¹ Cf. Ker, W. C. A. in his Loeb edition (rev. 1968), i. xiGoogle Scholar; Friedlaender, L., Martialis Epigrammaton Libri (Leipzig, 1886), p. 16Google Scholar; Bridge, R. T. and Lake, E. D. C., Select Epigrams of Martial, Books I–VI (Oxford, 1908), pp. x–xiGoogle Scholar; Izaac, H. J. in the Budé ed. (Paris, 1930), intro. p. xxiiiGoogle Scholar, who also quotes Butler, H. E., Post-Augustan Poetry, p. 274Google Scholar: ‘M. was a child-lover before he was a man of letters’; Simcox, G. A., A History of Latin Literature (New York, 1883), ii. 112Google Scholar ‘Martial stands almost alone in Roman literature in his appreciation of mere girlhood.’

² E.g. W. C. A. Ker (above, n. 1), index s.v. Erotion; Lessing, G. E., Sämtliche Schriften³, ed. Lachmann–Muncker, (1895), xi. 242Google Scholar, quoted by Kenney (below, n. 3), p. 78 n. 2; Johnson, S., ‘The Obituary Epigrams of Martial’, CJ 49 (1953–1954), 268.Google Scholar

³ Kenney, E. J., ‘Erotion Again’, G & R 11 (1964), 77–81Google Scholar, Lloyd, L. J., ‘Erotion: a Note on Martial’, G & R 22 (1953), 39–41Google Scholar. Lloyd (p. 40) finds the sentiments in 1–17 insincere, so that the ending of the poem is not too great a let-down: ‘where sentiment is lacking sarcasm will do’, Kenney, though regarding the poet's grief for Erotion as genuine, feels that he has this grief in perspective, and so is able to mirror playfully in his language the baby-talk in which he used to engage with the little girl.

⁴ Line 4 ‘cui nee lapillos praeferas…’ might possibly, though not necessarily, suggest a following verb in the 3rd person.

⁵ Cf. Longus, , Daphnis and Chloe 1.16, 24fGoogle Scholar.; Theoc. 10.26ff., 11.20ff.; Virg. Ecl. 7.37f.; Catull. 17.15f.; Ov. Met. 13.789ff.; A.P. 5.48 (also a list of comparisons ending in a punch line), 121, 118, 270.3; Misener, G., ‘Iconistic Portraits’, CP 19 (1924), 97ffGoogle Scholar., esp. pp. 122–3. For comparisons as a feature of the literary language of love, see Nisbet, R. G. M. and Hubbard, M., A Commentary on Horace, Odes Book I (Oxford, 1970)Google Scholar on Od. 1.19.6 and Citroni, M., M. Valerü Martialis Epigrammaton Liber I (Florence, 1975), pp. 336fGoogle Scholar. (on Mart. 1.109.1).

⁶ Cf. 10.67.6 ‘sita prurit in sepulchro’ though this is obviously parodic. Contrast Statius, Silv. 2.1.2, where the same idea occurs, but in a serious epikedion: the words ‘adhuc vivente favilla’ are to be taken in a straightforward way.

⁷ Above, n. 3, p. 40.

⁸ Above, n. 3, p. 79.

⁹ As evidence Kenney cites the comparison ‘concha Lucrini delicatior stagni’ (line 3) which, in his opinion, carries the implication that Erotion is good enough to eat – a further clue that the words are addressed to a child. Although he is certainly right in taking this as a reference to oysters, for which the Lucrine lake was famous (cf. Pliny, N.H. 9.168), rather than to pearls, which according to Pliny (N.H. 9.106) came from the Indian Ocean and Arabia, I cannot agree with his conclusion that, given Lucrine oysters were a delicacy, the comparison suggests that the girl is good enough to eat. It is more likely to be a play on two senses of delicatus: a delicacy (in the gastronomic sphere) and a puella delicata (on which see further below: this is another example of a phenomenon to be discussed later, whereby the unexpected revelation that the poet's puella is very young is anticipated by the choice of language). Kenney might have also pointed to the nitella = a type of dormouse (Plin. N.H. 8.224 distinguishes between the glis and the nitella, though they are obviously closed related, both hibernating in winter): dormice were a delicacy on the Roman table (Plin. loc. cit., Varro, R.R. 3.15, Apicius, 8.9, Petr. 31, Toynbee, J. M. C., Animals in Roman Life and Art (London, 1973), p. 204 n. 22).Google Scholar

¹⁰ Examples of poems where this technique is used include 1.41, 109.1–5, 115; 3.65; 4.13; 8.33; 9.57; 11.18, 21, 84. See also Citroni, op. cit., p. 130 on Mart. 1.41 intr. This does not necessarily mean that the lists are non-serious in themselves (e.g. in 4.13, a wedding poem, the felicity of the pair is expressed through a series of conventional comparisons; although there is a witty twist at the end, this does not detract from the sincerity of what has gone before).

¹¹ Cf. Varro, R.R. 3.6.2, Plin. N.H. 10.43, Ov. Met. 13.802, Sauvage, , Etude de thèmes animaliers dans la poésie latine (Collection Latomus vol. 143, Brussels, 1975), p. 269.Google Scholar

¹² 17 qualified, 14 unqualified.

¹³ Of 16 poems where a series of comparisons is used, the only clear case where unqualified examples are more numerous is 1.115.

¹⁴ On the hyperbolic nature of qualified comparative expressions and their use in parodic contexts, cf. Citroni on Mart. 1.115 ‘loto candidior cycno’, though in this example the application of lotus to a swan seems amusing in itself.

¹⁵ Like many of the effects in the poem it is only fully appreciated in retrospect, or at a second reading.

¹⁶ For a full discussion of the phrase, see appendix.

¹⁷ The procedure is, of course, assisted by the ambiguity of the Latin word puella which can be used equally appropriately of a very young child or (in love poetry) of the poet's mistress.

¹⁸ See TLL 2194.34ff. s.v. dulcis.

¹⁹ Cf. Plato, Phaed. 84e–85; Nisbet and Hubbard on Hor. Od. 2.20.10; Arnott, G., ‘Swan Songs’, G&R 24 (1977), 149–53Google Scholar; Sauvage, above n. 11, pp. 237–9.

²⁰ Servius on Georg. 1.181 identifies it with Virgil's exiguus mus: though Virgil is probably referring to mice or rodents in general (see R. Mynors in his recent commentary [Oxford, 1990]), the fact that Servius makes the identification indicates that he thought of the nitela as small.

²¹ Perhaps these little creatures were kept as pets: so Toynbee, above, n. 9, p. 293, a suggestion based on the use of amabilis in this passage; cf. also Keller, O., Die Antike Tierwelt (Leipzig, 1909–1913), p. 181Google Scholar. It might also be relevant that elsewhere when young girls (i.e. those just approaching puberty) are compared to animals it is to those of the larger sort such as heifers and fillies: see Nisbet and Hubbard on Hor. Od. 2.5 intro.

²² Cf. Citroni, op. cit., p. 352, on Mart. 1.115.3. Moreover, the phrase ‘non tactum’ suggests she will one day be tacta: see my discussion on the sexual resonances of the relationship between Martial and Erotion, see below.

²³ Cf. Ov. Am. 2.5.39f. (‘…ne longis flavescere possit ab annis./Maeonis Assyrium femina tinxit ebur’); cf. Plin. N.H. 8.7 (the whiteness of an elephant's tusks is an indication of its youth).

²⁴ In an age lacking sophisticated dental care, the freshness of the breath must have decreased with age and increasing tooth decay, so that really sweet breath would be a sign of youth. Cf. A.P. 5.118 where the breath of a youthful girl is said to be sweeter than spikenard; Long. Daphnis and Chloe 1.25; Mart. 3.65.1: the fragrance of the boy's kisses is likened to ‘quod spirat tenera malum mordente puella’.

²⁵ Compare for example Prop. 4.5.59f. ‘vidi ego odorati victura rosaria Paesti/sub matutino cocta iacere noto’ and see K. F. Smith on Tibullus 1.4.29, Nisbet and Hubbard on Hor. Od. 2.3.13, etc. Roses were also connected with fleeting beauty and early death: cf. Van Dam, H.-J., P. Papinius Statius Silvae Book 2: A Commentary (Leiden, 1984), p. 122Google Scholar, on Stat. 2.1.106–9.

²⁶ Mart. 3.65.5, 11.8.6; Juv. 6.573f., 11.50f.; Ov. Met. 2.365f.; cf. Plin. N.H. 37.30, 50–1; it has this smell because it is composed of fossilised pine resin: see Plin. N.H. 37.42 with Eicholz, D. E. (Loeb edn, 1962)Google Scholarad loc. See also P. Watson, ‘Balls of Crystal and Amber: Fact or Fiction?’ in LCM forthcoming.

²⁷ Perhaps one might also see in the use of rapta a reference to the ‘cito rapta’ formula common in tomb inscriptions for those who die young.

²⁸ Other examples include 1.53, 3.93, 4.14, 8.33, 9.57, 11.21 and 11.84. Another poem where the name of the subject (Diadumenus) is delayed is 3.65, but there the surprise is not as great because the comparisons are appropriate to a puer delicatus. And at 11.8 the punch line revolves round Martial's unwillingness to reveal the boy's name. In 1.109, although the name is given at the beginning, we do not find out Issa's real identity until line 5.

²⁹ Pace Bell, A. A. Jr. (‘Martial's Daughter?’, CW 78 (1984), 21–4)Google Scholar who suggests that Erotion was Martial's daughter by one of his slave women: his argument is based partly on the idea that the sentiments expressed about Erotion are too strongly felt to be merely those of a master for a slave.

³⁰ On Martial's sexual attitudes as generally reflecting the norm of society (apart from his own apparent preference for boys over women) see Sullivan, J. P., ‘Martial's Sexual Attitudes’, Philologus 123 (1979), 288–302.CrossRefGoogle Scholar

³¹ See Mohr, J. W., ‘Age Structures in Pedophilia’ in Adult Sexual Interest in Children, ed. Cook, M. and Howells, K. (London, 1981), p. 44Google Scholar; K. Plummer, ‘Constructing a Sociological Baseline’, ibid. p. 225.

³² Though the distinction was made by Freud, most modern studies of the subject have concentrated on the deviant form. K. Howells emphasises that paedophilia proper represents only a small proportion of adult–child sexual relations (‘Adult Sexual Interest in Children: Considerations Relevant to Theories of Aetiology’ in Cook and Howells, above, n. 31, pp. 55–94).

³³ K. Plummer, above, n. 31, pp. 236–8; K. Howells, above, n. 32, pp. 79–80. Freund, K., McKnight, C. K., Langevin, R. and Cibiri, S., ‘The Female Child as Surrogate Object’, Arch. Sex. Behav. 2 (1972), 119–33CrossRefGoogle ScholarPubMed, found from experiments that even subjects classed as sexually normal show some degree of erotic response to female children from the age of five upwards. Such feelings are sometimes even admitted to: cf. Evelyn Waugh's remark to Ann Fleming (I September 1952): ‘My sexual passion for my ten-year-old daughter [Margaret] is obsessive. I wonder if you'll come to feel like this about your son. I can't keep my hands off her’ (Letters, ed. Amory, M. (Penguin, London, 1980), p. 380)Google Scholar. (I would like to thank the CQ referee for the preceding quotation.)

³⁴ See Herman, J. L., Father–daughter Incest (Cambridge, Mass./London, 1981), pp. 12f.Google ScholarPubMed: results of surveys by Kinsey and others show ⅕–⅓ of all women reported some sort of childhood sexual encounter with an adult male; these figures may be lower than average because only white middle-class women were used as subjects.

³⁵ Summit, R. and Kryso, J., ‘Sexual Abuse of Children: A Clinical Spectrum’, Amer. J. Orthopsychiat. 47 (1977), 237–51.Google Scholar

³⁶ Mohr, J., Turner, R. and Jerry, M., Pedophilia and Exhibitionism (Toronto, 1964), p. 34Google Scholar; K. Plummer, above, n. 31, p. 226. For a contrary feminist view that children do not enjoy such encounters see Herman, above, n. 34, p. 28.

³⁷ Plummer, above, n. 31, pp. 229–30.

³⁸ Ariès, P., Centuries of Childhood (tr. R. Baldick, 1962Google Scholar; reprinted London, 1973), pp. 98ff.

³⁹ Ibid. p. 101.

⁴⁰ It might be argued that the Romans felt the same sort of sentimental attachment towards their children as we do, and would regard sexual play with a young girl with the same abhorrence. Though this may well have been true in the case of freeborn virgines, a distinction must be made between freeborn children and those of servile status.

⁴¹ Compare Patterson, O., The Sociology of Slavery (London, 1967), p. 42Google Scholar ‘the sexual exploitation of female slaves by white men was the most disgraceful and iniquitous aspect of Jamaican slave society. Rape and seduction of infant slaves; the raping of the common law wives of the male slaves…were the order of the day’; Finley, M. I., Ancient Slavery and Modern Ideology (London, 1980), pp. 95fGoogle Scholar. ‘…unrestricted availability in sexual relations. This is treated as a common-place in Graeco-Roman literature from Homer on; only modern writers have managed largely to ignore it’ (he lists some references but not to very young girls).

⁴² See e.g. Schmidt, J., Vie et Mort des Esclaves dans la Rome antique (Paris, 1973), pp. 61f.Google Scholar; Bradley, K. R., Slaves and Masters in the Roman Empire (Brussels, 1984), p. 118Google Scholar; Finley, above, n. 41; Courtney, E., A Commentary on the Satires of Juvenal (London, 1980)Google Scholar on Juv. 2.57; Vogt, J., Ancient Slavery and the Ideal of Man (tr. by Wiedemann, T., Oxford, 1974), p. 7Google Scholar; Nisbet and Hubbard on Hor. Od. 2.4 intro. p. 67.

⁴³ Slater, W. J., ‘Pueri, Turba Minuta’, BICS 21 (1974), 133–40.Google Scholar

⁴⁴ H.-J. Van Dam, above, n. 25, p. 73. Prescott, H. W., ‘Inorganic Rôles in Roman Comedy’, CP 15 (1920), 261Google Scholar, seems to identity delicia and delicati: ‘pueri delicati were a special type of handsome young voluptuaries…on intimate terms with their masters…conventionalized as pert youngsters furnishing entertainment with their saucy wit.’ An example is Paegnium in Persa who provides patter talk to pad out Act 2. In Comedy, then, the functions of delicium and delicatus are combined. Note the absence in Comedy of delicatae: inscriptional evidence provides a different picture (see further below).

⁴⁵ Herodian 1.17, Cassius Dio 67.15. Slater (above, n. 43) does, however, make mention of Ausonius' Bissula, and Mau, RE s.v. deliciae, includes Martial's Erotion, though neither Slater nor Van Dam mention her (Slater gives Erotion as the name of a delicium but without referring to Martial specifically).

⁴⁶ Via the computer concordance compiled by E. Jory and D. G. Moore (1974), published as CIL vol 6 part 7.

⁴⁷ I.e. deliciae (the literary equivalent of delicium) is used for delicatus in a sexual sense, e.g. Virg. Ecl. 2.2, 9.22, Plaut. Most. 15, Pers. 204.

⁴⁸ Conversely, the term delicium is occasionally applied to a person in their teens or early twenties (in the case of females, one is 23, four more between 13 and 17; of males one is 22, two are in their mid-teens), though in these cases the word could conceivably describe a function for which they were remembered but which they have outgrown.

⁴⁹ Male masters have more or less the same number of male as female delicia (25 female, 26 male); this contrasts with female owners of delicia who prefer female children (26 are by females for females, but only 12 for males by females). Though this last fact might suggest that there was no sexual element, it could equally well be that it was because of the possible sexual element that it was not thought as seemly for women to keep male (as opposed to female) delicia.

⁵⁰ Cf. (in the prose preface) ‘obscuritas’ … ‘arcana securitate’ … ‘verecundia’ … ‘erubesco’ and in the second epigram ‘pone supercilium’ (line 2) etc.

⁵¹ E.g. Ov. Met. 4.157 (Pyramus and Thisbe), Propertius 4.7.94. For a Greek epitaph on a young slave girl couched in erotically-tinged language, cf. Crinagoras' epigram on the 9-year-old Hymnis (A.P. 7.643) which begins Ὑμνίδα τήν Εὐάνδρου, ράσμιον αἰν ἄθυρμα|οἰκογενές, κούρον αἱμύλον εἰναέτιν.

⁵² Occasionally it is found as a man's name e.g. M. Erotio, a priest. It is not listed by Baumgart, J., Die römischen Sklavennamen (Diss. Breslau, 1936), p. 68Google Scholar among slave names at Rome, though this form of Greek diminutive in -ion (written in Latin -ium/io) is common e.g. Philematio/ium, Asterio, etc. It is found in first-century A.D. Greek inscriptions of slaves (see Reilly, L. C., ‘Slaves in Ancient Greece’, AE 1905, p. 196Google Scholar no. 11 and AE 1917, no. 319, p. 35).

⁵³ It is also a freeborn name. For other examples, see Fraser, P. M. and Matthews, E., A Lexicon of Greek Personal Names, Vol. 1 (Oxford, 1987), p. 168.Google Scholar

⁵⁴ J. Baumgart, loc. cit.; Plautus uses the ending -ium of women (mainly meretrices) and -io of men (except two who are pueri delicati).

⁵⁵ Slater, above, n. 43, p. 137 lists several names of delicia which are Greek: Euphrosyne, Thallusa, Methe, Apate etc. Cf. Lucretius' list of pet names at 4.1160ff. – mostly Greek forms are given, though note that two proper names (Palladium and Lampadium) are in the Latin spelling. Were delicia given Greek names as nicknames? All this assumes that Erotion is a real person: it is not impossible that the poems about her are jeux d'esprit, but I would not wish to argue this. On this point, see further below.

⁵⁶ Cf. Hor. Od. 2.5 and A.P. 5.124. Iff. on a young girl who is not quite ripe for sexual encounters.

⁵⁷ Above, n. 42, p. 118.

⁵⁸ See Lattimore, R., Themes in Greek and Latin Epitaphs (Urbana, 1962), pp. 295–7.Google Scholar

⁵⁹ Cf. the parodic Ov. Am. 2.6.3 ‘plangite pectora pinnis’ (addressed to the birds mourning the deceased parrot).

⁶⁰ For descriptions of women on epitaphs, see Lattimore, above, n. 58, pp. 275–80 and 295–7. There are no parallels in the CIL 6 index for any of these adjectives, with the sole exception of nobilis in CE 1136.3–4 (= 6.9693) cited earlier. There opulenta is also used. In the sense ‘of noble birth,’ nobilis is normally applied (in the classical period at least) only to men: in the case of women it carries only its basic sense ‘renowned’. See T. Hillard (whom I thank for this reference) in Garlick, B. et al. , Stereotype Attitudes Towards Women in Power (Greenwood Press, forthcoming), n. 22, pp. 58–9.Google Scholar

⁶¹ I.e. in both Stoic (prose) consolationes and also on real epitaphs (see Lattimore, above, n. 58, p. 217 for examples of the theme ‘you should stop lamenting’. Perhaps the tombstone element is uppermost since, as we have seen, in listing the virtues of his wife Paetus sounds like a tombstone inscription.

⁶² Cf. Sen. Ad Marc. 1.1, Ad Polyb. 4.1; Cons, ad Liv. 7f., 353f., 467ff.; Lattimore, above, n. 58, p. 219. Contrast the poetic epikedion, where excessive grief is the norm, cf. Van Dam, above, n. 25, p. 152.

⁶³ Cf. Sen. Ad Marc. 2.2, 12.4, 13.3, etc.; Ad Polyb. 14.3ff.; Cons, ad Liv. 59ff.

⁶⁴ See Manning, C. E., On Seneca's ‘Ad Marciam’ (Leiden, 1981), p. 42Google Scholar; cf. Seneca, Ad Polyb. 18.6, Ep. 63.1–3, 13–14, 98.18, 99.1–2, 15–21, Dial. 6.2, 11.4.3; also Hor. Od. 2.9.17–18 (with Nisbet and Hubbard ad loc).

⁶⁵ Cf. Austin on Aen. 6.220; Prop. 1.16.13 has it in the sense ‘weep to exhaustion’ and Ovid uses it of Orpheus grieving over Eurydice at Met. 10.12.

⁶⁶ Cf. CE 59.12–17; Lattimore, above, n. 58, p. 203 points out that a common theme on tombstones is the wish not to go on living: cf. Cons, ad Liv. 323 where the widow prays for death; Sen. ad Marc. 3.3 (giving in to grief shows loss of desire to live). This is also a topos of the literary epikedion: cf. Van Dam, above, n. 25, pp. 86f, 154 and on Silv. 2.1.24–5.

⁶⁷ I am not suggesting that there is no genuine sentiment behind the first section of the poem or that Martial's elaborate catalogue of Erotion's attractions is no more than an exercise in hyperbole. As Kenney rightly noted, Martial's genuine feelings provide a foil for Paetus' lack of sincerity. I am merely attempting to explain why the poet's grief is expressed in such hyperbolic terms.

⁶⁸ Above, n. 3, p. 80.

⁶⁹ Above, n. 29.

⁷⁰ For instance Martial's epigrams 1.88 and 101 and also the third poem on Erotion which – apart from the expression ‘exiguis manibus’ – could stand as a real grave inscription; cf. also Catullus' poem (101) on his brother's death.

⁷¹ A.P. 7.461; cf. 204, an epigram of later date for a pet bird.

⁷² On epitaphs for slaves as showing genuine affection see Kay, N. M., Martial Book XI: A Commentary (London, 1985)Google Scholar, on Mart. 11.91.

⁷³ It is no coincidence that the Latin word deliciae is used of both animals and slaves.

⁷⁴ Above, n. 3, p. 40. Howell, P., A Commentary on Book One of the Epigrams of Martial (London, 1980)Google Scholar on Mart. 1.115 views ‘lotis cycnis’ as humorous hyperbole and compares Kenney's remark on ‘senibus cycnis’.

⁷⁵ Above, n. 3, p. 79.

⁷⁶ He suggests that Martial either meant it to be amusing or else was being unwittingly funny through ignorance of this fact – but it seems scarcely credible that M. could be unaware of such a common topos. There is conceivably a reference to the fact that young swans (up to the second and occasionally the third year) have varying degrees of grey among their white feathers (cf. Peter, Scott, The Swans (London, 1972), p. 23)Google Scholar so that pure white plumage is associated with older (i.e. old) birds.

⁷⁷ E.g. Aristoph, . Wasps 1063ffGoogle Scholar. νν|δ᾽ οἴχεται, κύκνου τε πολιώτεραι δή|αἱ δ᾽ πανθοσιν τρίχες, Eur. Bacc. 1365, Ov. Tristia 4.8.1 ‘iam mea eyeneas imitantur tempora plumas’, Mart. 3.43.2 (of a man who dyes his hair) ‘tam subito corvus, qui modo eyenus eras’).

⁷⁸ For both ideas at once compare Eur. Her. 108f. στάλην|ἰηλέμων γόων [γέρων Nauck] οι|δς ὣστε πολις ρνις.

⁷⁹ The whiteness of the girl's complexion is referred to later in the poem (surely this must be the implication of the images in lines 4–6) but it would be suggested also to the reader by the reference to swans, in view of the fact that the whiteness of swans is commonly used in Latin poetry to evoke feminine beauty: see A. Sauvage, above, n. 11, pp. 232–4. For the sweetness to Martial of her voice, compare poem 34. Translations like that in Ker's Loeb (‘a maid, sweetervoiced to me than aged swans’) and the Budé edition of H. J. Isaac (‘fillette à la voix plus harmonieux à mes oreilles que le dernier chant du cygne’) are inadequate, since the implication of whiteness is missed and dulcis by itself cannot mean ‘sweeter voiced’.

⁸⁰ Cf. also Nemes. Cyn. 314. I assume Statius is borrowing from Martial rather than the reverse: the Thebaid was published in 90–1, whereas Martial's fifth book can be dated to 89 in view of several references to an edict of Domitian of that year (e.g. poems 8, 14, 23) and to the victory over Dacia of the same year (see poem 3).

⁸¹ Cf. Lucan 5.739 ‘vita mihi dulcior’ [sc. coniunx].

⁸² On the singing of swans see especially G. Arnott, above, n. 19, pp. 149–53. The presence of the idea here is confirmed by the fact that when Martial uses the phrase ‘senes cycni’ at 9.42.2 the context is about song (Apollo); cf. also Statius, Theb. 5.341 referred to above. The parallel with Eur. Her. 109f. (see n. 78 above) is also relevant. On the sweetness of the swan's voice, cf. also Isid. Orig. 12.7.18 ‘olor est avis, quam Graeci kuknon vocant…a canendo est appellatus, eo quod carminis dulcedinem modulatis vocibus fundit’.

⁸³ I would like to thank Dr L. Watson and Professor D. Konstan for help and encouragement; also the anonymous CQ referee.