Lynne Spellman-Substance and Separation in Aristotle (2002)

Substance and Separation
in Aristotle
Substance and Separation

in Aristotle
LYNNE SPELLMAN
University of Arkansas
CAMBRIDGE
UNIVERSITY PRESS
Published by the Press Syndicate of the University of Cambridge

The Pitt Building, Trumpington Street, Cambridge CB2 1 RP
40 West 20th Street, New York, NY 10011-4211, USA
10 Stamford Road, Oakleigh, Melbourne 3166, Australia
Cambridge University Press 1995
First published 1995
Printed in the United States of America
Library of Congress Cataloging-in-Publication Data
Spellman, Lynne, 1948Substance and separation in Aristotle / Lynne Spellman.
p. cm.
ISBN 0-521-47147-8
1. Aristotle Views on substance in philosophy. 2. Substance
(Philosophy) 3. Aristotle - Views on separation of substances.
4. Separation (Philosophy) I. Title.
b491.s8s64 1995
HIM -dc20
94-31450
CIP
A catalog record for this book is available from the British Library.
ISBN 0-521-47147-8 hardback
for James
Contents
Acknowledgments
page ix
Introduction
I The Separation of Platonic Forms
II Referential Opacity in Aristotle
III A Theory of Substance
IV Substance and Aristotle's Epistemology
V The Separation of Substance
VI Substance and Teleology
1
5
21
40
63
83
100
Bibliography
Index
123
129
vn
Acknowledgments
In the course of the argument of this book I draw heavily on recent

scholarship. I have been aided in understanding the contemporary debates by the Ancient Greek Philosophy Workshops at the University of
Texas, the Conference on Aristotle's Metaphysics at Florida State University in 1983, the Institute on Aristotle sponsored by the National
Endowment for the Humanities and the Council for Philosophical Studies in 1988, and an academic year at Cambridge University as the Fulbright College of Arts and Sciences Faculty Fellow at Lucy Cavendish
College in 1990-91. I am grateful to the University of Arkansas and
Lucy Cavendish for the year abroad, as well as, in the case of Arkansas,
for support for the academic year 1985-86, during which work was
begun. Chapters II and III are revisions of two published papers, "Referential Opacity in Aristotle," History of Philosophy Quarterly 7 (1990),
and "Specimens of Natural Kinds and the Apparent Inconsistency of
Metaphysics Z-H," Ancient Philosophy 9 (1989), and I thank the editors
of these journals for permission to reuse this material. Much of Chapter
V was presented at the meeting of the Society for Ancient Greek Philosophy held in conjunction with the American Philosophical Association, Central Division, in the spring of 1994. Finally, thanks are owed
to James Spellman, who read the aforementioned papers in many versions, for his philosophical insight and unfailing support for the project,
and to the referees for Cambridge University Press for their helpful
comments on the penultimate draft of the manuscript.
Introduction
We are sufficiently assured of this, then, even if we should examine it

from every point of view, that that which entirely is is entirely knowable.
(Republic 477a) 1
This is a study of Aristotle's theory of substance, more precisely of his

theory of sublunary substance. Although some philosophers, upon reading the Metaphysics, see the influence of Aristotle's biology,2 others and I am one of them - see Plato. Indeed (although Aristotle would
not have put the point in this way), I would go so far as to say that
Aristotle can be seen as attempting to offer a defensible version of
Platonism. What I mean when I say "a version of Platonism" is that for
Aristotle, as for Plato, there is something which is first in knowledge,
definition, and time, and that for Aristotle, as for Plato, whatever is
knowable must be eternal and unchanging. In the case of Plato, it is,
of course, the Forms which are intended to meet these requirements.
But Aristotle finds the Forms problematic on both metaphysical and
epistemological grounds, and while Plato himself certainly struggled
with some of the difficulties that Aristotle complains of, Aristotle believes that Plato's solutions fail, chiefly on account of separation. Specifically, Aristotle seems to believe that separation creates a gap that
recollection cannot fully bridge and that Plato's blurring of the distinction between universality and particularity not only leads to regress
but casts doubt upon the very intelligibility of Forms. What I intend to
argue, however, is that despite all his criticisms Aristotle's own account
of substance is nevertheless very like Plato's Theory of Forms but for
the denial - or more accurately, the reassessment - of separation.
1 All quotations from Plato follow the translations in Edith Hamilton and Huntington
Cairns, eds., The Complete Dialogues of Plato (Princeton, N.J.: Princeton University Press,
1961). Emphasis in all quotations follows the sources cited unless otherwise indicated.
2 This view was most recently explored in Furth (1988).
SUBSTANCE AND SEPARATION IN ARISTOTLE
In Chapter I, a chapter that lays the groundwork for Aristotle's theory, I examine Aristotle's criticism of Plato for separating the Forms,
arguing that by 'separation' Aristotle has in mind numerical distinctness,
and I cite passages to show that he believes that the numerical distinctness of the Forms from sensible objects causes insoluble metaphysical
and epistemological problems. To see Aristotle's theory as a response
to Plato inevitably raises questions about the accuracy of Aristotle's
presentation of Plato's Theory. Even though in Chapter I and elsewhere
I do from time to time sketch, in a very broad way, various interpretations of Plato's own views, in a sense the question is irrelevant to my
project - if Aristotle's theory is a response to what he took Plato to be
saying, the impact on his own views will be the same regardless of his
skill as an interpreter. Nevertheless I must admit that upon reading
Plato and Aristotle, I find the view that Aristotle misunderstood or failed
to appreciate Plato's Theory to be largely false. Rather, I agree with
those3 who say that it is just inherently implausible that one of the finest
philosophers who ever lived should, after twenty years in Plato's company, have failed to grasp his views and the issues that underlie them.
But, as I have said, the cogency of my project does not depend on
agreement with this claim.
Having discussed Aristotle's criticisms of Plato in Chapter I, I turn
to Aristotle's own views in the subsequent chapters. If Aristotle is to
say, as his criticism of Plato makes it plausible that he should, that
substances are not numerically distinct from sensible objects, one might
reasonably expect that he holds them to be identical with sensible objects. But it has to be remembered that, like Plato, Aristotle wants
substances to be unchanging if they are to be epistemologically fundamental. One might suppose that if substances are forms and if forms
are universals this requirement could be met. But then again Aristotle
wants substances to be ontologically fundamental as well, a fact that
seems to argue for their being objects, not properties. In Chapter II I
discuss the problem of referential opacity in Aristotle, claiming that
Aristotle uses a distinction between numerical sameness and identity to
address many sorts of metaphysical problems, and in Chapter III I argue
that this distinction is the key to Aristotle's theory of substance. What
I hold is that substances are for Aristotle specimens of natural kinds,
where specimens, as particular forms lacking the accidents introduced
by matter, are numerically the same as sensible objects yet not identical
with them. While specimens of kinds are not eternal, within a kind they
are indistinguishable from one another, with the result that unlike sensible objects they are knowable.
3 One of them is Russell Dancy, to whom, as a result of a conversation in the summer
of 1988, this description of the stance and my confidence in its reasonableness are in
part owed.
INTRODUCTION
Thus by the conclusion of Chapter III, an account of substance has

been given which, despite the denial of separation, makes it possible
for Aristotle to say that substances are knowable objects. But, of course,
when Aristotle objects to Plato's Theory of Recollection, his difficulty
is not with the knowability of Forms in this sense - Forms are, after
all, eternal and unchanging in a way that even specimens of kinds are
not. His complaint is rather that it is by no means assured that Plato's
Forms can now be known by us. In Chapter IV I consider Aristotle's
epistemology, that is, his account of the progression from perception
to knowledge, and argue that what supports it is precisely a theory of
substance of the sort I have proposed, namely, one characterized by
rejection of the Platonic separation of form. Indeed I argue here too
that Aristotle's lack of concern for certain skeptical questions can be
explained quite naturally by the fact that his epistemology is addressed
specifically to Platonic problems arising from separation.
Yet even if it is agreed that Aristotle intends substances to be specimens of natural kinds and even if it is conceded that specimens are
not only knowable but such as to make possible a credible account of
the acquisition of knowledge, there remain two problems. The first is
that, although I have tagged separation as the crux, Aristotle himself
says that substances must be separate. In Chapter V I address this
question, suggesting that by 'separation' what Aristotle endorses is what
I call the ontological counterpart of separation in definition. That is to
say, while he wants more than separation in definition (something only
conceptually separate from sensible objects could hardly be ontologically more fundamental than they), the separation of Aristotle's substances is not, I argue, the numerical distinctness characteristic of the
Forms.
Finally, in Chapter VI, I address what has to be the most serious
internal challenge for my interpretation. The problem is just that it is
by no means obvious that specimens of natural kinds will be ontologically
fundamental, as Aristotle's criteria for substance require. Here I take
on the question and argue that Aristotle does believe, contrary to our
inclinations, that specimens of natural kinds are more fundamental than
sensible objects. The argument in this chapter is admittedly more speculative in that it attempts to assess how it is, if I am right about his
theory, that Aristotle could think that something like a specimen of the
kind lion is more fundamental than a given individual lion in all its
peculiarity, that sensible object with which it is numerically the same.
In this chapter I argue that the grounds for Aristotle's view are teleological, a case I try to make more plausible by drawing some parallels
with art before embarking on a general discussion of Aristotle's agentless teleology and the understanding of the good which sustains it.
To summarize, the project is to defend the following claims:
(i)
Rejection of Platonic separation is the starting point for Aristotle's

account of substance.
(ii) In order to avoid separation while keeping the Platonic criteria according to which substances must be first in knowledge, definition,
and time, Aristotle distinguishes between numerical sameness and
identity.
(iii) Having done so, he holds that substances can be specimens of natural
kinds.
Yet even as I have been writing, others have also, and the two most
recent accounts of substance in Aristotle, Michael Loux's Primary 'Ousiai' (Cornell University Press, 1991) and Frank Lewis's Substance and
Predication in Aristotle (Cambridge University Press, 1991), came after
I had essentially completed this manuscript. As it happens, both Loux
and Lewis argue for forms as universals while my argument requires
them to be particulars, yet I have not attempted to provide an exhaustive
examination of all the texts that bear on this long-standing controversy.
Instead I have tried to consider a somewhat different cluster of issues
in such a way that they illuminate one another. For what I want to
contend is that, if read as criticism and revision of Plato in the way I
propose, Aristotle has a coherent view which, even if different from our
own, is nevertheless a philosophically challenging response to the experienced world.
I
The Separation of Platonic Forms
And Socrates gave the impulse to this theory, as we said before, by means
of his definitions, but he did not separate them from the particulars; and
in this he thought rightly, in not separating them. This is plain from the
results; for without the universal it is not possible to get knowledge, but
the separation is the cause of the objections that arise with regard to the
Ideas.
(Metaphysics XIII 9 1086b2-7)1
That Aristotle criticizes Plato for separating the Forms is a fact known
to every reader. However, what exactly it is that Aristotle wants to
criticize has, until recently, seldom been discussed explicitly and at
length, and indeed as exploration of the question has occurred, views
have differed considerably. It has been proposed by some interpreters
that when he criticizes Plato for separating the Forms, by 'separation'
Aristotle means their independent existence,2 that is, their capacity for
existing even if there were no sensible objects. But other interpreters
have held that he means their numerical distinctness from sensible objects,3 and some writers have thought that he means both of these.4
Thus despite the considerable importance Aristotle places on Plato's
separation of the Forms, there is disagreement about just what he is
objecting to. Moreover, besides our uncertainty about what Aristotle
meant, there is a further problem. For even as he criticizes Plato, Aristotle tells us that substances must be separate (Metaphysics VII 1
1028a34). Only after an account of Aristotle's theory of substance is
1 Unless otherwise noted, all quotations from Aristotle follow The Complete Works of
Aristotle: The Revised Oxford Translation, ed. Jonathan Barnes (Princeton, N.J.: Princeton
University Press, 1984). The page, column, and line numbers have where necessary
been corrected against the edition of the Greek text prepared by Immanuel Bekker
for the Berlin Academy, published in 1831. Emphasis follows Barnes except where
indicated.
2 Fine (1984); Hardie (1936), 73; Irwin (1977), 154.
3 Mabbott (1926). See also Morrison (1985), esp. 138-39 and 149-50.
4 Allen (1970), 131-32.
given will it be possible to decide whether the separation he intends to

assert is the same as that he intends to deny and, if so, whether in light
of his criticism of Plato, the assertion of it is problematic. These tasks
will be undertaken in later chapters. In this chapter, however, I want
to try to clarify what Aristotle means when he says Plato separated the
Forms and why he should think it a cause for objection. For I plan to
take seriously the idea that what Aristotle rejected in Plato's Theory
was principally the separation of the Forms5; indeed, as I have said, my
overall argument will be that Aristotle's account of substances can best
be seen as an attempt to preserve much that is Platonic by rethinking
separation.
I
In Prior Analytics 111 Aristotle says, "For there to be forms or some
one thing apart (para) from the many is not necessary if there is to be
demonstration; however, for it to be true to say that one thing holds
of many is necessary" (77a5-7). In addition to the many passages such
as this where Aristotle implicitly criticizes Plato for separating the
Forms, there are, fortunately, several where he tries to explain why
Plato separated the Forms and two where he explicitly describes the
priority of Forms over phenomena, a relation that has sometimes been
thought to explain what is meant by separation. I will begin with the
latter group.
At Metaphysics V 11 in the course of a number of definitions of
priority, Aristotle says:
Some things then are called prior and posterior in this sense, others in respect
of nature and substance, i.e. those which can be without other things, while the others
cannot be without them, - a distinction which Plato used. (1019al 4; emphasis
added; boldface indicates Barnes's emphasis)
This sense of priority - the priority of whatever can exist without other
things which in turn cannot exist without it - seems to be illustrated at
Eudemian Ethics I 8 in Aristotle's discussion o f the Form o f the Good.
There he says:
We must then examine what is the best, and in how many senses we use the
word. The answer is principally contained in three views. For men say that the
good per se is the best of all things, the good per se being that whose property
is to be the original good and the cause by its presence in other things of their
being good; both of which attributes belong to the Idea of good (I mean by
'both' that of being the original good and also the cause of other things being
good by its presence in them); for good is predicated of this Idea most truly
(other things being good by participation in and likeness to this); and this is the
5 Morrison (1985) suggests a similar line. See 149-50. See also Mabbott (1926).
SEPARATION OF PLATONIC FORMS
original good, for the destruction of that which is participated in involves also the
destruction of that which participates in the Idea, and is named from its participation
in it. But this is the relation of the first to the later, so that the Idea of good
is the good per se\ for this is also (they say) separable [choristen] from what
participates in it, like all other Ideas. (1217b 1-16; emphasis added)
Following Gail Fine, let us call the capacity to exist without the existence of some other thing the capacity for independent existence with
respect to that thing.6 Citing the passage just quoted, Fine argues that
when Aristotle talks of 'separation' he means a capacity for independent
existence7 and thus that in his criticism of Plato he (in the main, wrongly,
as she sees it) attributes to Plato the view that Forms are separate from
sensible objects in the sense that they are able to exist independently
of them. She says:
Aristotle is probably correct to say that at least some Forms, in some dialogues,
are separate. But he and others are incorrect to suggest that Plato, beginning
with the Phaedo, heralds separation as a new feature of Forms. On the contrary,
so far from this being the case, Plato never even says that Forms are separate;
it proves surprisingly difficult to uncover any commitment to separation; and
commitment to it emerges in unexpected ways and in unexpected cases.8
Fine understands Plato's Forms to be universals, and thus when she
denies that the independent existence of Forms is a key component of
Plato's Theory of Forms, what she is denying is that it is especially
important to Plato that the Forms be able to exist uninstantiated. But
whatever Forms are, Fine's conclusion about the role of separation is,
as she admits, very surprising. T. H. Irwin, for example, who agrees
that Forms are universals, thinks that even though Plato does not formulate it clearly and even though his arguments justify only the claim
that Forms are not defined in terms of sensible properties and are not
identical with such properties (which leaves open the possibility that
they might be identical with nonsensible properties of sensible objects),
he does in fact believe that he has established their capacity for independent existence.9
Since my concern is with the nature of separation, Irwin's conclusion
need be pursued only if it is plausible to hold that Forms are universals
6 Fine (1984), 35.
7 See Fine (1984), 33.
8 Fine (1984), 33-34. According to Fine the most likely candidates for separate Forms
are Forms of artefacts (as in the Cratylus and Republic X) since there would have been
a time when they existed which was prior to the work of human artisans, that is to say,
they would have existed in the absence of instances (76). If the account of creation in
the Timaeus is taken in such a way that, while Forms have always existed, there was a
time at which there were no sensibles, then, says Fine, many Forms are separate (79).
However, even in that case not all Forms would be so. For even before creation there
are some Forms that would have instances; traces of fire are found in the chaos, the
demiurge exemplifies justice and goodness, and so on (79).
9 See Irwin (1977), 154-55.
and that * separation' does in fact mean a capacity for independent

existence. That Forms are not universals will be argued in a later chapter. But as to the question of whether a capacity for independent existence is what 'separation' means, it seems to me that the case has not
been proved. In Metaphysics V 11 natural priority is defined in terms
of independent existence and then the definition is illustrated by an
appeal to Plato, that is to say, by the Forms, and there is no mention
at all of separation. Admittedly, in Eudemian Ethics 18 when independent
existence and natural priority are said to be characteristic of the Form
of the Good, that the Idea of the Good is also said to be separable is
added in a way which suggests its connection with these other notions.
But the difficulty is that 'separation' need not mean a capacity for independent existence for separation and natural priority to be linked;
rather, if a capacity for independent existence were merely entailed by
the nature of the Forms, that is to say, by the attributes that Forms are
able to have in virtue of being separate, the connection between separation and the natural priority of the Forms would also follow.10
I have argued against Fine that neither Eudemian Ethics I 8 nor Metaphysics V 11 clearly identifies a capacity for independent existence
with separation. But there is also another reason for holding that a
capacity for independent existence is not what Aristotle means. For if
one takes Aristotle to attribute to Plato the separation of the Forms,
meaning by that their capacity for independent existence, then Aristotle
seems to attribute to Plato an argument for separated Forms which
would have no persuasiveness at all. That is to say, as will be made clear
in the passages to be discussed in Section II of this chapter, Aristotle
takes flux to be Plato's primary motive for the postulation of separated
Forms. Whether he is right in this assessment is disputed.11 But if being
in flux is supposed to be an obstacle to being knowable and indeed fully
real, then given that there are in fact things that meet these criteria,
10 For example, in his commentary on 12l7bl4-15 Woods (1992), 68, refers the reader
to his discussion of 1218al15 where he says of separation: "[Unlike Nicomachean
Ethics I 1096a34-b5, in Eudemian Ethics I 8] there is mention also of the status of
the Form as something separate. Does that mean that it does not depend for its
existence on particulars? Or is it rather that it has to be conceived of as a distinct
good? The argument does seem to assume that the Form of the Good is itself a good,
and argue from that that it will be a good after a different fashion from other goods,
and hence not the common character" (80).
11 Owens (1963), 199, cites Plato's concentration on definitions. Cornford (1939), 74,
and, more cautiously, Burnyeat (1979), 59, cite recollection as the motive; Mabbott
(1926), 74, thinks that separation is entailed by the fact that Forms are originals of
which phenomena are copies. However, if the reason the originals must be separate
is that only in this way can they be perfect, for example, and if, as Plato sees it,
whatever is perfect and thus a suitable object of definition must be eternal and unchanging, then it seems that Aristotle is still right: the basis for the separation of the
Forms is the argument from flux. As will become clear in Chapter IV, I take recollection to be a consequence of the inaccessibility of Forms to our present experience.
what is proved is that there exist some things that are not in flux. That
is to say, what has been proved is, at most, that there are things that
exist besides - in addition to - those that are in flux. Whether they
would exist even if those that are in flux did not do so is clearly another
question.
In short, I have no inclination to deny that Plato intended the Forms
to have a capacity for independent existence and indeed neither would
I deny that Aristotle thought that this was Plato's intention. But if a
capacity for independent existence is what is meant by 'separation', then
it would seem that Aristotle must attribute to Plato an argument for
the existence of Forms which is flagrantly invalid. In fact Fine offers a
way to avoid this conclusion. What she suggests is that Aristotle did not
take Plato to be arguing directly from flux to Forms capable of existing
independently but rather to be using the assumption that there is knowledge despite the existence of flux to establish only that there are Forms
which are universals, not that they are separate. Recalling that separation is taken by Fine to be a capacity for independent existence, why
then would Aristotle accuse Plato of separating the Forms? What Fine
proposes is that these universals or Forms are, in Aristotle's interpretation of Plato's ontology, the only available candidates for substances,
and substances, so Aristotle holds, are separate.12
Thus as Fine understands it, the argument for the separation of the
Forms - which, if Forms are universals and 'separation' means a capacity
for independent existence, is, as I have said, just an argument that
universals can exist uninstantiated - need not after all be invalid; it is
not so because the need for something which is not in flux is not by
itself supposed to be sufficient to establish that Forms are separate. Yet
even if the argument is not invalid, neither is it in Aristotle's view sound.
Rather what Aristotle holds, according to Fine, is that Plato errs in his
belief that Forms, being universals, can be substances.
In fact I would agree with Fine that Aristotle believes that universals
cannot be substances. Nevertheless I disagree with her understanding
of separation and in consequence with her construal of how, from the
fact of flux, Aristotle supposes Plato to have arrived at separate Forms.
Indeed the passage in Metaphysics XIII 9 from which the quotation at
the head of the chapter was taken and upon which Fine founds her
argument would not seem most naturally to support her interpretation.
Aristotle says:
as regards those who believe in the Ideas one might survey at the same time
their way of thinking and the difficulties into which they fall. For they at the
same time treat the Ideas as universal, and again as separable [choristas] and
individual. That this is not possible has been shown before. The reason why
12 Fine (1984), 51-53.
10
those who say substances are universal combined these two views in one, is that
they did not make them the same13 [autas] with sensible things. They thought
that the sensible particulars were in a state of flux and none of them remained,
but that the universal was apart [para] from these and different. And Socrates
gave the impulse to this theory, as we said before, by means of his definitions,
but he did not separate them from the particulars; and in this he thought rightly,
in not separating them. This is plain from the results; for without the universal
it is not possible to get knowledge, but the separation is the cause of the objections that arise with regard to the Ideas. His successors, treating it as necessary, if there are to be substances besides the sensible and transient substances,
that they be separable, had no others, but gave separate existence to these
universally predicated substances, so that it followed that universals and individuals were almost the same sort of thing. (1086a31-bll)
Passages such as this will be mined for another purpose in Chapter
V. At present, however, my point is that when Aristotle says that Forms
are both separate and universal, his diagnosis of Plato's error would
seem to be not, as Fine claims, that he made the Forms universals while,
if they are to be substances, they must be separate and therefore particular. Rather Aristotle's complaint is that Plato failed to make Forms
the same as sensible things (1086a36). To be sure, this claim could mean
that he failed to make them the same in kind - in other words, that he
failed to make them particulars - but since in fact Aristotle here also
says that Plato makes Forms both particular and universal, this would
be, to say the least, a peculiar complaint and, anyway, the manner in
which he continues suggests a different explanation. For if the reason
for postulating something not the same as sensibles is that the latter
are in flux, this reason tells neither for nor against the universality or
particularity of Forms - it counts only for the postulation of something
that does not have the property of being in flux. Thus their numerical
distinctness from things in flux would seem to be what is asserted.
II
Even though Plato's Forms do have the capacity to exist independently,
as I have argued, a capacity for independent existence would not seem
to be what Aristotle has in mind when he says that Plato's Forms are
separate. Rather, as I see it, the attribution of a capacity for independent
existence to the Forms is an assertion of the ontological priority they
have over the phenomena from which they are held to be separated.
But if this is so, it is necessary to establish what 'separation' does mean,
and I have already given a reason for favoring numerical sameness. In
Metaphysics I 6 987a32-blO when Aristotle gives his account of the
13 Barnes (1984) has 'identical'. My reason for preferring 'same' will become apparent
in Chapter II.
11
origin of Plato's Theory of Forms and addresses the question of separation, he says:
For, having in his youth first become familiar with Cratylus and with the Heraclitean doctrines (that all sensible things are ever in a state of flux and there
is no knowledge about them), these views [Plato] held even in later years. Socrates, however, was busying himself about ethical matters and neglecting the
world of nature as a whole but seeking the universal in these ethical matters,
and fixed thought for the first time on definitions; Plato accepted his teaching,
but held that the problem applied not to any sensible thing but to entities of
another kind - for this reason, that the common definition could not be a definition
of any sensible thing, as they were always changing. Things of this other sort, then, he
called Ideas, and sensible things, he said, were apart from these, and were all called
after these; for the multitude of things which have the same name as the Form
exist by participation in it. (emphasis added)
The same story seems to be told in Metaphysics XIII 4 1 0 7 8 b l 2 - 1 0 7 9 a 4
where Aristotle begins by saying:
The supporters of the ideal theory were led to it because they were persuaded
of the truth of the Heraclitean doctrine that all sensible things are ever passing
away, so that if knowledge or thought is to have an object, there must be some
other [heteras] and permanent entities, apart [para] from those which are sensible; for
there can be no knowledge of things which are in a state of flux. (1078bl2-17;
emphasis added)
After a discussion of Socrates, Aristotle then explains the difference
between his views and those of Plato:
For two things may be fairly ascribed by Socrates - inductive arguments and
universal definition, both of which are concerned with the starting-point of
science. But Socrates did not make the universals or the definitions exist apart
[chorista]; his successors, however, gave them separate existence, and this was
the kind of thing they called Ideas. (1078b27-32)
What are we to make of these passages? I have argued against Fine
that Eudemian Ethics I 8 does not show 'separation' to mean a capacity
for independent existence and that, because an argument from flux to
independent existence (unless taken in something like the way in which
Fine takes it) would be invalid, a capacity for independent existence is
not likely to have been what Aristotle meant. Of course, it is possible
that Plato's argument for the separation of the Forms, assuming that
Aristotle is right about the origins of the Theory, just is invalid. Nevertheless, as I have already said, the difficulty with the proposal that Plato
argues from the need for something that is not in flux to the need for
something that can exist without the existence of whatever is in flux is
that the two notions seem too obviously unconnected for fallacious
argument to occur. Besides, further and, I think, decisive evidence
12
against the supposition that a capacity for independent existence is what

is meant by 'separation' is found as XIII 4 continues. Aristotle says:
Therefore it followed for them, almost by the same argument, that there must
be Ideas of all things that are spoken of universally, and it was almost as if a
man wished to count certain things, and while they were few thought he would
not be able to count them, but made them more and then counted them; for
the Forms are almost more numerous than the groups of sensible things, yet
it was in seeking the causes of sensible things that they proceeded from these
to the Forms. For to each set of substances there answers a Form which has
the same name and exists apart [para] from the substances, and so also in the
other categories there is one character common to many individuals, whether
these be sensible or eternal. (1078b32-1079a4)
The passage is parallel to Metaphysics I 9:
But as for those who posit the Ideas as causes, firstly, in seeking to grasp the
causes of the things around us, they introduced others equal in number to these,
as if a man who wanted to count things thought he could not do it while they
were few, but tried to count them when he had added to their number. For
the Forms are practically equal to or not fewer than the things, in trying to
explain which these thinkers proceeded from them to the Forms. For to each
set of substances there answers a Form which has the same name and exists
apart from the substances, and so also in the case of all other groups in which
there is one character common to many things, whether the things are in this
changeable world or are eternal. (990a34b8)
What Aristotle intends to attribute to Plato when he says that the
Forms exist apart from sensible things becomes in this cheap shot at
his predecessor very clear. If Plato's Theory is almost as if a man wished
to count certain things, and while they were few thought that he would
not be able to count them, but made them more and then counted
them, separation - the target of Aristotle's ridicule must be the supposed numerical distinctness of Forms from sensible things.
Ill
In considering those passages in which Aristotle criticizes Plato for
separating the Forms, I have argued that, while Plato did take the Forms
to be capable of independent existence and while Aristotle thought that
he did so, such is not the basis of the latter's complaint. Rather, what
the argument from flux, the argument that Aristotle cites as the origin
of the Platonic Theory, can reasonably be supposed to show and what
Aristotle takes it to be supposed to show - is that Forms are numerically
distinct from sensibles. But if this is so, then when Aristotle objects to
13
Plato's (as opposed to Socrates') view of Forms, it is most plausible to

conclude that by 'separation' Aristotle means numerical distinctness.14
Of course I have not tried to prove that Aristotle's interpretation of
Plato, as I understand it, is correct, and it would take me far beyond
the scope of this work to try to do so. Nevertheless since it is prima
facie more plausible to suppose that Aristotle did understand Plato's
theory than that he did not, it cannot be denied that Plato's actual views
are relevant, and although Fine denies that separation is a central tenet
of Plato's Theory, even she concedes that her stance is unusual. More
troublesome therefore is the fact that interpreters of Plato who do take
separation to be important do not always have the same thing - or even
one thing - in mind when they say that the Forms are separate. Richard
Patterson, for example, says of the Form of the Good, "First, the Good
is separate from all sensible things. It is not located where they are, not
contaminated by any admixture with them, nor dependent on them for
being what it is."15
None of these characteristics is, as it happens, what Fine suggests
Aristotle takes the separation of the Forms to mean; however when
Gregory Vlastos argues that Plato's proposal that the Forms exist "themselves by themselves" is equivalent to Aristotle's claim that Plato's Forms
exist separately16 he goes on to endorse Fine's interpretation.17 Yet what
is to be noticed is that all the features mentioned by both Patterson
and Vlastos could be seen as consequent upon the nature of separated
Forms if 'separation' were understood as numerical distinctness. One
of Vlastos's concerns can make this point clear. When he argues that
the separation of Forms means that Forms can exist even if uninstantiated, Vlastos finds some difficulty in the fact that Plato also wants to
hold that the relation between Forms and phenomena is asymmetric,18
something that independent existence as such does not entail; the soul,
which is said to be separate from the body, can exist without the body,
and the body, for a short time at least, can also exist without the soul.19
But if separation means, as I have suggested, numerical distinctness,
the symmetric relation found in the case of body and soul is unproblematic, while what needs to be said of the asymmetric character of the
relation between the Forms and phenomena is that it is not to be accounted for by separation as such. Rather just as with a capacity for
14 Numerical distinctness seems also to have a part in the passage quoted from Woods
(1982), 80, in my n. 10. That is, if the Form of the Good is "a distinct good" and
"good after a different fashion from other goods, and hence not the common character," it is hard to see how it could be anything other than numerically distinct.
15 Patterson (1985), 123; see also 129. Spatial separation may also be found in Else (1936).
16 Vlastos (1991), 256.
17 Vlastos (1991), 265.
18 What Vlastos (1991), 259-60, says is that the relation is "antisymmetric." What he
seems to have in mind is a modal formulation of asymmetry according to which x can
exist without y and not conversely.
19 Vlastos (1991), 260.
14
independent existence, it is at precisely this point that there must be

appeal to other facts about Forms, that they are necessarily eternal or
that they are models, for example. But if this is so, then even when
interpreters argue for - or assume - a different account of separation,
there seems to be nothing in such views that is inconsistent with the
claim that numerical distinctness from phenomena is what the separation of the Forms means.
To return then to Aristotle, we have seen that Aristotle says quite
clearly that separation is the cause of the objections that arise with
regard to the Ideas. (In Metaphysics XIII 4 and I 9 the term in question
is para, often translated as ''exists apart." Where chbristos does occur,
as in the passages from Metaphysics VII 1 and Eudemian Ethics I 8, there
is no reason to think any different sort of separation is intended; Metaphysics XIII 9, as we have seen, uses both.20) Of course, I certainly do
not contend that every criticism of the Theory of Forms Aristotle offers
is connected with the separation of the Forms. To take only Metaphysics
I 9, Aristotle there objects that Plato will be saddled with Forms of
negations, that there are unacceptable consequences to taking Forms
to be numbers, and so on. Nevertheless Aristotle repeatedly indicates
that separation is the crux, in the sense that it is peculiarly characteristic
of Plato's Theory and peculiarly troublesome. Thus if I am right that
separation is numerical distinctness, the question that must be considered is just this: Why, exactly, should Aristotle think the numerical
distinctness of Forms from phenomena to be objectionable?
IV
When Aristotle wants to object to the numerical distinctness of Forms
from sensibles, what he sometimes says is that whatever is numerically
distinct will be particular and therefore unknowable; at other times he
offers a variation on this objection, arguing that Forms, being knowable,
must be universals, in which case if they are also particulars, there results
incoherence ("this is not possible" (1086a34)). That whatever is numerically distinct is particular is a claim Aristotle more often asserts
than argues for; perhaps it is thought to follow from the very concept
of numerical distinctness. But the claim leads to some harsh descriptions
of Plato's Theory. For example, at Metaphysics III 2 997b5-12 Aristotle
says:
the most paradoxical thing of all is the statement that there are certain things
besides those in the material universe, and that these are the same as sensible
things except that they are eternal while the latter are perishable. For they say
there is a man-in-himself and a horse-in-itself and health-in-itself, with no further
20 So does Metaphysics XI 2. See 1060a3-b2.
15
qualification, - a procedure like that of the people who said there are gods,
but in human form. For they were positing nothing but eternal men, nor are
they making the Forms anything other than eternal sensible things, (cf. VII 16
1040b32-34)
This complaint is summarized in Nicomachean Ethics I 6 in a way that
makes the consequences clear:
And one might ask the question, what in the world they mean by 'a thing itselF,
if in man himself and in a particular man the account of man is one and the
same. For in so far as they are men, they will in no respect differ; and if this is
so, neither will there be a difference in so far as they are good. (1096a34-b3)
Thus Aristotle, beginning from their numerical distinctness, takes
Forms to be particulars and proposes that, if intelligible at all, Forms
are no more than eternal sensibles. Moreover in the Third Man Argument he seems to claim that the numerical distinctness of Forms will
lead to an infinite regress as well. Although the Third Man Argument
is no more than alluded to in Metaphysics I 9 (990bl7), there is a brief
discussion in the Sophistical Refutations. There Aristotle says: "Again,
there is the argument that there is a third man distinct from man and
from individual men. But 'man', and indeed every general predicate,
signifies not an individual, but some quality, or quantity or relation, or
something of that sort" (178b36-39). Even though the steps leading to
the troublesome third man are not spelled out, it seems evident that
Aristotle believes that for Plato 'man' signifies a particular and that this
is cited as the cause of a regress of distinct Forms. But since a lengthier
version of the Third Man Argument as given in his lost essay Peri Idem
is preserved in Alexander of Aphrodisias's commentary on the first book
of the Metaphysics, I will turn to the presentation of the Third Man
Argument in Alexander.
At 84, 22-85, 3 Alexander records the following objection to the
Theory of Forms:
If what is predicated truly of more than one thing is also [some] other thing
apart from the things of which it is predicated, being separated from them (for
this is what those who posit the Ideas think they are proving; for the reason
why, according to them, there is something, man-himself, is because 'man' is
predicated truly of particular men, who are more than one, and is other than
particular men) - but if this is so, there will be some third man. For if [the
'man'] predicated is other than those of whom it is predicated and subsists by
itself, and 'man' is predicated both of particular men and of the Idea, there
will be some third man apart from both particular men and from the Idea. And
in this way there will be still a fourth man, the one predicated of the third man
and of the Idea and of particular men, and similarly a fifth, and so on ad
infinitum.2X
21 Pages 121-22 of the Dooley translation of Alexander of Aphrodisias (1989). The
bracketed material is supplied by Dooley.
16
It has sometimes been found puzzling that in this argument, unlike

the 'Third Man" of the Parmenides, separation is one of the premisses.
But the explanation, it seems to me, is just that since the goal of the
argument is to establish that Plato's Theory leads to regress, there is
no reason not to begin with Plato's desired conclusion, namely, the
existence of a (first-level) Form for man, numerically distinct from sensible particulars.22 Moreover if the humanity shared by individual human
beings is supposed to lead to the conclusion that the Form for man is
something numerically distinct from phenomena and if to be a numerically distinct human being is to be a particular, then something like
self-predication seems at least a plausible charge. To be sure, the Third
Man Argument has been analyzed in immense detail, and some interpreters think that what has been called self-predication need not be
part of it.23 But even if they are right, as long as it is agreed that Aristotle
believes that Plato can be forced to admit that his views about the Forms
will result in regress, the connection between the Third Man Argument
and my claim that * separation' means numerical distinctness clearly
remains.
Following an argument in Section III for the conclusion that when
Aristotle criticizes Plato for separating the Forms what he has in mind
is their numerical distinctness from sensible objects, I have in this section considered passages where Aristotle links the numerical distinctness
of Forms from sensibles with their particularity and that with the Third
Man Argument. But particularity, Aristotle thinks, also creates another
problem, namely, the unknowability of the Forms. This difficulty will
be the subject of the next section.
In Metaphysics VII 15 Aristotle says:

As has been said, people do not realize that it is impossible to define in the
case of eternal things, especially those which are unique, like the sun or the
moon. For they err not only by adding attributes after whose removal the sun
would still exist, e.g. 'going round the earth' or 'night-hidden' . . . but also by
the mention of attributes which can belong to another subject; e.g. if another
thing with the stated attributes comes into existence, clearly it will be a sun;
the formula therefore is general. But the sun was supposed to be an individual,
22 Aristotle's objection at 990b 15 to Ideas of relations, as explained in Alexander, also
begins with Forms: "Again, if the equal is equal to an equal, there would be more
than one Idea of the equal; for the equal-itself is equal to the equal-itself, for if it
were not equal to anything it would not be equal at all" (83, 26-28). Dooley, trans.
(1989), 120. For competing explanations of the presence of separation in the premisses
of the Third Man Argument, see Fine (1982), 161-69, and Code (1985), 104-10 and
323-26.
23 See Code (1985), 106, and Lewis (1991), 15-24, 33-43.
17
like Cleon or Socrates. Why does not one of the supporters of the Ideas produce
a definition of an Idea? It would become clear, if they tried, that what has now
been said is true. (1040a27-b4)
Aristotle's support for the claim that particulars are unknowable is
in part the alleged impossibility of counterexample: Just try to produce
a definition of Cleon or the sun or an Idea. But, of course, one reason
a definition of Cleon cannot be produced is that Cleon is transitory.
Earlier in the same chapter Aristotle has said:
there is neither definition nor demonstration of sensible individual substances,
because they have matter whose nature is such that they are capable both of
being and of not being; for which reason all the individual instances of them
are destructible. If then demonstration is of necessary truths and definition
involves knowledge, and if, just as knowledge cannot be sometimes knowledge
and sometimes ignorance, but the state which varies thus is opinion, so too
demonstration and definition cannot vary thus, but it is opinion that deals with
that which can be otherwise than as it is, clearly there can neither be definition
nor demonstration of sensible individuals. (1039b27-1040a2)
Certainly the correlation found in this argument between different

states of mind - knowledge, opinion, and ignorance - and distinct objects is reminiscent of Republic V.24 For Aristotle as for Plato, what is
knowable must be necessary, eternal, and unchanging (e.g., Posterior
Analytics 7lb9-16, 73a21-3, 74b6, 88b30-89al0; Nicomachean Ethics
1139bl8-23, 1140a31-b3), constraints that, as we will see, significantly
affect his account of substance. But it is hard to see that this sort of
objection affects the knowability of either the sun, which, unlike Cleon,
is supposed to be eternal and unchanging, or of Platonic Forms; indeed
even in the case of the sun, the removal of accidents contemplated in
Metaphysics VII 15 would seem to be counterfactual. Moreover, whatever one says about the sun, it is not clear that one can use such a
reason - namely, the presence of attributes that might, even if only in
imagination, be removed - to claim that Forms are unknowable.25
In short, in the case of Forms, it seems most reasonable to suppose
the obstacles to definition arise not from the possible removal of attributes but rather from the fact that any attributes one gives could
belong to more than one thing - not just individually, for that would
not be troublesome, but taken together. In other words, the complaint
seems to be that if one gives a definition, that definition might be
24 Although Aristotle, like Plato, moves freely between knowledge of eternal truths (for
example, that the diagonal is incommensurable) and knowledge of objects, in the
Metaphysics it is evident that it must be the knowledge of objects - of substances that is fundamental.
25 One would, of course, be able to "remove properties" if Forms were described in a
sufficiently peripheral way, for example, "the first Form recollected by the slave," but
perhaps Plato could say that such descriptions do not correspond to properties of
the Form.
18
satisfied by more than one thing. That this is so would seem to be

confirmed in yet another passage in VII 15:
Nor is it possible to define any Idea. For the Idea is, as its supporters say, an
individual, and can exist apart; and the formula must consist of words; and he
who defines must not invent a word (for it would be unknown), but the established words are common to each of a number of things; these then must apply
to something besides the thing defined; e.g. if one were defining you, he would
say 'an animal which is lean' or 'white', or something else which will apply also
to some one other than you. (1040a8-14)
If the name or the definition of the Form can apply to only one thing,
its name is like inventing a word - or to put the point another way, its
name would be a proper name.
I have argued that Aristotle seems to believe separated Forms are
particulars that in virtue of their uniqueness would be unknowable. Of
course to suppose as Aristotle does that the name would apply only to
the Form overlooks - or contests - the supposed relation of likeness
between Forms and sensible particulars. But that much of the rest of
Aristotle's epistemological attack on the Theory also hinges, directly or
indirectly, on separation is apparent. In Metaphysics I, for example, he
says: "But again [Forms] help in no way towards the knowledge of the
other things (for they are not even the substance of these, else they
would have been in them) . . ." (991al2-13). His concern, expressed
in Nicomachean Ethics I 6, about the unattainability of the Form of the
Good and its uselessness even for knowledge of attainable goods illustrates the general problem. Here Aristotle says:
And similarly with regard to the Idea; even if there is some one good which is
universally predicable of goods or is capable of separate and independent existence, clearly it could not be achieved or attained by man. . . . Perhaps, however, some one might think it worth while to have knowledge of [the Form of
the Good] with a view to the goods that are attainable and achievable; for having
this as a sort of pattern we shall know better the goods that are good for us,
and if we know them shall attain them. This argument has some plausibility,
but seems to clash with the procedure of the sciences; for all of these, though
they aim at some good and seek to supply the deficiency of it, leave on one side
the knowledge of the good. Yet that all the exponents of the arts should be
ignorant of, and should not even seek, so great an aid is not probable. (1096b311097a8)
While numerical distinctness alone would not explain the unattainability
of The Good (goodness in carpentry is distinct from goodness in music
and both are attainable), the difference in nature between phenomena
and separated Forms can plausibly be said to offer such an explanation;
after all, even though Forms are said to be patterns for phenomena,
nevertheless if they are to be eternal, they must in significant ways be
unlike them. Moreover, having doubted the usefulness of knowledge of
19
the Form of the Good on the grounds that craftsmen do not use it,
Aristotle then seems to suggest that their lack of attention to the Form
should undermine our confidence in its existence.
In the two passages just cited, Aristotle uses the separation of Forms
as grounds for doubting that knowledge of Forms, even should there
be such, would yield knowledge of the sensible world. But in Metaphysics
I 9 he also mounts a general attack on the acquisition of knowledge of
the elements of things, an attack that seems to be aimed at Forms:
And how could we learn the elements of all things? Evidently we cannot start
by knowing something before. For as he who is learning geometry, though he
may know other things before, knows none of the things with which the science
deals and about which he is to learn, so is it in all other cases. Therefore if
there is a science of all things, as some maintain, he who is learning this will
know nothing before. Yet all learning is by means of premises which are (either
all or some of them) known before, - whether the learning be by demonstration
or by definitions; for the elements of the definition must be known before and
be familiar; and learning by induction proceeds similarly. But again, if the science
is innate, it is wonderful that we are unaware of our possession of the greatest of sciences.
Again, how is one to know what all things are made of, and how is this to be
made evident? This also affords a difficulty; for there might be a conflict of
opinion, as there is about certain syllables; some say za is made out of s and d
and a, while others say it is a distinct sound and none of those that are familiar.
Further, how could we know the objects of sense without having the sense in
question? Yet we should, if the elements of which all things consist, as complex
sounds consist of their proper elements, are the same. (992b24-993al0; emphasis added; boldface indicates Barnes's emphasis)
The problems attendant to the Theory of Recollection will be considered in Chapter IV. But it is already apparent that in arguments such
as these Aristotle is contending, in effect, that in addition to all the
metaphysical problems of the Theory, the Forms cannot accomplish
what he believes to be Plato's aim, namely, to explain how there can
be knowledge of anything in a world that seems to be full of flux. Yet
even if Plato's Theory is thought by Aristotle to fail, the questions it
was intended to answer remain, and it is from within their presuppositions that Aristotle too has to wrestle with problems of universality,
particularity, ontological priority, and knowability. As he says in Metaphysics III:
If [the first principles] are universal, they will not be substances; for everything
that is common indicates not a 'this' but a 'such', but substance is a 'this'. And if we can actually posit the common predicate as a single 'this', Socrates
will be several animals - himself and man and animal, if each of these indicates
a 'this' and a single thing. - If, then, the principles are universals, these results
follow; if they are not universals but of the nature of individuals, they will not
be knowable; for the knowledge of anything is universal. (1003a715)
20
Thus it is that Aristotle also will try to say what substances or forms are
and how they have the universality and particularity he claims is necessary. Likewise he will try to explain what the relation between substances or forms and sensibles is, and how, by encountering sensible
objects, one knows substances. The examination of his views in the
chapters to come will test my contention that what separates Plato and
Aristotle is - separation.
II
Referential Opacity in Aristotle
At the beginning of Metaphysics VII 6 Aristotle inquires whether each

thing and its essence are the same or different, an inquiry that, he claims,
is of use for the investigation of substance. In Chapter III I will argue
that VII 6 does indeed tell us much about substance. What is of relevance in this chapter, however, is Aristotle's preparedness to raise such
a question. What Aristotle says of the inquiry into the sameness of each
thing and its essence is that each thing is thought to be not different
from its substance and that the essence is said to be the substance of
each thing (1031al5-19). It is easy enough to suppose that 'is not
different from' means 'is the same as* and that 'is the same as' means
'is identical with'. If this assumption is made,1 there would not seem
much need for inquiry: If each thing is identical with its substance and
its substance is identical with its essence, then obviously each thing is
identical with its essence.
Nevertheless Aristotle does inquire about the relation between each
thing and its essence, a way of proceeding that suggests that the sort
of sameness that interests him may in fact not be identity. It is because
problems of sameness are made evident by the occurrence of referential
opacity that we focus on that concept in this chapter.2 After presenting
several responses to referential opacity in Aristotle, I offer my own
analysis of why, in Aristotle, failures of substitutivity occur. In Chapter
III I will consider the significance of referential opacity for Aristotle's
1 It needs also to be assumed that Aristotle does not intend to make a distinction between
'substance' and 'substance of. I believe that he does not, for reasons that will become
apparent in Chapter III.
2 Though I couch the discussion in terms of referential opacity, my intention is to give
a description of certain problems without making any assumptions that would prejudge
questions about the kinds of sameness to be found in Aristotle. That is to say, I do
not intend that 'referential opacity' be denned in the manner of Lewis (1991), 90: "A
context, A, is referentially opaque at a given position, if and only if for expressions a
and (3, such that a and 0 denote the identical entity, a context, A', is like A except that
A contains a in that position but A' contains /3 there, and A and A' do not denote the
same thing (or if A and A' are sentences, A and A' do not have the same truth value)."
21
22
alternative to Plato's Theory of Forms, namely, the account of substance

given in Metaphysics VII-VIII.
It is W. V. O. Quine who seems to have introduced the term 'referential

opacity'. In "Reference and Modality" he points out that a name may
occur in a statement, for example, 'Tully' in Tully denounced Catiline',
and yet not occur referentially in a longer statement which is formed
by embedding the statement in the context 4s unaware that . . .' or
'believes that . . .', as for example in 'Philip is unaware that Tully denounced Catiline'. Quine proposes that we speak of contexts such as
'is unaware that . . .' as "referentially opaque," 3 and because substitutivity is to be expected where there is identity, referential opacity can
be puzzling. As Quine says:
One of the fundamental principles governing identity is that of substitutivity
or, as it might well be called, that of indiscernibility of identicals. It provides that,
given a true statement of identityy one of its two terms may be substituted for the other
in any true statement and the result will be true.4
Not all failures of substitutivity are, of course, genuinely problematic.

Though 'Cicero' and 'Tully' refer to the same person, in " 'Cicero'
contains six letters" replacement of one name with the other changes
the sentence from true to false,5 yet here the paradox is not deep; it
need only be said that the statement is not about Cicero at all but about
a word that is a name for him. But the difficulty is that the opacity
found in modal and intentional contexts is not so easily dismissed. After
all, if a and b are identical, all the properties of a would seem to be
properties of b, and if all the properties of a are properties of b, then
surely it does seem reasonable that in any true statement in which something is predicated of a, if what is predicated of a had been predicated
of by truth should be preserved.
This is not the place to describe contemporary solutions, except insofar as they also have been proposed for the interpretation of Aristotle.
But the problem of referential opacity itself is of major significance in
Aristotle just because there are so many contexts - and such important
ones - that seem to be opaque. To take some examples, we can know
Coriscus but fail to know the masked man (who is Coriscus); Socrates
3 Quine (1953), 142. Quine says that the term 'referentially opaque' is roughly the opposite
of Russell's 'transparent' in Appendix C to Principia Malhematica, 2d ed., vol. I.
4 Quine (1953), 139. See also Carnap (1947), 98, who gives a clear statement of the
principle even though he rejects it. He says: "If two expressions name the same entity,
then a true sentence remains true when the one is replaced in it by the other; in our
terminology: the two expressions are interchangeable (everywhere)."
5 Quine (1953), 139.
REFERENTIAL OPACITY IN ARISTOTLE
23
is accidentally the same as the seated man, but Socrates, and not the
seated man, is essentially the same as his essence or form; the builder
is a cause of the house, and the musician is not (even if the builder is
a musician); the man, but not the pale one, becomes cultured. This list
- to which could be added '. . . is a substance', '. . . is per se [roughly,
'. . . is necessarily'] F', '. . . is a species o f . . . ' , ' . . . is the genus of. . .',
and so on, as well as \ . . moves', '. . . builds', * . . . sculpts', and all
other action verbs - includes not only the modal and propositional
attitude contexts that many philosophers besides Aristotle have thought
to be opaque, but also contexts that might seem to be paradigm cases
of the transparent and contexts involving Aristotle's technical vocabulary.
So far I have explained what is meant by 'referential opacity' and I
have described why, where identity is concerned, failures of substitutivity would seem to be problematic. But in fact whether Aristotle had
such a notion as identity is disputed. In the Topics, he himself expresses
what appears to be the principle of the indiscernibility of identicals or,
as it is often called, Leibniz's Law:
Moreover, examine them in the light of their accidents or of the things of which
they are accidents; for any accident belonging to the one must belong also to
the other, and if the one belongs to anything as an accident, so must the other
also. If in any of these respects there is a discrepancy, clearly they are not the
same. (152a33-37)
Or again:
Speaking generally, one ought to be on the look-out for any discrepancy anywhere in any sort of predicate of each term, and in the things of which they
are predicated. For all that is predicated of the one should be predicated also
of the other, and of whatever the one is a predicate, the other should be a
predicate as well. (152b25-29)
These passages and others from the Topics (for example, 132a2728) are usually taken as good evidence that, at least at the time of that
treatise, Aristotle held a standard view of identity.6 But there is also
reason to have qualms. At De Interpretatione 1121a 7-14, for example,
Aristotle says:
6 In his definition of 'numerical sameness' in Topics I 7, Aristotle distinguishes three
senses of the expression, namely, sameness where there is an alternative name or
definition ("its most literal and primary use"), second, sameness rendered by a nonessential property peculiar to that kind of thing, and third, sameness drawn from
accident, as in describing Socrates as "the creature who is sitting" (103a23-31). While
these senses may seem to foreshadow later distinctions, 152a33-37 and 152b25-29,
just quoted, make it clear that, with regard to substitutivity, no use is made of them
in the Topics. Moreover Aristotle's example of the first and most literal sense is at least
peculiar since the example (doublet and cloak) is not, as it stands, an example of
numerical sameness.
24
Of things predicated, and things they get predicated of, those which are said
accidentally, either of the same thing or of one another, will not be one. For
example, a man is white and musical, but * white' and * musical' are not one,
because they are both accidental to the same thing. And even if it is true to say
that the white is musical, * musical white' will still not be one thing; for it is
accidentally that the musical is white, and so 'white musical' will not be one.
Moreover in Sophistical Refutations 24 Aristotle clearly says that numerical sameness is not sufficient for substitutivity, a claim that raises
serious problems of interpretation. Nicholas White, for example, holds
that Aristotle's grasp of identity weakens7; Gareth Matthews argues that,
on the contrary, the restrictions Aristotle introduces show that he is
moving toward an understanding of the notion.8 What is clear is that
Aristotle thinks or comes to think that there are two varieties of numerical
sameness, essential sameness and accidental sameness {Metaphysics V 9
10l7b27-1018a9), such that substitutivity is said to hold in the first case
in certain contexts where it fails in the second. The problem therefore
is this: If numerical sameness just is identity, as elucidated by Leibniz's
Law, what can the distinction between essential and accidental sameness
come to and how can it be used to explain certain failures of substitutivity?
On the other hand, if either accidental sameness or essential sameness
(or both) is other than identity, what is their relation to identity?
II
One strategy for dealing with failures of substitutivity is that of Frege.
When we are presented with a belief context or a modal context in
which substitutivity seems to fail (for example, 'it is necessary that the
Morning Star is the Morning Star' is true but 'it is necessary that the
Morning Star is the Evening Star' is false), Frege's response is to say
that, in such sentences, 'the Morning Star' does not, despite appearances, refer to the Morning Star and 'the Evening Star' does not, despite
appearances, refer to the Evening Star. That is, in modal and belief
contexts, Frege claims, 'the Morning Star' refers not to the ordinary
referent of the expression 'the Morning Star' (namely, Venus) but rather
to the sense of that expression.9 Since 'the Morning Star' in 'it is necessary that the Morning Star is the Morning Star' does not, on Frege's
view, refer to the Morning Star but to its sense and since the sense of
'the Morning Star' is different from the sense of 'the Evening Star', it
is hardly surprising that substituting one for the other fails to preserve
7 White (1971), 177. I will not argue against White's view that, after the Topics, Aristotle
confuses 'X and Y are the same' with 'X and Y are one', where the latter is taken to
mean that both X and Kmake up a single entity (see 187). However both Miller (1973)
and Matthews (1982), 230-35, do so effectively.
8 Matthews (1982), 233.
9 See Frege (1952), 67.
25
truth. Thus, on Frege's view, Leibniz's Law is preserved; the Morning

Star and the Evening Star do indeed have all the same properties.
Even if the sorts of referents Frege suggested are not suitable, the
idea that when failures of substitutivity occur the referents are distinct
can be found in Aristotelian scholarship as well. Frank Lewis, for example, argues that substitutivity fails in sentences of the sort described
just because there is not an identity between, for example, Socrates and
the generous (the generous one); the generous one, claims Lewis, is a
different entity from Socrates. Specifically, when the generous one is said
by Aristotle to be accidentally but not essentially the same as Socrates,
what this means, according to Lewis, is that the generous one is an
accidental compound of which Socrates is a constituent.10 As Lewis says,
''When we do ontology, we can set Socrates down as one, single object,
and discover later that there remains a second object to be counted,
namely, Socrates + pale."11
Although Matthews is more cautious than Lewis and says that Socrates and Socrates seated "are not two people, nor, indeed, two of
anything else,"12 pointing out that for Aristotle there is not even a
univocal sense of the verb 'to be' in which they can both be said to be,
in the end he too argues along similar lines, saying:
For [Aristotle], after all, 'Coriscus' and 'the masked man' are not really coreferential expressions at all. The one picks out a kooky object [that is, an
accidental unity] that perishes when Coriscus takes off his mask; the other
doesn't. To be sure, the masked man is accidentally the same as Coriscus. But
accidental sameness is not identity and accidental sameness does not guarantee
that every attribute of Coriscus is an attribute of the masked man.13
That Aristotle continues, as Lewis admits he does continue, to consider essential and accidental sameness varieties of numerical sameness
would seem to be at least puzzling on the view Lewis presents.14 Moreover, the ontology Lewis and Matthews attribute to Aristotle may itself
seem surprising. Despite Aristotle's talk of compounds as coming into
being and passing away when attributes are gained and lost (Physics I
7 190a20), the usual explanation of Socrates' becoming pale is not that
some numerically distinct entity comes into being or ceases to exist but
that an enduring entity, Socrates, has undergone a qualitative change.
However, what needs to be considered is whether on such an account
failures of substitutivity can be explained.
10 Lewis (1982), 2, 4-5, 7, 24. Accidental compound theory is developed further in
Lewis (1991), 85-140.
11 Lewis (1982), 22.
12 Matthews (1982), 226.
13 Matthews (1982), 227-28.
14 Lewis (1982), 17. Matthews seems to acknowledge that Aristotle's claim that accidental
sameness is a variety of numerical sameness is problematic. See Matthews (1982), 239.
26
I have described a view that explains away referential opacity by the

claim that, where substitutivity fails, the referents of the terms in question are distinct entities, at least one of which is an accidental unity.
These accidental unities include both objects that are referred to by
the conjunction of a proper name and some property (for example,
Socrates seated) and objects referred to simply by means of the property
had (such as the seated one), where the latter are to be analyzed in
terms of the former. But there is also an alternative account according
to which Socrates and the seated one (where Socrates is seated) are not
in any way distinct objects; 'the seated one' just refers to Socrates. On
this view, which can be found in Richard Sorabji, Julius Moravcsik, Fred
Miller, Christopher Kirwan, J. L. Ackrill, and others, accidental unities
turn out to be sensible objects to which reference is fixed by means of
a definite description.15 That only certain claims are illuminating or
explanatory in the way that seems to Aristotle proper for science is,
they hold, the reason why so many contexts in Aristotle are opaque.
In the Physics (195a32-35), Aristotle argues that, properly speaking,
it is the sculptor that is the cause of the statue and not Polyclitus (or,
as Aristotle would think worse still, a musician), even if the sculptor is,
in a given case, Polyclitus (who is also, let it be supposed, a musician).
That is to say, it is in virtue of his skills as a sculptor that Polyclitus is
able to sculpt; for sculpting, his musicianship is irrelevant. As Sorabji
explains:
The cause of some statue may be specified by naming the sculptor, Polyclitus.
But it is not because Polyclitus is Polyclitus that he produced a statue, and it
is only indirectly explanatory to refer to him. For it is only so long as we can
assume other things about Polyclitus (e.g. that he is a sculptor) that we get an
explanation of the outcome.16
Likewise, Aristotle says that the doctor might cure himself but that it
is not insofar as he is a patient that he possesses the art of medicine
(192b23-26). That is, the patient, in a given case, may happen to be a
doctor, but having the skills of a doctor is not part of what it is to be
a patient. As a consequence, these attributes are not always found together (192b26-27). (Compare Metaphysics VI 1026b35-1027al2.)
That 'the builder builds' is true in all possible cases of building may
seem enough to justify Aristotle's preference. Whether all referentially
opaque contexts in Aristotle could be similarly accounted for, however,
is surely problematic. Besides, there is also a more serious problem. For
to say only that 'the builder builds' is explanatory and that 'the musician
builds' is not does not settle the status of 'the musician builds'. That is
15 Sorabji (1980), 5; Moravcsik (1975), 633; Miller (1973), 484; Kirwan (1971), 181;
Ackrill (1981), 38-39.
16 Sorabji (1980), 11. See also Kirwan (1971), 181.
27
to say, is 'the musician builds' supposed to be false or merely, from the

perspective of a demonstrative science, imperspicuous? This question
is of crucial importance. For if 'the musician builds' is imperspicuous
but nonetheless true, there is no referential opacity to be explained.
On the other hand, if 'the musician builds' is false, to say that the claim
is not explanatory fails to get to the source of the problem. If 'the
musician builds' is false, why is it so? That is to say, if, as this account
presupposes, the musician is identical with the builder, how could 'the
musician builds' be false?
To this point I have considered two sorts of proposals. According
to one, 'the builder', 'Socrates seated', and other similar expressions
are said to refer to accidental unities, where accidental unities are entities that are numerically distinct from sensible objects, thus explaining
away the opacity. On the alternative view these expressions refer to
sensible objects such as Socrates. But as we have seen, the second proposal, taken by itself, does nothing to remove the problematic character
of failures of substitutivity. But there may be yet another possibility. For
some philosophers have thought that referential opacity in Aristotle could
be explained and explained without the multiplication of entities found
in Lewis's view. According to Russell Dancy, for example, Socrates and
a given just thing are numerically the same yet formally distinct. What it
means to say that Socrates and that just thing are formally distinct, Dancy
tells us, is that what it is to be x is not what it is to be y.17
Dancy's account is suggestive, but one may still wonder why, exactly,
entities that are formally distinct should be such that substitutivity fails.
Irwin suggests a solution, namely, that Aristotle can claim that things
are identical but differ in being if he means "not that the being of this
sculptor is different from the being of this baker, but that the being of
sculptor differs from the being of baker."18 That is to say, on Irwin's
view, this baker and this sculptor do not differ in properties where
Callias, say, does both sculpting and baking (the being or essence of
both is the same, namely, what it is to be a human being), and so as
regards reference to Callias there is no referential opacity to be explained. Nevertheless on Irwin's interpretation Aristotle's remarks do
seem at least misleading. The reason is that, whereas the sculptor and
the baker, being in fact this sculptor and this baker, should be essentially
the same, this is just what Aristotle appears to deny, while, on the other
hand, the essence of sculptor and the essence of baker are, like sculpting
and baking, too obviously different to make Aristotle's assertion that
they are different necessary.
We have now seen several proposals for understanding referentially
opaque contexts in Aristotle. In the course of a discussion of agency
17 Dancy (1975), 367.
18 Irwin (1986), 73.
28
and patiency in Physics III, Aristotle himself seems to offer an explanation. Here he says:
Nor is it necessary that the teacher should learn, even if to act [to teach] and
to be acted on [to learn] are one and the same, provided they are not the same
in respect of the account which states their essence (as raiment and dress), but
are the same in the sense in which the road from Thebes to Athens and the
road from Athens to Thebes are the same. . . . For it is not things which are in
any way the same that have all their attributes the same, but only those to be
which is the same. (202bl0-16; cf. 202al8-20)
However, as has already been said with regard to Dancy's presentation
of it, the problem with Aristotle's explanation of referential opacity,
namely, that only to things the same in definition do all the same properties belong, is that it seems to be nothing more than a restatement
of the difficulty; that is, there are cases in which a is apparently identical
with b where substitutivity fails. In what way is the road from Athens
to Thebes the same as the road from Thebes to Athens? In the next
section I propose an answer to this question.
Ill
We have examined several proposals for understanding referentially
opaque contexts in Aristotle, including Lewis's and Matthew's suggestion that in such cases the referents are in fact not the same. In a passage
of the sort noticed by Lewis and Matthews, Aristotle says:
For to be a man is not the same as to be unmusical. One part survives, the
other does not: what is not an opposite survives (for the man survives), but notmusical or unmusical does not survive, nor does the compound of the two,
namely the unmusical man. (Physics I 7 190al7-21; cf. Prior Analytics I 47b2934, and On Generation and Corruption I 319b25-31)
Aristotle seems to want to say that the unmusical human being's ceasing
to exist is something distinct from a human being's ceasing to be unmusical; the unmusical (that is, this case of it) does not survive and the
unmusical human being does not survive. On the other hand, that the
accidental is "obviously akin to not-being" (Metaphysics VI 2 1026b21)
and "practically a mere name" (1026bl3-14) is said to be shown by
the fact that, although unmusical Coriscus ceases to exist, there is no
process by which he does so (1026b22-24). Thus what seems to be
needed is a view that is intermediate between saying, on the one hand,
that accidental unities are entities numerically distinct from the sensible
objects that have the accidents in question and saying, on the other
hand, that they just are those sensible objects, inappropriately described. When Coriscus comes to be musical, Aristotle seems to think
there comes to exist not just an instance of the quality musicality but
29
musical Coriscus, which is something objectlike. Similarly, what ceases

to exist as a result of the changing characteristics of Coriscus is not just
the instance of the quality unmusicality; something objectlike, we are
told, also ceases.
The identification of a subsidiary object does, of course, occur elsewhere in Aristotle, for example, in his account of geometry, and what
I want to argue is that what Aristotle says about geometrical entities is
useful here. Geometry, Aristotle says at Physics io,4a9-ll, "investigates
natural lines but not qua natural [lines]" (cf. 1077bl7-34). The concern
of the geometer, in other words, is not so-called mathematical entities
but rather, for example, doors and tables. Nevertheless the geometer
considers these things only with regard to their boundaries their shape
is relevant while their color, their condition, their use, and so on are
ignored. We might say that not everything true of a certain wooden
structure qua door is true of it qua rectangle; as Jonathan Lear says in
his discussion of mathematics in Aristotle, 'qua' is a "filter" for the
elimination of irrelevant properties. 19
To be sure, Lear's metaphor is not entirely apt for all referentially
opaque contexts in Aristotle since in some, as will become apparent,
properties are gained as well as lost. What is to be emphasized, however,
is Aristotle's insistence that the geometer investigates natural lines even though a rectangle would cease to exist should a door become
warped, the world does not contain both doors and rectangles. Indeed
any temptation to multiply entities should be diminished by the realization that 'qua </>' is detachable; that water qua water is not transparent
{De Anima II 7 418b7) 20 is equivalent to 'water is not transparent qua
water' and to 'it is not qua water that water is transparent'.21
To generalize then from mathematical contexts to passages of the
sort that interest Lewis and Matthews, it seems that Aristotle can deny
that there is anything numerically distinct from Socrates and Callias and
so on to which 'the musician' refers, while still referring to both Socrates
and the musician. Just as a geometer refers to rectangles as well as
(although not qua geometer) to tables, so, for example, a psychologist
researching the nature of musical talent can refer to musicians as such.
Moreover if our language had no expressions such as 'rectangle' and
'musician', such reference could be accomplished using 'qua', as in 'this
door-qua-rectangular object' or 'this human being-qua-musical one'
19 Lear (1982), 168. For 'qua' as a creator of opacity, see Wiggins (1967), 23-4. Compare
Aquinas, who argues in Summa Theologica III Q. 16, Art. 10 that 'Christ as man is a
creature' [Christus secondum quod homo, est creatura] is true even though 'Christ is a
creature' is false. I want to thank my colleague Sandra Edwards for directing me to
several medieval discussions.
20 I will cite De Anima by its more commonly used Latin title rather than the English
On the Soul, its title in the Barnes edition of Aristotle's works.
21 I owe this point to Richard Grandy.
30
(where the hyphens indicate that now 'qua' is part of a referring expression), and because 'qua' when used referentially is a filter of properties,
the result again is generality. That is, just as it is true that rectangles
have four right angles regardless of the color and age of the tables, in
other words, that for purposes of geometry two tables are, assuming
their size to be irrelevant or the same, indistinguishable, so it is as well
in other cases. Since 'Socrates-qua-musician' abstracts from all the properties of Socrates except musicality and 'Callias-qua-musician' does the
same, then even though they are, in virtue of being numerically the
same as two different sensible objects, two specimens, Socrates-quamusician and Callias-qua-musician are qualitatively indistinguishable.
In short, on the account which I am proposing, Socrates and
Socrates-qua-builder are numerically the same. They are not, however,
identical. Because not all the properties of a are true of a-qua-0, in a
move which is of crucial importance for his theory of substance, there
is a distinction to be made between numerical sameness and identity.
Postponing until later chapters all questions regarding the status of
kinds and the naturalness of any given kind, I want to call x-qua-0 a
specimen of the kind 0, and what I am suggesting is that what Aristotle
calls "accidental unities" are in fact specimens of kinds. That is to say,
what I take Aristotle to hold is that Callias is numerically the same as
the musician (and also musical Callias). Nevertheless Callias is not identical with either of these, and it is just because 'the musician' and 'musical Callias' refer not to Callias but in both cases to a specimen of the
kind musician, where a specimen of a kind has only the properties essential to members of that kind, that failures of substitutivity occur.
I have contended that 'the builder', 'Socrates seated', and other such
expressions for accidental unities are not, for Aristotle, definite descriptions that fix the reference to someone (namely, Socrates) who is
short, snub-nosed, a philosopher, and also a builder and seated. On the
other hand, I have also claimed that to deny that 'the seated one' refers
to Socrates does not have to mean that the seated one is, as Lewis
thinks, an entity numerically distinct from Socrates, even though Socrates and the seated one (where Socrates is seated) differ in properties
and so are nonidentical. To be sure, the seated one (that is to say,
Socrates-qua-one-seated) is identical with what it is to be one seated,
and this is what provides the generality required for explanation. But
from the fact that Socrates-qua-one-seated is identical with what it is
to be one seated, it of course cannot be concluded that Socrates and
what it is to be one seated are so. Likewise it does not follow that the
definition of the kind, even though it is a kind to which the sensible
object can be said to belong, is necessarily the definition of the sensible
object. If Socrates, for example, belongs to the kind (it is to be remembered that I am not committed to supposing that it is a natural
31
kind) seated thing, then Socrates-qua-one-seated is a specimen of the

kind seated thing, but the definition of the kind seated thing, though the
definition of the specimen of that kind, is not the definition of Socrates.
To return then to Aristotle's examples in Physics III, if the road from
Athens to Thebes is a specimen of the kind road uphill and the road
from Thebes to Athens a specimen of the kind road downhill, then the
road from Athens to Thebes is numerically the same as but not identical
with the road from Thebes to Athens. It is for the same reason true
that the teacher may teach someone who learns, but even if the activities
of teaching and learning are the same, a specimen of the (event) kind
teaching is not a specimen of the (event) kind learning.22
Admittedly, I have not given a defense on philosophical grounds for
the distinction I think Aristotle makes between sensible objects and
specimens of kinds; though I have hinted that it will be at the core of
Aristotle's account of substance, a metaphysics that confounds counting
is, to say the least, problematic. In fact the task of defense is, in large
part, outside the scope of this project. But in Chapter VI I will show
an interesting use that a contemporary philosopher has found for a
distinction between sameness and identity, a strategy I hope is sufficient
to establish that the distinction may be of philosophical as well as, if I
am right, historical interest. Moreover, as it happens, the same example
will prove suggestive of grounds for maintaining, as Aristotle certainly
would, that some kinds are ontologically more fundamental than others
and also for a defense of what I will take to be his claim that specimens
of natural kinds are ontologically more fundamental than the sensible
objects with which they are numerically the same. Now, however, I want
to argue that sameness, causal, and knowledge contexts, all of which
are opaque, can also be successfully analyzed in terms of reference to
specimens of kinds.
IV
In Physics III, as we have seen, Aristotle says that it is not things that
are in any way the same that have all their attributes the same, but only
those to be which is the same (202bl4-16). But that this is so will, of
course, affect substitutivity in statements to the effect that essential
sameness or accidental ("in some way") sameness itself obtains. Thus
if, as I have claimed, descriptions that occur in opaque contexts are
names for specimens of kinds, that is to say, for entities that are numerically the same as but not identical with sensible objects, and if the
properties of specimens of kinds are limited to those contained in or
entailed by the definition of the kind, it is easy to see why, for claims
22 Charles (1984), 10-15, 29, claims that for Aristotle processes having different definitions are nonidentical members of an equivalence class.
32
of essential sameness, sameness in definition is required to preserve

truth. In 'Socrates is essentially the same as a specimen of the kind
human being' and (tautologously) 'the builder is essentially the same as
a specimen of the kind builder9, reference to specimens prevents substitution for either term flanking 4is essentially the same as', unless what
is substituted is the same in definition as that which it replaces (for
example, 'the kind rational animaV for 'the kind human being9).
In the case of accidental sameness, the analysis is more complicated
but similar. As we would expect, that (1) 'Socrates is accidentally the
same as a specimen of the kind human being9 is false. That (2) 'Socrates
is accidentally the same as a specimen of the kind builder9 is true, and
likewise that (3) 'Socrates is accidentally the same as a specimen of the
kind philosopher9 is true. However, just as a contemporary philosopher
might assert that John may in fact believe both that Cicero denounced
Catiline and that Tully did but that his having one belief cannot be
inferred from his having the other, so because of the restrictions on
substitution which reference to specimens introduces, (3) is unobtainable by inference from the fact that Socrates is both a builder and a
philosopher and (2).
Of course Aristotle sometimes presents the relation of accidental
sameness as holding not between, say, Socrates and a specimen of the
kind builder, but between two specimens of such kinds, for example, the
housebuilder and the fluteplayer (Physics II 5 197al4-15). Since I have
said that specimens of kinds have only the properties given in the definition of the kind, such examples of accidental sameness may seem to
undermine my view that accidental unities just are specimens of kinds.
However, I believe that it does not do so. Rather, it is sufficient to
dissolve the problem to say that in Aristotle there are really two kinds
of accidental sameness, one a relation between a sensible object and a
specimen of a kind and another a relation between a specimen of one
kind and a specimen of another kind, where both specimens are accidentally the same as the same sensible object. Though, as we would
expect, he analyzes it in terms of compounding, Lewis has suggested
just such a distinction between accidental sameness and what he calls
accidental sameness*, respectively.23
Whether, as Lewis thinks, accidental sameness* is a relation between
two entities both of which are compounds or, as I think, between entities
both of which are specimens of kinds, that Aristotle holds not just that
Coriscus is accidentally the same as the musician but that the housebuilder is accidentally the same* as the musician is clear. As he says,
Things are said to be (1) in an accidental sense, (2) by their own nature. (1) In
an accidental sense, e.g., we say the just is musical, and the man is musical and
23 Lewis (1982), 9.
33
the musical is a man, just as we say the musical builds, because the builder
happens to be musical or the musical happens to be a builder; for here 'one
thing is another' means 'one is an accident of another*. So in the cases we have
mentioned; for when we say the man is musical and the musical is a man, or
the white is musical or the musical is white, the last two mean that both attributes
are accidents of the same thing

1017b27-28; emphasis added)
(Metaphysics V 7 1017a7-16; cf. 1015bl6-27,
In this section I have analyzed the relations of essential sameness,

accidental sameness, and accidental sameness*. However, it is also the
case that Aristotle's remarks about causation can be understood in terms
of these relations. The builder builds' is true and self-explanatory because specimens of the kind builder are essentially the causes of building.
'Socrates builds' is imperspicuous (because Socrates is not essentially
the cause of building) but true (because Socrates is accidentally the
cause of building, that is to say, he is accidentally the same as that which
is essentially the cause of building, namely, the builder). The fluteplayer
builds' is likewise imperspicuous (because the fluteplayer is not essentially the cause of building) but true (because the fluteplayer is accidentally the same* as that which is essentially the cause of building in
virtue of the fluteplayer's being accidentally the same as Socrates, where
Socrates is accidentally the same as the builder).
Finally my account of accidental sameness provides an explanation
of Aristotle's treatment of knowledge claims in Sophistical Refutations.
There, in the Masker Paradox, Aristotle says that we know Coriscus but
not the one approaching, even though the one approaching is Coriscus.
Aristotle's explanation is brief and cryptic: "For only to things that are
indistinguishable and one in substance [that is to say, one in essence]
does it seem that all the same attributes belong" (Sophistical Refutations
i7ga37-39). The similarity between Aristotle's solution to the Masker
Paradox and his remark in Physics HI 3 that ". . . it is not things which
are in any way the same that have all their attributes the same, but only
those to be which is the same" (202bl4-16) is plain. Moreover, as in
the Physics, Aristotle's rule in Sophistical Refutations 24 raises a number
of questions. What is the substance of the one approaching, if not
Coriscus? Which expressions can be substituted for one another in contexts involving knowledge and which cannot? And, even more important, why should substitutivity be limited if the terms substituted refer
to the same entity?
In fact, of course, it is hard to know why the paradox is, for Aristotle,
a paradox. If, as Aristotle usually thinks, all that can be known of individuals is their essence as given in the definition and the definition
of Coriscus is what it is to be a human being, such a definition, while
enough to distinguish him from some things that might be approaching
(for example, a dog), is not enough to distinguish him from other ap-
34
proaching human beings.24 But this problem does not affect the general
structure of the solution I propose. For even if contrary to Aristotle's
usual view, there were some way to know Coriscus as Coriscus and even
if the kind were not approaching thing but something so narrow that the
only approaching thing which could be a member of that kind would
be Coriscus (perhaps something like the kind thing now within sight and
approaching the viewer from the only direction from which he can now be
approached),25 still, as long as being a member of this kind is not Coriscus's essence, to know a specimen of this kind is not to know that
Coriscus is the same as this specimen.
In short, despite the peculiarity of the paradox, how my account of
accidental unities would deal with knowledge contexts is clear. Coriscus
is a specimen of the kind approaching thing, and we can know what it
is to be a thing approaching. But since the definition of what it is to
be approaching is not the definition of Coriscus, to know Coriscus is
not to know that it is he who is approaching.
I have argued that certain failures of substitutivity in Aristotle can be

explained on the supposition that many opaque contexts involve predicating something of specimens of kinds rather than of the sensible
objects with which they are accidentally the same, and I have proposed
to use instances of 'x-qua-</>' to refer to specimens of kinds, which specimens would have only the properties essential to the kind. That Lewis
believes that musical Coriscus, the musical one, and so on are accidental
compounds, entities constructed from individuals and their accidents
and therefore not identical with the individuals in question, has already
been said. In the Masker Paradox and likewise in Aristotle's accounts
of change, persistence, and causality, there is no need, Lewis argues,
to suppose that reference is made not to Coriscus but, in the manner
of Frege, to the sense of the name Coriscus; rather where there is
24 Aristotle's claim that knowledge is only of the universal (or, as I will argue, of what
is equivalent to it) is discussed in Chapters III and IV.
25 As I have not committed myself to the claim that all kinds are natural kinds, any group
of things collected by virtue of a common property will be a kind. However I do not
seriously intend to include, as here, kinds which can have only one member. That, as
I have said, is made necessary in the case of the Masker Paradox not because of
anything having to do with failures of substitutivity but only because, contrary to his
usual view, Aristotle here considers the possibility of knowing Coriscus in a way that,
once he is in sight, would make a difficulty of our not knowing that it is he who is
approaching. It should be noted also that, as I interpret it, there is no equivocation
between "knowledge by acquaintance" and "knowledge that" in either the Paradox
or its solution. The knowledge involved is always knowledge of an entity, either Coriscus or a specimen of some kind, and thus is knowledge by acquaintance.
35
accidental sameness substitutivity fails just because identity does not

obtain.26 But since Lewis also considers - and rejects - the proposal
that, by isolating the intensional component of the sentence, 'qua' (as
in 'the doctor qua doctor heals') might provide yet another alternative,
I want to make it clear that there is a difference between what I have
advocated and the approach Lewis criticizes. Specifically, what Lewis
concludes is that, since 'qua' cannot be eliminated without the occurrence of intensionality in the proposed paraphrase, there is no advantage in its introduction.27 Nevertheless my account is unlike the case
Lewis has in mind. For I have made use of 'qua' not in an attempt to
reconcile referential opacity with the ontology of common sense but to
refer to a specimen of a kind, an entity which is not identical to the
sensible object with which it is nonetheless numerically the same.
To summarize the concerns of this chapter, it may be said that when
faced with referential opacity there seem to be roughly four options:
(1) to deny that some things true of a subject really are properties, (2)
to introduce new nonidentical referents for the expressions in question
while not distinguishing numerical sameness from identity, (3) to revise
one's understanding of identity, and (4) to postulate varieties of numerical sameness other than identity. The first alternative might look
promising for predicates such as '. . . is a substance', but it seems implausible for, say, '. . . builds'. The second alternative has been illustrated by Frege and, in the case of the interpretation of Aristotle, by
Lewis and Matthews.
Ways of accomplishing the third option are varied and problematic;
Dancy's suggestion, mentioned earlier in this chapter, that things numerically the same might be formally distinct, may be an example.28 It
is likely that another is offered by Irwin when he says that in Metaphysics
VII 6 Aristotle intends to assert the identity of the individual and the
particular form, an identity eased by including in Socrates' essence or
form properties of his character and personality which distinguish him
from Callias.29 Here it seems clear that Irwin wants to distinguish such
properties from others - that Socrates is tenth in line for the theater
or that he has a bit of lint on his sleeve, for example. But if this is so,
that is to say, if Irwin would agree that not every trivial property of
Socrates is a property of Socrates' essence or soul, then the identity
Irwin asserts to obtain between Socrates and Socrates' soul is not identity, as usually understood.
26 See Lewis (1991), 131-40, 199-221.
27 Lewis (1991), 208; for the entire discussion, see 199-210.
28 Dancy (1975), 367, says: "So Aristotle would not say, simply, that in some sentences
the substitutivity of identity fails because those sentences are referentially opaque: he
would have things to say about identity."
29 Irwin (1988), 255. See also 252.
36
Thus it would seem that Irwin too attributes to Aristotle a nonstandard concept of identity;30 Sandra Peterson may do so as well. What
Peterson argues is that Aristotle's handling of opaque contexts is close
to that of Carnap, in whose system a distinction is made between identity
of extension and identity of intension.31 For Carnap substitution in
modal contexts requires sameness in intension, that is, sameness in all
state descriptions, while substitution in propositional attitude contexts
requires both intensional equivalence and intensional isomorphism. 32
Aristotle does not make the latter distinction. Nonetheless he does claim
that essential sameness (that is, the sort of numerical sameness in which
there is sameness in definition) is required for substitutivity to hold in
modal and propositional attitude (knowledge) contexts, and that, Peterson believes, is enough to justify the comparison.
However, if my interpretation is correct, the views of Aristotle and
Carnap are not especially close. In the first place, while Carnap's individual concept Socrates - the referent of modal contexts in which
there is seeming reference to Socrates - includes that which distinguishes Socrates from Callias in all state descriptions, on Aristotle's
account, as we have seen, Socrates-qua-human being and Callias-quahuman being turn out to be qualitatively indistinguishable. Moreover,
where Carnap would consider 'the fluteplayer' and 'the housebuilder'
(in a given state description) as extensionally equivalent definite descriptions, Aristotle would, I have argued, take them to refer to specimens of different kinds. Finally even though whatever is true of Socrates
is true of that which is numerically the same as a specimen of the kind
human being or a specimen of the kind builder, to refer to Socrates-quahuman being or to the builder is not to refer to Socrates but to a
specimen of a given kind, where, as I have said, what is characteristic
of specimens is that they (unlike the sensible objects with which they
are numerically the same) have only the defining characteristics of the
kind. But if this is so, then the numerical sameness that holds between
Socrates and Socrates-qua-</> is not identity, and Aristotle's distinction
30 Hartman (1977) may be yet another example. He says: "As his use of qua suggests,
Aristotle distinguishes entities according to their descriptions; thus an entity under a
certain description is not fully identical to what would normally be considered the
same entity under a different description. In effect, Aristotle makes mutual logical
dependency a requirement for true identity . . . " (73). To distinguish between true
or full identity, on the one hand, and accidental identity (still said not to involve
numerical distinctness) (75), on the other, would seem to revise our understanding
of identity.
31 Peterson (1969), 145-50. Hussey (1983), 69-70, argues that Aristotle's treatment of
referential opacity anticipates parts of Frege's, but Hussey's account is still like Peterson's. Hussey says: "Aristotle's point is that . . . 'Leibniz's Law' need not be true
unless there is sameness 'in being' or 'in definition', on top of ordinary identity" (69).
32 See Carnap (1947), 46-59, 9-11, 100.
37
between essential sameness and accidental sameness is not a distinction

between types of identity.
I have just described three ways of understanding referentially
opaque contexts. The fourth option, namely, that there are varieties of
numerical sameness which are distinct from identity is of course the
one that I have argued best describes Aristotle's position. In particular,
what I have claimed is that both substances and accidental unities should
be understood to be specimens of kinds, where specimens of kinds are
numerically the same as but not identical with sensible objects and thus
that, when substitutivity fails, it does so because what has been referred
to is not a sensible object but a specimen of a kind. Where the specimen
is a specimen of a kind whose definition is also the definition of the
sensible object with which it is numerically the same, the relation between the sensible object and the specimen is that of essential sameness;
in the case of a specimen of some other kind, there is accidental sameness. However, not even essential sameness is identity, and, as will become apparent, for both essential and accidental sameness failures of
substitutivity can occur. Of course, since specimens of kinds have only
the properties given in the definition of the kind, and since, in the case
of accidental sameness, that definition is not the definition of the sensible object with which the specimen in question is numerically the same,
it is not surprising that for accidental sameness there are far more
contexts in which substitutivity fails.
VI
At the beginning of this chapter, I said that Aristotle's grasp of identity
had been brought into question and understandably so, for, as we have
seen, there are many contexts in Aristotle which are referentially
opaque. In the course of the chapter, however, I have proposed to
explain a number of these by claiming that many opaque contexts involve predicating something of specimens of kinds rather than of the
sensible objects with which they are accidentally the same. Of course
insofar as it is numerical sameness without identity that explains failures
of substitutivity in Aristotle, no reason has been provided to agree that
Aristotle's grasp of the indiscernibility of identicals weakens. On the
other hand, if numerical sameness, not identity as determined by Leibniz's Law, has become the more common use of 'is' in Aristotle, there
would seem to be truth to the idea that Aristotle loses interest in identity.
Yet how much truth there is in the claim that identity loses importance for Aristotle depends on whether my belief that essential sameness
too is numerical sameness without identity is correct. When Aristotle
says in the Physics that only to things to be which is the same do all the
38
same properties belong (202bl5-16), his primary concern seems to be

to explain away certain cases of referential opacity involving accidental
sameness. Whether there are other cases of referential opacity - that is,
whether his conditions are sufficient for substitutivity or only necessary
- is a question that does not get asked. Yet it may be that it is answered
nonetheless in Metaphysics VII 6. Here Aristotle argues:
But in the case of so-called self-subsistent things, is a thing necessarily the same
as its essence? E.g. if there are some substances which have no other substances
nor entities prior to them - substances such as some assert the Ideas to be? If
the essence of good is to be different from the Idea of good, and the essence
of animal from the Idea of animal, and the essence of being from the Idea of
being, there will, firstly, be other substances and entities and Ideas besides those
which are asserted, and, secondly, these others will be prior substances if the
essence is substance. And if the posterior substances are severed from one
another, there will be no knowledge of the ones and the others will have no
being. (By 'severed' I mean, if the Idea of good has not the essence of good,
and the latter has not the property of being good.) For there is knowledge of
each thing only when we know its essence. (1031a28-b7)
Toward the end of the chapter he restates the point:
The absurdity of the separation would appear also if one were to assign a name
to each of the essences; for there would be another essence besides the original
one, e.g. to the essence of horse there will belong a second essence. (1031b28-30)
If the thing and its essence are not the same, then its essence must itself
be an individual and thus has itself to have an essence, which, because
it is not the same as the essence of the thing, is itself another individual,
and so on, resulting in regress.
Read as an objection to Plato, this argument seems very strange;
Aristotle has given us no reason to conclude that a Form and its essence
are not identical. But Aristotle uses his argument to lead us to a different
conclusion. The argument, he tells us, does not apply to Platonic Forms
alone; each primary and self-subsistent thing is one and the same as its
essence (1032a4-6; cf. 1031bl2-14). As he puts it, "Yet why should
not some things be their essences from the start, since [or if] essence
is substance?" (1031b31-32). To put the point another way, if the distinctness of a thing and its essence can for some cases be denied, why
introduce Platonic Forms at all? On the other hand, because for Aristotle as for Plato things with accidents and in flux cannot be known,
apart from Platonic Forms, what sorts of things could Aristotle be thinking of? What sorts of things, in other words, can Aristotle hold to be
the same as their essences?
It might seem that the answer is to be found at the end of Metaphysics
VII 6 itself; Aristotle, after all, says: "Now the sophistical objections to
this position, and the question whether Socrates and to be Socrates are
39
the same thing, are obviously answered in the same way . . ." (1032a68). Yet if this is his answer to the question, it is an answer which proves
elusive. For that the question is to be answered in the same way does
not, unfortunately, tell us what the answer is; some interpreters have
thought that Aristotle means us to conclude that Socrates and to be
Socrates are the same,33 and others that they are not.34 But since Socrates has many accidental properties, if (as I take VII 6 to assert) he is
the same as his essence, then in this case, which is certainly a case of
essential sameness, there will nevertheless be a failure of substitutivity.
A defense of the claim that, though not identical, Socrates and his
essence are the same - and with it an interpretation of substance in
Metaphysics VII-VIII - is the subject of the next chapter.
33 See, for example, Hartman (1977), 57-87, esp. 63; Sellars (1967), 115; Rorty (1973),
402; Irwin (1988), 218.
34 See Furth (1985), 114.
Ill
A Theory of Substance
Aristotle's account of substance involves yet another case - indeed the

most central case - of his use of numerical sameness without identity,
and in this chapter I offer an interpretation of Aristotle's views about
substance which depends on that distinction. The task of interpreting
VII-VIII has, of course, been undertaken many times, and yet nothing
approaching a consensus has been reached. My strategy is to argue that
Aristotelian substances are specimens of natural kinds, where such specimens are numerically the same as but not identical with sensible objects.
I maintain that, if a distinction between numerical sameness and identity
is posited, Aristotle's view is consistent, his claim about the separation
of substance is intelligible, and his requirement that substances have
ontological and epistemological priority is satisfied. This chapter begins
with a brief discussion of the Categories and proceeds to consideration
of how Aristotle's position in that work is affected by the demand in
the Metaphysics for the epistemic priority of substances; separation and
ontological priority will be considered in later chapters.
I
In the Categories, an early work, Aristotle makes a distinction between
what is present in a subject, what is said of a subject, what is both, and
what is neither (la20-lb6). Although this classification is unclear with
respect to whether the properties present in a subject are or are not
particulars, that dispute need not be entered into here.1 For to give an
1 For the claim that the properties present in a subject are particularized, see Anscombe
(1961), 8; Ackrill (1963), 74-5; and Heinaman (1981a). For alternatives, see Owen
(1965) and Frede (1987a), 58. I will assume that properties in the Categories are not
particularized or that, if they are, Aristotle has abandoned this view by the time of the
Metaphysics. However, should this assumption be false, my views about substances could
be reconstructed by saying that although properties are particulars (and hence collectible into kinds), since the existence of properties depends on the existence of substances, properties are dependent particulars while substances are specimens of independent natural kinds.
40
THEORY OF SUBSTANCE
41
account of substance it is the "what is neither" which is of importance.

That is to say, for Aristotle that of which things are said or in which
they are present is a subject, for example, an individual human being
or horse, and such things, we are told, are substances, "most strictly,
primarily, and most of all" (2all-12).
It seems then that subjects can be identified by what we call count
nouns (although not all count nouns would refer to such subjects), and
they are such that the grounds for counting or individuating them make
no reference to things other than themselves. For example, if one tile
were set alongside another of exactly the same color, despite the sameness of color, there are still two things, namely, two tiles; likewise Socrates and Callias standing arm in arm are still two human beings.
Thus sortal terms carry with them criteria for individuation - to know
what it is to be a human being, for example, is to know how much of
a clearly viewed and uncrowded perceptual field counts as one human
being2 - and they make it possible for Aristotle to distinguish between
subjects, on the one hand, and what is present in them (and, in the case
of differentiae, said of them), on the other. Yet as the preceding example
illustrates, it is also true that individuation requires classification, and
therefore the question of the relation of such particulars to one another
cannot be ignored. Of Callias and Socrates, Aristotle says in the Metaphysics that they are the same not in number but in kind (1034a7-8).
But that this is so leads only to further questions, namely, what is a
kind? And what is it for two individuals to be of the same kind?
In the case of Socrates and Callias, what we want to say, of course,
is that both are members of the same biological species. But not all
objects are living things. Some are artefacts, and others are chunks of
stuff or what Aristotle would regard as combinations of substances and
their properties, such as philosophers (human beings who engage in a
certain activity) and even white things (white rabbits, white tables . . .).
Aristotle of course gives primacy to living things, and some reasons for
his preference will be given in a later chapter. Here, however, I want
to focus on what all sortals have in common, namely, that similar particulars are gathered together. That is, while admitting that for Aristotle
only some kinds will be natural (with the result that their members are
neither said of nor present in a subject), I want in this section to restrict
attention to the features of kinds simply in virtue of their being kinds.
Like a class, a species - and, more generally, a kind - would seem
to have members and to be some sort of collection of those members,
2 There are some count nouns that do not carry internal criteria of individuation
'patch', for example. But insofar as patches can be counted, it is just because they are
physically separated, as for example, two patches of white paint on an otherwise blue
surface. That is to say, for patches to be distinguishable, their shapes have to be made
definite by what surrounds them.
42
for which reason it might be suggested that kinds just are classes or
sets. But even if this were so, not all classes would be kinds. For a class
is commonly defined as any collection of objects, 3 while at the very least,
not any arbitrary collection could be a kind; the members of a kind
need be alike with respect to certain nontrivial properties. Some classes,
of course, do meet this condition. All members of the class of white
things are white; all members of the class of human beings are rational.
Yet I would argue that this fact disguises a fundamental difference. For
whereas the identity of any class or set is solely a function of its membership, the identity of a kind, by contrast, would not seem to depend
upon its extension; 4 that for Aristotle kinds could have had other members seems beyond doubt. But if this is true, then even if, for example,
the class of human beings includes all past, present, and future human
beings, it would not be the same as the kind human being. More generally,
for any kind, it seems that there could have been other members of
that same kind.5
I have said that if kinds are not arbitrary collections there must be
some grounds that determine which objects are members of a given
kind. In fact what would seem to determine membership in a kind is
the presence in any object of certain properties; Socrates, for example,
is a member of the kind human being in virtue of his being rational or
two-footed. But - and this is crucial - even though both the kind and
the properties that determine membership in it are counted by Aristotle
as said of a subject, a kind is not identical with the properties essential
for membership in that kind. Laboring without the advantage of an
indefinite article (which is lacking in Greek), in the Categories, Aristotle
explains how kinds are both like and unlike properties. He says:
Every substance seems to signify a certain 'this' [tode ti\. As regards the primary
substances, it is indisputably true that each of them signifies a certain 'this'; for
the thing revealed is individual and numerically one. But as regards the secondary substances, though it appears from the form of the name - when one
speaks of man or animal - that a secondary substance likewise signifies a certain
'this', this is not really true; rather, it signifies a certain qualification \poion ti\
- for the subject is not, as the primary substance is, one, but man and animal
are said of many things. However, it does not signify simply a certain qualification [poion ti], as white does. White signifies nothing but a qualification [poiori],
whereas the species and the genus mark off the qualification of substance they signify substance of a certain qualification (3b 10-21).
3 See Massey (1970), 355. See also Mates (1965), 29.
4 To be sure Quine (1970), 8, thinks otherwise. He says: "There is no call to reckon
kinds as intensional. Kinds can be seen as sets, determined by their members. It is just
that [because not all classes or sets are such that their members share certain properties]
not all sets are kinds."
5 See Wolterstorff (1970), 258.
THEORY OF SUBSTANCE
43
When Aristotle says that man or animal is poion ti but not simply poion
ti (or not simply poion, according to some manuscripts) (3b 18), the
grounds for the distinction are as follows: since its name is applicable
to many things, a species is not a particular, yet because of the relation
between species and their members, species are nevertheless not properties. Even if it is too much to claim that any ontology involving objects
and properties must be committed to the existence also of collections,
the reasonableness of Aristotle's distinguishing kinds from essential
properties and including both in his ontology cannot be denied.
To this point I have said that, since like a class a kind can have
members, a kind would seem to be some sort of collection. Moreover
as it is a collection whose membership is determined by the presence
of certain properties, there could have been other members of the same
kind.6 Finally, because a kind is a collection, it is easy to see that most
properties of its members will not be characteristic of the kind. Of
course, given that Aristotle distinguishes between properties and kinds,
it will follow also that, while for Aristotle there are many human beings,
each with the property rationality, there would be, on his view, no
property man.1 Rather, as we have seen, 'man' ('human being') is the
name for a kind,8 and a kind is not a predicable universal but a collection.
II
We have examined the distinction between kinds and properties which
underlies the ontology of the Categories. Within this ontology, as we
have seen, Aristotle identifies that which is neither said of nor present
in anything else as primary substance. That is to say, primary substances,
6 Compare Mondadori (1978), 37, who accommodates the shifting membership of natural
kinds into possible world semantics as follows: "the view according to which natural
kind terms (rigidly) designate the corresponding kind (or species, or substance) can be
represented in modal semantics by assigning as extension to a given natural kind term
/ the corresponding kind (or species, or substance), and as intension a constant function
taking each possible world into a kind (a species, a substance). The latter, in turn, can
be represented by means of a function taking each possible world into a set of objects
(the set of tigers, or lemons, or samples of gold, as the case may be)." As a result of
this two-step definition, Mondadori can say both that natural kind terms are rigid
designators and that the class of tigers in one possible world is different from that in
another (see 38).
7 To be sure, Aristotle in De Interpretation 7 offers "man" as his example of a universal
(that which is of such a nature as to be predicated of many subjects), but the distinction
between properties and kinds is not made (or needed) in that context.
8 Pellegrin has marshaled evidence to show that in the biological works and elsewhere,
genos and eidos can function at very different levels of generality. If this variability
affects Aristotle's metaphysical views, it seems to me to complicate but not undermine
the account of substance which I want to give. That is to say, I will argue that substances
are specimens of natural kinds; if natural kinds can for Aristotle equally well be set at
various levels of generality, specimens of those kinds will vary in level accordingly. See
Pellegrin (1986), esp. ch. 2.
44
in this scheme, are the ultimate subjects. In the Metaphysics, however,

the ultimate subjecthood of the individual human being or the individual
horse is threatened by matter. Matter (concerning which more will be
said in Chapter V) is introduced in the Physics and is opposed to form.
The problem for the Metaphysics is that if the matter of the horse underlies the horse in something like the way in which the horse underlies
its attributes, it would seem that, unless substance is matter, being a
subject is no longer the criterion for being a substance. Yet even though
Aristotle may find it necessary to refine his understanding of being a
subject, that that which underlies a thing primarily is thought to be in
the truest sense its substance is reiterated in VII 3 (1029al). On the
other hand, it seems that matter cannot be substance because substances, as we are told (1029a27-28), have both separability (chdriston)
and thisness (tode ti).9
Thus the need to accommodate the distinction between matter and
form is one difference between the Categories and the Metaphysics. It is
not the only difference, however; the role of substance in Aristotle's
theory has expanded as well. In the Categories substance was said to
stand in a relation of priority to the other categories in such a way that
they are said to depend on it (and not conversely), while in the Metaphysics the priority is explicitly said to be not only ontological but
epis temological.
In fact views of the relation between the Categories and the Metaphysics
have varied widely, ranging from Montgomery Furth's claim that the
Categories is a model of Aristotle's metaphysical theory using "a cutdown and simplified conceptual apparatus"10 to Daniel Graham's claim
that the discontinuities are so great as to comprise two incommensurable
systems.11 Between these extremes, Michael Loux has argued that even
in the Categories Aristotle's essentialism threatens the asymmetry between individuals (thought to be primary substances in virtue of being
unanalyzable subjects) and their species, thereby encouraging a different
account of substance in the Metaphysics,12 while Lewis offers what he calls
a "reconciliationist" view.13 Though nothing in the Metaphysics exactly
fits the subject criterion of the Categories,14 in the Metaphysics Aristotle's
plan, claims Lewis, is "to knit together the earlier criteria on substance
. . . with the new ontology of form and matter"15 by "modifying the old
ingredients of his metaphysical thinking from the Categories and moulding
them to fit the requirements of their new metaphysical context."16
9 In the Complete Works of Aristotle (1984) Barnes translates 'tode ti* as "individuality."
Smith (1921) argues that to be tode ti is to be both singular and possessed of a universal
nature, a view that would also suit my interpretation.
10 Furth (1988), 4.
11 Graham (1987), esp. 84-118.
12 Loux (1991), 48.
13 Lewis (1991), 267 and 269, n. 12.
14 Lewis (1991), 274.
15 Lewis (1991), 267.
16 Lewis (1991), 269.
THEORY OF SUBSTANCE
45
Of course one's view of the relation between the Categories and the
Metaphysics depends in large part on what one takes Aristotle's theory
of substance in the Metaphysics to be. What I myself want to claim is
that, apart from the addition of the matter of which objects are composed, there is a certain continuity between Aristotle's ontology in the
Categories and the Metaphysics. However, since in the Metaphysics 'sub-
stance' ipusia) is an honorific title to be awarded to whatever is first in

knowledge, definition, and time (1028a32-33), even that claim, supposing that it is correct, does not answer the question "what is to count
as substance?" Nevertheless to discover that on a given interpretation
Aristotle's ontology in the Metaphysics provides a candidate that meets
the criteria for substance would be some evidence in favor of attributing
that ontology to Aristotle; to discover, moreover, that the same element
in that ontology disarmed in a plausible way the charge of inconsistency
would provide further evidence in favor of that ontology and the candidacy of that element. My strategy in this chapter will be to begin by
offering an account of substance in the Metaphysics. In Section V I will
then suggest that even in the Categories what Aristotle says of substance
is more subtle and less at odds with the Metaphysics than might at first
be supposed.
In Metaphysics VII 3 Aristotle offers a list of candidates for substance;
in this chapter the essence, the universal, the genus, and the substratum
are said to be the substance of each thing. But substratum can in turn
be understood as matter, form, or the composite. Having dismissed
matter (at least as matter is understood in VII 317) as neither tode ti nor
choriston and the composite as posterior and obvious, of this trio there
remains the form or essence, for these are identified with one another
in VII 7 (1032bl-2).18
The claims of the universal, including the genus, will be addressed
in the course of the present discussion, and more will be said about
both the composite and matter in later chapters. But as for form, it
seems to me that in VII-VIII there are several senses of the term
(including its synonyms such as 'formula' and 'shape'). In VII 4 only
the form of the genus is said to have an essence (1030all-12); in VII
8 Socrates and Callias are said to be the same in form (1034a8). These
I, along with some (but not all) other interpreters, take to be examples
of form as species. However by 'form' Aristotle more often means not
the species or, more broadly, the kind, but rather, as in VII 12 where
the last differentia is said to be the form (1038a26), the properties that
enable something to be a member of a given kind. The being glued of
17 How matter in VII 3 is to be understood has generated controversy that need not be
considered here. For two recent discussions see Loux (1991), 54-71, and Lewis (1991),
282-99.
18 This claim too is controversial, but see Irwin (1988), 239-40.
46
the book or the blending of honey water (1042bl6-18), the principles

that make the elements a and b a syllable and fire and earth into flesh
(1041bl2-19) - in short the arrangement of the matter - each of these
I take to be an example of form as the differentia (1042b32), that is,
of form as a thing's essential properties.
It seems reasonable then to think that 'form' sometimes means the
kind, that is, a collection of similar particulars which could have had
other members, and sometimes its defining properties. Yet that the
existence of collections depends on the existence of objects that are
members of the collection is evident, 19 while properties are said to be
themselves not separable entities (1028a23-25) but to exist pros hen
(1030a35-bl). Thus because substances are supposed to be first in
knowledge, definition, and time, and in order to be first in time, to be
tode ti and choriston (that is, to have thisness and separation), and because
substance is said to be form (1033bl7; 1037a29), if Metaphysics VIIVIII is to make sense, it must be the case that 'form' sometimes has
neither of these meanings. Rather, if in VII 6 Socrates is said to be in
some sense the same as his essence (and form is, as I have said, the
essence), the term 'form' must in this case refer in some way to Socrates.20 That is to say, Aristotle's strategy, as I see it, is to argue that
members of kinds can in some way be both first in time and first in
knowledge.
I have proposed that in Metaphysics VII Socrates is said to be the
same as his essence or form, in which case 'form' must sometimes mean
neither the kind nor its defining characteristics but, in some way, a
member of that kind. The problem, however, is that since any member
of a kind, such as Socrates, has characteristics that are unknowable, if
the particular form of Socrates were thought to include the accidental
features that distinguish him from other human beings, Socrates' form
would be no more knowable than Socrates.
This implication has been denied by David Balme and accepted (but
softened) by Irwin, but neither of their proposals seems to me satisfactory. What Balme suggests is that Socrates is a process which, when
considered at a moment of time, has no matter, so that Socrates at tn
(Socrates' form) can be known by grasping a complete description of
19 For the dependence of kinds, consider for example a club: while it need not have any
of its actual members, it could hardly be said to exist if it had no members.
20 Although it is not for the present purpose significant, there is also a fourth sense. In
VII 7 the building art is said to be the form of a house (1032bl3-14), and the form
of health is said to be found in the mind of the doctor (1032b22-23). In these cases
'form' seems to mean something like a set of rules or ideas that govern the production
of something. In saying that there are three or even four senses of 'form', I do not
mean that the term is simply ambiguous. When the term 'lion' is used of individual
lions, the kind lion, and even pictures of lions, it is so used not because of ambiguity
but because of the relations among these things. With 'form' (and 'substance') the
situation is similar.
THEORY OF SUBSTANCE
47
every detail at that moment.21 Irwin, on the other hand, concedes that
particular forms are only indirectly objects of knowledge but argues
that this is acceptable. He says, "Aristotle cannot . . . allow that the
particular is in itself an object of definition and scientific knowledge;
but he can still fairly insist that scientific knowledge and definition apply
to a particular form."22 "On this view," he continues, "particular forms
partly satisfy the requirements for being objects of knowledge. Not
being universals, they are not the primary objects of knowledge; but
they are among the objects that scientific knowledge applies to. We
might say they are what science is about, though not what it is o/]"23
That Balme's proposed ontology is not Aristotle's seems clear. As for
Irwin's proposal, the difficulty is that it leaves Aristotle with the dilemma
he identifies in Metaphysics III 6: universals are knowable but are not
substances while particulars are not knowable (1003a5-17). To put the
point another way, if particulars are primary ontologically and universals
are most fundamental epistemologically, there is nothing that meets the
requirements for substance.
I have argued that if Socrates' form includes accidental features such
as Socrates' snub-nosedness, Socrates' form will not be knowable. On
the other hand, if Socrates' form does not include, for example, being
snub-nosed,24 then it is open to Aristotle to distinguish what is true of
Socrates and what is true of Socrates' form in such a way that the latter
is knowable. However, if he is to avoid Platonism, Aristotle must at the
same time deny (and if Socrates and to be Socrates are said to be the
same, he does deny) that Socrates and to be Socrates are numerically
distinct entities. I will now argue that the notion of a specimen of a
kind, introduced in the preceding chapter as a solution to the general
problem of referential opacity, shows how this can be accomplished.
Ill
Earlier I proposed to use the expression 'Socrates-qua-musician' in such
a way that by abstracting from all properties of Socrates except musicality it designates a specimen of the kind musician. Though my use of
4
Socrates-qua-</>' to designate a specimen of a kind is an interpreter's
term of art, it is one clearly derived from Aristotle. Indeed there are
many occurrences of 'qua' in Aristotle; some typical examples are: It is
not qua water or qua air that water or air is transparent (De Anima II
21 See Balme (1984), 5.
22 Irwin (1988), 263.
23 Irwin (1988), 263-64.
24 For a view of particular forms such that Socrates and Callias do not have qualitatively
different forms, see Frede (1987a), 63-71 and (1987b), 78. Sellars (1967), 112, says
that the particular form of this bronze sphere is the bronze sphere itself simply qua
sphere; see also Sellars (1957), 698. Hartman (1977) holds that the particular form
of Socrates is Socrates qua man. See 57-87.
48
7 418b7),25 "what produces the impression is a signet of bronze or gold,

but not qua bronze or gold" (De Anima II 12 424a20-21), "but neither
is there friendship towards a horse or an ox, nor to a slave qua slave"
(Nicomachean Ethics VIII 11 1161b2-3), "but if anyone actually thought
of [the snub not qua snub but] qua hollow he would think of it without
the flesh in which it is embodied" (De Anima III 7 431bl3-15), "firstly
and in the proper sense [there is friendship] of good men qua good"
(Nicomachean Ethics VIII 4 1157a30-31), and, most famously, "being as
[qua] being" (1003a21).
As these examples show, the expression that follows 'qua' can be a
mass term, an adjective, a sortal term, or an abstract noun, and while
in some cases the same term either occurs twice or is replaced by another
term definitionally related to it (e.g., 'snub' and 'separately existing
hollow'), in others this is not so. For example in Physics III 1 Aristotle
says:
It is the fulfilment of what is potential when it is already fulfilled and operates
not as itself but as movable, that is motion. What I mean by 'as' is this: bronze
is potentially a statue. But it is not the fulfilment of bronze as bronze which is
motion. For to be bronze and to be a certain potentiality are not the same.
(201a27-32)
Clearly it is not part of what bronze is that it serve well for statuary
("to be bronze and to be a certain potentiality are not the same"),
though were bronze not suitable for statues, the qua relation could not
hold. Likewise in the case of Socrates, the applicability of 'qua human
being' to some true statements about him reflects the fact that Socrates
is identical with a given human being and not, say, a given elephant.
Of course there are also cases such as "[friends] wish well alike to
each other qua good" (Nicomachean Ethics VIII 3 1156b8-9) in which
what follows 'qua' is adjectival. However in such cases - to return to
my example, let us suppose the adjectival case to be 'qua musical' clearly the point is to consider neither musicality (for here 'qua' would
serve no purpose) nor musical Socrates (for in this case the goal of
knowability would be defeated). Rather the expression 'Socrates qua
musical', by abstracting from all properties of Socrates except one,
leaves us with 'Socrates qua musician', an analysis that, it can be seen,
depends on the distinction between kinds and properties drawn in Section I of this chapter.
In short, Socrates qua musician is Socrates as - and only as - a
member of the kind musician. More generally, any particular as and only
as a member of some kind is a thing having only the properties given
in the definition of that kind, which definition is necessarily true of all
25 Barnes {Complete Works of Aristotle, 1984) translates: "Neither air nor water is transparent because it is air or water."
THEORY OF SUBSTANCE
49
its members. Such a thing - that is, a particular as a member of some

kind - is what I have been calling a specimen of a kind. As for the
nature of specimens, in the preceding chapter two further points were
made. The first, and here I appealed to Lear's metaphor of 'qua' as a
filter, was that whatever is true of 'Socrates-qua-musician' is true of
'Callias-qua musician'; put generally, whatever is true of any specimen
of a given kind is true of every specimen of that same kind. The second
point was that, even though 'Socrates-qua-human being' refers to a
specimen of the kind human being, if identity and numerical sameness
are distinguished, the use of 'qua' need not introduce a multitude of
entities. Of course if Socrates-qua-human being is for Aristotle a substance, then certainly Socrates-qua-human being exists and indeed, so
Aristotle seems to think, exists in a more fundamental way than Socrates.
How Aristotle might defend such a claim is the subject of Chapter VI.
Nevertheless since Socrates and Socrates-qua-human being are not numerically distinct, even in this case, it will be true that there do not
exist two entities, Socrates and Socrates-qua-human being.
To recapitulate, on my account Socrates, Socrates-qua-builder, Socrates-qua-human being, and so on are numerically the same. Nevertheless - and this is the point of the hyphenated expressions - x and xqua-0 are not identical; not all the properties of x are true of x-qua-</>.
That this is so is what makes specimens of kinds different from simply
members of kinds. Socrates is a member of the kind human being (and
the kinds philosopher, builder, and so on); Socrates-qua-human being and
Socrates-qua-philosopher are specimens of those kinds, respectively.
As it happens, there is a distinct expression to refer to Socrates-quahuman being, namely, 'Socrates' soul'. What Aristotle says in Metaphysics
VIII 3 is that Socrates' soul and Callias' soul are distinct but not as
souls distinguishable; soul and to be soul (Aristotle can in this case mean
"identical") are said to be the same (1043b2). As for man, in other
words, a given human being such as Socrates, and to be man, Aristotle
tells us that they are not the same (1043b3), unless Socrates and his
soul are identified. To put the point another way, not everything said
of Socrates is said of Socrates-qua-human being salva veritate.
A similar example is given in the summary at the end of Metaphysics
VII 11:
but in the concrete substance, e.g. a snub nose or Callias, the matter also will
be present. And we have stated that the essence and the individual thing are
in some cases the same; i.e. in the case of primary substances, e.g. curvature
and the essence of curvature, if this is primary. (By a primary substance I mean
one which does not imply the presence of something in something else, i.e. in
a substrate which acts as matter.) But things which are of the nature of matter
or of wholes which include matter, are not the same as their essences, nor are
50
accidental unities like that of Socrates and musical; for these are the same only
by accident. (1037a32-1037b7)
Of course Aristotle does not mean that curvature exists without matter.
The point is rather that, since its matter is not part of what it is, between
curvature, that is to say, the geometrical shape, and its essence, there
will be identity. But like a snub nose, on the other hand, Socrates is a
whole that includes matter. Therefore although Socrates is essentially
the same as what it is to be a human being and accidentally the same
as the musical, in neither case is there identity.
I have defended the thesis that, as a result of the restrictive function
of 'qua', it is possible to refer not only, for example, to Socrates, but
also, by using the expression 'Socrates-qua-human being', to a specimen
of the kind human being. I have argued too that, despite the differences
between any two members of the same kind, since specimens of a kind
include no accidental properties there is no difference at all in the
qualities had by specimens of the same kind. But of course it is just this
indistinguishability that makes specimens of kinds, unlike the sensible
objects with which they are numerically the same, knowable. For at least
in the Metaphysics,26 the goals of science and metaphysics are limited to
the necessary and the universal. The result is that even if for Aristotle
the ideal candidates for knowledge would be the Platonic Forms, understood as eternally existing, necessarily unchanging patterns whose properties are all essential, specimens of kinds can be an acceptable approximation of the Forms without their separation.
I have argued that Socrates-qua-musician is knowable for Aristotle
because there is, in some sense, a definition for what it is to be musical;
of course, to say that x-qua-0 is a knowable object is not necessarily to
say that x-qua-0 is first in knowledge. Yet even though Aristotle would
say that qualities (and hence specimens of kinds derived from qualities)
have an essence and a definition in a secondary sense (VII 4), I have
as yet given no criteria that would establish that musician is a kind derived
from the properties of human beings, rather than conversely. To do so
would make it apparent that the requirement for epistemological priority is not one that is independent; to be not just knowable but first in
knowability will turn out to be, in other words, a characteristic not of
specimens of just any kind. Only specimens of natural kinds can be
specimens that are first in knowledge because it is only they that are
ontologically fundamental.
26 In fact Aristotle did realize that certain features not common to all members of a
species can be genetically transmitted. For a clear discussion of how this fact affects
the notion of form and the limits of knowability in the biological works, see Sharpies
(1985), esp. 120-22.
THEORY OF SUBSTANCE
51
IV
Although Aristotle says often enough that knowledge is belief about
things that are universal and necessary (e.g., Nicomachean Ethics VI 6
1140b31-32), in Section III I argued that for Aristotle substances specimens of natural kinds - are not universals but nevertheless are
knowable just because they, as it were, mimic the universal. However
if this is so, the inconsistency of Metaphysics VII-VIII can be resolved.
That in VII-VIII Aristotle asserts that substance is form, that form is
universal, and that no universal is substance, there is no doubt. 27 Some
philosophers have believed that the inconsistency cannot be overcome,
and others, taking Aristotle's approach to be aporetic, have supposed
that he might not be committed to all three claims. But most have sought
to avoid these conclusions either by reinterpreting one of the three
statements or by declaring one of the three constituent terms to be
ambiguous.
While every permutation has probably been tried, 28 recent solutions
have tended either toward particular forms or toward taking some universals to be substance. My view, that forms are specimens of natural
kinds, has evident affinity to the view that there are particular forms, 29
although also to a view according to which substances are neither particulars nor universals.30 An alternative takes form (substance) to be a
universal, perhaps something like a formal cause, 31 thereby denying
Aristotle's apparent claim in VII 13 that no universal is substance.
What Aristotle says in VII 13 is, in part, as follows:
For it seems impossible that any universal term should be the name of a substance. For primary substance is that kind of substance which is peculiar to an
individual, which does not belong to anything else; but the universal is common,
since that is called universal which naturally belongs to more than one thing.
Of which individual then will this be the substance? Either of all or of none.
But it cannot be the substance of all; and if it is to be the substance of one,
this one will be the others also; for things whose substance is one and whose
essence is one are themselves also one. (1038b815)
He then goes on to claim that substance means that which is not predicable of a subject but that the universal is predicable of some subject
27 For a list of relevant passages, see Lesher (1971), 169. Lesher does not distinguish
between passages that say only that the substance of a thing is form and those that
say that it is substance which is form.

28 For an overview of various interpretations, see Lesher (1971). For discussion of more
recent papers, see Lewis (1985), 65-7.
29 See Sellars (1967) and (1957); Hartman (1977), 57-87; Frede (1987a), 63-71, and
(1987b).
30 See Owens (1963), esp. 347-99, 426-34.
31 See, for example, Driscoll (1981); Lewis (1985) and (1991), esp. Pts. III-IV; Code
(1986); Loux (1979); and Furth (1988), 192-201.
52
always (1038bl5-16), and later concludes, "If, then, we view the matter
from these standpoints, it is plain that no universal attribute is a substance, and this is plain also from the fact that no common predicate
indicates a 'this', but rather a 'such' " (1038b34-1039a2), a conclusion
he repeats in VII 16 at 1040b23-24. 32
It might appear that VII 13 is decisive evidence in favor of a view,
such as mine, which denies that forms are universal. Nevertheless John
Driscoll, Alan Code, and both Loux and Lewis, among others, have
argued that the VII 13 claim is limited to universals that are predicated
of particulars while substances or forms are universals that are predicated of matter,33 in which case substance can be universal after all. As
Driscoll says, "Since the formal cause or essence taken without any
material substrate is definable, it cannot be a particular (cf. Z, 15,
1039b20-22, 1040a5-7), and since it cannot be a particular, it must
be a universal in the broad sense that every nonparticular is a universal."34 Though the arguments of Loux's and Lewis's recent books have
more complexity than I can here do justice to, I will indicate some
points where there is disagreement.
What Loux argues is that in the Categories primary substances are
the familiar concrete particulars that belong to substance-kinds; a certain horse and a human being such as Socrates are his examples. As
Loux says:
The fact that a basic subject falls under its species cannot rest on or be grounded
in some prior instantiation of the said-of or the present-in relation; and, as I
have said, these two relations exhaust the tools Aristotle has at his disposal in
the Categories for ontological analysis or reduction. But, then, the result of the
interaction of [the basic-subject criterion for ousiahood and the doctrine of
essentialism] is that a basic subject's falling under or belonging to its lowestlevel substance-kind is a primitive, bedrock fact about the world, a fact not
susceptible of further analysis. For want of a better name, let us call this result
the Unanalyzability Thesis.35
However, in the Metaphysics an analysis of the unanalyzable is, Loux

thinks, just what occurs. Indeed in the Metaphysics the subjects of the
Categories are found to be analyzable composites of matter and form,
and it is form that in the Metaphysics is primary substance. As Loux says,
"It is, then, because form is the only thing that is both the ousia of
other things and its own ousia that the thesis that form is ousia is the
dominating claim of the middle books." 36 He continues:
32 Loux (1991), 203,findsseven arguments in VII 13, but of these thefirst(1038b9-
15) gets most of his attention. His interpretation of it is discussed later in this chapter.
33 See Driscoll (1981), 150-52; Code (1986), 413; Loux (1979), 23; and Lewis (1985),
66-67. This way of reading VII 13 is derived from suggestions in earlier papers:
Woods (1967), 216, and Albritton (1957), 705.
34 Driscoll (1981), 151.
35 Loux (1991), 35.
36 Loux (1991), 159.
THEORY OF SUBSTANCE
53
Were it not a necessary fact about a thing that what it is to be that thing and
the thing itself are one and the same, it could not play the kind of explanatory
role that the Aristotle of Z and H37 wants to reserve for form. Form is to provide
us with a final answer to the question why things are what they are, and only
if a thing is necessarily the same with what it is to be that thing do we have the
requisite guarantee that there is no explanation in terms of something else for
that thing's being what it is, for its being the kind of thing it is. Only if a thing
is its own essence can it be its own ousia. For a thing of this sort, the question,
Why is it that kind of thing? has a ready answer that precludes the possibility
of any further questions. It is the kind of thing it is because it is one and the
same as being that kind of thing; and it could not be otherwise.38
Loux distinguishes form predication and species predication in such

a way that a substance species is predicated of its instances, this being
the less fundamental sort of predication, while form is predicated of
matter,39 a view that allows him to come to the conclusion about VII
13 already described, namely, that Aristotle's intent in that chapter is
only to exclude universals predicated of particulars from the ranks of
primary substances.40 But this conclusion having been reached, Loux
also offers other arguments, both philosophical and textual, against
particular forms. These include claims that ousiai explanations are inherently general, that the metaphysics of particular forms is "bizarre,"
and that if individual forms come into being and pass away universal
forms will still be needed in which case individual forms do no work in
the theory.41 His textual argument is that there is no passage in VIIVIII that demands a reading of forms as particulars; the statement in
V 18 that Callias is kath* hauto Callias and what it is to be Callias
(1022a26-27) and also the claim in XII 5 that your matter and form
and moving cause are different from mine but the same in universal
formula (1071a27-29) can, he says, refer to the general form.42 As for
Lewis's interpretation, he shares Loux's account of VII 1343 and the
distinction between kinds of predication which makes it possible, arguing for the even stronger conclusion that because in the Metaphysics
kinds have no "independent" place in Aristotle's theory sentences that
apparently assign an individual to a kind can be understood in terms
of the predication of form of matter.44
37 In Aristotle's works books are referred to sometimes, as I have done, by numbers and
sometimes by letters of the Greek alphabet. In the case of the Metaphysics, correlation
is complicated by the fact that Book I is A and Book II a. Thus Z and H refer to
Books VII and VIII, respectively.
38 Loux (1991), 159-60.
39 Loux (1991), 109-46.
40 Loux (1991), 202-10.
41 Loux (1991), 225, 227-28.
42 Loux (1991), 230, 233.
43 Lewis (1991), 309-21.
44 Lewis (1991), 190-91; see also 184, n. 22, where Lewis argues against Frede and
Patzig's analysis of VII 8 1033a2831, which they believe shows that a form can be
metaphysically predicated of the composite.
54
Thus in both Loux and Lewis we find a thoroughgoing defense of

the claim that substance is form and form universal, a defense made
possible by an interpretation of VII 13 according to which forms, being
predicated of matter, are not the target of the chapter. The chief advantage of this view, it seems to me, is that it makes it easy to explain
how forms can be knowable. Yet I have argued that particular forms
that are qualitatively indistinguishable from one another are also knowable, and if this is right, it seems to me that this approach is to be
preferred. For there is a serious difficulty with the interpretation according to which forms are universal, namely, that on such a view that
which expresses a form is, as Loux says, in its "depth logic" adjectival.45
That a form is "how" its matter is46 is something Loux does not take
to be a problem. Recognizing that form is said to be a this (that is to
say, not metaphysically predicated of anything else), Lewis, on the other
hand, proposes that in the Metaphysics what counts as a this (Jtode ti)
changes47 in such a way that some universals are thises, not just derivatively or "by courtesy" (that is, in virtue of their relation to what is
properly called a this), but rather because being a "such" is to be restricted to things that are universal in relation to thises,48 something
that forms, being universal in relation to matter, are not.
Thus Lewis uses his account of predication to explain how form can
be called a this. Yet I do not find his explanation of how form is to be
a this, in other words, a subject of which everything else is predicated
(VII 3 1029a27-30, VIII 1 1042a28-29) convincing, and Lewis himself
admits that it is problematic. He says:
In general, then, form is that in virtue of which the compound substance works
out its characteristic style of life. This influence manifests itself in two distinct
ways. The fact that form is predicated of matter explains how the creature is a
member of its kind in the first place: this gives matter as one kind of subject.
But the fact that a compound substance is a member of a kind is what makes
it a fit subject for accidents. It is thanks to form, then, that the subjects that
properly Have49 accidents can do so. This gives a second class of subjects, namely,
compound material substances.
All of this can make it seem that - without too much exaggeration - almost
anything except form can count as a subject. But it may be possible to manufacture out of these materials a somewhat precarious sense in which form too
is a subject. Form has a central role in placing a thing in its kind, so that the
thing is a suitable subject for accidents. Indirectly, then, and in a suitably extended sense, form too perhaps is a subject. I do not claim tofindthis reasoning
fully convincing. But, outside of the hypothesis of individual form, it is one of
45 Loux (1991), 8.
46 Loux (1991), 8.
47 Lewis (1991), 325.
48 Lewis (1991), 327-29.
49 The capitalized 'Have' for indicating accidental predication follows the terminology
of Paul Grice and Alan Code.
THEORY OF SUBSTANCE
55
the only two ready alternatives I know of for explaining Aristotle's express
assertion that forms are subjects.50
In short it seems to me difficult to see how if form is universal, form
is a this in any way that allows it to be ontologically fundamental. Moreover it may be that there is also another problem. For in Chapter V it
will be noted that some interpreters have argued that matter in Aristotle
is such that its having the form it has is essential to it and even that, as
matter, matter has no properties other than its potential for form.
Although both Loux and Lewis respond by suggesting that Aristotle
may have had more than one way of thinking about matter,51 insofar
as such views of matter have plausibility, Loux's and Lewis's claim that
form is predicated accidentally of matter is in jeopardy.
Having considered Loux's and Lewis's defense of the claim that~form
is universal, let us return to the problem with which this section began,
namely, that of the apparent inconsistency of Metaphysics VII-VIII. If
one reads VII 13 in such a way that forms are universals predicated of
matter, it is clear how inconsistency has been avoided; substance is form,
form is universal, and some universals, namely, those that are predicated
of matter, are substances after all. But if, as I argue, substances are
specimens of natural kinds, the inconsistency is to be resolved rather
by reinterpretation of the claim that form is universal.
Of course, because substance is said to be form, and because there
is, or so I have conceded, a sense of 'form', namely, the differentia, in
which form is a universal, the triad could also be made consistent by
maintaining that substance is formj while form2 is universal. Nevertheless, as I see it, Aristotle intends his argument to be more cohesive than
a solution that appeals to equivocation would produce. For Aristotle's
point in saying that form is universal is that substances must be knowable, and I have argued that, even though Socrates, the sensible object
with which Socrates-qua-human being is numerically the same, is not
knowable, Socrates-qua-human being is knowable. That is to say, on my
proposal, it is, strictly speaking, false that form (in the appropriate sense
of the term) is a universal. Likewise it is false that the form - a given
specimen of a kind - does not come into being and pass away, though,
of course, it is true that the species does not do so. What is true (and
what is intended by the claim that form is universal) is that particular
forms have in virtue of their indistinguishability the epistemic virtues
of universals.
V
I have been arguing that in the Metaphysics Aristotle takes substances
to be specimens of natural kinds that are numerically the same as but
50 Lewis (1991), 303-4.
51 See Loux (1991), 180-83, and Lewis (1991), 250-58.
56
not identical with sensible objects, and I have claimed that specimens
of kinds are, as for Aristotle they need to be, knowable. Indeed it seems
to me an interesting fact that even in the Categories, where there are
no epistemological issues, Aristotle offers as examples of primary substances neither (as we might have expected) 'Bucephalus' and 'Socrates'
nor even 'this horse' and 'this human being' but the very general ho tis
hippos and ho tis anthropos (lb4-5) - an individual horse and an individual
human being.
To be sure, most interpreters of the Categories, including, as we have
just seen, Loux, take Aristotle to intend by primary substances ordinary
sensible objects such as Socrates, objects which in the Metaphysics are
considered to be composites of matter and form and not primary substances. Yet I am far from sure that the Categories is so straightforward.
Of course I would not claim that at the time of the Categories Aristotle
had already worked out the implications of distinguishing a human being
as such from Callias, Socrates, and so on; clearly if he had done so he
would have found it necessary to distinguish the subject of which something is said from the subject of which some things are said and in which
others, namely, accidents, are present. Rather, the point is just that
even in the Categories the differences among individual horses or individual human beings seem not to be of importance.
But if in the Categories the differences between one human being and
another are taken to be less interesting than the similarities and in the
Metaphysics it cannot be otherwise as there substances are supposed to
be knowable, what I want to conclude from this fact is, as I have said,
that Aristotle differs from Plato chiefly in rejecting the numerical distinctness of Forms from sensible objects. Whether for Aristotle a thing
and its essence or form are numerically distinct from one another is,
as I indicated at the end of the last chapter, the question of Metaphysics
VII 6; indeed it is well known that G. E. L. Owen once argued that in
VII 6 Aristotle rejects what is usually called the Nonidentity Assumption
in cases of strong (essential) predication, thereby providing an answer
to the Third Man Argument.
In fact there does seem to be a connection between the Third Man
Argument (or more generally, the Theory of Forms) and VII 6; specifically, if 'separation' means, as I argued in Chapter I, numerical
distinctness, a denial of numerical distinctness would just be the rejection of one of its premisses. According to Owen, however, the rejection
of the Nonidentity Assumption leads Aristotle to "embarrassment":
[for Aristotle] if we take any primary subject of discourse and say just what it
is, we must be producing a statement of identity, an equation which defines the
subject. And this in turn helps to persuade him that the primary subjects of
THEORY OF SUBSTANCE
57
discourse cannot be individuals such as Socrates, who cannot be defined, but

species such as man.52
M. J. Woods reached a similar conclusion:
When we use a proper name like 'Socrates' to pick out an individual man, what
we pick out is always the form; though we pick it out as it occurs in a particular
piece of matter. The essence of Socrates is simply the form man, an essence
which he shares with Callias.53
Again, Woods says: "the essence of Callias will be the essence of the
species man, which, by the doctrine of Z, will be identical with the species
itself."54 But if I am right, the result of rejecting the premiss that produces separated Forms (and that conjoined with other assumptions results in a regress of them) need not be as Owen and Woods suppose.
For I have argued that what Aristotle wants to say of Socrates is that
he is a member of the species human being and that he is numerically
the same as but not identical with a particular form that is a specimen
of the kind human being.
To say that Aristotle had a certain understanding of Plato's Theory

and its problems is not, of course, to say that Plato held the view Aristotle
attributes to him, and, as the interpretation of Plato's Theory of Forms
is as difficult and as much disputed as anything in Aristotle, it is certainly
outside the scope of this work to investigate the issue. Nevertheless,
here as in Chapter I where the question of separation already arose, I
will say again that I do not find it intrinsically plausible to suppose that
Aristotle misunderstood Plato's Theory, or that, in a matter so central
to his own concerns, he intentionally caricatured it. Therefore that some
mainstream interpretations of Plato are not only in accord with Aristotle's criticisms of the Forms, discussed in Chapter I, but can be shown
to be such that my claim about the relationship between Platonic Forms
and Aristotle's substances are sustained would be important.
VI
Although Tlatonism' in philosophy and mathematics has come to stand
for realism about universals - for any theory, that is, which postulates
the existence of universals that are extramental and such that they exist
even if uninstantiated - the view that Forms are universals is problematic
and has, furthermore, been held in several quite different versions.
According to one account, Plato began by conceiving the Forms as
Socratic universals, came to think of them as separate paradigms, and
52 Owen (1966), 136-37.
54 Woods (1975), 168.
53 Woods (1975), 177.
58
never unraveled the confusion.55 More recently, the opposite sequence

has seemed more promising. According to this view, Plato began by
thinking of Forms as particulars, that is, as exemplars, paradigms, or
perfect instances of universals, and also, however incompatibly, as universals. In the Third Man Argument he uncovers his error and thereafter abandons the notion that Forms are paradigms and along with it
the Theory of Recollection. As Colin Strang says, "In the interval between the second act and the third, which begins with the Parmenides,
the paradigmatic eidos and its brother, recollection, have been unmasked
as imposters and quietly buried."56
If Strang and others are right, in the late dialogues Forms are
straightforwardly universals, universals that are (perhaps) still separate
in the sense of being such that they exist even if uninstantiated, and
they are known through dialectic.57 But that Plato's Theory of Forms
is a theory of universals at any point has proved more difficult to reconcile with the texts than Strang thinks; it requires that the Timaeus be
a middle dialogue as well as the explaining away of certain undeniably
late references, such as Statesman 286a, to Forms as paradigms.58 Nevertheless Patterson has offered an interesting solution to this difficulty.
According to Patterson's view, Plato at all periods intended the Theory
of Forms to be a theory of abstract separate essences, and his describing
sensible objects as "images" should be understood as a metaphor for
emphasizing how unlike their models sensible objects are.59
Despite Patterson's efforts, however, the view that Plato's Theory
just is a realist theory of universals remains, I think, problematic. In
the first place, Plato's own terminology vacillates between the abstract
noun - Largeness - and a construction that consists of the article plus
adjective or common noun - The Large Itself. At Phaedo 65d4-13, for
example, Plato uses both the adjectival expressions Just, Beautiful, and
Good and the abstract nouns Largeness, Health, and Strength for referring to Forms. Second, if Forms are straightforwardly universals, the
Third Man Argument should not even be interesting. Assuming that
Plato was not himself very confused, what reason could there be for
his having Parmenides suppose that Shuttlehood is a shuttle or that
Justice is just, and why should he have Socrates agree to it? Or is one
to suppose that, while Plato himself was not confused, Aristotle was?
But if that were the case, why did Plato not do more to answer the
Third Man Argument?
Terminology and the Third Man Argument then are two reasons for
doubting that the Forms are universals. A third reason is that such a
55 See Ross (1951), 35-36.
56 Strang (1963), 162.
57 See Strang (1963), 162, and Owen (1953).
58 See Cherniss (1957), 248-49, and Rees (1963), esp. 170-73.
59.Patterson (1985), esp. 25-62, 145, 159-64.
THEORY OF SUBSTANCE
59
view would require explaining away Plato's exploitation of the language

of patterns and images; the descriptions in the Phaedo of phenomena
as falling short, for example, are hardly suggestive of universals. Similarly, the lover of sights and sounds in Republic V seems to be someone
who does not simply mistake one thing for another but who makes
mistakes about ontological status. Indeed the metaphor of ascent is
emphasized not only in the Republic's Divided Line and Cave, but also
in the Symposium, Moreover, in a passage whose echo in the Statesman
has already been mentioned, the Phaedrus at 250a-d says that some
Forms such as Beauty have clear visual images, while others, such as
Justice, do not, even though they too have likenesses. Then too, however
it is dated, there remains the Timaeus, where the demiurge looks to the
Forms as patterns for the created world.
In short the metaphor of patterns and copies is pervasive, and it is
difficult to suppose, as the view that Forms are universals requires, that
such passages could be part of an analogy for understanding the relation
between universals and particulars. In his account of the defectiveness
of phenomena, what Patterson says is that the point of the image analogy
is that an image of F is not an F at all;60 a reflection of a shuttle is not
a shuttle. This seems right, and for Plato it means that a phenomenal
shuttle is not really a shuttle. Nevertheless there is a difficulty with
Patterson's understanding of the image analogy. On his interpretation,
the Form for shuttles, which in terms of the analogy should be that
which really is a shuttle, turns out to be the essential nature of a shuttle
and not a shuttle at all. At the same time ordinary shuttles, which by
the logic of the metaphor should be relational entities, on Patterson's
interpretation are not so. That is to say, that element in the metaphor
which is an independent particular is, as Patterson understands the
application of the metaphor to Plato's theory, universal,61 and that element in the metaphor which is a relational entity is, in its application
to Plato's theory, a particular.
In light of the problems with understanding Forms as universals, it
is not surprising that many interpreters have understood Plato's theory
not to be a theory of universals at all. R. E. Allen, for example, denies
that Forms are what he calls commutative universals on the grounds
that universals "clearly cannot be identified with standards and paradigms; for the latter are things characterized, not characters." 62 Likewise, P. T. Geach suggests that it is The Lion, The Bed, and The Master,
not Lionhood, Bedhood, and Mastery, which are Platonic Forms,63 and
60 Patterson (1985), 33.
61 Patterson (1985) declines to identify what he calls Plato's essential natures with universals, but his point seems to be that Forms are not just any separated universals
but only those determined by the Form of the Good. See 134-35, 155.
62 Allen (1960), 157.
63 Geach (1956), 75.
60
R. S. Bluck understands a Form to be not a determinate universal but

an instance of a determinate universal.64
If Forms are understood as paradigm cases, that is to say, as perfect
examples of the properties or sorts in question, it is easy to see how
the Third Man Argument might arise.65 On the other hand, in the case
of many Forms, it is clearly impossible for them to have the relevant
properties of phenomena. For example, The Large Itself cannot be of
any size for, like all Forms, it is not in space. Nor can The Shuttle Itself
be of any given shape or size or color, and even to say that it falls within
a range of sizes and shapes is problematic. Yet if The Shuttle Itself lacks
these characteristics, how can it be an instrument for weaving?
I have sketched the problems that arise on the one hand in taking
Forms to be universals and then, on the other, in supposing that they
are particulars. To deny that The Shuttle is a shuttle may seem a return
to the view that Forms are universals after all, but it is possible to make
a distinction that avoids this conclusion. For a universal is such as to
be predicated; that is, for a universal to exist is for it to be possible for
something to be F, even if in fact the universal exists uninstantiated.
What can be said of Forms, however, is that, though the Form for Fness
is not an F (Plato's reply to the Third Man Argument), it is nevertheless
itself the preeminent F and not predicable of other Fs, but rather something like a type.66 That is to say, if Forms are types, they do indeed
straddle the distinction between universal and particular, and it is not
perplexity or even imprecision but rather ontology that leads Plato to
use '/ness' and 'The F' indiscriminately to designate a Form.67
Were this primarily a study of Plato's Theory of Forms, much more
would need to be said by way of clarification and evaluation of the
proposal that Forms are not universals but patterns that do not have
but are that for which they are the Forms.68 Here, however, I will restrict
myself to three points. The first is to note the similarity of what I have
said to Harold Cherniss's view that the Form for x is "what is x". As
64 Bluck (1957), 121.
65 Geach (1956), 74, and Allen (1960), 151, avoid it by means of the Wittgensteinian
claim that the standard yard is not a yard long since it is that against which things a
yard long are to be measured; Allen then argues that phenomena are resemblances of
Forms. See 155.
66 For types, see Peirce (1933), 423, and Wollheim (1968), 64-68.
67 For the charge of perplexity, see Vlastos (1954). I disagree too with Penner (1987),
who argues that Plato wants to oppose reductively identifying Forms (for example,
what Beauty is) with sensible objects (things that are beautiful), but that otherwise he
has no precise account of the sort of abstract entities he wants Forms to be. See 60,
72-73, 89, 94-95.
68 Of course, Plato's most central examples of Forms are not The Bed and The Shuttle
- plausible candidates for types - but The Beautiful and The Equal. Nevertheless
when Geach (1956), 76, suggests that The Equal is a pair of equal things, presumably
the point of 'things' is to abstract from the sorts of things equally long sticks, equally
heavy stones - which can be equal, thereby minimizing the difference.
THEORY OF SUBSTANCE
61
he says: "of any character or property, x, that a particular has, the reality
is ho estin x, which it could not be if it were had by anything and which
therefore must be independent or 'separate' from all manifestations of
itself as a property."69 The second is that, once Forms are so described,
what is striking is the degree to which, but for Plato's inclusion of Forms
for properties, Plato's Forms so understood would indeed be similar to
Aristotle's substances, except as regards separation. Finally, the third is
to suggest that it is at least arguable that Aristotle is right about the
consequences of separation, that is to say, that Cherniss's assessment
notwithstanding,70 Aristotle's criticisms are very much on target after all.
Of course to say that Plato's Theory can, not implausibly, be taken
as straddling the distinction between particular and universal is not to
say that this is how Aristotle took it, anymore than to say that this is
how Aristotle understood the Theory shows Plato's intent. Nevertheless
that such an interpretation of Plato is not implausible and that it is
remarkably compatible with Aristotle's allegations, discussed in Chapter
I, remains to the point. Indeed in Metaphysics VII 14 there is what I
would claim is an acute analysis (at the level of the genus) of the problem
with types, if these are understood as real and yet, as Plato supposes,
numerically distinct from their tokens:
For if the Forms exist and animal is present in man and horse, it is either one
and the same in number, or different. . . . If there is a man-in-himself who is a
'this' and exists apart, the parts of which he consists, e.g. animal and two-footed,
must indicate a 'this' and be things existing apart and substances; therefore
animal too must be of this sort. Now if animal, which is in the horse and in
man, is one and the same, as you are one and the same with yourself, how will
the one in things that exist apart be one, and how will this animal escape being
divided even from itself? Further, if it is to share in two-footed and manyfooted, an impossible conclusion follows; for contrary attributes will belong at
the same time to it although it is one and a this. (1039a26-b4)
In short, as I understand Aristotle, the crux of his proposal, worked
out in his own theory of substances as a response to the Theory of
Forms, is just this: that the straddling of universal and particular which
is characteristic of Plato's Forms, and indeed which is necessary if what
is ontologically fundamental is also to be knowable, can be made unproblematic only if Plato's separation of Forms from sensible objects
is rejected. However, if Platonic separation is rejected, so Aristotle
argues, forms, though particular in virtue of their numerical sameness
with different sensible objects, have on account of their indistinguishability the knowability characteristic of universals.
In this chapter I have argued for understanding Aristotle's substances
as specimens of natural kinds and have claimed that, so understood,
69 Cherniss (1957), 261. See also 260.
70 See Cherniss (1944), esp. 211.
62
Aristotle can be seen as responding to the Theory of Forms by denying

the numerical distinctness of forms from sensible objects. However,
since substances are also not identical with sensible objects, it remains
to be seen whether this solution is tenable and whether it will in fact
solve the problems Aristotle identifies as consequent upon separation.
For example, even if it is the case that Aristotle's specimens (having a
sort of generality in virtue of being indistinguishable within a kind) are
knowable objects, there remain questions about whether they can be
known by us, a desideratum without which Aristotle could not count
his theory a significant advance. In the next chapter I will first explore
the difficulties of describing the relation between Platonic Forms and
sensible objects in such a way that from our experience of the latter a
plausible account can be given of how knowledge can be had of the
Forms. Then I will argue that Aristotle's explanation of how we come
to have knowledge is addressed to Plato's problems and matched to his
own account of substances as specimens of natural kinds. The ontological aspects of Aristotle's view will be examined in the final chapters.
IV
Substance and Aristotle's Epistemology
On the interpretation of Aristotle's account of substance I have just

offered, substances are specimens of natural kinds that are numerically
the same as sensible objects without being identical with them. Further,
in virtue of their indistinguishability from one another (within the same
kind), specimens of kinds have the knowability characteristic of the
Forms without the separation; indeed it would not mischaracterize Aristotle's theory to say that it treats each specimen of the kind human
being as if it were The Human Being - each specimen of a kind, in other
words, as if it were the Platonic Form.
But even if specimens of natural kinds meet the requirement set forth
in Metaphysics VII 1 that substances must be first in knowledge and
definition, mere knowability on the part of substances does nothing to
establish that they can be known by us. After all, one of the most serious
difficulties with the Theory of Forms is that even though Forms are said
to be eternal, unchanging, and the objects of definition, knowledge of
them remains problematic. I contend that Aristotle believes his rejection
of the separation of Forms can address this issue also - that is to say,
he believes that his view enables him to explain how it is that there is
knowledge. I begin with Plato's Theory.
I
Besides the metaphysical problems concerning the nature of Plato's
Forms (see Chapters I and III), there are, as Plato himself recognizes,
epistemological problems as well. Given that our experience now is not
of Forms, how are Forms to be known? This problem is, of course,
addressed by Plato in the Theory of Recollection.1
1 In the Phaedo (73a-76e), but not the Meno, Plato allows for remembering by unlikeness,
the transition to knowledge seems not to occur in stages, and recollection begins from
sense experience. I will assume that Plato intends these to be additions to the Theory,
but for a more critical discussion, see Gulley (1954), 19799. Although Forms do not
explicitly appear in the Meno, the problems discussed there are obviously relevant to
the Theory and certainly Aristotle took them to be so.
63
64
In the Meno Plato is faced with the paradox of discovery; how, the
dramatic character Socrates asks, paraphrasing a complaint raised by
Meno, can one try to discover either what he knows or what he does
not know? One would not seek what he knows, for since he knows there
is no need of inquiry, nor what he does not know, for in that case he
does not even know what he is to look for (80d). Plato's answer is to
argue that we seek what we in some sense know. Plato illustrates his
proposal, the theory of recollection, by eliciting from an uneducated
slave knowledge of geometry. Challenged to find a way to double the
area of a square, the slave twice proposes procedures that give the wrong
result, is led by Socrates to see that these are wrong, and then, having
conceded that he does not know, is able to follow Socrates' explanation,
worded as an interrogation, of how it is to be done. This knowledge,
the knowledge the slave comes to have, is said to result not from teaching
but from questioning (85d).
Plato's claim that Socrates does not teach the slave can look tendentious, but perhaps it is not so, given that it is intended to cover a priori
and not empirical knowledge; if Socrates supplies the solution to the
problem, he nevertheless expects the slave to see why what he proposes
concerning squares is true.2 To be sure, the example also suggests that
the recovery of knowledge may require a guide, and even at that what
is recollected may at first have a "dreamlike" quality (85c). Nevertheless,
Socrates says, "if the same questions are put to him on many occasions
and in different ways, you can see that in the end he will have a knowledge on the subject as accurate as anybody's" (85c-d) - he will be able
to produce the proof himself and to see why it is to be done in this
way. His true opinions, Socrates tells us, will then have been turned
into knowledge (86a). In fact it seems that not only will he have knowledge but he will then know that he knows.
With the doctrine of recollection Plato has produced a solution to
the paradox of discovery - he has claimed that we in fact know what
we do not now seem to know - but it is a solution beset by a variety of
difficulties. On any theory of knowledge there is need to explain acquisition and verification, yet that the objects of knowledge are for Plato
not now readily available for inspection causes special problems. In the
Meno it is said that the soul, since it is immortal and has seen all things
both here and in the other world, has learned everything that is (81c);
in the Phaedrus Socrates speculates that our souls were then like charioteers and their horses that rushed after the gods through the heavens
and saw the Forms along the way (246a-248b). But it is interesting that
in the Phaedrus Plato himself worries that we might have been unable
to attend properly to our experiences (248b), a state of affairs that
2 See Moravcsik (1971), esp. 65, and Vlastos (1965).
SUBSTANCE AND ARISTOTLE'S EPISTEMOLOGY
65
would doom at the outset the prospects for successful recollection.

Retention, which is mentioned in the Theaetetus, presents related difficulties; if our minds are like soft or crowded wax (194e-195a), what
we experienced may not wholly govern what remains.3
Whether we did acquire knowledge and whether we did retain what
we learned are serious questions whose answers cannot be ascertained
without criteria for verification. How do we know that what we recollect
- that which purports to be knowledge - is after all knowledge and not
error? Perhaps in the Phaedo (72e-76e) Plato hoped that it was sufficient
to say that knowledge is like remembering; to maintain that what we
have within us is a trace of previous experience can, after all, be seen
as a causal - that is to say, an externalist - justification of the belief.
Or perhaps the fact of remembering is being supposed to be supplemented by some recognition that provides assurance that one is remembering correctly. White, for example, says of Plato's view: "sometimes one simply has the feeling (if 'feeling' is the appropriate word)
that one is remembering, which one trusts as a warrant for saying that
one has correctly recollected what has come to consciousness accompanied by that feeling."4
Yet that what is accompanied by the feeling can nevertheless be wrong
is a fact too obvious for Plato to have overlooked, and in the degrading
of images impressed upon soft wax Plato himself has given a reason
why one should not trust such a feeling. Norman Gulley too believes
that recollection is meant to guarantee what is recollected, but he is
vaguer about just what provides the guarantee. He says:
The fact that this process is described as a process of recollection presumably
means that it is anamnesis which provides the recognition that 'this assumes that*
or 'this follows from that'. It must also guarantee the finality of the results of
analysis by affording the recognition that they are thus final in their correspondence with reality: this completes the conversion from opinion to knowledge.5
"Once the final truth is recognized," Gulley adds, "it cannot 'run away'.
We know that we have found the truth."6
Problems about the acquisition of knowledge and its verification
arise, as I have said, on any account of knowledge; they are hardly
unique to Plato. But the point is that because for Plato the objects that
we are said to know are not now experienced and because, indeed, we
for the most part are not even aware of having knowledge of them, the
3 Guthrie (1978), 113, claims that in the Theaetetus Plato cannot be thinking of the Theory
of Recollection because at 197e he says that the aviary is empty at birth. I take this to
be a feature of the story which is without philosophical significance for determining
Plato's overall concerns.
5 Gulley (1954), 195.
4 White (1976), 51-52.
6 Gulley (1954), 195.
66
means of verification for what is recollected are diminished. Moreover,

as we have just seen, what is crucial and yet uniquely difficult for Plato's
Theory is how retrieval of what we learned of Forms "in a previous
life" can be made to happen. In the Theaetetus at 197b-d, Socrates
proposes that one can have knowledge in the sense of possession (what
are sometimes now called unactivated beliefs) without having access to
or control of that knowledge (activated beliefs).7 But if that knowledge
is to be of the Forms, then, just as in the earlier dialogues, it is necessary
to explain how the first sort of knowledge can become the second.
To be sure, if there are enough activated beliefs on the part of either
the inquirer or a teacher to serve as a starting point, inquiry seems
possible; given enough pieces of knowledge, dialectic might produce
more. On the other hand, where prior control over all relevant knowledge is lacking, inquiry - that is to say, investigation of a predetermined
topic - seems impossible. Even though one might come to have some
piece of knowledge, there is no procedure for coming up with a given
piece. Philosophical activity at least at the outset will be like reaching
blindly into a birdcage.
Though I have argued that Plato's views about the Forms undermine
inquiry, one might take issue with this assessment. After all, even if one
has no occurrent knowledge of a given Form, still there are present
experiences of phenomena, and surely these (as is emphasized in the
Republic and the Symposium as well as in the account of recollection in
the Phaedo) may serve as cues. On the other hand, phenomenal images
are transitory, and the nature of space is such that the representations
of the Form which can exist in that medium bring no assurance of
knowledge. Certainly one and the same Form can be variously mirrored.
For all these reasons, it is little wonder that phenomena are by different
observers judged differently - that is to say, are judged to be images
of different Forms - and little wonder if, beginning with such faulty
images as phenomena, Forms are imperfectly recollected. In the Phaedo
even as Plato says that recollection begins from sense experience, he
says also that the senses are hindrances and distractions to the soul in
its search for reality (65a-67b; cf. 81b-d, 82d-83b).
How we are to have knowledge of Forms is, of course, a problem
Plato addresses again and again. Whether the part of the dialectical
process known in the later dialogues as "collection" begins from Forms
or sensibles is controversial,8 yet only those interpreters who believe
that Forms in these dialogues are no longer separate can suppose that
collection and division have replaced recollection.9 But sometimes Pla7 See, for example, Goldman (1986), 202.
8 Gulley (1954) seems to think that it begins with sensibles. See 201. Guthrie (1975),
428-29, believes that it is Forms which are collected.
9 See Gulley (1954), 209-10, for a refutation of the replacement view. See also Rees
(1963), 173-74.
67
to's attempts at a solution are more wide-ranging. In the Cratylus Plato

seems to consider whether ideal names might not be more satisfactory
than phenomena as cues. In Statesman, he suggests that Forms without
visible images might be known by analogy; weaving might provide perceptual cues for understanding statesmanship, for example. Yet since
ordinary language is not perfect, the route of the Cratylus seems quite
as inaccessible as its goal, and that the problems with perceptual cues
would only be compounded by their use in analogy is evident.
The obstacles to knowledge posed by the Theory of Forms would be
irrelevant to my topic unless Aristotle too identified them. But of this
there is no doubt. We have already come upon his claim that craftsmen
do not look to the Forms - as well as his biting remark that if the science
is innate, it is "wonderful" that we are unaware of our possession of
it, after which he adds: "Again, how is one to know what all things are
made of, and how is this to be made evident?" (992b34-993a3). Likewise
at Prior Analytics 67a22-26 Aristotle explicitly criticizes the Theory of
Recollection when he says: "For it never happens that a man has foreknowledge of the particular, but in the process of induction he receives
a knowledge of the particulars, as though by an act of recognition. For
we know some things directly; e.g. that the angles are equal to two right
angles, if we see that the figure is a triangle. Similarly in all other cases"
(cf. Posterior Analytics 71al7-b8).
Thus from passages such as these, it becomes apparent that Aristotle
does see a problem with retrieval. Indeed as D. W. Hamlyn says of the
passage in the Prior Analytics:
Seeing a triangle is ipso facto knowing that it is afigureof a certain general kind.
If we put this into relationship with the exposition of the doctrine of recollection
in the Meno, we can see that Aristotle is construing the problem presented in
that dialogue as how one comes to recognise that particular figure as the one
which . . . (that square as the one which has an area twice that of a given square).10
That is to say, even supposing that the slave somehow has knowledge
of geometry, his knowledge is of the universal and not of this square
in the dust; how then does he use this square in the dust to activate his
knowledge of squares - without which he cannot see that a square drawn
on the diagonal is the solution to the problem?
Of course, where geometry is concerned, the relation between token
and type is unusually transparent, and Aristotle's own example reflects
this fact. Nevertheless, as we have seen, the question remains: How out
of all one's knowledge is one able to access a triangle (or to return to
the example of the Meno, how is the slave able to access the correct
square)? While Aristotle himself agrees that knowledge of the instances
comes by recognizing something as an instance of a given type, he rejects
the thought that foreknowledge is either necessary or sufficient for
10 Hamlyn (1976), 170. See also McKirahan (1983).
68
recognition. Rather he seems to believe that an account of human knowing can be given which has an immediacy - and also a credibility lacking in Plato's. In the sections that follow, Aristotle's account will
be examined.
II
The goal of the Theory of Recollection is to explain how learning occurs.
Plato chiefly proposes that, starting from perceptual cues, we regain
access to knowledge already within us. In Posterior Analytics II 19 Aristotle himself attempts to explain how, beginning from the perception
of sensible particulars, scientific knowledge - that is to say, knowledge
of universals resting on indemonstrable first principles - can be attained.
That his target is the Meno is plain from the outset. In the first chapter
of Book I at 71a 24-30, he has said:
Before the induction, or before getting a deduction, you should perhaps be
said to understand in a way - but in another way not. For if you did not know
if it is simpliciter, how did you know that it has two right angles simpliciter? But
it is clear that you understand it in this sense - that you understand it universally
- but you do not understand it simpliciter. (Otherwise the puzzle in the Meno
will result; for you will learn either nothing or what you know.)
Although the goal is said to be knowledge of "first principles,"
whether by first principles Aristotle means primitive concepts or indemonstrable propositions, or whether he vacillates between them, is
unclear. But, as several interpreters have recently pointed out, since a
first principle is for Aristotle a definition and a definition is the articulation of a concept, it really makes no difference.11 Having argued that
knowledge of principles is not, on the one hand, innate (Posterior Analytics II 19 99b26-7) and that, on the other, it is impossible for such
knowledge to come about if we are ignorant of it and have no capacity
for it at all (99b30-32), Aristotle concludes that we must have some
capacity from which to derive our knowledge of the principles (99b3234). Of this capacity, he says:
And this evidently belongs to all animals; for they have a connate discriminatory
capacity, which is called perception. And if perception is present in them, in
some animals retention of the percept comes about, but in others it does not
come about. Now for those in which it does not come about, there is no knowledge outside perceiving (either none at all, or none with regard to that of which
there is no retention); but for some perceivers, it is possible to grasp it in their
minds. And when many such things come about, then a difference comes about,
so that some come to have an account from the retention of such things, and
others do not. (99b34-100a3)
11 See Modrak (1987), 161-64, Kahn (1981), 387-97, and Barnes (1975), 249, 259-60.
69
Aristotle then continues by considering the steps leading from retention

to the account:
So from perception there comes memory, as we call it, and from memory (when
it occurs often in connection with the same thing), experience; for memories
that are many in number form a single experience. And from experience, or
from the whole universal that has come to rest in the soul (the one apart from
the many, whatever is one and the same in all those things), there comes a
principle of skill and of understanding - of skill if it deals with how things come
about, of understanding if it deals with what is the case. (100a3-9)
Next he stakes out the difference between his position and Plato's,
offering what has become a famous simile. He says:
Thus the states neither belong in us in a determinate form, nor come about
from other states that are more cognitive; but they come about from perception
- as in a battle when a rout occurs, if one man makes a stand another does and
then another, until a position of strength is reached. And the soul is such as
to be capable of undergoing this. (100al0-14)
Finally Aristotle repeats and amplifies the simile as follows:
What we have just said but not said clearly, let us say again: when one of the
undifferentiated things makes a stand, there is a primitive universal in the mind
(for though one perceives the particular, perception is of the universal - e.g.
of man but not of Callias the man); again a stand is made in these, until what
has no parts and is universal stands - e.g. such and such an animal stands, until
animal does, and in this a stand is made in the same way. Thus it is clear that
it is necessary for us to become familiar with the primitives by induction; for
perception too instils the universal in this way. (100al4 100b5)
The simile of ending a rout is far from transparent. W. D. Ross and
Jonathan Barnes do not even attempt to correlate it in any detail with
the progression from perception to knowledge. 12 Hamlyn points out
that it is certainly irrelevant that the movement is a regrouping; otherwise knowledge would be recollection, after all. 13 Richard McKirahan
says of it:
This simile has favorably struck some commentators and has given rise to much
speculation over the circumstances imagined in the simile and its precise application to the matter at hand. However, like some other famous similes of
Aristotle's, this one does nothing to make the ideas clearer.14
Deborah Modrak, however, attempts an analysis. She says:
Both the cognitive process and the rout involve a movement from discrete
individuals considered as such to an orderly arrangement of individuals in which
individual differences are irrelevant. During the rout, each soldier is an isolated
12 See Ross (1949), 673-78, and Barnes (1975), 252-56.
13 Hamlyn (1976), 178.
14 McKirahan (1992), 244.
70
individual, as are sensible particulars when they are initially apprehended in

perception. When the reversal of the rout begins and some soldiers have resumed their battle positions, an orderly whole begins to emerge of which the
soldiers are constituents. Similarly, the generalization that is grasped through
experience orders a number of individual phantasmata - at least to the extent
of specifying similarities. When the battle formation is completely restored, the
formation can be fully specified in terms of the number, position, and location
of the soldiers; individual soldiers of the same type are interchangeable. 15
In Modrak's account it is the formation, the orderly whole, which
comes to be fully specified, and its specification is achieved by detailing
the relevant features of the soldiers; that is to say, on her account it is
the formation that comes to be known and therefore corresponds to
the universal. She continues:
The battle formation under a general description constitutes an ordering of
individuals and thus is analogous to a universal principle. The cognition of a
universal principle orders the particular states of affairs that are subsumed under
the universal, and the peculiar features of the phantasmata that embody these
states of affairs are irrelevant to their role in the cognition (cf. Metaphysics
1078a22-31, [De Memoria et Reminiscentia] 450al-4). There is no battle formation without soldiers, nor is there a noetic apprehension of a universal without the sensory representation of (some of) its instances; nonetheless, the first
member of each pair cannot be reduced to the second member. 16
It may be that W. K. C. Guthrie agrees. He says:
The process is illustrated by a vivid simile: it is as in a battle, when an army has
been routed, if one man has the courage to turn and make a stand, his example
fires another and then another, until their original order (arche) is restored.
From our earliest years we are bombarded with a confused mass of sensations.
A great many we forget at once: they slip away and flee from us. But there
comes a time when one remains in our memory, then more and more. Gradually
we are becoming experienced. Finally, being creatures possessed of reason, we
become aware of the arche, which is nothing more or less than the 'one beside
the many' [hen para to polla], a universal of which all the separately remembered
particulars are examples, and are enabled to produce that definition which is
one of the archai of the scientific or apodeictic syllogism. (90b24) 17
What is first to be noticed is that on any interpretation of the simile
it has to be agreed that Aristotle has told us nothing about how the
transition from perception to knowledge is to take place, how, that is
to say, it comes about that the soldiers start not just running around
but (re)grouping. Yet while the simile presents the activity entirely in
relation to the achievement of its result, it is precisely the possibility of
that achievement, the critic can seemingly protest, which remains so
puzzling. To put the point another way, if Aristotle can get away with
15 Modrak (1987), 169-70.
17 Guthrie (1981), 182.
16 Modrak (1987), 170.
71
saying, as he does at 100al3-14, that the soul is such as to be able to

reach the universal, why should not Plato, with equal justification, say,
"the soul is such as to be able to recollect"?
Perhaps in Aristotle's defense it can be said that his remark to the
effect that the soul just is able to extract the universal is intended to
make only the point that not all creatures have knowledge (animals do
not). Yet the absence of a psychological mechanism is not the only or
seemingly even the most serious problem for interpreting the simile.
Guthrie, as we have seen, chooses not to specify what in the metaphor
corresponds to the universal (the one beside the many), while Modrak
claims that it is the formation that does so. Indeed since definitions are
the archai of the scientific or apodeictic syllogism (90b24) and definitions
are of the essence or universal, if what is reached or restored is an
arche, it is natural to suppose, as she does, that what is known is the
collection of individuals in a certain arrangement. However, knowledge,
as Aristotle understands it, is of what it is to be a human being, for
example, and what it is to be a human being is something that exists
within each individual human being, while on the other hand the formation is not in each soldier but just is, as I have said, a collection of
them.
Thus although I agree with Modrak that the interchangeability of
individual soldiers is important, the reason why it is so, it seems to me,
is that it is the soldiers engaged in the characteristic activities of soldiers
and not, as Modrak concludes, the formation that needs to be what is
known. One can see the point more clearly if one applies the metaphor
to itself so that what is to be learned is what a soldier is. In a rout the
soldier is not functioning as a soldier; yet even though (in the case of
a hoplite), for soldiering the formation is required, nevertheless in coming to know what it is to be a soldier, it is not the formation that is
what is to be known.
In short, if the analogy is to be useful, what the formation has to
correspond to, it seems to me, is not what is known but simply the state
of having reached a condition of knowledge. What is needed from Aristotle's comparison with the ending of a rout, that is to say, is an
awareness of each particular in such a way that one is aware of it in its
typicality. Whether the analogy can be so interpreted, however, is a
question that I will not pursue. Yet even if, as McKirahan says, the
metaphor is just not helpful, nevertheless from Posterior Analytics and
De Anima I think it can be seen why Aristotle believes he can respond
to Plato's epistemological problems in a way that is successful.
Ill
In De Anima II 6 Aristotle tells us that the proper objects of perception
are sense qualities or sensible forms. These qualities, if unique to a
72
particular sense (as color is to sight), are called proper sensibles; if

distinguished by two or more senses (as are motion and shape), they
are called common sensibles. Objects such as this man and this horse
are said by Aristotle to be perceived incidentally.
Sense perception, whether of qualities or objects, is of the individual
(Posterior Analytics I 18 81b6; De Anima II 5 4l7b22). Yet as Analytics
Posterior II 19 makes clear (cf. Metaphysics I 1), we proceed from sensation to memory (sensation retained in a phantasm), to the unification
of memories in experience, and then to a grasp of the universal. However, if the starting point is sense perception or, even more narrowly,
the sensation of colors, shapes, and the like, sensory input may seem
insufficient for knowledge; in the gap between perception and knowledge there is much to be explained. One solution would be to hold that
when we perceive, for example, Callias, besides perceiving certain colors
and shapes, we perceive him as a human being. Of this Barnes says:
Yet it is not clear how we are to apprehend man in the first place. Aristotle's
theory of perception divides the objects of perception into two classes, essential
and incidental (cf. An [De Anima] B 6). Essential objects are either proper to a
given sense (e.g. colours to sight, sounds to hearing) or common (e.g. motion,
shape, size). Incidental objects cover everything else; if X is an incidental object
of perception, then I perceive X only if there is some essential object Y such
that I perceive Y and Y is X. Individuals are the prime examples of incidental
objects (An B 6, 418a21; T 1, 425a25). There is very little evidence for man but
what there is makes it an incidental object (An T 6, 430b29); and it is in any
case hard to see how man could be either a proper or a common sensible. Man,
then, is not directly implanted in our minds by our senses, as Aristotle's words
in B 19 suggest; but in that case we need an account, which Aristotle nowhere
gives, of how such concepts as man are derived from the data of perception.18
Charles Kahn repeats Barnes's concerns:
As Barnes has noted, the full account of aisthesis in De an. II makes it difficult
to understand how we could properly perceive a universal like man at all - or
even an instance of this universal. The proper objects of perception are aistheta,
sense qualities or sensible forms, and these are of two types: (1) qualities unique
to a particular sense (color for sight, sound for hearing, hot-cold, wet-dry, and
others for touch, plus the various tastes and smells), and (2) the "common
sensibles" distinguished by two or more senses: motion, rest, number, figure,
and magnitude (De an. II 6). These are the only properties that are sensible per
se, recognized by sense perception as such. All other information received
through the senses, such as the recognition that this pale shape is a man or the
son of Diares, is only an incidental object of sense: "one perceives this incidentally, because this property happens to belong to the white object one perceives. Hence the perceiver (i.e. his sense of vision) is not acted upon at all by
the incidental sense object as such." (418a21-24)19
18 Barnes (1975), 255.
19 Kahn (1981), 401-2.
73
Both Barnes and Kahn cast doubt on the idea that for Aristotle we
can know what a human being is by perceiving Callias and others as
human beings. Aristotle's psychology does not, they believe, allow Callias
to be perceived in such a highly conceptualized way. But in that case
how do we come to know what a human being is? Kahn's suggestion is
that for Aristotle it is the active intellect that permits us to learn from
experience, by acting on us so as to actualize our potential intellect.20
Indeed, according to Kahn, far from being a version of empiricism,
Aristotle's alternative to Plato is "superrationalism." As he says,
As far as the active intellect is concerned, the rationalist doctrines of innate
ideas and infallible intuition convey too weak a picture of its complete and
unwavering grasp of all noetic truth. As far as we are concerned, however, the
rationalist model simply does not fit, since our potential intellect is not stocked
at birth with noetic principles and does not acquire them by any act of direct
intuition that we can perform. One may, if one wishes, speak of the active
intellect as continuously intuiting the forms and essences of the natural world.
But we can enjoy such an intuition only to the extent that we succeed in realizing
its activity in our own thought and knowledge.21
Despite the fact that, as Kahn understands Aristotle, we have no innate
ideas, "if the intellect can grasp the noetic forms and essences in the
phantasms provided by experience, that is because it already knows or
is these very forms before experience." 22
Thus Kahn's view of Aristotle can, I think, not unfairly be called
Platonic - the forms can be known from experience because the forms
were known before experience. Nevertheless if Kahn is right about the
role of the active intellect and if he is right to suppose the active intellect
to be something apart from us,23 how is it that the knowledge had by
the active intellect comes to be ours as well? What Kahn offers on
Aristotle's behalf falls very short of being an explanation. He says:
And this process of learning and exercising science, although it has a metaphysical cause and even a metaphysical guarantee in the super-rationalism of
the active intellect, must be achieved in our own experience by the ordinary
processes of induction and hard work: there is no epistemic button we can push
in order to tune in on the infallible contemplation of noetic forms by the active
intellect.24
If Kahn's interpretation is correct, then in the De Anima as much as

in Posterior Analytics II 19, how Aristotle's view is an improvement over
Plato's is hard to see. We engage in the hard work of epagoge - but
20 Kahn (1981), 410.
21 Kahn (1981), 411.
22 Kahn (1981), 407.
23 See Guthrie (1981), 322-30, for a summary of the evidence on both sides of this
question.
24 Kahn (1981), 411.
74
what is the connection between that and the engagement of the active
intellect on our behalf? Whether the forms are buried within our memories or whether they are known by an active intellect that is distinct
from us, the problem of access is the same. Nevertheless it must be
remembered that the problem the involvement of the active intellect is
intended to solve arose because Kahn, in rejecting the idea that we can
perceive Callias as a human being, takes Aristotle to be committed to
a perceptual content of only proper and common sensibles. But some
interpreters believe Aristotle intends a more generous perceptual content than that which Kahn allows. One such proposal will be the subject
of the next section.
IV
According to Kahn's account of the movement from perception to
knowledge, incidental objects of perception fall outside the scope of
perception. Modrak, however, uses Aristotle's remark that though one
perceives the particular, perception is of the universal to suggest a way
in which universals might be grasped through perception after all. She
says:
[At 100al5-b3 Aristotle makes] the crucial point that there is a sense in which
universals are grasped through perception. The universals in question are those
that determine the features of the sensible particular; these include features
perceived kata sumbebekos as well as those perceived kath * hauta. Indeed, in the
case of substances, the single most important universal, namely, the substance
sortal that the particular falls under, is perceived kata sumbebekos. What is incidental at one cognitive level becomes essential at another (cf. 89a33-37). The
essential features of a concrete particular are general characteristics. Hence the
perceptible individual is not only the object of perception but is also the vehicle
for the apprehension of the universal.25
She explains how for Aristotle this is possible:
The sensible particular is a token of a type, and we apprehend the type in virtue
of apprehending the particular. In Aristotle's metaphysics, the substance type
is ontologically prior to the token; the essential characteristics of an individual
human being are determined by the substance type, the species anthropos. We
perceive a particular man (i.e., a token of a certain type or species), but the
perception is of man (i.e., the type is the ultimate determinant of the content
of the perception). Many of the distinctive perceptible features of a particular
object are type-dependent. For instance, the difference between a cat and a
dog is more easily recognized through perception than the difference between
one cat and another, (cf. Physics 184a24-bl4)26
One might conjecture that Barnes and Kahn would find Modrak's
view problematic; Kahn, it will be recalled, doubted that we could per25 Modrak (1987), 168.
26 Modrak (1987), 168.
75
ceive either a universal such as man or an instance of the universal.

What should be noticed, however, is that the position Barnes and Kahn
explicitly contest is stronger than the one Modrak defends. Both Barnes
and Kahn deny that Aristotle is entitled to say, "we perceive Callias as
a man." But Modrak does not say that we do. Rather what Modrak
claims is that many of the distinctive perceptible features of a particular
object are type-dependent. That is to say, it is consistent with Modrak's
view to say merely that on various occasions we perceive similar clusters
of proper and common sensibles.
In short, some interpreters try to find middle ground between perceiving colors and shapes and perceiving Callias as a human being.27 In
fact it seems to me not implausible to suppose that Aristotle thinks that
without knowing what a human being is we perceive a cluster of qualities
that belong to a particular human being and that this is a sufficient
starting point for coming to know what it is to be a human being. In
fact Barnes's definition of incidental perception (if X is an incidental
object of perception, then I perceive X only if there is some essential
object Y such that I perceive Y and Y is X) seems to contain just such
a suggestion. To say that Y is X is not to say that I know that Y is X or
even that I perceive Y as an X, the latter being the claim that Barnes
himself finds problematic.
I have argued that Barnes's definition presents a more modest and
defensible view of incidental perception than that which Barnes himself
criticizes Aristotle for holding. But given a more modest interpretation,
Kahn's remarks, quoted in Section III, to the effect that it is difficult
to understand how we could properly perceive a universal, that man is
only an incidental object of sense, that the perceiver is not acted upon
at all by the incidental sense object as such, and that the sense faculty cannot
be directly affected by properties that are not sensible in the strict sense
(all the italics are mine) would seem to be either vacuously true - incidental perception is not nonincidental perception - or tendentious.
That is to say, Kahn has said nothing to show that incidental perception
of the sort prescribed in Barnes's definition is not available to Aristotle.
On the contrary, such a view would seem to be required by what Kahn
himself says of Aristotle's understanding of even animal awareness:
"[sense perception] permits a dog to recognize its master and distinguish its master from the master's horse, without knowing that its master
27 Others who have attempted to find middle ground include Ross, Sorabji, and Cashdollar. Ross (1949), 678, says of this passage: "we perceive an individual thing, but
what we perceive in it is a set of qualities each of which can belong to other individual
things." Sorabji (1992), 197, argues that for Aristotle some perceiving is propositional
(we perceive that something is the case), in other words that "perceiving that" need
not be described as an inference of reason. As he puts it, "Coincidental does not
mean inferential." Cashdollar (1973), 156, 161, argues that incidental perception,
even though predicative, is nevertheless a case of perception.
76
is a man or that the horse is a horse."28 Nor has Kahn shown that (for
suitable perceivers) from such a beginning the universal cannot be attained.
Thus I want to agree with Modrak that Aristotle uses incidental perception to explain how perceptible particulars can be the vehicle for
knowledge of the universal and that it is reasonable for him to do so.
Specifically, what Modrak argues is that it is the capacity for representation which allows us to reach knowledge of the universal from sensory
awareness of particulars. When the content of a perception is retained
in a phantasm or memory image, the image, being an image of Callias,
will include some of his accidental features. Nevertheless once images
of other human beings and other images of Callias are retained as well,
then (in the human soul) it is possible for even the essence to be derived.
This progression need not, Modrak contends, involve a superimposition
of the images into a confused general image; that, she maintains, is a
view that is neither philosophically plausible nor true to Aristotle. 29
Rather, the proposal is that any of these particular images can be reused
to represent the others. She says:
if Callias is perceived as tall, thin and balding, the phantasma of Callias employed
in thinking about the essence of human beings would include (some of) these
idiosyncratic characteristics, but the thought would ignore them. To grasp an
intelligible form is to reinterpret the content of an appropriate phantasma.
Nevertheless, the phantasma is a necessary component of the thought. Just as
in the extramental world essences inhere in matter, essences-in-thought inhere
in the phantasmata that serve as their material substrata.30
We have seen that on Modrak's view the gap between the phantasm,
which is still particular, and knowledge of the universal is closed by the
capacity for representation. That is to say, it is representation that takes
us from the phantasm to thought (432al2-14). In this it seems to me
that she is right. I will not pretend that no philosophical problems arise
in understanding how representation is supposed to work, nor will I
try to determine whether Aristotle wants to appeal to a general image. 31
But what needs to be emphasized, I think, is the connection between
28 Kahn (1981), 402. Aristotle's motive for considering incidental perception to be a
case of perception seems in fact to be animal intelligence; animals, he wants to hold,
lack reason and yet they respond appropriately. At Metaphysics I 980b25-27, a passage
not unlike Posterior Analytics II 19, Aristotle claims: "The animals other than man live
by appearances and memories, and have but little of connected experience." Of this
passage Ross (1949), 677, says: "I.e. experience is a stage in which there has appeared
ability to interpret the present in the light of the past, but an ability which cannot
account for itself; when it accounts for itself it becomes art." Sorabji (1992), 195209, has an excellent account of the problem posed for Aristotle and other ancient
philosophers by animal intelligence.
29 Modrak (1987), 166.
30 Modrak (1987), 169.
31 D. Frede (1992), 291, seems, unlike Modrak, to believe that Aristotle does make use
of a general image and that his doing so is not unsatisfactory.
77
Aristotle's metaphysics and his epistemology. I will argue next that the
fit between Aristotle's epistemology and his metaphysics, on the interpretation of the theory of substance which I have offered, should be
taken as a reason for thinking that interpretation to be correct.
I have argued that Aristotle wants his substances to be as much like

Plato's Forms as possible, but for separation, and that Plato's Forms
can best be seen as types. But although metaphysical issues fall outside
the scope of her book, in an earlier paper cited in it Modrak gives an
account of substances which, though it may differ to some degree from
the view I want to propose, nevertheless illustrates why there should
be said to be a close connection between Aristotle's metaphysics and
his epistemology. For what Modrak claims is that Aristotelian substances
are types. Specifically, she suggests that the relationship between the
form and the individuals having that form is like that between a word
and its occurrences; the form is like 'cat' as the arrangement of the
letters 'c,a,t', while the individual is that form or formula in matter, as
in the ink marks on the page. Similarly in the case of the human form
and human individuals, the form is the functional organization characteristic of human beings and the individual is a particular organized
body such as Socrates.32
I myself am not entirely comfortable with Modrak's examples. Although she says that the form, being that which provides individuation,
is not a property, nevertheless "the arrangement of the letters 'c,a,t' "
and likewise functional organization are propertylike ('c' is to the left of
'a'); thus I would contend that even the formula in itself must be the
letters arranged (not, of course, inked letters arranged) and not their
arrangement, if one is to preserve the ontological priority required of
substances. But of course in its standard use the distinction between
types (for example, The Grizzly Bear and The American Flag) and tokens
(grizzly bears and American flags) is such that what types are derived
from are sortals.
My point is that, if Modrak's examples were made to conform to her
suggestion that forms are types, Aristotle's intention would, I believe,
be clear. That Aristotle would think that The Grizzly Bear - that is to
say, the Form supposed to be numerically distinct from sensibles cannot exist is undeniable. Yet I maintain that the goal of his metaphysics and epistemology is to achieve the virtues of Plato's Forms
without separation and that this goal is accomplished by the postulation
of specimens of kinds that are numerically the same as sensible objects
32 Modrak (1979), 375-76.
78
but do not have the accidental properties of the latter. However, to say
that Aristotle proposes that substances are, as it were, unseparated types
and thus that he must explain how there can be knowledge of these
would be to hold the same view. Moreover, the result of the process
that Modrak has described, namely, the reuse of a phantasma to arrive
at knowledge of the universal, or more precisely, at knowledge of an
unseparated type, is just knowledge of a specimen of a kind. Thus
Aristotle's epistemology, as Modrak describes it, would match his metaphysics as I understand it to be.
In Aristotle's account of the movement from perception to knowledge, it has often been noticed that he does not address skepticism of
the sort that concerns us - or even Plato. As his starting point, for
example, Aristotle claims that in perception the perceiving and the
object perceived are the same (431b20-432al); having postulated that
in perception the form of the object is taken in without the matter in
something like the way a piece of wax takes on the imprint of a signet
ring (424al7-24), Aristotle does not then concern himself with whether
there might not really be a surface to be seen or whether the white
human-shaped surface might nevertheless not be a human being.
In other words, Aristotle never doubts that we have knowledge. But
if he is addressing the problems introduced by Plato's Theory of Forms,
the otherwise puzzling limits of his epistemological interests make sense.
To start from experience and assume it to be experience of something
outside oneself is to do no more than Plato did in granting us awareness
of the phenomenal world. The difference between Plato and Aristotle
arises rather in the relation between what is in flux and what is unchanging. If, contrary to Plato, what is perceived is numerically the
same as what is to be known, then, Aristotle believes, the world is such
that what is perceived is also, in a way, what is both knowable and known.
That is to say, if what is perceived is numerically the same as what is to
be known, then Plato's peculiar problem of retrieval on the basis of
vague and ambiguous phenomenal cues vanishes, making inquiry - rational investigation from a base of experience - possible. As T. EngbergPederson explains Aristotle's theory, ''[One can attend to] particular
cases with the consequence that insight into some universal point is
acquired."33
Thus even if we may go wrong on a particular occasion, even if nous
is not presumed to guarantee the truth of whatever is grasped,34 by his
claim that substances are numerically the same as sensible objects,
Aristotle believes that he has eliminated reasons of the sort that confront
Plato for doubting that we can and do know that which is ontologically
fundamental. To put the point another way, in not considering Cartesian
33 Engberg-Pederson (1979), 305.
34 See Engberg-Pederson (1979), 308, and Lesher (1973), esp. 58-65.
79
grounds for doubt, it can fairly be said that Aristotle just assumes that
there is an intelligible world and that we can have knowledge of it.
Nevertheless what needs also to be said is that as a consequence of these
assumptions, Aristotle takes it to be a criterion for a successful metaphysical theory that it propose a world where such knowledge can plausibly be obtained and that by his rejection of separation he believes that
this is just what he has done.
VI
I have agreed with Modrak that Aristotle's account of the acquisition
of knowledge depends on the reuse of a phantasm in such a way that
any human being can represent any other. Of course, there are many
features of Aristotle's epistemology and philosophy of mind which I
have not considered. Nevertheless I believe that I have said enough to
show that his epistemological view is encouragingly parallel to the metaphysical account of substance given in earlier chapters. For there I had
argued that, as a result of the restrictive function of 'qua', it is possible
to refer not only to Socrates but also, by using the expression 'Socratesqua-human being', to a specimen or paradigm of the kind human being.
I had argued too that since accidental properties are eliminated there
is no qualitative difference among specimens of the same kind, with the
result that for Aristotle substances are knowable despite not being universals. Now I have argued in this chapter that, in the extraction of the
knowable from perception, what are known are in fact specimens of
kinds.
But if the metaphysical grounding of Aristotle's epistemology is as I
have argued it to be, then contrary to what is sometimes thought, Posterior Analytics II 19 and the parallel account of perceiving and thinking
in the De Anima are consistent with what Aristotle says in Metaphysics
XIII 10. In that chapter Aristotle describes the limits of knowability,
saying:
The statement that all knowledge is universal, so that the principles of things
must also be universal and not separate substances, presents indeed, of all the
points we have mentioned, the greatest difficulty, but yet the statement is in a
sense true, although in a sense it is not. For knowledge, like knowing, is spoken
of in two ways - as potential and as actual. The potentiality, being, as matter,
universal and indefinite, deals with the universal and indefinite; but the actuality,
being definite, deals with a definite object - being a 'this' [tode ti], it deals with
a 'this'. But per accidens sight sees universal colour, because this individual colour
which it sees is colour; and this individual a which the grammarian investigates
is an a. (1087al0-21)
The contrast between potentiality and actuality in this passage is
admittedly problematic. Walter Leszl supposes that it is the contrast
80
between, say, Socrates' knowledge of geometry when he has studied the

subject but is not now thinking about or contemplating geometrical
theorems (potential knowledge) and his now thinking, accurately, about
geometry (actual knowledge). 35 Yet Robert Heinaman, on the other
hand, has argued that the relevant contrast between potentiality and
actuality is not this familiar one; rather by 'potential knowledge' he
takes Aristotle in this context to mean having knowledge of one object
(Heinaman says one proposition) in virtue of having knowledge of some
other object (or proposition). 36
Heinaman calls the contrast between dispositional and occurrent
knowledge of the same object a contrast between potentiality! and
actuality! and the contrast where the objects differ that between
potentiality 2 and actuality 2. His claim then is that it is a distinction
between potentiality 2 and actuality 2 which is at issue in XIII 10, and in
this it seems to me that he is right. However his application of this
distinction is less persuasive. What Heinaman says is:
The problem Aristotle is concerned to answer in M 10 is the problem of how
principles (substantial forms) can be known if they are not universals. And if we
understand his answer in terms of the distinction between actual2 and potential*,
knowledge he must be saying the following: the knowledge of the universal
constitutes potential knowledge of the individual principles, the individual substantial forms. So, for example, the knowledge of the universal human soul
constitutes potential knowledge of Socrates' soul, Plato's soul, etc.37
He concludes:
In M 10 Aristotle is not giving up his belief that knowledge of the individual
requires knowledge of the universal (cf. 1036a8, 1086b5-6, 32-7), for the
individual is known by actual2 knowledge which is the actuality of the potential.,
knowledge which consists in knowing the universal. Rather, Aristotle is denying
that this fact entails that the universal which is known is a substance. So in order
to know that Socrates' soul is ABC I must know that the universal definition
of human soul is ABC. But knowledge of the universal is not knowledge of a
substance, except potentially. It constitutes potential., knowledge of individual
substances.38
The difficulty with Heinaman's view is that it is hard to see how, on his
account, the individual - not the universal - has epistemological priority.
To be sure, Heinaman can say that the relevant universals cannot themselves be objects of actual2 knowledge since there are no other propositions knowledge of which could constitute potential knowledge of
them.39 Nevertheless this seems, if anything, to establish their priority,
not their lack of it.
35 Leszl (1972), 312.
37 Heinaman (1981b), 70.
39 Heinaman (1981b), 70, 67.
36 See Heinaman (1981b), 67, 70.

38 Heinaman (1981b), 75.
81
Thus if one is to preserve the priority of the individual, it seems

necessary to suppose that what Aristotle is saying is that, even though
the numerical distinctness of specimens of the same kind prevents claiming that in actually 2 knowing any individual of a given kind one actually 2
knows the others, one does potentially 2 know those others and that to
do so just is what it is to know potentially 2 the universal and indefinite.
This view, it seems to me, is consonant with Aristotle's earlier discussed
criticism of the Meno on the grounds that knowledge of the instances
comes along with epagbge. In any case, of the argument in XIII 10 itself,
Joseph Owens says the following:
The form as seen in the thing is not universal. It is this definite form. That same
form, however, is seen in other singular things, indefinitely numerous. Yet the
knowledge of the form as seen in the one definite instance in a singular, is able
to be applied indefinitely to all things in which that form happens to be found. 40
A given a may be written with ink or chiseled in stone; it may be scrawled
or flourished. Yet if it is to be known in such a way that to know it is
potentially to know all other a's, it must be known as and only as an a,
that is to say, as a member of its kind. 41
Letters of the alphabet are, of course, hardly substances. Nonetheless
the analogy between knowing a's and knowing individual human beings
is clear. If one knows the universal by knowing an a as an a, so with
Socrates and Callias; Socrates may be a snub-nosed, penniless, exasperating yet inspiring Athenian philosopher, but to know what it is to
be a human being, that is to say, potentially to know Callias and others
by knowing Socrates, one must know Socrates (and indeed on the whole,
for Aristotle this is the only way in which one can know an individual
human being) just as a human being.
Thus, even though in Metaphysics XIII 10 Aristotle says that actual
knowledge is of the individual, that Socrates is not in all his peculiarity
a knowable object is, I think, as true in this chapter as in the Posterior
Analytics. The point of Aristotle's account of substance, however, is that,
having admitted as much, one need not agree with Leszl, for example,
when he says: "As to the 'this' which is supposed to be known, it is
something individual and is not the actual content of knowledge (which
is the universal, or rather the universal connection, which it instantiates),
but rather what knowledge is about." 4 2 For even given that for Aristotle
40 Owens (1963), 429.
41 At De Anima 4l7a24-29 also, Aristotle uses the example of letters: the person who
has a knowledge of grammar has a potentiality in the sense that he can reflect when
he wants to, while the person who is already reflecting is a knower in actuality of this
A. In other words, the former has knowledge of the letter type, where this is potential
knowledge of a token of that type. However what Aristotle says in De Anima is only
half the story; what De Anima II 6 and Posterior Analytics II 19 address is the other
half, namely, how knowledge of the letter type comes to be.
42 Leszl (1972), 307.
82
universals do not exist apart from individuals, to hold as Leszl does that
the content of knowledge is the universal is to preserve what Cherniss
has called "the discrepancy between the real and the intelligible,"43 and,
as I said in Chapter I, it is this very discrepancy that Aristotle worries
about in the aporia in Metaphysics III. His solution, I have claimed, is
that Socrates' form, a specimen of the kind human being, is, by virtue
of being numerically the same as Socrates, a particular and yet knowable
just because it is indistinguishable from other specimens of the same
kind. But if knowledge is of the universal only in the sense that it is
indeterminable which specimen of a given kind it is that one knows,
then Metaphysics XIII 10 and the passages in Posterior Analytics and De
Anima which have been discussed are consistent.

In this chapter I have argued that Aristotle attempts to show that a
plausible account of coming to know can be had once Platonic separation is abandoned. Even if he is vaguer than we would like in his
description of the psychological process by which knowledge is attained,
clearly Aristotle does believe that we move from our perception of
Socrates to knowledge of what it is to be a human being. Specifically,
for Aristotle we can come to have knowledge of Socrates-qua-human
being (and thus potentially also Callias-qua-human being and so on)
because the individual human beings that our phantasms represent are
numerically the same as specimens of the same kind.
Yet to argue, as I have done, that Aristotle's substances are like
Platonic Forms without separation - that is to say, that the rejection
of separation is the keystone of Aristotle's metaphysics and epistemology
- presents two problems. The first is that even if, as I maintain, Aristotle
takes separation to be the chief fault of Plato's Theory of Forms, as I
acknowledged in Chapter I, Aristotle himself says that substances must
be separate. The second concerns ontological priority. Aristotle may
want to say that, as learning proceeds from what is less knowable by
nature to that which is more knowable, it also proceeds from that which
has little or nothing of reality to that which has reality (1029b4-12; cf.
Posterior Analytics 7lb33-72a5); he may want to hold that the use of
'qua' in describing what comes to be known is strictly speaking not
abstraction from the real but the extraction of it.44 Nevertheless it is
by no means clear that specimens of natural kinds can plausibly be said
to be ontologically prior to the sensible objects with which they are
numerically the same. These objections will be addressed in the two
chapters remaining.
43 Cherniss (1944), 340.
44 See Guthrie (1981), 190.
V
The Separation of Substance
In the interpretation of Aristotle's account of substance I have proposed

thus far, I have claimed that Aristotle believes that by denying separation
he can uphold the epistemological, and, as I will argue in Chapter VI,
ontological priority of substances, where those requirements are understood in very Platonic terms. I have claimed further that my interpretation of the motivations for Aristotle's view of substance makes understandable his account of how we come to have knowledge.
Nevertheless, as I said in Chapter I, even as Aristotle criticizes Plato
for separating the Forms, he says of substances that they must be separate. In Metaphysics VII 1, for example, Aristotle says:
Now there are several senses in which a thing is said to be primary [proton]; but
substance is primary in every sense - in formula, in order of knowledge, in time.
For of the other categories none can exist independently [choriston], but only
substance. And in formula also this is primary; for in the formula of each term
the formula of its substance must be present. And we think we know each thing
most fully, when we know what it is, e.g. what man is or what fire is, rather than
when we know its quality, its quantity, or where it is; since we know each of these
things also, only when we know what the quantity or the quality is. (1028a31-b2)
For Fine that Aristotle should claim separation for his substances
while criticizing Plato for separating the Forms in no way presents a
problem. Platonic Forms, as Fine understands Plato, are universals, and,
according to her, are thought by Aristotle to be so.1 Aristotle's objection
to Plato, as she sees it, is that universals should not be said to have a
capacity for independent existence. Substance, on the other hand, which
is understood by Fine to be for Aristotle the composite, can be said to
have just this characteristic, existing independently of other substances
as well as of accidents.2
1 Fine was discussed in Chapter I. For Forms as universal, see (1984), esp. 45.
2 Fine (1984) argues that independent existence can apply to the composite in the sense
that composites are separate from their attributes. By this she means that although
83
84
But Charlotte Witt also takes substance to be the composite and holds
that for Aristotle substances are separate while nonsubstances are not;
she understands the separation of substances to mean that nonsubstances and only they exist in subjects, thereby making nonsubstances
ontologically dependent on substances.3 If Witt also believes that Platonic Forms are universals (a question on which she offers no opinion),
then one can assume that she would find no conflict between Aristotle's
criticism of Plato for separating the Forms and his views about the
separation of substance. That is to say, if Forms are such as to be
predicable, then Aristotle is justified in saying that they too must exist
in subjects, while substances do not.
Fine and Witt therefore offer views on which there is no difficulty
about the fact that Aristotle both criticizes Plato for separating the
Forms and, in the same sense, claims separation for his substances.
However, on my interpretation, as has no doubt become apparent,
Aristotle's claim to separation in the case of substances is problematic.
For what I have said is that Aristotelian substances, being forms in the
sense of specimens of natural kinds, are in fact rather like Platonic
patterns except as regards separation. I will need to propose, therefore,
a way of understanding the separation of substances which is different
from the separation had by Platonic Forms. To do so will be the task
of this chapter.
As we saw in Chapter I, Aristotle's claim that Plato separates the Forms

has been understood in several ways. Fine appeals to Metaphysics V 11
- a definition of natural priority - in order to argue that independent
existence is what is meant,4 while I have claimed that what Aristotle
disapproves of in Plato's Theory is the numerical distinctness of the
Forms from sensibles. But some of the same disagreements arise when
Aristotle says that substance is separate. When Aristotle says in Metaphysics VII 1 that none of the categories except substance is choriston
(1028a34), Barnes translates this as ''can exist independently," and indeed, support for Fine's view of separation can be found in the fact
Socrates cannot exist without matter or accidents, he need not have the exact matter
or accidents that he has. Morrison (1985), 132-33, claims that substances must be
separate from other substances and criticizes Fine by arguing that Socrates is not
separate from the sun since he cannot exist without it. Fine responds by agreeing that
each substance should be said to be separate from every other substance and argues
that Socrates is in fact separate from the sun as he can exist briefly without it. See Fine
(1985), 163, for discussion both of her original view and its clarification.
3 Witt (1989), 48, 51. Substances might seem to exist in matter and thus to lack separation,
but Witt argues, and I agree, that the relation of the composite to matter is not the
same as that between accidents and their underlying substance. See 53.
4 Fine (1984), 35.
SEPARATION OF SUBSTANCE
85
that VII 1 makes use of the categories and in the Categories an appeal
to independent existence is found: "Thus all the other things are either
said of the primary substances as subjects or in them as subjects. So if
the primary substances did not exist it would be impossible for any of
the other things to exist" (2b4-6).
Thus VII 1 alludes to some doctrines from the Categories. On the
other hand, there are problems with importing views about separation
from the Categories into the Metaphysics. In the first place, while the
view expressed in the Categories clearly states the priority of substance,
it does not state that substance could exist without the other categories,
only that they cannot exist without it, and if the former claim is intended,
it is hard to see how it can be sustained. In the second place, whatever
the view in the Categories and whatever its justification, it cannot be
assumed that Metaphysics VII 1 is simply a restatement of it; in the
Categories, after all, views about substances were uncomplicated by matter. For both these reasons, therefore, it seems that we cannot rely on
the Categories to explicate VII 1. It is likewise with Physics i85a31-32,
where Aristotle says, "For none of the others can exist independently
except substance; for everything is predicated of substance as subject";
since in Metaphysics VII 3 matter, form, and the composite are all said
to be a sort of substratum (1029a2-3), appeal to that notion does not
clarify the meaning of 'separation'.
It seems then that there is no recourse but to determine from the
Metaphysics itself what the separation of substances can be,5 yet in the
Metaphysics exegesis of the sense in which substances must be separate
is complicated by Aristotle's own terminology. Although Aristotle sometimes distinguishes separation in space from separation in definition,6
and the latter from simple or unqualified separation (VIII 1 1042a2831), at other times he just says 'separation' (1028a33-34). In VII 3
matter is rejected as a candidate for substance on the ground that it is
neither tode ti nor choriston (1029a27-28). Nevertheless here too what
that which is separate7 is to be separate from and in what way it is to
be so are not clarified.
5 Metaphysics XIII 10 1086b 14-20 might seem promising. Aristotle says: "Let us now
mention a point which presents a certain difficulty both to those who believe in the
Ideas and to those who do not. . . . If we do not suppose substances to be separate,
and in the way in which particular things are said to be separate, we shall destroy that
sort of substance which we wish to maintain; but if we conceive substances to be
separable, how are we to conceive their elements and their principles?" But unfortunately, how particular things are to be separate is, as we have seen, itself disputed.
6 De Anima 4i3bl4-15, 429al 1-12, 432*20; Metaphysics ioi6b2-3, 1048bl5, 1052bl7;
On Generation and Corruption 32ob24, 320b 12-14; Nicomachean Ethics no2a28-31; Physics i93b4-5.
7 Some interpreters prefer to translate choriston as 'separable' rather than separate. I do
not find it necessary or useful to make a distinction here, but some of the varieties of
separation that I consider would be described by others as varieties of separability.
86
To be sure, if one thinks that Aristotle's successful candidate for

substance is the composite, it is not hard to suppose that all the senses
of 'separation' commonly put forward - independent existence, numerical distinctness, or spatial discontinuity - apply. If, on the other
hand, as I think, substance is form, it is difficult to think that any of
these could capture whatever Aristotle intends. In Chapter III, it will
be recalled, I argued that, when Aristotle says that substance is form,
what this means is that substances are specimens of natural kinds that
are numerically the same as but not identical with sensible objects, and
the question at issue is how substances, so understood, could possibly
be separate. On the other hand, consider, for example, Aristotle's remarks about the whole line and the half. Aristotle says that in actuality
the whole line is prior to the half line and substance to matter (Metaphysics V 11 1019a8-10). But since, as we saw in VII 1, one of the
conditions offered for priority is separation, it follows that the whole
line is also separate from the half, as is substance from its matter. Here
separation can hardly mean independent existence or numerical distinctness or spatial separation.
In short, I want to claim that the separation of the whole line from
the half cannot be understood in terms of any of the senses of 'separation' already discussed. To be sure, one might think it reasonable to
suppose that the way in which the whole line is separated from the half
is only that it is separate in definition, another variety of separation
which, as we have seen, Aristotle mentions. But were this the only way
in which it is so, the analogy to substance and matter would have no
ontological force; more specifically, neither substance nor the whole
line could be used to illustrate priority "in respect of nature" (1019a2).
Yet if we suppose the separation of the whole line to be the ontological
equivalent to separation in definition, this problem would be resolved.
That is to say, to be separate in definition, so Aristotle tells us, is to be
such that in a definition of A no reference is made to B - that is, in
saying what A in itself is we do not need to say that it is (a) B. Likewise
what it means for A to be separate from B, if separation is the ontological
correlate to separation in definition, is that A would be such that B is
not at any time (part of) what A in itself is. What I now want to argue
therefore is that, when he says that substances must be separate, by
'separation' Aristotle in fact has in mind the ontological correlate of
definitional separation.
The variety of separation just defined I am going to call "independent
being" (I mean this to be something different from "independent existence"); what I need now to try to say is what has this sort of separation
from what. Independence in being - that is to say, separation, as I
believe it should be understood - will obtain among many entities,
including, of course, entities that are at the same ontological level. In
87
saying of this chair what it is, one need not say of the coat lying on the
chair what it is; having a coat thrown on it is not part of what a chair
in itself is. But for both Plato and Aristotle, the more interesting issue
is, as it were, the "vertical" separation of what is ontologically and
epistemologically fundamental from what is in flux, and it is on this
relation that I will focus in explicating and defending my interpretation.
That Plato's Forms have "independent being" in relation to phenomena is clear enough. To say what The Shuttle Itself is one need not
- and indeed cannot - refer to its phenomenal images. The problem
with Plato's Theory, I take Aristotle to be arguing, is that Plato erroneously supposes that that which has independent being in relation to
phenomena must also be numerically distinct from them. This point is
made most clearly, I believe, in Metaphysics VII 6. Beginning with the
Platonic Forms, Aristotle proceeds in VII 6 to undercut the case for
the Forms by arguing that each self-subsistent thing and its essence are
the same. The argument is as follows. If the essence of good is to be
different from the Idea of Good, there will be substances besides those
postulated - that is to say, the essence of the Idea of the Good will be
a substance numerically distinct from the Idea of the Good and prior
to it (1031a31-b3).
Using the example of the essence of horse, Aristotle also expands
the argument in VII 6 in such a way as to create a regress. If the essence
of the essence of horse is different from the essence of horse (and if
the essence of horse is different from horse, on what grounds would
the essence of the former not be different from it?), the same result namely, the existence of substances prior to presumed substances - will
again obtain (1031b28-30). But regress, Aristotle tells us, is not the
only difficulty. For if the Idea of the Good, for example, is numerically
distinct from its essence, it will be unknowable (1031b4). That is to say,
whatever we know, we will not know the Form (the presumed knowability
of which was one of the chief motives for its postulation), but something
else, its essence.
In the case of the Forms, since there is nothing that is a property
of the Form and not a property of its essence or conversely, to deny
the numerical distinctness of a Form and its essence is to assert their
identity. Presumably to do so is the course Plato would want to take.
But, Aristotle claims, his argument need not apply to Forms alone. That
is to say, whether there are Platonic Forms or not, all primary and selfsubsistent things are, Aristotle claims, the same as their essences
(1032a4-6), a position that, if defensible, would seem to eliminate the
need for Forms. Nevertheless it is clear that if Aristotle intends, as I
think he does, to imply that Socrates and to be Socrates are the same
(1032a8), they cannot be identical; unlike a Platonic Form and its es-
88
sence, Socrates and to be Socrates must differ in properties if to be

Socrates is to be any more knowable than Socrates.
Thus what the argument in VII 6 shows, as I see it, is that Aristotle
himself needs both to assert and to deny separation. That is to say, even
as he in VII 6 denies the numerical distinctness (and independent existence) of what is most real and most knowable from what is changing,
he must assert along with Plato that there is, in some sense, separation
of what is ontologically and epistemologically fundamental from what
is in flux. To put the point in the terms that I have been urging, a
specimen of a kind, although numerically the same as a given sensible
object, is separate from that sensible object; to define what a horse is,
one need not refer to any given horse. Of course, even though specimens of kinds are separate from sensible objects in the sense of having
independent being, the converse does not hold, and this is just what
should be the case if Aristotle's claim of the ontological priority of
substance is to be sustained. To take an example, any given horse is not
separate from a specimen of the kind horse; to say what Secretariat is,
one must say that he is a horse, where what it is to be a horse - the
essence of any individual horse - is, I have argued, not a property but
a specimen of the natural kind horse.
I have argued that specimens of kinds have independent being with
respect to sensible objects while sensible objects do not have independent being with respect to those specimens. On the other hand, both
Secretariat and a specimen of the kind horse are separate in this sense
from Secretariat's accidental properties. Secretariat may be brown, but
in saying what Secretariat is, brownness is not part of what he is, nor,
of course, is it part of what it is to be a horse. Conversely, nonsubstances
such as colors have independent being only in a derivative way. That is
to say, brownness has an essence, but the definition of any property
will make it clear that properties are properties of substances (or other
objects) (1030b4-13). Finally a specimen of a kind is not separate, even
in the sense of independence in being, from its essence nor is its essence
separate from it; indeed just because in this case there is no difference
at all in properties, the relation is one of identity. But having read
Aristotle's critical remarks in VII 6 about Platonic Forms, that, of
course, is just what we would expect.
II
I have described the relation of separation which I take to hold between
sensible objects and the essence or substance of those objects as "independent being." But that it is possible to understand 'separation' in
this way can perhaps also be seen from Aristotle's discussion of matter.
To be sure, any attempt to understand the sort of separation matter is
89
said to lack is complicated by disagreement among interpreters of Aristotle on how the notion of proximate matter is to be understood. One
might suppose that the bronze of the statue - I do not mean the bronze
from which a statue can be made, that is, the preexisting matter, but
the bronze from which an existing statue has been made - is an example
of matter and that as such it has various familiar properties, including
size, weight, hardness, color, location, and, of course, shape. Nevertheless this view may in fact not be Aristotle's. For many recent interpreters, including Balme, Dancy, and Michael Frede, believe that matter
as such has no properties. As Dancy explains it:
In Z 3, we could turn directly to the bronze, which is something (the elements
mixed in the ratio X:Y) in its own right, but thinking of it that way is not thinking
of it as the material cause of both the statue and the lump into which the statue
is beaten. Thinking of it in this latter way is not thinking of it in its own right
at all, but as what used to be pale and charioteer-shaped and is no longer either.
When we consider the transmutation of fire into air, the case is clearest. For
here, there is no answer to the question "what is it that used to be hot and dry
and is now hot and wet?" at all: there can be no question of turning directly
to the stuff and discussing its essential structure. But this is relatively unimportant. It was at the top of Aristotle's hierarchy that the distinctness of form
was clearest; still, it was distinct from the matter all the way down. So also here,
at the bottom of the hierarchy, the inseparability of the matter is clearest; yet
it is inseparable all the way up.8
If one wonders why Aristotle might have wanted to say that it is not
the bronze as such which is matter but rather the bronze understood
in relation to the statue, the motivation would seem to come from cases
where it is not so easy to identify the matter of the composite with the
preexisting material. Ackrill explains the problem that arises if one tries
to apply a distinction readily made in the case of many artefacts to living
things:
The timber, hinges, and screws can still be seen when the cupboard is built,
but the eggs and sugar are lost in the cake. If, as a result of cooking, a and b
combine to form a homogeneous stuff c, a and b are no longer there to be
picked out. We can refer to the a and b we started with, and perhaps we can
recover the a and b again by some process. But a and b are present now, if at
all, only potentially. Actual bricks constitute an actual wall, though those very
same bricks might not have done so. But here is quite a different story: potential
a and b are 'in' actual c, though they might have been actual a and b. Chemical
change, in short, which yields a new sort of stuff, cannot easily accommodate
an account tailor-made for other operations. . . . This is the difficulty for Aristotle with the basic living materials such as flesh and bone. They are produced,
as he explains in detail in the biological works, by processes like cooking. . . .9
8 Dancy (1978), 408.
9 Ackrill (1973), 132.
90
In the case of living things, as Ackrill says, *'until there is a living

thing, then, there is no 'body potentially alive'; and once there is, its
body is necessarily actually alive."10 Moreover, that the matter in the
thing cannot be isolated from its structure leads naturally enough to
another way of dividing matter from form, such that the form is function. When Aristotle says, "Suppose that the eye were an animal - sight
would have been its soul" (412b 18-19), sight is a property of a structured object. That is to say, the structure that in simple cases seemed
to be form, in opposition to the matter, now seems to be a property
of the matter, in opposition to the form.
Ackrill once saw these complexities as deep sources of trouble,11 but
if matter as matter has no properties in its own right, it is possible after
all to give a uniform account of the matter of various sorts of entities.
Bronze has various properties and bronze is the matter of the statue
(that is, the bronze understood as matter), but the matter that can come
to be the statue or that already is so does not itself have these properties.
The same can be said of the matter of living organisms. In both cases
the matter is matter only in relation to the form; in both cases matter
is potentiality for that form.
I have said that a number of interpreters have argued that proximate
matter must be understood not as stuff, existing before, in, and after
the object into which it is formed, but as a potentiality for form and
so as not having properties "in its own right." Not everyone entirely
agrees. Mary Louise Gill has claimed recently that there are in fact three
concepts of matter in Aristotle. First, there is the preexisting matter,
for example, bronze, which Gill believes to have, as matter, the various
properties of bronze. On Gill's view Aristotle comes to think that preexisting matter survives in the object only as a property; thus the statue,
for example, is said to be bronzen, a condition in which the matter is,
Gill says (this being Aristotle's second concept of matter), generic or
indefinite. The third concept of matter is functional matter, as in the
living body of an organism, and Aristotle's functional matter, Gill agrees,
is defined in terms of form.12
In short, although a more thorough consideration of matter leads
far from the topic at hand, it can fairly be said that many interpreters
believe that the matter of an object, as matter, does not have properties
except in relation to the form. It is to be noticed, moreover, that such
10 Ackrill (1973), 132. Recently Ackrill's influential paper has played a prominent role
in the debate over whether Aristotle is in some broad sense a functionalist; however
it has been cited on both sides of the question. For the antifunctionalist view, see
Burnyeat (1992); on the other side, see Cohen (1992), Wilkes (1992), and other papers
in the same volume.
11 Ackrill (1973), 133. Ackrill (1981), 76-77, offers a quite different account of body
and soul.
12 See Gill (1989), 96-97, 128-30, 149-63, esp. the summary on 163.
91
a view fits well with the account of separation I want to give. For, as
we have seen, in VII 3 Aristotle rejects matter as a candidate for substance on the grounds that it is neither tode ti nor chbriston. One might
suppose that by 'not separate' he means that it is not numerically distinct
from the composite, and On Generation and Corruption I 5 ("It is therefore better to suppose that in all instances of coming-to-be the matter
is inseparable, being numerically identical and one, though not one in
definition" (320bl2-14; cf. 320b22-25)) seems to suggest as much. But
Dancy says of Aristotle: "when he talks about separability he need not
be, and generally is not, thinking about questions of individuation, divided reference, or countability, about whether or not stuff can occur
in the absence of 'things'."13 Rather, according to Dancy, what is inseparable about matter is that, as matter, there is nothing that it is "in
its own right."14 As the potential for change in something, matter does
not "stand on its own";15 in other words, there is nothing that it is. But
if not having an essence is the ground for the inseparability of matter
as matter, then the sort of separation that matter is said to lack, the
sort of separation that is had by substance, would seem again to be the
ontological correlate of separation in definition.
Ill
In the first section of this chapter I argued that the separation of substances, which I have taken to be specimens of natural kinds, could be
understood as independence in being, a sort of separation not had by
attributes, matter, and (except in relation to accidents) the composite.
If this account of separation is right, the dispute between Plato and
Aristotle is not over whether what is ontologically fundamental is, in
some sense, separate, but over the implications of requiring separation
in the sense of independent being of that which is ontologically fundamental. In the case of the Forms, Aristotle tells us that Socrates sought
definitions but that it was Plato who was responsible for separation. Of
course, from the Socratic dialogues we learn that Socrates regularly
rejected definition by example, and in the Euthyphro he is even made
to talk of what he is seeking not just as what is common to all cases but
as that by which pious things are pious and that to which we can look as
a standard (6d-e). Nevertheless, even if we assume that Socrates made
claims such as these, it seems not implausible to believe that he left it
unclear what he meant by them;16 that it was Plato, not the historical
13 Dancy (1978), 401.
14 Dancy (1978), 407.
15 Dancy (1978), 400.
16 Thus I disagree with Allen (1970), 133-36, 147, 149. That Socrates did not separate
the forms might, he says, mean that Socrates did not distinguish forms from their
sensible instances or that, although he distinguished them, he did not regard them
as individuals or even that though he took them to be individuals he did not regard
them as independent of and prior to their instances. What Allen argues is that forms
were for Socrates separate in all these ways and that Socrates differed from Plato only
in that separation did not for Socrates make sensibles deficient.
92
Socrates, who attempted to clarify the characteristics had by a standard

just in virtue of its being the standard seems entirely credible. Specifically, by holding that Forms must be eternal, unchanging, intelligible
paradigms, it seems that Plato does contend that the existence of something in addition to - something numerically distinct from - what is in
flux is necessary in order to meet the Socratic requirements.
Thus I accept Aristotle's account according to which it is Plato who
separated the Forms, and I have argued that what is meant by this claim
is that he postulated eternal paradigms, numerically distinct from sensibles. However, to separate the Forms in this way is, Aristotle argues,
a fundamental error. That is to say, Plato's error, according to Aristotle,
is to think that, in order for there to be something knowable and unchangeable in the midst of flux, there must be something additional,
something numerically distinct from sensibles, something that exists
besides (para) particulars (1086b8) and that given its nature could exist
even if they did not do so. Indeed as I understand Aristotle, one might
say that Aristotle returns to a more Socratic view of separation. Yet to
say this does not do justice to Aristotle's thought, as he does not and
could not return to Socrates' unconcern about the ontological question.
Rather Aristotle must argue that, while being ontologically and epistemologically fundamental does require independent being, entities such
as what is horse can have independent being without numerical distinctness or independent existence from sensible objects, and I believe
that the central books of the Metaphysics contain such an argument.
To summarize, Aristotle believes there is a legitimate and an illegitimate sense in which substances might be held to be separate. What he
thinks separation needs to be, so I have argued, is independent being
- the ontological correlate of separation in definition. But Plato believes
that this weak sense of separation implies a stronger sense, namely,
numerical distinctness (a distinctness that, in the case of Plato's Forms,
is coupled with a capacity for independent existence) while Aristotle
denies the implication.
Thus on my interpretation Aristotle assigns to Plato not the blatant
error of arguing from flux to independent existence but rather what
he takes to be the more subtle error of supposing that, in order for
something to be different from what is in flux, it must be numerically
distinct from what is in flux. Another virtue of my view is that it allows
Aristotle's own understanding of the separation of substances to stand
in direct competition with Plato's. That is to say, where Plato asserts
numerical distinctness, Aristotle denies it. Yet, for both Plato and Aristotle, how there is to be something real and knowable in a world in
which what we experience is in flux must be explained. That is to say,
for Aristotle too there must be separation between the real and the
knowable, on the one hand, and what is in flux, on the other. The
difference between Plato and Aristotle on this point, therefore, comes
93
to this: that Aristotle believes that the separation of what is real and
knowable from flux can be achieved without its numerical distinctness
from what is in flux.
IV
My interpretation of separation in Aristotle differs from Fine's with
regard to the sort of separation that Aristotle attributes to Plato and
also the sort that is had by substances. My view also stands in sharp
contrast to the interpretation of the separation of Aristotelian substances offered by Donald Morrison. Morrison argues that in Aristotle
separation should be understood in terms of being outside the "ontological boundaries" of something else - or, equivalently, according
to Morrison, as numerical distinctness from another thing.17 But if a
thing is separate from that which is outside its ontological boundaries,
it is separate from that in relation to which it does not have ontological
influence. If this is so, then, as Morrison says, what substances are
separate from is other substances; they are not, that is to say, separate
from the accidents that inhere in them.18
On Morrison's view, therefore, separation as it applies within Aristotle's ontology is what I have called a horizontal relation; it is the
separation, as it were, of cats and dogs, and indeed of one dog from
another. Moreover, because a substance is naturally prior to its accidents
but is not separate from them while it is on the other hand separate
from every other substance but not prior or posterior to them, the
connection for Morrison between separation and natural priority is
more accurately a connection between natural priority and nonseparation. On my view, by contrast, since substances are said to be separate
in the sense of having independence in being from matter and accidents
and the composite, it is easy to see why, for Aristotle as for Plato, being
separate and being prior should be linked. Indeed, as I said at the close
of Section III, my interpretation makes Aristotle's view satisfyingly parallel to Plato's. For Aristotle recognizes that substances do need to be
separate from sensible flux, and he disagrees with Plato not about the
need for separation but about its nature.
Thus my account, unlike Morrison's, preserves the link between separation and priority as well as making evident the connection between
the Platonism Aristotle deplores and the Platonic requirements he
nevertheless endorses. There is also another difference. For it follows
from his account of the nature of separation that, as Morrison says,
17 Morrison (1985), 128, 138-39.
18 See Morrison (1985), 139-44. It should be noticed that by 'substance' Morrison at

this point in his exposition has in mind the composite, even though he says that the
other criteria for substance point in the direction of form. His claim, for reasons that
will become apparent below, is that the use of separation as a criterion for substance
"leads to significant strains within the theory" (126).
94
"[Aristotle's] claim that primary substance, form, is separate because it

is separate in definition, is a philosophical dodge that borders on being
a cheat." 19 He explains:
The situation Aristotle has gotten himself into is this: he has argued himself
into the view that the form is primary substance. The form is not separate, yet
a part of his goal is to show that whatever is substance satisfies all of his initial
criteria of substance. Ready to hand is the criterion 'unity of definition', another
name for which is 'separation in definition'. This latter criterion, form satisfies.
Thus by appealing to the notion 'separation in definition', Aristotle is able to
say that the form is separate, after all. However, if I have been right about the
way separation was supposed to function as a criterion of substance, Aristotle's
victory is only verbal. 'Separation' has become ambiguous, and form has been
shown to be 'separate' in a sense quite different from the original sense in which
substance was held to be separate.20
As Morrison understands Aristotle, form, being prior in definition,
has in this regard a better claim than the composite to be primary
substance, yet because it is in matter, form lacks the ontological boundaries in terms of which separation is to be understood. The result is
that Aristotle is forced, so Morrison thinks, to attribute to form a different sort of separation, separation in definition, which, not being the
real (ontological) separation characteristic of the composite, is a kind
of cheat. Though not everyone is so blunt as to call it a cheat, certainly
Morrison is not alone in his conclusion about the sort of separation
had by form. Guthrie says:
Yet here in the Metaphysics too, where essence becomes primary substance, the
requirement [of separate existence] is maintained, and in [Z] ch. 6 he argues
that a thing and its essence are the same, the essence being the substance of
the thing. We cannot get out of it that way. Lack of separability, after all, was
the main reason for rejecting the claim of matter to be substance. In what way
does essence possess it? It does so by being 'conceptually separate', separate in
thought or by definition. The difference is brought out at Metaphysics H
1042a26-31: "The substratum is substance, i.e. in one sense matter, potentially
but not actually a 'this', in another the logos or form, which as a 'this' can be
separated conceptually, and thirdly the product of the two, which alone undergoes generation and perishing, and is separate without qualification." The difference [from matter] is real. As essence is substance understood as the object
of scientific knowledge, so it counts as separate because, being intelligible and
definable, it can be abstracted mentally and thought of by itself.21
19 Morrison (1985), 154.
20 Morrison (1985), 155.
21 Guthrie (1981), 219-20. Code (1991), 6, sees the same problem but holds that for
Aristotle only god is a separately existing form. He says: "However, the substantial
forms of perceptible things, despite the fact that they can be denned without reference
to the matter in which they are instantiated (and hence are dejinitionally separate),
nonetheless cannot exist without their matter. Thus, contrary to the Platonist position,
the forms of perceptible things are not capable of separable existence." On my interpretation there is an important sense in which substances in the biological world
are separate and therefore truly substances after all.
95
Loux takes a similar view. He says:

But the cluster of themes surrounding the idea that the composite is separate
and subject to the "this something" schema is simply inapplicable in the case
of form. Substantial forms are not particulars and so cannot be individual subjects for the metaphysical predication of other things; unlike substance-kinds,
they are not predicated essentially of anything but themselves. A form is how
its matter is, not what it is; consequently, it does not exhibit the sortal logic of
substance-kinds. It does, however, have a determinate conceptual content, and
that content can be identified without reference to anything else. Accordingly,
we can say that form is separable in formula.22
A few lines later Loux explains why he believes that Aristotle could
find this sufficient:
The idea that form is separable in formula is the familiar one that, if it is to be
the primary reality, form must be a fundamental essence, one that can be defined
without reference to other essences. And the idea that form is "this something"
is tied up with its status as the primary ousia of things to which instances of the
"this something" schema in its primary sense are applicable. But, then, Aristotle's insistence that both form and composite be separable and subject to the
"this something" schema brings out the two dimensions of priority underlying
the use of the term 'ousia' in the two cases. Form has the priority in essence
its role in ousia explanations requires; composite particulars have the kind of
factual or existential priority that makes them the central point of focus in our
everyday commerce with the world.23
Thus Loux claims that the separation had by form is not, as Morrison
thinks, a cheat, but "an essential component in the theory of ousia
explanations and in the account of essences as constituting a structured
framework with a secure foundation."24 On my view, on the other hand,
the separation of form is not conceptual abstraction, but an ontological
notion, and it is one that allows Aristotle to deny the divergence between
"priority in essence" and "existential priority" that on all these accounts
divides the intelligible from the real. If what separation is for Aristotle
is the ontological correlate of separation in definition, it is clear that
form rather than the composite is what has it most of all. Therefore,
if separation is the ontological correlate of separation in definition,
Aristotle can distinguish what he believes to be his successful theory of
substance from what he takes to be Plato's unsuccessful theory.
I have argued that, when Aristotle criticizes Plato for separating the
Forms, he is arguing that Plato makes them numerically distinct from
22 Loux (1991), 263.
24 Loux (1991), 262, n. 28.
23 Loux (1991), 263-64.
96
sensibles, a move that, Aristotle claims, ontological and epistemological

priority do not require. There can be separation from flux without
numerical distinctness from what is in flux; for independence in being
- the ontological correlate of definitional separation - it is enough to
deny that there is identity between substance or form and sensible objects. Nevertheless there is an objection to my view which must be
addressed. It concerns passages in Physics II (193b4-5) and Metaphysics
VIII 1 where Aristotle says of the form that it is separate in definition
(or only in definition). This problem is most acute in VIII 1 where
Aristotle says not just that form is separate in definition but that the
composite is separate without qualification (choriston haplos) (1042a2831). In addition there are passages, mostly in De Anima, where Aristotle
asks, as if the dichotomy were exhaustive, whether the parts of the soul
are separate in definition or spatially separate {De Anima 42galO-b22,
413bl3-32, 432al9-b7); though in the last of these cases even the
question of separation in definition is quickly abandoned, in the others
Aristotle argues that only separation in definition applies, except, in all
likelihood, to that part which is capable of knowledge of the eternal.
Whatever Aristotle wants to say about the parts of the soul, nevertheless it seems clear that in general the dichotomy between separation
in definition, on the one hand, and, on the other hand, spatial separation
is not meant to be exhaustive; moreover, even in De Anima, there may
be another way to interpret separation in definition - that is to say, as
the ontological correlate to it. Indeed two passages from De Anima and
the Nicomachean Ethics are suggestive. In the first Aristotle says:
that which is the instrument in the production of movement is to be found
where a beginning and an end coincide as e.g. in a ball and socket joint; for
there the convex and the concave sides are respectively an end and a beginning
(that is why while the one remains at rest, the other is moved): they are separate
in definition but not separable spatially. (433b21-25)
Again at Nicomachean Ethics I 1102a26-32 he argues:

Some things are said about it, adequately enough, even in the discussions outside
our school, and we must use these; e.g. that one element in the soul is irrational
and one has a rational principle. Whether these are separated as the parts of
the body or of anything divisible are, or are distinct by definition but by nature
inseparable, like convex and concave in the circumference of a circle, does not
affect the present question.
The convex and the concave, which in the Ethics are compared to the
relation between the parts of the soul, are like the road up and the
road down of Physics III. That is to say, because the convex and the
concave surfaces of the joint can be said to be an end and a beginning,
at rest and moving, there are contexts in which 'the convex' and 'the
concave' will not be substitutable for each other, a state of affairs that
97
is characteristically said by Aristotle to be indicative of a difference in

being.
I have argued that "separation in definition" does not exclude its
ontological correlate; indeed, insofar as Aristotle is serious about an
ontology of specimens of kinds that includes even accidental unities, it
may actually be implied. There remains, however, the contrast in Metaphysics VIII 1 between separation in definition and separation haplos,
said to be true of the composite. What can be said first of all with regard
to VIII 1 is that it is in any case an anomalous passage as nowhere else
does Aristotle say 'chbriston haplos\ Moreover, it is to be noticed that
although Aristotle seems clearly in VIII 1 to be contrasting kinds of
separation, it is still the form, not the composite, which is said to be
tode ti, a characteristic of substance given along with separation in Metaphysics VII 1.
To be sure, there are elsewhere in Aristotle plenty of cases where

'haplbs* does mean what is most truly or properly X. There is, for example, simple incontinence as opposed to incontinence by analogy
(1149al-3); essence and definition belong simply to substance and to
the other categories in virtue of their relation to substance (1030b47). It is also true that there are cases where 'haplos' is contrasted with
the relative or the conditional - as in the good simply versus the good
for a particular person (1152b26-27). Nevertheless it is clear that 'haplos* is sometimes used in ways that are not so obviously evaluative. A
word such as 'earth' having no parts that signify (1457a31-32), is simple,
not compound; in similes (1412b35-1413a4), the simple is opposed to
the complex. Thus if separation haplos likewise is not, with respect to
the question "What is substance?" a preferred sort of separation, then
even in VIII 1, Aristotle's remarks about separation are compatible with
my view that the separation characteristic of the sublunar substances
of the central books of the Metaphysics is independence in being.
VI
Thus if Aristotle argues against Plato that there can be separation in
being without numerical distinctness, it can be true that substance is
form, despite the fact that Aristotelian forms - specimens of natural
kinds - are entities that are not numerically distinct from sensible objects. Moreover the plausibility of understanding the separation of substance as independent being, that is to say, as the ontological correlate
of separation in definition, in turn offers support for my interpretation
of substance as specimens of natural kinds; the views fit together.
But if the existence of specimens of kinds depends on the existence
of sensible objects, as I understand Aristotle's alternative to Plato to
assert, Aristotle's further claim that it is specimens of natural kinds and
98
not sensible objects that are ontologically fundamental may seem, at

best, arbitrary or contrived. After all, in the second chapter I argued
that there are not two distinct things in the universe, Socrates and
Socrates-qua-human being, and if Socrates-qua-human being has to be
a fictitious or psychological entity as the common rendering of 'qua' as
"considered as" suggests, the claim that Socrates-qua-human being is
a substance would not be at all credible.
In the next chapter I will argue that in fact there are in Aristotle the
materials for a defense against such charges. At this juncture, however,
my point is rather to emphasize the need for a defense. For Aristotle
must deny that his account of substance is inherently psychologistic or
linguistic; it is not open to him to adopt the sort of view taken in the
philosophy of science by Frederick Suppe, for example:
Scientific theories have as their subject matter a class of phenomena known as
the intended scope of the theory. The task of a theory is to present a generalized
description of the phenomena within that intended scope which will enable one
to answer a variety of questions about the phenomena and their underlying
mechanisms; these questions typically include requests for predictions, explanations, and descriptions of the phenomena. The theory does not attempt to
describe all aspects of the phenomena in its intended scope; rather it abstracts
certain parameters from the phenomena and attempts to describe the phenomena in terms of just these abstracted parameters. . . . As such the theory assumes
that the phenomena are isolated systems under the influence of just the selected
parameters.25
Although Suppe, like Aristotle, takes science to provide descriptions,

explanations, and predictions, that is to say, to provide knowledge, they
must differ over the relative ontological status of the objects of knowledge and the objects of experience. For Suppe, since actual phenomena
are rarely in fact isolated systems, theories do not give an accurate
characterization of actual phenomena. As he says, "what the theory
actually characterizes is not the phenomena in its intended scope, but
rather idealized replicas of those phenomena." 26
Clearly Suppe does not intend that his replicas be Platonic Forms.
But whatever they are - mental objects or fictional entities - the important point is that Aristotle's substances cannot be such. To suppose
that Aristotle thinks that accidental unities such as the musical one (Socrates-qua-musician) are fictitious entities might seem only charitable,
though even in this case, as we have seen, Aristotle's remarks about
coming to be and passing away would suggest that the supposition may
be wrong. But whatever exactly one takes Aristotle to hold about the
ontological status of accidental unities, I have argued that specimens
of natural kinds are substances, and that mental or fictional entities
25 Suppe (1974), 223.
26 Suppe (1974), 224.
99
would not be substances is clear. Indeed that Aristotle's view is far from
Suppe's is very evident if a claim of Owen's is correct. For what Owen
says is that in Aristotle it is a man or a plant that is a prime example
of substance, where a man and a plant "carry no namely-rider such as
Viz. Socrates or Callias or . . ,',"27 and this proposal is, I think, much
the same as the position which I have defended; Socrates-qua-human
being and Callias-qua-human being, being indistinguishable, might very
well be described as x-qua-human being - entities that are, that is to
say, without a "namely-rider."
To be sure, Owen's rendering appears no less paradoxical than mine;
that 'a man' means 'some man' and that 'some man', if taken to refer
at all, must refer to Callias or to Socrates or to whomever seems as
certain as that 'Socrates-qua-human being' refers to Socrates. And, of
course, in a way it does. Nevertheless Aristotle, unlike Suppe, has reason
to resist the impulse toward ontological reduction. What I intend
'Socrates-qua-human being' to exhibit is not that specimens of kinds
are psychological entities, in other words, ways of thinking about things
and thus idealizations or abstractions from them, but rather that specimens of kinds are not Platonic entities, existing in addition to sensible
objects.
Of course it is one thing to say that in Aristotle's ontology it is entities
such as 'a human being' which are substances, that these are numerically
the same as but not identical with sensible objects, and indeed that they
are ontologically more fundamental than the latter; it is another thing
to show that such a view is defensible or even intelligible. How can
there be numerical sameness without identity? Then too, even if one
grants that there are natural kinds, how can Aristotle take specimens
of these kinds to have ontological priority over the sensible objects with
which they are numerically the same? In the next chapter I will begin
by presenting a contemporary example of numerical sameness without
identity which is also suggestive for addressing the second of these
questions.
27 Owen (1978), 20. Owen's suggestion is endorsed by Lennox (1985), 83, in a paper
written from the perspective of the biological works.
VI
Substance and Teleology
In this book I have argued that the assumption that Aristotle distinguishes numerical sameness from identity provides a wide-ranging explanation of referential opacity in his works and makes possible an
interpretation of substance that sees Aristotle's theory as a response to
what he takes to be the flaws in Platonism. I have not attempted to
defend distinguishing between numerical sameness and identity on philosophical grounds or even to consider the philosophical implications
of such a view; as I said in Chapter II, the logic of a metaphysics that
confounds counting has to be, to say the least, problematic. It may be,
of course, that Aristotle adopted a position that cannot be made coherent or attractive, although such a conclusion would be disappointing.
Although I will not in this final chapter try to offer a philosophical
analysis or defense of the distinction, I will nevertheless describe an
interesting occurrence of it in the recent philosophical literature. But
the primary goal of this chapter is to argue that substances, understood
as specimens of natural kinds, can defensibly be said to be ontologically
prior to the sensible objects with which they are numerically the same,
and for that argument too the example now to be offered will prove
useful.
I
In an extraordinarily interesting book in the philosophy of art, The
Transfiguration of the Commonplace, Arthur Danto puzzles about the re-
lation between works of art and physical objects, putting the problem
into its sharpest focus by considering works of art that are perceptually
indistinguishable from ordinary things. To take Danto's example, Duchamp's Fountain is perceptually indistinguishable from a urinal. Following Danto, let us call the sensible object, in this case a given urinal,
100
SUBSTANCE AND TELEOLOGY
101
the "material counterpart" to the work of art.1 The question then is

this: What is the relation of a work of art to its material counterpart?
In the case of Duchamp's Fountain, Danto says:
But the question is whether the artwork Fountain is indeed identical with that
urinal, and hence whether those gleaming surfaces and deep reflections [properties of porcelain] are indeed qualities of the artwork. . . . But certainly the
work itself has properties that urinals themselves lack: it is daring, impudent,
irreverent, witty, and clever.2
What "detaches objects from the real world" and makes them part of
the art world, according to Danto, is that art works belong to a world
of interpreted things or representations, and the properties of a work
of art are just those that contribute to its interpretation. 3 For example,
whether gleams, a property of the porcelain, is a property of Fountain
would depend upon its relevance to Fountain's representational content;
even whether Fountain has a color is, according to Danto, to be decided
on the same grounds.4
In the present context it is not necessary to consider what properties
are lost and gained in the "transfiguration" of mere things into art or
even whether Danto's theory of art is correct. What is interesting to
me is that Danto distinguishes the properties of Fountain from the properties of its material counterpart. Fountain is witty, daring, and irreverent, but the urinal is not; the urinal is white, but Fountain may or
may not be so. It must be emphasized that such differences will arise
regardless of whether the material counterparts are everyday objects
such as a urinal or (another of his examples) a snow shovel5 - that is,
objects that have a use outside art - or simply painted canvases and
chunks of marble. Examples of the former sort contribute a certain
vividness to Danto's presentation of his theory, but that is all.
If a work of art and its material counterpart differ in properties, the
relation between them - a relation that Danto dubs the 'is' of artistic
identification6 - is not identity, given that Leibniz's Law is taken to
define that relation. Whether one describes it as a nonstandard identity
relation of some sort or whether one says that there are relations of
numerical sameness other than identity is perhaps not a matter of extreme importance. However, in specifying the extent to which for Danto
a work of art and its material counterpart differ, it is important to notice
that he has taken a stance that is more radical than the nonstandard
1 Danto (1981), 104. See also 99.
2 Danto (1981), 93-94.
3 Danto (1981), 135; see also 125. Danto argues that art works are distinguished from
other representations (for example, maps) by "[their use of] the means of representation
in a way that is not exhaustively specified when one has exhaustively specified what is
being represented" but which, in addition, "expresses something about [their] content"
(148).
4 Danto (1986), 38.
5 Danto (1986), 26.
6 Danto (1981), 126.
102
identity advocated by Allan Gibbard, for example, who uses the same
sort of case to argue that contingent identity can obtain between the
referents of proper names. What Gibbard maintains is that Goliath (a
statue) and Lumpl (a piece of clay), which come into existence simultaneously (by sticking two halves of the statue together) and cease to
exist simultaneously (by being smashed), are identical. Nevertheless,
Gibbard says, the identity is contingent. This is so because, although
the objects in question in fact begin and come to an end simultaneously,
had, for example, the wet clay once stuck together been squeezed into
a ball, the times at which the statue and the piece of clay ceased to exist
would have been different. That is to say, because Goliath could have
been destroyed while Lumpl survived, Goliath and Lumpl differ in
modal properties.7
Certainly Danto and Gibbard do not disagree merely about how the
relationship between Goliath and Lumpl, in the circumstances Gibbard
has described, is to be labeled. Rather what is evident is that Danto
would take issue with Gibbard's description of the case. Although Gibbard has said that Goliath and Lumpl have all the same actual properties,
for Danto, as we have seen, only Goliath would have representational
properties, and, furthermore, Goliath would have only representational
properties. On the other hand, in describing the relation between the
work of art and its material counterpart, Danto is no Platonist.8 Thus
since he does not take the work of art and its material counterpart to
be two distinct things and yet he denies their identity with each other,
Danto has offered a contemporary example of numerical sameness without identity.
The comparison between Danto and Aristotle will bear pressing. But
first it is useful to notice that, despite the term 'material counterpart',
if matter in Aristotle is understood as "in its own right" lacking properties, Danto's material counterpart is not, in Aristotle's sense, matter;
the material counterpart, so Danto tells us, has properties that are not
properties of the form and are unrelated to its potentiality for form.
On the other hand, there seems to be no reason why the material
counterpart should not be said to be a sensible object. In the case of
the urinal and the snow shovel, that it is so is obvious. To be sure, the
material counterpart of a painting is not assignable to any ordinarily
recognized kind since were it said to be a painting, it would have the
representational properties Danto denies to it. But even if the fact that
painting is a kind dependent on certain practices that endow its members
with (indeed on Danto's view with only) a very unusual sort of properties,
namely, representational ones, means that what the painted canvas must
be said to be is just a painted canvas and even if, as a result, the material
7 Gibbard (1975), 190-91.
8 See Danto (1981), 33-34, 153.
103
counterpart does not quite correspond to what in Aristotle's ontology

is called the composite, that it is a sensible object is still true and sufficient for my purposes.
II
Danto's 'is' of artistic identification is a contemporary example of the
usefulness of numerical sameness without identity. What makes comparison with Danto especially attractive, however, is that the relation
that Danto believes to hold between works of art and their material
counterparts is suggestive also of arguments to which Aristotle can appeal to establish the ontological priority of specimens of natural kinds.
Suppose, for example, that it is true of a painted canvas that it is 2272
X 29 inches. In such a case, the canvas is a specimen of the kind thing
that is 22lA X 29 inches, just as, given that it has certain other properties,
it is a specimen of the kind (indeed the only specimen of the kind) La
chambre de Wan Gogh a Aries. I do not, of course, claim that thing that
is 22lA X 29 inches is a natural kind. The point is rather quite the

opposite, namely, that at least a partial explanation of why kinds are
not all of the same status can now be given. For what can be seen as
significant is that the properties of a specimen of the kind thing that is
22V2 X 29 inches do not determine as many of the properties of the
sensible object with which it is numerically the same as does being a
specimen of the kind La chambre de Van Gogh a Aries. Indeed it would
seem to be because it is a specimen of the kind La chambre de Van Gogh
a Aries that the sensible object is 22^2 X 29 inches and thus a specimen
of the kind thing that is 22V2 X 29 inches, and not conversely.
Of course the reason a specimen of the kind La chambre de Van Gogh
a Aries is prior to the sensible object as well as to the specimens of
whatever other kinds that same sensible object is a specimen of is that
Van Gogh worked on the canvas as he did in order to bring La chambre
de Van Gogh a Aries into being. Indeed it is sometimes said that in the
best works of art the artist could not have changed anything - not a
single word or note or brushstroke - without diminishing the work. No
doubt this is an exaggeration. Nevertheless to say that it is as a byproduct of the artistic process that the material counterpart is, say, a
specimen of the kind thing that is 22lA X 29 inches is not unreasonable.
Indeed even many properties that are not entirely determined by the
artist's actions, for example, that a given canvas is now in the Musee
d'Orsay, can be seen to be consequent upon them. Since its being a
specimen of the kind thing in the Musee d'Orsay is not what was aimed
at by Van Gogh in the course of creation but is merely made possible
by what he did then, the specimen of the kind thing in the Musee d'Orsay
104
which is numerically the same as a given painted canvas also is not

ontologically prior to it.
I have argued for the priority of a work of art over its material
counterpart on the grounds that properties of the work of art govern
many of the properties of the material counterpart (and therefore many
of the other kinds to which the material counterpart belongs) in virtue
of the artist's intention. Of course one might object that the painting
exists only because the canvas was painted; indeed, some reductionistminded philosopher might want to say that a painting just is a piece of
painted canvas. Still, it seems to me that there is a case to be made for
having it the other way around. The painted canvas, one can maintain,
exists just because the painting does so; after all, it was because a painter
wanted a certain artistic result that he put these colors on a canvas of
this size and that he put them on it in this way.
As the grounds for the priority of specimens of one kind over another
will become important for my interpretation of Aristotle, I want to
restate what has been claimed. If Socrates is a builder, Socrates is a
specimen of the kind builder as well as a specimen of the kind human
being. About the consequences of this fact, some philosophers might
be happily egalitarian. H.-N. Castaneda, for example, holds that sensible
objects are bundles of guises,9 a guise being, like a specimen of a kind,
a particular having nothing but the properties specified in the definition
(for example, 'the round blue thing on the table' refers to the guise
"the thing which alone has nothing but the following properties: being
round, being blue, being on the table").10 For Aristotle, however, neither builder nor round blue sitting on a table thing would be a natural
kind. Thus if for Aristotle sensible objects can be said to be composed

from specimens of kinds, they must be said to be composed not from
specimens of just any kinds, but from specimens of natural kinds - that
is to say, from specimens that are ontologically prior to the sensible
objects in question just because they are specimens of kinds that are
natural. Specimens of these kinds are, I have claimed, substances.
Thus human being is for Aristotle a natural kind and builder is not.
Of course there is nothing about the relation of numerical sameness
without identity as such that justifies this claim; indeed the 'is' that
occurs in saying that Socrates is a specimen of this or that kind does
not necessitate any distinction between natural kinds and others, much
less tell us by what criteria the naturalness of any given kind is to be
assessed. Nor can the relation even show that a given sensible object
cannot constitute a specimen of more than one natural kind. To turn
once more to the analogy with art, by substituting the viewer's understanding for the artist's intention as what is determinative of the prop9 Castaneda (1977), 322.
10 Castaneda (1977), 315.
105
erties of a work of art, one could say that many different works of art
are artistically identified with the same material counterpart, as many
as there are interpretations. Nevertheless, even though the 'is' of artistic
identification does not mandate any distinction concerning priority and
does not make the selection, as we have seen, it does provide some
grounds for doing so. In particular, because the properties of the work
of art and the properties of its material counterpart are related so as
to make plausible the claim that certain properties of the one are explanatory of and indeed determinative of certain properties of the other
in light of the artist's intention, where art is concerned it is reasonable
to say that specimens of the kinds whose defining properties are of the
first sort are ontologically prior.
In this section the relation between a work of art and its material
counterpart has been used to explore the relation of being numerically
the same as but not identical with, a relation that I believe Aristotle
takes to hold between specimens of kinds and sensible objects, and to
consider the possibility that, given such a relation, one might be justified
in holding that ontological priority lies with specimens of certain kinds.
What I have said about the maker's intention determining the properties
of a work of art could be said of any more ordinary artefact, of course;
the reason for using art as my example has been only that Danto in his
discussion of that subject actually proposes a relation of numerical sameness without identity.
To be sure, there are limits to the analogy - even if a work of art
or a specimen of some artefact-kind can be prior to the sensible object
that is numerically the same as it, works of art and specimens of artefactkinds are hardly ontologically prior simpliciter. Nevertheless, the implications are clear. If Aristotle could claim that some properties, more
precisely, the essential properties, of sensible objects are determined
by the properties of specimens of certain kinds, he could reasonably
maintain that specimens of those kinds are ontologically prior to the
sensible objects with which they are numerically the same and also to
specimens of other kinds, in other words, to accidental unities. In particular, if such an argument were successful, Aristotle might be able to
conclude, for example, that specimens of the kind human being are and
that specimens of the kind builder are not ontologically prior to individuals such as Callias and Socrates.11 But, of course, on Aristotle's
account, nature is rather like a work of art - or at least like an artefact
11 Though his account of substance is different from mine, Irwin (1988) too makes a
case for the ontological priority of some subjects over others. He says: "If we see the
connexion between potentialities and persisting subjects of change, we can also see
why some subjects (e.g. men) are more genuine subjects than others (e.g. musical
men). A persistent subject needs a persistent potentiality explaining the changes that
happen to it. If we find that rather little is explained by reference to permanent
potentialities of musical men as such, and more is explained by reference to permanent
106
- in being itself teleological, even if it is teleology without agency. In

the remainder of this chapter therefore, I want to explore Aristotle's
account of teleological explanation, arguing that Aristotle sees teleology
as justifying the ontological priority of specimens of natural kinds and
that, on a certain view of goodness, teleology can in fact do so, even
in the absence of agency.
Ill
For Aristotle there are final causes for the parts or organs of living
things, the behavior of organisms, and also the development of the
organism from a seed or embryo to a mature functioning member of
its kind. But even though Aristotle's reliance on teleological explanation
in nature is explicit and defended, his use of it presents the interpreter
with several fundamental problems. One is to determine what is meant
by a final cause since teleology in Aristotle, like teleological explanation
in modern biology, has been analyzed in many ways. A second problem
is to determine the relation for Aristotle between final and material
causes, and a third concerns the soundness of whatever arguments Aristotle offers in defense of his use of final causes. These issues and their
relevance for the ontological primacy of specimens of natural kinds will
be discussed in this and the sections which follow.
In his summary of teleological explanation at Parts of Animals I 1
640a33-b4, Aristotle says:
The fittest mode, then, of treatment is to say, a man has such and such parts,
because the essence of man is such and such, and because they are necessary con-
ditions of his existence, or, if we cannot quite say this then the next thing to
it, namely, that it is either quite impossible for a man to exist without them,
or, at any rate, that it is good that they should be there. And this follows: because
man is such and such the process of his development is necessarily such as it is; and
therefore this part is formed first, that next; and after a like fashion should we
explain the generation of all other works of nature, (emphasis added)
As this passage and many others make clear, the presence of teleological claims in Aristotle's writings is undeniable. However, the analysis
potentialities of men as such, we have some reason for thinking that men are more
genuine subjects than musical men are. Aristotle claims in Metaphysics vii 4-6 that
some, but not all, apparent subjects are thises and primary subjects, identical to their
essences. If the persistence of explanatory potentialities determines the basic subjects

and essences, an account of potentialities should help us to see which essences and
which subjects satisfy his conditions. For these reasons we expect the essence of a
persistent subject to be at least partly constituted by its persistent potentialities for
change. If, then, the primary subjects that are substances are identical to their essences,
they should be at least partly identical to their persistent potentialities. Moreover, if
form is to be identified with essence and subject, it should be some sort of potentiality"
(224).
107
and justification of teleological claims in biology is a subject still in

dispute. The difficulty is that teleological claims appear to ascribe intentional states where none seem to be present. If John goes to the
store in order to get bread, presumably John desires bread and believes
that the store is a place from which to obtain it. If the spider is said to
spin a web in order to catch flies, does the spider desire flies and believe
that web spinning will promote its coming to have them? If a hammer
is for pounding nails, presumably someone designed it with that in mind,
but if the heart is for pumping the blood, can the same be said?
There have been some scholars, of course, who have thought that
Aristotle's views about teleology are to be explained by an appeal to a
valuing mind, in which case teleology in nature is not very unlike the
purposiveness of human actions and the functions of artefacts. On this
view processes in nature are supposed to be guided, if not by intentions,
at least by unconscious desires. For example, the developing embryo
might be thought to be an unconscious agent that guides the development of its matter toward maturity. R. G. Collingwood exemplifies
this tradition when he says:
The seed only grows at all because it is working at becoming a plant; hence the
form of a plant is the cause not only of its growing in that way but of its growing
at all, and is therefore the efficient as well as the final cause of its growth. The
seed grows only because it wants to become a plant. It desires to embody in
itself, in material shape, the form of a plant which otherwise has a merely ideal
or immaterial existence. We can use these words 'want' or 'desire' because
although the plant has no intellect or mind and cannot conceive the form in
question it has a soul or psyche and therefore has wants or desires, although it
does not know what it wants.12
Yet literally internalist interpretations of Aristotle - that is to say,
analyses of teleology which rely on the ascription of intention - such
as that given by Collingwood, have fallen into disfavor in recent years.
Perhaps in part this change has been motivated by the thought that
Collingwood's view would make Aristotle's teleology less interesting to
us, but it is also true that on the whole Aristotle, like most contemporary
biologists, seems to want to avoid speaking of nature's striving or desiring. When Aristotle says that nature is like shipbuilding but like shipbuilding without a shipbuilder (199b28-30), he seems to be doing something more radical than just considering the shipbuilder as part of the
ship.
Thus it seems likely that Aristotle's teleological explanations do not
depend on an appeal to agency. Of course even some philosophers who
reject a literally internalist notion of teleology in nature have held that
teleological claims occur as a result of metaphorical associations with
12 Collingwood (1945), 84-85.
108
human activities and artefacts. But the view that teleological claims are
metaphorical is often also the starting point for unpacking and justifying
their explanatory worth in terms of features other than the presence
of a mind. Working from the assumption that for Aristotle intention
or mind is to be excluded from the account, in the remainder of this
section I will consider one such externalist account, 13 that of Larry
Wright, in order to see how it might be adapted to Aristotle's needs.
Wright, who takes himself to be giving the content of what he says
is a dead metaphor, analyzes goal statements as follows:
S does B for the sake of G iff
1) B tends to bring about G
2) B occurs because (i.e., is brought about by the fact that) it tends to
bring about G.14
Thus, for example, spiders spin webs for the sake of catching flies if
and only if spinning webs tends to bring about the catching of flies and
spinning webs occurs because it tends to bring about the catching of
flies. Function statements are analyzed in similar fashion:
The function of X is Z iff:
1) Z is a consequence (result) of X's being there
2) X is there because it does (results in) Z.15
But Aristotle, as we have seen, also thinks of organic development teleologically; still for such cases it seems that Wright's schema could be
extended in the following way:
E exists for the sake of O iff:
1) E tends to develop into O
2) E exists because it tends to develop into O.
Wright's analysis, according to which "X is there because it results in
Z," sounds not unlike Aristotle's claim that the fittest mode of explanation is to say that such and such parts are necessary for an organism's
existence. Additionally, Andrew Woodfield argues persuasively that,
since externalists such as Wright in their talk of goal-directed systems
identify only some outcomes as G or Z (for example, those that contribute
to the survival of the organism), the outcomes in question have been
assumed to be good. 16 In fact Wright does talk of "resultant advantage," 17 and, as we will see, that Aristotle would hold that goodness
cannot be a coincidental feature of the outcome is beyond doubt. Thus,
for an analysis at least of Aristotle, something about goodness needs to
be added to the schemata, along the following lines:
13 The distinction between internalist and externalist analyses comes from Woodfield
(1976), 104-6.
14 Wright (1976), 39.
15 Wright (1976), 81.
16 See Woodfield (1976), 100-11, 113-23, 130-40, 203-6.
17 See Wright (1976), 85.
109
S does B for the sake of G iff:

1) B tends to bring about G
2) B occurs because it tends to bring about G
3) B tends to bring about G because G is beneficial to S.18
Again,
The function of X is Z iff:
1) Z is a consequence of X's being there
2) X is there because it does (results in) Z
3) X results in Z because Z is beneficial to S.
As for the case of organic development, since the very notion of development seems to assume the worth, at least from some point of view,
of the resulting organism, although one could add as a third claim "E
develops into O because its development into O is good," one could
argue also that it is not necessary.
It is one thing, of course, to produce an analysis of how an author
uses teleological claims, another to agree that he is entitled to use them
as he has. In Wright's analysis, this question turns on whether 'because'
is justified. Woodfield attributes our willingness to say that X occurs
because it results in Z to our imagining practical reasoning to have occurred.19 Wright himself, on the other hand, justifies its use on the
ground that explanation is contextual.20 But neither of these justifications is available to Aristotle, who argues, after all, that teleological
explanation is for nature the fittest mode of explanation. Nor is the
currently popular evolutionary justification of teleology available to
him,21 given his belief in the permanence and unchangingness of species
(Generation of Animals II 1 731b24-732al).22 To be sure, it could be
that Aristotle's attempt to make nature purposive in the absence of
purpose in fact fails. However, in the following sections I will argue
that Aristotle is aware of the need to justify teleological explanation in
nature and addresses himself to the question in a way that, given certain
assumptions about goodness, is successful. In Section IV I will consider
how benefit and agency are connected in the craft analogy.
18 Cooper (1982), 197, says: "[Aristotle] understands by a goal (hou heneka) whether
natural or not, something good (from some point of view) that something else causes
or makes possible, where this other thing exists or happens (at least in part) because
of that good."
19 Woodfield (1976), 135.
20 See Wright (1976), 61.
21 See Millikan (1984), 17-49. The same is true also of Bigelow and Pargetter (1987)
insofar as their propensity theory (". . . what confers the status of a function is not
the sheer fact of survival-due-to-a-character, but rather, survival due to the propensities the character bestows upon the creature" (192)) is superimposed on the assumption of biological change across generations.
22 See Cooper (1982), esp. 202-5.
110
IV
Although Aristotle knows that teleological explanations in nature at least
appear to imply agency, he seems to reject the implication. In Physics
II he says:
Now action is for the sake of an end; therefore the nature of things also is so.
Thus if a house, e.g., had been a thing made by nature, it would have been
made in the same way as it is now by art; and if things made by nature were
made not only by nature but also by art, they would come to be in the same
way as by nature. The one, then, is for the sake of the other; and generally art
in some cases completes what nature cannot bring to a finish, and in others
imitates nature. If, therefore, artificial products are for the sake of an end, so
clearly also are natural products. The relation of the later to the earlier items
is the same in both. This is most obvious in the animals other than man: they
make things neither by art nor after inquiry or deliberation. (199al 1-21)
Again, he declares:
It is absurd to suppose that purpose is not present because we do not observe
the agent deliberating. Art does not deliberate. If the ship-building art were in
the wood, it would produce the same results by nature. If, therefore, purpose
is present in art, it is present also in nature. The best illustration is a doctor
doctoring himself: nature is like that. (199b26-32)
Since crafts are practiced by agents with conscious desires and plans
and since this is precisely the sticking point in the application of teleology to nature, the comparison of nature to craft is usually dismissed
as unhelpful. But recently Sarah Broadie has argued that the analogy
could be made interesting if it were agreed that craft is not animated
by desire. Just as a tree that desired to be a tree would be more than
a tree, so, the analogy goes, a builder qua builder already pursues the
builder's ends.23 On this view of it, the purpose of the analogy is not
to attribute desire to nature but rather, having denied its presence in
craft, to show that nature like craft exhibits regularity and also that it
makes correct moves - that craft qua craft does not err.24
Thus Broadie defends the craft analogy from the usual charge against
it. She also contends, however, that even so, the comparison fails. The
reason why it fails, she argues, is that Aristotle in his use of craft to
illuminate nature has so distorted the notion of craft that it is almost
unintelligible; whereas the real craftsman aims at a good that answers
some human need or interest recognized as a result of practical wisdom
or reflection and refines his product accordingly,25 to serve the purpose
of the analogy, craft knowledge must be thought of as developed with
a view to its purpose only and not as a capacity that could be put to
23 Broadie (1987), 43-44.
25 Broadie (1987), 48.
24 Broadie (1987), 42-43.
111
other use or exceed what is needed.26 Indeed if one thinks of craft as

static or achieved in the way Aristotle for the sake of the analogy must,
craft becomes, so Broadie argues, nothing but an automaton.27
In short, Broadie wants to deny that the main issue in Aristotle's
teleology is whether it introduces mind into nature. One might object
that to describe depsychologized craft and thus also nature as enddirected automata would seem, if that description is taken seriously, to
reintroduce intentionality since the automata familiar to us (setting aside
nature on the grounds that what is at issue is whether nature should
be so called) are manufactured by beings who do have such ends in
mind. But be that as it may, Broadie's concern that if there is no context
of human needs and interests there will be no grounds for judging the
effectiveness and value of the activity is a charge to which it is necessary
to see whether Aristotle can reply.
It must be admitted that there are certain kinds of background to
which Aristotle does not appeal. For example, if each kind of organism
were said to contribute to a larger ecosystem, how the existence of the
organism would be a good is evident; nevertheless, on the whole, Aristotle does not move in this direction, and anyway such a move might
be said merely to invite the question anew. Nor does Aristotle offer
external validation in the manner of Plato, when the latter argues that
the Forms that exist are those that participate in the Form of the Good;
indeed Aristotle claims in Nicomachean Ethics I 6 that a uniform notion
of goodness across the categories is not even possible (1096al7-29).
Yet despite these self-imposed limitations, it is undeniable that Aristotle wants to discuss the good for kinds. To take an example, the
good for human beings is to lead a fully human life and in the Ethics
Aristotle tries to say what that would be; that he would also think that
it is good for dogs to lead dogs' lives there is little reason to doubt.
That it seem to cats or to human beings or even to dogs good that dogs
lead dogs' lives is not of course necessary; likewise whether the good
for dogs is good for cats or human beings or anything else (except dogs)
seems not to be the point.
To summarize, what Aristotle holds, I think, is a view of the good
that is objective, noninstrumental, and nonrelational. Admittedly such
a view is, to say the least, problematic. For example, a scalpel is good
for doctoring (it contributes to good doctoring). But that something is
good for doctoring - or even that one is a good doctor - is independent
of the question whether doctoring is a good. Picklocks can be good for
thieving - and one can be good at thievery - without thievery being a
good. More generally, we can distinguish between what is good for Xing
and even being a good x, on the one hand, and, on the other, whether
26 Broadie (1987), 43.
27 Broadie (1987), 49.
112
Xing is a good. Indeed when Broadie criticizes the craft analogy on the
ground that without a context of human needs and interests there is
no background against which to judge the value of the craft activity,
perhaps her point can be put by saying that, insofar as the craft analogy
is intended to be its support, Aristotle's appeal to final causes in nature
fails; without a context, nature too becomes no more like good doctoring than like bad doctoring and no more like doctoring of any sort
than like thievery.
But even if the craft analogy is not very helpful for the justification
of his use of teleological explanation in nature, the ways in which the
notion of craft must be stretched in order for there to be an analogy
are themselves illuminative, and it is not self-evident that Aristotle's
view of nature is false. The health of an oak tree is, after all, something
that is objectively determinable, and insofar as health (over a certain
span of years) is the norm and disease often overcome, surely nature,
in this case the nature of oaks, can be said to be more like good doctoring
than like inept or corrupt doctoring. Of course, it might be objected
that, since healthy ivy flourishes at the expense of trees, in the case of
ivy nature is rather like thievery, but even here it is not clear that
Aristotle would have no recourse. For thieves are persons who work
against the flourishing of their fellow human beings and indeed, on an
Aristotelian view, against the flourishing of their own true selves, while
ivy is just doing what ivy flourishes by doing; though thieves have an
unactualized good, ivy is not separated from the good for ivy. To put
the point another way, only if thieves were, like human beings, dogs,
and ivy, a natural kind would the objection be well taken.
I did not promise a defense of Aristotle's view that there are natural
kinds, and beyond what was suggested in the second section of this
chapter I could not deliver on such a promise; neither do I have any
novel criteria for judging of any given kind whether it is natural. But
that Aristotle thinks there are natural kinds, that he gives some thought
to which they might be (even if he does not give us entirely successful
criteria), and that we in reading him can see, at least for many cases,
what kinds he thought natural is true nonetheless. Metaphysics VII 4 is
a case in point. There Aristotle, needing to say what sorts of things have
an essence, distinguishes compounds such as white man from things
that are something in their own right (kath' hauto) and tries to explain
what is characteristic of the former. He says:
One kind of predicate is not said of a thing in its own right because the term
that is being defined is added to something else, e.g. if in denning the essence
of white one were to state the formula of white man; another because something
else is added to it, e.g. if 'cloak' meant white man, and one were to define cloak
as white; white man is white indeed, but its essence is not to be white. (1029b311030a2)
113
To say that one classifies wrongly if to a subject in its own right

something is added (this being what Aristotle would no doubt say of
' thief) or omitted may not seem helpful. After all it is an explanation
that depends on one's having already a grasp of which sorts of things
are subjects. But Aristotle does at least tell us how the determination
ought not to be made. For what one cannot do, he says, is simply to
follow language; even if there were a single word 'cloak' (defined, let
it be imagined, as "white man") cloak would not be a natural kind.
What Aristotle goes on to argue in VII 4 is that it is species of a
genus which have an essence (1030al 1-13). In the Metaphysics perhaps
it is a bit unclear whether he wants to limit natural kinds to biological
species; that is to say, it is hard to be certain whether he intends the
bronze sphere (VII 8) and other such products of art to be merely
simplified models for illustrating the relation between matter and form
or whether he really counts specimens of such kinds among the substances. But that the paradigm cases of natural kinds in the sublunary
world will be biological there is no doubt. Indeed, as we have seen, the
clearest indication of the naturalness of these is the capacity of their
members to reproduce - to produce, that is to say, other members like
themselves.
In this and the last section I have argued that Aristotle's use of final
causes involves neither agency nor metaphor nor evolution but rather
is tied to his claim that there are natural kinds. However, even as I have
emphasized that the good for Aristotle is both objective and internal
to a kind, it cannot be denied that he sometimes explicitly takes a
broader view. At On Generation and Corruption II10 336b27-29 Aristotle
says, for example, that nature always desires what is better and that
being is better than not being. Once Aristotle has said that the realization of the life for which members of a kind are suited is, objectively,
their good, it is not surprising if their realizing their good is sometimes
taken to be a good.
In short, whether goodness is occasionally applied more broadly or
whether it remains entirely internal to kinds, that an organism undergoes change (in part) because of what is good for it seems to be for
Aristotle incontrovertible.28 Kahn, for example, puts the point in this
way:
It is true that Aristotle also speaks of actuality (energeia, entelecheia) as the goal
in teleological contexts, and we can give a partial explication of his notion of
telos in terms of degrees of actuality, as the more or most complete realization
of a thing's potentialities. But this strategy can give an adequate account of
28 Gotthelf (1988) dissents, believing that the account of what it is to be an end can be
given in terms of Aristotle's conception of an actuality alone, that is to say, without
reference to the good. See 115.
114
Aristotle's conception only if we build into the analysis of actualization the

required normative component.29
I will argue next for an interpretation of Aristotle's argument from

chance to teleology which supports Kahn's view, a conclusion I will
thereafter claim has implications for the relation between final and
material causes and, ultimately, for what I have claimed is the ontological
priority of specimens of natural kinds.
In Physics II, as we have seen, Aristotle says that the accidental includes
the conjunction of a substance and an accidental property (the musical
man) or of a pair of properties (musicality and housebuilding), as well
as the causal connection between an agent described by means of some
conjoined property and an outcome (the fluteplayer builds a house).
As he says of this last case, 'Tor just as a thing is something either in
virtue of itself or accidentally, so may it be a cause. For instance, the
housebuilding faculty is in virtue of itself a cause of a house, whereas
the pale or the musical [human being] is an accidental cause" (196b2427).
But chance too, Aristotle thinks, is part of the accidental (198a5-7)
since it involves both causation and conjunction. More specifically, Aristotle classifies chance as that part of the accidental where there is an
outcome that is unintended but beneficial. As he says, "Hence it is clear
that events which belong to the general class of things that may come
to pass for the sake of something, when they come to pass not for the
sake of what actually results, and have an external cause, may be described by the phrase 'from chance"' (197bl8-20). Unlike the narrower
category of luck (he tuche), for something to occur by chance (to automaton) it is not necessary that it be done by or happen to a being capable
of choice; if a man goes to the marketplace and happens to meet his
debtor, that is luck (196b33-197a3), but if a tripod falls and lands
upright (197bl6-18) or a stone falls and strikes someone (197b30-32),
it is chance.30
With this analysis of chance, Aristotle is ready to argue against those
who claim that the structure of plants and animals is due to the random
29 Kahn (1985), 197.
30 Though like Ackrill (1981), 36-41, I prefer 'luck' and 'chance', it should be noted
that the Greek terms are variously translated. Barnes (1984), uses 'chance' where there
is a being capable of choice and 'spontaneity' for the broader category where there
need not be. Charlton (1970) contrasts 'luck' with 'the automatic'. I find 'luck' very
natural for the case involving an agent; however, I am less happy with either 'the
automatic' or 'the spontaneous' for the broader category because neither captures
the central point: that something has happened which because it has a good result
has the appearance of having been done for the sake of that result.
115
arrangement of matter. For Empedocles, the existence of viable organic

combinations is the result of the random interactions of elements; by
nature there were man-faced ox progeny as well as human beings and
oxen, but only the latter survive (198b29-32). 31 Aristotle replies by
considering the distribution of a given feature in a population: "For
teeth and all other natural things either invariably or for the most part
come about in a given way; but of not one of the results of luck or
chance is this true" (198b34-36). Thus the combination of human faces
on human bodies or ox faces on the bodies of oxen, if they were infrequent, could be said to be by chance, but being frequent they cannot
be so. More generally, since invariable or normal outcomes cannot in
any case be attributed to chance, then, when those outcomes are also
good, teleology is needed.
It has seemed to many readers that the argument of Physics II is
obviously inadequate; why does Aristotle not consider the possibility
that mechanical, in other words, material, necessity might be enough
to explain the results he attributes to teleology? It is hard to believe,
however, that such an evident omission could simply be an oversight;
after all, Aristotle knows that iron is hard, that water is clear and fluid,
and so on. What is more likely, I think, is that Aristotle believes that
mechanical necessity is not the right sort of explanation to be considered
- in other words, that, as Aristotle sees it, mechanical necessity could
(at most) explain the occurrence of an invariable outcome, but not that
what has occurred is something that is good. As he says in his discussion
of his predecessors in Metaphysics I 7:
For surely it is not likely either that fire or earth or any such element should
be the reason why things manifest goodness and beauty both in their being and
in their coming to be, or that those thinkers should have supposed it was; nor
again could it be right to ascribe so great a matter to chance and luck.32 When
one man said, then, that reason was present - as in animals, so throughout
nature - as the cause of the world and of all its order, he seemed like a sober
man in contrast with the random talk of his predecessors. (984bll-18)
In its concern that goodness be explained, this passage is, of course,
reminiscent of Socrates' view in the Phaedo. There it is said that upon
hearing of Anaxagoras's contention that it is mind that produces order
and is the cause of everything, Socrates had expected that by assigning
a cause to each phenomenon separately and to the universe as a whole
Anaxagoras would make clear what is best for each and what is the
31 In his explanation, Empedocles seems to have failed to account for regularity across
generations - why are the offspring of dogs generally dogs and not cats or trees or
monsters? The survival of viable individuals does not explain why their offspring should
be in nearly all respects like the parents.
32 Barnes (1984) has 'spontaneity' and 'luck'. For the reasons for 'chance' and 'luck',
see note 30.
116
universal good, a hope in which he was disappointed (97c-98b). But if

such passages are relevant to the discussion of teleology in Physics II,
the point has to be that there is something about the structure, behavior,
and development of organisms which earth and fire - mechanical necessity - cannot explain. Just as Socrates says that it is absurd to try to
understand his sitting in prison in terms of the rigidity of bones and
the contraction and relaxation of sinews while neglecting to mention
that Athens thought it better to condemn him and that he thought it
better to submit to whatever penalty the city ordered (98c-e), so also
in nature, both Plato and Aristotle believe, it is true that explanation
which appeals only to the material cause can be obtuse.
Thus while Aristotle does not suppose that teleological explanation
is to be understood in terms of the workings of a Anaxagorean Mind,
and indeed neither does Plato, nonetheless there is for Aristotle in
nature a good for kinds, and it is, I would claim, his counterpart to
Plato's Form of the Good. For Plato the Forms and phenomena that
are exist and are as they are (in part) because of their relation to the
Form of the Good. The Form of the Good is the explanation of their
existence and nature in the sense of being their metaphysical ground;
likewise without appeal to the Form of the Good there cannot be an
adequate explanation of phenomena and Forms in the sense that there
cannot be adequate understanding.
Of course, in Aristotle forms are not separated in the way that the
Forms are for Plato. Nevertheless it remains true that what is good is
(in part) metaphysically explanatory of the behavior, structure, and development of living things and that to fail to appeal to it is to fail to
have an adequate understanding of them. To be sure, unlike mechanical
necessity (Socrates' *'bones and sinews"), in calling something luck or
chance there is also an acknowledgment of goodness. But the difficulty
with citing chance is that it is not really an explanation - as Aristotle
says, chance is not the cause without qualification of anything (197al4).
To put the point another way, in the case of chance what happens is
beneficial, but it does not happen because it is beneficial and its being
so is not (and cannot be) explained.
I have argued that the intended conclusion of the argument from
chance is that, where there is regularly a good outcome, teleological
explanation is required. This is so because material causes cannot explain the goodness of it, and where the outcome occurs regularly,
chance is not an adequate substitute for explanation. However if chance
is but a pseudo-explanation of the goodness of some outcome whose
occurrence (not its goodness) may in fact be explicable in other terms,
then it is also true that a charitable analysis of the argument from chance
- one that does not accuse Aristotle of simply denying or neglecting
mechanical necessity - is possible, and for understanding the role of
117
teleology in Aristotle this fact is important. For it has long been noticed
that Aristotle frequently offers two explanations, one involving the characteristics of matter, the other those of form. For example, he says that
the deer sheds its horns both for relief and because of their weight
{Parts of Animals III 2 663bl2-14), that human heads are hairy both
for protecting the brain from excessive heating and chilling and because
of the fluidity of the brain and the sutures in the skull, which are such
that heat and fluid produce the outgrowth of hair (II 14 658b2-10),
and that eyelashes exist for the protection of the eyes and because they
are located at the end of small blood vessels where moisture comes off
(II 15 658bl4-26). Likewise he holds that the ability of serpents to
turn their heads backward enables them to guard against attacks from
the rear and is a necessary consequence of their cartilaginous and flexible vertebrae (IV 11 691b31-692a5) and that the webbed feet of waterbirds are useful in enabling them to swim and are a necessary consequence of their residual earthy substance (IV 12 694a23-bl2).
These examples and others are too numerous to be ignored. Nevertheless from such passages alone it is unclear exactly what Aristotle
takes the relation between the two explanations to be. For example, if
the fluidity of the brain is a sufficient explanation for eyelashes, could
it not be objected that even if eyelashes did not serve the animal's good,
they would exist anyway? But if it were true that eyelashes would exist
regardless of their usefulness, then, even given that eyelashes are in fact
useful, the appeal to benefit would seem to have no force. To this
objection - namely, that where there is a sufficient material condition,
the value of the result is causally irrelevant - there are several possible
lines of reply. That Aristotle intends teleological explanation to be, like
chance, compatible with mechanical necessity and yet, unlike chance,
more fundamental than explanation in terms of matter will be argued
in the next section. There I will argue also that it is final causality that
grounds the ontological priority of specimens of natural kinds.
VI
In response to the worry that given a necessitating material cause the
teleological explanation is not really explanatory, Martha Nussbaum at
one time proposed that double explanations are pragmatic. Against his
mechanist predecessors she imagined that Aristotle would have argued
along the following lines: "If you were a shepherd in charge of the
flocks, which account [Democritus's or Homer's] would give you more
information that was relevant to your plans and precautions? From eight
lines of Homer I learn more that is general and valuable about the
118
behavior of lions than I would from two volumes of detailed atomcharts."33

Other interpreters have solved the problem by supposing that material causes, since they are said by Aristotle to be hypothetical (200al3),
are in fact not necessitating. Balme, for example, at one time suggested
that matter (the stuff which is the matter) itself sometimes behaves
erratically,34 in which case, given that there is regularity in nature, something else is needed to explain that regularity, and teleology comes to
hand.
If Aristotle does believe that matter is erratic, his reason for thinking
final causes ineliminable would be evident. However I am inclined to
doubt that Aristotle does hold that view. Besides, the assumption that
teleology can be effective - that is to say, that it can have more than a
pragmatic role - only if material causes are not necessitating has itself
been challenged; according to David Charles, even given the existence
of independent physical conditions sufficient to necessitate the occurrence of psychological or biological phenomena, Aristotle would still rightly - think that teleological explanation is explanatory.35 Only explanation in terms of goals, Charles claims, picks out those features of
the goal or end-state which are essentially or directly connected to the
survival of the organism; in addition, explanation in terms of goals and only it - can select certain stages in a developmental causal story
as significant and explain the appropriateness of the route from the
desire or potentiality to the goal.36 Moreover, he argues:
Since desires [potentialities] are essentially goal-directed states . . ., they cannot
be type-identical with a type of physical state which is not essentially goaldirected. Further, in Aristotle's account it follows from the failure of typeidentity that there is no token-identity between (e.g.) this desire and this physical
state.37
But Aristotle does not believe merely that final and material causes
are compatible (assuming that he does believe them to be so); he believes
teleological explanations to be more fundamental than material explanations. That is to say, beyond holding that the features explained by
a physical account and the features explained by a teleological account
are distinct, Aristotle claims further that material necessity is hypothetical; it is because the organism is what it is that it must be made of matter
having certain characteristics. Of Aristotle's view Charles says:
33 Nussbaum (1978), 71.
34 Balme (1939), 137-38. See also Cooper (1982), 211.
35 Charles (1988), 38. Charles falls roughly into the functionalist camp (even if physical
states are sufficient for the outcome the behavior could in principle have been differently realized). Concerns of the sort Ackrill (1973) presents need not undermine
this view. See Chapter V, n. 10, of the present book.
37 Charles (1988), 40.
36 Charles (1988), 38-39.
119
It is in virtue of possessing these goals that man has the nature he has; and it
is because he has this nature that he has the potentiality to <f> and must <p in C.
For if he failed to do so, he would fail to be the creature he is. So far from
teleological causation resting on efficient causation, it is rather the reverse. The
presence of goals makes the organism what it is, and its being that organism
explains why it must <>
/ in C. If this is correct, teleological causation is not
explained in terms of efficient causal necessitation. The one is irreducible to
the other. Teleological goals are taken as primitives. These latter concerns show
why Aristotle preferred the downwards-perspective. The nature of the kind is basic,
and this is fixed by primitive teleological factors.38
Charles has argued that the behavior of an organism is explained in

terms of its nature and that its nature is in turn determined by its goals.
Moreover, where the goal in question is organic development, it is, of
course, a goal that is generally realized, and its realization is, I have
argued, the good for beings of that kind. Thus I agree with Charles
that Aristotle has reason to prefer what Charles calls the downward
perspective. For if the upward perspective were taken, the goodness
for the organism of the developmental outcome could only be explained
as chance, and chance, Aristotle contends, cannot account for the regular occurrence of a good outcome.
In short, even if matter is not, as Balme thought, erratic, even if
material causes are sufficient for a sort of outcome that is regularly
good for organisms of a given kind, because for Aristotle the outcome
must not only be good but be as it is because it is good, the downward
perspective has to be preferred. Indeed from the downward perspective,
objections of the sort raised in the previous section to the effect that,
given a complete physical account, a teleological account would seem
to do no work can be addressed. If it is claimed that should eyelashes,
for example, not be good for the organism, they would exist anyway,
the reply is that if they were not good for the organism, they would not
regularly exist.
I have offered an account of Aristotle's justification of teleological
explanation in nature and indeed of his claim that teleological explanation is basic and material explanation hypothetical. Of course, it is
not the case that teleological explanation can account for all the properties of organisms. It cannot do so any more than the artist's intentions
account for all the properties of the material counterpart to a work of
art; certainly there will be features of living things which are, on the
downward perspective, entirely adventitious and which can only be explained, if explained at all, by appeal to the nature of matter. However
since my goal is to provide a justification for the claim that specimens
of natural kinds are ontologically more fundamental than the sensible
objects with which they are, though not identical, numerically the same,
38 Charles (1988), 43.
120
this examination of teleology would serve no purpose in my exposition

of Aristotle's account of substance unless it is also true that the application of final causality, the most fundamental variety of causation,
occurs primarily at the level of specimens of kinds. But this is actually
an easy case to make, and once again it is art that can make the connection clear.
Let us assume, as Danto claims, that a work of art has only the
determinate perceptual qualities relevant to its representational content
and that to understand that these are qualities of the work of art and
not just of its material counterpart requires the downward perspective.
Even so, on Danto's account of art, it will also be true that in art forms
where there can be multiple tokens, such as woodcuts or photographs
and in all the performing arts, what makes the tokens multiple and
distinguishable will be features of the material counterpart. One might
say that on Danto's view a work of art, as such, has the same implicit
generality, the same absence of a namely-rider, as was attributed to
specimens of kinds in earlier chapters; like 'a human being', Ansel Adams's Old Faithful Geyser, Yellowstone National Park, Wyoming, 1982 need
not refer to either this or that member of the kind, but to any member
of the kind, taken as representative of the others. The end - the work
of art - does not include that which differentiates one print from another.
In short my proposal is that, just as in the case of art there are
preferred explanations of the features of the work of art which explain
some but only some of the features of its material counterpart, so for
Aristotle an adequate explanation of the goodness of certain activities,
structures, and patterns of development is an explanation only of what
is characteristic of specimens of natural kinds, and not of the more
finely differentiated activities and structures of individual organisms.
But just because this is so, teleological explanation can reasonably be
said properly to apply to specimens of natural kinds rather than to the
individual organisms with which they are numerically the same. To be
sure, living a lion's life is done by some particular lion in some particular
way; this lion is at this moment stalking this zebra. Nevertheless such
detail is not part of the specification of the end, and just because it is
not, the final cause remains at a level of generality commensurate with
the generality of specimens of a kind, thereby in the case of specimens
of natural kinds, justifying their ontological priority.
VII
In these chapters I have defended a version of the view that substance
is form. On my view of Aristotle, what there most fundamentally is in
the universe is form; ontological priority belongs to specimens of natural
121
kinds, and it is only because the characteristics of the preexisting matter

produce objects distinguishable from one another within their kinds
that there are sensible objects as well. To put the point another way,
sensible objects for Aristotle truly are composites; they are derived from
matter and form not just conceptually or analytically, but ontologically.
But if on my view it is not too misleading to describe Aristotelian
substances as Platonic forms multiplied by matter, it cannot be denied
that in Aristotle the role of matter compromises Plato's requirements
for reality and knowability. The problem is not that every sensible object
has accidental properties; sensible objects, I have argued, are not substances. The problem is rather that, because they are numerically the
same as sensible objects, even specimens of kinds are not eternal. However, even though the same cannot be said of kinds of artefacts, biological species are, or so Aristotle thinks, eternal, and to say that they
are eternal is just to say that they always have members; indeed, it is
by reproduction that they partake in what is divine (De Anima 415a25b7). The result is that, since specimens of the same kind are indistinguishable, Aristotle can claim that even without separated Forms there
are objects that can be said to be first in knowledge, definition, and since teleological explanation is necessary and explanatory of specimens
- time.
That the importance Plato and Aristotle assign to permanence in assessing what is real and knowable can in retrospect seem puzzling must
be admitted. What I myself am inclined to suspect is that Plato finds flux
unsatisfactory on what might be called aesthetic as well as epistemological
grounds. In a telling metaphor at the end of the Cratylus, Socrates says
to Cratylus that he cannot believe that all things leak like a pot or that
the world is like a man who has a running at the nose (440c-d). A leaky
pot or a runny nose is not merely something that allows change but
something that is deficient just on that account. That is to say, what
underlies Plato's demand for unchangeability may in the end be not only
an epistemological requirement but a view about the nature of perfection.
Yet whatever their motivation, it is clear that the metaphysical and epistemological assumptions Plato has are ones Aristotle largely shares.
What I have tried to show is that it is these assumptions, combined with
his rejection of separation, which shape Aristotle's metaphysics and
epistemology.
Bibliography
The Greek texts that I have used for Plato and Aristotle are the Oxford
Classical Texts.
Ackrill, J. L. 1963. Aristotle's Categories and De Interpretatione. Trans, with notes.
Oxford: Clarendon Press.
1973. "Aristotle's Definitions of Psuche." Proceedings of the Aristotelian Society
73: 119-33.
1981. Aristotle the Philosopher. Oxford: Oxford University Press.
Albritton, Rogers. 1957. "Forms of Particular Substances in Aristotle's Metaphysics." Journal of Philosophy 54: 699-708.
Allen, R. E. 1960. "Participation and Predication in Plato's Middle Dialogues."

Philosophical Review 69: 147-64.
1970. Plato's 'Euthyphro' and the Earlier Theory of Forms. New York: Humanities
Press.
1961. "Aristotle." In Anscombe and P. T. Geach, Three Philosophers.
Balme, David M. 1939. "Greek Science and Mechanism: 1. Aristotle on Nature
and Chance." Classical Quarterly 33: 129-38.
1984. "The Snub." Ancient Philosophy 4: 1-8.
Barnes, Jonathan. 1975. Aristotle's Posterior Analytics. Trans, with notes. Oxford:
Clarendon Press,
ed. 1984. The Complete Works of Aristotle: The Revised Oxford Translation. 2
vols. Princeton, N.J.: Princeton University Press.

Bigelow, John, and Robert Pargetter. 1987. "Functions." Journal of Philosophy
84: 181-96.
Bluck, R. S. 1957. "Forms as Standards." Phronesis 2: 115-27.
Broadie, Sarah. 1987. "Nature, Craft and Phronesis in Aristotle." Philosophical
Topics 15: 35-50.
Burnyeat, M. F. 1979. "The Virtues of Plato," a review of Irwin's Plato's Moral
Theory. New York Review of Books 26, no. 14: 56-60.
1992. "Is an Aristotelian Philosophy of Mind Still Credible? A Draft." In

Essays on Aristotle's 'De Anima'. Ed. Martha Nussbaum and Amelie Rorty.
Carnap, Rudolf. 1947. Meaning and Necessity. Chicago: University of Chicago
Press.
123
124
BIBLIOGRAPHY
Cashdollar, Stanford. 1973. "Aristotle's Account of Incidental Perception."

Phronesis 18: 156-75.
Castaneda, H-N. 1977. "Perception, Belief, and the Structure of Physical Objects and Consciousness." Synthese 35: 285-351.
Charles, David. 1984. Aristotle's Philosophy of Action. Ithaca, N.Y.: Cornell University Press.
1988. "Aristotle on Hypothetical Necessity and Irreducibility." Pacific Philosophical Quarterly 69: 1-53.
Charlton, W. 1970. Aristotle's Physics Books I and II. Trans, with introduction
and notes. Oxford: Clarendon Press.
Cherniss, Harold. 1944. Aristotle's Criticism of Plato and the Academy. Vol. I.
Baltimore: Johns Hopkins University Press.

1957. "The Relation of the Timaeus to Plato's Later Dialogues." American
Journal of Philology 78: 225-66.
Code, Alan. 1985. "On the Origins of Some Aristotelian Theses About Predication." In How Things Are: Studies in Predication and the History of Philosophy and Science. Ed. James Bogen and James McGuire. Dordrecht: D.
Reidel.
1986. "Aristotle: Essence and Accident." In Philosophical Grounds of Rationality: Intentions, Categories, Ends. ed. Richard Grandy and Richard Warner.
1991. "Some Basic Elements of Aristotle's Metaphysical Essentialism." Michigan Philosophy News. Fall 1991: 2-6.
Cohen, Marc. 1992. "Hylomorphism and Functionalism." In Essays on Aristotle's

'De Anima'. Ed. Martha Nussbaum and Amelie Rorty. Oxford: Clarendon
Press.
Collingwood, R. G. 1945. The Idea of Nature. Oxford: Clarendon Press.
Cooper, John. 1982. "Aristotle on natural teleology." In Language and Logos.
Ed. Malcolm Schofield and Martha Nussbaum. Cambridge: Cambridge University Press.
Cornford, F. M. 1939. Plato and Parmenides: Parmenides' Way of Truth and
Plato's Parmenides. Trans, with an introduction and commentary. New York:
Humanities Press, reprinted Indianapolis: Library of Liberal Arts, BobbsMerrill, n.d.
Dancy, Russell. 1975. "On Some of Aristotle's First Thoughts about Substances." Philosophical Review 84: 338-73.
1978. "On Some of Aristotle's Second Thoughts about Substances: Matter."
Philosophical Review 87: 372-413.
Danto, Arthur. 1981. The Transfiguration of the Commonplace. Cambridge, Mass.:
Harvard University Press.

1986. The Philosophical Disenfranchisement of Art. New York: Columbia Uni-
versity Press.
Dooley, W. E., trans. 1989. Alexander of Aphrodisias, On Aristotle's Metaphysics
1. Ithaca, N.Y.: Cornell University Press.
Driscoll, John. 1981. "EIAH in Aristotle's Earlier and Later Theories of Substance." In Studies in Aristotle. Ed. Dominic O'Meara. Washington, D.C.:
Catholic University of America Press.
BIBLIOGRAPHY
125
Else, G. F. 1936. "The Terminology of the Ideas." Harvard Studies in Classical

Philology 47: 17-55.
Engberg-Pedersen, T. 1979. "More on Aristotelian Epagoge." Phronesis 24\
301-19.
Fine, Gail. 1982. "Aristotle and the more accurate arguments." In Language
and Logos, ed. Malcolm Schofield and Martha Nussbaum. Cambridge: Cambridge University Press.
1984. "Separation." Oxford Studies in Ancient Philosophy 2: 31-87.
1985. "Separation: A Reply to Morrison." Oxford Studies in Ancient Philosophy
3: 159-65.
Frede, Dorothea. 1992. "The Cognitive Role of Phantasia." In Essays on Aristotle's 'De Anima. Ed. Martha Nussbaum and Amelie Rorty. Oxford: Clarendon Press.
Frede, Michael. 1987a. "Individuals in Aristotle." In Essays in Ancient Philosophy.
Minneapolis: University of Minnesota Press.
1987b. "Substance in Aristotle's Metaphysics." In Essays in Ancient Philosophy.
Minneapolis: University of Minnesota Press.
Frege, Gottlob. 1952. "On Sense and Meaning." In Translations from the Philosophical Writings of Gottlob Frege. Ed. Peter Geach and Max Black. Oxford:
Basil Blackwell.
Furth, Montgomery. 1985. Aristotle Metaphysics Books VII-X: Zeta, Eta, Theta,
Iota. Trans, with notes. Indianapolis: Hackett.

1988. Substance, Form and Psyche: An Aristotelean Metaphysics. Cambridge: Cambridge University Press.
Geach, P. T. 1956. "The Third Man Again." Philosophical Review 65: 72-82.
Gibbard, Allan. 1975. "Contingent Identity." Journal of Philosophical Logic 4:
187-221.
Gill, Mary Louise. 1989. Aristotle on Substance: The Paradox of Unity. Princeton,
N.J.: Princeton University Press.
Goldman, Alvin. 1986. Epistemology and Cognition. Cambridge, Mass.: Harvard
University Press.
Gotthelf, Allan. 1988. "The Place of the Good in Aristotle's Natural Teleology."
In Proceedings of the Boston Area Colloquium on Ancient Philosophy. Vol. IV.
Ed. John Cleary and Daniel Shartin. Lanham, Md.: University Press of
America.
Graham, Daniel. 1987. Aristotle's Two Systems. Oxford: Clarendon Press.
Gulley, Norman. 1954. "Plato's Theory of Recollection." Classical Quarterly N.S.
4: 194-213.
Guthrie, W. K. C. 1975. A History of Greek Philosophy. Vol. IV. Cambridge:
Cambridge University Press.
1978. A History of Greek Philosophy. Vol. V. Cambridge: Cambridge University
Press.
1981. A History of Greek Philosophy. Vol. VI. Cambridge: Cambridge University
Press.
Hamilton, Edith, and Huntington Cairns, eds. 1961. The Collected Dialogues of
Plato. Princeton, N.J.: Princeton University Press.
Hamlyn, D. W. 1976. "Aristotelian Epagoge." Phronesis 21: 167-84.
Hardie, W. F. R. 1936. A Study in Plato. Oxford: Clarendon Press.
126
BIBLIOGRAPHY
Hartman, Edwin. 1977. Substance, Body, and Soul: Aristotelian Investigations.
Princeton, N.J.: Princeton University Press.

Heinaman, Robert. 1981a. "Non-substantial Individuals in the Categories/' Phronesis 26: 295-307.
1981b. "Knowledge of Substance in Aristotle." Jou rnal of Hellenic Studies 101:
63-77.
Hussey, Edward. 1983. Aristotle's Physics Books III and IV. Trans, with notes.
Irwin, T. H. 1977. Plato's Moral Theory: The Early and Middle Dialogues. Oxford:
Clarendon Press.
1986. "Aristotelian Actions." Phronesis 31: 68-89.
1988. Aristotle's First Principles. Oxford: Clarendon Press.
Kahn, Charles. 1981. "The Role of nous in the Cognition of First Principles in
Posterior Analytics II 19." In Aristotle on Science: The 'Posterior Analytics'. Ed.
Enrico Berti. Padua: Editrice Antenore.

1985. "The Place of the Prime Mover in Aristotle's Teleology." In Aristotle
on Nature and Living Things. Ed. Allan Gotthelf. Pittsburgh: Mathesis
Publications.
Kirwan, Christopher. 1971. Aristotle's Metaphysics Books F, A, and E. Trans.
with notes. Oxford: Clarendon Press.
Lear, Jonathan. 1982. "Aristotle's Philosophy of Mathematics." Philosophical
Review 91, 161-92.
Lennox, James. 1985. "Are Aristotelian Species Eternal?" In Aristotle on Nature
and Living Things. Ed. Allan Gotthelf. Pittsburgh: Mathesis Publications.
Lesher, James. 1971. "Aristotle on Form, Substance, and Universals: A Dilemma." Phronesis 16: 169-78.
1973. "The Meaning of N0T2) in the Posterior Analytics." Phronesis 18: 4 4 68.
Leszl, Walter. 1972. "Knowledge of the Universal and Knowledge of the Particular in Aristotle." Review of Metaphysics 26: 278-313.
Lewis, Frank. 1982. "Accidental Sameness in Aristotle." Philosophical Studies
42: 1-36.
1985. "Form and Predication in Aristotle's Metaphysics." In How Things Are:
Studies in Predication and the History of Philosophy and Science. Ed. James
Bogen and James McGuire. Dordrecht: D. Reidel.

1991. Substance and Predication in Aristotle. Cambridge: Cambridge University
Press.
Loux, Michael. 1979. "Form, Species and Predication in Metaphysics Z, H, and
0." Mind 88: 1-23.
1991. Primary 'Ousia': An Essay on Aristotle's Metaphysics Z and H. Ithaca, N.Y.:
Cornell University Press.
Mabbott, J. D. 1926. "Aristotle and the XQPISMOS of Plato." Classical Quar-
terly 20: 72-79.

Massey, Gerald. 1970. Understanding Symbolic Logic. New York: Harper and Row.
Mates, Benson. 1965. Elementary Logic. New York: Oxford University Press.
Matthews, Gareth. 1982. "Accidental unities." In Language and Logos. Ed. Malcolm Schofield and Martha Nussbaum. Cambridge: Cambridge University
Press.
BIBLIOGRAPHY
127
McKirahan, Richard. 1983. "Aristotelian Epagoge in Prior Analytics 2. 21 and

Posterior Analytics 1. 1." Journal of the History of Philosophy 21: 1-13.
1992. Principles and Proofs: Aristotle's Theory of Demonstrative Science. Princeton,
N.J.: Princeton University Press.
Miller, Fred. 1973. "Did Aristotle Have the Concept of Identity?" Philosophical
Review 82: 483-90.
Millikan, Ruth Garrett. 1984. Language, Thought, and Other Biological Categories.
Cambridge, Mass.: MIT Press.
Modrak, Deborah. 1979. "Forms, Types, and Tokens in Aristotle's Metaphysics. "
Journal of the History of Philosophy 7 7: 371-81.
1987. Aristotle: The Power of Perception. Chicago: University of Chicago Press.
Mondadori, Fabrizio. 1978. "Interpreting Modal Semantics." In Studies in Formal Semantics. Ed. Franz Guenthner and Christian Rohrer. Amsterdam:
North-Holland.
Moravcsik, Julius. 1971. "Learning as Recollection." In Plato: I. Ed. Gregory
Vlastos. Garden City, N.Y.: Doubleday.
1975. "Aitia as Generative Factor in Aristotle's Philosophy." Dialogue 14:
622-38.
Morrison, Donald. 1985. "Separation in Aristotle's Metaphysics." Oxford Studies
in Ancient Philosophy 3: 125-57.
Nussbaum, Martha. 1978. Aristotle's De Motu Animalium. Text with translation,
commentary, and interpretative essays. Princeton, N.J.: Princeton University Press.
Owen, G. E. L. 1953. "The Place of the Timaeus in Plato's Dialogues." Classical
Quarterly N.S. 3: 79-95.
1965. "Inherence." Phronesis 10: 97-105.
1966. "The Platonism of Aristotle." Proceedings of the British Academy 51: 1 2 5 50.
1978. "Particular and General." Proceedings of the Aristotelian Society 79: 1-21.
Owens, Joseph. 1963. The Doctrine of Being in the Aristotelian 'Metaphysics'. 2d
ed. rev. Toronto: Pontifical Institute of Medieval Studies.
Patterson, Richard. 1985. Image and Reality in Plato's Metaphysics. Indianapolis:
Hackett.
Peirce, C. S. 1933. Collected Papers. Ed. Charles Hartshorne and Paul Weiss.
Vol. IV. Cambridge, Mass.: Harvard University Press.
Pellegrin, Pierre. 1986. Aristotle's Classification of Animals. Trans. Anthony Preus.
Berkeley: University of California Press.
Penner, Terry. 1987. The Ascent from Nominalism. Dordrecht: D. Reidel.
Peterson, Sandra. 1969. The Masker Paradox. Unpublished PhD dissertation,
Princeton University.
Quine, W. V. O. 1953. "Reference and Modality." In From a Logical Point of
View. Cambridge, Mass.: Harvard University Press.
1970. "Natural Kinds." In Essays in Honor of Carl G. Hempel. Ed. Nicholas
Rescher. Dordrecht: D. Reidel.
Rees, D. A. 1963. "Plato and the Third Man." Proceedings of the Aristotelian
Society Suppl. Vol. 37: 165-176.
Rorty, Richard. 1973. "Genus as Matter: A Reading of Metaphysics Z-H." In
Exegesis and Argument. Ed. E. N. Lee, A. P. D. Mourelatos, and R. M. Rorty.
New York: Humanities Press.
128
BIBLIOGRAPHY
Ross, W. D. 1949. Aristotle's Prior and Posterior Analytics. A revised text with
introduction and commentary. Oxford: Clarendon Press.
1951. Plato's Theory of Ideas. Oxford: Clarendon Press.
Sellars, Wilfrid. 1957. "Substance and Form in Aristotle." Journal of Philosophy
54: 688-99.
1967. "Aristotle's Metaphysics: An Interpretation." In Philosophical Perspectives. Springfield, 111.: C. C. Thomas.
Sharpies, R. W. 1985. "Species, Form, and Inheritance: Aristotle and After."
In Aristotle on Nature and Living Things. Ed. Allan Gotthelf. Pittsburgh:
Mathesis Publications.
Smith, J. A. 1921. "TOAE TI in Aristotle." Classical Review 35: 19.
Sorabji, Richard. 1980. Necessity, Cause, and Blame: Perspectives on Aristotle's
Theory. Ithaca, N.Y.: Cornell University Press.
1992. "Intentionality and Physiological Processes: Aristotle's Theory of Sense
Perception." In Essays on Aristotle's 'De Anima.' Ed. Martha Nussbaum
and Amelie Rorty. Oxford: Clarendon Press.
Strang, Colin. 1963. "Plato and the Third Man." Proceedings of the Aristotelian
Society, Suppl. Vol. 37: 147-64.
Suppe, Frederick, ed. 1974. The Structure of Scientific Theories. Edited with introduction. Urbana, 111.: University of Illinois Press.
Vlastos, Gregory. 1954. "The Third Man in the Parmenides. "Philosophical Review
63: 319-49.
1965. "Anamnesis in the Meno." Dialogue 4: 143-67.
1991. Socrates: Ironist and Moral Philosopher. Ithaca, N.Y.: Cornell University
Press.
White, Nicholas. 1971. "Aristotle on Sameness and Oneness." The Philosophical
Review 80: 177-97.
1976. Plato on Knowledge and Reality. Indianapolis: Hackett.
Wiggins, David. 1967. Identity and Spatio-Temporal Continuity. Oxford: Basil
Blackwell.
Wilkes, Kathleen. 1992. "Psuche versus the Mind." In Essays on Aristotle's 'De
Anima.' Ed. Martha Nussbaum and Amelie Rorty. Oxford: Clarendon Press.
Witt, Charlotte. 1989. Substance and Essence in Aristotle: An Interpretation of 'Metaphysics' VII-IX. Ithaca, N.Y.: Cornell University Press.
Wollheim, Richard. 1968. Art and Its Objects: An Introduction to Aesthetics. New
York: Harper & Row.
Wolterstorff, Nicholas. 1970. On Universals: An Essay in Ontology. Chicago: University of Chicago Press.
Woodfield, Andrew. 1976. Teleology. Cambridge: Cambridge University Press.
Woods, M. J. 1967. "Problems in Metaphysics Z, Chapter 13." In Aristotle: A
Collection of Critical Essays. Ed. Julius Moravcsik. Garden City, N.Y.: Doubleday.
1975. "Substance and Essence in Aristotle." Proceedings of the Aristotelian
Society 75: 167-80.
1982. Aristotle's Eudemian Ethics Books I, II, and VIII. Trans, with commentary. Oxford: Clarendon Press.
Wright, Larry. 1976. Teleological Explanations. Berkeley: University of California
Press.
Index
accidental sameness. See accidental

unities; numerical sameness;
referential opacity
accidental unities
accounts of, 25-31
as specimens of kinds, 30
Ackrill,J. L., 26, 89-90
Alexander of Aphrodisias, 15
Allen, R. E., 59
Balme, David, 46-7, 89, 118
Barnes, Jonathan, 69, 72-5, 84
beliefs, activated and unactivated, 66
Bluck, R. S., 60
Broadie, Sarah, 110-12
Carnap, Rudolf, 36
Castaneda, H.-N., 104
chance, 114-17
Charles, David, 118-19
Cherniss, Harold, 60-1, 82
Code, Alan, 52
Collingwood, R. G., 107
Dancy, Russell, 27, 35, 89, 91
Danto, Arthur, 100-3, 105, 120
Driscoll, John, 52
Engberg-Pederson, T., 78
essence
as specimen of kind, 57, 88
and form, 45-6
relation to thing {Meta. VII 6), 21,
38-9
Fine, Gail, 7, 9-10, 13, 83-4, 93
final causality. See teleological explanation
flux
knowledge, and, 19, 78
related to separated Forms, 8-11, 92
substances, and, 96
form
and essence, 456
indistinguishability within kinds, 55, 61
predication (Lewis, Loux), 53-5
senses of, 45-6, 55
specimens of natural kinds, 47, 51
Form of the Good (Plato)
teleology of Aristotle's counterpart to,
116
Forms (Plato)
Aristotle's criticism of separation of,
5-20, 92
capacity for independent existence,
7-14
importance for Aristotle of their
unchanging nature, 3, 121
obstacles to knowledge, 16-19, 63-8,
87
as particulars and universals (Aristotle's
interpretation), 14-19
regress arguments against, 15-16, 38,
87
Third Man Argument, 15-16
as types, 61, 77
universal or particular, 57-61
Frede, Michael, 89
Frege, Gottlob, 24-5, 35
Furth, Montgomery, 44
Geach, P. T., 59
geometrical entities, 29
Gibbard, Allan, 102
129
130
INDEX
Gill, Mary Louise, 90

Graham, Daniel, 44
guises, 104
Gulley, Norman, 65
Guthrie, W. K. C, 70-1, 94
Hamlyn, D. W., 67, 69
Heinaman, Robert, 80
identity
distinct from numerical sameness, 2,
22-5, 27-8, 35, 40, 49, 100-3
indiscernibility of identicals, 22-4, 37
nonstandard concepts of, 35-7, 102
See also numerical sameness; referential
opacity
independent being
defined, 86
of Plato's Forms, 87
separation as, 86, 88-95
of specimens of kinds, 87-88, 91-2
independent existence
capacity of Forms for, 7-14
See also separation
Irwin, T. H., 7, 27, 35-6, 46-7
Kahn, Charles, 72-6, 113-14
kinds
features of, 41-3
the good for, 111, 116
as like and unlike properties, 42-3
natural, 112-14
See also specimens of kinds; specimens
of natural kinds
Kirwan, Christopher, 26
knowability
of Plato's Forms, 3, 14, 16-20, 63-7
of specimens of kinds, 2, 50-1, 56, 63,
79
knowledge
in Plato's Theory of Recollection, 63-6
potential and actual, 79-82
transition from perception to, 3, 6879,82
See also knowability; recollection;
unknowability
Lear, Jonathan, 29
Leibniz's Law, 23-5, 37, 101
Leszl, Walter, 79-82
Lewis, Frank, 4, 25, 32, 34-5, 44, 52-5
Loux, Michael, 4, 44, 52-5, 95
Masker Paradox, 33, 34
matter
accounts of, 55, 88-91
inseparability, 91
Matthews, Gareth, 24-5, 35
McKirahan, Richard, 69, 71
Meno's Paradox, 64
Miller, Fred, 26
Modrak, Deborah, 69-71, 74-7

Moravcsik, Julius, 26
Morrison, Donald, 93-4
Nonidentity Assumption, 56-7
numerical distinctness,
as meaning of Platonic separation,
12-14
of Plato's Forms, 12-14, 56, 92
numerical sameness
accidental sameness as variety of,
24-34, 37
essential sameness as variety of, 24,
37-40
relation to identity, 2, 23-4, 30-1, 49,
100-6
of substances with sensible objects, 61,
86
See also identity; 'qua'; referential
opacity
Owen, G. E. L., 56-7, 99
Owens, Joseph, 81
Patterson, Richard, 13, 58
perception
gap between knowledge and, 72-7
induction from, 68-71
Peterson, Sandra, 36
priority
link to separation, 6-8, 10, 86, 93
of specimens of natural kinds, 103-6,
120-1
properties
distinct from kinds, 42-3
forms as universal, and, 51-5
of sensible objects and particular
forms, 46-7, 50
of works of art, 101-5
'qua'
occurrences in Aristotle, 47-9
restrictive function of, 50, 79
use of, 29-30, 35
Quine, W. V. O., 22
recollection
criticism of Plato's Theory, 19, 67-8
doctrine of Plato, 63-7
referential opacity
meaning of, 22-3
options for understanding, 35-7
Ross, W. D., 69
separation
in definition, 95-7
independent being, and, 86-91, 93-5
independent existence of Forms, and,
7-10, 12-14
numerical distinctness of Forms, and,
10-16, 56-7, 91-3
See also independent being; priority
INDEX
Sorabji, Richard, 26
specimens of kinds
accidental unities as, 30-2, 34
definition, 48-9
knowability, 2, 50, 55-6, 61, 63
knowledge of, 77-82
sensible objects, and, 30-1, 37, 88
specimens of natural kinds, and, 30,
104-5
specimens of natural kinds
as epistemologically fundamental, 50
as ontologically fundamental, 3, 31,
97_9, 103-6, 119-21
substances as, 40, 50, 55-6, 61-3, 86,
104, 120-1
See also specimens of kinds
Strang Colin, 58
substances
alleged inconsistency of view, 45, 51-5
candidates for, 44-6, 83-6
in the Categories, 40-3, 56
criteria for, 44-5, 83-5, 91
as specimens of natural kinds, 40, 50,
55-6, 61-3, 86, 97-9, 120-1
See also specimens of natural kinds
substitutivity
accidental sameness, and, 25-8, 31-5,
37
essential sameness, and, 36-9
failures of, 22-8, 34-7
as principle governing identity, 22
Suppe, Frederick, 98-9
131
teleological explanation
art, and, 105
chance, and, 114-17
craft analogy, 110-12
downward perspective, 119-20
externalist accounts, 108-9
the good, and, 111, 113-16
internalist accounts, 106-7
material causes, and, 117-19
natural kinds, and, 112-13
Third Man Argument
as in Alexander of Aphrodisias, 15-16
nature of Forms, and, 57-60
separation, and, 56-7
types
Forms as, 60-1
substances as, 61-2, 74-88
universals
knowability of, 14, 19, 51, 61
knowledge of, 74-82
separated Forms as, 7, 9-10, 14-16,
57-60, 83-4
unknowability
of accidental characteristics, 14, 16-18,
46-7
recollection, and, 18-19, 63-8
Vlastos, Gregory, 13
White, Nicholas, 24, 65
Witt, Charlotte, 84
Woodfield, Andrew, 108-9
Woods, M. J., 57
Wright, Larry, 108-9

Lynne Spellman-Substance and Separation in Aristotle (2002)

Încărcat de

Informații document

Titlu original

Drepturi de autor

Formate disponibile

Partajați acest document

Partajați sau inserați document

Opțiuni de partajare

Vi se pare util acest document?

Este necorespunzător acest conținut?

Drepturi de autor:

Formate disponibile

Lynne Spellman-Substance and Separation in Aristotle (2002)

Încărcat de

Drepturi de autor:

Formate disponibile

Substance and Separation

Substance and Separation

Published by the Press Syndicate of the University of Cambridge

In the course of the argument of this book I draw heavily on recent

We are sufficiently assured of this, then, even if we should examine it

This is a study of Aristotle's theory of substance, more precisely of his

SUBSTANCE AND SEPARATION IN ARISTOTLE

Thus by the conclusion of Chapter III, an account of substance has

SUBSTANCE AND SEPARATION IN ARISTOTLE

Rejection of Platonic separation is the starting point for Aristotle's

SUBSTANCE AND SEPARATION IN ARISTOTLE

given will it be possible to decide whether the separation he intends to

SEPARATION OF PLATONIC FORMS

SUBSTANCE AND SEPARATION IN ARISTOTLE

and that * separation' does in fact mean a capacity for independent

SEPARATION OF PLATONIC FORMS

SUBSTANCE AND SEPARATION IN ARISTOTLE

SEPARATION OF PLATONIC FORMS

SUBSTANCE AND SEPARATION IN ARISTOTLE

against the supposition that a capacity for independent existence is what

SEPARATION OF PLATONIC FORMS

Plato's (as opposed to Socrates') view of Forms, it is most plausible to

SUBSTANCE AND SEPARATION IN ARISTOTLE

independent existence, it is at precisely this point that there must be

SEPARATION OF PLATONIC FORMS

SUBSTANCE AND SEPARATION IN ARISTOTLE

It has sometimes been found puzzling that in this argument, unlike

In Metaphysics VII 15 Aristotle says:

SEPARATION OF PLATONIC FORMS

Certainly the correlation found in this argument between different

SUBSTANCE AND SEPARATION IN ARISTOTLE

satisfied by more than one thing. That this is so would seem to be

SEPARATION OF PLATONIC FORMS

SUBSTANCE AND SEPARATION IN ARISTOTLE

At the beginning of Metaphysics VII 6 Aristotle inquires whether each

SUBSTANCE AND SEPARATION IN ARISTOTLE

alternative to Plato's Theory of Forms, namely, the account of substance

It is W. V. O. Quine who seems to have introduced the term 'referential

Not all failures of substitutivity are, of course, genuinely problematic.

REFERENTIAL OPACITY IN ARISTOTLE

SUBSTANCE AND SEPARATION IN ARISTOTLE

REFERENTIAL OPACITY IN ARISTOTLE

truth. Thus, on Frege's view, Leibniz's Law is preserved; the Morning

SUBSTANCE AND SEPARATION IN ARISTOTLE

I have described a view that explains away referential opacity by the

REFERENTIAL OPACITY IN ARISTOTLE

to say, is 'the musician builds' supposed to be false or merely, from the

18 Irwin (1986), 73.

SUBSTANCE AND SEPARATION IN ARISTOTLE

REFERENTIAL OPACITY IN ARISTOTLE

musical Coriscus, which is something objectlike. Similarly, what ceases

SUBSTANCE AND SEPARATION IN ARISTOTLE

REFERENTIAL OPACITY IN ARISTOTLE

kind) seated thing, then Socrates-qua-one-seated is a specimen of the

SUBSTANCE AND SEPARATION IN ARISTOTLE

of essential sameness, sameness in definition is required to preserve

REFERENTIAL OPACITY IN ARISTOTLE

are accidents of the same thing

(Metaphysics V 7 1017a7-16; cf. 1015bl6-27,

In this section I have analyzed the relations of essential sameness,

SUBSTANCE AND SEPARATION IN ARISTOTLE

I have argued that certain failures of substitutivity in Aristotle can be

REFERENTIAL OPACITY IN ARISTOTLE