STOP Early Journal Content on JSTOR, Free to Anyone in the World This article is one of nearly 500,000 scholarly works digitized and made freely available to everyone in the world byJSTOR. Known as the Early Journal Content, this set of works include research articles, news, letters, and other writings published in more than 200 of the oldest leading academic journals. The works date from the mid-seventeenth to the early twentieth centuries. We encourage people to read and share the Early Journal Content openly and to tell others that this resource exists. People may post this content online or redistribute in any way for non-commercial purposes. Read more about Early Journal Content at http://about.istor.org/participate-istor/individuals/early- journal-content . JSTOR is a digital library of academic journals, books, and primary source objects. JSTOR helps people discover, use, and build upon a wide range of content through a powerful research and teaching platform, and preserves this content for future generations. JSTOR is part of ITHAKA, a not-for-profit organization that also includes Ithaka S+R and Portico. For more information about JSTOR, please contact support@jstor.org. THE MANUSCKIPTS OF CATULLUS By William Gardner Haib In the Classical Review for April, 1906, I published a paper entitled "Catullus Once More." In this, I corrected certain readings which appeared to me erroneous in Ellis's new edition, and which I saw must cause R temporarily to be misjudged, sum- marized my convictions about the relations of R to other MSS, including M, and announced that a year's leave of absence for the re-examination of my work upon R and the study of the entire tradition had been granted me by my university. Later, the university appointed Mr. B. L. Ullman, long trained in my Catullus courses, as research assistant to aid me. The programme has been carried out in substance. Beginning in March, 1907, I have spent nine months in Europe (my time being necessarily reduced), and Mr. Ullman has spent over a year and a half, beginning in the summer of 1906. In addition, another student of the same training, Mr. Evan T. Sage, has rendered me the great service of devoting three months, in the spring of 1907, to the collation of Catullus manuscripts. Among us, we have collated all the manuscripts of the existence of which we have been able to get information. By far the larger part of the secondary MSS were collated by Mr. Ullman, some of them in consultation with me (in the course of which my great confidence in his perspicacity and judgment was fully confirmed), but many necessarily without consultation. I have myself, besides collating various secondary MSS, studied anew O, G, R, and M, with a large expenditure of time. Each of these four manuscripts, too, Mr. Ullman and I have finally gone over together, line by line, after he had made preparatory studies, and after many previous discussions of difficult individual points. My purpose in the present paper is to publish a preliminary list of MSS, asking for information in case any reader knows of other MSS, and to discuss a few points of especial interest. [Classical Philolooy III, July, 1908] 233 234 William Gabdneb Hale In the paper cited, as in earlier papers (thus in Hermes XXXIV, pp. 133-44) I have stated these principal convictions: G and R are not direct copies of Ver. (the "lost Verona manu- script"), but of a copy of this, which for convenience I have called alpha. This (likewise lost) manuscript was sent to Coluccio Salu- tati in consequence of the well-known letters to Benvenuto of Imola and Caspar of Verona written in 1374 and 1375 (the first and third on p. xvi of Schwabe's generally very careful edition are assigned to the wrong years). The text of a, not of G, was finished on October 19, 1375 — a date which fits sufiiciently well the date of the latest letter of Coluccio. All our other MSS (with the exception of T) are descended from O, G and R — principally from R, with a certain amount of crossing from O and G, and a great deal of crossing to and fro in the sub-families descended from R. Compare the following: cii. 1: ab antiquo OGR, ab amico W cett. Ixxviii. 9: Verum id non 0, Veruni non id G, Id uerum non R cett. Ixxvi. 11: offirmas OG, affirmas R cett. Ixxxviii. 4: quantum O G Rice. 606, quamtum D (the archetype of D and Rice. 606 was partly derived from G), tantum R, corrected to q"ntum {— quantum) by R^ (but in such a way that the long stroke of the letter q, which letter was meant to cover the ta, could easily be read as only a deleting stroke for the a), tantum cett. If this tenet is sound, only that part of the tradition which is found in O, G or R can possibly be genuine, and critical editions will ultimately give the readings of O, G and R alone. My present larger acquaintance with the MSS tends to confirm this conviction. Our editors have in general given us only scattering readings outside of and G, while maintaining or assuming that any MS cited by them might yield us, at this and that place, the genuine tradition or a hint of it, as against O and G. My opposing con- clusion was based upon complete collations of twenty-two MSS and two partial MSS, and it is to these that I refer in the "cett." above. But I desired a fuller knowledge, feeling that the first critical problem, namely what MSS we are to start from in recon- stituting the text, could not finally be settled until we .could control a much larger number. Ultimately, I could not content myself with less than all. Before my assistant and I began the work of The Manuscripts of Catullus 235 collation, I had made out, partly by the help of old editions of Catullus (as Sillig's), partly by searching old editions of TibuUus or Propertius, a list of 96 MSS (Heyse, 1855, mentions 50, and says that some 70 exist). This subsequently, in a few cases by chance information, but mainly by the ransacking of catalogues, rose to the number of 120 MSS (not including the Florilegia, etc.) a few of them fragmentary. I possess complete collations now of all but Paris. 8231 (fragmentary), seventeenth century; Ashb. 973, sixteenth century; Brancaccianus IV, A. 4. (fragmen- tary), seventeenth century; Vatic, lat. 7044, sixteenth century; and an eighteenth -century copy made for Santen (see list below). These, except the last, contain readings seemingly gathered from every quarter, and, in the press of time ' and in view of their com- plete worthlessness for my purpose, the five were not collated (specimens only being taken), though two seventeenth-century MSS (fragmentary), contained in Rice. 2242, were collated in full. New collations w«re made of the twenty-four manuscripts which had been collated by or for me before. Through the courtesy of M. L6on Dorez, Don Marcelino Me- n^ndez y Pelayo, and the editors of four English journals, I have advertised for information about any MSS outside of a list given for France, Spain, and Great Britain and Ireland respectively. But even for these countries, it is quite possible that my list is incomplete ; while it is more than possible that MSS still remain for Germany, Italy, Switzerland, or Russia. There is nothing from the two last-named countries. I should be greatly obliged for supplementary information, which should be addressed to me at the University of Chicago. I should be glad, also, to receive help with regard to any of the unidentified MSS or unfound material given at the end.^ I have been obliged to dispense with dealing with hand-notes found in the margin of printed editions and purporting to give 1 The mere task of collation occupied what corresponded to the entire work of one man for two years and six mouths. 2 1 take the opportunity to thank Professor Jacoby of Kiel for his courtesy. He had not happened to see my paper in the Classical Review. Before allowing a student of his, however, to go to Rome to study R for the purpose of writing a dissertation upon it and its relations to other MSS, he wrote to me to ask whether I was continuing my work, and, upon learning the facts, at once withdrew his student. 286 William Gardner Hale readings from MSS, though I have done enough to make it seem probable that nothing is to be gained from this source. If some- one should undertake the work with a different result, I should be glad to be found wrong. In the list that follows, the initials C T P stand for Catullus, TibuUus, and Propertius respectively. The sign + means that more or less from other literature, generally humanistic, is also con- tained in the MSS against which it appears. The other abbrevia- tions explain themselves. PRELIMINAKY LIST OF CATULLUS MSS AUSTRIA Vienna K. K. Hofbibliothek 224.— C. T. P. 3198.— C. Petron. T. +. Ann. 1M0(?). 3243.— C. (frag.) -f BELGIUM MONS Bibliofhique de la Ville 218. 109.— T. C. P. 361.— C. T. P. FRANCE Cabpbntras BibliotMque de la Ville Gbenoblb BibliotMque de la Ville 858.— G. T. P. Ann. 1472. Paris Bibliothbque Nationale 7989.— T. P. C. Petron. +. Ann. 1423. 7990.— T. C. P. 8071. — luu. C. (frag.) + 8231. — C. (frag.) +. Seventeenth century. 8232.— C. Verg. Priap. + 8233.— C. T. P. Ann. 1465. 8a34.—T. C. 8236.— F. T. C. Verg. Priap. 8458.— T. P. C. + 14,137.-C. Ann. 1375(?). The Manusoeipts op Catullus 237 GERMANY Berlin K. Bibliothek Diez. B. Sant. 36.— G. + Diez. B. Sant. 37.— G. +. Ann. 1463. Diez. B. Sant. 40.— G. Diez. B. Sant. 46. — C. (Made at end of the eighteenth century for Santen. Probably a copy of the Edinburgh MS, then belonging to the church of S. Giovanni a Carbonara, Naples.) Diez. B. Sant. 56.— G. Ann. 1481. Dresden K. Offentl. Bibliothek Dc. 133.— G. P. T. G5TTIN6BN K. Universitdts-Bibliothek Philol. llf'.—T. P. C. +. Ann. 14B6(?). Hamburg Stadthihliotheh Serin. 139. i".— T. P. C. Munich K. Hof- und Staatsbibliothek lat. 473.— G. lat. 7471. — C. (carm. xlix only). WOLFENBUTTBL Herzogliche Bibliothek 65. 2. MSS.—G. T. P. + Qud. 283.— G. Qud. 332.-1. G. + GREAT BRITAIN AND IRELAND Cheltenham (England) Phillipps Library 3400 (apparently Saibant. 323).— G. Dublin (Ireland) Trinity College Library 1078 (formerly Phillipps 9590).— F. G. Edinburgh (Scotland) Advocates' Library 18. 5. 2.—G. Ann. 1495(?). 238 William Gardner Hale LiscoNNAN (Ireland) Library of Samuel Allen, Esq. P. T. C. Verg. Priap. + (apparently PhilUpps 6433). Ann. 1467. London (England) British Museum 10, 386 (apparently Saibant. 329). — C. Ann. 1474. 11, 674.— T. C. 11, 915.— C. 12, 005.— Mart. C. (frag.). Burn. 133.— C. Harl. 3574.— T. P. C. + Harl. 2778.— P. C. Harl. 4094.— C. (frag.) + Lincoln's Inn, Library of Walter Ashburner, Esq. C. (to Ixi. 134) T. (from ii. 4, 13). Ann. 1451. Oxford Bodleian Library Lat. class, e. 3.—T. P. C. + Lat. class, e. 15 (formerly PhilUpps 3364). — C. + Lat. class, e. 17 (formerly PhilUpps 9591).— C. T. +. Ann. 1453. Canon, lat. 30. — C. Canon, lat. 33.— T. C. Canon, lat. 34. — C. T. Verg. Priap. Laud. lat. 78.— T. C. Eichmond Library of Sydney C. Cockerell, Esq. C. HOLLAND Leyden Universiieits-bibliotheek Voss. Lat. in Oct. 13.— T. P. C. + Voss. Lat. in Oct. 59.— T. C. +. Ann. 1453. Voss. Lat. in Oct. 76.— C T. Ann. 1451. Voss, Lat. in Oct. 81.— Verg. Priap. Verg. + Petron. C. T. P. ITALY Bergamo Biblioteca Civica X 2. 33 {3).—T. P. C. + Bologna Biblioteca Universitaria 2621.— C Ann. 1412. 2744.— C. The Manuscripts of Catullus 239 Brescia Biblioteea Civica Queriniana A VII. 7.— P. C. T. + Cesbna Biblioteea Comunale e Malatestiana XXIX. sin. XIX.— G. T. Florence Biblioteea delfu Signor Landau 306. 314.— T. C. R. Biblioteea Medieeo-Laurenziana Plut. XXXIII. cod. XI.— C. P. T. Pint. XXXIII. cod. XII.— C. T. Ann. 1457. Plut. XXXIII. cod. XIII.— C. Pers. Plut. XXXVI. cod. XXIII.— Ou. Fasti, C.+ Strozz. 100.— C. (carm. xlix only) + Ashb. 260 (apparently Saibant. 324).— C. Ashb. 973.— C. Sixteenth century. R. Biblioteea Nazionale Centrale Magi. VII 948.— PeTs. luu. C.+ . Ann. 1476. Magi. VII. 1054.— T. C. Magi. VII. 1158.— C. Panciat. 146.—YeTg. Priap. T. C. + . Ann. 1475. JB. Biblioteea Rieeardiana 606.— C. T. + 2242. No. 25. — C. (carm. Ixiv only) + . Seventeenth century. 2242. No. 25 bis.—C. (carm. Ixiv only) + . Seventeenth century. Genoa Biblioteea Civiea MS. D^i^ 4. 3. 5.— T. C. Milan Biblioteea Ambrosiana D 24 sup. — C. G 10 sup.—T. C. (frag.)+ H 46 sup.— P. T. C.+ J 67 sup.—C. P. T. M 38 sup.—C. R. Biblioteea Nazionale di Brera AD. XII. 37 No. 2.—T. C. Ann. 1450 (?). Naples R. Biblioteea Braneaeeiana IV. A. 4.—C (frag.)+- Seventeenth century. Biblioteea Oratoriana de' Oerolimini (Filipino) MSS. membr. XIII Pil. X. No. XXXIX.— P. C. T. Stat. Siluae. Ann. 1484. 240 William Gardner Hale B. Biblioteea Nazionale IV. F. 19.— C. T. P.+ IV. F. 21.— C. P. IV.F.61.—G.+ IV. F. 63.— Stat. Achill. Ou. C. (frag.) + Padua Biblioteea Capitolare C.77.—T.C.+ Palermo Biblioteea Comunale 2 Q. q. E. 10.— T. C.+ Parma B.. Biblioteea Palatina H H. V. i1 {716).—P. C. T. Ann. 1471. Pesako Biblioteea Oliveriana 1217.— C. T. P. +. Ann. 1471. Rome B. Biblioteea Casanatense Biblioteea Chigiana Biblioteea Corsiniana 15.— T. P. C. (frag.). H. IV. 121.— C. T. 43.D.20.—T.C.+ Biblioteea ApostoUca Vatieana Barb. lat. 34.-1. P. C.+ Ottob. lat. 1550.— C. P.+ Ottob. lat. 1799.— C. Ottob. lat. 1829.— C. Ottob. lat. 1982.— C. (frag.)+ Palat. lat. 910.— T. Ou. P. C. +. Ann. 1467 (?). Palat. lat. 1652.— T. C. P.+ Urb. lat. 641.— C. T. P. Urb. lat. 812.— C. Vatic, lat. 1608.— C. Verg. Priap. Ann. 1479. Vatic, lat. 1630.— Plant. C. Vatic, lat. 3269.— C. Verg. Priap. + Vatic, lat. 3272.— P. T. C.+ Vatic, lat. 3291.— hucr. Pers. Verg. Priap. C. T.+ Vatic. lat. 7044.— C Ann. 1520. S. Danible Biblioteea Civiea 56.— P. T. C. The Manuscripts op Catullus 241 Siena Biblioteca Comunale H. V. 41.— C. + Venice Biblioteca Nazionale Marciana lat. 12. 80.— C. lat. 12. 81.— T. C. lat. 12. 86.— Ou. C. + lat. 12. 153.— T. C.+ Museo Civico MS. VI. No. 117. 549.— T. P. C.+ VlCENZA Biblioteca Comunale Bertoliana Q. 2. 8. 12.— T. C. P.+. Ann. 1460. SPAIN El Escoeiale Real Biblioteca IV. C. 22.— T. C. P. + IV. C. 22. — C. (in same volume as the above). FLOKILEGIA AND EXTRACTS AUSTRIA Krakau Universitdts-Bibliothek No. 3244. DD. XII. i5.— Selecta Phalericorum {sic). Q. Valerii Catuli, Veronensis. FRANCE Marseilles Bibliothhque de la Ville 1283. — Extracts. Seventeenth century. Nice BibliotMque de la Ville In schol. ad luu. (See Beldame Rev. de Phil. VI. 76). ITALY Rome R. Biblioteca Casanatense 904. — Florilegium sententiarum ex latinis scriptoribus excerptarum. Biblioteca Apostolica Vaticana Vatic, lat. 7192. — Extracts. Sixteenth centtiry. 242 William Gardner Hale Verona Biblioteea Capitolare CLXVIII (155). — Flores moralium autoritatum. Ann. 1329. GREAT BRITAIN AND IRELAND London British Museum \21,908. — fol. 45 V. Benvenuto de Campesanis, Ad patriam, etc.] MSS AND OTHER MATERIAL NOT FOUND (OR NOT IDENTIFIED) MSS of D. Baldassarre Boncompagni (catalog, ed. E. Narducci, 1862) No. 219. — Extracts. Seventeenth century. Saibanti and Canonici MSS (sale catalogue, Sotheby, 1821): No. 95. — C. P. paper. Sixteenth century. No. 96.— C. P. paper. No. 431.— P. C. {PhilUpps 9590 f) Vatican, Rome (Miintz-Fabre, La biblioth^que du Vatican au xv s., Ecole frangaise de Rome 1887, p. 101), Poeta Caculi boronensis (sic). 2 MSS Collegii Jesuitarum, Rome (Heyse, p. 288 implies one. A MS note on the title-page of the Corradini edition in Paris, Bibliothfeque Nationale R6s. gYc 223, mentions two, describing the first as "olim Biburgiensis," paper, containing T. C. On first fol. v there is a note, "Tibullus et Catullus exemplare Michaelis Angeli e Decimis de Burgo Sancti Sepulchri." The second MS was of parchment, written in 1460 by Joannes Carpensis (is this the Carpensis referred to in MS notes to Cat. vii. 9 and elsewhere in this edition?). See also the Parma edition of 1794, p. vi). Cavrianeus (Heyse, p. 288). 2 Bossiani (Heyse, p. 288). Library of Nicolaus Trivisanus. — C.T. parchment (Tomasini, Bibliothecae Patavinae manuscriptae publicae et priuatae, 1639, p. 107). Library of Federico Ceruti Veronensis (Tomasini, Bibliothecae Venetae 1650, p. 98). Catullus num. 3 (Leyden Voss. lat. in oct. 76 formerly belonged to Ceruti). Tournay, Belgium: Inter MSS codices Dionysii Villerii .... nunc Eccle- siae Cathedralis .... (Bibliotheca Belgica, Sander, 1641, p. 219). London: in aedibus lacobaeis (MSS Angliae, T. ii p. 247, No. 8236). " Angeliani 2 simul conglutinati, memb., quorum primus integrum exhibet C, secundus non nisi tria eiusdem poetae elegidia cum nonnullis Ovidii poematiis, et veterum epigrammatum libello. Habuit hos olim in domestica bibliotheca Antonius Angelius, qui seculo xvi humaniores literas in Pisana urbe docuit " . . . . (C. T. P. Parma edition, 1794, p. vi). The Manuscripts of Catullus 243 C. T. P. Parma edition, 1794, p. v: "Res igitur nobis fuit pane ex integro conficienda: Catulli codices plures e pulvere et situ evocandi: dispiciendivetereslibri: in aliorum interpretamenta, aut coniecturas curiosius inquirendum Quae omnia quum f usius alibi simus exposituri, deque toto Veronensis poetae exornandi apparatu uber- tim dicturi . . . ." (was a second volume of notes ever published?). Codex Petri Flerardii (Rob. Titius, Loc. controv. lib. ix, c. 21). Paul Heyse, Nachlass iiber Catull. My preference would be to deal with the whole of my material first, testing completely my tenet that all the other MSS come from OGrR, and publishing the evidence. But this will cost a good deal of time, and my collation of R has been long delayed. I therefore shall assume that the result will be as I have thought, and, as announced some time ago, shall publish a continuous restored text of the lost Verona MS, with the readings of O G and R below. There will be a number of corrections of previous reports upon O, and a larger one of those upon G. I wish now to make a few remarks upon my paper in the Classical Review, and to add a supplementary report upon several other points of especial interest. Of the three great MSS, O is to my mind the oldest in its style of writing, R the next, and G the youngest. But of course the actual order of writing may have been different. An older man and a younger may, on the same day, write hands belonging to fashions separated by many years. Certain things in the ornamentation of the first initial in O, if contemporaneous with the text, preclude a date much before 1400. The question whether these are original could be settled only by expert students of illumination. With a feeling that the style of the writing allowed a date fifty years earlier than had been assigned to it, I surmised that O might have been Petrarch's copy, and made this suggestion in my paper in the Classical Review, discussing the matter no further. The principal reason for this ascription was the resemblance of the "Catullus Veronensis poeta" written (partly in abbreviations) at the top of p. 1 by a hand different from that of the corrector and 244 William Gardner Hale from that of the annotator. Several distinguished scholars have agreed that the writing might be Petrarch's. The light curving and rising final stroke of the e's, the long sweep of the uirgula, and the angle of the two strokes of the a's, are in Petrarch's manner. But other experts have differed; and I am myself now convinced that the movement of the writing is somewhat freer than Petrarch's. It was always a difficulty, too, that notes in Petrarch's style of comment were lacking in the margins. Further, several of Petrarch's readings in his quotations differ from those of O ; and Petrarch, if he had made the emendations involved, would have been very likely to note them in his manuscript. I therefore withdraw my suggestion. Less credit may in consequence at first be given to another assignment which I have for some years been inclined to make, and which, after very careful comparisons, I now publish with confidence. The "corrector" of R (R^) was its owner, Coluccio Salutati. The MSS most serviceable to me in the identification have been a MS of Seneca's tragedies and the Ecerinis of Mussato in the British Museum (11,987), written throughout by Coluccio and signed by him at the end, and a MS of Petrarch's De Viris Illustribus in the Vatican (Ottob. 1883), which, beside other marginal notes, has many by Coluccio. The resemblances are complete, not only in the style of the letters (some of which have very marked peculiari- ties), but in the manner of making the frequent braces, the draw- ing of the pointing hands, and the character of the many examples of .No. used for the same purpose. I am also convinced that Novati is wrong [Epistolario di Coluccio Salutati I. 222, footnote 2) in attributing the ownership of G, on the basis of notes which he does not specify, to Coluccio. The hands of G^ and Coluccio are similar, but not the same. G was perhaps, or probably, written in Coluccio's office, but it was not corrected by him. In the article in the Classical Review I gave specimens of the proof of what I had before said in several places, that M is a direct copy of R. I then supposed that the scribe of M had G also before him, and that he selected a few readings from it, though in the great majority of cases following R. The Manusceipts of Catullus 245 I also said, apropos of aii indisposition to welcome R which has shown itself in certain quarters, that "the actual facts with regard to the relation of G and R will prove to be so dramatically overwhelming in favor, not only of R's position as a primary MS, but of its high position, that I am enabled easily to bear the present prevailing skepticism." I further said that Gr and R had too many corrections in common for chance, and hinted that G had been corrected upon R, giving as an illustration the readings, xiv. 16, fit O G, sit G' R (both readings being wrong). To this I may here find space for a single other specimen. In Ixiv. 355, O h&s prostei^net (right) and G originally had it; but G is cor- rected to the wrong reading prosternens, which is the original and untouched reading of R (by the second hand, supplying verses omitted by the scribe). Cases like this outweigh a few others, in which R is corrected to the right reading, found also in G. But there is another possibility, of a still more disturbing kind, which must also be reckoned with, namely that G may have been corrected upon M; with which must go the conclusion that the text of M was not based upon R and G, but upon R alone. , The facts mentioned are equally well explained upon this hypothesis, since, in the main, M is a pretty exact copy of R. Thus in the passages mentioned M has sit and prosternens. We come to this question presently. Meanwhile, let it be understood that most of the descriptions about to be given bear upon it; and, further, that the present discussion is necessarily provisional, not final, and that the figures are subject to slight changes. In R, after the verses of Benvenuto, we read Catulli Veronensis liber incipit ad (perhaps Ad) Cornelium, written by R^ in one line, and apparently at one time. G has, by the first hand, Catulli Veronensis liber Incipit, with a period. To this G^ adds, in the same line, ad Cornelium, with an additional period. It will be remembered that, in many cases where there are no interstices between poems in O and G, G has a paragraph mark in the margin. O, in almost every such instance, has two light lines of equivalent force. In all these cases, R has an interstice, and, generally, a catch-title in black in the margin, and a title in red in the interstice, both by the second hand. 246 William Gardner Hale The exact agreement of O and G as regards the interstices makes it morally sure that they represent the arrangement of the poems in Ver. The departure of R in giving an interstice wher- ever there was a sign of division was then due to a deliberate plan. In one place, at xxxviii, R rightly separated a poem where G has no paragraph mark, though O has the two lines. In accord- ance with his usual custom in such cases, R indented three verses, and began with Ale est, putting an m in the margin to guide the illuminator. R^ made a condemnatory note upon this in the mar- gin, and, after certain erasures, filled in the reserved spaces in the verses. He then expunged the whole of the first verse and added a uacat {tia at the beginning and cat at the end of the verse). This being done, he wrote the verse anew in the interstice, thus closing up the page to the eye. The interest in the matter lies in the indication that G^ did not get his paragraph marks from his archetype, but from R or a descendant of R, and in the fact that, in doing his correcting, R^ wrongly repeated the est from verse 1 in filling up the reserved space in verse 2, thus making the read- ing Male est si, from which the reading in nearly all our manu- scripts comes. Several of the earlier titles in G are by the first hand. Thus De phasello was written in the interstice before iv. R has the catch-title de phasello in the margin by the first hand, and the title De faselo in the interstice by the second hand. In the inter- stice before v, the first hand in G wrote De lesbia, which is cor- rected by the second hand io Ad lesbiam. In R, the first hand wrote de lesbia in the margin, and the second hand wrote Ad lesbiam in the interstice — evidently a correction of the probably inherited title written by R'. Before vi, G wrote a title of which enough remains to make it morally sure that it was Ad se ipsum. G' changed this to Ad Flauium. R has in the margin ad se ip- sum, probably by the first hand, to which, farther out in the mar- gin, the second hand added ad flauium, evidently again a correction. In the interstice, the second hand wrote Ad Flauium (possibly with a small /) . For xvii, G has the paragraph mark in the margin, and R an interstice, but neither has any title. The Manuscripts of Catullus 247 For xxii, R^ wrote Ad suffenum in the margin, but afterward crossed out the second word, replaced it by Varum, and, when he came to fill in the titles in red, wrote Ad Varum in the interstice. G^ wrote ad Varum in the margin. For the last two words of xxxii. 1, W wrote a variant aV Ipsicilla. Opposite the interstice he wrote a catch-title Ad Ipsicillam in the margin, and in the interstice Ad ipsicillam. G^ wrote the title Ad Ipsicillam in the margin, though the last words of G's first verse are ipsi thila, corrected by G^ from ipsi thili [ipsi Thila in R, ipsi ilia in O). For XXXV, R^ wrote Ad libellum suum de cecilio in the margin (this comes down in substance in the CLA family), and Ad Cecilium iubet libello loqui in the interstice. G^ wrote Ad Cecilium Iubet libello loqui in the margin. For xlix, M and two unreported MSS which have come directly down from R by separate traditions, apart from the main MSS, show that the original title was pretty surely Ad romulum. This corresponds in length to an erased catch-title in the margin of R, and an erased title in the interstice. G doubtless had originally the same, since G^ and R^ everywhere else agree in the titles. From Ixii inclusive on, R^ wrote titles in the interstices only. Before Ixxvii, 0,G and R have an interstice. R has ad ruffum, probably by the first hand, in the margin, and Ad Bufum, by the second hand, in the interstice. G has Ad Bufum by the second hand in the interstice. The interest here lies in the appearance of a title by the first hand of R so late in the MS. In R, titles are lacking in the interstices only for vii [ad lesbiam in the margin), viii (ad se ipsum in the margin), xvii (nothing in the margin), and before the wrongly separated At pater ut summa, etc., Ixiv. 241, for which a hand later than R^ has written, in the margin, Fletus egey, which does not appear in M. G^ has put his regular paragraph mark in the margin, but gives no title. O has no indication of any division. Thus where R^ has a marginal title, but none in the interstice, G^ has written the same title in the interstice ; and where R^ has written no title, G^ has written none. To state conclusions briefly, it looks as if the titles written by 248 William Gaednee Hale the first hand and corrected by the second in G had been corrected from R or M, and as if the titles by the second hand in R, except those which appear in the first hand in the interstices in G or the first hand in the margin in R, had been invented by R^ (mainly on the basis of vocatives, which appear generally in the first lines of the poems in such cases), and copied' by G' from R, in the form which R^ ultimately decided upon, or from M, which everywhere follows the title finally adopted by Rl At first thought, it would seem to be easy to decide which of the two MSS G^ had used. But it proves to be difficult. The close connection of G^ and M is obvious enough, and appears in many places. Examples follow: liii. 3: crimina OGK; aV carmina G^; earmina M. Ixi. 61: Nichil OGK; Nil (¥ M; aV nihil M\ Ixiii. 13: dindimene OGR; dindimenee G^ M. Ixiii. 91: dindimei OGK; dindimenei G^ M; aV dindimei M?. Ixiv. 307: His OGR; aV hie G'; Hie M. Such cases are indecisive. M may have chosen the variant or corrected reading of G^ for the reading in his text, or G^ may have taken his variant or correction from the reading in M's text. From the very large number of cases (more than 70) where the corrected readings in the text of G and R agree with each other and with the readings in the text of M, there is likewise no con- clusion to be drawn for our question. G^ may have corrected G upon R, or upon M. The field where we should look for evidence is in the behavior of M (the scribe writing the text) where G and R (or R^) dif- fered. There are about 130 examples like the following:^ 1x1. 5: Hymen O; hymenee hymen ORM; O hymenee hymen G. Ixvi. 83: colitis OG; queritis KM. Ixviii. 51: Nam OKM; Non G. 1 Bonnet Remie critiqiie d'histoire et de littirature, 1877, 1, 57, rightly says that the titles in G must have come from some other MS, and suggests that their source was some other copy of Ver. Giri De Locis qui sunt aut habentur corrupti in Catulli carminiltms, p. 15, likewise says that G's titles were taken from another MS, and sug- gests B, or some other source of the same kind. Both were right in their main point. But B does not satisfy the conditions, and the source also turns out not to be an inde- pendent copy of Ver. 2 Examples where the differences are only in spelling are not included. But note xvii. 6, where M writes salt subsali, mistaking the ci of E's salt subseili for an a (M* adds aV subseili), and ixviii. 2, where OG have sarcinulis, and RM sarcinolis. The Manuscripts of Catullus 249 On the other side are to be considered the cases following, in which M agrees with G or G' against R or R^. I arrange them by groups: A. xxi. 6: experibis OR; experibus G M; aV bis G^ M^. xcv. landB: Zinir na OGB,M.; Zmirna^'M?. Similarly in 6: ZmiV- nam OGRM; Zmirnam E^Ml Ixiv. 52: Uttore, O; litore G; litora R; litora M; with e written above the a by W. B. xxxix. 12: lamiuinus OG; lanuuinus R; lamuinus or lamiunus (no apex present) M; lamiuinus M'. [Ixiv. 180: ne OGRM; ue R'M^; uf a late hand in R.] Ixiv. 211: uisere OR; uiscere GM; with the s deleted by a line by G^, and expunged by M'. C Ixi. 151: tuits O, and R^ correcting himself; tuis GW (R omitted the verse) M. Ixiv. 123: in memori OR: in memori G; aV nemori G^; in nemori M; corrected, by an added stroke, to in memori Ml Ixiv. 176: requisisset OGR; with an r written above the first s by G^; requirisset M; corrected to requisisset by M^. Ixiv. 282: pit OGRM^; parit G^M. D. Ixiv. 360: flumina OR; lumina G; aV flumine G^; flumine M; aV lumina .V. flumina M^. xcviii. 4: carpatias 0; carpatians R; in marg. carpatinas (no aV) R^; carpatias G; carpatians M; aV carpatias Ml The first example under D proves conclusively that the corrector of M had G before him, and, if this is so, M also may easily, though not necessarily, have had G before him. In the second under D, M^ probably m^ant carpatinas. Com- pare his sub tegmid, so written in Ixiv. 347, though elsewhere spelled out. The first example under A seems to indicate that M used G. Yet just as experibus was a sheer blunder in G, due to the fact that -ibus is a common tertaination, it may conceivably be merely a coincident blunder in M. M makes many worse blunders. More- over, the same reading experibus is found in Leidensis 76, a MS of the BAV family which stands in no close relation to G, and which therefore probably owes the reading to the repetition of the same blunder. 250 William Gardner Hale From the Zinirna, -am, examples under A, if the whole argu- ment turned upon them, M would seem to have had G before him. Still, a certain amount of individual decision must be allowed for in the case of every scribe. It was easy to see what the three words had originally been in R, and M may have been copying from R alone, but in the present place have preferred the uncor- rected readings to the corrected. This is as easy to suppose as that he preferred G's uncorrected to R"s corrected readings. In the third example under A, the correction by M^ is a simple and easy one, and in so far of less weight. The examples under B are inconclusive. The change from tie to ue in R in the second example is not in the style of R^ (it may conceivably be by M^, who did at any rate add a variant or two' in R; or it may be by another hand, not much later). Yet, even if one assigns the change to R^, M, as above, may have preferred the original reading of R to the obviously corrected one (made by a heavy stroke below, with no erasure of the full connecting stroke of the ii). The third example under B likewise means little, since spelling in these MSS is largely a matter of individual habit. A large accumulation of such cases would have weight. A single one has practically none. The first example is likewise not worth much, since lanuuinus is so written in R that it could easily be read lamiuinus. Indeed, Ellis in his recent edition cites R as having ^^fortasse" lanuinus (probably an error in transcrib- ing lamiuinus), though it certainly has the last-named reading. Similarly, all editors up to the present time have reported irru- masti for G in xxviii. 10, though it unquestionably has irruinasti. The first example under C indicates that M used G, unless Coluccio added the correction in R after M was copied, or unless (more probably) the tuis in M is an accident, of a kind that fre- quently recurs elsewhere. Thus in ix. 8 R wrote ut mos est tuis (corrected by R^), though the agreement of OG in the right read- ing makes it morally sure that a had tuus. The second and third examples under C indicate that M had G^ before him, unless G was corrected upon M before the corrections ^AV crude, Iv. 16, and aV tibi, Ixiv. 276, were not written by B?; and are com- pletely in tiie style of M^. The Manuscripts of Catullus 251 (which are sound and easy ones) were made in M by M^. The fourth example points to the same alternatives, though with less force, since the interpretation of the particular abbreviation used is always uncertain (cf. seperata and separata in certain MSS of the BAT family in xvii. 19). We have to chose, then, between the two following hypotheses : I. M used both R and G, after both had been corrected. II. a) M was copied from R (and possibly before the death of Coluccio, 1406, who in that case may have made a correction or two after M was written). b) The owners of M and G lent each other their MSS. The corrector of M (who may well have been the owner) took a reading or two from G, which MS may have been only recently written, or have been written earlier, and, like O, have been left uncorrected. I incline toward the latter hypothesis. The flourishes above the tall letters in the first line of each page point to the execution of the text in a chancelry. The most probable hypothesis is that G was copied in Coluccio's office upon the arrival of a; that he then, looking at it and being dissatisfied with the irregular arrangement of interstices and titles, gave orders to another scribe to make another copy, which was R ; that Coluccio adopted this, corrected it, and put upon it, in his regular manner, its number in his library and the indication of the number of folios contained ("71 carte 39," written in Coluccio's hand at the top of p. 1 of the text). c) After the corrector of M had used G, the owner of G in turn borrowed M, and the corrector (who may again have been the owner) thoroughly revised his MS upon the basis of the other, getting from it the paragraph marks, the changes in the text which are in his hand, and the variants. To these, he added certain corrections of his own, not found in M; and the corrector of M similarly added a few corrections in the latter after its return to him, perhaps casually in the course of a fresh reading. For the first alternative (I) as against the second (II), the only evidence of weight, unless I have overlooked something, lies in the first and second examples under A and the second under B (supposing that it was R^ who changed we to ue in the last- named). The readings here, as already said, may be otherwise 252 William Gardnee Hale explained. Against this alternative is the fact that the total pos- sible number of cases on this side is so small, while in all the remainder, out of a total of more than 140, M followed K or W, not G. In favor of the second alternative (II) there are several facts, or sets of facts. The hypotheses (under b and c) that M^ used G in an uncor- rected state, and that G^, after correcting G upon M, made further corrections in the former, is in harmony with the facts in several passages. Thus in xi, G R M have 23 and 24 written as one verse, while G^ separates them. In li, G R M have Te in 4, while G^ carries it back to the end of 3. In the same poem, R has Ille michi and M Ille mihi, while G^ has Ille mi, with a space after the second word which would easily catch the attention of anyone who was examining a manuscript for corrections to his text. It is hard to believe that M^, if he had foxmd these sound corrections in G, would not have adopted any of them. Similarly G^'s correc- tion of the second nichil in xlii. 21 to nil, though G^ selected the wrong word, would have been likely to set either M or M^ to thinking. — These, it should be noticed, are all metrical corrections, and the first three are sound. The hypothesis that G^ used M is in harmony with one strik- ing fact in the metrical notes, which in all three manuscripts are by the second hands. In general, M'' agrees with R^ even to the varying spellings and misspellings, as Faulecij endecasillabi by both for V (G^ Faleuticus Endecasillabus) , Faleucij endecasil- labi by both for vi (G'' Faleuticus endecasillabus) , and FaleuciU endecasiW by R^ and FaleuciU endecasillabum by M^ for xiv (G^ Faleuticu endecasillabU) . In the long note for xi, M^ agrees with R^ while G^ has a somewhat differing text, though the same in substance. But in the note for ii, G^ and M^ have Genus metri [MeP G^) Faleuticum endecasillabum (with a period after metriinM), while R^ has Faleuticum endecasillibum {sic). A careful examination shows that four lines have been erased in the metrical note in M. There are sufficient traces left to make clear that what was first written was the entire metrical note (a long one) for the first poem. The explanation is simple. In R the The Manusceipts of Catullus 253 poem of Benvenuto ( Ad patriam, etc. ) comes first, so that the metrical note for i stands opposite the second poem. M^, working hurriedly, copied this note against the second poem of his MS, saw his mistake, noticed that the first four words would correspond in meaning to R^'s metrical note for the second poem, erased the rest, and then copied the long note over again, placing it against the first poem, where it belonged. G^ copied the note thus left by M^ for the second poem. If he did this, clearly he got his metrical notes from M, though varying them in form where he chose. No other hypothesis will account for the agreement of G^ and M^ in the one place where M* has blundered, and the agree- ment of K^ and M'' in detail elsewhere. Another set of facts remains to be considered. G and R always write e, not ae or oe, for these diphthongs. In 11 places, beginning at xlvi. 9, G^ has added a hook to an e, to represent the diphthongal spelling. R^ has the hook in 7 places, 4 alone being in the text. In none of these is there coincidence with G"s hook. Neither corrector, then, got his suggestion from the work of the other. In all the places where G^ writes a hook, M has the full spelling, with separated letters. This cannot be an accident, in view of the close relations independently determined. One worker must have followed the other. Did M follow G^ or did G' follow M? M's spelling fluctuates. The commonest form is the single letter. But in 66 cases he writes a diphthong with separated letters, joined letters, or e with a hook (6 cases only being of the last kind) . Of these, 6 occur before G^ begins to write the hook, 44 in the space covered by his hooks, and the remainder, 16, after he has stopped all corrections.' Clearly, then, M was not accepting suggestions from G^ but G'' was accepting suggestions from M, that is, G' was using M. 1 G^ let 6 cases of diphthongal spelling in M go by (4 of quae, 1 of haec, and 1 of pene), added the hook first at a word (coetus) where M's letters were large and strik- ing, and then kept up his corrections intermittently until he stopped his work com- pletely, confining them, however, to the words caelum and caelestis, coetus, coepit, and proelium, in various inflectional forme, and haec. He covers every case of coetus (5), caelum and caelestis (1 each), coepit (2), proelium (1), and 1 case of haec out of 11, accepting no case of quae, which occurs 18 times in the same space along with abundant instances of que). 254 William Gardner Hale Finally, a general survey of the treatment of the variants may throw light upon our question. In order to hold that M used Gr and R, as against holding that Gr^ owes the mass of his work to M, and the rest to his own fancy, we should be driven into the following combination of hypotheses : M put the work of R above that of G, and followed the former in all but the dozen cases given above (p. 249), out of more than 140. But he at the same time set a singularly high value upon the judgment of the corrector of G, which led him in several cases to adopt the reading of this man's variant, as against the reading of the scribe of R — in other words, he used the MS G for the sake of its corrections, not for its text. He was also able to distinguish between the variants by G^ and those by G' (though no modern scholar succeeded in doing this before Bonnet), and to avoid adopting the latter, as in the case of aV -g- iii. 14, where G' alone was right. He was able also to distinguish where G^ had made changes in the text of G, and so to follow his lead as against R, as in the cases of dindimenee and dindimenei in Ixiii. 13 and 91 (though even so late as the time of Schwabe's first edition, 1866, no one had seen that these readings were not original). Of the corrector of R, on the other hand, M thought so badly that, in the very large number of cases where R' and G'' had the same variant, he felt that the alliance destroyed the value of G"s opinion, and therefore as a rule did not adopt the latter's variant as the reading for his text. But, though so acute in detecting the changes made by G^ in the text of G, he failed to see that R^ had made many changes in the text of R, and thus was deceived into regularly following R^'s corrected readings where those of G difPered. This combination, even if the facts in detail agreed with it, is too fantastic for acceptance. But they do not. M does not follow G^'s variant everywhere, e. g., in Ixiii. 19, where the latter corrected cedit to cedat, and added aP cedit above. He did not, in fact, follow the variant aV his ( = experi- bis) in the one case, xxi. 6, that has any real weight to support the theory that M had the MS G before him as he wrote. He did not think badly of E^; for in several places, as in the case of The Manuscripts of Catullus 255 tndignis, Ixvi. 86, and fratri, Ixviii. 91, he adopted E,"s variant in his text — and, further, he himself occasionally copied, as a variant, a variant by R^, as in the case of aV ere citatis, Ixiii. 18. Further, the theory that M set a superstitiously high value upon the judgment of Gr^ is made difficult to hold by the fact that, in the places where M's reading corresponds to G'^'s variant, G^ is obviously wrong, as in Ixiv. 298, gnatisque OGR, aVgratis G^ gratisque M. And the entire hypothesis that M gets some of his readings of this kind from G"s variants is crippled by the fact that precisely the same combination of a blundering or wrong-headed reading and a variant giving the right reading of the tradition appears in M a number of times, as in xi. 7, septem geminus OGR, septem geminis M, aV nus M^ (a per- fect analogue for the experibus example), without a corresponding combination in G, and occurs 11 times after G^ stops his work. It is obvious that the opposite theory accounts vastly more easily for the state of affairs. M had gone astray in a number of places, and M^ had generally corrected the error by giving the read- ing of the archetype R in the form of a variant above. G^ when M came into his hands, finding so much that was good and new in it, and perhaps working provisionally and in a hurry, had too much confidence in the manuscript, and here and there added M's blundering reading to G in the form of a variant. His work was never completed (in the last quarter of G he neither wrote variants nor made corrections, except the hitherto unreported Vitas to Multas ci. 1, doubtless done when the paragraph marks were put in) ; and these foolish variants were in consequence never erased. The acceptance of this theory will also explain certain correc- tions, which otherwise are puzzling. Thus: xxiii. 19: Quod cuius OE, with aV Quod cuius in the margin by R^; Quod cuius G, corrected to Quod cuius by G^ with aV cuius written above by G^; Quod cuius M, with aV cuius written above by M^. xxviii. 14: nobis O; nobis aV uobis R and R''; nobis G, changed to uobis with aV nobis written above, by G^; uobis with aV nobis written above, M and M^. Ixiv. 309: uitte OGRM (in M with a short first i, which could be taken for c); uicte G'. 256 William G-ardneb Hale It has hitherto seemed strange that Q'\ instead of writing the corrected reading as a variant above the word in the two passages first mentioned, should have corrected the text and written the former reading above. But this procedure is at once intelligible if G^ was correcting from M, who had here inverted the readings of E and E,^. The same hypothesis explains the change in the third example. I am at present forced, then, to think that G was corrected, not upon its archetype a, nor even upon its sister MS R, but upon M, one of the daughters of R. Further light may come with further study ; but it seems improbable that the result will be changed. If it stands, then our critical editions will ultimately give the readings of OGR and R^ alone, the work of G" having no value for the text. I regret the loss of this aid, in spite of the fact that every- thing of importance by G" is also found in the work of R or R^ The University of Chicago