¹ Lebeck, A., The Oresteia (Cambridge, Mass. 1971) 150Google Scholar.

² R. Wuensch, Defixionum Tabellae Atticae, appendix to IG iii (1897) (‘DTA’) and Auguste Audollent, Defixionum Tabellae (1904) (‘DT’) are the basic collections. See K. Preisendanz, ‘Die griechischen und lateinischen Zaubertafeln’, APF ix (1930) 119, 54, for a full bibliography to that date, and D. R. Jordan, ‘A survey of Greek defixiones not included in the special corpora’, GRBS xxvi (1985), for more recent work. I should like to thank Prof. Jordan for kindly allowing me full access to forthcoming and unpublished materials.

³ Detailed instructions for the manufacture and burial of defixiones are preserved in the magical handbooks of the third and fourth centuries AD, and seem to be in agreement with the archaeological evidence of the classical period. Preisendanz, K. and Henrichs, A., Papyri Graecae Magicae i (1973) andGoogle Scholar ii (1974) (‘PGM’) collect several recipes: PGM V 304; VII 394, 417; IX; XXXVI 1–35, 231 and LVIII. The recipe in PGM IV 335–84 provides the text of a curse which also appears on a tablet in the Cairo Museum (no. 48217) and on two tablets and a small pot in Cologne (Wortmann, D., ‘Neue magische Texte’, BJ clxviii [1968] 56–111Google Scholar nos 1 3). D. R. Jordan gives the most recent assessment of the archaeology of Attic lead curse tablets in a series of articles connected with his re-editing of the entire corpus of Greek defixiones: ‘Two inscribed lead tablets from a well in the Athenian Kerameikos’, AthMitt xcv (1980) 225–39Google Scholar; ‘Fourteen defixiones from a well near the southeast corner of the Athenian Agora’, Hesperia xliv (1985Google Scholar), and ‘New archaeological evidence for the practice of magic in classical Athens’ which will appear in the Πρακτικὰ του̑ 12^ou⊿ιεθνου̑ς Συνεδρίου Κλασικη̑ς Ἀρχαιολογίας.

⁴ Wuensch, R., ‘Eine antike Rachepuppe’, Philol. lxi (1902) 26–31Google Scholar; Audollent, , DT lxxviiGoogle Scholar; Dugas, Ch., ‘Figurines d'envoûtement trouveés à Délos’, BCH xxxix (1915) 413–23CrossRefGoogle Scholar; Jordan, ‘New evidence’ (n. 3) passim.

⁵ Tambiah, S. J., ‘Form and meaning of magical acts: a point of view’, in Horton, R. and Finnegan, R., eds, Modes of Thought (London 1973) 199 229Google Scholar, reinterprets the data on Zande magical practices collected by Evans-Pritchard and argues persuasively (see Lloyd, G. E. R., Magic, Reason and Experience [Cambridge 1979] 2–3, 7Google Scholar) against the prevailing theory that ‘sympathetic’ or ‘homeopathic’ magic was based on (poor) observation of empirical analogies. Tambiah differentiates, instead, between the operation of ‘empirical analogy’ used in scientific inquiry to predict future action, and ‘persuasive analogy’ used in magical ritual to encourage future action. The Azande prick the stalks of bananas with crocodile teeth while saying ‘Teeth of crocodile are you. I prick bananas with them. May bananas be prolific like crocodile teeth’ (Evans-Pritchard, , Witchcraft, Oracles and Magic among the Azande [Oxford 1937] 450Google Scholar quoted by Tambiah on p. 204). This magical act is not based on any (mistaken) empirical analogy between bananas and crocodile teeth, but rather on the hope that correct performance of ritual and incantation will ‘persuade’ the bananas to become analogous to crocodile teeth with regard to their plenitude.

⁶ Curses which mention only the name of the intended victim steadily decrease in frequency from the classical age until their total disappearance in the first century AD. The use of the more complex formulas, on the other hand, becomes more popular in the later periods. See Kagarow, E. G., ‘Form und Stil der Texte der Fluchtafeln’, Archiv für Religionswissenschaft xxi (1922) 496Google Scholar and Griechische Fluchtafeln, Eos Suppl. iv (1929) 44–9Google Scholar (Graph of formula frequencies p. 45).

⁷ Audollent, , DT lxxxixGoogle Scholar, included public proclamations against unknown thieves as a fourth category, and Kagarow 1929 (n. 6) 50 described five types, adding public curses and phylacteries. Moraux, P., ‘Une défixion judiciaire au Musée d'Istanbul’, Acad. Roy. Belg. Mém. liv (1960) 5Google Scholar, however, makes an important distinction between ‘imprécation’ (a public notice cursing unknown or potential evildoers) and ‘défixion’ (a buried private curse operating against specifically named individuals, who are presumably unaware of its existence). Phylacteries against lead tablets are defensive rather than offensive magical operations, and as such fall outside the definition of curse tablet.

⁸ Perhaps ‘athletic curses’ is a better designation for a category which also includes runners and wrestlers. See Wortmann (n. 3) no. 12, against two runners, and Jordan, ‘Fourteen defixiones’ (n. 3), five new curses against wrestlers and one against a runner.

⁹ E. Ziebarth, ‘Neue attische Fluchtafeln’, Nachrichten der K. Ges. d. Wiss. Göttingen (1899) 122, asserted that judicial curses were enacted by the losers of a lawsuit, after the decision had been rendered. He was refuted by Wuensch, , ‘Neue Fluchtafeln’, RhM lv (1900) 68Google Scholar, who argued that judicial curse formulas all seem to point to a future event and that they were therefore employed beforehand or while cases were still pending. Audollent, , DT lxxxviii ixGoogle Scholar n. 2, supported this view. Years later, Ziebarth, , ‘Neue Verfluchungstafeln aus Attika, Boiotien und Euboia’, SDAW xxxiii (1934) 1028 32Google Scholar, adopted the compromise view that a judicial curse was enacted while the trial was going on, but only after its author had come to the conclusion that he was about to lose his case. Moraux (n. 7) 42 reviews the debate and concludes that although none of the curses seem to have been enacted after the final outcome of the trial, it is impossible to know at what point during the trial the litigants wrote the curses. There seems to be a trade-off between the practical desire to inhibit damaging evidence as early as possible and the point at which the litigants realize that such action is necessary.

¹⁰ See Wuensch (n. 9) 248–59 for a detailed discussion of a Carthaginian circus curse from the third century AD (DT 242).

¹¹ Brutus 217; Libanius tells us in his autobiography (245–9) how at one point late in his life he became gravely ill and was no longer able to read, write or speak before his students. After a time, the twisted and mutilated body of a chameleon was found in his lecture room. Its head had been placed between its hind legs, one of its forefeet was missing and the other was ‘closing the mouth for silence’. Libanius says that he regained his health after the chameleon was removed. Bonner, C., ‘Witchcraft in the lecture room of Libanius’, TAPA lxiii (1932) 34–44Google Scholar interprets this as a form of envoûtement directed against Libanius' oratorical abilities; the cutting off of the one forefoot was directed against the hand with which the orator gesticulated and the position of the other attempted to silence him, as Libanius himself seemed to realize. Profs John J. Winkler and H. S. Versnel have independently brought to my attention a third-century BC inscribed pillar from the island of Delos, (IG xi.41299Google Scholar; Powell, I. U., Collectanea Alexandrina [Oxford 1925] 68–71Google Scholar; recently republished with full commentary by Engelmann, H., The Delian Aretalogy of Serapis, Ét. prelim, aux relig. orient, xliv [Leiden 1975]CrossRefGoogle Scholar) which preserves a rather lengthy account in epic hexameters of the successful founding of the cult of Serapis on the island of Delos. The central miracle in this aretalogy is the god's timely intervention in a lawsuit that threatened the existence of his newly constructed temple; he binds the men who are pressing charges and ruins their performance in court (lines 85–90):

Although there is no mention of curse tablets or any other kind of magical activity, the binding of the tongue and the paralysis of the witnesses is remarkably similar to the aims described in many defixiones (see Engelmann ad loc. for specific parallels).

¹² Orestes is informed about the approaching trial very early in theplay when Apollo explains what will happen in Athens: κἀκει̑ δικαστὰς τω̑νδε . . . εὑρήσομεν (81–2). Shortly thereafter the Erinyes receive the same information: δίκας δὲ Παλλὰς τω̑νδ᾿ ἐποπτεύσει θεά(224), and then echo the forensic vocabulary when they declare their own intentions: ἐγὼ δ᾿, ἄγει γὰρ αἱ̑μα μητρω̨̑ον, δίκας μέτειμι τόνδε φω̑τα(230–1). These two announcements of the forthcoming trial give ample forewarning to all the litigants involved and perhaps help to facilitate the audience's recognition of the ‘binding song’ as a judicial curse.

¹³ The repetition of this ephymnion and others in this stasimon may indicate a literary rendering of magical formulas. The MSS, however, only repeat the first ephymnion and the scholarly debate, spanning nearly a century, has produced no clear consensus on the advisability of repeating the second and third in our modern texts. For the most recent discussion of the problem see Scott, W. S., ‘Non-strophic elements in the Oresteia’, TAPA cxii (1982) 189–91Google Scholar. Repetition is a standard component of magical incantation, and literary renditions of charms commonly include wholesale repetition (Theoc. Id. 2; Verg. Ec. 8). See the introduction to Thomson's 1938 edition of the Oresteia (p. 4); Kranz, W., Stasimon (Berlin 1933) 132, 135Google Scholar; and Alexiou, M., The Ritual Lament in Creek Tradition (Cambridge 1974) 134–5Google Scholar. Recently, a papyrus fragment of a first- or second-century school exercise has been found to contain, amongst other things, two strophes of anapests which have been tentatively identified as part of a chorus from Aeschylus' lost play The Psychagogoi. The two strophes are part of the chorus' instructions to someone (Odysseus?) on the correct method of sacrifice necessary to summon the souls of the dead from the underworld. The combination of magical ritual and anapests might provide a parallel for the anapests in the binding song in the Eumenides. See Rusten, J. S., ‘The Aeschylean Avernus: notes on P.Köln 3.125’, ZPE xlv (1982) 33 8Google Scholar.

¹⁴ Mueller (1833) and Ahrens (1892) printed 303–4 as a declarative sentence, whereas Page (1972), the most recent editor, follows Palcy (1879), Kirchhoff (1880), Wecklein (1888), Weil (1909) (his final judgement, see below for an interim choice) and Wilamowitz (1914), and prints the two lines as a question. Blass (1907) printed the first half of 303 as a question and the rest of the couplet as a declarative statement. Burges (1822) made a misguided attempt to introduce the scholiast's words into the text and Weil adopted ἀποπτύσεις for a time (see Wecklein's ‘Appendix ad Eumenides’ ad loc.); both men were trying to restore the future tense to these lines in order to suit the context of the prologue (299–306) which predicts the effect of the binding song (ῥύσαιτ´, 300; δαίσεις, 305). In two later literary renderings of erotic curses, the text of the actual incantation is preceded by a dramatic prologue which includes a prediction of the incantation's effect, stated in the future tense; Theocritus, in his Pharmaceutria, prefaces the elaborate text of Simaetha's erotic spellwith a prediction of its result: νυ̑ν δέ νιν ἐκ θυέων καταδήσομαι (Id. 2.10), as Horace does in Epod. 5.77–8: mains paraho, mains infundam tibi fastidienti poculum. Virgil's Alphesiboeus is less confident: ut magicis avertere sacris experiar sensus (Ec. 8.66–7).

¹⁵ Cormack, J. M. R., ‘A tabella defixionis in the Museum of the University of Reading, England’, HThR xliv (1951) 25–34Google Scholar, citing various similarities between this tablet and the cache of anomalous Cypriot curses discussed below in n. 18, dates this Bithynian tablet to the third or fourth century AD.

¹⁶ Koenen, L., ‘Ein wiedergefundenes Archilochos-Gedicht?’, Poetica vi (1974) 500Google Scholar n. 38, discusses the very rare use of a matronymic in line 7 of the Cologne Epode and the parallels in magical incantations. He argues that the use of the matronymic suits the situation of the poem, which necessitates a clear distinction between two stepsisters, and he rightly rejects the possibility of any connection with magical lore. Jordan, D. R., ‘CIL VIII 9525.(8)2: QPVVLVA = q(uem) p(eperit) vulva’. Philol. CXX (1976) 127–32Google Scholar, lists all the instances of maternal lineage in Greek curse tablets and notes that maternal lineage is not used before the first century BC.

¹⁷ DT 50 is a fourth-century BC curse from Attica which does not explicitly mention its social context.

¹⁸ All DTA curses date in the third century BC or earlier (seen. 19). A survey of the dates and provenances of the Greek judicial curses listed by Audollent in his Index 5A to DT reveals only one set of judicial curse tablets which date to a period later than the second century BC. This group of fifteen Cypriot curses (DT 22–35 and 37) were all found in the same well and seem to have been written by the same person (MacDonald, L., ‘Inscriptions relating to sorcery in Cyprus’, Proc. Soc. Biblical Arch. xiii [1890–1891] 160–90Google Scholar). They have been dated in the Roman Imperial period and seem to be an anomaly, since all the other Greek judicial curses with the exception of a second-century AD Magnesian tablet (Dörner, F. K., Öjh xxxii [1940] 63–72Google Scholar) and a third- or fourth-century AD Bithynian tablet published in 1951 (see n. 15 above)—are much earlier. With regard to the oldest Greek judicial curses see Jeffery, L. H., ‘Further comments on archaic Greek inscriptions’, ABSA i (1955) 69–76Google Scholar, who lists 25 fifth-centuryGreek curses, most of them discovered in Sicily subsequent to Audollent's collection. Three of them (Jeffery 1, 2 and 11) are explicitly judicial curses. (See SEG xxvi 111–16Google Scholar for more fifth-century Sicilian examples.) Jordan, ‘New evidence’ (n. 3) adds three more late-fifth-century examples of judicial curses (see n. 20 below).

¹⁹ Wuensch's DTA provides the most extensive collection of Attic curses, including twelve judicial curses: DTA 63, 65 7, 81, 94 5, 103, 105–7 and 129. He was cautious, almost agnostic, in his dating of the tablets and assigned them all to the third century BC—and then only tentatively—unless some overwhelming evidence pointed to an earlier or later date (see his introduction p. 1); accordingly, of the Attic judicial curses enumerated above, he assigned only DTA 107 with confidence to the fourth century. Wilhelm, A., ‘Über die Zeit einiger attischer Fluchtafeln’, Öjh vii (1907) 105–26Google Scholar argued persuasively that Wuensch greatly underestimated the antiquity of the DTA curses and by way of example he redated a number of them (including one of the judicial curses, DTA 103) to the fourth century using a combination of paleographic and prosopographic evidence. The tablets themselves have since disappeared and as a result most of them have never been properly redated. Audollent gave seven examples of Attic judicial curse tablets, six from the fourth century BC (DT 49, 60, 62–3, 66–7) and one which he was unable to date (DT 77). Attic curse tablets published subsequent to these two major collections include ten judicial curses. Five have been assigned to the fourth century BC; Abt, A., Archiv für Religionswissenschaft xiv (1911) 143–56Google Scholar, no. 5; Fox, W. S., AJP xxiv (1913) 76–80, no. 2Google Scholar; Ziebarth 1934 (n. 9) 1028 32, nos 2 and 3, and Robert, L., Collection Froehner i (Paris 1936) no. 11Google Scholar. The remaining five have been dated in the late fifth/early fourth century: Peek, W., Kerameikos iii (Berlin 1941) 93–4Google Scholar, no. 4, and four of the inscribed lead dolls published by Trumpf and Jordan (see n. 20 below). Prof. Jordan reminds me, however, that there are some sixty unpublished third-century AD curse tablets from the Athenian Agora, most of them still rolled up and unread, and that we must not rule out the possibility that the group contains later examples of judicial curse tablets.

²⁰ In his comments on DT 60, Audollent gently criticized Wuensch's haste to identify the Demosthenes and the Lykourgos mentioned there with the famous orators: ‘non raro redeunt apud Graecos huius saeculi talia nomina’. Wilhelm (n. 19) 105–26, however, redated many of the DTA curses to the fourth century [see n. 19 above] using the prosopography of rarer names. Ziebarth 1934 (n. 9) 1023–7 published a long curse which lists the names of the intended victims along with their demotics (a relative rarity). With the added assistance of the demotics he was able to make clear identifications of the well known politician Demades and others in his political circle. Three lead dolls, whose inscribed lead ‘coffins“ bear a series of names and descriptive epithets such as ξύνδικος, ἀντίδικος and μάρτυς, have been unearthed in the Athenian Kerameikos. The first bears the name Mnesimachos and may be connected with a defendant in a case in the Lysian corpus: see Trumpf, J., ‘Fluchtafel und Rachepuppe’, AthMitt lxxiii (1958) 94–102Google Scholar. The other two will be published by Jordan, ‘New evidence’ (n. 3), who has established that all three of the lead dolls date to the late fifth or early fourth century BC.

²¹ Some of the curse recipes in the magical papyri include spoken and written curses working simultaneously (PGM IV 325 35; V 314 ff; VII 429 ff; XXXVI 161 ff). It is hard to imagine what sort of effect the simple inscription of a name on lead would have unless it were accompanied by some further kind of incantation and ritual performance. Tambiah's theory of ‘persuasive analogy’ (n. 5) assumes the interplay of performative utterances and manipulation of objects (in this case the rolling and piercing of the curse tablet). The efficacy of the entire operation depends on the ability of the spoken word to ‘persuade’.

²² Brown, P., ‘Sorcery, demons and the rise of Christianity’, in Witchcraft Confessions and Accusations, ed. Douglas, M. (London 1970) 25Google Scholar.

²³ Although the traditional date for the establishment of democratic juries in Athens is the reform of Ephialtes in 462/1 BC, more recent scholarship points to a much more gradual shift in power to the hēliaia beginning sometime after the turn of the century and culminating in a law attributed to Pericles in the fifties which granted pay for jury duty: Hignett, C., A History of the Athenian Constitution ² (Oxford 1958) 216–21Google Scholar; Sealey, R., ‘Ephialtes’, CPh lix [1964] 14–18Google Scholar = Essays in Greek politics (Woodhaven, N.Y. 1965) 46–52Google Scholar; MacDowell, D. M., The Law in Classical Athens (Ithaca 1978) 29–40Google Scholar; Rhodes, P. J., A Commentary on the Aristotelian Athenaion Politeia (Oxford 1981) 318–19, 338–9Google Scholar.

²⁴ All of the extant Attic judicial defixiones date to the late fifth and fourth centuries (see n. 19). The period of radically democratic juries continued down until 322 BC, the year of Demosthenes' suicide, when the Athenian constitution was amended (at the urging of Antipater) to include property qualifications which effectively disenfranchised nearly four-sevenths of the citizens of Athens. As a result the size and power of the juries were severely curtailed and some courts were completely disbanded: Beloch, J., Gr. Gesch. iiia (Strassburg 1904) 77–80Google Scholar; Ferguson, W. S., Hellenistic Athens (London 1911) 20–6Google Scholar; CAH vi² 459–60.

²⁵ Brown, A. L., ‘The Erinyes in the Oresteia’, JHS ciii (1983) 29–30Google Scholar.

²⁶ In the so-called Aretalogy of Serapis discussed above (n. 11), Serapis calms the fears of his priest with regard to the forthcoming trial (IG xi.4 1299.77–80):

The god clearly interprets the legal charges laid against his human ‘clients’ as a personal attack upon himself.

²⁷ I should like to express my gratitude to Profs John J. Winkler and David R. Jordan for their help and enthusiasm in general, and for their specific remarks on earlier drafts of this paper. I have also benefited from the comments of Profs H. S. Versnel and Marsh McCall Jr and two anonymous referees.

Addendum: Prof. B. M. W. Knox has independently noted some of the general correspondences between the Erinyes' binding song and judicial curses in a forthcoming article entitled ‘Black Magic in the Oresteia’.