¹ The paper is based on a lecture held at the Royal Asiatic Society on 12 November 1998, with added documentation. I take the opportunity to express my gratitude to Professors Boyce and Sundermann for their perceptive criticism of an earlier draft.

² Our knowledge of early Greek philosophy derives for the most part from the extant writings of Aristotle, as well as from much later authors who took most of what they have to say on the subject from the lost writings of Aristotle's pupil Theophrastus. The material is collected in Diels, H., Die Fragmente der Vorsokratiker (Berlin, 1906–1910 and subsequent editions).Google Scholar To what extent these archaic thinkers really had a clear-cut concept of the “principles” of the universe and to what extent the doxographic reports reflect Aristotle's own rationalisation of their teachings is the perennial and, one would think, insoluble question facing all students. Aristotle's description of Plato's theory of principles, as mentioned at the end of this paragraph, is not confirmed by Plato's extant writings, but it is possible that Aristotle heard it thus from his master's mouth.

³ The most comprehensive study of Zoroastrianism in the pre-Christian period is by Boyce, M. (in vol. 3 with F. Grenet), A history of Zoroastrianism, i–iii (Leiden and Cologne, 1975–1991),CrossRefGoogle Scholar with full discussion of the previous literature, supplemented by her Zoroastrianism, its antiquity and constant vigour (Costa Mesa and New York, 1992).Google Scholar For the theology of the Lommel's, Avesta H. Die Religion Zarathustras nach dem Awesta dargestellt (Tübingen, 1930)Google Scholar is still very much worth reading. Given the huge amount of modern literature on Zoroastrianism and the enormous degree of disagreement between the specialists on virtually every point, no attempt will be made here to discuss diverging views; I limit myself to outlining my own provisional understanding of the primary sources.

⁴ The translation by Bartholomae, Chr., Die Gatha's des Awesla (Strassburg, 1905)Google Scholar can perhaps still be regarded as representing the main stream of interpretation. The most recent complete translation is by Humbach, H. (in collaboration with Elfenbein, J. and Skjærvø, P. O.), The Gāthās of Zarathushtra and other Old Avestan texts, 2 vols. (Heidelberg, 1991).Google Scholar

⁵ Spiritus, Semitic rūḥ-, are all etymologically “breath, wind”, while mainiiu- belongs to the root man-, “to think”. “Breathing” and “thinking” are different physiological processes. For this reason the frequently expressed comparison between Ahura Mazdā's spənta- mainiiu- and the “spirit/breath of god” (rūăḥ ‘ělōhīm) who hovers over the face of the water in Genesis 1:2 is not really appropriate.

⁶ Yasna 30,3. For a recent survey of the wildly different translations that have been suggested for this stanza see Kellens, J. and Pirart, É., “La strophe des jumeaux: stagnation, extravagance et autres méthodes d'approche”, Journal asiatique, 285 (1997), pp. 31–72.CrossRefGoogle Scholar

⁷ Hoffmann, K., in his article “Avestisch š”, in Studia grammatica Iranica, Festschrift für Helmut Humbach (Munich, 1986), pp. 163–83,Google Scholar has shown that Avestan aša- assumes an Iranian ^*ár-ta, and is thus not the same formation as Vedic ๛-tá-. On the other hand, in Old Persian the juxtaposition of the spellings a-r-t- (in alphabetic script) and ir-tá- (in Elamite script) points to ๛ta-, not arta-. One must consequendy assume that in Old Indo- Iranian both forms were used with identical meaning.

⁸ The Vendidad (thus the traditional reading of the Parsis), or (more correcdy) Jud-dēw-dād, is translated, with extensive commentary, in Darmesteter, J., Le Zend-Avesta (Paris, 1892–1893; reprinted Paris, 1960), ii, pp. 1–293.Google Scholar

⁹ In a number of Zoroastrian texts from the post-Sasanian period it is stated that the visible world (gētīg) is entirely the creation of Ohrmezd and that his opponent's creation exists only in the invisible world (mēnōg); see Shaked, Sh., “Some notes on Ahremen, the Evil Spirit, and his creation”, in Studies in mysticism and religion presented to Gershom G. Scholem (Jerusalem, 1967), pp. 227–234.Google Scholar But this is clearly not the view of the authors of Vendidad. We must evidently reckon with a later theological development, perhaps in reaction, or rather overreaction, to the anti-materialism of Christianity, Neo-Platonism and Manichaeism.

¹⁰ In Vendidad 13 various species of canines are mentioned as the creatures of Ahura Mazdā, and tortoises (zairimiiaηura-) as the creatures of Aηra Maniiu.

¹¹ Diogenes Laertius, Vit. phil., proem. 8. For this often discussed passage see most recently de Jong, A., Traditions of the magi (Leiden, 1997), pp. 222–4.CrossRefGoogle Scholar

¹² Most of the primary sources for Zurvanism are quoted or translated (together with many other passages mentioning “time”, but otherwise without real relevance to the Zurvanite formulation of Zoroastrianism) in Part II (“Texts”) of the book by Zaehner, R. C., Zurvan, a Zoroastrian dilemma (Oxford, 1955).Google Scholar As a collection of source material Zaehner's book retains considerable value, but nonspecialist readers must be warned against attaching any particular importance to the highly speculative arguments that constitute the main part of the volume. The principal primary testimonia are the Armenian, Syriac and Christian Arabic authors cited in section “F” of Zaehner's collection (Eznik, Elišê, Theodore bar Kōnāy, Martyr Acts etc.); the anti-Zurvanite zand of Yasna 30,3 cited in Dēnkard (Zaehner's text F3a); the Manichaean (or Marcionite) polemical hymn (Zaehner's F3b and F7b; see now JRAS, 1998, pp. 481–5Google Scholar); the Manichaean Xwāstwānīft (Zaehner's F3C); the Neo-Persian ‘Ulamā’ i islām (Zaehner's Z37); the Greek texts by Damascius (Zaehner's G1, ostensibly citing Aristotle's pupil Eudemus, but the Zurvanite content of the passage goes back, I think, not to Eudemus but to Damascius), Photius (G2, quoting Theodore of Mopsuestia), and Basil (G5). The Muslim testimonia (of which Zaehner gives only a very meagre sample) are mostly secondary sources (see below). Everything else in Zaehner's collection has nothing (or very little) to do with Zurvanism.

¹³ First, it seems, Pherecydes of Syros, who flourished in the sixth century B.C. and produced a mythological account of the origin of the world which involved three deities who “always existed”: Zas, Chronos and the female Chthonie, who later takes the name Ge. The first two are, as has been noted, merely playful adaptations of Zeus and Cronos respectively; see the discussion in Kirk, G. S. and Raven, J. E., The Presocratic philosophers, 4th edition (Cambridge, 1963), pp. 54–7.Google Scholar There are naturally some who have posited an Iranian origin of Pherecydes’ cosmology, but this is wild speculation.

¹⁴ I.e. writers on kalām and niḥal wa milal, whose knowledge of non-Muslim religions derived, as it seems, entirely from written sources, and ultimately for the most part from Christian heresographic literature. By contrast, the very well informed tenth-century Muslim writer al-Mas'ūdī, , in his Kitābu t-tanbīhi wa l-'išrāf, ed. de Goeje, (Leiden, 1894), pp. 93–4 (translated in Zaehner as text F12),Google Scholar quite righdy points out that although “the theologians of Islam and the authors of books on doctrines” ascribe to the Zoroastrians the doctrine “that God had a thought and that evil, that is to say the devil, came into being from his thought”, etc., in fact the Zoroastrians (majūs) deny believing any such thing. This is a rare exception to general assumption by mediaeval scholars (in East and West) that book learning is more valid than empirical observation.

¹⁵ For an overview of the extensive primary and secondary literature see Mikkelsen, G. B., Bibliogmphia Manichaica: A comprehensive bibliography of Manichaeism through 1996 (Turnhout, 1997).Google Scholar

¹⁶ The Book is preserved in the British Library manuscript Add. 14658, apparently from the seventh century, together with philosophical writings of Sergius of Rēš'aynā and translations (at least some of them by Sergius) of Greek works by (or ascribed to) Porphyry, Aristotle, Plato, Isocrates etc. The Book was first published by Cureton in 1833; the standard edition is by Nau, F. in Patrologia syriaca II (Paris, 1907), col. 492–657,Google Scholar with a Latin translation, and notes by Th. Nöldeke. The Syriac text has been reprinted, with an English translation (but with hardly any commentary) by Drijvers, H. J. W., The Book of the laws of countries, Dialogue on Fate of Bardaiṣan of Edessa (Assen, 1965).Google Scholar

¹⁷ Studies of Bardesanes include Schaeder, H. H., “Bardesanes von Edessa in der Überlieferung der griechischen und der syrischen Kirche”, Zeitschrift für Kirchengeschichte, 51 (1932), pp. 21–74;Google Scholar reprinted in his Studien zur orienlalischen Religionsgeschichte, ed. Colpe, C. (Darmstadt, 1968), pp. 108–161,Google Scholar with the editor's bibliographic “Nachwort”, pp. 265–273;Google Scholar Drijvers, H. J. W., Bardaiṣan of Edessa (Assen, 1966)Google Scholar (with a detailed survey, and partial reedition, of the primary sources); Teixidor, J., Bardesane d'Édesse, la première philosophie syriaque (Paris, 1992).Google Scholar

¹⁸ The standard study of Marcion and Marcionism is von Harnack, A., Marcion: das Evangelium vom fremden Gott (Leipzig, 1921),Google Scholar based mainly on the early Greek and Latin sources. The primary material used by Harnack has now been supplemented mainly by S. Ephraim's prose refutations of Mani, Marcion and Bardaisan, ed. and trans, by Mitchell, C. W. (London, 1912–1921),Google Scholar and by the new edition of Des heiligen Ephraem des Syrers Hymnen contra Haereses, ed. and trans, by Beck, E. (Louvain, 1957).Google Scholar A useful critical survey of recent work on Marcionism, with a full bibliography, can be found in Aland's, B. article “Marcion”, in Theologische Realenzyklopädie, XXII/1 (1991).Google Scholar Harnack's book is still the only comprehensive synthesis and remains of great value, though it is doubtless tarnished by the author's infamous attempt to portray Marcion as the precursor of a de-Judaized German protestantism, dispensing with the Old Testament as a source of revelation; interestingly, a more recent writer, Hoffmann, R. J., in his Marcion: On the restitution of Christianity (Chico, 1984),Google Scholar has gone to the opposite extreme of declaring Marcionism to be “fundamentally a Jewish heresy” (p. 307)Google Scholar. In reality, Marcionism is neither Jewish nor anti-Jewish, but represents an attempt at a literal reading and “fundamentalist” acceptance both of the Old and the New Testaments.

¹⁹ The ancient polemical sources do not give us a clear picture of how, exactly, Marcion envisaged the business deal between the two gods, but we have from the fifth century a colourful account of Marcionite cosmology and eschatology by the Armenian author Eznik. According to this, the just god does not at first realise that Christ has come from a different realm, but thinks he is one of his own rebellious creatures and crucifies him as a criminal. Christ then descends into the underworld and preaches to the souls of the dead. Then he rises from the dead and confronts the creator. At last the creator realises that this person is not of his world and that he has killed him unjustly, thus violating his own law. The penalty for murder is death, an eye for an eye, but the good god does not wish the death of the creator. Instead he strikes a bargain, by which, in compensation for his crime the just god agrees to sell the souls that he has created. Christ then returns to his own heaven taking with him the souls that followed him out of the underworld, but he promises to return later to collect the others. By contrast, Ephraem implies (Prose refutations ii, pp. 88–95,Google Scholar unfortunately one of the many passages where it is difficult to distinguish between genuine Marcionite doctrine and Ephraem's own polemical extrapolation) that the deal between the two gods was struck at the time of the transfiguration, when Christ met with Moses and Elijah, the emissaries of the creator. This would mean that the creator was aware of Christ's extra-cosmic status before the crucifixion, that he crucified him not in error, but as a wanton act of cruelty towards the son of the other god, for the pleasure of which he willingly exchanged the souls of his own creatures. Another question that must have concerned Marcion is why, one hundred years after the resurrection, the good god still had not collected the souls that he had purchased. Although the redemption is an unprovoked act of grace by the good god, independent of the will or action of mankind. Marcion did apparendy believe that each individual soul has the choice either to follow his new owner to the upper heavens or to stay behind with his creator. So perhaps the good god wanted to give mankind time to choose which path to follow? Or else he, out of fairness towards his business partner, did not wish to retrieve his new property, and thus put an end to the world, until the creator had had time to fulfill the promises that he had previously made to mankind with regard to his Messiah?

²⁰ Adversus Marcionem I, 15, 5: “proinde et creatorem in loco facit, utique eadem condicione censendo, et materiam ei subicit, utique innatem et infectam et hoc nomine aeternam, ut dominum, amplius et malum materiae deputans, innatum innatae, infectum infectae et aeternum aeternae, quartum iam hie deum fecit.” The difficulty with Tertullian's account is that here, as often, the author does not draw a clear line between what the Marcionites actually believed and his own parodistic reductio ad absurdum of the Marcionite premises. His purpose in this passage is to prove that, although Marcion avowedly taught the existence of two gods, in fact he “unwittingly” (I, 15, 6: “licet nesciens”) believed in nine gods: first his good god; then the space occupied by the good god, which is coeternal with him and consequently a “god” as well; then the matter out of which he created the upper heaven (if he created it from nothing, then the just god should also have been able to create from nothing; if the just god created his world out of preexisting matter, then logic demands that the good god must also have created out of matter …); then the four gods in the lower world (mentioned above) and finally the two Christs. Given this context, one would be tempted to think that Tertullian's distinction between materia innata and malum materiae deputatum innatum innatae is mere polemical pedantry were it not for the fect that Hippolytus explicitly states that “some of the Marcionites” did regard hyle and evil as two separate principles. This shows that, at least in this point, Tertullian is actually recording Marcionite doctrine. On the other hand, Tertullian's imputation that the good god also made his heavens out of matter (presumably matter unaffected by evil) is an enticing one, but it is not confirmed by other testimonies. Moreover, Tertullian's argumentation on this point is based on what is manifestly a spurious analogy between the upper and lower world, but Marcion's point is precisely that the stranger is not like the just god. It is curious that Ephraem (Prose refutations, i, pp. 134–5Google Scholar) similarly accuses Marcion of positing three “essences” (īθyē) in die upper world, namely the Stranger, his heavens and the space that surrounds him, though this seems to imply that the upper heavens were not created by the stranger, but are coeternal with him. And this is perhaps really what Marcion taught. Hippolytus (X, 19) presumably means the same thing when he says that all Marcionites agree that “the good one made nothing at all” that is, not even his own heaven.

²¹ For details, see the appendix.

²² Cf. Ephraem's, Prose refutations i, p. 122:Google Scholar “Those things which Bar Dayṣān makes five essences (īθyē), Mānē makes them from a single essenceīθy”. This seems to be the doctrine set out in most of the Manichaean cosmological texts. But the Arabic Manichaean source used by an-Nadīm, , Fihrist, ed. Tajaddud, , pp. 392–3,Google Scholar makes the elements the five pre-eternal “limbs” of the light-earth (’a‘dā’u ’arḍi n-nūr) which the Primal Man puts on as his armour when he sets out to do battle with the enemy.

²³ For the Manichaean use of Zoroastrian religious terminology see in particular Sundermann, W., “Namen von Göttern, Dämonen und Menschen in iranischen Versionen des manichäischen Mythos”, Altorientalische Forschungen, 6 (1979), pp. 95–133,CrossRefGoogle Scholar with copious references.

²⁴ For general orientation the work by Runciman, S., The medieval Manichee, A study of the Christian dualist heresy (Cambridge, 1947)Google Scholar is still worth reading, though in many points it has been superseded by subsequent research, and the author's staunch acceptance of the universal validity of the teachings of the Church of England will not be to the taste of all readers. For the Paulicians, still the most enigmatic of these denominations, see now Lemerle, P., “L'histoire des pauliciens d'Asie Mineure d'après les sources grecques”, Travaux et mémoires, V (1973), pp. 1–144;Google Scholar the interpretation by Garsoïan, N. G., The Paulician heresy (The Hague, 1967),CrossRefGoogle Scholar has not met with general approval.

²⁵ See Tacchella, E., “Les anciens pauliciens et les modemes bulgares catholiques de la Philippopolitaine”, Le Muséon, XVI (1897), pp. 68–90, 113–129, 209–223.Google Scholar I regret that the more recent literature in Bulgarian is not accessible to me.

²⁶ The Byzantine theologian Psellos (De operatione daemonum, 3; the passage is reproduced in Zaehner's Zurvan as text G7) ascribes a similar doctrine to the Christian sect known as Euchites (or Messalians), but it is possible that with “Euchites” he actually means Bogomils. There is no plausibility in Zaehner's contention that Psellos is talking about Zurvanites.

²⁷ For the Manichaeans and pseudo-Manichaeans in classical Muslim civilisation see in detail my article “Zindīk” in the Encyclopaedia of Islam.

²⁸ The of the Greek and Coptic sources is, of course, “air”, but in the Manichaean system this is not ordinary air, but the first and highest of the elements. The Middle Persian Manichaean texts render it with frawahr, a Zoroastrian term for the immortal soul, but translate the other four elements in the expected way as wād, rōšn, āb and ādur. The Arabic sources call the first element nasīm, “breeze”, one of the usual meanings of Syriac āar

²⁹ The element “light” is different from the principle “light” that is eternally opposed to “darkness”. This duplication of terminology is typical of Manichaeism.

³⁰ The identity of the Bardesanite and Manichaean theories of the five pure elements is an important discovery by Schaeder (see above, footnote 17), but not accepted, or indeed even mentioned, by Drijvers and Teixidor. The reconstruction offered here departs from Schaeder mainly in two points: I do not see any reason to identify Bardesanes' āar with aθrā, “space”, nor can I accept that there is any evidence for Stoic (as opposed to Peripatetic) influence on Bardesanes' system of elements.

³¹ E.g. Hippolytus, , Refutatio X, 32,Google Scholar outlining his own interpretation of the “true Christian doctrine”, gives a list of the four sub-celestial elements in which stands for “air”, and later, in Zoroastrian Middle Persian lists of the four (Greek) elements (as opposed to the seven creations of the Avesta), “air” is represented by wād; see the passages quoted by Bailey, H. W., Zoroaslrian problems in the ninth-century books (Oxford, 1943; reprinted 1971), pp. 88–9.Google Scholar

³² Drijvers, H. J. W., “Marcion's reading of Gal. 4,8: philosophical background and influence on Manichaeism”, in [Barg-i sabz] A green leaf, papers in honour of Professor Jes Asmussen, P. (Acta iranica, 28) (Leiden, 1988), pp. 339–48.Google Scholar

³³ Hymni contra haereses (see footnote 18), 48,2.

³⁴ rūḥā; Beck's rendering “Luft” is a lapsus.

³⁵ art. cit., p. 346.

³⁶ See Jansma, T., “Ephraems Beschreibung des ersten Tages der Schöpfung”, Orientalia Christiana periodica 37 (1971), pp. 294–316.Google Scholar