¹ Hermann, K. F., Geschichte und System der Platonischen Philosophie (Heidelberg, 1839), pp. 538–40Google Scholar.

² For critical discussion of Krohn and other separatists, see Zeller, E., Philosophie der Griechen, ii.l⁴, 556–62Google Scholar, and Raeder, H., Platonsphilosophische Entwickelung (Leipzig, 1905), pp. 287ffGoogle Scholar.

³ Gellius, Aulus, Nodes Alt. 14.3Google Scholar: ‘quod Xenophon incluto illi operi Platonis, quod de optimo statu reipublicae…scriptum est, lectis ex eo duobus fere libris, qui primi in vulgus exierant, opposuit contra conscripsitque [etc.].’ Gellius' source is unknown, and there are problems with his claim that the Cyropaedia is a response to the first part of the Republic. See Wilamowitz, , Platan (2nd ed., Berlin, 1920), ii.181fGoogle Scholar.

⁴ For full discussion, see Adam's Appendix I to Book 5, in The ‘Republic’ of Plato (Cambridge, 1902), i.345–55Google Scholar. Adam concludes that Plato probably had Aristophanes in mind when he wrote Book 5 (p. 354).

⁵ See Ritter's results summarized in Brandwood, L., The Chronology of Plato's Dialogues (Cambridge, 1990), pp. 67–73CrossRefGoogle Scholar. There has, however, been a revival of the analytical enterprise in Thesleff, H., Studies in Platonic Chronology (Helsinki, 1982), pp. 101ffGoogle Scholar.

⁶ Vlastos, G., Socrates (Cambridge and Ithaka, 1991), p. 250CrossRefGoogle Scholar. According to Friedlānder, Plato ii.305 n. 1)Google Scholar, the name Thrasymachus was given to the hypothetical dialogue by F. Dülmmler in 1895.

⁷ Friedlander (ii.50ff.) gives some other arguments, less serious in my opinion: for example, that the conversation of Book 1 begins too late in the day. ‘If the Republic had been planned as a whole, the discussion probably would have begun early in the morning, as in the Laws’ (p. 51). As if Plato, when he supposedly re-used an earlier work, could not have made such a change if he had regarded it as of any importance.

⁸ Wilamowitz, , Platon ii. 181Google Scholar.

⁹ Ibid. 184. Cf. vol. i.209–11.

¹⁰ Friedländer, P., Plato, ii (New York, 1964), p. 50Google Scholar.

¹¹ Referring to von Arnim's second stylometric publication in 1912, on which Wilamowitz was relying, Brandwood (p. 220) quotes with approval Ritter's judgement that this ‘immense amount of labour’ was ‘wasted’.

¹² The latest group (Sophist, Statesman, Philebus, Timaeus-Critias, Laws) was identified by Campbell in 1867, independently identified by Blass in 1874, then again independently by Dittenberger in 1881, and finally confirmed by Ritter in 1888. (See the report in Lutoslawski, W., The Origin and Growth of Plato's Logic (London, 1897), pp. 84ff., 101, 103–5, 124f.)Google Scholar. The place of the Republic, Phaedrus, and Theaetetus in the middle group was recognized by Campbell in his original study, and confirmed by Ritter in 1888. (See Lutoslawski, pp. 93 and 122.) The only definitive result achieved after 1888 was the recognition that the Parmenides belongs to the same middle group, a discovery made independently by Campbell and von Arnim in 1896 (Lutoslawski, pp. 137f.).

For the current state of the question I rely on Brandwood (although he is more optimistic about the possibilities of a linear ordering). Thesleff (see n. 5 above) combines considerations of style and content, which tends to obscure the division into three groups. In the latest exercise in pure stylometry, Recounting Plato (Oxford, 1989)Google Scholar, Ledger, G. R. recognizes ‘the sharp difference between early and late works’ (p. 225)Google Scholar, and his summary conclusions reflect the division into Groups II (Republic-Phaedrus) and III (Philebus-Critias), on pp. 224f. But this result is less perspicuous because of Ledger's (to my mind, misguided) attempt to obtain a linear sequence for all the dialogues.

¹³ The Cratylus has 8 late traits, the Phaedo 7, although each of these dialogues is almost twice as long as Republic 1 (which has 9 late traits). The Lysis also has 8 late traits. The variation for other members of Group I (from 3 late traits to 5) cannot be significant. Other members of Group II, on the other hand, all have a substantially larger number of late traits, like Republic 2–10: Parmenides, 17, Phaedrus, 21, Theaetetus, 25. See the data in Brandwood, pp. 66–74. Unfortunately, Ledger does not provide separate statistics for Republic 1.

¹⁴ See Brandwood, pp. 107f. Von Arnim claimed (on slender grounds) that the Lysis was the latest dialogue in Group I, immediately preceded by Laches, Republic 1, Phaedo, and Symposium. The Cratylus was placed by him just before this group.

¹⁵ A similar observation was made by Raeder, Platons philosophische Entwickelung, p. 202. For a systematic attempt to explain the stylistic peculiarities of Book 1 in literary rather than chronological terms, see Vretska, K., ‘Platonica III’, Wiener Studien 71 (1958), 32–6Google Scholar.

¹⁶ This is, I take it, the truth behind Jaeger's exaggerated claim that when Plato, ‘wrote the first words of his first Socratic dialogue, he knew the whole of which it was to be a part. The entelechy of the Republic can be quite clearly traced in the early dialogues.’ Paideia, tr. Highet, G. (Oxford, 1944), ii.96Google Scholar.

¹⁷ Compare Rep. 1.333e–334b with Hi. Mi. 365c–368a, and Republic 335b–336a with Crito 49a–c.

¹⁸ Reeve, C. D. C., Philosopher-Kings: The Argument of Plato's ‘Republic’ (Princeton, 1988), pp. 22–4Google Scholar. Cf. Irwin, T., Plato's Moral Theory (Oxford, 1977), p. 184Google Scholar.

¹⁹ See the Afterword on Prolepsis, below.

²⁰ Cf. Reeve, , Philosopher-Kings, p. 6Google Scholar: ‘Cephalus is an attractive character, portrayed with delicacy and respect’ (p. 6). ‘Within the Kallipolis men of his natural type will reliably achieve the highest level of virtue of which they are capable’ (p. 7).

²¹ The connection between Cephalus' remark about the afterlife and the concluding myth of Book 10 has often been noted. See, e.g. Wilamowitz, , Platon ii.181Google Scholar; Raeder, , Platons philosophische Entwickelung, p. 199Google Scholar (who also cites Gomperz for this point); cf. Lutoslawski, , op. cit., p. 273Google Scholar.

²² Similarly Raeder, p. 200. Reeve (p. 12) asks the right question, although he gives a different answer: ‘why Plato has him (Thrasymachus) reject an obvious defence in favor of an arcane one, and why he takes such pain (in the Cleitophon episode) to advertise that rejection.’

²³ Glaucon does not mention the tyrant, but he describes the case of perfect injustice as a τ⋯χνη (2.360e5–361a5), and specifies that such a man will rule in the city (362b2).

²⁴ The hint of tripartition is noted by Lutoslawski, pp. 273f. and Raeder, p. 200.

²⁵ Adam, i.56. Cf. Raeder, p. 201.

²⁶ David Sedley reminds me that the ἒργον concept was introduced earlier, in connection with the function of justice at 332e ff.

²⁷ But the meeting with Polemarchus leads us to the conversation with Cephalus, and items 2–4 show that this episode has been ‘reworked’ with the Republic in view.

²⁸ See Adam's, note on 380e: ‘The doctrine… enunciated here foreshadows, but does not presuppose, the metaphysical predominance of the Good in Book 6’ (i. 119)Google Scholar.

²⁹ I am indebted to David Sedley and Mary Hannah Jones for comments on an earlier draft.