¹ The problematic nature of the modern word ‘witch’ must be acknowledged here. The English term ‘witch’ conveys only a rough sense of a woman's possession of supernatural abilities and is an imprecise label for determining a character's nature, disposition, or motivation. One need only consider the disparity between Theocritus' Simaetha and Lucan's Erictho to get a sense of the term's broad scope. This polyvalence of the term ‘witch’ is also found in the Italian strega (see Barchiesi, A., ‘Poetica di un mito sessuale: la strega giambica’, in Vicende e figure femminili in Grecia e a Roma: atti del convegno Pesaro [Ancona, 1995], 335–42)Google Scholar; the Spanish maga/bruja (Meléndez-Valdés, P.M., ‘Sobre las magas romanas’, in Peláez, J. [ed.], El dios que hechiza y encanta: magia y astrología en el mundo clásico y helenístico: Actas del I Congreso Nacional, Córdoba 1998 [Cordoba, 2002], 233–43)Google Scholar; the French magicienne/sorcière (Tupet, A.-M., La Magie dans la poésie latine: des origines à la fin du règne d'Auguste [Paris, 1976]Google Scholar); and the German Hexe (Luck, G., Hexen und Zauberei in der römischen Dichtung [Zürich, 1962]Google Scholar). Although this terminology is problematic, it will be necessary to use the very word ‘witch’ repeatedly to avoid an overuse of awkward periphrases. In each instance, ‘witch’ will refer to a female character to whom a variety of powers have been ascribed. These powers may include the ability to effect reversals of natural phenomena and changes in interpersonal relationships or the knowledge and/or use of materia magica (effigies, rhombi, defixiones, hippomanes, etc.), efficacious language (carmina, voces magicae), or gestures (the evil eye). These women will also be characters whom multiple scholars have repeatedly and consistently labelled as witches, e.g. Circe, Medea, Deianira, Simaetha, the anonymous practitioner of Virgil's eighth Eclogue, Canidia, Dipsas, Acanthis, Meroe, Pamphile, Petronius' strigae, Martina, Erictho.

² See Burris, E.E., ‘The terminology of witchcraft’, CPh 31 (1936), 137–45Google Scholar, for an extensive survey of Latin descriptive terms for magical practices, as well as those applied to male and female magical practitioners. Although I depart from his generalizations regarding the deployment of these terms, the assembled material is an invaluable catalogue of the Latin witch vocabulary.

³ The dual meaning of prae as both a temporal and spatial designation complicates this interpretation, however, and is an indication of the problems attendant upon appeals to etymological explanations.

⁴ Johnston, S.I., Restless Dead (Berkeley, CA, 1999), 161–99Google Scholar.

⁵ The general imprecision of Latin (or Greek) terms for magic and magical practitioners has been a phenomenon widely observed, though rarely with prolonged attention to the Roman witch. See Dickie, M., Magic and Magicians in the Greco-Roman World (New York, 2001), 12–17Google Scholar, 124–41; Faraone, C., Ancient Greek Love Magic (Cambridge, 1999), 15–30Google Scholar; Graf, F., Magic in the Ancient World (Cambridge, 1997), 36–60Google Scholar.

⁶ The lamia of Hor. Ars 340 is glossed by Porphyrion as a being invoked to terrify children, and in his comments on lama at Hor. Ep. 1.13.10 he explains the etymology of lamia by stating that, just as bogs (lamae) typically swallow things down (ingluvies), lamiae are so named because they are devourers of children (devoratrices puerorum).

⁷ Cicero, for instance, often refers to the charge of poisoning levelled against his client Cluentius as venefici crimen (Clu. 166), and Tacitus describes the famed poisoner Martina as both infamem veneficiis (Ann. 2.74.5) and famosam veneficiis (3.7.6).

⁸ Graf (n. 5), 46–9, outlines the evolution and double nature of the term in republican Rome and later examines epigraphic testimonia for the untimely dead by way of veneficia/pharmaka, in Graf, F., ‘Untimely death, witchcraft, and divine vengeance: a reasoned epigraphical catalog’, ZPE 162 (2007), 139–50Google Scholar.

⁹ Cic. Div. 1.65: sagire enim sentire acute est; ex quo sagae anus, quia multa scire volunt, et sagaces dicti canes. is igitur, qui ante sagit, quam oblata res est, dicitur praesagire, id est futura ante sentire.

¹⁰ Col. 1.8.6: haruspices sagasque, quae utraque genera vana superstitione rudis animos ad inpensas ac deinceps ad flagitia conpellunt, ne admiserit.

¹¹ There is a similar situation at Tib. 1.5.9–16, where an anus is present and incanting (praecinuisset, 12) as the poet himself acts as a pseudo dream purifier (procuravi ne possent saeva nocere | somnia, 13–14). There is also a tendency to extrapolate from drunken witches such as those in Martial a wider trend towards inebriation. While a demonstrable and long-lived literary trope of old women and their love for wine exists (Rosivach, V., ‘Anus: some older women in Latin literature’, CW 88 [1994], 107–17Google Scholar, at 113–14), this occasional overlap between older women and drunken witches has collapsed into a false assumption of drunkenness as a prevalent attribute among Roman witches. Pollard, E.A., ‘Witch-crafting in Roman literature and art: new thoughts on an old image’, Magic, Ritual and Witchcraft 3 (2008), 119–55CrossRefGoogle Scholar, demonstrates this anachronistic tendency to associate drunken old women with witches in her description of a particular Anus Ebria statue from the Capitoline Museum: based on the figure's advanced age and possession of a wine jug, she posits that the old woman may have been a witch.

¹² Fest. p. 213.23–6 Lindsay. Curiously, the single other extant use of simulatrix is found in Statius (Theb. 4.551) to describe Circe, where the term evokes not purificatory abilities but her practice of transforming or ‘shifting’ her guests into animals.

¹³ It is likely that it is this dreadful aspect of the saga that is captured on a child's epitaph, apparently stolen (and presumably killed) by a saga's hand (CIL 6.19747). While this interpretation of the epitaph is commonly accepted (see Ogden, D., Magic, Witchcraft and Ghosts in the Greek and Roman Worlds [New York, 2002], 119Google Scholar; Lefkowitz, M. and Fant, M. [edd.], Women's Life in Greece and Rome [Baltimore, MD, 2005], 294)Google Scholar, one must note that the inscription technically leaves the malefactor's gender ambiguous (eripuit me saga manus) and thus permits a reading of a masculine or feminine agent behind the child's disappearance. That said, the comparison (suggested by Buecheler, F. and Riese, A., Anthologia Latina, sive Poesis Latinae Supplementum [Leipzig, 1897], 456)Google Scholar of the phrase saga manus in the inscription to the mala manus of Petron. Sat. 63.8 may be indicative of implied feminine agency in the inscription, since the relevant scene in the Satyrica deploys the phrase mala manus in the context of the covert abduction of a child by supernatural, feminine forces (strigae, mulieres plussciae).

¹⁴ This type of professional practitioner of erotic magic qua female pimp (lena) is one of the most easily recognized and therefore affords a generic (though mistaken) association of witches and prostitution that can result in a readiness to identify witches as prostitutes/pimps (or vice versa) on the basis of little verifiable evidence: see Dickie (n. 5), 178–91, wherein Horace's Canidia is cast as an ageing prostitute (Epod. 5, 17), and an old woman performing a rite ‘to bind hostile mouths’ at the Feralia (Ov. Fast. 2.571–83) is identified as ‘a lena and her girls’ based solely on their consumption of wine at the ceremony (see also Ogden [n. 13], 129). The fourth-century grammarian Nonius provides a significantly earlier example of this conflation in his citation of the Lucilian line ‘in age and aspect like a saga and a good go-between’ (Lucil. Sat. 7.271: aetatem et faciem ut saga et bona conciliatrix). Neither saga nor conciliatrix need specify a woman acting as a pimp in this instance; the terms may just as readily pertain to a saga who offers erotic magic for reconciliatory purposes, especially given that it is only in Tib. 1.5 that a saga is ever explicitly identified as a lena. Nevertheless, Nonius glosses saga by explaining, ‘Women are called sagae who hunt to satisfy the lust of men’ (sagae mulieres dicuntur feminae ad lubidinem virorum indagatrices, 23.1).

¹⁵ Keulen, W.H., Apuleius Madaurensis Metamorphoses: Book I (Groningen, 2007), 205Google Scholar, goes so far as to suggest that the ‘Apuleian sagae epitomize sorcery as an evil discipline’.

¹⁶ Var. Log. 15.1: ut faciunt pleraeque, ut adhibeant praecantrices nec medico ostendant. Many of the above citations have been noted by Pease, A.S., ‘M. Tulli Ciceronis De divinatione. Liber primus. Part I. With commentary’, University of Illinois Studies in Language and Literature 6 (1920), 159–326Google Scholar, at 210, and addressed further by Dickie (n. 5), 167, who finds that the persistent application of the term saga to such a wide range of characters is not proof of ‘a failure to mark a distinction … [but] that persons calling themselves prophets would at the drop of a hat turn their hand to magic and that their ability to foresee the future was thought to be a part of their magical powers’. It is my opinion that the characterization of the saga as a figure in possession of virtually limitless magical abilities and social functions tends to confound rather than clarify the role that she might play.

¹⁷ Ogden (n. 13), 124–5. Examples of these sketches may be seen at Tib. 1.2.43–66; Prop. 1.1.19–24; Ov. Her. 6.83–94; Ov. Am. 1.8.1–20.

¹⁸ Leinweber, D.W., ‘Witchcraft and lamiae in The Golden Ass’, Folklore 105 (1994), 77–82CrossRefGoogle Scholar, provides an excellent rationale for understanding Meroe as a folkloric demon along the lines of the Greek Lamia and Mormo. Her draining sexual relationship with Socrates somewhat resembles that of a succubus and her victim, and her removal of Socrates' blood through his neck, the emphasis on crossing the threshold to Socrates' room, and the illusion of Socrates' restored health disappearing upon contact with running water all suggest a close association with vampirism.

¹⁹ There are few instances of witches committing murder for reasons other than enacting a spell or acquiring magical ingredients. One of the most famous involves the woman Martina and the curious circumstances surrounding the death of Germanicus (Tac. Ann. 2.69–70). Given Martina's ambiguous label of venefica it is unclear whether she was supposed merely to have administered lethal poison or to have placed the materia magica that Germanicus found so terrifying. In either case, it is unlikely that Martina was acting on her own behalf. Likewise, the witch at Apul. Met. 9.29 summons a ghost to slay a baker, but she, too, was committing a murder for the benefit of someone other than herself. I am aware of only one instance of a witch committing murder with a sword: Medea's execution of her children (and, in some versions, her brother). However, in both cases the murders themselves are not the final goal but have an ultimate purpose beyond the simple extermination of the victim. This is not the case with Meroe and Socrates.

²⁰ Circe is a notable exception to this rule. However, while the specific nature of Circe's divinity is uncertain, she is explicitly referred to by Hermes as a goddess (Hom. Od. 10.297). Apuleius provides another exception in that he also attributes this power to the mortal Pamphile, though she is the next most prominent witch of the Metamorphoses after Meroe. Beyond Apuleius, other mortal examples are unknown.

²¹ Bushala, E.W., ‘Laboriosus Ulixes’, CJ 64 (1968), 7–10Google Scholar, interprets Canidia's implied demands on the narrator of Hor. Epod. 17 as being sexually exhaustive; hence the narrator plays the laboriosus Vlixes opposite Canidia's insatiable Circe, and the relationship thus results in the narrator's increasingly poor health (fugit iuventas et verecundus color | reliquit ossa pelle amicta lurida, | tuis capillus albus est odoribus, 21–3). This reading would provide a literary antecedent to the draining influence that Meroe has on Socrates' health, but the interpretation is not without its problems. At any rate, Canidia most certainly does not jugulate the narrator.

²² The unnamed boy at Hor. Epod. 5.98 calls his tormentors obscenas anus, and in Apul. Met. 2.22 the cantatrices anus are labelled nequissimae and likened to harpies (23).

²³ I am indebted to Fritz Graf and the anonymous referee at CQ for their observations regarding the potential Egyptian associations with Meroe's name. While Egypt had long been considered a source of potent magical knowledge in the ancient world (as early as the Odyssey we find Helen deriving her knowledge of pharmaka from Egypt [4.220–6]), there is unfortunately no single, standard image of the Egyptian magical practitioner that would inform a reading of Meroe here. For the many and varied aspects of Egyptian magical practice, see Nectanebo (Ps.-Callisthenes' Alexander Romance, 1–14), the anonymous, elderly necromancer of Heliodorus' Aethiopica (6.14–5), or the sorcerer Moeris of Virgil's Eclogues (8.95–9). The name could also have brought to mind merum (unmixed wine) (cf. Aus. Epigr. 41), and thereby conjured an association with drunken witch-procuresses such as Acanthis (Prop. 4.5) or Dipsas (Ov. Am. 1.8), yet Meroe wields considerably more power than either of these figures, has no apparent ties to prostitution, and is not characterized by drunkenness. She is in one instance described as ‘tipsy’ (temulenta, Met. 1.10), but in context this adjective does not typify her behaviour so much as provide an explanation for why she would have disclosed the details of her magical practices to Socrates.

²⁴ It is tempting to doubt the sincerity of this positive physical description of Meroe for two reasons: first because Socrates' interlocutor, Aristomenes, later dubs her a ‘leathery whore’ (scortum scorteum, Met. 1.8), and second because Socrates himself may have chosen his words carefully owing to his fear of Meroe's ability to eavesdrop from afar. On the other hand, one must also keep in mind that Aristomenes formulated his colourful assessment of Meroe's appearance without ever having seen her, and that Socrates' characterization of his relationship with Meroe as a ‘sickening, long-lasting disease’ is hardly the language of a man afraid to speak his mind, albeit quietly. It is my opinion that these final points obviate any hesitancy in taking Socrates at his word here.

²⁵ Apul. Met. 1.8: ‘saga’ inquit ‘et divini potens caelum deponere, terram suspendere, fontes durare, montes diluere, manes sublimare, deos infimare, sidera extinguere, Tartarum ipsum inluminare.’ Note that the list of natural inversions ascribed to Meroe here corresponds to the generic ‘thumbnail sketches’ outlined by Ogden (n. 13), 124–5.

²⁶ Lucil. Sat. 15.484 (Lactant. Div. Inst. 1.22.13): lamiae are mentioned as objects of scorn on a par with bogeymen (terriculas). At Sat. 30.1065 (Non. 117.29) an old woman named Lamia is an unwelcome glutton (gumia).

²⁷ Of Apuleius' usage of lamia at Met. 1.17, Scobie, A., Apuleius: Metamorphoses I (Meisenheim am Glan, 1975), 116Google Scholar, remarks that such ‘names for maleficae were often used loosely and indiscriminately’.

²⁸ Oliensis, E., Horace and the Rhetoric of Authority (Cambridge, 1998), 64–101CrossRefGoogle Scholar, skilfully reads Canidia as a bringer of Horatian impotence, and as a dark, quasi-alter-ego of Maecenas, whose presence pervades the Epodes as a whole. This reading, while constituting a considerable advancement in the scholarship on Canidia, does at times overlook discrepancies in her character for the sake of presenting her as a unified constant whose influence is felt throughout the Epodes. For more on Canidia, M.T. Paule, ‘Canidia: a literary analysis of Horace's witch’ (Diss., Ohio State University, 2012) is a comprehensive and recent contribution. See also Eitrem, S., ‘Exkurs: die 5. und 17. Epode des Horaz’, SO 12 (1933), 30–8CrossRefGoogle Scholar; Herrmann, L., ‘Canidia’, Latomus 17 (1958), 665–8Google Scholar; Ingallina, S.S., Orazio e la magia (Palermo, 1974)Google Scholar; Hill, B., ‘Horace, Satire 1.8: whence the witches? Thematic unity within the satire and within the Satires of Book 1’, in Deforest, M. (ed.), Woman's Power, Man's Game: Essays on Classical Antiquity in Honor of Joy K. King (Mundelein, 1993), 257–63Google Scholar; Wallinger, E., Hekates Töchter: Hexen in der römischen Antike (Vienna, 1994), 38–50Google Scholar, 66–84; Mankin, D., Horace: Epodes (Cambridge, 1995), esp. 299–301Google Scholar; Watson, L., A Commentary on Horace's Epodes (Oxford, 2003), esp. 175–91Google Scholar, 534–43; Pagán, V.E., Rome and the Literature of Gardens (London, 2006), 37–63Google Scholar.

²⁹ Canidia's appearances in Sat. 2.1, 2.8 and Epod. 3 are too brief to establish firm details of Horace's depiction of her therein.

³⁰ Hahn, E.A., ‘Epodes 5 and 17, Carmina 1.16 and 1.17’, TAPhA 70 (1939), 213–30Google Scholar, at 214 n. 11, finds in Canidia's name a potential resemblance to canities (white hair), an interpretation that would further weaken the impact of a reference to her as an anus.

³¹ D. Mankin (n. 28), 131, suggests this as a possible reading and offers the lively translation of venefica as a ‘dime-store witch’.

³² Pace Dinter, M.T., Anatomizing Civil War: Studies in Lucan's Epic Technique (Ann Arbor, MI, 2012), 68Google Scholar, who finds in Erictho an accretion of many witch-figures who preceded her, e.g. Medea, Lamia, Hecate, Canidia, and Allecto.

³³ If the venefica of line 79 is not to be understood as the same character referred to as saga on line 29, then the two practitioners (one a venefica and the other a saga) are cast as being similarly skilled and thus both potential suspects of the poet's impotence.

³⁴ Porph. In Hor. Epod. 17.58: ibi in aggeribus religiones et sepulcra erant, inter quae noctibus illa malefica sacra conficiebat. Malefica may just as plausibly be read as an adjective with sacra, instead of as the subject illa malefica. In either case, Porphyrion seems to identify Canidia with the practices of a malefica.

³⁵ Prior to Porphyrion, I am aware of no instance in which the feminine malefica may be read as a substantive noun. The adjective itself is used by and large as a generic pejorative and is broadly applied to animals (Plin. HN 9.50; Var. Rust. 3.16), people (both men and women: Plaut. Trin. 551; Sen. Troad. 752; Tac. Ann. 4.21), and practices (Suet. Ner. 16.2). In some cases it is used to describe supernatural activities of ill-repute (Apul. Met. 3.16; Tac. Ann. 2.69), but is used far less frequently of these events than of the merely mundane.

³⁶ Ps.-Acro In Hor. Sat. 1.8.20: sagas; 1.8.24: veneficam; In Hor. Epod. 3.8: malefica. This interchangeability of terms is most evident in Pseudo-Acro's comments on Epod. 5: he explains that the poem is written against the venefica Canidia (in Canidiam veneficam scribit, 1) and contains a description of necromancy carried out by the maga Canidia (necromantiam factam detestatur a Canidia maga). He later labels both the second and third witches as ‘other veneficae’ (Sagana; nomen veneficae alterius, 25; Veiam etiam nomine aliam veneficam, 29), then dubs the fourth witch ‘another maga’ (magam aliam significat, 41).

³⁷ Johnston, S.I., ‘Defining the dreadful: remarks on the Greek child-killing demon’, in Meyer, M. and Mirecki, P. (edd.), Ancient Magic and Ritual Power (Leiden, 1995), 361–87CrossRefGoogle Scholar, at 371.

³⁸ On the numerous recent attempts to define a unifying rubric for the ancient witch, see Clerc, J.-M., Homines Magici: étude sur la sorcellerie et la magie dans la société romaine impériale (Bern, 1995), 197–200Google Scholar; M. Schons, ‘Horror and the characterization of the witch’ (Diss. University of California at Los Angeles, 1998); Gordon, R., ‘Imagining Greek and Roman magic’, in Ankarloo, B. and Clark, S. (edd.), Witchcraft and Magic in Europe: Ancient Greece and Rome (Philadelphia, PA, 1999) 161–268Google Scholar; G. Luck, ‘Witches and sorcerers in classical literature’, in Ankarloo and Clark (this note), 95–158; Stratton, K., Naming the Witch: Magic, Ideology, and Stereotype in the Ancient World (New York, 2007), 71–105CrossRefGoogle Scholar; Dickie (n. 5), 176–92; P.M. Meléndez-Valdés (n. 1), 235–43; Ogden, D., Night's Black Agents: Witches, Wizards, and the Dead in the Ancient World (New York, 2008), 45–55Google Scholar.

³⁹ The disparity between literary and ‘historical’ representations of magical practices has been frequently observed; see, for example, Graf (n. 5), 175–85.