Bryn Mawr Classical Review

This is a learned, acute and thought-provoking book of an unusual kind. In essence it is a new edition of Catullus: after 20 introductory pages discussing recurring issues and principles, several hundred deviations from Mynors' OCT are listed on pages 21-32. For all the qualities of the apparatus (in particular the innovative work done on recording humanistic conjectures), Mynors' text was a poor one, conserving many corrupt and improbable readings. The best text currently available is probably Goold's, and a good proportion of his changes are accepted here; but many more are advocated (including the occasional return to a MS reading). The arguments for all deviations from Mynors are given on pages 33-301, in a commentary of a strictly textual kind. In each case, Trappes-Lomax cites some lines of text, as published by Mynors, and then goes through the individual problems before printing his preferred solution. There follows an admirably full list of works referred to.

Some wise observations are made about textual judgements, such as this on the limitations of manuscript authority (p. 19): "at 64.80 V's angusta is not prima facie evidence that Catullus wrote angusta rather than augusta; it is merely prima facie evidence that Catullus wrote either angusta or augusta". Similarly on 76A.11, he draws attention to the low evidential value of an apparent parallel when the anomaly could easily be removed in that case too. However, when he claims (117) that ocio in O at 44.15 could as easily stand for an original ocimo ("basil": Hermolaus Barbarus's attractive conjecture, revived by Thomson) as for an original otio, he overstates the case: ocio is simply one way in which scribes write otio.

A particular strength is the large number of excellent but forgotten conjectures that are revived, e.g.: 6.8 oliuo] amomo man. rec. in B.L. Add. 11915, odore Peiper; 6.9 et hic et illic recc., Vossius; 9.4 bonamque Palmer; 10.27 minime Pontanus; 10.33 salsa Schrader; 13.14 ut del. Fruterius; 25.10 natesque Scaliger; 26.4 ducenta recc., Gronovius (ducena is perhaps more idiomatic); 29.19 qua nitescit aurifer Tagus van Jever; 37.8 unum Pleitner; 40.1 Auite Fröhlich; 44.13 hinc Bergk; 51.11 gelida Baehrens; 56.5 in puella Schwabe; 61.25 umore] alimento Postgate; 61.122 uidens Schwabe; 62.54 maritae Heinsius; 64.94 miseros agitans Markland (rightly preferred by Trappes-Lomax to his earlier deletion); 64.103 frustra] superis Fea; post 64.121 lacunam Laetus; 64.187 intentant Orelli; 64.212 classi] castae recc.; 64.275 purpureaeque Lennep + nascente Baehrens; 64.283 interstinctis Heinsius; 64.350 lacerum Garrod; 67.33-4 del. Maffei; 68.2 conspersum Schrader, hoc del. Muretus; 69.10 cum Fröhlich: 84.2 hinsidias Arriu' si insidias; 108.1 populi arbitrio, Comini Fröhlich; 110.4 saepe] turpe Auantius.

There are a striking number of compelling new conjectures (a few published in previous articles, others suggested by Robin Nisbet): 17.3 insulis; 38.7-8 treated as a separate fragment; 52.2 Nonius (already referred to at Plin. Nat. 37.81)] non homo; 63.82 rugienti; 64.246-7 funesta domus ingressus morte paterna/ tecta (for tecta ... morte); 64.372 auidi Nisbet; 66.63 a deleted; 67.23 ipse sui (also suggested by Norbert Lain in a paper delivered at the APA meeting in 2002); 84.5-6 deleted; 90.6 flammam; 109.1 proponit; 110.2 quae] tum; 110.3 mentita es amicam Nisbet. Also attractive are the following (some of them simple improvements, others diagnostic): 8.5 nobis] tantum (as in the repetition at 37.12); 50.19 respuas; 52.1, 4 qui moraris; 62.60 quia; 63.9 deleted; 64.37 deleted; 64.58 iuuenis pellit uada <caerula> remis; 64.120 dulcem] tum; 64.122 aut ut <securo> deuinctam lumina somno; 64.149 medio uersantem in turbine; 64.183 curuans in; 79.4 amicorum; 82.3 ei] id; 83.6 prurit et obloquitur; 114.3 omne genus] siluas. But almost everything advocated here will need thought from future editors before they could decide not to include it in an apparatus.

For all the book's strengths, there are weaknesses. The Introduction speaks with some disdain of the apparatus criticus (p. 3) and the repertory of conjectures (p. 6), though the one is vital for exposing how many and what kind of errors scribes make, and the other is now perhaps the greatest need in Catullian textual criticism. The very first page proclaims the criterion si melius est, Catullianum est; and a little later "choosing the better reading should at least lead us nearer to the truth." This gloss is misguided: in a case where the choice is between emendation of a verse and deletion, either may be better (i.e. more probably right) than the transmitted text, but if we make the wrong choice (as we sometimes will) we are moving further from the truth. Even the initial formulation is problematic, partly (a) because the criteria for judging quality still need to be determined, and partly (b) because it leads Trappes-Lomax on to a series of binary discussions.

(a) Some of Trappes-Lomax's judgements are based on strangely general criteria: discussion of 63.16, for example, begins from the unqualified assertion "Catullus is a poet of balance and antithesis", 63.50 with "the alternation between mei and mea is inartistic", and on p. 88 we find the iambic poem 29 described as "exquisite lyrics". Again, the proposal to transpose 68.57-62 (with 59-60 deleted) after 66 is dependent on a view of Catullus as always desiring clarity. To me the transmitted order is superior precisely because of its complexity, in this poem full of complex and surprising similes: at first the image of the mountain stream describes Catullus's tears; it is only when we return to the context that we find its main purpose is to evoke the poet's grateful relief at the assistance given by Allius.

(b) The notes often argue for one reading without discussing alternatives, or even mentioning them. All of the following deserved acknowledgement or counter-argument: 8.14 nulli Rossberg; 9.1 e] o Baehrens; 10.17 beatiorum Auantius: 29.21 Lamum Slater (CR 26 (1912), 206-7; CQ 7 (1913), 127-8); 61.122 domino ed. Cantabrig. 1702; 61.179 bonis Passerat; 62.6 consurgere recc.; 62.56 innupta Weber; 63.51 misera Fröhlich; 66.15 an quod auentum Munro; 68.145 tacita Lain (HSCPh 90 (1986), 155-8); 73.4 <taedet,> taedet Auantius; 73.6 quae Birt; 76B.21 ei mihi Lachmann; 95.9 sunt recc.; 96.4 iunctas Baehrens; 109.1 continuum Camps (AJPh 101 (1980), 442). See also below, on 27.7.

Moreover, many of the notes seem to exist in isolation from the broader context, even of Catullus's corpus. For example, the note on 61.76 ignores the allusion to Callimachus (see Fedeli (Amsterdam, 1983) ad loc.; and MD 33 (1994), 51-3); that on 64.18 does not mention R.L. Hunter, " 'Breast is Best': Catullus 64.18", CQ 41 (1991), 254-5; and the case for deleting 37.5 is based on the observation that hirci are sexual animals and cannot properly be contrasted with those who have mentulae; but Catullus makes an explicit link between billy-goats and sexually unattractive men in poem 69 (and, less directly, 71). The point is not in itself decisive against the deletion (putare dependent on putatis is odd, and nothing else in the context suggests smell is an issue), but it is disappointing not to find it brought into the debate, especially as Hermann's putere (printed by Lee) solves both these additional problems. The same note also shows the extent to which the book is conceived as relevant only for the specific passages it discusses: there is a short collection of ellipses of verbs of sexual intercourse, but as the volume has no index, and the ellipse here depends on the deletion, the list will be missed by all bar the most careful readers, as will the conjectures made in passing on other authors: e.g. Terence, Eunuchus 217 offirmari (p. 251); Ovid, Ibis 224 uaga (p. 271); Propertius 3.1.35 meme (p. 282).

Particular interests of the volume are orthography, deletion of suspected interpolations, and the restoration of lost examples of ecthlipsis and prodelision, and I shall deal with each of these before running through the text with responses to a selection of the notes.

Arguments are presented for the restoration of a number of Greek forms and spellings, e.g. 4.1, 17 phaselos; 4.8 Rhodon, and Thrascia (J.A.K. Thomson); 10.26 Sarapin; 12.13 mnemosynon. Though these are guesses, they are based on clear routes of corruption we can spot in MSS. For Latinate orthography, things are not so clear: there seem to be archaizing tendencies when scribes write certain republican authors, and we cannot be confident that details go back to the author himself: it is not necessarily significant for the poet's own practice that MSS have male si for male( e)st, e.g. But there is a further point: though Catullus may have written loedere, not ludere, such forms are unlikely to have died out 30 years later, and to choose one orthography for Catullus and another for the Augustans risks making his style seem more alien from theirs than is valid. In a world where knowledge of Latin is not increasing, fiddling with unorthodox spelling seems more likely to irritate readers than to educate them.

Trappes-Lomax is keen to delete any unnecessary lines, if they exhibit unusual metrical form (especially 63, which has 7 lines removed) or contain any awkward phrasing. So, in poem 32 verse 4 et si iusseris, illud adiuuato is expelled, but not verse 2 or 9. This may be right, but (as transmitted) a lot of the power of the poem comes from the double contrast between the careful courtesy of verses 4 and 9 and the desperate, basic sexuality of 8 (nouem continuas fututiones) and 11 (pertundo tunicamque palliumque). Despite Trappes-Lomax's scepticism, we can translate 4-5 "if you do give the order, please help as follows: let no one bar the panel on the threshold, etc." Some of the deleted lines contain material unlikely to be interpolated, e.g. 6.12, where the lightly corrupted nil ..., nihil with two verbs is a classic Catullian combination (17.21, 42.21, 64.146). In 36, he argues for removal of verses 9-10, 15, 17, and draws attention to the "neat balance of 8 + 8 lines" that results (p. 103); but as transmitted the poem has an equally neat balance of 10 + 10. The removal of verses 9-10 is based on two grounds: (i) that they are generally acknowledged to be corrupt (but they can hardly have been written in this form by anyone); (ii) pessima puella in 9 repeats the epithet from pessimi poetae (6): but this is surely the point -- Catullus throws the demeaning word applied previously to him back at the girl, but in a context that softens the sense to "naughty", and immediately pays apparent compliments too with the adverbs iocose and lepide. These are then artfully matched by si non illepidum neque inuenustum est of his own behaviour in 17. In poem 21 Trappes-Lomax deletes both 2-3 and 13, but in neither case explains why the verses have been added. If someone wanting to embellish the lines drew on 24.2-3 why did he not simply write non harum modo, sed quot aut fuerunt/ aut posthac aliis erunt in annis? The similarity of that to 49.2-3 shows Catullus's willingness to repeat such hyperboles, and in the polymetrics, once a vocative takes up the opening line, it is more his style to prolong it further than "to get down to the main point of the poem immediately": against 27.1, 30.1, 36.1 we may set (besides 24 and 49) 2.1-8, 3.1-2, 9.1-2, 11.1-14, 17.1-4, 23.1-4, 25.1-5, 28.1-3, 31.1-3, 32.1-2, 37.1-2, 43.1-5, 44.1-5, 47.1-2. Above all the similarity between 21 and 24 looks pointed: it helps establish the identity in 21 of Juventius and in 24 of Aurelius, each carefully unnamed in those poems. As for 21.12-13, the repetitiveness is pointed, and leaves the poem (just like 16) with a closing obscenity: quare desine dum licet pudico,/ ne finem facias, sed irrumatus, "So cease while you can and still keep your chastity, lest you finish, but after you're buggered" [not of course the literal sense of irrumatus]. Even more implausible is the removal of 22.6-8: a positive account of Suffenus's writing or publishing practices is needed after nec sic ut fit in palimpsesto(n). For myself I have no difficulty supplying illi sunt with what follows ("but they were not, as happens, written down on a palimpsest; <he has> imperial sheets of new papyrus .."), but if Trappes-Lomax is right to be suspicious, something else must have been replaced by the interpolated lines.

One repeated source of change is the restoration of lost examples of ecthlipsis and prodelision. Discussing 39.9, Trappes-Lomax asserts "Catullus would certainly have pronounced monendus es as a trisyllable", and refers to p. 8. But on p. 8, the discussion leaps from the presence of ecthlipsis in the text to the expectation (nothing more) that Catullus would have also used the forms amatust etc. that we find in Plautus. If we are to give weight to the comments by the 4th-century grammarian Victorinus about the pronunciation of audiendus est as audiendust (GLK 6.22), it will affect far later poets than Catullus. Conceivably, Trappes-Lomax is right: but the case needs to be argued far more carefully than it is here if we are to allow if to affect our texts. However, I am now persuaded that he is right to add a number of others to the single Catullian ecthlipsis generally accepted (116.8: an obvious echo of Enn. Ann. 95 Sk.): Muretus's ultu' peccatum at 44.17; Trappes-Lomax's own omnibu' at 54.4; Fröhlich's factu' maritu' at 67.6, Brunér's maximu' multo at 115.7; and I would add Forsyth's euitabimu' missa at 116.7. The general case is well presented on pages 6-8, including the implicit conclusion that Catullus admitted the licence in poems of an invective nature and not in those of a higher stylistic level.

The production is good, but there are a few slips: p. 38, read loedere in the text advocated for 2.9 [so on 17.1]; on 64.237, final line, read "elision" not "hiatus" (and the reference should be to p. 10); p. 245, lines 12 and 13 manent should be bona; p. 272 something has gone wrong to produce "Alexandrian Antimachus". Contrasting "however" is often not marked off with commas. In the review copy there is a flaw in the printing of p. 59, and two lines of the note on 11.6 are not legible.

All in all this is a fundamental contribution to the textual criticism of Catullus, at times overpowering in the confidence of the author's assertions, but asking fundamental questions and regularly producing persuasive answers. The following notes on particular passages attempt to build on this work.

1.8-10 quare habe tibi quidquid hoc libelli/ qualecumque; quod, <o> patrona uirgo,/ plus uno maneat perenne saeclo: Trappes-Lomax replaces the transmitted patrona (seen as a Christianizing error) with Thaleia; but he does not attempt to show that Thaleia uirgo is a plausible combination. Attractive though the introduction is of the most frivolous of Muses, this is not likely to be correct: the only parallel I have found for vocative proper name with uirgo in apposition is Ovid, Metamorphoses 14.570 Lauinia uirgo, and that is a narrative address that picks up on Vergilian use of the phrase (Aen. 7.72, 11.479): the fact that Lavinia is a marriageable girl is at the heart of the story, whereas that aspect of Thalia's identity is quite extraneous here. The patrocini ergo of Gratwick (CQ 52 (2002), 305-20) remains the conjecture to beat. (I don't understand the reference on p. 34 to Goold's article "O patrona uirgo" in J.A.S. Evans (ed.), Polis and Imperium: Studies in Honour of Edward Togo Salmon (Toronto, 1974), 253-64: this argues for qualecumque quidem, patroni ut ergo.)

In keeping quod, <o> Trappes-Lomax also fails to deal with the metrical problem: Catullus does not elsewhere have as strong a pause between the short elements of the choriamb as that supposed here, and the need to make an alteration immediately after quod makes the vulgate text doubly suspect. qualest cumque does not help lighten the break. Moreover, though Ovid's use of the tmesis at Ex Pont. 3.4.6 (quale tamen cumque est) and 4.13.6 (qualis enim cumque est) may look allusive (in passages where he evokes Catullus's mock modesty), Ovid is under metrical constraints that do not apply to Catullus in a poem where he has two other opening trochees. Catullus does not elsewhere separate cumque. Perhaps read qualecumque quidem est.

3.10: Trappes-Lomax follows Birt in advocating titiabat for pipiabat, on the basis of Suet. fr. 161 passerum est titiare. The conjecture should certainly be in the apparatus, but given the multiplicity of onomatopoeic words it seems misguided to remove pipiabat from the text: should we replace every sonic "peep" with "cheep"?

6.16-17 uolo te ac tuos amores/ ad caelum lepido uocare uersu: Trappes-Lomax follows Nisbet (who "disposes of uocare"): "Catullus cannot 'call' anybody to the sky unless he is already there himself." In the first place, one might wonder whether Catullus sees himself as raised to heaven by his own poetic love affair; more seriously, the statement about the use of uocare is simply not true: cf. Mucianus speaking to Vespasian at Tacitus, Hist. 2.76.2 ego te, Vespasiane, ad imperium uoco. According to Nisbet's rule, only an emperor could say this. (Similarly B. Krostenko, Cicero, Catullus, and the language of social performance, (Chicago, 2001), 250-1, esp. n. 38.)

8.6: Trappes-Lomax follows G. Luck, Latomus 25 (1966), 280, and E. Fraenkel, JRS 51 (1961), 51 n. 20, in preferring tum (R mg) to cum, on the grounds that Catullus "never postpones cum to fourth place"; but he does at 13.13 have it in third (as it can be here); and at 64.80, which Fraenkel excluded as not "comparable", it does in fact stand fourth. And in his lecture handouts Oliver Lyne countered by pointing out that tum elsewhere in Catullus occurs in first, second or third position, except at 64.56 (utpote fallaci quae tum) and 68.131 (aut paulo cui tum), where the tum rides along with the delayed relative. cum also produces an effective chiasmus for the sequence 3-8.

8.15: manere + dative, though rare, is not as difficult as Trappes-Lomax implies (see ThLL s.v. manere 288.64-8, and Heyworth, Cynthia (Oxford, 2007), 195-6 on the sense), and given the presence of uae + accusative in Plautus and the Apocolocyntosis there seems little reason for denying it to Catullus. Moreover, "the next six questions all begin with the interrogative pronoun", as he notes, and they are all short and simple, so it seems misguided to interrupt the question by writing uae tibi and reordering.

17.23: Trappes-Lomax argues for Palmer's hunc uelis uolo, where GR have the intelligible nunc eum uolo de tuo ponte mittere pronum and O the manifest dittography nunc uolo uolo. It is true that Catullus does not intend to do the throwing himself, but as he does expect his words will have the effect of casting the husband from the bridge, there seems no problem in his saying uolo ... mittere. Trappes-Lomax does not illustrate the combination uelis uolo, and I have not been able to find a parallel; nor is there any special attraction to the sequence of corruption uelis uolo to uolo uolo [A, O] to uolo [X] to eum uolo [GR]. O has errors involving assimilation at 6.15, 31.13.

21.9: The Aldine's atqui needs the support of other elided occurrences: I have not found any in verse after Terence.

27.2: Verses 5-7 begin with a change of direction (at) before bidding farewell to water. This is rather awkward if the opening lines have already instructed the attendant to pour out wines with less water, meraciores, as Trappes-Lomax wants (with Sabellicus). The agreement of Gellius 6.20.6 with OGR gives us good warrant for preserving amariores, even if we cannot be quite sure what its force was for Catullus (probably "drier"; see H. Wilson, Wine & words in Classical Antiquity and the Middle Ages (London, 2003), 202-3; but with the programmatic implication "more bitter" (T.P. Wiseman, Catullan Questions (Leicester, 1969), 7-8), which Trappes-Lomax follows Thomson in denying).

27.7: Trappes-Lomax argues for Pleitner's sit; he seems to be so excited by the possibility of introducing another ecthlipsis that he fails to ask why the Thyonian wine has suddenly become masculine. Goold's "this man is an unqualified devotee of Bacchus" makes admirable sense in context, and plays neatly on merum: a shame to have no account of why this is rejected, nor the conjectures that preserve the iambic metre in 29.20, such as Munro's et huicne Gallia et metet Britannia? (printed by Lee): the discussion here starts from timetur as if that were the transmitted text.

28.2 Trappes-Lomax may be right to accept apti (recc.) sarcinulis et expediti (Wakefield); but there are parallels for aptis (Ov. ep. 4.24 sarcina ... non sedet apta) and expeditis (Petronius 99.4 expedite sarcinulas).

30.11: It would be easier to accept the old conjecture at di meminere, at meminit Fides, if we were given examples of the elision of the -ere ending (likewise at 64.383, on which see below): there are none transmitted in Catullus.

31.12-14: Trappes-Lomax wishes to construct these lines as addressed to three separate vocatives (Sirmio, the waters of the lake, and quidquid est domi cachinnorum). But quantum est hominum uenustiorum at 3.2 makes a fine subject for lugete in its context, whereas quidquid est domi cachinnorum produces an appropriate internal accusative with ridete; and one that subtly generalizes the effect from the waters of the lake to the inhabitants of the house. Moreover, cachinni are associated with waves also at 64.273. If we read ero gaude/ gaudente with Bergk, gaude is given a suitable cause -- it is the return not the identity of the master that gives the place reason to rejoice; Sirmio is the subject of two imperatives, and undae of one (despite Trappes-Lomax's claim on p. 96, at end); the waters of the lake are told to smile or laugh, ridere being a verb applied to waters (Lucretius 1.8, 5.1005), as gaudere is not; and we do not need to conjure up an alternative for the transmitted uos quoque more radical than uosque. (This is one of many places where the failure to provide a translation makes it harder than it should be to see how the recommended text is understood.)

34.22-3: Fowler's Romulam (apud Lyne, CQ 52 (2002), 604 = Collected Papers 287-8) is attractive, though hardly necessary; it does not deal with the serious problem in these lines, the sense of antique. As Trappes-Lomax argues, the adverb means "in an old-fashioned way", not "of old"; it is therefore out of place here. He suggests antehac (found in poetry at Enn. Ann. 203 Sk. and Lucr. 1.541); this is not impossible; but either of the renaissance conjectures Ancique and antiquam (both cited by Ellis) does the job far more plausibly, each giving a sense of historical antiquity that Romuli (or Romulam) gentem does not have on its own. Ancus was reckoned to have added the Aventine, site of the temple of Diana, to the city (Livy 1.33.2), though the temple itself is attributed to the reign of Servius Tullius (Livy 1.45). In support of antiquam, Bert Lain draws my attention to Cic. Leg. 3.4 omnes antiquae gentes regibus quondam paruerunt.

35.13: A scholar who conjectures nam ex quo at the start of a line has a duty to point to the parallels for the elision of the opening monosyllable: in hendecasyllables cf. 13.11 nam unguentum; 15.18 quem attractis.

36.3-8: Trappes-Lomax illustrates the absence of subject for daturam (cf. also 42.4) but goes on to make the (apparently quite separate) claim that Catullus will have written the invariable future infinitive daturum, as Gellius argues earlier Latin writers regularly did (1.7.15). A little more scepticism would be appropriate here: Gellius was reliant on one manuscript over 200 years old for the notion that Cicero at Verr. 5.167 preferred the archaic form to the participial form he normally uses, and the difference is between futurum and futuram. And this, the latest text to which the infinitive is attributed, was written rather earlier, and by an author rather older than Catullus.

42.17: Trappes-Lomax argues that, like other verbs meaning to extract something from somebody or something, exprimere requires either simple dative or ablative with a preposition; he therefore follows Maehly in reading ruborem/ ferreo canis exprimamus ori. However, the only examples he cites with ruborem/es as object have dative of a pronoun (cuiuis, Rhet. Her. 4.14; sibi, Sen. Ep. 11.7); there is a general tendency for ablatives accompanied by epithets not to need a preposition; and there is a possibility that the sense here is not "from" but "on": the face is, after all, not separated from the blush; cf. Verg. Geo. 1.430 suffuderit ore ruborem.

44.21 is deleted; but inadequate attention is paid to Baehrens' fecit, printed by Goold and now admirably explicated by J. Diggle, "On the text of Catullus", MD 57 (2006), 85-104, at 91-3. Generalizing from one instance is what poets do: cf. 12.1-3/11.

56.5-6: It is well worth reviving Schwabe's in puella; but it is unlikely that the following participle is not to be taken with pupulum, whether we read the transmitted trusantem ("thrusting"?) or the attractive humanistic conjecture crisantem ("moving the buttocks like a woman having sex"), which is adopted by Trappes-Lomax. I find 68 separate instances of hunc ego on the PHI disk, and in the only three where the combination does not begin the clause (Hor. Serm. 2.2.112; Ov. Met. 14.235; [Quint.] dec. min. 317.9), the punctuation is not in doubt as it would be here.

61.164: unus is also defended by H. Tränkle, MH 38 (1981), 256-8, who follows Ellis and Oksala in taking it as setting the solitary bachelor against the couple. But the placing is odd, enough is said to describe the groom, and intus makes a sharp contribution at the moment when the bride is imagined looking inside the new home she is about to enter.

62.63 Trappes-Lomax argues for tertia pars patri, pars est data tertia matri, but without any discussion of the unbalanced phrasing or whether the parallelism between father or mother is thematically desirable.

63.54: Nothing very sharp is said in support of the deletion of this line. Recently (MD 57 (2006), 98-102) Diggle has suggested et erae leonum adirem fremibunda latibula, but it seems unfortunate to anticipate in this wild locale the entry of the yoked lion that Cybele sends against Attis; and the double genitive is difficult ("the lairs of the lions of the mistress"); we may contrast cases where the second genitive is required for understanding the first as in dominae dearum ad aures (advocated by Diggle for 75 and a more attractive solution than Muretus's deletion, accepted by Trappes-Lomax). (In 74 Diggle's preferred text is again persuasive: hic [Itali] ... sonus editus [Fröhlich].)

64.52: Trappes-Lomax supports the humanistic alteration of namque fluentisono to namque fluctisono, but without addressing the metrical oddity (see on 64.136: it is especially odd to lengthen a word like namque to produce an opening spondee) or considering whether Catullus might have had good reason for introducing the unprecedented form: movement and sound are paired as remarkable features of this unruly ecphrasis. On the other hand, he might have noted cedentem in the line below as a possible origin for the corruption supposed.

64.109: The tradition offers prona cadit lateque cum eius obuia [or omnia] frangens; Trappes-Lomax rejects other corrections as "metrical fillers" and improves on Ellis's late quaeuis cumque with late quae sunt cumque. Agreed, this is an improvement; but it is hard to see that it is not a "metrical filler": what does it add to obuia? ruina is applied to trees at Statius, Thebaid 6.85, Silius 4.621 and Hercules Oetaeus 1630, so we might read Schwabe's lateque ruinis, perhaps with frangit.

64.119: Trappes-Lomax deletes. Removal of the line seems attractive at first, but here are two considerations that may discourage assent: would Catullus write omnibus his (120) if he is summing up just the immediate line and a half 117-18? Cf. 11.1-13, 64.63-6, 399-405, 115.1-5? And where has the transmitted nonsense quae misera in gnata deperdita l(a)eta come from? Does he imagine it emanates from what was originally a complete line? If so, what? For a clear explication of the context, including a case for Conington's compelling conjecture lamentata est, see W.Clausen, "Ariadne's leave-taking: Catullus 64.116-20", ICS 2 (1977), 219-23.

64.136: Trappes-Lomax conjectures crudele flectere mentis/ consilium, where the tradition has crudeles ... mentes and editors print crudelis ... mentis. This was worth considering. However, he rightly points out the ways in which the conjecture raises suspicion (no examples in this poem of a short syllable made heavy before mute + liquid [though Nisbet's rapida is attractive in 395], and none in thesi in classical dactylic verse), and fails to observe that though one of his starting points ("the sandwiching of the genitive is typical") can be backed up by his six examples from 64 (and more), there are several cases of genitive phrase preceding the noun on which it depends: 75, 128, 346, including at 147 a case that looks an adequate parallel to defend the vulgate text here: cupidae mentis ... libido.

64.260: Trappes-Lomax deletes orgia quae frustra cupiunt audire profani as "a pedestrian and irrelevant commonplace". Despite OLD 8 he is troubled by audire in the sense "to be informed about", and argues that it is sight not learning that matters, quoting Theoc. 26.13-14 ὄργια βάκχω, ἐξαπίνας ἐπιοῖσα, τὰ δ'οὐχ ὁρέοντι βέβαλοι. However, Apuleius, Metamorphoses 11.23 provides multiple examples of audire in a discussion of mystery revelation, e.g. cognosceres, si liceret audire. In any case uidere would be an easy substitution, and the echo of Theocritus gives enough warrant for the presence of the line.

64.263: OGR have multi raucisonos efflebant cornua bombos; the vulgate text is multis raucisonos efflabant cornua bombos. Trappes-Lomax points out that the possessive dative is not a straightforward example; there is also the oddity of the position: Catullus does not have many spondaic words in initial position, and possessive datives are normally placed less emphatically. He observes that multi is as likely to be a corruption of multae and proceeds by logical steps to argue for multae raucisonis inflabant cornua bombis. However, this changes four separate words, and it would be more reasonable to see efflabant and bombos as protecting one another: multaque would avoid the opening spondee, or we might try multae raucisonos efflabant cornibu' bombos (cf. Propertius 3.3.41 rauco praeconia classica cornu/ flare), though I doubt whether ecthlipsis belongs in 64.

64.273: Though Lennep may have been the first to put them together, leni is cited by Mynors from the theta group of MSS (with resonant) and que sonant goes back to O.

64.383: The palaeographical point made in favour of Baehrens' cecinere e would be stronger if there were not another place where the tradition has replaced -erunt with -ere: 64.16. Presumably the -unt ending had been written as a suspension above the line.

64.403-4: Trappes-Lomax follows Nisbet in reading improba for the first impia, but the attachment of improba and impia to the same noun is far less attractive than Reeve's conjecture (Phoenix 34 (1980), 183): ilia for the second impia, with se deleted.

66.9: One of the weakest notes in the book ends by proposing regina omnibu' diuis for multis illa dearum (Callimachus has πάσιν ἔθηκε θεοῖς). Haupt's simple cunctis illa deorum is rejected on the basis that the supposed corruptions are irrational: can we really give no weight to the appearance of the goddess Triuia in 5, the gender of illa, and the fact that the sequence of minims in cunctis (if that was the original) would have been preceded by the m of quam?

68.39: Trappes-Lomax (like Goold) accepts Nisbet's hucusque for utriusque in the final couplet of 68A. The only argument offered against utriusque is that Catullus has been asked by his friend for a single thing, love poetry, described in verse 10 with the hendiadys munera et Musarum et Veneris. This is a false claim: Manlius has asked for two separate things, as is shown by the articulation et ... et (and the correspondence to the previous two couplets), and then by Catullus's double explanation of what he cannot provide. First he talks of his previous acquaintance with Venus (15-18), then of how his brother's death has taken away such pursuits (19-30). The poem is not explicit about what munera Veneris might be; there are perhaps hints that they might share a bed, but the more solid implication is what we might call "the life of love", partying and whatever that might lead to. Catullus cannot offer haec munera (32) because after his brother's death he cannot lead the life of love. The second part of the response comes introduced by nam quod (37). nam does not here make a link to the previous sentence (there is no causal connexion), but takes us back to the initial refusal of the request; it is the resumptive use found a number of times in Cicero's lists and sometimes misunderstood there. (In his commentary on pro Caelio 4.23 R.G.Austin (Oxford, 3rd edition, 1960) gets it right, but A. Dyck goes badly astray through trying to find a causal sense for nam in his comment on DND 1.27 (Cambridge, 2003).) Verses 33-6 make the quite separate point that Catullus has no stock of books with him, so cannot send munera Musarum. Of course in writing the poem he does give Manlius both. (Most of this is already available in a note by A.J. Woodman, "A reading of Catullus 68A", PCPhS 29 (1983), 100-6.)

90: The deletion of 3-4 is compelling, but the retention of gnatus from OGR seems misguided: the subject of ueneretur is the same as the subject of nascatur and discat in 1-2. A new nominative substantive implies a new subject, and, given that Gellius himself is his mother's son, gnatus is very confusing. gratus, on the other hand, goes well with accepto ... carmine: cf. the pairing of the epithets at 96.1.

91.5: Trappes-Lomax rightly argues against sed neque quod: neque is displaced, and Catullus is not as intent as Ovid on getting a dactyl in the first foot. But the couplet has another oddity: the late placing of tibi, last in its clause, misleadingly just before the relative. tibi belongs in the hexameter, with matrem and germanam: I suggest Catullus wrote sed tibi nec matrem nec germanam esse uidebam/ hanc cuius meme(t) magnus edebat amor. The et of verse 7 now properly links the two parts of the positive account of the past.

96.4: It would be wrong to deny the possibility that renouamus might be a corruption of Peiper's reuocamus, advocated by Trappes-Lomax; but it is absurd for him to him to claim "no amount of weeping ... will renew a love which Death has sundered", when the poem ends with the assertion that the dead Quintilia gaudet amore tuo.

98: A good case is made that the epigrammatic thought culminates with 4; but rather than deleting 5-6 or seeing them as a cohering addendum we might think of them as a separate epigram. hiscas ("open your mouth wide") -- and you'll kill us all: the paradox of this simple weapon of mass destruction is the point. The repetition of omnino is not obviously pointed, however, and we might look for a replacement in one line (nos Romanos?).

100.6: For the transmitted est igitur Palmer conjectured igni tum, which Goold (e.g.) prints; Trappes-Lomax objects to tum, "for which there is no MS evidence": what then is igitur?