MATERIAUX POUR L'HISTOIRE publies par l'Ecole des chartes ------4------

LA COLLABORATION DANS LA PRODUCTION DE L'ECRIT MEDIEVAL

Actes du XlIIe colloque du Comite international de paleographie latine (Weingarten, 22-25 septembre 2000)

reunis par HERRAD SPILLING

PARIS ECOLE DES CHARTES 2003

c) l I/( C) Lt

SIMULTANEOUS COPYING OF CLASSICAL TEXTS 800-1100: TECHNIQUES AND THEIR CONSEQUENCES

BENJAMIN VICTOR

We all know about the stemma codicum, the family-tree of manuscript sources used to trace their history and weigh their evidence. Editors of texts are agreed that the stem- ma is often useful, but that its usefulness knows limits; it is on the nature and extent of the limits that disagreement starts. I should like to consider a neglected aspect of the problem, namely a certain technique used by preference to copy poetical texts in the earlier Middle Ages. It is a fact of life among philologists that stemmata can be drawn for some authors, for others not. Indeed, it is such a fact of life that we have scarcely paused to think about it. Why do the manuscripts of one writer align themselves in a tidy genealogy, every one of them known to be cousin or brother to some other, whereas those of a different writer remain foundling children, their parentage only to be guessed at ? Our first thoughts are apt to run like this: «It's all to do with popularity: the more manuscripts of a text were out and about, the more chance to correct them and contaminate them. That means that the rare authors, like Lucretius, can be put into stemmata; the common ones, like Virgil, cannot ». Such an answer would be most incomplete. It does not explain why the chief manuscripts of Sallust admit readily of a stemma, while those of Statius, less numerous, resist. Now certain traditions are notoriously difficult to sort, but it turns out that patterns of relation can be seen even in them, though only by fits. Alfred E. Hausman remarked of Lucan : « The manuscripts group themselves not in families but in factions; their dissi- dences and agreements are temporary and transient, like the splits and coalitions of po- litical party» 1. Similar complaints are made of Virgil in the Middle Ages and of Juve-

* The research presented here has been supported by a general grant of the Social Sciences and Hu- manities Research Council of Canada, administered by the Universire de Montreal. My thanks to the fol- lowing for answering queries, offering comments or providing other assistance: Franca Arduini and I. G. Rao (Biblioteca Medicea Laurenziana, Florence), Robert Babcock (Yale University Library, New Haven), A. T. Bouwman (Universiteitsbibliotheek Leiden), Veronika von Büren (IRHT, Paris), Renaud Gagne (Harvard University, Cambridge), Peter Gumbert (Leiden), Lise Otis (MeGill University, Montreal). .

1. M. Annaei Lucani Belli civilis libri decem, Oxford, 1926, p. vii.

LA COLLABORATION DANS LA PRODUCTION DE L'ECRIT MEDIEVAL, actes du xlIr colloque du Comite international de pateographie latine (Weingarten, 22-25 septembre 2000), reunis par Herrad Spilling, Paris, 2003 tMateriaux pour l'histoire publies par I'Ecole des chartes, 4). 348 BENJAMIN VICTOR

nat2. Textual variants in the latter have been carefully tabulated so as to plot the chang- ing patterns of agreement among manuscripts; the shifts are numerous and sometimes radicat3. The tradition of Terence is perhaps most instructive of all. There mistakes in the division of verses remained and compounded each other right through the Middle Ages, for the versification of Roman comedy was not understood until modem times; by analysing such colometric errors it is possible to see through layers of contamination toward a detailed stemma. Nine Terence manuscripts of the ninth and tenth centuries have been studied in that way: at least one of them is seen to change affiliation every 255 lines on average". Such a state of affairs must be due to copying in teams, from multiple exemplars. To get to the bottom of it we must consider all the ways, in princi- ple four, to produce a book by team.

I. - THE RELAY

One scribe writes, then stops .. another continues. The models may be single or mul- tiple. In (I), each scribe writes until he is judged to have done enough or until he is needed for some other task; another then takes over. Though such a practice allows chores to be divided equitably, it saves no time. It will be at work in those books, which we have all seen, where the hands change in mid-page; it is also a possibility where hands change at the beginning of a structural unit. That sometimes the new scribe wrote from a different model is possible, though there was no advantage to doing S05. Now it may well happen that a copy must be produced quickly, as for instance when one foundation supplies books to another. In that case, other techniques were available.

2. For example, Leighton D. Reynolds in Texts and Transmission. A survey of the Latin , Ox- ford, 1983, p. 436 and Richard Tarrant, ibid., p. 201. 3. John G. Griffith, « A taxonomic study of the manuscript tradition of Juvenal », in Museum Helveti- cum, vol. 25 (1968), pp. 101-138. 4. Benjamin Victor (with Bruno Quesnel), « The colometric evidence for the history of the Terence- text in the early Middle Ages », in Revue d'Histoire des Textes, vol. 29 (1999), pp. 139-166; note also earlier work along the same lines by John N. Grant (« Contamination in the mixed manuscripts of Te- rence: a partial solution? », in Transactions of the American Philological Association, vol. 105 [1975], pp. 123-153). 5. This may have happened in Sankt Gallen, Stiftsbibliothek, 877, containing among other texts the verse of Sedulius copied by five scribes. In scribe D's portion (pp. 201 to 203), from Ill. 89 of Cannen paschale to the end (with the verses «A solis ortu ... »), it stands noticeably closer to manuscripts Land Z than in the preceding sections (or was a different model used for the end of the text because the principal model had lost leaves there ?). Note also Cittä del Vaticano, BibI. Apost. Vat., Reg. lat. 1703 (MS. R of Horace) : it has been argued that the two scribes who wrote fol. 125v line 7 - fol. 143 of this manuscript (= Sat. 11. 1.l6-end) followed a slightly different model from that of the preceding sections: Friedrich Klingner, Q. Horatius Flaccus. Opera, , 1939, pp. XIII-XIV. SIMULTANEOUS COPYING OF CLASSICAL TEXTS 349

11. - THE SYSTEM OF THE VATICAN

There is only one model. It is unbound and the quires distributed among the scribes. Once complete, their work is assembled into a volume. System (IT),like (D, puts several scribes to work, but now they write at the same time. They must take care, however, to return with one quire for each quire of their model. The method has been well studied by codicologists, thanks to cases where both copy and original survive", The best known book created by it is the Vatican Livy to which reference has been made: Cittä del Vaticano, BibI. Apost. Vat., Reg. lat. 762, copied about 800 by a large team of scribes at Tours; the model, an ancient uncial co- dex, is kept in the Bibliotheque Nationale de France as lat. 5730. The method carried serious disadvantages: not only the difficulty of reproducing eight leaves on no more and no less than eight leaves, but also the bother of undoing, then re-doing, a binding. Two further techniques, without these drawbacks, could be brought to bear.

Ill. - DIVISION BY BOOK OR WORK

Copying is done simultaneously from multiple models .. labour is divided according to the natural units of the corpus. Corpora of different works were no doubt so copied from time to time. However, the result is so hard to distinguish from a recueil assembled after the fact that clear instances are hard come b/. One such is Bruxelles, Bibliotheque Royale, 5337-5338 : earlyelev- enth century, of Gembloux provenance and (most certainly) origin, belonging among the distinctive products of Olbert's abbacy. Hand A has written Statius' Thebaid, ending it on the last page of a binion (fol. 146v), and also the first page of the Achilleid (fol. 147) ; B has copied the remainder of the Achilleid. The text of the Thebaid is note- worthy for occasional agreements with the codex Puteanus (Paris, BibI. Nat. de France, lat. 8051), which carries a rare and valuable strain of the tradition; these continue to the end of the Thebaid8 but cease in the Achilleid, though the Puteaneus there retains its

6. Best by Jean Vezin, «La repartition du travail dans les 'scriptoria' carolingiens », in Journal des savants, 1973, pp. 212-227. 7. For example, it would be tempting to adduce Firenze, BibI. Medicea Laurenziana, San Marco 257, a collection of philosophical works by apparently copied at Corbie in the mid-ninth century. The first six quires of the text proper (foIs. 1-50), containing De natura deorum and De divinatione, are the work of one scribe, the following quires (foIs. 51-90, Timaeus, De fato, Topica, Paradoxa stoicorum, Lucullus, De legibus) of another. Both models, as it happens, still exist today: Leiden, Universiteitsbiblio- theek, Voss. lat. F. 84 and ibid. Voss. Iat. F. 86. The first scribe of the San Marco manuscript copied Voss.lat. F. 86, the second Voss. lat. F. 84 (recognized ever since Paul Schwenke, «Apparatus criticus ad Ciceronis libros De natura deorum », in Classical Review, vol. 4 (1890), pp. 347-355, with continua- tions ; see esp. p. 349). The identical page-layout of the two parts would seem to indicate that a single volume was planned. But what to make then of the quire-signatures, which begin again at «I» in the se- cond part? 8. E.g., XII. 128 «it »,241 «comitis »,252 «e » om., 287 «vitasse », 467 «fletibus » ; on the Brus- sels manuscript and its congeners see D. E. Hill, « The manuscript tradition of the Thebaid », in Classical Quarterly, vol. 16 n. s. (1966), pp. 333-346. 350 BENJAMIN VICTOR distinctive character. The Brussels manuscript will thus have drawn the two poems from different sources; the intention to make one volume, not two, is seen in the care taken to present the same hand on the facing pages at fols. 146v-147. Whether the copyists aimed to save time or simply had no model containing both texts, the technique and its consequences are the same.

IV. - DIVISION BY CALCULATION

Copying is done simultaneously from multiple models .. the division of labour does not correspond to the natural divisions of the corpus. Here copying is from multiple models as in (ill) and simultaneous as in (11)and (Ill), but the division of labour corresponds neither to the natural divisions of a corpus nor to the division of quires in the original. The director of the scriptorium must calculate how much of the original is to be fit onto each quire of the copy; each scribe must then act accordingly, so as to end his share of the project at the bottom of the last verso of a quire. A book created by method (IV) will look much like one done by method (11)or even method (I) (in the variant of [I] where the writers write whole quires). It is only by analysing both the codicological indications and the character of the text that a case for 9 (IV) can be made • I have lately been studying manuscripts of pagan classics and Christian poets (I hope also to include Christian prose-writers in the survey) with a view to the methods of pro- duction during the monastic period, particularly those methods involving multiple ex- emplars. The research is limited to texts meeting certain criteria: they must have been widely copied between 800 and 1100; their traditions must have been studied with some thoroughness and the principal manuscripts reported in detail by the editors. Manuscripts of the following have been examined thus far: Horace ; Justin ; Juvenal ; Juvencus ; Lucan ; Macrobius, Saturnalia ; Martianus Capella; Palladius ; Prudentius ; Sallust ; Sedulius, Carmen paschale ; Solinus ; Statius, Thebaid ; Terence ; Vegetius. 156 manuscripts, including most of those cited in the major critical editions, have been seen to date; their shelfmarks are listed in an appendix. Some were examined on mi- crofilm, others by autopsy. Their affiliations from one part of the text to another were determined from existing apparatus critici and other published scholarship, in a few cases from my own collations. Whenever I suspected a multiple-exemplar method of production, I traveled to the library and studied the original. Now it is method (N) and the multiple-exemplar variant of (I) that interest us here, for only they can spawn the textual chaos of Lucan and the like. Fifty manuscripts of my sample were found to be written by more than one scribe; the figure excludes obviously later supplements and contributions of less than one quire by an additional hand'", Two

9. I say no more than «a case can be made », The multiple-model variant of (I) often remains a pos- sibility. Moreover, no multiple-exemplar method can be deduced if the models were too closely related in text. 10. However, cases where the additional hand wrote all but the first recto and/or the last verso of a quire are included; see below on Düsseldorf, Universitäts- und Landesbibliothek, F 1 and on Leiden, Universiteitsbibliotheek, Voss. lat. Q. 38. SIMULTANEOUS COPYING OF CLASSICAL TEXTS 351 of these multiple-hand manuscripts, perhaps executed from multiple exemplars by method (I), have been signaled above at note (5). For the rest, the following have shown the strongest indications of being copied from multiple exemplars by (IV) or by (I) :

• Paris, BibI. Nat. de France, lat. 8671, manuscript M (Willis's siglum) of Martianus Capella. Ninth century, French, the work of four scribes writing (in most cases) whole quires. Certain changes of hand correspond to palpable changes of text-type. In quire III for example (fols. 17-24, containing De nuptiis Il, 202 - Ill. 298) M is a twin of R (Karlsruhe, Badische Landesbibliothek, Aug. LXXIll), sharing very numerous errors with it alone; this quire is begun by hand B and finished by A (the hand of quires 1-11). Quire IV on the other hand (fols. 25-30 = De nuptiis Ill. 298 - Illfin.), the first gathering to be written entirely by hand B, presents a strikingly different text. M there belongs to the group ABDMR but is no closer to R than others of the group. This situation persists through quires V, VI and VII, written by hands A, C and B, then by A again. In quire VID (fols. 57-64 = De nuptiis VI. 656 - Vll. 740, written by hand B) M suddenly shows, in addition to agreements with ABDR, many errors unique to itself and manuscript G (Bruxelles, Bibliotheque Royale, 9565-9566) as well as others characteristic of the group EFG. Such errors cease almost entirely in quire IX (fols. 65-72 = De nuptiis VII. 740 - VIll. 810), the only quire written by hand D, where M becomes a pure member of ABDMR. Clearly we must imagine for M a complicated genesis, from more than one model. Quires ending in crowded or in blank pages exclude method (1).

• Paris, BibI. Nat. de France, lat. 10314, an important ninth-century Lucan executed by four scribes; ex-libris of Echtemach ; goes by Z in the editions. The first scribe wrote just one quire, ending it with line 475 of book I. He will hardly have been replac- ing a loss, as his hand is quite contemporary to the others ll. In the next quire, a change in the character of the text is noticeable at once: Z becomes the twin of MS. M (Montpellier, Fac. de Medecine, 113), and continues such through most of the poem, despite other changes of hand in Z12. Two scenarios suggest themselves. Perhaps all of Z was produced by a multiple-model method; among its models were three very close to M and one less close; the scribe of the first quire used the model least close to M. Alternatively we may imagine a hybrid method to have been used: scribe 1 wrote from one model, the other three scribes from a single other model by method (I) or (IT).The problem is a typical one in this sort of investigation: the multiple models available in a given library will often have been closely related, making it difficult or impossible to tell where scribes change from one to another. In any case, multiple models were at work in at least part of Z.

• Paris, BibI. Nat. de France, lat. 10317, executed about the year 1000 at Echter- nachl3, manuscript Q of Statius' Thebaid. The text itself was copied by three hands, as it

11. It could also be supposed that he added his quire after the rest was copied from an acephalous an- tigraphon, but nothing about the presentation indicates that quire 11ever stood at the head of the volume. 12. The change of text-type, first noticed by Housman, has been amply demonstrated: Harold Gotoff, The text of Lucan in the ninth century, Cambridge (Massachusetts), 1971, pp. 44-58. 13. One of the annotating hands is otherwise known at Echtemach : Jean Schroeder, « Bibliothek und 352 BENJAMIN VICTOR seems to me'"; there are also scholia and later supplements by others. Q shifts its textual allegiance around line 700 of book IV, that is to say at the end of quire X (fols. 81v-82), where scribe A completes his second portion and C begins his first. At this point it be- comes the twin of Bamberg, Staatsbibliothek, Class. 47, with which it had no particular affinity until then. We may imagine either of the scenarios just sketched for Paris, BibI. Nat. de France, lat. 10314.

• Düsseldorf, Universitäts- und Landesbibliothek, F 1, a German Prudentius of the late ninth century (Provenance: Essen1\ Here the textual re-alignment, and its correla- tion to the structure of the manuscript, has already drawn notice: Maurice Cunningham recorded it in his edition of the poet". Two scribes were responsible for this manu- script: A wrote fols. 1-56 (and the very first lines of fo1. 56va), that is to say all but a page of the first seven quires; B wrote fols. 57-69v (quaternio + ternio) and almost all of fo1. 56v (no doubt so that the facing pages at fols. 56v-57 would present the same appearance). B's section, containing Peristephanon from X. 931 to the end and De opusculis suis, stands textually with the E)-class of Prudentius-manuscripts, A's, con- taining the rest of the corpus, with r.

Two more ambiguous cases deserve mention.

• Oxford, Bodleian Library, Auct. F.VI.27, tenth century, bearing a record of early readers connected with the late Ottonian court'? and an erased late-medieval ex-libris of Augsburg (the name of the house not decipherable with certainty). Hand B's first por- tion (quire ITand all but the last page of rn, or fols. 9-22) contains Andria 295-end. About where B begins, the text draws very close to that of another South-German ma- nuscript, namely Wien, Österreichische Nationalbibliothek, Cod. 85 (likewise tenth century). However, both the Oxford and the Vienna manuscript are nearly free from errors in the thirty lines following Andria 295, making it uncertain where the text-type changes. It is also curious that division of verses in the Oxford manuscript is similarly inaccurate both before and after the change of hands. So much room for doubt.

• Leiden, Universiteitsbibliotheek, Voss. lat. Q. 38, an illustrated Terence of the tenth century, attributable to the Loire Valley on art-historical grounds'" and carrying the medieval ex-libris of St. Maurice, Angers. Hand A wrote everything but quire VI

Schule der Abtei Echtemach um die Jahrtausendwende », in Publications de la section historique de l'Institut grand-ducal de Luxembourg, vol. 91 (1977), pp. 201-377, esp. 250-252. 14. Others have counted four (Alfred Klotz, P. Papini Stati Thebais, Leipzig, 1908, pp. XII-XIV) or one (J. Schroeder, loc. cit.). 15. See esp. Gerhard Karpp, «Bemerkungen zu den mittelalterlichen Handschriften des adeligen Da- menstifts in Essen (9.-19. Jahrhundert) », in Scriptorium, vol. 45 (1991), pp. 163-204, on p. 175. The manuscript may well have been written at Werden. 16. Tumhout, 1966 (Corpus Christianorum. Series latina, 126), pref. p. XIV. 17. Full discussion in Claudia Villa, La 'lectura Terentii'. Vol. 1: Da Ildemaro a Francesco Petrar- ca, Padova, 1984 tStudi sui Petrarca, 17), pp. 99-136. 18. See esp. Leslie W. Jones - Charles R. Morey, The miniatures of the manuscripts of Terence prior to the thirteenth century, Princeton, 1931, text vol., pp. 130-151. SIMULTANEOUS COPYING OF CLASSICAL TEXTS 353

20 (fols. 39_45)19,and wrote the first recto of that quire, to0 , presumably to ensure an attractive vis-a-vis at fols. 38v-39 (though it is odd that he did not write the last verso as well). Quire VI is a seven-leaf gathering (fo1. 39 conjoint with a stub). Quire VI also belongs to a different strain of the textual tradition than do its neighbours. The colomet- ric data in this stretch of text (Eunuchus 588-839) show occasional agreements of Voss. lat. Q. 38 with the ö-class as a whole, none with any of its subclasses (despite abundant evidence for them) ; the other quires all have the colometry of a certain small subclass within Ö or of i1• Now what to think of all this? It is possible that a different scribe was assigned the bulk of quire VI and given a different model to work from, in order to gain speed. But why just one quire, and why such a clumsy job of it ? Perhaps the chief model was lacunose but was transcribed anyway while a supplement was awaited, then quire VI added in when the missing text had been found in another source. Signatures might help decide the matter, but none are preserved.

A preponderance of poetical texts in this list will be noticed, however the doubtful points may be judged. I suspect that is because these manuscripts by and large result from method (IV), and that poets were this method's favourite objects, since the divi- sion of their text into verses simplified the calculation of link-ups. To produce manu- script M of Capella must have required troublesome planning, and the result, with its quires not altogether regular and its writing crowded and spread by turns, is anything but a handsome book. In a verse-text, on the other hand, the advantages of method (IV)- speed and no bindery work - could be had easily and esthetically. The traditions that most defy stemmatic description are indeed those of poets: Lucan, Juvenal, Virgil and Terence have already been mentioned; to them may be added Persius, Statius, Juvencus and Sedulius. They must often have been copied from several books at once. I return momentarily to Terence for one last observation. Not all manuscripts of this author were written as verse. Three of ninth/tenth-century date and of especial impor- tance were copied without any indications of colometry : these are C (Cittä del Vatica- no, BibI. Apost. Vat., Vat. lat. 3868), p (Paris, BibI. Nat. de France, lat. 10304), and Y (Paris, BibI. Nat. de France, lat. 7900, not yet used in critical editions but discussed amply in the literature"). Significantly, they are, along with just two others", the ma- nuscripts of Terence with the stablest textual affiliation. C and Y are twins throughout the text, being everywhere nearer to each other than to any other source; the group of their next closest kin always includes P (Paris, BibI. Nat. de France, lat. 7899), this group's size and configuration varying with the caprices of other manuscripts. p always

19. B, the hand of fols. 39v-45v, writes much like A but is distinguished among other things by his use of Tironian « et » rather than ampersand, the reverse of A's practice. 20. Misstated in my article of 1999 (n. 4, above), p. 151, where I implied that hand B wrote the whole quire. 21. Again, see the article of 1999 (n. 4, above). Unfortunately independent error in Voss. lat. Q. 38 makes its affiliation at the very beginning of quire VI unknowable (in that sense the diagrams on p. 151 are somewhat misleading). 22. Robert H. Webb, «An attempt to restore the 'Yarchetype of Terence manuscripts » in Harvard Studies in Classical Philology, vol. 22 (1911), pp. 55-110; John N. Grant, Studies in the textual tradition ofTerence, Toronto, 1986, pp. 136-159. 23. One of these two others (« D » = Firenze, BibI. Medicea Laurenziana, PIut. 38. 24) is itself written partly as prose. 354 BENJAMIN VICTOR stands apart from CYP and always belongs to a group which has D as a member (again, the composition of this group is otherwise unstablej'". Poetical texts written as prose behave like prose. I am well aware that I have suggested much but demonstrated much less. Ideally, a study such as this should rest on more data and answer to statistical tests, and it is hoped that research can be continued toward that goal. But for the time being, the above will have to do.

BENJAMIN VICTOR, Universite de Montreal.

ApPENDIX

MANUSCRIPTS SAMPLED

M = Executed by multiple hands - i.e., at least one quire is by a second contemporary hand (with the proviso given in note 10 above). Data on multiplicity of hands regard only the text in question, not other texts that the volume may hold; thus a book containing Solinus by one hand and Vegetius by several will be marked 'M' under Vegetius only. Obvious recueils factices (again, within the text in question) have been excluded from study.

Ho race

Bern, Burgerbibliothek, 363. Leiden, Universiteitsbibliotheek, BPL 28. M London, British Library. Harl. 2725. M Milano, Biblioteca Ambrosiana, 0 136 sup. M Paris. BibI. Nat. de France. lat. 7900A + Hamburg. Staats- und Universitätsbibliothek, Scrin. 53b. Paris, BibI. Nat. de France.Jat, 7971. M Paris. BibI. Nat. de France. lat. 7972. M Paris. BibI. Nat. de France. lat. 7974. Paris. BibI. Nat. de France. lat. 7975. M Paris. BibI. Nat. de France.Iat. 10310. Ciua del Vaticano, BibI. Apost. va., Reg. lat. 1703. M

Justin

Giessen, Universitätsbibliothek, 79. M Leiden. UniversiteitsbibIiotheek, Voss. lat. Q. 32. Paris. BibI. Nat. de France, lat. 4950. Paris. BibI. Nat. de France. nouv. acq. lat. 1601.

24. Details in the articles cited in note 4 above and in J. N. Grant. Studies .... pp. 97-159. SIMULTANEOUS COPYING OF CLASSICAL TEXTS 355

Juvenal

Cambridge, Trinity College, 1241. Firenze, Biblioteca Medicea Laurenziana, Plut. 34. 42. Leiden. Universiteitsbibliotheek, BPL 82. Leiden, Universiteitsbibliotheek, Voss.lat. Q. 18. M London, British Library, Add. 15600. Montpellier, Fac. de Medecine, 125. München, Bayerische Staatsbibliothek, Clm 408. M Oxford, Bodleian Library. Canon. class. Iat. 41. Paris, BibI. Nat. de France, lat. 7900A. Paris, BibI. Nat. de France, lat. 8071. Paris, BibI. Nat. de France, lat. 9345. M Cittä del Vaticano, BibI. Apost. Vat., Urb. lat. 661.

Juvencus

Karlsruhe, Badische Landesbibliothek, Aug. CXII. Karlsruhe, Badische Landesbibliothek, Aug. CCXVII. fols. 1-39. M Karlsruhe. Badische Landesbibliothek, Aug. CCXVII, fols. 138-161. Laon, Bibliotheque Municipale, 101. London, British Library, Roy. 15 A XVI. Montpellier, Fac. de Medecine, 362. München, Bayerische Staatsbibliothek. Clm 6402. M Paris, BibI. Nat. de France, lat. 9347. M

Lucan

Bern, Burgerbibliothek, 45. Bruxelles, Bibliotheque Royale, 5330-5332. Leiden, Universiteitsbibliotheek, Voss.lat. F. 63. M Leiden, Universiteitsbibliotheek, Voss.lat. Q. 16. M Leiden, Universiteitsbibliotheek, Voss.lat. Q. 51. Montpellier, Fac. de Medecine, 113. M Montpellier, Fac. de Medecine, 362. M Paris, BibI. Nat. de France, lat. 7502. Paris, BibI. Nat. de France, lat. 7900A. Paris, BibI. Nat. de France, lat. 10314. M

Macrobius, Satumalia

Bamberg, Staatsbibliothek, Class. 37. Napoli, Biblioteca Nazionale, V.B.I0. M Oxford, Bodleian Library, Auct. T.II.27. M Padova, Biblioteca Antoniana, 27. Paris, BibI. Nat. de France, lat. 6371. Cittä del Vaticano, BibI. Apost. Vat., Reg. lat. 2043.

Martianus Capella

Bamberg, Staatsbibliothek, Class. 39. Besan<;on. Bibliotheque Municipale, 594. Bruxelles, Bibliotheque Royale, 9565-9566. 356 BENJAMIN VICTOR

Karlsruhe, Badische Landesbibliothek, Aug. LXXIII. Leiden, Universiteitsbibliotheek, BPL 87. Leiden, Universiteitsbibliotheek, Voss.lat. F.48. London, British Library, HarI. 2685. Oxford, Bodleian Library, Laud lat. 118. Paris, Bibi. Nat. de France, lat. 8669. M Paris, BibI. Nat. de France, lat. 8670. Paris, BibI. Nat. de France, lat. 8671.M

Palladius

Cambridge, University Library, Kk.5.13. Laon, Bibliotheque Municipale, 426 bis. Leiden, Universiteitsbibliothek, BPL 102. Montpellier, Fac. de Medecine, 305. Paris, BibI. Nat. de France, lat. 6830E. Paris, BibI. Nat. de France, lat. 6842B. Paris, BibI. Nat. de France, nouv. acq.lat. 1730. Wien, Österreichische Nationalbibliothek,lat. 148.

Prudentius

Bruxelles, Bibliotheque Royale, 9987-9991. Düsseldorf, Universitäts- und Landesbibliothek, F 1. M Leiden, Unversiteitsbibliotheek, Burm. Q. 3. Montpellier, Fac. de Medecine, 220. M Paris, BibI. Nat. de France, lat. 8087. Paris, BibI. Nat. de France, lat. 8305. M Saint-Omer, Bibliotheque Municipale, 306. Cittä del Vaticano, BibI. Apost. Vat., Reg. lat. 321. Wolfenbüttel, Herzog-August-Bibliothek, Aug. 4° 56.18.

Sallust

Basel, Öffentliche Bibliothek der Universität, AN.lV.11. M Berlin, Staatsbibliothek zu Berlin - Preußischer Kulturbesitz, Phillipps 1902. M Leipzig, Universitätsbibliothek, Rep. I.foI. 4. München, Bayerische Staatsbibliothek, Clm 4559. München, Bayerische Staatsbibliothek, Clm 14477. M Paris, Bibi. Nat. de France, lat. 5748. Paris, BibI. Nat. de France, lat. 6085. Paris, BibI. Nat. de France,lat. 10195. Paris, BibI. Nat. de France,lat. 16025. Cittä del Vaticano, BibI. Apost. Vat., Pal. lat. 887. Citt1i del Vaticano, BibI. Apost. Vat., Pal. lat. 889. ZUrich, Zentralbibliothek, Car. C 143a. M

Sedulius

Bern, Burgerbibliothek. 267. M Karlsruhe, Badische Landesbibliothek, Aug. CCXVII, fols. 40-67. M Karlsruhe, Badische Landesbibliothek, Aug. CCXVII, fols.lO 1-117, 122-127, 162-169. Montpellier, Fac. de Medecine, 362. München, Bayerische Staatsbibliothek, Clm 18628. SIMULTANEOUS COPYING OF CLASSICAL TEXTS 357

Paris, BibI. Nat. de France,lat. 9347. M Paris, BibI. Nat. de France, lat. 13377. Paris, BibI. Nat. de France,lat. 14143. Sankt Gallen, Stiftsbibliothek, 197. Sankt Gallen, Stiftsbibliothek, 242. Sankt Gallen, Stiftsbibliothek, 877. M Trier, Stadtbibliothek, 1464. Cittä del Vaticano, BibI. Apost. Vat., Ottob.lat. 35. Zürich, Zentralbibliothek, C 68.

Solinus

Heidelberg, Universitätsbibliothek, Palat. lat. 1568-1. K~benhavn, Det Kongelige Bibliotek, GI. KgI. S. 444 2°. Leiden, Universiteitsbibliotheek, Voss.lat. Q. 87. M Paris, BibI. Nat. de France,lat. 6810. Paris, BibI. Nat. de France, lat. 7230. M Paris, BibI. Nat. de France, lat. 7230A. Sankt Gallen, Stiftsbibliothek, 187. Cittä del Vaticano, BibI. Apost. Vat., Vat. lat. 3342. Citta del Vaticano, BibI. Apost. Vat., Vat. lat. 3343. M Wolfenbüttel, Herzog-August-Bibliothek, Gud. lat. 163.

Statius, Thebaid

Bamberg, Staatsbibliothek, Class. 47. Bern, Burgerbibliothek, 156. M Bruxelles, Bibliotheque Royale, 5337-5338. Cambridge, St. John's College, 012. Leipzig, Universitätsbibliothek, Rep. I.fol. 12. M London, British Library, Roy. IS C X. Paris, BibI. Nat. de France, lat. 8051. Paris, BibI. Nat. de France, lat. 10317. M Paris, BibI. Nat. de France, lat. 13046. Paris, BibI. Nat. de France, nouv. acq.lat. 1627. Wolfenbüttel, Herzog-August-Bibliothek, Gud.lat. 54. M

Terence, written wholly or almost wholly as verse

El Escorial, Biblioteca Real del Monasterio de San Lorenzo, S.III.23. Firenze, Biblioteca Riccardiana, 528. Leiden, Universiteitsbibliotheek, BPL 109. Leiden, Universiteitsbibliotheek, Lips. 26. M Leiden, Universiteitsbibliotheek, Voss.lat. Q. 38. M Madrid, Biblioteca Nacional, Vitro 5-4. Milano, Biblioteca Ambrosiana, G 130 inf, Milane, Biblioteca Ambrosiana, H 75 inf, M Oxford, Bodleian Library, Auct. F.VI.27. M Oxford, Brasenose College, 18. Paris, Bibl. Nat. de France, lat. 7899. Paris, BibI. Nat. de France, lat. 7900A. Paris, BibI. Nat. de France, lat. 7902. Paris, BibI. Nat. de France, lat. 9345. M Paris, BibI. Nat. de France, lat. 14755. Valenciennes, Bibliotheque Municipale, 448. 358 BENJAMIN VICTOR

Terence, written as prose

Leipzig, Universitätsbibliothek, Rep. 1.4° 37. Paris, BibI. Nat. de France, lat. 7900. Paris, BibI. Nat. de France, lat. 10304. Citta del Vaticano, BibI. Apost. Vat., Vat. lat. 3868.

Terence, written as a mix a/verse and prose

Firenze, Biblioteca Medicea Laurenziana, 38. 24. M Firenze, Biblioteca Medicea Laurenziana, S. Marco 244.

Vegetius

Bern, Burgerbibliothek, 280. M Laon, Bibliotheque Municipale, 428. Leiden, Universiteitsbibliotheek, Perizon. F 17. München, Bayerische Staatsbibliothek, Clm 6368. Paris, BibI. Nat. de France, lat. 6503. Paris, BibI. Nat. de France, lat. 7230. M Paris, BibI. Nat. de France, lat. 7230A. Paris, BibI. Nat. de France, lat. 7231.