63.1–5

In line 4, uagus is usually construed with animi or animis depending on how the editor corrects amnis. WhiteFootnote ³ has suggested construing animi as a limiting genitive with furenti rabie, citing Cic. Tusc. 3.26.63, but this explanation could do with a stronger defence, as it is a novel interpretation. First, when uagus describes persons with no suggestion of physical motion it refers to vacillation or fickleness (for example Vell. Pat. 2.76.2; Prop. 1.5.7; Mart. 2.90.1), but for Attis to have crossed the ocean ‘eagerly’ and castrated himself means that he was of a fixed will, the opposite of what uagus should mean here. The phrase furenti rabie animi is also more characteristic of Catullus’ idiom than uagus animi/s. While the limiting genitive or ablative with uagus is generally difficult to parallel (commentators usually cite Verg. Aen. 8.228 furens animis, although FordyceFootnote ⁴ cites ferox animi and similar phrases in favour of animi), the pleonastic genitive of sphere with animi and its synonyms is particularly common in Catullus. Line 38 of the same poem deserves special attention as the phrase is almost the same (rabidus furor animi), but there are eight other instances (2.10, 64.136, 64.147, 64.194, 64.372, 65.4, 68a.26, 102.2). It may be objected that the hyperbaton of uagus is unusual, but it can be paralleled with tua Footnote ⁵ in line 9, as long as one does not accept Trappes-Lomax's deletionFootnote ⁶, which I do not. This hyperbaton slows down the rhythm, but it could equally be argued that the unusual word order reflects Attis’ mental state. The corruption to animis (ignoring the miscounting of minims that gave the archetypal amnis) would be caused by dittography of final -s from uagus.

The reading transmitted in line 5, deuoluit (‘rolled down’), is too mild for this extremely violent context, and so it is accepted to be corrupt. Haupt's deuolsit has won the most approval as it involves the simplest change, but Stephen Harrison has pointed out to me (viva voce) that deuello is not the vox propria for this context, as it describes the act of plucking flowers, hair, feathers and similar inanimate objects (see OLD s.v.). The act of slicing off the genitals with a stone cannot be compared very precisely to plucking. Indeed, the TLL gives only this passage for extending its meaning to body parts (5.1.849.14–15). I would also add that deuello is a rare word. The PHI disc gives seven examples, none of them contemporaneous with Catullus. The closest in date are Lucilius (30.1027 Marx) and Statius (Theb. 2.604). As deuolsit is restored by conjecture, we should question whether the vulgate reading is the best one possible.

The violence suggests di- rather than de-, whether it be diuellit (Aldine edition, 1515) or diuulsit.Footnote ⁷ The exchange of di- and de- is extremely frequent and it produces a word appropriate and common to dismemberment (compare, for instance, 64.257 pars e diuulso iactabant membra iuuenco, ‘some brandished about limbs torn from a bullock’), and yet I have not found it printed in any edition since 1805.Footnote ⁸ Argument from usage therefore strongly favours diuellit over deuellit.

The appropriate form of the perfect still needs determining. KokoszkiewiczFootnote ⁹ advocates Werthes's deuellit.Footnote ¹⁰ He infers from its use in Laberius (inc. 137 Bonaria), who used many vulgar forms, and from its rarity in most writers (except Lucan and Seneca) that -uolsi is colloquial. However, the form -uulsi is attested in Ovid (Met. 12.300) and Propertius (3.15.13), so this idea appears not to be valid.Footnote ¹¹ As the form is unobjectionable and it is slightly closer to the transmitted reading, diuulsit is preferable to the second Aldine's diuellit.

I accept Currie'sFootnote ¹² arguments in favour of ipse over ili for the troubled -iletas. But my reason for discussing this conjecture is its attribution, which is complex and goes back further than Palmer,Footnote ¹³ to whom Kiss attributes it.Footnote ¹⁴ Palmer attributed it in turn to Ellis and Passerat, though I can only find it in the latter.Footnote ¹⁵ Passerat attributed ipse to a ‘uetus liber’, which seems to be Muretus's edition.Footnote ¹⁶ Muretus printed ipse without comment, so it is possibly not his own conjecture. A similar reading, iste, is found in Avantius's 1502 Aldine edition; the 1515 Aldine edition and Avantius’ Trincavelli edition read ille. Avantius’s 1502 was enormously influential,Footnote ¹⁷ so it seems that it was used as Muretus’s base text and that ipse was printed by mistake, albeit a fortuitous one.

63.13–16

Dindymene is a rare Greek word that makes only two other appearances in Latin, both of them in Greek forms.Footnote ¹⁸ Catullus’ manuscripts are not to be trusted on orthography, and indeed Catullus does use Greek morphology. In the case of first-declension genitives, he uses the Greek genitive for Cybele and Cybebe, in accordance with all other Classical writers, and there is also Arsinoes at 66.54. As it is a rare word, and therefore one unlikely to have been naturalized to a Latin form through common usage, and given Catullus’ free use of Graecisms, it is likely that Catullus would have preferred the Greek form Dindymenes.

In this case the homoeoteleuton should also be considered, as this conjecture removes it, and it may be considered deliberate. In this poem there is homoeoteleuton in 28 out of 93 lines, so it is common, but 18 of these involve a short a and should probably be thought of as consequences of the limited ways of producing long sequences of short syllables.Footnote ¹⁹ Some of the others may be considered deliberate: in lines 59–60 the repeated final long -o recalls the exclamation o, in line 64 it may emphasize the force of fui, in lines 74 and 86 it may be onomatopoeic, and in lines 91–2 it may imitate ritualistic incantation. The effect in this case may be similar to lines 92–3, but I prefer the exoticism of the Greek morphology.

63.29

Here Catullus describes the instruments’ sounds, but unlike caua, which refers to the cymbal's resonance (cf. Lucr. 2.620), leue is irrelevant to sound. A far more appropriate word for this context is graue, for which we can compare a dramatic fragment describing the same thing: agite turba Phrygia nunc et quatite grauia tympana (‘Come on now, Phrygian crowd, shake the low-pitched tympana’, Anon. apud CGL 6.144). Both caua and graue should be taken adverbially, referring to the instrument's low-pitched tone (OLD s.v. grauis 9a). The collocation graue mugire is well attested for both musical instruments and animals, and so this also has the advantage of reinforcing the pervasive animal imagery.Footnote ²⁰ The earliest and most interesting parallel is in Lucretius: tuba depresso grauiter sub murmure mugit | et reboat (‘the horn booms and bellows with a low-pitched tone in a low murmur’, 4.543–4 = 4.545–6). These two writers are the first who attest to the verb reboat, and one may be imitating the other.Footnote ²¹ This is a polar error motivated by a reminiscence of leue typanum from line 8. A parallel is found at Sen. HF 1117, where the manuscript A gives leues for graues because of leui in line 1113.

63.52–4

This passage is a locus conclamatus. The transmitted text gives a hemistich that lacks anaclasis, which is agreed not to be possible in Catullus’ handling of this metre.Footnote ²² In poetry earum is avoided except in Lucretius and Horace's Sermones,Footnote ²³ and is far too weak to balance ferarum. omnia has also been suspected. Nisbet asks ‘why should Attis tour all the lairs?’Footnote ²⁴ But Postgate disputes this, as it can perhaps be taken as a piece of emotional exaggeration.Footnote ²⁵ More recently, Diggle and Nisbet have also cast doubt on furibunda.Footnote ²⁶ It is not an easy epithet to apply to latibula, so it is usually taken to be feminine singular, applying to Attis, but the word order makes this awkward. As for the conjectures, the only unobjectionable suggestion is Heyworth's ut for et,Footnote ²⁷ as the temptation to restore anaphora is difficult to resist when it is so frequent, especially in lines 20–5 and 62–72.

Trappes-Lomax cuts the Gordian knot by deleting the line,Footnote ²⁸ but this is not very attractive, as these two lines are closely complimented by another pair in lines 70–1,Footnote ²⁹ and deleting line 54 will upset this structural balance. The fact that a line is repetitive often argues in favour of interpolation, but repetition of all kinds is such a pervasive device in this poemFootnote ³⁰ that it argues just as much in favour of authenticity.

The most frequent type of conjecture in recent discussions has been to look for a genitive plural behind earum to describe latibula in some way and balance ferarum. The combination Phrygiae … Idae in the corresponding passage at lines 70–1 supports this. Diggle's erae leonum introduces Cybele's lions too early, and gives an awkward double-genitive construction. Heyworth's aprorum opaca Footnote ³¹ is better in some ways, but it is unlike Catullus to balance a general word for beast with a specific species. He groups together terms more similar in meaning than ferarum and aprorum, such as domum Cybebes and nemora deae in line 20, mei creatrix and mea genetrix in line 50, and Idae and Phrygiae in lines 70–1. Mueller's opaca is attractive in sense, as the cold is then complemented by the dark, but palaeographically it is rather unlikely. A more general word for ‘beast’ would be more convincing, but none has yet been suggested.

More approaches need to be tested, so I will offer some here in the hope that it will stimulate fresh debate. The first way is to reassign the dens to inhabitants other than animals, and the best alternative is the Maenads.Footnote ³² The only suggestion with any palaeographic plausibility that I can offer is ut Maenadum animo Footnote ³³ adirem. There are two potential paths of error. Saut du même au même may have produced the arrant nonsense ut m[a]en adirem, omīa would have been a reasonable guess for maen in Gothic script, where e and o are often confused, and earum would be a metrical filler. Otherwise, m[a]enadum, although it was only lightly corrupt in line 23 and transmitted correctly in line 69, may have been corrupted to earum because of its unfamiliarity,Footnote ³⁴ and animo is simply an anagram of omnia. animo has the considerable advantage of easing the difficulty of furibunda. The idea of line 53 is inspired by epigrams in which a Gallus shelters in a cave and meets a lion ([Simon.] Anth. Pal. 6.217 = 918–27 FGE = [116] Sider; Antipater of Sidon 6.219 = 608–31 HE; Dioscorides 6.220 = 1539–54). This would be a novel development of that idea.Footnote ³⁵

The second option is a word that can complement ferarum without combining the general with the specific as all previous efforts have done. My best effort is beluarum, in which furibunda can again be taken as a transferred epithet. The combination is attested elsewhere in prose and verse,Footnote ³⁶ so the sense is not objectionable, but it may be felt that there is not enough distinction in meaning to justify the repetitiveness. The corruption is explicable but involves several stages, so it will not convince everyone. beluarum would have been corrupted to ueluarum by betacism (for which we can compare uerum uera for uerbera in line 81) and rationalized as uel earum. uel would then be deleted as nonsensical and omnia interpolated to fill the metre.

The third approach is to use a different kind of genitive altogether, one that defines where the latibula are rather than whose they are. This is attested with latibulum, but not in elevated verse.Footnote ³⁷ As an example I suggest ut nemorum opaca adirem. In epigrams by Alcaeus of Messene and Antistius, a Gallus meets a lion in the woods rather than in a cave (Anth. Pal. 6.218 = 134–43 HE, 6.237 = 1101–8 GP), which could have inspired the variation. However, this does not alleviate the awkwardness of furibunda, and the repetition of nemorum after nemora two lines above has no point. This option seems the least attractive.

One last option is to supply more epithets to complement furibunda, which would then relieve some of its awkwardness, though on the other hand a transferred epithet would do this more effectively. An example along these lines is et ut animam agens adirem,Footnote ³⁸ which would repeat the epithets in line 31. animam for omnia (omīa for anīā) and the loss of ut after et would be simple errors, less so the omission of agens before adirem and interpolation of earum. A point against this is that it leaves latibula isolated, which also applies to ut beluarum adirem, but perhaps it may be taken as gloss on stabula, which is first attested here for the dens of non-domesticated animals.

Of all these options I have a slight preference for ut Maenadum animo adirem, but these are intended as diagnostic conjectures only.

63.61–2

I start with an analysis of the structure of Attis’ speech. As far as I know, it has so far been treated as a single unit,Footnote ³⁹ but it should be divided into two halves with a new paragraph starting at line 61.

The change in addressee from the patria in line 50 to the animus is significant: it marks the difference between addressing the public consequences of his self-imposed exile and the private ones. The two halves broadly run in parallel. First is the apostrophe (49–51a, 61), then a series of anaphoric rhetorical questions focussing on Attis’ present situation (51b–5, 62). This is followed by a statement that expands on the rhetorical questions (56–7, 63–7),Footnote ⁴⁰ though in different ways. The fourth element is another set of anaphoric rhetorical questions that speculate on what the future will hold (58–60, 68–72), in both cases dwelling on how Attis can live on Ida. The speech is then rounded off with a short conclusion (173), though the apostrophe to the animus contains a similar statement (querendum est etiam atque etiam, 61).

The first half focusses on Attis’ self-imposed exile, the second on his changed social status. Attis’ epithets for the patria—creatrix and genetrix—are pregnant, as they evoke the debt of pietas that Attis owes to it, even though he has abandoned it. If we compare a fragment of Varro's Menippean Satires, he seems to have even violated it (fr. 236 Cèbe = 235 Astbury):

This retrospectively transforms Attis’ migration into betrayal. He becomes like a runaway slave, the lowest of the low (51–2), and he will now always be without any of the benefits of having a patria (59–60). The second half expands on the material losses of his desertion and the soliloquy becomes much more introspective, as Attis dwells on the status he had in his patria (lines 64–6) and what it must become now (68–9).

As for textual matters, in line 62 the manuscript reading non quid abierim is impossible. quid is nonsensical, and it requires ab with abierim. obierim has stood in most modern editions with the sole exceptions of Baehrens, who tries to defend abierim,Footnote ⁴¹ and Palmer,Footnote ⁴² who reads habuerim. This goes back to Statius, who read habu[e]rim in one of his manuscripts.Footnote ⁴³ The best way to understand obierim is as ‘accept’ or ‘undertake’ (OLD s.v. 5), though there is no parallel for its application to physical states, only to something more abstract. The strangeness of obierim is apparent from Thomson's gloss, who translates obierim more literally as ‘passed into <on my way>’.Footnote ⁴⁴

habuerim must be revived. At the very least it belongs in the apparatus criticus. Palaeographically it is at least as easy as obierim, especially after the repeated abero (lines 59–60), as it involves little more than the loss of an aspirate.Footnote ⁴⁵ It is also much easier in sense. Unlike obierim it is regularly used for physical features (OLD s.v. 3), for example habent figuram litterae V (‘they have the shape of the letter V’, Varro, Ling. 5.117), so its usage here is unremarkable.

63.63–4

mulier is dubious grammatically because it requires an ellipsis of sum as well as of fui. Footnote ⁴⁶ It is also implausible psychologically. In these lines Attis shows a deep regret for his lost youth, whereas in lines 68–9 the tone is one of self-loathing for his present lack of manliness. Here mulier weakens this contrast. I prefer Rossberg's iuuenis,Footnote ⁴⁷ but recently Thomson and GooldFootnote ⁴⁸ have printed Scaliger's puber.

The purpose of this note is to improve on Scaliger's conjecture, even though I reject it. puber gives the right sense but is of doubtful Latinity. In surviving literature and in inscriptions only the form pūbēs, which does not scan here, is securely attested. According to Priscian (CGL 2.249), Caesar's De analogia, which is contemporary with this poem, advocated pūbis pūberis. Other grammarians attest to the form pūber (for example CGL 5.418.10; Serv. Aen. 5.546), but they are much later in date and quote no examples. This was probably formed by levelling, and the lack of quoted parallels may indicate that it was a spoken form. It is uncertain whether Caesar supported pūbis pūberis by analogy or usage,Footnote ⁴⁹ but perhaps the former is more likely, as ‘Cerēs Cereris is the only parallel for a nominative pūbēs’Footnote ⁵⁰. Against this are cinis cineris, puluis pulueris and so on. Garcea also cites an apparently unique form pūbĕs (Cic. Rab. Perd. 31), but this is long by position before -que. It is possible that others used pūbis pūberis but that it has been eliminated from the manuscript record by normalization. Whatever the case, the evidence indicates that pūbis is more likely to be a Republican form than puber, and should therefore be preferred to Scaliger's original conjecture.

The tense variation of fui and eram has been suspected by Baehrens, Nisbet and HeyworthFootnote ⁵¹ because of the lack of parallels for such a variation that disregards potential differences in nuance. Heyworth's otherwise neat solution of eram gyminasii flos must be rejected because the genitive form -ii is not attested until the Augustan period. ThomsonFootnote ⁵² says that fui carries over from the previous line, but the parallelism between the two phrases in line 64 is too close for such an explanation. In this case it is essential that any parallel should involve the same verb. I offer the following: Arcesilas tuus, etsi fuit in disserendo pertinacior, tamen noster fuit; erat enim Polemonis (‘Your friend Arcesilaus, although he was rather too obstinate in discussion, was still one of ours, for he was one of Polemon's’, Cic. Fin. 5.94), uictor erat quamuis, aequus in hoste fuit (‘although he [Minos] was victorious, he was fair to the enemy’, Prop. 3.19.28), and parua mea sine matre fui; pater arma ferebat; | et duo cum uiuant, orba duobus eram (‘As a small child I was without my mother; my father was under arms, and though the two lived, I was bereft of the two’, Ov. Her. 8.89–90). The latter two cases are due to metrical constraint,Footnote ⁵³ the first to avoid excessively repeating fuit.

63.68–9

DiggleFootnote ⁵⁴ has doubted uir, since Attis is no longer a man (euirastis, 17). He could adduce no parallels for the phrase uir sterilis, and the fact that Attis describes himself with feminine forms (54, 58, 68; Maenas, 69) shows that he does not consider himself a man anymore. But surely the oxymoron is pointed here, and the tone is bitterly ironic. Diggle's solution is ui, and he intends ui sterilis to mean ‘deprived of sexual potency’, but this idea has already been conveyed by mei pars, and the meaning he ascribes to the singular ui properly belongs to the plural (OLD s.v. uis B20). The singular could be justified if procreandi were understood in ellipsis (OLD s.v. uis A16), but this usage belongs to philosophical discourse.

There is, in fact, an excellent parallel for uir sterilis in Mart. 3.91.5 hoc steriles sensere uiri (‘when the sterile men sensed this’), which also describes the Galli. As these are the only two places where sterilis occurs with uiri attributively, Martial has likely taken the phrase from Catullus.

63.74–7

It is difficult to say what the point of laeuum (lenum O) is, and there has not been much analysis as to what it might mean. The transmitted text is usually accepted, but here Thomson is a notable exception. Since Cybele is addressing one of a number of lions, Catullus may have wanted some specification of which lion Cybele is addressing. The only possible defence for the paradosis I can conceive of is that Catullus imagines that there are two lions, the mythological figures of Atalanta and Hippomenes (Ov. Met. 10.703–4). Ovid's account is the only one which mentions Cybele, but he may have drawn from a lost Hellenistic source, since aitia and narratives about localized deities are typically Hellenistic interests. Catullus may be referring to Atalanta because of her speed, but this would be cryptic, and I can see no reason why laeuum should denote Atalanta as opposed to Hippomenes, unless this were clarified by the putative Hellenistic source. This is so far the best defence for the text, but it is still far too weak for me to take seriously, and one does not need to be a radical critic to accept Puccius's saeuum, a standard epithet well attested with hostis. Manuscripts often oscillate between laeuus and saeuus (TLL 7.2.890.64–5), so the error is trivial.

63.78–80

This is a passage that has not been treated in much depth, as Butterfield has said.Footnote ⁵⁵ Indeed, the vulgate text has remained the same for 300 years. However, there is a problem with the Latinity of agedum age <i> fac. For age I will defend D'Orvill's <i>.Footnote ⁵⁶ The variation in hortatory or jussive particles like agedum age is unparalleled. The normal idiom can extend an imperative to a maximum of one particle and two paratactic imperatives (the first always either age or i), and either the particle or the first imperative can be geminated. The closest parallel to the vulgate text is age age, i puere, duc me ad patrios fines (‘Come on, come on, boy, go and lead me to the country's borders’, Caecilius, fr. 100 Ribbeck), where the particle is geminated, but that is not a parallel for the variation agedum age. The only way to make the vulgate resemble idiomatic Latin is to print a strong sense pause before i, but age, even if one accepts that agedum age is possible, is meaningless as a main verb without an object, and the pause in this position is not seen elsewhere in this poem. If Bancks and Annesley's <agitet> is right, one may argue that the consonance may be deliberate, but it is still an unparalleled expression and should be treated with great suspicion.

All of this makes emendation of age necessary to restore good Latin. D'Orvill's agedum inquit i is far more idiomatic, and the loss is easy to account for palaeographically. After the long succession of minims in inquit the loss of another minim would be simple, and age is an inept supplement drawn from agedum or age.

After ferox the transmitted text is a syllable short. Scaliger's <i> has stood without challenge in all modern editions, and the gemination of the imperative is attested in poems of a high style.Footnote ⁵⁷ It is more pointed than doubling the hortative particle, as the lion encourages itself (line 85), whereas Cybele (caede, 81; quate, 83) and the narrator focus on motion (uadit, fremit, refringit uirgulta pede uago, 86), which the geminated i would develop. Finally, the gemination of i is stylistically supported by the gemination of fac, so the vulgate reading here is more than adequate.

ferox is usually printed without commas, so it could be taken as adverbial or as a vocative. In line 83 it must be vocative: Cybele does not need to tell her lions how to shake their manes, as it is an intrinsically threatening gesture. In this line I take it in the same sense, so commas would be appropriate in both lines to make this clear. If we take ferox as semantically identical in both lines, this would be complemented by i … i and fac ut … fac ut.

The end of line 78 is also defective, probably owing to its unusual length.Footnote ⁵⁸ Butterfield's <acuat> is more pointed than the vulgate agitet. His note is brief, and since <agitet> has won such unquestioning support it will do no harm to reinforce his arguments. His conjecture produces the correct meaning (OLD s.v. 4) whilst adding to the animal imagery seen elsewhere in the poem.Footnote ⁵⁹ Attis is compared to an animal in line 33 and he quite possibly compares himself to wild animals in line 72.Footnote ⁶⁰ The consonance with fac ut … ictu is also appealing, more so than the consonance with agedum produced by agitet, it is entirely in this poem's style, and it would imply an etymological link between acuo and icio. This is a device Catullus uses elsewhere, for instance in 64.156, where Scylla rapax suggests σκύλαξ. agitet, which has stood since Lachmann, is not as choice, but it is used with a similar meaning in 63.93 and 64.94 heu miseros agitans [dub. Markland; misere exagitans V] immiti corde furores [‘alas stirring up wretched madness with his cruel heart’]).

There is yet another syllable missing from the transmitted text in the next line. The most commonly printed emendation is Lachmann's uti, which is only securely transmitted in Catull. 116.2. That line is unique in having ecthlipsis and being holospondaic, both of which imitate Ennian practice. In that context uti suggests itself as another archaizing form. The manuscripts’ fac ut hunc furor <acuat> fac ut furoris is strong evidence for a more than usually complex figure of repetition that involves gemination of the imperativeFootnote ⁶¹ varied by a change in subject in the subordinate clause, thus producing polyptoton (furor … furoris) straight after the imperative (fac ut). The polyptoton of hunc furor … hic furoris is more elegant than the prosodic variation ut … uti. The second option introduces a new variation without a point, whereas the first reinforces an already existing figure, and the repetition of the pronoun draws attention to Attis’ exceptional behaviour (for a Gallus) of regretting his castration. If one follows Goold in believing that Cybele should speak of Attis as a female, the beginning of line 79 should read ut haec, whereas Goold prints uti.