Escolar Documentos
Profissional Documentos
Cultura Documentos
DE LAPENSEE
~
reunis par
MONIQUE CANTO-SPERBER
et
PIERRE PtLLEGRIN
Ouvrage publie
avec le concours du Centre National du Livre
Photo. Herve Morel
Paris
Les Belles Lettres
2002
ARISTOTELIAN RELATMTIES
Aristotelian Relativities
DAVID SEDLEY
325
Aristotle's point is metaphysical, not linguistic. It is important not to be misled into thinking that he is in any way appea2. Thus e.g. Simplicius, In Ar. Cat. 197.7-9. At Met. N I, 1088a 27-35, relatives
are argued by Aristotle to be the least substantial of all things.
.:>..::;u
327
ARISTOTELIAN RELATMTIES
DAVID SEDLEY
3. 8b 3-4, xa.l rrl ThlV xa.9' exacrTa. BE. BfjA.ov TO TOtOUTOV. This is not, I think, a
move from universals to particulars, i.e. from secondary to primary substances, so
much as one from the general principle to its further defence via exam pies :
cf. 2a 34-6. The examples may perhaps involve particulars (depending how one
interprets 1:6Be n at 8a 38, b 4), but at any rate the relatives in question are universals which are said of those particulars.
4. I am assuming that no difference is intended between ~ptcr[Jkvwc and
O:<{>wptcr[livwc;.
U.t\.VIU :)E.ULEY
ARISTOTELIAN RELATMTIES
But as for head and hand, and each of the things of this kind which
are substances, one can know definitely what thing each itself is
without its being necessary to know definitely what it is spoken of relatively to. tFor it is not possible to know definitely to what it is that this
head or hand belongst ( TLVO(; yap <XUTY) Xe<f>aA.r) i} TLVO(; xetp oux
crnv d6vat wptcr{.LEVW(;, 8b 18-19). So these things will not be relative. And if they are not relative, it will be true to say that no substance
is relative.
3~Y
emendation. I see no way in which <ivaytea!ov could be left for the reader to supply
from its occurrence in the preceding sentence at 8b 18, as these translators may
be supposing it is.
6. Cf. Porphyry, In Ar. Cat. 126.10-14, Ammonius, In Ar. Cat. 79.17-19, Simplicius,lnAr. Cat. 200.7-9,27-9
7. Ackrill's implicit assumption that primary substances are at issue (op. cit.
pp. 102-103) is prefigured in the ancient commentary tradition, starting with
Porphyry, In Ar. Cat. 126.10ff. Mignucci (art. cit., pp. 121-2) avoids this mistake,
but is consequently driven to devise a bizarrely ingenious equivalent of my showercurtain example, where someone knows what a paw as such is (i.e. the secondary
substance) without ever having found out what it belongs to.
;JJU
DAVID SEDLEY
ARISTOTELIAN RELATIVITIES
331
reasonable enough way to designate an Aristotelian seconsubstance10 - I believe that the main point of the expresis to draw attention to a shift that has occurred between
the preceding examples and the present one. In those earlier
pies, the relative terms under consideration functioned
as predicates : knowing that x is double , is more beauriful ,etc. The problematic case now introduced, by contrast,
takes the supposedly relative term as itself the subject of predi:tation : with regard to head, etc., taken in their own right as
subject~ 1, we can know definitely what thing each itself is
(nun) f.LEV orrp crnv, 8b 16-17), but not necessarily know definitely what it is relative to. That is, there is a cognitive asymmetry between knowing what the (supposedly) relative item
itself is and knowing what its correlative is.
And why the asymmetry? Aristotle appears to be absolutely
right that, when taking as your subject the secondary
substance head, you can know definitely that a head is such
and such a kind of thing (say, a bodily part with certain specialised functions in sense-perception and the ingestion of
food), yet that you cannot- not just that you need not- know
definitely what a head is the head of. This latter is because
what a head is the head of is, properly speaking, simply the
headed. And one who claims to know that what a head
belongs to is the headed is no more professing definite knowledge than one who claims to know that what the more beautiful is more beautiful than is whatever is less beautiful than it.
We may sum up Aristotle's argument as follows. Consider
those secondary substances which are parts of further secondary substances, e. g. head taken as a universal rather than
as a particular. Such items may appear to belong to the category of relativity, and thus to cross a categorial boundary
which should at all costs be respected. However, we can avoid
10. Secondary substances are c:ton and yevT}, and it is widely accepted that their
demotion to the second rank has a directly Platonic reference ; so the Platonising
language is not totally out of place, even if Aristotle may well not use the Platonic
locution elsewhere in precisely this way.
11. Cf. Plato, Phd. 74b-c, where o:tmx rO: 'tmx is similarly used to pick out as
subject what was previously the predicate, while at the same time serving as a standard Form-locution.
,'),')~
Much unnecessary darkness has been shed on this question by the assumption, a commonplace in the ancient,
medieval and modern commentators 1 , that the former defiI 2. Ammonius, In Ar. Cat. 77.28ff., Simplicius, In Ar. Cat. I 98. I 7ff., Philoponus,
In Ar. Cat. l08.3I-109.31, Olympiodorus, In Ar. Cat. 100.4-20, Ackrill, op. cit. p. 10I
(where it is even said to be undeniable, although his own further discussion of
the passage does not exploit it.), Oehler op. cit. pp. 248ff., Zanatta op. cit.
pp. 592 ff., M.Erler, Relation. I. Antike , in Historisches Wirterbuch der Philosophie
vol. VIII (Darmstadt I992) 578-86, p. 580, and F. Morales, Relational attributes
in Aristotle, Phronesis 39 (1994), 255-75. The honourable exceptions known to
me are Porphyry, In Ar. Cat. I25.6-23, Mignucci art. cit. pp. 107-8, and F. CaujolleZaslawsky, Les relatifs dans les Categories , in P. Aubenque (ed.), Concepts et categories dans la pensee antique (Paris I980), I57-95, pp. I85-7.
ARISTOTELIAN RELATMTIES
333
nition focuses on the way things are spoken of, the latter on
the way they actually are. On this view, Aristotle's tentative
solution rests on a deliberate shift from what we may for
convenience call a de dicta to a de review of relativity 13 The
principle of cognitive symmetry will somehow be one which
follows only from the de re view of relativity. There are,
however, severe difficulties attached to this view.
If the de dicta version of relativity were to be understood in
a way which did not imply the cognitive symmetry principle,
that as far as I can see could plausibly only be either because
on this definition relativity is a purely linguistic matter, and
therefore does not necessarily map onto how things actually
are, or because, given that linguistic expressions are equivocal, some apparently relativising expressions fail to express
genuine relativities. But we have seen that, when Aristotle
asserts that e. g. some cow is not spoken of as some cow of
something (8a 17-18), that is already a metaphysical thesis,
not a report of actual linguistic usage, which can hardly
exclude expressions like some cow of some farmer .
Outside this passage, in fact, one might be forgiven for
understanding the Categories as treating the way things are
and the way they are spoken of as entirely co-extensive : to say
how_ a thing is spoken of is to give the description under
which it is proper to speak of it- proper precisely because the
description picks out the appropriate aspect of what that
thing is. To put it another way, the legomena which the categories classify (1b 25ff.) are predicates, and Aristotle is unconcerned with any distinction between linguistic and metaphysical predicates, since he assumes them to be both. Hence in
the Categories, as elsewhere in Aristotelian metaphysics, there
is simply no exploitable gap between how a thing is spoken of
and what that thing is. Indeed, even within our present
passage, just where Aristotle is supposed to be exploring the
advantages of shifting from the de dicto to the de redefinition
of relativity, he seems to have no qualms about slipping back
into the de dicta locution- at 8b 14-15,
I3. The actual terminology adopted in the medieval tradition (Aquinas.
Scotus. Occam etc.) is the horrible secundum dici and secundum esse.
J.:.J'-1:
B liv EU>n
EU)vat 14
DAVID SEDLEY
Tt(;
TWV rrp6c
Tt
ARISTOTELIAN RELATMTIES
335
Clearly episteme (6b 3) will fit Aristotle's initial, weaker definition, which I have called soft relativity. As he explains
(6b 5), any case of epistememust be episteme of something. But,
we may now observe, episteme does not consist purely in the
relation between the understander and the understandable.
It is in addition an internal disposition of the soul 16, whose
various species Aristotle therefore classes as qualities (8b 2930, etc.). Hence episteme is a soft relative. By contrast,
double is a hard relative. Doubleness is a property which
consists purely in a relation between two correlative items,
the double one and the half one. There is no internal qualitative disposition of the double item which can in any way be
identified with its doubleness.
The classification of some other relatives as hard relatives
may look less straightforward to justify. Take slave and master
(7a 31-b 9). Being a slave does appear in Aristotle's view to
consist purely in the relation of having a master 17 Yet there is
at least one internal qualitative disposition that is entailed by
it, namely being human. 'Why does this not make slave no
more than a soft relative ? Because the intrinsic state of being
a human being can in no way be identified with slavery, in the
way that the intrinsic mental state of an understander can be
identified with their understanding. Aristotle's singular way
of putting this (7a 31-b 1) is that, with regard to being a slave,
the property of being human is merely accidental 18 That
is why we can safely conclude that slave is a hard relative.
rather be that they are relative because they are genera, requiring as differentiae
items which are external to themselves : understanding must be understanding of
something, e.g. letters, a state (of mind) must be a state regarding something, and
virtue and vice (added as relatives at 6b 16) must be differentiated into species by
reference to one's being good or bad at something. This reading would however
imply, again in line with 11a 20ff., that their species are not normally relatives,
and, if so, we would be forced to take the species of 9crtc listed at 6b 11-14 not to
be themselves relatives other than (as 11a 28-9 explains) X<XTCx yvo~;.
16. Cf. 1b 1-2, r, ErrtO"nl[LT} EV UITOXEL(.LEVCf [LEv ecrn Tij t(JUXU
17. There is no sign yet of Aristotle's later doctrine that some people are slaves
by nature, which the findings of Peter Garnsey (Ideas of Slavery from Aristotle to
Augustine [Cambridge 1997}, pp. 107-28) suggest may be unique to the Politics.
18. Cf. DA 418a 20-3, 425a 26-7 for a similar use of accident.
.J.:>U
UAVIU
S~DLEY
ARISTOTELIAN RELATIVITIES
337
.LJL"l.V UJ
.;)CULE.Y
ARISTOTEElAN RELATMTIES
339
.:>':I:V
UAVIU SEDLEY
ARISTOTELIAN RELATMTIES
341
342
DAVID SEDLEY
J-:rJ
assume that a two-way relative disposition is at issue, but his formal account of the
criterion of hard relativity does not require it): that a virtuous state of mind
should be specifically courage depends wholly on the presence of external
danger, but the danger can hardly be said to depend on the presence of courage.
TI1is may look like a difference from Aristotle, who treats hard relativity as reciprocal, and bases his cognitive symmetry principle on just that thesis. But if it is a
difference there is no necessity to conclude that it is a difference in their notions
of hard relativity. Aristotle's reciprocity rule is a rule about relativity in general
(Cat. 6b 28ff.). It is because he tentatively equates relativity with hard relativity that
he expects the same reciprocity rule to apply to the latter as well. It need not
follow that the notion of hard relativity which he adopted was already defined as
a reciprocal relation.
28. Sextus Empiricus, PHil 81.
29. Mignucci (art. cit. [n. 24 above]), pp. 166-8, also argues against a Stoic
dependence on Aristotle here. His arguments, however, are based largely on a
doubt, which I do not share, as to whether the Aristotelian and Stoic concepts are
sufficiently similar.
30. T11is issue has been very powerfully explored by F.H. Sandbach, Aristotle
and the Stoics (Cambridge 1985).
31. Although the Categories, along with the De interpretatione, appears in two
versions of a list of Aristotelian works held to derive from a late third-century BC
source (one in Diogenes Laertius, see V 26, the other in Hesychius), it is generally
agreed that they are a later interpolation there: seeP. Moraux, Les Listes anciennes
des ouvrages d'Aristote (Louvain 1951), pp. 131, 186-90, 313, I. During, Aristotle in
the Ancient Biograj1hical Tradition (Goteborg 1957), pp. 69, 90.
ARISTOTELIAN RELATMTIES
345
to know the one without the other. Hence the one must be included
in the definition of the other.
to mark off the hard relatives within it. But actually, as for instance 146a 36-7
makes clear, xo:e' m1T6 relative merely contrasts with relative xo:nx To y8voc ,
where, for example, geometry satisfies the latter description but not the former.
The TofJ. VI 8 passage makes it unambiguous that he is speaking of all relatives, not
a subset of them.
34. One sign that this may predate the Categories is that in the Tbp. VI 8 passage
Aristotle asserts that all relativity is hard relativity without yet letting this narrow
the range of items included as relative : his two examples of relatives are in fact
episteme (correlated with the episteton) and boulesis (correlated with the good).
347
DAVID SEDLEY
ARISTOTELIAN RELATMTIES
346
348
DAVID
SEDL~Y
largely in the past tense, as if he knows of them only indirectly. My best guess is that these are fellow Academics whom
the very independent-minded Eudorus chose to criticise in
his Categories commentary. Probably, as is typical of in-school
polemic, he did not deign to name them. If so, Simplicius in
his turn will be relying on Eudorus' report, and will know no
more about them than Eudorus has chosen to vouchsafe.
If we follow this hypothesis, how should we date them? It
is very hard to find room in the history of the first-century BC
Academy for Platonist Academics earlier than Eudorus
himself. Eudorus is, after the Academy's long sceptical phase,
the very first Platonist we hear of with a recognisably Platoni38
sing metaphysics , and the only one before Plutarch to call
himself an Academic. It is easier to suppose that Eudorus was
drawing his material from the Academy's early Platonist
phase, which he must in some sense have seen himself as resuming.
If so, we have to ask whether a categorial doctrine of this
kind is likely to have been developed in the early Academy.
The answer is emphatically yes. There is no doubt that there
were strong categorial interests in the fourth-century
Academy. Xenocrates is said to have adopted a version of the
simple two-category scheme, absolute (xae' auTo) and
relative>> (rrp6c; Tt) bequeathed by Plato himself 39 , and
Hermodorus to have produced his own more elaborate classification of relatives 40 Hermodorus' scheme, as reported,
does not itself include soft and hard relativity. Nevertheless,
the categorial interests, the size and the doctrinal diversity of
38. Harold Tarrant (Scepticism or Platonism ? TI~e Philosophy of the Fourth Academy
(Cambridge 1985) has sought to maximise the Platonist leanings of Philo of
Larissa, who may well have been Eudorus' own Academic teacher, and does go so
far as to offer Philo at least an epistemological use of Platonic Forms ; but the
proposal is highly speculative.
39. Cf. Plato, Sph. 255c-d, Tizt. 160b, Phlb. 5lc; and for a probably Peripatetic
attribution to Plato of the formal two-category scheme, see DL II 108-9.
40. Simplicius In Ar. Phys. 247.30ff. = Hermodorus fr. 7 Isnardi. For a judicious
assessment of the evidence for relativity in the Academy, see Gail Fine, On Ideas
(Oxford 1993), pp. 176-82.
ARISTOTELIAN RELATMTIES
349
350
DAVID SEDLEY
ARISTOTELIAN RELATIVITIES
46. For Andronicus see Simplicius, In Ar. Cat. 63.22-4, 134.5-7; for Eudorus, ib.
174.14-16,206.10. The same practice is found in Anon. In Plat. Tht. (LXVIII 1-15),
who is probably datable to this same period.
47. This practice was adopted by the early Peripatetic commentator Boethus
(ib. 163.16-17, 202.1ff.), who actually wrote a book on the rrp6(;' nand the rrp6(;' d
TT<U(; itxov ; by Ariston of Alexandria ( ib. 202.lff.); and in the Categories of ps.Archytas (see esp. p. 28 Thesleff, where there are clear signs that it is meant to
designa~e hard relativity).
48. This eventual fusion is apparent not only in the many Platonist commentators on the Categories, but even in the 2nd~century AD Stoic Hierocles, who
casually uses a version of rrp6(;' r 7tcu(;' itxetv to express the undoubtedly soft-relative oiheiosis relation (El. Eth. li 7-9).
49. Mignucd, art. cit. (n. 24), pp. 197-8, makes some important further suggestions about the Academic background to Stoic relativity theory.
50. Cicero, Ac. I 24-5, 27-8. There is no good reason to take any of this passage
to be a mere retrojection by Antiochus of Stoic thought onto the early Academy:
much of it, including the precise use made of quality, is un-Stoic, even though
it is easy to see in it the antecedents of Stoicism. I argue this briefly in Lucretius
and the Transformation of Greek Wisdom (Cambridge 1998), pp. 75-8, and fully in
~The origins of stoic god~~, in D. Frede, A. Laks (ed.), Traditions of Theology
(Leiden, forthcoming).
51. Cf. Smp. 199e, a8eA.<f>6(;', o:1ho roue Brrep F-crrv, F-crr nvo(;' &5eA.<j>o(;' 1) ou;
The Platonic derivation was obvious to the early Peripatetic commentator Boethus
(Simpl., In Ar. Cat. 159.12-22, cf. Porph., In Ar. Cat. 111.27-9), and was disregarded
xov.
351
352
DAVID SEDLEY
* *
It is an honour to be able to present the above study as part of a
thoroughly merited tribute to Jacques Brunschwig. His seminal
writings on ancient philosophy have always served me as both a
model and an inspiration. Quite apart from his ground-breaking
studies of Aristotle, his La thiorie stoi'cienne du genre supreme et
l 'ontologie platoniciennrl2 magnificently illuminates the Platonic
background to Stoic metaphysics. I have no right to assume that he
will agree with my own speculations in the same general area, but I
hope he will take them, as they are intended, as a further exploratory
journey down paths he has done more than anybody to open up53
by most later commentators only because they thought that the emphasis in Aristotle's first definition is on being spoken of,, as opposed to actually being,,,
while the SofJhist formulation is in terms of being: see Simpl., lac. cit. As a matter
of fact, however, Plato's own account of relativity switches indifferently between
being spoken of>> (255c 13) and being >> (255d 7), just as Aristotle's does !
52. InJ. Barnes, M. Mignucci (ed.), MatterandMetafJhysics (Naples 1988), 19127; English translation in J. Brunschwig, PafJers in Hellenistic Philosophy
(Cambridge 1994), 92-157.
53. TI1is paper started life as an oral presentation at the conference on F.H.
Sandbach's book Aristotle and the Stoics, held at Cambridge in May 1986. I am
grateful to a number of participants for criticisms which helped me in writing it
up, and I can only apologise that it took me ten years to get round to doing so.
The present version was written for an October 1996 seminar at the Institute of
Classical Studies in London, and the first part was also presented at the Moral
Sciences Club at Cambridge in March 1997, at the University of Bologna in March
1997, and to a seminar on the Categories at Cambridge in May 1997. My thanks for
invaluable comments received on all these occasions, and also to Barrie Fleet,
Luca Caffaro and Chiara Palu for further comments and help. An Italian version
appeared under the title Relativita aristoteliche >>in Dianoia2 (1997), 11-25 and
3 (1998), 11-23.