^{page 79 note 1} Sophistes and Politicus, introd.; essays in Republic (ed. Jowett, and Campbell, ), vol. ii; C.R. x, 1896, pp. 129–36.Google Scholar

^{page 79 note 2} Campbell and Lutoslawski, Raeder and Constantin Ritter have at different times written as though, even if Astdoes not list all occurrences of a word, he does name all the dialogues in which it occurs; this is quite false (cf., for example, p. 84, n. 4 infra). He does not even list all Plato's words.

^{page 79 note 3} Origin and Growth of Plato's Logic, chap. iii.

^{page 79 note 4} This was the sheet-anchor of stylo-metrists who were not content with such broad groupings of the dialogues as that accepted by Taylor (Plato, the Man and his Work, p. 19). Yet there is no external or internal evidence which proves that the Laws or even some section of it was later than every other work: cf. p. 82 infra.

^{page 80 note 1} Here the attempts of Schanz, Ditten-berger, and Constantin Ritter to measure the relative frequency of synonyms were theoretically sound. But a study of the Phaedrus (cf. p. 81 infra) proves that Plato adopted the ‘late’ synonyms in passages of elevated style earlier than elsewhere. In fact, when Plato is said to be dropping one synonym for another he is commonly bor rowing from poetry (Campbell, , Rep. ii, pp. 50–51Google Scholar), and to find these borrowings either in speeches for whose poetic vocabulary Socrates apologizes (Phdr. 257 as) or in a work ‘in Inhalt und Form mit der Poesie wetteifernd’ (Wilamowitz on the Timaeus) is obviously not the same thing as finding them in dialogue proper.

^{page 80 note 2} J. Philol. xxxv, 1920, pp. 225–56.Google Scholar

^{page 80 note 3} p. 250. The distribution of end-rhythms in the Tm. closely matches that in the middle and early dialogues. Thus the rhythms which are dominant (65–85 per cent.) from the Soph. digression onwards total 45·6 per cent. in Tm., the same in Crito, and 2–3 per cent. below in (for example) Phdo., Rep. 6 and 10.

The graph for later works is interesting (but to be used with care): in the Phdr. these rhythms steadily recede; the overall figure (37·7 per cent, or, omitting Lysias'speech, 36·9 per cent.) matches that of the first part of the Parm. (38·1 percent.); in the Tht. it rises, reaching 50 per cent, from Protagoras’ speech (165 e) with brief further rises (e.g. in the discussion of the ); the Crat., for those who want it here, is higher (52·4 per cent.), and thereafter the rise is steep. (My figures are approximate to the extent that Billig's rules for assessing interjections are not precise.)

^{page 80 note 4} Wiener Studien, xxvi, 1904, p. 190.Google Scholar

^{page 80 note 5} Platans Epinomis (1939), p. 13, n. 1.Google Scholar

^{page 80 note 6} Commentary on Plato's Timaeus, pp. 4–5.

^{page 80 note 7} Plato's Cosmology, p. 12, n. 3: cf. now Hackforth, , Plato's Phaedrus, p. 3;Google ScholarSkemp, , Plato's Statesman, p. 238.Google Scholar

^{page 80 note 8} Blass, , Att. Bered., p. 458.Google Scholar By Janell's count the figure for the Phdr. is little more than half that for the Parm. (23·9 and 44·1 per page of Didot, respectively).

^{page 81 note 1} Comm., p. 4.

^{page 81 note 2} For which cf. esp. Ritter, , Untersuchungen, pp. 2–33, 56–59Google Scholar (with corrections in Platan, i, pp. 236–7),Google Scholar 70, n. 1; Lina, , de praep. usu platan., p. 12;Google ScholarCampbell, , Rep. ii, pp. 53–55.Google Scholar But these critics draw no distinctions within the Phdr., and sometimes we shall correct their totals.

^{page 81 note 3} Lysias' speech, which I shall not consider, tallies by present tests with the dialogue proper.

^{page 81 note 4} Ritter's figures, after large corrections in Platan and articles in Bursian's Jahresbericht, remain untrustworthy: e.g. in Phdr. he under-estimates cases of (2 excluding 257 a 3), before a vowel (2, 1), etc., in rel. clause (2), (18).

^{page 82 note 1} e.g. in the coining of adjectives in and the Tht. and Parm. are characteristic of the late dialogues, and the Tm. of the middle period (Lutoslawski, p. 115). Of Billig's other criteria some are discussed above and one, the greater frequency of after its noun, is not a late form (cf. Lutoslawski, pp. 131–2: in the Rep. it is much higher than in the Tm. and as high as in the Soph. and Pol.): B. may have confused this with the predominance of c. acc. over c. gen.

^{page 82 note 2} Cf. pp. 84, n. 4; 93, n. 3.

^{page 82 note 3} Here it seems on stronger ground than recent post-datings of the Cratylus; but the stylistic evidence on that dialogue (like the arguments so far given for its lateness) can and should be pruned and supplemented.

^{page 83 note 1} ‘Parmenides, Zeno and Socrates’, Proc. Ar. Soc. xvi, 1916, pp. 234–89;Google ScholarPlato's Par menides, intro., p. 26.

^{page 83 note 2} A Study in Plato, pp. 96–97.

^{page 83 note 3} That of arguing as though, because the relation between copy and original is not simply resemblance, it does not include resemblance; for if it is included Parmenides’ regress follows at once. The most one could maintain on Taylor's lines is that, if to predicate X of A is to assert that A is not only like but copied from a Form, then (by definition of ‘Form’) it is a contradiction to predicate X of the Form that A allegedly resembles in respect of X But then no such resemblance between A and the Form can be maintained, nor a fortiori can A be the Form's copy; so this serves Parmenides' ends by wrecking the account of predication. But the evidence is against this line of argument (see next note).

^{page 83 note 4} e.g. (i) such uses of the terminology as at Rep. 501 b where the legislator is a painter with his eye on the and able to make a direct comparison between sitter and portrait (cf. Phdo. 76 e 2); (ii) the fact that on the old theory of Forms the property represented by the Form was predicated without qualms of the Form itself: Justice just, Holiness holy (Prot. 330 c–e), Largeness large (Phdo. 102 e 5), where the predicate-expression is used unambiguously of Forms and particulars, as is proved, for example, by (Prot. 330 d 8); (iii) Aristotle's use of the premiss that the was common to Forms and particulars (e.g. Met. 997 ^b 10–12; E.E. 1218 ^a 13–15). So Plato did not suppose the paradeigmatic function of the Form of X, any more than its being , to rule out the assertion of resemblance between Form and in respect of X And this position is not modified in the Tm. Hence Parmenides' regress is the exactly appropriate criticism of the theory.

^{page 83 note 5} Aristotle's Criticism of Plato and the Early Academy, pp. 297–9.

^{page 83 note 6} Ibid., pp. 295–7; cf. Apelt, , Beiträge, PP. 52–53.Google Scholar

^{page 83 note 7} In the Tm. the resemblance of to is introduced to prove not the uniqueness of the Form but that of the , given that of the

^{page 83 note 8} A.C.P.A., p. 293.

^{page 83 note 9} e.g. Met. 990b 17, 991 ^a 2–5, 1032 ^a 2–4.

^{page 83 note 10} As to the answer which Cherniss constructs for Plato, certainly Plato later concluded that the eiδos should be regarded as ‘being that which the particular has as an attribute’ (A.C.P.A., p. 298)—the necessary type-distinctions are forced by Parmenides' first regress (132 a i–b 2) and sketched in Tht. 156 e, 182 a–b; but to expound in the idioms of resemblance and copying is just to show that one has not yet grasped these type-distinctions.

^{page 84 note 1} Plato's Theory of Ideas, pp. 89, 230–1.Google Scholar

^{page 84 note 2} Ibid., p. 127.

^{page 84 note 3} Ross, (P.T.I., pp. 228–30)Google Scholar has collected occurrences of the idioms by which the relation between Forms and particulars is de scribed in the dialogues. From his data he infers that ‘there is a general movement away from immanence towards transcen dence’ (sc. towards the –idioms). But his list does not bear this out. Of the dialogues taken to follow the Phdr., the Tm. is alone in using the –idioms, and uses them exclusively and almost exhaustively. Tht. 176 e 3–4 is no exception (as Ross agrees, p. 101), for the context (the ‘digression’) is strongly metaphorical, and the twin cannot be because the at least has no place in the which is the soul's proper habitat (177 a 5). Ross does not note the following po nts: (a) the special term used to describe knowledge of the seems to be confined to the Rep. and Tm., except for its occurrence at Crat. 407 b 4 and 411 d 8 where the particular form is required by the etymology. Since such knowledge was a dyadic relation between minds and Forms, it seems likely that the old expression was shelved when the Tht. had proved (199 c–200 c) that knowledge and error were not a matter of bare recognition and misidentification. (b) The term introduced in the Phdr. myth and Parm, and subsequently often used for etc., is not found in the Tm., which here too confines itself to the vocabulary of the Rep. (e.g. seems to be peculiar to these two works). But the word occurs in Crat. 434 a (Ast omits this, so it has eluded Campbell and Lutoslawski).

^{page 84 note 4} Contrast with the refutation of the the less intimidating arguments brought against the so-called ‘immanence’ version of (Parm. 131a 4–e 5). In the Phil. (15b) it is these arguments alone that are quoted as needing an answer if the are to be saved. Professor Skemp (P.St., p. 238) thinks that, since is explicitly located in the fourfold classification in Phil. 23 c–27 c, the cannot have been superseded either. The plain fact, whatever one makes of it, is that this classification of does make room for the and does not make room for I do not quote this on behalf of my position, but it scarcely tells against it.

^{page 85 note 1} Taylor, , Comm., p. 32.Google Scholar Taylor says that the Tm. maintains this incompatibility ‘from first to last’ in sharp contrast to the Phil. theory of , but contradicts himself in a note on 31 b 3 by importing an allusion to the Phil, and so leaving the Tm. inconsistent on a key-doctrine; he is corrected by Cornford ad loc. (P.C., p. 42, n. 1). The Tm. does not in fact (and does not promise to) adhere always to the special usage proposed in 37 e–38 b and discussed in this section: naturally, since (as Plato came to see) its adoption is ruled out by logical absurdities. The point is that if he had seen this when writing the Tm. the proposal made in 37 e–38 b would never have been made.

^{page 85 note 2} Laws 894 a 5–7, Phil. 26 d 8, 27 b 8–9, 54a–d (cf. de gen. an. 640 ^a 18), Soph. 248 a–249b, Parm. 163 d 1–2, and passages discussed above. Phil. 59 a and 61 d–e are not parallels to the Tm. disjunction, because the Tm. says not only (as the Phil, does) that some things exist without changing but (as the Phil, does not) that some things change without existing; this step, the outcome of the Republic's muddles about existence, is not entailed by the commonplace distinction between , and it is this which is refuted in the Tht.

^{page 85 note 3} Cf. Diès, , Philebe (Budé), pp. xxviii–xxix.Google Scholar

^{page 85 note 4} e.g. Cornford, , Plato's Theory of Know ledge, p. 101;Google ScholarCherniss, , A.C.P.A., p. 218, n. 129.Google Scholar

^{page 85 note 5} Cf. Robinson, , Phil. Rev. lix, 1950, PP. 9–10.Google Scholar

^{page 85 note 6} , 182 c 10: this argument defeats the lame plea of the Tm. (49 d–e) that even if we cannot say what any mere is we can describe it as (cf. Tht. 152 d 6). In a similar argument the Crat. makes the point so explicitly alone would vindicate its place in the critical group.

^{page 86 note 1} Cornford, misconstruing the previous argument, can naturally make nothing of the fact that this final refutation hinges on the He is reduced, first to seeing an ambiguity in , finally to making the argument turn on the denial of to (P.T.K., pp. 108–9).

^{page 86 note 2} Apollodorus sets his floruit in 368–365. The theory was in any case presumably published before he left Athens for the final task of legislating for Cnidus (D.L. 8.88), and this in turn must be some years before his death in 356–353. Cf. Harward on Ep. 13. 360 c 3 (The Platonic Epistles, p. 234).

^{page 87 note 1} Comm., p. 211.

^{page 87 note 2} As this implies, if at Tm. 40 b 8 signifies a motion I accept Cornford's account of it as compensatory rotation (P.C., pp. 130–1).

^{page 87 note 3} Eudemus ap. Simpl. in De Caelo 292 ^b 10 (488. 20–24, cf. 492. 31–493. 32).

^{page 87 note 4} Cornford, , P.C., pp. 106–12.Google Scholar

^{page 87 note 5} Taylor, , Comm., p. 202.Google Scholar

^{page 87 note 6} In fact it represents part of the source of Plato's complaint against empirical astronomy in Rep. 530 a 3–b 4—a passage which clearly prefigures the Tm., and not only in introducing the Equally, it explains why Plato's astronomy throughout depends for its precise expositi on the manipulation of an orrery (e.g. Tm. 40 d 2–3).

^{page 87 note 7} P.C., p. 116.

^{page 87 note 8} Cornford in this connexion wrongly quotes the number 27 (which includes the motions of sun, moon, and stars); but even 27 is no

^{page 87 note 9} It is sometimes said (e.g. by ProfessorSkemp, , T.M., p. 79)Google Scholar that the Tm., like the Laws, condemns the description of the planets as This is not so. It says merely that they are so called ( 38 c 5–6) and goes on to define the (40 b 6). Cf. Simplicius in De Caelo, 489. 5–11.

^{page 88 note 1} Class. Phil. xxvii, pp. 80–82;Google ScholarPlato's Thought, p. 14a.

^{page 88 note 2} A.C.P.A., p. 409, n. 337.

^{page 88 note 3} P.C., p. 62. The parenthesis hardly deserves refutation. If such ‘ancient critics’ as Xenocrates and Crantor ever attended to the Sophist in constructing their divergent interpretations, it was notoriously not their ‘sole clue’: cf. Taylor, , Comm., pp. 112–15.Google Scholar Xenocrates’importation of motion and rest was presumably grounded in the Tm. itself (57 d–e), and attempted to reconcile the Tm. with the definition of given in the Phdr.

^{page 88 note 4} P. T.K., p. 50 and n. 2: cf. especially his introduction of ‘visual fire’ and ‘fiery particles’ which ‘interpenetrate and coalesce’.

^{page 88 note 5} P.T.K., pp. 124, n. 2; 327, n. 2.

^{page 88 note 6} P.C., pp. 59–66.

^{page 88 note 7} P.C., pp. 62–64, 102. It might have been glossed by the Phdr. myth (247 c–e) in which the that represents is contrasted with that which is

^{page 88 note 8} P.C., pp. 65–66.

^{page 88 note 9} Cf. the determining of harmonic intervals in the world-soul and the mathematical idioms in Tm. 31 c 4, 36 c 5–7.

^{page 88 note 10} I think Plato may have seen conclusive reasons for excluding of existence, identity, and difference before he saw the general objection to making the Forms then the readmission of existence, etc. as in the Sophist would mark the revised function of the But this falls outside the present paper. In the Tm. Plato does not commit himself and should not be committed by his commentators.

^{page 89 note 1} P.C., p. 96.

^{page 89 note 2} P.C., p. 95, n. 1.

^{page 89 note 3} “Timaeus always talks of the in the old undiscriminating fashion familiar to us from the fifth book of the Republic’ (Taylor, , Comm., p. 32).Google Scholar

^{page 89 note 4} P.C., p. 98, n. 4.

^{page 89 note 5} To try to give it the former use is to try to say what is (238 c 10); correspondingly 4). For a further refutation of Cornford's account of the Sophist see A. L. Peck, ‘Plato and the of the Sophist', C.Q. xlvi, esp. pp. 35–38.Google Scholar Though I think Dr. Peck's positive thesis mistaken (viz. that the Soph, has primarily the local virtue of beating certain sophists on their own ground), I take it to be at least partly prompted by the very real problem why the Soph, differs markedly from the Tm. in its terminology and interests (cf., for example, op. cit., pp. 39, 53). My own answer to this will be evident.

^{page 89 note 6} Soph. 258 c 2–3, Pol. 284 b 8, 286 b 10.

^{page 90 note 1} Cf. Rivaud's notes on Tm. 17 c–19 a; he does not remark Tm. 18 b 3 = Rep. 419 a 10 or the deliberate use of for the State marriages (a word apparently confined to Rep. 460 a 9 and Tm. 18 d 9).

^{page 90 note 2} As to dramatic date, surely the reason why the Tm. could not be set after the Rep. (i.e. two days after the Bendidea) is just that when writing the earlier work Plato had not yet formulated the plan of the later and therefore had not seen the need to introduce any speaker of Timaeus' powers among either Cephalus' guests or Socrates'(presumed) auditors next day. Hence a further recital had to be invented. To infer from this that ‘the design of the [Timaeus] trilogy is completely independent of the Republic' (Cornford) is to invert the natural inference.

^{page 90 note 3} Barker's paradox, that the Republic is ‘uncompromisingly hostile to law’ and that this hostility is relaxed in the Politicus (Greek Political Theory, p. 271), hardly needs refutation. The Republic does not repudiate any ‘system of law’; it contends only that continuous piecemeal legislation and litigation will be eliminated (425 e), since then the Guardians will know Even if the of the Republic were ‘unwritten ordinances’, the Politicus censures immutability in written and unwritten alike (295 e 5); but in fact it is only the that will not have written legislation (Rep. 425 a–b). No punishment for crime is considered because Plato concentrates on the Guardians, whose crimes will disrupt the constitution and make punishment unavailable and unavailing. If it is true of this that ‘its government is the result of its nature’ (op. cit., p. 204), it is conversely true that its nature is the result of the prescribed by which are irrevocable (424 b–d).

^{page 91 note 1} Taylor, , P.M.W., p. 397.Google Scholar

^{page 91 note 2} Probably under the influence of Laws 4. 713 a–714 a, on which see p. 93, n. 4. For another refutation of Taylor's interpretation see Skemp, J. B., P.St., p. 52.Google Scholar

^{page 92 note 1} The possibility is not cancelled by the concession that men do not credit it and that at present no such natural autocrat is to be found (301 c–e). At this point Professor Grube's analysis breaks down (Plato's Thought, chap, vii). Against Barker he rightly points out that the Republic never supposes, what the Politicus affirms, that ‘the best laws, even those enacted by the philosopher-king himself, are inevitably imperfect’ and that law is a (Pol. 300 c 2). But he thinks that now the philosopher-king has risen ‘so high [sc. above law] as to join the gods’ (p. 281), and is consequently puzzled that ‘the final definition of statecraft seems to imply the philosopher's knowledge all over again’ (p. 284).

^{page 92 note 2} Politicus, intro., pp. xxi–xxvi.

^{page 93 note 1} Taylor seems to be right in saying that ‘we are apparently to think of the authorities of [Plato's] “city” as needing less than a generation for the experience which would justify them in declaring their institutions definitely inviolable’ (The Laws of Plato, intro., p. xxxii).

^{page 93 note 2} There is perhaps another in 12. 945 b– 948 b where certain political abuses described in Pol. 298 e–299 a are eliminated by arrangements for the election and scrutiny of magistrates. In 6. 773 a–c the marriage of complementary characters recommended in Pol. 310 a–311 a is independently defended. On 4. 713 a–714 a see p. 93, n. 4 infra.

^{page 93 note 3} Suggested by Taylor, Diès, Field, and Ross, inter alios. There is no direct evidence that any part of the Laws was written after every other dialogue. The work certainly followed the Republic (Aristotle, Pol. 1264 ^b 28). But Diogenes' remark that it was left on the wax does not certify even that it occupied Plato to his death, much less that nothing else was written at the same time. (Who would argue that the works which Descartes or Leibniz left in manuscript must have been their last?) The connexion of the with Plato's work at Syracuse (Ep. 3. 316 a) does not show that the technique first suggested itself to him there or in the year 360 (Taylor, , P.M.W., pp. 464–5;Google Scholar cf. Burnet, , G.P., p. 301.Google Scholar But note that in the Tm. (29 d 5) the contrast between has the musical connotation found in the Rep. (531 d8), not the later legal sense). Taylor arbitrarily and inconsistently assumes a ‘block’ Laws in arguing that, if Laws 4. 711 a–b (describing as if from personal knowledge the powers of a tyrant, which the wise legislator may hope to harness) should be dated after Plato's last return from Syracuse, ‘the work must therefore belong to a date later than 360’ (Laws, intro., p. xii). In any case (a) the optimism of the passage hardly accords with Taylor's dating and (b) the personal experience (of a tyrant's power to shape a State for good or evil) could clearly have been gained earlier.

^{page 93 note 4} This would more easily explain the form of a myth in Laws 4. 713 a–714 a which bears a superficial similarity to that in the Politicus. The moral wrongly imported by Taylor into the Pol. (namely that the ideal ruler independent of laws is not an historical possibility) is in fact the moral of the allegory in the Laws, which can be regarded as a briefer and less sophisticated version corrected, in the light of later political theories, by the Pol. For whereas in Laws 4 the ‘divine shepherd’ and the supremacy of law are presented as a simple disjunction (713 e–714 a) and law is itself the (714 a 2), the Pol. insists on the tertium quid, the independent ruler with . And in the Laws this possibility does not seem to be entertained before Book 9.

^{page 94 note 1} Cf. Cornford's, development of Raeder's suggestion, P.C., pp. 6–8.Google Scholar

^{page 95 note 1} Hackforth, , P.P., p. 75Google Scholar