1. That Plato quite intentionally leaves us in the dark and forces every one of us to start putting the pieces of the puzzle together again and again in order to prevent a passive absorption of his doctrine, is the conclusion to be drawn from the famous passage which explains the inferiority of the written against the spoken word in the Phaedrus (274b–7a).

2. An example of such a hint can be found in the Euthyphro where Socrates gives us a clue as to what all gods love: what they regard as noble, good and just (7e6: “…”); or in the Meno the distinction between knowledge and opinion as Daedalus' flightly children who need to be tied down (97d). In Rep. 1 the ‘functional argument’ (cf. 352eff. on the specific ergon and the proper virtue) at least foreshadows the later definition of justice as ‘doing your own thing’. It is not difficult to detect such hints; why Plato does not pursue them must remain a matter of speculation.

3. For a discussion of Plato's own explicit treatment of the conception of aporia cf. Mackenzie, M. M., ‘Impasse and explanation: from the Lysis to the Phaedo’, AGph. 70 (1988) 15–45; esp. 15–24CrossRefGoogle Scholar.

4. I have to gloss here over the troublesome details of Plato's distinction between the souls' faculties. For a detailed analysis of Plato's treatment of sensation and the conception of a unified perceiving consciousness see Burnyeat, M., ‘Plato on the grammar of perceiving’, Classical Quarterly 26 (1976) 29–51CrossRefGoogle Scholar; Frede, M., ‘Observations on perception in Plato's later dialogues’, Essays in ancient philosophy (1987) 3–8Google Scholar. I share the view that Plato treats the senses as devoid of any power of judgment. The opposite view that perception involves at least the ‘labelling’ of the percepts is defended (with reservations) by Cooper, J., ‘Plato on sense-perception and knowledge, (Theaetetus 184–186)’, Phronesis 15 (1970) 123–46CrossRefGoogle Scholar. This view is based on the assumption that we always immediately label our sensations, such as ‘red’ or ‘bitter’. Against this assumption it should be pointed out that the habitual labelling may be undeniable for such routine experiences but not, eg., for nuances of colours (reseda-green), unusual tastes, and not at all for sounds. Few of us are able to label a sound as ‘b-flat’ immediately!

5. For this achievement Theaetetus gets high praise. He is called beautiful, not ugly (185e3), because he liberates Socrates from the heavy task of having to argue for the dichotomy between what the soul does by itself and what it does through the capacities of the body.

6. That only highly abstract terms are mentioned may be a matter of sheer argumentative economy: Socrates would, at this point, find it hard to convince Theaetetus that less abstract concepts like ‘dog’ or ‘colour’ are not the product of perception but of the mind ‘by itself’. Plato is silent about the question where these concepts belong. Perhaps we may assume that we learn the names ‘dog’ or ‘colour’ through perception but not the concepts themselves. The problem is compounded by the fact that Plato does not tell us in the Theaetetus what faculty procures a synthesis of a complex perceptible object.

7. He even apologises for the fastidiousness and verbal precision necessary in this investigation (184c).

8. It has been debated whether sensation is a mere passive affection of the soul. For this claim cf. the arguments by M. Frede (n.3). Against it cf. Burnyeat (n.4), who points out why it is not necessary to read Plato this way (cf. p.42f). Against Burnyeat's argument that we intentionally employ our senses to gain access to and to investigate sensible qualities (43), it should be noted that the employment is not determined by the affection itself. Nothing hinges on this question for my further interpretation, but it should be noted that since the soul in sensation seems merely to receive the pathēmata, a passive reading would seem more appropriate. The difficulties for the view that Plato already assigns rudimentary judgments to the senses (‘this is red’) come to the fore in Modrak's defence (‘Perception and Judgment in the Theaetetus’, Phronesis 26 (1981) 35–54CrossRefGoogle Scholar). She ascribes an implicit possession of concepts to the senses (cf. 43: ‘the notions are indifferently and incompletely given in the perceptions of all senses’; 44: ‘It uses concepts that are implicitly given in the sensuous representation’). There is, however, absolutely no textual evidence for any such ‘implicit containment’ of concepts in the senses, and all evidence speaks for the view that the judgment ‘this is red’, since it asserts being, is a function of the mind itself.

9. On the much-debated question of the meaning of‘being’ (ousia) in this passage cf. Burnyeat (n.4) 44–5. I follow his recommendation that when Plato says that the mind determines ‘being’ or ‘not-being’ of perceptions he should be understood in the sense that ‘each of them is and is not because they are and are not various things’. At times it seems best to translate ousia by ‘nature’ as at 186b7: the nature of their contrariety, i.e. what it is. Cf. also Kahn, C., ‘Some philosophical uses of “to be” in Plato’, Phronesis 26 (1981) 105–34, esp. 119–27CrossRefGoogle Scholar.

10. The extreme Heraclitean theory of sense-perception is here treated as discarded. Socrates now assumes that perceptions have a definite nature and put us in touch with independently existing objects.

11. This revokes Socrates' assumption at 179c that with respect to the actual affections both perceptions and doxai might always be true. Perceptions seem to remain incontestable (materially), while the doxai depend on the correctness of the analogismata.

12. I will not make any decisions about the difficult question whence the soul gets the concepts of reflection, since Plato does not tell us anything about them beyond the fact that the mind has them and uses them. For a criticism of Cornford's and Cherniss' claim that the contemplation of the Forms of Being, Likeness … themselves is at stake cf. Cooper 125/6.

13. Cf. Farm. 129d–e: ‘likeness and unlikeness, plurality and unity, rest and motion and all that’. Being, Sameness, and Difference are also said to be parts of the world-soul in Tim. 37a–c, and its capacity to reason rests on them. The differentiation of sameness and difference is called the ‘road to reason’ at 44a–c. No further help will be sought from these parallels in other dialogues for the interpretation of the Theaetetus, not only because this would go beyond the bounds of a paper but because each dialogue should, as far as possible, be treated as a self-contained and -explanatory unity.

14. ‘All but a definition’ may be an overstatement. At best we have here a formal indication that knowledge must consist in the reflective employment of the common concepts; it does not tell us anything about the kind of employment, the extent or the degree of exphcitness that would guarantee the achievement of ‘being and truth’. As will be seen later this question is indeed crucial for the problem of the distinction between knowledge and opinion. Apart from the crucial passage at 186b6–9 which indicates that Plato has a systematic procedure in mind, he also gives at least a hint that progress has been made towards the depiction of epistēmē at 187a: ‘We have made enough progress to stop looking for it in perception altogether and look for it in whatever one calls what the mind is doing when it is busying itself, by itself, about the things which are.’

15. Cf. n.7. That there is indeed a wide gap between full knowledge and perception, and in what sense the distinction between knowledge and true opinion fills this gap, will emerge in the further discussion of the difficulties that stand in the way to a proper distinction.

16. Socrates in fact seems to check Theaetetus at this point (186b2: ‘hold off’ – “”), who takes it that the ‘calculations’ are extrapolations from the past and present to the future, since future states of affairs cannot be grasped by perception. Theaetetus sees here a reference to the question of the knowledge of future states of affairs discussed earlier (178c–9b), but Socrates is not concerned with such practical calculations but with that concerning the ousia of the perceived qualities.

17. The article by Bedu-Addo, J. T., ‘Plato on the object of knowledge’ in Gotthelf, A. (ed.) Aristotle on nature and living things (1985) 301–11)Google Scholar, argues also for such a double reading. He takes it, however, that Plato clearly wants us to opt for the alternative which consists in ‘pure reasoning’ (with a Kantian flavour, 309) about the ‘real nature’ of the objects of the mind, and sees therefore no difference between the koina and the Forms of the middle dialogues. His exclusive identification of ousia with ‘real nature’ seems to be misguided, however, and leads to a very one-sided reading of the text.

18. Paideia certainly need not be ‘the way out’ of the Cave (Rep. 514a; 518b7); but it is hard to believe that Plato meant by the paideia that helps achieve ousia and ōpheleia the bare learning of speech. Earlier in the dialogue Theaetetus had identified paideia with the liberal arts (145a; cf. also the distinction between paideia and the demiurgic' disciplines in Soph. 229d2, suggested by Theaetetus himself!). The employment of mogis reminds one of Socrates' claim in Rep. 7 that the Form of the Good is ‘just barely seen’ – (517c1, cf. Phdr. 248a4 where it is said of the soul that it is ). It is therefore most likely that Plato has the paideia in mind which characterises the philosopher who does not spare any trouble to find truth, in the excursus on the true philosophic nature (172–7).

19. Socrates even indicates that he thinks they ought to ‘wipe out everything that had been agreed on before’ (), 187b1.

20. Socrates sometimes indicates when they have made progress, cf. (187a ) The Wax-Tablet is vindicated later as constituting true (and thus by implication also false) opinion, but not more than that (290c).

21. The ‘all-or-nothing position’ underlies Meno's paradox. It is exploited in the Euthydemus.

22. Cf., eg., Phaedo 76b5. Most significant is the fact that in the Theaetetus itself the ability to give an account is mentioned as a decisive characteristic that distinguishes the true philosopher from the forensic orator, the man ‘with a small and sharp mind’ (175c).

23. More will be said later concerning the attribution and interpretation of the Dream-Theory.

24. See the account of dialectic in Rep. 535b/c and Phdr. 263bff.

25. Cornford, F. M., Plato's theory of knowledge (1935) 102–9Google Scholar. For a critical discussion of this view cf. Cooper (n.4) 123–6.

26. To keep the discussion within reasonable limits only those contributions are mentioned of the extensive secondary literature that are directly relevant to this interpretation.

27. The term koinon is not used in that sense again. For what is called koinon in the final argument (208d8; 209a10) are common, as opposed to differentiating, properties.

28. That Theaetetus is not quite taken in is indicated at 191b. It may indeed be often his easy temperament and readiness to learn (144a) that makes him go along with Socrates instead of insisting on his own, often quite justifiable insights.

29. For a detailed discussion cf. McDowell, , Plato's Theaetetus (1973) 194–209; 235–9Google Scholar. McDowell, however, goes on the assumption that Plato did not see the fallaciousness of his own arguments, although he allows that there are indications that Plato ‘is not happy with them’, (208).

30. Initially it is supposed to improve the deficiency of (1) and (2) by explaining how allodoxia, ie. thinking that one thing is another, is possible, but it ends by asserting its impossibility.

31. If Plato is concerned with the conditions of applying common terms it is clear why he is dealing with terms of judgement rather than states of affairs, as McDowell (n.29) expresses it (198).

32. That it is the unclarity about a thing's identity that causes the trouble seems indicated by the phrases and (188a7; b8). It leaves room for the possibility that I may know the object under one description but not under another, or that I may not recognise it in the present circumstances, while being otherwise quite familiar with it.

33. This problem is not really cleared up in the Tht.; but the indication of an ambiguity in ‘not being’ (188d9/10) points forward to the solution in the Sophist. Plato presumably wanted to provoke some reflections here on the question in what way ‘thinking what is not’ differs from ‘seeing or hearing what is not’.

34. A more sophisticated depiction of the problems in modern jargon (de dicto and de re, referential opacity) has been worked out by Bostock, D., Plato's Theaetetus (1988) 161–76Google Scholar.

35. The discussion focuses almost exclusively on mistakes of identity (the exception is the second paradox which deals with the ambiguity of being and not-being). It is not clear whether Plato concentrated on them because he thought them the most obvious, easiest to demonstrate, or whether he thought that all judgements involve the establishment of identity before any of the other koina can be applied. Establishing the identity of an item would then always be the first problem. Cooper (n.4) 134 n.14) expresses some puzzlement as to what Plato means by self-identity. Plato may have had no more than the identification of the object in mind; but he was also aware of other problems, such as the question of identity over time. Is, eg., the Theaetetus who has outgrown Socrates the same as he was before? For sameness and difference in change cf. Tim. 35a–36d.

36. In the Sophist the doctrine that a dianoia is the result of a silent dialogue is treated as established (263e–4a).

37. One difficulty lies in Socrates' not spelling out explicitly whether the mistake is propositional or not. Cf. Lewis, F., ‘Foul play in Plato's aviary’, in Lee, et al. (eds.) Exegesis and argument (1973), 262–84Google Scholar.

38. If the reader is somewhat relieved to see Socrates debunk the Aviary then this is largely due to the theoretical implications of this model: it seems to presuppose a very primitive theory of predication, as some interpreters have suggested, that Plato was to overcome only in the Sophist (cf. McDowell (n.29) 219ff.). But I do not think that this interpretation can be right, if it assumes that Plato regards all cases of false judgement as cases of mistaken identity. For the list of common concepts suggests that he envisaged quite a host of mistakes: mistakes about being, sameness, difference, likeness, unlikeness, number, goodness, beauty, usefulness.

39. Cf. 200c/d. I regard this as not a mere subterfuge to end the unsuccessful discussion.

40. For a discussion of this passage cf. McDowell (n.29) 228–31. That Theaetetus only now remembers this important definition can be explained as a case of recollection prompted by Socrates' criticism of the forensic orator's inability to procure knowledge: Socrates had mentioned his inability to ‘teach sufficiently’ the truth about the matter in his race against the water-clock (201b3). Theaetetus would have realised that a ‘proper account’ could make up for this deficiency (). The shift is thus not quite as impromptu as some interpreters have assumed but represents a classical case where anamnēsis is caused by the right kind of questioning. It must remain an open question whether for Plato any amount of additional logoi could turn the witness' testimony into knowledge.

41. For a comprehensive discussion of the Dream and the unlikeliness that Plato addresses another author cf. Burnyeat, M., ‘The material and sources of Socreates' dream’, Phronesis 15 (1970) 101–22CrossRefGoogle Scholar; McDowell (n.29) 234–9.

42. Socrates here (perhaps half in jest) uses the elements of his own description (202a2) to indicate that even this much cannot be properly said of the elements.

43. Cf. Burnyeat (n.41) 120.

44. Nor does, therefore, the polythrylēton apply here that Plato does not distinguish between ‘knowing by acquaintance’ (connaitre) and ‘knowing that’ (savoir) (cf. McDowell (n.29) 231ff.). You may claim to ‘know’ any number of facts about the elements, according to the Dream Theory, without knowing them. Socrates does not forbid Theaetetus to say other things about them, as indeed he does soon afterwards; he says that S belongs to the consonants and is ‘only a sound’, 203b; both esti and monon had been excluded in 201e/202a as ‘adding something’.

45. Cf. the article by Fine, G., ‘Knowledge and logos in the Theaetetus’, Philosphical Review 88 (1979) 366–97CrossRefGoogle Scholar. She realises that the salient point is that a logos must be proprietory (oikeios) but wrongly supposes that this means that the logos must say something ‘towards analyzing or fully explaining it’ (374). No theory of what a full explanation would have to be like lurks behind the Dream Theory. It merely rules out that anything oikeion can be said about the elements either by (any) attribution or by pointing to parts.

46. The list of attributes ruled out by the Dream Theory is only partially identical with the common concepts listed in 184–6; but this need only indicate that the misunderstanding is not specifically turned against Plato's own candidates as alien additions (although being and not-being stick out as prime candidates), but against anything that can be regarded as ‘common’, even pronouns.

47. Any interrelational account at this point would be ruled out a fortiori by our inability to give an account by even as much as referring to the elements, let alone by differentiating them from one another, in the way Plato had described as necessary for the proper employment of the common concepts. Theaetetus rightly recognises that his descriptions of the letters (203b) cannot, given the conditions of the dilemma, be regarded as logoi. The establishment of an ‘interrelational’ account of ‘account’ that Fine sees in this passage (380) is here then at best foreshadowed.

48. An example for such a nexus would be the thesis of the letters (206a7). One could not attribute ‘common descriptions’ like ‘this next to that’ to describe the order of the letters in the syllable in order to define it. In defending the soundness of the dilemma I do not subscribe to the fairness of all its moves; especially the claim in the mathematical example that the sum and the whole are the same (204b/c) is highly dubious. Plato may have wanted to signify by its very dubiousness that once the Dream Theory is discarded, definitions in terms of constitutive parts can be resurrected (cf. McDowell (n.29) 244; 249).

49. Cf. esp. (202a5) with the phrase at 186a3, where it is said of being that it ‘goes with everything’ as if it were a separate item (.

50. Socrates and Theaetetus are quite aware of the dire consequences of the addition-theory, for Theaetetus agrees that despite his ‘definition’ of S as a hissing consonant, and of B as a mute one, they remain aloga (203b).

51. The method of teaching consists in the identification of each letter in a word by pointing to them in succession. We thus avoid beginner's confusion which arises when they are spoken or written in connection (thesei: we should remember the extra confusion caused by the fact that there was no space between words in ancient Greek). Socrates' depiction of the instruction in spelling (, 206a6/7) shows more clearly than that in music that the one-by-one diagnosis is primary. It is therefore a mistake if Fine regards this description as a justification of the ‘interrelational method’ (385/6). At stake is not yet the system of the notes on strings but the name of the note on each string. How the notes are combined in the different modes is a much more complicated affair, as everyone knows who has tried to study the arcana of Greek music.

52. Cf. Phil. 17a–18d; Soph. 252e–3a; a technē is necessary to determine what combinations are possible.

53. Cf. 206b5 11 cf. a6; b7).

54. A different interpretation is attempted by Desjardins, R., ‘The horns of the dilemma’, Ancient Philosophy 2 (1979) 109–26Google Scholar. Her ‘conciliatory’ solution seems to me untenable. Plato does not call what unifies the whole its ‘differentiating form’ (120; nor can one see that ‘true opinion’ contains the elements of knowledge while the ‘account’ contains the differentiating form (of the true opinions?) (12); nor does Desjardins show how this should help to improve Plato's last example: how we can know (vs. truly believe) that this is Theaetetus on account of his snub-nose.

55. The necessity to establish systematically the active and passive capacities of the subject matter in order to determine the nature of something is also stressed in Phdr. 270d (, …).

56. Cf. 208c7 .

57. I agree here with G. Fine's position that no particular true statement taken in isolation can be regarded as knowledge (n.45) 95 et pass.

58. Reflection on the Forms is in fact concerned with the application to sensibles already in the Phaedo (74aff) and the Republic (cf. ‘the fingers’ 523aff).

59. If this sounds like an over-interpretation of a light remark we should keep in mind Plato's assertion in the Sophist (227a–6), that nothing is too little to interest and serve the true dialectician. There is no reason why Plato should not ascribe the same type of technical knowledge to the draughtsman that he grants to the musician and the linguist.

60. If one applies these possibilities to the seminal discussion of the koina at 184–6, one can argue for a triple entendre instead of the double entendre advocated above. There is (1) the ordinary dialogue which unreflectively and ‘unaccountably’ employs the koina, and does not achieve knowledge but only true opinion. There is (2) the dialogue of the specialist which reflects on the common concepts insofar as they apply to the special science. And there is (3) the dialectician's dialogue which reflects on the common concepts and their interrelations themselves, as discussed later in the Sophist (cf. 253b–e: ).

61. At the very worst one can say that Theaetetus' looks represent a philosophically unexciting example since it does not incite us to further questioning, just as the finger in the example of the Republic does not (523c).

62. This was explicit in the Aviary, cf. 197e3; 198a1; a7; 199b2. It is rather significant that Socrates there speaks of knowing arithmetic as having ‘knowledges’ – , 198b1.

63. For their special, technical, way of handling their subject matter, cf. 207b3: , 207c2: … ….

64. Cf. the criticism of Fine's interpretation by D. Bostock (n.34) 216–18; 243–50. Fine has to import some of the principles from other dialogues, e.g. her KL condition, i.e. of what constitutes a proper logos. She does not realise that the interrelational account is quite justifiable if one takes it as an explication of what Plato had indicated in 184–6: the need for a proper and systematic employment of the koina.

65. For a discussion that covers the same ground and emphasises the open-endedness of Plato's method cf. the monograph by Heitsch, E.. Überlegungen Platons im Theaitet, (1988)Google Scholar and the same author's article ‘Platons Dialoge und Platons Leser’, Rheinisches Museum 131 (1988), 216–38Google Scholar. They came into my hands too late to permit a fruitful comparison.

66. An ancestor of this paper was read as part of a panel at the convention of the American Philosophical Association in Portland/Oregon in March 1988. Later versions were presented to the Oxford Philological Society and to the Cambridge Philological Society. The paper has greatly benefited from the discussions. I want to thank all those whose penetrating questions have forced me to tighten my arguments and to strengthen my evidence, especially Jonathan Barnes, Lesley Brown, David Charles and Michael Woods in Oxford; and Myles Burnyeat, Geoffrey Lloyd, M. M. Mackenzie and David Sedley in Cambridge.