Studies on Early Arabic Philosophy [1 ed.] 1472420268, 9781472420268

Philosophy in the Islamic world from the 9th to 11th centuries was characterized by an engagement with Greek philosophic

295 66 67MB

English Pages 344 [345] Year 2015

Report DMCA / Copyright

DOWNLOAD FILE

Polecaj historie

Studies on Early Arabic Philosophy [1 ed.]
 1472420268, 9781472420268

Table of contents :
Cover
Series Page
Half Title
Title Page
Copyright Page
Table of Contents
Preface
Acknowledgements
Publisher's Note
Reception of Greek Philosophy
I: Arabic Philosophy and Theology Before Avicenna
II: Aristotle in the Arabic Commentary Tradition
III: The Last Philosophers of Late Antiquity in the Arabic Tradition
Galen and al-Rāzī
IV: Galen and al-Rāzī οn Time
V: Galen on Void
VI: Platonic Pleasures in Epicurus and al-Rāzī
VII: Abū Bakr al-Rāzī on Animals
Τhe Baghdad School
VIII: The Arabic Sea Battle: al-Fārābī on the Problem of Future Contingents
IX: Κnowledge of Universals and Particulars in the Baghdad School
X: Yaḥyā Ibn ʿAdī and Averroes on Metaphysics Alpha Elalton
Ανicenna
XI: On Knowledge of Particulars
XII: Avicenna and His Commentators on Human and Divine Self-Intellection
XIII: From the Necessary Existent to God
Index Nominum
Thematic Index
Index of Arabic Terms

Citation preview

Also

ίn

the

Varίorum

Collected Studίes

Serίes:

PETERADAMSON Studies οη Plotinus and al-Κindϊ

DIMITRI GUTAS Orientations of Avicenna's Philosophy Essays οη his Life, Method, Heritage

RICHARD SORABJI Perception, Conscience and Will

ίη

Ancient Philosophy

JOHNDILLON The Platonic Heritage Further Studies in the History of Platonism and Early Christianity

CHARLESBURNETT Arabic into Latin in the Middle Ages The Translators and their Intel\ectual and Social Context

RICHARD

Μ.

FRANK

Philosophy, Theology and Mysticism in Medieval Islam Texts and Studies οη the Development and History ofKalam, Vol. 1

RICHARD

Μ.

FRANK

Early Islamic Theology: The Mu 'tazilites and a\-Ash ' arϊ Texts and Studies οη the Development and History ofKalam, Vol. 11

RICHARD

Μ.

FRANK

C\assica\ Islamic Theology: The Ash 'arites Texts and Studies οη the Development and History ofKalam, Vol. 111

JULES JANSSENS lbn

Sϊna

and his ln:fluence

οη

the Arabic and Latin World

STEPHEN GERSH Reading Plato, Tracing Plato From Ancient Commentary to Medieval Reception

DIMITRI GUTAS Greek Philosophers

ίη

the Arabic Tradition

JOHN MARENBON Aristotelian Logic, Platonism, and the Context ofEarly Medieval Philosophy in the West

VARIORUM COLLECTED STUDIES SERIES

Studies

οη

Early Arabic Philosophy

Peter Adamson

Studies

οη

Early Arabic Philosophy

1~ ~~o~;~;n~~~up LONDON AND N EW YORK

First published 2015 by Ashgate Publishing Published 2016 by Routledge 2 Park Square, Milton Park, Abingdon, Οχοη 52 Vanderbilt Avenue, New York, ΝΥ 10017 Routledge

ίs

an

ίnψt·ίnt

of the

Taylω·

ΟΧ1 4

& F,-ancis

4RN

Groiιp,

an inforrna

biιsin ess

This edition © 2015 Peter Adaιη son Peter Adamson has asserted his right under tl1e Copyright, Designs and Patents Act, 1988, to be identified as the author of this work. All rights reserved. Νο part of this book may be reprinted or reproduced or utilised ίη any form or by any electronic, mechanical, or otl1er means, now known or !1ereafter invented, including photocopying and recording, or ίη any information storage or retήeval system, without permission ίη wήting from the publishers. Notice: Product or corporate names may be trademarks or registered trademarks, and are used only for identification and explanation without intent to infringe.

British Library Cataloguing ίη Publication Data Α catalogue record for this book is available from the British Library. The Library of Congress has cataloged the printed edition as follows: 2014960001

ISBN 13: 978-1-4724-2026-8 (hbk)

VARIORUM COLLECTED STUDIES SERIES CS\054

CONTENTS Preface

Vll

Acknowledgements

ΧΙ

RECEPTION OF GREEK PHILOSOPHY

Arabic philosophy and theology before Avicenna

58-82

The Oxford Handbook of Medίeval Phίlosophy, ed. J. Marenbon. Oxford: Oxford Unίversίty Press, 2012 ΙΙ

The Oxford Handbook ofArίstotle, ed. C. Oxford Universίty Press, 2012 ΠΙ

645- 664

Aristotle in the Arabic commentary tradition Shίelds.

The last philosophers oflate antiquity tradition Entre

the Arabic

1-38

et lα scίence greco-romaίnes, eds U. Rudolph and R. Goulet (Entretίens sur l Άntίquίte classίque, vol. LVΠ). Vandoeuvres : Fondatίon Hardt, 2011 Orίent

et

ίη

Oxford:

Occίdent: lα phίlosophίe

GALEN AND Αι-RΑzϊ

IV

Galen and

al-Razϊ οη

time

1-14

Thought: Essays ίn Honour of Frίtz Zίmmermann, eds R. Hansberger and C. Burnett. London: Warburg lnstίtute, 2012 Medίeval Arabίc

V

Galen

οη

197- 211

void

Phίlosophical

Themes ίn Galen, eds Ρ Adamson, R. Hansberger and J. Wilberdίng. London: Jnstίtute of Classίcal Studies, 2014

VI

Platonic pleasures in Epicurus and al-Raz1

71-94

ln the Age of al-Farabϊ: Arabίc Phίlosophy ίn the Fourth/Tenth Centuιy, ed. Ρ Adamson. London: Warburg lnstίtute, 2008

VII

Abίi

Bakr al-Raz1 οη animals

Archίv fiίr Geschίchte

der Phίlosophίe 94, 2012

249- 273

CONTENTS

νι

ΤΗΕ BAGHDAD ScHooL

VIII

The Arabic sea battle: future contingents Archίv fiir Geschίchte

ΙΧ

al-Farabϊ οη

the problem of

163- 188

der Phίlosophίe 88, 2006

Κnowledge ofuniversals and particulars Baghdad school Documentί

e Studί sulla

ίη

the

141- 164

Tradίzίone Fίlosofica Medίevale

18,

2007 Χ

Yal:ιya ibn Elatton Documentί

Άdϊ

and Averroes

οη Metaphysίcs

Alpha

343-373 e Studί sulla

Tradίzίone Fίlosofica Medίevale

21,

2010 ΑνιCΕΝΝΑ

ΧΙ

Οη

ΧΙΙ

273- 294

knowledge of particulars

Proceedίngs

of the Arίstotelίan

Socίety

105, 2005

Avicenna and his commentators οη human and divine self-intellection

The Arabίc, Hebrew and Latίn Receptίon ofAvίcenna s Metaphysίcs, eds D.N. Hasse and Α. Bertolaccί. Berlίn: Walter de Gruyter, 2011 ΧΙΙΙ

From the necessary existent to God

97-122

170- 189

lnterpretίng Aνίcenna, Cambrίdge

ed. Ρ. Adamson. Cambridge: Unίνersίty Press, 201 3

Index nominum

1-5

Thematic index

1- 2

Index of Arabic terms

1-6

This volume contains xii + 330 pages

PREFACE The history ofphilosophy in the Islamic world can usefu\ly be divided into two unequal parts. 1 The larger, yet less studied part, began in the Eastern heartlands ίη the 12th century or so, and is marked by a fusion of Avicennan philosophy with Islamic theology (kalam) and other intellectual cuπents, such as Sufism. By contrast, in what I would cal\ the "formative period," that is, the time up to Avicenna, we see extensive direct engagement with the Greek sources that had been rendered into Arabic during the great translation movement under the Άbbasids. These sources were not entirely forgotten after the 11th century; most notably, there would be a resurgence ofinterest in Graeco-Arabic texts ίη Safavid Iran. But typica\ly, thinkers ofthe post-formative period were usually engaging directly or indirectly with Avicenna. Αη exciting trend ίη recent research has been an increase in the attention paid to this later period. 1 can in ηο way claim to have been in the forefront of this development. One piece included here (§ΧΙΙ) does look at the commentaries οη Avicenna's Poίnters and Remίnders by two key figures ofthe 12th- 13th centuries, Fakhr al-Dϊn al-Razϊ and al-Tίisϊ. But for the most part my publications have dealt with thinkers up to, and including, Avicenna. My first forays into the field dealt with early texts, focusing οη the Arabic reception ofNeoplatonism and οη the fιrstfaylasίif to write in Arabic, al-Κindϊ. Papers οη these subjects have been collected in a separate volume in the Variorum series, Studίes on Plotίnus and al-Kίndϊ (Farnham, 2014). The present volume brings together contributions οη subsequent figures of the formative period. Most of the leading thinkers of the era appear: the provocative doctor-philosopher Abίi Bakr al-Razϊ, the so-called "Baghdad school" of Aristotelians including al-Farabϊ, and Avicenna himself. Though I usually prefer nowadays to speak of "philosophy in the Islamic world," for the title ofthis volume I have used the phrase "Arabic philosophy." This is intended to highlight the importance ofGreek-Arabic (or in some cases Greek-Syriac-Arabic) translations for the papers included here. Ιη every one

' This divi sion is used in my History of Philosophy podcast, which (i f a\\ goes well) will appear in print version with Oxford Uni versity Press as Α His tωy of Philosophy Without Any Gaps: Phίlosophy ίn the Jslamίc World. There I lr eat pbi\osophy in Anda\usia as a third, distinctive part of the story.

PREFACE

νιιι

of these articles, the influence of Greek philosophical sources is relevant; in a number of cases it is a chief object offocus. Unsurprisingly, a number ofthe pieces look at how Aristotle was received by formative-period philosophers. §11 is a general introduction to this topic, but usually I am discussing the reception of specific passages and themes from the Aristotelian corpus. These papers mostly resulted from an interest in the Aristotelian material itself, and curiosity about how a given idea was received in the Arabic tradition. For instance I wound up writing §VIII, which looks at how al-Farabϊ understood the famous "sea battle" argument for determinism, simply because I went looking for a dedicated treatment of al-Farabϊ's commentary οη it and couldn't find one. Likewise, 1 was struck by the (to us rather counterintuitive) thesis of Aristotle's Posterίor Analytίcs that knowledge properly speaking must be universal in scope. This led me to write papers §ΙΧ and §ΧΙ, both of which examine the impact of this notion in the formative period. Ιη the latter article, 1 argue that it forms the crucial background for understanding Avicenna's infamous view οη God's knowledge of particulars. By contrast, the discussion in §Χ of two Arabic commentaries on Metaphysίcs Alpha Elatton was provoked more by the close attention given to that text in the Islamic world, where it was (for reasons that still remain unclear) treated as the first book ofthe Metaphysίcs. For me, part ofthe appeal ofthe formative period is the continuity between the \ate antique and Arabic traditions. The commentators οη Aristot\e feature ίη several ofthe papers included in this volume, and §ΠΙ is devoted to showing how two major philosophical debates ίη \ate antique philosophy lived οη in Arabic. This is, of course, not to say that the thinkers of the Islamic world were derivative or unoriginal. Το the contrary: §VΠΙ argues that al-Farabϊ provides insights about the sea batt\e problem that had been absent, or at least far less explicit, in the \ate antique commentaries. Ιη §ΠΙ we see how a change of cultural context \ed to a different way of understanding the classic Aristotelian idea that logic is an "instrument" and not "part" ofphilosophy. Still, the authors of the Baghdad school, and their like-minded Anda\usian heir Aveπoes, are recognizably working within a commentary tradition that extends back to antiquity. The same cannot be said for Avicenna, or at least not without significant caveats. As Robert Wisnovsky has nicely remarked, Avicenna stands at the end of a synthetic approach to philosophy that can be traced back to the school of Ammonius ίη late antiquity, and begins a new period grounded ίη his own, original synthesis. 2 There were a few significant, and pointed,

2

The

R. Wisnovsky, ' Άvicenna and the Aristotelian Tradition," in Ρ. Adarnson and R.C. Taylor, to Arabίc Phίlosophy (Carnbridge, 2005), 92-136, at 132. For a

Cambrίdge Companίon

PREFACE



exceptions: Aveπoes and Άbd al-Laμf al-Baghdadϊ both lamented the way that Avicenna had changed the philosophical landscape, and sought to return to a more authentically Aristotelian paradigm. But their reactionary project was a failure as concerns the Islamic world (within the Jewish and Latin Christian traditions, Aveπoes was of course wildly successfu\). This was despite the fact that Avicenna himself c\osely engaged with Aristotle and his commentators, as has been shown masterfully by Amos Bertolacci and, much more modestly, by two ofthe articles included here (§ΧΙ and §ΧΙΙ). 3 With item §ΧΙΙΙ, though, we see Avicenna striking out more or less οη his own to devise his new conception of God as the Necessary Existent, which will be one of his most important legacies. Το this extent, the picture offered ίη this col\ection of papers is a broad\y familiar one. Aristotle, mediated by late ancient commentators, has an unparalleled impact οη philosophers of the formative period, who respond to him in original and sometimes surprising ways. That picture is accurate as far as it goes, but needs to be complicated ίη at \east two ways. First (and Ι have certainly not been alone ίη urging this), we should not underestimate the importance and intrinsic philosophical interest of kalam ίη the early period. We are accustomed to contrast the falsafa and kalam traditions, even if we may adrnit that this contrast becomes bluπed to the point of non-existence ίη the post-formative age. But in an earlier paper, 4 Ι argued for the wide-reaching relevance of Μυ 'tazilite theology ίη the thought of al-Kindϊ, and §Ι attempts to generalize the point for other formative-period thinkers. Kalam was always an irnportant foil and inspiration for the falasifa, even those who came before Avicenna. This piece also emphasizes the need to integrate Christian and Jewish philosophers into our naπative , a point also reflected ίη §ΙΧ. The other complication has to do with a kind of secondary transmission of philosophy from Greek into Arabic. The Greek protagonists in this story are not Aristotle and the Neoplatonists, but Plato and Galen. This became clear to me only when Ι undertook several preparatory studies for a book Ι am writing about Αbϋ Bakr al-Razϊ. (Or at least, Ι keep telling people Ι am writing it.) Because al-Razϊ saw the Greek legacy from the viewpoint of a doctor, rather than through the Aristotelian spectacles worn by an al-Farabϊ or Ibn Άdϊ, his conception ofphilosophy was deeply Galenic. It was through Galen that al-Razϊ came to admire Plato, whom he names as the main inspiration for his famously further example oftbe continuity in the commentary tradition, see a recent study Ι co-authored with Wisnovsky: 'Ύa~ya Ibn ' Αdι on the Location of God," Oxford Studies ίn Medieval Philosophy 1 (2013), 205- 28. 3 Α. Bertolacci, The Receptίon of Aristotle's Metaphysίcs ίn Avίcenna's Kitiib al-Sifli': α Milestone of Western Metaphysical Thought (Leiden, 2006). 4 Studies on Ρ/οtίnιιs and al-Kindί, §ΧΙΙ.

PREFACE

χ

unorthodox cosmology. Four papers included here emerged from my research into this phenomenon. §IV and §V deal with the Arabic evidence concerning Galen himself, and effectively lay the groundwork for understanding al-Razϊ's theory of time and place. §VI and §VII focus more οη al-Razϊ himself. §VI challenges a prominent interpretation of him as a hedonist, by arguing that he ίη fact adopts Plato's understanding of pleasure as that which accompanies a transition from a harmful to neutral state. §VII explores al-Razϊ's attitude towards animals, urging that he should receive due credit for his remarks regarding the ethical treatment of animals. 1 try to understand these remarks against the background of his medical and psychological theories. This piece also emphasizes the relevance of the kaliim background for understanding alRazϊ.

Of course, the papers collected here were written for various purposes and over a number ofyears (from 2006 to 2014). Nonetheless I hope that together, they offer an illuminating panorama of philosophy ίη the formative period. This is not just because of the variety of figures discussed, but also because of the variety of philosophical topics that arise. These range from problems ίη Iogic (§ΠΙ, §VIII) and epistemology (§ΙΧ, ΧΙ), to natural phi\osophy (§IV, V), to ethics (§VI, VII), to metaphysics and philosophical theology (§Χ, §ΧΙΙ, §ΧΙΙΙ). Ιη this respect the col\ection exemplifies a major advantage ofworking ίη the history ofphilosophy. lt is simply assumed that historians will be ready and willing to tackle such various areas ofphilosophy, whereas the high degree of specialization ίη contemporary philosophy means that non-historians will often focus οη just one or two branches of the field. Since pre-modern figures lived before the era of philosophical specialization, such naπow focus makes ηο sense for the historian. 1 can't claim to be equally interested ίη al\ areas of philosophy: political philosophy is a notable absence ίη both this and my other Variorum volume. But for whatever reason (short attention span?), 1 do find it intriguing and enjoyable to explore different philosophica\ problems ίη a historical setting, and I hope that my enthusiasm comes across ίη the papers collected here. PETER ADAMSON Munίch,

January 2015

ACΚNOWLEDGEMENTS

Grateful acknowledgement is made to the following persons, institutions, journals and publishers for their kind permission to reproduce the papers included in this volume: Oxford University Press (for articles Ι and ΙΙ); Fondation Hardt, Vandoeuvres (ΠΙ) ; Warburg Institυte, London (IV, VI); Wiley Publishing and the Institute ofC\assical Studies, London (V); Walter de Gruyter GmbH, Berlin (www.reference-global.com) (VII, VIII, ΧΙΙ); SISMEL Edizioni del Galluzzo, Florence (ΙΧ , Χ) ; the Aristotelian Society, London (ΧΙ); and Cambridge University Press (ΧΙΙΙ) . Every effort has been made to trace all the copyright holders, but if any have been inadvertently overlooked the publishers will be pleased to make the necessary aπangement at the first opportunity. Thanks a\so to Fedor Benevich for the preparation ofthe indices.

PUBLISHER'S

ΝΟΤΕ

The articles in this volume, as in al\ others ίη the Variorum Collected Studies Series, have not been given a new, continuous pagination. Ιη order to avoid confusion, and to facilitate their use where these same studies have been refeπed to e\sewhere, the original pagination has been maintained wherever possible. Each article has been given a Roman number in order of appearance, as listed in the Contents. This number is repeated οη each page and is quoted ίη the index entries.

I

ARABIC PHILOSOPHY AND THEOLOGY BEFORE AVICENNA

IN the year 833 CE, year 218 of the Islamic calendar, the caliph al-Ma'mίιn (reigned 813-833) wrote a letter laying down a point of Islamic doctrine. According to the caliph, the Qur'an is created, and not eternal as held by some theologians of the time. Judges and scholars were to be interrogated as to whether or not they accepted this doctrine of the Qur'aήs createdness. Thus began the infamous miJ:ιna, a 'test: or more sinisterly, 'inquisition: during which the political authority of the Άbbasid caliphs would lend support to what seems to be a rather abstruse question ίη Islamic theology. The policy was abandoned by the later Άbbasid caliph al-Mutawakkil (r. 847-61), and thereafter it became orthodox to hold that the Qur'an is co-eternal with God. The miJ:ιna was an atypical event ίη the history of Islam. Thereafter, theological orthodoxy ίη sunnί Islam was normally settled by the consensus of scholars and the community, rather than by the unilateral decision of any political ruler. ' This may seem to have little to do with philosophy. Yet it was under the same caliph, al-Ma'mίιn, that the political elite sponsored the massive translation movement that rendered works of Greek philosophy and science into Arabic. Later sources, influenced by al-Ma'mίin's propaganda, tend to give him credit for initiating this translation movement. Ιη fact, the second Άbbasid caliph, al-Man~r (r. 754-75), already began to put serious resources behind the translations.2 Still, there is ηο doubting that philosophy and theology flourished under al-Ma'mίin. During his reign al-Κindί (d. after 866), usually recognised as the first philosopher of Islam, began his career. Al-Ma'mίιn's policy of support for the translation

Reproduced by permission ofOxford University Press, USA.

I 59

ARABIC PHILOSOPHY AND THEOLOGY BEFORE

Α VICENNA

movement was continued by his successor al-Mu 'ta~im (r. 833-42). He does not seem to have had much of an intellectual bent himself, but al-Κindi served as tutor to his son Aηmad, and the caliph was the addressee of al-Kindί's most important work, On First Philosophy. Whether he appreciated its intricate metaphysical arguments is, to put it mildly, doubtful. AI-Mu 'ta~im also continued the mi!Jna, albeit somewhat half-heartedly, and it was imposed with enthusiasm by his son and successor al-Wathiq (r. 842-7) before al-Mutawakkil put a stop to the policy in 848-9. Of course, it was not the caliphs who came up with the idea that the Qur' an was created rather than eternal. It was a thesis held by a range of theologians, especially those we call the Mu 'tazila. These were rationalist thinkers who seem to have been influenced by Greek philosophical ideas. The ninth century was their heyday, as part of a more general cultural flowering under the Άbbasids. 3 After its brief period of influence, Mu 'tazilism would never again have the kind ofbacking it received from these 'Abbasid caliphs. Ιη the later history oflslam, we never find such a confluence of centralised political power, Greek-inspired philosophy and rationalist theology as occurred in the ninth century. Yet into the tenth century and beyond, philosophers and theologians were in intellectual competition. They often reached similar conclusions οη the same topics, albeit by following different paths. This chapter will sketch some prominent issues ίη both traditions, looking especially at cases where philosophers of the ninth to tenth centuries responded directly to theological discussions. First, a word about this contrast between 'philosophy' and 'theology'. Ι use the word 'philosophy' narrowly, as a translation of the Arabic falsafa. As you might guess by looking at this term, it is itself a loan word from the Greek philosophia. The etymology is not misleading:falsafa means direct engagement with the Greek works translated into Arabic under the ' Abbasids. Of course, one could instead use the word 'philosophy' more broadly, as including all texts and authors that strike us as having philosophical interest. Ιη this broad sense, much of the theological material of our period is 'philosophical'. Theologians of the period argued about atomism, free will, morality, divine attributes and so οη . 4 The narrower sense, though, is faithful to the outlook of the medieval authors themselves, who distinguish between theologians and 'philosophers (jaliisifa)'. The Arabic expression Ι render as 'theology' is ' ilm al-kaliim (or just kaliim, with theologians referred to as mutakallimun ). Literally this means 'science of the word; which may be an allusion to the debate over the status of the Qur 'an as God's word, though the exact origins of the term are obscure. 5 Το some extent the boundary between falsafa and kaliim is easy to mark: the faliisifa explicate Greek sources and use these as the basis of their doctrines, whereas the mutakallimun explicate and use the Qur ' an and !Jadιth (reports about the sayings and deeds of the Prophet). But the line is often blurry. Some figures seem not quite to belong to either camp, for instance, one of the thinkers we will examine below, Αbϊι Bakr al-Razί (d. 925). Other authors used philosophical materials to defend distinctive theological positions within Islam, especially a branch of shί Ί Islam called Isma'ϊlism . 6 Furthermore, 'philosophical' authors like a\-Κindi do cite and explicate

I 60 the Qur'an, while Greek influence is often detected ίη kalam . Admittedly, ίη kalam this influence is subtle, whereas Greek ideas animate the entire output of philosophers like al-Κindi or al-Farabi. More fundamental than this question of sources, but also more controversial, is the question of how philosophical discourse relates to theological discourse, especially when the two traditions address themselves to the same topics. As we will see, the Aristotelian thinker al-Farabi (d. 950) draws a strong contrast between falsafa and kalam, identifying the latter as 'dialectical' and hence inferior to the 'demonstrative' discipline offalsafa. Yet in the wake of Avicenna οη the falsafa side and al-Ash ' ari (d.935) οη the kalam side, it becomes increasingly common to see authors combiningfalsafa and kalam into a hybrid discipline that, for lack of any better expression, we may as well call 'philosophical theology'.

1. ΤΗΕ Mu'τAZILA AND ΤΗΕ AsHΆRITE CRITIQUE The Mu 'tazila were known as 'the upholders of justice and oneness'. Ιη this slogan, 'justice (' adl)' stands for their c\aim that God does ηο wrong, and Όneness (taw]:ιid)' for their claim that God is one. Of course, no Muslim would reject either claim, when put this baldly; what is distinctive about the Mu 'tazila is the way they interpret justice and oneness. Το begin with justice, the Mu 'tazila held that human reason is capable of discriminating good from evil. For instance, even without revelation we can know that we should acknowledge and show gratitude to God. 7 This still sounds rather innocuous. The position becomes more controversial when the Mu ' tazila point out that if moral strictures are knowable by reason, then even God is bound by them. Some Mu ' tazilite thinkers held that God cannot do what is unjust, others that he can do so but never does. Opposing views were held οη this question by the two greatest exponents of Mu ' tazilism in the first half of the ninth century, Αbϊι 1- Hudhayl (d. 841 ?) holding the latter view and his student and nephew a\-Na~am (d. 836) holding the former.8 Either way, it is not up to God to decide what will constitute justice. Rather, God performs justice because it is just. Το put this in the language of contemporary philosophy of religion, this means that the Mu 'tazila reject a divine command theory of ethics. The Mu 'tazila highlight one particular moral requirement that God must observe. It would, they say, be unjust for him to punish creatures for sinning, if they lacked the ability (qudra) to refrain from sin. Thus, their position οη justice leads directly to another signature doctrine, their endorsement of human free will. The Mu 'tazilites anticipate modern libertarian conceptions of freedom to a striking degree, even insisting that moral responsibility requires the presence of alternative possibilities. Thus, we find al-Na~~am saying that a genuine choice (ikhtiyar) requires the presence of two inclinations (khafirani), one that would lead the agent to perform the action, the other leading the agent to refrain from ίt. 9 His point

I 61

ARABIC PHILOSOPHY AND THEOLOGY BEFORE Α VICENNA

would seem to be that, if the agent has only a motivation to act, he will inevitably act, and this inevitability will eliminate freedom of choice. 10 Yet alongside these resonances with contemporary philosophical debates about free will, we find a preoccupation with distinctively theological concerns. Α great deal of attention was directed, for instance, to the question of whether or not human actions are 'created' by God. If so, then God would wind up creating evil actions; but if not, then humans wou1d seem to be usurping God's unique status as Creator. The Mu 'tazila avoid the first horn of this clilemma but, as their opponents pointed out, risk being impa1ed οη the second. 11 The Mu 'tazilite position οη God's oneness was ηο less contentious. As with divine justice, divine oneness or taw/:ιϊd ίs a common commitment of all Muslims. The shahada itself, the affirmation ofMuslim belief, asserts that 'there is ηο God but God: God's uniqueness rules out the existence of other entities that have a divine status. God can have ηο 'partners: and it is unbelief to recognise such partners (hence the term shirk, for polytheism, from the verb sharaka meaning to share or associate with). From this fundamental teaching of Islam, the Mu ' tazila inferred that there can be ηο multiplicity within God. Here Christianity was a clear target. Trinitarian doctrine was seen as an obvious violation of taw/:ιϊd, and Muslim theologians of a11 persuasions delighted in comparing their opponents' positions to the doctrine of the Trinity. The Mu 'tazila and some other theologians of the period argued that accepting multiple clivine attributes (~ifat), such as justice, knowledge or power, wou1d violate taw/:ιϊd just as much as belief ίη the Trinity. There was a further reason for the Mu ' tazila to be suspicious of rea1 divine attributes: it says ίη the Qur'an (42:9) that God 'has ηο like' among created things. Obviously a man may be just, knowing and so οη . Το avoid saying that a just, knowing man is 'like' God, it seems we must say either that God is not just and knowing, or that if he is just and knowing, it is in a quite different way from what we find in humans, so clifferent, indeed, that there is ηο Ίikeness: Some mutakallimίΊn, such as .Qiriίr ibn Άmr, Mu 'ammar, and al-Na~~iίm, adopted a version ofthe former solution. Anticipating the 'negative theology' of the great Jewish thinker Maimonides, they explain that apparently positive divine predications are, in fact, concealed negations: to say that God is knowing is to deny that he is ignorant, for example. 12 Obviously this doctrine tends to make God unknowable since even positive statements about him are unmasked as mere denials. This consequence was highlighted by opponents of the Mu 'tazila, who accused them of going so far in avoiding tashbϊh (the 'assimilation' of God to creatures) that theyfell into ta ' tϊl ('voiding' God's attributes). 13 But Mu 'tazilite authors adopted a variety of ways to strike a ba1ance between ta ' tϊl and tashbϊh, to affirm divine attributes without violating taw/:ιϊd. Here the most prominent view is that of ΑbίΊ 1-Hudhayl, who held that God's attributes are identica1 with God Himself. He said that God 'is knowing through a knowledge which is He (huwa 'alim bi-' ilm huwa huwa): 14 and likewise for the other attributes.15 This may seem to be a merely verbal so\ution-a suspicion that arises not infrequently when \ooking at kalam doctrines. That is, it seems ΑbίΊ 1-Hudhayl wants to preserve words like 'just' and 'knowing' because of their usage ίη the Qur' iίη, and he simply asserts that God is the

I 62

same as his attributes to prevent any multiplicity from arising. But there is more to the doctrine than meets the eye. Consider, for example, the implications of ΑbίΊ 1-Hudhay\'s view for divine justice. If God's justice is identical to God, then it will not be contingent, chosen from among various possibilities. Rather, divine justice will be nothing other than God. This helps explain why ΑbίΊ 1-Hudhayl rejected the divine command theory. The point also shows that ΑbίΊ 1-Hudhayl is not vulnerable to a criticism often made of the Mu 'tazilite theory of justice, namely, that if the laws of justice are discoverable by reason, they must be something independent of God that He must obey.16 The Mu 'tazilite doctrines of justice and oneness were, then, c\osely related. Taken together, they undergird further doctrines οη a wide range of theological issues, including the one that was at stake in the miiJna: the createdness of the Qur ' an. Again, the connection may not be immediately obvious. The origins of the controversy seem to lie ίη the eighth century, when some theologians designated the Qur ' an as God's word or 'speech: As such, the Qur 'an belongs intimately to God. Certain other theologians, forerunners of the Mu 'tazila, detected a violation of tawiJϊd. If God's word is distinct from God, and yet uncreated, then it will be a second divine thing a\ongside God himself-which is polytheism. It was only at this point that the debate was put in the terms we find in al-Ma ' mun's letter, namely, that the Qur' an is not 'eternal (qadϊm): 17 The point here is to assert the absolute contrast between God and everything other than God, by preserving 'eternity' as the prerogative of the divine and calling all other things 'created: Mu ' tazilite thinkers would also have seen the issue of human freedom as relevant. This is because some of Mu.! Jammad's opponents and detractors are singled out by name ίη the Qur ' an as sinners who will be punished in hell. If the Qur ' an were eternal, then it would seem to follow that these individuals did not sin freely but were determined from eternity to sin. Ιη that case, punishing them would violate God's justice. Of course, not al\ theologians ίη the ninth/tenth century subscribed to these doctrines. For one thing, the Mu 'tazila were only one group of theologians and were bitterly opposed by other factions even before the searching criticism of the lapsed Mu ' tazilite theologian al-Ash ' ari. Ιη particular, their rather negative approach to divine simplicity was diametrically opposed to the 'traditionalist' view, represented most prominently by Ibn J:Ianbal, who ran afoul of the miiJna when he refused to accept the Qur' an's createdness. 18 Traditionalists not only accepted the standard divine attributes, but they also took more or less at face value anthropomorphic passages ίη the Qur ' an that talk of God's having a face or sitting upon a throne. Mu 'tazilite interpretations of such passages, of course, were \ess literal. Furthermore, different theologians who are retrospectively grouped under the rubric of'Mu 'tazilism: ίη fact, disagreed οη a range of issues. Theological controversy was as likely to take place between opposed Mu 'tazilites as between Mu 'tazilites and traditionalists. Finally, it was common for theologians to adopt some 'Mu 'tazilite' views whi\e rejecting others.19 It actually took several generations for a group of rationalist mutakallimun to group themselves together around the banner of 'justice and oneness: Mu ' tazilism is usually said to begin with a stance adopted by a man named Wa~il ibn Άμ ' (d. 748). The question at issue was whether a 'grave sinner' should be considered to be a

I 63

ARABIC PHILOSOPHY AND THEOLOGY BEFORE Α VICENNA

'believer' or a 'hypocrite: Wii~l adopted the conciliatory view that such a sinner is ίη an 'intermediary position: that is, neither a believer nor a hypocrite. Ίhe name 'Mu 'tazila: meaning 'those who withdraw: is traced to a supposed event where Wii~il asserted this doctrine and hence 'withdrew' from those gathered around a1-I:Iasan al-Ba~rι (d. 728). 20 Wii~il's 'intermediate position' was named by Abu 1-Hudhayl as one of the five chief doctrines of the Mu' tazila, along with the commitments to God's oneness and justice, the 'promise and the threat' (i.e., of divine reward and punishment) and the command to do good and avoid evil. But this belies the complexity of the subsequent tradition. Even leaving aside the spread of Mu ' tazilite doctrines beyond the heart of the 'Abbiisid empire ίη Iraq, 21 the ninth century saw the development of two Mu ' tazilite 'schools: one ίη Baghdad and one ίη Basra, which differed οη numerous points despite their agreement to the basic doctrines described above. We are better informed about the tradition ίη Basra. After the early theorists Abu 1-Hudhayl and al-Na~~am, who may take much of the credit for the development of Mu' tazilism into a sophisticated theology, we have something more like a Basran 'school' initiated by the father and son Abu ' Αlι (d. 915) and Abίi Hiishim ibn al-JubbiiΊ (d. 933). His followers included the Qiigί Άbd al-Jabbiir (d. 1024), whose voluminous Mughni fi abwab al-tawlJid wa-l- 'adl ( Comprehensive Work on Topics of Oneness and Justice) displays mature Mu 'tazilism in all its scholastic glory. 22 Α few Mu ' tazilite authors from the earlier period have left extant works. 23 But unfortunately, most of the authors just mentioned are known to us chiefly through the reports of later more or less hostile heresiographers. Among these hostile authors, the most important is al-Ash 'arι.24 He began his theological career as a student of Abu Άlι al-JubbiiΊ but then turned against Mu 'tazilism. He and his followers, the Ash ' arites, accepted the existence of divine attributes that are really distinct from God. Ίhese attributes are eternal but, because they are ίη some sense dependent οη God, the doctrine leaves room for contingency. Ιη fact, for Ash 'arite mutakallimίίn God's actions are wholly unrestricted, and the nature of justice is determined by His choices rather than vice versa. Ίhis Ash ' arite tendency to make all things dependent οη God's unconstrained will has consequences for human freedom . Ίhe Ash ' arites were uncomfortable with making humans effectively the 'creators' of their own actions. Ίhey instead developed a doctrine of 'acquisition (kasb); according to which God creates our actions but we 'acquire' them and thus take upon ourselves the responsibility for those actions. 25 It is tempting to understand Ash 'arism as little more than reverse Mu 'tazilism, an acceptance of everything al-Ash ' arι's masters had rejected and vice versa. But οη several points, we can find continuity between the two schools. For example, both sorts of kalam had a tendency towards irenicism. As we saw, the very name 'Mu 'tazila' may refer to a conciliatory position taken οη a politicised theological issue. For aJI the controversy that raged around various Mu 'tazilite views, most members of the school sought to resolve intellectual conflict rather than to stoke it. 26 (Αη exception should probably be made for al-Na~~iim, who is noted for his bold and provocative theses.) Consider the most sensitive issue of the ninth century, namely, the claim of unique political and religious authority made by the descendants of 'ΑΗ, the cousin

I

of the Prophet. Like the caliph al-Ma ' mun, Mu 'tazilites tended to make conciliatory gestures towards ' Alid sympathies, without actually endorsing the doctrine of the shϊ ' a, or 'party' of Άlϊ. 27 Another example is the Mu 'tazilite habit of giving merely 'verbal' answers to theological dilemmas. Such answers may annoy philosophicallyminded readers, but they need to be seen ίη context. These careful formulations were, arguably, designed to satisfy as many people as possible, much as the studied ambiguity of a political manifesto enables it to represent the diverse views of party members. We find a similar tendency among Ash 'arites, who sought a broad consensus among the sunni community οη theological issues, often by leaving questions open rather than insisting οη any particular solution. It is in this light that we might understand their use of the slogan bi-lii kayf 'without [saying] how', which was applied, for instance, to the way in which descriptions of God ίη the Qur ' an hold true. 28 Another manifestation of this irenic sensibility is the emphasis both schools placed upon reason ('aql) . Of course, the Qur'an and IJadϊth were common ground for all Muslim scholars, even if there was controversy about which IJadϊth were genuine. Kaliim, like the Islamic jurisprudential tradition with which it was intertwined, rarely strayed far from expounding these sources. But the interpretation of religious texts lay at the heart of various disputes, and reason offered an independent means of finding a resolution. The Mu 'tazilites duly placed great trust ίη reason, regarding it as a gift from God that empowers humans to discern many things even without the aid of revelation. They held that humans grasp good and evil through reason alone, as we've seen, and οη other issues, too, they appea\ed to universa\ intuitions as well as revealed sources. 29 They devoted considerable attention to the rules of reasoning, engaging ίη detailed technical disputes that one can only describe as logical in character. 30 With their tactic of bi-lii kayf and their conviction that good and evil are determined only by God's will, the Ash 'arites seem to have been more impressed by reason's limits than by its powers. But Ash ' arism has its own rationalist commitments, as shown by its most famous exponent, al-Ghazalϊ (d. 1111). AI-Ghazalϊ's famous corrective to philosophy consisted in a careful delineation of what reason can achieve, not in a rejection of rationality. One sign of this is his disdain for those who reject the utility of Aristotelian logic. 3 1 Mu ' tazilite and Ash ' arite kaliim is also unified by similar ideas about the metaphysics of created objects. As usual, the details differ from thinker to thinker, but for almost al\ the mutakallimun mentioned above, created things consist of 'substances (jawiihir)' that have 'attributes (§ifiit)', often called 'accidents (a 'rii4)'. Despite the Aristotelian terminology, this was a distinctly un-Aristotelian theory, according to which created substances are atoms ('indivisible parts') to which God joins accidental properties. Substances are preserved ίη existence moment-to-moment by divine power, so that God's creative activity continues at all times rather than being limited to an initial act that brings the world into existence. This general framework was broadly agreed, but the Basran Mu 'tazilite school ίη particular engaged ίη refined analysis of the different sorts of attributes that can inhere ίη substances. 32 The Ash ' arites, meanwhile, used the

I 65

ARABIC PHILOSOPHY AND THEOLOGY BEFORE Α VICENNA

atom/attribute ontology to emphasise once again the complete freedom and power of God. For them, God may join attributes to substances however he wishes, so long as he does not violate the law of non-contradiction (he cannot make something both a\ive and dead) . As made c\ear ίη a\-Ghazalϊ's famous discussion of miracles in his Incoherence of the Philosophers, on this view the stability of the created world depends οη the fact that God's habits never change, not οη the intrinsic features of created beings. 33

2. ΤΗΕ MUSLIM FALASIFA Given the political resonance, intellectual sophistication and central geographical location of the kalam activities just surveyed, it is hardly surprising that the philosophers of ninth-to-tenth century Islam felt obliged to respond to the mutakallimun. One could multiply examples, but Ι will concentrate οη the three most famous Mus\imfalasifa before Avicenna, namely, al-Κindι, al-Farabι and al-Razι. (Some might hesitate to call al-Razι a 'Mus\im' faylasuf; Ι will argue that one should instead hesitate to call him a Muslim 'faylasuf.) Ιη Section 3, Ι will discuss how two philosophers from other faiths, but within the same cultural milieu, engaged with Islamic kalam. As already mentioned, al-Kindι's career peaked under the very caliphs who imposed the Mu 'tazilite doctrine of the Qur ' an's createdness upon their subjects. 34 Al-Κindι never explicitly alludes to the doctrine in his extant writings. Nor, ίη fact, does he speak directly about kalam as an intellectual phenomenon.35 Nonetheless, al-Κindι's work was implicitly engaged with the Mu ' tazila. 36 Α particularly clear case is the topic of divine simplicity. Al-Kindϊ is best known for coordinating a circle of (mostly Christian) translators who rendered texts such as Aristotle's Metaphysics, Plotinus's Enneads and Proclus's Elements of Theology into Arabic. 37 Already in the Kindϊ circle version of the Enneads, we find a preoccupation with the problem of God's 'attributes ($ifiit)'. Plotinus's stress οη the ineffability of the One naturally lent itself to the Kindϊ circle project of making Greek philosophy relevant to contemporary theological concerns. Ιη the Arabic Plotinus, numerous passages recast Plotinus's doctrine of ineffability as a denial of attributes. Using terminology that has attracted much discussion, the translator identifies the Plotinian One-now called 'the Creator' -as 'nothing but being (anniyya faqaO: with ηο attributes superadded to divine being or essence. 38 God's unity is a central issue in al-Kindϊ's own works, too, especially the aforementioned On First Philosophy. 39 Sections 3 and 4 of the extant first part dea\ with God as the 'true One', by first proving the existence of a principle of unity, and then showing that this principle is not subject to any form of predication. 40 This understanding of God is strikingly similar to the Mu 'tazilite conception, οη which God's unity (tawήιd) ru\es out the application of attributes to the divine essence. Al-Κindϊ's

I 66 refutation of the world's eternity ίη the previous section of On First Philosophy should be understood ίη the same context. 41 The Mu ' tazila distinguished God from all other things (including the Qur ' an) by saying that whereas the latter are created, God alone is eternal. Al-Kindi follows their lead, even at the price of disagreeing with Aristotle οη the topic of the world's eternity. 42 However, al-Kindi's procedure throughout On First Philosophy is selfconsciously philosophical, with ηο recourse to revealed texts.4 3 Only at the end of Section 4 does al-Κindi reveal, with a flourish of Qur 'anic epithets, that the true One is to be identified with the almighty Creator of Islam. Prior to that, he mostly employs ideas and terminology drawn from basic works of Aristotelian logic, especially Porphyry's Isagoge. He uses the same materials elsewhere to refute the Christian doctrine of the Trinity.4 4 Αη even more striking deployment of Aristotelian materials ίη a 'theological' context is al-Κindϊ's epistle On the Prostration of the Outermost Sphere.4 5 Taking as his text the Qur ' anic statement (55:6) that the stars 'prostrate' themselves before God, al-Κindi offers something new ίη the burgeoning literature of Qur 'anic commentary (tafsιr): he explains a revealed text purely ίη terms of falsafα .4 6 Even though philosophy establishes the same truths that are contained ίη revelation, the Qur 'an delivers these truths ίη a more powerful way. This point is made ίη another treatise, when al-Κindϊ rather surprisingly digresses from enumerating Aristotle's works to explain a Qur'anic passage (36:79-82) about God's power to create and resurrect the dead. He states that an expert philosopher will provide the same answer as was provided by the Prophet οη this topic.4 7 He then goes οη to use materials drawn from the Christian philosopher John Philoponus, as he expounds the Qur ' anic idea that God creates by saying to a thing, 'be!' 48 Ιη another digression, this time in a work οη meteorology, he explains the superiority of prophets to philosophers by explaining that the former are spared the need to exert effort and long study since God reveals the truth to them directly.4 9 There is ηο suggestion here that prophets know things that philosophers cannot. Al-Κindi wanted to show that Greek sources could provide a foundation for understanding Islamic texts and doctrines. Though the doctrines he supported ίη this way were frequently reminiscent ofMu 'tazilism, his sources and methods were quite different. Still, al- Κίηdϊ was not necessarily trying to replace kalam with falsafa. More likely, he sought simply to show the power and usefulness of the Greek inheritance. Α work like On First Philosophy aimed to show al-Kindi's contemporaries (including the caliph!) what could be achieved by supporting the translation movement. His motivation is made clear ίη On First Philosophy, ίη an impassioned criticism of those who reject wisdom just because it comes from a foreign culture. 50 This need not be taken, as has been suggested, as a broadside aimed at the Mu 'tazila. Rather, the targets were unnamed theologians (they engage ίη 'speculation' and 'traffic ίη religion'), perhaps traditionalists, who opposed ideas imported from the Greeks. 51 First- and second-generation students of al-Kindi's were also quite open to kalam. This further undermines the notion that falsafa, as al-Kindi understood it, involved hostility towards the whole enterprise of rationalist theology. 52 Our

I 67

ARABIC PHILOSOPHY AND THEOLOGY BEFORE Α VICENNA

admittedly incomplete evidence never shows him distinguishing falsafa from kalάm or arguing that the former is superior to the latter. Before long, though, philosophers did seek to differentiate themselves more explicitly from the mutakallimun. One reason for this was, no doubt, the increasing specialism of falsafa as more texts entered into the Arabic milieu. The members of the so-called 'Baghdad school' of peripatetic philosophers in the tenth century were 'professionals' in a much narrower sense than al-Kindϊ, with his immense breadth of interests, had ever been. They were experts in logic and the exegesis of Aristotelian texts, who saw falsafa as a separate and legitimate enterprise of its own. Το some extent they were also οη the defensive. Clearly, al- Kindί's programme of integrating Greek wisdom into the Arabic-speaking world had its detractors, such as the theologians he rails against ίη On First Philosophy. But a man can afford to be confident when he is in a position to address a treatise to the caliph personally. The members of the Baghdad school wrote in a less advantageous context. They were mostly Christians, and their founder, the logician and translator Αbϊι Bishr Matta, had been embarrassed ίη a famous clash with the grammarian al-Sϊrafi. He heaped scorn οη the falasifa for resorting to pretentious foreign learning in place of indigenous Arabic disciplines like grammar. 53 lt was in this rather more fraught atmosphere that al-Farabϊ, a Muslim student of Αbϊι Bishr's and the most famous member of the Baghdad school, decided to put kalam and other non-philosophical disciplines firmly in their place. He discusses kalam ίη several passages, including the conclusion of his Enumeration of the Sciences (Ιη~α ' al- ' ulum). 54 This was not a case ofsaving the best for last. AI-Farabϊ's treatment of kalam ίs designed to show that even when done well, theology is a second-rate discipline. He does recognise a legitimate, but limited, purpose for theology, namely, defending religion from its detractors. Unfortunately, the mutakallimun of aJ-Farabϊ's day do not, according to him, stay within this apologetic remit. They try instead to build up positive theories, and they engage ίη heated disputes with one another as to which of these theories is correct. Some theologians, observes a horrified al-Farabϊ, even resort to threats when their argumentative skills are insufficient. ΑΙ- Farabί's stance is grounded in his understanding of the types of argumentative discourse set forth ίη Aristotle's logicaJ works. The best type is demonstration, as described in the Posterior Analytics.55 Demonstration provides genuine knowledge (' ilm, here translating the Greek episteme) and certainty, by proceeding from absolutely certain first principles and restricting itself to that which is necessary and universa\.56 Other forms of discourse differ from demonstration chiefly ίη the sorts of premises they use. Poetic discourse uses symbolic images; rhetoric chooses its premises on the basis of what is persuasive rather than what is certain; and diaJectic begins from whatever premises command assent in a given context. These premises may be widely accepted or taken from esteemed authorities (what Aristotle calls endoxa), or alternatively, they may simply be agreed οη by the parties to the discussion at hand. With these distinctions ίη hand, al-Farabί explains how religion (milla or dιn) and kalam relate to philosophy. Philosophy is demonstrative. Proceeding from

I 68 indubitable first principles, it generates knowledge through syllogistic proof. Religious discourse, by contrast, is rhetorical or dialectical. Α text like the Qur'an conveys the truth non-demonstratively, aiming at mere conviction.57 Kalam, meanwhile, is purely diaJectical. Hence, al-Farabi's claim that its proper purpose is apologetic: the aim of dialectic is to refute one's opponent, not to demonstrate the truth from first principles. The mistake made by the mutakallimun is that they begin from the rhetorical and poetical basis of the Qur ' an, occasionally adding dialectical premises of their own. Yet they think their discipline is certain and systematic, as if they were philosophers.58 Here one can draw an analogy to the contrast between logic and grammar. Grammarians think that by studying the expressions of one language, such as Arabic, they can establish the rules for correct speech. But it is actually logic, the universal science of truth and meaning, that shows which statements are certainly true, and which are merely 'correct' by convention. 59 This condescending attitude towards kalam, which is taken over and amplified by Averroes ίη his famous Decisive Treatise, 60 suggests that al-Farabi would have spent little energy keeping up with the theological disputes of his day. Yet he shows a reasonable knowledge of contemporary kalam, and especially of Mu 'tazilism. 61 He alludes to sιιch details as their definition of knowledge, their non-literal interpretation of the Qιιr ' an, and their conviction that some things are known immediately by human reason. The latter example shows how theology operates dialectically. Instead of seeking indubitable first principles, the mutakallimun proceed from what is 'acknowledged at first glance. 62 Another example is al-Farabί's discussion of the problem of logical determinism ίη his commentary οη Aristotle's On Interpretation.63 Not only does al-Farabi assert that if determinism were true, God would be ιιηjιιst ίη meting out reward and pιιnishment, bιιt he raises the problem of whether God can do evil. 64 He sees that this involves the question of whether a thing can be possible, yet never occur-something that we would now take for granted but is arguably excluded by Aristotle's modal notions. 65 These glancing references to al- Farabi's contemporaries amongthe mutakallimun show that he was well aware of intellectual debates beyond the confines of falsafa, and that he recognised the philosophical interest of such debates. This is entirely consistent with his criticism of kalam. His complaint is not that theologians never raise issιιes of philosophical significance, bιιt-to the contrary-that they stray into philosophical territory, overstepping their proper apologetic role, and thus sow discord and confusion. Where al-Kindi sought merely to show the relevance of philosophy for Islamic theological problems, al-Farabi adopted a more aggressive posture, claiming that key theological qιιestions can only be answered definitively by the philosopher. Consider what may be his best-known work, The Principles of the Opinions of the Inhabitants of the Perfect City. 66 Al-Farabi would have us believe that, insofar as his discussion is 'philosophicaI: it rests upon indubitable first principles. But he tackles themes also prevalent ίη contemporary kalam-his, it has been argued, is key to understanding the purpose of the Principles.67 lts structure and subject matter are strikingly parallel to works on the 'principles of religion (u?ul al-dϊn )' written by theologians like al-Ash ' arί. Of coιιrse, al- Farabί was not engaging

I 69

ARABIC PHILOSOPHY AND THEOLOGY BEFORE Α VICENNA

ίη kalam. He hoped, rather, to supplant kalam, with demonstrative philosophical knowledge of such matters as the afterlife, the soul and God's relation to the created universe. Despite the influence of al-Farabϊ's stance οη Averroes, his position οη these matters was very unusual for the period we are considering (or any period oflslamic thought, for that matter). Few, if any, other thinkers of the time were so condescending towards revealed religion, and so disdainful of kalam attempts to explicate that religion. For some readers, one name may leap to mind: the great medical clinician and controversial philosopher Abίi Bakr al-Razϊ. Al-Razϊ is known principally for two things, apart from his influential and sophisticated work ίη medicine. First, his theory of the 'five eternals', according to which the world is produced not solely by God, but out of an interaction between God and four other eternal principles (soul, matter, time and place). Second is his scathing critique of revealed religion. We learn about this especially from al-Razϊ's contemporary, the Isma'ϊlϊ philosopher Abίi I:Jatim al-Razϊ (ηο relation: 'Razϊ' means someone from Rayy, a city ίη Northern Persia). Abίi I:Jatim tells us of a debate between himself and al-Razϊ and quotes from a book ίη which al-Razϊ purportedly denied the validity of prophecy, rejected prophetic miracles and sneered at revealed religion.68 This portrayal of al-Razϊ has occasioned both dismay and fascination . Whereas Abίi 1-!iΊtim refers to him simply as 'the heretic; ίη recent times scholars have admired him as a 'free thinker' who boldly departed from his religious and philosophical context. 69 Ι would contend, however, that al-Razϊ's engagement with religion was more subtle and \ess scandalous than it may seem. Abίi J:Iatim's evidence is thrown into doubt by testimony from a less hostile witness, the great theologian Fakhr al-Dϊn al-Razϊ (also from Rayy!). 7°Fakhr al-Dϊn quotes al-Razϊ as citing the Qur 'an and other prophets in support of his own views, and he gives us a detailed account of his dispute with the Mu 'tazilite theologian Abίi 1-Qasim al-Balkhϊ (known as al-Ka ' bϊ) over the correct reading of certain passages ίη the Qur 'an. This dispute concerned theodicy-the question of why God allows evil in the world. Pain and suffering clearly weighed heavily ίη al-Razϊ's philosophical considerations, perhaps because ofhis experiences as a doctor. This seems to have been a primary motivation for his theory of the 'five eternals: which allowed him to explain evil ίη terms of sou\'s foolish choice to become associated with matter, rather than by admitting that a perfectly wise God chose to create an imperfect world. It seems difficult to reconcile this al- Razϊ with the heretic portrayed by Abίi J:Iatim. Ι suspect that Abίi I;Iatim deliberately misdescribed al-Razϊ's position as a thoroughgoing rejection of Islam, indeed of all revealed religion. The target was, ίη fact, narrower: he took issue with groups who relied οη miracles to validate Mu):ιammad's prophecy, who accepted anthropomorphic descriptions of God, and above all whose religious beliefs rested οη uncritica\ acceptance of authority ( taqlιd) rather than rational inquiry (na;ar). 71 'Traditionalists' like Ibn J:Ianbal and his followers would have been among the offenders. Also under attack were the Isma' ϊlϊs, who were certainly not guilty of anthropomorphism, but whose theological system coώd be taken as the ultimate endorsement of taqlιd. 72 Their teaching was built around the need for infallible

I 70 imiΊms, whose religious teaching is authoritative. This helps to explain why ΑbίΊ I:Iatim and other Isma ' ilis are such rich sources of (hostile) testimony about al-Razi. Stung by his criticism, they set about refuting him with great vigour. Ιη this context they summarised his vίews, not without the occasional strategic distortion. AI-Razi's relationship to the Mu 'tazila was more nuanced. The report of ΑbίΊ I:Iatim indicates numerous points of possible agreement between al-Razi and the Mu 'tazila. These include his suspicion of anthropomorphism and taqlϊd, and to some extent his rejection of miracles. Ιη the more reliable context ofhis own authentic works, al-Razι also extols reason ('aql) as God's greatest gift to mankind and the basis for all correct ethical teaching. 73 Rationalism was common ground between him and the Mu 'tazila, something he exploited ίη the debate about theodicy. His solution to the problem of evil emphasises that a wise Creator would not permit pointless evil and suffering. Some theologians of medieval Islam, and some philosophers of religion nowadays, would simply reply that God's ways are unknowable for mankind. But against a Mu'tazilite opponent like al-Ka ' bι, al-Razi's argument had considerable force. For the Mu ' tazila likewise emphasised God's perfect wisdom and justice and assumed that human reason is capable of discerning the requirements of this justice. Ιη fact, al-Razi's thought οη these issues seems to have developed precisely through his dialectical engagement with Mu 'tazilite kaliΊm. 74 ΑΙ- Razι also reacted to the falasifa of his day and their Greek sources. Ιη one area where he at first seems to adopt a kalam view, he turns out to be more dependent οη the Hellenic tradition. He differs from the falasifa in accepting an atomic view of bodies, as did the theologians. But his version of atomism has more to do with the preSocratic type of atomism attacked by Aristotle than with the more mathematical version of atomism we find ίη kalam, where atoms are akin to geometrical points.75 As should be clear from this example and the theory of the five eternals, al-Razι was ηο Aristotelian. Among the Greeks, his main authorities were Plato and the medical writers, especially Galen. Ιη keeping with his disdain for taqlid, he criticised even these favourite Greek authorities-one Razian work has the self-explanatory title Doubts About Galen . But he was entirely unimpressed by the Aristotelians, especially when it came to matters of physics.76 For this reason, al-Razι cannot neatly be classified as a faylasuf, and still less was he a mutakallim. His primary vocation was medicine, not philosophy or theology. The delight he took ίη disputing with both philosophers and the theologians ηο doubt says as much about his personality as about his intellectual formation: al-Razι was above all a man who loved a good argument.

3. JEWISH AND CHRISTIAN

FALSAFA

For a more harmonious marriage of falsafa and kalam, we may turn to a contemporary Jewish thinker, Sa 'adia Gaon (882-942). Sa 'adia hailed from Egypt but moved to Iraq where he headed a rabbinical school near Baghdad. He integrated Graeco-Arabic

I 71

ARABIC PHILOSOPHY AND THEOLOGY BEFORE

Α VICENNA

philosophical ideas and Muslim kaliim into Judaism, adopting a number of doctrines directly from Mu 'tazilism; he is sometimes described as a 'Jewish Mu 'tazilite:77 Α particularly clear example is his embrace of free will as a pre-condition for moral responsibility, so that God cannot punish us justly if we are unfree. 78 He also assumes that reason is able to discover the necessity for God's commanding and prohibiting as he does, 79 going so far as to c\aim that it is rationally necessary for God to supply humankind with guidance from prophets, a position we also find ίη contemporary Mu ' tazilism.80 Sa' adia's most extensive development of a Mu 'tazilite theme is his treatment of divine attributes. This dominates the second treatise of his Book of Doctrines and Beliefs. Sa 'adia has already laid out a qualified empiricist epistemology that recognises various paths to knowledge but insists that all methods are grounded ίη sensation.81 When he comes to discuss divine attributes, he sticks to his empiricist guns and argues that God transcends the limits of what we can grasp because the progress of our knowledge ends with the most 'subt\e' (la(if) of sensible things. 82 Since God is not a body, but rather the maker (~ni' ) of all bodies, and since all human knowledge is based ultimately οη our experience ofbodies, God is beyond human knowledge. Those who speak as if God can be known, or described ίη human language, are ίη effect implying that he is a body: 'those who seek to give Him motion, rest, wrath, delight, or anything of the sort, have ίη truth sought to make Him a body, ίη terms of meaning if not ίη terms of expression: 83 We are thus forced to adopt a negative theology, ίη which God's attributes are apparently denied, so as to avoid tajsιm : speaking as if God were corporeal (from jism, 'body'). Το the extent that Scripture seems to engage ίη tajsιm, this is to be understood as providing believers with 'an approximation and image ( taqrιb wa tamthιl)' of God. Scripture must use such metaphorical language because we could say essentially nothing about God if we restricted ourselves to literal truth: 'if we kept ourselves to His attributes according to strictly true expression, then we would have to deny that He is hearing, seeing, merciful, and willing, so that we would be restricted to His being a\one (anniyya faqat) :84 Yet there is a tension ίη Sa ' adia between a general ban οη divine attributes and a commitment to certain privileged attributes. 85 Scholars disagree even about how many attributes Sa 'adia recognises (if any). Sa 'adia says that revelation teaches us five things about God: He is one, living, powerfu1, knowing and unlike any of his creatures. 86 The last of these is not really an attribute, and the attribute Όηe' could also be understood ίη a negative way, as indicating that God lacks the multiplicity characteristic of bodies.87 So perhaps we have only three positive attributes. Unfortunately, this still seems to be three attributes too many, given the wholly negative theology set out by Sa ' adia in close proximity to these passages. Sa ' adia tries to solve the difficulty by emphasising that the three attributes simply spell out the implications of God's being a maker. It is only the exigencies of language that prevent us from using a single word to express the notion that a maker must be living, powerful and knowing; in the mind, they are grasped 'all at once'. Furthermore, these attributes imply ηο multiplicity ίη the divine essence (dhiit) but merely explicate the fact that things are made by Him.

I 72 It has thus been suggested that Sa 'adia is aligning himself with Mu ' tazilites who accepted 'attributes of action' but rejected 'attributes of essence'. 88 Attributes of essence express what God is. Attributes of action express what God does and are far less troublesome because they can be taken as referring indirectly to created things. Ιη favour of this interpretation is the fact that Sa ' adia thinks we know of God's existence only because the physical world requires an external cause: 'there is ηο means of proof that is not οη the basis of creation'. 89 Οη the other hand, Sa ' adia elsewhere speaks of attributes of essence (mα ' άnϊ dhάtiyya), which include the three just mentioned.90 For clarification we may turn to Sa 'adia's lengthy discussion of biblical names and statements about God. The material considered so far \ooks indebted to Mu ' tazilite kalάm rather than Aristotelian philosophy. But here Sa 'adia shows the falsafa side of his thought, classifying Scriptural language in accordance with Aristotle's ten categories. It has been suggested that the categories simply give Sa' adia a convenient way of organising his materia\. 91 More \ikely, he chooses this strategy precisely because the categories analyse language as it applies to physical objects.92 Of particular interest for the question regarding attributes of essence and attributes of action is Sa' adia's treatment of the category of action. He stresses that divine agency is very different from physica\ agency, for instance, ίη needing ηο time, place, instrument or motion. Still, we can say that God acts ίη sσme sense, however disanalogous divine action might be to the actions we normally experience. For instance, when Scripture speaks of God being vengeful, this refers obliquely to his actions ίη the created world. Sa ' adia adds, 'all these names that function ίη this way go back to the things that are made. This is the difference between the names of essence and the names of actions'. 93 Sa 'adia's theory of divine attribution is, then, more complex and more positive than it at first appears. Most terms are to be denied ofGod because they would immediately imply His corporeality and fall into the trap of tajsϊm . Α second type of term seems to apply to God but, ίη fact, applies to his works ίη the created world. These are names or attributes of action. Finally, a third type of term expresses the divine essence itself. There are only a few such 'essential attributes': 'knowing', 'powerful' and '\iving' -perhaps also Όηe' and 'inimitable; to the extent that these have a positive content. But these terms introduce ηο multiplicity, and hence ηο corporealisation. For we grasp them all simultaneously, upon grasping that God is a Creator. This would seem to be the limit of what we can know about God: that he must have these features, if he created the world we see. Sa ' adia was not the only Jewish philosopher to write ίη Arabic ίη the ninth and tenth centuries. For instance, there was Isaac Israeli (d. 955), whose extant works show pervasive influence from al-Κindi. 94 But in this early period, Christian thinkers were even more central to the falsafa tradition. They played a crucial role as translators of philosophical and scientific works from Greek into Arabic (sometimes by way of Syriac), and a\so as philosophers ίη their own right. We have already met al-Farab1's teacher, the founder of the 'Baghdad schooI: ΑbίΊ Bishr Matta. Little of his thought is preserved for us, unfortunately. We are much better informed about a thinker who reportedly studied with both ΑbίΊ Bishr and al-Farab!: Ya\:ιya ibn ' Αdί

I 73

ARABIC PHILOSOPHY AND THEOLOGY BEFORE

Α VICENNA

(d. 974) .9 5 lbn Άdϊ was nicknamed 'the logician (al-manfϊqϊ)' for his expertise in Aristotelian logic. He wrote extensively οη this and other aspects of falsafa . As a Christian, he also used Aristotelian ideas in apologetic works defending the doctrine of the Trinity. 96 Unlike Sa 'iΊdia, lbn Άdϊ was not deeply influenced by kalάm . His interest in Islamic theology was that of a mildly interested outsider, which is perhaps unsurprising. Mutakallimun rarely found common cause with Christians and excoriated the doctrines of the Trinity and Incarnation. Yet we do find lbn ' Αdϊ addressing distinctively kalam themes.9 7 Ιη one short epistle, he criticises the recently developed kalam theory of 'acquisition:98 the idea that God creates human actions that are appropriated or acquired by human agents. This text is evidently the exception that proves the rule, given that lbn ' Αdϊ begins by remarking that he takes up the subject only at the behest of the epistle's addressee. Accordingly, he gives the theory rather short shrift in the subsequent discussion. Α more nuanced engagement with 'theological' issues-again in the form of the problem of divine attributes-can be found ίη his little-studied treatise On Unity. 99 The treatise explores the sense ίη which Όneness (tawhϊd) ' applies to God; he does not say explicitly that On Unity is meant to support a Christian understanding of divine unity, but this is clearly his intention. He proceeds by proposing and then refuting, ίη considerable detail, a whole series of possible ways to understand God's oneness. For instance, he considers the claim that God is one ίη the sense that nothing is like him. Ibn ' Αdϊ says that this must mean either that God is not like anything else in any respect, or that God is not like anything else in every respect. But the first is impossible: if Α and Β are distinct, they will always have at least something in common. For one thing, they will, ironically, both have ίη common that they are dissimilar to the other. Furthermore, God must have something ίη common with, say, man, in that neither God nor man is a horse. ιοο As for the second suggestion, lbn Άdϊ points out that equally, nothing is like anything else in every respect. If Α is distinct from Β, Α must have some feature or other that Β lacks (so he endorses the principle now called the 'identity of indiscernibles').10 1 Another position considered by Ibn Άdϊ is that God is Όηe' in virtue of a unity (wahda) that belongs to God. This is ruled out through a complex treatment of the relationship between the supposed 'property (ma ' na)' of unity and God's essence or self (dhat). For instance, Ibn Άdϊ considers whether the unity would be identical to the essence, or created by God and hence posterior to the essence. 102 This highly scholastic discussion of divine unity has obvious resonances with kalam discussions of the problem of attributes, and ίη style and content it is also reminiscent of al-Kindϊ's On First Philosophy. ' 03 But when Ibn ' Αdϊ sets out his own view, he argues for a conclusion that al-Kindϊ would reject: God must be in a sense one, and in a sense many, as neither pure unity nor pure multiplicity are possible. 104 Even so, most of the ways in which a thing can be one or many are inapplicable to God. 105 By process of elimination, Ibn ' Αdϊ concludes that God will be Όηe' only ίη respect of definition. However, this does not rule out God's being multiple. Ιη fact, we know that he is multiple because our experience of created things proves that

I 74 God possesses the attributes (?ifiΊt) of generosity, wisdom and power. 106 This is strikingly similar to what we have seen in Sa 'adia, albeit that lbn Άdi has the attribute of 'generosity (jϊιd)' instead of Sa ' adia's Ίiving: However, Sa ' adia was at pains to deny that the three attributes constitute a genuine mu1tiplicity and, ίη fact, criticises Christians precisely for believing that the three attributes constitute a genuine multiplicity. Sa ' adia explains that the doctrine of the Trinity arose precisely from a misunderstanding of this point οη the part of the more sophisticated Christians, who understood that God is not a body but still believed that his multiple attributes mean that he is three as well as one.107 Sa ' adia could almost have written this with Ibn 'Adi in mind. It was clearly the latter's intention to leave room for a genuine, and triune, multiplicity in God. The materials surveyed here wou1d still allow us to divide the thinkers who wrote in Arabic prior to Avicenna into two camps, thefaliΊsifa and the mutakallimun. It is legitimate, as well as useful, to say that al-Kindi and al-Farabi were 'philosophers' whereas Αbϊι 1-Hudhayl and al-Ash 'ari were 'theologians: But this shou1d not blind us to the complex and productive exchanges between the two groups, or to the variety of attitudes that philosophers adopted towards kaliΊm. The picture is complicated still further by the activity of Jewish and Christian thinkers who pursued theological concerns of their own, sometimes drawing οη Islamic kaliΊm. Matters are equally complex once we get to Avicenna. Recent attempts to emphasise the kaliΊm influence οη Avicenna himself remain controversia1 but seem plausible in light of the considerable importance of kaliΊm for philosophers ίη previous generations.108 In the Eastern Islamic heartlands, the criticism of Avicenna by al-Ghazalι and others did not lead to an abandonment of falsafa. 109 We instead find, alongside other developments, a fusion of Avicennian (rather than Aristotelian) philosophy with sunnί kaliΊm. This later tradition stands as a living rebuke to al-Farabί. For many centuries, authors would try their hand at developing systems that were clearly theological, and admittedly dialectical-yet also philosophical. 110

NOTES 1. Οη the mi~na see Patton 1965. Οη the broader issues, see Crone and Hinds (2003). 1 speak here of sunnI, as opposed to shi'J Islam- the shI 'a being those who believed that political authority is passed down through the family of the Prophet and ίη particular through his cousin ' AII. Theological groups could cross this divide; many later Mu 'tazilite authors were shi'J. 2. As shown by Gutas (1998) . On the translations see also Endress (1987- 92) and Gutas's appendix in Pasnau (2010). For translations of Aristotle, still useful is Peters (1968). 3. For a lively account of the period, see Kennedy (2005). See also Young et al. (1990). 4. Thus, Marenbon (2007, 60) rightly designates kaliΊm as a type of'Islamic philosophy'. 5. The most important study of theology in this period is the monumental van Ess (1991- 95); see, more briefly, van Ess (2006). Also important are numerous publications by

I 75

ARABJC PHJLOSOPHY AND THEOLOGY BEFORE AVICENNA

R. Μ . Frank, including Frank (1978), as well as Nader (1984) ; Watt (1998). For a briefer survey, see Blankinship (2008). 6. Οη Isrniϊϊli philosophers, see De Srnet (1995), Walker (1993, 1999); for Isrna'ilisrn rnore generally, Daftary (1996). 7. Al-Sharastani (1984, 42). See also Frank (1971, 7). 8. For al-Na~?iirn, God rnust do what is beneficial for his creatures, and he cannot, for instance, increase or decrease the suffering of the darnned. See Nader (1984, 77-8). Against the possibility that God could perforrn evil, he argues that even the possibility (tajwϊz) of evil is evil, so this possibility is incornpatible with God's justice (report by al-Shahrastani 1984, 49). Abίi 1-Hudhayl adrnits that it is 'inconceivable' that God do evil, because of his good nature but insists that he does have the power to do so. See van Ess (1991-5), vol.5, 401-2 (texts ΧΧΙ.79-80) . Cf. Daiber (1975 , 233) οη the question ofwhether God rnust create, if creation is good. 9. Report by al-Shahrastani (1984, 52). 10. See further Frank (1982) . 11. Abίi 1-Hudhayl, for instance, firrnly denies that God has power over the actions perforrned by hurnans. See van Ess (1991-5), vol.5, 402-4 (texts ΧΧΙ.82-86). 12. See Daiber (1975), 132-6. 13. For instance, al-Shahrastani's overview of Mu 'tazilite views οη divine predication by rernarking that they 'entirely (a~a/an) reject the eternal attributes' (al-Shahrastani 1984, 41). 14. van Ess (1991-95), vol. 5,392 (text ΧΧΙ.56) . 15. See furth er Frank (1969). See Frank (1978, 53); Frank (1979, 74-5). 16. More problernatic is the apparent result that God's justice will be identical to his power and his knowledge since all of these are identical to God . Ιη another seerningly Ύerbal' solution, he clairns that the attributes are 'neither identical nor not-identical' to one another. See van Ess (1967, 112). 17. For all this see Madelung (1974) . 18. See van Ess (1967, 103), and 1nore generally Melchert (2006) . 19. Α prorninent exarnple would be I;>iriir ibn Άrnr (d. 815) who is norrnally classified as a Mu 'tazilite, but who accepted that God creates hurnan actions- view shared with the 'deterrninist' Jahrn ibn ~afwan (d. 745)- while insisting that hurnans are a/so the agents of their actions. 20. For a surnrnary and critical discussion of this account, see Watt (1998, 209ff) . 21. Nader (1984, 45 - 6) . 22. For studies see Heernskerk (2000); Hourani (1971); Peters (1976); Reynolds (2004) . 23. Particularly al-Κhayyat, author of al-Intψir, and the 1nulti-faceted literary genius al- Jaφ~ (d. 869 ), a student of al- Na~rn. Other irnportant early witnesses include al-Nashi ', for whorn see van Ess (1971) . 24. Despite his rejection of Mu 'tazilite doctrines, he surnrnarises the1n ίη his invaluable Μαqά/άt, a rnajor source for early theological teachings. Other key sources include al-Shahrastanί (1984), al-Baghdadi (1919 - 35). 25. See further Allard (1964); Girnaret (1990); McCarthy (1953) . 26. The fifth 'principle' to cornrnand good and evil is usually taken as a call to political activisrn, but it was honoured 1nore ίη the breach than ίη the observance. See 'Mu ' tazila; Encyclopedia of Islam, new ed., 12 vols (Leiden, the Netherlands: Brill, 1960- 2004) . 27. The milJna itself is often seen ίη this context. For Mu ' tazilite views οη the question of the imάm (political and religious leader), see el-Ornari (2007). 28. Not dissirnilar is the approach of al-Maturίdi, founder of the other rnain postMu'tazlite school ofSunni kalάm, which spread ίη Central Asia. See further Rudolph (1997) .

I

29. For instance, our freedom of action is known to us innately and also confirmed by revelation, as remarked by al-Shahrastanϊ (1984, 44). 30. Schδck (2005). 31. See his autobiography, the Deliverer from Error ( trans. Watt 1994), and further Griffel (2009); Shihadeh (2005). 32. Οη which see Frank (1978); Dhanani (1994). 33. See Μ . Marmura, Άl-Ghazalϊ; ίη Adamson and Taylor (2005, 137-54). His occasionalist interpretation, which I follow here, challenges the reading ofFrank (1992). 34. For a comprehensive translation ofhis works, see Adamson and Pormann (2011). 35. He uses the word mutakallimun at least once, but not for theologians. The term appears, of all places, in his treatise explaining why air is colder at higher altitudes (al-Kindϊ 1950-53, vol.2, 100). The word mutakallimun here refers simply to people who have expressed views οη such scientific phenomena. 36. Adamson (2003) . The topic was previously discussed by lvry (1974, 22-34). 37. For the Arabic version of Proclus, called in Arabic Book on the Pure Good, and ίη the Latin translation ofthis Arabic version, the Book ofCauses, see DΆncona (1995); Endress (1973). Οη the Kindϊ circle more generally, see Endress (1997). 38. See Adamson (2002, 165-70), and further DΆncon a (2000), Frank (1956), Taylor (1998). 39. Ed. al-Kindϊ (1950-3, 97-162); Rashed and Jolivet (1998, 9-99); English translations Ivry (1974)(, Ada1ηson and Pormann (2011). 40. See Adamson (2007a, 47- 57). 41. Or so I have argued ίη Adamson (2007a, 98- 105). 42. For the topic more generally, see Davidson (1987) . 43. Al-Kindϊ wrote a work, now lost, called On Oneness by way of Commentaries (Fi l-tawήid bi-tafsiriit); Ivry (1974, 30) suggests that this might have grounded the same theory ίη religious texts. Information about al-Kindϊ's many lost treatises is found in Ibn al- Nadϊm's Fihrist (translation ίη Dodge 1970). For lists of al-Kindϊ's writings, see McCarthy (1962); Adamson and Pormann (2011). 44. For the work see Rashed and Jolivet (1998, 122- 7); Adamson and Pormann (2011). 45. Ed. al-Kindϊ (1950- 53, vol. 1, 244- 61); Rashed and Jolivet (1998, 177- 99); English translation Adamson and Pormann (2011). 46. Janssens (2007). 47. Al-Kindϊ (1950- 53, vol. ι, 373). 48. Adamson (2003, 57- 66), Adamson (2007a, 43- 4 and 63- 6). Al-Kindϊ may not have known that Philoponus was a Christian but did know he was a fierce critic of Aristotle. 49. Al-Kindϊ (1950- 53, vol. 2, 93). 50. Al-Kindϊ (1950- 53, vol. ι, 103- 4). 51. Adamson (2007a, 23- 4), against Ivry (1974, 33). 52. See Adamson (2007b). 53. See Endress (1986), Margoliouth (1905). 54. al-Farabi (1931, 131- 8). 55. This work was simply called The Demonstration (al-Burhiin), and al-Farabϊ wrote a paraph rase of the work with the same title. See Fakhry (1987). English translation by Fakhry in Nasr and Aminrazavi (1999, 93 - ιιο). 56. See Black (2006); Adamson (2007c). 57. He makes this point in several works, but especially the Book of Religion (Kitiib al-Milla), ίn al-Farabϊ (1968, 41 - 66). French translation by Mallet (1989, 117- 45). English translation by Butterworth (2001, 93- 113).

I 77

ARABIC PHILOSOPHY AND THEOLOGY BEFORE

Α VICENNA

58. See Mahdi (1972); Reisman (2005, 65- 9). 59. For the parallel see Ν . Lahoud and Ε . Gannage in Mέlanges de /'Universitέ SaintJoseph 57 (2004), following a suggestion by F. W. Zi1n1nermann. 60. Averroes (1997). English translations include Averroes (1961). 61. See Rudolph (2007). 62. Letter on the Intel/ect, ed. al-Farabί (1938, 7- 8). English translation in McGinnis and Reisman (2007, 68- 78). See Rudolph (2007, 70), which is my source for the other allusions to Mu 'tazilite doctrines just mentioned. 63. Ed. al-Farabί (1960 ), English translation in Zimmermann (1981). See Adamson (2006). 64. al-Farabί (1960, 98- 100). 65. See further Wisnovsky (2003, 219- 25); and Knuuttila's paper in this volume. 66. Ed. and trans. al-Farabί (1985). 67. Rudolph (2008). 68. Ιη his work The Tokens of Prophecy (ed. Αbϋ I:Iatim al-Razί 1977). 69. See Stroumsa (1999); Urvoy (1996, 142- 52). Οη him see further Druart (1996, 1997). 70. As discussed ίη two important articles, Rashed (2000, 2008). 71. See Αbϋ J:Iatim al-Razί (1977), 191 (οη miracles), 114 (οη anthropomorphism, tashbϊh), 13 (οη taqlϊd) . 72. The same criticism is levelled at the Ismaϊlίs by al-Ghazalί. See Watt (1994, 45-56), and further Frank (1991-2). 73. See the opening ofhis Spiritual Medicine, ed. Αbϋ Bakr al-Razί (1939), 15-96, trans. Arberry (1950). 74. Here Ι agree wilh the conclusions ofRashed (2000). 75. See Baffioni (1982, ch. 2). 76. Even more so if we accept the authenticity of On Metaphysics: see Αbϋ Bakr al-Razί (1939, η6-34); studied in Lucchetta (1987). 77- For example, by Alexander (1984, 22). 78. Arabic and Hebrew (ed. Saadia Gaon 1880, 150-51). English translation by Saadia Gaon (1948). See further Efros (1974, ch.6); Adamson (2010a). 79. Saadia Gaon (1880, η4) . 80. Saadia Gaon (1880, η8); cf. El-Omari (2007, 51). 81. Saadia Gaon (1880, 12-14); cf. Efros (1942). 82. Saadia Gaon (1880, 73ff). 83. Saadia Gaon (1880, 78, cf. 92-3). 84. Saadia Gaon (1880, 97). Note his use of the phrase anniyya faqa\, which has been mentioned above as appearing ίη texts from al-Kindi's circle. 85. Efros (1974, 50-60); Kaufmann (1967); Neumark (1928); Rawidowicz (1943). 86. Saadia Gaon (1880, 79). 87. This is how it is taken by Kaufmann (1967, 16-17). 88. Efros (1974, 52-3). 89. Saadia Gaon (1880, 80-81). Cf. Efros (1974, 49). 90. Saadia Gaon (1880, ηο) . 91. Kaufmann (1967, 55). 92. So also Neumark (1928, 184). 93. Saadia Gaon (1880, 106). Cf. Kaufmann (1967, 68-9 η . 127) . 94. Altmann and Stern (1958). 95. See Endress (1977); Platti (1983). Though there is evidence for his having studied with al-Farabί, recent studies have suggested that intellectually he may be closer to al-Kindί than al-Farabί οη some points. See Rashed (2009); Adamson (2010b).

I

96. Perier (1920). 97. This was already pointed out by Endress (1986, 204- 5). 98. Pines and Schwarz (1979). Cf. Adamson (2010a, 405 - 6). 99. Ed. Ibn ΆdΙ (1988, 375- 404). 100. Ibn Άdι (1988, 377- 8). 101. Ibn Άdι (1988, 379) . 102. Ibn Άdι (1988, 381-3). 103. For instance, the last argument is comparable to al-Kindι's argument that nothing can cause itself (dhatuhu) to exist (al-KindI 1950-3, 123-4). 104. Ibn ΆdΙ (1988, 390-92). But compare al-KindI (1950-53, 132-41) for a demonstration that created things must be both one and many. 105. He rules out unity with respect to genus, species, relation, continuity and undividedness (Ibn ΆdΙ 1988, 394). 106. Ibn ΆdΙ (1988, 404). 107. Saadia Gaon (1880, 86). See further Wolfson (1977). 108. See especially R. Wisnovsky (2003); Rudolph (1996). 109. Important recent studies include Eichner (2007); Gutas (2002); Michot (1993); Shihadeh (2005). 110. Ι wou]d like to thank Nadja Germann for her helpful comments οη a previous draft, and the Leverhulme Trust for their support as the chapter was completed.

BIBLIOGRAPHY Adamson, Ρ. 2002. 1he Arabic Plotinus: Α Philosophical Study of the '1heology of Aristotle'. London: Duckworth. - -. 2003. AI-KindI and the Mu'tazila: Divine Attributes, Creation and Freedom . Arabic Sciences and Philosophy 13:45-77_ __2006. The Arabic Sea Battle: Al-FarabI οη the Problem ofFuture Contingents. Archiv fur Geschichte der Philosophie 88:163-8. - -. 2007a. Al-Kindϊ. New York: Oxford University Press. - -. 2007b. 'The Kindian tradition: The structure of philosophy ίη Arabic Neoplatonism: In Libraries of the Neoplatonists, ed. C. D'Ancona, 351-70. Leiden, the Netherlands: Brill. - -. 2007c. Knowledge ofUniversals and Particulars ίη the Baghdad School. Documenti e Studi sulla Tradizione Filosofi.ca Medievale 18:141-64. - -. 2010a. 'Freedom and determinism: !η Pasnau 2010, vol. 1, 399-413. - - . 2010b. Ya]:ιya Jbn ' AdI and Averroes οη Metaphysics Alpha Elatton. Documenti e Studi sulla Tradizione Filosofi.ca Medίevale 21:343- 74. Adamson, Ρ. and Ρ. Ε . Pormann. 2011. 1he Philosophical Works of al-Kindϊ. Karachi: Oxford University Press. Adamson, Ρ. and R. C. Taylor, eds. 2005. 1he Cambridge Companion to Arabic Philosophy. Cambridge: Cambridge University Press. Alexander, Ρ. S. 1984. Textual Sources for the Study of Judaism . Chicago: University of Chicago Press. Allard, Μ . 1964. Le probleme des attributs divins dans Ια doctrine d' al-Ash ' arϊ et de ses premίers grands disciples. Beirut: Imprimerie Catholique.

I 79

ARABIC PHILOSOPHY AND THEOLOGY BEFORE AVICENNA

Altmann, Α . and S. Μ . Stern. 1958. Isaac Israeli: α Neoplatonic Philosopher of the Early Tenth Century. Oxford: Oxford University Press. Arberry, Α. J. 1950. The Spiritual Physick of Rhazes. London: John Murray. Averroes. 1961. On the Harmony of Religion and Philosophy, trans. G. F. Hourani. London: E.J.W. Gibb. - - . 1997. Ρα~ al-Maqal, ed. Μ. Α. Jabarί. Beirut: Markaz Dirasat al-WaJ:ιda al- ' arabiyya. Baffioni, C. 1982. Atomismo e antiatomismo nel pensiero islamico. Naples: Instituto Universitario Orientale. al-Baghdadι. 1919-35. Moslem Schisms and Sects. 2 vols. New York: Columbia University Press. Black, D. L. 2006. Κnowledge ( Ίlm) and Certainty ( Yaqϊn) ίη al-Farabi's Epistemology.

Arabic Sciences and Philosophy 16 : η-45. Blankinship. 2008. 'The Early Creed'. !η The Cambridge Companion to Classical Islamic Theology, ed. Τ. Winter, 33-54. Cambridge: Cambridge University Press Butterworth, C. 2001. Alfarabi. The Political Writings. Selected Aphorisms and Other Texts. Ithaca, ΝΥ: Cornell University Press. Crone, Ρ. and Μ. Hinds. 2003 . God's Caliph: Religious Authority in the First Centuries of Islam. Cambridge: Cambridge University Press. Daftary, F. 1996. Mediaeval Isma ·ili History and Thought. Cambridge: Cambridge University Press. Daiber, Η . 1975. Das theologisch-philosophische System des Mu 'ammar Ibn Ά.bbad as-Sulamι. Beirut: Franz Steiner. DΆncona, C. 1995. Recherches sur le Liber de Causis. Paris: J. Vrin. - - 2000. Tinfluence du vocabulaire arabe: causa prima est esse tantum'. !η ulaboration du vocabulaire philosophique au Moyen Age, ed. J. Hamesse and C. Steel, 51-97. Turnhout, Belgium: Brepols. Davidson, Η. Α. 1987. Proofs for Eternity, Creation and the Existence of God in Medieval Islamic and Jewish Philosophy. New York: Oxford University Press. De Smet, D. 1995. La quietude de l'intellect: neoplatonisme et gnose ismaelienne dans l'reuvre de Ifamid ad-Din al-Kirmdni (Xe/Xle s.). Leuven, Belgium: Peeters. Dhanani, Α . 1994. The Physical Theory of the Kalam. Leiden, the Netherlands: Brill. Dodge, Β. 1970. The Fihrist of al-Nadϊm . New York: Columbia University Press. Druart, Τ.-Α . 1996. Al-Razi's Conception of the Soul: Psychological Background to his Ethics. Medieval Philosophy and Theology 5:245-63 . - -. 1997. The Ethics of al-Razi. Medieval Philosophy and Theology 6:47-71. Efros, 1. 1942. Saadia's Theory ofKnowledge. The Jewish Quarterly Review 33:133-70. - -. 1974. Studies in Medieval Jewish Philosophy. New York: Columbia University Press. Eichner, Η. 2007. Dissolving the Unity of Metaphysics: from FalJr al-Din al-Razi to Mulla ~adra al-Sirazi. Medioevo 32:139- 98 . el-Omari, R. 2007. Άbu 1-Qasim al-Balkhi al-Ka 'bi's doctrine of the Imama'. !η Α Common Rationality: Mu ' tazilism in Islam and Judaism, ed. C. Adang, S. Schmidt, and D. Sklar, 39- 57. Wίirzburg, Gerιnany: Ergon. Endress, G. 1973. Proclus Arabus: Zwanzig Abschnitte aus der Institutio Theologica in arabischer ϋbersetzung. Beirut: Steiner. - -. 1977. The Works ofYaήya Ibn Άdϊ. An Analytical lnventory. Weisbaden, Germany: Ludwig Reichert. - -. 1986. 'Grammatik und Logik. Arabische Philologie und griechische Philosophie im Widerstrei t'. !η Sprachphilosophie in Antike und Mittelalter, ed. Β . Mojsisch, 163-299. Amsterdam: B.R. Grίiner. Κ.

I 80 - - . 1987 and 1992. 'Die wissenschaftliche Literatur'. Ιη Grundriss der arabischen Phi/ologie, ed. Η. Gatje, vol. 2, 400- 506, and vol. 3 (supplement), 3- 152. Wiesbaden, Germany: Ludwig Reichert. - - . 1997. 'The circle of al-Κindi: !η Ίhe Ancient Tradition in Christian and Islamic He/lenism, ed. G. Endress and R. Kruk, 43- 76. Leiden, the Netherlands: Research School CNWS. Fakhry, 1987. Mantiq 'inda Ί-Fiiriibi: Kitiib al-Burhiin wa-Kitiib Sharii ' it al-Yaqin. Beirut: Dar el-Machreq. al-Farabί. 1931. ΙΙ;ι~α · α/- ' u/um, ed. U. Αmϊη . Cairo: Matba 'a al-Sa 'ada. - - . 1938. Risiila fi /- 'aql, ed. Μ . Bouyges. Beirut: Imprimerie Catholique. - -. 1960. Commentary on Aristotle's Peri Hermeneias, ed. W. Kutsch and S. Marrow. Beirut: Catholic University Press. - -. 1968. Kitiib al-Milla wa-nu~u~ ukhrii, ed. Μ. Mahdi. Beirut: Dar al-Mashriq. - -. 1985. On the Perfect State, ed. and trans. R. Walzer. Ox.ford: Oxford University Press. Frank, R. Μ. 1956. The Origin of the Arabic Philosophical Term Anniyya. Cahiers de Byrsa 6:181- 201. - -. 1969: The Divine Attributes According to the Teaching of ΑbίΊ Ί-Hudhayl al-Allaf. Le Museon 82:451- 506. - -. 1971. Several Fundamental Assumptions of the Ba~ra School of the Mu 'tazila. Studia Islamica 33:5- 18. - - . 1978. Beings and Ίheir Attributes: Ίhe Teachings of the Basrian School of the Mu ' tazila in the Classical Period. Albany: SUNY Press. - - . 1979. 'Kalam and Philosophy, Α Perspective from One Probleni'. Ιη Islamic Philosophical Ίheology, ed. Ρ. Morewedge, 71- 95. Albany: SUNY Press. - - . 1982. The Autonomyofthe Human Agent ίη the Teaching ofΆbd al-Jabbar. Le Museon 95:323- 55. - - . 1991- 2. AI-Ghazalϊ οη Taqlϊd. Zeitschrift fur die Geschichte der arabisch-islamischen Wissenschaften 7:207- 52. - - . 1992. Creation and the Cosmic System. Heidelberg, Germany: Carl Winter. Giιnaret, D. 1990. La doctrine de al-Ash ' αrϊ. Paris: Cerf. Griffel, F. 2009. Al-Ghaziili's Philosophical Ίheology. New York: Ox.ford University Press. Gutas, D. 1998. Greek Ίhought, Arabic Culture: the Graeco-Arabic Translation Movement in Baghdad and Early Society (2nd- 4th /Sth-1oth Centuries). London: Routledge. - - . 2002. 'The heritage of Avicenna: The golden age of Arabic philosophy, 1000- ca. 1350'. !η Avicenna and His Heritage, ed. J. Janssens and D. De Smet, 81- 97. Leuven, Belgium: Leuven University Press. Heemskerk, Μ . Τ. 2000. Suffering in the Mu ' tazilite Ίheology: Άbd al-Jabbiir's Teaching on Pain and Divine Justice. Leiden, the Netherlands: Brill. Hourani, G. F. 1971. Islamic Rationalism: Ίhe Ethics ofΆbd al-Jabbiir. Ox.ford: Clarendon Press. Ibn

Yaryya. 1988. Philosophical Treatises, ed. S. Khalifat. Amman, Jordan: University Press of Amman. Ivry, Α . L. 1974. Al-Kindi's Metaphysics. Albany: SUNY Press. Janssens, ). 2007. Al-Kindϊ: The Founder of Philosophical Exegesis ofthe Qur 'an. Journal of Qur' anic Studies 9:1- 21. Kaufmann, D. 1967. Geschichte der Attributenlehre in der judischen Religionsphilosophie des Mittelalters von Saadja bis Maimuni. Amsterdam: Philo. Kennedy, Η . 2005. Ίhe Court of the Caliphs: When Baghdad Ruled the World. London: Phoenix. ' Αdϊ,

I 81

ARABIC PHILOSOPHY AND THEOLOGY BEFORE Α VICENNA

al-Kindϊ.

1950-53. Rasii'il al-Kindϊ al-falsa.fiyya, ed.

Μ. Αbίι Rϊda,

2 vols. Cairo: Dar al-Fikr

al-Άrabϊ.

Lucchetta, G. Α . 1987. La natura e la sfera: la scienza antica e le sue metafore nella critica di Riizϊ. Bari, Italy: Milella. Madelung, W. 1974. 'The origins of the controversy concerning the creation of the Koran'. Ιη Orientalia Hispanica vol. 1, ed. J. Μ . Barra\, 504- 25. Leiden, the Netherlands: Brill. Mahdi, Μ. 1972. Alfarabi οη Philosophy and Religion. Philosophical Forum 4:5-25. Mallet, D. 1989. Farabi. Deux traitέs philosophiques. Damascus, Syria: Institute Francais de Damas. Marenbon, J. 2007. Medieval Phi/osophy: an Historical and Philosophical Introduction. London: Routledge. Margoliouth, D. S. 1905. The Discussion between Abu Bishr Matta and Abu Sa'id al-Sirafi οη the Merits of Logic and Grammar. Journal of the Royal Asiatic Society: 79- 129. McCarthy, J. 1953. Ίhe Ίheology of al-Ash ' αrϊ. Beirut: Impri1nerie Catholique. McCarthy, R. J. 1962. Al-Ta~iinif a/-Mansuba ilii Faylasuf al- ' arab. Baghdad: Matba'a al-'ani. McGinnis, J. and D. C. Reisman, ed. and trans. 2007. Classical Arabic Philosophy: An Anthology of Sources. Indianapolis, ΙΝ : Hackett. Melchert, C. 2006. Aήmad ibn Ifanbal. Oxford: Oneworld. Michot, Υ. J. 1993. La pandemie avicennienne au VI' /ΧΙ!' siecle. Arabica 40:287-344. Nader, Α. 1984. Le systeme philosophique des Mu ' tazila. Beirut: Dar El-Machreq Sarl. Nasr, S. Η. and Μ. Aminrazavi 1999. An Anthology of Philosophy in Persia, vol. 1. Oxford: Oxford University Press. Neumark, D. 1928. Geschichte der jiidischen Philosophie des Mittelalters ΙΙ .2. Berlin: Walter de Gruyter. Pasnau, R., ed. 2010. Ίhe Cambridge History of Medieval Philosophy. 2 vols. Cambridge: Cambridge University Press. Patton, W. Μ . 1965. Aήmed ibn Ifanbal and the Mihna. Leiden, the Netherlands: Brill. Perier, Α . 1920. Petits traitέs apologέtiques de Yaήyii ben Άdι. Paris: J. Gabalda. Peters, F. Ε. 1968. Aristoteles Arabus. Leiden, the Netherlands: Brill. Peters, J. R. Τ. Μ . 1976. God's Created Speech. Leiden, the Netherlands: Brill. Pines, S. and Μ. Schwarz. 1979. Ύa]:ιya Ibn 'Adϊ's refutation of the doctrine of acquisition (Iktisab)'. Ιη Studia Orientalia Memoriae D.H. Baneth Dedicata, ed. J. Blau, S. Pines, Μ. J. Kister and S. Shaked, 49- 94. Jerusalem: Magnes. Platti, Ε. 1983. Yaήyii Ibn Άdϊ: Ίheologien chretien et philosophe arabe. Leuven, Belgium: Departement Orientalistiek. Rashed, Μ. 2000. Αbίι Bakr a\-Razϊ et la kalam. Mέlanges de /'institut dominicain dέtudes orientales du Caire 24:39-54. - - . 2008. Αbίι Bakr al-Razϊ et la prophetie. Melanges de l'institut dominicain dέtudes orientales du Caire 27:169- 82. - - . 2009. Οη the Authorship of the Treatise on the Harmonization of the Opinions of the Two Sages Attributed to Al-Farabϊ. Arabic Sciences and Philosophy 19:43- 82. Rashed R. and J. Jolivet, ed. and trans. 1998. Oeuvres Philosophiques & Scientifιques d'al-Kindϊ. Volume 2, Mέtaphysique et cosmologie. Leiden, the Netherlands: Brill. Rawidowicz, S. 1943. 'Saadya's purification ofthe idea ofGod'. Ιη Rosenthal, 1943, 139- 65. al-Razϊ, Αbίι Bakr. 1939. Rasii'ilfalsa.fiyya (Opera philosophica), ed. Ρ. Kraus. Cairo: Paul Barbey. al-Razϊ, Αbίι f:Jatim. 1977, Kitiib Α Ίiim al-nubuwwa, ed. S al-Sawϊ. Tehran: Imperial Iranian Academy of Philosophy.

I 82 Reisman, D. C. 2005. Άl-Farabi and the philosophical curriculum'. In Adamson and Taylor 2005, 52- 71. Reynolds, G. S. 2004. Α Muslim Ίheologian in α Sectarian Milieu: Άbd al-Jabbiir and the Critίque of Chrίstian Origins. Leiden, the Netherlands: Brill. Rosenthal, Ε. 1. J., ed. 1943. Saadya Studies. Manchester, UK: Manchester University Press. Rudolph, U. 1996. Jbn Sina et le Kalam. Bulletίn d'Etudes Orientales 98:131 - 6. - - . 1997. Al-Miiturfdf und die sunnίtische Ίheologie in Samarkand. Leiden, the Netherlands: Brill. - - . 2007. Άl-Farabi und die Mu'tazila'. In Adang et al., 2007, 59- 80. - - . 2008. 'Reflections on al-Farabi's Mabiidi ' iirii' ahl al-madfna al-fiiqila'. In ln the Age of al-Fiiriibf: Arabic Philosophy in the Fourth!Tenth Century, ed. Ρ. Adamson, 1- 14. London: Warburg Institute. Saadia Gaon. 1880. Kίtiib al-Amiiniit wa Ί-/' tίqiidiit, ed. S. Landauer. Leiden, the Netherlands: Brill. - -. 1948. Ίhe Book of Beliefs and Opinion, trans. S. Rosenblatt. New Haven, CT: Yale University Press. Schδck, C. 2005. Koranexegese, Grammatik und Logik: Zum Verhiiltnis von arabischer und arίstotelischer Urteils-, Konsequenz-und Schlusslehre. Leiden, the Netherlands: Brill. al-Sharastani. 1984. Muslim Sects and Divisίons, trans. Α. Κ . Kazi and J. G. Flynn. London: Kegan Pau.l. Shihadeh, Α . 2005. From al-Ghazalί to al-Razί: 6th/12th Century Developments ίη Muslim Philosophical Theology. Arabic Sciences and Philosophy 15:141-79. Stroumsa, S. 1999. Freethinkers of Medieval Jslam . Leiden, the Netherlands: Brill. Taylor, R. C. 1998. Aquinas, the Plotίniana Arabica and the Metaphysics of Being and Actuality. Journal of the History of Jdeas 59:241-64. Urvoy, D. 1996. Les penseurs libres dans /Ίslam c/assique. Paris: Albin Michel. van Ess, J. 1967. Ibn Kullab und die Mi!)na. Oriens 18- 19:92- 142. - -. 1971. Fruhe Mu ' tazilίtische Hiiresίographίe. Beirut: Orient-Institut der Deutschen Morgenliίndischen Gesellschaft. - -. 1991- 5. Ίheologie und Gesellschaft im 2 . und 3. Jahrhundert Hidschra: Eine Geschichte des religiosen Denkens im fruhen 1s/am. Berlin: Walter de Gruyter. - -. 2006. Ίhe Flowering of Muslim Ίheology. Cambridge, ΜΑ : Harvard University Press. Walker, Ρ. 1993. Ear/y Philosophica/ Shiism: Ίhe lsmaili Neoplatonism of Abiι Ya'qiιb a/-Sijistiinf. Cambridge: Cambridge University Press. - -. 1999. f./amfd al-Dίn al-Kirmiinf: lsmaili Ίhought in the Age of a/-Ifiikim. London: Ι.Β. Tauris. Watt, W. Μ . 1994. Ίh e Faith and Practίce of al-Ghaziilf, 2nd edition. Oxford: Oneworld. - -. 1998. Ίhe Formative Period of Islamic Ίhought, 2nd edition. Oxford: Oneworld. Wisnovsky, R. 2003. Avicenna's Metaphysics in Context. London: Duckworth. Wolfson, Η. Α. 1977- 'Saadia οη the Trinity and Incarnation'. Ιη Wolfson, Studies in the Hίstory of Phi/osophy and Religion, vol. 2, 394- 414. Cambridge, ΜΑ: Harvard University Press. Young, Μ. J. L., J. D. Latham and R. Β. Serjeant, eds. 1990. Religίon, Learning and Science in the Άbbiisid Period. Cambridge: Cambridge University Press. Zimmermann, F. W. 1981. Al-Farabi's Commentary and Short Treatise on Aristotle's De

lnterpretatione. Oxford: Oxford University Press.

II

ARISTOTLE ΙΝ THEARABIC COMMENTAR Y TRADITION

ΙΝ late antiquity, the commentary became the most prominent genre of philosophical writing. Aristotle was the author who received the lion's share of attention, even though the commentators, beginning with Porphyry, were Platonists. They did produce commentaries οη Plato as well as Aristotle, 1 but the latter's writings are the subjects of a sizeable majority of extant Greek philosophical commentaries.' This fact is often explained ίη light of the pedagogical purposes for which commentaries were written. Since Aristotle was seen not only as harmonious with Plato, but as more suitable for initial study in philosophy, commentaries for the use of students were naturally more often devoted to his works than to Plato's. Α similar point applies to the proportion of extant commentaries οη different parts of Aristotle's corpus.3 Because logic was the first part of the philosophical curriculum, followed by physics and psychology, and only then finishing with metaphysics, we have many commentaries for Aristotle's organon, and quite a few for the physical works. By contrast, ίη the post-Plotinian period there are only two extant commentaries οη the Metaphysics, one by Syrianus, which covers only Metaphysics III-IV and XIII-XIV, and another by Asclepius which covers Metaphysics I-VII. 4 The practice of writing commentaries οη Aristotle, and the curriculum the commentaries were meant to support, cut across confessional lines. For instance Porphyry, Syrianus, the head of the Alexandrian school Ammonius,

Reproduced by permission of Oxford University Press, USA .

II

and his student Simplicius were all pagans. Indeed Porphyry and Simplicius were bitterly hostile to Christianity.5 Yet the commentary tradition continues smoothly into Christianity with Christian members of the school of Ammonius, and also into Latin with the works of Boethius. All of this applies, mutatis mutandis, to the Arabic commentary tradition. Here commentaries οη Plato are even more marginal. This can be explained not only by the same issue of Aristotle's role ίη the philosophical curriculum, but also by the fact that the transmission of Plato into Arabic was patchy, whereas most of the works of Aristotle we know today were also available ίη Arabic. 6 (The latter may itself result from the former; that is, one reason Plato was not much translated is that his works were not standardly studied as part of a broad philosophical education.) Here too, commentaries focus οη the earlier parts of the Aristotelian curriculum, with most emphasis οη the logical and physical works. Only the greatest commentator of the Arabic tradition, Averroes, commented extensively on the De Anima or the Metaphysics.7 As in the Greek tradition, confessional divides were ηο obstacle to continuous and even co-ordinated efforts to understand Aristotle.8 Thίs ίs best shown by the group of commentators known as the 'Baghdad school', who will be discussed below. For most, but not all, of the members of thίs 'school' were Christians, including its founder Abίi Bishr Matta. There are at least three good reasons to study the Arabίc tradition of commentary οη Aristotle. Fίrst, ίt can enhance our understanding of the Greek traditίon, not only by providing us with meticulous and ίnsightful readίngs of Arίstotle, as do the Greek commentarίes, but also because otherwise lost Greek material ίs preserved ίη Arabic translatίon.9 Second, ηο attempt to write the history of philosophy ίη the Islamίc world can afford to ignore the commentators. Two of the most ίmportant Muslim philosophers, al-Farabϊ and Averroes, produced numerous extant commentarίes, and ίη the case of Averroes at least there is little doubt that he considered his commentaries οη Aristotle to be the most ίmportant ofhis philosophίcal works. Third, some of the most interesting philosophical ίdeas produced ίη the medίeval Arabic tradition are to be found ίη commentarίes . Perhaps the best example ίs Averroes' mature posίtion οη ίntellect, whίch is presented ίη his Long Commentary on the De Anima. Before exploring the output of the Arabic commentators themselves, a word about lost Greek material preserved ίη Arabίc. At one time this was seen as the chief interest of the Arabίc tradίtion for the historίan of philosophy. Now this ίs ηο longer the case, given the growing interest taken ίη Arabic philosophy ίη its own right. And ίt is ίmportant not to overstate the extent to whίch lost Greek works are preserved in Arabic. But the fact remaίns that the Arabίc tradίtίon does preserve important works, and thίs is especίally true for the Greek commentators. Το gίνe just a few examples, the great Aristotelian commentator Alexander of Aphrodisίas wrote numerous works which are lost ίη Greek but extant in Arabic, such as his treatise οη providence and another work οη the prίncίples of the universe.'° Themistius' commentary οη Metaphysics ΧΙΙ is preserved ίη Arabic, Latίn, and Hebrew, but not ίη Greek. 11 Portions of Philoponus' commentary οη the Physics

II ARISTOTLE

ΙΝ ΤΗΕ

ARABIC COMMENTARY TRADITION

are preserved only ίη Arabic," ίη the so-called 'Baghdad Physics' which is further discussed below. Averroes is a major additional source for several Greek commentators, especially Alexander and Themistius. The fact that so many Greek commentaries were translated into Arabic, and the fact that Averroes and others actually quote the Greek commentators ίη their own commentaries, already shows the degree of continuity that exists between the Greek and Arabic philosophical traditions.'3 ln fact we can establish a more or less unbroken chain of authors who did philosophy by commenting οη Aristotle, spanning from Greek through Syriac and into Arabic. The late antique philosophical tradition terminates with the Alexandrian commentators in the sixth century. Some of these commentators were Christian, as mentioned above, and their work would be taken οη by two further Christian traditions. Ιη Byzantium, commentaries οη Aristotle continued to be produced in Greek. Elsewhere ίη the East, a monastic tradition of Syriac literature includes translations of, and commentaries οη, Aristotle, though this seems to have been limited to the logical corpus.' 4 When the translation of Aristotle and other Greek philosophical and scientific works began ίη earnest under the Άbbasid caliphs, there was a direct link to this Syriac tradition.' 5 This is, not least, because the translators often hailed from Syria or had a Syrian background. For instance Ibn Na'ima al-f:lim~ί, who wrote partial translations of the works of Plotinus and Philoponus' Physics commentary, was as his name indicates from Emesa (l;ιim~) ίη Syria.16 The great f:lunayn lbn Isl;ιaq (d. c. 873) hailed from Iraq, but he and his school translated from Greek via Syriac into Arabic. His speciality was medical literature, and he sought out and translated works of Galen. But the circle gathered around him, and especially his son Isl;ιaq Ibn f:lunayn, produced many of the Arabic versions of Aristotle that would be used by subsequent philosophers. ΑΙΙ these translators just mentioned were among the Greek-speaking Christians who were handsomely paid by the Muslim intelligentsia and political rulers to create an Arabic version of the Greek scientific corpus. Greek commentators were, as we have seen, being translated into Arabic right along with the works of Aristotle himself. So it was natural for philosophers writing original works in Arabic to understand Aristotle in light of these commentaries, and to produce commentaries of their own. This is shown already by the output of the first man to engage with the Greek tradition by writing original philosophical works in Arabic, al-Kindί (d. c. 870).17 Al-Kindί was deeply influenced by the Greek commentators, though it seems that he engaged more with independent treatises written by commentators like Philoponus and Alexander, rather than with their commentaries. The most prominent example is al-Kindί's proof that the world is not eternal, for which he drew οη Philoponus' critique of Aristotle's Physics and On the Heaνens. ' 8 His doctrine οη divine providence is indebted to the writings of Alexander οη this topic. 19 Al-Kindί also followed the example of these authors by writing commentaries of his own, but unfortunately his apparently rather extensive writings ίη this area are almost entirely lost. According to the list of his books provided by the Fihrist (or 'List') of the tenth-century author Ibn al-Nadίm, al-Kindί was especially engaged with logic. He wrote, for instance,

II

an account of the purpose and topic of the Categories, ' 0 a summary of Porphyry's Isagoge, and abridgments or paraphrases that related to other works of the organon like the Sophistical Refutations. We are also told that he wrote a work called On the Physics (Sam' al-kiyiίn), and this may have been a commentary. 11 Though the works just mentioned are all lost, we get a good idea of al-Kindί's knowledge of Aristotle from his On the Quantity ofAristotle's Books, which itemizes each work in the corpus and briefly explains the topic of each." This itemization again shows the influence of the late ancient curriculum, as al-Kindί emphasizes the need to study Aristotle's works in the correct order. He begins, as is traditional, with the logical works, and it is these works about which he is best informed. He thus devotes more discussion to the organon than to works οη physics, psychology, or ethics (his knowledge of the ethical works and the Politics seems to be especially sketchy; they are mentioned at the very end almost as an afterthought). This is unsurprising given the centrality of the logical works in the Syriac tradition, which will remain a feature of the Arabic reception of Aristotle. If we go ahead a generation or two, we find that philosophers who were influenced by al-Kindί also tended to concentrate οη logic.'3 His student al-Sarakhsί (d. 899) produced a voluminous corpus of philosophical works which are almost entirely lost. Again, our information shows that he wrote treatises, paraphrases, or commentaries οη almost the whole organon.' 4 Α second generation student of al-Kindί, al-Άmirί (991), also seems to have written commentaries οη Aristotle: it is ηο surprise that the only one for which we (perhaps) have extant fragments is a commentary οη the Categories.' 5 This same period was also the time of the most significant group of commentators οη Aristotle to work in the heartlands oflslam: the Peripatetic 'school' founded by ΑbίΊ Bishr Matta (d. 940). 26 As mentioned above, this school was made up mostly ofChristians, but its most famous representative was the Muslim al-Farabί (d. 950). Most of the extant commentaries οη Aristotle derive either from this group, which was based in Baghdad in the 1oth-11thcenturies, or from two commentators who lived later in Muslim Spain (Andalusia): Ibn Bajja (d. 1139), known in Latin as Avempace, and Ibn Rushd (d. 1198), known in Latin as Averroes. Averroes was certainly the most important Arabic commentator οη Aristotle, and was simply called 'the Commentator' by Aquinas and other philosophers in Christian Europe, whom he deeply influenced. But it is not always realized that Ibn Bajja and Averroes were hearkening back to the project of the Baghdad school when they wrote their commentaries. Ιη particular, they admired and followed al-Farabί, agreeing with the assessment of Avicenna (Ibn Sίna, d. 1037) in seeing him as the preeminent member of the Baghdad school, perhaps in part because he was a Muslim, but mostly because ofhis outstanding expertise in logic. It is in these two bursts of activity, and especially in the works of al-Farabί and Averroes, that we find the highpoints of the Arabic commentary tradition. Ιη fact, though, it is misleading to speak only of'commentaries'. Averroes' treatments of Aristotle were in fact written in a range of different styles, and only a few of these works can be described as commentaries in any strict sense. Traditionally, his expositions of Aristotle are divided into short, middle, and long commentaries.

II ARISTOTLE ΙΝ ΤΗΕ ARABIC COMMENTARY TRAD!TION

649

But as Dimitri Gutas has pointed out, this division 'fails accurately to depict what Averroes actually did, but rather seems to reflect European perceptions of the Latin Averroes'.' 7 What Averroes actually did was in large part to follow the lead of al-Farabί and Ibn BiΊjja. All three of these authors produced independent treatises οη topics drawn from Aristotle-for instance all three wrote brief works οη the nature of the intellect, which are tied to Aristotle's De Anima thematically but are not presented as expositions of the text.' 8 All three authors furthermore wrote epitomes of Aristotelian works, which summarize and even rearrange the source text. Particular energy was devoted to logic: al-FiΊriΊbί wrote epitomes of the entire organon (including not only Porphyry's Isagoge but also the Rhetoric and Poetics' 9 ), Ibn Bajja wrote commentaries οη these epitomes, 30 and Averroes composed an 'abstract' of al-Farabί's epitomes.31 He wrote epitomes ofhis own for other Aristotelian works, such as the Metaphysics.3 ' These epitomes need to be distinguished from longer expositions which clarify and convey the main points of an Aristotelian work. These expositions have a fairly fluid structure, and referring to them collectively as 'paraphrases' or 'middle commentaries' suggests more uniformity than is displayed by the texts themselves. Still, a good comparison here would be the so-called 'paraphrase commentaries' of Themistius. Examples of this form ίη the Arabic tradition include Ibn BiΊjja's exposition of the Physics and the works that go under the rubric of'middle commentaries' ίη Averroes.33 Finally, there are the works that actually fulfill our expectations of a proper 'commentary', ίη which the Aristotelian work is quoted in lemmata, each lemma being discussed ίη detail. Like their Greek models, these commentaries are massive and extraordinarily detailed, and would be of use only to a highly specialized and well-informed reader. So it is unsurprising that these lengthy commentaries are often lost. One example is al-FiΊriΊbί's commentary ad litteram on the Nicomachean Ethics, which is ηο longer extant but was still known in Andalusia.34 We do have examples from the Baghdad school, however: al-Farabί's commentary οη On Interpretation, the logical commentaries oflbn al-Tayyib, and the Baghdad Physics (all discussed below). History has been rather unkind to the 'long' commentaries-i.e., lemmatized commentaries-of Averroes, which are mostly lost ίη their Arabic originals. We have Arabic texts for his lemmatized commentaries οη the Posterior Analytics and Metaphysics, 35 but the commentaries οη the De Anima, De Caelo, and Physics are extant only ίη Latin or ίη Latin and Hebrew. As we will see below, the fact that many of Averroes' treatments of Aristotle are lost in Arabic is an important clue about the way that philosophy ίη the Muslim world was already moving away from a focus οη Aristotle in the 12th-13th centuries. As this survey has already suggested, logic was central to the activities ofboth the Baghdad school and the Andalusian revival. This focus οη logic can be traced back to the founder of the Baghdad Aristotelians, ΑbίΊ Bishr MattiΊ, who is perhaps best known for being οη the losing side of a debate over the relative merits of logic and grammar. This debate is an oft-cited example of early resistance to the incursion of Greek ideas into Muslim intellectual culture.36 ΑbίΊ Bishr, who also translated philosophical works from Syriac into Arabic, wrote several commentaries οη

II

Aristotle's logical works. These are unfortunately lost, as are most of the commentaries ofYabya Ibn Άdί (d. 974), who studied with both Abιl Bishr and al-Farabί.37 We do have an opportunity to compare al-Farabί and lbn Άdί as commentators ίη the case of On Interpretation, because al-Farabϊ's commentary and a short exposition of this text are extant,3 8 and lbn Άdί appends a commentary οη the famous chapter 9 ofthe work (concerning the 'sea battle' argument for determinism) to an otherwise independent treatise οη the nature of possibility.39 The last representative ofthe Baghdad school, Abιl 1-Faraj Ibn al-Tayyib, produced two extensive commentaries which have come down to us, οη Porphyry's Isagoge and the Categories.40 ΑΙΙ these commentaries show the extensive influence ofthe Alexandrian commentators.41 They are still quoted at length by Ibn al-Tayyib, and ίη a remarkable commentary οη the Physics produced jointly by several members of the school. This 'Baghdad Physics', extant ίη only a single manuscript, includes the Arabic translation of the Physics by Isbaq Ibn Jjunayn, passages from the Physics commentaries of Alexander and Philoponus ίη Arabic translation, and collated comments by several members of the Baghdad school, including Abιl Bishr, Ibn Άdϊ, the latter's student Ibn al-Samb, and Ibn al-Tayyib. 42 This manuscript is a valuable document for understanding the activities of the Baghdad school. One particularly striking feature is that the Greek commentators, especially Philoponus (called Yabya ίη the text, creating potential confusion with Yabya Ibn Άdί), are treated as interlocutors, fellow interpreters whose exegesis is useful though not beyond criticism. For example, both lbn Άdί and lbn al-Samb leap to defend Aristotle's original view against the innovations of Philoponus, respectively οη the topics of place and motion.43 The latter issue of motion is especially interesting: Philoponus had introduced the un-Aristotelian idea of 'impressed power', which has been compared to the modern theory of impetus. 44 Aristotle had argued that the speed of a moving body is inversely proportional to the density of the medium through which the body moves. Thus motion through a void would, absurdly, be infinitely fast (Physics IV 9). Philoponus disagrees, holding that motion through a void is possible, its speed being determined by its impressed power or impetus. Whereas Ibn al-Samb rejects this innovation, the later Andalusian commentator Ibn Bajja builds upon ίt, holding that the medium serves only to slow down the moving body.45 The process of innovation and retrenchment to Aristotelian orthodoxy was repeated when Averroes rejected the impetus theory ίη the version put forward by Ibn Bajja, just as Ibn al-Samb had rejected the version found ίη Philoponus. 46 Formal aspects of the Baghdad Physics and the lengthy commentaries of Ibn al-Tayyib also showthe impact of the Greek tradition. 47 For instance, some late Greek commentaries, such as those by Olympiodorus, quote lemmata from Aristotle and then summarize the gist of each lemma (the theδria) before moving οη to a detailed discussion, which proceeds through the text phrase by phrase (the lexis). This is imitated by Ibn al-Tayyib. 48 Another overt debt to the Alexandrian tradition is the practice of beginning commentaries with a series of standard questions about the Aristotelian text: why is it titled as it is, what is its place ίη the corpus, what is its

II ARISTOTLE

ΙΝ ΤΗΕ

ARABIC COMMENTARY TRADITION

subject-matter (topos), and so οη. Ibn al-Tayyib again provides a good example. Ιη his prolegomenon to the Categories, he not only covers these standard questions, but also follows the Alexandrian practice of devoting a special discussion, at the beginning of commentaries οη the Categories, to certain questions regarding philosophy as a whole. For example, how many philosophical schools are there? What is the goal of philosophy? How should Aristotle's corpus be divided? And so οη . 49 Another example of Alexandrian influence is the interest members of the Baghdad school take ίη issues of methodology and the philosophical curriculum. Ιη several ofhis works, al-Farabι emphasizes the need to study the philosophical sciences ίη the correct order, and shows how one science naturally leads οη to another ίη the systematic way sketched ίη Aristotle's Posterior Analytics.50 Α striking difference between al-Kindι's aforementioned On the Quantity of Aristotle's Books and al-Farabι's own summary of the Aristotelian corpus is that al-Farabι makes a more serious effort to describe the corpus as a structured curriculum.5' Al-Kindι too says that Aristotle's works must be studied ίη the correct order, as we have seen. But he does not explain the links between the various works ίη any detail, perhaps ίη part because his knowledge of these works is less complete than al-Farabι 's. This question of how the branches of philosophy interrelate is not merely an abstract methodological issue. lt can and does have specific consequences ίη deciding the meaning of Aristotle's texts. For instance, ίη his discussion of the deterministic argument of On Interpretation 9, al-Farabι chastises Greek commentators who thought that Aristotle was trying to prove that determinism is false. This cannot be right, because determinism is not an issue that can appropriately be settled ίη a logical work. Logic, he says, does not tel1 us 'regarding the natures of things that exist, how they exist.' Rather, according to al-Farabι, Aristotle simply assumes the falsity of determinism ίη order to make a properly logical point about the 'indeterminacy' of assertions about the future.5' The most important such procedural question to arise ίη the Arabic tradition concerned the relation of physics to metaphysics. Though it was widely agreed that physics is subordinate to metaphysics-after all, metaphysics is 'first philosophy'-there was considerable disagreement about which science is responsible for proving God's existence. The central figure here, as so often ίη the history of philosophy ίη the Muslim world, was Avicenna. His self-consciously original approach to philosophy is shown by the fact that instead of writing commentaries οη Aristotle, he composed self-contained works covering more or less the same ground as Aristotle's corpus (and then some, since for instance mathematics was also included).53 The most voluminous of these Avicennan works is The Healing (al-Shifii'), a massive collection of original treatises οη every area of philosophy. But he wrote others, including the enigmatic Pointers and Reminders (al-Ishάrάt wa-l-tanbϊhiit). If Avicenna intended these works effectively to supplant Aristotle with a new system, then his intentions were largely fulfilled. As we will see below, ίη the post-Avicennan tradition philosophical commentary was mostly devoted to expounding Avicenna, rather than Aristotle. Avicenna did weave Aristotelian themes into his system, so that for instance the Metaphysics or Divine Science

II 652

(Ilahiyyat) of the Healing is related to the Metaphysics of Aristotle ίη an articulated and coherent way.54 But his handling and placement of these themes was, again, self-consciously innovative. Α prime example is this question Ι have just raised: which branch of philosophy proves God's existence, physics or metaphysics? Aristotle certainly seems to prove the existence of the First Mover ίη Physics VIII, based οη the need for an eternal, immaterial mover to explain the eternal circular motion of the heavens. But Avicenna developed a famous proof for God which requires ηο physical premises. Rather, his proof begins from the 'immediate' notions of existence, necessity and contingency. Avicenna argues that the existence of contingent objects (things 'possible ίη themselves') requires a cause external to the aggregate of these objects, which will be an existent that is necessary ίη itself. He further argues that this Necessary Existent would have the attributes we associate with God.55 Ιη his commentaries, Averroes repeatedly criticizes this strategy of proving the existence of God ίη metaphysics, and of using a proof that supposedly proceeds from first principles. Rather, as he says ίη numerous contexts including his Long Commentary on the Metaphysics, it is physics that proves the existence of God οη the basis of eternal motion.56 Here he follows not only Aristotle but also al-Farabί, whose Attainment of Happiness likewise claims that natural philosophy leads οη to metaphysics by proving the existence of immaterial movers. This methodological debate must be understood within the context of a broader problem about the subject-matter of metaphysics itself. For late ancient Platonists, metaphysics or first philosophy was the study of divine principles. Aristotle sometimes seems to support this view as well, especially in an influential passage at Metaphysics VI 1. Here he says that theoretical philosophy is divided into mathematics, physics, and 'first science' or 'theology (theologike)', the latter dealing with things that 'exist separately and are immovable'. There is a long tradition of reading the entire Metaphysics in light of this passage, as devoted specifically to divine things (God and the celestial movers, and perhaps also human souls). We find this attitude in al-Kindί's On the Quantity of Aristotle's Books and also in Ibn Άdί's commentary οη Metaphysics ΙΙ.57 This would align the project of the Metaphysics nicely with the theological preoccupations ofNeoplatonism. Such an understanding of Aristotle's project makes it easier to see how the Arabic version of Plotinus' Enneads could be harmonized with the Aristotelian corpus and even mistaken for a work by Aristotle himself. But al-Farabί puts forward a more subtle view of 'first philosophy' ίη his short work On the Aims of the Philosopher, which explains the purpose of the Metaphysics and each of its books.58 He claims that the subjectmatter of first philosophy or metaphysics is not God and other divine entities, but being as such. This includes the study of divine principles-which are the causes of other beings-but also of material substances, as well as universal principles of reasoning (such as the principle of non-contradiction) that apply equally to all beings. Avicenna follows al-Farabί's identification of the subject-matter of metaphysics: being as such, and not the divine.

II ARISTOTLE

ΙΝ ΤΗΕ

ARABIC COMMENTARY TRADITION

But what does all this have to do with the question of whether physics or metaphysics proves the existence of God? The connection lies ίη another methodological principle taken from the Posterior Analytics, namely that ηο science proves the existence of its own subject-matter. This rule makes a certain amount of sense. After all, if a subject-matter fails to exist, then there will be ηο science of this subject-matter. And it seems circular that a science should prove that there is a need for itself. Aristotle puts the rule into practice elsewhere, for instance, ίη Book Ι of the Physics when he states that the student of physics need not prove the existence of motion. Aristotle goes οη to do so anyway, but this, he says, is simply because of the philosophical interest of the question, not because the task is incumbent upon the student of physics (Ι 2 184b25-185a20). By denying that God is the subject-matter of metaphysics, Avicenna was able to reconcile his 'metaphysical' proof of God with Aristotle's methodological stricture. But the manoeuvre did not meet with Averroes' approval. Though he too was deeply influenced by al-Fiiriibϊ's understanding of the Metaphysics, 59 ίη at least some passages he adopted a more 'theological' reading of the work. He was ηο doubt at Ieast partially motivated by the desire to retain Aristotle's proof of God as the first cause of motion ίη the Physics. For by holding that God and the other celestial movers are at least part of the subject-matter of metaphysics, Averroes could show that a properly metaphysical proof of their existence was illegitimate. Rather, God can be approached only from the lower science of physics, using a proof that moves from empirical observation of God's effects to His existence, rather than proceeding οη the basis of immediately evident first principles.60 Οη this and οη other topics, Averroes intended to provide an authentic reading of Aristotle shorn of Platonist and Avicennan presuppositions. This meant a meticulous reading of Aristotle's texts and an unashamed acceptance of passages where Aristotle rejects Platonist metaphysical commitments. Significantly, for Averroes the most important previous commentator οη Aristotle was the Peripatetic Alexander, whereas the Baghdad school was influenced primarily by the late Alexandrian Neoplatonist school (though they too read and made use of Alexander). Averroes' reading of Aristotle is, ίη short, a far cry from the Platonizing and harmonizing approach of al-Kindί. Ιη his determination to achieve fidelity to Aristotle, Averroes also departed from and indeed frequently attacked the innovations of Avicenna. More difficult is the question of the extent to which Averroes' reading is presaged by al-Fiiriibί . He is credited with having written On the Harmony of the Two Sages, which systematically undermines reasons for thinking there is any significant disagreement between Plato and Aristotle.6' If the Harmony ίs indeed a work of al-Fiiriibϊ, it shows that he was fully committed to compatability of Plato and Aristotle, a thesis usually associated with Greek Neoplatonists (even though it was clearly accepted only with reservations, if at all, by some Neoplatonists, such as Syrianus). Furthermore, his own systematic works make extensive use of such standard Neoplatonic notions as emanation. However, ίη other contexts al-Fariibϊ has ηο qualms about mentioning, and agreeing with, Aristotle's hostility towards Platonic Forms. For this and other reasons it has recently been questioned whether

II

the Harmony is genuinely a work of al-Farabϊ. If it is not, it may nonetheless represent the attitude of other Baghdad Aristotelίans. For there are reasons to think that Ibn Άdϊ adhered more closely to a Kindian understanding of Greek philosophy as a single, harmonious body of doctrine. 6' Ιη light of Averroes' determination to 'de-Platonise' the Aristotle handed down to him, it is often thought ironic that he developed a theory of intellect which seems to go beyond any ofhis predecessors ίη a Platonising direction. The theory, sometimes misleadingly called 'monopsychism' (misleading because psuche ίη Greek means 'soul,' and this theory concerns not the soul but specifically the intellect), is the most notorious interpretation of Aristotle presented ίη any Arabic commentary. But despite its notoriety it is often misunderstood, and one misunderstanding that needs to be avoided is the notion that the theory grows out of a Platonist tradition. Rather, it grows out of Averroes' sustained engagement with Aristotle and specifically the De Anima. Averroes' views οη the correct reading of the teaching οη intellect in the De Anima evolved throughout his career, so that we find three different interpretations ίη the epitome ('short commentary'), the exposition ('middle commentary'), and the lemmatized Long Commentary. 63 His final position is found ίη the Long Commentary, which defends the controversial and seemingly preposterous claim that there is only one intellect for all of mankind. The first thing that should be said here is that Averroes was far from being the first to posit a single intellect to explain all human thought. Many, indeed most, readings of Aristotle from late antiquity through the Arabic tradition suppose that there is a single intellect which is a repository of universal and intelligible forms, and which enables us to grasp these intelligibles. For such authors as al-Farabϊ and Avicenna, this single intellect is to be found already ίη Aristotle: it is the entity discussed ίη the enigmatic fifth chapter of De Anima ΠΙ. They refer to it as the 'active intellect'.6 4 It has a dual role ίη the Farabian and Avicennan systems. Οη the one hand, the active intellect emanates forms into suitably prepared matter, and this accounts for the generation of material substances such as animals and individual humans. Οη the other hand, the intellect illuminates a suitably prepared human soul with intelligible forms. This is meant to explain how humans can achieve an adequate grasp of necessary and universal intelligibles, which might be difficult to account for by appealίng to the empirical process of abstraction.6 5 The theory sounds rather Platonist, and indeed there is a clear similarity between this version of Aristotelian noetics and the doctrine of Plotinus, for whom the universal intellect is just identical with the realm of Platonic Forms. But al-Farabϊ and Avicenna saw their embrace of the active intellect as perfectly consistent with Aristotle's anti-Platonic strictures. For the intelligible forms ίη this intellect are thoughts ίη a mind, not separately existing paradigms. Averroes, then, was not saying anything unusual ίη arguing that a single, universal intellect is involved ίη human thought. Where he departs from the tradition is rather ίη affirming that there is only one material or potential human intellect. Al-Farabϊ and Avicenna, like the Greek commentators, had assumed that each human has an intellective or rational soul which is unique to him or her. It is this

II ARISTOTLE ΙΝ ΤΗΕ ARABIC COMMENTARY TRADITION

655

numerically distinct intellect which has the potential to receive intelligibles from the single active intellect. Against this, Averroes points out that there cannot be two numerically distinct potentialities for exactly the same actuality, unless this potentiality is seated ίη a numerically distinct body. But we know from De Anima ΠΙ 4 that the human intellect has ηο bodily organ. If there is ηο material object to individuate the human capacity for thought, then there is ηο way of individuating two distinct instances of such a capacity. Rather, there is only one capacity which we all share, and it is permanently and fully actualized by the universal active intellect. This explains why two people who are thinking about the same intelligible object are quite literally having the same thought, and not two distinct thoughts that are very similar to one another. Το put it another way, there is only one correct scientific understanding of any intelligible, as opposed to a distinct understanding for each person who comes to grasp the intelligible ίη question. The extent to which this position is anti-Platonist is shown by the fact that Averroes adheres to an empiricist, 'bottom up' account of how universal intellection is generated. Individual humans have lower psychological faculties (memory, imagination, and what he refers to as 'cogitation') seated in their brains, and these serve as the basis for a universal abstractive grasp of intelligibles. The involvement of these lower faculties also explains how it can be that, if the universal intellect belongs to all of us and is always thinking about everything, each human has only intermittent experiences of intellecting certain intelligibles to the exclusion of others. It is only when my imagination, memory, and so οη are used as the basis for universal intellection that Ι have the experience of intellecting. Thus Averroes already anticipates and answers a complaint of Thomas Aquinas, who attacks the teaching of the unity of the material intellect by saying that it cannot account for the fact that this man thinks. 66 While it is true that there is only one intellect for all of us, there is an explanation for why this man has the subjective experience of thinking and that man does not. It is just that this explanation is grounded in psychological faculties lower than the intellect. 67 Averroes' theory of the material intellect sparked a great deal of controversy in Latin Christian philosophy, as did his acceptance of the world's eternity. But even those Christians who, like Aquinas, opposed his teaching οη these two issues still saw him as the most reliable and important commentator οη Aristotle. Similarly, in the Jewish tradition Maimonides (d. 1204) commends the use of Averroes as a guide to Aristotle, while Levi Ben Gerson (Gersonides, d. 1344) and his students composed super-commentaries, i.e., commentaries οη Averroes' commentaries. 68 Ιη both Hebrew and Latin, the translations of Averroes' commentaries were an important means of transmitting the Aristotelian text itself. 69 For when the long commentaries were translated, the lemmata of the Aristotelian text were translated along with them. ΑΙΙ of this is in stark contrast to the fortunes of Averroes in the Arabic-speaking Muslim world. It is telling that, as mentioned above, many of his commentaries are lost ίη Arabic but extant ίη Latin, Hebrew, or both. Even his magisterial Long Commentary on the Metaphysics is preserved in only a single Arabic manuscript.

II

The reasons are not far to seek. For one thing, Averroes worked οη the geographical fringes of the Muslim world, in Andalusia, and unlike his fellow Andalusians Ibn Άrabϊ and Maimonides he never travelled to the East. Το this one can add the technical and even forbidding nature of his commentaries. It's not surprising that the more accessible Decisive Treatise and Incoherence of the Incoherence, Averroes' response to al-Ghazalϊ's Incoherence of the Philosophers, are indeed extant ίη Arabic and were read to some extent in the later Muslim tradition. Yet a more fundamental reason for Averroes' lack of influence was the very nature of his commentatorial enterprise. As we have seen, Averroes rejected the innovations of Avicenna and tried to return to a more authentic and detailed reading of Aristotle. This may have been an avant garde project when al-Farabϊ and his colleagues undertook it in tenth-century Baghdad, but by the late twelfth century it was decidedly old hat. The Muslim philosophical tradition had moved οη to grappling with the thought of Avicenna, and although Avicenna was far from universally accepted or admired, his philosophy demanded a detailed engagement rather than the reactionary criticism he provoked from Averroes. Α dramatic illustration of this is the fact that the tradition of philosophical commentary not only continued ίη the eastern Islamic world, but actually blossomed; yet almost none of the commentaries expounded Aristotle. There is a vast extant corpus of commentaries that fuse philosophical discussions with discussions of the Muslim creed or themes ίη Islamic theology, as well as commentaries οη philosophical works.7° Within the latter category there are many commentaries devoted to Avicenna's writings, especially his Pointers and Reminders, whose compressed and allusive writing style invited commentators to try their hand at expounding the text.71 The most important commentators οη this text were Fakhr al-Dϊn al-Razϊ (d. 1210) and Na~ϊr al-Dϊn al-Tίisϊ (d. 1274). Al-Razϊ, an important theologian ίη his own right, fashioned his commentary into a complex dialectical critique of Avicenna, which provoked a defensive and generally approving commentary from al-Tίisϊ. These two commentators are only the best-known representatives of a tendency in the Muslim philosophical tradition, from the eleventh century onwards, to engage primarily (if often critically) with Avicenna and not with Aristotle.7' There are exceptions, notably Άbd al-Latif al-Baghdadϊ (d. 1231), who wrote a paraphrase commentary of Metaphysics ΧΙΙ and whose hostility to Avicenna was easily a match for that shown by Averroes.73 Generally speaking, though, the direct engagement with Aristotle ended rapidly ίη the East once Avicenna came οη the scene, and the surge of interest ίη Andalusia with Ibn Bajja and Averroes was a short-lived exception. Despite this, philosophy in the Islamic world continued to engage with Aristotelian ideas and themes for centuries to come. Indeed this engagement persisted for at least as long as ίη the West, where Aristotle continued to be a more directly dominant figure through the Renaissance. If we consider a later development like the so-called school of Isfahan, a group of thinkers who revived Neoplatonism under the Safavids ίη Iran during the 16-17th centuries, we do not find any commentaries οη Aristotle himself.74 Yet it has recently been argued that

II ARISTOTLE

ΙΝ ΤΗΕ

ARABIC COMMENTARY TRADITION

their most famous representative, Mulla ~adra (d. 1640), structured his magnum opus The Four Journeys around themes drawn from Aristotle's Metaphysics.7 5 More broadly, the terminology and problems of philosophy in this later period are often reminiscent of Avicenna, who despite all his innovations served as a conduit for Aristotelian thought. Α standard question discussed in the later tradition, for instance, was 'does the Necessary Existent have knowledge of particulars?' That is, does God know about each of the things that exist in the created realm? This is a difficulty first discussed explicitly by Avicenna, posed in Avicennan language. Yet that very language preserves Aristotelian vocabulary. In our example, the Arabic word for 'particulars' would be al-juz' iyyάt, corresponding to ta kath' hekasta. Furthermore, Avicenna's reasons for denying that God knows particulars as such are rooted in Aristotle.76 So it was Avicenna, more than any other thinker, who was responsible for putting an end to an interfaith culture of Arabic commentary on Aristotle. But ίt was also Avicenna, more than any other thinker, who passed on Aristotelian concepts, terms, and difficulties to the later philosophical and theological tradition in the Islamic world.77

NOTES For Porphyry as the starting-point of this tradition see G. Karamanolis, 'Porphyry: the First Platonist Commentator οη Aristotle; ίη Ρ. Adamson, Η . Baltussen, and M.W.F. Stone eds., Philosophy, Science and Exegesis in Greek, Arabic and Latin Commentaries, 2 vols (London: 2004) , vol. 1, 97- 120. Porphyry also commented οη Plato, though these commentaries are lost (unless we ascribe to him the anonymous Commentary on the Parmenides) . Despite the fact that all subsquent commentators οη Aristotle, with the arguable exception ofThemistius, were card-carrying Platonists, Olympiodorus (d. 565 AD) is the only Greek commentator for whom we have extant commentaries οη both Plato and Aristotle. Of course the tradition of commentary οη Plato is dominated by Proclus, from whom we have commentaries οη the Cratylus, First A/cibiades, Parmenides, Republic, and Timaeus. 2 . Indeed commentaries οη Aristotle, and independent works by these same commentators, constitute approximately half of all directly preserved Greek philosophical works, by the reckoning of R. Goulet, 'La conservation et la transmission des textes philosophiques grecs; ίη C. D'Ancona ed., Libraries of the Neoplatonists (Leiden: 2007), 29-61; see the remarkable pie chart at p. 60. According to this chart, works by Neoplatonists make up approximately 80 percent of all directly extant Greek philosophical literature. 3. Of course many more commentaries were produced than are extant today, both in the Greek and Arabic traditions. Thus our surviving evidence is an imperfect guide to the priorities and interests that held sway ίη these traditions. Οη the other hand, such factors as teaching needs can also help explain not only why certain commentaries were written, but also why they were, or were not, preserved down to the present day. The points that follow, regarding the dominance oflogic for instance, or the dominance of Aristotle over Plato in the Arabic tradition, are largely borne out by our information about lost commentaries as well as by the extant evidence. But we know that Plato 1.

II

was ίη fact read ίη Arabic, at least ίη paraphrase versions, more often than would be suggested by the texts we possess now. 4. For English translations, see J. Dillon and D. O'Meara (trans.), Syrianus: On Aristotle Metaphysics 3-4 (London: 2008); and Syrianus: On Aristotle Metaphysics 13-14 (London: 2006). There was also a commentary by Themistius, which is ηο longer extant ίη Greek; see note 11. 5. See recently Μ. Edwards, 'Porphyry and the Christians; ίη G. Karamanolis and Α. Sheppard, eds., Studies on Porphyry (London: 2007). Η. Baltussen, Philosophy and Exegesis in Simplicius. Ίhe Methodology of α Commentator (London: Duckworth, 2008) emphasizes the anti-Christian polemic ίη Simplicius' work. Anti-Christian sentiment has also been suspected ίη Proclus' work: see H .D. Saffrey, Άllusions antichretiennes chez Proclus, le diadoque platonicien; Revue des Sciences

philosophiques et theologiques 59 (1975), 553-63. Όη the Knowledge of Plato's Philosophy ίη the Islamic World; Islamic Culture 14 (1940), 387-422; F.E. Peters, Aristoteles Arabus (Leiden: 1968). The only

6. See F. Rosenthal,

significant extant commentary οη Plato from the Arabic-speaking world is Averroes' commentary οη the Republic. See R. Lerner, trans., Averroes on Plato's Republic (Ithaca, Ν . Υ.: 1974).

7. There are, however, notes οη the De Anima and Metaphysics ΧΠ by Avicenna; see note 53. We also have evidence that Ibn al-Tayyib, οη whom see below, wrote a commentary οη the Metaphysics. And we have an extant commentary, again οη book ΧΙΙ, by the mathematician and philosopher Thabit Ibn Qurra. Οη this see now D.C. Reisman and Α. Bertolacci, 'Thabit Ibn Qurra's Concise Exposition of Aristotle's Metaphysics: Text, Translation and Commentary', ίη R. Rashed and Μ. Rashed, eds., Sciences and Philosophy in 9th Century Baghdad. Ίhabit Ibn Qurra (826-901) (Berlin: 2009), 715-76. 8. This is one reason I speak here of an Arabic tradition of commentary οη Aristotle, rather than an Islamic tradition. 9. It should also not be forgotten that Arabic translations of Aristotle and of Greek commentaries are sometimes several hundred years earlier than the earliest Greek manuscripts we have. They provide an important resource for establishing the text of even extant Greek works. 10. For the former see the translations ίη Ρ. Thillet, Alexandre d'Aphrodise: Traite de /α providence (Lagrasse: 2003); and S. Fazzo and Μ. Zonta, Alessandro di Afrodisia: La Provvidenza (Milan: 1998). For the latter, see C. Genequand, Alexander of Aphrodisias on the Cosmos (Leiden: 2001). 11. See the translation of R. Brague, Ίhemistius: Paraphrase de /α metaphysique d'Aristote: Livre Lambda (Paris: 1999). 12. For an English translation of these sections see Ρ. Lettinck and J. Ο. Urmson, trans., Philoponus: On Aristotle Physics 5-8; Simplicius: On Aristotle on the Void (London: 1994). The Arabic tradition also provides important evidence concerning Philoponus' critique of Aristotle's position οη the eternity of the world. See, for instance, S. Pines, Άη Arabic summary of a lost work ofJohn Philoponus', Israel Oriental Studies 2 (1972), 320-52. 13. See the useful overviews of what was translated ίη C. DΆncona, 'Greek Sources ίη Arabic and Islamic Philosophy', online at the Stanford Encyclopedia of Philosophy; and ίη DΆncona, 'Greek into Arabic: Neoplatonism ίη Translation', ίη Ρ. Adamson and R. C. Taylor, eds., Ίhe Cambridge Companion to Arabic Philosophy (Cambridge: 2005), 10-31.

14. See, e.g., Η. Hugonnard-Roche, Tintermediaire syriaque dans la trasmission de la

philosophie grecque a l'arabe: le cas de l'Organon dΆristote; Arabic Sciences and Philosophy 1 (1991), 187-209; S. Brock, 'The Syriac commentary tradition', ίη

II ARISTOTLE

ΙΝ ΤΗΕ

ARABIC COMMENTARY TRADITION

C. Burnett, ed., Glosses and Commentaries on Aristotelian Logical Texts: The Syriac, Arabic and Medieval Latin Traditions (London: 1993). There are numerous extant Syriac commentaries οη the Isagoge, Categories, On Interpretation , and Prior Analytics; see the useful list at Brock, 11-15. Αη issue that arises here is how much of the organon was studied ίη the Syriac tradition. Οη this see further D. Gutas, 'The Άlexandria to Baghdad' complex of narratives. Α contribution to the study of philosophical and medical historiography among the Arabs; Documenti e Studi sulla Tradizion e Filosofica Medievale 10 (1999) , 155-193. 15. For the translation movement see D. Gutas, Greek Thought, Arabic Culture: The GraecoArabic Translation movement in Baghdad and early society (2nd-4th!Bth-1oth centuries) (London: 1998). 16. For his translation of Plotinus, part of which was mistaken for a work by Aristotle and called the Theology of Aristotle, see Ρ. Adamson, The Arabic Plotinus (London: 2002). 17. Οη whom see Ρ. Adamson, Al-Kindϊ (New York: 2007). 18. See Η . Α . Davidson, 'John Philoponus as a Source ofMedieval Islamic and Jewish Proofs ofCreation', Journal of the American Oriental Society 89 (1969) , 357-91; and

Adamson, Al-Kindϊ (see previous note), ch. 4. Η. Wiesner, Άlexander of Aphrodisias ίη the Κίηdί Circle and ίη al - Kindϊ's Cosmology', Arabic Sciences and Philosophy 3 (1993), 119-53; and Adamson, Al-Kindί (see note 17), ch. 8. 20. This is clearly related to the set of propaedeutic questions asked about each Aristotelian work by Greek commentators, which I discuss further below. 21. For a list of al-Kindϊ's writings see R. J. McCarthy, al-Ta~anif al-mansuba ila Faylasuf al- 'arab (Baghdad: 1962); and for an English translation of the whole Fihrist see Β. Dodge, The Fihrist of al-Nadim (New York: 1970), with the list of al-Kindi's works at

19. See S. Fazzo and

615- 26. 22. See Μ. Guidi and R. Walzer, Uno Scritto Introduttivo allo Studio di Aristotele (Rome: 1940). 23. For this line of thinkers, whom I have elsewhere called the 'Kindian tradition;

see Ρ. Adamson, 'The Kindian Tradition: the Structure of Philosophy ίη Arabic Neoplatonism; ίη C. DΆncona ed., Libraries of the Neoplatonists (Leiden: 2007), 351-70. 24. See F. Rosenthal, A~mad b. at-Tayyib (New Haven: 1943), 54. 25. See Ε. Wakelnig, 'Philosophical Fragments of al-Άmirί Preserved Mainly ίη al-Tawbϊdί,

Miskawayh and ίη the Texts ofthe ~iwiin al-l;ιikma Tradition', ίη Ρ. Adamson, ed., In the Age of al-Fariibί: Arabic Philosophy in the Fourth!Tenth Century (London: 2007), 220. 26. Οη whom see Ρ. Adamson, 'Knowledge ofUniversals and Particulars ίη the Baghdad School', Documenti e Studi sulla Tradizione Filosofica Medievale 18 (2007), 141-64.

27. D. Gutas, Άspects ofLiterary Form and Genre ίη Arabic Logical Works', ίη Burnett, ed., Glosses and Commentaries (see note 14), 29-76, at 55. 28. We already find a work ofthis kind by al-Kindί : his On the Intellect, οη which see J. Jolivet, ι:Intellect selon Kindϊ (Leiden: 1971); Ρ. Adamson, Al-Kindϊ (see note 17 above), 118-27. For al-Farabϊ's On the Intellect see the English translation ίη J. McGinnis and D. C. Reisman, ed. and trans., Classical Arabic Philosophy: an Anthology of Sources (Indianapolis: 2007), 68-78. For lbn Bajja's On Conjunction see Μ. Αsίη Palacios, 'Un texto de Avempace sobre la uniόn del intelecto con el hombre', ίη Al-Andalus 7 (1942), 1-47. Averroes wrote several epistles οη the topic; see Κ. Ρ. Bland, trans., Averroes. Epistle on the Possibility of Conjunction (New York:

II 660 1982); Μ. Geoffroy and C. Steel, Averroes. La Beatitude de /Ame. Editions, traductions et etudes (Paris: 2001) . 29. Οη this inclusive version of the organon see D. L. Black, Logic and Aristotle's Rhetoric and Poetics in Medieval Arabic Philosophy (Leiden: 1990 ). 30. See Μ. Fakhry, ΤαΊ'ίlϊq Ibn Bάjja 'α/ά man(iq Αriψ:ι (Beirut: 1994). 31. Οη this see again Gutas, Άspects ofLiterary Form' (note 27 above), esp. 47-50, 54- 6. 32. This work is an example of the fact that epitomes could radically rearrange the source text: see R. Arnzen, Ίbη Rushd οη the Structure of Aristotle's Metaphysics: Documenti e Studi sulla Tradizione Filosofica Medievale 21 (2010), 375-410. 33. For Ibn Biijja οη the Physics see Ρ. Lettinck, Aristotle's Physics and It's Reception in the Arabic World (Leiden: 1994). It's worth noting that Averroes wrote 'commentaries' of this sort οη Galen as well as Aristotle; we can also put his commentary οη Plato's Repub/ic into this category. 34. It is referred to by Ibn Biijja, Averroes, and Maimonides: see Α. Α . Akasoy and Α. Fidora, The Arabic Version of the Nicomachean Ethics (Leiden: 2005), 49-52. 35. For the Posterior Analytics commentary see Α . Badawί, ed., Ibn Rushd. Grand commentaire et Paraphrase des Seconds Analytiques d'Aristote (Kuwait: 1984). For the Metaphysics commentary see Averroes, Tafsϊr Μά Ba'd at-Tabϊ'at, ed. Μ. Bouyges, 3 vols. (Beirut: 1938-52). 36. For a translation of the report of this debate, see D. S. Margoliouth, 'The Discussion Between Abu Bishr Matta and Abu Sa'id al-Sirafi οη the Merits ofLogic and Grammar', Journal of the Royal Asiatic Society (1905), 79-129. 37. See G. Endress, The Works of Υα~yά Ibn 'Αdϊ. An Analytical Inventory (Weisbaden: 1977). 38. F. W Zimmermann, Al-Farabi's Commentary and Short Treatise on Aristotle's De Interpretatione (Oxford: 1981); Α . Hasnawi, 'Fariibί et la pratique de lexegese philosophique (remarques sur son Commentaire au De Interpretatione d'Aristote)', Revue de Synthese 3rd series, 117 (1985), 27-59. 39. For which see C. Ehrig-Eggert, Ύal;ιya ibn Άdi: ϋber den Nachweis der Natur des Mδglichen; Zeitschrift fur Geschichte der arabisch-islamischen Wissenschaften 5 (1989 ), 283-97 (Arabic text 63-97). Part of this work is translated ίη McGinnis and Reisman, Classica/ Arabic Philosophy, 128-39. (See note 28.) Οη the topic see further Ρ. Adamson, 'The Arabic Sea Battle: al-Fariibi οπ the Problem of Future Contingents', Archiv fur Geschichte der Philosophie 88 (2006), 163-188, and 'Freedom and Determinism; in R. Pasnau, ed., The Cambridge History of Medieval Philosophy, 2 vols (Cambridge: 2010), νο!.1, 399-413. 40. See Κ. Gyekye, ed., Ibn al-Tayyib's Commentary on Porphyry's Eisagoge (Beirut: 1975); id. (trans.), Arabic Logic: Ibn al-Tayyib's Commentary on Porphyry's Eisagoge (Albany: 1979); C. Ferrari, Die Kategorienkommentar von Abiί 1-Farάj Άbdallάh Ibn al-Tayyib (Leiden: 2006). 41. ΑΙ- Fiiriibi's indebtedness to the Alexandrians is emphasized in Ρ. Vallat, Farabi et l'ι!cole d'Alexandrie (Paris: 2004). 42. Οη this commentary see Lettinck, Aristotle's Physics (see note 33). Lettinck summarizes and discusses the comments of the Baghdad school alongside the later Physics commentary oflbn Biijja. See further Ε. Giannakis, Philoponus in the Arabic Tradition of Aristotle's Physics, D. Phil. Thesis (Oxford 1992); id., 'The Structure of Αbίι 1-1:fusayn al-Ba~rί's Copy of Aristotle's Physics: Zeitschrift fur Geschichte der arabischislamischen Wissenschaften 8 (1993), 251-58.

II ARISTOTLE

ΙΝ ΤΗΕ

ARABIC COMMENTARY TRADITION

661

43. See the summary ofLettinck, Aristotle's Physics (see note 33), 19, 21; Ε. Giannakis,

ibn Άdί Against John Philoponus οη Place and Void; ίη Zeitschrift fur Geschichte der arabisch-islamischen Wissenschaften 12 (1998), 245-302. 44. Οη this see F. W. Zimmermann, 'Philoponus' Impetus Theory ίη the Arabic Tradition', ίη R. Sorabji, ed., Philoponus and the Rejection of Aristotelian Science (London: 1987), 121-9; R. Sorabji, Matter, Space, and Motion (London: 1988). 45. See S. Pines, 'La dynamique dΊbn Biijja; ίη Mέlanges Alexandre Koyrέ Ι (Paris: 1964), 442-68; Lettinck, Aristotle's Physics, 342; id., 'The Transformation of Aristotle's "Physical Philosophy" ίη Ibn Bajja's Commentaries; ίη F. J. Ragep and S. Ρ. Ragep, eds., Tradition, Transmission, Transformation (Leiden: 1996), 65-70; J. Puig Montada, 'Philosophy ίη Andalusia: Ibn Bajja and Ibn Tufayl', ίη Ρ. Adamson and R. C. Taylor, eds., The Cambridge Companion to Arabic Philosophy (Cambridge: 2005), 155-79, at Ύal;ιya

160-1. 46. See Lettinck, Aristotle's Physics, 343-4. 47. Ibn al-Tayyib wrote a very large number of commentaries οη Aristotle, most of which are lost; see Ferrari, Der Kategorienkommentar, 30-1 (see note 40). 48. See Ferrari, Der Kategorienkommentar, 44; Hasnawi, 'Fiirabί et la pratique', 53. Ι have proposed seeing this as a structuring principle oflbn Άdί's commentary on Metaphysics α: see Ρ. Adamson, Ύal;ιyii Ibn Άdί and Averroes οη Metaphysics Alpha Elatton', Documenti e Studi sulla Tradizione Filosofica Medievale 21 (2010),

343-74. 49. See Ferrari, Der Kategorienkommentar, 95ff. Cf. L. G. Westerink, 'The Alexandrian

50.

51.

52.

53.

Commentators and the Introductions to their Commentaries', ίη R. Sorabji, ed., Aristotle Transformed (London: 1990 ), 325-48. Α similar set of opening questions was applied to the medical writings of Galen and Hippocrates. See Η. Biesterfeldt, 'Palladius οη the Hippocratic Aphorisms', in Libraries of the Neoplatonists (see note 2), 385-97, at 391-2. Α good example can be found ίη his Attainment of Happiness. See al-Fariibί, Tal:ψl al-sa 'άda (Hyderabad: 1346 Α . Η.), English version ίη Μ. Mahdi, trans., Alfarabi: Philosophy of Plato and Aristotle (Ithaca, Ν.Υ. : 1962). See for instance his Philosophy of Aristotle: al-Farabί, Falsafat Ari~tutάlis, ed. Μ. Mahdi (Beirut: 1961) . For an English version see again Mahdi, A/farabi: Philosophy of Plato and Aristotle (see previous note). See Ρ. Adamson, 'The Arabic Sea Battle' (see note 39), 167-72. Al-Fariibί further explains that the existence of 'possibility' (i.e., contingency) is a 'first principle; whose certainty cannot be denied. Again this suggests that the rejection of determinism would belong to metaphysics, which is similarly responsible for discussing-but of course not proving-the principle of non-contradiction (Metaphysics IV). All of this contrasts interestingly to Ibn Άdϊ's handling of the same topic. Already the title of his treatise οη the subject, On Establishing the Nature of the Possible (see note 39), shows that he is linking the argument of On Interpretation 9 to a more 'metaphysical' discussion. Ιη addition, this treatise argues for, rather than merely asserting, the reality of contingency. Αη exception was a work called The Fair Judgement, which contained notes οη Metaphysics ΧΠ as well as the pseudo-Aristotelian Theology. There is also a series of notes οη the De Anima: see D. Gutas, Avicenna and the Aristotelian Tradition (Leiden: 1988), 130-40.

II 662 54. See Α. Bertolacci, Ίhe Receptίon of Aristotle's Metaphysics ίn Aνicenna's Kitab al-Shifa'. Α Milestone of Western Metaphysical Ίhought (Leiden: 2006). 55. Α distinctive feature of the Healing is that the metaphysical section does not set out this argument as it is found ίη other texts, including Pointers and Reminders. Rather the explicit proof for God ίη the Healing is based οη more general causal regress arguments, inspired by Metaphysics Π; this is part and parcel of the Healing's avowed dependence οη the Aristotelian tradition. 56. Most of this commentary is not yet translated into English, but for the commentary οη Met. ΧΙΙ see C. Genequand, trans., Ibn Rushd's Metaphysics. Α Translation with Introductίon of Ibn Rushd's Commentary on Aristotle's Metaphysics, Book Lάm (Leiden: 1984). 57. See Ρ. Adamson, 'The Kindian Tradition' and Ύa]:ιya lbn Άdι and Averroes οη Metaphysics Alpha Elatton' (see notes 23, 48) . 58. Οη this text and its impact οη Avicenna, see D. Gutas, Avicenna and the Aristotelian Tradition (see note 53), 238-53. See further Α. Bertolacci, 'From Al-Kindι to Al-Farabί: Avicenna's Progressive Knowledge of Aristotle's Metaphysics According to His Autobiography', Arabic Sciences and Philosophy 11 (2001), 257-95. 59. See Arnzen, Ίbη Rushd οη the Structure of Aristotle's Metaphysics' (note 32 above) for a reconstruction of metaphysics according to Averroes, based especially οη the

Epitome. 60. The foregoing draws heavily οη Α. Bertolacci, Άvicenna and Averroes οη the Proof of God's Existence and the Subject-Matter of Metaphysics', Medioeνo 32 (2007), 61-97. 61. For a translation see C. Ε. Butterworth, Alfarabi. Ίhe Political Writings: Selected Aphorisms and Other Texts (Ithaca, Ν.Υ.: 2001). 62. For all this see Μ . Rashed, Ά New List of al-Farabi's Writings and the Author of the

Harmonization of the Opinions of the Two Sages Plato and Aristotle; Arabic Sciences and Philosophy 19 (2009 ), 43-82. For a study of the Harmony, including an Italian translation and a defence of its authenticity, see C. Martini Bonadeo, Al-Farabi. Larmonia delle opίnioni dei due sapienti il divino Platone e Aristotele (Pisa: 2008). 63. For the epitome there is a Spanish translation: S. Gόmez Nogales, trans., La Psicologίa de Averroes. Comentario al libro sobre el alma de Aristόteles (Madrid: 1987). For the exposition see Ι. lvry, trans., Averroes. Middle Commentary on Aristotle's De Anima (Provo: 2002). For the Long Commentary, preserved only ίη Latin, see R. C. Taylor, trans., Aνerroes. Long Commentary on the De Anima of Aristotle (New Haven: 2009 ). 64. There is a precedent ίη the Arabic tradition ίη al- Kindι's On the Intellect (see note 28). He refers to the entity as the 'first intellect: 65. Οη this topic ίη al-Farabι see his Letter on the Intellect, translated ίη McGinnis and Reisman, Classical Arabic Philosophy (see note 28); Ρ. Adamson, 'Knowledge of Universals and Particulars' (see note 26); D. L. Black, 'Knowledge (Ί/m) and Certainty (Yaqϊn) ίη al-Farabί's Epistemology; Arabic Sciences and Philosophy 16 (2006), 11-45; Τ.-Α. Druart, Άl-Farabi and Emanationisrn', ίη J. F. Wippel, ed., Studies in Medieval Philosophy (Washington DC: 1987), 23-43. For the topic ίη Avicenna see Ρ. Adamson, 'Non-Discursive Thought ίη Avicenna's Commentary οη the Ίheology of Aristotle', ίη ). McGinnis, ed., Interpretίng Avicenna: Science and Philosophy in Medieval Islam (Leiden: 2004), 87-lll; D. Gutas, Ίntuition and thinking: the Evolving Structure of Avicenna's Epistemology', and D. Ν . Hasse, Άvicenna οη Abstraction', both ίη R. Wisnovsky, ed., Aspects of Avicenna (Princeton: 2001), 1-38 and 39-72. More generally see also Η .Α. Davidson, Alfarabi, Avicenna and Averroes on Intellect (Oxford: 1992).

II ARISTOTLE

66.

ΙΝ ΤΗΕ

ARABIC COMMENTARY TRADITION

663

the criticisms made by Aquinas see his On the Unity of the Intellect, trans. R. Mclnerny (West Lafayette: 1993). For discussion of more favourable reactions to Averroes ίη the Christian tradition see D. Ν. Hasse, 'The Attraction of Averroism ίη the Renaissance: Vernia, Achillini, Prassicio', ίη Ρ. Adamson, Η . Baltussen, Μ . W. F. Stone, eds., Philosophy, Science and Exegesis in Greek, Arabic and Latin Commentaries, 2 vols (London: 2004), vol. 2, 131-147 (see note 5); id., Άrabic philosophy and Averroism; ίη J. Hankins, ed., The Cambridge Companion to Renaissance Philosophy (Cambridge: 2007), 113-36. 67. The literature οη Averroes' position is extensive. See for instance D. L. Black, 'Conjunction and the Identity ofKnower and Known ίη Averroes', American Catholic Philosophical Quarterly 73 (1999), 159-184; Α. Ivry, Άverroes οη Intellection and Conjunction', Journal of the American Oriental Society 86 (1986), 76-85; R. C. Taylor, Άverroes οη Psychology and the Principles of Metaphysics', Journal of the History of Philosophy 36 (1998), 507-523; id., Άverroes: Religious Dialectic and Aristotelian Philosophical Thought', ίη Adamson and Taylor, The Cambridge Companion to Arabic Philosophy, 180-200 (see note 45); and id., 'Separate Material Intellect ίη Averroes' Mature Philosophy', ίη Words, Texts and Concepts Cruising the Mediterranean Sea (Leuven: 2004), 289-309. 68. See R. Glasner, 'Levi ben Gershom and the Study oflbn Rushd ίη the Fourteenth Century', Jewish Quarterly Review 86 (1995), 51-90. 69. S. Harvey, 'The Greek Library of the Medieval Jewish Philosophers', ίη Libraries of the Neoplatonists (see note 2), 493-506, comments: 'The Jewish Aristotelians knew Aristotle very well, but their knowledge for the most part came from Averroes' commentaries' (504) . 70. Οη this corpus, essentially untouched by modern scholarship, see R. Wisnovsky, 'The Nature and Scope of Arabic Philosophical Commentary ίη Post-Classical (ca. 11001900 AD) Islamic Intellectual History: Some Preliminary Observations', ίη Adamson, Baltussen, and Stone, Philosophy, Science and Exegesis (see note 5), vol. 2, 149-191. 71. Wisnovsky's list (see previous note) itemises ηο fewer than 30 extant commentaries and super-commentaries οη the works of Avicenna, more than half of them devoted to the Οη

Pointers. 72. See D. Gutas, 'The Heritage of Avicenna: The Golden Age of Arabic Philosophy, 1000-ca. 1350', in J. Janssens and D. De Smet, eds., Avicenna and his heritage (Leuven: 2002), 81-97; G. Endress, 'Reading Avicenna in the Madrasa. Intellectual Genealogies and Chains of Transmission of Philosophy and the Sciences in the Islamic East; in J. Ε. Montgomery, ed., Arabic Theology, Arabic Philosophy. From the Many to the One: Essays in Celebration of Richard Μ. Frank (Leuven: 2006), 371-423; Α. Shihadeh, 'From al-Ghazalι to al-RazJ: 6th/12th Century Developments ίη Muslim Philosophical Theology', Arabic Sciences and Philosophy 15 (2005), 141-179. 73. See Α. Neuwirth,

Άbd al-Latif al-Baghdiίdίs Bearbeitung von Buch Lambda der aristotelischen Metaphysik (Weisbaden: 1976). Οη him see also D. Gutas, 'Philosophy ίη

the Twelfth Century: One View from Baghdad, or the Repudiation of al-Ghazalί; ίη Ρ. Adamson, ed., In the Age of Averroes: Arabic Phi/osophy in the Sixth!Twelfth Century (London: 2011), 9-26. 74. The exception that proves the rule is that commentaries were written on the pseudoAristotelian Theology of Arίstot/e: see S. Rizvi, '(Neo)Platonism Revived ίη the Light ofthe Imams: Qaqι Sa'Jd Qummϊ (d. ΑΗ 1107/Αο 1696) and his Reception ofthe

II

Ίheologia Aristotelis; ίn Ρ. Adamson, ed., Classical Arabic Philosophy: Sources and Reception (London: 2007), 176-207. 75. R. Arnzen, 'The Structure of Mulla ~adra's al-/:ιikma al-muta 'a/iya fι 1-asfar α/- 'aqliyya al-arba'a and his Concepts ofFirst Philosophy and Divine Science. Απ Essay; ίη Medioevo 32 (2007), 199-240. 76. Or so Ι have argued ίη Ρ. Adamson, Όη Knowledge of Particulars', Proceedings of the Aristotelian Society 105 (2005), 273-94. 77, Ι would like to thank Amos Bertolacci for his very helpful comments οη an early draft of this paper.

BIBLIOGRAPHY Ύai)ya Ibn Άdι and Averroes οη Metaphysics Alpha Elatton, Documenti e Studi sulla Tradizione Filosofica Medievale 21, 343-74. Akasoy, Anna Α., and Alexander Fidora (2005) Ίhe Arabic Version of the Nicomachean Ethics (Leiden: Brill), 49-52. Bertolacci, Amos (2006) Ίhe Reception of Aristotle's Metaphysics in Avicenna's Kitab al-Shifa'. Α Milestone of Western Metaphysical Ίhought (Leiden: Brill). Black, Deborah L. (1990) Logic and Aristotle's Rhetoric and Poetics in Medieval Arabic Philosophy (Leiden: Brill). Burnett, Charles Μ., ed. (1993) Glosses and Commentaries on Aristotelian Logical Texts: Ίhe Syriac, Arabic and Medieval Latin Traditions (London: Warburg Institute). Ferrari, Cleophea (2006) Die Kategorienkommentar von Abiί 1-Faraj Άbdallah Ibn al-Tayyib (Leiden: Brill). Genequand, Charles (1984) Ibn Rushd's Metaphysics. Α Translation with Introduction of Ibn Rushd's Commentary on Aristotle's Metaphysics, Book Lam (Leiden: Brill). Gutas, Dimitri (1988) Avicenna and the Aristotelian Tradition (Leiden: Brill). - - (1998) Greek Ίhought, Arabic Culture: Ίhe Graeco-Arabic Translation movement in Baghdad and early society (2nd-4th!Sth-1oth centuries) (London: Routledge). Gyekye, Kwame, ed. (1975) Ibn al-Tayyib's Commentary on Porphyry's Eisagoge (Beirut: Dar al-Mashriq). - - trans. (1979) Arabic Logic: Ibn al-Tayyib's Commentary on Porphyry's Eisagoge (Albany: SUNY Press). Hugonnard-Roche, Henri (1991) Tintermediaire syriaque dans la transmission de la philosophie grecque a l'arabe: le cas de l'Organon dΆristote', Arabic Sciences and Philosophy 1, 187-209. Lettinck, Paul (1994) Aristotle's Physics and Its Reception in the Arabic World (Leiden: Brill). McGinnis, Jon, and David C. Reisman, ed. and trans. (2007) C/assical Arabic Phi/osophy: An Anthology of Sources (Indianapolis, Ind.: Hackett). Peters, Francis Ε. (1968) Aristoteles Arabus (Leiden: Brill). Reisman, David C., and Amos Bertolacci (2009) 'Thabit Ibn Qurra's Concise Exposition of Aristotle's Metaphysics: Text, Translation and Commentary', ίη Sciences and Philosophy in 9th Century Baghdad. Ίhabit Ibn Qurra (826-901), ed. R. Rashed and Μ. Rashed (Berlin: de Gruyter), 715-76. Zimmermann, Fritz W (1981) Al-Farabi's Commentary and Short Treatise on Aristotle's De lnterpretatione (Oxford: Oxford University Press).

Adamson, Peter (2010)

III

ΤΗΕ

LAST PHILOSOPHERS OF LATE ANTIQUITY ΤΗΕ ARABIC TRADITION

ΙΝ

The classical texts of Greek philosophy reached the Islamic world through a double filter. One filter was the Syria tradition, which is discussed by Prof Hugonnard-Roche in the present volume. Α second filter was provided by late antique Greek authors immediately preceding, and in some cases contemporaneous with, the scholars who produced a philosophical literature in Syriac. 1 One could point ω a similar historical continuity in the case of Latin medieval philosophy, where late ancient authors like Boethius and Augustine exercised enormous influence for many centuries. But at least some of the differences between Arabic and Latin philosophy of, say, the 9'h- 10 th centuries CE, can be ascribed ω the very different fortunes of the late antique philosophical corpus in the Latin- and Arabic-speaking worlds. Whereas the output of the Greek commentators was largely unknown in Latin, it was ω an amazingly large extent retained in Arabic. This was thanks ω the translation movement of the Άbbasid era. Without going into great detail about the translation movement in general, 1 will begin this paper by recalling a few basic and well-known points about the reception of the commentators. 1 will then focus οη two topics which display how late antique philosophy was received in the Islamic world: the question of how logic 1 Consider that Sergius of Resh'ayna died closure of the Platonic school ίη Aτhens.

ίη

536 CE, only 7 years after the

III 2

relates to philosophy and the dispute over the world' s eternity. Ιη both cases, 1 will not only highlight the decisive inf1uence of late ancient philosophy οη authors writing in Arabic. 1 will also try to show how these authors reacted to contemporary cultural pressures by reshaping, extending, and departing from their Graeco-Arabic sources.

The Commentators in Arabic

The Commentaria in Aristotelem graeca (CAG), a monumental edition of the commentators produced in Germany over a century ago, was read rather sparingly until relatively recently. Α renewed interest in this corpus has represented one of the biggest shifts in the study of ancient philosophy over the past 30 years. Largely this has been instigated by the Ancient Commentators Project led by Richard Sorabji. As a result, now in 201 Ο we can say that we have just about caught up with, and perhaps even surpassed, the knowledge of late antique philosophy that was achieved in Baghdad in the 10 th century CE. Not coincidentally both endeavors involved a huge effort of translation, as well as thoughtful analysis of the commentators' ideas. If we consider the fruits of the Graeco-Arabic translation movement, we can see immediately how keen was the interest taken ίη Greek commentaries under the Άbbasids. 2 Medieval book lists, above all the invaluable Fihrist of Ibn al-Nadϊm, tel1 us of numerous translations of commentaries by Alexander of Aphrodisias, Themistius, Porphyry, and members of the Alexandrian 2 See τhe very useful τable provided by D. GUTAS, "Greek Philosophical Works Τ ranslated into Arabic", ίη The Cambridge History ofMedieval Philosophy, ed. by R. PASNAU, 2 vols (Cambridge 2010), vol. 2, 802-814. See also C . D'ANCONA, "Greek into Arabic. Neoplatonism ίη Translation", ίη The Cambridge Companion to Arabic Philosophy, ed. by Ρ. ADAMSON and R.C. TAYLOR (Cambridge 2005), 10-31, with a table at 22-23 focusing specifically οη Neoplatonists; and furτher Ιο., "Greek Sources ίη Arabic and Islamic Philosophy", ίη τhe online Stanjόrd Encyclopedia of Philosophy, ed. by Ε.Ν. ZALTA: hτφ://plato. stanford.edu/entries/arabic-islamic-greek.

III ΤΗΕ

LAST PHILOSOPHERS OF LATE ANTIQUITY

3

school like Ammonius, Simplicius, Philoponus and Olympiodorus. Α smaller number of commentaries and works by Greek commentators are preserved, wholly or partially, in their Arabic versions. Complerely preserved commentaries are rare; here a prominent example would be the paraphrase of the De Anima by Themistius. We also have him οη Metaphysics Book Lambda, as I will discuss later οη. More common are commentaries known partially or as fragments. For instance we have large parts of the Arabic version of Philoponus' commentary οη the Physics. Also extant are Arabic fragments of some of Alexander' s commenraries, for instance rhose οη rhe Physics, Metaphysics and On Generation and Corruption. Ιη some cases these supplement what can be known from the Greek rradition. There are surviving Arabic versions of independent treatises from some of rhe same authors, especially Alexander. ln addition we can, even ίη rhe absence of Arabic versions of the commentaries, observe the direct influence of commentators οη Arabic philosophy. The works of al-Kindϊ, the first Hellenizing philosopher to write in the Islamic world, already betray extensive influence from Greek commentators from the Alexandrian school, as well as from earlier authors like Alexander.3 But rhe influence of rhe Alexandrian commentators peaks in the lO th - 11 th centuries with the so called "Baghdad Peripatetics", a group of mostly Christian Aristotelian philosophers who produced rheir own commentaries and trearises based οη Arisωrle. They often imitate rhe very form of rhe 3 For a good example see S. FAZZ0 and Η. WΙESNER, 'Άlexander of Aphrodisias ίη the Κίηdι Circle and ίη al-Kindι's Cosmology", ίη ASPh 3 (1993), 119153. The influence οη al-Κindι from the Alexandrian tradition is clear from his use of Philoponus ίη treating the world's eternity (see below), and from his discussion of the nature of philosophy at the beginning of On First Philosophy. As detailed by Α. IVRY, Al-Kindi's Metaphysics (Albany 1974), 115-118, this passage and the related defιnitions of philosophy ίη al-Κindι's On the Definitions and Descriptions of Things extensively parallel authors like Ammonius, Elias and David.

III 4

commentaries, as well as reproducing the comments made by their Greek forebears - sometimes verbatim. Among the Baghdad Peripatetics, the author who follows the Alexandrians most closely is the last representative of the school, Αbϊι 1-Farag ibn al-Tayyib. We have extant commentaries from his pen οη Porphyry' s Isagoge and Aristotle' s Categories. 4 Other members of the school also engaged extensively with the Greek commentators. This includes not only the famous al-Farabϊ, but also his student Yabya ibn Άdϊ. 5 As we will see in a moment, the logical writings of these authors provide eloquent testimony of the impact of the commentary tradition. Another fιne example is the so-called "Baghdad Physics", which incorporates comments by numerous members of the Baghdad school, including Ibn Άdϊ, as well as excerpts from Alexander' s and Philoponus' commentary ίη Arabic translation. 6 Also worth noting is Ibn Άdϊ' s commentary οη Metaphysics Alpha Elatton, which seems to imitate the Greek commentaries in its theoria and lexis structure. 7 There are, then, a large number of texts displaying the impact of the Greek commentators οη philosophy ίη Arabic ίη 4 For rhe former see Κ. GΥΕΚΥΕ (ed.), Jbn al-Tayyib's Cornmentary on Porphyry's Eisagoge (Beirut 1975); trans. in Κ. GΥΕΚΥΕ, Arabic Logic. Jbn al-Tayyib's Commentary on Porphyry's Eisagoge (Albany 1979). For the latter, see C . FERRARJ, Der Κategorienkommentar von Αbϊι l-Farάg 'Abdallιih ibn at-Tayyib (Leiden 2006). 5 His works are edited in S. KHALIFAT, Yabya ibn Άdϊ, The Philosophical Treatises (Amman 1988). See furrher G. ENDRESS, The Works ο/ Yabya ibn 'Αdϊ. An Analytical Inventory (Wiesbaden 1977). 6 See the edition in Α. BADAWΊ, Aristϊιtάlis : al-Tabίa, 2 vols (al-Qahira 1964-1965). The commentary's contents are summarized ίη Ρ. LΕΠΙΝCΚ, Aristotle's Physics and its Reception in the Arabic World (Leiden 1994). See further several studies by Ε. GIANNAKIS : Philoponus in the Arabic Tradition ο/Aristotle's Physics, D. Phil. Thesis (Oxford 1992); Ιο . , "The Srructure of ΑbίΊ 1-l:Iusayn al-Ba~rΊ's Copy of Aristorle's Physics", in ZGAIW 8 (1993), 251-258; ID., "Fragments from Alexander's lost Commentary οη Aristotle's Physics", in ZGAIW 10 (1995-1996), 157-187. 7 Ρ. ADAMSON, 'Ύabya ibn ΆdΊ and Averroes on Metaphysics Alpha Elatton," in DSTFM 21 (2010), 343-374. See also C. ΜΑRΤΙΝΙ BONADEO, "Un commento ad alpha elatton 'sicut litterae sonant' nella Baghdad del Χ secolo", ίη Medioevo 28 (2003), 69-96.

III ΤΗΕ

LAST PHILOSOPHERS OF LATE ANTIQUITY

5

the 9'h- 11 ' 11 centuries CE. Το some extent, their influence wanes thereafter, as philosophical commentary tends to be directed towards Avicenna rather than Aristotle. Of course the revival of Aristotelianism ίη Andalusia, above all ίη the commentaries of Averroes, constίtutes a major exception. Averroes ίη fact preserves for us some of the fragments mentioned above, for authors like Alexander. But ίη what follows here I will be focusing οη the early period. Before delving ίηtο specific topics, Ι' d like to note three points of continuity between the Greek commentary tradition and the early Arabic tradition of Hellenizing philosophy (falsafa). These are not novel points, but well worth repeating. First, the Arabic tradition carries οη Greek attitudes towards the prospects of reconciling Aristotle with Platonism. This is not to say that all authors take a harmonizing attitude, but rather that a harmonizing attitude is the norm, yet admits of exceptions. Good examples of harmonization (not without nuance) can be found οη the Greek side ίη authors like Porphyry and Simplicius, and οη the Arabic side ίη al-Κindϊ and the author of the work On the Harmony of the Two Sages, whose ascription to al-Farabϊ has recently been a matter of dispute. 8 Other authors are less optimistic about the prospects for a consistent Platonic Aristotelianism. Here one might think of Alexander or Syrianus οη the Greek side, and al-Razϊ οη the Arabic side. Second, the Greek tradition already involves the passing of Platonism and Aristotelianism - more or less fused ίηtο a 8 For a skeptical view see Μ . RASHED, 'Όη the Auτhorship of the Treatise οη τhe Harmonization of the Opinions of τhe Two Sages Attributed to Al-Farabι", ίη ASPh 19 (2009), 43-82, following τhe lead of LAMEER, Al-Farιibi and Aristotelian Syllogistics. Greek Theory and Islamic Practice (Leiden 1994), 30-39. Οη the topic of harmonization see further C. D'ANCONA, "The Topic of the Ήar­ mony Between Plato and Aristotle'. Some Examples ίη Early Arabic Philosophy", ίη Wissen uber Grenzen. Arabisches Wissen und lateinisches Mittelalter, hrsg. νοη Α. SPEER (Berlin 2006) , 379-405; Α. BERTOLACCI, 'Όifferent Attitudes to Aristotle's Authority ίη the Arabic Medieval Commentaries οη the Metaphysics" , ίη AntPhilos 3 (2009) , 145-163.

J.

III 6

single tradition, as just mentioned - from pagans to Christians. The last philosophers of Greek antiquity were mostly pagans, but the very last philosophers of Greek antiquity were often Christians. Most prominent here is of course John Philoponus, but one thinks also of the associates of Olympiodorus, Elias and David - who are often among the closest models for the commentaries of Ibn al-Tayyib. Οη the Arabic side, it's already been mentioned that the Baghdad Aristotelians were almost all Christians, and the central role of Christians in the translation movement hardly needs emphasis. Indeed jάlsajά was an admirably ecumenical enterprise. Not only do we find devout Muslims like al-Kindϊ working together with the Christian translators and scholars who were so indispensable to the translation movement, but we find Christian authors like Ibn Άdϊ engaging politely with Jews in philosophical exchanges. 9 Third, authors writing ίη Arabic took over from late antique authors not only philosophical ideas, but a philosophical curriculum. Ιη the case of Porphyry' s Isagoge, a work by a Neoplaωnist was acωally added to the standard Aristotelian curriculum. 10 lt became such a standard part of philosophical education that al-Kindϊ uses it as a basis to refute the Trinity because, he says, it is well-known to his Christian opponents.11 More generally and more fundamentally, the late ancient ordering of Aristotle's works and of the philosophical sciences (which of course go hand-in-hand) penetrated deeply 9 See S. ΡΙ Ν ΕS, 'Ά T enth Cenωry Philosophical Correspondence", in PAAJR 23 (1954), 103-136. ιο Whether the Isagoge itself is a Neoplaτonic work is less clear; for an affirmative answer see R. CΗΙΑΙιΑDΟ ΝΝΑ, "What is Porphyry's Isagoge?", in DSTFM 19 (2008), 1-30. 1 1 See Α. PERIER, "Un traite de Yal:ιya ben Άdι. Defense du dogme de la Trinite contre les objections d 'al-Kindi", ίη ROC 3rd series, 22 (1920-1921), 3-21. AI-Κindϊ's arguments, without the response of Ibn Άdϊ, are translated ίη R. RASHED and J. Jοιι vΕΤ, CEuvres Philosophiques & Scientifiques d'al-Kindi. Volume 2, Metaphysique et cosmologie (Leiden 1998), and ίη Ρ. AoAMSON and Ρ . Ε . ΡΟΚΜΑΝΝ (trans.), Al-Kindϊ's Philosophical Works (Karachi 2011 , forthcoming).

III ΤΗΕ

LAST PHILOSOPHERS OF LATE ANTIQUITY

7

into Arabic philosophy. 12 This question of how philosophical disciplines relate to one another provides a common link between the two topics I will be discussing in the remainder of this paper.

Logic as an Instrument of Philosophy ln Greek commentaries οη the works of the Organon Porphyry's Isagoge plus Aristotle's 'logical' works - it became standard to discuss the question whether logic is a part or an instrument of philosophy. 13 The very term organon (instrument) shows which way the Aristotelians were inclined to see the issue. For them, logic was not strictly speaking a philosophical science, but rather a discipline or art (τέχνη) which contributes instrumentally to philosophy. They developed this view in opposition to the Stoics. For the Stoics, logic constituted one of the three parts of philosophy, alongside ethics and physics. Alexander of Aphrodisias objected to this, οη the grounds that logic has a different subject-matter (υποκείμενον) and goal (τέλος) from philosophy. 14 lts subjectmatter is "statements and propositions (άξιώματα χαί προτά­ σεις)", and its goal is "to prove that, when propositions are 12 See G. ENDRESS (ed.), Organizing Knowledge. Encyclopaedic Activities in the Pre-eighteenth Century Islamic Wor!d (Leiden 2006) ; D . GUTAS, "The Άlexan­ dria to Baghdad' Complex of Narratives. Α Conτribution to the Study of Philosophical and Medical Hisωriography among the Arabs", in DSTFM 10 (1999), 155-193 ; Ρ . AoAMSON, "The Κindian Tradition. The Structure of Philosophy in Arabic Neoplatonism", in Libraries of the Neoplatonists, ed. by C . D'ANCONA (Leiden 2007), 351-370. 13 For τhe issue a good place to start is R. Sοι~ΒJΙ (ed.), The Philosophy ofthe Commentators 200-600AD. Α Sourcebook, 3 vols (London 2004), vol. 3, §l(b). See further Κ. I ERODΙAΚONOU, "Aristotle's Logic: an Instrument, Not a Part of Philosophy", in Aristotle on Logic, Language and Science, ed. by Ν. AVGELIS and F. PEONIDIS (Thessaloniki 1998), 33-53 ; T .S. LEE, Die griechische Tradition der aristotelischen Syllogistik in der Spiitantike (Gottingen 1984), ch. 2; A.C. LLOYD, The Anatomy ofNeoplatonism (Oxford 1990), 17-21. 14 Alexander, In An. Pr. Ι, ed. Μ. WALLIES (Berolini 1883), 1.18-2.2.

III 8

compounded with one another in certain ways, something may be deduced by necessity from what is posited or conceded (ή γάρ ταύτης πρόθεσις το διά της ποιας των προτάσεων συνθέσεως έχ. των τιθεμένων τε χ.αί συγχωρουμένων έξ άνάγ­

χ.ης τι συναγόμενον δειχ.νύναι)". 15 Elsewhere in the same com-

mentary, Alexander develops the idea that logic studies argument jόrms, that is, syllogistic forms, to which terms stand as matter. This is why a certain syllogism can be represented schematically (indeed Alexander uses the word σχήμα), with letters instead of terms (e.g. "All Α is Β , all Β is C, therefore all Α is C"). As has been pointed out, Alexander here seems to be taking a step towards what we might call "formal" logic. 16 Better, one might instead say that he is acutely observing the significance of Aristotle' s own steps towards treating logic schematically and "formally". However, things are not so simple. One reason Alexander and other commentators disagree with the Stoic view ίs that for them logic is defined by its instrumental role, in that the study of logic (when done properly) ignores "useless" yet valid inferences. lt is not germane to point out that 'Ίf Α, then Α; Α; therefore Α" is a valid inference. Rather the serious logician is interested ίη argument forms that can be used to advance philosophical science. 17 This differentiates the Aristotelian outlook from the modern understanding of logic as the study of purely formal systems (since trivial inferences belong to the system just as much as "useful" inferences) . Το put it another way, the commentators seem to be interested more ίη soundness than validity. This becomes clear when they say, ίη a phrase that will reappear frequently in the Arabic tradition, that the role of 15 Translation from Alexander of Aphrodisias, On Aristotle Prior Analytics 1.1-7, aans. J. BARNES et al. (London 1991). 16 T.S. LEE, op. cit. (η. 13), 38-39, citing Alexander, in Pr. An. 53-54 for the form/matter idea and the importance of substituting letters for terms. 17 Thus T.S. LEE, op. cit. (η. 13), 49-50 speaks of Alexander envisioning a "working logic". Cf. Κ. IERODIAKONOU, op. cit. (η. 13), 36. She also points out (p. 38) that the Peripatetics underestimated the Sroics' own stress οη the utility of logic. See also A.C. LLOYD, op. cit. (η. 13), 18-19.

III ΤΗΕ

LAST PHILOSOPHERS OF LATE ANTIQUITY

9

logic is to "distinguish the true from the false and the good from the evil". Actually, if we look at the formulation ίη a preserved excerpt from Elias οη the Prior Analytics, we see that he is a bit more nuanced than this. What he says ίs that "philosophy uses logic (χέχρηται τ;~ λογιχiJ) to show, ίη the theoretical domain, what is true and what false , and ίη the practical domain what is good and what is bad". 18 Note that logic does not establish the true, false, good and bad. Rather ίt is used to establish these things. This is an important qualification, because the commentators need to hold οη to the idea that logic is merely instrumental, even if ίt is an indispensable instrument. The goal is to devise arguments which establish truth. And logic is merely necessary, not sufficient, for the grasp of truth. As Katerina Ierodiakonou has argued, the commentators would have a principled reason for insisting οη this point. Logic studies the expression (φωνή) , not the thing itself (πραγμα). But each philosophical science has some range of objects - real things out ίη the world which it studies. For instance, physics studies things subject to motion and rest. Since logic deals with the words which refer to things rather than the things referred to, logic is not a proper philosophical science. 19 This is, of course, consistent with its merely necessary and non-sufficient character - logic as such is pre-philosophical, precisely because it does not by itself establish truth. Even commentators with more ambitious views of logic acknowledge this. Here Ι am thinking particularly of Ammonius, who endorses what he identifies as the Platonic valorization of logic as both part and instrument of philosophy. This is because he wants to bring Aristotelian logic ίηtο close relation with Platonic dialectic (as described ίη the middle books of the Republic, for instance), which is clearly much more than an 18 Ed. by L.G. WESTERINK, 'Έlias οη rhe Prior Analytics", (1961) , 126-139: 134.23-24. 19 κ. IERODIAΚONOU, op. cit. (η. 13), 46.

ίη

Mnemosyne 14

III 10

instrument. 20 But Ammonius is able to distinguish "mere" logic from truth-yielding dialectic by using Alexander' s idea of syllogistic form which needs to be fιlled in by matter. Logic is a mere instrument because the form is "empty", but once the arguments (λόγοι) are "taken together with real things" logic becomes a part of philosophy.21 Ammonius might as well have said that, once applied to the real things, logic is simply identical to philosophy, insofar as philosophy consists of demonstrations with a logical form. Ιη any case, the 'Ώon-scientifιc" status of logic did not prevent it from being every student' s introduction to philosophy in the late ancient teaching curriculum. And for good reason: as we've seen, logic is merely instrumental, but it is an indispensable or necessary instrument. One can ηο more do philosophy without logic than one can do carpentry without tools. This attitude passed into the Arabic tradition along with the textual tradition of commentary οη the Organon. As has been noted by numerous scholars - notably Gerhard Endress in his study of the standing of logic in Arabic culture - numerous Arabic texts repeat, even verbatim, the commentators' remarks about logic' s instrumentaliry. 22 As usual Ibn al-Tayyib adheres closely to the Alexandrian commentators here, remarking for instance that logic is "the instrument for philosophy (add li-l-falsafa); without the instrument, the agent can do nothing." 23 Various members of the Baghdad school also classify logic as an art in terms of its subject-matter and goal, and they reproduce Greek 20 See Ρ . ΗΑοοτ, "La logique, partie ou instrument de la philosophie?", ίη Simplicius. Commentaire sur les Catιfgories, dir. Ι. H ADOT, fasc. 1: Introduction, (Leiden 1990), 183-188, who connects Ammonius' view to the treatment of dialectic ίη Plotinus, Enneads 1.3. 21 ΑΜΜΟΝ., Jn An. Pr. Ι, ed. Μ. W ALLI ES (Berolini 1889), 10.38-11.3. For him the Plaronic view is the reasonable middle ground between the extreme positions of the Sroics and Peripatetics. Cf. also T.S. Lεε, op. cit. (η. 13), 40. 22 See G . ENDRESS, "Grammatik und Logik. Arabische Philologie und griechische Philosophie im Widerstreit", ίη Sprachphilosophie in Antike und Mittelalter, hrsg. νοη Β . MOJSISCH (Amsterdam 1986), 163-299. For a useful overview of the relevant sources see also C . ΗΕΙ Ν , Definition und Einteilung der Philosophie (Frankfurt a.M . 1985), 153-162. 23 in. Cat., ed. by C. FειυιΑιυ, op. cit. (η . 4) , 10.25 .

III ΤΗΕ

LAST PHILOSOPHERS OF LATE ANTIQUITY

11

ideas about this, saying for instance that its subject-matter is expressions (alfd;;,). 24 The Arabic commentators, however, are in a rather different dialectical situation than the one faced by their Greek predecessors. The latter were opposing a Stoic (and Platonic) tradition which gave too much weight to logic, by making it a full-blown part of philosophy. The former, by contrast, are defending the merits of logic against detractors who argue that it is worthless. ln particular, they are confronted by the rival claim of grammar to be the fully adequate study of expressions. 25 So famous as hardly to need mention is the debate before the vizier Ibn alFurat, between the grammarian al-Sϊrafϊ and the father figure of the Baghdad school, the Christian Abu Bisr Matta. This event, and related criticisms, provoked several responses from members of the Baghdad school. Al-Farabϊ thematizes the relationship between logic and grammar in his Enumeration of the Sciences Ub1a' al-'uliim), and Ibn Άdϊ wrote a treatise On the Diffirence between Logic and Grammar which expounds the subject-matter and goal criteria for demarcating each art. 26 24 As G. ENDRESS, art. cit. (η. 22), 207, points our, laft. renders φωνή ίη IsJ:ιaq b. J::lunayn's Arabic translarion of On Interpretation. Al-Κindι shows less awareness of rhe Greek trearment of rhis issue. Ιη his On the Quantity ο/ Aristotle 's Books, he unblinkingly makes logic one of four broad areas of rhe Arisrorelian corpus, without mentioning its merely instrumenral srarus. He does, however, implicirly set logic apart ίη much rhe way suggesred by Κ. IERODIAKONOU, ίη rhar he names types of entities srudied by physics, intermediare science (which he here idenrifιes as psychology, elsewhere as marhemarics) , and metaphysics. Logic has ηο type of entity assigned ro it. See Μ. Gιποι e R. W ALZER, Uno scritto introduttivo allo studio di Aristotele (Roma 1940), §ΙΙ . 2 . 25 Οη rhe grammar vs. logic debare see G. ENDKESS, art. cit. (η. 22); Ιο., "La conrroverse enrre la logique philosophique er la grammaire arabe au remps des khalifs", ίη JHAS 1 (1977), 339-351; Α. ELAMMNI-]AMAL, Logique aristotelicienne et grammaire arabe: etude et documents (Paris 1983). The accounr of rhe debare from al-Tawbιdϊ is edired and rranslared inro English ίη D. S. MAKGOULIATH, "The Discussion Berween Abu Bishr Marra and Abu Sa'id al-Sirafι οη rhe Merits ofLogic and Grammar", in]RAS (1905), 79-129. 26 Ed. by S. KHALIFAT, op. cit. (η . 5), 414-24. Trans. ίη Α. ELAMMNI-JAMAL: "Grammaire er logique d'apres le philosophe arabe chrerien YaJ:ιya b. Άdϊ (280364 Η/893-974)", ίη Arabica 29 (1982), 1-15. G. ENDRESS, art. cit. (η. 22) provides annorated translarion of rhe debate as recounred by al-TawJ:ιϊdϊ and Ibn Άdϊ's rreatise, at p. 238-296.

III 12

Because the aim of these philosophers is ω extol the importance of logic, they understandably give a rather different impression of logic' s status than the Greek commentators, who focus οη its mere instrumentality. Admittedly, the members of the school do faithfully repeat that logic is an instrument. Αbϊι Bisr, for instance, is quoted in the report of the debate as saying that logic is "one of the instruments (ιildt) by which one knows correct from faulty speech, and unsound from sound concept (ma"nιi) - like a balance (mϊzιin), with which I may know the more from the less".27 But formulations like this might easily leave one with the impression that logic sorts out the true from the false οη its own. This can be seen from Ibn al-Furat's purported characterization of Αbϊι Bisr' s view when introducing the debate. According ω the vizier, Αbϊι Bisr claims that "there is ηο way to knowledge of the true and false (la sabϊl ilιi ma"rifat al-baqq wa-l-batil), the right and wrong, or the good and bad, apart from logic". This is the sentiment we found in Elias' commentary οη the Prior Analytics, but stripped of its nuance: now logic is the way to know truth, goodness and so οη, rather than something that is used by philosophy to know these same things. Of course we can hardly rely οη Ibn al-Furat (or those who are reporting the debate ω us) ω capture Αbϊι Bisr's view accurately. But an almost identical statement is given by Ibn Άdϊ' s student Ibn Zur'a in his own defense of logic: "it is clear and obvious, ω anyone who knows about logic or follows what its adherents have said, that logic is an art whose goal (garar/) comprises the sorting out (tablϊ1) of true from false in speech, and the discrimination (tamyϊz) of good from evil ίη action". 28 This is in at least superficial disagreement with Ibn al-Tayyib, who 27 D.S. MARGOULIATH, art. cit. (η . 25), 93. 28 Ν. R.ESCHER, 'Ά Tenth-Century Arab-Christian

Apologia for Logic", ίη

Islamic Studies 2 (1963), 167a9-l l. Ibn Zur'a is arguing against unnamed opponents who accuse logic of undermining religion rather than of being superfluous. He takes the rather surprising line that logic instead lends an important support to religion, because ίt distinguishes the possible from the impossible. This allows us to define a miracle as that which is (naturally) impossible; hence without logic there can be ηο concept of miracles!

III ΤΗΕ

LAST PHILOSOPHERS OF LATE ANTIQUITY

13

following a different strand of the Greek tradition (found ίη Ammonius, Elias and others), says rhat rhe end (gaya) of logic is demonstration (burhdn). 29 Ibn Άdϊ's On the Dijference between Logic and Grammar also makes demonstration the defιnitive end of logic. Logic's subject-matter is "expressions insofar as they refer ω universal things (al-umur al-kulliyya)", 30 and the reason the things ίη question must be universal is that the goal is demonstrarion. For demonstrarion concerns irself only with universals, not particulars. 31 Α demonstration is, more precisely, a composition of universal referring expressions into a syllogism which is (necessarily) ίη accord with the way things really are. Ibn Άdϊ rhus goes οη ω give rhe following overall defιnirion of logic: "rhe art which is concerned wirh expressions which refer ω universal things for the sake of composing [those expressions] ίη a way that agrees with the things ω which rhey refer".32 Again, one could be forgiven for rhinking rhat someone who has perfecrly mastered logic can dispense with the rest of philosophy. If I am already ίη possession of syllogisms rhat confer universal knowledge of how things are, what else remains ω be done? Alternatively, ω put it ίη terms of the formulation used by Ibn al-Furat and Ibn Zur'a, if logic tells me the difference between true and false, good and evil, isn't it a sufficient condition for philosophical wisdom, rather than a mere necessary instrument? As I say, ίt ίs natural rhat rhe philosophers might give this impression, given their need ω stress the value of logic. But ίη fact, the defense of logic requires them only ω show that ίt is a necessary and not suffιcient means of reaching truth. Whereas rhe Greek commentators needed to emphasize, against the Stoics, that logic is only an instrument, the Baghdad school Κ. GΥΕΚΥΕ, op. cit. (η. 4), §52. S. KHALIFAT, op. cit. (η. 5), 422. Of course Ibn ΆdΙ is thinking here of Porphyry's understanding of the Categories as studying word.s insofar as they signify things. 31 Οη this see Ρ . AoAMSON, "Κnowledge of Universals and Particulars ίη the Baghdad School", ίη DSTFM 18 (2007), 141-164. 29

30

32

s.

ΚHALIFAT,

op. cit.

(η .

5) , 423.

III 14

needs to emphasize, against al-Sϊrafϊ and his ilk, that it is an instrument one cannot do without. Thus our philosophers owe us an account of why logic is a necessary tool for reaching the goals they mention - discerning truth from falsehood and good from evil, producing demonstrations - without being by itself capable of reaching those goals. Το my knowledge the first adequate answer to the question in the Arabic tradition is suggested by al-Farabϊ, and further taken up by Ibn Άdϊ. For al-Farabϊ the key text is found in his Enumeration of the Sciences: 'Άmong the objects of the intellect, there are some things about which the intellecr cannot err at all. These are the things man perceives by himself as if he were naturally endowed with knowledge of them and certainty regarding them - for example that the whole is greater than the part, and that every three is an odd number. About other things it is possible to err, and to deviate from truth to untruth. These are the rhings which are such as to be grasped with ratiocination (fikr) and consideration (ta'ammub, by argument and proof. So regarding these, but not regarding the things [known immediately], the man who seeks to arrive at cerrain truth about what he is inquiring ίηtο needs the canons (qawanϊn) of logic."33

Al-Farabϊ unfortunately omits to explain further, proceeding instead to a comparison between logic and grammar which is rather unflattering to grammar (logic deals with objects of the intellect, grammar with linguistic expressions). But for greater illumination, we can turn to Ibn Άdϊ - not his study of logic in relation to grammar, but the more rarely studied On the Four Scientific Questions Regarding the Art of Logic. 34 This

33 Al-Farabι, Catάlogo de las ciencias, ed. y trad. por Α. GONZALEZ PALENCIA (Madrid 1953), 22.5-14 ίη the Arabic text. 34 Μ. ΤϋRΚΕR, 'Ύa]:ιya ibn Άdι νe Nesredilmemis, bir Risalesi", ίη Ankara ϋniversitesi Dil ve Tarih-Cografya Fakultesi dergisi 14 (1956), 87-102, Arabic edition at 98-102. Trans. ίη Ν. RεSCHER and F. SHEHADI, 'ΎaJ:ιya Ibn Άdι's Treatise Όη the Four Scientifιc Questions Regarding the Art of Logic"', ίη ]ΗΙ 25 (1964) , 572-578. Ι quote from the Rescher and Shehadi translation, with some modifιcations, giving the page and line number from the Tίirker edition.

III ΤΗΕ

LAST PHILOSOPHERS OF LATE ANTIQUITY

15

little treatise implicitly raises the issue of logic's necessity (that is, instrumentality) as opposed to its suffiency, and gives a persuasive account of why it is merely necessary. Ibn Άdϊ defines logic in now-familiar terms as "an instrumental art by which one discriminates between truth and falsehood in theoretical science, and between good and evil in practical science" (98.19-20). He then explains each term in the definition, one by one, in a manner reminiscent of his somewhat pedantic approach to commenting οη lemmata in Aristotelian texts. His explanation of the term "instrument (ιila)" is that it is something intermediary (mutawassita) between the artisan and his subject (99.1-2). This is reprised in the next paragraph, in which Ibn Άdϊ stresses the absolute need for logic (99.11-13): "The good obtained rhrough [logic] and apprehended by rhe intermediary [of logic] (bi-tawassu(iha) is beyond any parallel, since it [sc. rhis good] is complete happiness. There is ηο happiness more complete for rheory rhan belief in rhe truth, and it is rhrough it [sc. logic] rhat rhis is apprehended; and in action ηο [happiness] more complete than acquiring the good, without which ίt cannot be possessed."

As with other passages we've examined, this could give the impression that logic is all one needs in life, whether in theory or in practice. But a more careful reading shows that Ibn Άdϊ is careful to describe it as an instrument and intermediary to the end of happiness. Again, the question is how exactly it serves to bring us to truth and the good, without doing all the work itself. His answer ίs given shortly thereafter (99.14-100.7), and has a clear relationship to what we have seen in al-Farabϊ. The relevant passage is too long to quote in its entirety, so Ι summarize : whatever ίs known (ma'ru.fi ίs known either with ηο need for proof, because it is self-evident, or known by means of proof. Things known without proof are either sensible forms (which may be essential or accidenral) , or immaterial and grasped directly by the intellect. Of the latter, there are simple things known by stipulation (warf,; and definition, and there

III 16

are composed things we know as "immediate premises (muqaddimdt gayr rjawdt awJdt)" (99.25) . As for that which is known by proof, knowledge "is obtained by resorting to [logic] from a knowledge of things other than it, with a need for prior, antecedent knowledge in making it known. This type of knowledge-acquisition is called proof, argument, and demonstration" (99.27-9) . He illustrates the point by referring to the way mathematicians derive previously unknown essential truths from the properties of things like lines and numbers. Ibn Άdϊ' s explanation of logic' s instrumental role must be understood in the context of the foundationalist epistemology of Aristotle' s Posterior Analytics. As explained there, there are two kinds of knowledge involved in science. First, there are the things we know by proving them - Aristotle calls these things objects of "demonstrative knowledge (έπιστήμη)" (ΙΙ.19, 99b15-17). But we cannot demonstrate everything, because this would lead to an infinite regress (see Post. An. Ι.3, 72b18-25). Thus there must be some things known immediately - in Post. An. Π . 19 the state of knowing such things is not called έπιστήμη but rather νους. 35 Ibn Άdϊ, making more explicit what is surely already implicit in the passage from al-Farabϊ's Enumeration, says that logic enables us to extend our immediate knowledge so as to produce demonstrations. Ι take it that he means something like this : Ι have immediate knowledge that man is animal. Ι also know immediately that every animal is mortal. 36 Ι then use a logical 35 Οη this chapter see Ρ . ADAMSO N, "Posterior Analytics ΙΙ . 19 : a Dialogue with Plato?", ίη Aristotle and the Stoics Reading Plato, ed. by V. HARTE, Μ . Μ . McCABE, R.W . SHAΚI'LES and Α. SHEI'l'ARD (London 2010), 1-19. 36 Of course there is a puzzle about how such things could be known "immediately", unlike al-Farabϊ's examples ("the whole is greater than the part"). But Ibn Άdϊ has made a place for principles like this by mentioning items of knowledge grasped directly through sensation, and in this he is true to the account ίη Post. An. ΙΙ.19. Note that "immediately" should not be taken to mean something like "instanιly" or "from the beginning", as al-Farabϊ suggests ίη ιhe case of fundamental rules of reasons. Rather, ίt means wiιhout any antecedent premises. This is consistent with ιhe idea that a fιrst principle might be grasped only after a lengthy process of induction based οη sensation.

III ΤΗΕ

LAST PHILOSOPHERS OF LATE ANTIQUITY

17

scheme (All Α is Β; All Β is C; Therefore All Α is C) to infer that man is mortal. Without the logical scheme I would be unable to extend my knowledge any further than what I know immediately. Recall that Ibn Άdϊ has defιned "instrument" as an intermediary, and now it ίs clear why: logic is instrumental because it is the intermediary used to arrive at mediated knowledge. Notice that οη this account, we can in fact have certain knowledge of many things without using logic. (So Ibn al-Tayyib is exaggerating when he says, as we saw him saying above, that the philosopher can do nothing without logic unless he holds, rather implausibly, that immediate knowledge is not part of "philosophy". 37) The indispensability of logic does not consist in its supplying us with the principles of knowledge, but with completing our knowledge and thus conferring total happiness, that is, the fulfιllment of our rational capacity. Obviously this raises further questions. For instance, we might be willing to admit that logic, when added to immediate knowledge of principles, could in theory lead us to theoretical perfection. But to say the same about practical perfection would imply a highly rationalist account of ethics, in which we reach the practical good simply by reasoning correctly from fιrst practical principles. 38 Leaving this aside, it is worth asking whether Ibn Άdϊ and like-minded philosophers genuinely adhere to a purely instrumental vision of logic, in which it does nothing but to extend 37 Incidentally it is worth noting an echo of Ibn Άdι's account ίη Ibn al-Tayyib. Ιη the Categories commentary (ed. by C. FERΙ~RΙ, op. cit. (η . 4), 18.21ff), he contrasts things grasped without error and immediately by sensation, as well as things grasped as principles by the intellect, with those that are grasped only by prior knowledge. His examples of the former are more like alFarabι' s: that equal things are equal ω the same thing and that the whole is greater than the part. But since other things do stand ίη need of proof, "this logical art is intended precisely to give us a way (tariq) and method by which we may adequately reach hidden things by means of evident things : namely demonstration" (19.15-17). 38 Indeed this sort of view is put forth by al-Farabι in other contexts, as Ι have pointed out at Ρ. ADAMSON, art. cit. (η. 31), 149.

III 18

knowledge from what is grasped immediately. 1 think this depends οη what we understand as falling under "logic". The logic Ibn Άdϊ has in mind is basically syllogistic, so that his account applies primarily to logic as we fιnd it ίη the Prior Analytics. lt could certainly be adapted to explain the need for texts like the Topics and Sophistical Rejutations as well, since these help us avoid invalid, and valid but non-demonstrative, inferences. But the Organon touches οη many themes other than validity and the criteria required for demonstration. The Categories commits Aristotle to a division, at least, of words into 10 types - and for the post-Porphyrian tradition, this division has ontological signifιcance insofar as the Categories studies "words as they refer to things". Even restricting ourselves to the criteria for demonstration, consider a claim Ibn Άdϊ highlights ίη On the Difference between Logic and Grammar: that demonstrative knowledge is always of universals. This notion ίs put to use ίη such robustly non-logical contexts as Avicenna's proof of the immateriality of soul and his discussion of God' s knowledge of particulars. 39 Ιη short, Aristotle' s organon may include some metaphysics and certainly includes what we would call epistemology. So even if Aristotle' s Greek and Arabic interpreters manage to show that logic is an instrument, not a part, of philosophy, they have little hope of showing that the Organon is instrumental for, rather than part of, Aristotelian philosophy.

Physics or Theology? Arguments for and against the World's Eterni'ty

Consider the following two ancient Greek arguments for the eternity of the world: 39 Ι have discussed these issues ίη Ρ. ADAMSON, "Correcting Plotinus: Soul's Relationship to Body ίη Avicenna's Commentary οη the Theology ofAristotle", ίη Philosophy, Science and Exegesis in Greek, Arabic and Latin Commentaries, ed. by Ρ . ADAMSON, Η. BALTUSSEN and M .W.F. STONE, 2 vols (London 2004), vol. 2, 59-75; Ιο., 'Όη Knowledge of Particulars", ίη PAS 105 (2005), 273-294.

III ΤΗΕ

LAST PHILOSOPHERS OF LATE ANTIQUITY

19

Assume there is a fιrst motion. Then what is potentially moved either (a) comes ίηtο existence at some moment, or (b) is potentially moved for an eternity before actually moving. Ιη either case, a prior motion is required: ίη the case of (a), to make the movable exist, or ίη (b) to actualize the movable's potentiality for motion. Thus the supposedly fιrst motion is not after all fιrst, which shows that the assumption of a fιrst motion is incoherent. Similarly, motion cannot end, because whatever destroyed the last mover would itself need to be destroyed. The maker of the cosmos is eternally generous, and creates through his generosity; therefore the cosmos that he creates is eternal. Furthermore, if he went from not creating to creating or vice-versa, he would change, but the maker of the cosmos is immutable.

The first summarizes Aristotle' s argument ίη Physics VIII.1 (251a9-b10; 251b28-252a6), the second Proclus' opening argument ίη his On the Eternity of the World. 40 Both arguments attempt to prove the same conclusion, namely that the physical cosmos is eternal ex parte ante and ex parte post. Despite this they are importantly different ίη strategy. Aristotle' s argument is presented alongside other proofs of the world' s eternity given ίη Physics VIII. For instance, he argues that there cannot be a first moment of time, because this is incoherent (251bl0-251b28), and that since heavenly motion is circular, it can be beginningless and endless, having ηο contrary (264b9-265al2; cf. De caelo 1.3). These arguments all turn οη Arisωtle's conception of motion: either motion ίη general, or time which measures motion, or the motion of the 40 H .S. LANG and A.D. MACK0 (ed. and trans.), Proclus, On the Eternity of the World (Berkeley 2001). The first argument is preserved only ίη Arabic, and is translated ίη the volume by J. McGΙNNIS. For the Arabic text see also Α. BADAWI (ed.), Neoplatonici apud arabes (Cahirae 1955), 34.4-35.8. For a French translation see G.C. ANAWATI, "Un fragment perdu du De Aeternitate Mundi de Proclus", ίη Melanges de Philosophie Grecque offerts ά Mgr. Dies (Paris 1956), 23-25. For a German translation by Ρ. ΗΕΙΝΕ see Μ. BALTES, Die Weltentstehung des platonischen Timaios nach den antiken Interpreten , volume ΙΙ (Leiden 1978), 134-136. The extant Greek portions of Proclus' text may be found ίη John Philoponus, De aeternitate mundi contra Proclum, ed. Η. RABE (Lipsiae 1899).

III 20 heavens. This is ηο accident, for physics is the study of motion. These are, then, properly physical arguments for the eternity of the world. While Aristotle believes that the eternal motion of the cosmos does require an explanatory principle that ίs itself unmoving, a "fιrst mover", this principle is not fιrst in the order of discovery. What we grasp fιrst is the nature of motion; then we grasp that motion must be eternal; and only then do we argue from the eternity of motion to an eternal cause of that motion. Proclus uses a very different strategy, proving the eternity of the cosmos by appealing to the eternity and generosity of its maker, which he simply assumes (he is of course thinking of the Demiurge in Plato's Timaeus). Α number of Proclus' arguments in On the Eterni'ty ofthe World have this structure, which as we will see is representative of Proclus' way of understanding the world' s eternity. For him the world is not eternal in its own right, so to speak, but rather because it ίs the effect of an eternally active cause. Thus the most appropriate way to understand or demonstrate the world' s eternity would be by reasoning from the nature of its cause. We could call this a theological approach. The purpose of this second section of my paper is to trace these two contrasting approaches to the question of the world's eternity through some authors of late Greek and early Arabic philosophy. The issue is whether the world' s eternity is properly answered by physics or by metaphysics. Here "metaphysics" is understood to mean the study of the ultimate causes of things. This terminology is perhaps anachronistic for some of the Greek authors discussed below (and certainly for Aristotle, who of course does not use the word or title "metaphysics"). But it captures the way that early Arabic philosophical works tend to treat physics and metaphysics. For instance Αbϊι Sulayman al-Sigistanϊ, a student of Ibn Άdϊ, says : concerning rhe conjunction of effects wirh causes has two aspects: the fιrst, insofar as ίt ascends through their connections to their cause; the second, insofar as the power of the cause 'Ίnquiry

III ΤΗΕ

LAST PHILOSOPHERS OF LATE ANTIQUITY

21

pervades its effects. lnquiry ίη the fιrst mode belongs to the physicist; ίη the second, to the science of metaphysics." 41

Applying this contrast ω the problem of the world's eternity, the "physical" approach means arguing for or against eternity from what we actually observe about the world, especially about motion and things that are ίη motion, since for Aristotle this is the subject-matter of physics. We may then, following Arisωtle, use the eternity of the world and its motion ω argue that the world must have an eternal moving cause. The "metaphysical" or "theological" approach, by contrast, answers the same question by appealing to necessaty truths governing the causation excercised by world's principle(s). Το understand the Peripatetic treatment of these problems one can do ηο better than to turn once again ω Alexander. As ίt happens one of the most important texts for establishing his thought οη the eternity of the world is not preserved ίη Greek, but only ίη Syriac and Arabic translations: this is the text known as On the Principles of the Cosmos. 42 After an introductory section, the Principles begins by explaining (§4) that ''natural" bodies are bodies that have principles of motion. lndeed an internal (fi rjdtihi) principle of motion is what we mean by "nature" (tabϊ'a). The heavens have such a principle, and are thus natural bodies. Furthermore, as Alexander remarks, discussions elsewhere have shown that the heavenly bodies are "divine, ungenerated and imperishable" (§4). Thus at the very outset of the work, Alexander has indicated his adherence ω what I have called the physical approach ω 41 The passage appears ίη On the First Mover, τranslation from J.L. KRAEMER, Philosophy in the Renaissance of Islam. Abu Sulayman al-Sijistani and his Circle (Leiden 1986), 291. 42 C. GENEQUAND (ED.), Alexander of Aphrodisias on the Cosmos (Leiden 2001). Citations are to Genequand's section numbers. See further G. ENDΚESS, 'Άlexander Arabus οη the First Cause: Aristotle's First Mover ίη an Arabic Treatise attributed to Alexander of Aphrodisias", ίη Aristotele e Alessandro di Afrodisia nella tradizione araba, a cura di C. D'ANCONA e G. SERRA (Padova 2002) , 19-74.

III 22

the question of the world' s eternity. The heavens are natural, despite their divinity, and their nature ensures their eternity (see further §46, §57). Other arguments for eternity in the Principles are taken from the Physics. Alexander uses Aristotle' s argument against the possibility of a first motion (§66-69), and also rehearses the argument that there cannot be a first instant of time (§70-71), since any instant has time both before and after ίt - an allusion to the Aristotelian doctrine that the "now" or instant is not a part of time, but is without extension, a division between past and future. 43 Recognizing the eternity of motion allows us to establish the eternity of an unmoved mover. Alexander says both οη his own behalf and οη that of Aristotle that we know the first cause is eternal because we know its effects are eternal, rather than vice-versa (§49, §89; compare Aristotle, Physics 259a6-7). Another passage in the Principles likewise reveals Alexander' s commitment to the physical approach. He has just pointed out that sublunar bodies constantly change into one another, and are thus generable and perishable. He then continues: "This kind of perishability existing ίη the universe is not something happening to ίt by the will and resolution of some other being, 1 mean by that the divine things, but it is something inherent ίη its proper nature. For ίt does not fit the divine nature to will that which is not possible, just as ίt is not possible either, according to the opinion of those who profess the doctrine of creation, that perishability should attach to what has not been generated at all." 44

Even though the divine causes (the first cause and the heavens) do bring about change in the sublunar world, the proper nature of sublunar bodies is in a sense independent of those causes. What is possible for the sublunar bodies ίs determined by their nature, and the same is true for the heavens: since they 43 See also R.W. SHARPLES, 'Άlexander of Aphrodisias, On Time", ίη Phronesis 27 (1982), 58-81. This ηοtίοη of the instant is affirmed ίη the passage translated as §11-13, at p. 62-63. 44 §139-140, translation from C. GENEQUAND, op. cit. (η. 42) .

III ΤΗΕ

LAST PHILOSOPHERS OF LATE ANTIQUITY

23

are not generable, they must also be incorruptible.45 Alexander believes we can come to know the corruptibility of sublunar bodies and the eternity of the heavens (and thus of the cosmos as a whole) by studying their natures, that is, by understanding the intrinsic principles of their motions. This ίs the case even though, as he says elsewhere, the eternal motion of the heavens has the prime mover as an extrinsic fιnal cause (§52). This should be compared to one of Alexander' s Quaestiones (1.18), where Alexander argues "that it ίs not possible for the world to be incorruptible through the will of God, if it is corruptible by its own nature". 46 This quaestio has attracted attention for its discussion of modal notions : Alexander holds that ίt ίs impossible for S to be Ρ when S ίs prevented from being Ρ, or when S has ηο natural disposition towards being Ρ. 47 Ιη light of this, Alexander is able to refute the Platonist position that "according to its own nature", the world ίs disposed towards corruption and not eternity, but that the world nevertheless possesses an eternal existence given to ίt extrinsically by God (as we will see below, this is roughly the position that will be taken up by Proclus) . Employing his notion of impossibility, Alexander argues (31.25-32.3) that if the world has ηο innate disposition towards eternity, it cannot possibly be eternal. Even God cannot make such a thing eternal, since, as he says, "what ίs impossible ίη this way, since it ίs impossible for all, is impossible even for the gods" (32.3-4). The Platonists' position is, he might say, like holding that water has ηο innate disposition to be dry, but could be made dry by the gods. Alexander' s own view ίs of course that the Platonists are exactly wrong: the nature of the world is such that it has only a dispo45 For rhis commonly held principle ίη Greek thoughr, see L. JUDS0N, "God or Narure? Philoponus οη Generabiliry and Perishabiliry", ίη Philoponus and the Rejection ofAristotelian Science, ed. by R. S0RABJI (Irhaca 1987), 179-196. 46 Alexander of Aphrodisias, Quaestiones, ed. Ι. BRUNS (Berolini 1887), 30.23-24. 47 See rhe cirarions provided ίη Alexander of Aphrodisias, Quaestiones 1.12.15, trans. R.W. SHARPLES (London 1992), at p. 66-70.

III 24

sition to be eternal, not to be corrupted. Thus it is impossible - purely because of the world' s nature - that the world be generated or corrupted.48 Things look a bit different in the works of Themistius. His views οη the world' s eternity are again largely faithful to those of Aristotle, as may be gleaned not only from his paraphrases of the Physics and De caelo, 49 but also from his paraphrase of Metaphysics Book Lambda, which ίs preserved ίη Arabic, Hebrew and Latin. 50 Ιη Book Lambda, chapter five, Aristotle had appealed to the eternity of the world as a premise for proving the existence and nature of the first cause. He did not, however, attempt to prove here that the physical world ίs eternal - as we would expect, if it is right to say that he sees this as a matter for physics. Ιη his paraphrase of this chapter Themistius mostly confines himself to expanding οη allusions to the world' s eternity by inserting versions of the Physics VIII arguments, e.g. οη time (at 12.13 ff.) and the impossibility of a first motion (13.5 "there is ηο origination of motion except through [another] motion") . 48 Admittedly Alexander does also use arguments rhat are not straighrforwardly "physical". For example, he alludes repearedly ίη rhe Principles (e.g. at §23) ro rhe fact rhat the heavens "imitate" rhe eternity of the fιrst cause. 49 Ιη his paraphrase of Physics 8.1 , for example, Themistius follows Aristorle closely ίη arguing for rhe impossibiliry of a beginning of motion (210.3ft) or time (2 l l .34ff, concluding with the point rhat "if rime is eternal, then so is motion", 212.8), and for rhe impossibiliry of an end to motion (212.10ft). Like Aristorle and Alexander he says rhat rhe eternity of the fιrst mover follows from the ererniry of motion (233.14-17). For the Physics paraphrase see the edition of Η . SCHENKL (Berolini 1900); for τhe exτant Laτin version of the De caelo paraphrase see τhe edition of S. LANDA U E!ι (Berolini 1902). 50 Arabic ediτion in Α. BADAWΊ, Aristu 'inda l-'arab (al-Qahira 1947). Latin and Hebrew editions by S. LANDAUE!ι (Berolini 1903). Badawι's Arabic texτ must be read alongside the τextual variants supplied in R.M. Flv\NK, "Some Texrual Notes οη the Oriental Versions of Themistius' Paraphrase of Book 1 [sic] of the Metaphysics", in Cahiers de Byrsa 8 (1958-1959), 215-230. French translation by R. BMGUE in Themistius, Paraphrase de la Metaphysique dΆristote (livre Larnbda) (Paris 1999). Οη rhis rext see also S. ΡΙΝΕS, "Some Disrincrive Metaphysical Conceptions ίη Themisrius' Commentary οη Book Larnbda and Their Place ίη rhe Hisrory of Philosophy", ίη his Collected Works, volurne ΠΙ Oerusalem 1996), 267-294.

III ΤΗΕ

LAST PHILOSOPHERS OF LATE ANTIQUITY

25

So far, so Peripatetic: he merely makes explicit the physical arguments that Aristotle has given elsewhere. But consider the following: "We say that motion cannot cease or come to be. If ίt were originated, then its mover would be prior to ίt. But how can we imagine that ίt has a mover, which is eternal (azaltJ, but that ίt does not come to be from it for all of eternity (dahr)? There is nothing to prevent its coming to be from ίt. And there is nothing that begins to be ίη a state, such that by being ίη this state ίt would originate [the motion], since all that originates only originates from [the first mover], but there is nothing other than [the first mover] that could hinder it or awaken its desire. Nor is it possible for us to say that ίt was first incapable of bringing its effect about, and then capable, [or that ίt first did not will and then willed, or did not know and then knew,] 51 because this would require change, which would ίη turn require that there be something else that changes ίt [sc. the first mover]. But if we say that there is something that prevents ίt [from causing the motion], then ίt would follow that there is some other cause more powerful than ίt." 52

At the beginning of the passage Themistius is alluding to the argument of Physics VIII.1 with which we began this section. But he soon veers off into an argument that is more reminiscent of Proclus': the first cause is immutable, unique, and cannot be made to act by anything else. Thus ίt ίs the nature of the cause that determines the eternity of its effect. So in this passage, at least, we can see Themistius going further than Alexander in a Platonizing direction, using metaphysical argumentation in the midst of a generally physical treatment of the world' s eternity. Let us turn now to Proclus, whose discussion of the eternity of the world in his Commentary on the Timaeus is among the most clear and detailed expositions of the problem from a Neoplatonic point of view. 53 The fact that this exposition is 51 This phrase ίη brackets appears ίη the version reported by SharastanI: see R.M. FRANK, art. cit. (η. 50), 220 η. 73. 52 Α. BADAWI, op.cit. (η . 50) , 12.18-13.5; cf. 14.21-32 ίη the Latin version. 53 Procli Diadochi in Platonis Timaeum commentaria, ed. Ε. DIEHL (Lipsiae 1903), 276.8-296.12. See further J. F. PHILLIPS, "Neoplatonic Exegeses of Plato's Cosmogony (Timaeus 27C-28C)", ίη JHPh 35 (1977), 173-197; R. SORABJI,

III 26 found in the context of interpreting the Timaeus is ηο accident. Just as the decisive texts οη the world's eternity were, for the Peripatetics, two unambiguously physical works from Aristotle' s corpus (the Physics and De Caelo), so for the Platonists the question of the world' s eternity arises in the context of discussing the Timaeus.54 Since Platonists saw the Timaeus as the dialogue in which Plato sets out his views οη the physical world, 55 we might initially expect Proclus to pursue exclusively physical arguments in his Commentary. And in fact, in this very section οη the world's eternity, Proclus reproaches his predecessor Severus for "bringing mythic obscurities into natural philosophy" (Ι 289.14-15), going οη to add, "these exegetical points, being unrelated to physics, must not be admitted" (290.2-3) . However, matters are complicated by the fact that for Proclus, the Timaeus is also a work of theology, insofar as the dialogue sets out how the physical cosmos is fashioned by a demiurge. 56 Thus ίt ίs as a practicioner of "natural philosophy" in a rather different sense than Aristotle' s that Proclus addresses himself to the infamous interpretive diffιculty raised by Plato's claim, at Timaeus 28b, that the physical world "has Time, Creation and the Continuum (London 1983), chapters 13-15; G . VERBEΚE, "Some Late Neoplatonic Views οη Divine Creation and the Erernity of the World", ίη Neoplatonism and Christian Thought, ed. by D.J. O'MEARA (Albany Ι 982), 45-53. 54 As H.S. LANG and A.D. MAcRo, op. cit. (η . 40), remark, "for the Platonists the problem of the eternity of the world is indistinguishable texrually from Plato's Timaeus and its account of how rhe world, or cosmos, is made" (21) . 55 Proclus himself announces at rhe beginning of his commentary ( 1.17-18) that the ropic of the dialogue is the philosophy of narure. Οη this see now Μ . MARTIJ N, Proclus on Nature (Leiden 2010) . 56 This becomes clear especially from Proclus' use of the Timaeus in his Platonic Theology. See Proclus, Theologie Platonicienne, vol. 1, texte eτabli eτ τrad. par Η . Ο. SAFFREY eτ L.G . WESTERINK (Paris 1968), e.g. at 19.6-8 (the Timaeus is one of the dialogues that studies "divine rhings" from a mathematical, erhical, or physical point of view); 24.17 (ίt is one of the most important dialogues for Plato's theology); 25.8-11 (ίt deals specifιcally with rhe intellectual gods, the demiurgic monad, and rhe encosmic gods, see furrher 27.17-18); 29.24-30.3 (the Timaeus is about physics, but "for rhe sake of narural philosophy" must also deal wiτh τhe noetic gods, since one k.nows images rhrough τheir paradigms). See also rhe use of τhe Timaeus at Platonic Theology V.15-20.

III ΤΗΕ

LAST PHILOSOPHERS OF LATE ANTIQUITY

27

been generated, beginning from some starting-point (γέγονεν , άπ' άpχ_ης τινος άpξά.μενος)".

Αη earlier Platonist reading of this passage, associated with Atticus and Plutarch, understands Plato to be saying that the world has a beginning (άpχ_ή) ίη time, and takes γέγονεν to be a reference to that temporal beginning. Proclus rejects this, arguing that γέγονεν instead echoes the immediately preceding contrast (Timaeus 27d-28a) between the realm of becoming (το γιγνόμενον) and rhe intelligible world of being ( το ον). The cosmos, holds Proclus, is perpetually in a state of coming-to-be, and thus it may said always to be undergoing generation. As for the word άpχ_ή at 28b, it refers not to a temporal beginning but to an "external cause", namely the Demiurge himself (see In Tim. Ι 279.23-25) . Proclus praises his Neoplatonic predecessors Plotinus, Porphyry, and Iamblichus for likewise seeing the passage as a reference to the fact that the world has an external cause, rather than a beginning in time (Ι 277.11-14). Proclus gives a series of arguments against the idea that the world has a temporal beginning. Most of them are textual; that is, Proclus tries to establish Plato' s meaning by citing other passages in the Timaeus and beyond. For our purposes his final two arguments (Ι 288.14ft) are more interesting. These proceed, like the first argument of his On the Eterni'ty of the World, from the nature of the Demiurge. Unlike the cosmos the Demiurge belongs to the realm of being, and thus he must always be doing whatever he does. But, "if he always makes (οημιουpγει), what ίs made always exists too" (Ι 288.16-17). Οη the other hand, the eternity (&ιδιότης) possessed by the world is not the same as the timeless eternity (αίων) of the realm of being; rather the world ίs eternal ίη the sense of lasting for infinite time. And it is preserved eternally only by the constant renewal of its existence by its cause (Ι 278.19-21). So the generation of the world is not the generation of something that comes to be and later passes away, or that begins moving and later completes its motion. It is not, that is, the sort of

III 28

generation studied in physics. 57 Το Proclus' mind, it is part of Plato' s superiority over Aristotle that Plato grasps the need to ground the world' s continued existence in theological, rather than physical, principles (1 295.22-27). Ιη arguing that the question of the world' s eternity is decided above all by its relationship to an external cause, Proclus opens the door for anti-eternity arguments that belong to the same theological or metaphysical arena. The opening was exploited by John Philoponus in his works οη the eternity of the world. We are in possession of fragments of a work rebutting Aristotle' s arguments for the eternity of the world in De caelo and the Physics (Against Aristotle), 58 and a monumental treatise which repeats, and then refutes, Proclus' arguments in On the Eternity of the World (Against Proclus). 59 Philoponus seems also to have written an independent treatise arguing that the world cannot be eternal because it does not possess infinite power; this work was known ίη the Arabic tradition. 60 Philoponus uses different strategies in responding to Aristotle' s physical arguments and Proclus' theological arguments. Against Aristotle, he argues that since the heavens are finite (by virtue of the fact that they are bodies), they cannot move for an infinite period of time by their very nature. 61 Instead - and here Philoponus could find some common ground with Proclus - if the heavens move 57 See the remark of J.F. PHILLIPS, art. cit. (η . 53), 178, that "to aei gignomenon means for Proclus rhat which comes to be by a cause external to ίt (in Tim . Ι 279, 24f.) ... rhis relationship to its higher cause makes the cosmos a special sort of geneton to which the concepts of the natural sciences do ηοτ apply". 58 Preserved in Simplicius' commentaries οη De caelo and the Physics; see Philoponus, Against Aristotle on the Eternity of the World, trans. C . WILDBERG (London 1987). 59 Ed. by Η . RABE (Leipzig 1899; reprinted Hildesheim 1984). 60 See S. ΡΙΝΕS, "Απ Arabic Summary of a Lost Work of John Philoponus", in 105 2 (1972), 320-352; reprinted in his Studies in Arabic Versions of Greek Texts and in Medieval Science, vol. 2 Oerusalem 1986). This may or may not be the same as the work discussed by Simplicius at in Phys. 1326-36. See Philoponus and Simplicius, Place, Void, and Eternity, trans. D. FURLEY and C. Wιιo­ BERG (Iιhaca 1991), 107-128. 61 SIMPL., in Cael., ed. J.L. HEIBERG (Berolini 1894), 79.2-8; Fragment ΙΙ.49 ίη Philoponus, Against Aristotle, trans. C. WILDBEKG, op. cit. (η. 60).

III ΤΗΕ

LAST PHILOSOPHERS OF LATE ANTIQUITY

29

eternally they can do so only by receiving the infinite power required to do so from an external cause. There is ηο possibility, then, of a physical proof of the heavens' eternity. Similarly, in refuting the arguments given by Aristotle in Physics 8, Philoponus argues rhat Aristotle has illegitimately assumed rhat rhe production of the world must be like the production of anything in nature. But this is false, because the world is created by God, and God' s creative act need not obey the laws of nature. 62 When arguing against Proclus, Philoponus must of course take a different tack. Here it will not be sufficient simply to say that the physical world' s nature is compatible with a beginning ίη time. He must dispute Proclus' claims about rhe way in which God, or the Demiurge, in fact creates the world. Only this will tel1 us whether the world is eternal or not. As I have pointed out elsewhere, Philoponus seems to be aware rhat his task is to show that even the Demiurge cannot create an eternal cosmos. 63 For instance, an eternal world would require rhere to be an actual infinity, which is impossible, as Aristotle and Proclus both admit. Likewise, an eternal world would be equal to its Creator in eternity, but this is impossible, for what is brought into being must be lesser than its cause. These are not physical impossibilities that have to do wirh the nature of the created cosmos: rather rhey are absolure impossibiliries, which limit the possible outcome of God' s creative power and generosity. Philoponus recognizes that he and Proclus in a sense agree, insofar as both accept rhat rhe world' s remporal duration ίs determined by God, not by its nature: "if Proclus agrees wirh Plato about these doctrines [sc. that the world acquires its being from an external cause, is in itself generable and corruptible, and ίs finire], rhen he makes ίt clear rhat rhe world ίs corruptible by its nature, while incorruptibility belongs to it 62 SIMPL., in Phys., ed. Η. DIELS (Berolini 1882, 1895), 1141 . 12-16; Fragment VI. 115 ίη Philoponus, Against Aristotle, trans. C. WILDBERG, op. cit. (η. 60). 63 See Ρ. ADAMSON, Al-Kindi (New York 2007) , 84-85.

III 30 from above nature, supplied by some superior power" (§29, 240.19-23). Recent scholarship has suggested that the Arabic tradition continues this trend towards establishing the world's eternity through theological or metaphysical arguments, rather than physical ones. As we might expect given Philoponus' influence οη him, al-Κindϊ is a good example. 64 Ιη fact al-Kindϊ seems even happier than Philoponus to assume that physical considerations are not decisive in settling the matter. He enthusiastically endorses Aristotle' s cosmology, according to which the heavens consist of an ungenerable and indestructible fιfth elemen t - but adds that, of course, the heavens can only exist for as long as God wills. 65 He thus casually discards as irrelevant the entire debate between Philoponus and Aristotle. And reasonably so, if physical considerations are ίη any case overriden by theological ones. Similarly, a recent study by Marwan Rashed showed that al-Farabϊ saw Aristotelian arguments for the world' s eternity drawn from physics as falling short of demonstrative status.66 Some such arguments are merely dialectical. Others can be used to prove the jάct that the world is eternal, without attaining a demonstration, because a proper demonstration argues from cause to effect (in this case, from God to the world). Thus only a theological argument can be demonstrative. 67 1 have elsewhere suggested that al-Κindϊ was motivated to disprove the world' s eternity because he adhered to the contemporary theological contrast (as seen most prominently in the controversy over the nature of the Qur'an) between the Or so Ι have argued ίη ibid., 88. See his On the Nature of the Celestial Sphere, ar Μ. ΑΒΟ RioA (ed.), Rasa'il al-Kindi al-falsafiyya, 2 vols (al-Qahira 1950/53), vol. 2, 40-46: 46. 66 Μ. RASHED, 'Άl-Farabι's Losr Trearise On Changing Beings and rhe Possibility of a Demonsrrarion of rhe Ererniry of rhe World", ίη ASPh 18 (2008), 19-58 : 21 for Maimonides' disrincrion between narural and rheological proofs. 67 Jbid. , 44. Such proofs would proceed, for insrance, from rhe erernity and unchanging acrivity of God ro rhe erernity of His effecr, namely rhe world. 64

65

III ΤΗΕ

LAST PHILOSOPHERS OF LATE ANTIQUITY

31

eternal and the created. 68 He does not consider the possibility that something could be both created and eternal. This uncritical acceptance of the created-eternal dichotomy could not last for long, and indeed it was already rejected by al-Κindϊ's contemporary Thabit Ibn Qurra. His short treatise explaining Aristotle' s Metaphysics has recently been edited and translated by David C . Reisman and Amos Bertolacci.69 It deals with, among other things, the manner in which God bestows existence οη things in the physical cosmos. Thabit sees divine causation of motion as tantamount to divine causation of existence: "The First Mover is the cause of the form that gives subsistence to the substance of all the things that are properly in motion. Thus the subsistence of the substance of each one of them does not belong to it in itself, but rather is from something that is the first ground (sabab) for its motion ... The First Mover, then, ίs the principle and cause for the existence (wugud) and perdurance of the forms of all corporeal substances. For, when we imagine the removal of the existence of natural motion [from corporeal substances] ... their substance undoubtedly corrupts." 70

For Thabit' s Aristotle, the fιrst mover does not merely cause the motion of a cosmos whose existence ίs taken for granted, but rather causes the cosmos to exist precisely by making it move. The argument proceeds by supposing that for any natural object to exist is for it to have a form, and for it to have a form is for it to have a proper motion. Thabit' s Aristotle does not however say that God is an efficient cause of that existence, only that He ίs a fιnal cause : God causes motion as an object of desire (sawq). Thabit nonetheless ascribes to Aristotle the view that God' s relationship to the See Ρ . ADAMSON, op. cit. (η . 63), 98-105. D .C. R.EISMAN and Α. BERT0LACCI, "Thabit ibn Qurra's Concise Exposition ofAristotle 's Metaphysics: Text, Τ ranslation and Commentary", ίη Thabit ibn Qurra: Science and Philosophy in Ninth-Century Baghdad, ed. by R. RASHED (Berlin 2009) , 715-776. Ι am grateful to the authors for allowing me to see this important study ίη pre-print form. Their translations, with some minor modifι­ cations. 70 D .C. R.EISMAN and Α. BERT0LACCI, art. cit. (η. 69), §2. 68

69

III 32 world is one of "willful making (fan" iradt)" (§ 6). Ιη light of this it is still unclear what sort of proof should be given regarding the world's eternity. Thabit's Aristotle may still want to say that the eternity of motion ίs implied by the intrinsic nature of the heavens or of bodies in general. Ιη that case the final causality exercised by God will be only a necessary condition for the persistence and motion of these bodies. As it turns out, though, Thabit' s Aristotle makes ηο appeal to the nature of the physical cosmos when proving the world' s eternity: Treatise, §4: "What Arisrorle says is rhat rhe most excellent [state] for the First Principle is that ίη which ίt is the cause from eternity of rhe exisrence of everything that exisrs ... without having become like that only at some time, after not being like rhat ... So rhis is more excellent rhan rhat rhe First Cause is rhe cause of rhe existence of rhe universe at some rime." 7 1

This has nothing to do with the impossibility of a first motion or moment of time. Rather, the argument is that for God to be most excellent (afrf,al), He must have an eternal, rather than changing, relationship with the cosmos. And of course this could only be the case if the cosmos is itself eternal. Thabit' s Aristotle continues by observing that, if God went from not being a cause of existence to being a cause of existence, there would have to be some second, further cause to explain this change. But this is impossible, since there is nothing else that could either assist or hinder God ίη bringing existence to the world. We have seen this sort of argument before. lt is very like Themistius' argument from the paraphrase of Lambda, altered so as to make God a cause of existence, as well as motion. 72 But unlike Themistius, Thabit gives only this theological argument for the eternity of the world, proceeding then to deal with possible objections. The fιrst objection is that if the 71

Ibid. , §4.

72 D.C. REISMAN and Α. BEIΠOLACCI,

op. cit. (η. 69), demonstrate Thabit's reliance οη rhe paraphrase of Themistius ίη rheir commentary οη rhe text, rhough rhey do not cite Themistius as a source for rhis particular passage.

III ΤΗΕ

LAST PHILOSOPHERS OF LATE ANTIQUITY

33

world is eternal, then it has ηο cause - in other words, alKindϊ' s assumption that the eternal and created are exhaustive and mutually exclusive. Thabit rejects this in his own voice (§ 5), and goes οη to ascribe to Aristotle a rebuttal, which again relies οη the idea that the first cause cannot change so as to start bringing the world into existence. The second objection is that if the world is eternal then God can have ηο will (or at the very least that He need not have a will, since He can serve as a final cause without actually doing anything at all). This ίs refuted by reverting to the claim that God must be as perfect as possible - thus God does have will, but does not have desire, since causing without will or out of desire would both imply imperfection. Thabit's discussion of the eternity of the world is thus strikingly theological in character. This is despite the fact that Thabit' s Treatise is ω be located much more in the Peripatetic than the Neoplatonic tradition. The theological approach can also be discerned in early Arabic writings that are overtly hostile to the Aristotelian tradition. One example is a work of uncertain authorship entitled On Metaphysics. It was discovered by Paul Κraus in an Istanbul manuscript, where it is ascribed to the famous philosopher and physician Αbϊι Bakr al-Razϊ. Despite his suspicions about its authenticity, Κraus included it in his edition of the works of al-Razϊ. 73 The work has since found its supporters and detractors as an authentic Razian document.74 Ι myself am increasingly convinced that it is authentic. One reason is that the treatment of the world' s eternity chimes well with remarks οη the same topic in al-Razϊ's certainly authentic Doubts About Galen, which Ι will mention below. On Metaphysics is only partially ed. Ρ . Kι~us (Al-Qahira 1939), at 116-134. See Α. BADAWI's chapter οη al-Razι ίη Α History ofMuslim Philosophy, ed. by Μ.Μ . SHARIF, νοl. 1 (Wiesbaden 1963), at 440-441. He cites previous views, mostly noncommittal (to this group may be added the remarks ίη Α. BAUSANI, Un jilosofo 'Ίaico" del medioevo musulmano: Abu Bakr Mu/Jammad ben Zakiryya Razi [Roma 1981], at 14). The most significant study of the work accepts its authenticity: G.A. LUCCHEΠA, La natura e la sftra: la scienza antica e le sue metafore nella critica di Razi (Lecce 1987). 73 Al-Razι, Rasa'ilfάlsafiyya,

74

III 34

extant and has a rather disjointed structure, with ηο smooth transitions between the three main sections of the extant work. 75 These three sections are: a general attack οη philosophers' claims that "nature" explains motion and operates teleologically (116.2-124.6), a section disputing philosophical theories of how the human fetus is generated (124.7-128.2), and a cosmological section, dealing inter alia with the eternity of the world, the possibility of multiple worlds, and the question of whether the world is infinitely large (128.3-134.11). The author' s discussion of the eternity of the world follows οη from the attack οη teleology. So it is ηο surprise that he fιnds unconvincing the physical arguments that have been adduced in favor of the world's eternity. He repeats (128.3-8) Aristotle' s argument in Physics VIIl. l against the possibility of a fιrst motion, and then says rather dismissively (128.8-9): "What we say is that the body and the motion are originated together, and we have undermined [the above argument] already, by saying that the Creator, the great and exalted, possesses an act that operates without His having changed."

Interestingly, the response seems designed to anticipate and rebut even the revised version of Aristotle' s argument, as it appears ίη Themistius and Thabit, ίη which the argument is supposed to turn οη the immutability of God as an agent, rather than οη the conceptual impossibility of a fιrst motion.76 75 lt ends abruptly ίη the midst of a discussion of whether the cosmos must be of fιnite size. There are also internal references, which may be to other sections of the same work; if so the original text could have been quiτe extensive. (See especially the reference aτ 120.11 to a "secτion οη the soul (bab al-naft)"; and also 124.5-6, 128.16-17, 129.11-12.) 76 This is despite the fact that the argument is ηοτ set out that way by τhe author: 'Άristotle gives several proofs that the world 's motions are eternal. Ιη one of these he assumes that the world has always exisτed (anna al- alam qadim) . What he says is that if mοτίοη began in τime, then the body [of the cosmos] sτayed unmoving for an infinite τime, and then moves. If [τhe body of τhe cosmos] has a mover τhaτ has always exisτed, which moves ίt, τhen eiτher ίt changes or τhe body τhaτ ίτ moves changes. Whichever of τhe τwο moves, τhere was a motion before τhaτ motion" (128.3-6).

III ΤΗΕ

LAST PHILOSOPHERS OF LATE ANTIQUITY

35

For the author insists that God could create body and motion de novo without Himself changing. Ιη responding to Aristotle's argument from the nature of time, 77 by contrast, the author is willing to meet the philosophers οη their own ground. He interprets the Aristotelian view (correctly) as follows: there cannot be a first instant of time, because there must be a time before and after every instant. But, he argues (128.17-129.5), if this is taken to imply that ηο period of time can actually begin or end, then neither could there be a first moment of Sunday or last moment of Saturday! Furthermore, time is analogous to space, 78 so if there is a time before and after every time, then there must be a spatial extension beyond every spatial extension, and therefore the world ίs infinitely large, contrary to what the philosophers believe (129.6-9). Because of the author' s dialectical strategy he is happy to confront specific arguments of the upholders of eternity (ahl al-dahr) 79 οη the empirical level of physics, when it suits him. But the dominant feature of his treatise ίs an impatience with appeals to 'Ώature", whether this be to explain apparent intelligent design in the world or the formation of human embryos, or to undergird the opponents' chosen cosmology. Απ appeal to nature cannot settle any of these issues, because nature is subordinated to divine action. The author' s philosophically 77 Paraphrased ar 128.11-16, and said ro appear nor in rhe Physics, bur in Metaphysics book Lambda. This facr and rhe aforemenrioned response ro rhe

argumenr from morion suggesr rhar rhe aurhor may, like Thabir, have consulred rhe Arabic version of Themisrius' paraphrase of Lambda. Tanralizingly, rhe aurhor says he has elsewhere responded ro a similar argumenr in a refurarion of Proclus. 78 For rhis assumprion see also 132.5-7, and compare al-GhazalΊ, The lncoherence of the Philosophers, ed. and rrans. by Μ . Ε . MARMURA (Provo 1997), 33. 79 The expression firsr appears ar 125 .1 ίπ rhe conrexr of rhe argumenr over rhe formarion of rhe embryo; ar rhis srage ahl al-dahr is perhaps used as a general rerm for rhose who pay insufficienr reverence ro rhe divine as cause of rhe world (e.g. marerialisrs). Bur ir may also suggesr rhe conrinuiry of rhe aurhor's polemic: jusr as his opponenrs rhink rhar narure is a sufficienr cause of human formarion, so rhey believe ir is self-sufficienr and ίπ πο need of a crearor, and rherefore erernal.

III 36

astute skepticism about natural philosophy comes out most strongly in passages where he attacks the Aristotelians' use of induction from sense-experience. When one group of opponents attempts such an argument (the world ίs infinitely large, because we never see a body without another body beyond it), he responds by pointing out the weakness of such inductive inferences (127.17-128.2). 1s an Mrican who has only met black people entitled ω think all people are black? Or ίs someone from a landlocked area entitled ω think that all land is surrounded by further land? Our polemicist is οη ω something here. He realizes that counter-examples 80 carry a weight that positive generalizations made from experience cannot, since such generalizations could always be falsified with the discovery of a counter-example in the future. Here the author moves beyond rejecting physical arguments for the opponents' various cosmological claims, and suggests the ultimate inefficacy of Aristotelian physics as a whole. 81 Let us now turn briefly ω al-Razϊ's Doubts About Galen, which also contains a discussion of the world' s eternity. 82 This is provoked by Galen' s treatment of the issue in the lost On Demonstration, where ίt was argued that such things as the heavens and oceans have never been known ω change from 80 Such as he presenrs elsewhere in rhe rexr, e.g. when he says rhar semen cannor be a suffιcienr cause of pregnancy, because if ir were women would conceive a child whenever ir were presenr (125.3-6). 81 He is similarly scarhing abour Arisrorle's appeals ro common opinions, or endoxa, which represenr anorher imporranr srarring-poinr for rhe Periparerics. Ιη response ro Arisrorle's claim "rhar rhere is ηο need ro give a proof (dali[) of narure, owing ro irs obviousness, and rhe facr rhar everyone recognizes ir and granrs irs exisrence" (116.3-4), he says, "somerhing is nor rrue jusr because everyone granrs ir, jusr as somerhing is nor wrong jusr because everyone denies ir ... Proof is unnecessary only for immediarely evidenr rhings (al-asya' al-muJahada), and for rhe inrellecrual firsr principles of demonsrrarions; bur narure is nor grasped by rhe senses, nor is rhe knowledge of ir a principle ίη rhe inrellecr" (116.9-10; 14-16). 82 For an edirion of rhe work see AI-Razϊ, Kitab al-Sukϊιk 'alιi Galinϊιs, ed. by Μ. MOHAGHEGH (Tehran 1993). The relevanr secrion is rranslared in J. McGιN­ NIS and D.C. RειSΜΑΝ (ed. and rrans.), Classical Arabic Philosophy. An Anthology ofSources (Indianapolis 2007), 51-53.

III ΤΗΕ

LAST PHILOSOPHERS OF LATE ANTIQUITY

37

their state ( 'an balihi) . Since these large-scale features of the cosmos are unchanging over time, we can infer that the cosmos as a whole is eternal (3.18-21). Al-Razϊ contrasts this passage unfavorably ω the more agnostic treatment of the same topic ίη other Galenic works, and exposes certain weaknesses ίη the argument. Particularly interesting for us is his point that things can be destroyed without displaying change or decay over time. For instance, a glass vessel might persist as ίt is for some time, and then suddenly be struck by a rock (4.23). Could the same sort of thing happen ω the cosmos? Yes, at least according ω some : "it is ίη this way that the world is destroyed, according ω those religious believers (mutadayyinϊn) who speak of the world's destruction" (5.2-3). Al-Razϊ need not be endorsing this suggestion. Rather, he's pointing out that it is a possibility which is not eliminated by Galen' s inductive argument. This is reminiscent of On Metaphysics, especially a passage (which 1 take ω be dialectical ίη just the same way) where the author proposes that the phenomena explained as ''natural" by philosophers could just be the result of direct divine action. 83 Ιη both contexts, natural experience is shown ω be non-demonstrative once the possibility of divine action is considered.

Conclusion 1 have here examined two topics which display the continuity of late antique and early Arabic philosophy. This continuity is hardly surprising, given how closely the fιrst philosophers of the Arabic tradition engaged with the last philosophers of antiquity. Οη the other hand, we should always be mindful of the context of that engagement. With respect ω the instrumental status of logic, we saw that members of the Baghdad 83 Al-RazI, op. cit. (η . 73), 116.17-18: 'Όη what basis do you deny that God , great and exalted, is all by Himself (bi-cfιitihi) the one who necessitates the powers of all other acts, and the natures of things?"

III 38 School retained the commitments of their Greek authorities while also formulating a response to the attack of contemporary critics. Οη the topic of the world's eternity, slightly earlier authors like al-Kindϊ, Thabit and al-Razϊ carry οη the late ancient process by which physical arguments are shown to be inadequate to determine whether the world is eternal or not. Their discussions must also be read in contemporary context, in this case provided by kaldm authors who strictly opposed an eternal God to created things, and proposed occasionalist views οη which God acts directly in the world, obviating the need for stable natures. 84 Such developments threw into doubt the apodeictic pretentions of Greek science. Then again, subtle reflection οη the epistemic status of logic, physics, and other philosophical disciplines had always been a part of Greek science itself. Aristotle taught al-Farabϊ the difference between the demonstrative and the merely dialectical; Galen taught al-Razϊ the difference between the certain and the merely probable. Such distinctions invited authors writing in Arabic to stand in judgment over their favorite authorities, and to decide not only what should be retained from the antique tradition - but also where there was room for improvement. 85

84 For more discussion of the relation between kaliim and the first generations of philosophy ίη Arabic, see Ρ. ADAM SO N, 'Άrabic Philosophy. Falsafa and the Kaliim Tradition before Avicenna", ίη The Oxford Handbook ο/ Medieval Philosophy, ed. by J. MARENBON (forthcoming). Ι should perhaps clarify that ίη speaking of a "theological" approach to the eternity debate Ι do not mean a kaldm approach, but rather an approach proceeding from premises about God and His relaιion to ιhe world. Ιη ιhis sense " ιheology" is simply ιhe part of philosophy dealing with God. 85 Ι gratefully acknowledge the Leverhulme Trust's support for my research ίηtο naιural philosophy ίη ιhe Islamic world.

IV

Galen and al-Razϊ οη Time* !η 1955, Shlomo Pines pointed out a striking resemblance between Abίi Bakr alRazrs famous theory of 'absolute time' and a position οη time which we find ascribed to Galen ίη the Arabic tradition. 1 As far as Ι know, ηο one since has tried to determine in any detail whether and how Galen's and al-Razϊ's treatments of time were related. This is, perhaps, unsurprising. Galen voiced his opinions οη time in the context of a criticism of Aristotle's celebrated definition of time. The critique was part of his On Demonstration, which is unfortunately now lost and known only through fragments and testimonia in the Greek and Arabic traditions. As for al-Razϊ, he expounded a theory of absolute time as one of his 'five eternals', setting this doctrine out in several treatises. But these treatises too are lost, and his views are likewise known only through the (mostly hostile) reports of other authors.2 Thus, ίη trying to establish a link between Galen and al-Razϊ οη the topic of time, we are ίη the unpromising position of investigating the influence of one lost work οη other lost works. Some reason for optimism is nonetheless provided by the fact that al-Razϊ himself is a major source for Galen's On Demonstration. He cites extensively from the work towards the beginning of his Doubts About Galen (hereafter Shukίίk), and even refers (albeit briefly) to Galen's position οη time in this section of the Shukίίk (see below, Τ3). 3 Furthermore, Pines was right that there are strong parallels between al-Razrs views οη time and the position οη time we find ascribed to Galen ίη a letter to Ya]:ιya ibn Άdϊ by a Jewish philosopher named Ibn Αbϊ Sa'ϊd al-Maw~ilϊ (see below, Τ4) .4 Both al-Razϊ and the Galen we find ίη Ibn

* Ι am very grateful to Riccardo Chiaradonna, Marwan Rashed, Richard Sorabji and the late David C. Reisman, for comments on a previous draft of this paper. Ι would further like to thank the Leverhulme Trust for their support of this research. 1. S. Pines, Ά Tenth-Century Philosophical Correspondence', Proceedings ofthe American Academy for Jewish Research, 24, 1955, pp.103-36. 2. The evidence is mostly gathered in Abίi Bakr Mui)ammad b. zakariyyii' al-Riizϊ, Rasiϊil falsafiyya (Opera philosσphica), ed. Ρ. Kraus, Cairo, 1939. 3. Mul:iammad b. zakariyya' al-Razϊ, Kitab al-Shukiik 'ala]alϊniis, ed. Μ. Mohaghegh, Tehran, 1993, p. 8, U. 8-9ff. There is a complication here: the Arabic version of On Demonstration by 1:funayn ibn Ιsi)aq apparently did not contain book VIII, which accordiηg to the Greek tradition was the section where Galen discussed time and void. Yet al-Riizϊ clearly knows this part of On Demonstration - unless there was another part of the work where these issues arose. Οη this see Μ . Rashed, Άlexandre et la "magna quaestio". Rδle et independance des scholies dans la tradition byzantine du corpus aristotelicien', Les έtudes classiques, 63, 1995, pp. 296-351 (repr. in Μ. Rashed, LΉeritagearistotelicien. Textes inedites de lΆntiquite, Paris, 2007, pp. 85-141), n. 95. He rightly says that this matter 'demanderait une veritable etude'. 4. For which Ι cite the Arabic edition ίη Yai)ya ibn Άd1, The Philosσphical Treatises, ed. S. Khalifat, Amman, 1988, pp. 314-36. The section ofthe letter setting out Galen's supposed view of time is at pp. 318-19.

IV

reject Aristotle's definition of time; both claim to be following Plato 5; and both hold that time 'subsists through itself' independently of body.' The word 'duration (mudda)', which we are told al-Razϊ used to refer to absolute time, is also used in Ibn Αbϊ Sa'ϊd's report of Galen. 7 Perhaps most striking of all, in the Shukίik al-Razϊ has Galen calling time a 'substance (iawhar)' . This is paralleled ίη both the evidence concerning al-Razϊ's own theory of time and in Ibn Αbϊ Sa'ϊd's presentation of Galen. 8 So we have a strong prima facie case for thinking that al-Razϊ was influenced by Galen when he developed his own conception of time, and that in doing so alRazϊ took himself to be echoing Plato's views οη time. That would fit well with what we know about al-Razϊ, whose primary vocation was medicine, who admired Galen as a paramount medical authority, and who considered himself to be an adherent of Platonism, which he understood through the filter of Galen's presentation of Plato (especially the Timaeus).' But pursuing this plausible hypothesis any further brings us to another difficulty. The version of Galen's remarks on time that we find ίη the Arabic tradition is at variance with the version we can find in Greek sources. As we will see, the Greek tradition suggests that Galen was (at most) making a purely epistemic or methodological point ίη criticizing Aristotle. The Arabic tradition, though - as already implied by the evidence mentioned above - has Galen making a bold metaphysical claim about time, namely that time is a 'self-subsistent substance'. !η what follows, 1 want to look at this discrepancy ίη some detail. 1 will argue that the Arabic tradition does provide strong evidence that Galen made significant metaphysical claims about time, although not every aspect of this evidence is reliable. Ι further want to show that al-Razϊ was influenced not only by the metaphysical import of Galen's view, but also by the epistemic remarks about time that were made ίη On Demonstration.

Αbϊ Sa'ϊd

The Greek Galen and the Arabic Galen The most important Greek source for Galen on time is Themistius's paraphrase of the Physics. 10 Ιη two passages, he refers to Galen's critique of the Aristotelian definition of time. Simplicius, who repeats the material ίη his own commentary οη 5. For Galen see Κhalifat, Philosophical Treatises (η. 4 above), p. 318, ll. 13- 14; for al-Razι see e.g. al-Razι, Rasa'il (η. 2 above), p. 305, ll. 7-8. 6. Κhal1fat, Philosophical Treatises (n. 4 above), p. 318, 1. 9; al-Razι, Rasa'il (n. 2 above), p. 196, 1. 6 of §3. 7. Κhal1fat, Philosophical Treatises (n. 4 above), p. 318, 1. 14; al-Razι, Rasa'il (η. 2 above), p.195, 1. 5. 8. Al-Razι, Shukuk (η. 3 above), p. 8, 1. 8; al-Razι, Rasii'il (n. 2 above), p. 266, 1. 3 of §4; Khalifat, Philosophical Treatises (η. 4 above), p. 318, 1. 9. 9. For an example of this see Ρ. Adamson, 'Platonic Pleasures ίη Epicurus and al-Razί, in In the Age of alFarabί: Arabic Philosophy ίn the Fourth/Tenth Century, London, Warburg Jnstitute, 2008, pp. 71-94. See also Μ. Μ. Bar-Asher, 'Quelques aspects de l'ethique dΆbίi Bakr al-Razι et ses origines dans l'oeuvre de Galien', Studia Islamίca 69, 1988, pp. 5-83 and 70, 1989, pp.119-47. 10. Themistii ίn Aristotelίs physίca paraphrasis, ed. Η. Schenkl, Berlin, 1900. The relevant sections of the paraphrase are translated ίη Themistius, On Aristotle Physics 4, transl. R. Β. Todd, London, 2003.

2

IV GALEN AND AL-RAZΪ ΟΝ ΤΙΜΕ

the Physics, tells us that the critique was to be found ίη On Demonstration Book Here is the second of the two passages ίη Themistius: 12

νιιι. 11

We must not align ourselves with Galen ίη his belief that time is defined through itself (άφορίζεσθαι δι' αύτου). For after fully listing numerous significations of 'before' and 'after' he says that none coincide with the definition [of time] except the one ίη respect of time, so that time is [defined as] 'the number of change ίη respect of time'.

Τ1

The point of this is fairly clear. Aristotle claimed that time is the number of motion ίη respect of before and after. Against this Galen points out that 'before and after' can, as Aristotle himself might have put it, be said ίη many ways. !η this case the relevant sense of 'before and after' is a temporal one. Thus Aristotle has defined time ίη terms oftemporal priority and posteriority, and this is circular. 13 The earlier passage ίη Themistius also shows Galen taking aim at Aristotle's definition of time ίη terms of motion: 14 Clearly there is ηο time without change - and without change not in the way Galen believes, because we think of time as we undergo change (for this is what he believes Aristotle to be saying), but because the conceptions of time and change are interdependent. Why did he pointlessly contest this by trying to argue against it? Ίη fact', he says, 'we think of unchanging things as we undergo change, e.g. the poles of the cosmos and the centre of the earth, and still these are not accompanied by change.'

Τ2

This is somewhat harder to interpret than Tl. 1 take the point to be that we are aware of time even as we consider things that are not merely at rest, but necessari/y 11. Simplicii in Aristotelis physicorum libros quattuor priores commentaria, ed. Η. Diels, Berlin, 1882, p. 708, 11. 27-8. The relevant sections ofthe commentary are translated in Simplicius, OnAristotle Physics 4.1-5 and 10-14, transl. J. Ο. Urmson, London, 1992. Simplicius seems to draw mostly οη Themistius for his

knowledge of Galen, but the information that the original Galenic source is On Demonstration Vlll is not found in our version ofThemistius. 12. In Phys. p. 149, 11. 4-7 (transl. Todd, modified): Γαλην δέ ού προσεκτέον οίομένψ τον χρόνον άφορίζεσθαι δι' αύτου· πολλά γάρ έξαριθμησάμενος σημαινόμενα του προτέρου τε καl ύστέροu τά μέν

iiλλα ούκ έφαρμόζειν φησl τ όρισμ, τό κατά χρόνον δέ μόνον, ωστε είναι τον χρόνον αριθμόν της κινήσεως κατά χρόνον.

13. This accusation has an afterlife in modern discussions of Aristotle. For a sophisticated attempt to show that Aristotle succeeded in defining temporal priority in terms of priority of position (and more generally, priority within a change), see U. Coope, Time for Aristotle: Physics ΙV.1 0-14, Oxford, 2005, ch. 4. Themistius's own response is interesting, and seems to admit that there is a kind of circularity ίη Aristotle's definition, while insisting that the circularity is notvicious (p. 149, 11. 15- 19). 14. ln Phys. p. 144, 11. 23-9 (transl. rodd): δηλον οδν, ώς ούκ εστιν ό χρόνος iiνευ κινήσεως, iίνευ κινήσεως δέ ούχ δν τρόπον οιεται Γαληνός, έπειδή κινούμενοι νοουμεν τον χρόνον· οϋτω γiφ ύπολαμβάνει λέγειν

Άριστοτέλην, άλλ' έπειδή τfi της κινήσεως έννοί~, σuνήρτηται ή τού χρόνου. τί οδν μάτην ηθλει προς τούτο δηθεν άντιλέγειν έπιχειρών; 'καi γiφ τiι άκίνητα', φησί, 'νοούμεν κινούμενοι, οίον τούς πόλους τού κόσμου

και τό κέντρον της γης, καi ομως ούκ εστι ταυτα μετά κινήσεως'. There is a slight problem here in that Simplicius has κινούμενον in the quotation of Galen, where Themistius has κινούμενοι. R. Chiaradonna, Ίe Traite de Galien Sur /α demonstration et sa posterite tardo-antique'. in Physics and Philosophy ofNature in Greek Neoplatonism, eds R. Chiaradonna and F. Trabattoni, Leiden, 2009, pp. 43- 77, at p. 55, suggests emending Simplicius to bring the text into line with Themistius.

3

IV

unmoving. For the centre of the cosmos is eternally, hence necessarily, unmoving and this according to Aristotle's own cosmology, which gives the example its dialectical bite. So we have here an example of time without motion and without even potential motion, and Galen thinks this example disproves the Aristotelian definition. 15 If we accept the Galenic critique, there seem to be two options before us. First, we might offer a different, non-circular definition of time. There is ηο hint ίη the (admittedly scanty) Greek tradition that Galen's 0n Demonstration included any attempt to do this. Second, we might just admit that time cannot be defined ίη terms of non-temporal concepts: it is epistemically primitive. This is the conclusion of an excellent recent discussion of these passages by Riccardo Chiaradonna, who writes: !Όbjection de Galien prenait place au sein d'une discussion consacree a la definition, dont le but etait de montrer qu'il existe des entites primitives et manifestes (dont le temps) qu'il n'est ni necessaire ni possible de definir. 16

In fact, the Greek evidence does not allow us to go even this far with any certainty. Perhaps Galen was making no claims of any kind, even epistemic ones, with regard to the nature of time. He may have been satisfied to demolish Aristotle's definition and then move on, leaving as an open question whether time might be non-circularly defined ίη some other way. Ιη that case the purpose was simply to illustrate the importance of avoiding circularity when defining. But Simplicius can be read as supporting the more interesting version of Galen's strategy. He says that for Galen time is 'self-indicating', for which he uses a variety of expressions (δι' έαυτοu δηλοuσθαι; δι' έαυτοu δεικνύμενος; δι' έαυτου σημαίνεται : ln Phys. pp. 718, 1. 14; 719, l. 11; 719, 1. 16). Unfortunately there is a question whether these phrases refer to a positive view about time on Galen's part, or simply gloss what Galen was saying about Aristotle's view of time, namely that it fails by defining time through itself. 17 For the sake of argument, though, let us suppose that Galen was making at least some claim of his own and not merely attacking Aristotle. !η that case, what the Greek evidence suggests is that his claim was purely epistemic, and concerned the impossibility of defining time in non-temporal terms. This is remarkably different from what we find ίη the Arabic evidence. We can start with the earliest source, which is none other than al-Razϊ's Doubts about Galen:18 [Galen] declares that in his view, time is a substance (jawhar), since it can bear quantity.

Τ3

15. Compare the similar interpretation ofRashed, Άlexandre et la "magna quaestio"' (n. 3 above), n. 81. 16. Cruaradonna, Ίe Traite de Galien' (n. 14 above), p. 73. !n the same volume, R. Μ. van den Berg interestingly discusses a range of other ancient texts which appeal to 'cornmon notions' of time and how these were, or were not, used to reach a non-circular definition of time. 17. Thus Chiaradonna, Ίe Traite de Galien' (n. 14 above), p. 61. 18. Shukiik (η. 3 above), pp. 8, 11. 7-9, 1. 12. 4

IV GALEN AND AL-RAZΪ ΟΝ

ΤΙΜΕ

The letter of Jbn Αbϊ Sa'ϊd mentioned above contrasts Galen's views on time to those of Alexander of Aphrodisias. Here is the passage detailing Galen's view: 19 Also, let me know whether you hold that time necessarily depends οη motion, such that there can be ηο time except through motion, and the latter is the cause of the former's existence. For Aristotle does hold this. Or, do you rather hold that time has an existent nature [of its own], and is a substance subsisting through itself (jawhar qa'im bi-nafsihT), which motion merely measures and determines, the way a surveyor measures the earth with a cubit? For according to the testimony of Alexander in his treatise refuting Galen οη the subject of place and time, Galen held this opinion, but Alexander refuted him on this point. For Galen held that time is subsistent through itself and has ηο need for motion ίη its existence, and he said that Plato was of the same opinion οη this topic, that is, that he held that time is a substance - by which he means 'duration (al-mudda)' - and motion merely measures and determines it. So Galen says that motion does not produce time for us, 20 but rather produces only day, month and year for us. Time, though, is existent in itself, and is not 21 an accident dependent upon [motion].

Τ4

Ibn Bajja's commentary on the Physics also reports views of Galen on time: 22 TS Some people hold that time is something divine (amr i/ahi). For instance Galen says that one must content oneself with what people know about it, and that its substance ljawhar) eludes the mind. If you consider what he says about time ίη this section (faJ/), it is obvious that he fails to understand [time], conceiving it as some thing or other among the types of existents. Galen's situation in this case is as Alexander says: that he speaks of things without realizing that he is not saying anything about them.

Another set of remarks by Jbn Bajja has Galen accusing Aristotle of making time a product of the imagination, based on our experience of motion.23 This seems to relate to a survival of the material from Galen also reported by Themistius. We find this material also in a passage in Averroes' Long Commentaιy on the Physics, which reads: 24 Galen thought that Aristotle meant we grasp time only insofar as we change (non comprehendimus tempus nisi cum mouemur), and we grasp change through the imagination; this indicates that there is ηο time without change (tempus non est extra

Τ6

19. Ed. Khalifat, Philosophical Treatises (η. 4 above), pp. 318, 1. 6-319, 1. 3. Ά Tenth-Century Philosophical Correspondence' (η. 1 above), pp.111-13.

Cf. the translation offered ίη Pines,

20. Omitting the first innamii. 21. r follow Pines in supplying a negation here. 22. Ρ. Lettinck, Aristotle's Physics and its Reception ίn the Arabic World, Leiden, 1994, p. 694, 11. 13-19. The passage is also briefly summarized at p. 374. 23. See Lettinck's summary, ibid., p. 382: 'Galen has opposed the proposition of Aristotle that motion belongs to the essence of time because we cannot imagine time without motion. Galen thought that Aristotle meant to introduce time by means of the imagination, ίη the same way as when one imagines Zayd one imagines him dressed ίη black'. 24. Aristotelis Opera cum Averrois Commentariis, Venice, 1562 (repr. Frankfurt a. Μ., 1962), IV, p. 177Μ.

5

IV

motum). Because Galen thought this, he opposed Aristotle, saying that we often grasp things that are at rest while we are changing due to the motion of the imagination, for instance when we grasp the poles and centre of the cosmos.

In addition to these testimonia, there are two Graeco-Arabic works that should be taken into consideration. One is Galen's own paraphrase of the Timaeus, which is lost in Greek but extant in an Arabic translation. Its only direct discussion of time is a close paraphrase of Timaeus 38-9:25 Next, he discusses the nature of time, saying that it is measured (yuqaddaru) by the revolutions of the wandering stars and the whole of the celestial sphere. For night and day both arise from the motion of the sphere, while the months arise from the revolutions of the moon as it traverses its course and catches up to the sun (cf. Tim. 38d), and years arise from the sun's traversing its course. Then, he says that each one of the wandering stars has a motion specific to it, about which most people fail to inquire (cf. Tim. 39c-d), yet all these motions have a single thing as their goal (qaJd), namely the completion of the whole year.

Τ7

Another passage in the paraphrase mentions in passing that heavenly bodies are 'instruments of time (al al-zaman)'." Earth too is mentioned as one of these instruments, which provokes the aside, 'in my opinion this is an error'. It is unclear to me what we should make of this. Perhaps Galen is suggesting that the earth is not in fact an instrument of time because it does not move, which would vaguely connect the passage to the evidence ofT2. The other Graeco-Arabic text is Alexander's On Time, which is likewise lost in Greek but extant in Arabic (as well as Latin).27 This treatise, a defence and interpretive exposition of Aristotle's account of time, never mentions Galen by name. But it may be identical with Alexander's Against Galen on Time and Place, which is mentioned in the Fihrist; if so, the section on place is now missing.28 In our version of On Time the opponent confronted by Alexander remains unnamed, but even without the evidence from the Fihrist there are powerful reasons for thinking Galen is the target. The opponent's basic position is set out negatively. He denies that time can be identified with the celestial sphere itself, its motion, or the number of its motion (§1, p. 19, 11. 6-7 Arabic). This fits reasonably well with the 25. PlatoArabus Ι. Ga/eni compendium Tίmaei Platonίs, eds R. Walzer and Ρ. Kraus, London, 1951, §V, p. 8, 11. 610 Arabic. 26. Ιbid., §VΙ(g), p. 11, U. 5-6 Arabic, based on Tim. 42d. 27. For discussion and translation ofthis text see R. W. Sharples (with notes on the Arabic version by F. W. Zimmennann), Άlexander of Aphrodisias, On Time', Phronesίs, 27, 1982, pp. 58-81. The Arabic text is edited in Ά. Badawϊ, Commentaires sur Arίstote perdus en grec et autres epitres, Beirut, 1971, pp. 19-24. 28. So Sharples, Άlexander of Aphrodisias, 0n Time' (η. 27 above), p. 72. For a sceptical approach to the AJexander νs. Galen motίf in the Arabic tradition, see 5. Fazzo, Άlexandre contre Galien: la naissance d'une legende', Phi/osophie antique. Problemes, renaίssances, usages, 2, 2002, pp. 109-44. See also the nuanced and wideranging discussions in Rashed, Άlexandre et la "magna quaestio'" (η. 3 above), and F. W. Zimmermann, ΆI­ Farabi und die philosophische Κritik an Galen von Alexander zu Averroes', in Akten des 7. Κongresses fiir Arabίstik und Islamwίssenschaft, ed. Α. Dietrich, Gδttingen, 1976, pp. 401-14. 6

IV GALEN AND AL-RAZΪ ΟΝ ΤΙΜΕ Galen we know from Themistius, who was trying to undermine the Aristotelian definition of time as the number of motion. Furthermore, the opponent says that time would continue even if motion (e.g. the motion of the celestial sphere) were to cease, a claim which Alexander flatly rejects (§5, p. 20, 11. 3-7 Arabic). This would fit well with Τ2, ίη which Themistius tells us that Galen thought time would pass even for that which is not moving, giving the example of the centre and poles of the cosmos. That very example arises ίη On Time, at §23, where Alexander bites the bullet and admits that such things are indeed atemporal. 29 Given that Τ4 by !bn Αbι Sa'ιd actually refers to a text in which Alexander refuted Galen οη the nature of time, the obvious inference is that Τ4 is based οη this treatise, presumably a more complete version of it which included a discussion of place too, and which identified Alexander's opponent as Galen. If we now summarize the position ascribed to Galen ίη all these Arabic sources, we get something like the following. Time considered ίη itself is self-subsistent, instead of being dependent οη the individual motions that 'measure' it by dividing ίt into specific, limited durations. The determined times corresponding to such motions are for example a day, a month, or a year (mentioned in Τ4 and Τ7, and also in Alexander, On Time §27, p. 24, 11. 3-6). These measures of time, which we produce ίη accordance with the motions of the celestial sphere, create the illusion that time is actually dependent οη those motions. But ίη reality we can conceive of time passing without even the possibility of motion (On Time §5 and Τ6, confirmed by Themistius). Time is self-subsistent, a 'substance Uawhar)' (Τ3, Τ4, Τ5) and even 'divine' (Τ5). AII of this sounds remarkably similar to al -Razϊ's doctrine of absolute time. But is it a theory that can plausibly be ascribed to Galen? The most promising idea is that he could have developed the view as part of his interpretation of Plato. Sharples pointed out that Galen's paraphrase of the Timaeus clearly rejects the eternity of the world - a rejection which incidentally is confirmed by quotes from On Demonstration found at the beginning of al-Razϊ's Shukίίk and ίη Philoponus's Contra Proclum. 30 But he accepted that there was a 'disorderly motion' prior to the demiurge's imposition of proportion οη the receptacle. This is clear from the following passage in the Tίmaeus paraphrase: 31 By the words 'something undivided' and 'something divided in bodies', he means the innate motion in matter, which he says a little later on possesses eternity. for soul is the principle of motion, according to him, but matter is moved by itself. So it is clear that [matter] is ensouled, but that soul which is in it is confused and moved without any definite order. Thus the Creator, the blessed and exalted, wishing to

Τ8

29. Οη this see Rashed, Άlexandre et la "magna quaestio'" (η. 3 above), p.111, who ingeniously proposes a parallel to Alexander's reported vίew that the outermost celestial sphere has ηο place. 30. See Shukίik (η. 3 above) pp. 3, 11. 17- 6, 1. 6, translated ίη C)assical Arabic Philosophy: an AnthologyofSources, eds and transl. J. McGinnis and D. C. Reisman, Indianapolis, 2007, pp. 51- 3; Philoponus, Agaίnst Proe/us on the Eternίtyofthe World 12-18, transl.J. Wilberding, London, 2006, pp. 89- 91. 31. §IV, p. 7, 11.1-7 Arabic.

7

IV

turn [matter] towards arrangement and order, put the soul into it which is from the nature of something that remains always in a single condition.

Perhaps, then, Galen forrnulated the view that there is an eternal tirne which corresponds to the eternal disorderly rnotion prior to the derniurgic ordering of the cosrnos. The derniurgic irnposition of order and proportion then produces the definite tirnes that correspond to orderly rnotions, such as the revolutions of the spheres (e.g. one revolution of the sun around the earth would constitute a single day). This is, as far as it goes, quite plausible, especially since Ιbη Αbϊ SaΊd rnentions that Galen clairned cornrnon cause with Plato. But we need to be cautious. Neither Τ7 nor Τ8 go so far as to clairn what Ibn Αbι Sa'ϊd attributes to Galen ίη Τ 4, narnely a tirne which is self-subsistent and independent of rnotion.32 Οη the reconstruction Ι just offered, Galen would be saying that we need tirne before the cosrnos precisely because there is already rnotion, narnely the disorderly rnotion of rnatter. Adrnittedly, that leaves open the possibility that tirne is a necessary condition for rnotion, yet can also (ίη principle) exist without rnotion. But we find ηο such conception of tirne as self-subsisting in the Timaeus paraphrase. Nor is it sornething the opponent, presurnably Galen, seerns to envision ίη Alexander's On Time. Rather, it seerns to appear ίη the Arabic tradition only in Ibn Ab1 SaΊd's Τ4, after which it is picked up again by Ibn Bajja in TS. Let us, then, consider rnore carefully each of the significant clairns about tirne ascribed to Galen ίη the Arabic tradition, but not ίη the Greek tradition: (a) Tirne is a substance (iawhar, presurnably translating ούσία) [Τ3, Τ4 , TS]. (b) Tirne is not an accident [Τ4] (c) Tirne is self-subsistent [Τ4] (d) Tirne is rneasured by rnotion, not vice-versa [Τ4, Τ7, Alexander On Time] (e) Tirne exists independently of motion [Τ4, Alexander On Time] (t) Tirne is divine [TS] The evidence at our disposal suggests strongly that (a) is accurate. As we have seen, Ibn Αbι SaΊd's report οη Galen's view seerns to derive frorn its refutation by Alexander; our On Time ίs probably a partial version of this refutation. We know that al-Razι, by contrast, had the luxury of reading Galen's On Demonstration directly. These represent, therefore, two independent strands of evidence concerning Galen's understanding of time. The characterization of time as a 'substance' is sornething we find ίη both al-Razι's Τ3 and Ibn Αbϊ Sa'ϊd's Τ4. Unless we assurne contarnination of one of these traditions by the other (Ι will consider and reject this possibility in the next section), the two bits of evidence can thus be taken to confirrn one another. Clairn (a) is fairly secure. 32. This poίnt has also been made by R. Sorabji, Time, Creation and the Continuum, London, 1983, p. 83. 8

IV GALEN AND AL-RAZΪ ΟΝ ΤΙΜΕ

Sirnilar reasoning renders (b) and (c) at least plausible. Although only Ibn Αbϊ Saίd in Τ4 makes these two claims explicitly, 33 al-Razϊ gives implicit support to both. Starting with (b), in Τ3 al-Razϊ directly quotes Galen as having said that time has quantity. Indeed he makes claim (b) the basis of claim (a): time is the subject of quantity, therefore it is a substance. Α plausible way of reconstructing the full argument would be, 'tirne is the subject of quantity; quantity is an accident; that in which an accident adheres is not an accident but a substance; therefore time is a substance'. If Galen argued in this way, he would have said explicitly (in the third premise) that tirne is not an accident. On the other hand, it may have been an inference that was drawn only in the Arabic tradition, or even foisted on Galen by Alexander, on the basis that anyone who says that time is a 'substance' must also deny that time is an accident. As for (c), it is again only Τ4 that has Galen saying time is 'self-subsistent'. But this is a claim that al-Razϊ made about time on his own behalf. 34 So if Galen did not actually call time 'self-subsistent', it would seem that both al -Razϊ and Ibn Αbϊ Sa'ϊd (or his source) independently inferred the self-subsistence of time from its being a substance. We cannot, then, rule out that Galen explicitly said time was self-subsistent, and not an accident. But I think an at least equally plausible account would be the following. Once Galen was on record as saying that time was an ούσία, sympathetic authors like al-Razϊ, unsympathetic readers like Alexander, and officially neutral readers like Ibn Αbϊ Sa'ϊd, all naturally assumed that time must have certain features associated with Aristotelian ούσία. 35 And two hallmarks of Aristotelian substance, of course, are that it is contrasted to accidents and that it is self-subsistent. But by calling time ούσία, Galen may have meant no more than that time really exists, as opposed to being the subjective result of the soul's measuring motion. 36 This will have been Galen's understanding of the notorious remark of Aristotle at Physics IV.14, 223a21-6, that without soul there would be ηο time.37 Now go back to the idea most prominent in our Greek evidence, that according to Galen time cannot be defined in terms of anything else. This point is an epistemic or methodological one, but one that might easily be misunderstood. It would have been natural to confuse a claim about epistemic primitiveness ('selfindicating') with a claim about metaphysical primitiveness ('self-subsistence'), especially ίη light of Galen's having said that time is an ούσία. So on balance, 1 33. And it is worth bearing ίη mind that we need to correct the text to get claim (b) explicitly even ίη Τ4; see η. 21 above. 34. See n. 6 above. 35. This is not, of course, to say that al-Riizϊ's own positive theory identifιes time with an Aristotelian primary or secondary substance. However, Ι think that his commitment to the metaphysical primacy of time does go beyond anything he could explicitly find ίη Galen, and part of this may have been his stress on the 'self-subsistence' of time. 36. As Riccardo Chiaradonna has pointed out, Galen elsewhere seems to take ούσία to mean nothing more than ίiπαρχις, 'existence'. See Chiaradonna, 'Le Traite de Galien' (η. 14 above), p. 63. 37. On which see U. Coope, Time fσr Aristσtle (n.13 above), ch.10. 9

IV

think we should take Galen to have made the more modest metaphysical claim that time exists, without having gone οη to insist οη its self-subsistence and nonaccidentality. This sorts better, too, with what we know about On Demonstration. lts aim was not to set out a metaphysical account of time or anything else. It was to lay down the rules of proper argument, giving examples and criticizing other authors when they violated these rules. 38 Next, we have two claims about the relationship between time and motion: time (d) is measured by motion and (e) can exist independently of motion. Claim (d) is one of the most strongly attested ίη the Arabic tradition, especially since Galen's own Timaeus paraphrase says that time is 'measured (yuqaddaru)' by orderly motion. So this is fairly reliable. Things are a bit murkier with claim (e). lt is similar to what we find already ίη the Greek tradition, with Galen's idea that time would pass for us even as we think about necessarily unchanging things like the centre of the cosmos (ΤΖ). But again, it is telling that the epistemic dimension of this claim has largely been lost in the Arabic tradition (the exception is Averroes' Τ6) . What Galen probably did say is that time is a necessary condition for motion, rather than the other way around. He may have had ηο interest in the question of whether time could, in theory, exist in the complete absence of motion. 39 For, as mentioned above, the Timaeus as he understood it was committed to an eternal disorderly motion prior to the cosmos. The question whether time could pass without any motion at all would be theoretical and idle - the sort of question Galen took pride in ignoring. Here we can draw a parallel between On Demonstration's handling of time and its discussion of the eternity of the world. Galen's main aim in both cases was to attack Aristotle's poor argumentation, but he was willing to stray from this purpose to endorse positive claims he found in the Timaeus. The Arabic tradition provides ηο strong evidence that he strayed further. Finally, Ι think we should certainly reject the ascription of claim (f) to Galen. This idea that time is 'divine' appears only very late, in Ibn Bajja, and is presumably a hostile gloss οη Galen's commitment to the eternity of time,4° one that rnisunderstands him completely. Galen thought that time, like matter and motion, is eternal, since this is required by the Timaeus. But there is ηο hint in the Timaeus that time is itself divine, nor is there any reason to think Galen would have said so. Indeed, not even al-Razϊ says that time is divine. For him, as for Galen, it is eternal, and a 38. See again Chiaradonna, Ίe Traite de Ga1ien' (n . 14 aboνe). 39. rf this is right, it would give us a point of contrast between Galen and al-Razϊ. We're told that the latter asked common folk for their intuitions about time and place, and learned that they thought time would persist even if the celestial spheres were removed. He took this to show that one may coherently imagine t ime without any body whose motion is meas ured by time. See al -Razι, Rasaϊl (n. 2 above), p. 264; translation in Pines, Beίtriige zur ίslamίsch en Atomenlehre, Berlin, 1936, rr, p. 54 . 40. Ιn the late antique tradition, indeed, Christian authors like Zacharias and Procopius argued against the eternity of the world on the basis that the doctrine would compromise God"s unique claim to divinity. See Richard Sorabji's contribution in the present volume, item (iii) under 'Christian objections to a co-eternal world'.

10

IV GALEN AND AL-RAZΪ ΟΝ ΤΙΜΕ

necessary condition for the formation of the ordered cosmos we see around us. The same could be said for matter, but it clearly is not divine.

Thiibit ibn Quπa and the metaphysical primitiveness of time So far, then, 1 have argued that the Arabic tradition is of use in reconstructing Galen's position on time, so long as it is used with caution. It shows that Galen did call time an ούσία, though he may have meant less by that than later authors assumed. It shows also that for Galen, time is not only epistemically prior to motion but is also measured by motion, rather than vice-versa. None of this is particularly surprising: the c!aim of time's existence was part of his antiAristotelian polemic, and the idea that time is measured by motion was taken from his reading of the Timaeus. Galen's view was then adopted, and given a stronger metaphysical valence, by a1-Razϊ in his theory of the five eternals. Ι now want to consider an alternative story, namely that al-Razϊ may have played some role in distorting the Arabic interpretation of Galen, and in particular, been responsible for exaggerations we find in Ibn Αbϊ Sa'ϊd. After all, it is very striking how similar Ibn Αbϊ Sa'ϊd's Galen is to al-Razϊ. Both say that time is an eternal, self-subsistent substance, independent of but measured by motion. 41 Furthermore, ibn Αbϊ Sa'ϊd makes a point of using the Razian term mudda ('duration') to refer to time in itself. Could it be, then, that al-Razϊ played some role in developing the metaphysical interpretation of Galen's treatment of time, which was then taken on by Ibn Αbϊ Sa'ϊd and Ibn Bajja? If so, then al-Razrs theory of absolute time would not be a borrowing from Galen, but rather a view of his own which he then fathered upon Ga]en. Indeed there is some evidence that al-Razϊ's theory had a very different provenance. According to the Ziid a/-Musiifirϊn of Na~ir Khusraw, which is one of our major sources for al-Razϊ's thought, the theory of absolute time was taken over by al-Razϊ from a mysterious thinker named al-ϊranshahrϊ. There are some similarities between the statements about time that Na~ir Khusraw imputes to al-Ϊranshahrϊ and the statements put into Galen's mouth by ibn Αbϊ Sa'ϊd. 42 It is difficult to say who is Al-Riϊzϊ,

41. Ιn addition to the passages already cited, see al-Razι, Rasa'il (n. 2 above), p. 198, ΙΙ.10-21 for the clairn that absolute time is not elirninated by the elirnination of the rnotion that 'rneasures' it,just as absolute place is not elirninated by the elirnination of what is contained within it. 42. See al-Riizι, Rasiϊίl (n. 2 above), pp. 266-7, translated frorn Kraus's Arabic version: 'The sect of philosophers who said that rnatter and place are eternal also affirrned that tirne is a substance (jawhar; the terrn is also used ίη the original Persian). And they said that tirne is an extended and eternal substance. They rejected the staternent of those philosophers who said that tirne is the nurnber of the rnotions of the body, and said that if this were so, then it would be irnpossible for rnoviηg thiηgs to rnove at the sarne tirne with different nurnbers. Al-ϊranshahrι the Philosopher said that tirne, eternity, and duration (mudda; the terrn is also used in the oήginal Persian) are nothing but narnes whose rneanings reduce to a single substance. He further said that tirne is a proof of God's knowledge, just as place is a proof of His power, rnotion is a proof of His action, and body is proof of his potentiality. He said that each of these four is infinite and eternal, and that tirne is a substance which proceeds without standing still. As for Mu~arnrnad Ιbη Zakariyyii' al-Riizι, his discussion follows al-ϊranshahrι: he said that tirne is a substance which flo,vs'. On this passage see further S. Pines, Beitrαge (n. 41 above), p. 52, and on al-Transhahrι, ibid., pp. 56-60. 11

IV

influencing whom ίη all of this, especially because we know essentially nothing about al-ϊranshahrϊ other than what Νaψ Khusraw tells us. And there is some reason to be sceptical. Like most sources which tel1 us about al-RazΓs ideas, Na~ir Khusraw is hostile towards him, and ίη this case the hostility manifests itself ίη the claim that al-Razϊ stole his ideas from somebody else. But if we do take this evidence at face value, it becomes tempting to conclude that al-Razϊ read Galen's On Demonstratίon under the influence of al-ϊranshahrϊ. The resulting distortions might then have been repeated by later authors. But there are at least two problems with this story. First, 1 stress again that ibn Αbϊ Sa'ϊd and Jbn Bajja refer to the refutation by Alexander when they summarize Galen's view. The basis for their understanding of Galen was Alexander's On Time, not the original arguments of On Demonstration, which were al-RazΓs source. Since there is ηο text connecting al-Razϊ to Alexander's critique of Galen, it is implausible to ascribe to al-Razϊ a pivotal role ίη the interpretation of that critique. Second, there is a text which suggests that the reading of Galen we find ίη Ibn Αbϊ Sa'ϊd predates al-Razϊ. Α letter by Thabit ibn Qurra remarks, rather in passing: 43 Again, if someone considers the essence (dhat) of time and place according to Galen's doctrίne, he will find it impossible for them to receive any changes whatever.

Τ9

One cannot really call this a testimony for Galen's views οη time, since the reference is so oblique; this is why Ι have not considered it above. Still, it suggests that the sort of interpretation of Galen we find in Τ4 was already known to Thabit. His idea might be that, if as Galen believed time and place were self-subsisting substances which exist independent of change, they themselves would be unchanging. Jt is worth emphasizing that Thabit refers to Galen's views οη both time and place. That suggests a link to Alexander's refutation of Galen, since as we have seen, our version of On Time is most probably only part of a larger text that dealt with both time and place (as mentioned ίη the Fihrist). We should not put too much weight οη this though, since Galen apparently also discussed place in On Demonstration. One might counter that Thabit's allusion to Galen comes in the context of criticizing 'those who hold that there are first causes other than [God] for motions and actions'. If Thabit is implying that Galen made time a principle which is even οη a par with God, that would be reminiscent of the views οη time we find ascribed to al-Razϊ and al-Ϊranshahrϊ. Might Thabit already be reacting to his younger contemporary al-Razϊ? 44 Ι think it much more likely that we have an independent tradition of reading Alexander's On Time - and not On 43. Α. 1. Sabra, 'Thabit ibn Qurra on the Infιnite and Other Puzzles', 7-eitschrift fiir Geschichte der arabischWissenschafien, 11, 1997, pp. 1-33; §12, at p. 10, 11. 3-4 ίn the Arabic edition (Sabra's translation). 44. AI-RazI died in AD 925, Thabit already in AD 901. Here it is worth noting that the treatise On Metaphysics (edited at al-RazI, Rasiί'i/ [n. 2 above], pp. 116-34), whose attribution to al-RazI is disputed, has a section (pp. 129-31) refutίng Thabit on the topics of celestial motion and infinity, which is the main topic of islωnischen

the letter ίn which Thabit alludes to Galen's views on time.

12

IV GALEN AND AL-RAZΪ ΟΝ

ΤΙΜΕ

Demonstration - already witnessed here ίη Thabit, later taken up by Ibn Αbϊ Sa'ϊd and then Ibn Bajja. According to this tradition, Alexander was confronting a Galen who made time a self-subsistent substance.45 There is ηο reason to think that alRazϊ had anything to do with this tradition. Rather, he read On Demonstration itself and used it as inspiration for his own theory of time. Whether or not he also took over ideas from the shadowy al-Ϊranshahrϊ remains, of course, a moot point. Al-Riizι and the epistemic primitiveness of time We can securely establish one more link between the Galenic and the Razian treatments of time, namely a commitment to its epistemic primitiveness. This is an aspect of al-Razis theoιy that is well testified across a range of sources. For example:

TlO Mu]:ιammad b. Zakariyya' al-Razϊ, drawing οη the ancient Greeks, has spoken of

five principles: the Creator, praise be to Him, universal soul, prime matter, and absolute place and time. Οη this he builds up his teaching (madhhab), which is based οη [these principles]. He distίnguishes between time (zamiin) and duration (mudda) οη the basis that number is applied to one but not the other, because that to which number attaches is finite. Just as the philosophers say that time is the duration of what has a beginning and an end, but eternity (dahr) is the duration of what has ηο beginning or end. And he says that the five [eternals] ίη this existence are necessarily exίstent. The sensible things ίη ίt are matter which has receίved form through composition, and this [sc. matter] is ίη place; so doubtlessly there is place. The variatioη ίη its [sc. the seηsible's] states is due to their beiηg subject to time (mίn lawiizίm al-zamiin), for some of them [sc. the states] come earlier and some later; through time are kηοwη the old and the ηew, what is relatively older and newer, and what is simultaneous (wa-ma 'an). So there can be ηο doubt about [time] (fa-lii

budda mίn-hu) . 46 Tll One of them [sc. the heretics] meηtioηed an account, taken from the ancients,

that eternity (al-dahr) and void subsist process of argument (istidlii/). 47

ίη

the natures of the intellects without any

Know that amoηg those who affirm duratioη, there are two groups. Οηe holds that knowledge of its existence is knowledge that arises necessarily ίη us, without any proof or demonstration. The other attempts to establish it through proof and demonstration. The first group includes Mu]:ιarnmad Ibn zakariyya' al-Razϊ and others.48 Τ12

TlO is a summary of the whole doctrine of the five eternals and hence raises many questions. Ι want here simply to point out its emphatic claim that one can be 45. Or more complicated stil]: Ιbn Αbϊ Saιd is drawing on multiple sources ίn his letter. Sharples, of Aphrodisias, On Time' (η. 27 above), p. 74, suggests that he 'could have combίned ίnformation that he took from our treatise ... with other ίnformation that he had about Galen's views from elsewhere'. The other source could be On Demonstration itself, with or without an intermediary interpretation. 46. al-Razϊ, Rasa'il (η. 2 above), p. 195, 11. 1- 9 (from a]-Bϊrίinϊ, a/-Hind). 4 7. Ιbid., p 198, 11. 5- 6 (from al-Ι~fahanϊ, Κ. a/-Azmina wa-1-amkina). 48. Ιbid., p. 272, 11. 4-6 (from Fakhr al-Dϊn al-Razϊ, Matίilib al-'aliya). Άlexander

13

IV

ίη 'ηο doubt' about the necessary existence of the principles, and that time is that through which we know things measured by time (the Ό]d', the 'new', and the simultaneous), rather than vice-versa. 49 Elsewhere al-Razϊ is quoted as saying, 'the nature of time has certain existence (min ta 'akkud al-wujud)' (al-Razϊ, Rasii'il, p. 199, 11. 9-10). Tll and Τ12 are more explicit still: time is not grasped by the intellect οη the basis of any argument or proof. Rather knowledge of it arises necessarily. It is worth adding that the evidence Na~ir Khusraw provides about al-ϊranshahrϊ does not suggest the latter was concerned with epistemic issues regarding time.50 Thus al-Razϊs emphasis οη the epistemic primitiveness of time is a good candidate for Galenic influence, uncomplicated by other additional sources. Admitted\y, our passages do not focus οη what was apparently most important to Galen in On Demonstration, namely the impossibility of defining time non-circularly. Rather the epistemic point appears in the more general guise of time's immediacy to the intellect. But al-Razϊs lack of emphasis οη the problem of definition as such is balanced by the presence of genuinely Galenic themes: time is grasped not through motion but directly; 'before and after' (or Όlder and newer') are understood ίη virtue of time rather than vice-versa; specific measurements of time like the day come into play only when time is measured by a particular motion. So it is likely that al-Razϊs assertion of the epistemic primitiveness of time represents a survival of Galen's discussion of time ίη On Demonstration. Ιη addition, as we have seen, al-Razϊ drew οη Galen's view that time is a substance measured by motion. We may thus conclude that On Demonstratίon played a significant role in the development of the Razian doctrine of the five eternals.

49. For this idea see also ibid., p. 200, 1. 11: "They said that time is the duration (mudda) οη account of which (read min ajli or li-ajli) one grasps before and after'. 50. As Pines, Beitrίige (n. 41 above) has argued (pp. 56-8), his overriding concern seems rather to have been explicating the divine attributes.

14

V

GALEN

ΟΝ

VOID*

Given how many of Galen's works survive to us, it may seem churlish to lament the loss of others. But historians of philosophy have reason to regret that one treatise ίη particular is no longer extant: Galen's On demonstratίon. It is known to us only through later reports and quotations. These are sufficient to show that this text represented one of Galen's ιnajor engageιnents with philosophical theιnes. 1 ln it, he laid down and defended methodological strictures, and criticized Aristotle and other thinkers for failίng to abide by these strictures. By way of example, he addressed important philosophical issues which remain unexplored ίη the voluminous extant corpus, for instance the nature of time and the eternity of the world.2 Another topic mentioned in On demonstratίon was void. What remains of that discussion is rather meagre, but it can be suppleιnented by passages from other Galenic works. One might woπy, though, that Galen would disapprove of anyone's writing about his views on void. \η ΡΗΡ, he mentions arguments about void as a prime example of fruitless philosophical inquiry: Ιη philosophy it is not surprising that most disagreements have not been resolved, as the rnatters it deals with cannot be clearly judged by an ernpirical test, and therefore some say that the universe did not have a beginning, others that it had, and again sorne say that there is nothing outside surrounding it, others that there is something, and of the latter, some say that what surrounds it is a void that has ηο substance ίη it .. . Such disagreements cannot be settled by clear sense-perception (PHPV.766-67 Κ). 3

* My thanks to Riccardo Chiaradonna and Peter Ε. Pormann for helpful discussion. This paper was written with the support of the Leverhulme Trust, which Ι gratefully acknowledge here. 1 For the fragments of On demonstratίon see Ι. von Mίiller, ' ϋber Galens Werk vom wissenschaftlichen Beweis, ' Abhandlungen der kδnίglίch bayerischen Akademίe der Wissenschaften 1. Klasse 20.2 (J 897). See also Β. Morison , 'Logic,' in The Cambrίdge companίon to Galen, ed. R. J. Hankinson (Cambridge 2008) 66-115, and R. Chiaradonna, ' Le tτaite de Galien Sur la demonstration et sa posterite tardo-antique', in Physics and philosophy of nature ίn Greek Neoplatonism, ed. R. Chiaradonna and F. Trabattoni (Leiden 2009) 43-77. The present paper is strongly influenced by Chiaradonna's study and will, Ι hope, Jend a small bit of confirmation to his general assessιnent ofGalen's purposes in Οη den10nstι·atίon.

For the evidence concerning Galen ' s discussion of time, see Ρ. Adamson , ' GaJen and al-Razϊ οη time,' ίη Medieval Arabic thought: essays ίη honour of Frίtz Zimmermann, ed. R. 1-!ansberger, Μ. Α. al-Akiti , and C. Burnett, Warburg Studies and Texts 4 (London 2012) 1-14.

2

3

See further Chiaradonna' s discussion ofthe passage in his contribution to the present volume.

V 198

It would seem that ίη taking up the topic of void, we are ίη danger of quite literally making much ado about nothing. Απ obvious retort would be that Galen here objects, not to inquiry into void as such, but ίηtο the presence of void outside the universe. The passage concerns what can, and cannot, be determined through sense-perception. Galen's stricture here ιηay therefore derive from the impossibility of going to the edge of the universe to look whether there is any empty space outside it, rather than ruling out discussion of whether a void can exist at all . He thus implies a para11el between the question of void outside the universe and the question of the universe's temporal beginning: we can ' t go back in time to find out whether it did or not. Indeed, Galen did discuss void in ιnedical contexts, where ίt ιnight seem that senseperception can be brought to bear. Υ et it remains the case that void is a topic which provokes Galen to methodological reflection. The parallel question - whether the world has a beginning - is one that Galen did think worthy of detailed discussion, in order to i11ustrate methodological issues. As already mentioned, he did so in the lost On demonstratίon. We know this from the same source that informs us ofGalen ' s a11usion to the void in that work: Doubts about Galen, by the ninth-to-tenth-century doctor and philosopher al-Razϊ. 4 We also find brief reports about Galen's remarks on void in the ancient commentators Themistius and Simplicius. They do not explicitly say they are referring to On demonstratίon , but I wi11 suggest that this is the likely source. In the rest of this paper, then, 1 will first \ook at the topic of void ίη Galen's extant corpus, which means looking at a dispute with Erasistratus. With this background in mind, we will be ίη a good position to reconstruct Galen's more methodologically oriented treatment of void ίη On demonstratίon.

Galen

agaίnst

Erasistratus on

νοίd

Galen's anatomical theories were often presented as polemica\ refutations of his predecessors. The best-known example is of course his refutation of the cardiocentrism of Aristotle and the Stoics. Another is his engagement with Erasistratus (first half of the third century), the great Alexandrian physician and anatomist. Not uncharacteristica11y, Galen ' s polemic is apt to obscure a degree of agreement with, even indebtedness to, Erasistratus. Ιη Nat.Fac. and several shorter works, Galen refutes Erasistratus's theory regarding the motion of pneuιna and blood within the body. 5 This theory depends heavily οη the principle that adjacent material will flow into an eιηptied space to prevent a void from forming in that space. Galen frequently describes this theory using the phrase 'following what is emptied (προς το κενούμενον ακολουθία) '. Ιη this paper 1 wi11 refer to it as the

There are two editions, neither satisfactory: al-Razι, Kίtab al-Shukiik Όlα Jalιniis, ed. Mohaghegh (Tehran 1993); ed. Μ . L. Άbd al-Ghanι (Cairo 2005). For a translation of the section οη the world 's eternity, see Classίcal Arabic phίlosophy: an anthology of soιιrces, ed. and trans. J. McGinnis and D. C. Reisman (lndianapolis 2007) 51-53. 4

Μ.

Οη the topic see 1. Μ . Lonie, ' Erasistratus, the Erasistrateans, and Aristotle', Bulletίn of the History of Medίcίne 38 (1964) 426-43; L. G. Wilson, Έι-asistratus , Galen and the pneuma', Bulletίn of the History of Medicίne 33 (1959) 293-314, and the overview at D. J. Furley and J. S. Wilkie, Galen on respίratίon and the arterίes (Princeton J 984) 26-37. 5

V GALEN

ΟΝ

VOID

199

νοίd prίnciple. 6 Ιη the modern context, a ιηedical exaιηple leaps to ιηίηd: when the plunger of a hypodermic needle inserted into the arιη is pulled backwards, blood flows upwards into the needle. Sucking through a tube (like a drinking-straw) provides a more mundane example. 7 Ιη the theory of Erasistratus, the void principle is used to explain why the dilating heart draws pneuma in from the lungs. If it did not, the increased space in the heart would be empty, and a void would be created. Erasistratus here made the analogy of a bronzesιηith's bellows being expanded, which causes air to flow ίηtο the be!Jows so that ηο void can form. 8 The valve of the heart then closes so that the pneuma does not rush back into the lungs when the heart contracts again. From there, pneuma is pushed along into the arteries by the pumping of the heart. 9 And notoriously, it is thus pneuma and not blood that fills the arteries. This claim may seeιη to us ridicuJous, οη the basis that a pricked artery immediately gives forth blood, not a rush of pneuma. It seemed ridiculous to GaJen, too. He records however that Erasistratus had a ready answer, whicb again invokes the void principle: 'when the pneuma is emptied out, either there will be a massed empty space, or that which is continuous with it will tol\ow it (κενουμένου του πνεύματος ητοι κενός εσται τόπος άθρόος η τό συνεχές ακολουθήσει)' (Art.Sang. 3.1, IV.710 Κ, trans. Furley/Wilkie). Ιη other words, puncturing the artery releases the pneuma and, to prevent a void from forming in the artery, the blood that is 'continuous with it' is pulled along behind. The blood first fills the artery, and then proceeds to leak out ofthe wound. ' 0 Galen has chosen his target we\l, for Erasistratus's theory does seem weak οη various points. Let us begin with the idea that the dilating heart draws ίη pneuma from the lungs. This raises a problem:

Erasistratus says that the heart cannot even draw the air out of the lung during the stoppage of breathing, because ίη these conditions the same volume of expansion is maintained by the organs of breathing. Hence, if the heart had taken some part of the air, the place of the air that was taken would have become empty (κένος), and that is iιηpossible (Ut.Resp. 2.1, IV.473 Κ, trans Furley/Wilkie). But obviously the heart keeps beating when we hold our breath, as Galen is not slow to point out ( Ut.Resp. 2.2, IV.474-75 Κ) . Erasistratus's defenders saw the difficulty and claimed that the heart must therefore draw back pneuma from the artery it has just filled. lt is often referred to as tbe principle that 'nature abbors a vacuum', or by the Latin pbrase horror For a complaint that this is to some extent anacbronistic see Α. Debru, Le corps respίrant: la pensee physiologίque chez Galίen (Leiden 1996) 53 η.26 . 6

vacuί.

7

Discussed e.g. at Nat.Fac. 3.14, ΙΙ.206.6-11 Κ.

8

ΡΗΡ Υ.549 Κ. Οη tbis see Wilson, 'Erasistratus' (η.5, above) 297-98. Galen refers to tbe bellows at UP 6.15 , ΠΙ.480-81 Κ; ΡΗΡ 8.9, V.713-16 Κ; Nat.Fac. 3.15, 11 .206.15-19 Κ.

exaιηple

9 See Wilson, 'Erisistratus' (η . 5, above) 303 and Lonie, 'Erisistratus' (η.5, above) 433 , which points out tbat the needed mechanism of 'squeezing (ekthlίpsis) ' is used in Erasistratus's theory to explain digestion (Nat.Fac. 3.4, JI.156 Κ).

ιο The blood is pulled into the artery tbrough junctions with the veins (this process is called This is one reason Erasistratus was so insistent on the existence of such junctions, also recognized by Galen. See Wilson, Έrasistratus ' (η.5, above) 303. anastoιnosis).

V 200

But this causes problems of its own, for instance, the presence of fυrther membranes that would prevent this from happening even οη Erasistratus's own anatomical theory (Ut.Resp. 2.3-4, IV.475-76 Κ). Without going further into the details, we can see that the void principle invoked by Erasistratus is both a strength and weakness of his theory. Οη the one hand, it can explain transmission of pneuma from lung to heart to arteries through faιηiliar ' mechanical' 1ηeans. Οη the other hand, under anomalous conditions, ίt implies anatomica11y unwelcome results, like backwards reflux into the heart. As for the arteries, here Erasistratus would seem to be οη even shakier ground. Leaving aside the fact that we do not notice pneuma flowing from wounds, it seems that all the pneuma in the body would have to be evacuated before the blood drawn in behind the escaping pneuma would reach the wound. This would presumably take a wbile, and even if it didn't because pneuma moves so quickly, it would surely have catastropbic effects οη tbe animal, something we do not see each time an artery is cut. Galen makes a11 these points, and also observes that the artery wa11s could simply co11apse as the artery is emptied . This objection is raised at Art.Sang. 3.l , IV.7l0 Κ as if it were Galen's brainchild, but a parallel text at Nat.Fac. 2.1, ΙΙ.75 Κ shows that he has lifted the objection from Asclepiades. 11 The two passages even share the same logical structure: Erasistratus says that there are only two options, the creation of void or the drawing ίη of blood; Asclepiades and Galen point out that this ignores the third option of artery collapse. There is ιηοre at stake here than simple questions of anatoιηy. This becomes clear from Nat.Fac., long sections of which contend with Erasistratus in an effort to show that mechanical explanations like the void principle are insufficient to explain the construction and functioning of the animal bodies. Rather, we must suppose that there are various 'inborn faculties' given to animals by nature (Nat.Fac. 2.3, ll.80 Κ) . Among these faculties is the power of ' attraction ' ( όλκή). For instance, the kidneys attract urine, and the linings of the arteries attract the 1ηost nutritious part of the blood. This is of course a familiar Galenic clai1η , but it is interesting to see how he defends it in Nat.Fac. preciseJy ίη opposition to Erasistratus's use of the void principle. ln this case, the main point of contention is not respiration or the function of the arteries, but nutrition. Erasistratus explicitly clai1ηed that 'there seems to be ηο attraction belonging to the stomach (όλκη μεν ούν της κοιλίας ούδεμία φαίνεται είναι)' (Naι.Fac. 1.16, 11.60 Κ). lnstead, he invoked the void principle to explain the dispersal ofnutriment. Galen attacks this claim rather indirectly. He first argues that the kidneys must have a power of attraction for urine. Again, this is against the view of Erasistratus who, as we might by now expect, cJaimed that urine is drawn into the kidney to prevent a void forming when previously present urine is evacuated. And again, Galen is able to attack this by mentioning anomaJous situations. If Erasistratus were correct, then there could never be a dangerous build-up of urine ίη the kidney, since the void principle would onJy cause the kidney to take ίη as much as it expelled. But ίη fact we see people die of this condition (Nat.Fac. 2.16, ΙΙ.64 Κ) . Armed with this observation, Galen is able to argue by analogy for attraction ίη the case of digestion:

11 Furley and WiLkie, Galen on respίratίon (η.5 , above) 260 η.\ 7, also point out the pa1·al\el.

V GALEN

ΟΝ

VOID

201

The dispersal of nutriment does not require us to invoke the filling of what is voided, so long as we concede the attractive power of the kidneys (ή δέ τής τροφής άνάδοσις ουδέν δειται τής προς το κενούμενον άκολουθίας απαξ γε τής έλκτικής δυνάμεως έπι των νεφρών ώμολογημένης) .

11.62-63 Ιη

(Nat.Fac. 1.16,

Κ)

fact, 1ηore generally, If the movement of urine does not depend on the filling of what is voided, it is clear that neither does that ofthe blood nor that ofthe bile, or ifthat ofthese latter does so, then so also does that ofthe former. For they must all be accomplished in one and the same way, even according to Erasistratus himself. (άλλ' εί μή των οϋρων ή φορά τij προς το κενούμενον άκολουθίg γίγνεται, δήλον, ώς ούδ' ή του αϊματος ούδ' ή τής χολής η είπερ εκείνων και τούτου· πάντα γαρ ώσαύτως άναγκαιον έπιτελεισθαι και κατ' αύτον τον Έρασίστρατον.)

(Nat.Fac. 1.17, 11.73

Κ, trans. Brock, modified). 12

If the hypothesis of a natural attractive power is our only option ίη the case of the kidneys, then parity of reasoning demands that we use it in other cases as well. Galen does also argue more positively for an attractive power ίη the stomach, though, and ίη so doing gives us a better reason to see the cases as analogous. He points out that we s01ηetimes have a dangerous build-up (plethos) in the stomach, which is analogous to the case of the kidney's retaining too much urine. As with the kidney, this would not happen if we had only the void principle as an explanatory device, but the power of attraction can cause the build-up when it becomes too dominant (Nat.Fac. 2. 1, 11.76 Κ). 1 3 Nonetheless, a modern reader is apt to be sy1ηpathetic to Erasistratus here, if only because the 'attractive power' mentioned by Galen is rather mysterious. One is put ίη mind of Plato's Phaedo and Socrates' safe but simple-minded explanation of properties: largeness is explained by the presence of the large. Similarly, attraction is explained by the presence of a power to attract - safe enough, but is it really explanatory? One is tempted to say that Erasistratus is invoking an attractive power of a kind, namely the power possessed by a place which is being emptied and which therefore draws in whatever is nearby. Το pursue the comparison with the Phaedo, this would be analogous to a ' clever' cause: snow is a cleverer explanation of cold than coldness, and in the same way emptying is a clever explanation of attraction. Apparently, Erasistratus himself did not see things quite in this way, since as we just saw, he says that the stomach seems to lack attraction (όλκή). But Galen admits that motion explained by the void principle would count as a type of 'attraction': There are two kinds of attraction, that by which a void becomes refilled and that caused by appropriateness of quality; air is drawn into bellows in one way, and iron by the magnet ίη another (ώς δύ ' έστιν όλκής είδη, το μεν τij προς το Translation from Galen , On the natιιralfaculties, trans. Α. J. Brock, LCL (London and New York 1916).

12

13 Galen ιηakes the reverse point also: if nutrition occurred thanks to the void pήnciple, e1ηaciated people could never gain weight! See Nat.Fac. 2.6, 11.103-06 Κ.

V 202 κενούμενον άκολουθίQ , τό δ' οίκειότητι ποιότητος γιγνόμενον· έτέρως μεν γαρ είς τάς φύσας ό άήρ , έτέρως δ' ό σίδηρος ύπό της ήρακλείας έπισπiiται λίθου).

(Nat.Fac. 3.15, ΙΙ.206

Κ,

trans. Brock, modified)

The two kinds of attraction are importantly different, because one involves qualitative discrimination and the other does not. 14 Galen's favorite example of qualitative attraction is the one mentioned here: magnets. Back in Book 1 of Nat.Fac. he \avishes what at first may seem to be a surprising amount of attention οη magnetism. He discusses Epicurus's atteιηpt to explain how magnets work οη the basis of entangling atoms (Nat.Fac. 1.14, 11.47-51 Κ), 15 and argues ίη detai\ that any such theory is bound to fail. Rather, we must ascribe a power of attraction to the magnet, which is of course exerted not οη whatever is adjacent, but specifically οη things like iron . Galen returns repeatedly to the parallel between bodily organs and the magnet, comparing it for instance to sperm's ability to attract b\ood (Nat.Fac. 2.3, ΙΙ.85 Κ; see also 1.15, 11.59-60 Κ, 2.7, ΙΙ.106 Κ). Why is Galen so insistent οη this? lt is not because he rejects the void principle. Οη the contrary, as we wiJ\ see shortly, he makes use ofthis principle in his medica\ theories, even in Nat.Fac. itse\f. Rather, he objects to Erasistratus's theory because it exclιιsίvely invokes mechanisms such as the void principle, without ascribing natural powers to the parts of animal bodies. The power of attraction possessed by the kidneys, stomach, sperm and so οη is a sign ofnature's design, as is shown by the sensitivity ofthis power for what is 'appropriate' to it. We can see that this is a motive for Ga\en's po\emic at several points, especially towards the beginning of Nat.Fac. Book 2, where he criticizes Erasistratus for failing to follow through on his supposed teleological commitments. 16 Οη the one hand he does 'regard al\ the parts ofthe body as having been well and truly placed and shaped by nature' (Nat.Fac. 2.3, 11.81 Κ, trans. Brock), and he says explicitly that nature is ' artistic' (τεχνική). Yet he persists ίη invoking mechanical explanations, like the size of various channels (πόροι). Elsewhere it is explicit that Galen opposes Erasistratus's use of the void principle owing to his own commitment to teleology: 'we can \eave off speaking ofthe filling ofwhat is voided, as unsuitable for a man who assumes nature to be artistic (της προς τό κενούμενον άκολουθίας άποστάντες, ώς ού πρεπούσης άνδρί τεχνικήν ύποθεμένcρ την φύσιν, Nat.Fac. 2.7, 11.106 Κ, trans. Brock, modified).' Nonetheless, Galen is willing to ιηake use of the void principle so long as this principle is at work within an appropriately teleological theory. Nature is the primary explanation ίη our medical theory, but this does not prevent our invoking the void principle as a secondary explanation. This is perhaps why ίt is only towards the end of Nat.Fac. - once teleology has been thoroughly established - that Galen begins to invoke the principle himse\f, even sparing a kind word for Erasistratus. 1n discussing the fact that the same channel can accommodate motions ίη two directions, e.g. swallowing and vomiting, he says:

14

This point is well rnade by Furley and Wilkie, Galen on respίratίon (η.5, above) 33.

See R. Α. Fritzsche, ' Der Magnet und die Athrnung in antiken Theorien ', Rheίnίsche5· Mιιseum 57 (1902) 363-9 l. 15

16 See G. Carnbiano, ' Galeno, Erasistrato e la teleologia di Aristotele,' in St11dί su Galeno: scίenza, filosofia, retorίca efilologίa, ed. D. Manetti (Florence 2000) 45-57.

V GALEN

ΟΝ

VOID

203

Everything possessing a large and appreciable cavity must, when it dilates, abstract matter from all its neighbours, and, when it contracts, must squeeze matter back into them. This should all be clear from what has already been said in this treatise and from what Erasistratus and we too have demonstrated elsewhere about the filling of what is voided. ( ώς ούκ ένδέχεταί τι των αίσθητήν και μεγάλην έχόντων εύρύτητα μή ούκ ητοι διαστελλόμενον ελκειν έξ άπάντων των

πλησίων η έκθλίβειν αύθις είς ταύτα συστελλόμενον εκ τε των ηδη προειρημένων έν τφδε τφ λόγ