ARISTOTLE, ANTISTHENES OF RHODES, AND THE MAGIKOS* Among the many lost works that ancient writers attributed to Aristotle is a piece entitled Magikos. The only substantial discussion of this work is that of Valentin Rose published a cen- tury and a half ago; since his analysis rests on some questionable assumptions that have never been closely examined, it is perhaps time for a reconsideration of the evidence. Although the evidence is scanty, we can nevertheless reach a better understanding of the work’s probable nature and, in turn, of the tradition to which it belonged. Rose assigned five fragments to the Magikos.1 F 32 comes from Diogenes Laertius: “Aristotle says that a certain magus, who came from Syria to Athens, made several observations about Socrates, most notably that he would have a violent end”.2F 33 is from the Suda, which, in its entry on Antisthenes, says that “he composed ten volumes; first is Magikos. It tells about Zoroaster, a certain magus who discovered wisdom. But certain people at- tribute this to Aristotle, and others to Rhodon”.3F 34 comes from the elder Pliny’s discussion of magic: “Eudoxus, who wanted it [i.e., the magical art] to be considered the most illustrious and use- *) I owe thanks to the Editor, an anonymous reader, and my colleague Jer- emy Trevett for their suggestions, and especially to Robert Phillips for all his advice and encouragement. 1) I follow here the numbering of V.Rose, Aristotelis qui ferebantur libro- rum fragmenta (Leipzig 31886), in which he prints the same passages in the same order as in his Aristoteles pseudepigraphus (Leipzig 1863); I give also the number- ing of O.Gigon, Aristotelis Opera III: Librorum deperditorum fragmenta (Berlin/New York 1987). 2) Diog. Laert. 2.45 = Aristotle F 32 Rose = F 663 Gigon: fhs‹ d' ÉAris- tot°lhw mãgon tinå §lyÒnta §k Sur¤aw efiw ÉAyÆnaw tã te êlla katagn«nai toË Svkrãtouw. ka‹ dØ ka‹ b¤aion ¶sesyai tØn teleutØn aÈt“. 3) Suda A 2723 Adler = Aristotle F 33 Rose = F 665 Gigon: otow sun°grace tÒmouw d°ka: pr«ton magikÒn: éfhge›tai d¢ per‹ Zvroãstrou tinÚw mãgou, eÍrÒntow tØn sof¤an: toËto d° tinew ÉAristot°lei, ofl d¢ ÑRÒdvni énatiy°asin. 36 J.B.Rives ful of the systems of wisdom, asserted that this Zoroaster lived six thousand years before the death of Plato, and Aristotle agrees”.4 The last two are again from Diogenes Laertius, this time from the discussion of ‘barbarian philosophers’ in his preface. He opens by surveying the various groups to whom the role was attributed (F35): “among the Persians there were the magi, and among the Babylonians or Assyrians the Chaldeans, and the Gymnosophists among the Indians, and among the Celts and Galatians the ones called Druids and Semnotheoi, as Aristotle says in the Magikosand Sotion in book twenty-three of his Succession”.5 A little later, he has a more detailed if somewhat rambling account of the magi in particular: they spend their time in the worship of the gods; they make fire and earth and water the origins of the gods; they con- demn the use of images; they consider cremation impious but not intercourse with their mothers or daughters, a piece of information for which he again cites Sotion’s twenty-third book. After further discussion of their customs, he says (F36) that “they know noth- ing of goetic mage¤a, as Aristotle says in the Magikosand Dinon in the fifth book of the Histories”.6 The first thing to note about these passages is that only three of them cite the Magikosby name; Diogenes in F32 and Pliny in F34 refer merely to Aristotle. The attribution of these two pas- sages to the Magikosis thus prima facie quite uncertain. Rose him- self more or less admitted that his assignment of F34 was arbitrary, and Werner Jaeger argued cogently that it should be assigned instead to Aristotle’s lost dialogue Per‹ filosof¤aw.7 In contrast, Rose had very specific reasons for assigning F32 to the Magikos; indeed, it was central to his overall interpretation of that work. 4) Plin. NH 30.3 = Aristotle F 34 Rose = F 664 Gigon: Eudoxus, qui inter sapientiae sectas clarissimam utilissimamque eam [i.e., artem magicam] intellegi voluit, Zoroastrem hunc sex milibus annorum ante Platonis mortem fuisse prodidit, sic et Aristoteles. 5) Diog. Laert. 1.1 = Aristotle F 35 Rose = F 661 Gigon: gegen∞syai går parå m¢n P°rsaiw Mãgouw, parå d¢ Babulvn¤oiw µ ÉAssur¤oiw Xalda¤ouw, ka‹ Gum- nosofiståw par' ÉIndo›w, parã te Kelto›w ka‹ Galãtaiw toÁw kaloum°nouw Dru˝daw ka‹ Semnoy°ouw, kayã fhsin ÉAristot°lhw §n t“ Magik“ ka‹ Svt¤vn §n t“ efikost“ tr¤tƒ t∞w Diadox∞w. 6) Diog. Laert. 1.8 = Aristotle F 36 Rose = F 662 Gigon: tØn d¢ gohtikØn mage¤an oÈd' ¶gnvsan, fhs‹n ÉAristot°lhw §n t“ Magik“ ka‹ De¤nvn §n tª p°mpt˙ t«n ÑIstori«n. 7) See below, n.26. Aristotle, Antisthenes of Rhodes, and the Magikos 37 Taking his cue from the attribution of the Magikosto Antisthenes, Rose proposed that it was a dialogue whose scenario was set out in F32: a magus visits Athens and converses with Socrates. This interpretation also determined the order in which he placed the fragments: he put F32 first because in his view it provided the cru- cial evidence about the nature of the work, and he put F33 second because the attribution of the work to Antisthenes corroborated his assumption about its nature. Rose’s interpretation is ingenious and apparently coherent, and has won the assent of other scholars.8Yet it depends on prob- lematic assumptions. First, he assumes that the attribution of the work to Antisthenes is evidence that it took the form of a Socratic dialogue. Since Antisthenes apparently also wrote in other formats, however, the attribution in itself is hardly cogent evidence. More- over, the work was also attributed to Aristotle, who so far as we know did not write Socratic dialogues at all, and there is no reason why the attribution to Antisthenes should carry more weight.9Sec- ondly, Rose assumes that F32 describes the scenario of a dialogue, although there is nothing in the passage itself to suggest this; on the contrary, it reads just like all the other anecdotes that Diogenes Laertius records. In short, Rose’s hypothesis that the Magikoswas a dialogue has no basis in the evidence.10But without it, there is no compelling reason to assign F32 to the Magikosat all. The mere fact 8) E.g., F.Wehrli, Die Schule des Aristoteles, Supplementband II: Sotion (Basel 1978) 66–7. 9) Rose, Pseudepigraphus (above n.1) 50 asserts that “scilicet ex Suida (s. ÉAntisy°nhw) discimus eundem librum ab accuratioribusAntistheni adscriptum fuisse” (emphasis added), but I can find nothing in the Sudaentry itself to support this characterization. 10) As further support, Rose cited other examples from the fourth century BCE of dialogues between Greek philosophers and representatives of ‘alien wis- dom’: Heraclides Ponticus’ dialogue Zoroaster(Plut. Adv. Colot. 14, 1115a = Hera- clides F 68 Wehrli) and Clearchus’ dialogue On Sleep, in which Aristotle, as one of the chief interlocutors, describes his encounter with a Jew (Jos. Ap.1.175–82 = Clearchus F 6 Wehrli). But the former is far too poorly known to provide a mean- ingful parallel (H.B.Gottschalk, Heraclides of Pontus [Oxford 1980] 111–12), and in the latter it is by no means clear that Clearchus included an actual dialogue between Aristotle and the Jew. Rose also noted the possible influence of Aristo- xenus’ story of a conversation between Socrates and an Indian (Eus. Praep. Evang. 11.3 = Aristoxenus F 53 Wehrli), but did not point out that this parallel, which is the most exact of those he cited, weighs more in favor of F 32 being simply an anec- dote rather than the scenario of a dialogue. 38 J.B.Rives that it mentions a magusis hardly decisive, since as I shall discuss below Aristotle referred to magi in other contexts. Moreover, a wide range of spurious works circulated under his name; we need merely consider the paradoxography of On Marvelous Things Heard to get some sense of what sorts of information might be attributed to him.11 Given our limited knowledge about both the genuine lost works of Aristotle and these pseudepigrapha, we can only guess about Diogenes’ source for his anecdote; although the Magikosis a not unreasonable guess, it is no more than that. F32 is thus a very uncertain foundation for an interpretation, and we would do better to build instead on the definite references to this work, i.e., F33, 35, and 36. The best place to begin is with its title: indeed, the title is the only thing to indicate that these three passages all refer to the same work.12The titles given to prose works generally followed certain conventions, so that a title of a particular sort is usually a good indi- cation that the work to which it was given belonged to a particular genre. Consequently, the title Magikosprovides a clue to the nature of the work itself. This is true even if the author himself was not responsible for the title. It is in fact very unclear whether the authors of prose works gave them specific titles at all, at least before the Hellenistic period; what we regard as titles may in practice have been no more than conventional ways of referring to particular types of works.13If so, titles ought to have even more predictive value, since they would have depended not on the whims of individual authors but on conventions widespread among the sorts of people who referred to or quoted from publicly available texts. 11) Some form of this text was attributed to Aristotle at least by the time of Athenaeus (12.541a); see further A.Giannini, Studi sulla paradossografia greca, II, Da Callimaco all’età imperiale: la letteratura paradossografica, Acme 17 (1964) 99–140 at 133–5. I owe this reference to Robert Phillips. 12) It is even possible that the reference in the Suda(F 33) and those in Dio- genes Laertius (F 35 and 36) do not concern the same work, although the Suda’s mention of possible Aristotelian authorship and the similarity in the allusions to its contents make this a very strong probability. 13) On titles in general, see E.Nachmanson, Der griechische Buchtitel: einige Beobachtungen (Göteborg 1941) and J.-C.Fredouille, M.O.Goulet-Cazé, P.Hoffmann, and P.Petitmengin (eds.), Titres et articulations du texte dans les œu- vres antiques: Actes du Colloque International de Chantilly 13–15 décembre 1994 (Paris 1997); I owe thanks to my colleague Jeremy Trevett for his guidance on this issue. Aristotle, Antisthenes of Rhodes, and the Magikos 39 The title Magikosis presumably short for ı magikÚw lÒgow. The method of referring to a prose work by means of a masculine adjective in agreement with lÒgowdates back to the first half of the fourth century BCE. Thus Plato describes Socrates’ speech in Phae- drusas an §rvtikÚw lÒgow(Phdr.227c) and his speech in Menexenus as an §pitãfiow lÒgow (Men. 236b). Although in these cases the terms function more as names for types of speeches than as titles for specific speeches, by the latter part of the fourth century BCE they seem to be well on the way to becoming actual titles. So for exam- ple Isocrates, in his Philippus of 346 BCE, can refer to his pan- hgurikÒw (Isocr. 5.9 etc.): he clearly expects that his readers will know that this is not just any panhgurikÚw lÒgow, but the famous Panegyrikos of 380 BCE. This development is even clearer in Aristotle’s Rhetoric, in which he can casually cite “Isocrates in the Panegyrikos” or “Socrates in the Epitaphios”.14Aristotle also pro- vides the earliest evidence for naming speeches after a particular place, such as the Messeniakos of Alcidimas (Rhet. 1.13,1373b18) or the Olympikos logos of Gorgias (Rhet. 3.14,1414b31). By the Hellenistic period this convention was clearly well established, so that even now it is customary to refer to the Areopagitikosand the Panathenaïkos of Isocrates, or the Olynthiacs and Philippics of Demosthenes. But this form of title was used chiefly for speeches; for other genres, other conventions became established. Dialogues, for example, were typically either named after one of their major characters or given a descriptive title in which a noun denoting the topic was the object of the preposition per¤; by the first century BCE it was apparently conventional to assign dialogues a double title that included each type.15Prose treatises, for their part, were usually assigned titles either in the per¤ form or in the form of 14) Rhet. 3.7, 1408b15 (cf. 3.17, 1418a31): ÉIsokrãthw §n t“ panhgurik“; Rhet. 3.14, 1415b30: Svkrãthw §n t“ §pitaf¤ƒ. 15) According to Diogenes Laertius (3.57), Thrasyllus used double titles for all of Plato’s dialogues, e.g., EÈyÊfrvn µ per‹ ıs¤ou, Fa¤dvn µ per‹ cux∞w, Yea¤thtow µ per‹ §pistÆmhw; J.Mansfeld, Prolegomena: Questions to be Settled be- fore the Study of an Author, or a Text (Leiden 1994) 71–3, convincingly argues that Thrasyllus only generalized an existing tendency. That the practice of giving dia- logues a double title was already conventional is suggested by the titles of Varro’s logistorici, e.g., Marius de fortuna(Macr. Sat. 3.18.5) and Messalla de valetudine (Prob. Buc. 6.31), and by Cicero’s Cato maior de senectute(Cic. Amic. 4; cf. Div.2.3 and Att. 14.21.3). 40 J.B.Rives neuter plural adjectives, such as we find in the extant works of Aristotle, e.g., Per‹ poihtik∞w or Politikã. It was relatively rare, however, for philosophical works of any sort to be given titles in the logos-form that was used for speeches. There were various ex- ceptions to this general trend, of which by far the most important was the title Protreptikos.16 This title was assigned to works by a range of philosophers, the best known and most important being the Protreptikos of Aristotle. In this case, there is an obvious explanation for the title, since the work was written in the form of a prose address, i.e., a written speech.17These conventions in titles were by the first century BCE so well established that when Ci- cero, following the lead of Aristotle, wrote a protreptic work of his own but cast it in dialogue form he gave it a title of the appropri- ate type, Hortensius.18Given these conventions, we would expect a dialogue between Socrates and a magus, such as Rose proposed, to have been given the title Magosrather than Magikos. It is of course true that some philosophical dialogues were as- signed titles in the form of a singular masculine adjective. The most obvious example is Plato’s Politikos, which might at first glance seem an obvious parallel to Magikos. But the parallel is deceptive, since in the Platonic title politikÒwis undoubtedly meant to be in agreement with énÆr, not lÒgow, as the symmetry with its com- panion piece SofistÆw reveals. It is extremely unlikely that the same was true of magikÒw, since the existence of the noun mãgow made the periphrasis magikÚw énÆr unnecessary. In other cases, however, it is not so easy to decide whether a title implies énÆror 16) Two of Xenophon’s prose treatises have titles in this form, Kynegetikos and Hipparchikos. We may also note that Diogenes Laertius credits various philoso- phers with works entitled ÉErvtikÒw, e.g., Eucleides of Megara (2.108), Aristotle (5.43), Theophrastus (5.43), Demetrius of Phalerum (5.81), and Heraclides Ponticus (5.87); the title perhaps looks back to Socrates’ §rvtikÚw lÒgowin the Phaedrus. 17) Diogenes Laertius assigns protreptiko¤to Aristippus (2.85), Plato (3.60, as an alternative title for Clitophon), Aristotle (5.22), Theophrastus (5.49), Demetrius of Phalerum (5.81), and Epicurus (10.28). The earliest definite reference to Aristotle’s work under this title is in Alexander of Aphrodisias (In Top.110a23 = F 51 Rose = F 55.1 Gigon), although W.Jaeger, Aristotle (Oxford 21948) 58–60, ar- gued that the reference to “those who write protreptiko‹ lÒgoito their friends” in Ad Demonicum([Isocr.] 1.3) was in fact a specific reference to Aristotle by one of Isocrates’ pupils. Whether or not one accepts his argument, the passage does point to the origins of this title as a descriptive term for a particular type of speech. 18) I follow the observation of Jaeger (above n.17) 55; for the title, see Cic. Div.2.1. Aristotle, Antisthenes of Rhodes, and the Magikos 41 lÒgow. The two most striking occur among the works of Xeno- phon: Oikonomikosand Tyrannikos(an alternative title for Hiero). It is certainly possible that whoever assigned these titles to these dialogues intended the reader to understand lÒgow, and many readers may have done so even if it was not intended.19 But although we cannot exclude the possibility that a dialogue could have a title in this form, the weight of the evidence remains strong- ly against it. It thus seems best to conclude that, on the evidence of its title, the Magikosis much more likely to have been an address or prose treatise than a dialogue. We may now consider its contents, by examining in more detail the three definite references to the work. Although brief and gener- al, they do provide some indication of its overall subject and gener- al approach. In F 35, Diogenes Laertius says that the work present- ed the magi as a caste of foreign wise men comparable to the Chaldeans, the Gymnosophists, or the Druids, and implies that it presented these wise men as the originators of philosophy. This idea eventually became commonplace, and was often elaborated by sto- ries that this or that Greek philosopher acquired his wisdom by studying with this or that group of foreign wise men.20Its origins, however, seem to lie with the early peripatetics, and to some extent with Aristotle himself. In Metaphysics, Aristotle treats the magias sages who combine philosophical and mythical language in their descriptions of the cosmos; the fact that he associates them with Pherecydes of Syrus suggests that he placed them in the earliest stages of philosophical development.21 In his lost dialogue Per‹ filosof¤aw he apparently developed this view at greater length. 19) For the title Tyrannikos, see Athen. 3.121d and 4.171e, and Diog. Laert. 2.57.S.Pomeroy, Xenophon’s Oeconomicus: A Social and Historical Commentary (Oxford 1994) 213, prefers to understand lÒgowwith these titles, although she does not give any specific arguments in its favor. There is a similar ambiguity with the Physikosof Antisthenes, known by that title to Philodemus (Antisthenes F 39A De- cleva Caizzi = F 179 Giannantoni) and Cicero (ND 1.32), although in this case our information is too meager to indicate whether this was a prose treatise or a dialogue. 20) The bibliography on this topic is extensive. For two important ancient discussions, see Diog. Laert. 1.1–11 and Clem. Alex. Strom. 1.15, 66–72; for a com- prehensive survey of the evidence concerning Plato, with much further bibliogra- phy, see H.Dörrie, Der Platonismus in der Antike, Bd.2: Der hellenistische Rah- men des kaiserzeitlichen Platonismus (Stuttgart 1990) 166–218 and 425–505. 21) Met. 14.4, 1091b8–12: §pe‹ o· ge memeigm°noi aÈt«n ka‹ t“ mØ muyik«w ëpanta l°gein oÂon FerekÊdhw ka‹ ßtero¤ tinew tÚ genn∞san pr«ton êriston tiy°asi ka‹ ofl Mãgoi. 42 J.B.Rives According to Diogenes Laertius, “Aristotle in the first book of Per‹ filosof¤awsays that the magiare older than the Egyptians, and that according to them there are two principles, a good daimon and an evil daimon, the one called Zeus and Oromasdes and the other called Hades and Areimanios”.22 Since we know from another citation that Aristotle also talked about the Orphic poems in this work, it seems likely that he began with a discussion of the same sort of semi- mythical philosophical works that he mentions in Metaphysics.23 The statement in the Suda(F33) that the Magikosdealt with Zoroaster, whom it depicted as a “discoverer of wisdom”, fits very well with the evidence from Diogenes Laertius that it treated the magi as barbarian philosophers. The Greek belief that Zoroaster was the founder of magian tradition apparently dates back to Xan- thus the Lydian, who according to Diogenes Laertius assigned Zoroaster a date of 6000 years before Xerxes’ invasion of Greece and made him the founder of the succession of magi.24From Pliny, it appears that Eudoxus and Aristotle modified Xanthus’ date by substituting the death of Plato for the invasion of Xerxes, but agreed with him in making Zoroaster the originator of the magian ars, which Eudoxus allegedly regarded as “the most illustrious and useful of the systems of wisdom”.25 Aristotle’s discussion of 22) Diog. Laert. 1.8 (continuing from the passage quoted above in n.6) = F 6 Rose = F 6 Untersteiner: ÉAristot°lhw d' §n pr≈tƒ Per‹ filosof¤aw ka‹ pres- but°rouw e‰nai t«n Afigupt¤vn: ka‹ dÊo kat' aÈtoÁw e‰nai érxãw, égayÚn da¤mona ka‹ kakÚn da¤mona: ka‹ t“ m¢n ˆnoma e‰nai ZeÁw ka‹ ÉVromãsdhw, t“ d¢ A·dhw ka‹ ÉAreimãniow. Similar views are attributed to Eudoxus (Diog. Laert. 1.8 = F 341 Lasserre), Eudemus (Dam. Princ. I 319 Ruelle = F 150 Wehrli), and Clearchus (Diog. Laert. 1.9 = F 13 Wehrli), as well as Theopompus (Diog. Laert. 1.8–9 and Plut. Is. et Os.47, 370b–c = FGrHist 115 F 64a and 65). 23) Orphic poems: Philoponus, In De anima 186.24–6 and Cic. ND 1.107 = F 7 Rose = F 7 Untersteiner; for the opening of the Per‹ filosof¤aw,see Jaeger (above n.17) 128–36. 24) Diog. Laert. 1.2 = FGrHist 765 F 32: Jãnyow d¢ ı LudÚw efiw tØn J°rjou diãbasin épÚ toË Zvroãstrou •jakisx¤liã fhsi, ka‹ met' aÈtÚn gegon°nai polloÊw tinaw mãgouw katå diadoxÆn; for discussion, see J.Bidez and F.Cumont, Les Mages hellénisés (Paris 1938) I 5–8 and II 7–9. 25) Plin. NH 30.3 = Eudoxus F 342 Lasserre, quoted above in n.4; accord- ing to Diogenes Laertius (1.8 = F 341 Lasserre), Eudoxus also agreed with Aristo- tle on the two principles of magian tradition. For the suggestion that Eudoxus may have played a key role in bringing Zoroaster and magian tradition to the attention of fourth century BCE philosophical circles, see Jaeger (above n.17) 131–2 and Bidez and Cumont (above n.24) I 11–12; A.Momigliano, Alien Wisdom (Cam- bridge 1975) 144, expresses doubts. Aristotle, Antisthenes of Rhodes, and the Magikos 43 Zoroaster’s date was probably part of the survey of quasi-mythical proto-philosophy with which he opened his dialogue Per‹ filosof¤aw, since as we have seen he there discussed the antiquity of magian tradition in general.26 Other writers and thinkers in Aristotle’s circle also apparently treated Zoroaster as an archaic philosopher. Heraclides Ponticus, for example, wrote a work that was known to Plutarch by the title Zoroasterand that perhaps con- cerned questions of physical philosophy.27This view of Zoroaster as a philosopher later became a commonplace. Lastly, according to Diogenes Laertius (F 36), the Magikosre- jected the idea that the magiknew anything about ‘goetic mageia’, that is, magic in the sense of ‘sorcery’ or ‘Zauberei’. This polemi- cal assertion was undoubtedly a response to the colloquial use of the word mãgowas a synonym for gÒhw, a usage that dates back to the fifth century BCE. Several writers of the fifth and fourth cen- turies use the word mãgow and its cognates to describe wandering religious specialists; some associate it with fãrmaka, wonder- working, and interactions with the dead, and two explicitly treat it as equivalent to gÒhw.28Given these tendencies, anyone who want- ed to discuss the Persian magias philosophers or religious special- ists might be tempted to make an explicit distinction between their 26) This is the key observation in Jaeger’s argument that Rose’s F 34 belongs not to the Magikos, but to the Per‹ filosof¤aw: see Jaeger (above n.17) 135–6, followed by Bidez and Cumont (above n.24) I 15–16, M.Untersteiner, Aristotele: Della Filosofia (Rome 1963) 88, and J.Bollansée, FGrHist IV A: Biography, Fasci- cle 8: Hermippus of Smyrna (Leiden 1999) 436 n.22. While granting the force of his argument, I would add that there is no reason that the Magikoscould not have con- tained a similar discussion. 27) Plut. Adv. Colot. 14, 1115a with the discussion of Bidez and Cumont (above n.24) I 80–4; although Gottschalk (above n.10) 111 rightly points out that their more detailed conclusions go far beyond the evidence, the general context of Plutarch’s reference strongly suggests that the work dealt in some way with natur- al philosophy. 28) Religious specialists in Heraclitus DK 22 B 14 (if genuine), Soph. OT 387–8, Hippoc. Morb. Sacr. 1; association with fãrmakain Eur. Supp.1109–11 and Or. 1497–8, Pl. Plt. 280d–e, Theophr. Hist. pl. 9.15.7 (if genuine); skill in dealing with daimones, apparently spirits of the dead, in PDerv col. iv. Note especially the fragment of Sosiphanes of Syracuse, active in the 330s and 320s BCE, in which “every Thessalian maiden” is said to know how to bring down the moon mãgoiw §pƒda›w(TrGF I2261–3). Gorgias (Hel. 10) treats mage¤aand gohte¤aas equiva- lents; Aeschines (In Ctes.137) uses the phrase mãgow ka‹ gÒhw. For recent discus- sions, see F.Graf, Magic in the Ancient World (Cambridge, Mass.1997) 20–35 and J.Bremmer, The Birth of the Term ‘Magic’, ZPE 126 (1999) 1–12 at 1–9. 44 J.B.Rives teaching and the gohte¤a with which some people associated the term mãgow. Diogenes Laertius claims that the fourth-century BCE historian Dinon was of the same opinion as the author of the Magikos. Dinon discussed the magi in his Persika, and since he evidently presented them as religious authorities and wise men, it would not be surprising if he sought to distance them from the colloquial associations of the word mãgow.29One of the most elab- orate examples of this distinction occurs in Philo. Philo describes “the true magic” as “the scientific vision by which the facts of nature are presented in a clearer light”, and says that it is held in such reverence that no one may become Great King of the Persians unless he has mastered this lore. “But”, he continues, “there is a counterfeit of this, most properly called a perversion of art, pur- sued by charlatan mendicants and parasites and the basest of the women and slave population, who make it their profession to deal in purifications and disenchantments and promise with some sort of charms and incantations to turn men’s love into deadly enmity and their hatred into profound affection”.30 We may thus conclude that the Magikos was a prose treatise or address dealing with the Persian magi, whom it presented as dis- ciples of Zoroaster and the bearers of a quasi-mythical physical philosophy that had no connection to popular ideas about mage¤a. Although it is obviously impossible to know what else the work may have contained, we may reasonably suppose that it was not wildly inconsistent. One result of this re-examination of the evi- dence is that Diogenes Laertius’ ascription of the Magikosto Aris- totle no longer seems prima facie absurd. Although virtually every scholar has dismissed it out of hand, it is by no means intrinsically impossible or even implausible.31On the contrary, there is nothing 29) Dinon agreed with the Platonist Hermodorus that ‘Zoroaster’ literally meant ‘star-worshipper’, and claimed that the magiconsidered fire and water to be the only images of the gods: see, respectively, Diog. Laert. 1.8 and Clem. Alex. Protr. 5.65.1 = FGrHist 690 F 5 and F 28. He also followed Herodotus and others in presenting the magias experts in dream-interpretation: Cic. Div.1.46 = FGrHist 690 F 10. 30) Phil. Spec. Leg. 3.100–1, in the Loeb translation of F.H.Colson. Note also Suda M 13: MagikÆ: taÊthn §feËron M∞doi ka‹ P°rsai, ≤ diaf°rousa t∞w gohte¤aw ka‹ aÈt∞w farmake¤aw. 31) It is regarded as spurious by, e.g., Rose, Pseudepigraphus (above n.1) 50– 2; E.Heitz, Die verlorenen Schriften des Aristoteles (Leipzig 1865) 294; Jaeger (above n.17) 135 and n.1; Bidez and Cumont (above n.24) II 8 n.2; P.Moraux, Les Listes anciennes des ouvrages d’Aristote (Louvain 1951) 266–7, and Momigliano