1. Some dispute that Forms are, or are only, universals; with Aristotle, I assume that Forms are at least universals. Though all Forms are universals, the converse is not true; Forms appear to be a subclass of universals – non-sensible ones. (I assume that not all universals are nonsensible – redness, e.g., is not.) I discuss some features of Forms in more detail in my ‘The One over Many’, Philosophical Review 89 (1980) 197–240CrossRefGoogle Scholar; and in ‘Relational entities’, Archiv für Geschichte der Philosophie (1983)Google Scholar.

2. I thus align myself with those who believe that in the Met. Aristotle acknowledges individual forms, and counts them as his primary substances. This view is, of course, highly controversial; and my defense of it here is at best partial. That the primary substances of the Met. are individual forms is also defended by Hartman, E., Substance, body, and soul (1977)Google Scholar, esp. chapter 2; and Sellars, W, ‘Aristotle's Metaphysics: an interpretation’, in Philosophical perspectives (1959) 73–124Google Scholar. If this view is rejected, then many of the things I say about Aristotle's criticism of, and alternative to, Plato would need to be revised. On the other hand, if I sketch Aristotle's criticisms correctly, that supports the picture I paint of his alternative. I do not deny, by the way, that Aristotle countenances forms (or entities called ‘eidē;’) other than individual forms (see esp. n. 7). But unless otherwise noted, whenever I speak of Aristotelian forms, I shall mean the individual forms that count as primary substances.

3. Cousin, D. R., ‘Aristotle's doctrine of substance’, Mind, n.s. 42 (1933) 319–37; 43 (1935) 168–85CrossRefGoogle Scholar; Dancy, R., Sense and contradiction: a study in Aristotle (1975) 95ffCrossRefGoogle Scholar.

4. It is, of course, sometimes disputed that Plato accords sensibles essences or that, if he does, their essences are Forms. For a partial defense of the view assumed here, see ‘Relational entities’.

5. Cat. 2b7–14; 2b29–37. A second reason they count as secondary substances is that they are the second best subjects. I discuss both reasons further below, in III and V.

6. I say ‘at best’ because not all essences count as secondary substances; only the essences of primary substances do. Redness, or color, are the essences of various nonsubstances; but they are not substances at all.

7. ‘eidos’ is one word Plato uses for PFs. Aristotle uses ‘eidos’ in at least the following ways: (a) for individual forms (e.g. Socrates' soul); (b) for species (e.g. man); and (c) for the universal form or essence of that species (e.g. rationality). This multiple usage need not import any confusion. The distinction between (b) and (c) is made much of by Driscoll, J., in ‘EIDE in Aristotle's earlier and later theories of substance’, in Studies in Aristotle, ed. O'Meara, D. J. (1981) 129–159Google Scholar. He argues that in the Met., the primary substances are eidē of sort (c); I favor (a). Driscoll is right to argue that (a) is not the only alternative to (b), which he agrees is not primary substance in the Met.; but there are other reasons for preferring (a) to (c).

8. Owen, G.E.L.; ‘The Platonism of Aristotle’, Proceedings of the British Academy 51 (1965) 125–50, 136–7Google Scholar. In a later article, ‘Particular and General’ PAS 79 (1978–1979), esp. 14–15Google Scholar, however, he seems to have retracted this view in favor of individual forms, though the claim is not pressed.

9. By, e.g., Woods, M. J., ‘Substance and essence in Aristotle’, PAS 75 (1974–1975) 167–80Google Scholar; cf. also Kirwan, C., Aristotle's Metaphysics: Books Γ, Δ, and E (1971), esp. 100–101Google Scholar.

10. Cf. Ackrill, J. L., Aristotle's Categories and De Interpretatione (1963) 89–90Google Scholar. The translation is also Ackrill's; all other translations are my own.

11. Thus, in Z.4 (1030a5–6) he says that only substances are thises; and in Z.3, matter is ruled out as (the sole) substance on the grounds that, inter alia, it is not a this (1029a28–30). The notion of a tode ti is difficult, and has been explicated in many different ways. Cf., e.g., Smith, J. A., ‘TODE TI in Aristotle’, CR 35 (1921) 19Google Scholar; Owen, , ‘Particular and general’, 2Google Scholar; Barnes, J., Aristotle (1982), 43Google Scholar. Barnes remarks that tode ti is ‘an unorthodox phrase which Aristotle nowhere explains’.

12. I am unsure whether (1) is tacit in the Met. Perhaps it plays a role in the restriction Aristotle sometimes makes of substances to natural substances, i.e. to those with internal sources of change. In any case, I shall largely ignore (1) when speaking of the Met.

13. I take both ‘chōris’ and ‘chōriston’, as used in connection with substance and PFs, to indicate actual, and not merely possible, separation; and I take actual separation to be the capacity for independent existence. Most of the claims I make in this paper about separation are defended in detail in ‘On the separation of Platonic Forms’, forthcoming in Oxford Studies in Ancient Philosophy (1984).

14. At 1028a32–3, I read: kai physei kai logō(i) kai gnōsei – adding ‘physeī’ to, and deleting ‘chronō(i)’ from, Jaeger's text. This emendation is not essential to my view.

15. I say that in the Cat., Aristotle denies that universals are separate. To be sure, this follows from his remark that universals are ultimately dependent on primary substances; but it is interesting to note that Aristotle does not use ‘chōris’ or its cognates in pressing this claim. Notice, in this connection, the account of natural priority in Cat. 14b10–22. Here, in striking contrast to Met. Z.1, A can be naturally prior to B without being separate from B; indeed, separation is actually precluded, since natural priority is defined so as to obtain only between items that reciprocate as to existence. In general, the Cat. has considerably less to say about separation than the Met. does. ‘Chōris’ and its cognates occur only once, at 1a25.

16. Such, at any rate, appears to be the implication of Z. 1. Even if (5) is only necessary, however (and it is clearly at least that), the problems I raise for Aristotle over (5) would remain (see below, V (5)). One might find it tempting to weaken (5) so that it requires only that ousiai be mentioned (as opposed to defined) in every definition. I know of no place where priority in definition is explicated in this weaker way, however. But even if it is, neither would this remove the difficulties canvassed below; for substance must still be definable, even if its definition need not actually occur in definitions of other things, and that is sufficient for the worries I later air. Analogous remarks to those made in this note apply to (6).

17. Aristotle, of course, believes that Plato intends PFs to be efficient causes of change (Met. 991b3–9 = 1080a2–8; cf. G.C. 335b7–15, 18–24), and he criticizes them on that score. Aristotle has recently been defended by Annas, J., ‘Aristotle on inefficient causes’, Philosophical Quarterly (1982) 311–326CrossRefGoogle Scholar. But I was unpersuaded by her argument, which seems to assume that if Plato does not explicitly distinguish between efficient and other sorts of causes, he must be confused about their differences. A more sympathetic reading of Plato is provided by Vlastos, G., ‘Reasons and causes in the Phaedo’, Philosophical Review 78 (1969) 291–325CrossRefGoogle Scholar.

18. Owen, , ‘Particular and general’, 2Google Scholar. This is Owen's explication of what Aristotle means by ‘subject’. If Owen is right, then PFs are subjects in Aristotle's sense; but I doubt that Owen is right.

19. The issue is disputed; but for one plausible defense of the possibility of uninstantiated universals, see Butchvarov, P., Resemblance and identity (1966) 186Google Scholar–97.

20. See Owen, , ‘Particular and general’, 14Google Scholar; contrast the passage cited above from ‘Platonism’.

21. ‘Self-predication’, as traditionally construed, is the claim that the Form of F is itself an F thing, a member of the class of F things; the Form of Large, e.g., is itself large. If S. Peterson's account of self-predication is accepted, there would be even less reason to suppose that self-predication implies particularity. See Peterson, S., ‘A reasonable self-predication premise for the Third Man Argument’, Ph. R. 82 (1973) 451–470Google Scholar.

22. Indeed, the Met. account of thisness has Platonic antecedents: cf. Crat. 439c6–140cl; Tht. 181–3; So. 261–3; Tm. 48e2–52d 1. For some discussion, see Zeyl, D., ‘Talk of a world in flux’, HSCP 79 (1975), esp. 146–8Google Scholar; Driscoll, J., ‘The Platonic ancestry of primary substance’, Phronesis (1979) 253–269Google Scholar. Interestingly, Plato never seems explicitly to say that PFs are thises, so conceived; and like Aristotle he at least sometimes uses similar phrases of sensible particulars. But his notion of being a this is not defined in terms of sensible particulars; and PFs satisfy his notion, even if he does not himself explicitly make the point.

23. See esp. Met. M.9; I discuss the passage in detail in ‘Separation’.

24. Aristotle may, however, intend to foreclose this possibility in Met. Z. 13.

25. I assume mereological essentialism is false.

26. Notice that this problem for ASs is not alleviated by claiming that they are universals, not particulars; for as we have seen, Aristotle consistently denies that universals are separate, and he consistently associates separation with particularity. On the other hand, T. H. Irwin has suggested to me that ASs – conceived as individual forms – are, contrary to my suggestion, separate, at least from universals: whenever Aristotle defines universals, he defines them in terms of being predicated of many things. Suppose that only Socrates – one man – exists. Then the universal, man, does not exist; its existence requires the existence of many men. But if Socrates exists, and the universal, man, does not, then Socrates is separate from man. A parallel argument shows that Socrates is separate from all universals. He is not, of course, independent of, e.g., his individual pallor or height; but perhaps these are not other than him, and so he need not be separate from them. (An analogous argument might be used to show that neither is he separate from his matter; for he is not other than it in the requisite way.) But if Socrates is separate from all universals so, too, is every other man and, analogously, every other AS.

I am unpersuaded by this argument, however. Even if it is accepted, it gives us no reason to prefer ASs to PFs; it shows at most that universals and particulars are on an equal footing.

27. This might actually be disputed. Suppose, as is sometimes believed, that there are Forms only corresponding to the accidental properties of things. Then even if Socrates is, e.g., tan, in virtue of participating in the Form of Tan, he is not dependent on the Form, since he can exist without it; he would of course then cease to be tan, but that would not impugn his continued existence. However, at least in the Timaeus there are Forms corresponding to some of the essential properties of things (there is, e.g., a Form of Man), and so at least there sensibles do depend on (some) Forms: Socrates could not exist if he were not a man; hence he is dependent on the Form of Man, since he can be a man only by participating in it. I discuss this issue more fully in ‘Relational entities’.

28. Notice that if separation here is definitional separation (and not IE), then Aristotle apears to believe that even definitional separation implies particularity. And at least one line of thought does yield this result: in Z. 1, Aristotle argues that every definition must include a definition of ousia. If all ousiai are substance particulars, then definitions of everything else, including universals, must include definitions of them. But then no universal can be definitionally separate; each must include a definition of a substance particular. Hence if something is definitionally separate, it is a particular. I rather suspect, however, that separation here is IE.

29. This is, of course, a controversial account of Aristotle that I do not have the space to defend here.

30. Hume, D., A treatise of human nature, ed., Selby-Bigge, L. A. (1888), Book I, Pt. 1Google Scholar; Enquiry concerning human understanding, ed. Selby-Bigge, L. A. (1902), section 2Google Scholar.

31. Leibniz, G. W., New essays on human understanding, translated and edited by Remnant, P. and Bennett, J. (1981), Book I, chapter 1, passimGoogle Scholar.

32. An earlier version of this paper was read at the Oxford Philosophical Society in November, 1982; I thank the audience on theat occasion for a number of helpful comments. I especially wish to thank my commentator, Professor J. L. Ackrill, for generous and helpful written and oral comments; and also Terry Irwin, M. M. Mackenzie, and Jennifer Whiting. I am also grateful to the National Endowment for the Humanities for awarding me a fellowship for 1982-3, and to the Principal and Fellows of Brasenose College, Oxford, for electing me to a Visiting Fellowship for 1982-3.