¹ E.g. Fine, G., ‘Plato on naming’, PhQ 27 (1977), 289–301Google Scholar; Anagnostopoulos, G., ‘Plato's Cratylus: the two theories of the correctness of names’, Rev. Met. 25 (1971–1972), 690–736Google Scholar; Ketchum, R. J., ‘Names, forms and conventionalism: Cratylus 383–395’, Phron. 24 (1979), 133–47CrossRefGoogle Scholar; Kretzmann, N., ‘Plato on the correctness of names’, APhQ 8 (1971), 126–38Google Scholar; Schofield, M., ‘The dénouement of the Cratylus’ in Language and Logos, ed. Schofield, /Nussbaum, (Cambridge, 1982), 61–81CrossRefGoogle Scholar. Contrast the line taken by Gosling, J., however, Plato (London,1973), 200Google Scholar, who characterises the arguments as ‘polemical’.

² Kahn, C. H., ‘Language and ontology in the Cratylus’ in Exegesis and Argument, ed. Lee, /Mourelatos, /Rorty, (Assen, 1973), 152–76Google Scholar and Weingartner, R. H., The Unity of the Platonic Dialogue (Indianapolis, 1973), 15–43Google Scholar both suggest that the middle-period theory of forms is anticipated in the Cratylus. The general view appears to have been that the dialogue is early: e.g. Guthrie, W. K. C., Socrates (Cambridge, 1971), 122CrossRefGoogle Scholar suggests that it is Socratic. Ross, , Plato's Theory of Ideas (Oxford, 1951), 4Google Scholar says that the date of the dialogue is ‘open to serious doubt’, but tends towards a pre-Republic date, following Von Arnim, Raeder and others. We may compare Crombie, I. M.'s compromise, An Examination of Plato's Doctrines (London, 1963), ii. 376Google Scholar. Ryle, , however, Plato's Progress (Cambridge, 1966), 2Google Scholar, assumes that it belongs with other critical dialogues.

³ Cf. e.g. Weingartner, op. cit. 38.

⁴ I have further argued, in ‘Paradox in Plato's Phaedrus’, PCPS, n.s. 28 (1982), 64–76Google Scholar, that the Phaedrus proposes paradox-mongering as a serious procedure for inquiry, which is then carried on into later dialogues in practice. In what follows, I attempt to show that the Cratylus is a case in point.

⁵ Cf. above, n. 1.

⁶ This, and its parallel for Cratylus, is observed by Williams, B. A. O., ‘Cratylus’ theory of names and its refutation’ in Language and Logos, 83–94Google Scholar.

⁷ Though cf. Schofield, ‘A displacement in the text of the Cratylus’, CQ 22 (1972), 246–53CrossRefGoogle Scholar, on this passage and its place in the argument. My own interpretation requires that 385b–c is in the right place in the text.

⁸ Cf. e.g. Democritus, DK 68B8; Gorgias 483ff.; Ar. Soph. El. 173a; and a passage rich in epistemological overtones, Pindar Ol. 2.83ff. I am grateful for the last reference to Penny Wilson.

⁹ L.S.J. explain πρόσσω κα⋯ ⋯πίσσω in terms of the future (behind us, and so inscrutable) and the past; but there seems no reason not to prefer a more literal interpretation here, with the phrase out of context.

¹⁰ We can, I take it, avoid needless Eleatic pitfalls here by insisting that ‘is’ is incomplete, equivalent to ‘is F’. Cf. Owen, , ‘Plato on Not-Being’ in Plato II, ed. Vlastos, (New York, 1971), 223–67CrossRefGoogle Scholar.

¹¹ It is, I take it, important to except these individuals from flux. Cf. below, pp. 138ff.

¹² Cf. Burnyeat, , ‘Protagoras and self-refutation in Plato's Theaetetus’, PhR 85 (1976)Google Scholar.

¹³ Cf. remarks by Burnyeat, , ‘Protagoras and self-refutation in later Greek philosophy’, PhR 84 (1975)Google Scholar.

¹⁴ Lee, E. N., ‘Hoist with his own petard’ in Exegesis and Argument, 225–61Google Scholar.

¹⁵ ‘Everyone’ or ‘in every respect’ — this text is as ambiguous as Phaedo 74b8; but I take it that the point is clear enough. Although the Greek of 386d4 would allow the translation ‘anything is likewise to everyone’, which would imply total stability, it must mean ‘everything considered altogether is likewise to everyone’ — total indeterminacy. For the sequel (d5–6) glosses: ‘virtue and vice are likewise to everyone always’ and is decisive for total indeterminacy. In my discussion of this passage below, p. 140, I shall find it convenient to symbolise this as a universal quantification over contradictory predicates, even though the gloss gives us contradictory subjects with the same predicate.

¹⁶ Complex enough for the Sophist cf. Fine, op. cit.

¹⁷ Where do the primary names come from? The gods? ’Ware god's eye view; cf. below. Barbarians? ’Ware ‘barbarian souls’; cf. Heraclitus fr. 107 (DK).

¹⁸ Binder, G. and Liesenborghs, L. ‘Eine Zuweisung der Sentenz (οὐκ ἔστιν ⋯ντιλέμειν) an Prodikos von Keos’ in Sophistik, ed. Classen, (Darmstadt, 1976), 452–64Google Scholar.

¹⁹ Cf. Theaet. 191a ff., the wax tablet. I suspect a strong connexion between the Cratylus and the Philebus (39a ff.) here.

²⁰ Thus at 430d6 the καί is inferential, on this reading — equivalent to ἄρα.

²¹ Echoed, I suggest, at Phil. 15a ff. Once again, I hedge my bets on dating, though I suspect the traditional late dating of the Philebus is correct. But cf. Waterfield, R., ‘The place of the Philebus in Plato's dialogues’, Phronesis 25 (1980), 270–305CrossRefGoogle Scholar.

²² But n.b. the controversies noted by Gosling, J., Plato: Philebus (Oxford, 1975), 185–206Google Scholar. For the view I suggest, it is crucial to observe two points: (a) that the only ontological commitment made in the passage is to the mixed class, composite of πέρας and ἄπειρον (vide the intrusion of ɸύσις at 25c11, the introduction of the composite; γενέσεις at 25e4, cf. 7–8; 26c7 etc.). (b) This is in contrast to the explanatory claim made for the πέρας/ἄπειρον pair. Reading back from the emphasis on explanation at 26e ff., we may conclude that material-sounding terminology (⋯ν τοῖς οὖσι, 26c6; ⋯κ 26d8; ⋯περμάζεσθαι, 25e–26a etc.) can in fact be cashed formally, as explanatory factors, not constitutive elements. If, then, we give due emphasis to Plato's interest in explanation, which need not require him to be any more profligate, ontologically, than Aristotle, then the formal aspect of the Philebus passage fits well with my present argument about Plato's discussion of meaning.

²³ Waterfield, op. cit., suggests that the Philebus conflates the polar approach to relations with the idea that relations are prone to excess (cf. 52c); whereas the Politicus differentiates the two. The latter passage does indeed make the requisite distinction. However, there is more: the Politicus distinguishes between the polar notion in speech (283d11) and excess in words and deeds (283e4). The latter, like the excessive pleasures of the Philebus, are certainly real — entities or events. The former, however, is a formal point, that when we say ‘larger’ we imply ‘than some smaller’, which tells us not only about the composition of the world, but about our conceptual structure. If Waterfield's early dating of the Philebus is rejected, we may see in the Politicus passage support for the view that the Philebus also contains this formal argument, that some terms are to be understood only in relation to a polar structure. This polarity, not some real indeterminacy, is their flux (Phil. 24d).

²⁴ Some irony, perhaps, in Socrates' remark about ‘coming to the truth later than we should have done’ (433b1)?

²⁵ Cf. above, p. 133; presumably ‘many’ will be predicates, although the realist late-learners must also deny a plurality of properties to any one thing.

²⁶ Cf. Owen, G. E. L., ‘Plato on Not-Being’ in Vlastos, Plato I (New York, 1971), 223–67CrossRefGoogle Scholar.

²⁷ Cf. Bumyeat, M. F., ‘The material and sources of Plato's dream’, Phronesis 15 (1970), 101–22Google Scholar, on dream imagery in Plato. It is important to notice that a dream could reveal an important truth; but equally it could postulate a false view of the world — that this hypothesis of the Forms is Socrates’ dream does not vouch for his commitment to it.

²⁸ Cf. e.g. Luce, J. V., ‘The theory of ideas in the Cratylus’, Phronesis 10 (1965), 21–36CrossRefGoogle Scholar.

²⁹ The text is tricky here; so is its interpretation. Burnet transmits: αὐτό τοίνυν ⋯κεῖνο σκεψώμεθα, μ⋯ εἰ πρόσωπόν τί ⋯στιν καλ⋯ν ἥ τι τ⋯ν τοιούτων, κα⋯ δοκεῖ τα⋯τα πάντα ῥεῖν ⋯λλ' αὐτό, ɸ⋯μεν, τò καλòν οὐ τοιοὐτον ⋯εί ⋯στιν οἷόν ⋯στιν; The grammar is awkward. Luce, op. cit. 33, wishes to delete κα⋯…ῥεῖν, and with it any reference to flux; so also Schanz. We do, however, have the flux hypothesis already on offer, 411 and 439c, so such a deletion is of little effect. Stallbaum and Méridier interpret the clause as a further hypothetical — that is (a) suppose there are particulars and (b) suppose them to be in flux. Jowett goes for epexegesis: suppose there are particulars — to be a particular is to be flux-ridden. Given my strictures against finding some ‘real’ Platonic doctrine lurking in the puzzles, either of the last two possibilities will do for the final argument to go through, using as its premiss the complex hypothesis that forms and particulars in flux are the (only) two sets of things that exist. Thus the final argument is based on the antithesis between completely stable entities and totally flux-ridden entities. So the conclusion, rather than ignoring the tertium quid of modified flux, might be seen to point to it, on the collapse of the original antithesis.

³⁰ ‘It’ refers back to some form, postulated at d5. 5 is inferred from 4 (αὐτó at d8); but the nature of the inference is extraordinarily obscure. Socrates' procedure, however, appears to be to suppose that some entity (not a particular, because they are already hypothesised as flux-ridden) can be such as to sustain total stability, partial stability/partial change (cf. 439e2), or total flux. He then explores the cognitive reliability of such an entity. ‘It’ here, then, could equally be understood as ‘something’, provided we recognise that this ‘something’ could be totally stable; on the hypothesis, this could only be a form.

³¹ The dialogue has explored the denial of falsehood, and the denial of dialectic that follows from that (438a ff.). Saying something correctly, then, is directly connected to knowing or understanding it. So the move from ‘saying correctly’ to ‘knowing’ which the argument makes so readily is warranted by the dialogue as a whole. Consequently step 7 may be understood as a separate strand in the argument, not a further supplement.

³² Cf. Theaet. 183a–b.

³³ Socrates moves swiftly but clearly from ‘if it ever is the same, at that time it does not change’ (e2) to ‘if it always stays the same, it never changes’. The possibility of partial stability — the tertium quid — is thus mentioned but not explored. The argument is operating, then, on the explicit disjunction of total stability and total flux.

³⁴ Here the subject is taken readily from the preceding sentence, and not — as others take it — from the one before that. The object in question is now characterised as totally stable (e3), and as such it is argued to be unknowable. οὐδέ at e7 is then to be read as intensified — ‘not even’ — or as answering δέ at e3. ⋯λλ⋯ μήν at 37 corresponds, not to π⋯ς ἄν at el, but to π⋯ς ἄν at e4, thus making sure of the connexion between e3–5 and e7.

³⁵ The verb used here is the aorist passive, implying being acted upon towards an end, rather than being in a completed state. Cf. Lyons, J., Structural Semantics (Cambridge, 1963), 112ff.Google Scholar

³⁶ The genitive absolute here could, as the traditionalists aver, simply have temporal significance: the object is always different, whenever the knower comes along. However, the causal inference I suggest — since the knower comes along, the object becomes different — is supported by the ἄμα of e7, since the simultaneity of knower coming along and the change in the object seems to be significant. There is nothing in the text, moreover, to rule out such a reading.

³⁷ This, of course, reminds us of the section on Euthydemus; cf. above, pp. 132f., below, p. 140.

³⁸ Here, for the sake of clarifying my reading of the argument, I have reordered the steps. In the text 11 is followed by 13, then 14 and finally 12. I expand below how I suppose these moves to develop an epistemological dilemma.

³⁹ The change of tense–future in 13, present in 14–emphasises the fact that coming to know is what is at stake here.

⁴⁰ The tense variations noticed above make it quite clear that the problem concerns coming to know — cf. discussion of the god's eye view, below; and compare the recollection solution at Meno 85d.

⁴¹ Cf. Burnyeat, M. F., ‘Socrates and the jury: paradoxes in Plato's distinction between knowledge and true belief’, PAS (Supplement) 1980,173–91Google Scholar.

⁴² Irwin, T. H., ‘Plato's Heracliteanism’, PhQ 27 (1977), 1–17Google Scholar; Nehamas, A., ‘Predication and Forms of opposites in the Phaedo’, Rev. Met. 26 (1972–1973), 471–91Google Scholar and ‘Plato on the imperfection of the sensible world’, APhQ 12 (1975), 105–17Google Scholar; Jordan, W. R., Plato's Arguments for the Forms (Cambridge Philological Society Supplement, 1983)Google Scholar.

⁴³ Thus Plato need not, to get his own arguments going, be committed to substantial flux. It remains true, however, that substantial flux, conversely, implies cognitive unreliability, as Theaet. 181–3 makes clear. So the compresence of opposites is a weaker problem than substantial flux.

⁴⁴ This is a common enough device in Plato — cf. e.g. Phil. 16c; 30d; 43a; but it is impossible to tell whether, in any of these contexts, the ancients are cited with approval or otherwise.

⁴⁵ For example, our own actions, 387a ff.; cf. below, n. 49.

⁴⁶ Theaet. 152d5 can be read as a problem of compresence; vide McDowell, J., Plato: Theaetetus (Oxford, 1973)Google Scholar, ad loc. However, the tense here is future, as opposed to the present tense of Phaedo 74b8; and, crucially, the problem is then explained by means of a theory of change — the theory of perception — which implies a diachronic problem.

⁴⁷ With the Timaeus we may compare the Philebus, which contrasts the fixity of the monads (15b) with the indeterminacy of particulars, and gives us a two-tier epistemology to match (62a). Beware the ‘indeterminacy’ of the Philebus, however — the particulars of 15b appear to be only numerically indeterminate; while the ἄπειρον of 24d is in flux, indeed, but perhaps only because it covers any polar spectrum, rather than describing some set of indeterminate real objects. Philebus 24d will be revisited below. On dating, cf. here Waterfield, R., ‘The place of the Philebus in Plato's Dialogues’, Phronesis 25 (1980), 270–305CrossRefGoogle Scholar.

⁴⁸ ‘Paradox in Plato's Phaedrus’.

⁴⁹ The argument about function, 387d ff., is generally construed as a teleological account of naming. It is, however, clearly associated with the claim that things have some fixed nature; and would provide us with direct, personal evidence for that without recourse to empirical data which are external to us, and so question-begging, on Euthydemus' thesis.

⁵⁰ That the Theaet. is not making a relativist point here is made clear by the naturalist arguments to support the secret doctrine at 153a–d.

⁵¹ Although cf. n. 67 below on stylometric considerations.

⁵² Cf. above, nn. 1–2.

⁵³ Cf. Owen, G. E. L., ‘Plato & Parmenides on the timeless present’ in The Presocratics, ed. Mourelatos, (New York, 1974), 290Google Scholar.

⁵⁴ Lofty or divine (σεμνόν) immobility — cf. Phdr. 275d6, an ironical term? Cf. below on the god's eye view.

⁵⁵ The connective at 248e6 indicates a break between two arguments.

⁵⁶ Cf. here Vlastos, G., ‘Am ambiguity in the Sophist: Appendix I’ in Platonic Studies (Princeton, 1973), 309–17Google Scholar on the enthymemes here.

⁵⁷ So, of course, would a modified version of the Theaetetus' theory of perception which insisted only on numerical non-identity between one event of perceiving and the next; Plato's silent assimilation of locomotion and qualitative change in both Soph. and Crat. is rendered suspect by Theaet. 181c–d.

⁵⁸ Cf. Vlastos, op. cit.; Owen, , ‘Dialectic and eristic in the treatment of the Forms’ in Aristotle on Dialectic, ed. Owen, (Oxford, 1968), 103–25Google Scholar; Keyt, D., ‘Plato's paradox that the immutable is unknowable’, PhQ 19 (1969), 1–14Google Scholar.

⁵⁹ Cf. Keyt, , ‘The mad craftsman of Plato's Timaeus’, PhR 81 (1971)Google Scholar.

⁶⁰ Cf. Geach, P., God and the Soul, 71ff.Google Scholar

⁶¹ Whether or not Plato actually believes this account of perception, it should be distinguished from his middle-period view where the properties of phenomenal objects, albeit compresent with their opposites, are seen as natural to the object (cf. Symp. 211 etc.).

⁶² Cf. Hackforth, R., ‘Notes on Plato's Theaetetus’, Mnemosyne (1957), 130Google Scholar.

⁶³ Cf. McDowell, ad loc.

⁶⁴ Cf. Nehamas, op. cit.

⁶⁵ The Second T.M.A., Parmenides 134, could be construed thus.

⁶⁶ Perhaps reflection on Plato's critical approach to the god's eye view might give us pause about Heraclitus: cf. Burnyeat, , ‘Reading Heraclitus’, N.Y. Rev. Books, 1982Google Scholar.

⁶⁷ Cf. the figures for the avoidance of hiatus used by Cherniss in ‘The relation of the Timaeus to Plato's later dialogues’ (in Studies in Plato's Metaphysics, ed. Allen, [London, 1965], 344ff.)Google Scholar. Here the group which includes Phaedrus, Parmenides, Theaetetus and Cratylus is strikingly different, in its failure to avoid hiatus, from Timaeus, Sophist, Philebus and Politicus. This is not, of course, to concede that stylometric considerations are or can be decisive in the dating of Platonic dialogues. After all, here we have a highly literate author, who may well achieve some of his effect by the deliberate echoing of the style of an earlier work. We cannot, that is, make any definite or plausible claims about where his development is unconscious and where he employs conscious allusions. It follows from this that the stylometric tests, which purport to examine unconscious development, beg the entire question of the method of Platonic composition.

* Parts of this paper were delivered at the Oxford meeting of the Southern Association for Ancient Philosophy in 1983. I profited greatly from the discussion that followed; and I should like to thank members of that audience, and also Jonathan Lear, Malcolm Schofield, and R. W. Sharples, for their helpful comments.