The history of the text in the Middle Ages is the history of who read Catullus, where, when, and how; potentially, therefore, it is an important element of cultural history. However, though Catullus continued to be known through late Antiquity and is mentioned casually by the likes of Ausonius in the fourth century, Martianus Capella in the fifth, and Boethius in the sixth, he is essentially unknown in the so-called Dark Ages and after till nearly 1300, apart from the ninth-century copy of Catullus 62; for nearly everyone in Europe during the Middle Ages Catullus was merely a name and a few citations, no more accessible than his contemporaries Calvus and Cinna are to us today, and no extant library catalogue names him. Thus Stoppard’s Jowett is off the mark in referring to ‘‘a thousand years’’ or ‘‘thirty generations’’ of copying, for the copying of Catullus was a rare occurrence indeed.
Nevertheless, numerous claims have been made on the basis of alleged ‘‘echoes’’ that Catullus was known to some medieval author or other, but this is a precarious category of evidence that cannot be used effectively without carefully considering alternative sources, or possible intermediaries (especially in the poetry of late Antiquity or the early Middle Ages, seldom taken into account by classicists) - or even the role of mere coincidence. Past claims that are now largely rejected or at least ignored are discussed in Ullman (1960: 1029-35). More recently, Giuseppe Billanovich has claimed (1974: 46ff.) that a Brescian monk named Hildemar shows familiarity with Catullus in a poem composed around the middle ofthe ninth century, but this is rightly disputed at Tarrant (1983: 43 n. 1), and the other claims advanced by Billanovich in the same work are no more persuasive. No more certain is the alleged imitation of Catullus 68 in a work of Agius of Corvey composed in 874 alleged in Nisbet (1978: 106-7). As to the claims of Guido Billanovich (1958) that
Catullus was known to and imitated by Lovato Lovati near the end of the Middle Ages, these were greeted with skepticism in Ullman (1960) and conclusively refuted in Ludwig (1986).
The history of the text in the Middle Ages is also a matter of practical interest to the editor, and therefore of at least slight importance to the reader. Italian scholars were consulting V for as much as a century before any extant MS was copied, and their quotations and references to Catullus therefore constitute a fresh ‘‘secondary tradition’’ that can occasionally be used along with the principal MSS to reconstruct V. However, they are less reliable for this purpose than the MSS, since they worked in a tradition of excerpting that allowed quotations to be adapted to specific, often moralistic, purposes, and the collections that contain them of course have their own MS traditions in which accidental errors can occur.
Our chief evidence for what happened to the text of Catullus in the Middle Ages lies of course in the manuscript tradition. Nearly 150 MSS containing some or all of Catullus exist today in libraries and private collections (for a full enumeration see Thomson 1997: 72-91), but it is likely that only three of them have independent value for reconstructing the text (i. e., only these three do not appear to derive directly or indirectly from another MS that survives). The three MSS, all written in northern Italy near the end of the fourteenth century, are:
O (= Oxford, Bodleian Library, Canonici class. lat. 30);
G (= Paris, Bibliotheque Nationale lat. 14137, dated 1375, and probably written by Antonio da Legnano);
R (= Vatican library, Ottobonianus lat. 1829).
G was the first of these to be used by editors, but because they had no way to appreciate its importance, they used it along with a number of fifteenth-century MSS of no value at all except as the occasional source of emendations. O was first used by Robinson Ellis, but he had no inkling of its value, which it was left to Emil Baehrens to appreciate (Stoppard has Jowett taunt him over this in a memorably humorous sequence that is inspired by Housman’s ridicule of Ellis as having ‘‘the mind of an idiot child’’). R (the source of most of the fifteenth-century copies, which were only rarely influenced by O or G) was discovered in 1896 by W. G. Hale, who soon realized that a fresh proliferation of errors shared by GR but not by O showed that they derived from the Veronensis through a lost copy, which editors designate X. The reconstruction of V from these witnesses is not entirely straightforward; while it is true that the agreement of O with either G or R against the other will normally give the reading of V, disagreements of O against GR are more difficult to resolve, and the entire enterprise is complicated by the corrections, conjectures, and variant/alterna-tive readings that arose at various stages. Even the exact relationship between Vand OGR has been disputed. Though it was long thought that O and X were copied directly from Vitself, McKie (1977) and Thomson (1997) have argued convincingly that another lost copy (I shall use Thomson’s siglum and call it A) must lie between O and Xand the Veronensis (Thomson 1997: 26-7). Moreover, the very nature of Vis also disputed. This was formerly assumed to be a MS of late Antiquity in some script such as semiuncial (see, for example, Thomson 1997: 23; I wonder whether an even more difficult script such as a North-Italian pre-Carolingian minuscule might be involved), but Billanovich and Thomson now use v to designate that copy, employing Vto identify a hypothetical MS in Gothic script copied from it about 1280 (Giuseppe Billanovich 1988; Thomson 1997: 24-5). Thomson defends this view on the grounds that certain errors of OGR were induced by the misreading of Gothic script or of late abbreviations, but the former were argued away in Clausen (1976), and for the latter see Tarrant (1983: 45 n. 17); hence the hypothesis that a Gothic Vwas copied from v and served as the source from which A was later copied may be unnecessary - we need no more than a single intermediary, A, between the early Veronensis and OX.
At only two points in the Middle Ages can knowledge of Catullus be documented securely in the form of an actual copy, at (probably) Tours in the ninth century, and at Verona in the thirteenth or fourteenth. The evidence for Tours is the text of Catullus 62 contained in Paris, Bibliotheque Nationale lat. 8071, cited by editors of Catullus as T. T was copied in the third quarter of the ninth century, probably at Tours; it is an anthology, but contains nothing else by Catullus (there is an unresolved scholarly controversy as to how T is related to the eighth-century MS Vienna 277, with similar contents but no Catullus). The lines in which Tand Veither share an error or contain two variations of the same error show beyond a doubt that they have a common source (for the former cf. 9 uisere, 45 tum, 58 cura, 59 nec, for the latter 7 imbres T: imber V, 8 certes. i. T, certe si V, 63 patris T: parspatri V).
As to Verona, the earliest evidence for the presence of Catullus there is in the writings of Rather, a bishop ofVerona, around 966. Again ‘‘echoes’’ of Catullus have been alleged in his writings, but those adduced in Reece (1969) in his Index citatio-num do not survive critical inspection, not even the claim that Catullus 58b.2-3 served as the source of non angelico. . . subvectu, non pennigero, ut poeticus ille noster,... volatu (30.2.7ff.): there is no reason to take pennigero... uolatu as an echo of Catullus’ Pegaseo... uolatu when neither penniger nor uolatus is a rare word and (as Gaisser 1993: 283 n. 68 notes on the authority of a colleague), ‘‘a more likely source for Rather is Jerome: ‘et quasi pennigero volatu petulcam animal aufugit’ (Vita Pauli 8).’’ Thus the only evidence that Rather read Catullus is his own claim that he did so: in an awkwardly long sentence in a sermon on Mary and Martha (too long to quote in full), he refers to reading both Catullum numquam antea lectum (‘‘Catullus, never before read’’) and Plautum... iam olim... neglectum (‘‘Plautus long neglected’’). It is far from certain whether Rather meant that Catullus had been read by no one at all or only by himself, or that Plautus had long been neglected by everyone or only by himself; but the context does imply that these readings distract Rather from his pastoral duties, suggesting that they are not simply a hypothetical possibility but a factual one.
Of course scholars have tried to explain how an identical strain of text could be found in both ninth-century Tours and tenth - and/or thirteenth-century Verona: was it originally at or near Tours, then taken to Verona after Thad been copied? Was it originally at Verona? And if it was, does T derive its text of Catullus 62 from the Veronensis itself (on some unexplained westward journey) or from a copy of it, or perhaps from a copy of Catullus 62 alone? Such questions are intimately tied up with the notorious epigram by Benvenuto Campesani of Vicenza (d. 1323) that is found in MSS G and R. Entitled ‘‘On the resurrection of Catullus, Veronese poet,’’ it appears to claim that, thanks to a fellow countryman, Catullus has now returned to his homeland ‘‘from a distant country,’’ his ‘‘paper’’ no longer shut up beneath a bushel basket:
Ad patriam uenio longis a finibus exul: causa mei reditus compatriota fuit, scilicet a calamis tribuit cui Francia nomen quique notat turbe praetereuntis iter. quo licet ingenio uestrum celebrate Catullum, cuius sub modio clausa papirus erat.
An exile, I come to my homeland from a distant country: the cause of my return was a compatriot, namely a scribe to whom France assigned a name and who marks the route of the passing crowd.
Celebrate with whatever talent you can your Catullus, whose paper was shut up under a bushel-basket.
Thomson (1997: 26) well remarks that ‘‘The meaning of Campesani’s epigram, and the facts underlying it, are the greatest puzzles in this whole question of the resur-rectio Catulli.'’'’ The traditional literal reading takes it to convey that a notary (a calamis) named Franciscus discovered a copy of Catullus under a bushel-basket somewhere (perhaps in France, since that is where T was copied), then brought it to Verona. More fancifully, some scholars in the twentieth century entertained the notion that the words scilicet a calamis tribuit cui Francia nomen refer not to a notary but to Can Grande, a fourteenth-century lord of Verona, via a pun on French canne, meaning much the same as Latin calamus, and even that quique notat turbe praetereuntis iter refers to an equestrian statue of him set up in public (for these theories see, conveniently, Zaffagno 1975). A conspicuous weakness of such views, however, is that there was no need for an Italian to pun on French canne when his own language offered canna, with exactly the same meaning. Most recently, Giuseppe Billanovich (1988) has tried to solve the enigma (and to keep Catullus in Verona) by reference to contemporary Italian politics: the MS involved is not V but X, and the ‘‘distant country’’ is no further from Verona than nearby Padua.
Whatever Campesani meant to convey, any understanding of the epigram must surely take into account its clearly metaphorical and indeed humorous nature: after all, if the rebirth is literally true, then the return from exile is not, and if the return is literally true, then the resurrection is not - it is surely not implied that Catullus was resurrected and repatriated and rescued from under a bushel-basket. In fact, the ‘‘resurrection’’ of the title is only a metaphor, and applied with a light touch, since the concept does not recur within the poem. Instead, this begins with a fresh metaphor, the return from exile, which is itself supplanted by a third metaphor (the only one that has been recognized for what it is), the biblical notion of ‘‘hiding one’s light under a bushel’’ - hence the invitation now to celebrate Catullus’ genius (cf. Matthew 5:15, Mark 4:21, Luke 11:33; papirusoften means ‘‘wick’’ rather than ‘‘book’’ in medieval Latin). In the end, only the allusively identified notary seems ‘‘real’’; it may be relevant that Guglielmo da Pastrengo praised Campesani himself as a ‘‘remarkable poet and notary’’’ (poeta et scriba mirabilis). I suggest that Campesani’s ‘‘riddle’’ commemorates a notary named Francesco who metaphorically restored Catullus to life for his home-city by undertaking the challenge of copying the ancient Veronensis into a script comprehensible to his contemporaries - in other words, creating A, the copy of V from which OX and thus all the other MSS derive, metaphorically taking the text out from under the bushel that had concealed it and making it accessible to readers and potential admirers. As to T, there seems to be no evidence to establish which is more likely, that a complete or partial copy of Catullus was taken from Verona to Tours, or that both T and V derive independently from an archetype that was originally found in France and only later made its way to Verona.
Whatever the truth behind the epigram, the text of Catullus was indeed restored to European civilization in Verona not long before the year 1300, but in a seriously corrupted and somewhat confused condition in which many words had been mis-copied, divisions between poems blurred, and (nearly) all titles lost. The very earliest medieval allusions clearly presuppose a text without titles. Giuseppe Billanovich (1988: 38) has drawn attention to a MS of Terence closely connected to one belonging to Petrarch (London, British Library Harley 2525) in which c. 52.1 is cited ‘‘apud Catullum prope finem primi operis’’ (‘‘in Catullus near the end of the first part of his work’’); instead of confirming the theory that Catullus circulated on three rolls in Antiquity (cf. Heyworth 1993: 132), this merely shows that this medieval reader lacked titles to cite but noticed the sequence lyrics/long poems/ epigrams and used it as the basis of his citation, with the lyrics being the obvious ‘‘first part’’ of the corpus. When Hieremias de Montagnone, working between 1275 and 1320, quotes Catullus, his text again seems to lack titles and is instead divided into 12 large sections that he calls capitula - a division that, according to Ullman’s persuasive arguments, Hieremias introduced himself (Ullman 1973; Ullman also demonstrated that the remains of these divisions can be traced in O). Around 1310, Benzo of Alessandria quotes Catullus 35.1-4, in the corrupt version of the Veronensis, in his Cronica (cf. Clausen 1976: 41 f.); he introduces the quotation with ‘‘Catullus writes to his friend Aurelius,’’ which Ullman has shown derives from 21.1, the first line of the large block in which 35.1-4 appeared. A florilegium compiled in Verona in 1329 quotes 22.19-21 under ‘‘Catullus ad Varum,’’ but Ullman again has shown that this is another deduction from a faultily divided text.
These citations by capitulum or by improvised title suggest a shortage or, more likely, total absence of titles in V. The issue of whether titles were present in the Veronensis or were invented exclusively in its descendants is not merely an academic matter of reconstructing a lost MS; it is vital to the entire question of how Catullus arranged his poetry. For example, an assumption about the antiquity of the titles in the MSS is fundamental to Wiseman’s claim that the entire corpus as we have it was assembled in its current order by Catullus himself: ‘‘One might expect, a priori, that a collection entitled Catulli Veronensis liber and beginning with a dedication poem ought to be Catullus’ own arrangement’’ (Wiseman 1985: 265-6). But what if Catulli Veronensis liber (found in O only as an addition by a later hand) is nothing but the improvisation of a thirteenth - or fourteenth-century Italian?
Certainly there is no compelling reason to believe that the titles found in the descendants of V derive from Catullus himself. For one thing, ancient poetry collections seem not to have given titles to their individual components, but medieval and Renaissance MSS often supplied them, and those in OGR most frequently conform to the patterns ‘‘To X’’ and ‘‘About X’’ characteristic of those late inventions. For another, they occur only in places where the MS tradition indicates a division between poems and never elsewhere, no matter how clear it is that a new poem is beginning. These misdivisions are a relatively common phenomenon, and what we now know as a corpus of (more or less) 113 poems comprised no more than 53 in the archetype, and quite probably fewer. Mynors’s OCT (pp. xiv-xv) offers a ‘‘Carminum in arche-typo discriptio’’ which is supposed to tabulate the poem-divisions and titles of the Veronensis, but it really represents a compilation of the divisions and titles marked in one or more of OGR (see the rightly skeptical observations of Heyworth 1993: 133 n. 46); many of these - conceivably even all of them - were introduced first in A, in O, in X, or in G or R. In any case, at the most generous estimate, V indicated new poems only at (the first line should be understood when no further indication is given) 2, 4, 5, 6, 7, 8, 9, 11, 12, 13, 14, 15, 17, 21, 22, 23, 25, 26, 27, 28, 29, 30, 31, 32, 34, 35, 36, 37.1, 37.17, 40, 49, 50, 51, 52, 53.5, 54.6, 56, 59, 61, 62, 63, 64.1, 64.323, 65, 68, 69, 72, 76, 80, 89, 92, 100, and 101.
Some of the most revealing titles, of course, are those that cannot possibly be authentic under any interpretation. One class of these comprises titles introduced at incorrect divisions, such as ‘‘Ad Egnatium’’ at 37.17 at the head of a poem that combines 37.17-20, addressed to Egnatius, with 38, addressed to Cornificius, and 39, addressed again to Egnatius; ‘‘De Octonis capite’’ at 53.5; and ‘‘Ad Camerium’’ at 54.6. Other clearly inauthentic titles misidentify the contents. The title ‘‘Argonau-tia’’ bestowed on 64.1-322 is based on the same hasty misunderstanding of the poem’s opening as the note that stands in O stating that ‘‘here he tells the story of the golden fleece’’ (narrat hic ystoriam aurei velleris); probably the same thirteenth-or fourteenth-century scholar was responsible for both. Most revealing of all, however, is the title to Catullus 36, ‘‘Ad lusi cacatam’’ (i. e., ‘‘Ad lusicacatam,’’ ‘‘To Lusicacata’’), based on its opening line as corrupted in V, anuale suo lusi cacata carta (modern editions read annales Volusi, cacata carta).
Beyond confirming that we have lost any ancient titles that Vcontained, OGR have little or no value for the debate over how Catullus’ poetry circulated in Antiquity. It has been claimed that the unusually large gap that follows Catullus 60 in O, where it ends near the bottom of f. 14v and 61 starts afresh at the top of f. 15r, is ‘‘a survival perhaps of the ancient division of Catullus’ work into libelli’’ (Thomson 1997: 7), but the gap is not substantially larger than some others left by this scribe, who may simply have been unwilling to leave a ‘‘widow’’ comprising only a line or two on f. 14v. It has also been observed that, while O has very few decorated initials early in the corpus (only at 1 and 2, in fact), they have been supplied fairly consistently in the elegiac parts of the corpus (65, 68, 69, 72, 77, 80, 89), and it has been suggested that this too reflects the construction of the corpus from different sources. Another explanation, however, is that the owner of O simply took a greater interest in the epigrams than in the polymetrics, perhaps because the meter was more familiar.
The only title of significance is ‘‘explicit epithalamium,’’ which is found between the end of Catullus 61 and the beginning of 62, though only in O, where it is written not as a heading but as just another line of text in a single unbroken poem comprising both 61 and 62. The obvious interpretation - that this is the remnant of an ancient title identifying 61 as an epithalamium - is unlikely to be correct, though we are certainly dealing with a survivor from Antiquity. As we have seen, the title ‘‘epithalamium’’ belongs to 62, not to 61: it is 62 that Quintilian cites as the Epithalamium, 62 that bears that title in T. Hence I suggest that ‘‘explicit epithalamium’’ combines scraps of two originally separate notices: ‘‘explicit’’ survives from a heading such as ‘‘explicit liber hendecasyllaborum’’ that stood at the end of 61, ‘‘epithalamium’’ from ‘‘incipit epi-thalamium’’ at the head of 62 (an arrangement like ‘‘liber hendecasyllaborum explicit epithalamium incipit’’ could explain even better what we find in O). (G, on the other hand, calls Catullus 62 ‘‘exametrum carmen nuptiale’’; this is not ancient, being the definition of‘‘epithalamium’’ found in the medieval Vocabularium of Papias.)
Editing Catullus is, in effect, the process of repairing the damage that befell the text during Antiquity and the Middle Ages. It begins with the reconstruction of V; then, once the readings of the archetype are known, with whatever variants or notes were present, editors must evaluate them and judge whether they are authentic or corrupt. As to the quality of the text offered by V, it had clearly suffered considerably from scribal error in its transmission, chiefly (one expects) in the copying of Vitself from some difficult older script (see Thomson 1997: 23 for speculation about these early stages) and in the copying of A from V. The text of Catullus 62 in T shows that much of the corruption was already present in the ancestor of both Tand Vand had occurred before the ninth century, and allows us to judge how much additional corruption was introduced in V. An idea of the total extent of corruption in V can be gained by examining an edition such as Mynors’s OCT, the standard text of Catullus for the English-speaking world, with Thomson (1997) its only possible rival, for its more accurate apparatus criticus. In some respects, the picture can look quite rosy to the unwary. For one thing, Mynors rejects no lines as interpolations, though such interpolations have occurred in other authors, including Ovid and Propertius. This probably reflects both the infrequency with which Catullus was read and copied and the difficulty of reproducing the largely unfamiliar metrical schemes of the lyrics. In addition, Mynors regards only two passages as involving a disruption in the order of lines (58b.2-3; 64.377-80). In three passages, he supplies a refrain omitted by the archetype, and in one passage he omits a refrain transmitted erroneously by the archetype (64.378); in 12 other places he indicates lines lost either within a poem or between poems. In 26 passages, a line is noted as metrically deficient, with the missing word(s) supplied by conjecture when possible. In 17 lines, Mynors prints the transmitted text with one or more obeli (f) to show that it is hopelessly corrupted, with no convincing remedy yet proposed. That these cases total under 100 will perhaps convey an impression of optimism for a corpus of well over 2,000 lines. But these cases are greatly outnumbered by the passages where Mynors acknowledges an error in the archetype and replaces it with a correction that he regarded as certain. In the first 100 lines (1.1-7.2), for example, he corrects 36 such errors - just over one in every third line - while in the last 100 lines (99.11-116.8) he corrects 45 - nearly one in every second line. The text of Catullus clearly suffered from substantial corruption on the infrequent occasions when it was copied; among ancient Latin poets, perhaps only the text of Propertius was corrupted to a comparable degree.
World History
