Bryn Mawr Classical Review 97.1.8


E.A. Duke, W.F. Hicken, W.S.M. Nicoll, D.B. Robinson, and J.C.G. Strachan (edd.), Platonis Opera, t. I. Tetralogias I-II continens. Oxford: Oxford University Press, 1995. Pp. xxxi, 572. $32.00. ISBN 0-19-814569-1.


Reviewed by David J. Murphy, The Nightingale-Bamford School.

Students of the text of Plato have been waiting for a new Opera Omnia for at least two generations. It was not long ago that E.R. Dodds could lament that "to this day, no one can say with certainty how many primary witnesses to the text there are, or how they are related to each other and to the secondary MSS." (Plato. Gorgias [Oxford 1959] 34). Much work has since been done by Dodds and others to isolate the primary witnesses, and the results have now served the team of scholars (one of whom, Winifred F. Hicken, lamentably did not live to see the publication of this first volume) who have begun the new OCT of the ancient author most written-about in our time.

Burnet's knowledge of Plato was deep, and lovers of his excellent text may wonder, why replace it? Despite Burnet's acumen, however, the inaccuracies of his apparatus are well known. Burnet was the first editor to make substantial use of what came to be considered the three most important primary MSS: Clarkianus 39 (=B, A.D. 895); Ven. Marc. gr. append. cl. IV.1, coll. 542 (=T, s. x); Vind. suppl. gr. 7 (=W, s. xi). Although he made some consultations of B in Oxford and of photographs of T, he obtained most of his MS. reports from others. Burnet had but limited acquaintance with six other MSS. that subsequently have been shown to be primary witnesses. Numerous papyri and indirect testimonia, as well as Aristippus' Latin translation of the Phaedo, have been published since his day. Finally, Burnet selected many readings that rest on weak or no MS. authority. While the Budé series represented certain advances, those volumes are of uneven quality. Therefore, although the text of Plato has not been in bad shape, we have needed an Opera that rests on a truly scientific basis.

In the preface, the new editors lay out their program for meeting this need (p. v-xxi). Except for the Theaetetus, for which Miss Hicken was responsible, Elizabeth A. Duke investigated the indirect tradition, and William S. M. Nicoll tracked down and collated the MSS. In this latter task, Duke and David Robinson assisted by examining various MSS. by autopsy. Nicoll edited the Euthyphro, Apology and Crito; Christopher Strachan edited the Phaedo; Nicoll and Duke edited the Cratylus; Robinson edited the Sophist (assisted by Nicoll) and the Politicus. The collaborators discussed and considered each others' opinions, however, so that the result is a shared work.

The primary MSS. divide into the three well-known families clustered around B, T and W. Each of these MSS. offers all the dialogues of tetr. i-ii. Closely related to B but independent of it are Ven. Marc. gr. 185, coll. 576 (=D, s. xi-xii), and Tueb. gr. Mb14 (=C, s. xi), which latter MS. for Vol. I transmits only Euthphr., Crito and Phdo. The editors believe that these three MSS. descend in a tripartite stemma from some hyparchetype "Beta", but acknowledging the effects of contamination and chance, they deny that one can safely take the agreement of two of these MSS. to represent Beta.1 T is the sole independent member of the second family. Alongside W, the third family numbers Vat. Pal. gr. 173 (=P, s. x, containing Apol., Phdo. and excerpts), Vat. gr. 225 (=V, s. xiv?2 independent only in tetr. i), Par. gr. 1813 (=Q, s. xiii, containing Phdo. and Crat. of this volume), and Par. suppl. gr. 668 (=S, s. xi, containing Crito and parts of Phdo. and Crat.). To these must be joined the putative exemplar (=L) of Aristippus' Latin translation of the Phaedo (A.D. 1156) and that of the Armenian translation of Euthphr. and Apol., as well as the fragment of Apol. contained in Ven. Marc. gr. 511 and the corrections assignable to B2 (presumably Arethas, the codex's owner). The hyparchetype of this third family is designated "Delta". The editors also believe that the corrector T2 is an independent "Delta" witness, but they rightly avoid making absolute claims about particular T corrections. The editors follow Carlini's division of the third family in Phdo. between VWL and PS, adding Q to the latter, and they see similar bifurcation elsewhere.

Evidence for the independence of these MSS. can be found in works cited in the preface. The hardest MSS. to classify are C and D, for they share very few correct readings with T or "Delta" against B. I have expressed elsewhere my reasons for believing in the independence of D.3 Additional separative B errors include Euthphr. 8e4 E(/KASTO/N] E(KA/STWN B; Apol. 19d7 TA)=LLA] POLLA\ B; 38c3 ME om. B; Crat. 399c4 O)/PWPE] PWPE sic B; 417a8 E)/OIKE om. B; Soph. 247d5 H(MW=N om. B.4

The editors omit to cite readings of papyri or ancient authors when they deem the MS. text indubitable (p. xvii), and the index testimoniorum (pp. 561-72) catalogs only those cited in the apparatus, excepting scholia and Platonic commentaries that follow the sequence of the text. A modern bibliography is a desideratum for new OCT's; there is a bibliography of editions that have provided the indirect testimony (xxvi-xxxi).

The OCT team claims to have investigated the other MS. candidates to independent status and to have determined that none is such. They have deemed it unnecessary to make complete collations of all the MSS. (xvii-xviii). Strictly speaking, one must collate every MS. in its entirety to be sure of its place in the tradition. Copyists may use different exemplars; a late MS. may contain readings, even corrections, of ancient authority. I have examined a large number of Platonic manuscripts without finding any primary witnesses that are unknown to the OCT team, however, and I believe it unlikely that we shall identify any other known MS. as independent in the earlier tetralogies. Although Platonic manuscript work is not finished, the results of the editors' recension are likely to stand for a long time.

Now to the apparatus. It is remarkably accurate, and it corrects errors of Burnet that escaped the Bude editors. I have consulted my collations in Phdo. and Cra. and have verified anew from photographs only these errors: Phdo. 59 d6 EI)SH/EIMEN sic re vera codd. (not EI)SH=|MEN); 67e3 OU) GELOI=ON P Socrati tribuens (not GELOI=ON); 78a1 and 97d3, not corrected by Q1;5 78b4 A)NEREI=SQAI P (not A)NERE/SQAI); 91c6 ME Q (not ME/N); Cra. 385b2 TI/ Qlpc (not TI); 413b2 ME A)POPIMPLA/NAI Q (OCT says Q omits ME). There are few misprints: Phdo. 99c3 A)/N POTE TDelta (not A(/N POTE); 99a6 A)/LLA O(/SA WPLVQ (not A)LLA\ O(/SA, and the OCT omits reference to Q); 108d3 OU)=N Q (not OU)/N); Soph. 244d11-12 "Socrates" is a slip for "hospes"; Plt. 281e4 instead of "SUNAITI/OUS BTW P2 Stob." the true variant is ME/N BTW P2: ME/N OU)=NStob., and the index lacks the citation (Stob. IV.413-14 Wachsmuth-Hense); 285 "b8" should read "b7"; 297a2, b2 dots for iota subscript appear.

The editors elect to pass over in silence many obvious or insignificant MS. errors (xix; cf. Nicoll [supra, n. 1] 31). Space limitations of course obtain; yet, students of papyri and MSS. will still need to consult collations in earlier editions. A good policy would have been to list all errors shared by more than one primary witness. For example, at Phdo. 105c3-5, BD omit by homoioteleuton the words NOSE/SEI ... E)GGE/NHTAI, which are supplied in the margins by B2 and by an early corrector of D. C on the other hand omits c3-4 SW/MATI ... A)RIQMW=|. BCD being related, it would be worthwhile to cite the omissions, since they prove that C cannot be a direct copy of Bac or B2. To cite Phdo. 59c2 TO\N SWKRA/TH] TW=(sic) SWKRA/TEI QS would help to evidence the QS link. Sometimes an error in one MS. goes unmentioned when the same error in others is cited, even across MS. families: e.g. Phdo. 69a6 GA\R om. T atque V; 95c4 A)NO/NHTON WL atque D; Cra. 388d13 AU) TA/] TAU=TA W atque Q. Some readings are presented as though they rest on authority of all MSS. when they do not: e.g. Phdo. 113c1 O(\ (the MSS. and Burnet read O(\N), Tht. 156a4 E)C H(=S (only in bW2), Soph. 244b9 A)POKRINE/SQWN Simpl.: A)POKRINE/SQWSAN bTW (Burnet's report is correct). Since T is the only representative of its family, one misses fuller collations of this codex. Especially useful would be citation of omissions of lines of text like Phdo. 103d9-11 PA/NU A)POLEI=SQAI om. T. T's value as a witness can be overestimated by those who have been led into a false view of the fidelity to his model of its scribe, Ephraim Stoudios.

A special problem is posed by Lambda, for the citations would be more useful if doubtful cases were represented by the quoted Latin and not as e.g. Phdo. 59a8 "TOTE\ TPQS, ?Lambda"; better to print "? Lambda (nunc)" or simply to give the Latin reading. On the other hand, more Lambda readings could have been included, as e.g. 90d1, where Aristippus' talibus most likely presupposes TOIOUTOISI\ with BCDWPV and not TOIOUTOIS(I\) TISI\ with Q and T. But the OCT makes good use of Aristippus. His quid equidem is rightly represented as TI/ DH\ at 105d13, and his version is cited as authority for readings like 70e1 I)/DWMEN (videamus) or 85e3 H(= DH\ (prout), a conjecture of Forster (Burnet misstated the reading of W).

The reader is warned that not all the primary witnesses of the first or third family necessarily contain the readings ascribed to the hyparchetypes (p. xix). The editors frequently enumerate the witnesses singly when the MSS. differ within a family, but a better policy would be to cite the putative hyparchetype only when all its witnesses agree. This is especially important in the case of B2, which does not offer all the distinct Delta readings. When the apparatus shows Delta as differing from Beta, should we suppose that Arethas made the Delta correction or not? This is worth knowing, since he is the earliest Delta witness.

The new editors have found MS. authority for many correct readings which in Burnet rested on the authority of later MSS. or on that of printed editions or scholarly conjecture. These readings fall into several categories. A) Readings printed by Burnet without the authority of a primary MS., in support of which the new OCT cites at least one primary MS.: e.g. Crat. 426e1 TRE/XEIN Q; Tht. 175e5 O( D' P and Iamblichus; Plt. 281d2 A)\N ... PERIE/LWMEN D; etc. B) Readings in support of which the new OCT cites a source of superior but not independent stemmatic position: e.g. Euthphr. 15d1 PROSSXW\N Vind. phil. 80 ("scripsi" Burnet); Apol. 23b2 EI) add. man. r. Laur. 85.6 ("add. Stephanus" Burnet); Crat. 385a2 O(\ A)\N FH\|S KALH=| Ven. app. cl. 4.54pc (coniec. Hirschig); etc. The editors think that these readings are probably conjectures (p. xix). We owe many of them to the first scribe of Par. gr. 1808 (cf. e.g. Plt. 274d2, 275b5, etc.). Flor. Laur. gr. 85.6 and Ven. Marc. gr. 186 are also well represented. The editors deserve credit for tracking down the sources of readings that Burnet ascribed only to "al.", "recc.", "vulg." and the like, and for identifying which critics first proposed many modern conjectures. At Crito 46b4, the new OCT cites a newer transcription (IG XIV 318 no. 1214) for the reading NU=N PRW=TON.

To the text itself. The editors acknowledge that Delta is closer to Beta than to T in Apol., that TW agree in Plt. after 288, etc., but they rely on editorial acumen rather than stemmatic considerations when deciding what to print: the tradition is heavily contaminated, MSS. switch sources, and the connections between the MS. families are quite entangled (p. xiv-xvii; Nicoll [supra, n. 1] 34-36).6 In broad outline, an eclectic approach is the only possible one for the "open" Platonic tradition. A correct reading is often found in only one witness, and there is no best branch of the tradition. Nevertheless, in cases of relative indifferentia like minor word order differences, particles, etc., stemmatic criteria can create a presumption of probability along the lines suggested by Boter (1987) 151, and if the editors would admit this weak claim, they should say so. For example, at Tht. 162d2 TAXU\ TW: TA/XA Beta (so Burnet), the W reading seems to have helped make the decision. The same impression comes from Apol. 26a8 DH=LON H)/DH DT et fort. B: H)/DH DH=LON Delta (so Burnet), Plt. 303d1 NU=N BetaW: NU=N DE\ T (so Burnet), etc. Actually, although Nicoll (1995) 36 has rejected Boter's suggestion, the two seem not to disagree: Boter invoked probability; Nicoll denies certainty.

As to the passages that differ from those printed by Burnet,7 the largest number are those for which Burnet did not cite third family readings. Cf. Apol. 24e10, where the third family along with T1(?)pc agree with BCD's OI(/DE OI( against OI( DE| of Tac, which Burnet printed. At Euthphr. 6a7, on the other hand, knowledge that V agrees with BetaT in including O( seems to have helped the editors reject Burnet's KAI\ QEOFILH\S of W. The editors readily opt for a minority reading; cf. Crito 53a1, where Nicoll follows the superior BARBA/RWN of T against BARBARIKW=N of BetaDelta and Eusebius (Burnet had almost made this choice; cf. ad loc.). The agreement of TDelta is by the same token often preferred to a first family reading, esp. in Phdo., where the stemma appears tripartite. Cf. e.g. 69b6 KAI\ B2TDelta Iambl. Stob.: om. Beta (Burnet bracketed). Some superior readings rest only on "Delta", as e.g. Crat. 407e6 TO\ GA\R against KAI\ TO\ of BDT; 408d7 DE\ TW=N TOIW=NDE confirms previous editions; 418a8 DIIO\N W essentially anticipates Bekker's I)O\N. The editors uncover places where first family MSS. too are decisive, as e.g. at Apol. 26a8, where D backs up what seems to have been an original agreement of B with T in reading DH=LON H)/DH E)STI\N against the third family's different word order, which was favored without strong reason by Burnet. Cf. also Apol. 38c3 ME (omitted by Burnet, but support for TDelta comes from D); Plt. 288e4, where Burnet's "al.", which read PROSAGOREU/WMEN against Burnet's preferred indicative mood of T, turn out to be BDW! More accurate collations provide gems like Crat. 421d4 A)\N EI)/H EI) H( T2, where Burnet was led into unnecessary deletion of A)/N by his mistaken belief that BT read only EI)/H and not EI)/H H(. Broader knowledge of papyri and the indirect tradition has also helped the editors support good readings: e.g. Apol. 41c2 OI(=S DIALE/GESQAI E)KEI= (where Pack2 1387 confirms T); Phdo. 101d3 E)/XOITO confirmed by Pack2 1389; Crat. 405c7 W(/SPER ... A)/KOITIN confirmed by P.Oxy. XXXIII 2663 (s. ii); Soph. 223d6 KAPHLIKH\ Themistius.

The new OCT Plato accordingly is far better supported than any predecessor. The editors' choices from among this plethora of readings have created a text that reads smoothly and lays out logical steps lucidly. Robin Waterfield has already used it for his translation of the Statesman. As noted above, the editors make excellent use of W and its relations, and that MS. should now come under closer scrutiny as representing what is often the most philosophically sophisticated branch of the tradition.8 On the other hand, the last century's deference to the Clarkianus has finally been abandoned. The text in tetr. i (less so Apol.) remains fairly close to Burnet's, and I agree with the tendency to retain words that had been excluded without sufficient reason as interpolations. In tetr. ii, however, the new OCT differs much more, chiefly through emendations that often are unnecessary or unconvincing. On the other hand, the editors solve some difficult problems, and in their turn they are right to restore certain MS. readings that Burnet had abandoned unnecessarily: e.g. Crat. 386b5 A)/NQRWPOI om. T (Meridier also retained), Tht. 183c6 DEI=, Soph. 224a3 KAI\ PIPRASKOME/NHN (Dias retained), Plt. 264e12 A)/RTION A)RIQMO/N. I proceed now to a few of the many passages that merit discussion.

Euthphr. 14c3-4. Burnet printed TO\N E)RW=NTA TW=| E)RWME/NW|, but the new editors rightly prefer ... E)RWTW=NTA... E)RWTWME/NW| with Schanz et al. Against Burnet, the latter participle is attested by VArm, and E)RWTW=NTA is the better attested of the first pair of alternatives. The passage concerns following the argument where it leads (cf. b9), and the dialogue does not present Euthyphro as an erotic object. I suspect that a common source read TO\N E)RWTW=NTA TW=| E)RWME/NW| (these verbs are often confused).

Apol. 27e6. The new editors wrongly agree with various precedessors in excluding OU) along with Ven. Marc. 184. Burnet's reasoning in following the primary MSS. is sound: Socrates is arguing that the same man must believe in both divine and "daimonic" activities, and the same man again (AU)=) who believes in the activities cannot deny the existence of gods or daimons (cf. Burnet's comment ad loc.).

Apol. 29c4 and 30c1. Nicoll prints A)/N with future in agreement with the MSS., but cf. E. de Strycker-S.R. Slings (Plato's Apology of Socrates [Leiden 1994] ad locc.).

Phdo. 69a1. With Verdenius (Mnemosyne 11 [1958] 204-205), I concur that the strong adversative A)LL' O(/MWS of the MSS. and indirect tradition is needed here, against D' OU)=N of the Arsinoe papyrus (Pack2 1388), which Strachan and others prefer.

Phdo. 83b7-9. The new OCT is right to print H)\ LUPHQH=|, for the latter reading was lacking only in Beta. The words KAI\ FO/BWN at b7, however, are absent in Pack2 1388, Iamblichus and Tac, and Burnet was probably right to bracket them as an addition motivated by desire to balance the four verbs of b8-9 with four correlative nouns (cf. Verdenius 218; R. Loriaux, Le Phedon de Platon2 I [Namur-Gembloux 1981] 184-185).

Against deletion of FASIN at Phdo. 83e6, originally suggested by Jachmann, cf. L. Taran (Gnomon 48 (1976) 764-65) and C. Rowe (Plato. Phaedo [Cambridge 1993] ad loc.). On "normalization" in papyri of s. i-ii, cf. P.J. Sijpensteijn (Mnemosyne 19 [1966] 231) with reply of Verdenius (ibid. 269).

Phdo. 92d4. Relying on Pack2 1389, M. Haslam has suggested the punctuation E)A/N TIS AU)TOU\S MH\ FULA/TTHTAI EU)= MA/LA, E)CAPATW=SIN (ZPE 89 [1991] 11ff). Strachan adopts this improvement. Phdo.100d6. Strachan prints Wyttenbach's conjecture, PROSAGOREUOME/NH, but reports that Pack2 1389 does not support this reading. Ueberweg's PROGENOME/NOU is more likely to be right, as Cornford and others have thought. A scribe early in the tradition made the participle agree with PAROUSI/A EI)/TE KOINWNI/A, but the thing that is added is TO\ KALO/N, not "presence" or "communion." For attempts to defend the MSS., cf. Loriaux (II, 97) and Vicaire.

Phdo. 104d3. This whole passage is very vexed, and Hackforth even supplied several lines. Stallbaum's AU)= TW|, which the OCT reads here, is paleographically the best suggestion (AU)TW=| BD) and fits the context: some X (e.g. "3") has to receive the form of some other thing, Y (e.g. "odd") that is opposite in turn to something else ("the even"), not opposite to "the opposite in question" (against Burnet ad loc.).

Crat. 385b2-d1. The words FE/RE DH/ ... PW=S GA\R OU)/ have been bracketed by Nicoll and Duke under the influence of Malcolm Scofield (CQ n.s. 22 [1972] 246-53), although they rightly do not follow his transposition to 387c5. The passage establishes that a name can be true or false inasmuch as it is part of an argument or statement (LO/GOS), which itself may be true or false. This conclusion prepares the way for Socrates' question (385d2) whether any name is valid for the person who uses it, which question in turn leads to the references in 385e ff. to Protagoras' relativism. Since steps in arguments in Platonic dialogues sometimes wander in ways that befit conversation but can violate our sense of economy, it is risky to excise a passage because it does not match a philosopher's view of what Plato should have written. Here, the MS. tradition receives support from Proclus, and although it is impossible to show that this passage or almost any other cannot be an interpolation, Socrates' view of a statement's truth as its correspondence to reality (TA\ O)/NTA, 386e4), on which his attack on Protagoras depends, relies on the enunciation of this notion in 385b7 ff. in the bracketed passage.

Crat. 425c4. Unlike Nicoll, I am not convinced that Reinhard's TEXNIKW=S is a plausible emendation of TI XRHSTO/N, but the apparatus offers enough material for the reader to make an informed choice of readings.

Crat. 432d8. Against conjectures of Heindorf and Burnet, the editors correctly print the MS. reading, viz. "and neither of the two of them (i.e. the pair of doubles at d7-8) could say which was the thing itself and which was the name." The things are personified as the subjects; cf. d6-7.

Crat. 438a1 ff. TA/DE DE\ ... OU)/ MOI DOKEI=, found only in W, is labelled by the editors Versio A, and they follow Kapp in adding 438b4-7 TI/NA OU)=N ... O)NOMA/TWN to it. Their Versio B, contained in all the MSS., extends from 438a1 E)PANE/LQWMEN to 438b3 OI(=A/ E)STI The editors are not sure whether Kapp was right to delete b4, and they even propose that "posses ... H)/ MEMAQHKO/TES H)\ HU(RHKO/TES H)=SAN (a11-b1) scribere si a11-b3 ad versionem A accomodare velles." The versions both make the point that the first creators of names had to possess knowledge of the things named in order to create the names. For this reason, I concur that one should not read the two versions consecutively as though Version A simply dropped out of BetaT. Version B better fits the context, for in Version A, Socrates derives the first claim from Cratylus' earlier statement that the lawgivers possess techne (428e ff). But Cratylus has already enunciated the first claim at 436 ff., so that there is no need to derive it again. Robinson's view that 438b4-7 should not be attached to Version A seems right; it fits in the traditional place with no real difficulty. The editors are partial to Kapp's view that both versions go back to Plato. My suspicion is that Version A was created by a philosophically sophisticated reader who thought that techne needed to be brought in at this spot. Q begins to agree with BD after 435b or so, so that it offers no help.

Crat. 439d8-440a5. Robinson suspects e1-5 is displaced, and Strachan suspects a lacuna after e2. But there are no insuperable problems with the received text. Socrates wants to argue that objects of knowledge must be unchangeable realities and not changeable images (up to 439d6), and he moves on therefore to difficulties that will result if objects of knowledge themselves are ever-changing (d8 ff), as Cratylus believes all things are.

Tht. 147d4. With Dias and now Waterfield (Plato. Theaetetus [London 1987] 23 and ad loc.) and against Burnet and Hicken, I would retain A)POFAI/NWN of BetaW and Anon. Berol. Like many single words that are lacking only in T, this is not essential to the sense. But T too often stands alone in omissions, reversals of word order, and the like, and editors can be led to accord it too much weight. This caveat applies as well in Crat. where Burnet, followed by the new editors, unnecessarily bracketed words lacking only in T, e.g. 386b5 A)/NQRWPOI, 388d12 EI)=NAI, 389a1 E)STI/N, etc.

Tht. 151c7. Along with what was said above, I disagree with the OCT's omission of E(/TOIMOI EI)=NAI with Plutarch. The idiosyncracies of Plutarch's quotations of earlier authors are well known. Hicken may have been thinking lectio difficilior ("as literally to bite"), but why follow Plutarch against all the MSS. when there is nothing wrong with the received text?

Tht. 175d4. LSJ cite this passage for BATTARI/ZWN, "stammering," but the reading is really a conjecture of Pierson based on Themistius' use of the word in what are at best only allusions to Plato (Themistius 21.252c, 22.278a). The MSS.' BARBARI/ZWN makes sense (cf. Campbell ad loc.) and should be retained in default of real support for the other.

Tht. 179a1-3. Hicken follows Schleiermacher in deleting AU(TW=| (a3). This reading has Protagoras gain disciples by persuading them that no one makes better predictions than he. Fitting the context less well are readings that by emending MH/ and retaining AU(TW=| seek to make Protagoras retain his disciples by keeping them persuaded that the individual is not his own best judge of the future. Protagoras is not claiming to be a better predictor than his disciples, as Campbell thought, for the disciples know their lack already. Protagoras claims to be more expert than other experts.

Tht. 180e1. Cornford (CR 49 [1935] 122-23) believed that this is a different fragment of Parmenides than DK28 B. 8.38, but for us, the issue is what Plato wrote. The new OCT prints OI)=ON A)KI/NHTON TELE/QEI TW=| PANTI\ O)/NOM' EI)=NAI. The MS. traditions of Plato and Eusebius and two references to Parmenides in Simplicius read OI(=ON as against OI)=ON of Anon. Berol. If Plato wrote OI)=ON, as Cornford thought (cf. OU)=LON, fr. 8 and MO/NON, Simpl. in Phys. 29.17), we would have a faulty hexameter unless we read T' E)/MENAI TW=| PA/NT' or the like from fr. 8. But against elision of iota in dative, cf. B.M. Perry (AGP 71 [1989] 2. n. 3). I incline to think that Plato meant OI(=ON, not as part of the quotation but as a transitional word (so Dias et al.). In any case, the apparatus gives much useful material.

Tht. 209e5. I do not see the need for adoption of conjectures by Badham (EI) DE/ GE) and Hicken (W(S E)RW=N <E)/T'> E)PU/QOU). BetaW's EI)/ GE DH\ could have arisen by confusion of maiuscule letters from EI)PE/ DH\, the reading of TWim, which I would print. BD contain many such errors.

Soph. 221a3. Robinson chooses Herodian's TERA/MWSIN, "reeds," following the theory of Wilamowitz (Hermes 35 [1900] 544-45) that KALA/MOIS is a gloss (W. spelled TERA/MOUSIN). Unlike the case of Tht. 151c7 (cf. supra), the indirect tradition should be followed, for Herodian expressly cited this rare word.

Soph. 230c7. Robinson esteems the authority of Stobaeus less highly than did Burnet. Here, he prints the MSS.' perfectly good E)N AU)TW=| against Stobaeus' E)NTO/S. Cf. similarly at 263d7, where GE/NH is preferred to Stobaeus' GE.

Soph. 235e6. The OCT rightly adopts Badham's KW/LWN, "... the true proportion of the limbs," over KALW=N of the MSS. and many editors. The Stranger at 236a4-6 makes the point that the symmetry or proportion itself of the skilfully represented figure seems beautiful at a distance. Beauty is attributed to the symmetry, not to the various parts severally; although the parts may be beautiful in themselves, the MSS.' reference to the beauty of the symmetry of beautiful parts seems unnecessarily redundant. Confusion of alpha and omega can occur even in maiuscules.

Soph. 247a5. The OCT prints Cornford's supplement H)\ FRONH/SEWS, motivated by the presence of this term in the argument at 247a3 and b1, but it is not necessary. DIKAIOSU/NHS serves as synecdoche for the virtuous state of soul that Plato has in mind (cf. TH=S A)/LLHS A)RETH=S, b2). Against Campbell, to oppose the plural E)NANTI/WN (a7) to DIKAIOSU/NHS is not un-Platonic.

Plt. 259d4-6. Although I disagree with many of Robinson's emendations, I concur that these lines are out of place. The inference in d4-6 does not follow from what immediately precedes it but from 259a6-b5. What is more, it breaks up the discussion of the ruler's knowledge that began at c1, without adding anything that has not been concluded previously. Instead, d4-6 form a bridge from the practitioners of statecraft (through 259b5) to the single knowledge according to which they are all classified in the same category. The Stranger will then go on to ask questions about that knowledge. Robinson's transposition to follow b6 or Sandbach's slightly different solution (ICS 2 [1977] 50-52) obviates these problems. It is worrisome, however, to lack an obvious cause of this dislocation; there is no real haplography, and we are not dealing with discrete verses of poetry, which can easily become displaced in a codex.

Plt. 269e4. The argument of 269d-270a seeks to explain why the universe at times revolves backwards. As Campbell saw, A)NALU/KLHSIS = TO\ A)NA/PALIN I)E/NAI (or POREU/ESQAI) (269d2, 270a7), i.e. the change of direction which occurs when the cosmos is released by god to begin to move backwards on its own (cf. E)C A)NA/GKHS ... GE/GONE, 269d2-3, with EI)/LHXEN, e4, and 269e5-270a9). AU)TOU= of the MSS. and Eus. cannot stand, for A)NAKU/KLHSIS is not an alteration of the universe's movement of itself. Robinson [supra, n. 7] 45 gives a good review of the problems, but Baltes' A)PO\ QEOU= (cf. 269c5) is paleographically more plausible than R.'s PRO\ TOU=. Or perhaps just AU)TOU=?

Plt. 284d7. We are told in d7-8 that "if one of these two does not exist, neither of the two will ever exist." What are "the two"? The MSS. give us more than two candidates in d6-7: TOU/TOU TE GA\R O)/NTOS (i.e. TO\ ME/TRION) E)KEI=NA E)/STI (i.e. the greater and the less, d5), KA)AEI/NWN OU)SW=N [AI( TE/XNAI, d4] E)/STI KAI\ TAU=TA (i.e. the greater and the less). But it seems intuitive that E)KEI=NA/KA)KEI/NWN should denote the same antecedents; for if not, the greater and less become derived both from the mean (TOU/TOU) and from the crafts (E)KEI/NWN, reading the MSS.' TAU=TA), a conclusion that would lead nowhere. We know from c1-3 that the mean is necessary for crafts, and the feminine participle in d7 is more likely to be genuine than is E)KEI=NA E)/STI, given MS. confusion between E)STI/N and EI)SI/N. Robinson makes the right choice to print E)KEI=NAI EI)SI/ with Madvig and TOU=TOU with Cornarius and earlier editors. This makes the crafts and the mean interentailing (cf. J. Annas and R. Waterfield, Plato. Statesman [Cambridge 1995] 43 n. 39).

Plt. 290a3. More interesting than TINE/S GE of Robinson's emendation is the MSS.' TH=S GE: "I suppose they (i.e. the merchant class) might perhaps [dispute something] about the ruling art that pertains to commercial matters, at any rate."

Plt. 300a8. A trivial case of applying stemmatic principles: cf. T's A)LHQE/STATA/ GE, which Robinson prints, to BetaW's A)LHQE/STATA, preferred by Burnet. TW agree closely in this part of Plt., so that W's agreement with the b reading accords it probability. For superlative adverbs without GE as formulae of assent, cf. Soph. 241e6, 246e1, 249d5, Plt. 260d10, etc.

Plt. 301b7-8. Robinson transposes DIA\ (sic BDTW: DI' A(\ Laur. 85.6) ... GE/GONEN to follow DHMOKRATI/A at c7, presumably because the five types of constitution are enumerated there. But this is not sufficient reason, for analysis into five types was already begun at 291d ff. Waterfield, to whom Robinson had generously shown the as-yet unpublished Greek text, also does not follow this transposition.

A different problem is the text of b7-8, which is incoherent in the MSS. The best solution is to read DI' A(\ because the true king, the one who possesses knowledge, is still called BASILEU/S, the five names of previously-discussed constitutions collapse into one, i.e. king (the ruler stands for the constitution). This is because the ideal government is that run by a single expert. True kingship stands in contrast to the five other polities re-enumerated at c6-7, which are only second-best ones because they are ruled by sovereigns who rely on less than expert knowledge. Therefore these do not contribute to an examination of what the ruling art is (cf. 303c ff). Robinson's apparatus shows other solutions that amount to reading PA/NTA O)NO/MATA POLITAIW=N ... PE/NTE MO/NON GE/GONEN. This will not do, for there never were more than five names of polities, even though seven types of sovereign are discussed (cf. 302c).

Plt. 302e1, 303c9-d1. Robinson's emendation of DIPLOU=N to A(PLOU=N is unnecessary. The point is that even if rule of the many already has a two-fold name along the lines of rule of one or of few, the principle of division must still be whether government of the many follows the laws. Also unnecessary is the nominative SATURIKO/S TIS QI/ASOS.

Plt. 310b10. To emend GE/NH ("houses, family lines") to H)/QH with Stallbaum, under influence of H)=QOS at c9, loses sight of how the people contracting marriages in 310c-d are concerned to match family stock (not necessarily "dynasties," against Robinson [supra, n. 7] 41), a sphere of concern that includes other things besides character traits. These people stand in contrast to those in b7-8, who only consider external benefits of the marriage like wealth or political advantage. "People at present are too clannish in their alliances" (Campbell).

Finally, the OCT usually follows Burnet's orthography. Although the apparatus often shows where the MSS. have CUN- and CUG-, these variants deserve to be taken into the text as well, for the earlier MSS. are generally to be followed in matters of spelling, given later tendencies to regularize.9

Does this new volume put the text on a truly scientific basis? Substantially, yes. The accuracy of the apparatus is superior, and I agree with the editors' construction of MS. families. In particular, we profit from new and better knowledge of the W group. Not all the MSS. of tetr. i-ii have been collated and classified, however, so that this edition does not quite meet Dodds' challenge to the next editor of the Opera Omnia. I would have listed more errors shared by primary witnesses and would have given more weight to "Maasian" stemmatics when contamination is not likely. We will all agree or disagree variously with the editors' choices of readings, but they are carefully thought out, and the apparatus gives a wealth of material that will help the reader to make a judgement. The task of producing a new Plato is enormous, and the editors have greatly advanced our knowledge of his textual tradition.10


NOTES

1. For a fuller discussion of the editorial program, cf. W.S.M. Nicoll, "The Manuscript Tradition of Plato's Statesman," in Reading the Statesman. Proceedings of the III. Symposium Platonicum (=International Plato Studies 4) (Sankt Augustin 1995) 31-36.

2. This date has been proposed by G. Prato, "I manoscritti greci dei secoli xiii e xiv: Note paleografiche," in D. Harlfinger and G. Prato, edd., Paleografia e codicologia greca I (Alessandria 1991) 139-140.

3. Cf. Mnemosyne 45 (1992) 312 n.3 (with citations of previous discussions); BMCR 94.1.7; Murphy and Nicoll, Mnemosyne 46 (1993) 459 n.9.

4. For Plt., cf. Nicoll (1995) 33 n.4. For the benefit of those who have not yet bought the new OCT, I use Burnet's line numbers in this review.

5. On correctors of Q, cf. the remarks in Murphy and Nicoll (1993) 467 n.22.

6. Cf. similarly D. von Dornum and M. Haslam, ZPE 89 (1991) 13-14. On source switiches, cf. Nicoll CQ n.s. 25 (1975) 41-47, G. Boter, CodMan 13 [1987] 144-55 and now Nicoll (1995), who disagree on whether W or T changed sources; cf. Murphy and Nicoll (1993) 461-468 on Q. Regarding MS. relations, Boter argues that W shows affinity with B against T in tetr. i, Cra. and Plt. up to 288, affiinity with T against B in Plt. 288-end, and stands alone in Tht. and Soph., displaying frequent agreement with Anon. Comm. (as noticed by Diels and Schubart). A. Carlini held the tradition to be tripartite in Phdo.; cf. Studi sulla tradizione antica e medievale del Fedone (Rome 1972) 170-1.

7. In the 350 Stephanus pages of tetr. i-ii, excluding orthographical matters, I have counted six different readings in the new OCT Euthphr., 33 in Apol., eight in Crito, 60 in Phdo., 90 in Crat., 37 in Tht., 66 in Soph., and 77 in Plt. D. Robinson gives a list of those in Plt. in "The New Oxford Text of Plato's Statesman: Editor's Comments", in Nicoll (1995) 37-39.

8. I am eager to see how the editors will use it in Prm., for Brumbaugh (RevMeta 26 [1972] 142-44; 1982, passim) thought that W uniquely avoided many contradictions that lace the argument, and that some of these correct readings are also found in the Proclus lemmata and in Damascius. Of course, some ancient scholar may have "improved upon" what he read, so that W must be evaluated carefully.

9. Cf. L. Taran, Academica: Plato, Philip of Opus, and the Pseudo-Platonic Epinomis (Philadelphia 1975) 174-75.

10. Although neither of us has modified his views, I have found it profitable to exchange drafts with Michael Haslam after we had written our reviews, and I am grateful to him for pointing out some ill-considered expressions in mine. -- D.J.M.