Our knowledge of the manuscript tradition of Plato has been deepened greatly during the last generation as scholars have undertaken to collate and classify all the manuscripts of given dialogues. One thinks of the work of Carlini (fourth tetralogy and Phdo.), Moreschini (Prm. and Phdr.), Berti (Crito), Moore- Blunt (Epp.), Slings (Clt.), Boter (Rep.), and Jonkers (Tim. and Critias), among others. As these studies have been published, the stemmatic position of various codices has become clear, and several more primary witnesses have come to light. At the same time, advances have occurred in Greek paleography with the delineation of certain styles of writing and the identification of many hitherto unknown copyists. The study of the manuscripts continues in turn to expand our insight into the scholarly habits of known scribes and humanists. In this ambitious work, a revised version of his doctoral dissertation, Christian Brockmann seeks to approach the textual tradition of Plato from all of these starting points. The primary goal of his book is to identify all the manuscripts of the Symposium and work out their stemmatic relationships. As one might expect of a student of Dieter Harlfinger, however, Brockmann also approaches the MSS. as phenomena in themselves. He seeks to revise dates on paleographical grounds, to identify scribes or catalog identifications of scribes made by others, and to shed light on the history of Platonism by studying how copyists and correctors worked on the text. Engagingly written, containing fascinating excurses into the critical achievements of such figures as Planudes and Bessarion, the result addresses a wider audience than would a strictly stemmatological study. The book is worth owning for its contribution to the history of scholarship and for its compilation of information about the individual MSS. It comes as a disappointment, then, to discover that Brockmann’s arguments for his proposed stemma are often incomplete and in some cases reach untenable conclusions. Future editors of the dialogue will have to check his results.
The book is divided into four sections: an introduction and statement of method; a short history of the most notable scholars who have worked on the text of Plato since the Renaissance (Part I); a list of the 55 MSS. and one papyrus that contain Smp., with codicological information and bibliography for each (Part II); and the bulk of the work (Part III), in which the filiation of MSS., divided into three families, is worked out. Brockmann also classifies the first four printed editions of Plato, as well as Ficino’s Latin translation. A final section of Part III explores the relationship of the primary witnesses to each other and to the papyrus, P. Oxy. 843. There follow bibliography, indices, plates of nearly every codex, and a stemma.
Brockmann collated most of the MSS. from photographs but checked Venetian MSS., the Clarkianus, and Zittaviensis A 2 by autopsy. Although he did search each MS. for erasures, corrections and the like, Brockmann restricted his collation of derivative MSS. to four sample passages of about two Stephanus pages each (there were a few exceptions) and collated about two-thirds of the text of the dialogue in the three MSS. which he regarded as primary witnesses (viz. Bodl. Clarke 39 [=B], Ven. append. cl. IV. 1 [=T], and Vind. suppl. gr. 7 [=W]). This weir may be broad enough to trap all the fish; time is short, and researchers need to judge when to stop collecting data. However, in a work that aims to provide the basis for a new text, one expects the collation at least of key witnesses in their entirety—especially when the author faults the accuracy of the apparatus of previous editions. Theoretically, the textual critic should collate all the manuscripts in full. Although that may be too much to ask in the case of a dialogue of Smp.’s length, some Platonic MSS. do display source switches in the middle of dialogues. Further, since the Platonic MSS. are heavily laced with horizontally transmitted variants, there may be important material lurking outside the portions of text that Brockmann did collate.
With this said, it is best to go directly to Part III. Brockmann is a remarkably accurate collator. I have checked the greater part of his citations of two MSS., B and Marc. gr. 185 (=D for most editors, although Brockmann does not use this siglum), and I have found no false reports. Second, Brockmann makes good use of layers of correction to trace the genesis of apographa, as he does for example with the copies of Par. gr. 1808 (p. 164ff.). He appeals skillfully to proofs based on physical evidence (e.g., on Par. gr. 1810 and apographa, p. 105f.).
On the other hand, Brockmann’s arguments for the derivation of one MS. from another are often incomplete. Only in about half of the cases does Brockmann catalog conjunctive errors that link the copy to its supposed model; one must otherwise presume that they exist. Brockmann’s tendency to regard insignificant errors as significant does not instill confidence, however; cf. e.g. his appeal to errors involving confusion of δή and δέ (p. 189), omission of τε (p. 155), or misreading of the compendium παρά (p. 192). Too often, Brockmann will merely assert that a given MS. Q is a copy of some MS. P and then list separative errors of Q. But these only prove that P is not a copy of Q. This is done for Perus. F 56, Lobcovicianus, Vat. gr. 1029, Ambr. D 56 sup, etc. Several times, Brockmann acknowledges the riskiness of appealing to omissions caused by homoioteleuton (pp. 57, 162, and elsewhere), but he appeals to them nevertheless, sometimes as significant errors. While theorists differ on whether textual dislocations occasioned by homoioteleuton can be used as evidence at all (I avoid so using them), Brockmann accords weight to them in an inconsistent fashion that leaves his theoretical standpoint obscure.
The bulk of Brockmann’s conclusions about the filiation of MSS. in Smp. match the findings of students of other dialogues, and despite my reservations about his methodology, I agree with the greater part of Brockmann’s results. Brockmann’s MSS. families are the familiar ones: 1) the group of B; 2) T and its apographs, all but two of which descend through Par. gr. 1808; 3) the family of W. However, some conclusions are either incorrect or not fully proved. In what follows, MS. readings come from my collations from microfilm of DTWP and from the photofacsimile edition of B.
Marc. gr. 185. Brockmann argues that D is a copy of B, made through a lost intermediary. This is a controversial claim, for the weight of scholarly opinion has shifted in recent years to hold that D and B are twins. Brockmann’s proof of D’s dependence is this (pp. 52-59): 1) he finds no separative errors in B against D and T, P or W; 2) he points to five errors in D, occasioned by homoioteleuton, for which the relevant key words in B stand one or two lines directly below each other, and argues that five are too many to have occurred by chance; 3) Brockmann sets aside the force of Schanz’ and Berti’s arguments based on incomplete but differing sets of stichometric numbers in the two codices by supposing that the numbers which are present in D but lacking in B may have been added in an intermediary. Now, Brockmann is not quite right to say that B presents no separative errors against D, for there are a few (excluding matters of orthography, etc.): 190a8 ἀπερειδόμενοι bTW: ἅπερ εἰδόμενοι D: ἅπερ εἰδόμεναι B; 195b3 ἔρως bDTWP: ἔρωτος B; 203c6 πένης bDTWP: πενίης B; 212e5 ἤλθομεν bDTW: ἤχθομεν B. The corrections in B are clearly not by the first scribe and appear to be made by a person later than the contemporary corrector, B2 (who was presumably Arethas). Even if they were made by B2, however, this would not help Brockmann’s thesis, for he is committed to the view that D’s source was copied before B was corrected. This is because D shares with Bac numerous silly errors that were corrected by B2 or later hands (p. 53ff.). Creating suspicion about D’s descent from B are errors of accents and word division of the kind that arise from early MSS. incompletely furnished in these regards, in places where B is substantially correct: cf. e.g. 187c8 δέῃ B: δὲ ἡ D; 189d7 ἀλλ’ ἀλλοῖα] ἀλλ’ ἀλλόι B: ἀλλάλλοια Dac; 193c4 ἐκτελέσαιμεν B: ἐκτελέσαι μὲν D; 195d4 ἐπ’ οὔδεος B: ἐπυδεος D2ac; 218c4 ἦ δ’ ὅς] ῆ δ’ ὄς B: ἦδος Dac; 219a5 τὰ μὲν B: ταμεν D. All of this material more consistently fits the thesis that D (or its source) was copied from B’s exemplar than it fits Brockmann’s view. The five textual dislocations mentioned above are striking, even granting that homoioteleuton is their cause, but they could have arisen from any MS. with lines of the same length as those of B. Brockmann should have adduced omissions that are not due to homoioteleuton but yet constitute entire lines in B or are otherwise caused by its text-form. I have collated all of B and D in Smp. as well as in various other dialogues and have not found any examples. Brockmann grants that D may be independent in other dialogues, but I do not believe that D is a remarkably close twin of B elsewhere and yet is a copy of it in Smp.; surely D’s extreme coincidence with B in even trivial errors over a wide range of dialogues shows that it must be one or the other. That B displays so few separative errors in a dialogue of this length is indeed disquieting, but Brockmann has not proved his case, and evidence from other dialogues better supports the opposing position.
Vat. Pal. gr. 173 (=P). Brockmann is wrong to classify the fragmentary P with T in the second MS. family and not with W in the third family, where students of other dialogues have placed it. Brockmann’s three conjunctive errors are of no value, for they amount to omission of δή, ἐπί and τε (p. 154f.). No codex omits more small words than does P, a scholar’s working copy that is marked by a concerted effort to abbreviate. It is premature for Brockmann to dismiss a conjunctive WP error reported by Vicaire-Laborderie (220c8 τελευτῶντες] καὶ τελευτῶντες WP) for, despite his statement that the ink at this spot is almost completely faded, I found when working on P in situ that it was possible to read many words that do not show up in photographs, since traces of ink or even the impression made by the pen are often still discernible. Brockmann does not seem to have visited the Vatican during his research; the apparatus of Vicaire-Laborderie may yet be right here. Second, I cannot find discussion of four other conjunctive WP errors that Vicaire-Laborderie report (I have checked and corrected their readings): viz. 211a6 αὐτῷ BDT Oxy: αὐτὸ WP; 211b1 ἐν τῳ TWit Oxy: ἐν τῷ BD: ἐν τινι WslP; 211c3 ἐπαναβασμοῖς T Oxy: ἐπαναβαθμοῖς WP: ἐπ’ ἀναβαθμοῖς B: ἀπαναβαθμοῖς re vera D; 216d6 γεγλυμμένος T2: ἐγλυμμένος re vera B sine accentu, DT: ἐγγεγλυμένος dWP. One must regard P as a relative of W not T, a conclusion which may influence editors in their choice of certain readings.
W and Lobcoviciensis VI.Fa.l. Brockmann accepts the argument of Lidia Perria that W and Lobc were copied in the eleventh century by the same scribe, and he views Lobc as a copy of W (pp. 237-247). Brockmann is correct on the derivation of these MSS. and on the date of W, but his view on the identity of their scribes cannot be right. Fol. 256, offering Smp. 173c2-175bl, is inserted along with several other leaves in W by a hand (W3) that is at least s. xiv, and Brockmann himself inclines to s. xv (p. 237; he claims to have discovered the insertion of fol. 256, but it had already been reported by Post). Brockmann asserts without proof that W3 copied this leaf from Lobc. Actually the relationship between W3 and Lobc is the other way round, as Carlini and others have recently shown. In fact, W3 copied Esc. Y.I.13, s. xiii ex., save in Tht., where the source of W3’s fol. 139, which Lobc still copied, is D or a copy of it. Proof of the latter assertion must await another occasion, but cf. one tiny correction which the first scribe of Lobc made at Tht. 201e1: οἱονπερεὶ Lobc1pc: οἷονπερεὶ sic T: οἷονπερ BDW3Lobcac. The scribe erased the circumflex and added the grave accent and a hyphen below the line—the pen, ink, and type of stroke are clearly his. Lobc is not W3’s source but vice versa. W3 has been shown to postdate A.D. 1314, so that Lobc must be later. The contributions of Carlini et al. (note 2 above) were published after Brockmann’s book, but the fact that the text of W3Lobc lurches into the tradition of Esc in Smp. 173c-175b should have alerted Brockmann to the implausibility of the progression W1-Lobc-W3.
Perusinus F 56. This little-known codex, written in s. xv by Theodore Gaza, contains only Smp. Brockmann asserts that it is descended from T independently of Coisl. 155 or Par. gr. 1808 (p. 160f.). The only evidence Brockmann offers of this, however, is an omission in Perus occurring at 212a4-5, occasioned by homoioteleuton and word position in T. Otherwise, he cites only separative errors that show that T is not a copy of Perus. Brockmann’s citation of -αν instead of -ας at 173c3 does not really prove anything. Perus deserves further study, for Gaza may have incorporated interesting readings into the parts which Brockmann did not collate.
Brockmann’s derivation of Marc. gr. 189 from Marc. gr. 590 in Smp. creates no problems. However, I am not convinced that 189 was copied in s. xiv (p. 126), for in Chrm. it appears to be a copy of the fifteenth-century Laur. 85.9. This would explain why 189’s scholia in Smp. share separative errors with those of 85.9, without the need for Brockmann’s hypothesis that 189 got them from a lost intermediary that stood between 85.9 and its source, Laur. 59.1 (pp. 129f, 231). Enlarged, rounded letters in mid-line, nu’s that do not end in a pure point, and absence of final sigma of the form ς are features that make 189 look earlier than s. xv, but some later hands still display them; all these features occur in the script of Ioannes Rhosus, Bessarion’s amanuensis. To dispose of the need for an intermediary between 59.1 and 85.9 makes it easier to accept the thesis suggested by Aubrey Diller and elaborated by James Hankins, that 85.9, Ficino’s principal Plato exemplar, was copied from 59.1 in Italy, probably during the time of the Council of Florence.
Many readers will be most interested in Brockmann’s conclusions about the lines of the tradition that extend back before the earliest MSS. Brockmann finds that TWP (their presumed source = ϕ) stand in a group against B, as they do in other dialogues, so that we have a bipartite stemma (p. 248ff.). Brockmann is right to point out that for this reason, the readings of B (I would say BD) and ϕ have equal weight from the point of view of recension. The text of Smp. was not heavily reworked in antiquity, and Brockmann is basically right to aver that the greater part of the variants that divide B(D) from ϕ arise from misreading of maiuscule letters. Although he mentions Oxy, I wish that Brockmann had discussed what seems to me, from a perusal of Burnet’s apparatus, to be a bias on the part of the papyrus in favor of the readings of B; if true, that would point to a split in the tradition before s. ii-iii. Brockmann theorizes that φ was an early minuscule MS., which must have been copied before c. A.D. 950 (T’s approximate date) and which could have been the lost first volume of Par. gr. 1807 (=A; p. 251f). Whether WP were copied from A is speculative, and in this connection, I would have liked to see Brockmann step outside the boundaries of Smp. and address the question, why T and W differ so widely from dialogue to dialogue in the degree of their respective divergence from Β. Nicoll and Boter had already published on this problem prior to 1990.
Brockmann is at his most stimulating when he delves into the history of Platonic textual work. In addition to cataloging scribal identifications made by others, Brockmann comes up with about eight of his own: e.g., Matthew of Ephesus copied Albinus in Vat. gr. 225, Leonardo Bruni appears to have made marginal notes in Lobc, Gemistus Plethon annotated Marc. gr. 189. I am not convinced that the scribe of Lobc also copied Vat. gr. 1029. One of Brockmann’s most provocative claims is the view that Marsilio Ficino translated the Symposium, not from Laur. 85.9, as he seems to have done in other dialogues, but from Ricc. 92 or its exemplar, which Brockmann holds was a lost copy of Malat. D 28,4. Brockmann suggests that this lost copy may have been the manuscript “in carta bombicina” which Ficino borrowed from Amerigo Benci (pp. 220-29). His examination of Ficino’s tendency to translate one Greek word with two Latin words will be useful to those working on Ficino’s versions. Brockmann’s analysis of the textual scholarship of Moschopoulos, Planudes, George Pachymeres, Bessarion and Musurus reveals these men to be the sources of numerous well-known variants, most of which Brockmann takes to be conjectures, although some bespeak an origin in other exemplars, as, for example, variants which suggest that Bessarion had access to T and Coisl. 155. Not surprisingly, Brockmann finds in the work of Bessarion the highest point reached by pre-Aldine scholarship.
Part I, although it opens the book, actually brings this history of textual scholarship up to our own day. Brockmann’s aesthetic and textual judgements of the early editions are well balanced. Stephanus comes into criticism for citing variant readings in the margin of his edition with the sign γρ(άφεται) or “in vet. lib.”, when they actually are transmitted only in Arlenius’ edition and probably came to Stephanus from there. We are treated to instructive and gently amusing accounts of the controversies that flared up between Schanz and Jordan or Schanz and Wohlrab. Although this is an engaging part of the book, it addresses the history of scholarship more than it does the constitution of the text of Smp., and one feels that the space might have been better devoted to more complete treatment of the relations between the MSS. The same can be said of a five-page digression on the alternative version of the text at Cra. 438a and the views of Jachmann (p. 64-68).
Part II’s list of MSS. provides date, material, number of folia, size, contents, scribe if known, together with name of the person who made the identification, and short bibliography. Together with the lists in the recent works of Boter and Jonkers, this list begins to lay the ground for a Plato MS. “Repertorium”. It would have been useful had the respective page numbers of Part III, where each MS. is discussed, been included in each entry; the plate numbers do appear. It also would have been easier for students of Plato if Brockmann had used commonly accepted MS. sigla such as D, Y, etc., for some of the better known codices.
Philologists and paleographers all too often fail to appreciate each others’ methods fully or to make proper use of each others’ results. Brockmann is to be congratulated for setting out to combine both perspectives. He would have made a greater contribution to our effort to establish the text of Plato had his collations been much more complete and had he marshalled his evidence more judiciously. One looks forward to future work by this promising scholar upon the MSS. of ancient philosophers and upon what they reveal of our own intellectual history.
 W.S.M. Nicoll, whose team is preparing the new OCT, informs me of one: Tht. 187c7-8 ἆρ’ … λέγεις om. D (50 letters). However, he has found other omissions of the same length in that dialogue and others which cannot have arisen from copying B. Nicoll points out to me that D’s omission of Smp. 194a2-4, a passage that Brockmann uses to prove D Cannot be a direct copy of B (p. 59), is of this same length (52 letters).
 Cf. D.J. Murphy, “The Independence of Parisinus Gr. 1813 in Plato’s Phaedrus, Hipparchus and Alcibiades II,” Mnemosyne 45 (1992), 312 n. 3 (with citations of earlier contributions by Carlini, Moreschini and Berti); D.J. Murphy and W.S.M. Nicoll, “Parisinus Graecus 1813 in Plato’s Cratylus,” forthcoming in Mnemosyne 46 (1993), n. 9; also Praefatio of forthcoming OCT, edited by Duke, Hicken, Nicoll, Robinson and Strachan.
 W.S.M. Nicoll informs me that he too was able to read many faded words in P by autopsy, which do not show up in photographs.
 A. Carlini, “Le vicende storico-tradizionali del Vind. W e i suoi rapporti con il Lobcoviciano e il Ven. gr. Z 185,” Studi su codici e papiri filosofici (Firenze 1992), 26-30. For the controversy over Lobc as a whole, see the positions of Perria and Canart vs. Carlini, Berti, et al., in those Studi, 11-143, and D.J. Murphy, “The Plato Manuscripts W and Lobcovicianus,” GRBS 33 (1992) 99-104. Nigel Wilson is planning further work on the MS.
 W. Marg, Timaeus Locrus, De natura mundi et animae (=Philosophia antiqua 24), Leiden 1972, 20.
 Cf. D.J. Murphy, “The Manuscripts of Plato’s Charmides,” Mnemosyne 43 (1990) 330-31. Significant, common errors link 189 to 85.9 against the latter’s exemplar, Laur. 59.1. 189 displays additional errors against 85.9; it is correct against 85.9 only in cases of easy corrections. There is no evidence that the two MSS. are twins. A. Carlini found 189 to be “molto probabilmente” from 85.9 in Hipparch. and Rivales—Platone. Alcibiade, Alcibiade secondo, Ipparco, Rivali (Torino 1964), 37 n. 86. It is copied by the same scribe throughout.