¹ For convenience, and except where otherwise indicated, we cite the text and sigla of Mynors' OCT of Catullus (corrected reprint, 1960) in this article, though we often modify or amplify his apparatus. The following editions of Catullus are referred to by author's name only: Ellis, Robinson, Catulli Veronensis liber (Oxford, 1878)Google Scholar, Fordyce, C. J., Catullus (Oxford, 1961)Google Scholar, Goold, G. P., Catullus² (London 1989)Google Scholar, Kroll, W., Catull³ (Stuttgart 1959)Google Scholar, Lee, Guy, The poems of Catullus (Oxford 1990)Google Scholar, Syndikus, H-P., Catull (3 vols., Darmstadt 1984–1990)Google Scholar, Thomson, D. F. S., Catullus (Toronto 1997)Google Scholar.

² ‘Notes on the text and interpretation of Catullus’, PCPS 24 (1978) 92–115Google Scholar = Collected papers on Latin literature (Oxford 1995) 76–100Google Scholar. We would be inclined to accept the following conjectures: 10.25–6 quaero … istaec; 22.6 nouae bibli; 64.359 densis; 66.74 imi. Less successful, perhaps, are the suggestions made at 11.13, 17.3, 61.111, 69.9. All the other discussions are worth close consideration; and some of the passages will be revisited in what follows, a homonymous homage to that ground-breaking article.

³ Zicàri, M., Scritti catulliani (Urbino 1976) 160–79Google Scholar.

⁴ One quaestor per (pro)praetor was the norm; but the (pro)praetor of Sicily, a province with two major centres, had two (Cicero, Ver. 2.2.11), and it may be that the double province of Bithynia/Pontus had two when established by Pompey in 63–2 B.C., for which the evidence is very scanty; for what can be known cf. Marek, C., Stadt, Ära und Territorium in Pontus–Bithynia und Nord-Galatia (Tübingen 1993), 34–47Google Scholar.

⁵ mihi nec ipsi was already conjectured by Statius.

⁶ Here, as more generally, I owe a debt to Dr David McKie, with whom I discussed the passage as an undergraduate.

⁷ Prof. Kenney suggests miser for die, also attractive.

⁸ But note that primum is transmitted in line 17.

⁹ Catullus is one of the few poets to use ei; but on the one occasion where the MSS have the form (at 82.3) it is scanned as a single long syllable. The 9 instances in Lucretius are all spondaic, in the final foot of the hexameter; and it is not until the pseudo-Ovidian Halieutica (v. 34), and Germanicus, (Arat. 333, 457)Google Scholar that we find secure examples of iambic scansion. See Axelson, B., Unpoetische Wörter (Lund 1945), 70Google Scholar; Richmond, J. A., The Halieutica Attributed to Ovid (London, 1962), 42Google Scholar, TLL 7.2.1.457.35-62.

¹⁰ comestur was already conjectured by Wiman.

¹¹ Or ‘has increased’, if we read auxit (see above).

¹² Rossberg's nulli at 8.14 gives us another instance of dative of the agent with a finite verb in Catullus (note 7.9 for confusion between ei, an alternative spelling for long i, and a in the tradition). None of the supposed parallels for nulla stand up; at 17.20, for example, it functions as a complement (= ‘non-existent’).

¹³ JPh 22 (1894) 88Google ScholarPubMed = Classical Papers (Cambridge 1972) 317Google ScholarPubMed. His rejection of the earlier correction haec (Itali) on the grounds that the Annales are still addressed, I consider valid.

¹⁴ As Prof. Jocelyn points out.

¹⁵ He supports this view by a cross-reference to 21.11, where he improbably reads a te mi puer for the obviously corrupt me me puer of the MSS; Goold more persuasively reads Froelich's a temet puer. Kroll cites Terence, Andr. 596 corrigi mihi gnatum porro enitere for the possessive interpretation of mi, but there the dative is at least partly that of advantage.

¹⁶ See Colin Macleod's brief but important analysis (n. 25 below).

¹⁷ The others are Cat. 99.6 (where one might think of reading e.g. taetrae; tibi already supplies a second person), Ov. Am. 2.16.24, Sen., HO 1513Google ScholarPubMed.

¹⁸ The role of the intermediaries also reflects the practice of Roman divorce – see Mayer, R. G., CQ 33 (1983) 297–8CrossRefGoogle Scholar.

¹⁹ Goold eases this by transposing 16–17 after line 23.

²⁰ For Iupanar of a person (the collective intensification of lupa) cf. Apuleius, Apologia 74.6.

²¹ Gratwick argues for ‘how’ as a possible alternative here, as elsewhere; but such a marker of intensity is less well suited to sneezing than to the ut perii of Ecl. 8.41.

²² Gratwick refers to earlier discussions by Wackernagel, J., Glotta 14 (1925) 64–6Google Scholar, A. S. F. Gow on Theocritus 2.82 and Timpanaro, S., Contributi di filologia e di storia della lingua latina (Rome, 1978), 219–88Google Scholar.

²³ For the close association of amor/Amor and amo, cf. 64.335 nullus amor tali coniunxit foedere amantes, 40.7–8 quandoquidem meos amores ∣ cum Ionga uoluisti amare poena, Propertius 1.2.8 Amor non amat.

²⁴ There seems little doubt that this is Licinius Calvus, given the closeness implied by poems 14 and 53, and the fact that Calvus is a poet writing in a variety of metres like Licinius here.

²⁵ Macleod, C., ‘Parody and personalities in Catullus’, CQ 23 (1973) 294–303CrossRefGoogle Scholar, at 294 = Collected Essays (Oxford 1983) 171Google ScholarPubMed. He, like others, rightly draws attention to the erotic tinge to much of the poem's diction.

²⁶ The decasyllable is simply the hendecasyllable with the central choriamb (– ∪ ∪ –) replaced by a molossus (– – –).

²⁷ The integration of the two pieces is tempting, given that they share the same metrical peculiarity, a theme and an addressee; but Catullus often writes more than one poem on the same theme, and it may be that 58b is simply a fragment of another poem (perhaps unfinished: cf. 60, the essence of which got turned into 64.153–7).

²⁸ Goold, alert as ever to abnormalities, is an exception, as has been said.

²⁹ Harrison's conjecture; see next note, as also for the text of v. 11.

³⁰ The vulgate text follows the MSS in reading tanto te in fastu; but in, which produces a far less idiomatic phrase than the simple ablative, was rightly ejected by Schuster, and the frequency of the collocation te in in verses 3–5 makes the corruption easy to comprehend.

³¹ I.e. the plural of the imperative cedo (‘hand over’).

³² en can be used thus with imperatives other than those of verbs of perception, provided that an accompanying action is referred to (here prendi): cf. Vergil, , Ecl. 6.69 en accipe, Georg. 3.42 en age, Silius 4.281 disce en nunc, TLL V 547.26ffGoogle Scholar.

³³ I differ from my co-author in regarding it as neither possible nor desirable to achieve alternation of decasylables and hendecasyllables in this poem.

³⁴ Nisbet here conjectures uariatum for the difficult uelatum, generally doubted by editors (Collected Papers 386, not in the original article).

³⁵ See R. D. Williams (Oxford 1960) ad loc. for other instances. The example here clearly does not fit into the idiomatic group given by OLD 2 (‘never mind! very well!’).

³⁶ BICS 16 (1969) 40Google Scholar.

³⁷ Or his alternative operta; but opaca occurs in verses 3 and 32, and this is a poem in which repetition of diction seems more aimed at than avoided – see Fordyce's introduction to this poem.

³⁸ I am also inclined to get rid of the superfluous et in 60, so that patria, bonis, amicis, genitoribus abero will be followed by abero foro, palaestra, stadio, gyminasiis, not et gyminasiis.

³⁹ The last to mention iuvenis was apparently Palmer, A., Catulli Veronensis liber (London 1896) p. xlGoogle Scholar.

⁴⁰ PCPS 24 (1978) 100Google ScholarPubMed = Collected papers 86.

⁴¹ See Bentley on Horace. Carm. 3.11.18, Axelson, B., Unpoetische Wörter (Lund 1945) 72Google Scholar.

⁴² Bentley contrasts functional usages such as that at Prop. 4.6.67–8 Actius hinc traxit Phoebus monumenta, quod eius ∣ una decem uicit missa sagitta rates.

⁴³ olim is used by Catullus in the same sedes at 67.47 and as a marker of the previous generation at 67.4.

⁴⁴ For such shortenings see Hartenberger, R., De O finali apud poetas latinos ab Ennio usque ad Iuvenalem (Diss., Bonn 1911)Google Scholar.

⁴⁵ Wölfflin, E., ALL 14 (1906)Google Scholar made this conjecture in the modern era, but it had already been made by Markland, as reported earlier by Palmer, A., ‘Unedited conjectures of Markland’, Hermathena 4 (1883) 154Google Scholar; we are grateful to Prof. Kenney for his kind help in tracking down this reference.

⁴⁶ Thomson claims that ‘the elision of a long o at the end of a dactyl in the first foot would be awkward’, failing to observe that this is precisely what is offered in verse 2 by the text he prints (as well as at 77.4).

⁴⁷ Beyond the editions, note especially the article by Biondi, G. G., L&S 11 (1976) 409–25Google Scholar.

⁴⁸ For the sentiment compare Propertius 4.7.93–4 (spoken by Cynthia's shade to the poet) nunc te possideant aliae, mox sola tenebo: ∣ mecum eris et mixtis ossibus ossa teram.

⁴⁹ The central parts of the argument are as follows. The fifty lines of poems 72 to 76 were collected together in the lost archetype under the heading Ad Lesbiam. The limits of the poems within this collection are for the editor to judge. The canonical 76 is an anomaly, an elegy of 26 verses amidst a host of epigrams. Of the other poems between 69 and 116 only 99 exceeds 12 verses, and that extends to a mere 16. The final ten verses of 76 are a separate and complete poem, a prayer for relief from sickness. The gods, addressed in the opening words, are returned to in the final line. Verses 1–16 are also complete in themselves. There is overlap of theme and diction between 76a and 76b as there is between other epigrams, including adjacent epigrams, but that does not force us to consider the lines a unity.

⁵⁰ We would like to acknowledge much helpful discussion of these problems with the fifty or so graduate students, visiting students and final-year undergraduates who have attended our joint class on the textual criticism of Catullus in Oxford in the period 1993–1998. Above all, this article is the fruit of a stimulating lustrum of collaboration between the two of us; though conjectures and discussions are here assigned to us individually, many of the notes have been jointly drafted, and the thoughts of each of us on almost all these passages have been much influenced by the other.