Monday, February 22, 2010


Version at BMCR home site
Gaia Clementi, La filologia plautina negli Adversaria di Adrien Turnèbe. Studi e ricerche 76. Alessandria: Edizioni dell'Orso, 2009. Pp. xxxvi, 384. ISBN 9788862741460. €30.00 (pb).
Reviewed by Michael Fontaine, Cornell University

Better known as Turnebus (long penult), the Parisian professor Adrien Turnèbe (1512-1565) bequeathed us a gigantic miscellany of observations and interpretations of classical texts titled Adversaria, "random notes" (vols. I/1564, II/1565, III/1573). The most recent reprint, the 1604 Geneva edition, comprises 694 double-columned quarto pages plus indices (available online). All the good information in the Adversaria has long since been absorbed into our dictionaries, commentaries, grammars, and handbooks, the bad long since discarded. The book's chief claim on our interest today is that in it Turnebus makes repeated reference to a valuable manuscript of Plautus that is no longer extant. This is the famous "codex Turnebi" (also called codex Senonensis or fragmenta Senonensia), designated T, which antedated our earliest representatives of the Palatine family of manuscripts (B, C, and D). Clementi's new monograph, a revised Pisa dissertation, aims to check the readings commonly attributed to it. She brings an enviable set of skills and impressive determination to her project, which is a welcome contribution to Plautine studies.

T was a 9th-century parchment manuscript. It was already fragmentary when, at some point before writing his Adversaria, Turnebus obtained access to it, collated its variants, and apparently entered them as marginalia in a printed edition of Plautus in his possession. It is to this collation that he refers in the Adversaria when quoting some of its readings for Asinaria, Captivi, Casina, Curculio, Mostellaria, Poenulus, Pseudolus, Rudens, and Stichus. But Turnebus is never very clear about T. He constantly changes the words he uses to describe it, using such vague and varied expressions as antiqua volumina, antiqua scriptura, chartae, libri veteres, membranae, optimus liber, pergamenae, schedae, vetus liber, vetera exemplaria, and so on. It is obvious that some of these expressions could refer to other manuscripts or to printed editions--this is of course a common problem in dealing with humanist scholars (see Silvia Rizzo's Lessico filologico degli umanisti [Rome, 1973], rightly often cited by Clementi)--and it means we risk attributing to T readings that it did not contain.

That is not all. In 1898 W. M. Lindsay announced that he had discovered a new collation of T in the margins of a book in the Bodleian library; or, more precisely, he had discovered a flawed transcript of Turnebus's original collation of T.1 Moreover, these two sources--Turnebus's quotations and this new collation--supplemented each other. As Lindsay recognized, however, this new collation was larded with worthless variants from another source, and it is not always clear which are which; and there are a number of other problems as well.

Since Clementi presupposes familiarity with Lindsay's arguments, which for reasons both historical and stylistic are no longer easy to follow, I summarize them here, taking this opportunity--missed by Clementi--to at last assign sigla to the various sources and place them in a stemma. I begin by reserving the name T for the lost codex Turnebi itself, suggesting t for "Turnebus's collation of T," and τ for "Turnebus's references to T in the Adversaria".

Lindsay's new collation is more complicated. I designate the printed edition of Plautus in the Bodleian Library (the "Oxford Gryphius") that contains it "γ," and the entire mass of the variants in its margins, from any source, the "γ marginalia". Later copies of these γ marginalia also exist, but they are merely transcripts of γ itself and relatively inconsequential.2

All the γ marginalia that record only a single variant reading go back to an unimportant manuscript now in the British Library ("the Burney MS" or "MS Burneianus"), which I designate β.3 For our purposes neither β itself nor these single variants in γ that come from it are of any positive interest--this is the lard that I referred to earlier.

Sometimes the γ marginalia offer two variants on the text, sporadically distinguished by one of two marks indicating provenance.4 Those marked poict. go back to β and are more lard.

The other variants, sporadically marked dr. (or do. or du.), are the valuable "new collation" of T. I designate them δ, for François D(o)uaren (1509-1559), the Frenchman who added them to γ's margins.5 His source was not T directly, but the collation I call t, which he may have obtained from Turnebus himself.

In sum:

T: the lost codex Turnebi, for our purposes the Holy Grail

t: Turnebus's collation of T

β: the Burney manuscript, of value only in identifying γ marginalia as non-δ

γ: the Oxford Gryphius volume annotated by Duaren

δ: Duaren's marginalia in γ that go back to T

τ: Turnebus's citations of T in the Adversaria, made from t

For a stemma see here, wherein only sources above the horizontal line are of value in reconstructing T. 6

Our two sources of T are thus τ and δ, and, as noted above, they are not coextensive: we have τ for the nine different plays listed above (Asinaria-Stichus), whereas δ are available only for parts of Bacchides, Persa, Poenulus, Pseudolus, and Rudens--but in these latter plays, δ contains some variants where τ is silent. This situation would be ideal for supplementing our knowledge of T, except that--and herein lie two massive problems--(1) τ itself sometimes seems to misquote T, and (2) Duaren evidently did a poor job of transcribing Turnebus's collation, so that δ is not always a reliable guide to T, either. Among the more salient problems with δ is the difficulty in determining what exactly constitutes a δ variant--are these any variants in γ not found in β? More troubling, in many places δ-readings are missing where τ implies we should find them, and most worrisome of all, in many other places δ and τ are actually contradictory.

Thus the game of sorting all this out begins, which brings us at last to Clementi's new discussion.

In pp. 3-22 ("The Adversaria of Adrien Turnèbe" and "Work on Turnèbe's Plautine Philology") Clementi begins by contextualizing the nature and publication history of the Adversaria (3-15). By studying all the early printed editions of Plautus that were conceivably available to him (about 30!), Clementi has found that Turnebus did not use any single one exclusively when citing Plautus in the Adversaria (19-20). This negative conclusion is an excellent first step, and it leads her in the following sections to discuss the convoluted questions of how τ relate to T.

In 23-33 ("The Adversaria as a source of the lectiones in T") Clementi lays further groundwork by trying to determine when one of Turnebus's manifold terms for "old books" or "ancient manuscripts" (etc.) unequivocally refers to T and not to some other, less valuable, manuscripts or early printed editions--the latter of which, she just proved, Turnebus made promiscuous use of. Now by isolating all the instances in which Turnebus discusses the text of Plautus and also uses one of these expressions, she is able to show that in addition to all of the printed editions Turnebus also used a range of Plautine MSS of differing ages (p. 30). Ugh. This means that by relying on the Adversaria alone we have probably attributed some lections to T that come from elsewhere. How can we separate the wheat from the chaff?

In pp. 35-82 ("A comparison of the Adversaria and Bodleian Marginalia in restoring the lectiones of T") she turns to δ for help in checking τ. This ground has been covered before by Lindsay himself and by A. Klotz,7 but what distinguishes Clementi's effort as superior is that she compares δ only with those τ-passages in the Adversaria in which--as she established in the previous sections--Turnebus is explicitly referring to t or T--i.e., to those passages that are definitely τ. These amount to 22 test cases (all from Pseudolus, Poenulus, and Rudens). When δ and τ do not match, which is often, she subjects each to a rigorous analysis from multiple angles (philological, paleographic, linguistic, metrical, psychological, typographical, etc.) to decide which is the truer witness. Her usual system is first to check each τ quotation against all the printed editions of Plautus theoretically available to Turnebus, then against δ, then against any other available MSS (including those of the Italian recension), and lastly to pronounce her opinion. (Occasionally she goes on to give etymological information about rare or unusual words that Plautus uses in the relevant extract, but I am not sure what value this adds.)

Clementi is an extremely competent and skilled scholar, and as she powers her way through each passage I found myself acceding to her analysis almost every time. Only a few times did I query her presentation, viz: On p. 40n20 she raises as a mere possibility that Turnebus noticed that an obviously unmetrical reading in a senarius would not scan. This seems overly cautious. Clementi does not say what Turnebus knew of the meters of Roman Comedy, but the fact that he automatically changes migdilibys (unmetrical) to migdilibs (metrical) in Poen. 1033 suggests that he could scan senarii (on p. 50 Clementi says she isn't certain why Turnebus suggests this change), and other passages in the book indicate that Turnebus knew at least the trochaic septenarii, too. The only misstatement I noticed in this section is when (p. 36) she says that at Ps. 737-8 δ has hircum ab alis; in her photograph (#1, of 6) δ clearly reads hircum ab alijs; and this is not so much a real mistake as it is an overly compressed report of the facts (cf. Lindsay 1898, 26 for a clearer presentation).

Much as I often agree, I am left with a nagging concern. As with so many reconstructions of lost texts, special pleading has to be brought in to account for anomalies. That is to be expected. Like Lindsay and Klotz before her, Clementi believes that Duaren (or Turnebus himself) miscopied his lost source text in various fashions, sometimes at more than one remove; or she speculates that Duaren or Turnebus silently adjusted a reading, perhaps unconsciously. All of her hypotheses are individually reasonable, perhaps even likely, and I stress that she may be totally right. But in their cumulative effect the differences between δ and τ stand out sharply; and when presented with a conclusion that contradicts the given--i.e. that we should believe a δ reading over Turnebus's explicit testimony about T's text in the Adversaria (so Clementi, p. 47 on Poen. 530, cf. 42; so Lindsay--though not Clementi--on Poen. 413, p. 61-2), or when asked to believe that Duaren not only miscopied a reading, but also left out half of it and forgot to mark its provenance as well (on Ps. 1051, p. 39), Occam's razor comes to mind, bringing with it a larger worry: viz., if it was τ itself that Lindsay used to establish that the δ marginalia go back to T, the attempt to decide whether δ or τ is a better witness to T seems open to a charge of circularity.

Early on Clementi says "we know" (sappiamo, p. 30) that δ go back to T; that is, she takes Lindsay's argument for granted. In fact E. A. Sonnenschein (1851-1929), twice editor of Rudens, came to doubt Lindsay's hypothesis that the entirety of δ actually go back to T.8 He reasoned instead that δ represents not a collation of T but "a heterogeneous collection of variants, not from any one MS. but from several (including one or more of the Palatine MSS.), and of comments derived from various sources" (1899, 223, emphasis original). Many of his arguments remain unanswered, and I must say his conclusion tempts me. Clementi knows some of Sonnenschein's criticisms9, but she merely tucks them away in an early footnote (p. 33n29). If she or someone else addresses this concern in a future essay, this portion of her examination will rest on a more secure foundation and may alleviate these doubts.

In the remaining 43 pages (83-126, "An attempt to restore the lectiones of T without the help of the Bodleian marginalia;" general conclusions) Clementi examines the remaining τ passages where δ are unavailable, i.e. Turnebus's discussions in the Adversaria of Asinaria-Stichus, in which the readings of libri veteres, optimus liber, antiqua scriptura, and the like are invoked. Using the same exhaustive procedure as before, she shows that some of these readings definitely go back to humanist manuscripts rather than T, and on pp. 125-126 she reaches the following conclusion: when Turnebus refers to membranae, membranulae, chartae, schedae, and pergamenae, he means T, and probably also does on two occasions when he refers to a vetus liber and optimus liber. By contrast, his references to libri veteres, antiqui, vetusti, and autographi, and to vetera exemplaria, antiqua volumina, and antiqua scriptura all refer to humanistic manuscripts; which, she finds, Turnebus used only for the first 'eight' comedies (copies of which were numerous in his time) and not for the latter '12' comedies (which were much rarer). She also concludes, though she does not list the passages, that some of the passages quoted by Lindsay as evidence of T10 do not belong, and she generally would privilege the text of τ over δ (based on my remarks above, I would say we should always privilege τ over δ, but perhaps I am simply missing something). If, like me, you are persuaded by her arguments, then six allegedly τ passages (Aulularia 438, Captivi argumentum 3 and vv. 27, 31, 59, and 641) are false. Otherwise her conclusions tend to confirm Lindsay's, so that for all her work we have a surer foundation for future work than a number of surprising novelties. Can we extrapolate from her results? Clementi suggests that it may be time to revisit Lindsay's entire list of T-passages in T and scrutinize them in the same way. Any takers?

In part 2 (pp. 127-358) Clementi scrupulously edits into an alphabetical catalogue all the references in the Adversaria that mention Plautus in any way, including orphan quotations and other misfit Plautina. The catalog should be a resource of great general utility. Space precludes an analysis of Turnebus's manifold literary interests, but these include diction, religious customs, pots and pans, jokes, the duration of pregnancy (p. 138), catamites, French kisses (154, 155), etymology, puns, Atkins-breath kisses (242), denominations of money (342), obscenity (278-279) or its lack (305), and more. More surprisingly, though he knows the sort of interventions that (after Fraenkel) we tend to call 'Plautine elements' (p. 200 on Casina 524, 339 on Truculentus 269), he is refreshingly reluctant to appeal to them to explain seemingly Roman allusions (p. 145 on Asinaria 124, 147 on Asinaria 199, 219 on Curculio 393, 248 -- if I follow him--on Miles 211-212).

I remark in closing how impressively organized the book is, deceptive in its simplicity. It is hard to indicate, even to guess, how much time and labor must have gone into writing it--the checking and rechecking of scores of early printed editions, manuscripts, organization of material, etc., etc. (p. 97 gives just one heart-stopping glimpse). So it may seem uncharitable of me in doing the review's duty to complain about a stylistic feature of Clementi's presentation. Excessive familiarity with her material occasionally leads her to refer to manuscripts variously and expansively (e.g. "Palatinus Latinus 1615" instead of "B"). If she had assigned MSS sigla to her various sources as I have done or included a stemma, she could have shortened and clarified the book appreciably. The same goes for people: Turnebus is "the French humanist" nearly half the time, Duaren "the jurist," Nonius Marcellus "the author of the De compendiosa doctrina," and so on. Elegant or not, eliminating this antonomasia would have made a demanding book more accessible.

On the other hand, the copy-editing is superhuman,11 and on the whole, this will be the new place to go for anything Turnebus has to say about Plautus. You will never be in doubt about what he says, or where he says it.


1.   W. M. Lindsay, The Codex Turnebi of Plautus. Oxford: Clarendon Press, 1898 (get it here), consolidating earlier work.
2.   See Lindsay 1898 passim. These include a 1535 Gryphius edition of Plautus now in the Bibliothèque National in France; a 1522 Aldine Plautus in the same library; the vetus codex Scaligeri, now in the Bodleian library, and its various progeny in Leiden and elsewhere. Their relations are shown here.
3. Clementi neglects to include the Burney MS (British Library, Codex Burney 228, saec. XV) in her list of codicum sigla (pp. xxv-xxvi).
4. The signs remain mysterious; on p. 32n25 Clementi forgets to mention Lindsay's revised speculations about du. (cf. "The Codex Turnebi of Plautus and the Bodleian Marginalia," CR 13 (1899), 254-264 on 260.2-261.1).
5. Clementi (p. 31) mistakes Lindsay's mannered statement that some of the γ marginalia "bear the impress of Turnèbe's hand" (1898, 5) to mean that Turnebus himself wrote out these marginalia in γ, but Lindsay means their intellectual quality is such that only a Turnebus could be their author. Thus Lindsay later refers (1899, 262 col.1) to "a characteristically Turnebian remark" regarding the content of a note shortly after disparaging (1899, 259n1) Turnebus's penmanship ("The best that can be said of it is that it is not quite so illegible as Lambin's").
6. I sidestep here the case of Lambinus' veteres libri. Lindsay (1898) believes these variants do go back to t and even via a different transcription than δ, but that their value as a witness to T is for several reasons limited.
7. Alfred Klotz, "Die Plautushandschrift des Adrianus Turnebus," Philologus 95 (1942), 121-141.
8. E. A. Sonnenschein, "Lindsay's Codex Turnebi," CR 13 (1899) 222-224, answered by Lindsay (1899) (n. 4 above), promptly answered by Sonnenschein ibid. 264-265.
9. See also Sonnenschein in CR 20, 1906, 446-449 (not mentioned by Clementi).
10. Namely in W. M. Lindsay, Plauti codicis Senonensis lectiones, Philologus Supplementband 7 (Leipzig, 1899), 117-132 (get it here).
11. I find only these mistakes: p. 18.8 read praeclare, 83n1.9 independent, 145.14 e), 243.19 Troiam, 291.27 Iudaei. Is verisilmente 'prolly' (26.10, 55.5, 56.8) an accepted variant of verisimilmente?

No comments:

Post a Comment