1. The designation puella is a potent signifier of elegy. By contrast, the most polite term Horace can muster for his ‘partner’ is mulier, in some contexts (often generalising ones) perfectly respectable but insulting in his line, and not an erotic word. It does occur in love poetry, but only rarely, often in a quasi-formulaic usage with uir, and not infrequently with less-than-pleasant undertones. Cf. for example Ov. AA 3.765 turpe iacens mulier multo madefacta Lyaeo. For a comparison of the word with femina, which shows a strong preference against mulier in Augustan poetry, see Axelson, B., Unpoetische Wörter (Lund 1945), 56Google Scholar. Axelson’s technique is refined as regards puella by Watson, P., ‘Axelson Revisited: The Selection of Vocabulary in Latin Poetry’, CQ 35 (1985), 430–48CrossRefGoogle Scholar, at 433f. She argues that in epic contexts puella retains its diminutive force (usually for the sake of pathos), a force which is also a factor in its erotic use. She holds that genre, and not just context, is a significant factor in the use of puella in elegy.

2. For a reading of the parallel rising and falling of Am. 1.1 see Kennedy, D.F., The Arts of Love: Five Studies in the Discourse of Roman Love Elegy (Cambridge 1993), 59Google Scholar.

3. Don Fowler has made the entertaining suggestion that the ‘real reason’ for Ovid’s impotence might in the end be seen as that of which he is accused by his partner: aut alio lassus amore uenis (‘or you are coming to me when worn out by making love to someone else’, 80), which changes the whole nature of the poem—or does it?

4. The link is discussed by Baeza Angulo, E.F., ‘Ovidio, Amores III.7’, Faventia 11.1 (1989), 25–58Google Scholar.

5. See also Prop. 2.17.6 and Ov. AA 2.606.

6. It will be argued below that the subject matter is in fact highly pertinent to elegy, albeit sub-versively, although its expression is unconventional.

7. See Adams, J.N., The Latin Sexual Vocabulary (London 1982), 224Google Scholar, who describes the poem as nevertheless euphemistic.

8. A substantial analysis of the poem’s relationship with other poems on impotence is made by Baeza (n.4 above), who explores in detail parallels between Am. 3.7 and various Hellenistic epigrams, together with Augustan and a considerable amount of later material. He is not concerned specifically with poems about the speaker’s own impotence. Examples later than Ovid are to a greater or lesser extent dependent on our poem, or indeed on that of Horace (see below), as is the case for Martial’s rejection of a woman at 11.97. An amusing parody of the opening lines of our poem can be found in Priapeum 80, which begins: at non longa benest, bene non stat mentula crassaj et quam si trades, crescere posse pules?/ me miserum! cupidas fallit mensura puellas,/ non habet haec aliquid mentula maius eo. (‘But is it not truly long? Does the prick not stand good and thick?/ And isn’t it the sort that you would think would grow if you treat it right?/ Alas! Its size deceives desirous girls,/ for this prick has nothing greater than that.’)

9. Richlin, Amy, The Gardens ofPriapus: Sexuality and Aggression in Roman Humour (Oxford 1992), 117–19Google Scholar.

10. See Henderson, J.G.W., ‘Suck It and See (Horace, Epode 8)’, in Whitby, M., Hardie, P. and Whitby, M. (eds.), Homo Viator: Classical Essays for John Bramble (Bristol 1987), 109–18Google Scholar, for the construction of the ‘matey male’ poet and reader implicated in such misogynistic invective.

11. On these poems see Fitzgerald, W., ‘Power and Impotence in Horace’s Epodes’, Ramus 17 (1988), 176–91CrossRefGoogle Scholar, and Oliensis, E., ‘Canidia, Canicula, and the Decorum of Horace’s Epodes’, Are-thusa 24 (1991), 107–38Google Scholar.

12. The notion of poetic memory and the inherent failure of the recusatio will be discussed further below.

13. For a consideration of metapoetic readings of Ovid generally, see Hinds, S.E., ‘Arma in Ovid’s Fasti’, Arethusa 25 (1992), 84–153Google Scholar.

14. ‘Present’, rather than present, because as ever the presence of the author is a pretence at the moment of reading.

15. See AA 2.197 and Sharrock, A.R., Seduction and Repetition in Ovid’s Ars Amatoria II (Oxford 1994)Google Scholar, Chapter 4.

16. Kennedy (n.2 above), 59

17. Kennedy (n.2 above), 58. The sexual member has become identified with the whole man (see Adams [n.7 above], 46), in a way which almost leads, in this metapoetic reading, to a physical identification between the man and his poetry (as happens for example in Hor. Carm. 2.20): for there must, I think, be an echo in the inutile pondus (‘useless lump’, 15) as which Ovid describes his penis-self and the inutile lignum (‘useless log’, Am. 1.12.13) on which the ‘failed’ poem returned to the poet with the addressee’s rejection. No doubt the statue of Priapus in Hor. Sat. 1.8.1 is also active here: olim truncus eramficulnus, inutile lignum. That ‘useless log’ has now become highly utilis—‘useful’ in the sense of ‘sexually potent’.

18. Impotence is a nefas, a thing not to be spoken about. It is typically Ovidian to talk about a subject while protesting that it is not a fit subject for words. See for example the pretence at dismissal of the Muse at the bedroom door in AA 2.703, and Sharrock, A.R., ‘Ovid and the Politics of Reading’, MD 33 (1994), 97–122Google Scholar. I am grateful to Duncan Kennedy for the link between Ovid and Tantalus.

19. On this point see Adams, R.M., ‘Soft Soap and the Nitty Gritty’, in Enright, D.J. (ed.) Fair of Speech: The Uses of Euphemism (Oxford 1985), 44–55Google Scholar. The book also contains an essay on ancient euphemism by Jasper Griffin.

20. On this see Derwent May, ‘Euphemisms and the Media’, in Enright (n.19 above), 122–34.

21. The tendency to self-ridicule is a fascinating feature of euphemism. Surely there is something of that in the name ‘Health Alteration Committee’, given to itself by a CIA assassination unit.

22. See also Kennedy (n.2 above), 61–63, on appropriateness. There does seem to be some slippage in discussion of euphemism between ‘not saying taboo words’, ‘not speaking about taboo subjects’, and something in between in the essay of Joseph Epstein, ‘Sex and Euphemism’, in En-right (n.19 above), 56–71.

23. But not always! I recently thought someone was telling me that a scholar we wanted to come and speak at a conference had died. In fact he meant she had turned from the subject of the conference to some other field, but he said ‘you know she has moved on’ in a rather hushed voice.

24. Such is the force of Lateiner’s article on Am. 2.19. See Lateiner, D., ‘Ovid’s Homage to Callimachus and Alexandrian Poetic Theory (Am. ii.19)’, Hermes 106 (1978), 188–96Google Scholar. The idea has precedents in Callimachus (for example. Epigram 30) and in Catullus (for example, poem 50).

25. Kennedy (n.2 above), 58–63.

26. A neat example comes from Tacitus: tragoediam disposui iam et intra meformaui (‘I have set out a tragedy already and formed it within me’, Dial. 3).

27. See TIL 6.1.1109.35–72.

28. It is used for the composition of poems for example at Hor. Carm. 4.2.32, Prop. 4.1.135, Mart. 12.94.9. The semantic range of the word would suggest that the metaphor is from physical shaping, but also with associations of mental invention and imagining.

29. Cf. Cic. de Orat. 2.79: inuenire quiddicas, inuenta disponere.

30. See for example Prop. 3.24 and 25, and Ov. Am. 2.18.5f., 3.11, 3.14 and 15.

31. Horace is particularly fond of the word. Cf. Sat. 1.4.7, Carm. 4.2.11, Ep. 1.18.59, 2.1.158, 2.2.144, AP 74, 270. Elsewere see Cat. 50.5, Virg. Eel. 9.45, Aen. 6.646, 9.776, Ov. Am. 1.1.1, 19, 27, 1.15.28, 3.1.28, Rem. 381, Fast. 2.6, 109, Met. 14.520, Tr. 2.331, 4.1.6, 87, 4.10.25, Pont. 2.5.1, 69 (with neruos, to be discussed later), 4.2.30, 4.8.73, 4.14.2, 4.16.11.

32. For impotence as programmatic for elegy, see briefly Wyke, M., ‘Taking the Woman’s Part: Engendering Roman Love Elegy’, Ramus 23 (1994), 110–28CrossRefGoogle Scholar, at 120. This important essay will feature again below.

33. The river is itself open to both sexual and poetic interpretations.

34. On the softness of elegy see Kennedy (n.2 above), 31–33; Oliensis (n.ll above), 125; Wyke (n.32 above), 117–21.

35. On the association of virility with the ‘real man’ see Henderson, J.G.W. ‘Satire Writes “Woman”: Gendersong’, PCPS 35 (1989), 50–80Google Scholar, at 50–53. See also Skinner, M.B., ‘Ego Mulier: the Construction of Male Sexuality in Catullus’, Helios 20 (1993), 107–30Google Scholar, at 111: ‘Any loss of physical vigor due to old age, infirmity, or overindulgence in carnal pleasure, any analogous lapse of moral reserve, or any diminution of social standing, can weaken the bulwarks of masculinity and cause reversion to a passive, “womanish” condition.’

36. Cf. Tib. 2.5.106 (inermis Amor) and Prop. 4.6.32 (testudineae carmen inerme lyrae) for elegy as opposed to higher genres.

37. This notion that self-control is threatened by love is central to Veyne’s reading of elegy. See Veyne, P., Roman Erotic Elegy: Love, Poetry and the West, tr. Pellauer, D. (Chicago 1988)Google Scholar.

38. Ancona, R., ‘The Subterfuge of Reason: Horace, Odes 1.23 and the Construction of Male Desire’, Helios 16 (1989), 49–57Google Scholar.

39. For an important discussion of the workings of the pharmakon see generally Derrida, J., Dissemination (London 1981)Google Scholar. On magic and love see Tupet, A.M., La Magie dans la poésie Latine (Paris 1976)Google Scholar; Fauth, W., ‘Venena Amoris: die Motive des Liebeszaubers und der erotischen Verzauberung in der augusteischen Dichtung’, Maia 32 (1980), 265–82Google Scholar; Luck, C., Arcana Mundi (Baltimore 1985)Google Scholar; Sharrock (n.15 above), ch. 2. Since erotic spells belong so intimately in erotic discourse, they tend to be performatively self-refuting, for the lover seeking a magical way out of love is generally driven further into it. Dido’s behaviour in Virg. Aen. 4 could be seen in this way, as could that of Tibullus in changing the content of the spell at the crucial moment (1.2.63f.).

40. This discussion occurs briefly, in a different form, in Sharrock (n.15 above), 75f. It would perhaps be possible to read haec nocuere mini as referring to the complaints which Tibullus goes on to make about the lena, but since she is most graphically a witch the point remains the same. Horace’s Canidia is, of course, a witch. It is no accident that there is magical involvement in the impotence-poems in the Epodes.

41. See also Faraone, C.A., ‘Sex and Power: Male-targetting Aphrodisiacs in the Greek Magical Tradition’, Helios 19 (1992), 92–103Google Scholar, who shows how spells were applied to men aiming to subjugate them by manipulating their desire, the magic-assisted seduction of Zeus by Hera in IL. 14 being a classic example. Magical love enervates.

42. See Cunningham, M.P., ‘Ovid’s Poetics’, CJ 53 (1958), 253–59Google Scholar.

43. See Kennedy (n.2 above), 34–39, and on Roman moralising discourse see Edwards, C., The Politics of Immorality in Ancient Rome (Cambridge 1993), 70–72CrossRefGoogle Scholar.

44. Classically loving and impotent is the weak and wimpish adulescens in comedy.

45. For example Prop. 1.8b, 2.7, 2.15, Ov. Am. 1.5.

46. The standard works on seruitium amoris are Copley, F.O., ‘Seruitium Amoris in the Roman Elegists’, TAPA 27 (1947), 285–300Google Scholar, and Lyne, R.O.A.M., ‘seruitium Amoris’, CQ 29 (1979), 117–30CrossRefGoogle Scholar, but neither of them really brings out the ambiguous gender-relations of the conceit. See also Veyne (n.37 above), 132–50.

47. See particularly Prop. 2.1.4 ingenium nobis ipsa puellafacit. Maria Wyke’s analysis of the beloved as the poetry, ‘Written Women: Propertius’ Scripta Puella’, JRS 77 (1987), 47–61Google Scholar, is the seminal article on this subject.

48. An important analysis of this poem as a paradigm for reading the elegiac woman can be found in Wyke, M., ‘Reading Female Flesh: Amores 3.1’, in Cameron, A. (ed.), History as Text: The Writing of Ancient History (London 1989), 111–43Google Scholar. Wyke shows how the traditions of classical scholarship have sought to garner evidence for real women’s lives from the writings of the elegists, and argues that this not only fails to acknowledge the poetic significance of the details given of elegiac ‘women’, but also that the tradition perpetuates the contemporary Roman mindset which presents women as erotic and poetic objects. See also Gold, B.K., ‘The “Master Mistress” of my Passion: The Lady as Patron in Ancient and Renaissance Literature’, in DeForest, M. (ed.), Woman’s Power, Man’s Game: Essays on Classical Antiquity in Honor of Joy K. King (Wauconda 1993), 279–304Google Scholar, particularly 287, where it is argued that the mistress as inspiring Muse fulfils the role of patron which the avowed patrons (Tullus, Maecenas) cannot play because of the nature of the genre. While I am not convinced that the functions of inspiration and patronage are so closely aligned, the reading may contribute to our understanding of the generic significance of the beloved as inspiring force.

49. See Wyke (n.32 above), 121–24, on the puella speaking, and also Hallett, J.P., ‘The Role of Women in Roman Elegy: Counter-cultural Feminism’, Arethusa 6 (1973), 103–24Google Scholar.

50. See chapters 3 and 4 of Sharrock (n.15 above) for more discussion of Callimacheanism and the rhetoric of the disclaimer.

51. On the recusatio see Wimmel, W., Kallimachos in Rom: Die Nachfolge seines apologetischen Dichtens in der Augusteerzeit (Wiesbaden 1960)Google Scholar, Hardie, P.R., Virgil’s Aeneid: Cosmos and Imperium (Oxford 1986), 86fGoogle Scholar.

52. See Conte, G.B., The Rhetoric of Imitation: Genre and Poetic Memory in Virgil and other Latin Poets, ed. Segal, C.P. (Ithaca and London 1986), particularly 32–39Google Scholar.

53. Cf. Virg. Ecl. 7.19 (alternos Musae meminisse uolebant, ‘the Muses desired to recall alternate [song]’), Aen. 7.645 (et meminisse enim, diuae [sc. Musae], et memorare potestis, ‘you are able, goddesses, to remember and recount’). After Virgil’s explicit expressions, a reference to memory can easily remind us of the power of the Muses. As the quotation from Aen. 7 shows, the link with memorare is an important element in the connection between memory and poetry. It is particularly evident in the self-expression of story-tellers in the Metamorphoses and Fasti.

54. See particularly Ecl. 9.38, 45, 52. See Alpers, P., The Singer of the Eclogues: A Study of Virgilian Pastoral (Berkeley & Los Angeles 1979)Google Scholar, particularly 142f. Ovid’s exile poetry is also full of ‘memory’.

55. The poet of the AA, posing as the lover speaking from experience, tells us that all he did in the earlier poem was to mess up the beloved’s hair, and his absurd self-degradation was just elegiac excess, or a pose (or both), but also that the girl used his minor offence to gain power over him, by claiming damages. So be warned. Cf. also AA 2.551, 3.659.

56. For the identification of Gallus see Ross, D.O., Backgrounds to Augustan Poetry: Gallus, Elegy and Rome (Cambridge 1975), 83fGoogle Scholar., and Sharrock, A.R., ‘Alternae Voces—Again’, CQ 40 (1990), 570fCrossRefGoogle Scholar.

57. This is not quite the same as, but does have affinities with the type of ‘euphemism’ which replaces obscene words with others directly signifying the same thing, like ‘fanny’ or ‘holy of holies’ for the female genitalia. Is the taboo word avoided—or not?

58. In both cases, of course, the words are in the ablative. For angustus as a Calli-machean/elegiac term cf. also Prop. 2.34.43. A few lines after the Callimachean angustum pectus, Propertius talks of himself and Cynthia angusto uersantes proelia lecto (2.1.45). As in Ovid’s case, the narrowness of the bed is programmatic. The Callimachean import of the narrow bed is noted also by K. McNamee, ‘Propertius, Poetry, and Love’, in DeForest (n.48 above), 215–48, at 232f.

59. See chapter 4 of Sharrock (n.15 above) for the interplays of big and little in Ovidian programme.

60. McNamee (n.58 above), in a thorough-going analysis of love in the Propertian monobiblos as a metaphor for poetic composition, also makes the specific connection between love as poetry and a statement of Callimachean allegiance. My own reading, while appreciating the validity of her metapoetic examples, would seek to retain in addition an important role in elegy for examination of issues of gender and human relationships.

61. Our ideas about the ancient Roman male’s self-image and psyche are challenged by Barton, C.A., The Sorrows of the Ancient Romans: The Gladiator and the Monster (Princeton 1993)Google Scholar, even if one is uncertain (as I am) about the validity of some of her propositions and examples. Hallett, J.P., ‘Women as Same and Other in Classical Roman Elite’, Helios 16 (1989), 59–78Google Scholar, has shown (tangentially to her main argument) that while it was sometimes socially acceptable for women to exhibit, or to be said to exhibit, the characteristics associated with men (usually their fathers), it was never acceptable for men to display those associated with women. On the problems of finding male identity in Roman society, see Skinner (n.35 above), 121, where the acquisition of male gender is argued to be a process which takes considerable time and requires continual vigilance: ‘The Roman compulsion to reappropriate the outlawed “feminine”, however spuriously, offers crucial insights into the destructive effects on the male personality, first, of polarized gender roles and, second, of conflating sexuality and power.’

62. On double entendre in the discourses of sex and rhetoric see Freudenburg, K., The Walking Muse: Horace on the Theory of Satire (Princeton 1993), 164fGoogle Scholar., and ‘Horace’s Satiric Program and the Language of Contemporary Theory in Satires 2.1, AJP 111 (1990), 187–203.

63. See Am. 1.1.19 and Kennedy (n.2 above), 59.

64. Cf. also Ov. Pont. 2.5.69: meis numeris tua dat facundia neruos. nerui may also gain poetic significance from reference to the strings of a lyre, as at Virg. Aen. 9.776, Prop. 3.3.4 and 35. It is not of course relevant to these examples that ‘“stringing your bow” seems to be a colloquial way in Latin of saying “have an erection” ’ (Kennedy [n.2 above], 59)…

65. Cf. e.g. Sen. Con. l.pr.8, Tac. Hist. 2.76, Plin. Pan. 18.3.

66. Adams (n.7 above), 49, says that latus is sometimes suggestive of the male genitalia but is too general simply to be a substitute for it. Cf. e.g. Cat. 6.13, Hor. Epod. 11.22, Ov. Am. 2.10.55, 3.11.14, AA 2.413, Priap. 26.11, 68.34.

67. Cf. Cic. Ver. 2.52, me dies, uox, latera deficiunt, si hoc nunc uociferari uelim; 4.67, quae uox, quae latera, quae uires huius criminis querimoniam possunt sustinere?; de Orat. 1.114, linguae solutio, uocis sonus, latera, uires.

68. See Kennedy (n.2 above), 58–63, 83.

69. I have not been able to consult Helios 17.2 (Autumn 1990), which contains discussions by a number of feminist classical scholars of Phyllis Culham’s ‘Decentering the Text: The Case of Ovid’, the opening essay in that number.

70. Gamel, M.-K., ‘Non sine caede: Abortion Politics and Poetics in Ovid’s Amores’, Helios 16 (1989), 183–206Google Scholar.

71. It would be fair to say that since Gamel is not primarily interested in authorial intention and authority, my concerns are perhaps tangential to her argument, but I want to assert the power of difference in reading, and the impotence of authors to control meaning.

72. See Wyke, M., ‘Mistress and Metaphor in Augustan Elegy’, Helios 16 (1989), 25–47Google Scholar, at 42.

73. Wyke (n.32 above).

74. Gold (n.48 above), 288, speaks of the pretence of turning gender-roles upside down; cf. also Hallett (n.49 above), who rightly sees counter-cultural effects in the male elegist taking on female roles. But I suggest it is more confused and problematical than that. If I may be allowed a generalisation, there is perhaps a tendency for the group of American scholars who have featured in this discussion to offer more fixed and, so to speak, ‘powerful’ readings of their texts, while the British are more reticent and more concerned with difference. But this suggestion is open to challenge. For an analysis of similar issues but structured around the difference between famous French and Anglo-American feminist thinkers, see Gold, B.K., ‘ “But Ariadne was Never There in the First Place”: Finding the Female in Roman Poetry’, in Rabinowitz, N.S. and Richlin, A. (eds.), Feminist Theory and the Classics (London and New York 1993), 75–101Google Scholar.

73. I am very grateful to Don Fowler, Duncan Kennedy and James Zetzel for very helpfully criticising a draft of this essay.