¹ Die Ueberlieferungsgeschichte von Ovids Carmina amatoria. Verfolgt bis zum 11. Jahrhundert (Tübingen, 1910), 26–32Google Scholar.

² P. Ovidi Nasonis Amores ed. Munari, F. (Florence, 1951), xv–xviGoogle Scholar; different pagination in later eds.

³ These have been drawn from the MSS themselves. For earlier descriptions see Tafel, 4–5, and Munari, 's collations in SIFC 23 (1948), 113–152Google Scholar [= Kl. Schriften (Berlin, 1980), 32–71Google Scholar]. I am indebted to Professor Kenney for access to materials used largely in the second part of the article and to Dr Denis O'Brien of the Centre National de la Recherche Scientifique for additional information. Throughout I have gratefully accepted suggestions made by Professors Brink, Kenney, and Reeve.

⁴ On these titles see Kenney, E. J., ‘The manuscript tradition of Ovid's Amores, Ars Amatoria, and Remedia Amoris’, CQ 12 (1962), 7 n. 3CrossRefGoogle Scholar. It will be convenient to refer henceforth to this article, which contains much essential information, by page number alone. Roman numerals refer to the same author's OCT preface.

⁵ Corbie is suggested for P: Bischoff, B., Mittelalterliche Studien i (Stuttgart, 1966), 59Google Scholar.

⁶ See, e.g., the summaries by Barsby, John, Ovid Amores I (Oxford, 1973), 33Google Scholar and Tarrant, R. J. in Reynolds, L. D. (ed.), Texts and Transmission (Oxford, 1983), 260 (Am.), 269 (Her.)Google Scholar.

⁷ Tafel, 31; Kenney, 7, n. 2.

⁸ The total remains at 12 if two of these lines (2.2.23–4) are not genuine; cf. the other suspect couplet supplied only by the β MSS at 1.13.33–4.

⁹ E.g. (31): ‘Die Annahme, dass P aus einem Teil von R abgeschrieben wäre, bleibt also möglich, aber sie ist nicht zu beweisen.’

¹⁰ Stemma on 7, General Conclusions on 24–9.

¹¹ Of Parce tuns in me perdere uictor opes, the tear has left, at the beginning, Par [and, at the end,] uictor opes. On the preceding side two letters and the tops of 9 or 10 others survived from line 20, also the top of the capital letter N from line 21.

¹² Tafel, 29; Munari, xx; Kenney, 6. There is no possibility (cf. p. 219, above) of allowing, as Kenney suggests, one line for a title, which would take us only to 1.3.1.

¹³ Pp. 3–4.

¹⁴ The case is in fact worse than this: A.A. and Rem. occupy together some 50 leaves of R; Am. and Her. each occupy some 50 leaves of P. So the portion for sale was much closer to ⅔ of the whole.

¹⁵ The inscription ‘Ex libro primo Amorum finis elegiae secunda’ is patently recent.

¹⁶ The chances that P set out to reproduce the appearance of these lines just as they survived in the torn (or recopied) portion of R seem too remote to take into account and would still do nothing to explain his placing of them at the very top of the page.

¹⁷ See, for example, Plate LXXXIV, where fol. 32^r of the 10th-century cod. ɸ of Horace (Par. Lat. 7974) ends with ‘Q. HORATII FLACCI CARMIN .LI .II./EX . INCIPIT.III PRAGMATICE.’ The other instances cover prose texts as well as verse.

¹⁸ On these errors, see further pp. 233–4. There is no need to suggest, as an alternative solution, that P, like R, omitted lines 1–2 of poem 1, which would reduce the amount of text before 1.2.51 to 82 lines and so leave two lines free for the book-heading at the top of the first page. The other MSS descended from the α parent, Y and S, both contain the couplet and, for what it is worth, R uses three lines rather than two for the heading. The only editor of the Amores to have made allowance for the loss of the first three pages in P is Munari, , SIFC 23 (1948), 115Google Scholar, = Kl. Schriften, 34; cf. edition, xvi n. 6. But no mention is made of the book-heading.

¹⁹ Tafel's fall-back position (p. 31), that R and P were each copied from the two parts of an exemplar divided, deliberately or accidentally, after 1.2.50 is untenable for all of the same reasons. The damage was in any case clearly sustained by R itself.

²⁰ It is in the end impossible to discover with sufficient precision exactly how many years may separate the two. It is worth recalling that the received 9th-century dating for R and 9/10th-century for P is only an approximation, designed perhaps to facilitate the derivation of P from the second part of R. Bischoff (above, n. 5) identifies P as one of the Corbie MSS associated with Hadoard and thus to be placed securely within the 9th century (at most not later than the third quarter). R and P would then appear to be closely contemporary descendants of the same source.

²¹ Cf. the remark (ix) ‘Recensionem habemus, ut Pasqualiano more loquar, apertam’.

²² Tafel, 31–2, Kenney, 7,24, placing Her. in hyparchetype α; Goold, 3–4, placing them also in an archetype, Ω.

²³ Note that Kenney (p. 7) allows for an intermediary (ρ) between R and a, to accommodate the stemtnatic requirements of A.A. and Rem. For the purposes of Amores, however, this MS and a come to one and the same thing and only α will be referred to as source of R and P.

²⁴ Advanced by Müller, Lucian, De Re Metrical ¹ (1861), 43ffGoogle Scholar.

²⁵ As it is, for example, in the extensive, if parlous, reconstruction of an early archetypal stage of the Ovidian corpus by Luck, Georg, Untersuchungen zur Textgeschichte Ovids (Heidelberg, 1969)Google Scholar.

²⁶ Dörrie, H., ‘Untersuchungen zur Überlieferungsgeschichte von Ovids Epistulae Heroidum Teil I, Nachr. Akad. Wiss. Gött., Phil.-Hist. Kl. 1960, 124ffGoogle Scholar.

²⁷ Gnomon 33 (1961), 479Google Scholar.

²⁸ Pp. 186–8.

²⁹ There is a similar loss of a double leaf in the next quaternio, this time of the two middle leaves, where 5.97–6.49 (112 lines) are missing between foil. 9 and 10. There are 111 lines of text missing together with an interstice of one line between poems, P's regular practice in Her.

³⁰ Even if a smaller gathering of 4 leaves (Am. and Her. occupying different sides of the same leaf) or a gap of 3 complete leaves between the two works is posited (which might in itself suggest a separate source), it is difficult to see how the first 18 lines of Her. came to begin some way down on the verso of a leaf. Another work to fill this gap (if not also one or more complete gatherings in between) has to be postulated (below, with n. 34).

³¹ A conclusion happily congruent with identification of P as one of the classical texts assembled by Hadoard at Corbie (above, n. 20).

³² Y, it should be noted, is also mutilated at the end, having lost its final two leaves containing the last 4 lines of Am., which were later replaced in the 12th century. Munari, , Il Codice Hamilton 471 di Ovidio (Rome, 1965), 12 n. 4Google Scholar, while preferring to believe that fol. 70 and the remainder of fol. 69 were blank (hence, most probably, the reason for their loss), allows for the possibility that another work may have followed, but this, together with Oliver, R. P.'s suggestion, Studies Presented to Ben Edwin Perry (Illinois, 1969), 163Google Scholar, that the work was Her., now becomes less likely.

³³ Dörrie, 187.

³⁴ In case Ovid's ‘paruus libellus’, the Medicamina Faciei Femineae, should be suggested for inclusion between Am. and Her. on the specious grounds that a appears to have contained 25 lines to the page (Kenney, 24) and that 100 lines, or one bifolium of a, are all that survive of the work, it should be said that (1) the tradition of the work when it surfaces in MSS of the 11th century and later bears no relation to our own, and (2) that this is to divorce the work from the spurious Nux, by which it is almost ubiquitously accompanied.

³⁵ In these, as can be seen from Kenney's list (3–6), Her. occur in conjunction with various works by Ovid in widely differing orders. The order A.A., Rem., Am., though it occurs in 3 MSS and one florilegium (F, P_a, W;, P₂), seems by that time to be quite random and retains nothing sacrosanct about it.

³⁶ Above, p. 220; Kenney, 9.

³⁷ The 14/15th-century MSS are too interpolated to have any textual significance.

³⁸ The Classical Text (Berkeley, 1974), 134Google Scholar.

³⁹ PCL, pl. XCI.

⁴⁰ Which much not be taken to preclude the possibility of double readings higher up in the tradition (below, p. 237).

⁴¹ For which evidence emerges below, n. 53.

⁴² Munari, edition, xvi; Kenney, 8.

⁴³ Reference given above, n. 32.

⁴⁴ MSS P_a P_b P_r (Explicit only). O_b, X, Z.

⁴⁵ A_b,P_b, V_a, W.

⁴⁶ H: ‘Incipit ille liber cuius non nomen habetur’.

⁴⁷ ‘Incipit ouidius sine titulo’ F, ‘incipit ouidius amorum’ F².

⁴⁸ N (ut uid.): ‘Liber ouidii amorum siue de sine titulo’; T: ‘Liber ouidii sine titulo qui uocatur amorum’, conflations of correct with incorrect title redolent of the 14/15th-century interpolated MSS.

⁴⁹ NI for M is a constant misreading in capital script. The scribe thought he saw ANIORVM and adjusted accordingly. False reading or an abbrevation would not explain the error.

⁵⁰ Fol. 75^r: ‘P. OVIDI NASONIS ARTISSIMATORIAE/LIBER II EXPLICIT.INCIPIT LIBER III.’

⁵¹ See further n. 56.

⁵² Deciphered by Lenz, for his collation of S in Rend. Ist. Lombard. 69 (1936), 635Google Scholar.

⁵³ These notes affect only the first six pages in S and may explain why the lines omitted in α were not supplied in S when so many other β readings were (above, p. 232). The marginal additions come to an end too soon to include these lines, and suggest that, as with the title and initial letters, the scribe never succeeded in completing his MS.

⁵⁴ E.g. ‘Ouidius…ait in ouidio sine titulo’: cod. Bern. 411, a 12/13th-century accessus to the Metamorphoses. See Alton, E. H., ‘The mediaeval commentators on Ovid's Fasti’, Hermathena 44 (1926), 121Google Scholar; cf. also ibid. 95 (1961), 74.

⁵⁵ Cordoba, Arch. de la Cat. 150, 13/14th century (Rubio no. 43).

⁵⁶ titulus Y and β MSS, titulo F¹, titulos R. The α MSS made nonsense of the beginning of the line: deue cerem R deueterum Y, a corruption which reduces still further any chance (above p. 234) that a was in a position to recover the correct title for himself.

⁵⁷ Though it is perhaps important that he leaves three lines unfilled at the division between books 1 and 2 and four lines at the division between 3 and 4.

⁵⁸ The same seems likely also to have happened in the Matritensis (M) of Manilius, where the lack of a book-heading among its descendants indicates the absence of one in M itself, now missing its first leaf. Quite possibly Poggio's scribe left the heading for insertion later in red or by a hand more accomplished than his own.

⁵⁹ Ghisalberti, F., Memorie dell'Istituto Lombardo 24 (1932), 166Google Scholar.

⁶⁰ The very close relationship proposed by Oliver (152), that β represents the parent of a mutilated in some way after α had been copied from it, is ruled out by the explanation given above for the loss of β's titles. As mutilation does not appear to be involved, β demands at least one separate copying of a parent it holds in common with α. Nor does the Somnium, transmitted as 3.5 by α, but in various positions (or omitted) by the β MSS, shed any light on the matter. Even if it could be shown to have been in the archetype and, if so, in what position, this would not help to date the divergence of α and β.

⁶¹ No classes among the β MSS can be established by these divisions; as usual, contamination has made it possible for us only to classify readings, not the MSS themselves.

⁶² It is not totally impossible that scribes on the β side would draw correct readings from α MSS and still prefer not to add to or alter their own ‘Sine titulo’, but this remains much less likely than that a composite title would already have emerged, as it did in later MSS.