (1982) and Courtillot & Vink (1983)
have described a simple model of rift propagation.
They envisage that a rift would probably form along
existing lines of weakness in a continent, such as old
sutures or hotspot tracks. Under extensional forces
 CONTINENTAL RIFTS, CONTINENTAL MARGINS AND SEDIMENTARY BASINS
(a)
Fig. 10.22 Model of continental
splitting:(a) continental lithosphere
under regional tension resulting
from the trench suction force acting
at opposite edges of a continental
plate;(b) thinning of the
lithosphere and increased tension (b)
arising from the development of a
hotspot in the upper mantle;
(c) dyke intrusion initiates
continental splitting, which rapidly
propagates laterally to the plate
X?X?X?X?Xm
boundaries. The tensile stresses
then give over to compressional
stresses near the split as the ridge- •x*x*x*x*x*x*x*x*.j.j.j.j.j.j*j.j.j.j.j.i
X%j.j.j.x.j.x<<<«X*X*X*X*X*X*X*X*I
push force from the new spreading
centre develops (redrawn from
Bott, 1982b, with permission from
Elsevier). (C)
the lithosphere deforms and thins by normal faulting
at shallow levels and plastic necking at depth, to
produce an extension ofa few tens ofkilometres over
a few tens of millions of years. Thinning reduces the
strength of large tracts of the future plate boundary
and rifting begins. However, the likely nonunifor-
mity of the lithosphere would cause the plates to
remain attached at a number of locked zones. Even
tually the extensional forces increase to the point at
which the tips of the rift begin to enter the locked
zones. The sequence of events during the rupture of
a locked zone is shown in Fig. 10.23. In Fig. 10.23a
the tip of a rift has reached both ends of a locked
zone. Oceanic crust is produced along the rift, lead
ing to stretching ofthe locked zone (Fig. 10.23b). In
Fig. 10.23c the rifts have penetrated the locked zone,
so that its outer parts are carried away by each plate.
The part ofthe locked zone not penetrated by the rift
227
Continental lithosphere In tension
Enhanced tension
wKw>>>>>w>?r!%wwW.v.wKwKv.''
Hotspot
Continental split
tension relieved
NNV.NNM«.5S%?.%M
XvX*X*X<*X
XjXWXfX
.W.V.%W.%%%V.V.I.5
X*X*X%*XvX*X*c
''%%%%%%%%%'X'X'X'X'X'X'X^'X**
Xt%:jIj.>x<<<<»X<*X<<*X*X*X*X;I;%J;I;I;I;
continues to stretch (Fig. 10.23d) until it too be
comes invaded by the rifts (Fig. 10.23e). The process
continues by incremental invasion of the rifts until
the tips of the rifts meet and the locked zone is
ruptured to produce two distorted zones on either
side of the ridge (Fig. I0.23f). The model is easily
modified to encompass continuous, rather than dis
crete, step-wise penetration of the locked zone. The
magnetic character of the deformed zones will be
different from that of typical oceanic crust, and may
well produce a magnetic quiet zone at the new
continental margin.
Evidence for this model is provided from a study of
the evolution of the Gulf of Aden (Fig. 10.24),
whose history can be deduced by the fitting of mag
netic anomalies on either side of the central ocean
ridge (Section 4.6). The ridge can be seen to propa
gate at a rate of 30 mm a"^ in a westerly direction as
228 CHAPTER 10
Locked zone
(c)
(d)
(e)
(f)
Fig. 10.23 Schematic illustration of the incremental
splitting of a Mocked zone'in a continent by a propagating
rift (redrawn from Courtillot & Vink, 1983).
the Gulf opens. Moreover, the magnetic anomalies
are interrupted at the coastal magnetic quiet zone so
that the oceanic margins become younger to the
west. The model ofa propagating rift is supported by
Continental crust
the seismicity of the region, in which earthquakes
cluster along the rift and extend into the Afar region
(Fig. 10.21) into which the rift is penetrating. Fur
ther evidence is provided by tholeiitic basalts in Afar,
which show strong affinities with ocean ridge basalts.
Another example ofa propagating rift may be present
in the GulfofCalifornia, but the intense interruption
of the ridge in this area by transform faulting makes
the magnetic anomalies difficult to interpret.
 CONTINENTAL RIFTS, CONTINENTAL MARGINS AND SEDIMENTARY BASINS 229

(a)10 Ma

(c)Present

Fig. 10.24 Reconstructions of the

Gulf of Aden at (a) 10 Ma;(b) 7 Ma;
(c) present. Magnetic anomalies are
shown in a schematic fashion. Solid
ornament represents active volcanic
ranges ahead of the propagating
ridge, diagonal ornament regions of
overlap (redrawn from Courtillot &
Vink, 1983).
230 CHAPTER 10
It is significant that the palaeogeographic recon
structions ofFig. 10.24a,b show an area ofoverlap of
the coastlines of Arabia and Africa to the NW of the
propagating rift tip. This slight misfit is partly a
consequence of the model, as the continental mar
gins would have formed diachronously as the rift
propagated westwards. The reconstruction thus at
tempts to fit features of slightly differing age. The
overlaps represent the stretching of the continental
crust which occurred as the crust was ruptured by the
propagating rift. Similar overlaps are present in the
reconstruction of the continents around the Atlantic
of Bullard et al. (1965) shown in Fig. 3.2, and are
apparent along the juxtaposed margins of eastern
South America and western Africa. It is possible that
these result from the opening of the South Atlantic
by rift propagation from south to north (Fig. 10.25),
accompanied by southerly rift propagation in the
North Atlantic (Bonatti & Crane, 1984). This would
be in accord with the observation that the oldest
magnetic anomalies are found only in the southern
most Atlantic. Courtillot (1982) has calculated that
the mean separation between these overlaps is some
700 km,and they tend to be about 400 km long and
150 km wide. Perhaps the best way of depicting
continental reconstructions taking continental dis
tortion on rifting into account is, according to Vink
(1982), to represent the pre-rift configuration by
matching the point ofinitial rifting and allowing the
overlap to increase in the direction of rift propaga
South America South America
tion (Fig. 10.26). The overlap would then be equiva
lent to the amount of extension of the continental
edges.
Models of continental rifting have largely been
derived from studies of old passive margins or conti
nental regions ofextension where sea floor spreading
has not yet commenced. There is, however, a re
stricted number of areas where the transition from
rifting to spreading can be observed in action. One
such area, for which geophysical data have recently
become available, is the Woodlark Basin off north
eastern Papua New Guinea (Taylor et a/., 1995). In
many respects,the scenario there is in accord with the
model of continental splittirig described above, that
is, the progressive truncation of magnetic lineations
and uplift above the propagating rift. Other observa
tions, however, indicate greater complexity than im
plied by the model. Rather than the rift tip's
propagating continuously, it appears that short
spreading segments can develop instantaneously,
probably reflecting the contrasting strengths of dif
ferent parts of the continental lithosphere. Fiarth-
quake activity along the developing passive margin is
not immediately localized at the plate boundary as
spreading commences. During its evolution, there is
evidence that the developing ocean ridge reoriented
instantaneously by about 5° by synchronous jump
ing during the recent past, possibly in response to
rapidly reorienting stresses. The data even cast some
doubt on the postulate that oceanic transform faults
Fig. 10.25 Opening of the South
Atlantic by a propagating ridge
(a) at 105 Ma,(b) at 100 Ma
(redrawn from Bonatti & Crane,
1984).
 CONTINENTAL RIFTS, CONTINENTAL MARGINS AND SEDIMENTARY BASINS
Amount of
extension
Original
continent "
(a)
Initial
rifting
h Rifting
(b)
Fig. 10.26 Propagating rift model for continental break
up (in plan),(a) Continent under tension,(b)Initial
rifting, parallel lines show amount of extension,
(c) Mid-rifting, with sea floor spreading in lower half and
extension in upper half,(d) Rifting complete,(e) 3 Ma
always reflect the original shape of the rift (Section
7.2). Clearly,then,our models ofcontinental rifting,
although generally reasonable, are not perfect in
detail. It is unlikely that perfection will be achieved
even with the availability of these new data, because
the continental lithosphere is so variable in character
that it is doubtful that a single model could satisfy all
cases.
10.3.6 Volcanic margins
It follows from the theory of continental splitting
(Bott, 1982b) described in Section 10.3.5 that the
continental margins ofan ocean formed by continen
tal rifting should be of two types: those associated
with volcanic products, situated close to the thermal
anomalies responsible for the initial rifting, and
those with no associated volcanism, which represent
231
;i Mid-rifting = 3 Ma after
complete
(C)
Reconstruction
Amount of
overlap <
complete
Closure
<d)i
after rifting complete,(f) Reconstruction of pre-rift
configuration. Extension due to rifting gives rise to an
apparent overlap (redrawn from Vink, 1982, with
permission from the American Geophysical Union).
the more normal areas of continental crust across
which the rift subsequently propagated. The former
have been termed volcanic margins.
An elegant comparison ofthe two types of margin
has been provided by the work of White & McKenzie
(1989; see also White, 1993), who have described
sections across the rifted continental margin of the
northeastern Atlantic (Fig. 10.27a) derived from
seismic surveys. The profile across Goban Spur
(Fig. 10.27c), situated at the mouth of the English
Channel, shows that the continent-oceanic transi
tion is marked by very little volcanic activity and that
continental rifting was dominated by extensional
faulting and the tilting of blocks ofcrust. The profile
at Hatton Bank (Fig. 10.27b),at the Atlantic margin
between Scotland and Greenland, shows the pres
ence ofextensive lava flows,5 km in thickness, which
dip gently out to sea. At the base ofthe crust there is
232 CHAPTER 10

■ Sediment

Oceanic crust - ''

'''" ' ''
' %Xi
^ V Vw V V yX - - ■! ; Continental :
, , ^y ,V, Intrusive
. V .
,,V ^\ crust \ ~
\v „ igneous y
V rocks V
ICELAND
^ V V y V\
^ V ^

Sediment

; Fault
Tilted ^ -- : ; , ' r, : ■
fault X z'' ~''i
', block ^ ^ - X ■ 'X
\\ 'j^ ' L~, ^ Continental
- ,x- ' -' ; crust

Fig. 10.27 (a) Location map. Profiles across (b) the
volcanic margin at Hatton Bank; (c) the non-volcanic
margin at Goban Spur (after White, 1993).
an even larger body of igneous rock some four times
greater in volume than the volcanics. Subsequent
work has shown that very large tracts of the north
eastern Atlantic continental margin are associated
with these supra- and infracrustal igneous bodies
(Fig. 10.28).
The presence of extensive bodies of igneous rocks
at volcanic margins is explained by their proximity to
the hotspot responsible for the initial breaching of
the continental lithosphere. McKenzie & Bickle
(1988) have demonstrated that the amount of mol
ten rock generated as mantle material ascends and
melts due to the reduced pressure is strongly depen
dent upon the initial temperature of the mantle
material. Only very small increased temperatures, of
the order of 100-200 °C, are necessary for the
amount of molten material produced by this decom
pression melting to increase very dramatically. In
deed, it has been estimated that some 10^ km^ of
igneous rocks were generated when the Atlantic first
rifted. It is thus the presence of the causative hotspot
that is responsible for the differing nature of the
continental margins formed by rifting. Although the
hotspot, or mantle plume, was probably quite small,
of the order of a few tens of kilometres in horizontal
section, as it ascended in the mantle, modelling of its
behaviour as it reached the cold boundary layer at the
base of the lithosphere indicates that it would spread
considerable distances horizontally (Courtney 8c
White, 1986), producing a layer of hot material at
the top of the asthenosphere with a diameter of some
2000 km; in the North Atlantic the plume would
have been centred on Iceland. This behaviour ex
plains why the igneous products associated with the
plume occupy a broad area around it. The partition
ing of these products into volcanic and intrusive
fractions depends on their relative density; the major
ity of the igneous rocks are relatively dense and only a
small proportion are of sufficiently low density to rise
buoyantly thtough the crust to the surface. The
 CONTINENTAL RIFTS, CONTINENTAL MARGINS AND SEDIMENTARY BASINS 233

\^\roU_of_zoneof
^oornilousi^T^

Hotspot

Labradbr

Massive
layered
basalts

Abundant
volcanic
rocks

Vpring Plateau

j ^—v.^Hotspot
Fig. 10.28 (a) Reconstruction of
the North Atlantic region at 57 Ma.
(h)The present North Atlantic.The
locations of massive layered basalts, Hatton
Bank
abundant volcanic rocks,the
hotspot and its associated region of
anomalously hot mantle are shown
(after White & McKenzie, 1989).

thickening of the crust also results in less subsidence ic margins are present at many continental margins
than seen at non-volcanic margins because of isos- world-wide at the location of mantle plumes,and the
tatic effects (Section 2.11). concept of mantle plume-related volcanic margins
Subsequent investigations have shown that volcan has been widely embraced. However, there are some
234 CHAPTER 10
^o^vwacting
B C
Incipient
subduction
^^'Oanding oc®^^
(a) (b)
Fig.10.29 The Wilsonian cycle,(a) Continents AB and
CD separated by expanding and contracting oceans AD
and BC,respectively,(b) Collision of AB and CD as BC
closes completely; initiation of subduction beneath A and
D.(c) Splitting ofsupercontinent ABCD into two new
continents AB' and CD and the growth of a new
expanding ocean B'C. Ocean AD starts to contract.
inconsistencies to this simple idea. The volcanic
sequences do not thin with distance from their re
lated hotspot as might be expected. The effects ofthe
hotspots appear to diminish very rapidly with time
after rifting and normal oceanic crust is found imme
diately adjacent to volcanic margins; it might be
expected that the effects of the hotspot would have
been longer-lived. There are also locations where
volcanic sequences appear to be present with no
accompanying hotspot, for example the passive mar
gin of the eastern USA (Holbrook & Kelemen,
1993). Indeed, in this location the generation of the
observed lower crustal seismic velocities would seem
to require a hotspot at 550°C above normal. It thus
appears that the division of passive continental mar
gins into volcanic and non-volcanic types is too
simplistic.
Also associated with mantle plumes are thick,
extensive flood basalts (Fig. 5.9), such as the Deccan
Traps of western India and the Siberian Traps of
northern Russia. These represent the volcanic expres
sion of plumes which were often associated with
continental rifting (Section 10.3.4). Cox (1989) has
demonstrated that the topography associated with
such plumes is still in evidence today, and is respon
sible for characteristic drainage patterns,for example
New
Collision expanding
BC ocean
Incipient rift
(c)
the predominantly eastern flow of rivers in India is
related to drainage down the eastern flank of a dome
associated with the plume responsible for the Deccan
Traps.
10.4 The Wilsonian cycle
The Pacific Ocean opened about 300 Ma ago and will
close completely in a further 200 Ma, so that the
lifetime of an ocean basin is about 500 Ma. Plate
tectonic processes, however, have probably been
active at least since Early Proterozoic times, some
2500 Ma ago; consequently, there must have been
several cycles of ocean creation and destruction dur
ing the Earth's history. Very little remains of these
ancient oceans, although fragments are preserved in
ophiolite assemblages (Section 2.5), but their exist
ence is implied by continental reconstructions.
The periodicity of ocean formation and closure
is known as the Wilsonian cycle (Dewey & Burke,
1974), named after J. Tuzo Wilson in recognition
of his major contribution to the theory of plate
tectonics.
Figure 10.29a shows a section through an Earth
on whose surface exist two continents AB and CD.
Ocean AD is expanding and BC contracting. Some
time later (Fig. 10.29b) ocean BC closes completely,
and this phase of plate movements ceases with the
collision of margins B and C, the construction of a
Himalayan-type mountain range at the suture BC
and the formation of a supercontinent AD. Ocean
AD must develop subduction zones at its margins to
maintain constant global surface area, although the
reasons for the initiation of subduction are un-
 CONTINENTAL RIFTS, CONTINENTAL MARGINS AND SEDIMENTARY BASINS
known; sediment loading does not appear to be the
cause (Cloetingh et al.^ 1984). The trench suction
force (Section 11.6) places the supercontinent AD
under tension.Ifother conditions are extant(Section
10.3.2), the supercontinent will rift (Fig. 10.29c) at
B'C, not necessarily along the previous suture BC.
Two continents AB' and CD now separate across a
new expanding ocean B'C, and to compensate for
the newly formed oceanic lithosphere, ocean AD
must start contracting by the continued develop
ment of subduction zones at continental margins A
and D. This phase will continue until A and D
collide. Vink et al. (1984) have demonstrated that
the strength of the continental lithosphere decreases
as the thickness of the crust increases. Conse
quently, the large crustal thicknesses associated with
suture zones make them likely locations for future
rifting. There will thus be a tendency for new oceans
to open along the suture where a former ocean
closed.
Present day analogues of the oceans involved in
the first and last phases ofFig. 10.29 are: Fig. 10.29a,
AD = Atlantic Ocean, BC = Pacific Ocean; Fig.
10.29c, AD = Pacific Ocean, B'C = Red Sea.
The Wilsonian cycle will continue to operate until
the concentration of radioactive elements within the
Earth becomes too low to provide sufficient heat
energy for the driving of plate motions.
Gurnis(1988) has made numerical simulations of
continental aggregation and dispersal, and has postu
lated that plate motion during the Wilsonian cycle is
episodic rather than continuous. He suggests that
continents tend to aggregate over cold downwellings
in the mantle, where they act as an insulating blanket
(Anderson, 1982). The mantle consequently heats
up, altering the convection pattern, and the super-
continent splits in response to the resulting tension.
The continental fragments then move towards the
new cold downwellings resulting from the changed
convective regime. Gurnis points to the fact that the
continents, except Africa, are currently moving to
cold regions of the mantle, characterized by geoid
lows, few hotspots and high seismic velocities. Pan-
gaea was probably positioned, 200 Ma ago, over
what is presently the Atlantic-African geoid high.
Since Africa has moved only slowly with respect to
the hotspot reference frame, it appears that Pangaea
may have been situated over a geoid high prior to
break-up, in accord with the model. It would thus
appear that the pattern of mantle convection may be
235
controlled, to some extent, by the formation of
supercontinental assemblages.
10.5 Vertical movements
associated with plate tectonics
10.5.1 Introduction
The study of plate tectonic processes is frequently
concerned with the horizontal movements of plates
over the Earth's surface. It must be recognized,
however, that associated with these horizontal mo
tions are vertical movements, which play a very im
portant role in the evolution of the Earth. Many of
these vertical movements are implicit in the phenom
ena described in previous chapters of this book, and
will be briefly listed. Others have not been previously
considered, in particular the predominantly down
ward motion of the continental lithosphere associ
ated with sedimentary basin formation, and these
will be discussed in more detail.
Hotspots (Section 5.4) produce vertical move
ments of both continental and oceanic lithosphere
because of thermal expansion and because they iso-
statically support positive, topography by a form of
Pratt-type compensation (Section 2.11.3). In the
ocean basins, passage of the lithosphere over a hot-
spot creates an aseismic ridge, which progressively
subsides as it moves away from the hotspot. Beneath
continents, hotspots create pleateau uplifts which
may subsequently rupture to produce rift valley
systems(Section 10.1.6) and, ultimately, aulacogens
(Section 10.2)or complete lithospheric rupture(Sec
tion 10.3).
In the oceans, the elevation of ocean ridges is
similarly supported isostatically by a hot,low density
region in the upper mantle (Section 6.3), and the
gross topography ofthe sea floor is controlled by the
progressive cooling of the lithosphere as it moves
away from the ridge (Section 6.9). Similarly, the
juxtaposition of oceanic lithosphere ofdifferent ages
at fracture zones produces abrupt scarps on the sea
bed (Section 7.3). The morphology of subduction
zones (Section 8.1) is controlled by plate motions.
The accretionary wedge(Section 8.7)is progressively
built up by the stacking of thrust slices of sediment
scraped off the top of the downgoing slab by the
overriding plate. Island arcs maintain their positive
topography by the continuous construction of
the volcanic edifice by the addition of magmatic
236 CHAPTER 10
products produced in the upper mantle immediately
above the BeniofF zone. Marginal seas and back-arc
ridges owe their negative and positive topographies
to tension in the rear-arc area generated as a result of
plate movements (Section 8.9).
On the continents, vertical movements can occur
in a number of different environments. At Andean-
type plate margins (Section 9.2) a mountain range
develops as the result of the upward migration of
magmas from the Benioffzone in a fashion similar to
island arcs but accompanied by large volumes of
acidic material. In collisional orogens (Section 9.3)
Himalayan-type mountain ranges grow vertically by
the stacking of thrust sheets in the upper crust and
ductile movements at depth. In the distal regions
of such orogens, impactogens (Section 10.1.3) may
be developed in response to tensional stresses associ
ated with collision. Transform and transcurrent
faults which have a sinuous trend may be associated
with vertical movements resulting from transpres-
sion and transtension (Section 7.4.2) giving rise to
small compressional uplifts and pull-apart basins,
respectively.
10.5.2 Sedimentary basin formation
A category of structures not covered in the previous
section concerns those sedimentary basins which
occur in intraplate environments out of the direct
influence of plate margin processes. The problem of
their formaLtion was first recognized in the 19th
century when it was observed that more than 12 km
shallow water sediments had been deposited in the
Appalachian geosyncline. It can be shown, however,
that the sediment loading itself cannot cause the
basin subsidence, as described below.
In xhc gravity loading hypothesis it is assumed that
the lithosphere reacts by faulting or flexing solely in
response to the sediment load. By using either an
Airy-type model, in which isostatic equilibrium is
maintained by movement along vertical faults, or a
flexural model, it is possible to predict, if the densi
ties of sediment, crust and mantle are known, the
depth to basement at any stage in the filling of the
basin (Steckler & Watts, 1978). It has been found
that the Airy-type model predicts a sediment fill of
only about 2.4 times the initial water depth before all
the available water is displaced by sediment. Since the
shallow water sediments were deposited in water less
than 150 m deep, a thickness of only about 350 m
would be expected. More sediment can be accom
modated in the flexural model, as the margins ofthe
basin are able to prograde seawards(Walcott, 1972).
However, both models suggest that gravity loading
causes a basin subsidence which is far smaller than
the observed sediment thickness.
Watts & Ryan (1976) applied this method to the
Gulf of Lion, and progressively backstripped the
sediment layers to determine the basement depth
throughout the history ofthe basin. When compared
with known basement depths (Fig. 10.30), they
found that the subsidence had been much greater
than would have arisen from the sedimentary load
ing. This additional loading was attributed to some
form oftectonic subsidence or'driving force',so that
the main role of the sediment loading is to amplify
the subsidence produced by this primary mechanism.
In many cases this driving force originates directly or
indirectly from plate motions.
Sleep (1971) and Sleep & Snell (1976) attributed
the subsidence of basins situated on continental
margins to simple thermal contraction following
continental rupture. Thermal expansion ofthe litho
sphere would have taken place during rifting as a
thermal anomaly develops under the continent. This
expansion causes a plateau uplift which is isostatically
supported by the hot,low density region in the upper
mantle. As rifting is completed and an ocean grows,
the continental margin becomes removed from the
hot mantle, and subsidence takes place. Some thin
ning of the crust is necessary for subsidence to
continue below sea level, as otherwise the crust
would only sink back to its pre-rifting level. This, in
fact, is a major criticism of this mechanism of basin
subsidence as there is often no evidence of the
extensive supra-crustal erosion necessary to produce
the observed subsidence, and sub-crustal erosion
seems an unlikely process.
Bott (1971) has presented a different model for
the development ofsedimentary basins on the conti
nental shelf. Their development is viewed as the
response of the crust after rifting to the progressive
loss ofgravitational energy associated with the juxta
position of thick continental crust and thin oceanic
crust. At that time, geophysical observations had
suggested that the transition from continental to
oceanic crust, which was quite abrupt immediately
after rifting, actually takes place over a distance of
50-100 km,although more recent data indicate that
this is only the case for continental margins not
 CONTINENTAL RIFTS, CONTINENTAL MARGINS AND SEDIMENTARY BASINS 237

Tramontine Mistral Autan

Age(Ma) Age(Ma) Age(Ma)

20 10 0

Local
loading

E 1
Flexural ^
loading g

Subsidence due to
Ocean ridge
sediment loading

\//\ Due to driving forces

Fig. 10.30 Summary curves for the subsidence history
recorded in boreholes in the Gulf of Lion. Gaps in the
data correspond to the Mediterranean salinity crisis
between 7 and 5 Ma ago. Local and flexural loading
models are shown (redrawn from Watts & Ryan, 1976,
with permission from Elsevier).
affected by extensive volcanism (Mutter et al.^
1988b). Bott suggested that the continent-ocean
transition becomes gradational with time, and the
readjustment allows the development ofsedimentary
basins parallel to the continental margin. When
rifting takes place, the whole region is uplifted in
response to the hot,low density material in the upper
mantle. Regional subsidence then occurs as the crust
moves away from the thermal anomaly. The high
temperatures existing at this time permit a broaden
ing of the continent-ocean transition by hot creep
from the base of the continental crust towards the
ocean (Fig. 10.31). The juxtaposition of thick, low
density continental crust and thin, dense oceanic
crust causes the continental crust to be thrown into
tension. To maintain isostatic equilibrium as lower
crustal material moves oceanwards, the continental
shelfsubsides and the continental slope and adjacent
oceanic crust rise. These crustal movements are bal
anced by the flow of asthenosphere inwards beneath
the continent. The brittle upper crust responds to
these movements by faulting and the development of
basins on the continental shelf by wedge subsidence.
238 CHAPTER 10

Rise Slope Shelf

Brittle layer

creep; Ductile layer

"E Crust-mantle mix

Fig. 10.31 Development of

Q f Lithosphere
continental margin basins by
oceanward creep of the lower
continental crust and graben
formation in the brittle upper crust
(redrawn from Bott, 1971, with
1000 km permission from Elsevier).

The development of these basins is thus associated and the intrusion of granites during the extensional
with the maintenance of isostatic equilibrium as the episoide associated with continental rupture. The
potential gravitational energy associated with the intrusion of igneous material would have weakened
margin is gradually dissipated. the continental lithosphere and provided a zone of
The above mechanism cannot provide an explana regional stretching in which cratonic basins formed
tion for the formation of basins far from plate mar almost simultaneously over a wide region.
gins. Haxbyef a/. (1976) have proposed a model for
these intraplate basins with reference to the Michigan
Basin. Basin subsidence is attributed to the flexural
response of the continental lithosphere to a load
arising from the effects of a thermal anomaly in the
upper mantle. The process is initiated by the action
Hotspot,
of a hotspot which thins the lithosphere. so that
asthenospheric material ascends to the base of the
crust (Fig. 10.32a). The hotspot then causes thermal
metamorphism of the lower crust in which gabbroic
material is converted to garnet eclogite. Garnet Asthenospnere/
eclogite is denser than normal lower crust, and the
additional mass is initially supported by the negative
buoyancy of the hotspot. When the hotspot decays,
the low density support is removed and the litho
sphere subsides by flexing (Fig. 10.32b). Thus, basin
subsidence in this model is caused by the excess
weight of eclogite in the lower crust, and no major
deformation should be visible at the surface.
Klein & Hsui (1987) have noted that cratonic
basins of North America, Europe, Africa and South
America all formed during the period 550-500 Ma
and have similar sedimentation histories and dates of
regional unconformities. They suggest that their
Fig. 10.32 Model for the formation ofthe Michigan
common dates offormation may have arisen by their Basin:(a) development of a hotspot in the upper mantle
originating during the break-up of a Late Precam- and its ascent to the base of the crust which it
brian supercontinent (Section 12.2). The supercon- metamorphoses to eclogite;(b) removal ofthe hotspot
tinent may have acted like a heat lens which caused and sinking of the basin in flexural response to the load of
partial melting of the lower crust and upper mantle. eclogite (after Haxby era/., 1976).
 CONTINENTAL RIFTS, CONTINENTAL MARGINS AND SEDIMENTARY BASINS
Basin
Progressive subsidence Progressive denudation
aTid infiil by sediments
Sediment
illili Ductile inflow towards mountains ?
Fig. 10.33 Sedimentary basins paired with an eroding
mountain range (redrawn from Bott, 1964, with
permission from Nature 201, pp. 1082-4. Copyright ©
1964 Macmillan Magazines Ltd).
An alternative mechanism of intracratonic basin
subsidence has been proposed by Bott (1964) in
which sedimentary basins are paired with mountain
ranges(Fig. 10.33). The uplift ofa mountain range is
followed by rapid erosion, with the clastic sediments
transported into adjacent basins. Erosion causes the
mountain range to become overcompensated and
out of isostatic equilibrium. To regain equilibrium,
the mountain range rises and mantle material flows
beneath it. Renewed uplift promotes further erosion,
and so the cycle continues until the mountain range
disappears or the stress differences become too small
to cause further uplift. This behaviour differs slightly
from classic Airy isostasy in that the mantle flows in
from localized regions only, rather than from the
whole region beneath the depth of compensation,
which may be due to the mantle's possessing a small
finite strength, making unrestricted flow impossible.
The model was subsequently modified by Bott(1976)
in the recognition that only the upper crust can
behave in a brittle manner. Rather than occurring
within the mantle, the inflow of material at depth is
believed to occur in the ductile lower
(Fig. 10.34), which stretches and thins, producing
tension in the upper crust which responds by faulting.
One of the most widely accepted models of basin
formation is that ofMcKenzie (1978), which views
239
Mountain range Basin
Progressive subsidence
and isostatic uplift and infill by sediments
.Sediment
~ Crust"'~
, 2,8 Mg m~3
Mnhn \"/r O ^ 'i""'V \
[ Ductile inflow :
Upper mantle
0 50 km ij
3.3 Mg m~^
the subsidence as the response to stretching of the
continental lithosphere and subsequent cooling as it
relaxes to its predeformation state. On the initiation
of basin formation a unit length of lithosphere is
abruptly stretched to a length P (Fig. 10.35). Thin
ning of the lithosphere gives rise to normal faulting
and rapid subsidence, and is accompanied by an
upwelling of the asthenosphere. As the thermal per
turbation associated with the upwelling of the asthe
nosphere decays,the lithosphere thickens and further
slow subsidence takes place which is not accompa
nied by faulting. The crust remains permanently
thinned even after the decay of the thermal perturba
tion. Calculations based on an isostatically compens
ated system show that subsidence only takes place for
crustal thicknesses which were initially in excess of
18 km. The amount of subsidence and the heat flow
depend only on the amount of stretching.
Figure 10.36 shows the relationship between the
thermal component of subsidence and the square
Sediments
Brittle layer
jMotio
crust 1 -»■ Mantle
Fig. 10.34 Graben formation by faulting in the brittle
upper crust and ductile flow in the lower crust (redrawn
from Bott, 1976, with permission from Elsevier).
240 CHAPTER 10
r(°c)
(a) 0 1000
r\
(b)
.o[
pa-
a/p
I
•VSlli
t-^co
(c)
Fig. 10.35 Subsidence model combining stress and
thermally based mechanisms,(a) At time r = 0 a block of
continental crust is extended by a factor p, causing
subsidence,(b) Hot asthenosphere wells up to replace the
thinned lithosphere.(c) Further subsidence as the thermal
perturbation decays and the lithosphere thickens(redrawn
from McKenzie, 1978, with permission from Elsevier).
root oftime(P = oo corresponds to complete rupture
of the continental lithosphere and the formation of
oceanic lithosphere). The subsidence can be con
verted into sediment thickness by using an Airy or
flexural model of isostatic equilibrium. It requires,
however, knowledge of the sediment density and
degree of compaction so that such computations
must be performed separately for each basin. The
subsidence history of a basin determined from bore
holes can then be compared with the model of a
stretching event of given rate and duration (Jarvis &
McKenzie, 1980). Figure 10.37 shows the relation
ship between heat flux and time, which is similarly
dependent only on p. Consequently, if p is known
from investigations of subsidence history using
hJTime(Ma)
Fig. 10.36 Thermal subsidence as a function of the
square root oftime for the model shown in Fig. 10.35,
shown in terms of both water depth and sediment
thickness (redrawn from McKenzie, 1978, with
permission from Elsevier).
100-
E
§
E
(0
X
a>
p = 1.25
40 60
Time(Ma)
Fig. 10.37 Heat flux as a function oftime for various
values ofP for the model shown in Fig. 10.35 (redrawn
from McKenzie, 1978, with permission from Elsevier).
boreholes or from seismic refraction experiments
over stretched and adjacent unstretched areas, the
time variation of heat flux can can be estimated. This
has important ramifications, as heat flow controls
the reactions between organic materials which lead
to their maturation into hydrocarbons, and so
knowledge of p makes possible the prediction of
 CONTINENTAL RIFTS, CONTINENTAL MARGINS AND SEDIMENTARY BASINS
those basins which are likely to contain accumula
tions ofoil and gas(MacKenzie & McKenzie,1983).
The model does not take account ofthe radioactiv
ity of the continental crust and adopts a simplified
version of the variation of temperature with depth.
However,the inclusion of these effects is unlikely to
alter the general form of the results. The model has
been refined by White & McKenzie (1988), who
postulate a stretching ofthe lithospheric mantle over
a wider area than the crust in order to explain the
post-rift sedimentary onlap observed in many basins.
A major objection to the model is that observa
tions of sedimentary basins do not reveal the large
amounts of extension required. McKenzie (1978)
has presented evidence from several regions which
strongly suggests that the dip ofthe high angle faults
seen at the surface decreases with depth. In particu
lar, seismic reflection sections reveal that rotation of
the sedimentary reflectors has occurred, indicating
that the faults are listric. The curved nature of these
faults prevents accurate estimates being made of
their associated extension from surface observations
and it is proposed that listric faults conceal large
extensions.
Barton & Wood (1984) have tested McKenzie's
model in the evolution of the Cfentral Graben of the
North Sea (see also White, 1989), which is probably
an aulacogen (Section 10.2) associated with the
opening of the Atlantic. Stretching estimates were
obtained by seismic refraction experiments over
stretched and unstretched crust, and independently
from the history ofsubsidence revealed by well data.
Both methods suggest a P value of about 1.6 for the
last extensional event. Seismic refraction experi
ments in the Aegean Sea, a back-arc basin (Section
8.9) of the Hellenic Trench, suggest a P value of
about 2. These data and model computations sug
gest that large extensions, of the order of p = 2, are
required to produce a basin fill of 4.5 km thickness.
The lack of a proposed mechanism for the origin of
the regional stresses responsible for the extension is
the least satisfactory aspect of McKenzie's model.
Consideration ofthe forces acting on plates (Section
11.6) and the mechanism of continental splitting
suggests that there are only two environments in
which regional tensile stresses may occur in the
continental lithosphere: first, in the area on the
landward side of an Andean-type mountain range,
where the trench suction force creates regional ten
sion by a similar mechanism to that proposed for the
241
formation of marginal seas (Section 8.9); and sec
ond,in the neighbourhood ofan incipient continen
tal rift, by the mechanisms described in Section 10.3.
Consequently, the extensional stresses required in
McKenzie's model would only be expected to occur
within a few hundred kilometres of destructive or
incipient accretionary plate margins, or perhaps in
the vicinity of passive continental margins by the
mechanism proposed by Bott(1971). It is difficult to
envisage, however, how the mechanism could apply
to basins within the interior of continental plates
unaffected by indentation tectonics, especially since
focal mechanism solutions of earthquakes in this
environment indicate that they are subject to com-
pressional stresses.
McKenzie's model of basin formation involves
brittle failure of the upper crust and ductile (pure
shear)stretching ofthe lower crust and mantle litho
sphere. Wernicke (1985) has proposed a different
model which entails simple shear only, possibly af
fecting the whole ofthe lithosphere, along low angle
normal faults. The model was derived in explanation
of the structure of the Basin and Range Province of
the western USA (Wernicke, 1981), but it may also
be applicable to rifting in general (Section 10.1.6)
and the evolution of passive continental margins
(Lister et 1986). The model arose from the
recognition in the Basin and Range area of a region
ally extensive low angle detachment beneath normal
fault systems, flanked by an adjacent zone of ductile
deformation whose maximum elongation is in the
same direction as that implied by the normal faults.
The majority of ductile deformation was observed to
pre-date the episode of extension.
The model for the development of the simple
shear model is shown in Fig. 10.38. The undeformed
section of Fig. 10.38a is affected by a major, shal-
lowly dipping (10-30") fault penetrating the crust
and, perhaps, the whole lithosphere (Fig. 10.39).
Within the sedimentary basin so formed, coarse
clastic sediments accumulate. After continued fault
displacement (Fig. 10.38c), the sedimentary basin
and its basement undergo penetrative brittle defor
mation with the formation of a normal fault duplex,
having the internal geometry of a number of highly
rotated faults and thin fault blocks. Extension is not
sufficient to expose the basal detachment. The thin
ning ofthe hanging wall ofthe detachment resulting
from this extension causes the tectonics of the shear
zone to be raised into the zone of brittle behaviour. A
 242 CHAPTER 10

Time at_
)mm a~^ o
Future upper plate
of core complex

Future lower plate

exposed in core complex

K-15 km-H

Incipient faults

3 Ma

Imbricately 25km^-
extended range

Future "core-complex
8 Ma upper plate

32 km- H

"Breakaway" Large fault-block ranges

Core-complex"
range Basin range Basin ^ Ra«in
Ifi 17.5 19

14Ma
Denudation = 20 km

Depth = 25 km

Preorogenic datuhri

^
Orogenic c.astics: Firie
hoarseclastic-lacustrine
clastic
Total extension = 72 km (100%)
irrmn Highly attenuated'
rocks
Ductile shear zone

16.5 Paleo depth of "core-complex" lower plate

Fig. 10.38 Model of an extensionai simple shear system Wernicke, 1985, with permission from the National
in the upper and middle continental crust (redrawn from Research Council of Canada).
 CONTINENTAL RIFTS, CONTINENTAL MARGINS AND SEDIMENTARY BASINS
Upper
crustal Thinnest upper crustal 'Moho Topographic
breakaway Extensional crust
allochthons
Proximal
Mantle
80
Uthos
Asthenosph
+ +
Zone A ZoneB
Fig. 10.39 Model of an extensional simple shear system
involving the entire lithosphere (redrawn from Wenicke,
1985, with permission from the National Research
Council of Canada).
part of the upper plate may begin to extend beyond
adjacent areas (Fig. lO.SSd) and, if the extension is
sufficiently large, folds may form in the detachment
as a result of the isostatic rebound of the unloaded
footwall. Old rocks then become exposed in the
cores of anticlines. These 'core-complex' ranges are
highly extended regions >vhich passed through the
stage of imbricate normal faulting before becoming
thinned by low angle normal faulting to expose the
footwall of the detachment. Large fault-block ranges
occur adjacent to the core complex, which exhibit
little internal distortion and ride passively on the
hanging wall.
Examples ofall the stages ofdevelopment of Basin
and Range structures are found in the western USA.
The model is supported by deep seismic reflection
profiling, which has identified a fault which may
extend to a depth of 20 km with a displacement of
30-60 km. This model, unlike the pure shear model
ofMcKenzie(1978),can account for the close juxta
position of brittle deformation and ductile deforma
tion and also for the temporal separation of these
different types of deformation.
243
Limit of
significant
extension hinge' Culmination
^DiscrepantO Ju
zone' 0 I
Distal
Tectonic:
Thermal:
ZoneC Zone D ZoneE
Wernicke has indicated the possibility that the
whole lithosphere may have failed by simple shear
(Fig.10.39). Five different zones can be postulated
on the basis of their subsidence/uplift history.
Because mantle lithosphere is denser than the asthe-
nosphere, its thinning should result in uplift. Con
versely, thinning of the lower density crust should
result in subsidence. The subsidence resulting from
extensional strain is referred to as tectonic subsidence^
and that related to subsequent cooling ofthe mantle
lithosphere as thermal subsidence. In zone A the
lithosphere is unstrained, but a small amount of
uplift may occur from the isostatic edge effect result
ing from the unloading ofthe crust by the breakaway
fault, to which zone A is coupled.In zone B the crust
is extended but not the mantle lithosphere, and so
subsidence takes place. The distal boundary of zone
B is defined by the point at which the shear zone cuts
the mantle. In zone C both crust and mantle are
extended, producing a gradual change from subsid
ence to uplift. In zone D only the mantle lithosphere
is thinned, producing uplift and broad doming of
the surface. Erosion may thin the crust so that the
subsequent thermal subsidence produces an overall
subsidence of the zone. Zone E is unstrained, but
may be uplifted by its coupling with zone D and may
eventually undergo overall subsidence because of
erosion. Magmatism may play an important role in
244 CHAPTER 10
the evolution of such a region, as the extension in
proximal areas may induce melting ofthe middle and
lower crust and the rest of the lithosphere.
Wernicke has applied this model to Arizona and
Utah and also to the Red Sea, where the uplift of the
Arabian plateau to the northeast may be the result of
an extensional fault system dipping in that direction.
Two rather different models of basin formation
have been described which entail both an initial
tectonic subsidence and a later thermal subsidence.
The model of McKenzie (1978) thins the lower
lithosphere homogeneously by ductile extension be
neath an upper crust thinned by faulting, and super
imposes a thermal subsidence on an early fault-
bounded basin. The model of Wernicke (1985) off
sets the regions of lower lithosphere stretching and
thermal subsidence. Coward (1986) considers that
these models represent end members in a range of
possible mechanisms. Coward suggests that there
will often be an overlap between the zones of upper
and lower lithosphere stretching, which is likely to be
Mountain
(a) Forebulge Foreland
range
basin
'7/////////(/77-^
OrCist' ^;
Mantle
Lithosphere
Asthenosphere
Forebulge
About
2-4 km
About 50-100 km
heterogeneous and patchy. In this case there would
be localized uplift and subsequent thermal subsid
ence within any particular faulted basin. This may
explain several problems associated with the end
member mechanisms such as the differing stretching
estimates for the same basin provided by different
techniques, the variations in dip of detachment
zones in large basins, the regional uplift associated
with parts ofsome basins which has led to the erosion
ofearly fault blocks and the presence ofvolcanism in
basins with only small upper crust extensions. Cow
ard proposes that concentrated stretching of the
lower lithosphere below an extensive zone of upper
crustal extension may develop instantaneously or via
an initial stage of uniform extension throughout the
lithosphere,followed by extension ofthe lithospheric
mantle and lower crust over a relatively narrow area
while the upper crust extends over a much wider
region. The model appears to explain satisfactorily
structures in the Basin and Range Province and the
northeastern European continental shelf.
Orogenic
Fig. 10.40 (a) The formation of a
wedge
foreland basin by lithospheric
flexure associated with the weight of
a mountain range (after Coward,
1994a).(b)Sketch of the
stratigraphic pattern of a foreland
basin, showing the onlap at
positions Xi to corresponding
to chronostratigraphic time lines Tj
to (redrawn from Allen & Alien,
1990).
 CONTINENTAL RIFTS, CONTINENTAL MARGINS AND SEDIMENTARY BASINS
. External forelands
Fig. 10.41 Foreland basins associated with Alpine
mountain ranges (redrawn from Allen etal., 1986).
10.5.3 Foreland basins
Continental collision results in a Himalayan-type
mountain range with a thickened crust(Section 9.3).
This additional mass represents a load which causes
flexure (Section 2.11.4) of the adjacent lithosphere
and creates a depression peripheral to the mountain
range, bounded on the side distant from the moun
tains by a forebulge (Fig. 10.40a). An example ofthis
is the development of the Indo-Gangetic Plain Basin
adjacent to the Himalaya. This flexural depression is
termed a foreland basin or foredeep, and it is the
depository for the sedimentary material (molasse)
that pours off the newly elevated mountain range as
it experiences erosion and as the active thrust belt of
245
Foreland basins
the collision transports material laterally. The fore
land basin extends much further into the surround
ing craton than the margin of the load and forms in
advance of the thrust front of the mountain range.
Foreland basins associated with the Alpine chain are
shown in Fig. 10.41.
The shape of the basin depends on the flexural
rigidity (Section 2.11.4) of the lithosphere beneath
and surrounding the mountain belt; a low value,
arising in young, hot, weak lithosphere, gives rise to a
narrow, deep basin, whereas a high value, arising in
old, cool, strong lithosphere, produces a wide basin
with a better developed forebulge (Coward, 1994a).
Varying flexural rigidity of the lithosphere can thus
cause the character of the foreland basin to change
along strike.
Unlike basins formed by lithospheric stretching
(Section 10.5.2), foreland basins do not experience
a temporal diminution in the rate of subsidence.
246 CHAPTER 10
(a)
SW
\
H Upper Cretaceous Jurassic and
Lower Cretaceous
HI Trlasslc
and subsidence persists over a long period of time
(Coward,1994a).Ifthe lithosphere rheology is visco-
elastic rather than elastic, the flexural rigidity would
decrease with time. This would cause the basin
to narrow and deepen and the forebulge to become
higher (Allen & Allen, 1990).
As a result of the formation mechanism, the
sediment thickness in a foreland basin decreases
away from the mountain front to a feather edge on
the forebulge. Close to the mountain range the
sediments are coarse grained and deposited in a
shallow water or continental environment; at the
feather edge they are fine grained and often tur-
biditic. The sediments thus form a wedge-shaped
unit whose stratigraphy reflects the subsidence his
tory of the basin as it grows and migrates outwards
as convergence continues (Allen & Allen, 1990).
Figure 10.40b shows the progressive onlap of sedi
ments onto the foreland. The stratigraphy is thus
characterized by units which thin laterally, overstep
older members or may be truncated by erosion.
As convergence continues in the mountain range,
small sedimentary basins may form on the top of
Pin line
V NE
^ \ D
CD
h5 "S
-10
Pin line
Fig. 10.42 (a) Present day
structure ofthe Broad Fourteens
Basin derived from a seismic
reflection profile,(b)The same
section restored to its form before
5 km
inversion in the Upper Cretaceous
(redrawn from Hayward & Graham,
V=H 1989).
moving thrust sheets in addition to the foreland
region. Such basins are referred to asplayback basins.
10.5.4 Inversion
Many regions have undergone reversal in the sense of
vertical movement at different stages in their evolu
tion. This reversal is known as inversion., and is
referred to as positive, the most common type, when
a region of subsidence becomes one of uplift and
negative when uplift is followed by subsidence. In
version can arise from several mechanisms(Coward,
1994b). Plate motions can cause inversion as a result
of the tectonism associated with continent-
continent collision, resulting in the uplift and de
formation ofa sedimentary basin. At major strike-slip
faults, inversion can occur as a result oftranspression
and transtension (Section 7.4.2) at bends in the
fault. Inversion can also occur as blocks rotate,
narrow and shorten. Isostatic mechanisms can also
result in inversion. Isostatic rebound (Section
2.11.5) can cause uplift as a surface load, such as an
icecap or mountain range, is removed by melting or
 CONTINENTAL RIFTS, CONTINENTAL MARGINS AND SEDIMENTARY BASINS
erosion, causing the structure to be overcompen-
sated and followed by progressive uplift in an at
tempt to regain isostatic equilibrium. On a smaller
scale, the diapirism of evaporites can cause uplift as
the evaporites mobilize and ascend because of their
low density. Lithospheric heating can cause uplift by
doming above a hotspot, which is followed by sub
sidence as the effects of the hotspot are removed.
Extensional faulting can cause uplift as the litho-
sphere is thinned and rises isostatically (Section
10.5.2).
In orogenic belts,inversion is frequently caused by
the reversal of motion on major dip-slip fault sys
tems. Figure 10.42 shows sections through the
Broad Fourteens Basin (offshore northwest Holland)
before and after inversion. The bounding normal
faults of the graben present in Upper Cretaceous
times have been reactivated as thrusts in response to
subsequent compression, leading to uplift and fold
ing of the graben fill. It is clear that such inversions
are indicated by breaks in the stratigraphic record.
Similarly, thrust systems may be reactivated as nor
mal faults to produce negative inversion.
Further reading
Allen, P.A. & Allen, J.R. (1990) Basin Analysis, Principles
and Applications. Blackwell Scientific Publications, Ox
ford.
Bailey, D.K.(1983)The chemical and thermal evolution of
rifts. Tectonophys. 94,585-97.
Barberi, P., Bonatti, E., Marinelli, G. & Varet, J. (1974)
Transverse tectonics during the split of a continent; data
from the Afar Rift. Tectonophys. 23,17-29.
247
Beaumont, C. & Tankard, A.J. (eds) (1987) Sedimentary
Basins and Basin-Forminpf Mechanisms. CSPG Mem.12/
AGS Spec. Publ. 5. Can. Soc. Petrol. Geol.
Bosworth, W. (1987) Off-axis volcanism in the Gregory
Rift, east Africa: implications for models of continental
rifting. Geolopfy 15,397-400.
Bott, M.H.P.(1982) The mechanism of continental split
ting. Tectonophys. 81,301-9.
Burke, K. (1977) Aulacogens and continental breakup.
Ann. Bsv. Earth planet. Sci. 5, 371-96.
Burke, K. & Dewey, J.F. (1973) Plume generated triple
junctions: key indicators in applying plate tectonics to
old rocks.J.pieol. 81,406-33.
Courtillot, V.& Vink, G.E.(1983) How continents break
up. Sci. Am. 249(1),40-7.
Coward, M.P. (1986) Heterogeneous stretching, simple
shear and basin dc\c\opmcnt.Earth planet. Sci. Lett. 80,
325-36.
McKenzie,D.P.(1978)Some remarks on the development
ofsedimentary basins. Earth planet. Sci. Lett. 40,25-32.
Oxburgh, E.R. & Turcotte, D.L. (1974) Membrane tec
tonics and the East African Rift. Earth planet. Sci. Lett.
22,133-40.
Watts, A.B. 8c Ryan, W.B.F.(1976) Flexure of the litho-
sphere and continental margin basins. Tectonophys. 36,
25-44.
Wernicke, B. (1985) Uniform sense simple shear of the
continental lithosphere..C/»w./. Earth Sci. 22,108-25.
White, R.S. 8c McKenzie, D.P.(1989) Magmatism at rift
zones; the generation of volcanic continental margins
and flood bzszdts.J.pfeophys. Res. 94,7685-730.
White, R.S., Spence, G.D., Fowler, S.R., McKenzie, D.P.,
Westbrook, G.K. 8c Bowen, A.N.(1987) Magmatism at
rifted continental margins. Nature 330,439-44.
Whitmarsh, R.B., Avedick, F. 8c Saunders, M.R. (1986)
The seismic structure of thinned continental crust in the
northern Bay of Biscay. Geophys. J. Roy. astr. Soc. 86,
589-602.
11/ The mechanism of plate tectonics
11.1 Introduction
The mechanism behind the motion of plates is still
controversial. Older theories for the origin of the
major structural features ofthe Earth's surface which
relied on the supposed contraction or expansion of
the Earth have now been discounted. The most likely
mechanism of heat transfer from depth appears to be
thermal convection. The form ofthis convection and
the manner in which the thermal energy is utilized in
driving the plates are discussed in this chapter.
11.2 Contracting Earth hypothesis
In the 19th century it was believed that, since its
formation, the Earth had been cooling due to heat
loss by thermal conduction. Computations by Lord
Kelvin on the rate of cooling of large iron spheres
provided the first, erroneous, estimates of the age of
the Earth of 20-80 Ma. As a corollary, it was sug
gested that the accompanying contraction of the
Earth on cooling might provide a mechanism for
mountain building. It was estimated that the circum
ference of the Earth had decreased by 200-600 km
since the Earth's accretion. The discovery of radio
activity at the end ofthe 19th century negated much
of the early work as it provided a precise method of
Region of
tangential compression
Level of
zero strain
Region of
tangential tension
Fig. 11.1 Contracting Earth model.
248
dating rocks and also demonstrated that the Earth
possesses its own internal sources of heat.
The contraction hypothesis envisaged that the cen
tral region of the Earth underwent most rapid cool
ing and contraction and was placed in a state of
tangential tension. Above a horizon of no strain, the
outer shell ofthe Earth was then subjected to tangen
tial compression as it collapsed inwards upon the
shrinking centre (Fig. 11.1). The lithosphere is too
thick to respond to this compression by buckling,
but would yield by thrust faulting, producing moun
tain ranges by the stacking of thrust slices.
A contracting Earth is no longer recognized as a
possible mechanism for tectonic activity for two
convincing reasons:
1 The Earth is not cooling sufficiently rapidly to be
consistent with contraction, and modern evaluations
of cooling rates imply a total contraction of only a
few tens of kilometres. Consequently, the contrac
tion hypothesis cannot account for the many thou
sands ofkilometres ofcrustal shortening which must
have occurred in mountain belts throughout geolog
ical time.
2 The hypothesis implies that the lithosphere is
everywhere in compression, and cannot provide an
explanation for the multitude of phenomena which
must have originated in tensional regimes, such as
normal faults, ocean ridges and rift valleys.
11.3 Expanding Earth hypothesis
The expanding Earth hypothesis was first proposed
in the 1920s and was subsequently adopted by sev
eral geologists as the mechanism behind the break-up
of continents, the formation of continental rifts and
the presence of extensional features such as normal
faults (Carey, 1976; 1988). Their proposal was that
the continental lithosphere was originally continu
ous over the surface ofan Earth ofsmaller radius and
that, as the Earth expanded and its surface area
increased, the continental lithosphere fragmented

Complete shell of Continental Oceanic
continental lithosphere lithosphere
lithosphere
Radius 3500 km Radius 6370 km
Fig. 11.2 Expanding Earth model.
and dispersed, while mantle material welled up into
the consequent gaps to form the oceans (Fig, 11,2).
Independent evidence for the expanding Earth
hypothesis appeared to be provided by certain theor
etical physicists, who suggested that the universal
gravitational constant was decreasing with time as
the universe expanded and its constituent matter
became more widely dispersed. Gravitational forces
are responsible for binding the Earth into a spherical
form, and since the gravitational constant directly
controls the magnitude of the force of attraction
between masses, its decrease would imply a progres
sive relaxation of the binding forces and an increase
in the Earth's radius.
The most recent versions of the expanding Earth
hypothesis correlate the period of rapid expansion
with the break-up and fragmentation of Pangaea
since 200-300 Ma. These argue that continental
reconstructions can be arranged more accurately on a
globe ofsmaller radius, and propose that during this
period the surface area of the Earth increased by a
factor of three while the radius increased from an
initial value of70% ofits present value. This calcula
tion was based on the surface area of the present
continents, and provides a mean radial expansion
rate of about 6 mm a"^.
There are two methods available which can be used
to test the expanding Earth hypothesis directly.
11.3.1 Calculation of the ancient
moment of inertia of the Earth
The moment of inertia of a rigid body about a given
axis is defined as where m is the mass of each
THE MECHANISM OF PLATE TECTONICS 249
small element of the body and r the distance of the
element from the axis of rotation. The moment of
inertia of a uniform sphere is given by IMR^S,
where M is the mass of the sphere and K its radius.
Newton's laws of mechanics for linear motion state
that the momentum (mass x velocity) of a system is
conserved unless an external force acts upon it.
These laws apply equally to angular (rotational)
motion, in that angular momentum (moment of
inertia x angular velocity) is conserved unless the
system is acted upon by an external torque.
Since the Earth's mass remains constant,any deter
mination of the ancient moment of inertia of the
Earth would allow a calculation of its ancient radius
and so demonstrate if any expansion had occurred.
The theory behind any such determination is com
plicated by the fact that momentum is conserved
within a system comprising the Earth and the Moon.
Throughout geological time the angular momentum
inherited from the fragments that accreted to form
the Earth and Moon has been progressively parti
tioned between the two bodies, by a mechanism
known as tidal interaction, in such a way as to reduce
the rotational energy of the system. At present the
stage has been reached in which the Moon spins very
slowly and must consequently lie at a greater distance
from the Earth than in the past,so that momentum is
conserved in its orbital motion. The tidal interaction
of the Moon on the Earth is similarly causing the
latter's angular rotation to decelerate. Since the
number of rotations in a complete orbit of the Sun
determines the number of days in a year, the year in
the past would have consisted of more days than at
present. This also implies that the length of the day
has progressively increased.
Transfer of angular momentum from Earth to
Moon thus causes an increase in the length of the
day. A further contributor to this phenomenon
would be an increase in the Earth's moment of
inertia, which would allow angular momentum to be
conserved by a slower rate of rotation. Knowledge of
the length of the lunar month would allow an esti
mate to be made of the lunar contribution to the
Earth's rotational deceleration and allow any change
in its moment ofinertia to be isolated.
Information on the rotational history of the
Earth-Moon system has been provided from a
detailed examination of fossil organisms whose
patterns of growth are strongly affected by diurnal
effects. In particular, certain rugose corals of Middle
250 CHAPTER 11

Devonian age (375 Ma) have been shown to exhibit

epithecal banding which can be attributed to daily,
monthly and yearly cycles of growth (Scrutton,
1967). Such studies have indicated that the Middle
Devonian year comprised 400±7 days, and was
divided into 13 months of 30.5 days. The average
increase in the length of day up to present times is
24sMa-\
The length of the lunar month in Devonian times
allows an estimate to be made ofthe Moon's angular
momentum at that time, and hence the deceleration
of the Earth's rotation resulting from tidal friction.
The deceleration not accounted for in this way can be
used to provide an estimate ofthe Devonian moment
of inertia of the Earth, which is found to be 99.4-
99.9% ofits present value. Given the uncertainties in
the calculation,the moment ofinertia does not seem
to have altered significantly. The expansion of the
Earth required to cause continental drift implies that Fig. 11.3 Parameters used in estimating the palaeoradius
the Devonian moment of inertia would have ofthe Earth from palaeomagnetic data.
to have been only 94% of its present value. Con
sequently, such rapid expansion can be ruled out.
the expanding Earth hypothesis can definitely be
ruled out.
11.3.2 Calculation of the
The expanding Earth hypothesis clearly does not
ancient radius of the Earth
stand up to direct testing. Also, indirectly the
A rather less involved method of testing the expand hypothesis cannot account for presently observable
ing Earth hypothesis entails determining the palaeo- phenomena. If continental drift results from this
radius ofthe Earth using palaeomagnetic techniques mechanism there would be no necessity for subduc-
(Ward, 1963). tion zones for the consumption of oceanic litho-
The method involves selecting sampling sites of sphere, and no explanation is provided for extensive
the same age, and differing as much as possible in zones affected by collisional tectonics. The majority
palaeolatitude, on a landmass which has been stable of plates are presently spreading in an east-west
since the time the sites acquired their primary re- sense. If such a pattern results from an expanding
manent magnetizations(Fig. 11.3). Determining the Earth it would imply a progressive increase in the size
palaeolatitudes (<t)i, (|)2) ofthe sites then provides the of the equatorial bulge, which is not occurring. An
angle originally subtended at the centre of the Earth expansion of the Earth would imply the existence
(<t>i + <t>2)' The known separation of the sites {d) can of extensive zones subjected to membrane stresses
then be used to calculate the palaeoradius of the (Section 10.3.1) as plates attempt to adjust to the
Earth {R^ according to the relationship R^ = d/ increasing radius ofcurvature ofthe Earth, and these
((|)i + (|)2), where angles are expressed in radians. The do not exist. Finally, the theory does not provide a
present wide availability of palaeomagnetic data now mechanism for the continental drift which is known
allows the results of such determinations to be sub to have occurred in pre-Mesozoic times (Sections
jected to statistical analysis, and an assessment by 12.1 and 12.2).
McElhinny et al. (1978) has provided a 95% con
fidence level estimate of the radius of the Earth
400 Ma ago of 102±2.8% of the present radius. A 11.4 Implications of heat flow
very slight expansion, or, indeed contraction, of the The average vertical thermal gradient at the Earth's
Earth could be tolerated by this analysis, but, cer surface is about 25 ®C km"^.Ifthis gradient remained
tainly, the very large increase in radius required by constant with depth, the temperature at a depth of

100km would be ISOO^C. This temperature is in
excess of the melting temperature of mantle rocks at
this depth, and so a fluid layer is implied. Such a
molten layer does not exist as S waves are known to
propagate through this region (Section 2.1.3). Two
possibilities exist in explanation of this pheno
menon: first, that heat sources are concentrated
above a depth of 100 km; and second, that a more
efficient mechanism than conduction operates below
this depth whereby heat is transferred at a much
lower thermal gradient. Distinction between these
processes can be made by considering the variation in
heat flow over the Earth's surface in conjunction
with the variation in content of radioactive minerals
of different crustal types.
Heat flow generally decreases with the age of the
crust (Sclater et 1980). Within the oceans heat
flow decreases from the ocean ridges to the deeper
water flanking basins and it has been shown (Section
6.9) that this cooling correlates with a progressive
thickening ofthe oceanic lithosphere and an increase
in water depth. Similarly, the heat flow of marginal
seas (Section 8.9) decreases with age, with the pres
ently active basins exhibiting the greatest heat flow.
Within continental regions the heat flow generally
decreases with increasing time since the last tectonic
event. Consequently, Precambrian shields are char
acterized by the lowest heat flow and young fold
mountain belts by the highest.
The representation of the global pattern of heat
flow is difficult because the density of the obser
vations is highly variable so that the location of
contours can be greatly biased by only a small num
ber of measurements. Chapman & Pollack (1975)
Fig. 11.4 Pattern of global heat
flow represented by spherical
harmonic analysis. Contour interval
40 mW m"^ (redrawn from Bott,
1982a, with permission from
Edward Arnold).
THE MECHANISM OF PLATE TECTONICS 251
have overcome the problem of limited observations
in some areas by predicting the heat flow in those
areas on the basis ofthe correlation of heat flow with
the age of the oceanic lithosphere and the age of the
last tectonic event to affect continental crust. In
Fig. 11.4 their results are presented by a spherical
harmonic analysis of the heat flow measured or
predicted in 5"x5® grid areas of the globe. This
procedure imparts a certain smoothing of the true
pattern, so that variations with wavelengths of less
than about 3300 km are not represented. Figure 11.4
illustrates the high heat flow associated with the
ocean ridge system and the youngest marginal basins
of the western Pacific. Low heat flow values are
associated with old oceanic crust and with Precam
brian shields.
Histograms of heat flow measurements from
oceans and continents (Pollack & Chapman, 1977a;
VoWzcket 1993) are presented in Fig. 11.5. The
significantly greater number of heat flow values mea
sured in the oceans reflects the relative ease oftaking
measurements at sea (Section 2.13). The greater
dispersion of the oceanic values, and their positive
skewness, reflects variability arising from localized
extreme values at the crests of ocean ridges. By
contrast, there are fewer extreme high or low values
present in the continental values. The mean of oce
anic heat flow measurements is 65 mW m~^. How
ever,this only represents the heat loss by conduction,
and ignores the heat reaching the surface by the
discharge of hot fluids such as water and lava. It is
now recognized that the hydrothermal contribution
accounts for about a quarter of the global heat loss,
and the average oceanic heat flow is increased to
252 CHAPTER 11
c 1200
.2
> 1000
Continents
Heat flow(mW m )
101 mW m"^ (Davies, 1980; Sclater et al.^ 1980).
The mean continental heat flow is 57 mW m~^, in
cluding the small contribution from lavas. The mean
continental heat flow is thus only about 60%
of the oceanic mean. The global average heat flow is
87 mW m-2.
The majority of the heat escaping at the Earth's
surface originates from the decay oflong-lived radio
active isotopes of uranium, thorium and potassium
(Section 2.13) which have half-lives of the same
order as the age of the Earth. These isotopes are
relatively enriched in the upper continental crust,
and it has been estimated that their decay contributes
18-38 mW m~^ to the observed heat flow (Pollack 8c
Chapman, 1977b). Consequently up to about 70%
ofthe heat flow in continental regions may be gener
ated within the upper 10-20 km of the crust. The
oceanic crust, however, is virtually barren of radio
active isotopes, and only about 4 mW m"^ can be
attributed to this source. Over 96% of the oceanic
heat flow must originate from beneath the crust, and
so different processes of heat supply must act beneath
continents and oceans (Sclater 8c Erancheteau,
1970).
Thus, a large proportion of the continental heat
flow is from sources concentrated at a shallow depth,
and only a small sub-crustal component is required.
Conversely, the majority of oceanic heat flow must
Fig. 11.5 Comparison of the heat
flow from continents and oceans
(redrawn from Pollackefa/., 1993).
originate at sub-crustal levels. Because of the melting
problems discussed above, this heat must be trans
ported under the influence ofa low thermal gradient.
The mechanism of heat transfer by convection is the
only feasible process conforming to these con
straints. Therefore, although heat transfer by con
duction takes place within the rigid lithosphere, heat
transfer by convection must predominate in the
sublithospheric mantle. Indeed, conduction cannot
occur to any great depth as the rate of heat transfer by
this mechanism is much slower than required. The
feasibility and form ofsuch convection is discussed in
the following sections.
11.5 Convection in the mantle
11.5.1 The convection process
The nature of convective flow in the mantle is prob
lematic. Analytical solution is difficult because of the
complex theological structure, including the pres
ence of a transition zone (Section 2.8.5), because of
the presence of heat sources within the convecting
layer as well as beneath it, because of the unknown
influence of an overlying rigid lithosphere on the
pattern of convection, and because the convecting
layer has the form of a spherical shell. However, by
making use ofsimple hydrodynamic theory and con-

straints on the pattern of convection supplied from
past and present plate motions,it is possible to derive
considerable information on the convective process.
Reviews of ideas concerning mantle convection are
given in Phillips & Ivins(1979) and Peltier (1985).
Convection takes place in a fluid when, as a result
of heating from below, a density inhomogeneity is
created, causing the heated material to attain a nega
tive density contrast with respect to its surroundings
(Fig. 11.6). In the mantle this inhomogeneity arises
because ofthermal expansion, but in other contexts a
chemical change can also create a negative density
contrast. The heated material rises in the fluid and,
on reaching the top, travels laterally, losing heat all
the time. Eventually its heat loss and consequent
contraction will be such that the material assumes a
positive density contrast with its surroundings and
sinks. The cycle continues with the material travel
ling along the base of the fluid layer absorbing heat
and subsequently reascending. If the heat lost at the
surface is the same as that supplied from below, a
steady pattern of convection is established. Indeed,
should the temperature regime of the system be
disturbed, the rate ofconvection, which depends on
the viscosity of the fluid which in turn is controlled
by temperature, alters in such a way that stable
convection is again established (Tozer, 1972). The
rate at which heat is supplied to the mantle, however,
must have decreased throughout geological time as
the quantity of long-lived radioactive isotopes de
creased exponentially by radioactive decay. This has
probably caused a slight cooling ofthe Earth over its
lifetime of 230-S00®C.
11.5.2 Feasibility of mantle convection
In order to gain insight into the feasibility and nature
of convection within a spherical, rotating Earth, it is
convenient to assume that the mantle approximates a
Newtonian fluid. Although this assumption may be
erroneous, it does allow simple calculations to be
made on the convective process.
The condition for the commencement of thermal
convection is controlled by the magnitude of the
dimensionless Rayleigh number (R), which is de
fined by:
R = a^^fd'^/kv
where a is the coefiicient ofthermal expansion, P the
superadiabatic temperature gradient (the gradient in
THE MECHANISM OF PLATE TECTONICS 253
Temperature
nsowuuwuuuuuuwwuwuuu^
OSS by conduct
eat supply by conduction
Fig. 11.6 Convection model and its associated thermal
gradient.
excess of that expected to be associated with the
increasing pressure),^ the acceleration due to grav
ity, d the thickness of the convecting fluid, k the
thermal diffusivity (the ratio of the thermal conduc
tivity to the product of density and specific heat at
constant volume) and v the kinematic viscosity (the
ratio of dynamic viscosity to density). For a two-
dimensional fluid with a free surface, convection will
commence when R exceeds 658 and the convection
cells will have a horizontal dimension of about
2yj24. For a mantle 3000 km thick this corresponds
to convection cells only 155 km wide. However,for
convection in the mantle, which is a spherical shell
overlain by a rigid lithosphere, the Rayleigh number
corresponding to the onset of convection increases
to 2380.
Table 11.1, from Bott (1982a), shows calculated
values of R for the upper, lower and whole mantle.
Although the various parameters used to compute R
are not known exactly, they have been estimated
sufficiently well for R to be better than 50% accurate.
The value of the kinematic viscosity has only been
determined relatively recently from detailed studies
of isostatic rebound following deglaciation (Section
2.11.5). Incorrect earlier estimates implied that con
vection would be restricted to the upper mantle, but
it appears now that the viscosity ofthe mantle is fairly
uniform (Cathles, 1975; Davies, 1977). Table 11.1
indicates that convection in the upper mantle is not
far from marginal, and would only become vigorous
in isolated regions where the superadiabatic temper
ature gradient is anomalously high and the viscosity
anomalously low. The Rayleigh numbers for convec
tion in the lower mantle or throughout the whole
mantle are well in excess of the critical value and so
vigorous convection would be expected to occur
within these regions.
254 CHAPTER 11
Table 11.1 Computed Rayleigh numbers for the mantle and its subdivisions assuming a superadiabatic temperature
gradient p = 0.1 Km"^ and^ = 10 m s"^(from Bott, 1982a, with permission from Edward Arnold).
Thermal expansion Layer thickness
(a(K-')) (<<(m))
Upper mantle 2.5x10-® 3.0x10®
Lower mantle 1.5x10"® 2.0 X 10®
Whole mantle 1.9x10"® 2.7x10®
The nature of the flow ofthe convecting fluid can
be judged by the magnitude ofthe Reynolds number
(i^), which allows discrimination between laminar
and turbulent flow. Re is defined.
Re=vd/v
where v is the velocity of flow. Taking v = 200 mm
a"^ = 6 X 10~^ms"^, //= 3000 km = 3 x lO^m and
V = 2 X 10^^ m^ s"^, jR^ = 9 X 10"^°. This very low
value indicates that the flow is laminar, and that the
amount of heat transferred by convection never in
creases to the level at which turbulent flow occurs
and the regular pattern of convection cells disap
pears.
The effect of the Earth's rotation on convection
can be judged by the magnitude of the Taylor num
ber(T), which is defined:
r= {2wd^/v)^
where w is the angular velocity of rotation. Putting
= 7.27 X 10"® rads"^ and other values as above,
T«4xl0"^^.A value ofTless than unity implies no
significant effect ofrotation on convection and so the
Earth's rotation should have no effect on the pattern
of mantle convection. However, at the present time
the majority of plates are moving approximately
east-west (Solomon et 1975), and during the
Mesozoic the predominant movements were north-
south. It is possible, therefore, that there is some
poorly understood mechanism responsible for this
phenomenon (Doglioni, 1993).
The efficiency of convection is measured by the
Nusselt number (N«), which is the ratio of the total
heat transferred to that transferred by thermal con
duction. Elder(1965) has computed experimentally
the relationship between Nu and R, and has found
that at values of R appropriate to marginal convec
tion Nu is unity and very little heat can be transferred
by convection. At R values appropriate to vigorous
Thermal diflusivity Kinematic viscosity Rayleigh number
(4(m^s->)) (vKs-i)) m
1.1x10"® 3x10^® 6x10®
1.3x10"® 2x10'^ 9x10®
1.3 X 10"® 2x10'^ 4x10®
convection of 10®-10^, Nu is about 10, indicating
the predominance of heat transfer by convection.
Exact solution ofthe form of mantle convection is
not possible, although approximate solutions (Par-
mentier et al., 1976)reveal that the pattern offlow is
as shown in Fig. 11.7a. Heat is transferred to the
convection cell by conduction at the base of the cell
and is removed by conduction at its top. Within the
fluid most of the heat is transported around the
margins of the cell, with an isothermal core rotating
in sympathy with these movements but not contrib
uting to the heat flow. Parmentier et al, (1976) have
also demonstrated that the form ofconvection is not
radically diflferent if the mantle acts as a non-
Newtonian fluid (Fig. 11.7b), the only major effect
being to increase the critical Rayleigh number. Con
sequently, for non-marginal convection, the nature
ofthe creep mechanism in the mantle is not of great
significance in determining the flow pattern of con
vection.
The mantle transition zone (Section 2.8.5) must
influence the vertical extent of convection in the
mantle. If this zone represents a change in chemical
composition, convection currents could not cross it.
In this case convective circulation would be in the
Fig.11.7 Streamlines (dashed) and isotherms (solid
lines) for(a) Newtonian and (b) non-Newtonian
convection (redrawn from Parmentier 1976, with
permission from the American Geophysical Union).

Lithosphere
(a) Core-mantle boundary
Lithosphere
700 km
(b) Core-mantle boundary
Fig. 11.8 Possible influence of the mantle transition zone
on the pattern ofconvection,(a) The transition zone
represents a phase boundary which can be traversed by
convection currents,(b) The transition zone represents a
chemical boundary prohibiting its crossing by convection
currents (redrawn from Bott, 1982a, with permission
from Edward Arnold).
form of paired cells (Fig. 11.8b), with heat trans
ported across the separating transition zone by con
duction. The mechanism seems unlikely, as
convection in the upper mantle is not far from
marginal so that it could not cope with the amount
of heat transported from below by much more vigor
ous convection. This idea is discussed more fully in
the following section.
11.5.3 The vertical extent of convection
The vertical extent of mantle convection is still
controversial, and there is no general agreement
whether convection is mantle wide (Fig. 11.8a) or
THE MECHANISM OF PLATE TECTONICS 255
takes place in paired cells above and below the mantle
transition zone (Fig. 11.8b).
The nature of the mantle transition zone is equi
vocal, but the majority view appears to be that it
represents a region in which solid state phase changes
take place whereby the mineralogy ofmantle material
changes to higher pressure forms with depth rather
than representing a change in chemical composition
(Section 2.8.5). For example. Watt & Shankland
(1975) have shown, from an inversion of velocity-
density data, that the mean atomic weight of the
mantle shows no change across the transition zone.
Ifthis is the case, convection currents could cross the
transition zone as long as the phase changes take
place very rapidly, and convection cells would then
be mantle wide (Fig. 11.8a). The phase changes
would have two important effects on convection.
The phase changes,such as from olivine to spinel, are
temperature dependent. The change from low pres
sure to high pressure forms takes place at shallower
than average depths in the cold descending currents
and at greater than average depths in the hot ascend
ing currents. Consequently,low density minerals are
created deeper on ascent than high pressure forms on
descent, and their negative and positive buoyancies
then help to drive the convection cells. Also, the
phase changes are associated with a release or absorp
tion oflatent heat, the high to low pressure reaction
being endothermic and the low to high pressure
reaction being exothermic. This causes steepening of
the thermal gradient across the transition zone, so
that the temperature in the lower mantle would need
to be 100-150"C higher than if the zone did not
exist.
Ringwood & Irifune (1988) have examined the
physical and chemical processes occurring in the re
gion of the mantle transition zone, in particular the
density relationships of the mantle and components
of the oceanic lithosphere at the pressures corre
sponding to this region. It appears that when young,
relatively thin oceanic lithosphere is subducted, the
various phase changes associated with increasing
pressure and temperature cause both the oceanic
crust and lithospheric mantle to be trapped gravita-
tionally. Growth of this boundary layer with time
would lead to its thickening until it occupied a shell
between about 600 and 700 km depth. If no other
phenomena operated,this layer would cause a barrier
to the descent of young lithospheric slabs and isolate
256 CHAPTER 11

convection systems in the upper and lower mantle.
However, the interiors of thick, mature, oceanic
plates have a high thermal inertia and viscosity so that
when the tip of the slab meets the boundary layer at
a depth of about 650 km it buckles and thickens
plastically to form what Ringwood & Irifune term a
megalith, which grows with time. The megalith is
neutrally buoyant and in isostatic equilibrium at this
level, so that it 'floats' straddling the boundary layer
(Fig. 11.9). If the megalith lies in the region of a
descending convection current in the lower mantle,
it would become entrained and gradually mix into
the lower mantle; this must be accompanied by the
movement of an equal volume of lower mantle into
the upper mantle. The model envisages a dynamic
equilibrium between the subduction of young litho-
sphere,that creates and maintains a boundary layer at
about 650 km depth, and the subduction of mature
lithosphere which gives rise to megaliths which pen
etrate the layer and allow the exchange of material
between the upper and lower mantle. Thus,although
whole mantle convection is suggested, the process
need not be continuous and depends upon the age of
the oceanic lithosphere being subducted. This model
has many important ramifications for mantle pro
cesses, including the petrogenesis of ocean ridge
basalts and intraplate volcanic associations and the
nature ofthe convective regime in the Archaean; it is
clearly testable by geophysical methods, and the
outcome of such investigations will be awaited with
interest.
Tackley et al. (1993) have numerically modelled
mantle convection in three dimensions with an en-
dothermic phase change at the base of the transition
zone. They suggest that, periodically, cold down-
welling material accumulates above 670 km and then
rapidly flushes into the lower mantle. This sugges
tion is supported by seismic tomographic imaging

Island arc
Calc-alkaline volcanism Basaltic crust

Uprise of contaminated diapirs and ——

Lithosphere
partial melting
Peridotite contaminated by jff\
partial melt of outer water-rich
- Depleted pyrolite
Brittle regime

jI /'^^^^^^eso''btion
W Resorbtion ofof
y/depleted pyrolite
'IM, 400 km discontinuity
'1^//1 V
'Buckling and plastic
thickening of crust

vCpol highly viscous megalith;^^
550 km discontinuity
"^Ancient oceanic
- Former harzburgite
- Former basaltic crust
lithosphere

Fig. 11.9 Formation of a layer of

former harzburgite and basalt
('ancient oceanic lithosphere )
between 600 and 700 km and the
production of a megalith as a
\Descending convection / result of subduction (redrawn
^ \ current in pyrolite / from Ringwood & Irifune, 1988,
\ \ lower mantle / with petmission from Nature
\ \ \ I I 331, 131-6. Copyright© 1988
I t ) 1 t Macmillan Magazines Ltd).

(Section 2.1.8) of the Tonga subduction zone (van
der Hilst, 1995)in which the descending iithosphere
apparently penetrates the transition zone (see Fig.
8.18).
Thus, the transition zone may not be a barrier to
mantle-wide convection, and a number of workers
have presented evidence in accord with this premise.
Kanasewich (1976) has noted an organized distribu
tion of plates, in which the Pacific and African plates
are approximately circular with the smaller plates
having an approximately elliptical form and arranged
systematically between these two large plates. Kana
sewich attributes this organization to convection that
is mantle-wide. Davies (1977) has conducted model
experiments and has concluded that only extreme
viscosity contrasts would restrict convection to the
upper mantle, and maintains that such contrasts do
not exist. Elsasser et al. (1979) have employed a
scaling analysis in which the depth of convection is
derived as a function ofknown parameters, and have
concluded that this depth is consistent with convec
tion throughout the entire mantle. Grand(1994)has
demonstrated, by seismic tomographic analysis
(Section 2.1.8), that oceanic Iithosphere descends
beneath western America into the lower mantle, so
that at least one convection cell is mantle-wide.
Morgan &Shearer(1993)have derived the buoyancy
distribution in the mantle from seismic tomographic
maps and concluded that there must be significant
flow between the lower and upper mantle. Other
workers, however, such as O'Nions (1987), suggest
that the geochemical and isotopic pattern of trace
elements found in oceanic volcanic rocks supports a
model in which portions of the mantle have been
chemically isolated for much of Earth history. This
would suggest that the mixing implied by whole
mantle convection has not occurred,and that layered
convection is more likely. Loper (1985), however,
has been able to produce a model of whole mantle
convection in which the preservation of chemically
distinct reservoirs is possible. The topography on the
base of the mantle transition zone has an amplitude
of about 30 km (Shearer & Masters, 1992), which is
an order of magnitude lower than predicted for a
chemical, rather than a phase, change at this depth.
It would seem, therefore, that convective circula
tion is most likely to be mantle-wide and not con
strained by the transition zone. The presence of
earthquakes to depths of680 km certainly indicates
that the Iithosphere circulates almost to the base of
THE MECHANISM OF PLATE TECTONICS 257
the transition zone, and is therefore further evidence
in favour ofdeep convection. The existence ofearth
quakes with compressive focal mechanism solutions
at great depth in BeniofF zones (Section 8.3) was
once cited as evidence that a barrier to slab penetra
tion existed at a depth of about 700 km which
limited convection to the upper mantle. This argu
ment no longer holds, as Isacks & Barazangi(1977)
have explained the compressive nature of these
events in terms of an unbending, rather than a
stopping, of the slab. The cessation of earthquake
activity probably results from the loss of identity of
the downgoing slab as it heats up to the same
temperature as the surrounding mantle.
It is possible that the controversy between whole
mantle and layered convection is artificial and that
both modes may be possible. Stein & Hofmann
(1994) have described a model in which the different
types of convection alternate during the Earth's his
tory. During the present phase of plate motions, the
majority of new continental crust is formed by volca
nic processes at subduction zones. However, in the
past, geochemical studies suggest that some large
igneous terrains appear to have formed in environ
ments different from subduction zones, and at a
much higher rate than at present. It has been pro
posed that these ancient terrains formed when the
bulbous head of a mantle plume approached the
surface, partially melted and gave rise to copious
basaltic volcanism. The Cretaceous, when there were
many events of flood basalt volcanism, may have
been one of these episodes. The ascent of material
from the lower mantle in the plume would have
replenished the incompatible elements in the upper
mantle removed during continental crust formation.
Replenishment by the subduction of continental
crust would be inconsistent with the known isotopic
composition of the upper mantle. The new model
suggests that periods of layered convection, during
which incompatible elements were depleted in the
upper mantle by crust formation, were followed by
whole mantle convection, when the incompatible
elements were replenished. This would explain both
the episodic nature ofcontinental crust formation in
the geological record and the isotopic composition
of the upper mantle. The model is still speculative
and a true understanding of mantle convection will
only come when the nature of the transition zone is
fully characterized.
The motion ofthe plates must be compensated by
258 CHAPTER 11
return flow at depth in the mantle,so that the mantle
material displaced by downgoing slabs is balanced by
material flowing towards spreading centres. Several
workers have discussed the nature ofthis return flow
(e.g. Garfimkel, 1975; Harper, 1978; Chase, 1979;
Alvarez, 1982), and consider that generation and
consumption of plates is not balanced around single
plates, but globally. For example, Garfimkel (1975)
and Alvarez (1982) suggest that material at present
displaced by the circum-Pacific subduction zones
streams towards other oceans which are increasing in
size. The form of the return flow must change con
siderably with time as the plate configuration changes,
and so its geometry must be complex. These studies
implythatflow in the mantle does not take the form of
regular circuits linked to the overlying plates.
11.6 The forces acting on plates
In order to understand the structural styles and
tectonic development of plate margins and interiors,
it is necessary to consider the nature and magnitude
ofall the forces which act on plates. Forsyth & Uyeda
(1975) have solved the inverse problem ofdetermin
ing the relative magnitude of plate forces from the
observed motions and geometries of plates. Since the
present velocities of plates appear to be constant,
each plate must be in dynamic equilibrium, with the
driving forces being balanced by inhibiting forces.
Forsyth & Uyeda (1975) used the corollary of this,
that the sum of the torques on each plate must be
zero, to determine the relative size of the forces on
the 12 plates which they assumed make up the
Continent
Frp- Ridge push I?r -Ridge resistance
Fnb- Negative buoyancy He - Bending resistance
Fsp -Slab pull Rs -Slab resistance
-Trench suction Ro -Overriding plate resistance
Hdo- Mantle drag under ocean
Hoc- Mantle drag under continent
Earth's surface. The asthenosphere's role in this
scenario was considered to be essentially passive.
A similar set of computations based on a similar
method and providing similar results was made by
Chappie & Tullis (1977). The following description
of forces is based on the extensions of the work of
Forsyth & Uyeda (1975) made by Bott (1982a).
At ocean ridges the ridge push force JFpj,
(Fig. 11.10) acts on the edges of the separating
plates. This derives from the buoyancy of the hot
inflowing material causing the elevation of the ridge
and hence an additional hydrostatic head at shallow
depths which acts on the thinner lithosphere at the
ridge crest. It may also arise from the cooling and
thickening of the oceanic lithosphere away from the
ridge (Section 6.9), which exerts a pull on the ridge
region. The ridge-push force may be two or three
times greater if a mantle plume (Section 5.4) under
lies the ridge (Bott, 1993), because of the increased
pressure in the asthenosphere at the ridge crest. The
separation of plates at ocean ridges is opposed by a
minor ridge resistance which originates in the
brittle upper crust, and whose existence is demon
strated by earthquake activity at ridge crests. The
resisting forces are small so that the net effect is the
presence of a driving force.
Beneath plate interiors a mantle drag force acts
on the base of both the oceanic and continental
lithospheres ifthe velocity ofthe underlying astheno
sphere differs from that of the plate. If the asthenos
phere velocity exceeds that of the plate, mantle drag
enhances the plate motion (Pdo^ but if the
asthenosphere velocity is lower, as shown in
Ocean
Fig. 11.10 Some ofthe forces
acting on plates (developed
from Forsyth & Uyeda(1975)
by Bott, 1982a, with permission
from Edward Arnold).
 THE MECHANISM OF PLATE TECTONICS 259

Fig. 11.10, the mantle drag tends to resist plate 01

movement(Rjyo-> Rr>c)- Mantle drag beneath contin
ents is about eight times the drag beneath oceans;
this may be due to the increased viscosity ofthe sub
continental lithosphere arising from somewhat lower
geothermal gradients in the sub-crustal mantle.
(a)
At subduction zones the major force acting on
plates results from the negative buoyancy (JPnb) of
the cold, dense slab of descending lithosphere. Part
of this vertical force is transmitted to the plate as the
slab-pull force F^p. The density contrast, and hence
JPnb) is greatly enhanced at depths of 200-300 km
where the olivine-spinel transition occurs in the slab.
Fsp is opposed by a slab resistance (JR-s), which
mainly acts on the leading edge of the descending
plate where it is five to eight times greater than the
viscous drag on its upper and lower surfaces. Under-
thrusting involves a downward flexure of the litho
sphere in response to Pnb,and since it behaves in an
elastic manner in the top few tens of kilometres,
flexure is opposed by a bending resistance (-Rb). A
further resistance to motion at subduction zones is
the friction between the two plates. This overriding
plate resistance (Rq) is expressed in the intense
earthquake and tectonic activity observed at shallow
depths at destructive plate margins. The downgoing
slab achieves a terminal velocity when is nearly
balanced by Rg -i- Rq. If F^p exceeds Rg + Rq, the
slab descends at greater than the terminal velocity
and throws the slab into tension at shallow depths. If
Fsp is less than Rg+Ro the slab is thrown into
compression. The balance between driving and resis
tive forces may thus control the distribution ofstress
types, as revealed by earthquake focal mechanism
solutions, within downgoing slabs (Section 8.3).
In the region on the landward side of subduction
zones the overriding lithosphere is thrown into ten
sion by the trench suction force (Rsu)- There are
several possible causes ofthis force (Fig. I I.I la).(I)
It may arise because the angle ofsubduction becomes
progressively greater with depth (Fig. 11,11a). Ten
sion would then arise as the overriding plate collapses
towards the trench. (2) The tension could result
from the 'roll-back' of the underthrusting plate Fig. 11.11 Possible sources ofthe trench suction force
(Fig. I I.Ilb) that is, the downgoing slab retreats (after Forsyth & Uyeda, 1975, with permission from the
from the overriding plate. (3) Tension could be Royal Astronomical Society),
generated by secondary convective flow in the region
overlying the downgoing slab (Fig. 11.lie). This sphere (Section 11.5.2). (4) Tension may arise
would require a relatively high geothermal gradient from any of several mechanisms proposed for the
giving rise to a relatively low viscosity in the astheno- formation of marginal seas on the landward side of
260 CHAPTER 11
subduction zones (Fig. 11.1 Id) as described in Sec
tion 8.9. However, once back-arc spreading com
mences the landward plate becomes decoupled from
the trench system (Fig. ll.lle).
When two plates of continental lithosphere are
brought into contact after the complete consump
tion of an intervening ocean at a subduction zone,
the resistance to any further motion is known as
collision resistance. The mechanism ofthis resistance
is complex because it takes place both at the suture
between the plates and within the overriding plate
(Section 9.3). Finally, transform fault resistance af
fects conservative plate margins in both continental
and oceanic areas. The resistance acts parallel to the
faults and gives rise to earthquakes with a stick-slip
mechanism (Section 2.1.5) confined to a shallow
depth. More complex resistance is encountered when
the fault trend is sinuous so that motion is not purely
strike-slip (Section 7.4.2).
The relative magnitude of the forces acting on
plates and their relevance to the driving mechanism
of plate tectonics will be discussed in Section 11.7.
11.7 Driving mechanism of
plate tectonics
The energy available to drive plate motions is the
heat generated in the core and mantle which is
brought to the surface by convection in the mantle.
It now remains to consider the manner in which this
Ocean
ridge Mantle drag Continent
Tension-^Compression Comoression
Slab pull Suction
Compression ^Tondnn
thermal energy is employed in driving the litho-
spheric plates. The proposal by Morgan (1971;
1972b) that plates are driven by the horizontal flow
of material brought to the base ofthe lithosphere by
hotspots is discounted (Chappie & Tullis, 1977), as
the lateral flow would probably be equal in all hori
zontal directions and thus would not apply a direc
tional force to the plates. Two models have been
proposed. The classical, or mantle drag, model con
siders that the upper, cool, boundary layer of the
convecting system is represented by the upper part of
the asthenosphere, and that plates are driven by the
viscous drag of the asthenosphere on their bases. In
contrast, the edge-force model recognizes the litho
sphere itself as the upper,cool, boundary layer ofthe
convection cells and proposes that the plates are
driven by forces applied to their margins. The two
models thus differ in the importance placed on
the various forces acting on plates (Section 11.6)
described by Forsyth & Uyeda (1975).
11.7.1 Mantle drag mechanism
Mantle drag was the first driving mechanism to be
proposed, and envisages plate motion in response to
the viscous drag exerted on the base of the litho
sphere by the lateral motion of the top of mantle
convection cells in the asthenosphere (Fig. 11.12a).
The convection cells would consequently rise be
neath oceanic ridges and descend beneath trenches.
Ocean
► Mantle drag ridge
Tension
Fig. 11.12 Two concepts of
Ridge plate driving mechanism: (a)
push cellular convection, with the
Comoression cells exerting a mantle drag on
the lithosphere; (b) Orowan-
Elsasser-type convection, with
plates driven by edge forces
(redrawn from Bott, 1982a,
with permission from Edward
Arnold).

being largely absent beneath continental regions.
This mechanism predicts that the oceanic lithosphere
would be in a state oftension at the ocean ridges and
compression at the trenches.
The region of contact between convecting mantle
and lithosphere lies at the low velocity zone at the
top of the asthenosphere (Section 2.8.4). Because
the material in this zone is at a temperature close to
its melting point, the LVZ must possess a very low
viscosity. Consequently, the coupling between LVZ
and lithosphere is not efficient (Lister, 1975), and it
has been calculated that to propel the lithosphere at a
velocity of40 mm a"^ the asthenosphere must move
at 200 mm a"^, a rate rather too high to be geologi
cally reasonable, as the very small relative motions of
hotspots (Section 5.4) argues against rapid horizon
tal velocities in the upper mantle.
Because of their relationship to accretive and
destructive plate margins,the horizontal dimensions
ofthe convection cells powering mantle drag would
be expected to be about half the width of an ocean,
that is, 2500 km. This great lateral extent implies
that the cells should have a relatively simple form. It
is consequently difficult to explain how cells of sim
ple geometry could drive plates with irregularly
shaped margins, such as the Mid-Atlantic Ridge at
equatorial latitudes where it is offset along a suite of
transform faults. Also, the constant geometry of the
convection cells cannot explain the relative move
ments between plate margins, such as is happening
between the Mid-Atlantic and Carlsberg ridges. The
large horizontal dimensions of the cells cannot ac
count for the movements ofsmall plates, such as the
Caribbean and Philippine plates, which can hardly
be powered by their own individual convective sys
tems.
A further factor that argues against the mantle drag
mechanism comes from the rate at which energy is
dissipated in the LVZ of the asthenosphere. The
model can only account for about 10% of the heat
from the mantle, and so is clearly inadequate. It
would therefore seem that the classical mantle drag
mechanism is not the main process causing the
mobility of plates.
It is possible, however, that our views on mantle
drag are biased by the fact that the present continents
are dispersed. Ziegler(1993)argues that mantle drag
may have been a significant mechanism during super-
continent break-up and indeed, Phanerozoic plate
motions appear to require this mechanism.
THE MECHANISM OF PLATE TECTONICS 261
11.7.2 Edge-force mechanism
In this mechanism the oceanic lithosphere represents
the top of the convection system, and the plates
move in response to forces applied to their edges
(Fig. II.12b). The mechanism was first proposed by
Orowan (1965) and Elsasser (1969; 1971) and is
sometimes referred to as Orowan-Elsasser-type con
vection (see also Davies & Richards, 1992).
Only a small percentage of the energy supplied
from the mantle is available to drive the plates, but
this fraction is adequate to power the present plate
motions (Bott, 1982a). The energy is utilized by the
lithosphere to drive the plates in several ways. The
ridge-push force (Section 11.6) originates from the
expansion of hot material from the asthenosphere
which is injected at ocean ridges. This provides a
lateral push to the rear of accreting oceanic litho
sphere. The slab-pull force (Section 11.6) arises from
the negative buoyancy of the downgoing slab at
trenches, and is assisted by the phase changes to
denser forms which affect minerals in the slab at
increased pressure. The slab-pull force is potentially
some four times larger than the ridge-push force,
although in practice much of this force is probably
utilized in overcoming slab resistance (Chappie &
Tullis, 1977). The trench suction force (Section
11.6) originates from the geometry of the downgo
ing slab and also provides a significant driving force.
The edge-force mechanism can account for many
phenomena more satisfactorily than the mantle drag
mechanism,in particular:
1 It is more acceptable thermodynamically and is
much more effective in transporting heat from the
mantle.
2 It is consistent with the observed pattern of intra-
plate stress. As discussed in Section 11.7.1, the
mantle drag mechanism implies tension at ocean
ridges and compression at trenches. The edge-force
mechanism would give rise to the opposite stress
configuration, and this is in accord with the stress
regime indicated by focal mechanism solutions of
intraplate earthquakes.
3 It is reconcilable with the present plate motions,in
particular with the observations of Forsyth & Uyeda
(1975)that:
(a) plate velocity is independent of plate area
(Fig. 11.13a). If mantle drag were operative it
would be expected that the greatest velocities
would be experienced by plates with the greatest
262 CHAPTER 11

mantle drag inhibits the motion of such plates

(a)
rather than driving them.
_100i The mechanism also provides a reasonable expla
E
cc
nation of the motions of small plates.
(O
O
2< g <
Consequently, the edge-force mechanism of plate
< movement appears to be much more successful in
S 50 CO
u.
CO
explaining all observed phenomena, and has been
txi
<
cc
generally adopted by most workers.
<

50 100 11.8 Areal distribution of

Velocity(mm a~^)
(b)
convection ceiis
Convection cells are believed to exist beneath oce
!§ 30%-, anic regions, where their presence is required to
<D
O O) explain the heat flow that cannot originate from the
c c
2? o
0 O)
oceanic crust. The continental crust produces a large
I 20% proportion of the heat flow observed over continen
3 o
E ■§ tal regions, and the heat flow originating at sub-
o o
® t: 10%
crustal levels is much smaller. Thus, only a subsidiary
iS B
CL u
type of convection is believed to occur beneath the
C
c
continents, and this is manifested in the presence of
o
■ li ■ I spasmodic 'hotspot' volcanic activity through which
50 100 the heat building up beneath continental regions is
(0 Velocity (mm a~^)
periodically released.
Since the edge-force model of plate movement
E
(Section 11.7.2) implies that the oceanic lithosphere
(O
50 n represents the upper, cooling portion of convection
O
cells, it would be expected that the distribution of
these cells correlates with the configuration of the
oceanic plates. Thus, the hot ascending limbs of
convection cells would be situated beneath ocean
ridges and the cool descending limbs beneath sub-
duction zones.
—I "— McKenzie et al. (1980) and McKenzie (1983)
c
50 100
o
o
Velocity (mm a"^)
have attempted to map the areal distribution of
convection cells in the Pacific Ocean from their
Fig. 11.13 Correlations of plate parameters with plate gravity anomalies. Conventional gravity anomalies
velocity: (a) plate area; (b) plate circumference connected measured from ships are dominated by short wave
to downgoing slab (open bar, total length; filled bar, length components of crustal origin, with long wave
effective length); (c) continental area of plate (redrawn length errors induced by gravimeter drift and faulty
from Forsyth & Uyeda, 1975, with permission from the
calibration. Consequently, these studies relied on the
Royal Astronomical Society).
effect that large scale, deep structures would have on
the sea level surface or geoid. The geoid would be
area over which the mantle drag would act; warped up over mass excesses and down over mass
(b) plates attached to downgoing slabs move more deficiencies. Such mass anomalies would also be
rapidly than other plates (Fig, 11.13b). This is in expected to be reflected in the bathymetry of the
accord with the slab-pull force being greater than oceans; data have been corrected for the subsidence
other forces affecting the plates; of the oceanic lithosphere with age (Section 6.9).
(c) plates with a large area of continental crust The geoid was mapped by radar altimeters carried
move more slowly (Fig. 11.13c). This implies that in orbiting satellites. Knowing the precise position of
 THE MECHANISM OF PLATE TECTONICS 263

the satellites, the height ofthe sea level surface can be
determined from the two-way travel time of a radar
pulse. The geoid and bathymetric data were subse
quently filtered to remove variations with wave
lengths of less than 300 km, which originate within
the lithosphere, in order to isolate anomalies caused
by more regional phenomena(Fig. 11.14). Although
it might be expected that the low density region
associated with a rising plume would give rise to a
negative gravity anomaly, this effect is swamped
by the bathymetric bulge over the plume. Con
sequently, hot plumes are associated with positive
geoid anomalies and cold descending plumes with
negatives. The correlation between Figs 11.14a and
11.14b is further demonstrated by the profiles shown
in Fig. 11.15, which also show the temperature dis
tribution and flow pattern in a two-dimensional layer
which might give rise to the observed bathymetry,
gravity and geoid anomalies.
Examination of Fig. 11.14a shows that the sea
floor around Hawaii is shallower than suggested by
its age and is associated with a positive geoid anom
aly. This suggests that the Hawaiian Islands are
underlain by a hot ascending plume, which is in ac
cord with other observations suggesting this mode of
origin (Section 5.4). Other regions are too deep for
their age, and must thus be underlain by cool, desc
ending mantle material. There is no correspondence
between positive geoid anomalies and ocean ridges;
indeed, some ridges are associated with negative
anomalies. The study suggests that accretive plate
margins are not necessarily associated with the ascend
ing limbs ofconvection cells, and so sea floor spread
ing in some areas is merely a passive phenomenon

30''N 20" N
10" N

10"S

20"S

(a) 170"E 180"W /160"W 140"W

2500

30" N 20" N

10" N

Hawaii

10"S

Fig. 11.14 (a) Filtered residual

depth map ofthe Pacific Plate, 20"S
contour interval 500 m.(b) Filtered
gravity anomaly map ofthe Pacific
Plate, contour interval 200 g.u.
(redrawn from McKenzie, 1983). (b) 170"E 180"W 160"W 140"W
264 CHAPTER 11

the A;-axis of Fig. 11.14b, which corresponds to the

direction of plate motions, and appear to become
increasingly extended as plate velocity increases.
The geoid anomalies suggest that mantle convec
tion is taking place whose horizontal scale is smaller
than the plates. The elongation parallel to spreading
direction has been suggested by theoretical calcula
tions and experimentation (Richter, 1973; McKen
zie 8c Weiss, 1975; Richter 8c Parsons, 1975) to
indicate, at least for fast moving plates, the presence
of two scales of convective circulation (Fig. 11.16).
The smaller scale, with horizontal dimensions of
about 500 km normal to the spreading direction,
takes the form of elongate rolls extending from the
base ofthe lithosphere to about650 km depth whose
axes parallel the spreading direction. These are be
lieved to be necessary to explain the observed heat
flow through the older oceanic lithosphere (McKen
zie 8c Weiss, 1975). These small secondary cells are
not predicted to cause significant nett drag on the
base of the lithosphere and would thus not affect
o OJ
plate motions. The large-scale circulations, with hor
izontal dimensions of some 5000 km, consist of the
plates themselves and the complementary return
1000 2000 3000 4000 5000 flow at depth. No information is provided by these
Distance(km)
studies on the depth of the base of the convection
Fig. 11.15 Numerical simulation of mantle convection: cells, although the geometry ofthe anomalies is most
(a) flow pattern;(b)temperature contours at 100°C easily explained by instabilities ofa relatively shallow,
interval;(c) variation in ocean depth resulting from hot boundary layer.
convection;(d) variation in gravity;(e) variation in geoid The ability to map convection cells from their
height(redrawn from McKenzie, 1983).
associated geoid anomalies clearly represents a very
important step in the understanding of the convec
whereby mantle material is drawn into a zone of tive mechanism. Several of the discoveries of this
melting(Houseman,1983; Klein 1988). Most method,such as the non-coincidence ofocean ridges
of the geoid anomalies are elongated parallel to and hot mantle plumes, are in conflict with previ-

Fig. 11.16 Schematic illustration

oflarge-scale flows in the vicinity
of an ocean ridge and
superimposed small scale
longitudinal rolls (redrawn from
Richter 8c Parsons, 1975, with
permission from the American
Geophysical Union).

ously accepted models. It must be stated, however,
that there is no universal agreement that the plan-
form of mantle convection is evident in the geoid
(Sandwell & Renkin, 1988). We must now await
further studies of this kind in other ocean basins to
see if results from the Pacific Ocean are generally
applicable.
Much important information on the three-
dimensional structure of the mantle has been
supplied by seismic tomography (Section 2.1.8).
Convection is driven by lateral differences in
temperature and density. These variables affect seis
mic velocity, which decreases with decreasing density
and increasing temperature (Dziewonski & Ander
son, 1984). By mapping velocities in the mantle it is
possible to infer the differences in temperature and
density which are a consequence ofconvection. Also,
by mapping seismic anisotropy both vertically and
laterally it is possible to obtain estimates of the
direction of mantle flow.
Woodhouse & Dziewonski(1984) have derived a
three-dimensional velocity model of the upper
mantle by the inversion of a large dataset of shear
wave arrivals. Because of the wavelengths involved
the resolution is about 2500 km horizontally and
200 km vertically. They found that the velocity struc
ture is only related to surface tectonic features down
to a depth ofabout 200 km. Ocean ridges are under
lain by low velocity mantle, shields by very fast
velocities and old ocean basins by high velocities.
High velocities probably reflect the presence of ma
ture mantle materials which are colder and denser
than average. Between 200 km and 400 km there are
some remnants of the surface features. Most shield
areas are still underlain by positive velocity anoma
lies, but the velocity contrasts are lower than near the
surface. The origin of this high velocity material
beneath shields is not known;it could reflect material
that is translating with the overlying shield,implying
that shields maintain a continuous identity to this
depth; it might equally well represent cold, sub
ducted material overridden by the continent. Within
this depth range only a small proportion ofthe ocean
ridge system retains anomalous velocities. The fast
spreading East Pacific Rise is underlain by nearly
normal mantle material below 200 km. However,
isolated regions beneath ocean ridges are underlain
by low velocities extending to 400 km, and there is
evidence that some ridge segments are fed by lateral
transport of material at shallow depth. Below about
THE MECHANISM OF PLATE TECTONICS 265
300 km new velocity features develop which become
fully established in the transition zone. Their pattern
is completely different from that near the surface. For
example, there are two broad, high velocity zones,
one extending from the western Pacific under south
ern Australia to the Antarctic coast, the other be
neath South America,the South Atlantic and parts of
Africa. At this depth the localized low velocity zones
associated with hotspots near the surface change to
high velocities in the transition zone. This suggests
that hotspots originate at shallow depths. However,
they might be fed through conduits from much
greater depths which are too narrow to be resolved by
the model.
A velocity model of the lower mantle has been
derived from P wave arrival times by Dziewonski
(1984). In general, velocity anomalies in this region
are much smaller than in the upper mantle. The
Pacific is surrounded by a ring ofhigh velocities, and
there is some correlation between hotspots and ve
locity lows.
Figure 11.17 shows four global cross-sections
illustrating the velocity models for both the upper
and lower mantle(note that the vertical exaggeration
ofthe upper mantle model is twice that ofthe lower
mantle). Unfortunately the resolution at a depth of
670 km is quite low, and this makes it difficult to
trace features between the two models.Figure 11.17a
shows an equatorial section. The high velocity
regions beneath the Brazilian and African shields are
separated by the low velocities of the Mid-Atlantic
Ridge near the surface. At about 400 km the highs
merge to form a broad high velocity region which
appears to continue into the lower mantle at a lower
angle. The low velocities associated with the Carls-
berg Ridge extend into the uppermost lower mantle,
while the Pacific Basin is underlain by low velocity
material extending to the core-mantle boundary.
The marginal seas of Indonesia are underlain by low
velocities to a depth of200 km.The underlying high
velocity zone, which appears to penetrate the transi
tion zone, may be related to subducted material.
Figure 11.17b shows a section across the Pacific
approximately parallel to the spreading direction. As
expected, lithospheric velocities increase with in
creasing age. At the western margin ofthe Pacific the
high velocity material is probably related to subduc-
tion, and merges with the high velocity material
under Eurasia to form an extensive positive region in
the lower mantle extending to the core-mantle
266 CHAPTER 11

670

(km)

I + 3.0%
I + 0.75%

Fig. 11.17 Four vertical

sections through the model of
the mantle derived by seismic
tomography. For the upper
mantle the velocity
perturbations(up to ±3%)are
for shear wave velocities; in the
lower mantle they are P wave
o liii
velocity perturbations (up to
±0.75%)(redrawn from
Woodhouse & Dziewonski,
1984, with permission frtim
the American Geophysica I
Union).

boundary. The profile crosses the velocity low be
neath the Atlantic which, as in Fig. 11.17a, divides
the high velocity material beneath South America
and Africa which converge at depth. The low velocity
anomaly beneath the Red Sea may continue into the
lower mantle.
Figure 11.17c runs along the ridges ofthe east and
south Pacific. It illustrates the variability ofthe depth
extent of the low velocity material underlying the
ridges, and the persistence of the high velocities of
the North American and Eurasian plates to about
400 km. Figure 11.17d crosses the low velocity re
gion south of Hawaii, and shows it to be continuous
with an anomaly extending to the core-mantle
boundary. The Tonga-Kermadec Trench is not, sur
prisingly, associated with high velocities in the upper
mantle, but is underlain by a high velocity column in
the lower mantle.
Unfortunately, because of the lack of resolutiiin at
the depth of the transition zone, these models can
not provide any information on the vertical extent of
convection.
More recent tomographic studies of the mantle,
which have been summarized by Montagner (1994),
have greater resolution. All investigations sho^v ve
locity anomalies in the upper 400 km which correlate
with surface tectonic features. The upper 200 km
have slow velocities below plate boundaries. Only
fast spreading ridges are visible between 200 and
400 km and the corresponding low velocity zones are

Fig.11.17 {Continued).
often displaced laterally. High velocities are only
present beneath large continents. Below 400 km ve
locity anomalies decrease in size, with the largest in
the western Pacific, below the Alpine-Himalayan
mountains and below the Atlantic; these may be
related to subducting lithosphere. The anomalous
velocity zones of spreading centres are no longer
visible, but large lows are present beneath the central
Pacific and eastern Africa.
Information on the actual directions offlow in the
mantle can be obtained by tomographic analyses of
anisotropy in the mantle. Most mantle minerals are
elongate and so, if subjected to recrystallization,
shearing or flow they can assume a predominant
orientation which is retained long after the processes
THE MECHANISM OF PLATE TECTONICS 267
I + 3.0%
I + 0.75%
r...r^
I + 3.0%
I + 0.75%
causing it have ceased. The preferred orientation of
crystals causes seismic waves to propagate faster in
the direction of elongation than transverse to it, that
is, it imparts an azimuthal anisotropy to the mantle.
Also, transverse, or polarization, anisotropy can be
studied in which S waves propagate horizontally
(SH)with a different velocity from vertically(SV). By
making use ofthe very accurate seismic data available
from digital recording systems it is now possible to
map past and present flow lines using azimuthal
anisotropy, whereas transverse anisotropy can be
used to determine the relative importance of hori
zontal and vertical flow and to map the boundaries of
convection cells, as SH is greater than SV for waves
travelling in the direction of flow. Results are still
268 CHAPTER 11
only preliminary, and more data are required before
the details of mantle convection can be mapped.
Tanimoto & Anderson (1985) have studied azi-
muthal anisotropy. Their results are consistent with a
low velocity channel in the upper mantle,and show a
fair correlation between fast directions and plate
motions. Nataf et al. (1984) have studied polariza
tion anisotropy. Horizontal flow appears to be re
lated to plate motions under the ocean basins and
continents. Upward vertical flow of warm material
appears to take place beneath ocean ridges, while
downward vertical flow of cold materials appears to
occur beneath the oldest ocean basins and subduc-
tion zones.
The methods of studying the distribution of con-
vective flow ofthe mantle are still in their infancy. It
is anticipated that, with refinements of methodology
and the accumulation of more data,they will provide
important new information in the future.
Further reading
Bott, M.H.P. (1982a) The Interior of the Earth, Its Struc
ture, Constitution and Evolution (2nd edn). Edward Ar
nold, London.
Davies, G.F. & Richards, M.A.(1992) Mantle convection.
J.£ieoL 100, 151-206.
Elsasser, W.M.(1969) Convection and stress propagation
in the upper mantle. In Runcorn,S.K.(ed.) The Applica
tion ofModem Physics to the Earth and Planetary Interiors,
pp. 223-46. Wiley (Interscience), New York.
Forsyth, D.W. & Uyeda,S.(1975) On the relative import
ance of driving forces of plate motion.Geophys. J, Roy.
astr. Soc. 43,163-200.
McKenzie, D.P. (1983) The Earth's mantle. Sci. Am.
249(3), 50-67.
Peltier, W.R.(1985) Mantle convection and viscoelasticity.
Ann. Rsv. Fluid. Mech. 17, 561-608.
Pollack, H.N. & Chapman, D.S.(1977) The flow of heat
from the Earth's interior. Sci. Am. 237(2),60-76.
Ringwood,A.E.&Irifune,T.(1988)Nature ofthe 650 km
seismic discontinuity: implications for mantle dynamics
and differentiation. Nature 331,131-6.
12/ Implications of plate tectonics
12.1 Palaeozoic plate tectonics
Palaeogeographic maps for the Mesozoic and Ceno-
zoic can be computed by the fitting together of
continental margins or oceanic lineations of the
same age either side of an ocean ridge (Chapters 3
and 4). The location of the palaeopoles can be
determined from palaeomagnetic measurements
(Section 3.6) and so the only unknown in these
reconstructions is the zero meridian oflongitude.
These techniques cannot be used for reconstruc
tions in pre-Mesozoic times, as in situ oceanic crust is
lacking. The existence of plate tectonic mechanisms
is well documented in the Palaeozoic rock record
from the preservation ofoceanic lithosphere in alloch-
thonous ophiolite complexes, from the presence of
orogenic belts which mark ancient Himalayan-type
mountain belts at the sutures between collided
continents and from the evidence provided from the
past distributions of flora and fauna. The only
method of quantifying plate motions in pre-
Mesozoic times is by the use of palaeomagnetic data.
Ancient plate edges, albeit somewhat distorted, are
jnarked by orogenic belts. Palaeomagnetic data from
each of the plates are rotated relative to some arbi
trary fixed datum, usually the African continent, and
projected as a map. The continents are then succes
sively shifted along fixed latitudes, that is, rotated
about the Eulerian pole, until the overlap of conti
nental margins is minimized.
Morel & Irving (1978) and Smith et aL (1981)
have presented palaeogeographic maps for Palaeo
zoic times using this method. Morel & Irving have
tentatively identified the major stages in global evo
lution during this time (Fig. 12.1). At the beginning
of Palaeozoic times the plates were arranged into a
supercontinent (Fig. 12.1a). In Cambrian times the
supercontinent fragmented (Bond et 1984),
with the opening of the first proto-Atlantic Ocean
(Fig. 12.1b,c). This closed, and the supercontinent
reassembled, in Late Silurian times, producing the
Caledonian Orogeny (Fig. 12.Id). In Early and
Middle Devonian times internal adjustments oc
curred and transcurrent movement between the
northern and southern parts of the supercontinent
produced the Acadian Orogeny (Fig. 12.1e). A sec
ond phase of opening in the Late Devonian created
the second proto-Atlantic (Fig. 12.If), which closed
in Middle Carboniferous times producing the Varis-
can Orogeny and the arrangement of the plates
into the supercontinental assembly of Pangaea
(Fig.l2.1g).
Although the palaeomagnetic data do not provide
a unique sequence of reconstructions, they clearly
indicate the gross trends of plate movements during
this time. More detailed inferences on the evolution
of particular regions can be made from their geology
viewed in terms of the plate tectonics concept (see,
for example, Dewey, 1969; Phillips ,1976).
12.2 Precambrian plate tectonics
12.2.1 Introduction
In considering the nature of global tectonic activity
in Precambrian times, three approaches may be
adopted (Kroner, 1981). First, a strictly uniformitar-
ian approach can be taken in which Precambrian
tectonics originated by the same mechanism of plate
tectonics active in Phanerozoic times. Second, a
modified uniformitarian approach can be postulated
in which plate tectonic processes in the Precambrian
were somewhat different from present because the
physical conditions affecting the crust and mantle
have changed throughout geological time. Third, a
completely different tectonic mechanism can be in
voked for Precambrian times. Only the first two of
these approaches will be considered, as, unless there
are cogent reasons for doing so, the invocation of
mechanisms unrelated to those of the present day
appears to involve unjustified complication (Glikson,
1981).
269
270 CHAPTER 12

600 Ma 500 Ma

450 Ma 400 Ma

375 Ma 350 Ma

Laurasia

Acadian
FB

\ /\
Gondwana
 IMPLICATIONS OF PLATE TECTONICS 271

325 Ma

Appalachian
Hercynian

LA-Laurentia
SN-Chlnes'eShieid
SB-Siberian Shield
BA-Baltica
B-Bohemian FB

Fig. 12.1 {Facin^f pa^e and above) Palaeomagnetically- (e) Early-Middle Devonian;(f) Late Devonian;
derivcd continental reconstructions for the Palaeozoic: (g) Middle Carboniferous. FB,fold belt(redrawn from
(a) end-Precambrian;(b) Late Cambrian-Early Morel & Irving, 1978, with permission from the
Ordovician;(c) Late Ordovician;(d) Late Silurian; University ofChicago Press).

12.2.2 Heatflow in the Precambrian

The most important physical parameter to have 400

varied throughout geological time is heat flow. The
majority of terrestrial heat production comes from
radioactive minerals dispersed through the core,
mantle and continental crust (Section 2.13) and e 300
because oftheir exponential decay rates the heat flow 5
in the past must have been considerably greater than
at present (Fig. 12.2). For an Earth model with a
K/U ratio derived from measurements of crustal I 200
(D
rocks, the heat flow at 4.5 Ga would have been some I

four-and-a-half times greater than at the present day

(Hargraves, 1981), and at 3.8 Ga about two-and-a-
100
half times the present value (Kroner, 1981).
It is probable that the geothermal gradient in the
Precambrian continental crust was not much greater
than at present. Windley (1981) argues that the J_ -L. -L. j
temperatures in the lower crust, presently up to 2 3 4 Present
about 500 ®C, could not have exceeded 800"C or Increasing age(Ga)
extensive partial melting would have occurred. There Fig.12.2 Variation ofsurface heat flow with time. Solid
is no evidence of the magmatic products expected line, based on a chondritic model; dashed line, based on a
from such melting in presently exposed Archaean K/U ratio derived from crustal rocks(redrawn from
granulites and studies of metamorphic mineral McKenzie & Weiss, 1975, with permission from the Royal
assemblages suggest temperature gradients of Astronomical Society).
272 CHAPTER 12
20-25° km"^ (Bickle, 1978). The conduction mech
anism of heat transfer in the continental crust is
inefficient; in contrast, the convection process opera
tive beneath the oceanic lithosphere is far more
effective in dissipating heat from the interior of the
Earth. Since the oceanic lithosphere probably repre
sents the upper layer ofthe mantle convecting system
(Section 11.7.2), the increased heat supply in the
mantle could be dissipated by increasing the length
ofthe ocean ridge system or by increasing the rate of
plate production by a factor of at least 6 with respect
to the present (Bickle, 1978). Assuming a non-
expanding Earth (Section 11.3),this implies a similar
increase in plate subduction rate and suggests that
mantle convection was very vigorous. Certainly,
these computations suggest that some form of plate
tectonics was taking place during the Precambrian at
a much greater rate than today.
The increased heat flow in the Precambrian also
has implications bearing on the nature of the plate
driving forces. The increased length of ocean ridges
or increased rate of plate accretion implies that either
the mean distance from ridge to trench, or the mean
age of oceanic lithosphere, decreased. The thickness
and density of oceanic lithosphere are a function of
age (Section 6.9), and it has been argued (Hargraves,
1981)that subduction only occurs when the oceanic
lithosphere has attained some critical density at
which it becomes negatively buoyant. The conclu
sion is that, in Precambrian times, this density was
not achieved so that the principal plate driving force
ofslab pull could not have operated. This conclusion
is questionable, however, as at the present time there
is little correlation between lithospheric age and
subduction. In the Pacific, oceanic lithosphere as
young as 5 Ma is being subducted while in the
Atlantic there are extensive regions of lithosphere
200 Ma old which are not being subducted.
Hargraves (1981) has also argued that the mantle
drag force beneath Precambrian oceanic lithosphere
would have been stronger than at present because of
the more rapid convective overturn, and could have
been of sufficient magnitude to drag down the posi
tively buoyant lithosphere. Hargraves has concluded
that the gravitational forces on a crust of density
contrast 0.25 Mgm"^ would, if less than 20 km
thick, exceed the critical value of 5xlO^Nm"^
(McKenzie & Weiss, 1975) at which it would be
dragged down into the mantle. This calculation,
however, assumes that the lithosphere consisted en
tirely of basaltic crust. It also seems unlikely that the
lithosphere could have been as thin as 20 km. More
over, the high temperature of the asthenosphere
implies a low viscosity, so that the viscous coupling
between lithosphere and asthenosphere would be
expected to be too low to create significant mantle
drag forces. If the Earth were not to have expanded
considerably during its early history, some sort of
subduction must have taken place. If plate motions
were not powered by mantle drag forces it is probable
that the edge-force mechanism (Section 11.7.2) was
active, as it is believed to be at present. Certainly
increased ocean ridge length/more rapid spreading
rates imply a powerful ridge-push force and it must
be concluded that the density of the lithosphere,
even though lower than at present, must have been
sufficient to allow the initiation ofsubduction. Once
the subducted plate reached depths of 70-80 km,
phase changes affecting the lithosphere would have
increased its density and facilitated further subduc
tion. The magmatic products of the underthrusting
would have increased the crustal thickness in island
arc environments, and the collisions of different arcs
would have created regions of high negative buoy
ancy which may have evolved into protocontinents.
As the radioactive heat sources decayed,the heat flow
diminished and the lithosphere would have thick
ened and become more dense. The thermal necessity
of extensive ocean ridges and/or rapid spreading
would gradually be removed, and the plate tectonic
regime would have gradually changed over to that
typical of Phanerozoic times. Given the differing
tectonic environments of the older and younger
Precambrian, it seems reasonable to postulate that
the changeover to modern-type plate tectonics took
place around the time of the Archaean-Proterozoic
boundary. It must be appreciated, however, that
conditions in the Early Precambrian are quite conjec
tural, so that these conclusions are speculative.
12.2.3 Archaean tectonics
The Archaean is dominated by two types of terrain
which must have formed in different tectonic
environments: the greenstone or greenstone-granite
belts and granulite gneiss belts (Windley, 1981).
Greenstone belts are typified by a synformal struc
ture, 40-250 km wide and 120-800 km long. They
exhibit a low pressure regional metamorphism of
the greenschist facies and are intruded by plutons

ranging in composition from tonalite through
granodiorite to potassic granite. They are divided
stratigraphically into three groups: a lower group of
komatiitic ultramafic and mafic volcanic rocks with
pillow structures which have a bulk composition
similar to ocean ridge basalts; a central group of
andesites and calc-alkaline volcanic rocks whose trace
and rare earth elements are similar to rocks found on
island arcs; and an upper group of clastic sediments,
such as greywackes, sandstones, conglomerates and
banded ironstones with cherts and limestones.
Granulite-gneiss belts exhibit a high grade of
metamorphism and comprise three components:
quartzofeldspathic gneiss oflargely tonalitic to gran-
odioritic composition; layered peridotite-gabbro-
anorthosite or leucogabbro-anorthosite complexes,
and metavolcanic amphibolites and metasediments.
Most of the regions in which granulite-gneiss belts
are situated were subjected to a major accretionary
event, indicated by isotope abundances, during the
period 2700-3100 Ma.
Workers adopting a non-uniformitarian view of
the Archaean propose that the crust of this time was
Ultramafic
group
(a)
Thin unstable
primitive crust,
inpartsialic
Gravity
deformation
(b)
Fig. 12.3 Diagrammatic model ofthe
evolution of an Archaean greenstone belt by
progressive sinking of a volcano-sedimentary
basin followed by compression (redrawn from
Windley, 1984, with permission from John
Wiley). (c)
IMPLICATIONS OF PLATE TECTONICS 273
developed from a continuous global scum of inter
mediate composition which was continuously re
cycled as a result of surface volcanism and remelting
at its base (Hargraves, 1981). Windley (1984) has
summarized the mechanisms proposed for the two
main crustal members preserved from this time. A
typical non-uniformitarian model for the formation
of greenstone belts is shown in Fig. 12.3. A down-
warp in the primitive crust was formed in response to
the load of injected ultramafic rocks, and subse
quently filled by sediments. Later the crust was be
lieved to have thickened and intruded by diapiric
granites which lead to compression of the belt.
Most workers who suggest a plate tectonic model
for the formation of greenstone belts propose that
they originated in a subduction-related setting. A
particularly convincing model has been proposed by
Tarney et al, (1976). Their studies of the Rocas
Verdes complex in southern Chile (Section 8.9 and
Fig. 8.38) lead them to postulate that greenstone
belts originated in ensialic marginal basins of this
type. Their model is shown in Fig. 12.4. It envisages
the greenstone belts as developing on thinned
Sediments
Calc-alkaline
volcanics
Mantle tapping
fracture
Conglomerates
Granite development
— crust thickens
Greenstone belt
stoped around its
margins by
granite invasion
274 CHAPTER 12

Volcanic
arc
Extension

.5555%%%?.%?;?;

® Deep source
»»»»&»»»»^
komatiite '»M*W

Volcanic phase

5!*%5!W.swX*••••*******^^^^

SS^S^lSSflSSSgSSS^^ IS? JSS555$iSS55$%aj^^

Sediment
Continental sediment + lavas
Sedimentary phase

Deformation 4-
synorogenic tonalite
Basin closure

•»;«X*Xw.

I%%%%wXv.v
5:S!%»»:oK*;w5Wi%%x*>>
w.vKv.w.?.!

Late K-granites
Fig. 12.4 Development of an Archaean
greenstone belt from a marginal basin, based
on a comparison with the Rocas Verdes
complex in southern Chile (redrawn from
WK«5!W
Tarney flj/., 1976, with permission from
v.%VA%v.v.v.v.v*:.;.:.v.v
John Wiley).

continental crust in a back-arc environment, covered
by sediments originating from the flanking continent
and volcanic arc, and subsequently becoming de
formed and intruded by granites as the basin was
subjected to compression.The structure,volcanosedi-
mentary rock associations,geochemistry and igneous
history of the Rocas Verdes complex are remarkably
similar to Archaean greenstone belts (Windley,
1984). The back-arc environment also explains how
Archaean oceanic crust came to be preserved;ifit had
originated in a normal ocean basin it would probably
have been completely destroyed by subduction.
Far less attempt has been made to propose tectonic
models for the formation of granulite-gneiss belts.
Clearly, though, any reasonable model must include
a mechanism for the production ofthe great quanti
ties of tonalite which make up the major part of
these belts. Windley(1981) has noted the geological
and geochemical similarities between Archaean
granulite-gneiss belts and the deeply eroded granitic
rocks associated with Andean-type subduction
zones. He considers this the only environment in
which voluminous quantities of tonalite can be pro
duced, and concludes that this represents a reason-

Sediments and lavas in Siaiic 'basement' of folded gneisses
extensional backarc (Amitsoq-type) with relic
basin metasupracrustais (Isua-type)
High grade terrain with tonaiitic
gneisses interthrusted and folded
with 'basement' gneisses and Greenstone belts
shelf sediments .
Fig. 12.5 Plate tectonic model for the development and
growth of continents in the Archaean,(a) Widespread
small continental plates with ensialic back-arc basins. The
Andean-type margins are invaded by mantle-derived
tonaiites in batholithic quantities,(b) Aggregation of the
able analogue for the formation of granulite-gneiss
belts.
Windley (1984) has argued that some form of
plate tectonics is the only mechanism capable of
continuously replenishing, over hundreds of mil
lions of years, the mantle-derived material which
enabled a large proportion ofthe present continental
crust to have been created during Archaean times.
The mechanism was probably not identical to
present day plate tectonics because the increased heat
flow in Archaean times would have led to a thinner
lithosphere and greater plate motions. It can, how
ever, provide a mechanism for the complementary
origin of both greenstone and granulite-gneiss belts
whereby the marginal basin and subduction-related
batholith models are combined (Fig. 12.5). This
mechanism views the Archaean crust as being made
up of a large number of small, relatively thin, siaiic
crustal blocks. Subduction of oceanic crust between
these gave rise to back-arc spreading and the forma
tion of oceanic-type crust (Fig. 12.5a). Subduction
beneath these blocks also gave rise to calc-alkaline
volcanics and the intrusion of large bodies of
tonaiitic magma (Fig. 12.5a). By the end of the
IMPLICATIONS OF PLATE TECTONICS
Tonaiites as horizontal Sheif-type
sheet intrusions in main arc sediments
Late K-granite
continents producing an extensive continental plate
consisting of greenstone belts and granulite gneiss belts
(redrawn from Windley, 1984, with permission from John
Wiley).
Archaean (Fig. 12.5b), these crustal fragments had
coalesced into composite siaiic blocks characterized
by alternate belts of greenstone and granulite gneiss
formed, respectively, from the oceanic crust and
eroded tonaiitic batholiths. This phase gave rise to
the larger, thicker plates which dominated litho-
spheric evolution in the Proterozoic.
12.2.4 Precambrian palaeomagnetic
studies
Potentially the most rigorous method of investigat
ing the presence of relative plate motions in Precam
brian times is that of palaeomagnetism (Section 3.6).
The application of this method is rather less straight
forward than for Phanerozoic times for three main
reasons (Dunlop, 1981). First, the error limits of
radiometric age determinations are larger. Second,
isotopic and magnetic records may be partially reset
during metamorphism to different degrees, and the
distinction between pre- and post-orogenic isotopic
and magnetic overprints is extremely difficult. Third,
overprints occur during post-orogenic cooling and
uplift, and the temperatures at which isotopic
276 CHAPTER 12
Fig. 12.6 Summary apparent polar wander path for the
proposed Proterozoic supercontinent between 2200 and
800 Ma. Dated points: approximate locations of Eulerian
poles of rotation between 2700 and 1000 Ma(redrawn
from Piper, 1987, with permission from the Open
University Press).
systems close and magnetizations stabilize are dif
ferent,so that the dates may be younger or older than
the magnetizations by intervals of tens of millions of
years. However, even given these uncertainties and
the gaps in the palaeomagnetic record arising from
the lack of suitable samples of certain ages, the data
should be adequate to determine the major features
of apparent polar movement relative to each craton
(Briden, 1976).
Piper (1987) has synthesized all available palaeo
magnetic data for the Precambrian. Although the
data are patchy both in space and time, he has shown
that the APW paths for the best studied continents
are similar in shape, while the poles of the more
sparsely sampled continents are not inconsistent
with APW paths of the same form. The APW paths
for the individual cratons can be brought into coin
cidence by a sequence of angular rotations about
Eulerian poles to provide the composite APW curve
shown in Fig. 12.6. From this Piper concluded that
in late Archaean times the individual cratons coa
lesced into a single supercontinent which remained
coherent, within the limits of palaeomagnetic detec
tion, throughout the whole of Proterozoic times.
The composite APW curve of the supercontinent is
illustrated as a broad swathe whose width reflects the
effects of a number of factors, including incomplete
subtraction of non-primary magnetizations, non-
dipole components of the geomagnetic field and
warping of the continents not evident in their geol
ogy (Piper, 1982). The APW path exhibits eight
 IMPLICATIONS OF PLATE TECTONICS 277

loops of consistent shape and a periodicity of about

200 Ma which radiate outwards from a common
location which seems to have been stable. From
2150 Ma until the Late Precambrian all the Eulerian
poles associated with motions ofthe supercontinent
are situated within it. This implies that the supercon
tinent underwent rotation rather than any horizontal
translation by large distances. The supercontinent
consequently appears to have remained in the same
location rather than being pushed over the Earth's
surface by the ocean basins which surrounded it. The
fundamental cause ofthis phenomenon is unknown,
but must be related to the pattern of mantle convec
tion of this time.
By applying the same rotations about the same
Eulerian poles to the continents as to their APW
paths and fitting them together to minimize over '^hazakstani
laps, the supercontinent can be reconstructed as
shown in Fig. 12.7. The success of this reconstruc
tion can be judged from the continuity of geological
features across the cratonic margins. Features which
can be traced in this way include greenstone belts, the
fundamental Archaean lineaments known as straight
belts, and metallogenic provinces. It may be relevant
to the origin of the supercontinent that both the
greenstone and straight belts appear to strike parallel
to its long axis.
A controversial alternative reconstruction of the
Precambrian supercontinent has been suggested by
Dalziel(1991; 1995), Moores(1991) and Hoffman
(1991), who have proposed that the southwestern
coasts of North America and eastern Antarctica were
juxtaposed in Late Proterozoic times. This has be
come known as the SWEAT (South West US and
East AnTarctica) hypothesis. It was suggested that
the rifting of this continent resulted in its turning
inside out (or 'extraversion') and the drifting of its
components into their more modern configuration. Australia
The motion of Antarctica would inevitably have
involved Australia, as the two were joined until
95 Ma. The hypothesis initially arose from the simi Fig. 12.7 Proposed Proterozoic supercontinent.
larity in the stratigraphy (Young, 1992) and struc Mollweide projection (redrawn from Piper, 1987, with
ture of the two Precambrian continental margins, permission from the Open University Press).
including the correlation ofEarly to Mid-Proterozoic
orogenic belts across them. tica. The reconstruction shows that the Grenville
Figure 12.8a shows the Late Proterozoic recon Province (a high grade metamorphic terrain with
struction of Hoffman (1991). In contrast to the anorthosite intrusions of about 1100 Ma age) of
reconstruction of Piper (1987), Laurentia (North eastern and southern North America continues di
America and Greenland)forms the core ofthe super- rectly into east Antarctica, and similar belts of this
continent and is flanked to the north by east Antarc- age can be traced over most of the Gondwana
ProtoPacific Ocean

88

+ (BaitiM

traversion of the
an. 1991h

fragments. The age of the oldest sediments associ
ated with break-up, and the provinciality of certain
animal groups across the split,suggest that the super-
continent fragmented at about 750 Ma (Storey,
1993), with the blocks now making up eastern
Gondwana (east Antarctica, Australia and India)
moving anticlockwise, opening the proto-Pacific
Ocean, and colliding with the blocks of western
Gondwanaland(Congo,west Africa and Amazonia).
The intervening Mozambique Ocean closed by the
pincer-like movements of these blocks and Gond
wana was created when they collided to form the
Mozambique belt of east Africa and Madagascar.
Eastern Gondwana then rotated clockwise away from
Laurentia about 200 Ma later. Southern Africa was
located at the pivot of these movements and Baltica
moved independently away from Laurentia, opening
the lapetus Ocean, which was subsequently to close
during the assembly ofPangaea. Figure 12.8b shows
the configuration at 500 Ma. It appears to be con
firmed by the similarity in the structural grains of
northwestern Canada,southern Australia and Adelie
land in Antarctica.
It is claimed that palaeomagnetic data are ofinsuf
ficient quality to disprove the proposed reconstruc
tion, and more rigorous testing will have to await
better data, should this become available, and an
improved geochronological database.
The process of fragmentation of the supercon-
tinent took only 200 Ma, implying rapid rates of
movement.It is possible that the process affected the
sudden changes to metazoan life experienced at the
beginning of the Phanerozoic, perhaps by the in
creased number of shelf seas surrounding the new
continents and climate changes associated with the
new tectonic regime.
12.2.5. Proterozoic tectonics
The recognition of a single supercontinent during
the Proterozoic places important constraints on the
types of tectonism which could have been active
during this time. It must be noted, however,that this
supercontinental assembly was probably not com
pletely rigid. The APW data are insensitive to rela
tively small movements and can accommodate
internal relative motions of up to about 1500 km
(McElhinny & McWilliams, 1977)and angular rota
tions of up to about 10®. Piper (1987)suggests that
the stable Archaean cratons were subdivided by mo
IMPLICATIONS OF PLATE TECTONICS 279
bile belts in which deformation was wholly ensialic
(Sutton & Watson, 1974) with no rock associations
that could be equated with ancient ocean basins.
Similarly, mineralization in the Proterozoic indicates
a lack of the exhalative volcanic processes related to
calc-alkaline volcanism which are so abundant in the
Phanerozoic. However, ophiolite complexes have
been recognized in the Proterozoic record, and these
Piper attributes to processes associated with subduc-
tion zones around the periphery of the supercon
tinent, a well-studied example being the Wopmay
Orogen at the northwestern margin of the Lauren-
tian Shield (Hoflfman, 1980). This model thus views
Proterozoic tectonism as occurring in two distinct
environments, interior mobile belts and marginal
subduction, which must have been of Andean-type
and similar to processes occurring in the Phanero
zoic. Indeed, Nakajima^f At/. (1990)have recognized
blueschists typical ofa subduction zone(Section 8.5)
in the Proterozoic of China.
An example of the type of ensialic orogeny envis
aged by this scenario is provided by the Labrador
Trough of northeast Quebec and Labrador, whose
geology is particularly well known. The Labrador
Trough is a linear belt of sedimentary and volcanic
rocks of Early Proterozoic age some 800 km long
and 100 km wide. To the west it is flanked by
Archaean rocks of the Superior Province which were
unaffected by the Hudsonian Orogeny (1800-
1600 Ma) which deformed and metamorphosed the
Labrador Trough. To the east lies the Archaean
terrain of the Churchill Province which was sub
jected to high grade metamorphism during the Hud
sonian Orogeny.
Dimroth (1981) has summarized the geological
history of the Labrador Trough as shown in
Fig. 12.9. Sedimentation was initiated by the deposi
tion of a continental red-bed sequence in a normal
fault-bounded elongate basin (Fig. 12.9a) which
was followed by slow subsidence and the deposition
of shelf sediments (Fig. 12.9b). There followed a
period of tectonic instability with faulting and ero
sion and the deposition of flanking conglomerates
(Fig. 12.9c). Subsidence then increased greatly with
the outpouring of voluminous basalts and the depo
sition of shale and greywacke (Fig. 12.9d). The re
gion then stabilized and a further cycle ofthe events
shown in Fig. 12.9b-d took place. The development
ofthe trough culminated in the Hudsonian Orogeny
during which 100-200 km of crustal shprtening
280 CHAPTER 12

Andesite and alkali-

iiiil Arkosic conglomerate
FTTyi Mass flow conglomerate of
' ^ dolomite and quartzite pebbles
MSii Arkose
I I Orthoquartzites,dolomites
IeeHeII Shales,greywackes
j^H Basalt
basalt
Thrustfault
Fig. 12.9 Evolution ofthe Labrador Trough:
S.L Sea level (a) continental fault basins;(b) deposition of
shallow marine sediments;(c) instability,
Archaean gneiss
faulting and deposition of high energy
Remetamorphosed sediments;(d) basaltic volcanic activity;(e) final
and deformed
Archaean gneiss stage of orogeny (redrawn from Dimroth,
1981, with permission from Elsevier).

occurred as the volcanosedimentary sequence was
folded and thrust on to the Superior block
(Fig. 12.9e).
Dimroth (1981) has demonstrated that the basalts
of the Labrador Trough are autochthonous and
formed by a process of vertical accretion and thus
differ markedly from ophiolite complexes. He has
suggested an intracratonic origin based on the crustal
delamination model of Bird (1979), shown in
Fig. 12.10. The trough was initiated when a crack in

the sub-crustal lithosphere developed and astheno-
spheric material rose and spread laterally at the crust-
mantle interface (Fig. 12.10a)(Section 2.10.3). The
flow of the asthenosphere was believed to be under
the influence of convection currents, which caused
lateral spreading of the asthenosphere to only one
side of the trough and dragged down the subcrustal
lithosphere. The low density asthenosphere invasion
caused uplift and the deposition of the continental
red beds. The uplift decayed after about 100 Ma and
shelf sediments were deposited. This phase was
followed by a massive intrusion of asthenosphere-
derived mafic and ultramafic magmas into the crust,
whose high density caused tectonic instability and
rapid subsidence as large quantities of basalts were
erupted on to the sea floor (Fig. 12.10b). Continued
sinking ofthe sub-crustal lithosphere, driven by con-
(a) Crust
Mantle Littiosphere
Asthenosphere ■
Mantle Lithosphere
Asthenosphere^
(c)
Mantle Lithosphere
Asthenosphere
Fig. 12.10 Intracratonic delamination model for the
formation of the Labrador Trough (redrawn from
Dimroth, 1981, with permission from Elsevier).
IMPLICATIONS OF PLATE TECTONICS 281
vection currents, then caused the final orogenic
phase (Fig. 12.10c), induced by weakening of the
crust by the intrusive activity or changes in the
velocity of the convection cells.
There are several points which should be consid
ered concerning this model. The main driving force
oftrough formation and deformation is proposed to
be convectional motion in the mantle. At the present
day, in order to explain observed heat flow values
(Section 11.4), it is believed that convection cells are
only present beneath the oceanic lithosphere. Given
that the radioactive content of the Proterozoic crust
would have been greater than today,the temperature
of the crust would have been higher and the contri
bution of additional heat from sub-lithospheric con
vection cells would have raised the temperature of
the lower crust even more. It is probable that this
increased heat flow would have led to anatexis ofthe
lower crust and the ascent of considerable quantities
of acidic magma. There is no geological evidence for
the presence of such material within the Labrador
Trough. The powering of trough evolution by man
tle convection implies that lithospheric motions were
driven by mantle drag forces. This appears unlikely
because, at the present time, mantle drag tends to
inhibit rather than enhance the motion of continen
tal plates (Section 11.7). Finally, the production of
magmas at the relatively shallow levels proposed for
the model requires that the melting temperature of
the mantle be considerably lowered in some way. In
modern subduction zones this achieved by the deri
vation of water from hydrous phases in the crustal
portion of the subducting lithosphere (Section 8.8).
Such a mechanism is not possible for a downgoing
slab which has been delaminated from its crustal
component, and no other mechanism is proposed.
Clearly, our understanding of mechanisms operative
in these so-called ensialic belts is incomplete. An
alternative mechanism is presented in Section 12.2.6.
The concept of a Proterozoic supercontinent, al
though accepted by many workers, is not without its
critics. For example, Dunlop (1981) accepts that
palaeomagnetic poles derived from cratons of indi
vidual continents tend to fall on a single APW
swathe, and thus that the mobile belts between these
cratons are the product of ensialic processes. He
suggests, however, that palaeopoles from different
continents do not follow the same APW path and
record large relative rotations and latitudinal shifts.
282 CHAPTER 12
For example, he argues that no match is possible of
the APW paths ofLaurentia and Africa for the period
2300-1750 Ma unless the former is more complex
or the latter less complex than indicated by available
data, and that, throughout most of the Proterozoic,
Laurentia, Africa and Australia drifted indepen
dently. Confirmation ofthis model or ofthe Protero
zoic supercontinent must await the accumulation of
high quality palaeomagnetic data in both space and
time. This would provide information on those era-
tons at present only poorly sampled and fill in the
gaps in the records of better sampled cratons by the
identification of suitable rocks of the desired ages.
The very major differences between the grouping of
the continents in the Late Precambrian proposed by
Morel & Irving (1978)(Fig. 12.1) and Piper(1987)
(Fig. 12.7) suggest that it is probably premature to
accept palaeogeographic reconstructions for the Pre
cambrian based on palaeomagnetic results.
12.2.6. Proterozoic plate tectonics
Given the uncertainties inherent in the palaeomag
netic investigation of Precambrian rocks, it is un
likely that it will ever be possible to reduce the width
of the APW swathe which represents a form of
confidence limit on the pole determinations. The
uncertainty means that the presence of narrow ocean
basins, of up to 1500 km width, between cratonic
areas cannot be discounted. Also, it must be remem
bered that palaeomagnetic methods give no control
on palaeolongitude (Section 3.6),so that linear inter-
cratonic regions which strike in the same direction as
the Eulerian pole used to bring the cratons into
juxtaposition are not constrained to have had any
particular width. The method of continental reas
sembly based on palaeomagnetic poles assumes that
these intracratonic regions would have the minimum
width possible when, in fact, they might have been
represented by wide ocean basins.
Palaeomagnetic data for the Precambrian thus can
not preclude the presence ofat least relatively narrow
ocean basins between the Archaean cratons. Conse
quently, rather than accepting that both intracra
tonic and plate marginal tectonic processes were
active in the Proterozoic, it is possible to postulate
that the mobile belts of this age were related to the
Wilsonian cycle of present day plate tectonics (Sec
tion 10.4), involving the opening and closure of
ocean basins up to 1500 km in width. Indeed, there
are many examples in the literature in which the
origin of Precambrian orogenic belts has been ex
plained in terms of modern plate tectonic processes.
Gibb et aL (1978; 1983) have proposed that the
structural province boundaries of the Canadian
Shield can be explained in terms of plate margin
interaction during Proterozoic times. Using a combi
nation of geological information and gravity inter
pretation, they argued that the structures marking
these suture zones, which are all characterized by a
remarkably similar gravity signature, were formed by
the continent-continent collision of the juxtaposed
provinces following the consumption of oceanic
lithosphere beneath the member of the pair whose
structure and radiometric age were more recently
overprinted at the time of the collision. Similar
gravity signatures have been described from crustal
province boundaries in the Precambrian of Australia
(Wellman, 1978; 1988), southern Africa (Reeves,
1976) and the Brazilian Shield (Lesquer et al.^
1981).
The nature of these province boundaries is exem
plified by studies of the Labrador Trough. Kearey
(1976) has proposed a model for the evolution of
this structure (Fig. 12.11) which differs markedly
from that of Dimroth (1981) described above. Be
fore the Hudsonian Orogeny the Superior and
Churchill provinces were separated by a small ocean
basin (Fig. 12.11a) whose initiation by rifting may
have created the environment of the early red-bed
sediments in the trough. Subsequently, a subduction
zone formed at the western margin of the Churchill
block (Fig. 12.11b). Shales and greywackes filled the
trench and forearc basin, while the eruption of volu
minous quantities of tholeiitic basalt, derived from
the upper mantle above the subducted lithosphere,
built up a volcanic edifice on the continental margin,
A briefcessation of movement would have permitted
shallowing of the forearc region and the deposition
of shelf sediments before subduction recommenced
with the deposition of further turbiditic sediments
and the further emplacement of basaltic rocks. Even
tually the ocean closed completely during the Hud
sonian Orogeny (Fig. 12.11c) and the Churchill
block and Labrador Trough were thrust over the
Superior craton. The Churchill block was then thick
ened and metamorphosed by the basement reactiva
tion mechanism of Dewey & Burke (1973)(Section
9.3.2). The orogeny would have ended as the sub
ducting oceanic lithosphere decoupled, and the sys-
 IMPLICATIONS OF PLATE TECTONICS 283

Superior Churchill

t^UTT] Continental
■.\-/".^M crust

Subcrustal
lithosphere
n;; Potassic
^ \ upper crust
V/ '/\ Refractory
Superior Churchill I/- 'A lower crust

\jrZ\ Migmatite

LT - Labrador Trough
S - Sediments
V - Basic igneous rocks
G-Granitic rocks
C - Conrad discontinuity
M-Moho

Superior Churchill PM-Partial melting

DP-Ductile flow
^ v\ \\
^\\
\\ \\ \\ V V V'N Present level of erosion
\\.\\a^.n^.xn \\ n\ ^-xnX'^ ^ -iy-y, V,?

iiiiiiiiiliiilllliii

Fig. 12.11 Plate tectonic model for the formation of the
Labrador Trough: (a) Late Archaean (pre-2560 Ma);
(b) Lower Aphebian (2560-1800 Ma); (c) Hudsonian
Orogeny (1800 Ma) (redrawn from Kearey, 1976, with
permission from Elsevier).
tem stabilized. Thomas & Kearey (1980) have
produced a further model for the Labrador Trough
which recognizes its similarity to an Andean-type
plate margin at an early stage in its development
(Fig. 12.12). Elsewhere in the Canadian Shield, Gibb
(1978) and Hoffman (1987) have recognized, from
the displacement of magnetic anomaly patterns, a
series of wrench faults associated with province
boundaries which bear a strong geometrical similar
ity to those apparent in the Himalaya which devel
oped in response to indentation tectonics (Section
9.3.6). If this parallel is correct, it implies that plate
tectonic forces remained active after Precambrian
collisions, and may have arisen from ridge-push
forces acting upon the opposite edge of the under-
thrusting plate.
There are many other regions where plate margin
processes appear to have been active in Proterozoic
times which do not necessarily coincide with the
margins of the Proterozoic supercontinent. Hoffman
(1980) has described a complete Wilsonian cycle
(Section 10.4) in early Proterozoic rocks in the
northwest Canadian Shield. Stauffer et al. (1975)
have described what they interpret to have been an
island arc sequence of andesitic lavas in the Amisk
Group of Manitoba. Krogh (1977) has presented
284 CHAPTER 12

□ Metasediments
m Basic igneous rocks
|+U[ Granitic rocks
□ Gneisses & migmatites
(a) Granulites
-200 n
Bouguer ^
anomaly ■
(g-u.) Zone of
Granite Dorset
-1200 compression
Site of Hudsonian belt Fold Belt
Trench
500^
Height (m) 0-

Main
20-
Depth (km)
40- 4tQ.30|

200 km
West Peruvian Trough Zone of
(b) Peru-Chile Coastal Batholith
East Peruvian compression
Present Trough
Trench
coastline
//
0

Depth (km) ^0 Shield

40 H

Basic intrusions

I+ + +I Coastal batholith

Fig. 12.12 (a) Gravity interpretation of a profile across interpretation of a crustal section across the central Andes
the Labrador Trough in which the Churchill Province is at 100 Ma (redrawn from Thomas & Kearey, 1980, with
made up of several fault-bounded blocks, each in isostatic permission from Nature 283, 61-3. Copyright © 1980
equilibrium. Density contrasts in Mgm"^, (b) Schematic Macmillian Magazines Ltd).

evidence for subduction and continent-continent
collision in the Proterozoic of western Norway.
Black et al. (1979) have proposed a cycle of ocean
opening and closure in the Late Precambrian of West
Africa and Abdelsalam & Stern (1993) a similar
Wilsonian cycle (Section 10.4) in the Proterozoic of
the Sudan. Watters (1976) has identified a possible
late Precambrian subduction zone in southwest Af
rica. Matthews (1972) has described an episode of
Precambrian obduction in southeastern Africa. Prot
erozoic subduction and ophiolite obduction have
been suggested for several regions of the Arabian
Shield (Leblanc, 1976; Al-Shanti & Mitchell, 1976;
Shanti & Roobol, 1979). It has also been proposed
that the Midcontinent Rift of North America repre
sents a continental rift developed in middle Protero
zoic times but which failed to develop significant
crustal extension (Chase & Gilmer, 1973; Green
1983). For further examples see Windley (1993).
It therefore appears that there is a large body of
evidence to suggest that true plate tectonic processes
were active in Proterozoic times, and that rather than
developing in ensialic environments, Proterozoic
mobile belts evolved in plate margin settings. The
major difference from Phanerozoic times was that
the ocean basins are palaeomagnetically constrained

to have been somewhat narrower than at the present
day. The imaging, by seismic reflection,ofa structure
similar to a suture zone in the 2700 Ma Superior
Province of the Canadian Shield (Calvert et
1995)suggests that modern-type plate tectonics may
also have been extant in the Archaean.
12.3 Plate tectonics and
economic geoiogy
12.3.1 Introduction
The application of plate tectonic theory has not yet
led to the discovery ofeconomically viable mineral or
hydrocarbon deposits. This is probably because it is
only relatively recently that attempts have been made
to relate the evolution ofeconomic deposits to plate-
controlled environments. Plate tectonics, however,
is important in this context as it provides exploration
geologists with a framework to which they can relate
the habitats and spatial relationships of economic
deposits and to trace their structural development
(Rona, 1977). Studies of this kind are expected to
become increasingly important in the future as
the search concentrates on smaller, more obscure
deposits.
Several publications in this field are available and
the following account will attempt to summarize the
most important findings. For more detail the reader
should refer to Mitchell & Garson (1976; 1981),
Rona (1977), Tarling (1981), Hutchinson (1983),
Sawkins(1984) and Evans (1987).
There are several types of economic deposit which
can be related to plate tectonic processes:
1 autochthonous deposits directly related to mag-
matism at plate margins and interiors;
2 allochthonous deposits related to plate margin
magmatism;
3 deposits related to sedimentary basins formed by
plate motions;
4 deposits related to climate and hence north-south
plate motions.
12.3.2 Autochthonous and
allochthonous deposits
The various plate tectonic environments in which
metalliferous deposits are found are shown in
Fig. 12.13. The initial rifting of a continent (Section
10.3) includes the emplacement of alkaline and per-
IMPLICATIONS OF PLATE TECTONICS 285
alkaline igneous rocks, and ore minerals are gener
ated from this magmatism. Carbonatites, which form
ring complexes within the alkaline rocks, are fre
quently associated with extensive mineralization.
The important elements found in this environment
are phosphorus (as apatite), niobium (pyrochlore),
rare earths(monazite, bastnaesite), copper, uranium,
thorium and zircon. Also found are magnetite, fiuo-
rite, barite, strontianite and vermiculite. Carbon
atites may also contribute to the sodium carbonate,
chloride and fluoride found in vast quantities in the
lakes of the East African rift system, although it is
possible that these derive from weathering of the
alkaline rocks. Directly related to the magmatism are
porphyry and vein-type molybdenum deposits associ
ated with subalkaline granites, copper-nickel depos
its associated with mafic intrusions and hydrothermal
copper deposits. Within the sediments related to
rifting, stratiform copper deposits of great volume
are associated with specific shale or sandstone hori
zons. These disseminated ores are believed to form
during the first marine transgression into the conti
nental interior and overlie red-bed horizons, and are
probably derived from the copper-rich rift basalts in
response to the elevated heat flow ofthe rift. Carbon
ate hosted lead-zinc-barite ores are also found in the
intracratonic rifts and passive continental margins,
typified by the deposits of the Upper Mississippi
Valley of North America.
With the development of a narrow ocean basin
between the rifted continental fragments, new min
eral deposits are created at the mid-ocean ridge. The
present day example of this environment is the Red
Sea. Here 13 pools of hot brines (Fig. 12.14) have
been located along the central ridge where it is
intersected by transform faults. These contain zinc-
copper-lead sediments of possible economic value,
for example the Atlantis II Deep, which contains
sulphide layers with zinc contents of up to 20%
which are 20 m thick and cover an area of over
50 km^. It is generally agreed that the metals are of
volcanic origin and have been concentrated into
brines by the thermally-induced circulation of sea
water through the volcanic rocks and thick evaporite
sequences found in this region (Cowan & Cann,
1988). As the ocean basin evolves, these deposits
may become buried by sediment and reappear in
collisional orogens where the tectonism obscures
their original setting. Also associated with this ad
vanced phase of rifting are sediment-hosted massive
 N)
00
O)

o
X
>
-0

(a) Intracontinental hotspot Sn, F, Nb (b) Aborted rift zone Carbonatite Nb, PoO-,,
REO, Ba, F
?Sullivan-type Pb-Zn-Ag j
Sn, F. Nb
}'\ iTv--1'I// J
-'r/x> ''r '"/»' V^

(c) Intercontinental rift zone Evaporitic layer (d) Ocean rise and Hawaii-type chain
. LIL-rich Potassic
Red Sea Mn '^^gnd'^m^ ds Mississippi-type LIL-depleted
I II ^ I . .basalt basalt basalt
^ ^' l\ Pb-Zn-Ba ?Fe, Mn,As,B,Zn,Pb,Hg seds
Cyprus-type Cu-F^ '

Alkali-Depleted

?i*A*i*i%*»*i'PodIfnrrri Cr Ni^

(e) Andean-type magmatic belt (f) Island arc and inter-arc basin
Mn nodules with Porphyry Sn, W , ^ Incipient Inactive Cyprus-type
Cu, Nl, Co Cu, Mo, Hg / f Kuroko Zn, Cu ^ riff inter-arc Cu-Fe
basin
I ^ 1.,^,V
(g) Magmatic and outer arc with marginal basin
D ~ Ophiolite Sn, W, Bi, Mo, F
Cyprus-type Cu-Fe~,^°"''°''7^
I ■ or mantle diapir C,p™.-W, J s.
Porphyry
^u-Au —7*^ .

SOj^Hg in serpentinite k*i&'.ml«M^^«Podiform CrgI

^Besslii-type Cu-Fe' S? NiS, Pt

^;vtVTW«

(h) Continent-continent collision belt Maamatic arc (i) Post-collision volcanism

Sn,W,F >Ni,Co minlalization .. Post-collision
molasse A / Jadeite Or,Pt,Cu-Fe ■^^Q'T^^t'carc mm. magma in rift zone
N ephrite
3k. ^ \J. .

• w-^-n r-ty t-rn A •> >w>>>w5Kww

Ffi-T i-wrAA?
S:60 Ikm iinHorthriiefJnr,
underthrusting Fe-Ti
^SS!5&s\ t*tVi*iVi*tViVi*i%*iV

(j) Continental collision with obducted ophiolites h ^ I Tonalite-granodiorite Oceanic crust

o
Podiform Cr, Pt ^ Cyprus-type Cu-Fe >
1+ +1 Granite-alkaline granite Upper mantle
nn Undersaturated I * I Anorthosite
' ' alkaline rocks
O)
I c I Carbonatite mi l I Limestone O
TI
Volcanic PPa o j. * flysch/ "D
1 1 rocks |gs2U Sediments: molasse I—

I II Large ionic lithophile >

Fv\^7^J Continental crust H
trace elements m
:! 60 km v.v.v.%v.v.%%%wAw.vw.fw.T^w.v.%%%v.v.%wrt : .V.....V.W.V.-.. • v...v.v.w..« 00 High temperature PP High pressure H
•X'XOX'X'X'X^^^^^ Z U U K m X*X«X«X*X«>>X»X»X;>X;>X»^^^^^ m
metamorphics metamorphics
O
H
Fig. 12.13 Schematic cross-sections through plate boundary-related tectonic settings of mineralization (redrawn from Mitchell & O
z
Garson, 1976). o
w

ts)
00
vj
288 CHAPTER 12

• Brine pool
A Metalliferous sediments

Oceanographer

Kebnt deep
Gypsum deep
Vema deep

Nereus deep
Thetis deep
Atlantis II deep
Valdivia deep Albatross deep
Discovery deep Erba deep
Sudan deep
Shagara deep
Suakin deep

;P

Saudi
Arabia

Fig. 12.14 Locations of hot brine

Sudan/
pools and metalliferous sediments in
I- •t-i.t. Inrliany
the Red Sea (redrawn from Bignell
etal., 1976, with permission from
1000 km
^^^12 the Geological Association of
Canada).

sulphide deposits which occur in thick continentaiiy-
derived clastic sediments on passive continental mar
gins. They comprise single or multiple lenses of
pyrite, galena and sphalerite ores with minor silver
and copper. These are not common and probably
reflect the influence of metal-rich formation waters
powered by long-lived geothermal systems
As an ocean basin continues to grow,contempora
neous mineralization takes place at the mid-ocean
ridge, and has been observed at certain locations
along the Pacific (Corliss et al.^ 1979), Atlantic
(Scott et 1973) and Indian Ocean ridges. The
mineralization is of hydrothermal origin and its loca
tion depends upon the availability ofoceanic crust of
high permeability overlying the magma chamber
which allows fluids to percolate with relative ease.
Hydrothermal processes of low intensity lead to the
formation offerromanganese nodules, and encrusta
tions of iron and manganese on pillow basalts at the
layer 1-layer 2 interface. Higher intensity hydrother
mal activity has been observed at some locations,
such as on the East Pacific Rise where discharge is of
two types. Black smokers are vents where pyrrhotite
particles are discharged, producing ores which may
be zinc- or iron-rich and containing lesser amounts of
cobalt, lead, silver and cadmium. At white smokers
little sulphide material is discharged, and the main
precipitate is barite.
In open ocean conditions ferromanganese nodules
and encrustations form on top of basalt or sediments
 IMPLICATIONS OF PLATE TECTONICS 289

where strong ocean currents prevent the accumula
tion of clastic sediments. These deposits are of
hydrous origin and accumulate slowly, sometimes
forming extensive pavements. As well as iron and
manganese, they also contain smaller amounts of
copper, nickel and cobalt.
In addition to the exhalative processes of mineral
ization described above, ore bodies may form within
the oceanic lithosphere as it is created. Much of our
knowledge of these deposits is derived from studies
of ophiolites (Section 2.5), which are interpreted as
allochthonous slices of oceanic or back-arc basin
lithosphere tectonically emplaced within continental
crust during collisional orogenesis. One of the most
intensively studied bodies ofthis type is the Troodos
complex of Cyprus(Fig. 12.15), and other examples
include ophiolites of northeastern Newfoundland,
the Semail Ophiolite of Oman and Ergani Maden in
Turkey. At high levels in the lithosphere, massive
sulphide deposits (marcasite, chalcopyrite and
sphalerite) occur on top ofor within the pillow lavas
oflayer 2. It has been suggested that these sulphides
formed either in a manner similar to the brines ofthe
Red Sea or by precipitation from hydrothermal solu
tions which became enriched in metals by circulating
within the volcanic rocks. The deeper, plutonic por
tions of ophiolites contain economic deposits of
chromite, which occur as podiform bodies and tabu
lar masses within the harzburgites and dunites ofthe
upper lithospheric mantle. These deposits may have
formed by partial melting of primitive mantle ma
terial or by crystal fractionation within the magma
chamber underlying ocean ridges (Section 6.10).
Similarly associated with the magma chamber are
nickel and platinum sulphides. The model for miner
alization of the oceanic lithosphere derived from
ophiolite studies is shown in Fig. 12.16 (Rona,
1984).
Several forms of mineralization are present in
subduction zone environments, their types depend
ing upon whether the overriding lithosphere is con
tinental or oceanic. Hedenquist & Lowenstern
(1994) have reviewed the role of magmas in the
formation of hydrothermal ores in such environ
ments. The most important mineralizations are the
porphyry coppers. These are very large, low grade
deposits which are gold-rich and molybdenum-poor
when associated with island arcs and gold-poor and
molybdenum-rich in Andean-type mountain ranges,
being found, for example, in the Andes themselves,
the Philippines, Taiwan, Puerto Rico and the Ryuku
and Burma arcs. The controls on the location of
porphyry coppers are poorly understood, but they
appear to be related to the primary calc-alkaline
magmatism and do not require the presence of any
specific crustal type. Solomon(1990)has noted that.

Umber
o
(/) Upper pillow lavas A AA
T3 Massive Fe-Cu sulphides
O Lower pillow lavas
2'Q.i5
Basal group
0

-D 0
^ X
"S Q- Diabase
B E
0 o
0 o 1 km
CO

- Plaglogranlte
Gabbro +
[ 0

Q.
E Layered gabbro
8 Pyroxenltes -
O
Fig. 12.15 Mineralization in the 'c Wehrllte Cu-Ni-Co-Fe sulphides
o
Troodos Ophiolite (redrawn from 3 □unite £- Podiform chromite
Searie & Panayiotou, 1980, with Q.

permission from the Ministry of

Agriculture and Natural Resources,
Cyprus).
290 CHAPTER 12
Sediment
Metalliferous sediment ^
(Cu,Fe, Mn,Pb,Zn,Ba,Co,Ag,Au)
Stratiform oxides, hydroxides,
silicates(Fe, Mn) >
Massive,disseminated,and
stockwork sulphides(Cu,Fe,Zn,Ag,Au)
Dykes
Gabbro and serpentinite
Cumulates —
Chromite-
Lherzolite and harzburgite —
Nickel and platinum sulphides -
Fig. 12.16 Schematic block diagram showing the
potential distribution of mineral deposits in the oceanic
lithosphere (redrawn from Rona, 1984, with permission
from Elsevier).
in the southwestern Pacific rim, porphyry copper-
gold deposits mostly form after a reversal of arc
polarity following a collisional event. Sometimes
associated with porphyry coppers are mercury depos
its (as cinnabar or quicksilver), which may have
originated in a similar manner.
Another important class of deposits found associ
ated with oceanic subduction zones (Fig. 12.17) is
stratiform massive sulphides ofzinc,lead and copper
known,after their type area ofoccurrence in Japan,as
Kuroko-type ores. They were deposited in a shallow
marine environment and occur interbedded with
pyroclastics and silicic calc-alkaline lavas. It is be
lieved that they are ofsyngenetic origin, and occur at
a late stage of arc evolution. Halbach et al. (1989)
suggest that they formed in a back-arc basin, and cite
the Okinawa Trough as a modern analogue. They
may have been deposited by saline submarine hot
springs arising from the separation of aqueous ore
fluids during the final stages of magmatic fraction-
ation or from the leaching of older volcanic rocks.
Kuroko-type ores may be incorporated into conti
nents during continent-island arc collisions, such as
at Rio Tinto in Spain, Umm Samiuki in Egypt and
the Buchans mine in Newfoundland.
There also exist other forms of stratiform massive
sulphides which differ in their depositional environ
Layer 1:
sediment
Layer 2:
mafic rocks
_ E
(basalt)
Layer 3:
mafic and
ultramafic rocks
-5
Upper mantle:
ultramafic rocks
ment from Cyprus- or Kuroko-type. They are associ
ated with intermediate to basic volcanic rocks with
carbonaceous mudstones, clastic limestones or
quartzites, all of which suggest a deep water environ
ment unlike ocean ridges, ocean basins or island arcs.
They have been termed Besshi-type deposits. They
may have formed in a trench or a tensional environ
ment, but their origin remains, as yet, obscure.
There are several types ofdeposit which are specific
to Andean-type subduction. These include strata-
bound copper sulphide deposits,such as are found in
Chile, which are closely related to episodic calc-
alkaline volcanism and occur within porphyritic
andesite lavas. The principal minerals are chalcosite,
bornite and chalcopyrite, and they contain signifi
cant amounts of silver. The intercalation of these
deposits with shallow marine and terrestrial deposits
suggests their formation in small lagoons. Tin and
tungsten mineralization occurs in the eastern Andes
of Peru and Bolivia on the landward side of the
porphyry copper belt. It appears to be derived from
the same Benioff zone region as the magmas, and
may owe its existence to the anomalously shallow
dip of the subduction zone in this region (Section
9.2.3).
In the back-arc environment of Andean-type sub
duction zones, such as in east and southeast Asia,
there occur granite belts with which are associated tin
and tungsten with lesser molybdenum, bismuth and
fluorite. The origin ofthe tin, in particular, is contro
versial. Tin is present in only minute quantities in the
oceanic crust, and is similarly absent in island arcs.
 IMPLICATIONS OF PLATE TECTONICS 291

Outer arc

-Podiform chromite Position of possible

spreading axis with
Melange with
oceanic sulphide - Besshi-type Cyprus-type deposits
deposit massive sulphides

Oceanic sulphide Porphyry copper ore

deposits
[Cyprus-type] Kuroko-type
Plutons with
being transported ■ ■ massive
Sn, W, Bi veins
in oceanic crust sulphides

Batholiths
7^^^~^^'Continentai
Oceanic crusT™®^ "beanie crust'
Descending lithospheric
plate
Ascending
calc-alkaline
magma

Asthenosphere

Fig. 12.17 Development and emplacement of mineral
deposits in a subduction-related setting (redrawn from
Evans, 1987, using data from Sillitoe, 1972a; 1972b, with
permission from the Economic Geology Publishing Co.
and the Institute of Mining and Metallurgy).
An oceanic origin of the tin appears unlikely. One
hypothesis is that tin is derived from deep in a Benioff
zone which is migrating away from a continent
during back-arc spreading. The fluorine originating
at these levels would extract tin from deep levels of
still hot granite plutons and deposit it at the surface
in their vicinity. Another hypothesis is that the gen
eration oftin requires the presence ofthick continen
tal crust, such as is present in the tin belts of the
major episode offaulting. They probably originate in
magmatically associated brines.
In the marginal seas of oceanic subduction zones
(Section 8.9) the crust is similar to oceanic, although
generated in a different fashion, and so mineral
deposits would not be expected to differ greatly from
those in oceanic crust. The mineralization would be
similar to that formed during the early development
of a spreading ridge, and would thus be related to
magmatic and exhalative volcanic processes. If the
greenstone belts of Precambrian terrains are the an
cient analogues of back-arc basins (Section 12.2.3),
then the copper-zine sulphides and gold deposits
found in them must have originated in this environ
ment.

Andes, Alaska and upper Burma, and a shallow
dipping Benioff zone which acts as a source of heat
and volatiles. In this case the tin may originate from
pre-existing concentrations in the lower continental
crust.
Within ensialic back-arc basins vein-type gold and
silver deposits are common,such as are found in the
Great Basin of Nevada. These are associated with
andesites, dacites and rhyolites and pre-date the
Collisional environments may contain allochtho-
nous members which contain deposits typical of the
earlier stages of crustal accretion, such as ophiolites,
ferromanganese nodules, subduction-related de
posits and mineralization related to early stages of
rifting. Additionally, deposits which originate in
continental collision are also present. Commonly
granite bodies are emplaced during and after the
collisional event, such as in the Variscan belt of
292 CHAPTER 12
northwest Europe. Associated with these granites are
tin-tungsten deposits of cassiterite and wolframite.
It is believed that the mineralization, like the gran
ites, is derived from anatexis ofthe lower continental
crust, either by magmatic concentration or the melt
ing ofexisting mineral accumulation. Also associated
with these granites are vein-type deposits ofuranium.
There are no economic deposits associated with gran
ites in the Himalaya,all known examples being older.
Some models of the evolution of collisional oro-
gens require the formation of an anorthositic lower
crust after the upward migration of potassium and
the lithophile elements(Section 9.3.2). The majority
of massive anorthosites are of Precambrian age and
so this hypothesis cannot be tested by modern ana
logues. Associated with anorthosite bodies are iron-
titanium deposits of magnetite and ilmenite, which
probably originated in the same partial melting
episode.
Oceanic transform faults are favourable environ
ments for mineralization because they may be associ
ated with high heat flow and provide highly fractured
and permeable conduits for both the downward
percolation ofsea water and the upward migration of
mineralizing fluids. Iron sulphide concretions have
been reported from the Romanche Fracture Zone of
the equatorial Atlantic which may have originated by
this mechanism. The brine pools of the Red Sea
appear to be located where transform faults intersect
the central ridge, and it is possible that the metals
ascend along these faults. Indeed, base metal depos
its are found along the continental continuation of
the faults. A similar mechanism has been postulated
for the brines of the Salton Sea, California. It is
probable that the ultramafic intrusions occurring in
fracture zones (Section 7.3) contain high propor
tions of nickel, cobalt and copper.
Very little mineralization appears to be associated
with continental transform faults at the present level
oferosion.
Precambrian mineral deposits can be widely inter
preted as forming in plate tectonic environments,
particularly those related to divergent plate margins
and subduction zones (Gaal & Schulz, 1992).
12.3.3 Deposits of sedimentary basins
The majority of fossil fuels are found within sedi
mentary basins whose formation can be related di
rectly or indirectly to plate motions. In addition to
the sedimentary environment,quite stringent condi
tions are necessary for the development and preserva
tion of these resources.
There are four principal criteria which must be met
for the development of petroleum and gas, herein
after referred to as hydrocarbons: layers rich in or
ganic matter within the sedimentary succession; a
source of heat applied for a time sufficient for the
maturation of organic materials into hydrocarbons;
permeable pathways which allow movement of the
hydrocarbons; and a porous reservoir whose top is
sealed by impermeable capping beds.
The main source ofthe disseminated organic mat
ter, or kerogens,in sediments is plankton. The abun
dance of plankton is controlled by climate, the
quantity of nutrients available and water body geo
metry. The first two factors are latitude dependent,
and the majority of oil basins originate at low lati
tudes. The latitude is obviously affected by the
north-south component of plate motion, while the
plate configuration at any given time determines
water body geometry. Organic material is especially
abundant along continental margins where there is
major river runoff into large deltas.
The preservation of kerogens requires conditions
which are not oxidizing. These are achieved along
continental slopes where the production of organic
matter exceeds the availability of free oxygen to
convert them to carbon dioxide, and in closed anoxic
basins. It follows that the shales and mudstones
produced in such environments are the most com
mon source rocks as they have the ability to absorb
kerogens and remove them from the effects of free
oxygen.
The temperature experienced by the kerogens after
burial is critical, and depends on the local geothermal
gradient. Temperatures of 70-85 are required to
develop liquids and 150-175 ®C for dry gas. It is
also important that a critical exposure time to these
temperatures is exceeded, so the basin must be free
from tectonism and uplift during this period.
After formation, the hydrocarbons undergo pri
mary migration from the fine-grained source rocks
and secondary migration as they concentrate and
accumulate in a reservoir of high porosity. Migration
occurs because ofthe buoyancy ofthe hydrocarbons,
and it follows that all hydrocarbon accumulations are
allochthonous. There are several types of oil trap,
including anticline, fault, stratigraphic, unconfor
mity and lithological traps, which have the effect of
providing a capping to the reservoir with an imperme
able cover which prevents further upward movement.

Trade Winds
Ocean current
Major oil basin
Mid-cretaceous
Fig. 12.18 Possible relation of oil basins to the Tethys-
Atlantic seaway in middle Cretaceous times (redrawn from
Irving et al., 1974, with permission from the National
Research Council of Canada).
Plate tectonics contrors the locations of reservoirs
in that it is responsible for the formation and preser
vation of the sedimentary basins in which hydro
carbons are generated and trapped. These include:
1 intracratonic basins formed by hotspot activity,
Paris and Michigan basins (Section 10.5.2);
2 basins associated with continental rifting, e.g.
the Gulf of Suez, Red Sea;
3 aulacogens (Section 10.2), e.g., the North Sea;
4 passive continental margin basins, e.g., the Ga
bon Basin;
5 ensialic back-arc basins, e.g. the Oriente Basin of
Ecuador and Peru;
6 marginal seas, e.g., the Andaman Sea;
7 accretionary prisms, e.g., the coastal oil-fields of
Ecuador and Peru;
8 forearc basins, e.g., the Cook Inlet of southern
Alaska;
9 pull-apart basins associated with wrench faults
(Section 7.4.3), e.g. the Los Angeles Basin, western
USA (Moody, 1973);
10 foreland basins (Section 10.5.3) of Himalayan-
IMPLICATIONS OF PLATE TECTONICS 293
mmmm
type mountain ranges, e.g., the Aquitaine Basin,
southwest France,
11 tensional basins associated with indentation tec
tonics (impactogens), e.g. the Rhine Graben.
Not only can plate tectonics create the habitat of
hydrocarbon deposits, it can also explain why certain
regions are particularly rich in these deposits. A large
proportion of the Earth's hydrocarbon reserves are
located in the Middle East, and the evolution and
preservation of these deposits has been a conse
quence of a specific pattern of plate interactions
(Irving 1974).
During Mesozoic and Early Cenozoic times two
large embayments existed on the continental shelf of
the Afro-Arabian continent on the southern side of
the Tethys Ocean (Fig. 12.18). Such embayments
around the Tethys, which also included the Gulf of
Mexico and the Persian Gulf, may have been con
nected via the proto-Mediterranean Sea, or Tethys-
Atlantic seaway, which was situated at low latitudes.
At about 100 Ma the rate ofspreading of the seaway
increased, maximizing the development of hydro
carbon source rocks because of the formation of
extensive, warm,shallow seas to which were supplied
large quantities of nutrients from the spreading
centre. When the Seaway subsequently began to
close following the development of a subduction
zone at its north margin, the geometry of the plate
294 CHAPTER 12
movements was such as to protect the Persian Gulf
from major tectonism. This arose because the rapid
northerly motion of the Indian Plate absorbed most
of the energy associated with the collision with the
Eurasian Plate. The Gulf of Mexico was similarly
protected by northeastward motion of the Greater
Antilles.
Coal is a combustible sedimentary rock containing
in excess of50% by weight ofcarbonaceous material.
It is formed by the decomposition, compaction and
diagenesis ofan accumulation ofterrestrial and fresh
water plant debris. Coals thus appear in the geologi
cal record from Devonian times when the first plants
appeared.
In order to prevent the total destruction of the
vegetable matter by biochemical decomposition,
very wet conditions are required to stop the decay by
the accumulation of toxic waste products. The con
ditions under which this process occurs are con
trolled by climate and topography. Normally a warm,
wet climate is required to promote luxuriant growth,
and this should be under the condition of constant
standing water. Although,in regions of high rainfall,
peat forms in upland areas, it is rarely preserved due
to the erosion experienced in this environment. The
prime conditions for coal formation are those offlat,
low-lying ground invaded by swamps with stagnant
water. The slow sinking ofthese regions preserves the
organic layers by progressive burial.
The process of coalification refers to the physical
and chemical changes experienced by the organic
matter after burial in response to rising temperature
and pressure. On compression, water and volatiles
are expelled and the deposit becomes enriched in
carbon. The degree ofcoalification is reflected in coal
of different ranks, varying from the low rank lignite
to high rank anthracite.
Plate tectonics affects coal formation in that it
controls the latitude of a region, since most coal
deposits originated at low latitudes. Plate tectonics
also creates the environments necessary for the pre
servation of organic matter, of which the most im
portant are passive continental margins. Deltas
formed on such margins produce the most favour
able conditions for coal formation, and swamps can
develop on a regional scale. Present day examples
include the Niger, Amazon and Mississippi deltas,
and ancient examples the Carboniferous coals of
North America and northwest Europe. Intracratonic
deltas,such as the Rhine,are similarly productive and
are likely to be preserved due to their stable sur
roundings. Coal deposits are also found in aulaco-
gens and ensialic back-arc basins. The tectonism
associated with collisional orogens provides an envi
ronment whereby coals increase in grade by high
pressure metamorphism.
12.3.4 Deposits related to climate
It has already been stated that the formation of
hydrocarbon and coal deposits is dependent upon
both climate and special conditions of sedimenta
tion. There are certain deposits, however, which
appear to be related solely to climate. Since climate is
largely dependent upon latitude, north-south plate
motion can be considered as controlling the forma
tion of such deposits. They include laterites and
evaporites.
The most important lateritic deposit is nickel later-
ite, which results from the extreme weathering ofthe
ultramafic parts of ophiolite bodies under tropical
conditions. The original nickel content offresh peri-
dotite becomes enriched by a factor of about seven
under the influence of such weathering by percolat
ing ground water. These deposits are becoming in
creasingly important sources of nickel, and are
exploited in the southwestern Pacific and the north
ern Caribbean.
A similar deposit is bauxite, a residual deposit
enriched in aluminium hydroxide, which provides
the vast majority of the world's aluminium. This
forms by the in situ weathering of aluminosilicate
minerals on stable peneplaned topography in a wet
tropical climate by the intense leaching of alkalis and
silica. Bauxite only forms within 30® of the equator
and requires high rainfall and high ambient tempera
tures. It is mined in Jamaica, northern Australia and
China.
Evaporites form in an arid climate by the evapora
tion ofsea water in semi-isolated basins which receive
periodic marine influxes. They cannot develop by the
evaporation ofa single isolated body of water, as this
could not explain the vast observed thicknesses of
evaporite deposits. The sequence of minerals pre
cipitated is calcium carbonate and sulphate, sodium
chloride and finally magnesium and potassium min
erals. Evaporites are important commercially in the
chemical industry, particularly for the potash salts.
They are also important in the generation of hydro
carbon traps because, being of low density, they

mobilize after burial and rise through the sedimen
tary layers. Such halokinesis provides fault traps
along the sides of rising salt masses and anticlinal
traps in the layers above the masses which are folded
during the ascent. This is an important process in the
North Sea, for example, which is underlain by salt
deposits ofPermian age.
12.3.5 Geothermal power
Geothermal energy can be effectively utilized for
power generation when the vertical thermal gradient
is several times its mean value of about 25®Ckm~^,
producing near-surface temperatures above 180®C.
This condition is achieved atconstructive and destruc
tive plate margins, as exemplified by the geothermal
power plants on Iceland and on Guadeloupe, West
Indies, respectively. Anomalously high geothermal
gradients are also present in intraplate areas where
they are frequently associated with granitic plutons.
The normal geothermal gradient can be utilized for
lower energy power generation, such as for space
heating, wherever a thick pile of permeable sedi
ments allows the circulation of fluids to depths of
several kilometres. An example of this type is the
Paris Basin, where space heating for over 20 000
dwellings is provided by deep fluid circulation.
IMPLICATIONS OF PLATE TECTONICS 295
Further reading
Bickie, M.J.(1978) Heat loss from the Earth: a constraint
on Archaean tectonics from the relation between geo
thermal gradients and the rate of plate production.Earth
planet. Sci. Lett. 40,301-15.
Evans, A.M.(1987) An Introduction to Ore Geology (2nd
edn). Blackwell Scientific Publications, Oxford.
Irving, E., North, F. & Couillard, R.(1974) Oil, climate
and tectonics. Can.J. Earth Sci. 11,1-I7.
Kroner, A.(1981) Precambrian plate tectonics. In KrOner,
A.(ed.)Precambrian Plate Tectonics., pp. 57-90. Elsevier,
New York.
Mitchell, A.H.G. & Garson, M.S.(1981)Mineral Deposits
and Global Tectonic Settings. Academic Press, London.
Piper, J.D.A.(1987)Palaeomagnetism and the Continental
Crust. Open University Press, Milton Keynes.
Smith, A.G., Hurley, A.H.& Briden, J.C.(1981)Phanero-
zoic Palaeocontinental World Maps. Cambridge University
Press, Cambridge.
Tarling, D.H.(ed.)(1981)Economic Geology and Geotecton-
ics. Blackwell Scientific Publications, Oxford.
Tarney, J., Dalziel, I. & De Wit, M.(1976) Marginal basin
'Rocas Verdes' complex from S. Chile: a model for
Archaean greenstone belt formation. In Windley, B.F.
(ed.) The Early History of the Earth., pp. 131-46. Wiley,
New York.
Windley, B.F.(1981)Precambrian rocks in the light ofthe
plate-tectonic concept. In KrOner, A.(ed.) Precambrian
Plate Tectonics., pp. 1-20. Elsevier, New York.
Windley, B.F. (1984) The Evolving Continents (2nd edn).
Wiley, New York.
Windley, B.F. (1993) Uniformitarianism today: plate tec
tonics is the key to the past./.geol. Soc. Land. 150,7-19.
Review questions
1 Summarize the subdivisions of the Earth's sili
cate mantle and metallic core which have been
deduced from seismological studies. Indicate
briefly the type ofseismological study or observa
tion which provides the evidence for each layer
that you mention.
Suggest a plausible temperature distribution
within the Earth that is consistent with the phys
ical state and theological properties of these
different layers.
2 Describe the way in which the seismic 'first mo
tion' at a number of seismological observatories
across the world can be used to obtain a focal
mechanism solution for an earthquake of suffi
ciently large magnitude.
Explain how, for near-surface earthquakes,
such solutions can be related to a simple theory
of faulting and hence to the principal stress
directions in the vicinity of the earthquake and
the sense of motion on the fracture.
3 Review critically the various lines of geophysical
evidence suggesting that there is a layer of
low strength (the asthenosphere) in the upper
mantle. How,ifat all, does the depth ofthis layer
vary beneath different geological provinces?
4 In the late 1960s four aspects of seismology
provided the basis for the formulation of the
concept of plate tectonics:
(a) more precise epicentral and focal depth deter
minations;
(b)the study of surface waves and free oscilla
tions;
(c) attenuation studies; and
(d)the determination of focal mechanism solu
tions.
Explain briefly the nature and significance of
each of these contributions. Hence show how a
nearly complete statement of plate tectonics can
be derived from seismology alone.
5 Explain the underlying assumptions ofthe palaeo-
magnetic method for determining the palaeo-
latitude and orientation of a continental area.
Outline the various stages involved in carrying
out such a palaeomagnetic study.
Describe two ways in which studies based on
the remanent magnetization of rocks have con
tributed to the verification of continental drift
and the formulation of plate tectonics.
6 (a) Define the term transformfault and illustrate
the six theoretical possibilities for active,
right-lateral transform faults which do not
terminate in triple junctions. Indicate the
way in which the geometry of each type
evolves with time and, where possible, cite a
recent or present day example.
(b)Explain the significance of transform faults
when deducing the relative motion between
adjacent lithospheric plates.
(c) Explain the origin and nature of pull-apart
basins and strike-slip/sheared continental
margins.
7 Explain the principles upon which the direction
and rate of relative motion between lithospheric
plates are deduced for the different types of plate
boundary developed at the Earth's surface. Re
view briefly the types of argument used to con
vert these 'relative' motions into 'absolute' plate
motions with respect to the Earth's deep interior.
To what extent do the resulting models for
absolute plate motions differ?
8 Briefly outline the basic assumptions involved in
making a palaeomagnetic study.
Explain qualitatively how the global palaeo
magnetic dataset for rocks formed during the
past 200 Ma may be analysed to test:
(a) the geocentric dipole model for the Earth's
magnetic field;
(b)the hotspot model for absolute plate mo
tions; and
(c) the suggestion that much of the Western
Cordillera of North America consists of'dis
placed' or 'suspect' terrains.
297
298 REVIEW QUESTIONS
In each case take care to state any assumptions
and limitations of the method.
9 Explain the following plate tectonic concepts,
illustrating your answer with specific examples:
(a) failed arms or rifts;
(b)triple junctions;
(c) back-arc spreading; and
(d)Benioff plane reversal.
10 Explain current ideas and models for the evolu
tion of passive continental margins. Pay particu
lar attention to those aspects of the models that
help to explain the observed diversity of such
margins at the present day.
11 Cite the empirical, i.e. geological and geophysi
cal, evidence for brittle fracture in the outer
layers of the Earth's crust. Explain the way in
which the nature and geometry of the fractures
reflect the state of stress in these layers at the
present day or in the past.
Focal mechanism solutions for certain earth
quakes can be used to give information on the
state of stress at the present day. Outline the
principles and the limitations of this method for
determining the principal stress directions.
12 Outline the McKenzie (1978) model for the
development ofsedimentary basins.
Describe some of the ways in which this
model has been modified subsequently to ac
count for discrepancies between the observed
subsidence history and thermal evolution of
sedimentary basins, and the predictions of the
simple model.
Indicate briefly the potential importance of
such variations in relation to the formation of
oil and/or gas in a particular sedimentary basin.
13 Explain how earthquake focal mechanism solu
tions can give an indication of the stress field in
the vicinity of the earthquake focus. Cite exam
ples of focal mechanism solutions and inferred
maximum and minimum principal compressive
stress directions for earthquakes occurring:
(a) at plate boundaries; and
(b)within plates.
14 Extension ofthe Earth's crust may be accommo
dated in a number of different ways. Write a
short essay outlining the different possibilities,
distinguishing between brittle and ductile re
sponses, and pointing out any thermal and/or
isostatic consequences of different models.
15 Explain the principle of isostasy and illustrate
your answer by outlining the calculations in
volved in the simple predictive models of:
(a) the maximum sedimentary loading of oce
anic crust;
(b)the subsidence history of extensional basins;
and
(c) the evolution of passive continental margins.
16 Formulate a possible 4ife cycle' for an ocean
basin within the framework of continental drift,
sea floor spreading and plate tectonics. Give
present-day examples to illustrate various stages
in your proposed cycle.
17 Compare and contrast island arcs and Andean-
type mountain ranges.
18 Discuss the statement by Wyllie that 'Plate tec
tonics has given us a framework of tectonic
environments within which various suites of
igneous rocks can be located'.
19 Continental rift zones are thought to form in a
number of different ways. Explain the different
models for the formation of such rifts, paying
particular attention to similarities and differences
regarding crustal structure and geological his
tory.
20 Some geologists and geophysicists claim that
studies of the structural setting, internal struc
ture and composition of ophiolites are highly
relevant to a better understanding of the pro
cesses occurring in subduction zones and at
ocean ridge crests. Review critically the basis and
implications of these claims.
21 Compare and contrast continental and ocean
lithosphere.
22 Explain why studies ofisostasy are important in
considering the Earth's dynamic behaviour.
23 Write an essay on the evolution of the theory of
plate tectonics.
24 Describe the data and techniques used in making
continental reconstructions.
25 In what ways have studies ofterrestrial heat flow
contributed to plate tectonic theory?
26 Explain the strengths and limitations of palaeo-
climatological techniques for determining past
plate motions and geographies.
27 Describe the evidence for plate tectonic processes
being operative in Palaeozoic times.
28 Relate metamorphism to plate tectonic environ
ment.
29 Discuss the role of supercontinents within the
basic framework of plate tectonics.
 REVIEW QUESTIONS 299

30 How and why does the Archaean tectonic regime tances from the source to place detectors if
differ from that ofthe present? critical-angle reflections from the interfaces are
31 Write an essay on energy resources and plate to be observed.
tectonics. 47 The following table lists data obtained from two
32 Discuss the evidence for the concept that suspect tectonic provinces in North America. Calculate
terrains played an important role in the tectonic the heat flow at each site and, using graphs or
development ofthe Western Cordillera of North calculations as appropriate, discuss the signifi
America. cance of the results:
33 Compare and contrast the evolution of the
Andes and the Himalaya.
34 Describe the scope for the development of thick Heat production Thermal Temperature
of surface rocks conductivity gradient
wedges of sediment within the context of plate
Site (lO-^Wm-3) (Kkm-^)
tectonic environments.
35 Discuss the evidence for convection in the man Cenozoic province (last thermal event at 20Ma)
tle. A1 4.28 6.6 15.0

36 Outline and discuss the data which have led to A2 5.50 4.9 22.5
A3 5.90 3.8 30.0
current theories of the causes of plate motions.
A4 3.78 3.9 23.9
37 To what extent can the occurrence of mineral
deposits be related to plate tectonic processes?
Caledonian province (last thermal event at 400Ma)
38 Explain the evidence for and role of suspect B1 2.14 2.0 25.0
terrains in the geological development of the B2 2.90 3.0 18.5
Himalaya. B3 3.57 3.0 20.2
39 What has been the impact of plate tectonics on B4 4.83 4.2 16.0
geological studies of continental areas?
40 Give an account of the principal mechanisms of
crustal extension and relate them to specific 48 Assuming that Airy's mechanism for isostatic
tectonic settings. compensation is perfectly obeyed, show that the
41 Describe the structural features of the upper thickness of continental crust at any point is a
crust which might be associated with transform linear function of the topographic height at that
and transcurrent faults. point.
42 Describe how polarity changes of the Earth's The Bouguer gravity anomaly map for a
magnetic field can be used for dating purposes. mountainous area of Utopia was corrected for
43 Discuss the statement 'All orogenesis is related isostatic compensation using an Airy model. In
to continental collision'. general the anomalies on the resulting isostatic
44 What geophysical measurements might assist in anomaly map were negligible, with the exception
elucidating the vertical extent of convection in of one area where a broad band of negative
the mantle? anomalies persisted. This was interpreted as
45 Describe the generalized morphology of a sub- arising from compensation for topography
duction zone, explaining all the features de which has since been removed by erosion. If this
scribed in terms of a subduction model. residual isostatic anomaly had an amplitude of
46 In a particular continental area, the crust is -1000 g.u. and the area is broad, estimate the
believed to consist ofan upper layer, ofthickness excess thickness ofthe root. To what thickness of
25 km and P wave velocity 6.0 km s"^, overlying eroded crustal material does this correspond?
a lower layer of thickness 15 km and P wave (G = 6.67 X 10"^^ m^ kg"^ s"^; 1 g.u. =
velocity 7.0kms~^. The upper mantle P wave 10"^ms~^;
velocity is about 8.2 km s"^ By calculating ar Mean density ofcrust = 2.85 Mg m~^;
rival time curves for refracted arrivals from a Density of upper mantle = 3.35 Mg m"^).
surface explosion,show whether or not the head 49 The thermoremanent magnetization of the Pali
wave from the lower layer of the crust can be sade Sill, which overlooks the Hudson River
expected as a first arrival. Deduce the best dis and the city of New York,is directed towards the
300 REVIEW QUESTIONS

north and has an inclination of 25.6° tion at T isy mm a"^, describe the way in which
downwards. Calculate the implied palaeomag- the situation evolves with time for each of the
netic pole position (in present day coordinates), following five cases:
and the geographic latitude ofthe sill at the time (a)y less than x\
ofits intrusion, approximately 190 Ma. (b)y equal tox;
Assuming that the difference between the (c)y between x and 2x;
present and the palaeolatitude ofthe sill is due to (d)y equal to 2x; and
the drift of North America with respect to the (e)y greater than 2x.
geographic pole, calculate the implied average
rate of drift during the past 190 Ma. Is this an
absolute, maximum or minimum average rate.>
Explain.(Present latitude and longitude ofsill =
40.5°N,75°W; 1 degree oflatitude = 111 km.) R
a:
50 In the plate tectonic situation illustrated below
three plates (A, B and C) are separated by a
symmetrically spreading ridge at R and a subduc- Hint: Consider plate C to be fixed and assign
tion zone at T. If the rate of spreading at R is velocities to A,R and B,relative to C,in terms of
X mm a"^ per ridge flank, and the rate ofsubduc- X andy.
Appendix / The geological timescale and
stratigraphic column

Epoch

Quaternary Pleistocene

Miocene

Oligocene
Cenozoic Tertiary

Paiaeocene

Cretaceous

Jurassic
Mesozoic
Triassic

Permian

Carboniferous

Devonian
Palaeozoic

Ordovician

Cambrian

(Eon)
Proterozoic
Precambrian

Archaean

'Ages from Harland etal.(1990)

References
Abdelsalam, M.G.& Stern, R.J.(1993)Tectonic evolution
ofthe Nakasib suture, Red Sea Hills, Sudan: evidence for
a late Precambrian Wilson cycle. J.£feol. Soc. Lond. 150,
393-404.
Aki, K., ChristofFerson, A.& Husebye, E.(1977) Determi
nation of three-dimensional seismic structures of the
lithosphere.J.^eophys. Res. 82, 277-96.
Allegre, C.J. et al. (1984) Structure and evolution of the
Himalaya-Tibet orogenic belt. Nature 307,17-22.
Allen, C.R. (1981) The modern San Andreas Fault. In
Ernst, G.(ed.) The Geotectonic Development of California.,
pp. 511-34. Prentice Hall, Englewood Cliffs, NJ.
Allen, P.A. & Allen, J.R. (1990) Basin Analysis Principles
andApplications.Blackwell Scentific Publications,Oxford.
Allen, P.A., Homewood, P. & Williams, G.D. (1986)
Foreland basins: an introduction. In Allen, P.A. &
Homewood, P. (eds) Poreland Basins. Spec. Pub. Inter-
nat. Assoc. Sedimentologists, 8, 3-12. Blackwell Scien
tific Publications, Oxford.
Al-Shanti,A.M.S.& Mitchell, A.M.G.(1976)Late Precam
brian subduction and collision in the Al Amar-ldsas
region, Arabian Shield, Kingdom of Saudi Arabia. Tec-
tonophysics 30,T41-7.
Alvarez, W.(1982) Geological evidence for the geograph
ical pattern of mantle return flow and the driving mecha
nism of plate lectomcs.J.pfeophys. Res. S7^ 6697-710.
Anderson,D.L.(1982)Hot spots, polar wander, Mesozoic
convection and the geoid. Nature 297,391-3.
Anderson,D.L.(1989)Theory ofthe Earth. Blackwell Scien
tific Publications, Oxford.
Anderson, D.L. (1994) Superplumes or supercontinents?
Geology 22, 39-42.
Anderson, D.L., Tanimoto, T. & Zhang, Y. (1992) Plate
tectonics and hotspots: the third dimension. Science 256,
1645-51.
Andrews, J.A. (1985) True polar wander: an analysis of
Cenozoic and Mesozoic palaeomagnetic poles./.geophys.
Res. 90,7737-50.
Arculus, R.J. & Curran E.B. (1972) The genesis of the
calc-alkaline rock suite. Earth planet. Sci. Lett. 15, 255-
62.
Argand, E.(1924) La tectonique de PAsie. 13th Intemat.
Geol. Cong. Brussels 1924, Comptes Rendus No. 5, pp.
171-372.
Artemjev, M.E. & Artyushkov, E.V.(1971) Structure and
history of the Baikal rift and the mechanism of rifting.J.
geophys. Res. 76,1197-211.
Ashby, M.F. & Verrall, R.A.(1977) Micromechanisms of
flow and fracture,and their relations to the theology ofthe
upper mantle. Phil. Trans. Roy. Soc. Lond. 288A,59-95.
Atherton, M.P. (1984) The coastal bathshift of Peru.
In Harmon, KS. & Barreiro, B.A. (eds) Andean
Magmatism: Chemical and Isotopic Constraints, pp. 152-
67. Shiva, Cheshire.
Atwater, T.(1981) Propagating rifts in seafloor spreading
patterns. Nature 290,185-6.
Auzende,J.-M. et al.(1989)Direct observation ofa section
through slow-spreading oceanic crust. Nature 337,
726-9.
Bailey, D.K.(1983)The chemical and thermal evolution of
rifts. Tectonophysics 94, 585-97.
Baker,B.H. & Wohlenberg, J.(1971)Structure and evolu
tion of the Kenya Rift Valley. Nature 229,538-42.
Baker,P.E.(1982)Evolution and classification oforogenic
volcanic rocks. In Thorpe, R.S. (ed.) Andesites, pp. 11-
23. Wiley, Chichester.
Ballard, R.D. & van Andel, T.H.(1977) Morphology and
tectonics of the inner rift valley at lat 36®50'N on the
mid-Atlantic ridge. Bull.geol. Soc. Am. 88,507-30.
Banks, R.J., Parker, R.L. & Huestis, S.P. (1977) Isostatic
compensation on a continental scale: Local versus re-
. gional mechanisms.Geophys.J. Roy. astr. Soc. 51,431-52.
Barazangi, M.& Dorman,J.(1969)World seismicity maps
compiled from ESSA, Coast and Geodetic Survey, epi
center data, 1961-1967. Bull, seism. Soc. Am. 59, 369-
80.
Barazangi, M. & Isacks, B. (1971) Lateral variations of
seismic wave attenuation in the upper mantle above the
inclined earthquake zone of the Tonga island arc: deep
anomaly in the upper mdinxXc. J.geophys. Res. 76,8493-
516.
Barazangi, M.& Isacks, B.(1979)Subduction ofthe Nazca
plate beneath Peru: evidence from spatial distribution of
earthquakes. Geophys. J. Roy. astr. Soc. 57,537-55.
Barber, A.J.(1982)Interpretations ofthe tectonic environ
ment ofsouthwest Japan. Proc.geol. Assoc. 93,131-45.
Barber, A. (1985) A new concept of mountain building.
Geology Today 1,116-21.
303
304 REFERENCES
Barberi, F., Bonatti, E., Marinelli, G. & Varet, J. (1974)
Transverse tectonics during the split of a continent; data
from the Afar rift. Tectonophysics 23,17-29.
Barker, P.P.(1979) The history of ridge crest offset at the
Faiklands-Aguihas Fracture Zone from a small circle
geophysical profile. Geophys,]. 59, 131-45.
Barton, P. & Wood, K (1984) Tectonic evolution of the
North Sea Basin: crustal stretching and subsidence. Geo
phys. J. Roy. astr. Soc. 79,987-1022.
Bath, M.(1979)Introduction to Seismology. Birkhauser Ver-
lag, Basel.
Batiza, R., Melson, W.G. & O'Hearn, T. (1988) Simple
magma supply geometry inferred beneath a segment of
the Mid-Atlantic Ridge. Nature 335,428-31.
Beaumont, C. & Tankard, A.J. (eds)(1987) Sedimentary
basins and basin-forming mechanisms. C.S.P.G.
Mem.l2/A.G.S. Spec. Pub. 5. Can. Soc. Pet. Geols.
Bechtel, T.D., Forsyth, D.W., Sharpton, V.L. & Grieve,
RA.F.(1990) Variations in effective elastic thickness of
the North American lithosphere. Nature 343,636-8.
Beck, M.E. Jr (1980) The palaeomagnetic record of plate-
margin tectonic processes along the >vestern edge of
North America. Res. 85,7115-31.
Beck, M.E. Jr (1986) Model for late Mesozoic-early Ter
tiary tectonics of coastal California and western Mexico
and speculations on the origin of the San Andreas fault.
Tectonics 5,49-64.
Beck, M.E., Drake, R.E. & Butler, R.F.(1986) Palaeomag-
netism of Cretaceous volcanic rocks from central Chile
and implications for the tectonics of the Andes. Geology
14,132-6.
Ben-Avraham, Z. 8c von Herzen, R.P. (1987) Heat flow
and continental brczk-up.J.geophys. Res. 92, 1407-16.
Ben-Avraham, Z., Nur, A., Jones, D. 8c Cox, P. (1981)
Continental accretion and orogeny. From oceanic pla
teaus to allochthonous terranes. Science 213,47-54.
Ben-Avraham,Z.,Nur,A.8c Jones,D.(1982)The emplace
ment of ophiolites by collision. ]. geophys. Res. 87,
3861-7.
Berggren, W.A., Kent, D.V., Flynn, J.J. 8c van Couvering,
J.A.(1985)Cenozoic geochronology. Bull.geol. Soc. Am.
96, 1407-18.
Bickle, M.J.(1978) Heat loss from the Earth: a constraint
on Archaean tectonics from the relation between geo-
thermal gradients and the rate of plate production. Earth
planet. Sci. Lett. 40,301-15.
Bicknell, J.D.,Sempere,J.-C., Macdonald, K.C. 8c Fox,P.J.
(1988) Tectonics of a fast spreading center—a deep-tow
and seabeam survey at EPR 19"'30'S.Marinegeophys. Res.
9, 25-45.
Bignell, R.D., Cronan,D.S. 8c Tooms,J.S.(1976) Red Sea
metalliferous brine precipitates. In Strong, D.F. (ed.)
Metallogeny and Plate Tectonics. Geol. Assoc. Can. Spec.
Paper 14, 147-79.
Bird, P. (1979) Continental delamination and the Colo
rado Vbxtzu.J.geophys. Res. 84,7561-71.
Black, R., Caby, R., Moussine-Pouchkine, A. et al.(1979)
Evidence for late Precambrian plate tectonics in West
Africa. Nature 278,223-7.
Bloxham,J. 8c Gubbins,D.(1985)The secular variation of
Earth's magnetic field. Nature 317,777-81.
Bonatti, E. (1978) Vertical tectonism in oceanic fracture
zones. Earth planet. Sci. Lett. 37, 369-79.
Bonatti, E. (1987) Rifting or drifting in the Red Sea?
Nature 330,692-3.
Bonatti, E. 8c Crane,K.(1984)Oceanic fracture zones. Sci.
Am. 250(5), 36-47.
Bonatti, E. 8c Honnorez, J.(1976)Sections of the Earth's
crust in the equatorial h.i\zni\Q.J.geophys. Res. 81,4104-
16.
Bonatti, E., Harrison, C.G.A., Fisher, D.E. et al. (1977)
Easter volcanic chain (southeast Pacific): a mantle hot
line.J.geophys. Res. 82, 2457-78.
Bond, G.C., Nickeson, P.A. 8c Kominz, M.A. (1984)
Breakup of a supercontinent between 625 Ma and
555 Ma: new evidence and implications for continetal
histories. Earth planet. Sci. Lett. 70, 325-45.
Borch, R.S. 8c Green,H.W.11(1987)Dependence ofcreep
in olivine on homologous temperature and its implica
tion for flow in the mantle. Nature 330, 345-8.
Bosworth, W.(1987) Off-axis volcanism in the Gregory
Rift, east Africa: implications for models of continental
rifting. Geology 15, 397-400.
Bott, M.H.P.(1964) Formation of sedimentary basins by
ductile flow of isostatic origin in the upper mantle.
Nature 201,1082-4.
Bott, M.H.P.(1967)Solution ofthe linear inverse problem
in magnetic interpretation with application to oceanic
magnetic anomalies. Geophys. J. Roy astr. Soc. 13, 313-
23.
Bott, M.H.P. (1971) Evolution of young continental
margins and formation ofshelf basins. Tectonophysics 11,
319-27.
Bott, M.H.P.(1976) Formation of sedimentary basins of
graben type by extension of the continental crust. Tec
tonophysics 36, 77-86.
Bott, M.H.P.(1981) Crustal doming and the mechanism
ofcontinental rifting. Tectonophysics 73,1-8.
Bott, M.H.P.(1982a)The Interior ofthe Earthy its Structurey
Constitution and Evolution (2nd edn). Edward Arnold,
London.
Bott, M.H.P.(1982b)The mechanism ofcontinental split
ting. Tectonophysics 81, 301-9.
Bott, M.H.P.(1983) Deep structure and geodynamics of
the Greenland-Scotland Ridge: an introductory review.
In Bott, M.H.P., Saxov, S., Talwani. M. 8c Theide, J.
(eds) Structure and Development of the Greenland-
Scotland Ridge, pp. 3-9. Plenum Press, New York.

Bott, M.H.P. (1992) The stress regime associated with
continental break-up. In Storey, B.C., Alabaster, T. &
Pankhurst, R.J.(eds)Ma^fmatism and the causes ofConti
nental Break-up. Geol. Soc. Spec. Pub. 68,125-36.
Bott, M.H.P.(1993) Modelling the plate driving mecha
nism./, Soc. Lond. 150,941-51.
Bott, M.H.P. & Kusznir, N.J. (1979) Stress distribution
associated with compensated plateau uplift structures
with application to the continental splitting mechanism.
Geophys.J. Roy. astr. Soc. 56,451-9.
Bott, M.H.P. & Kusznir, N.J. (1984) Origins of tectonic
stress in the lithosphere. Tectonophysics 105,1-13.
Bowin, C., Purdy, G.M., Johnston, C. et at. (1980) Arc-
continent collision in the Banda Sea region. Bull. Am.
Assoc. Petroleum Geols. 64,868-915.
Brace, W.F. & Kohlstedt, D.L. (1980) Limits on litho-
spheric stress imposed by laboratory experiments. /.
pfeophys. Res. 85,6248-52.
Brett, R.(1976) The current status of speculations on the
composition ofthe core ofthe Earth. Rev. Geophys. Space.
Phys. 14, 375-83.
Briden, J.C. (1976) Application of palaeomagnetism to
Proterozoic tectonics. Phil. Trans. Roy. Soc. Lond. 208A,
405-16.
Briggs, J.C.(1987)Biogeopfraphy and Plate Tectonics, Devel
opments in Palaeontology and Stratigraphy 10. Elsevier,
Amsterdam.
Brown, G.C. & Mussett, A.E. (1981) The Inaccessible
Earth. Allen & Unwin, London.
Brown, L.D., Chapin, C.E., Sanford, A.R., Kaufman,S. 8c
Oliver, J. (1980) Deep structure of the Rio Grande Rift
from seismic reflection profiling. /. geophys. Res. 85,
4773-800.
Brozena, J.M. 8c White, RS. (1990) Ridge jumps and
propagations in the South Atlantic Ocean. Nature 348,
149-52.
Bullard, E.G., Everett, J.E. 8c Smith,A.G.(1965)The fit of
the continents around the Atlantic. Phil. Trans. Roy. Soc.
Lond. 258A,41-51.
Bullen, K.E. 8c Bolt, B.A. (1985) An Introduction to the
Theory of Seismology (4th edn). Cambridge University
Press, Cambridge.
Burchfiel,B.C.(1980)Tectonics of non-collisional regimes
—the modern Andes and the Mesozoic Cordilleran oro-
gen ofthe western United States. In Burchfiel, B.C.(ed.)
Continental Tectonics, pp.65-72. National Academy of
Sciences, Washington D.C.
Burke, K.(1976) Development of graben associated with
the initial ruptures of the Atlantic Ocean. Tectonophysics
36,93-112.
Burke, K. (1977) Aulacogens and continental breakup.
Ann. Rev. Earth planet. Sci. 5, 371-96.
Burke, K. 8c Dewey, J.F. (1973) Plume generated triple
junctions: key indicators in applying plate tectonics to
REFERENCES 305
old rocks./. Geol. 81,406-33.
Burke,K., Kidd, W.S.F. 8c Wilson,J.T.(1973)Relative and
latitudinal motion of Atlantic hotspots. Nature 245,
133-7.
Burnett, M.S., Caress, D.W. 8c Orcull, J.A.(1989) Tomo-
graphic image ofthe magma chamber at 12®50'N on the
East Pacific Rise. Nature 339, 206-8.
Butler, R.W.H. (1986) Thrust tectonics, deep structure
and crustal subduction in the Alps and Himalayas.
Soc. Lond. 143,857-73.
Calvert, A.J.(1995)Seismic evidence for a magma chamber
beneath the slow-spreading Mid-Atlantic Ridge. Nature
377,410-14.
Calvert, A.J., Sawyer, E.W., Davis, W.J. 8c Ludden, J.N.
(1995) Archaean subduction inferred from seismic im
ages of a mantle suture in the Superior Province. Nature
375,670-4.
Campbell, A.C. et al. (1988) Chemistry of hot springs on
the Mid-Atlantic Ridge. Nature 335, 514-19.
Cande, S.C. 8c Kent, D.V. (1992) A new geomagnetic
polarity time scale for the late Cretaceous and Cenozoic.
J.geophys. Res. 97,13 917-51.
Cann,J.R (1970)New model for the structure ofthe ocean
crust. Nature 226,928-30.
Cann, J.R. (1974) A model for oceanic crustal structure
developed. Geophys. J. Roy. astr. Soc. 39, 169-87.
Cannat, M., Mevel, C., Maia, M.et al. (1995) Thin crust,
ultramafic exposures, and rugged faulting patterns at the
Mid-Atlantic Ridge (22'-24'N). Geolop^ 23,49-52.
Carbotte,S.M. 8c MacDonald,K.C.(1994)The axial topo
graphic high at intermediate and fast spreading ridges.
Earth planet. Sci. Lett. 128,85-97.
Caress, D.W., Menard, H.W. 8c Hey, RN.(1988) Eocene
reorganization of the Pacific-Farallon spreading center
north ofthe Mendocino fracture zone,.],geophys. Res. 93,
2813-38.
Carey,S.W.(1976)The Expanding Earth. Elsevier, Amster
dam.
Carey, S.W. (1988) Theories of the Earth and Universe.
Stanford University Press, Stanford, CA.
Carlson, RL. 8c Melia, P.J. (1984) Subduction hinge mi
gration. Tectonophysics 102, 399-411.
Carter, N.L. 8c Ave'lallemant, H.G.(1970) High tempera
ture flow ofdunite and peridotite. Bull.geol. Soc. Am.81,
2181-202.
Carter, W.E. 8c Robertson, D.S.(1986)Studying the earth
by very-long-baseline interferometry. Sci. Am. 255(5),
44-52.
Cathles, L.M. (1975) The Viscosity of the Earth's Mantle.
Princeton University Press, Princeton NJ.
Cawood, P.A. 8c Suhr, G.(1992) Generation and obduc-
tion of ophiolites: constraints from the Bay of Islands
complex, western Newfoundland. Tectonics 11,884-97.
Chang, C. 8c Cheng, H.(1973)Some tectonic features of
306 REFERENCES
the Mt. Jolmo Lungma area, southern Tibet. Sci. Sinica
16, 257-65.
Chang, C. et al, (1986) Preliminary conclusions of the
Royal Society and Academica Sinica 1985 geotraverse of
Tibet. Nature 323, 501-7.
Chapman,D.S. 8c Pollack, H.N.(1975)Global heat flow: a
new look. Earth planet. Sci. Lett. 28, 23-32.
Chappie, W.M. 8c Forsyth, D.W.(1979) Earthquakes and
bending of plates at tvtnchts. J.pfeophys. Res. 84,6729-
49.
Chappie, W.M. 8c Tullis, T.E. (1977) Evaluation of the
forces that drive phtcs.J.^feophys. Res. 82,1967-84.
Charlton T.R (1991) Postcollision extension in arc-
continent collision zones, eastern Indonesia. Geolqgy 19,
28-31.
Chase, C.G. (1978) Plate kinematics: the Americas, east
Africa, and the rest of the world. Earth planet. Sci. Lett.
37, 355-68.
Chase, C.(1979) Asthenospheric counterflow: a kinematic
model. Geophys.J. Roy. astr. Sac. 56,1-18.
Chase, C.G. 8c Gilmer, T.H. (1973) Precambrian plate
tectonics: the Midcontinent Gravity High. Earth planet.
Sci. Lett. 21, 70-8.
Choukroune,P., Francheteau,J. 8c Le Pichon,X.(1978)In
situ structural observations along Transform Fault A in
the FAMOUS area. Mid-Atlantic Ridge. Bull. £feol. Sac.
Am. 89,1013-29.
Christensen, N.l. 8c Fountain, D.M.(1975) Constitution
of the lower continental crust based on experimental
studies of seismic velocities in granulite. Bull, pfeol. Sac.
Am. 86, 227-36.
Christensen, N.I. 8c Salisbury, M.H. (1972) Seafloor
spreading, progressive alteration of layer 2 basalts, and
associated changes in seismic velocities. Earth planet. Sci.
Lett. 15, 367-75.
Christodoulidis, D.C., Smith, D.E., Kolenkiewicz, R,
Klosko, S.M., Torrence, M.H. 8c Dunn, P.J. (1985)
Observing tectonic plate motions and deformations from
satellite laser V2in^mg.J.^eophys. Res. 90,9249-63.
Clark, T.A., Gordon, D., Himwich, W.E., Ma, C.,
Mallama, A. 8c Ryan, J.W. (1987) Determination of
relative site motions in the western United States using
Mark 111 Very Lx)ng Baseline Interferometry. J.£feophys.
Res. 92, 12 741-50.
Cloetingh, S., Wortel, R 8c Vlaar, N.J. (1984) Passive
margin evolution, initiation of subduction and the Wil
son cycle. Tectonophysics 109, 147-63.
Cobbing, E.J. (1982) The segmented coastal batholith of
Peru: its relationship to volcanicity and metallogenesis.
Earth Sci. Rev. 18, 241-51.
Cobbing, E.J. 8c Pitcher, W.S.(1972) Plate tectonics and
the Peruvian Andes. Nature Rhys. Sci. 240, 51-3.
Coe, RS., Prevot, M. 8c Camps, P.(1995) New evidence
for extraordinarily rapid change of the geomagnetic field
during a reversal. Nature 374,687-92.
Coffin, M.F. 8c Eldholm, O. (1994) Large igneous
provinces: crustal structure, dimensions, and external
consequences. Geophys. 32,1-36.
Cohen, S.C. 8c Smith, P.E. (1985) LAGEOS scientific
results: introduction./.pfeophys. Res. 90,9217-20.
Collette, B.J.(1979)Thermal contraction joints in spread
ing seafloor as origin of fracture zones. Nature 251,
299-300.
Collier,J. 8cSinha, M.(1992)Seismic mapping ofa magma
chamber beneath the Valu Fa Ridge,Lau Bzsm.J.£ieophys.
Res. 97,14 031-54.
Condie, K.C.(1982)Plate Tectonics and Crustal Evolution.
Pergamon Press, Oxford.
Coney, P.J. (1989) Structural aspects of suspect terranes
and accretionary tectonics in western North America. J.
struct. Geol. 11,107-25.
Coney, P.J., Jones, P.L. 8c Monger,J.W.H.(1980) Cordil-
leran suspect terranes. Nature 288,329-33.
Cooper, A.K., Marlow, M.S. 8c Scholl, D.W.(1977) The
Bering Sea—a multifarious marginal basin. In Talwani,
M. 8c Pitman, W.C. Ill (eds) Island Arcs, Deep Sea
Trenches and Back-arc Basins., pp.437-50. Maurice
Ewing Series 1. Am. Geophys. Union.
Cooper, A.K., Marlow, M.S., Scholl, D.W. 8c Stevenson,
A.J. (1992) Evidence for Cenozoic crustal extension in
the Bering Sea region. Tectonics 11,719-31.
Corliss, J.B. etal.(1979)Submarine thermal springs on the
Galapagos Rift. Science 203,1073-83.
Courtillot, V. (1982) Propagating rifts and continental
breakup. Tectonics 1, 239-56.
Courtillot, V. 8c Besse, J. (1987) Magnetic field reversals,
polar wander and core-mantle coupling. Science 237,
1140-7.
Courtillot, V. 8c Vink, G.E.(1983) How continents break
up. Sci. Am. 249(1),40-7.
Courtney,RC.8c White,RS.(1986)Anomalous heat flow
and geoid across the Cape Verde Rse: evidence for
dynamic support from a thermal plume in the mantle.
Earth planet Sci. Lett. 79, 373-84.
Cowan, J. 8c Cann, J.(1988) Supercritical two-phase sepa
ration of hydrothermal fluids in the Troodos ophiolite.
Nature 333,259-61.
Cowan, D.S. 8c Silling, R.(1978) A dynamic scaled model
of accretion at trenches and its implications for the
tectonic evolutions of subduction complexes. J.£feophys.
Res. 83,5389-96.
Coward, M.P. (1986) Heterogeneous stretching, simple
shear and basin development. Earth planet. Sci. Lett. 80,
325-36.
Coward, M.P.(1994a) Continental collision. In Hancock,
P.L. (ed) Continental Deformation., pp. 264-88. Perga
mon Press, Oxford.
Coward, M.P. (1994b) Inversion tectonics. In Hancock,

P.L. (ed) Continental Deformation; pp. 289-304. Perga-
mon Press, Oxford.
Cox,A.(1975)The frequency ofgeomagnetic reversals and
the symmetry of the nondipole field. Rev. Geophys. Space
Rhys. 13(3), 35-51.
Cox, A.(1980) Rotation of microplates in western North
America.In Strangway,D.W.(ed.)The Continental Crust
and its Mineral Deposits. Geol. Assoc. Canada. Spec.
Paper 20, pp. 305-21.
Cox,A.& Hart,RB.(1986)Plate Tectonics. How It Works.
Blackwell Scientific Publications, Oxford.
Cox, A., Doell, R.R & Dalrymple, G.B.(1964) Geomag
netic polarity epochs. Science 143, 351-2.
Cox,A.v.,Dalrymple,G.B.& Doell,R.R (1967)Reversals
ofthe Earth's magnetic field. Sci. Am. 216(2),44-54.
Cox, K.G.(1989) The role of mantle plumes in the devel
opment of continental drainage patterns. Nature 342,
873-7.
Creer, K.M. (1965) Palaeomagnetic data from the
Gondwanic continents. In Blackett, P.M.S., Bullard, E.
& Runcorn,S.K.(eds)A Symposium on Continental Drift.
Phil. Trans. Roy. Soc. Lond. 258A,27-40.
Cross, T.A. & Pilger, R.H.(1982) Controls ofsubduction
geometry,location of magmatic arcs, and tectonics ofarc
and back-arc regions. Bull.^eol. Soc. Am. 93,545-62.
Crough, S.T. (1979) Hotspot epeirogeny. Tectonophysics
61,321-33.
Crough, S.T., Morgan, W.J. & Hargraves, R.B. (1980)
Kimberlites: their relation to mantle hotspots. Earth
planet. Sci. Lett. 50, 260-74.
Crowell, J.C.(1962) Displacement along the San Andreas
Fault, California. Spec. Paper 71, Geol. Soc. Am.
Crowell, J.C. (1974a) Sedimentation along the San
Andreas fault, California. In Dott, R.H. Jr & Shaver,
R.H. (eds) Modem and Ancient Geosynchrial Sedimenta
tion. Soc. econ. Pal. Mineral. Spec. Pub. 19, 292-303.
Crowell, J.C., (1974b) Origin of late Cenozoic basins in
southern California. In Dickinson, W.R (ed.) Tectonics
and Sedimentation. Soc. econ. Pal. Mineral. Spec. Pub. 22,
190-204.
Dalziel, I.W.D. (1991) Pacific margins of Laurentia and
East Antarctica-Australia as a conjugate rift pair: evi
dence and implications for an Eocambrian superconti-
nent. Geology 19, 598-601.
Dalziel, I.W.D. (1995) Earth before Pangea. Sci. Am.
272(1), 38-43.
Davidson, A. & Rex, D.C. (1980) Age of volcanism and
rifting in south western Ethiopia. Nature 283,657-8.
Davidson, J.P. (1983) Lesser Antilles isotopic evidence of
the role of subducted sediment in island arc magma
genesis. Nature 306, 253-5.
Davies, D.(1968) A comprehensive test ban. Sci. J. Lond.
Nov. 1968, 78-84.
Davies, G.F. (1977) Whole mantle convection and plate
REFERENCES 307
tectonics. Geophys.J. Roy. astr. Soc. 49,459-86.
Davies, G.F.(1980)Review ofoceanic and global heat flow
estimates. Rev. Geophys. Space Phys. 18,718-22.
Davies, G.F. & Richards, M.A.(1992) Mantle convection.
J. Geol. 100,151-206.
Davis, D.,Suppe, J. & Dahlen, F.A.(1983)The mechanics
offold-and-thrust beks.J.geophys. Res. 88,1153-72.
Davis, P.M.(1984) The median valley, a result of magma
fracture beneath mid-ocean ridges. Nature 308,53-4.
Debiche, M.G., Cox, A. & Engebretson, D. (1987) The
motion of allochthonous terranes across the North
Pacific Basin. Spec. Pap. 207, Geol. Soc. Am.
DeLong, S.E., Dewey, J.F. & Fox, P.J. (1977) Displace
ment history of oceanic fracture zones. Geolopfy 5, 199-
201.
DeMets, C., Gordon, R.G.,Stein,S.& Argus, D.F.(1987)
A revised estimate of Pacific-North America motion and
implications for western North America plate boundary
zone tectonics. Geophys. Res. Lett. 14,911-14.
DeMets, C., Gordon, RG.,Argus, D.F. & Stein, S.(1990)
Current plate motions. Geophys. J. Int. 101,425-78.
Detrick, R.S., Buhl,P., Vera, E.etal.(1987)Multi-channel
seismic imaging of a crustal magma chamber along the
East Pacific Rise. Nature 326, 35-41.
Detrick, R.S., White R.S. & Purdy, G.M.(1993) Crustal
structure of North Atlantic fracture zones. Rev. Geophys.
31,439-57.
Detrick, R., Collins, J., Stephen, R. & Swift, S.(1994). In
situ evidence for the nature of the seismic layer 2/3
boundary in oceanic crust. Nature 370, 288-90.
Dewey, J.F. (1969) Evolution of the Appalachian/
Caledonian orogen. Nature 222, 124-9.
Dewey,J.F.(1972)Plate tectonics. Sci. Am.226(5),56-68.
Dewey,J.F.(1976) Ophiolite obduction. Tectonophysics 31,
93-120.
Dewey, J.F. & Bird, J.M. (1970) Mountain belts and the
new global itQtomcs.J.pfeophys. Res. 75, 2625-47.
Dewey,J.F. & Burke,K.C.A.(1973)Tibetan, Variscan,and
Precambrian basement reactivation: products of conti
nental collision./. Geol. 81,683-92.
Dewey, J.F. & Burke, K.C.A.(1974) Hot spots and conti
nental break-up: implications for collisional orogeny.
Geology 2, 57-60.
Dewey,J.F. & Horsfield, B.(1970)Plate tectonics,orogeny
and continental growth. Nature 225, 521-5.
Dewey, J.F., Hempton, M.R,Kidd, W.S.F., Saroglu, F. &
§eng6r, A.M.C. (1986) Shortening of continental
lithosphere: the neotectonics of eastern Anatolia: a
young collision zone. In Coward, M.P. & Ries, A.C.
(eds) Collision Tectonics. Geol. Soc. Spec. Pub. 19,3-36.
de Wit, M.J. & Stern, C.R.(1981) Variations in the degree
ofcrustal extension during formation ofa back arc basin.
Tectonophysics 72, 229-60.
Dickinson, W.R.(1971) Plate tectonic models of geosyn-
308 REFERENCES
dines. Earth planet. Sci. Lett. 10,165-74.
Dickinson, W.R.(1974)Plate tectonics and sedimentation.
In Dickinson, W.R. (ed.) Tectonics and Sedimentation.
Soc. econ. Pal. Mineral. Spec. Pub. 22, 1-27.
Dickinson, W.R (1977)Tectonostratigraphic evolution of
subduction-controlled sedimentary assemblages. In Tal-
wani, M.& Pitman, W.C. Ill (eds)Island Arcs, Deep Sea
Trenches and Back-arc Basins, pp. 33-40. Maurice Ewing
Series 1. Am. Geophys. Union.
Dietz, R.S.(1961) Continental and ocean basin evolution
by spreading ofthe sea floor. Nature 190,854-7.
Dietz, R.S. & Holden, C.(1970) The breakup of Pangaea.
Sci. Am. 223(4), 30-41.
Dimroth,E.(1981)Labrador Geosyncline: type example of
early Proterozoic cratonic reactivation. In Kroner, A.
(ed.) Precambrian Plate Tectonics, pp. 331-52. Elsevier,
Oxford.
Dixon, T.H. (1991) An introduction to the Global Posi
tioning System and some geological applications. Rev.
Geophys. 29, 249-76.
Dixon,T.H.,Stern,R.J.& Hussein,I.H.(1987)Control of
Red Sea Rift geometry by Precambrian structures. Tecton
ics 6, SSl-72.
Doglioni, C.(1993) Geological evidence for a global tec
tonic polarity. Soc. Lond. 150,991-1002.
Drake, R. Vergar, M., Munizaga, F. & Vincente, J.C.
(1982) Geochronology of Mesozoic-Cenozoic magma-
tism in central Chile, lat 31®-36°S. Earth Sci. Rev. 18,
353-63.
Dunbar, J.A. & Sawyer, D.S.(1988) Continental rifting at
pre-existing lithospheric weaknesses. Nature 333,
450-2.
Duncan, RA. & Richards, M.A.(1991) Hotspots, mantle
plumes, flood basalts, and true polar wander. Rev.
Geophys. 29, 31-50.
Dunlop, D.J.(1981) Palaeomagnetic evidence for Protero
zoic continental development. Geophys. J. Roy astr. Soc.
301A,265-77.
du Toit, A.L.(1937) Our Wandering Continents. Oliver &
Boyd, Edinburgh.
Dvorkin,J., Nur,A., Mavko,G.& Ben-Avraham,Z.(1993)
Narrow subducting slabs and the origin ofbackarc basins.
Tectonophysics 227,63-79.
Dziewonski, A.M. (1984) Mapping the lower mantle:
determination of lateral heterogeneity up to degree and
order 6.J.pfeophys. Res. 89, 5929-52.
Dziewonski, A.M. & Anderson, D.L. (1984) Seismic to
mography of the Earth's interior. Am. Sci. 72,483-94.
Edgar, N.T. (1974) Acoustic stratigraphy in the deep
oceans. In Bur, C.A. & Drake, C.L.(eds) The GeoUtgy of
Continental Margins, pp. 243-6. Springer-Verlag, Ber
lin.
Elder, J.W.(1965)Physical processes in geothermal areas.
In Lee, W.M.K.(ed.) Geophys. Mono^r. No. 8, pp. 211-
39. Am. Geophys. Union, Washington DC.
Elsasser, W.M.(1969) Convection and stress propagation
in the upper mantle. In Runcorn,S.K.(ed.) The Applica
tion ofModem Physics to the Earth and Planetary Interiors,
pp. 223-46. Wiley-Interscience, London.
Elsasser, W.M.(1971) Sea-floor spreading as thermal con
vection./. Geophys. Res. 76,1101-12.
Elsasser, W.M.,Olsen,P.& Marsh,B.D.(1979)The depth
of mantle convection./,geophys. Res. 84,147-55.
Elthon, D. (1981) Metamorphism in oceanic spreading
centres.In Emiliani, C.(ed.) The Oceanic Lithosphere, The
Sea 7, pp. 285-303. Wiley, New York.
Elthon, D.(1991) Geochemical evidence for formation of
the Bay of Islands ophiolite above a subduction zone.
Nature 354,140-3.
Engebretson, D.C., Cox, A. & Gordon, R.G.(1985) Rela
tive motion between oceanic and continental plates in
the Pacific Basin. Geol. Soc. Am. Spec. Paper 206.
Engeln, J.F., Stein, S., Werner, J. 8c Gordon, R.G.(1988)
Microplate and shear zone models for oceanic spreading
center reorganisations. Res. 93,2839-56.
England, P.C. (1982) Some numerical investigations of
large scale continental deformation. In Hsu, K.G. (ed.)
Mountain Buildin£f Processes, pp. 129-41. Academic
Press, London.
England, P.C. 8c Houseman, G.A. (1986) Finite strain
calculations of continental deformation 2. Comparison
with the India-Asia collision zone./. geophys. Res. 91,
3664-76.
England, P.C. 8c McKenzie, D.P. (1982) A thin viscous
sheet model for continental deformation. Geophys.J. Roy.
astr. Soc. 70, 295-321.
England, P. 8c Molnar, P. (1990) Right-lateral shear and
rotation as the explanation for strike-slip faulting in
eastern Tibet. Nature 344, 140-2.
Ernst, W.G. (1973) Blueschist metamorphism and P-T
regimes in active subduction zones. Tectonophysics 17,
255-72.
Ernst, W.G. (ed.)(1975) Subduction Zone Metamorphism.
Benchmark Papers in Geology, Vol. 19. Dowden,
Hutchinson 8c Ross,Stroudsberg, PA.
Evans, A.M. (1987) An Introduction to Ore Geology (2nd
edn). Blackwell Scientific Publications, Oxford.
Faure, G. (1986) Principles of Isotope Geology (2nd edn).
Wiley, Chichester.
Fein, J.B. 8c Jurdy, D.M.(1986) Plate motion controls on
back-arc spreading. Geophys. Res. Lett. 12, 545-8.
Flower, M.F.J. (1981) Thermal and kinematic control on
ocean ridge magma fractionation: contrasts between At
lantic and Pacific spreading axes./.geol. Soc. Lond. 138,
695-712.
Force, E.R. (1984) A relation among geomagnetic rever
sals, sea floor spreading rate, paleoclimate, and black
shales. Eos (Trans. A.G.U.) 65,18-19.

Forsyth, D.W.(1975) The early structural evolution and
anisotropy of the oceanic upper mantle. Geophys, J. Roy.
astr. Sac. 43,103-62.
Forsyth, D.W. & Uyeda, S.(1975) On the relative impor
tance of the driving forces of plate motion. Geophys. J.
Roy. astr. Sac. 43,163-200.
Fowler, C.M.R. (1976) Crustal structure of the mid-
Atlantic ridge crest at 37°N. Geophys.J. Roy. astr. Soc. 47,
459-91.
Fowler,C.M.R.(1990)The Solid Earth. Cambridge Univer
sity Press, Cambridge.
Fox,P.J., Schreiber, E., Rowlett, M.& McCamy,K.(1976)
The geology of the Oceanographe Fracture Zone: a
model for fracture zones.J.geophys. Res. 81,4117-28.
Fox,P.J. & Stroup,J.B.(1981)The plutonic foundation of
the oceanic crust. In Emliani, C. (ed.) The Oceanic
Lithosphercy The Sea 7, pp. 119-218. Wiley, New York.
Frakes, L.A. (1979) Climates throughout Geologic Time.
Elsevier, Amsterdam.
Francheteau, J.(1983)The oceanic crust. Sci. Am. 249(3),
68-84.
Francis, T.J.G. & Porter, l.T. (1973) Median valley
seismology: the mid-Atlantic ridge near 45°N. Geophys.].
Roy. astr. Soc. 34, 279-311.
Frank, F.C. (1968) Curvature of island arcs. Nature 220,
363.
Frankel, H.(1988) From continental drift to plate tecton
ics. Nature 335, 127-30.
Fujita, K. & Kanamori, H.(1981) Double seismic zones
and stresses of intermediate depth earthquakes. Geophys.
J. Roy. astr. Soc. 66,131-56.
Fuller, M.(1987) A mechanism for reversals? Nature 326,
132-3.
Fyfe, W.S. & Londsdale, P. (1981) Ocean floor hydro-
thermal activity. In Emiliani, C.(ed.) The Oceanic Litho-
sphere, The Sea 7, pp. 589-638. Wiley, New York.
Gaal, G. & Schulz, K.J. (eds)(1992) Precambrian metal-
logeny related to plate tectonics. Precambrian Res. 58,
1-446.
Gao, S., Davies, P.M., Lin, H. et al. (1994) Asymmetric
upwarp ofthe asthenosphere beneath the Baikal rift zone,
Sihenz.J.geophys. Res. 99, 15 319-30.
Garfunkel, Z. (1975) Growth, shrinking and long term
evolution of plates and their tectonic implications for the
flow pattern in the mzntXe.J.geophys. Res. 80,4225-32.
Garland, G.D.(1979)Introduction to Geophysics(2nd edn).
W.B. Saunders Co., Philadelphia.
Garson, M.S.& Krs, M.(1976)Geophysical and geological
evidence ofthe relationships ofRed Sea transverse tecton
ics to ancient fractures. Bull.geol. Soc. Am. 87, 169-81.
Gass, l.G. (1980) The Troodos massif; its role in the
unravelling of the ophiolite problem and its significance
in the understanding of constructive plate margin pro
cesses. In Panayistou, A. (ed.) Ophiolites^ pp. 23-35,
REFERENCES 309
Geol. Surv. Cyprus.
Gente,P., Pockalny, RA.,Durand,C.etal.(1995)Charac
teristics and evolution of the segmentation of the Mid-
Atlantic Ridge between 20"N and 24°N during the last
10 million years. Earth planet. Sci. Lett. 129, 55-71.
Gibb, RA. (1978) Slave-Churchill collision tectonics.
Nature 271,50-2.
Gibb, RA., Thomas, M.D., Kearey, P. & Tanner, J.G.
(1978) Gravity anomalies at structural province bound
aries of the Canadian Shield: products of Proterozoic
plate interaction. GeoSkrifter 10, 21-57.
Gibb, RA., Thomas, M.D., Lapointe, P. & Mukhopad-
hyay, M. (1983) Geophysics of proposed Proterozoic
sutures in Canada. Precambrian Res. 19, 349-84.
Glatzmaier, G.A. & Roberts, P.H. (1995) A three-
dimensional self-consistent computer simulation of a
geomagnetic field reversal. Nature 377, 203-9.
Glikson, A.Y. (1981) Uniformitarian assumptions, plate
tectonics and the Precambrian Earth. In Kroner, A.(ed.)
Precambrian Plate Tectonics.^ pp.91-104. Elsevier,
Amsterdam.
Goetze,C.(1978)The mechanism ofcreep in olivine. Phil.
Trans. Roy. Soc. Lond. 288A,99-119.
Gorbatov, A., Suarez, G., Kostoglodov, V. & Gordeev, E.
(1994)A double-planed seismic zone in Kamchatka from
local and teleseismic data. Geophys. Res. Lett. 21,1675-8.
Gordon,RG.& Stein,S.(1992)Global tectonics and space
geodesy. Science 256, 333-42.
Grand, S.P. (1994) Mantle shear structure beneath the
Americas and surrounding oceans. J. geophys. Res. 99,
11591-621.
Green, D.H. & Ringwood, A.E. (1967) An experimental
investigation of the gabbro to eclogite transformation
and its petrological applications. Geochim. cosmochim.
Acta, 31, 767-833.
Green, H.W. (1994) Solving the paradox of deep earth
quakes. Sci. Am. 271(3), 50-7.
Green, J.C.(1983) Geologic and geochemical evidence for
the nature and development of the Precambrian Mid-
continent Rift ofNorth America. Tectonophysics 94,413-
38.
Green,T.H.(1970)High pressure experimental studies on
the mineralogical constitution of the lower crust. Phys.
Earth planet. Interiors 3,441-50.
Grow,J.A.(1973)Crustal and upper mantle structure ofthe
central Aleutian arc. Bull.geol. Soc. Am. 84,2169-92.
Gubbins, D.(1987) Mechanism for geomagnetic polarity
reversals. Nature 326,167-9.
Gurnis, M.(1988) Large-scale mantle convection and the
aggregation and dispersal of supercontinents. Nature
332,695-9.
Hadley,D.& Kanaiiiori, H.(1977)Seismic structure ofthe
Transverse Ranges, California. Bull. geol. Soc. Am. 88,
1469-78.
310 REFERENCES
Halbach, P. et al (1989) Probable modern analogue of
Kuroko-type massive sulphide deposits in the Okinawa
Trough back-arc basin. Nature 338,496-9.
Hallam, A.(1972) Continental drift and the fossil record.
Sci. Am. 227(5), 56-66.
Hallam, A. (1973a) A Revolution in the Earth Sciences.
Clarendon Press, Oxford.
Hallam, A.(1973b) Provinciality, diversity and extinction
of Mesozoic marine invertebrates in relation to plate
movements. In Tarling, D.H. & Runcorn, S.C. (eds)
Implications of Continental Drift to the Earth Sciences, 1,
pp. 287-94. Academic Press, Lx)ndon.
Hallam, A.(1975) Alfred Wegener and the hypothesis of
continental drift. Sci. Am. 232(2), 88-97.
Hallam, A. (1981) Relative importance of plate move
ments, eustasy, and climate in controlling major biogeo-
graphical changes since the early Mesozoic.In Nelson,G.
& Rosen, D.E.(eds) Vkariance Biopfeqgraphy. A Critique,
pp. 303-40. Colombia University Press, New York.
Ham, W.E.(1969) Regional Geolopfy ofthe Arbuckle Moun
tains, Oklahoma. Oklahoma Geol. Survey Guidebook.
Hamburger, M.W. & Isacks, B.L.(1988) Diffuse back-arc
deformation in the southwestern Pacific. Nature 322,
599-604.
Harding, A.J., Orcutt, J.A., Kappus, M.E. et al. (1989)
Structure of young oceanic crust at 13°N on the East
Pacific Rise from expanding spread profiles. J. £feophys.
Res. 94,12 163-96.
Harding,T.P.(1985)Seismic characteristics and identifica
tion of negative flower structures, positive flower struc
tures and positive structural inversion. Bull. Am. Assoc.
Petroleum Geols. 69,582-600.
Hargraves, RB. (1981) Precambrian tectonic style: a
liberal uniformitarian interpretation. In Kroner, A.(ed.)
Precambrian Plate Tectonics, pp. 21-56. Elsevier, Amster
dam.
Harland, W.B., Cox, A.V., Llewellyn, P.G., Picton,
C.A.G., Smith, A.G. & Walters, R (1982) A Geologic
Time Scale. Cambridge University Press, Cambridge.
Harland, W.B., Armstrong, RL., Cox, A.V., Craig, L.E.,
Smith, A.G.& Smith, D.G.(1990)A Geologic Time Scale
1989. Cambridge University Press, Cambridge.
Harper, G.D. (1988) Episodic magma chambers and a
magmatic extension in the Josephine Ophiolite. Geology
16,831-4.
Harper,J.F.(1978)Asthenosphere flow and plate motions.
Geophys.J. Roy. astr. Soc. 55,87-110.
Harrison, C.G.A. 8c Bonatti, E.(1981) The oceanic litho-
sphere.In Emiliani. C.(ed.) The Oceanic Lithosphere. The
Sea 7, pp. 21-48. Wiley, New York.
Harrison, C.G.A. 8c Sclater, J.G. (1972) Origin of the
disturbed magnetic zone between the Murray and Molo-
kai fracture zones. Earth planet. Sci. Lett. 14,419-27.
Hart, RS., Anderson, D.L. 8c Kanamori, M.(1977) The
effect of attenuation on gross Earth models. J. geophys.
Res. 82,1647-54.
Hasegawa, A., Umino, N. 8c Takagi, A. (1978) Double-
planed seismic zone and upper-mantle structure in the
northeastern Japan arc. Geophys. J. Roy. astr. Soc. 54,
281-96.
Hasemi,A.H.,Ishii, H.8c Takagi, A.(1984)Fine structure
beneath the Tohoku district, northeastern Japan arc, as
derived by the inversion ofP-wave arrival times from local
earthquakes. Tectonophysics 101,245-65.
Haxby, W.F.,Turcotte, D.L. 8c Bird, J.M.(1976)Thermal
and mechanical evolution of the Michigan Basin. Tec
tonophysics 36, 57-76.
Hayes, D.E. 8c Ewing, M.(1970) Pacific boundary struc
ture. In Maxwell, A.E.(ed.) The Sea. Ideas and Observa
tions on Progress in the Study ofthe Seas 4(2), pp. 29-72.
Wiley, New York.
Hayward, A.B. 8c Graham,RH.(1989)Some geometrical
characteristics ofinversion.In Cooper, M.A.8c Williams,
G.D.(eds)Thrust Tectonics. Spec.Pub. Geol.Soc.London
44,17-39.
Hedenquist, J.W. 8c Lowenstern, J.B. (1994) The role of
magmas in the formation of hydrothermal ore deposits.
Nature 370,519-27.
Heirtzler, J.R, Le Pichon, X. 8c Baron, J.G.(1966) Mag
netic anomalies over the Reykjanes Rdge. Deep Sea Res.
13,427-43.
Heirtzler, J.R, Dickson, G.O., Herron, E.M., Pitman,
W.C. Ill 8c Le Pichon, X. (1968) Marine magnetic
anomalies, geomagnetic field reversals and motions of
the ocean floor and contm^ats.J.geophys. Res. 73,2119-
36.
Heiskanen, W.A. 8c Vening Meinesz, F.A. (1958) The
Earth and its Gravity Field. McGraw-Hill, New York.
Helwig, J.(1973)Plate tectonic model for the evolution of
the Central Andes: Discussion. Bull. geol. Soc. Am. 84,
1493-6.
Herring, T.A. et al. (1986) Geodesy by radio inter-
ferometry: evidence for contemporary plate motion. J.
geophys. Res. 91,8341-7.
Herron,E.M.(1972)Sea-floor spreading and the Cenozoic
History of the East-Central Pacific. Bull. geol. Soc. Am.
83,1671-92.
Herron, T.J., Ludwig, W., Stoffa, P., Kan, T. 8c Buhl, P.
(1978) Structure of the East Pacific Rise crest from
multichannel seismic reflection data. J.geophys. Res. 83,
798-804.
Herron,T.J., Stoffa, P.L. 8c Buhl, P.(1980) Magma cham
ber and mantle reflections—East Pacific Rse. Geophys.
Res. Lett. 7,989-92.
Hess, H.H.(1962) History of ocean basins. In Petrologic
Studies: A Volume in Honor ofA.F. Buddington, pp. 599-
620. Geol. Soc. Am.
Hey, RN.(1977) A new class of pseudofaults and their

bearing on plate tectonics: a propagating rift model.
Earth planet. Sci. Lett. 37, 321-5.
Hey, R.N., Dunnebier, F.K., & Morgan, W.J. (1980)
Propagating rifts on mid-ocean ridges.J.pieophys. Res. 85,
3647-58.
Hey, R.N., Kleinrock, M.C., Miller, S.P., Atwater, T.M.&
Searle, R.C. (1986) Sea Beam/Deep-Tow investigation
of an active propagating rift system, Galapagos 95.5°W.
J.£feophys. Res. 91,3369-94.
Hey, R.N., Menard, H.W., Atwater, T.M. & Caress, D.W.
(1988) Changes in direction of seafloor spreading revis
ited. Res. 93, 2803-11.
Hilton,D.R.& Craig, H.(1989)A helium isotope transect
along the Indonesian archipelago. Nature 342,906-8.
Hirahara, K. (1981) Three-dimensional seismic structure
beneath southwest Japan: the subducting Philippine sea
plate. Tectonophysics 79,1-44.
Hirn, A., Jiang, M., Sapin, M. et at. (1995) Seismic ani-
sotropy as an indicator of mantle flow beneath the
Himalayas and Tibet. Nature 375,571-4.
Hirn, A., Lepine, J.C., Jobert, G. et at. (1984a) Crustal
structure and variability of the Himalayan border of
Tibet. Nature 307,23-5.
Hirn, A., Nercessian, A., Sapin, M. et at. (1984b) Lhasa
block and bordering sutures—a continuation ofa 500-km
Moho traverse through Tibet. Nature 307, 25-7.
Hoefs, J. (1980) Stable Isotope Geochemistry. Springer-
Verlag, Berlin.
Hoffman,K.A.(1979)Behaviour ofthe geodynamo during
reversal: a phenomenological model. Earth planet. Sci.
Lett. 44,7-17.
Hoffman, P. (1973) Evolution of an early Proterozoic
continental margin: the Coronation Geosyncline and
associated aulacogens of the NW Canadian Shield. Phil.
Trans. Roy. Soc. Lond. 273A,547-81.
Hoffman, P.P.(1980) Wopmay orogen: a Wilson-cycle of
early Proterozoic age in the northwest of the Canadian
Shield. Geol. Assoc. Can. Spec. Paperj 20, 523-49.
Hoffman, P.P. (1987) Continental transform tectonics.
Great Slave'Lake shear zone (ca 1.9 Ga), northwest
Canada. Geolo^ 15,785-8.
Hoffman, P.P. (1989) Speculations on Laurentia's first
gigayear (2.0 to 1.0 Ga). Geology 17,135-8.
Hoffman, P.P.(1991) Did the breakout of Laurentia turn
Gondwanaland inside out.> Science 252, 1409-12.
Hoffman, P., Burke, K.C.A. & Dewey, J.P.(1974) Aulaco
gens and their genetic relation to geosynclines, with a
Proterozoic example from Great Slave Lake, Canada. In
Dott, R.H. Jr & Shaver, R.H.(eds)Modem and Ancient
Geosyncline Sedimentation. Soc. econ. Pal. Mineral. Spec.
Pub. 19, 38-55.
Hogrefe, A., Rubie, D.C.,Sharp, T.G. & Seifert, P.(1994)
Metastability of estatite in deep subducting lithosphere.
Nature 372, 351-3.
REFERENCES 311
Holbrook, W.S. & Keleman, P.B. (1993) Large igneous
province on the US Atlantic margin and implications for
magmatism during continental breakup. Nature 364,
433-6.
Holmes, A. (1928) Radioactivity and continental drift
(Report of Glasgow Geological Society meeting 12
January 1928) Geol. Mag 65,236-8.
Hooper, P.R (1990) The timing of crustal extension and
the eruption of continental flood basalts. Nature 345,
246-9.
Houseman,G.A.(1983)The deep structure ofocean ridges
in a convecting mantle. Earth planet. Sci. Lett. 64, 283-
94.
Houseman, G. & England, P. (1993) Crustal thickening
versus lateral expansion in the India-Asia continental
co\\\s\f)n. J.geophys. Res. 98,12 233-50.
Howell, b.G.(1989) Tectonics of Suspect Terranes: Moun
tain Building and Continental Growth. Chapman & Hall,
London.
Hsu, V. (1990) Seismicity and tectonics of a continent-
island arc collision zone at the island of Taiwan. J.
geophys. Res. 95,4725-34.
Hsui, A.T. & Toksoz, M.N. (1981) Back-arc spreading:
trench migration, continental pull or induced convec
tion? Tectonophysics 74,89-98.
Hurley,P.M.(1968)The confirmation ofcontinental drift.
Sci. Am. 218(4), 52-64.
Hutchinson, C.S. (1983) Economic Deposits and their Tec
tonic Setting. MacMillan Press, London.
lilies, J.H.(1977)Ancientand recent rifting in the Rhinegra-
ben. Geologie Mijnb. 56, 329-50.
Irving, E.(1979)Pole positions and continental drift since
the Devonian. In McElhinny, M.M.(ed.) The Earth: its
Origin^ Structure and Evolution., pp. 567-93. Academic
Press, London.
Irving, E., North, P. & Couillard, R (1974) Oil, climate
and tectonics. Can.J. Earth Sci. 11,1-17.
Isacks, B.L. & Barazangi, M.(1977) Geometry of Benioff
zones: lateral segmentation and downward bending of
the subducted lithosphere. In Talwani, M. & Pitman,
W.C. Ill (eds) Island ArcSy Deep Sea Trenches and Back-
arc Basins^ pp.99-114. Am. Geophys. Union, Washing
ton D.C.
Isacks, B. 8c Molnar, P.(1969) Mantle earthquake mecha
nisms and the sinking of the lithosphere. Nature 223,
1121-4.
Isacks, B. 8c Molnar, P.(1971) Distribution of stresses in
the descending lithosphere from a global survey of
focal mechanism solutions of mantle earthquakes. Rev.
Geophys. Space Phys. 9, 103-74.
Isacks, B., Oliver, J. 8c Sykes, L.R.(1968) Seismology and
the new global tectonics.J.geophys. Res. 73,5855-99.
Isacks, B., Sykes, L.R. 8c Oliver, J. (1969) Pocal mecha
nisms of deep and shallow earthquakes in the Tonga-
312 REFERENCES
Kermadec region and tectonics of island arcs. Bull. £ieol.
Soc. Am. 80,1443-69.
Ito, E. & Sato, H.(1991) Aseismicity in the lower mantle
by superplasticity of the descending slab. Nature 351,
140-1.
Jackson, E.D.(1976) Linear volcanic chains of the Pacific
plate. In Sutton, G.H., Manghnari, M.H.& Moberly, R.
(eds) The Geophysics of the Pacific Ocean Basin and its
Marpfin. Am. Geophys. Union. Geophys. Mono^. 19, 319-
35.
Jackson, E.D., Silver, E.A. & Dalrymple, G.B. (1972)
Hawaiian-Emperor chain and its relation to Cenozoic
circumpacific tectonics. Bull.^feol. Soc. Am. 83,601-18.
Jacob, K.H., Nakamura, K.& Davies, J.N.(1977)Trench-
volcano gap along the Alaska-Aleutian arc: facts and
speculations on the role of terrigenous sediments for
subduction. In Talwani, M. & Pitman, W.C. Ill (eds)
Island ArcSy Deep Sea Trenches and Back-arc Basins.
pp. 243-58. Maurice Ewing. Series I. Am. Geophys.
Union.
Jacobs, J.A. (1975) The Earth's Core. Academic Press,
London.
James, D.E.(1971) Plate tectonic model for the evolution
of the central Andes. Bull.pieol. Soc. Am. 82, 3325-46.
James, D.E. & Snoke, J.A. (1990) Seismic evidence for
continuity of the deep slab beneath central and eastern
Vcm.J.geophys. Res. 95,4789-5001.
Jarvis, G.T. & McKenzie, D.P. (1980) Sedimentary basin
formation with finite extension rates. Earth planet. Sci.
Lett. 48,42-52.
Jeanloz, R. (1993) The mantle in sharper focus. Nature
365, 110-11.
Johnston, C.R. & Bowin, C.O. (1981) Crustal reactions
resulting from the mid-Pliocene to Recent continent-
island arc collision in the Timor region. B.M.R.J. Aust.
Geol. Geophys. 6, 223-43.
Jones, D.G., Silberling, N.J. & Hillhouse, J. (1977)
Wrangellia—a displaced terrane in northwestern North
America. Can.J. Earth Sci. 14, 2565-77.
Jones, D.L., Howell, D.G., Coney,P.J. & Monger,J.W.H.
(1983) Recognition, character and analysis of tectono-
stratigraphic terranes in western North America. In
Hashimoto, M. & Uyeda, S. (eds) Accretion Tectonics in
the circum-Pacific RepfionSy pp. 21-35. Terra Scientific
Pub. Co., Tokyo.
Jordan, T.E., Isacks, B.L., Allmendinger, R.W., Brewer,
J.A.,Ramos,V.A.& Ando,C.J.(1983)Andean tectonics
related to geometry ofsubducted Nazca plate. Bull.pfeol.
Soc. Am. 94, 341-61.
Jurdy, D.M. & Stefanick, M. (1983) Flow models for
back-arc spreading. Tectonophysics 99., 191-206.
Kamp, P.J.J. (1980) Pacifica and New Zealand: proposed
eastern elements in Gondwanaland's history. Nature
288,659-64.
Kanasewich, E.R (1976) Plate tectonics and planetary
convection. Can.J. Earth Sci. 13, 331-40.
Kanimori, H.(1971) Great earthquakes at island arcs and
the lithosphere. Tectonophysics 12,187-98.
Karig, D.E. (1970) Ridges and basins of the Tonga-
Kermadec island arc system.].pfeophys. Res. 75,239-54.
Karig, D.E. (1971) Origin and development of marginal
basins in the western Pacific. J.geophys. Res. 76, 2542-
61.
Karig, D.E.(1974) Evolution of arc systems in the western
Pacific. Rev. Earth planet. Sci. 2, 51-75.
Karig, D.E. (1982) Deformation in the forearc: implica
tions for mountain belts. In Hsii, K.J. (ed.) Mountain
Buildinpf Processesy pp. 59-71. Academic Press, London.
Karig, D.E. & Kay, R.W.(1981) Fate of sediments on the
descending plate at convergent margins. Phil. Trans. Roy.
Soc. Lond. 301A,233-51.
Karig, D.E. & Sharman, G.F. (1975) Subduction and
accretion in trenches. Bull.pfeol. Soc. Am. 86,377-89.
Karl, D.M., McMurty, G.M., MalahofF, A. & Garcia, M.O.
(1988) Loihi Seamount, Hawaii: a mid-plate volcano
with a distinctive hydrothermal system. Nature 335,
532-5.
Karson, J.A. et al. (1987) Along-axis variations in seafloor
spreading in the MARK area. Nature 328,681-5.
Kaula, W.M.(1975) Absolute plate motions by boundary
velocity minimizations. Res. 80, 244-8.
Kawakatsu, H.(1986) Downdip tensional earthquakes be
neath the Tonga arc: a double seismic zone? J. geophys.
Res. 91,6432-40.
Kay, R.W. & Kay, S.M. (1981) The nature of the lower
continental crust: inferences from geophysics, surface
geology and crustal xenoliths. Rev. Geophys. Space Phys.
20,735-60.
Kay, R, Hubbard, N.J. & Gast, P.W. (1970) Chemical
characteristics and origin ofoceanic ridge volcanic rocks.
J.pfeophys. Res. 75,1585-613.
Kearey, P.(1974) Gravity and seismic reflection investiga
tions into the crustal structure ofthe Aves Ridge,eastern
Caribbean. Geophys.]. Roy. astr. Soc. 38,435-48.
Kearey,P.(1976) A regional structural model ofthe Labra
dor Trough, northern Quebec, from gravity studies and
its relevance to continental collision in the Precambrian.
Earth planet. Sci. Lett. 28, 371-8.
Kearey,P.& Brooks, M.(1991) Introduction to Geophys
ical Exploration (2nd edn). Blackwell Scientific Publica
tions, Oxford.
Kearey, P., Peter, G. & Westbrook, G.K.(1975) Geophysi
cal maps ofthe eastern Caribbean./,pfeol. Soc. Lond. 131,
311-21.
Keen, C. & Tramontini, C. (1970) A seismic refraction
survey on the mid-Atlantic ridge. Geophys. J. Roy. astr.
Soc. 20,473-91.
Kellogg, J.N. & Bonini, W.E. (1982) Subduction of the

Caribbean plate and basement uplifts in the overriding
South American plate. Tectonics 1, 251-76.
Kennett, B.L.N. (1977) Towards a more detailed seismic
picture of the oceanic crust and mantle. Marine Geophys.
Res, 3,7-42.
Kennett, B.L.N., & Orcutt, J.A. (1976) A comparison of
travel time inversions for marine refraction profiles. J.
geophys. Res. 81,4061-70.
Kent, D.V. & Gradstein, F.M. (1985) A Cretaceous and
Jurassic geochronology. Soc.Am. 96,1419-27.
Kent, G.M., Harding, A.J. & Orcutt, J.A.(1990)Evidence
for a smaller magma chamber beneath the East Pacific
Rise at 9°30'N. Nature 344,650-3.
Kidd, R-G.W.(1977)A model for the process offormation
of the upper oceanic crust. Geophys.]. Roy. astr. Soc. 50,
149-83.
Kingma, J.T. (1958) Possible origin of piercement struc
tures, local unconformities, and sedimentary basins in
the Eastern Geosyncline, New Zealand. N.Z.J. Geol.
Geophys. 1, 269-74.
Klein, E.M., Langmuir, C.H.,Zindler, A., Standigel, H.&
Hamelin, B. (1988). Isotope evidence of a mantle con
vection boundary at the Australian-Antarctic Discor
dance. Nature 333,623-9.
Klein, G. de V. & Hsui, A.T. (1987) Origin of cratonic
basins. Geology 15,1094-8.
Kleinrock, M.C. & Hey, R.N. (1989) Detailed tectonics
near the tip of the Galapagos 95.5°W propagator: how
the lithosphere tears and a spreading axis develops. ].
geophys. Res. 94,13 801-38.
Knittle, E. & Jeanloz, R. (1991) Earth's core-mantle
boundary: results of experiments at high pressures and
temperatures. Science 251, 1438-43.
Krogh, E.J. (1977) Evidence of Precambrian continent-
continent collision in western Norway. Nature 267,
17-19.
Kroner, A.(1981) Precambrian plate tectonics. In Kroner,
A.(ed.)Precambrian Plate Tectonics., pp. 57-90. Elsevier,
Amsterdam.
Kurten,B.(1969)Continental drift and evolution. Sci. Am.
220(3), 54-65.
Kusznir, N.J. & Bott, M.H.P.(1976) A thermal study of
the formation of oceanic crust. Geophys.]. Roy. astr. Soc.
47,83-95.
Kusznir N.J. & Park, R.G. (1984) Intraplate lithosphere
deformation and the strength ofthe lithosphere. Geophys.
J. Roy. astr. Soc. 79, 513-38.
Langmuir, C.H., Bender, J.B. & Batiza, R. (1986) Petro-
logical and tectonic segmentation ofthe East Pacific Rise,
5°30'-14°30'N. Nature 322,422-9.
Larson, R.L.(1991a) Latest pulse of Earth: evidence for a
mid-Cretaceous superplume. Geology 19, 547-50.
Larson, R.L. (1991b) Geological consequences of super-
plumes. Geology 19,963-6.
REFERENCES 313
Larson, R.L. (1995) The mid-Cretaceous superplume epi
sode. Sci. Am. Feb., 66-70.
Larson, R.L. & Olson, P. (1991) Mantle plumes control
magnetic reversal frequency. Earth planet. Sci. Lett. 107,
437-47.
Larson,R.L.&Pitman, W.C.Ill(1972) World-wide corre
lation of Mesozoic magnetic anomalies, and its implica
tions. Bull.geol. Soc. Am. 83,3645-61.
Leblanc,M.(1976)Proterozoic oceanic crust at Bou Azzer.
Nature 261, 34-5.
Leggett,J.K.(1987)The Southern Uplands as an accretion-
ary prism: the importance ofanalogues in reconstructing
palaeogeography./.^^o/. Soc. Lond. 144, 737-52.
Leggett, J.K., McKerrow, W.S. & Casey, D.M.(1982) The
anatomy of a Lower Palaeozoic accretionary forearc: the
Southern Uplands of Scotland. In Leggett, J.K. (ed.)
Trench-Forearc Geology, Spec. Pub. Geol. Soc. Lond. 10,
495-520.
Lemiszki, P.J. & Brown, L.D. (1988) Variable crustal
structure of strike-slip fault zones as observed on deep
seismic reflection profiles. Bull.geol. Soc. Am. 100,665-
76.
Le Pichon, X.(1968) Sea-floor spreading and continental
drift. J.geophys. Res. 73,3661-97.
Le Pichon, X., Francheteau, J. & Bonnin, J. (1973) Plate
Tectonics. Elsevier, Amsterdam.
Le Pichon,X.,liyama,T., Boulegue,J. etal.(1987)Nankai
Trough and Zenisu Ridge: a deep-sea submersible survey.
Earth planet. Sci. Lett. 83,285-99.
Lesquer, A.,de Almeida,F.F.M., Davino, A., Lachaud,J.C.
& Maillard,P.(1981)Signification structurale des anom
alies gravimetriques de la partie sud du craton de Sao
Francisco (Bresil). Tectonophysics 76, 273-94.
Lister, C.R.B.(1975) Gravitational drive on oceanic plates
caused by thermal contraction. Nature 257,663-5.
Lister, C.R.B.(1980) Heat flow and hydrothermal circula
tion. Ann. Rev. Earth planet. Sci. 8,95-117.
Lister, G.S., Etheridge, M.A. & Simons, P.A. (1986) De
tachment faulting and the evolution of passive continen
tal margins. Geology 14, 246-50.
Livermore, R., Vine, F.J. & Smith, A.G. (1984) Plate
motions and the geomagnetic field—II. Jurassic to Ter
tiary. Geophys. J. Roy. astr. Soc. 79,939-61.
Logatchev, N.A. & Florensov, N.G. (1978) The Baikal
system of rift valleys. Tectonophysics 45, 1-13.
Long,R.E. 8c Backhouse,R.W.(1976)The structure ofthe
western flank of the Gregory rift. Part 11. The mantle.
Geophys.]. Roy. astr. Soc. 44,677-88.
Loper, D.E. (1985) A simple model of whole mantle
convection. Res. 90,1809-36.
Lowrie, A., Smoot, C. 8c Bartiza, R. (1986) Are oceanic
fracture zones locked and strong or weak? New evidence
for volcanic activity and weakness. Geology 14, 242-5.
Lyon-Caen, H. 8c Molnar, P. (1983) Constraints on the
314 REFERENCES
structure of the Himalaya from an analysis of gravity
anomalies and a flexural model of the lithosphere. J.
^eophys. Res. 88,8171-91.
Lyzenga,G.A., Wallace, K.S., Fanselow,J.L., Raefsky, A.&
Groth, P.M.(1986) Tectonic motions in California in
ferred from very long baseline interferometry observa
tions, 1980-1984.7.^^(^/;yj. Res. 91,9473-87.
MacDonald, K.C.(1982) Mid-ocean ridges: fine scale tec
tonic, volcanic and hydrothermal processes within a plate
boundary zone. Rep. Earth planet. Sci. 10,155-90.
MacDonald, K.C. & Fox. P.J. (1983) Overlapping spread
ing centres: new accretion geometry on the East Pacific
Rise. Nature 302, 55-8.
MacDonald, K.C., Sempere, J.-C., Fox, P.J. & Tyce, R.
(1987) Tectonic evolution of ridge-axis discontinuities
by the meeting, linking, or self-decapitation of neigh
bouring ridge segments. Geolq^y 15,993-7.
MacDonald, K.C., Fox, P.J., Perram, P.J. et at. (1988) A
new view of the mid-ocean ridge from the behaviour of
ridge-axis discontinuities. Nature 335,217-25.
MacKenzie, A.S. & McKenzie, D.P.(1983) Isomerization
and aromatization ofhydrocarbons in sedimentary basins
formed by extension. Geol. Mag. 120,417-70.
McCarthy, J., Mutter, J.C., Morton, J.L., Sleep, N.H. &
Thompson, G.A. (1988) Relic magma chamber struc
tures preserved within the Mesozoic North Atlantic
crust.> Bull.geol. Sac. Am. 100,1423-36.
McDougall,1.&Duncan,R.A.(1980)Linear volcanic chain
recording plate motions? Tectonophysics 63,275-95.
McElhinny, M.W.(1973)Palaeomagnetism and Plate Tec
tonics. Cambridge University Press, Cambridge.
McElhinny, M.W. & McWilliams, M.D. (1977) Precam-
brian geodynamics—a palaeomagnetic view. Tectonophys
ics 40, 137-59.
McElhinny, M.W., Taylor, S.R. & Stevenson, D.J.(1978)
Limits to the expansion of Earth, Moon, Mars and
Mercury and to changes in the gravitational constant.
271, 316-21.
McFadden,P.L. & Merrill, R.T.(1984)Lower mantle con
vection and geomagnetism. Res. 89,3354-62.
McKenzie, D.P.(1969) Speculation on the consequences
and causes of plate motions. Geophys.J. Roy. astr. Soc. 18,
1-32.
McKenzie,D.P.(1978)Some remarks on the development
ofsedimentary basins. Earth planet. Sci. Lett. 40,25-32.
McKenzie, D.P. (1983) The Earth's mantle. Sci. Am.
249(3), 50-67.
McKenzie, D.P.(1990)Spinning continents. Nature 344,
109-10.
McKenzie, D.P. & Bickle, M.J. (1988) The volume and
composition of melt generated by extension ofthe litho
sphere./. Petrol. 29,625-79.
McKenzie,D.P.& Morgan,W.J.(1969)Evolution oftriple
junctions. Nature 224,125-33.
McKenzie, D.P. & Parker, R.L.(1967) The North Pacific:
an example oftectonics on a sphere. Nature 216,1276-
80.
McKenzie,D.P.& Sclater, J.G.(1971)The evolution ofthe
Indian Ocean since the late Cretaceous. Geophys. J. Roy.
astr. Soc. 24,437-528.
McKenzie, D.P. & Weiss, N.(1975) Speculations on the
thermal and tectonic history ofthe earth. Geophys.J. Roy.
astr. Soc. 42,131-74.
McKenzie, D.P., Watts, A., Parsons, B. & Roufosse, M.
(1980) Planform of mantle convection beneath the
Pacific Ocean. Nature 288,442-6.
McKerrow, W.S. & Cocks, L.R.M.(1986) Oceans, island
arcs and olistostromes: the use offossils in distinguishing
sutures, terranes and environments around the lapetus
Occzn.J.geol. Soc. Lond. 143,185-91.
McKerrow, W.S., Leggett, J.K. & Eales, M.H. (1977)
Imbricate thrust model of the Southern Uplands of
Scotland. Nature 267, 237-9.
Maekawa,H.,Shozul, M.,Ishii, T., Fryer, P.& Pearce, J.A.
(1993) Blueschist metamorphism in an active subduc-
tion zone. Nature 364,520-3.
Martin, H.(1981) The late Palaeozoic Gondwana glacia-
tion. Geol. Rund. 70,480-98.
Mascle, J. & Blarez, E. (1987) Evidence for transform
margin evolution from the Ivory Coast-Ghana continen
tal margin. Nature 326,378-81.
Mason, R.(1985) Ophiolites. Geology Today 1,136-40.
Mason, R.G. & Raff, A.D.(1961) Magnetic survey off the
west coast of North America, 32®N latitude to 42"N
latitude. Bull.geol. Soc. Am. 72,1259-66.
Mattauer, M.(1986) Intracontinental subduction, crust-
mantle decollement and crustal-stacking wedge in the
Himalayas and other collision belts. In Coward, M.P. &
Ries, A.C.(eds) Collision Tectonics, pp. 37-50. Geol. Soc.
Spec. Pub. 19.
Matthews, P.E. (1972) Possible Precambrian obduction
and plate tectonics in southeastern Africa. Nature Phys.
Sci. 240, 37-9.
Maxwell, A.E., von Herzen, KP., Hsu, K.J. et al. (1970)
Deep sea drilling in the South Atlantic. Science 168,
1047-59.
Megard, F. & Philip, H.(1976) Plio-Quaternary tectono-
magmatic zonation and plate tectonics in the central
Andes. Earth planet. Sci. Lett. 33, 231-8.
Menard, H.W. (1964) Marine Geology of the Pacific.
McGraw-Hill, New York.
Menard, H.W.& Atwater, T.(1968) Changes in direction
ofsea floor spreading. Nature 219,463-7.
Menard, H.W. & Atwater, T. (1969) Origin of fracture-
zone topography. Nature 222,1037-40.
Milanovsky,E.E.(1981)Aulacogens on ancient platforms:
problems of their origin and tectonic development. Tec
tonophysics 73,213-48.

Minster, J.B. & Jordan, T.H. (1978) Present-day plate
motions,J.£feophys. Res. 83,5331-54.
Mitchell, A.H.G. & Garson, M.S.(1976) Mineralization at
plate boundaries. Sci. Efi0ineerin£f 8,129-69.
Mitchell, A.H.G. & Garson, M.S.(1981)Mineral Deposits
and Global Tectonic Settinpfs. Academic Press, London.
Mitchell, A.H.G. & Reading, H.G. (1969) Continental
margins,geosynclines,and ocean floor spreading./. Geol.
77,629-46.
Mitchell, A.H.G. & Reading, H.G.(1986) Sedimentation
and tectonics. In Reading, H.G.(ed.) Sedimentary Envi
ronments and Fades,, pp.471-519. Blackwell Scientific
Publications, Oxford.
Miyashiro, A.(1972) Metamorphism and related magma-
tism in plate tectonics. Am.J. Sci. 272,629-56.
Miyashiro, A. (1973) Paired and unpaired metamorphic
belts. Tectonophysics 17,241-54.
Miyashiro, A.(1986) Hot regions and the origin of mar
ginal basins in the western Pacific. Tectonophysics 122,
195-216.
Molnar,P.(1984)Structure and tectonics ofthe Himalaya.
Ann, Rev. Earth planet. Sci. 12,489-518.
Molnar, P. (1988) Continental tectonics in the aftermath
of plate tectonics. Nature 335, 131-7.
Molnar, P. & Atwater, T.(1973) Relative motion of hot
spots in the mantle. Nature 246, 248-91.
Molnar,P.& Chen,W.-P.(1982)Seismicity and mountain
building.In Hsu,K.G.(ed.)Mountain Building Processes,,
pp.41-57. Academic Press, London.
Molnar, P.& Stock,J.(1987) Relative motions of hotspots
in the Pacific. Atlantic and Indian Oceans since late
Cretaceous time. Nature 327, 587-91.
Molnar, P. & Tapponier, P.(1975) Cenozoic tectonics of
Asia: effects of a continental collision. Science 189,419-
26.
Molnar,P.,Freedman, D.& Shih,J.S.F.(1979)Lengths of
intermediate and deep seismic zones and temperatures in
downgoing slabs oflithosphere. Geophys.J. Roy. astr. Soc.
56,41-54.
Montadert L. et al. (1979) Initial Reports of the Deep Sea
Drilling Project Vol. 48. US Government Printing Office,
Washington DC.
Montagner, J.-P. (1994) Can seismology tell us anything
about convection in the mantle.> Rev. Geophys. 32, 115-
37.
Moody, J.D. (1973) Petroleum exploration aspects of
wrench-fault tectonics.Aw.Assoc. Pet. Geol. 57,449-76.
Moore, J.C.(1989) Tectonics and hydrogeology of accre-
tionary prisms: role of decollement zone.J. struct. Geol.
11,95-106.
Moore, J.C. & Silver, E.A. (1987) Continental mafmn
tectonics: submarine accretionary prisms. Rev. Geophys.
25,1305-12.
Moores, E.M.(1982) Origin and emplacement of ophio-
REFERENCES 315
lites. Rev. Geophys. Space Phys. 20,737-60.
Moores, E.M. (1991) South west U.S.-East Antarctic
(SWEAT)connection: a hypothesis. Geolopfy 19,325-8.
Morel, P. & Irving, E.(1978) Tentative paleocontinental
maps for the early Phanerozoic and Proterozoic./. Geol.
86,535-61.
Morgan,J.P.& Chen,Y.J.(1993)Dependence ofridge-axis
morphology on magma supply and spreading rate.
Nature 364, 706-8.
Morgan, J.P. & Shearer, P.M.(1993) Seismic constraints
on mantle flow and topography of the 660-km
discontinuity: evidence for whole-mantle convection.
Nature 36S, 506-11.
Morgan, W.J. (1968) Rises, trenches, great faults and
crustal blocks,J.£feophys. Res. 73,1959-82.
Morgan, W.J. (1971) Convection plumes in the lower
mantle. Nature 230,42-3.
Morgan, W.J. (1972a) Plate motions and deep mantle
convection. Geol. Soc. Am. Mem. 132,7-22.
Morgan, W.J. (1972b) Deep mantle convection plumes
and plate motions. Bull. Am. Assoc. Petroleum Geols. 56,
203-13.
Morgan, W.J. (1981) Hot spot tracks and the opening of
the Atlantic and Indian Oceans.In Emiliani, C.(ed.) The
Oceanic Lithosphere. The Sea 7, pp. 443-87. Wiley, New
York.
Morgan,W.J.(1983)Hotspot tracks and the early rifting of
the Atlantic. Tectonophysics 94,, 123-39.
Mount,V.S.& Suppe,J.(1987)State ofstress near the San
Andreas Fault: implications for wrench tectonics. Geoloigy
15,1143-6.
Mueller,S. & Landisman, M.(1966)Seismic studies ofthe
Earth's crust in continents 1: evidence for a low velocity
zone in the upper part ofthe lithosphere. Geophys.J. Roy.
astr, Soc. 10, 525-38.
Mutter,J.C., Barth, G.A., Buhl, P., Derrick, RS.,Orcutt,J.
& Harding, A.(1988a)Magma distribution across ridge-
axis discontinuities on the East Pacific Rise from multi
channel seismic images. Nature 336,156-8.
Mutter, J.C., Buck, W.R & Zehnder, C.M.(1988b) Con-
vective partial melting. 1. A model for the formation of
thick basaltic sequences during the initiation of spread
ing. Res. 93,1031-48.
Myers,J.S.(1975) Vertical crustal movements ofthe Andes
in Peru. Nature 254,672-4.
Nafe, J.E. & Drake, C.L. (1963) Physical properties of
marine sediments. In Hill, M.N.(ed.) The Earth beneath
the Sea. The Sea 3, pp. 794-815. Interscience Publishers,
New York & London.
Nakajima, T., Maruyama, S., Uchiumi, S. et al, (1990)
Evidence for late Proterozoic subduction from 700-Myr-
old blueschists in China. Nature 346, 263-5.
Nataf, H.C., Nakanishi, 1. & Anderson, D.L. (1984) An-
isotropy and shear-velocity heterogeneities in the upper
316 REFERENCES
mantle. Geophys. Bxs. Lett. 11,109-12.
Nataf, H.C.& VanDecar, J.(1993)Seismoiogial detection
of a mantle plume? Nature 364,115-20.
Neumann, E.R. & Ramberg, LB. (1978) Paleorifts—
concluding remarks.In Neumann,E.R.& Ramberg, LB.
(eds) Tectonics and Geophysics of Continental Rifts,
pp.409-24. D. Reidel, Dordrecht.
Nicolas, A.(1989) Structure of Ophiolites and Dynamics of
Oceanic Lithosphere. Kluwer Academic Publishers,
Dordrecht.
Nicolas, A., Boudier, F. & Ildefonse, B.(1994) Evidence
from the Oman ophiolite for active mantle upwelling
beneath a fast-spreading ridge. Nature 370,51-3.
Nicholson, C., Seeber, L., Williams, P. & Sykes, L.R.
(1986a) Seismic evidence for conjugate slip and block
rotation vv^ithin the San Andreas fault system, southern
California. Tectonics 5,629-48.
Nicholson, C.L., Seeber, L., Williams, P. & Sykes, L.R.
(1986b) Seismicity and fault kinematics through the
Eastern Transverse Ranges, southern California: block
rotation, strike-slip faulting and low-angle thrusts. J.
geophys. Res. 81,4891-908.
Niell, A.E., Ong, K.M., MacDoran, P.P. et at. (1979)
Comparison of a radiointerferometric differential base
line measurement with conventional geodesy. Tectono-
physics 52,49-58.
Nisbet, E.G. & Fowler, C.M.R.(1978) The mid-Atlantic
Ridge at 37 and 45'N:some geophysical and petrological
constraints. Geophys. J. Roy. astr. Soc. 54,631-60.
Nunns, A.G. (1983) Plate tectonic evolution of the
Greenland-Scotland Ridge and surrounding areas. In
Bott, M.H.P.,Sarno, V.S., Talwani, M.& Thiele, J.(eds)
Structure and Development of the Greenland-Scotland
Rid£fe, pp. 11-30. NATO Conference Series IV 8.
Plenum Press, London.
Nur, A. & Ben-Avraham, Z.(1982) Oceanic plateaus, the
fragmentation of continents and mountain building. J.
geophys. Res. 87,3644-61.
O'Connell, R.J. & Budiansky, B.(1977) Viscoelastic pro
perties of fluid-saturated cracked solids. J. pfeophys. Res.
82,5719-35.
Olivei, J. (1982) Tracing surface features to great depths:
a powerful means for exploring the deep crust. Tectono-
physics 81,2S7-72.
Oliver, J. & Isacks, B. (1967) Deep earthquake zones,
anomalous structures in the upper mantle, and the litho
sphere./. Res. 72,4259-75.
Olson,P. & Nam,I.S.(1986)Formation of seafloor swells
by mantle iphimts.J.geophys. Res. 91,7181-91.
O'Nions, R.K.(1987)Relationships between chemical and
convective layering in the Earth./. Soc. Lmd. 144,
259-74.
Open University Course Team (1989) The Ocean Basins:
their Structure and Evolution. Pergamon Press, Oxford.
Opdyke, N.D., Glass, B., Hays, J.D. & Foster, J. (1966)
Palaeomagnetic study of Antarctic deep sea cores. Science
154,349-57.
Opdyke, N.D., Burckle, L.H. & Todd, A. (1974) The
extension of the magnetic time scale in sediments of the
central Pacific Ocean. Earth planet. Sci. Lett. 22,300-6.
Orcutt, J.A., Kennett, B.L.N. & Dorman, L.M. (1976)
Structure of the East Pacific Rise from an ocean bottom
seismometer survey. Geophys. J. Roy. astr. Soc. 45, 305-
20.
Orowan, E. (1965) Convection in a non-Newtonian
mantle, continental drift, and mountain building. PhU.
Trans. Roy. Soc. Lond. 258A,284-313.
Oxburgh, E.R. (1972) Flake tectonics and continental
collision. Nature 239, 202-4.
Oxburgh,E.R.& Turcotte, D.L.(1974) Membrane tecton
ics and the East African Rift. Earth planet. Sci. Lett. 22,
133-40.
Packham, G.H. & Falvey, D.A.(1971) An hypothesis for
the formation of marginal seas in the Western Pacific.
Tectonophysics 11, 79-110.
Pakiser, L.C. (1963) Structure of the crust and upper
mantle in the western United States.].geophys. Res. 68,
5747-56.
Park, RG. (1983) Foundations of Structural Geology.
Blackie, London and Glasgow.
Park, RG.(1988) Geological Structures and Moving Plates.
Blackie, London and Glasgow.
Parker,E.N.(1969)The occasional reversal ofthe geomag
netic field. Astrophys. J. 158,815-27.
Parmentier, E.M.,Turcotte, D.L. & Torrance, K.E.(1976)
Studies of finite amplitude non-Newtonian thermal con
vection with application to the Earth's mantle./.geophys.
Res. 81,1839-46.
Parsons, B. & McKenzie, D.P.(1978) Mantle convection
and the thermal structure ofthe plates./,geophys. Res. 83,
4485-96.
Parsons, B. & Sclater, J.G. (1977) An analysis of the
variation of ocean floor bathymetry and heat flow with
2Lge.J.geophys. Res. 82,803-27.
Pavoni, M.(1993) Pattern of mantle convection and Pan-
gaea break-up, as revealed by the evolution ofthe African
p\2Xc.J.geol. Soc. Lond. 150,953-64.
Pearce, J.A. (1980) Geochemical evidence for the genesis
and eruptive setting oflavas from Tethyan ophiolites. In
Panayiotou, A.(ed.) Ophiolites, pp. 261-72. Geol. Surv.
Cyprus.
Peltier, W.R (1985) Mantle convection and viscoelasticity.
Ann. Rev. Fluid Mech. 17,561-608.
Peltier, W.R & Andrews, J.T. (1976) Glacial-isostatic
adjustment—1. The forward problem. Geophys. J. Roy.
astr. Soc. 46,605-46.
Perry, F.V., Baldridge, W.S. & De Paolo, D.J. (1988)
Chemical and isotopic evidence for lithospheric thinning

beneath the Rio Grande Rift. Nature 332,432-4.
Phillips, R.J. & Ivins, E.R. (1979) Geophysical observa
tions pertaining to solid state convection in the terrestrial
planets. Phys. Earth planet. Interiors 19, 107-48.
Phillips, W.E.A., Stillman, C.J. & Murphy. T. (1976) A
Caledonian plate tectonic model. Sac. Land. 132,
579-609.
Pigram, C.J. & Davies, H.L. (1987) Terranes and the
accretion history of the New Guinea orogen. B.M.R.J.
Austral.£feol. & Geophys. 10,193-211.
Pilger, R.H. Jr (1982) The origin of hotspot traces: evi
dence from eastern Australia./. Res. 87, 1825-
34.
Piper, J.D.A. (1982) The Precambrian palaeomagnetic
record: the case for a Proterozoic supercontinent. Earth
planet. Sci. Lett. 59,61-89.
Piper, J.D.A.(1987)Palaeomapfnetism and the Continental
Crust. Open University Press, Milton Keynes.
Pitcher, W.S.(1978) The anatomy of a batholith. J. ^eol.
Soc. Lond. 135,157-82.
Pitcher, W.S.,Atherton, M.P.,Cobbing,E.J. & Beckinsale,
KD.(1985). Ma£jmatism at a Plate Edpfe: the Peruvian
Andes. Blackie, London and Glasgow.
Pitman, W.C.Ill & Heirtzler, J.R.(1966)Magnetic anom
alies over the Pacific-Antarctic ridge. Science 154,1164-
71.
Pitman,W.C.Ill & Talwani, M.(1972)Sea-floor spreading
in the North Atlantic. Bull.pfeol. Soc. Am. 83,619-46.
Plank,T.& Langmuir,C.H.(1993)Tracing trace elements
from sediment input to volcanic output at subduction
zones. Nature 362, 739-43.
Platt, J.P. (1986) Dynamics of orogenic wedges and the
uplift of high-pressure metamorphic rocks. Bull.pfeol. Soc.
Am. 97,1037-53.
Plumstead,E.P.(1973)The enigmatic Glossopteris flora and
uniformitarianism. In Tarling, D.H. & Runcorn, S.K.
(eds) Implications of Continental Drift to the Earth Sci
ences., 1, pp.413-24. Academic Press, London.
Pollack, H.N. & Chapman, D.S.(1977a) The flow of heat
from the Earth's interior. Sci. Am. 237(2),60-76.
Pollack, H.N. & Chapman, D.S.(1977b) On the regional
variation of heat flow,geotherms, and lithospheric thick
ness. Tectonophysics 38,279-96.
Pollack, H.N., Hunter, S.J. & Johnson, J.R. (1993) Heat
flow from the Earth's interior: analysis of the global data
set. Rev. Geophys. 31,267-80.
Powell, C.McA. & Conaghan, P.J. (1973) Plate tectonics
and the Himalayas. Earth planet. Sci. Lett. 20,1-12.
Powell,C.McA.& Conaghan,P.J.(1975)Tectonic models
ofthe Tibetan Plateau. Geology 20,727-31.
Powell, C.McA., Johnson, B.D. & Veevers, J.J. (1980) A
revised fit ofeast and west Gondwanaland. Tectonophysics
63,13-29.
Purdy, G.M. (1987) New observations of the shallow
REFERENCES 317
seismic structure of young oceanic crust. J.geophys. Res.
92,9351-62.
Puzyrev, N.N., Mandelbaum, M.M.,Krylov, S.V., Mishen-
kin, B.P., Petrik, G.V. & Krupskaya, G.V.(1978) Deep
structure of the Baikal and other continental rift zones
from seismic data. Tectonophysics 45, 15-22.
Quennell, A.M. (1958) The structural and geomorphic
evolution of the Dead Sea Rift. J. geol. Soc. Lond. 14,
1-19.
Quennell, A.M.(1985)The Western Arabia rift system.In
Dixon, J.E. & Robertson, A.H.F. (eds) The Geological
Evolution of the Eastern Mediterranean. Spec. Pub. geol.
Soc. Lond. 17, 775-88.
Raff, A.D. & Mason, R.G.(1961) Magnetic survey off the
west coast of North America, 40°N latitude to 52°N
latitude. Bull.geol. Soc. Am. 72,1267-70.
Ramos, V.A. (1988) The tectonics of the central Andes;
30° to 33°S latitude. In Clark, S.P., Burchfield, B.C.
& Suppe, J. (eds) Processes in Continental Lithospheric
Deformation. Geol. Soc. Am. Spec. Paper 218, 31-54.
Ranalli, G.(1987) Rheology of the Earth. Allen & Unwin.
London.
Ranalli, G. & Murphy, D.C.(1987) Rheological stratifica
tion ofthe lithosphere. Tectonophysics 132, 281-96.
Reading, H.G. (1980) Characteristics and recognition of
strike-slip fault systems. In Ballance, P.P. & Reading,
H.G. (eds) Sedimentation in Oblique Slip Mobile Zones.
Spec. Pub. int. Ass. Sediment. 4,7-26.
Reck,B.H.(1987)Implications of measured thermal gradi
ents for water movement through the northeast Japan
accretionary prism.J.geophys. Res. 92,3683-90.
Reeves, C.(1976) The delineation of crustal provinces in
southern Africa from a compilation of gravity data. 20th
Ann.Rep. Research Institute ofAfrican Geology., University
of Leeds.
Renkin, M.L. & Sclater, J.G.(1988) Depth and age in the
north Vzcific.J.geophys. Res. 93,2919-35.
Richter, P.M. (1973) Convection and large-scale circu
lation of the mdmtic.J.geophys. Res. 78,8735-45.
Richter, P.M. & Parsons, B.(1975) On the interaction of
two scales ofconvection in the mznxX^.J.geophys. Res. 80,
2529-41.
Ringwood, A.E. (1974) The petrological evolution of is
land arc systems./.geol. Soc. Lond. 130, 183-204.
Ringwood, A.E. (1975) Composition and Petrology of the
Earth's Mantle. McGraw-Hill, New York.
Ringwood, A.E.(1977)Petrogenesis in island arc systems.
In Talwani, M. & Pitman, W.C. Ill (eds) Island Arcs,
Deep Sea Trenches and Back-arc Basins, pp. 311-24. Mau
rice Ewing Series 1. Am. Geophys. Union.
Ringwood,A.E.&Irifune,T.(1988)Nature ofthe 650 km
seismic discontinuity: implications for mantle dynamics
and differentiation. Nature 331, 131-6.
Ringwood,A.E.& Major,A.(1970)The system Mg2Si04-
318 REFERENCES
Fe2Si04 at high pressures and temperatures. Phys. Earth
planet. Interiors 3,89-108.
Ritger, S., Canson, B. & Snegge, E. (1987) Methane-
derived authigenic carbonates formed by subduction-
induced pore-water expulsion along the Oregon/
Washington margin. Bull.pfeol. Soc. Am. 98,147-56.
Roecker,S.W.(1982)The velocity structure of the Pamir-
Hindu Kush region: possible evidence of subducted
cmst.J.£feophys. Res. 87,945-59.
Rom, J. (1994) A new forerunner for continental drift.
Nature 367,407-8.
Rona, P.A. (1977) Plate tectonics, energy and mineral
resources: basic research leading to payoff. Eos {Trans.
AG.C/.)58,629-39.
Rona,P.A.(1984)Hydrothermal mineralization at seafloor
spreading centres. Earth Sci. Rev. 20,1-104.
Runcorn, S.K. (1962) Palaeomagnetic evidence for conti
nental drift and its geophysical cause. In Runcorn, S.K.
(ed.) Continental Drift., pp. 1-40. Academic Press, New
York & London.
Rupke, N.A.(1970) Continental drift before 1900. Nature
227, 349-50.
Sabadini, R.& Yuen,D.A.(1989)Mantle stratification and
long-term polar wander. Nature 339, 373-5.
Sabadini, R.,Yuen,D.A.& Boschi,E.(1982)Polar wander
ing and the forced responses ofa rotating, multi-layered,
viscoelastic p\2Sia..J.£feophys. Res. 81,2885-903.
Sager, W.W.& Bleil, U.(1987) Latitudinal shift of Pacific
hotspots during the late Cretaceous and early Tertiary.
Nature 326,488-90.
Salisbury, M.H. & Christensen, N.I. (1978) The seismic
velocity structure ofa traverse through the Bay ofIslands
ophiolite complex, Newfoundland, an exposure of oce
anic crust and upper m2int\t.J.£ieophys. Res. 83,805-17.
Sandwell, D.T. & Renkin, M.L.(1988) Compensation of
swells and plateaus in the north Pacific: no direct evidence
for mantle con\cction.J.£ieophys. Res. 93,2775-83.
Sawkins, F.J.(1976) Widespread continental rifting: some
considerations of timing and mechanism. Geok(gy 4,
427-30.
Sawkins, F.J. (1984) Metal Deposits in Relation to Plate
Tectonics. Springer-Verlag, Berlin.
Schaer, J.-P. & Rodgers, J. (eds)(1987) The Anatomy of
Mountain Ranpfes. Princeton University Press, Princeton,
NJ.
Schilling, J.-G., Anderson, R-N. & Vogt. P. (1976) Rare
earth, Fe and Ti variations along the Galapagos spreading
centre and their relationship to the Galapagos mantle
plume. Nature 261,108-13.
Scholz, C.H., Barazangi, M. & Sbar, M.L. (1971) Late
Cenozoic evolution of the Great Basin, western United
States, as an ensialic inter-arc basin. Bull, pfeol. Soc. Am.
82, 2979-90.
Schoonmaker,J.(1986) Clay mineralogy and diagenesis of
sediments from deformation zones in the Barbados accre-
tionary prism.In Moore,J.C.(ed.) Synthesis ofStructural
Fabrics in Deep Sea Drillinpf Project Coresfrom Forearcs.
Mem.^eol. Soc. Am. 166,105-16.
Schubert, G., Yuen, D.A. & Turcotte, D.L.(1975) Role of
phase transitions in a dynamic mantle. Geophys. J. Roy.
astr. Soc. 42,705-35.
Sclater, J.G. & Francheteau, J.(1970) The implications of
terrestrial heat flow observations on current tectonic and
geochemical models ofthe crust and upper mantle ofthe
Earth. Geophys. J. Roy. astr. Soc. 20,509-42.
Sclater, J.G.,Jaupart,C.& Galson,D.(1980)The heat flow
through oceanic and continental crust and the heat loss
of the Earth. Rep. Geophys. Space Phys. 18,269-311.
Scotese, C.R., Gahagan, L.M.& Larson, R.L.(1988) Plate
tectonic reconstructions ofthe Cretaceous and Cenozoic
ocean basins. Tectonophysics 155, 27-48.
Scott, M.R., Scott, R.B., Nalwalk, A.J., Rona, P.A. &
Butler, L.W. (1973) Hydrothermal manganese in the
median valley of the mid-Atlantic Ridge. Eos (Trans.
A.G.U.) 54,244.
Scrutton, C.T.(1967)Absolute time data from palaeontol
ogy.In Runcorn,S.K.etal.(eds)International Dictionary
of Geophysics I., p. 1. Pergamon Press, Oxford.
Scrutton, R-A.(1979)On sheared passive continental mar
gins. Tectonophysics 59,293-305.
Searle, D.L. & Panayiotou, A.(1980) Structural implica
tions in the evolution ofthe Troodos massif, Cyprus.In
Panayiotou, A. (ed.) Ophiolites^ pp. 50-60. Geol. Surv.
Cyprus.
Searle, R.C. (1970) Evidence from gravity anomalies for
thinning of the lithosphere beneath the rift valley in
Kenya. Geophys. J. Roy. astr. Soc. 21,13-31.
Searle, R.C. (1983) Multiple, closely spaced transform
faults in fast-slipping fracture zones. Geolqgy 11,607-10.
Searle, R.C., Rusby, R.L,Engein, J. et al. (1989) Compre
hensive sonar imagery of the Easter microplate. Nature
341,701-5.
Seely, D.R,(1979) The evolution of structural highs bor
dering major forearc basins.In Watkins,J.S., Montadert,
L. & Dickerson, P.W. (eds) Geological and Geophysical
Investigations of Continental Margins., pp. 245-60.
A.A.P.G. Mem. 29y Am. Assoc. Petroleum Geols. Tulsa,
Oklahoma.
Sempere, J.-C., Purdy, G.M. & Schouten, H.(1990) Seg
mentation ofthe Mid-Atlantic Ridge between 24®N and
30AO'N. Nature 344,427-9.
§eng6r, A.M.C.(1976) Collision of irregular continental
margins: implications for foreland deformation of
Alpine-type orogens. Geolo£fy 4,779-82.
Shackleton, R.M., Dewey, J.F. & Windley, B.F. (1988)
Tectonic evolution of the Himalayas and Tibet. Royal
Society, London.
Shanti, M. & Roobol, M.J. (1979) A late Proterozoic

ophiolite complex at Jabal Ess in northern Saudi Arabia.
Nature 279,488-91.
Shearer, P.M. (1990) Seismic imaging of upper-mantle
structure with new evidence for a 520-km discontinuity.
Nature 344,121-6.
Shearer, P.-M.& Masters, G.M.(1992)Global mapping of
topography on the 660-km discontinuity Nature 355,
791-6.
Shermer, E.R., Howell, D.G. & Jones, D.L. (1984) The
origin of allocthonous terranes: Perspectives on the
growth and shaping of continents. Ann. Rev. Earth
planet. Sci. 12,107-31.
Sillitoe, R-H.(1972a)Formation ofcertain massive sulphur
deposits at sites of sea-floor spreading. Trans. Inst. Min.
Metall. 81,B141-8.
Sillitoe, R.H.(1972b)A plate tectonic model for the origin
of porphyry copper deposits. Econ. Geol. 67,184-97.
Silver, E.A. & Reed, D.L. (1988) Backthrusting in accre-
tionary wedges. Res. 93,3116-26.
Silver, E.A. & Smith, R. (1983) Comparison of terrane
accretion in modern Southeast Asia and the Mesozoic
North American Cordillera., Geology 11,198-202.
Silver, E.A., Reed, D., McCaffrey, R. 8c Yoyodiwiryo, Y.
(1983) Back-arc thrusting in the eastern Sunda arc,
Indonesia; a consequence of arc-continental collision./.
geophys. Res. 88,7429-48.
Simpson,R,W.8c Cox,A.(1977)Palaeomagnetic evidence
for tectonic rotation ofthe Oregon Coast Range. Geology
5, 585-9.
Sinton, J.M. 8c Detrick, R.S. (1992) Mid-ocean ridge
magma ch^sahtxs.J.geophys. Res. 97,197-216.
Sleep, N.H. (1971) Thermal effects of the formation of
Atlantic continental margins by continental break-up.
Geophys.J. Roy astr. Soc. 24,325-50.
Sleep, N.H. (1975) Formation of oceanic crust: some
thermal constrzints.J.get^hys. Res. 80,4037-42.
Sleep, N.H. 8c Snell, N.S.(1976)Thermal contraction and
flexure of mid-continent and Atlantic marginal basins.
Geophys.J. Roy. astr. Soc. 45,125-54.
Smith, A.G. 8c Hallam, A.(1970) The fit of the southern
continents. Nature 225,139-44.
Smith, A.G., Hurley,A.M.8c Briden, J.C.(1981)Phanero-
zoic Paleocontinental World Maps. Cambridge University
Press, Cambridge.
Smith, D.E., Christodoulidis, D.C., Kolenkiewicz, K,
Dunn,P.J., Torrence, M.H. 8c Klosko,S.(1985) Global
plate motion results from satellite laser ranging (ab
stract). Eos (Trans. A.G.U.)66,848.
Smith, D.K. 8c Cann, J.R.(1993) Building the crust at the
Mid-Atlantic Ridge. Nature 365,707-15.
Smithson, S.B. 8c Brown, S.K. (1977) A model for the
lower continental crust. Earth planet. Sci. Lett. 35,134-
44.
Solomon, M. (1990) Subduction, arc reversal, and the
REFERENCES 319
origin of porphyry copper-gold deposits in island arcs.
Geology 18,630-3.
Solomon, S.C., Sleep, N.H. 8c Richardson, R.M.(1975)
On the forces driving plate tectonics: inferences from
absolute plate velocities and intraplate stress. Geophys. J.
Roy. astr. Soc. 42,769-801.
Spada, G., Ricard, Y. 8c Sabadini, R.(1992) Excitation of
true polar wander by subduction. Nature 360,452-4.
Spence, W.(1987) Slab pull and the seismotectonics of
subducting lithosphere. Rev. Geophys. 25,55-69.
Spohn, T. 8c Schubert, G.(1982) Convective thinning of
the lithosphere: a mechanism for the initiation of conti
nental rifting.J.geophys. Res. 87,4669-81.
Spudich, P. 8c Orcutt, J.(1980) A new look at the seismic
velocity structure ofthe oceanic crust. Rev. Geophys. Space
Phys. 18,627-45.
Stauder, W. (1968) Mechanism of the Rat Island earth
quake sequence of February, 1965, with reference to
island arcs and sea-floor spreading. J. geophys. Res. 73,
3847-58.
Stauffer, M.R., Mukherjee, A.G. 8c Koo, J. (1975) The
Amisk Group: an Aphebian (.>) island arc deposit. Can.J.
Earth Sci. 12, 2021-35.
Steckler, M.S.8c Watts,A.B.(1980)Subsidence ofa young
continental margin. Nature 287,425-9.
Stein, M. 8c Hofmann, A.W.(1994) Mantle plumes and
episodic crustal growth. Nature 372,63-8.
Stein, R.S.(1987) Contemporary plate motion and crustal
deformation. Rev. Geophys. 25,855-63.
Stewart, J.A. (1990) Drifting Continents and Colliding
Paradigms: Perspectives on the Geoscience Revolution.
Indiana University Press, Bloomington.
Stolper, E. 8c Newman,S.(1994) The role of water in the
petrogenesis of Mariana Trough magmas. Earth planet.
Sci. Lett. 121, 293-325.
Storey,B.C.(1995)The role of mantle plumes in continen
tal breakup: case histories from Gondwanaland. Nature
377,301-8.
Storey, B.C.(1993)The changing face oflate Precambrian
and early Palaeozoic reconstructions. J. geol. Soc. Lond.
150,665-8.
Stothers, R.B. (1986) Periodicity of the Earth's magnetic
reversals. Nature 322,444-6.
Stothers, R.B. (1993) Hotspots and sunspots: surface
tracers of deep mantle convection in the Earth and Sun.
Earth planet. Sci. Lett. 116,1-8.
Stow, D.A.V. 8c Lovell, J.P.B. (1979) Contourites: their
recognition in modern and ancient sediments. Earth Sci.
Rev. 14,251-91.
Snares, G., Molnar,P. 8c Burchfiel, B.C.(1983)Seismicity,
fault plane solutions, depth offaulting,and active tecton
ics of the Andes of Peru, Ecuador, and southern Colom
bia./. Res. 88,10 403-28.
Sun, S.S., Nesbitt, R.W. 8c Sharashin, A.Y. (1979)
320 REFERENCES
Geochemical characteristics of mid-ocean ridge basalts.
Earth planet. Sci. Lett. 44,119-38.
Sutton,J.& Watson,J.V.(1974)Tectonic evolution ofcon
tinents in early Proterozoic times. Nature 247,433-5.
Sykes, L.R. (1966) The seismicity and deep structure of
island zrcs.J.^reophys. Res. 71, 2981-3006.
Sykes, L.R.(1967) Mechanism of earthquakes and nature
offaulting on the mid-oceanic ridges./.£feophys. Res. 72,
2131-53.
Sykes, L.R. (1978) Intra-plate seismicity, reactivation of
pre-existing zones of weakness, alkaline magmatism and
other tectonism post-dating continental fragmentation.
Rev. Geophys. Space Rhys. 16,621-88.
Tackley, P.J., Stevenson D.J., Glatzmaier, G.A. & Shubert,
G.(1993) Effects of an endothermic phase transition at
670 km depth in a spherical model of convection in the
Earth's mantle. Nature 361,699-704.
Takagi, H.(1986) Implications of mylonitic microstruc-
tures for the geotectonic evolution of the Median Tec
tonic Line, central Japan./, struct. Geol. 8,3-14.
Talwani, M. & Watts, A.B.(1974) Gravity anomalies sea
ward ofdeep-sea trenches and their tectonic implications.
Geophys. J. Roy. astr. Sac. 36, 57-90.
Talwani, M., Le Pichon, X. & Ewing, M.(1965) Crustal
structure of the mid-ocean ridges 2. Computed model
from gravity and seismic reduction data./.£feophys. Res.
70, 341-52.
Tamaki, K. (1985) Two modes of back-arc spreading.
Geolqgy 13,475-8.
Tanimoto, T. & Anderson, D.L.(1985) Lateral heteroge
neity and azimuthal anisotropy ofthe upper mantle: Love
and Rayleigh waves \00-2S0s.J.£feophys. Res. 90,1842-
58.
Tapley, B.D., Schutz, B.F. & Eanes, R.J. (1985) Station
coordinates, baselines,and Earth rotation from LAGEOS
laser ranging: 1976-1984. Res. 90,9235-48.
Tapponier, P. & Molnar, P. (1976) Slip-line field theory
and large-scale continental tectonics. Nature 264, 319-
24.
Tapponier, P., Peltzer, G., Le Dain, A.Y., Armijo, R. &
Cobbing, P. (1982) Propagating extrusion tectonics in
Asia: new insights from simple experiments with plas-
ticene. Geolo£fy 10,611-16.
Tapponier,P., Lacassin, R.,Leloup,P.H.etal.(1990)The
Ailao Shan/Red River metamorphic belt: Tertiary left-
lateral shear between Indochina and South China. Nature
343,431-7.
Tarling, D.H.(ed)(1981)Economic Geolqgy and Geotecton-
ics. Blackwell Scientific Publications. Oxford.
Tarling, D.H.(1983)Palaeomagnetism. Chapman & Hall,
London.
Tarling,D.H.& Tarling, M.P.(1971)Continental Drift:A
Study ofthe Earth^s Moving Surface. Bell, London.
Tarney, J. & Windley, B.F. (1981) Marginal basins
throughout geologic time. PhU. Trans. Roy. Soc. Lond.
301A, 217-31.
Tarney, J., Dalziel, 1. & de Wit, M.(1976) Marginal basin
*Rocas Verdes' complex from S. Chile: a model for
Archaean greenstone belt formation. In Windley. B.F.
(ed.) The Early History ofthe Earth., pp. 131-46. Wiley,
London.
Taylor, B., Crook, K. & Sinton, J. (1994) Extensional
transform zones and oblique spreading centers./.geophys.
Res. 99,19 707-18.
Taylor, B., Goodliffe, A., Martinez, F. & Hey, R.(1995)
Continental rifting and initial sea-floor spreading in the
Woodlark Basin. Nature 374,534-7.
Taylor, F.B.(1910) Bearing ofthe Tertiary mountain belt
on the origin of the Earth's plan. Bull.geol. Soc. Am. 21,
179-226.
Taylor, R.T. & Scott, M.M.(1985) The Continental Crust:
its Composition and Evolution. Blackwell Scientific Publica
tions, Oxford.
Thatcher, W.(1979)Systematic inversion of geodetic data
in central J.geophys. Res. 84,2283-95.
Thomas, M.D. & Kearey, P. (1980) Gravity anomalies,
block faulting and Andean-type tectonism in the eastern
Grenville Province. Nature 293,61-3.
Thompson, G. & Melson, W.G.(1972) The petrology of
oceanic crust across fracture zones in the Atlantic Ocean:
evidence ofa new kind ofsea-floor spreading./. Geol. 80,
526-38.
Thorpe, R.S.(ed.){\9%2)Andesites. Wiley, London.
Thurber,C.H.& Aid,K.(1987)Three-dimensional seismic
imaging. Rev. Earth planet. Sci. 15,115-39.
Tissot, B.(1979)Effects on prolific petroleum source rocks
and major coal deposits caused by sea-level changes.
Nature 277,463-5.
Toksoz,M.N.,Minear,J.W.& Julian,B.R.(1971)Temper
ature field and geophysical effects ofa downgoing slab./.
geophys. Res. 76,1113-38.
Toomey, D.R., Purdy, G.M., Solomon, S.C. & Wilcock,
W.S.D. (1990) The three-dimensional seismic velocity
structure of the East Pacific Rise near latitude 9®30'N.
Nature 347,639-45.
Tozer, D.C.(1972)The present thermal state ofthe terres
trial planets. Phys. Earth planet. Interiors 6,182-97.
Turcotte, D.L. & Oxburgh, E.R (1978) Intra-plate volca-
nism. PhU. Trans. Roy. Soc. Lon. 288A,561-79.
Turcotte,D.L.,McAdoo,D.C.& Caldwell,J.G.(1978)An
elastic-perfectly plastic analysis of the bending of the
lithosphere at a trench. Tectonophysics 47,193-205.
Uyeda,S.& Kanamori,H.(1979)Back-arc opening and the
mode ofsubduction./.^^(^^yj. Res. 84,1049-61.
Vacquier, V. (1965) Transcurrent faulting in the ocean
floor. Phil. Trans. Roy. Soc. Lond. 258A,77-81.
Valentine, J.W. & Moores, E.M.(1972) Global tectonics
and the fossil record./. Geol. 80,167-84.

Valentine,J.W.& Moores,E,M.(1974)Plate tectonics and
the history of the oceans. Sci. Am. 230(4), 80-9.
Van der Hilst, R. (1995) Complex morphology of sub
ducted lithosphere in the mantle beneath the Tonga
Trench. Nature 374,154-7.
Van der Voo, K (1990)Phanerozoic paleomagnetic poles
from Europe and North America and comparisons with
continental reconstructions. Bsv. Geophys. 28,167-206.
Van Orman, J., Cochran, J.R., Weissel, J.K. & Jestin, F.
(1995) Distribution of shortening between the Indian
and Australian plates in the central Indian Ocean. Earth
planet. Sci. Lett. 133, 35-46.
Vening Meinesz,F.A.(1951)A third arc in many island arc
areas. Koninkl. Nederlandsch. Akad. Wetensch. Proc. Ser.
B. 54,432-42.
Vermeersen, L.L.A. & Vlaar, N.J. (1993) Changes in the
Earth's rotation by tectonic movements. Geophys. Res.
Lett. 20,81-4.
Vine, F.J. (1966) Spreading of the ocean floor: new evi
dence. Science 154,1405-15.
Vine,F.J.(1977)The continental drift debate. Nature 266,
19-22.
Vine, F.J. & Matthews, D.H.(1963) Magnetic anomalies
over oceanic ridges. Nature 199,947-9.
Vink,G.E.(1982)Continental rifting and the implications
for plate tectonic reconstructions. J. geophys. Res. 87,
10677-88.
Vink, G.E., Morgan, W.J. & Zhao, W.-L.(1984)Preferen
tial rifting ofcontinents: a source ofdisplaced terrains./.
geophys. Res. 89,10 072-6.
Vogt, P.R. (1975) Changes in geomagnetic reversal fre
quency at times oftectonic change: evidence for coupling
between core and upper mantle processes. Earth planet.
Sci. Lett. 25, 313-21.
Vogt, P.R, Ostenso, N.A. & Johnson, G.L. (1970) Mag
netic and bathymetric data bearing on sea floor spreading
north ofIceland./,geophys. Res. 75,903-20.
von der Borch, C.(1979)Continent-island arc collision in
the Banda arc. Tectonophysics 54,169-93.
Walcott, R.1. (1972) Gravity, flexure, and the growth of
sedimentary basins at a continental edge. Bull, pfeol. Sac.
Aw. 83,1845-8. , £ieol. Soc. Lond. 10, 275-90.
Ward, M.A.(1963) On detecting changes in the Earth's
radius. Geophys. J. Roy. astr. Soc. 8,217-25.
Watanabe, T., Langseth, M.G. & Anderson, R.N.(1977)
Heat flow in back-arc basins of the western Pacific. In
Talwani, M.& Pitman, W.C. Ill (eds)Island Arcs, Deep
Sea Trenches and Back-arc Basinspp.137-61. Maurice
Ewing Series 1, Am. Geophys. Union.
Watt J.P. & Shankland ,T.J.(1975) Uniformity of mande
composition. Geolopfy 3,91-4.
Watters,B.R.(1976)Possible late Precambrian subduction
zone in south west Africa. Nature 259,471-3.
Watts, A.B. & Ryan, W.B.F.(1976) Flexure of the litho
REFERENCES 321
sphere and continental margin basins. Tectonophysics 36,
25-44.
Watts, A.B. & Talwani, M.(1974) Gravity anomalies sea
ward of deep-sea trenches and their tectonic significance.
Geophys.J. Roy. astr. Soc. 36, 57-102.
Watts, A.B., Cochran, J.R. & Selzer, G. (1975) Gravity
anomalies and flexure of the lithosphere: a three-
dimensional study ofthe Great Meteor seamount, north
east Atlantic. Res. 80,1391-8.
Watts, A.B., Bodine, J.H. & Steckler, M.S.(1980) Obser
vations of flexure and the state of stress in the oceanic
lithosphere. Res. 85,6369-76.
Weertman, J. (1978) Creep laws for the mantle of the
Earth. PhU. Trans. Roy. Soc. Lond. 288A,9-26.
Wegener, A. (1915) Die Entstehunpf der Kontinente und
Ozeane. Vieweg, Braunschweig.
Weissel,J.K.(1981)Magnetic lineations in marginal basins
ofthe western Pacific. Phil. Trans. Roy. Soc. Lond. 300A,
223-47.
Weissel, J.K., Anderson,R.N.& Geller, C.A.(1980)Defor
mation of the Indo-Australian Plate. Nature 287, 284-
91.
Weldon,R.& Humphries,E.(1986)A kinematic model of
southern California. Tectonics 5,33-48.
Wellman,P.(1978)Gravity evidence for abrupt changes in
mean crustal density at the junction of Australian crustal
blocks. B.M.R.J. Aust. Geol. Geophys. 3,153-62.
Wellman, P. (1988) Development of the Australian Pro-
terozoic crust as inferred from gravity and magnetic
anomalies. Precambrian Research 40/41,89-100.
Wernicke, B.(1981) Low angle normal faults in the Basin
and Range Province-nappe tectonics in an extending
orogen. Nature 291,645-8.
Wernicke, B. (1985) Uniform sense simple shear of the
continental lithosphere. Can.J. Earth. Sci. 22,108-25.
Wernicke, B. & Burchfiel, B.C. (1982) Modes of exten-
sional tectonism./. struct. Geol. 4,105-15.
Westbrook, G.K. (1982) The Barbados Ridge complex:
tectonics of a mature forearc system. In Leggett, J.K.
(ed.) Trench-Forearc Geology: Sedimentation and Tectonics
on Modem and Ancient Active Plate Marpfins. Spec. Pub.
Westbrook, G.K., Mascle, A. & Biju-Duval, B. (1984)
Geophysics and structure of the Lesser Antilles forearc.
In Biju-Duval, B. & Moore,J.C.(eds)Init. Repts. DSDP
78A, pp. 23-38. US Government Printing Office, Wash
ington DC.
Westbrook, G.K., Smith, M.J., Peacock, J.H. & Poulter,
M.J. (1982) Extensive underthrusting of undeformed
sediment beneath the accretionary complex of the Lesser
Antilles subduction zone. Nature 300,625-8.
White, G.W.(1980) Permian-Triassic continental recon
struction of the Gulf of Mexico-Caribbean area. Nature
283,823-6.
322 REFERENCES
White, N.(1989) Nature of lithospheric extension in the
North Sea. Geolqgy 17,111-14.
White,N.& McKenzie,D.(1988)Formation ofthe 'steer's
head' geometry of sedimentary basins by differential
stretching ofthe crust and mantle. Geolqgy 16,250-3.
White, R.(1993) Breaking up is hard to understand. New
Scientist^ 3 April, 25-9.
White, R. & McKenzie, D.P. (1989) Magmatism at rift
zones: the generation of volcanic continental margins
and flood basalts./.^eophys. Res. 94, 7685-729.
White, R.S., Spence, G.D., Fowler, S.R., McKenzie, D.P.,
Westbrook, G.K. & Bowen, A.N.(1987) Magmatism at
rifted continental margins. Nature 330,439-44.
Whiteman, A.J., Naylor, D., Pegrum, R.M. & Rees, G.
(1975) North Sea troughs and plate tectonics. Tectono-
physics 26, 39-54.
Whitmarsh, RB.(1975) Axial intrusion zone beneath the
median valley ofthe mid-Atlantic ridge at 37®N detected
by explosion seismology. Geophys. J. Roy. astr. Sac. 42,
189-215.
Whitmarsh,R.B., Avedik,F.& Saunders, M.R.(1986)The
seismic structure of thinned continental crust in the
northern Bay of Biscay. Geophys. J. Roy. astr. Soc. 86,
589-602.
Wiens, D.A., McGuire, J.J. & Shore, P.J.(1993) Evidence
for transformational faulting from a deep double seismic
zone in Tonga. Nature 364,790-3.
Wilson, D.S. (1993) Confirmation of the astronomical
calibration of the magnetic polarity timescale from sea-
floor spreading rates. Nature 364, 788-90.
Wilson,J.T.(1963)Evidence from islands on the spreading
ofocean floors. Nature 197, 536-8.
Wilson, J.T.(1965) A new class offaults and their bearing
on continental drift. Nature 207, 343-7.
Wilson, J.T. (1988) Convection tectonics: some possible
effects upon the Earth's surface of flow from the deep
mantle. Can.J. Earth Sci. 25,1199-1208.
Wilson, J.T. & Burke, K.(1972) Two types of mountain
building. Nature 239,448-9.
Wilson, M.(1989) Ipfneous Petropfenesis. A Global Tectonic
Approach. Unwin Hyman,London.
Windley, B.F.(1981)Precambrian rocks in the light of the
plate-tectonic concept. In Kroner, A. (ed.) Precambrian
Plate Tectonics., pp. 1-20. Elsevier, Amsterdam.
Windley, B.F.(1983) Metamorphism and tectonics of the
Himalaya./.^^o/. Soc. Lond. 140,849-65.
Windley, B.F. (1984) The Evolving Continents (2nd edn).
Wiley, London.
Windley, B.F. (1985) The Himalayas. Geolqgy Today 1,
169-73.
Windley, B.F. (1993) Uniformitarianism today: plate tec
tonics is the key to the p2L$t.J.pfeol. Soc. Lond. 150,7-19.
Woodcock, N.H. (1986) The role of strike-slip fault
systems at plate boundaries. Phil. Trans. Roy. Soc. Lond.
317A,13-29.
Woodcock, N.H. & Fischer, M. (1986) Strike-slip
duplexes./, struct. Geol. 8,725-35.
Woodhouse, J.H. & Dziewonski, A.M. (1984) Mapping
the upper mantle: three dimensional modelling of Earth
structure by inversion of seismic waveforms./. pfeophys.
Res. 89, 5953-86.
Wyllie, P.J. (1981) Plate tectonics and magma genesis.
Geol. Rundsch. 70(1), 128-53.
Wyllie, P.J. (1988) Magma genesis, plate tectonics and
chemical differentiation of the Earth. Rev. Geophys. 26,
370-404.
Yorath, C.J. & Chase, R.L.(1981) Tectonic history of the
Queen Charlotte Islands and adjacent areas—a model.
Can.J. Earth. Sci. 18,1717-39.
Young,G.M.(1992)Late Proterozoic stratigraphy and the
Canada-Australia connection. Geolopfy 20,215-18.
Young, J.B., Lilwall, R.C. & Douglas. A. (1988) WM
Seismicity Maps Suitable for the Study of Seismic and
Geographical Regionalisation. Atomic Weapons Research
Establishment Report 0 7/87. HMSO,London.
Yungul, S.H. (1976) Tectonic implications of gravity
anomalies associated with some continental rift
valleys—a model study. Tectonophysics 32, 321-9.
Zandt, G. & Ammon,C.J.(1995) Continental crust com
position constrained by measurements of crustal Pois-
son's ratio. Nature 374,152-4.
Ziegler, P.A.(1993) Plate-moving mechanisms: their rela
tive importance./.^^o/. Soc. Lond. 150,927-40.
Zeil, W.(1979) The Andes: A Geological Review. Gebriider
Borntraeger, Berlin.
Zhao,W.,Nelson,K.D.& Project INDEPTH team (1993)
Deep seismic reflection evidence for continental under-
thrusting beneath southern Tibet. Nature 366, 557-9.
Zoback, M.D., Zoback, M.L., Mount, V.S. et al. (1987)
New evidence on the state of stress of the San Andreas
fault system. Science 238,1105-11.
Index

aborted rift zone, cross-section 286 Rocas Verdes region 170, 171, 172
absolute plate motion 80,81-3 seismicity 175-7
Acadian Orogeny 269, 270 structure 174-5, 176
accommodation zone 215 section 177
accretion 280 andesites 291
accretionary prism 293 angle of rotation 46
subduction zones 155-60 animals, evidence for continental drift 52-4
accretionary tectonics 198 anisotropy
accretionary wedge 139, 155,235 azimuthal 268
actualism 1 seismic 267
Afar rift system, French Somaliland 224 anorthosite 20, 292
Africa Antarctica
African Shield 265 earthquake epicentres 94
APW paths 282 and North America,SWEAT hypothesis 277
Ivory Coast-Ghana, sheared continental margin 135-6 anticrack faulting 141-3
separation from North America 74 Antilles see Lesser Antilles
southwest, Precambrian subduction zone 284 Aphebian 283
sec also East African Rift Appalachian Fold Belt 218, 223
Airy mechanism 21 apparent polar wandering(APW)curves 57-60, 276-7,281-2
hypothesis of isostatic compensation 35, 38, 39 appendix^ geological and stratigraphical timescale 301
Alaska arc-continent collisional mountain ranges 195-7
Cook Inlet 293 Archaean
Wrangellia 199-200 granulite-gneiss belts 273, 274
Aleutian island arc greenstone belts 272-3, 274-5
Andean-type orogeni? belt 150,160 plate tectonic model 275
earthquakes tectonics 272-5
focal mechanism solutions 147 Archimedes' Principle 36
foci 150 aseismic ridge 235
types 141, 146 Asia
free air gravity anomaly profile 139 deforming zone 194
magnetic lineation 167 distribution of tectonic styles 193
marginal seas 165 extrusion (escape) tectonics 187-94
Aleutian Trench, subduction 167 suture between India and Eurasia 180-2, 293-4
Alps asthenosphere 40-2
foreland basins 245 Atlantic Ocean
InsubricLine 128 aulacogens 221
section, compared with Himalaya-type orogens 195 cratons, correlations 50
Amisk group, Manitoba 283 FAMOUS area 103-4
ammonites, distribution 53 island arc systems 160-3
andalusite 151 North Atlantic
Andaman Sea 293 acoustic stratigraphy 22
Andean-type magmatic belt, cross-section 286 Great Meteor Seamount, free air anomaly 37
Andean-type mountain ranges 174-80 mid-ocean ridges see Mid-Atlantic Ridge
porphyry coppers 289 plate reconstruction and hotspot frame of reference 84
Andean-type orogenic belt 150,160 Tertiary igneous province 226
Andean-type subduction zones 207, 274-5, 290 Vema Fracture Zone 23
schema 207 rate of movement across plates 92
Anderson theory of faulting 13, 31 reconstruction of continents around 47
Andes seismic reflectors 23
Altiplano 174, 175, 176,177 attenuation 9,41, 140, 169
Andean plate, and island arc, continental and oceanic crust attenuation factor 140
section 175 aulacogens 216-18,222,293,294
BeniofFzone 175 Atlantic Ocean 221
formation 204-5 Australia
Nazca plate 175, 177 Banda Sea arc-continent collision 196-8
negative mass anomaly 34 earthquake epicentres 94
plate tectonic model 177-80 mammal migrations 55
Rocas Verdes complex 273, 274 marginal seas 165

323
324 INDEX

Australia-Atlantic Ocean section 175 breakaway fault 243

auxiliary plane 10 brine pool 285, 292
Aves Ridge 171,172, 173 brittle deformation in mantle 30-1
azimuthal anisotropy 268 brittle-ductile transition 32, 36
Broad Fourteens Basin, Holland 246-7
bulk modulus 8
P factor 239-41 Burman arc 189
p-phase, spinel 28
back-arc basin sec basins
back-arc ridge 140, 163, 236 Cache-Creek terrain 199
Bahama Platform 48 calc-alkaline volcanics 162, 211, 273-5, 290
Baikal Rift 209-12 Caledonian Orogeny 160, 269, 270
Banda Sea California
arc-continent collision 196-8 California-Colorado section, seismic refraction 40
structural elements 197 generalized fault maps 130
Barbados accretionary prism, eastern Caribbean 159 Salton Trough, pull-apart basin 132, 292
Barbados Ridge 171, 172 transform faults 128-30
basalts Transverse Ranges 131,133
lower crust 20 see also North America;San Andreas Fault
magnetization, Koningsberger ratio 68 Cambrian, Late, supercontinent, reconstruction 278
Basin and Range Province, USA Canada
basin formation 241, 244 Labrador Trough 279-85
crustal thickness 39 see also North America
East Pacific Rise 170, 209 Canadian Shield, Proterozoic interactions 282, 283
gravity anomalies 40 carbonates, and climate 52
basins carbonatites 285
back-arc basins (marginal seas) 163-70, 203, 290 Caribbean Sea 48
ensialic 293 Barbados accretionary prism 159
evolution 167-70 Barbados Ridge 171,172
spreading 275 see also Lesser Antilles
continental margin basins 236-7 Carlsberg Ridge 265
development by creep 238 Cascade arc 157,159
fault wedge basins 132-3 cassiterite 292
forearc basin 139 cataclasis 31
foreland basins 244, 245-6 cataclastic flow 31
formation 236-45 catastrophism 1
far from plate margins 238 Central America 48
inter-arc basin, cross-section 286 chemical remanent magnetization(CRM) 56
intracratonic basins 239 Chilean Andes
mineralization 288 Rocas Verdes complex 273, 274
models of formation 236-44 Rocas Verdes region 170, 171,172
piggyback basins 246 China, blueschist facies 279
pull-apart basins 131-3, 292, 293 chron, polarity 69
thermal subsidence 240 Churchill Block, Hudsonian Orogeny 282
volcanosedimentary basins 273 climate see palaeoclimatology
see also sedimentary basins coal
batholiths 175,291 coalification 53, 294
bauxite 294 see also hydrocarbons
formation 52 Coble creep 33,34
Benioff zones 78, 139-46, 291 collage tectonics 198
Andes 175 collisions
diapirism 169 arc-continent collisions 195-7
distribution of downdip stresses 149 collisional mountain ranges 180-95
length, and product of spreading rate and age 153 continent-continent collisions 287, 290
Bering Sea continental with obducted ophiolites, cross-section 286
evolution 202 post-collision volcanism 287
model 205 compensation depth 35
magnetic lineations 167 composite terrain 198
underthrust 141,167 compressional wave 11
Besshi-type ores 290 conductivity 44
black smokers 288 confining pressure 30
blueschist fades 150-1,153 Conrad discontinuity 17, 18
China 279 continent(s)
body (seismic) waves 8,15 arc-continent collisions, mountain ranges 195-7
botanical evidence for continental drift 53, 54 continent-continent collisions 287, 290
Bouguer anomalies correlation of stratigraphy, Gondwana 51
Gregory Rift model 212 fit around Atlantic Ocean 47
Himalaya 184, 186 supercontinents 278-80
isostasy 34, 35, 38, 39,40 see also hotspots
Brazil, kimberlites 87 continental crust
Brazilian Shield 175,265 igneous activity 26
 INDEX 325

properties 19-21 Deccan Traps 234

simple shear models 242, 243 Deep Sea Drilling Project(DSDP)see Ocean Drilling Program
strength profiles 32 deformation in mantle
transition to oceanic crust 236-8 brittle 30-1
velocity range, lower 20 ductile 31-3
sec also lithosphere delamination 182, 280
continental drift 46-61 depth of compensation 35
apparent polar wandering(APW)curves 57-9, 58, 59,60 desert deposits, and climate 52
continental reconstructions 46-9 detachment fault 214
Euler's theorem 46 detrital remanent magnetization(DRM) 56
formation of Tertiary mountain belts 3 deviatoric tension 221-2
geological evidence 49-51 diamonds 49
historical perspective 1-7,46 diapirs, mantle 160, 164, 167, 169, 247
and palaeoclimatology 51-2 diffusion creep 33
palaeomagnetism 55-60 dispersion of surface (seismic) waves 9, 15-16, 116
palaeontological evidence 52-5 displaced terrains 198-207
continental margin disrupted terrains 198
formation 220 ductile deformation in mantle 31-3
sheared 135-7 dunite 27
formation, model 136 dynamic viscosity 33, 253
continental margin basins 236-7
continental reconstructions 269, 270, 271
continental rifts 209-15 Earth
classification 209-11 composition 19, 20,43
East African Rift, model 226 Conrad discontinuity 18
evolution to continental margin 220 contracting/expanding Earth hypothesis 248-9
origin 212-15 methods of testing 249-50
petrology 210-11 core 28-30
structure 211-12 crust see continental crust; oceanic crust
velocity structure, P (seismic) waves 212 deformation in crust and mantle 30-4
see also rift zone geomagnetic field 56-7
continental rupture, and hotspots 224-5 geothermal energy 43-4
continental splitting 218-34 Gutenberg discontinuity 18-19, 29,43
model,schema 227 inertia, ancient moment of 249-50
propagating rift 228, 231 internal structure 8-45
triple junctions 216-18 K/U ratio of crustal rocks 271
continental tension, origin 223-5 mantle 26-8
continuum model 192 isostasy 34-40
contourites 21 mean atomic weight estimations 19, 20
contracting Earth model and hypothesis 248 Moho discontinuity 17, 18,43, 109-11, 117-18
convection, concept. Holmes, A. 3, 5 moment of inertia 19, 249-50
convection cells, areal distribution 262-8 radioactive decay 43
convection currents see mantle; thermal processes radius, calculation of ancient (palaeoradius) 250
Cordillera, North American, suspect terrains 198-207 rheology 43
Cordillera, Peru 174-5 terrestrial heat flow 42-4
Coastal Batholith 175 thermal gradient, average 250
see also Andes velocity structure 17-19
core earthquakes
size 28-9 Benioff zones 78, 139-46, 149, 153, 169, 175, 291
and convection currents, mantle 223-5 distribution 76-8
see also Gutenberg discontinuity elastic rebound theory 10
core-complex range 243 epicentre 8
Coriolis force 57,66 focal depths 77
Costa Rica Rift, eastern Pacific 23 focus 8
Coulomb-Navier hypothesis 31 hypocentre 8
eratons, correlations, southern Atlantic 50 island arc systems 139-48
creep 30,33-4 location 9-10
Coble creep 33, 34 magnitude 8
development of basins 238 mechanisms 10
diffusion creep 33 seismology 8-17
Nabarro-Herring creep 33, 34 Conrad and Moho discontinuities 17, 18
power-law (hot, recovery, Weertman) creep 33 focal mechanism (fault plane) solutions
Cretaceous, mid,superplume 89-90 10-12
Curie point 55,67 ambiguity 12-15
Cyprus,Troodos Ophiolite 289 great circle paths 17
location of earthquakes 9-10
P (seismic) waves 8-11, 15, 19
D" layer 28 S (shear seismic) waves 8, 18-19, 267-8
dacites 291 strike-slip fault 12
Dead Sea fault system 128, 132 teleseismic inversion geometry 16
pull-apart basin 132 terminology 8
326 INDEX

East African Rift 81, 220 first arc 139

continental rifting 221 flexural rigidity 36,42, 185
model 226 defined 36
Ethiopia 224-5 flexure of lithosphere 36-42,141, 236, 238,240, 259
Gregory Rift 212,214-15 flower structure 51,53
minerals 285 flysch 139,157
East Pacific Rise focal depth 9,77
axial magma chamber 109-11 focal mechanism (fault plane) solutions 10-15
axial region 108-10 focal sphere 11
Basin and Range Province, USA 170 focus, defined 8,9
hydrothermal circulation 114 fold belts 49,50
hydrothermal mineralization 288 forces
marine magnetic anomalies 71 acting on plates 30, 258-60
petrology 114-15 stress(a)and strain (e), defined 30
seismic refraction, gravity anomalies 106 forearc basin 139
spreading rate 103-4 morphology 139
velocity modelling 265 foredeep 244, 245-6
eclogite 20, 27 foreland basins 244, 245-6
Michigan Basin 238 fossils
economic geology 285-95 evidence for continental drift 52-4
Ecuador, Oriente Basin 293 invertebrates, correlations of diversity with time 54
edge-force mechanism 261-2 fracture zones see oceanic fracture zones
plate tectonics 261-2 France
clastic limit 30 Aquitaine Basin 293
elastic rebound theory 10, 14-15,92 Paris Basin 293,295
English Channel, Goban Spur 231-2 French Somaliland, Afar rift system 224
ensialic back-arc basins 293 fundamental fault 128
ensialic belts 281, 291
epicentral angle 8
epicentre 8,9 Gabon Basin 293
escape tectonics 187-94 Ganga Basin 183
Ethiopia, East African Rift 224-5 Gangamopteris flora 51, 53
Euler poles, jumping 93 garnet 30
Proterozoic 276-7 geoid 88, 202, 235, 262, 263
Euler's theorem 46, 78 geological timescale 301
Eurasia, suture between India and Eurasia 180-2, 293-4 geomagnetic field 56-7
Europe geomagnetic polarity timescale 69,70
Broad Fourteens Basin 246-7 geomagnetic reversals 64-7
separation from North Amierica 74 frequency 66
evaporites and TPW 88
and climate 52 mechanism 66
diapirism 247 time intervals 64,66, 73
economic importance 294 Vine-Matthews hypothesis 67-9
exotic terrains 198-207 geomagnetic timescale 73
extraversion 277 geosynclinal theory, historical perspective 6-7
extrusion (escape) tectonics 187-94 geothermal energy 43-4, 295
see also thermal processes
Ghana-Ivory Coast, sheared continental margin 135-6
failed rift 119, 120 glacial deposits, and climate 52
Falkland-Agulhas Fracture Zone 124 glaucophane 150
FAMOUS area, Mid-Atlantic Ridge 103-4 Global Positioning System (GPS)satellites 91
Farallon Plate 98, 170 Glossopteris flora 51,53
fault plane solutions 10 Goban Spur, English Channel 231-2
fault wedge basins 132-3 Gondwana
faulting correlation of stratigraphy between continents 51
detachment fault 214 and Laurentia 279
focal mechanism (fault plane) solution ambiguity 12,13 Pacifica 202
listric fault 157,214,241 Permo-Carboniferous glacial deposits 50, 52
principal stresses 31 reconstructions 48-9, 279
tectonic styles in Asia 193 alternatives 49
transformational (anticrack) 141-3 based on APWs 60
see dlso earthquakes, seismology; strike-slip faults; thrust faults; separation of India 187
transform faults graben
Federation of Digital Seismic Networks 9 formation by faulting 239
Fermat's principle 16 Rhine Graben 209-13
ferromagnetic substances 55 granulite-gneiss belts
ferromagnetism 56 Archaean 273, 274
ferromanganese nodules, hydrothermal mineralization 288-9, Precambrian terrains 273, 274
291 gravitational constant 249
Fiji, earthquakes, epicentres 142-3 Great Basin, Nevada 291
finite rotations 92-5 great circle paths 16, 17
 INDEX 327

Great Glen fault, Scotland 128 development, criteria 292

Greenland reserves. Cretaceous 90
HattonBank 231-3 sedimentary basins 293
ice cap mass 38 Tethys-Atlantic seaway 293
Greenland-Scotland Ridge 226 hydrothermal circulation 25, 112-14
greenstone belts, Archaean 272-3, 274-5, 291 hydrothermal metamorphism 25
Gregory Rift 214-15 hydrothermal mineralization 288
cross-section 215 hypocentre 8
model, Bouguer anomalies 212
Grenada Trough 171,172,173
Grenville belts 278 lapetus Ocean 160, 162, 279
Grenville Province 277-8 Iceland 48
Gulf of Aden area igneous activity 26
evolution 227-8 igneous provinces 49-50
plate tectonic framework 225 ilmenite 29
reconstructions 229 impactogens 210
Gulf of Bothnia, land rise 38 indentation tectonics 187-94
Gulf of California India
leaky transform fault 128 continued convergence 187-8
propagating rift 228 fit with southern continents 48,49
Gulf of Lion,sedimentary basin 236-7 northward drift with respect to Asia 187
Gulf of Mexico 48, 293 suture between India and Eurasia 180-2, 293-4
Gulf ofSuez 293 see also Himalaya
Gutenberg discontinuity (core-mantle) 18-19, 29,43 Indian Ocean
Guyana, Roraima Formation 49 hotspot tracks 86
northwest, marine magnetic anomalies 71
Indian Ocean-Tibet, section 175
halokinesis 294 Indonesia, plate 265
harzburgite 24, 256 induced magnetization 55
Hatton Bank 231-3 instantaneous and finite rotations 92-5
Hawaii, sea floor, geoid anomaly 263, 266 inter-arc basin 286
Hawaii-type chain, cross-section 286 interferometry(VLBI) 90-2
Hawaiian-Emperor chain 81-2 intracratonic basins 239
heat flow see thermal processes inversion 246-7
Himalaya 180-95 inversion geometry 16
Altyn Tagh Fault 189,191 invertebrates, correlations of diversity with time 54
Bangong-Nujiang suture 183,188 island arc systems 139-48
Bouguer anomalies 184,186 and Andean plate, continental and oceanic crust section 175
collision mechanism 194-5 arc-continent collision 195-7
comparisons offocal distributions beneath subduction Atlantic Ocean 160-3
zones 204 cross-section 286
deep structure 184-5 defined 138
evolution 185-7 early and late development phases 164
formation 187-94 and earthquakes 139-48
models 181 general morphology 138-9
geology 182-4 marginal seas 163-70
map 184 petrology 162
indentation tectonics 187-94 remnant arc 139
Plasticine experiments 190-1 Scotia island arc 160
Indus-Zangbo suture zone 183, 186, 188 volcanic and plutonic activity 160-3
Kangting Fault 188 see also subduction zones
Kokoxili suture 183,188 island chains, structure 81
Main Boundary Thrust 188 isostasy 34-40
Moho topography 185-6 Bouguer anomalies 35, 38
Red River Fault 188,191 isostatic rebound 37-8
section N-S 185 tests 38-40
historical perspective isostatic anomaly 38
contracting Earth hypothesis 248-9 isothermal remanent magnetization (IRM) 56
plate tectonics 1-7 Ivory Coast-Ghana,sheared continental margin 135-6
Holland, Broad Fourteens Basin 246-7 Izu-Bonin Trench 167
Hooke's law 30
hotspots 81-3
and continental rupture 224-5 jadeite 150
intracontinental section 286 Japan arc 140
reference frame 79,81-3,167, 170 convergent plate margins 153
see also mantle earthquakes, distribution 147
Hudson Bay 38 fractional velocity variations 145
Hudsonian Orogeny 279, 280, 282 Itoigawa-Shizuoka Line 154
hydraulic head loss 115 marginal seas 165
hydrocarbons Nankai Trough 160
and climate 52, 53, 294 paired metamorphic belts 151, 154
328 INDEX

Japan arc {cont) magnetite 62

section 145 magnetization
Java Trench 196-7 induced 55
Juan de Fuca Ridge, northeast Pacific 65,71 permanent 55,56
Juan Fernandez Ridge, southeast Pacific 203 see also palaeomagnetism
Jurassic-present, geomagnetic reversal timescaie 73 magnetohydrodynamics 57,65
magnetometers and magnetometry 56,62,69
magnetostratigraphy 69-74
Kamchatka 141,165 magnitude of earthquakes 8
evolution, model 205 Makran arc 157,159
Kelvin, contracting Earth hypothesis 248 mammal migrations
Kenya, Gregory Rift 212, 214-15 Australasia 55
Kenya Rift 209 placentals 55
kerogens 292 Manitoba, Amisk group 283
kimberlites 26 mantle 26-8,252-8
Brazil, Trindade hotspot 83, 87 anomalous mantle 107-8
kinematic viscosity 254 composition 27
Koningsberger ratio 68 convection currents 5,64, 223-5, 252-8
Kuroko-type ores 290 3-D velocity model 265
areal distribution of convection cells 262-8
azimuthal anisotropy 268
Labrador Trough effect of transition zone 255
evolution, model 282-5 feasibility 253-5
geological history 279-82 Newtonian/non-Newtonian convection 254
Lambert equal area net 12 numerical simulation 264
large ion lithophile elements(LILE) 210-11 Nusselt number {Nu) 254
Laser Geodynamics Satellite(LAGEOS) 91 process 252-3
laterite, formation 52, 294 Rayleigh number(R) 253, 254
Laurasia 2,49, 52, 53, 59, 202, 216 Reynolds number (R^) 254
Laurentia seismic tomography 15-17, 265-7
APW paths 282 Taylor number {T) 254
and Gondwana 279 vertical extent 255-8
Late Proterozoic supercontinent 278 whole mantle and layered convection 257
Ordovician 160 creep 33-4
Laurentian Shield, Wopmay Orogen 279 D" layer 28
leaky transform fault 127, 128 deformation, brittle and ductile 30-3
Lesser Antilles subduction zone 160,170-3 diapirs 160,164, 167, 169
Lesser Antilles Trench 155 drag force 259-61,281
limonite 52 hotspots 81-3
linear island chains 81 and continental rupture 224-5
listric fault 157,214,241 number and distribution 82,83
lithosphere isostasy 34-40
and asthenosphere 36-7,40-2,43 low velocity zone(LVZ) 18, 27-8,41-2
downgoing slab thermal structure 148 megaliths 256
flexure 36-42,141, 236, 238, 240, 259 metamorphism 25-7
isostatic rebound 37-8 mineral assemblages 29
low velocity zone(LVZ) 18, 27-8,41-2 peeling 182
mineralization 290 plumes
shear wave model 41 and geomagnetic reversals 67-9
stress distributions, model 148 mechanism 82
temperature and depth 41 superplumes 88-90
thickness beneath oceans 35,41 seismic structure 27
transform faults 75 transition zone 26, 28-9, 236-8, 254-5, 257
see also continental crust; mantle; oceanic crust viscosity 38
longitudinal wave 8 see also thermal processes
Los Angeles Basin 293 marginal seas see back-arc basins
Love waves 9, 15 Mariana arc 138, 157, 159
low velocity zone(LVZ) 18, 27-8,41-2 Mariana Trench
blueschists 151
lithosphere bending 146
magma chambers 109-12 marine magnetic anomalies 62-3
mechanism for development of median rift 115-17 marsupials 55
magmatic arc, cross-section 287 mass, anomalous see isostasy
magmatic belt, Andean-type, cross-section 286 median rift 115-17
magnetic anomalies, marine 62-3 Mediterranean Sea, salinity crisis 237
mid-ocean ridges 71 megaliths, mantle 256
variation with geomagnetic latitude 68 membrane stress 220, 224
magnetic cleaning 56 membrane tectonics, mechanism 224
magnetic lineation 62-3, 167 Mendocino Fault, offsets 122
magnetic quiet zone 72, 227 Mendocino Fracture Zone,San Andreas Fault 98
magnetic reversals see geomagnetic reversals Mesosaurus 52
 INDEX 329

mesosphere 42,43 Nabarro-Herring creep 33, 34

Mesozoic, Pacific Ocean 202 Nafe-Drake relationship 106
Mcsozoic-Cenozoic orogenic belts, suspect terrains 207 Nankai Trough subduction zone 202
metallogenic provinces 51 natural remanent magnetization(NRM) 55-6
metamorphic terrain 198 Nazca plate 175, 177,203
meteorites 19,27, 30 Nevada, Great Basin 291
Michigan Basin, model for formation 238,293 New Hebrides region, Tonga-Kermadec island arc system, model
microcontinents 174,187, 199 of deformation 169
microplates 202 New Zealand
Mid-Atlantic Ridge 103-9 Alpine Fault 95-6
alternative model 107-9 Taupo region 166
anomalous mantle below 107 Newfoundland,ophiolites 24
axial magma chamber 109-12 Niger Delta 48
course 124-5 nodal plane 11
FAMOUS area 103-4 North America
gravity anomalies and crustal structure 106-7 and Antarctica,SWEAT hypothesis 277
hydrothermal circulation 114 Appalachian Fold Belt 218, 223
magnetic anomalies 64,71 Canada, Labrador Trough 279-85
median valley evolution 105 Grenville Province 277-8
northern termination 124-5 Labrador Trough 279-85
segmentation 112 marine magnetic anomalies 62-3
seismic velocity structure 22-3, 265-6 mid-continent ridge 284
transform faults 124-5 Nevada, Great Basin 291
mid-ocean ridges 76,103-21 Oklahoma Aulacogen 218,222
anomalous upper mantle beneath ridges 107-8 Ouachita Fold Belt 223
bathymetric profiles 104 separation from Europe 74
depth-age relationship of lithosphere 116 southern major structures 223
fast and slow spreading 115-16 suspect terrains, map 200
Great Meteor Seamount,free air anomaly 37 Western Cordillera 198-207
heat flow and hydrothermal circulation 112-14 see also California; San Andreas Fault
heat flux 44 North Atlantic see Atlantic Ocean
magnetic anomaly profiles 71 North Sea, Central Graben 241
overlapping spreading centres(OSCs) 109-11 Nusselt number {Nu) 254
petrology 114-15
propagating ridges 118-21
ridge-axis discontinuities 112, 113 obducted ophiolites 286
sea floor spreading 64,69-74 obduction 24, 284
rates 72 objective function 47
separation rates 103 oblique subduction 165
segmentation, model 111-12 ocean(ic) basin, compensation condition, formula 35
structure Ocean Drilling Program (ODP) 23, 74-5
axial zone 103-4 confirmation of Vine-Matthews hypothesis 74-5
below ridges 104-7 oceanic crust 21-4, 25-7
shallow structure 108-12 comparison of structure with ophiolite complexes 24
transform faulting 127-8 depth-age relationship 116,127
see also named ridges differentiation from continental crust 26
mineralization magnetic anomalies 62-3,68
plate tectonics 286-92 metamorphism 25-6
schematic cross-sections 286-7 oceanic layers 1, 2 and 3 21-4
transition zone 29 origin 116-18
Mississippi Valley-type deposits 285 P and S wave velocity structure 22
Moho discontinuity 17, 18,43 seismic and elastic thickness compared 42
magma chambers 109-11,117-18 spreading centres
temperatures and densities 107 critical spreading rate 118
Moho reflections 186 overlapping spreading centres(OSCs) 109-11
Mollweide projection, Proterozoic supercontinent 277 propagating rifts 118-21
moment of inertia 19,249-50 ridge rotation model 119-20
moon,composition 20 Vine-Matthews hypothesis 67-9
mosaic tectonics 198 see also Benioffzones
mountain ranges 174-208 strength profiles 32
Andean-type 174-80 see also lithosphere
arc-continent collision 195-7 oceanic fracture zones
collisional mountain ranges 180-95 basement morphology 127
see also Himalaya Romanche Fracture Zone 292
indentation tectonics 187-94 transform faults 125-8
suspect terrains 198-207 oceanic island chains 81-2
Mozambique Ocean 279 oceanic plateaux, world distribution map 201
mud volcano 160 oceanic ridges see mid-ocean ridges
Murray Fault, offsets 122 oceanic rise, cross-section 286
Murray Fracture Zone,San Andreas Fault 99-100, oceanic subduction zones see subduction zones
124 oceanic trenches 155
330 INDEX

offscraping 157 Paris Basin 293, 295

oil reserves su hydrocarbons peridotite 27
Okinawa Trough 290 Permo-Carboniferous glacial deposits 52
Oklahoma Aulacogen 218,222 correlations 50
olivine 22, 27, 28, 29 perovskite 28
ophiolites 24-5, 279, 284, 289 Peru, Oriente basin 293
Himalaya 182 Peru-Chile Trench 155
Newfoundland 24 subduction zone 202
obducted 24, 286 Phanerozoic, apparent polar wander(APW) 59
stratigraphy 24 phosphorites 52
Troodos 289 piggyback basins 246
ore bodies 289 pillow lavas, hydrothermal metamorphism 25-6
Orowan-Elsasser convection 261 Pioneer Fault, offsets 122
overlapping spreading centres(OSCs) 109-11 placental mammals 55
oxygen. Earth's core 30 plagioclase 22
plants, evidence for continental drift 53, 54
plastering 157
P (seismic) waves 8-11, 18-19 plastic flow,defined 32-4
arrival times 265 Plasticine, indentation tectonics 190-1
oceanic crust 22 plate(s)
velocity structure, continental rift zones 212 absolute plate motion 80,81-3
Pacific Ocean arcuate fracture zones 93
eastern, Costa Rica rift, seismic boundary 23 divergence, palaeomagnetic signature 59
island chains 81 forces acting on plates 258-60
marginal seas, evolution 167-70 intraplate deformation 76
Mesozoic 202 parameters, correlations with velocity 262
mid-ocean ridges see East Pacific Rise; Juan de Fuca Ridge relative plate motion 78-81,90-2
northeast three plate problem 92
fracture zones 94 triple junctions 95-102
marine magnetic anomalies 63,65, 71 plate tectonics 76-102
seismic tomography 266-7 coal formation 294
spreading phases 93 current analysis and principal features 77, 79-81
spreading rate 103-4 earthquake distribution 76-8
subduction zones 138-73 economic geology 285-95
profiles 156 autochthonous and allochthonous deposits 285-92
western expanding Earth hypothesis, methods of testing 248-50
distribution of marginal seas 165 historical perspective 1-7
velocity anomalies 267 hotspots and absolute plate motions 81-3
Wilsonian cycle 234-5 implications 269-95
see also East Pacific Rise; Pacific Plate; Tonga-Kermadec Trench instantaneous and finite rotations 92-5
Pacific Plate mantle drag 260-1, 281
filtered gravity anomaly map 263 mechanism 248-68
filtered residual depth map 263 edge-force 261-2
Pacifica, Gondwana 202 see also mantle
paired metamorphic belts mineralization,schematic cross-sections 286-7
circum-Pacific region 155 Palaeozoic 269, 270
Japan 151, 154 plate margins 76
palaeobotany 51,53 evolution, comparison with aulacogens 217
palaeoclimatology 51-2 Precambrian 269-85
palaeomagnetism 55-60 Proterozoic 282-5
chemical remanent(CRM) 56 relative plate motions 78-81
detrital remanent(DRM) 56 direct measurement 90-2
natural remanent(NRM) 55-6 ■f^DOi^DC 258
Precambrian studies 275-9 rates 80
signature, plate divergence 59 ■^DOv^DC 259
thermoremanent(TRM) 55-6 ridge push force (Frp) 258
viscous remanent(VRM) 56 ridge resistance (Rr) 258
palaeontological evidence of continental drift 52-5 slab-pull force (Fsp) 194, 259
palaeoradius of Earth 250 superplumes 88-90
Palaeozoic plate tectonics 269, 270 trench suction force 259
Pangaea true polar wander 83-8
apparent polar wandering curves 57-9,60 vertical movements associated 235-44
breakup 90 Wilsonian cycle 234-5, 282
geoid high 235 see also mantle; thermal processes; transform faults
lapetus Ocean, closure 279 Plio-Pleistocene geomagnetic polarity timescale 70
map 270 plumes see mantle
present distribution of flora and fauna 54 Poisson's ratio (a) 21, 23, 36
reconstruction 50 polarity chron 69
Panthalassa 2 poles
paramagnetism 59 apparent polar wander (APW) curves 57-60, 276-7, 281-2
passive rift 210,226 Euler poles 78-9
 INDEX 331

of rotation 46 remnant arc 139

true polar wander(TPW) 83-8,95 review questions 297-300
porphyry coppers 289, 290 Reykjanes Ridge, marine magnetic anomalies 71
potassium Reynolds number {Re) 254
Earth's core 30 rheology of lithosphere 40-3
radioisotope decay 43,252 compositional vs theological layering 43
power-law (hot, recovery) creep 33 Rhine Graben 209-13
Pratt's hypothesis of isostatic compensation 35-6 rhyolites 291
Precambrian palaeomagnetic studies 275-9 Richter scale 8
Precambrian plate tectonics 269-85 ridge propagation 119-21
heat flow 271-2 ridge push force (Frp) 258
Precambrian terrains ridge resistance {R^ 258
granulite-gneiss belts 273, 274 ridge rotation model 119-20
greenstone belts 291 ridge-axis discontinuities 112,113
heat flow 44 rift zone
principal pressure 31 aborted, cross section 286
propagating rift 228, 231 intercontinental, cross-section 286
continental splitting 228, 231 see also continental rifts
Proterozoic rigidity
plate tectonics 282-5 flexural 36,42, 185
Sudan, Wilsonian cycle 284 shear modulus 8
tectonics 279-82 Rio Grande Rift 209, 211
Wilsonian cycles 282-3 Rocas Verdes region, Chilean Andes 170, 171, 273, 274
Proterozoic supercontinent roll-back 167
apparent polar wander path 276 Romanche Fracture Zone 292
Mollweide projection 277 root and anti-root 35
reconstruction 278 Roraima Formation, Guyana 49
proto-Atlantic Ocean 269, 270
proto-Mediterranean Sea 293
proto-Pacific Ocean 278 S (shear seismic) waves 8, 18-19, 267-8
pseudofault 119 Salton Trough, California 132, 292
pull-apart basins 131-2, 292, 293 San Andreas Fault 80,91
sketch map 133 Basin and Range Province, USA 170, 209
Pyrenean fault zone 128 initiation 170
pyroclastics 290 rate of displacement 92
pyrolite 27 strike-slip faults 128-30,134
pyroxene 30 triple junctions 98, 101-2
pyrrhotites, black smokers 288 see also California
satellite laser ranging (SLR) 91
Sayan-Baikal Fold Belt 209
Q-factor 140 Scotia island arc 160
quasars, VLBI 90-2 Scotland
Great Glen fault 128
Greenland-Scotland Ridge 226
radiointerferometry, GPS satellites 91-2 Midland Valley 160, 161
radioisotope decay 43, 252 Southern Uplands accretionary prism 161
short-lived radioisotopes 43 sea floor spreading 62-75
Rayleigh number(R) birth of plate tectonics 5-7,6
computed for mantle and subdivisions 254 defined 64
defined 253 magnetic anomalies 62-3
Rayleigh (seismic) waves 9,15 spreading rate 103-4
reconstructions transform faults 75
continental 46-9 Vine-Matthews hypothesis 67-9
around Atlantic Ocean 47 second arc 139
Gondwana 48-9,60, 279 secular variation 57
Gulf of Aden area 229 sedimentary basins
Pangaea 50 economic deposits 285-95
plates, and hotspot frame of reference. North Atlantic 84 formation 236-45
positions of continents, Wegener, A.4 gravity loading hypothesis 236
supercontinent heat flow 42-4, 239, 240, 250-2
Late Cambrian 278 see also thermal processes
Proterozoic 278 hydrocarbons 293
recovery creep 33 paired with eroding mountain range 239
red beds, and climate 52 sinking model 273
Red Sea seismic anisotropy 267
metalliferous sediments 288 seismic gap 145-6
minerals 285 seismic tomography 15-17, 265-7
oil 293 seismic waves see P and S waves
Red Sea-Gulf of Aden area, plate tectonic framework 225 seismograph 9
relative plate motions 78-81,90-2 seismology see earthquakes, seismology
remanent magnetization 56 shear modulus (rigidity), defined 8
332 INDEX

shear waves see S waves original configuration 207

sheared continental margin 135-7 stratigraphic terrain 198
SIAL 18 world map 206
Siberian Traps 234 suture zone 180
side-scan sonar 103 SWEAT hypothesis, Antarctica and North America 277
silicon. Earth's core 30
SIMA 18
slab-pull force (Fsp) 194, 259 Taylor, F.B., formation of Tertiary mountain belts 3
Snider-Pellegrini, A., continental drift 2 Taylor number(T) 254
sonar 103 tectonic subsidence 243
South America teleseism 16
Atacama Fault 199 teleseismic inversion geometry 16
Cordillera, Peru 174-5 terrain amalgamation and accretion 198-9
Nazca plate 175, 177 see also suspect terrain
Nazca Ridge 203 Tertiary mountain belts, formation by continental drift 3
subducting oceanic rises, Cenozoic volcanism and graben 203 Tethys
see also Andes contraction 187
South Atlantic Cretaceous 293
marine magnetic anomalies 71 transport processes 202
opening, propagating ridge 230 Tethys-Atlantic seaway, hydrocarbon reserves 293
spreading rate 69, 73 thermal diffusivity 33, 253, 254
South Sandwich island arc 167 thermal gradient. Earth's surface 43, 250
southern continents thermal processes 42-4, 250-4
fit with India 48,49 concept of convection (A, Holmes) 3,5
palaeoclimatic data 53 convection currents of mantle 5,64, 223-5, 252-8
spinel 28 see also mantle
spreading centres 67-9,109-11 geothermal energy 43-4, 295
spreading rate 69, 72, 73, 103-4, 115-16, 153, 275 heat flow 42-4
Vine-Matthews hypothesis 67-9 and average thermal gradient 250
steady-state necking 115 continents and oceans compared 44, 252
stratigraphic terrain 198 global patterns 250-2
stratigraphic timescale 301 H,formula 43
stress(a)and strain (e), defined 30 measurements, land and sea 44
strike-slip faults 12,128-36 Precambrian plate tectonics 271-2
idealized, map view 134 heat flux, average value 44
San Andreas Fault 128-30,134 thermal conductivity 44
trench-linked 199 thermal subsidence
subcretion 157 basins, models of formation 240-4
subduction, oblique 165 defined 243
subduction complex 155-60,161-4 thermoremanent magnetization(TRM) 55-6
subduction zones 138-73, 194 third arc 139
accretionary prism 155-60 tholeiites 162,211
Andean-type 274-5, 290 thorium, radioisotope decay 43, 252
schema 207 thrust fault see faulting
comparisons of focal distributions with Himalaya 204 Tibet
downgoing slab direction of motion 194
earthquake activity 141, 146 extrusion (escape) tectonics 189
thermal structure 148-50 faulting 193,194
element abundance, variation 162 Moho reflections 186
general morphology 138-9 welding of North and South Tibet 186-7
gravity anomalies 139 tidal interaction 249
greenstone belts 273-4 timescales
and megalith production 256 geological and stratigraphical 301
mctamorphism at convergent margins 150-5 geomagnetic polarity 69,70,73
Precambrian,southwest Africa 284 Timor Trough 196-7
subduction complex 155-60 tin, oceanic crust 290-1
volcanic and plutonic activity 160-3 TobagoTrough 171, 172,173
Sudan,Proterozoic, Wilsonian cycle 284 tomography 15-17, 265-7
sulphide deposits 289, 290 tonalites 273-5
supercontinents Tonga-Fiji earthquakes, epicentres, maps 143-4
Cambrian, reconstruction 278 Tonga-Kermadec island arc system
Proterozoic bathymetry 166, 167
apparent polar wander path 276 hypothetical section 144
reconstruction 278 Lau marginal basin 146, 166
superplumes 88-90 New Hebrides region, model of deformation 169
surface (seismic) waves 9, 15-16, 116 vertical section 144
suspect terrains 198-207 volcanic activity 160-3
composite terrain 198 Tonga-Kermadec Trench 96, 140-8, 266
disrupted terrain 198 lithosphere bending 146
metamorphic terrain 198 transcurrent faults 122
North America, map 200 transform continental margin,formation 220

transform faults 122-37
California 128-30
deep structure 133-4
defined 75,76
leaky, defined 127
oceanic fracture zones 125-8
origin 123-5
system 129
triple junctions 95-102
transformational (anticrack) faulting 141-3
olivine to spinel 141-3
transition zone 26, 28, 236-8, 254-5, 257
convection currents, mantle 255
mineralization 29
transpression and transtension 130-1
transverse wave 8
trench suction force 259
trench-linked strike-slip faults 199
trenches 76
evolution 95-6
trilobites, distribution 53
Trindade, hotspot, Brazil kimberlites 83,87
triple junctions 95-102
all possible, geometry and stability 100
continental splitting 216-18
defined 96
plume-generated, schematic origin 219
San Andreas Fault 98, 101-2
types and examples 98
Troodos Ophiolite, mineralization 289
true polar wander(TPW) 83-8,95
uniformitarianism 1
underplating 157
uranium, decay, heat escape 43, 252
INDEX 333
Variscan Orogeny 269,270, 291
velocity structure of Earth 17-19, 212
Vema Fracture Zone, North Atlantic 23
vents (black smokers) 288
vertical movements, associated with plate tectonics 235-44
very long baseline interferometry(VLBl) 90-2
Vine-Matthews hypothesis 67
deep-sea drilling(GDP)results 74-5
sea floor spreading 67-9
viscous remanent magnetization(VRM) 56
volcanic margin 231-4,273
volcanic and plutonic activity, island arc systems 160-3
volcanosedimentary basins 273
Washington-Oregon accretionary prism 160
Weertman creep 33
Wegener, A., reconstruction of positions of continents 4
Western Cordillera, North America 198-207
white smokers 288
Wilsonian cycle 234-5, 282-3
wolframite 292
Woodlark Basin, Papua New Guinea 230
world maps
oceanic plateaux 201
suspect terrains 206
World-Wide Standard Seismograph Network(WWSSN) 9
Wrangellia, Alaska 199-200
wrench faults 131
xenoliths 26
yield strength 33-4
Young's modulus(E) 36