* I would also like to thank Maxence Hernant for advice regarding the Codex Thuaneus and the fate of manuscripts in seventeenth-century France; Helen Dixon, Bruce Gibson, Michael Reeve, Edina Zsupán and the anonymous referee of Classical Quarterly for their helpful comments on earlier versions of this paper; the Ludwig-Maximilians-Universität München and the Irish Research Council for a Research Fellowship and a Government of Ireland Postdoctoral Research Fellowship, respectively, which enabled me to write this article, and for generous travel funding; and Cyril Simon and Diane Le Grand de Belleroche for their kind hospitality while I studied the Thuaneus in Paris.

¹ Vossius, I., Cajus Valerius Catullus et in eum Isaaci Vossii Observationes (London, 1684), 32–3Google Scholar. According to the printer's preface, Vossius' notes on the poet ‘lay in the dark for over thirty years’ (‘[p]er integros triginta & plures annos in tenebris iacuere’) before 1684, and his biographer Blok notes that he composed his commentary on Catullus ‘during the years 1652 and 1653’ (Blok, F.F., Isaac Vossius and his Circle: His Life until his Farewell to Queen Christina of Sweden, 1618–1655 [Groningen, 2000], 407Google Scholar). However, on p. 92 of his commentary Vossius writes of a conjecture that ‘nuper in Castigationibus ad Melam fecimus’, and his Observationes ad Pomponium Melam De situ orbis appeared in the Hague in 1658. Either Vossius continued to modify his commentary on Catullus after 1653, or else he was referring to the unpublished text of his notes on Mela.

² Vossius, I., Periplus Scylacis Caryadensis cum tralatione, & castigationibus Isaaci Vossii ... (Amsterdam, 1638)Google Scholar; Vossius, I., Iustini historiarum Ex Trogo Pompeio Lib. XLIV cum notis Isaaci Vossii (Leiden, 1640)Google Scholar. On Vossius' education, see Blok (n. 1), 20–3 and 38–43.

³ Blok (n. 1), 96–196; ter Horst, D.J.H., Isaac Vossius en Salmasius: een episode uit de 17de-eeuwsche geleerdengeschiedenis (‘s-Gravenhage, 1938), 15CrossRefGoogle Scholar.

⁴ Ter Horst (n. 3), 40; see also ibid., 15, and Blok (n. 1), 103–4.

⁵ Vossius (n. 1 [1684]), 126 and 150.

⁶ Mart. Sp. 7 and 8, as well as 14.26 and the title of 14.171. On Vossius' rendering of Mart. Sp. 17.1 and 32 see n. 7 below.

⁷ Correct: Catull. 62.7 (on p. 150); 62.9 (p. 152); 62.35 (p. 154); 62.40 (p. 154); 62.50 (p. 155); Mart. Sp. 7 and 8 (p. 127). Slips: Catull. 62.14 tota penitus for penitus tota (p. 152); 62.59 et tu nec pugna for et tua nec pugna (p. 156); Mart. 14.26 accendat for accendit in line 1 (p. 142); 14.171 title fichile for fictile (p. 229). Mistake: Catull. 62.45 dum ... dum cara suis, sed (p. 155), where Vossius conflates the version of the line as it is quoted by Quintilian (dum ... dum ... est), the reading of the Thuaneus (dum ... tum ... est), and that of OGR and their progeny (dum ... tum ... sed). At p. 126 Vossius notes that of two epigrams of Martial ‘ita reformanda sunt lemmata, prout concipiuntur in vetustis membranis Thuanaeis’, but on the next page he proceeds to give his own reconstruction of Mart. Sp. 17.1 and 32, rather than an accurate transcription of them as they stand in the manuscript.

⁸ Vossius stayed in Paris again between July 1643 and August 1644 (with interruptions), and in July and August 1650, but there are no indications that he studied manuscripts at the De Thou collection on those visits: see Blok (n. 1), 177–95 and 327–35.

⁹ For its most detailed description to date, see Vecce, C., Iacopo Sannazaro in Francia: scoperte di codici all'inizio del XVI secolo (Padua, 1988), 93–5Google Scholar, with further bibliography.

¹⁰ Bischoff's view of where and when the Thuaneus was copied has appeared three times in print: (i) in an undated letter mentioned by Schmidt, P.L., ‘Rezeption und Überlieferung der Tragödien Senecas bis zum Ausgang des Mittelalters’, in Lefèvre, E. (ed.), Der Einfluss Senecas auf das europäische Drama (Darmstadt, 1978), 12–73Google Scholar, at 60 with n. 78: it was copied in the second quarter of the ninth century; (ii) in a letter to Claudia Villa dated 20 January 1977 that is quoted by Vecce (n. 9), 95 n. 2: around the third quarter of the ninth century between Auxerre and Paris, probably closer to the former than to the latter; (iii) in Bischoff, B., Katalog der festländischen Handschriften des neunten Jahrhunderts (mit Ausnahme der wisigotischen), ed. Ebersperger, B., 3 vols. (Wiesbaden, 1998–2014)Google Scholar, 3.138: possibly in the third quarter of the ninth century, between Paris and Auxerre or somewhere nearby. In Bischoff, B., Manuscripts and Libraries in the Age of Charlemagne, ed. and trans. Gorman, M. (Cambridge, 1994)Google Scholar, at 145, Bischoff observes that the Thuaneus is ‘characteristic of the scribal activity in the area of Fleury and St Germain’ (i.e. between Auxerre and Paris) around the middle of the ninth century. I would like to thank Birgit Ebersperger and Vicente Flores Militello for enabling me to consult vicariously the final volume of Bischoff's Katalog.

¹¹ It is sometimes supposed that the anthology had been put together by the scribe of the Thuaneus. At best, this assumption would be unwarranted: the Thuaneus could also be a copy of an earlier anthology. In fact it is wrong, as is shown by the survival of part of an older relative of the manuscript, and of traces of a third Carolingian copy of the anthology: see pp. 349–50 below.

¹² Dated thus by Vecce (n. 9), 94.

¹³ The title is followed by Juvenal 3.317–22, but the scribe evidently regarded these lines as the start of Juvenal's first satire.

¹⁴ It is not significant that folio 32v ends with Mart. 4.79.1–6, and folio 33r starts with 4.79.9–10: lines 4.79.7–8 have been lost through haplography (lines 5 and 7 both start with et s...). No text has been lost between the other gatherings.

¹⁵ This title alludes to Peter's words to Jesus at Matthew 17:4 si uis, faciamus hic tria tabernacula, tibi unum, Moysi unum, et Eliae unum.

¹⁶ Harrisse, H., Le Président de Thou et ses descendants, leur célèbre bibliothèque, leurs armoires, et les traductions françaises de J.-A. Thuani Historiarum sui temporis d'après des documents nouveaux (Paris, 1905), 2–3Google Scholar.

¹⁷ Harrisse (n. 16), 28–9, 37 and 40.

¹⁸ Delisle, L., ‘Les manuscrits de Colbert’, Comptes-rendus des séances de l'Académie des Inscriptions et des Belles-Lettres 7 (1863), 296–304Google Scholar, at 300.

¹⁹ Pellegrin, É., ‘Membra disiecta Floriacensia’, Bibliothèque de l'École des Chartes 117 (1959), 5–56CrossRefGoogle Scholar, at 5–6 n. 3, reprinted in: ead., Bibliothèques retrouvées: manuscrits, bibliothèques et bibliophiles du Moyen âge et de la Renaissance. Recueil d'études publiées de 1938 à 1985 (Paris, 1998), 159–210Google Scholar, at 159–60 n. 3.

²⁰ tactus was prone to be corrupted here, as is shown by the readings of Catullus' codices recentiores in this passage. One hundred and seven read tactus; eleven read tractus (these are nos. 1, 5, 19a, 27, 29, 51, 61, 92, 97 and 99 Thomson-Kiss, and no. 13, where the word was subsequently erased and is hard to make out); four read fractus (nos. 46, 54, 75 and 86); in four of them the word was added in the margin by the scribe (nos. 118, 122 and 128), and in one by a later hand (no. 78); and two read factus (nos. 56 and 77). For a list of the surviving manuscripts of Catullus, see Thomson, D.F.S., Catullus Edited with a Textual and Interpretative Commentary (Toronto, 1997), 72–91Google Scholar, with Kiss, D., ‘Towards a catalogue of the surviving manuscripts of Catullus’, Paideia 67 (2012), 607–22Google Scholar. For the readings of Catullus' codices recentiores I have used the collations made by Berthold L. Ullman and others in the early twentieth century that are conserved among the Hale-Ullman Papers at the Department of Classics of the University of North Carolina at Chapel Hill. Cecil W. Wooten, then Chair of the Department, kindly allowed me to have the collations scanned in January 2010, and I have collated, photographed or obtained reproductions of several other manuscripts, so that here and below I can quote the readings of all known codices recentiores of Catullus that were copied before 1502.

²¹ Is the text of Catullus 11 in the Thuaneus likely to have offered any significant improvements at all on the vulgate text of the poem that was available to Vossius? There is only one place where it is very likely that that vulgate was corrupt, namely 11.11–12 horribiles et ulti- | mosque Britannos (thus already the editio princeps of 1472; Vossius writes horribilis). One would expect the Thuaneus to have written either horribilesque | ultimosque britannos with the principal manuscripts OGR (O writes uitimosque for GR's ultimosque, and all three put this word in line 12, before they were connected), or something else – perhaps the genuine reading, or something that could have helped us to reconstruct it. It is a pity that Vossius does not preserve its reading in this passage.

²² Gaisser, J.H., Catullus (Malden, 2009)CrossRefGoogle Scholar, 22 suggested that Catullus 62 may well have been added to the anthology in antiquity, when it circulated on a separate roll, given that in the Thuaneus the poem bears the title Epithalamium Catulli, which may well be ancient. But ancient titles could also have survived in early medieval manuscripts of the complete Catullian corpus; and poem 62 could also have received the title Epithalamium from any learned reader, including the person who compiled the anthology.

²³ It was already suggested by Ullman, B.L., ‘The transmission of the text of Catullus’, in Studi in onore di Luigi Castiglioni, 2 vols. (Florence, 1960), 2.1025–57Google Scholar, at 1036, that the Thuaneus represented a side branch of the OGR tradition.

²⁴ Eugenius of Toledo, c. 39: see Vecce (n. 9), 94.

²⁵ For a description of Vienna 277, part 3, see Vecce (n. 9), 95–9. Its contents correspond to two sections of the Codex Thuaneus; it does not contain any poems of Catullus. It has been controversial whether the Vienna manuscript is the exemplar or a sibling of the Thuaneus. For the latter view, see O. Zwierlein, Prolegomena zu einer kritischen Ausgabe der Tragödien Senecas. Akademie der Wissenschaften und der Literatur: Abhandlungen der Geistes- und Sozialwissenschaftlichen Klasse. Jahrgang 1983, Nr. 3 (Mainz, 1983), 15–23 and Vecce (n. 9), 106–9. Richmond, J., ‘The Relationship of Vindob. 277 and Paris. Lat. 8071’, Philologus 142 (1998), 80–93CrossRefGoogle Scholar argues that parts of the Thuaneus may well derive from a corrected copy of the Vienna manuscript rather than from the archetype. In my view the hypothesis of Zwierlein and Vecce accounts better for the codicological evidence of the two manuscripts. A third manuscript of the Florilegium Thuaneum appears to have been discovered by Jacopo Sannazaro in 1503 or 1504 at the monastery of Ile-Barbe near Lyons. This manuscript has gone lost, but before that Sannazaro evidently showed it to Aulo Giano Parrasio, who added several readings of Catullus 62 from it to the margin of his copy of the Reggio Emilia 1481 edition of Catullus, which is now Aberdeen, University Library, Incun. 165: see further Vecce (n. 9), 109–44, esp. 114–15, and Richardson, B., ‘Pucci, Parrasio and Catullus’, IMU 19 (1976), 277–89Google Scholar, at 285–8. It should be noted that Parrasio's notes on Catullus 11 in this incunable do not contain the variant fractus at line 24, nor any other reading that could have come from the Florilegium Thuaneum. I am indebted to Julia Haig Gaisser for this information.

²⁶ Vecce (n. 9), 96, cf. 99.

²⁷ See n. 10.

²⁸ Vecce (n. 9), 98–9.

²⁹ Vecce (n. 9), 99–101. Bischoff, Manuscripts and Libraries (n. 10), 145 suggests that the anthology transmitted in the Thuaneus was put together around Fleury and St Germain; on p. 139, his position on the Thuaneus and the Vienna fragment is less clear.

³⁰ Zwierlein (n. 25), 21.

³¹ Zwierlein (n. 25), 21–3.

³² Compare p. 348 above. On the composition of the anthology, see also Vecce (n. 9), 101–6.

³³ Rather confesses that he has been reading Catullus in Sermo de Maria et Martha 4, for which see Reece, B.R. (ed.), Sermones Ratherii Episcopi Veronensis (Worcester, 1969), 86Google Scholar and Reid, P.P.L. (ed.), Ratherii Veronensis Opera Minora. Corpus Christianorum Continuatio Medievalis, vol. 46 (Turnhout, 1976), 147–8Google Scholar. Fiesoli, G., ‘Percorsi di classici nel medioevo: il Lucrezio Bobiense. Raterio lettore di Plauto e di Catullo’, MR n.s. 15 (2004), 1–37Google Scholar, at 8–27, argues that Rather did not mean to say that he had actually read Catullus; against this position, see Kiss, D., ‘The lost Codex Veronensis and its descendants: three problems in Catullus' manuscript tradition’, in id. (ed.), What Catullus Wrote: Problems in Textual Criticism, Editing and the Manuscript Tradition (Swansea, 2015)Google Scholar, 1–27, at 6–8. We know about the return of Catullus to Verona around a.d. 1300 from an epigram of Benvenuto dei Campesani that is transmitted in Catullus' manuscripts GR. For the text, see e.g. Thomson (n. 20), 194; for a discussion, see Kiss, op. cit., 2–5.

³⁴ Thomson (n. 20), 25.

³⁵ Vossius (n. 1 [1684]), 9, 25, 122 and 215.

³⁶ Vossius (n. 1 [1684]), 121.

³⁷ Vossius (n. 1 [1684]), 241.

³⁸ Vossius (n. 1 [1684]), 284.

³⁹ Scaliger, I.I., Castigationes in Catullum Tibullum Propertium (Paris, 1577), 85Google Scholar.

⁴⁰ Mynors, R.A.B. (ed.), C. Valerii Catulli Carmina (Oxford, 1958), x–xi n. 2Google Scholar.

⁴¹ I know the readings of the Ambrosian manuscripts of Catullus through the collations that are preserved among the Hale-Ullman Papers at Chapel Hill, on which see n. 20 above.

⁴² Catull. 22.8 desecta Mediolanensis teste Vossio (n. 1 [1684]), p. 55: defecta aut desecta Mi.  58b.5 plumideas Med. Voss., p. 122: pulmideas Mi.  64.11 illa rudem cursu prora imbuit Amphitriten Med. Voss., p. 192: illa rudem cursu proram imbuit Amphitritae Mi.  64.194 postportat pectoris iram Med. Voss., p. 215; item Mi.  64.227 carbasus obscurata dicet ferrugine Ibera Med. Voss., p. 217: c. o. d. f. hibera Mi.  64.364 perfusae uirginis Med. Voss. p. 248: perfusa uirginis Mi.  68b.84 posset ab innupto uiuere coniugio Med. Voss., p. 293: p. ab in nupto u. c. Mi.  68b.143 ueta Med. Voss., p. 296; item Mi.  71.1 si qua uiro .....se.....obstitit hircus Med. Voss., p. 299; item, spatiis relictis, Mi¹ post correctionem.  99.10 tanquam committe spurca saliua guttae Med. Voss., p. 329; item Mi.  101.7 haec tu interea Med. Voss., p. 332: haec tum interea Mi.  Wherever I state that a characteristic reading of Vossius' ‘Mediolanensis’ is identical with that of one particular Milan manuscript, this should be taken to imply that the reading(s) of the others is or are significantly different.

⁴³ Catull. 63.5 deuoluit ilecto Mediolanensis teste Vossio (n. 1 [1684]), p. 160: d. ileto Mh.  55.16 Audacter committe crede licet Med. Voss., p. 121: a. hoc commite c. l. Mm.  92.3–4 Quo signo? Quia sunt totidem mea. Deprecor illam | assidue: uerum dispeream nisi amem Med. Voss., p. 316: thus, with amo for amem, Md.

⁴⁴ Catull. 3.10 piplabat Mediolanensis teste Vossio (n. 1 [1684]), p. 9; item Mh Mi Mm.  58b.4 cytæque bigæ Med. Voss., p. 122: citęque bigę Md¹ post correctionem, Mh: citeque bige Mm.  64.263 efflebant Med. Voss., p. 226; item Mh¹ post correctionem, Mi Mm.  64.334–7 suo loco habet Med. Voss., p. 241; item Mh Mm.

⁴⁵ Catull. 10.20 octomines Mediolanensis teste Vossio (n. 1 [1684]), p. 25: octo homines Mh Mm. Here Vossius may be reporting a garbled version of this reading, he may misattribute the reading in what is a rather complicated note, or there may be a mistake in the collations of the Milan manuscripts that I have been using.  102.3 meque inuenies Med. Voss., p. 332: meque esse inuenies Md Mh Mi. For the omission of a word by Vossius compare his omission of hoc at 55.16 (see n. 43).

⁴⁶ Blok (n. 1), 171–2.

⁴⁷ Isaac Vossius to Nicolaus Heinsius, in November 1651 (for the date see Blok [n. 1], 374 n. 131): ‘Mediolani operae fuerit describere inscriptionem quandam ... Velim quoque cures conferri Catulli quoddam exemplar satis vetustum, nempe illud, in quo Catullus solus continetur. Alterum, quod & ipsum satis est bonum, ipse olim contuli.’ For the text of the letter, see Burmannus, P. (ed.), Sylloges Epistularum a Viris Illustribus Scriptarum Tomus III ([Leiden], 1727), 632–3Google Scholar (quotation from p. 632).

⁴⁸ Thomson (n. 20), 79.

⁴⁹ Vecce (n. 9), 94. It used to be believed that Pierre Daniel had obtained many manuscripts from the Abbey of Saint-Benoît-de-Fleury in 1562, when the monastery was plundered by the Huguenots: thus e.g. Vidier, A., L'historiographie à Saint-Benoît-sur-Loire et les miracles de Saint Benoît: ouvrage posthume revu et annoté par les soins des moines de l'abbaye de Saint-Benoît de Fleury (Paris, 1965), 30–1Google Scholar. It was always a matter of controversy whether Daniel had acted as a saviour or a thief. However, Pellegrin has shown that he acquired the books from Fleury gradually between 1560 and 1569: see Pellegrin, ‘Membra disiecta Floriacensia’ (n. 19), 8 = ead., Bibliothèques retrouvées (n. 19), 162 and Mostert, M., The Library of Fleury: A Provisional List of Manuscripts (Hilversum, 1989), 29–30Google Scholar.

⁵⁰ The inventory is now Paris, Bibliothèque nationale de France, Nouvelles acquisitions latines 137, fols. 9r–26r; for a transcription, see Delisle, L., Notice sur plusieurs manuscrits de la bibliothèque d'Orléans (Paris, 1883), 70–83Google Scholar; id., ‘Notice sur plusieurs manuscrits de la bibliothèque d'Orléans’, Notices et extraits des manuscrits de la Bibliothèque nationale et autres bibliothèques 31 (Paris, 1884), 357–439Google Scholar, at 426–39; and Cuissard, C., Catalogue général des manuscrits des bibliothèques publiques de France – Départements 12 (Paris, 1889), vii–xviiiGoogle Scholar. On the circumstances in which it was drawn up, see Mostert (n. 49), 29.

⁵¹ Delisle, Notice ... (n. 50), 75; Delisle, ‘Notice ...’ (n. 50), 431; Cuissard (n. 50), xi.

⁵² Lowe, E.A. (ed.), Codices Latini Antiquiores: A Palaeographical Guide to Latin Manuscripts Prior to the Ninth Century, 11 vols. (Oxford, 1934–66)Google Scholar, vol. 6, no. 806, cf. p. xxi; see also Mostert (n. 49), 149, and the description of the manuscript in the Catalogue collectif de France, consulted online on 9 May 2014 (permanent URL: http://ccfr.bnf.fr/portailccfr/jsp/index_view_direct_anonymous.jsp?record=eadcgm:EADC:D18011388), both with further bibliography.

⁵³ At 240 × 158 mm these folios are smaller than those of the Thuaneus, which are about 290 × 202 mm, but irregularly shaped. But this difference in size does not rule out the possibility that the two manuscripts could have been bound together.

⁵⁴ Lowe (n. 52), vol. 6, 31 (on no. 804); Mostert (n. 49), 149.

⁵⁵ A title indicating the present contents of the Codex Thuaneus was added to fol. 1r in the tenth or the eleventh century: see p. 346 above.

⁵⁶ See Mostert (n. 49), 223.