E.A. Duke, W.F. Hicken, W.S.M. Nicoll, D.B. Robinson, J.C.G. Strachan (edd.), Platonis Opera, tomus I. Oxford University Press, 1995. Pp. xxxii + 572. $29.95. ISBN 0-19-814569-1.
Reviewed by Michael W. Haslam, UCLA, firstname.lastname@example.org.
Here we have vol.1 of the new OCT of Plato, replacing Burnet's. It is no longer a one-man undertaking, nor even an all-male one: this volume is the work of five editors, and the volumes still to come will involve yet others. But it is a unified project, and executed with notable editorial consistency.
Burnet's text has always been very serviceable, and could continue to be so. Its successor's claim to distinction lies less in its actual text than in its apparatus, which is more precise and more accurate than Burnet's, and cites more manuscripts. It must represent an unimaginable amount of labor. The textual rewards, however, are slender. Rarely do the additional manuscripts expand the range of attested variants, and the new editors' choice among them is usually the same as Burnet's. At the beginning of the Phaedo (57a6), where Burnet reported simply "E)/GW B: om. T" (B meaning B and W), the new edition reports "E)GW\ BCWPSVL: om. DTQ." Most users, if they pay the apparatus any heed at all, will be content merely to register that the tradition is split -- which is indeed the significant thing to register; some will fall into the trap of thinking that E)GW\ has more authority because more manuscripts are now listed as carrying it. The user who takes the trouble to consult the manuscript list that stands at the front of the dialogue will discover that D is held to belong to the same family as BC, and Q to the same family as WPSVL, and will thus begin to appreciate just how complicated things have become.
The manuscript situation is surveyed in the (latinized) preface. Much good work has been done on the manuscripts in recent years, and it is convenient to have the results encapsulated here -- or such of them as pertain to the dialogues of the first two tetralogies, for the preface rigorously confines itself to the contents of this one volume (Euth., Ap.Soc., Crito, Phaedo; Crat., Theaet., Soph.-Pol.). The exclusivity is understandable, but it seems a pity that the reader is given no indication of the fact that in this as in most sectors of the Platonic corpus the medieval tradition has a considerably narrower range of access to ancient readings than in those dialogues where the 13th-cent. F comes into play. Some contextualization would have been welcome.
The familiar three "families," B, T and W, have now become Beta, T, and Delta: the picture is more complex than it was, but remains basically the same. T, anachronistically described as heading the "altera" family, is still on its own (lots of descendants, but no siblings) but is now known to be earlier (mid-10th cent.) than used to be thought. Since it is not just a manuscript but a family -- that is, independent of Beta and Delta (but wholly, or only partly?) -- it might have been better to cite it in bold as T, to show its status. The renowned B, written for Arethas in 895, is joined by C (only some dialogues) and D, which tend nowadays to be thought -- on what grounds, the editors do not say -- to derive not from B itself but from B's exemplar, and are thus elevated to the status of "primary" manuscripts, to be cited no less systematically than B itself. It would be heresy to suggest that some manuscripts are more primary than others, but certainly the value of C and D is slight. The reason for not ignoring them, however, is surely not the one given by editors, namely that when they agree with B we probably(?!) have the text of their common source; on the contrary, their value should be in cases where B miscopies and CD do not; but since CD's differences from B may be due to adoption of readings from elsewhere ("contamination"), it is a slippery business. The editors use Beta to designate the hypothesized manuscript from which BCD derive. They are willing to assign to it readings not carried by all three, but it is not clear how they think they can then recognize Beta readings; the discomfort which they show on this point on p.vii has disappeared by the time they get to p.xix. In practice they appear to cite the manuscripts separately when there is any question of the Beta reading.
Even worse problems attend what is now designated the Delta family, composed of W and an interesting variety of other witnesses that variously come and go for various parts of the text. These supposedly derive from a single lost manuscript (Delta), a supposition which the editors do not substantiate. The earliest representative of this lot, if it is not an Armenian translation of Euth. and Apol., is actually B itself, or rather the contemporary variants entered in that manuscript; and a set of early corrections in T too is said to belong to the same family. Aristippus' Latin translation of the Phaedo is also assigned to it. The editors retain primacy of place for W, rather fatuously it might be thought, on the ground that it covers parts of the text that none of the other Delta witnesses do. They devote a disproportionate amount of space to an almost but not quite valueless 13th-cent. manuscript now labelled Q.
Relations among the three families are somewhat shifting, which to me suggests differential interplay among them in the initial phases of the medieval tradition. There are dialogues in which Beta and T are very close to one another, others in which they are far apart. And it might be suggested that the value of Delta fluctuates accordingly, for where Beta and T are split, Delta tends to be (if I dare say so) almost negligible. The editors clearly do not like "contamination." Instead of welcoming it as widening the medieval tradition's range of access to ancient readings (and as evidencing the dynamism of contemporary scholarship), they complain of it as obstructing one's view of the "real" relationships of manuscripts. To be sure, contamination prevents the construction of neat and tidy stemmata; but a manuscript has no less real a relation with one from which readings have been drawn by collation than with one that was directly copied. A statement such as "evenire potest ut codices, qui re vera aut nulla aut minima cognatione inter se adnexi sunt, tamen speciem quandam propinquae affinitatis praebeant" (xv) seems to rest on an unduly restrictive notion of what constitutes "cognatio" and "affinitas."
In fact the editors are so preoccupied with the problem of disentangling manuscript interrelations, a singularly unprofitable business, that they neglect to say anything at all about the medieval tradition as a whole, or about its relation to the text in antiquity, beyond a throwaway acknowledgment late in the preface that papyri and the indirect tradition sometimes preserve the truth against the testimony of the codices. The preface's starting-point is the business-like statement that this volume's dialogues rest on the testimony of manuscripts belonging to three families, as follows ..., and it is not until the preface is nearly over that the reader who has persevered so far will discover that the text does not rest exclusively on the testimony of these manuscripts but that there are other witnesses too, not only older but in places superior. Clearly it will take more than my pointing out that P.Oxy.2181 (admittedly fragmentary) "can fairly claim to be our best manuscript of the Phaedo" (ZPE 89  1) to change the prevailing state of affairs, but the editors' failure to address the question of the medieval tradition's representativeness and worthiness is regrettable.
Just what is meant by a "family" (a metaphor ripe for deconstruction)? What integrity does each of the three of them have? What authority? How unitary is the tradition? Virtually nothing is said of such matters, and assumptions tend to be hidden. One of them pops into sight in the conclusion to the presentation of the three families (xv), where we suddenly encounter an incidental reference to "the archetype itself" (sc. of the three families), as if the existence of such a thing went without saying -- which is far from being the case. The range and distribution of variants in the early medieval tradition seem to me to point unequivocally to direct utilization of more than one ancient manuscript, and this is confirmed by the scattered incidence of agreement with papyri and quotations; and since collation of manuscripts was fairly extensive in antiquity, the readings of any two ancient manuscripts cannot realistically be ascribed to a single source.
The account of the tradition, then, is very useful, but only within narrow limits, and anyone concerned with transmissional questions will be disappointed. But a more honest and well-informed overview of the current state of play could not be found.
The text itself, as was only to be expected, is not very different from Burnet's, though it has clearly been thought through afresh at every point, and on balance is probably better. As an editor Burnet was fairly conservative, and with the exception of Robinson, who carries the primary responsibility for the refreshingly edited Soph.-Pol., the new editors tend if anything to be more so. It is symptomatic that like Burnet they decline to restore XRH\ U(MA=S for XRH=N (or XRH\N) U(MA=S in the opening sentence of the Apology (17a6; cf. Crito 45d5), and that unlike him they do not even record PRI\N <A)\N> A)NAGKH\N at Phaedo 62c7 (which I have little doubt is right; for the haplography cf. e.g. 109e3, 72c1, 84d2). Such extreme faith in the tradition is hard to justify. Burnet, a reader of Plato second to none, made numerous amendments to the text; the new editors make few, though Robinson's ventures are notable; my sense is that the field for successful emendation is far from played out. The new editors are not generous in recording conjectures, suppressing many that Burnet saw fit at least to report; this is to some extent counterbalanced by the occasional mention and even the rare acceptance of suggestions not recorded by Burnet (mostly pre-Burnet, but some more recent or new), but still the general effect is to convey a greater confidence in the trustworthiness of the medievally transmitted text than I would have thought warranted. Similarly, many phrases suspected of having been interpolated now have all taint of suspicion lifted from them. Even when interpolations have not succeeded in permeating the entire tradition, as with the construction-easing EU)LABOU/MENOI at Phaedo 91c4 or still more damagingly A)LHQH= A)\N LE/GOI at 99a7 (for the ellipse cf. Rep. 575d, Protag. 325d), not a breath of suspicion is allowed against them.
One good feature of the new edition is that the relevant manuscripts, medieval and ancient alike, are listed at the head of each individual dialogue. ("Sigla codicum" was perhaps not the most well-chosen heading, when most of the ancient manuscripts are not codices.) Another is the index testimoniorum at the end of the volume, and indeed the painstaking deployment of the indirect tradition throughout; most of the work on the indirect tradition is credited to Duke, while the lion's share of the investigation and collation of the codices was undertaken by Nicoll, aided principally by Robinson.
A departure from normal editorial practice is that words deemed interpolated are not square-bracketed but eliminated, appearing only in the apparatus. Burnet often square-bracketed even material that was carried by only part of the tradition, so the contrast is stark. It is presumably by oversight that square-bracketed words here and there remain in the text (at Theaet. 196b3 the word is actually absent from part of the tradition). While defensible enough in principle -- an editor should aim to present a text as close as possible to the original, should she not? -- the new practice makes for a less neat relation between text and apparatus and has the unfortunate effect of concealing major editorial interventions. It is inconsistent of the editors to retain the use of angle-brackets to signal editorially added words, but I am glad they do.
Angle-brackets enclose not only modern scholars' additions but also (sometimes) readings found in individual codices, the implication being that they are byzantine or humanist conjectures; and many an emendation previously assigned to a modern scholar now has some manuscript cited for it instead. Some of the most important manuscripts for these purposes -- vehicles of contemporary conjectures rather than of independently inherited readings -- are listed in the preface (xviii), but with no indication of date or scholarly provenance; so all work on the text prior to the age of print is reduced to uniform impersonality. While unfailingly more precise than Burnet in its manuscript references, the new edition can be less informative: few will know that a given manuscript was Bessarion's. But it must be reckoned one of the major contributions of the new edition that readings which Burnet was content to assign merely to "al." or to leave unassigned are now pinned to the particular manuscripts in which (one hopes) they first appear.
Eliminated from the apparatus are manifest mistakes in one or another witness or family. An enormous quantity of Burnet-reported readings is thus swept off the scene. For an OCT's "brevis adnotatio critica" this is only proper, but the editors' rhetorical justification ("Quid enim prodest lectori ...?" xix) is easily answered. The reader of the new OCT gets no clear view of the nature of the tradition. The fact that B is transliterated from a pre-minuscule exemplar is now hidden, and the thought is less likely to occur that an A)\N should perhaps be DH\.
Discrepancies with Burnet's reports are perhaps less frequent than might have been expected. Presumption of correctness will lie with the new edition. But the presumption will not be quite safe: at Theaet. 206d7 the new edition prints AU)= without registering its absence from any manuscript: previous editions have reported it as absent from BT, and I have confirmed its absence from B. I have not spotted many other such errors, and I do not expect they are very numerous. Occasionally in evidence is a desire to give more than their due to minor manuscripts now ranked as "primary," betokening an understandable reluctance to admit the law of diminishing returns. At Soph. 221d4, for example, where PANTO\S is printed, the apparatus reports "PA/NTOS D: PA/NTWS BTW," which if correct is an unwarranted glorification of the wretched D: better "PANTO\S Winckelmann: PA/NTWS BetaTW" (cf. 237c2).
Burnet's punctuation has been intermittently revised, but not radically: the curious mix of medieval and modern is perpetuated. The extra pointing sometimes seems to me a bit on the fussy side, but punctuation is not a wholly scientific business, and my preference for less rather than more may be idiosyncratic. In e.g. "OU)KOU=N" FHSI/, "LE/GEIS...," however, the commas are otiose -- and worse than otiose, for is FHSI/ not enclitic? The new punctuation of Phaedo 92d4-5 is gratifying, but it makes me wonder whether there may not be other places where the correct articulation still awaits recognition.
To give an illustrative if not altogether typical sample of the new OCT vs. the old: In the first page of the Statesman the text printed is different at three points, a much higher rate than the average. (i) In the first line W's TE is accepted (TH=S TE QEAITH/TOU); the reading is not even reported by Burnet; it may or may not be right; (ii) at 257a3 GE, a supralinear addition in T, is accepted (bracketed by Burnet) -- a transmissionally bold choice, I would say probably wrong; (iii) at b6 ME\N OU)=N after PA/NU is eliminated, after Denniston (Greek Particles 480f.); the report of the deletion is very uneconomical, and inexcusably fails to record that the particles are present in P1, our earliest manuscript of the passage (a post-Burnet accession); this neglect of ancient evidence is fortunately abnormal. There are further differences in the apparatus, the most significant being (i) suppression of T's TOI for TOU= at a7, consistently with the policy of not mentioning clear errors, and (ii) report of a difference of word order in W at b5. The upshot is an arguably improved text (on the basis both of new data and of new thought), an evidently more accurate yet defective apparatus. Things may not have moved much, or enough, but they have not stood still.
The fact that Soph.-Pol. are a single dialogue (AJP 97  336-9) is not signalled. The sundering of the two parts is of course traditional, but no less distortive of Platonic literary form for that. Another regrettable habit of editors is that of plastering the names of a dialogue's speakers at the head of the text. And while I am about it, I take this opportunity of impugning the convention of inserting speaker-identifications within the body of the text. If the convention "is now regarded as essential" (N.G. Wilson, CQ 20  305), it ought not to be, for such extra-textual interference destroys the text's self-sufficiency and so does real damage to the way the text is read. In genuinely dramatic texts the practice is defensible, indeed highly desirable, inasmuch as it compensates for the degradation of the medium (the conversion of play to script), but Platonic dialogues are not dramatic but pseudo-dramatic, and identification of speakers is something that should be left to the reader to elicit from the text itself, part and parcel of the challenge built in to reading written dialogue. That said, the sad truth must be admitted that we modern readers are no longer capable of walking without the crutches that editors provide, and it would be deemed a dereliction of duty if OCT editors failed to provide them. But we should recognize that in a world true to Plato we would do without them.
There seems reason to hope that the examination and recension of extant textual witnesses have now been conducted with sufficient thoroughness that (an inevitable modicum of error apart) all significant surviving readings are on display and properly assigned. The selectivity of reporting, in conjunction with the contaminated and defective nature of the tradition, means that in principle practically any reported variant or conjecture has a chance of being the truth. If I have ventured some disagreements with the editors' textual choices,1 that does not mean that their judgment does not command respect. No edition is definitive, but unless there are major new papyrus finds, or the study of ancient Greek revives, it is unlikely that the world will ever have a better text of Plato than this.2
1. And could venture more. E.g., Soph. 264b6 PRAO/TERON (both T and Stobaeus) is not easy to explain as a corruption of the editorially preferred PRO/TERON, and seems to me right (they did not have to force their way through endless lines of defence, cf., the imagery of 261a-c, and for Theaetetus' PRAOTH/S Theaet. 144b4). While in this vicinity, at 265d6 I would put A)\N between XRO/NON and A)/LLWS.
2. For ordinary reading purposes, however, I have to say that I find the old OCT much better. This is simply because of the look of the text on the page.