ASP 2004 Vol.14 Issue.2

Arabic Sciences and Philosophy, vol. 14 (2004) pp.
171–174
DOI:10.1017/S0957423904000062 2004 Cambridge University Press
ABSTRACTS
Thābit ibn Qurra and the Composition of Ratios

Pascal Crozet
As the natural continuation of his work on the sector figure, the study
Thābit ibn Qurra devoted to the composition of ratios inaugurated a
tradition that traverses the history of classical mathematics. Initially
inspired by the desire to give all possible forms of writing Menelaus’
theorem, this study includes a combinatory aspect which, in particular,
leads the mathematician to use the composition of permutations. Neverthe-
less, other themes intersect one another in this work – the relations between
number and magnitude, the arithmetical treatment of magnitudes, the
concept of ratio, the influence of burgeoning algebra, the role of Euclid’s
Data, etc. – all of them indicative of a new way of apprehending the
objects bequeathed by Thābit’s Greek predecessors, and announcing future
developments.
The Arabic Original of Liber de compositione alchemiae

Ahmad Y. al-Hassan
The Liber de compositione alchemiae (The Book of the Composition of

Alchemy) is believed to have been the first book on alchemy that was
translated from Arabic to Latin (by Robert of Chester). J. Ruska claimed
that the Latin work was a compilation by an Italian Christian cleric
possibly as late as the 14th century. But recent research has shown that it
was a translation from an Arabic work. The article presents an edition
and a translation of the first part of this Arabic original, confronted to
the old Latin translation (quoted in the English translation provided by
L. Stavenhagen).
Yūh
* annā ibn Sarābiyūn: Further Studies into the Transmission of his
Works
Peter E. Pormann
Yūh
* annā ibn Sarābiyūn was an influential medical writer who composed
two Compendia in Syriac: the Small Compendium and the Large Compen-
dium. The Syriac text being mostly lost, we rely on translations into Arabic
and Latin for our knowledge about these works. In the first part we argue
that Yūh
* annā ibn Sarābiyūn lived in the second half of the ninth century,
Downloaded from https://www.cambridge.org/core. Tsinghua University, on 28 Mar 2018 at 13:17:06, subject to the Cambridge Core terms of use, available at
https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0957423904000062
172 ABSTRACTS / RÉSUMÉS
and that reports about a ‘‘Younger Serapion’’ (as opposed to the ‘‘Elder’’)
are erroneous. Secondly, the di#erent Arabic fragments of the Small
Compendium come under scrutiny, with specific emphasis on the problem of
determining which translation each manuscript represents. Thirdly, a
fragment from the Large Compendium will be edited for the first time,
making it possible to compare the two works. One conclusion is that
Yūh
* annā ibn Sarābiyūn displays certain Aristotelian leanings in the Large
Compendium which do not have the same prominence in the Small Compen-
dium; this tendency to use philosophical and logical procedures for the
analysis and classification of disease points to the scholarly tradition of late
antique Alexandria.
Plato’s Republic in Arabic: a Newly Discovered Passage

David C. Reisman
It is a commonly held fact that very little from the works of Plato were
translated into Arabic in the course of the Graeco-Arabic translation
movement and almost none of that meagre amount retain the dialogue form.
The major aim of this contribution is to present an editio princeps of a
hitherto unknown Arabic translation, in dialogue form, of an important
passage from Plato’s Republic (Book VI, 506 d 3–509 b 10), found in only one
of the two recensions of the Kitāb fı̄ Masā’il al-umūr al-ilāhiyya by the
little-known Abū H* āmid al-Isfizārı̄ (fl. mid-fourth / tenth c.). A brief intro-
duction recounts scholarly knowledge to date about Arabic translations of
Plato and of his Republic in particular, presents a general biographical
statement on al-Isfizārı̄ and his legacy, and describes the two manuscripts
used for the edition. This is followed by the edited passage of the Arabic
Republic along with a literal translation of the passage.
RÉSUMÉS
Thābit ibn Qurra et la composition des rapports

Pascal Crozet
Suite naturelle de son traité sur la figure-secteur, l’étude que Thābit ibn
Qurra consacre à la composition des rapports inaugure une tradition qui
traverse l’histoire des mathématiques classiques. Initialement suscitée par
le souci de donner toutes les formes d’écriture possibles du théorème de
Ménélaüs, cette étude comporte un aspect combinatoire qui conduit notam-
ment le mathématicien à utiliser la composition de permutations. Mais
d’autres thèmes s’y entrecroisent (relations entre nombre et grandeur,
RÉSUMÉS 173
traitement arithmétique des grandeurs, concept de rapport, influence de
l’algèbre naissante, rôle des Données d’Euclide, etc.), indiquant dès lors une
manière nouvelle d’appréhender les objets légués par les prédécesseurs
grecs et annonçant les développements futurs.
L’original arabe du Liber de compositione alchemiae

Ahmad Y. al-Hassan
Le Liber de compositione alchemiae (Le livre de la composition d’alchimie) est

considéré comme le premier ouvrage d’alchimie traduit de l’arabe en latin
(par Robest de Chester). J. Ruska pensait que l’ouvrage latin était une
compilation faite par un clerc italien, peut-être au XIVe siècle. Mais la
recherche récente a montré qu’il s’agissait bien d’une traduction à partir
d’un texte arabe. L’article présente une édition et une traduction de
l’original arabe, confrontée à l’ancienne traduction latine (citée ici dans la
traduction anglaise de L. Stavenhagen).
Yūh
* annā ibn Sarābiyūn: Études supplémentaires sur la transmission de ses
œuvres
Peter E. Pormann
Yūh
* annā ibn Sarābiyūn, auteur d’une grande influence dans le domaine
médical, écrivit deux compendiums en syriaque: le Petit Compendium et le
Grand Compendium. Le texte syriaque étant pour la plus grande part perdu,
nous dépendons des traductions arabes et latines pour notre connaissance
de ces œuvres. Dans la première partie de ce travail, nous montrons que
* annā ibn Sarābiyūn vécut dans la deuxième moitié du IXe siècle de notre
Yūh
ère, et que la notion d’un Sérapion le jeune (par opposition à un Sérapion
l’ancien) est erronée. Nous examinons ensuite les divers fragments arabes
du Petit Compendium, tout en essayant de déterminer quelles traductions
sont contenues dans les di#érents manuscrits arabes. En troisième lieu, un
fragment du Grand Compendium est édité ici pour la première fois, ce qui
nous permet de comparer les deux ouvrages. Une des conclusions à tirer de
cette comparaison est que Yūh * annā ibn Sarābiyūn montre certains pen-
chants aristotéliciens dans le Grand Compendium qui n’occupent point une
place aussi importante dans le Petit Compendium; cette tendance à em-
ployer des procédures philosophiques et logiques dans l’analyse et la
classification des maladies relève de la tradition de l’école alexandrine de la
basse antiquité.
La République de Platon en arabe: un texte nouvellement découvert

David C. Reisman
On croit généralement que le courant de traduction du grec vers l’arabe a

très peu concerné l’œuvre de Platon, et que les rares exceptions n’ont
174 RÉSUMÉS
quasiment jamais retenu la forme du dialogue. La présente contribution
se propose avant tout de donner l’editio princeps d’une traduction arabe
sous forme de dialogue, inconnue jusqu’ici, d’un passage important de la
République de Platon (Livre VI, 506 d 3–509 b 10), traduction qui ne se
trouve que dans l’une des deux recensions du Kitāb fı̄ Masā’il al-umūr
al-ilāhiyya, de l’auteur peu connu Abū H * āmid al-Isfizārı̄ (fl. milieu du
IVe / Xe siècle). Une brève introduction fait le point sur les connaissances
scientifiques actuelles concernant les traductions arabes de Platon et plus
particulièrement de la République; on présente ensuite une mise au point
biographique globale sur al-Isfizārı̄ et son héritage; enfin, on décrit les deux
manuscrits utilisés pour l’édition. Suit le texte édité de la République arabe,
accompagnée d’une traduction littérale.
Arabic Sciences and Philosophy, vol. 14 (2004) pp. 175–211
THA
z BIT IBN QURRA ET LA COMPOSITION DES
RAPPORTS*
PASCAL CROZET
Dans l’introduction à un texte majeur consacré à la compo-

sition des rapports et dans lequel on s’accorde à voir un pas
important dans l’élaboration de la notion de nombre irration-
nel, al-Khayyām remarque:
Quant à la composition du rapport dans l’ouvrage de Ptolémée connu
sous le nom d’Almageste, c’est une chose très importante, très utile, et
dont on fait très souvent usage. Sauf que Ptolémée a lui aussi [i.e. à
l’instar d’Euclide] postulé cette prémisse sans démonstration. Et il a
basé sur elle la figure sécante, et sur la figure sécante il a basé la plupart
de la science de l’astronomie; notamment ce qui a lieu en fait d’états, de
lois et de figures dans la sphère céleste et l’équateur. L’utilité de cela, je
veux dire de la composition des rapports, n’est donc pas des moindres! Et
de même l’ouvrage Les Coniques d’Apollonius qui est un préliminaire
important à la plupart des sciences géométriques, notamment les soli-
des. En général, beaucoup de choses très di$ciles et très importantes
dans la science de l’astronomie et la science de la géométrie sont basées
sur la composition du rapport.1
Dans la façon dont al-Khayyām attire l’attention de son lecteur
sur l’importance du sujet dont il entend traiter, en la mettant
en regard de ce qu’il considère comme une lacune dans l’héri-
tage de ses prédécesseurs grecs et hellénistiques – à savoir au
bout du compte l’absence d’une définition consistante de la
composition des rapports – on pourrait ne voir qu’un procédé
rhétorique habile. En réalité, ce passage témoigne bien, nous
* Cet article fait suite à celui d’Hélène Bellosta sur la figure-secteur paru dans
le dernier numéro (‘‘Le traité de Thābit ibn Qurra sur la figure secteur’’, 14, 1
[2004]: 145–68); il s’appuie sur la contribution que nous avons présentée sur ce
sujet lors du colloque Thābit ibn Qurra, savant arabe du IXe siècle, qui s’est tenu à
Paris à l’Institut du Monde Arabe les 14 et 15 décembre 2001, et sur un exposé
donné dans le cadre d’un séminaire du Centre d’Histoire des Sciences et des
Philosophies Arabes et Médiévales en janvier 2004; que les participants à ces
deux réunions soient remerciés pour les remarques qui ont permis d’améliorer ce
texte.
1
Roshdi Rashed et Bijan Vahabzadeh, Al-Khayyām mathématicien (Paris,
1999), p. 374.
176 PASCAL CROZET
semble-t-il, de la distance prise depuis les conceptions de ses
devanciers, en même temps qu’elle annonce la nature même de
son projet.
Pour mieux en comprendre les enjeux, il faut rappeler que la
composition des rapports ne fait l’objet d’aucune définition
explicite dans les Éléments,2 mais qu’une définition implicite se
dégage de l’utilisation que fait de la notion Euclide dans la
proposition VI-23 (les parallélogrammes équiangles ont entre
eux un rapport composé des rapports de leurs côtés): le rapport
A / B entre deux grandeurs A et B est dit composé des rapports
A / C et C / B par l’introduction d’un moyen terme C:
A A C
≠ J ;
B C B
A / C et C / B pouvant être égaux à d’autres rapports r1 et r2, on
pourra avoir également:
A
≠ r1 J r2 .
B
Cette définition implicite semble avoir été partagée par les
successeurs d’Euclide, qui font de la composition des rapports
le même usage: on retrouve ainsi une présentation similaire
chez Archimède (par exemple dans la proposition II-4 de la
Sphère et du Cylindre) ou chez Apollonius (par exemple dans
les propositions III-53–56 des Coniques). Les rapports composés
apparaissent encore dans la proposition VIII-5 des Éléments
(les nombres plans ont entre eux un rapport composé des rapports
de leurs côtés), qui est en quelque sorte le pendant numérique
de la proposition VI-23 et assure donc que:
m1 m2 m1 3 m2
J ≠ .
n1 n2 n1 3 n 2
L’abord calculatoire de la composition des rapports dans des

ouvrages comme l’Almageste de Ptolémée, où l’application du
théorème de Ménélaüs – autrement dit l’utilisation de la
‘‘figure-secteur’’ – se traduit par l’assimilation des grandeurs à
des nombres et, partant, à l’utilisation de la proposition VIII-5,
2
La définition VI-5, que nous commentons brièvement plus loin, est
généralement considérée comme interpolée.
THA
z BIT IBN QURRA ET LA COMPOSITION DES RAPPORTS 177
n’est sans doute pas étranger à l’insertion, à l’époque hellénis-
tique, d’une définition VI-5 qui apparaît comme radicalement
non-euclidienne dans son esprit: un rapport est dit être composé
de rapports quand les valeurs des rapports, étant multipliées
entre elles, produisent quelque chose.
En citant ici dans un même élan les noms de Ptolémée et
d’Apollonius, al-Khayyām ne se propose pas simplement de
réduire la distance entre des traditions qui pourraient paraître
inconciliables. C’est aussi la conception même que l’on a des
rapports et de la façon dont ils peuvent être manipulés qui sont
alors revisitées à la lumière des dernières avancées de la
recherche mathématique: développement de l’algèbre, en par-
ticulier application de la géométrie à l’algèbre grâce au choix
préalable d’une unité de mesure, extension de la notion de
nombre, etc.
Loin d’être isolée, la contribution d’al-Khayyām prend place
dans une histoire d’une étude de la composition des rapports
qui demeure souvent liée à l’examen de la figure-secteur – ou du
moins reste le plus souvent suscitée par lui – mais dont on
pressent qu’elle en dépasse le simple cadre en touchant aux
fondements mêmes des mathématiques classiques. De cette
histoire qui compte encore parmi ses acteurs des mathé-
maticiens comme al-Sijzı̄, Ah * mad ibn Yūsuf, Nas*ı̄r al-Dı̄n
al-T
* ūsı̄, mais aussi des auteurs latins comme Campanus,
Cardan et Maurolico, le traité que Thābit ibn Qurra consacre
au sujet constitue l’un des premiers jalons et peut être consi-
déré comme le point de départ de toute une tradition.
En dépit du fait que certains historiens ont tenté d’attirer
l’attention sur ce livre en en soulignant l’importance ‘‘in
preparing the extension of the concept of number to positive
real numbers’’,3 songeant alors surtout, nous semble-t-il, à sa
descendance chez al-Khayyām, le texte de Thābit sur la com-
position des rapports est resté longtemps méconnu: jusqu’à très
récemment, il n’en existait en e#et aucune édition, seule étant
disponible une traduction en russe par Rosenfeld et Karpova
précédée d’un bref commentaire.4 Richard Lorch en a fourni il
3
B. A. Rosenfeld et A. T. Grigorian, ‘‘Thābit ibn Qurra’’, Dictionary of
Scientific Biography (New York, 1976), vol. XIII, p. 293.
4
L. M. Karpova et B. A. Rosenfeld, ‘‘Traktat Sabita ibn Korry o sostavnykh
otnosheniyakh’’, Fiziko-matematicheskie nauki v stranakh Vostoka (Moscou, 1966),
pp. 5–41; voir également le commentaire des traducteurs dans Istoriya i
metodologiya estestvennykh nauk, 5 (1966): 126–30. Je remercie Irina Luther de
178 PASCAL CROZET
y a peu une première édition critique, agrémentée d’une tra-
duction en anglais et à laquelle nous renvoyons.5 Pour autant,
la portée historique de ce traité, son volume (une soixantaine
de pages) et surtout la richesse des questions qu’il soulève (sur
le concept de rapport, les relations entre nombres et grandeurs,
l’influence de l’algèbre naissante, l’aspect combinatoire qui est
le sien, etc.) justifient que l’on tente d’en donner une analyse
plus fouillée que ce que proposent les commentaires qui accom-
pagnent cette édition, ce à quoi nous nous essaierons ici.
DÉFINITIONS, EUCLIDIENNES ET NON-EUCLIDIENNES

Le livre de Thābit comprend trois chapitres d’inégales
longueurs qui sont ainsi résumés par l’auteur:
1. Le premier chapitre traite des rapports composés les uns avec les
autres.
2. Le second chapitre traite de la connaissance des grandeurs dont
les rapports sont composés les uns avec les autres.
3. Le troisième chapitre traite de problèmes résolus à l’aide de la
composition des rapports.
Le premier chapitre, le plus court, est ainsi destiné à expli-
citer les notions auxquelles Thābit va recourir par la suite:
rapport, composition et séparation des rapports, multiplication
des grandeurs.
Le concept de rapport
Estimant que ‘‘c’est la plus générale des choses qu’il m’est
nécessaire de traiter et que les explications que j’ai vues
m’avoir permis de franchir la barrière de la langue et de prendre connaissance de

cet article.
5
Thābit ibn Qurra, On the Sector-Figure and Related Texts, Edited with
Translation and Commentary by Richard Lorch, Institute for the History of
Arabic-Islamic Science at the Johann Wolfgang Goethe University (Frankfurt am
Main, 2001), pp. 167–326; l’édition du Livre sur la composition des rapports et sa
traduction anglaise occupent les pages 167 à 307. Nous donnerons prochainement
de ce texte une traduction française sur la base de notre édition réalisée pour le
colloque de 2001, dont la qualité de l’édition de Richard Lorch a rendu inutile la
publication (voir Roshdi Rashed [éd.], Thābit ibn Qurra: Savant et philosophe, à
paraître). Pour les citations du texte de Thābit, nous donnons dans ce qui suit les
références du texte arabe dans l’édition de Lorch, la traduction en français nous
étant bien entendu personnelle.
THA
ailleurs sur cette notion n’étaient pas claires’’, Thābit com-
mence son ouvrage en rappelant la définition du concept de
rapport. Mais, d’emblée, se pose un problème textuel qui n’est
sans doute pas anodin compte tenu de l’importance du sujet. La
traduction arabe qu’il propose de la définition V-3 des Éléments
est en e#et la suivante:
Le rapport est, pour deux grandeurs homogènes entre elles, une certaine
raison (qiyās) de l’une à l’autre selon la mesure (misāh
* a).6
Or, dans sa propre édition du traité euclidien, celle que l’on
désigne aujourd’hui sous le nom d’Ish
* āq-Thābit, on trouve:
Le rapport est, pour deux grandeurs homogènes entre elles, une certaine
relation (id
* āfa) de l’une à l’autre selon la grandeur (miqdār).7
Une même structure grammaticale est donc utilisée dans les
deux cas pour rendre la phrase grecque (´ ’ `
´
˜ ¢ ˜ ¢ ` ´ ` ´ ), la seule
di#érence résidant dans le double remplacement de id * āfa par
qiyās et de miqdār par misāh * a. La substitution de id * āfa par
qiyās, que nous traduisons ici par raison, n’a du reste pas
manqué d’être relevée par des lecteurs attentifs du traité de
Thābit, comme en témoigne l’inscription de la phrase suivante
dans la marge de l’un des trois manuscrits disponibles, de la
main même d’al-Sijzı̄:
La relation (id
* āfa) est une notion qui existe pour deux choses [au-dessus
de la ligne: des choses] dont les quiddités sont en raison (bi-al-qiyās)
d’autres (choses).8
6
On the Sector-Figure, éd. Lorch, p. 168.
7
Voir l’édition du livre V par John William Engro# Jr., The Arabic Tradition
of Euclid’s Elements: Book V, PhD Dissertation, Harvard University, 1980, p. 60.
Richard Lorch rend compte de quelques variations mineures observées sur des
manuscrits autres que les dix examinés par Engro#, mais elles ne portent pas sur
les points dont il va être question ici (On the Sector-Figure, pp. 11–13).
8
MS Paris, BN 2457, fol. 60v; nous restituons ici une phrase que Richard Lorch
n’était pas parvenu à saisir dans son intégralité (On the Sector-Figure, p. 169), en
remerciant Roshdi Rashed pour son aide.
180 PASCAL CROZET
Pour mieux juger de l’inflexion apportée ici par Thābit à la
définition euclidienne, citons le commentaire que celui-ci en
donne lui-même:
Ce qu’il [Euclide] a voulu dire par une certaine raison est la raison selon
la partie.9 Il a voulu dire, par homogènes entre elles, que les deux
grandeurs sont du même genre, n’étant pas, par exemple, l’une une
droite et l’autre un plan, ou l’une un plan et l’autre un solide. Il a voulu
dire, par selon la mesure, que la raison des deux grandeurs, de l’une à
l’autre, est selon leur mesure et non selon autre chose qui ressortirait à
leurs états.10
Il poursuit en commentant ce qu’il en entend par ‘‘selon la
partie’’, ce qui éclaire d’une certaine manière, on le verra, le
choix du mot mesure (misāh * a) dans sa traduction:
Cet énoncé, expliqué et démontré, devient donc le suivant: le rapport est
une raison selon la partie, pour deux grandeurs de même genre, de l’une
à l’autre, selon leurs mesures. Il est nécessaire, dans cet énoncé, de faire
en sorte que la raison soit selon la partie, car si l’on ne faisait pas cela,
il entrerait dans cet énoncé quelque chose qui n’est pas un rapport, à
savoir des raisons générales qui recouvrent d’autres raisons, comme la
raison d’une droite à une droite selon laquelle l’une serait plus grande
ou plus petite que l’autre: il s’agit d’une raison de deux grandeurs de
même genre, de l’une à l’autre selon leurs mesures, sauf qu’elle recouvre
de nombreuses raisons selon la partie, à savoir le double, le triple, la
moitié, le tiers et ce qui est similaire à cela. Quant au double, au triple,
à la moitié, au tiers et à ce qui leur est semblable parmi les rapports,
il s’agit d’une raison selon la partie spécifique, qui ne recouvre pas
d’autres raisons.11
Thābit n’évoque pas dans son traité la définition V-5 des
Éléments, qui fournit un critère d’égalité des rapports à l’aide
des équimultiples des grandeurs qui les déterminent. On sait
que cette dernière définition a été à l’origine d’importants
commentaires de la part de mathématiciens comme al-Māhānı̄,
al-Nayrı̄zı̄, Ibn al-Haytham ou al-Khayyām, qui lui ont souvent
préféré la définition anthyphérétique du rapport, à savoir sa
caractérisation par la suite des quotients obtenus grâce à
l’algorithme d’Euclide tel qu’il est exposé dans les propositions
2 à 4 du livre X des Éléments. Comme plusieurs historiens
l’ont déjà souligné,12 la définition anthyphérétique présente
9
Nous traduisons ainsi le mot juz’ı̄.
10
On the Sector-Figure, éd. Lorch, pp. 168–70.
11
Ibid., p. 170.
12
Voir Bijan Vahabzadeh, Trois commentaires arabes sur les concepts de rapport
& de proportionnalité, thèse de doctorat, Université Paris 7 – Denis Diderot, 1998,
p. xix.
THA
l’avantage de caractériser de façon indépendante chacun des
rapports d’une proportion – ce que ne peut faire à l’évidence la
définition V-5 – et ainsi de ‘‘donner un sens déterminé au
concept de rapport en soi, tel qu’il est défini en Euclide, Déf.
V-3’’.13 Or en insistant sur les parties des grandeurs, qui sont,
rappelons-le, d’après la définition V-1, des grandeurs plus
petites qui les mesurent, et en suggérant dès lors que le rapport
constitue une sorte de généralisation des notions de double, de
triple, de moitié et de tiers, Thābit nous semble précisément
donner à la définition V-3 un contenu e#ectif qui reste absent
en tant que tel de l’énoncé euclidien. Dans ce sens, il annonce
les travaux de ses successeurs sur la définition anthyphéré-
tique: le recours aux parties des grandeurs pour expliciter la
définition V-3 n’oriente-t-il pas du reste les recherches dans
cette direction puisque l’algorithme d’Euclide traite précisé-
ment de la mesure commune éventuelle de deux grandeurs,
autrement dit d’une troisième grandeur qui est une partie de
l’une comme de l’autre? Le témoignage d’al-Māhānı̄ lui-même
semble aller dans le sens d’une telle interprétation. Celui-ci
écrit en e#et au début de son traité Sur la difficulté relative à la
question du rapport:
Je me suis e#orcé à ce sujet de démontrer les multiples qu’Euclide a
produits au début du cinquième Livre à propos des grandeurs dont le
rapport est le même, ainsi qu’à propos du rapport plus grand, selon ce
qu’en avait prescrit Thābit ibn Qurra; à savoir que la perception du
rapport des grandeurs et de leur proportionnalité suivant les préceptes
est une chose que l’on pourra saisir en reconnaissant le rapport d’une
manière numérique et en partant des propositions qui se trouvent au
début du dixième Livre.14
S’il est pour l’heure di$cile de savoir à quel texte précis de
Thābit pourrait se référer ici al-Māhānı̄, il nous faut examiner
en quoi les ‘‘prescriptions’’ rapportées par ce dernier s’accor-
dent avec le commentaire qui nous occupe. Puisqu’il est clair
que ‘‘les propositions qui se trouvent au début du dixième
Livre’’ désignent ici l’algorithme d’Euclide, il nous reste donc
à comprendre plus précisément ce qu’al-Māhānı̄ entend par
‘‘manière numérique’’. Or la définition que celui-ci donne du
rapport, qui indique comment ‘‘percevoir’’ cette notion, est la
suivante:
13
Ibid.
14
Ibid., p. 3.
182 PASCAL CROZET
Le rapport entre deux grandeurs homogènes quelconques – ainsi
qu’entre deux nombres quelconques – est l’état qui a lieu pour chacune
d’elles lorsqu’elle mesure son conjoint ou que son conjoint la mesure.
Cette définition, pour n’être pas identique à la définition V-3
des Éléments, semble bien destinée à s’y substituer, à ceci près
qu’elle contient à l’opposé de celle-ci une détermination de
la notion de rapport, à l’instar de l’interprétation de Thābit
qui identifiait ‘‘une certaine raison’’ et ‘‘une raison selon la
partie’’. Ainsi, pour al-Māhānı̄, comme le remarque Bijan
Vahabzadeh, ‘‘le rapport exprime la mesure de l’une des deux
grandeurs par l’autre’’.15 Si l’on s’en tient à une stricte inter-
prétation du mot mesure (ici taqdı̄r), il s’agit donc de l’état
qui a lieu pour chacune d’elles lorsqu’elle est une partie de
son conjoint ou que son conjoint est une partie d’elle, ce qui,
ainsi formulé, ne peut que souligner une identité de vues entre
les deux mathématiciens. Mais al-Māhānı̄ développe ensuite
ce qu’il entend par mesure de l’une par l’autre d’une façon
qui n’est pas aussi restrictive et correspond sans aucun
doute, selon nous, à ce qu’entend Thābit par ‘‘selon la partie’’.
En e#et, selon al-Māhānı̄, cette mesure ‘‘a lieu selon trois
espèces’’:
– soit ‘‘la plus petite, lorsqu’elle mesure la plus grande,
l’épuise complètement par la mesure de sorte que rien n’en
reste’’;
– soit les deux grandeurs sont commensurables, ce que
l’auteur, sans utiliser ce dernier terme, traduit en décrivant
un algorithme d’Euclide qui aboutit à l’évanouissement du
dernier reste;
– soit l’algorithme continue indéfiniment, les deux grandeurs
étant dès lors incommensurables (le terme n’est pas non plus
utilisé).
Et, ajoute al-Māhānı̄, ‘‘les deux premières espèces de la
mesure ont lieu dans les nombres ainsi que dans les grandeurs,
mais la troisième uniquement dans les grandeurs’’. La
‘‘manière numérique’’ évoquée par le mathématicien ne semble
donc pas tant renvoyer ici à la notion de rapport elle-même qu’à
la façon d’appréhender les grandeurs pour définir le rapport:
par l’introduction d’une troisième espèce, on généralise les
résultats que l’on obtient sur les nombres, en suggérant que les
rapports quelconques ne sont qu’une extension des rapports
15
Ibid., p. xxv.
THA
exprimables numériquement et que l’on peut user du mot mesure
même si cette mesure fait apparaı̂tre un processus infini.
L’idée qu’il y a bien là un même concept qui ne fait que
s’étendre des nombres aux grandeurs nous semble précisément
soulignée par Thābit dans la dernière partie de son commen-
taire sur la notion de rapport. Celui-ci écrit en e#et:
Dans cette définition, par laquelle Euclide a défini le rapport, n’entrent
pas les rapports de nombres, puisqu’il a établi le rapport pour deux
grandeurs, que le nom de grandeur, pour lui, ne se dit pas pour le
nombre, et qu’il a établi la raison de l’une à l’autre selon la mesure alors
que les nombres ne sont pas munis de mesure.16 Or nous n’avons pas
trouvé, là où il a utilisé ce terme dans ses livres, qu’il se soit restreint à
cette acception, mais qu’il l’a également utilisé pour les rapports
d’angles, de nombres, de mouvements et d’autres choses selon ce qui est
habituel. Quant à nous, nous ne voulons traiter ici que des rapports
de grandeurs même si, pour tout ce que nous décrivons se rapportant
aux grandeurs, quelque chose de semblable peut en découler pour les
nombres. Si tu veux connaître cela, à chaque fois que nous disons
grandeur, il convient donc que tu comprennes en plus de cela ou
nombre.17
Mais le texte de Thābit donne à la ‘‘manière numérique’’ de
traiter les grandeurs, évoquée par son contemporain, une
dimension supplémentaire sur deux plans qui, nous le verrons,
sont liés. Le premier a trait à l’introduction du mot mesure
(misāh* a) dans la définition. Pour Thābit en e#et, comme il
l’expliquera à propos de la multiplication des grandeurs, on
connaît la mesure d’une grandeur par la médiation d’une unité
de mesure dont l’assemblage un certain ‘‘nombre’’ (‘adad) de
fois constitue cette grandeur (voir plus loin la citation com-
plète). Une telle formulation suggère que ce nombre est un
nombre entier et donc que l’unité de mesure est une partie de la
grandeur. Mais on ne peut que penser qu’il s’agit là d’une
formulation analogue à celle dont use al-Māhānı̄ lorsqu’il
définit le rapport comme l’état de deux grandeurs lorsque l’une
mesure l’autre, à savoir que l’idée introduite dépasse le cadre
strict de son énonciation: Thābit semble ici donner à nombre un
sens plus étendu que celui de nombre entier, de la même
manière qu’al-Māhānı̄ donne à mesure un sens plus étendu que
le simple fait pour une grandeur d’être une partie d’une autre
16
Comme dans la définition du rapport, Thābit utilise ici le mot misāh
* a, qui se
rapporte à la mesure des étendues et correspond donc à une longueur, une aire ou
un volume.
17
184 PASCAL CROZET
(l’idée générale est au demeurant la même chez les deux
auteurs). Dans ce contexte, l’emploi du mot misāh * a dans la
définition du rapport laisse entendre que la médiation d’une
unité de grandeur permet d’y envisager les grandeurs comme si
elles étaient caractérisées par des nombres. Ce qui nous amène
au second aspect de cette ‘‘manière numérique’’ de traiter les
grandeurs: dès lors qu’il était notamment question de compo-
sition de rapports, agir sur les rapports de grandeurs comme s’il
s’agissait de rapports numériques ne pouvait que conduire
Thābit à substituer à la proposition VI-23 des Éléments un
équivalent de la proposition VIII-5, et par conséquent à définir
une multiplication des grandeurs qui renforcera l’analogie
avec les nombres; nous y reviendrons.
Pour conclure sur la notion de rapport telle qu’explicitée par
le mathématicien, il faut relever que, comme il l’explique
lui-même en a$rmant qu’il ne veut traiter que du rapport des
grandeurs, Thābit tient ici à rester dans un cadre euclidien.
Mais ce cadre est surtout formel et, s’il semble respecter la
lettre de la définition de son prédécesseur malgré quelques
inflexions, son commentaire n’en respecte guère l’esprit et
annonce les développements ultérieurs.
La composition des rapports

La situation est inverse pour la composition des rapports, dont
Thābit donne une définition explicite parfaitement en accord
avec l’usage qu’en fait Euclide dans la proposition VI-23.
Contrairement à al-Khayyām, il ne fait en particulier aucune
référence à la définition apocryphe VI-5: s’il y a bien chez lui
une ‘‘manière numérique’’ de traiter les grandeurs pour définir
les rapports et si la composition de ceux-ci constitue bien
une opération, les rapports n’en sont pas pour autant des
nombres, pas plus que la composition des rapports n’est une
multiplication.
Dans un style qui demeure donc tout à fait euclidien, Thābit
commence par ajouter au livre V quelques définitions: tout
d’abord l’antécédent et le conséquent dans un rapport, termes
utilisés dans les Éléments mais dépourvus dans cet ouvrage de
définition quoique d’un usage dénué d’ambiguïté, puis, surtout,
la liaison de deux rapports:
– deux rapports sont dits liés (muttas*ilān) s’ils ont une
grandeur en commun;
THA
– ils sont liés par continuité (‘alā al-wilā’) si l’antécédent de
l’une est le conséquent de l’autre;
– ils sont liés par non-continuité (‘alā ghayr al-wilā’) dans le
cas contraire (identité des deux antécédents ou des deux
conséquents).
La composition de deux rapports consiste alors en leur liaison
par continuité, le rapport composé étant celui que l’on obtient
entre les extrémités une fois éliminée la grandeur commune.
Thābit n’explique pas comment on lie deux rapports par
continuité, c’est-à-dire, étant donné deux rapports, comment on
peut trouver deux autres rapports qui leur soient égaux et qui
soient liés par continuité. Il est clair cependant, parce qu’il
utilise par la suite abondamment le procédé, qu’il pense à
l’usage d’une quatrième proportionnelle. On retrouve donc là,
comme chez Euclide, un même postulat implicite d’existence
(l’existence d’une quatrième proportionnelle est démontrée
pour les seuls segments de droites dans Éléments VI-12, mais est
utilisée par Euclide dès le livre V). En outre, pour que la
définition soit consistante, il faudrait s’assurer, ce que Thābit
n’évoque pas non plus, que tous les procédés utilisés
pour lier les rapports donnent bien le même rapport composé:
A C A D C B
soit à composer et , posons par exemple = et = ; le
B D B E D F
A C
rapport composé J , tel que défini par Thābit, peut en e#et
B D
A C 18
être aussi bien que . Enfin, seront postulées implicitement
F E
aussi bien l’associativité que la commutativité de ce qui est
conçu comme une opération sur les objets mathématiques dont
il vient de commenter la définition. L’opération inverse est du
18
La proposition V-22 permet de s’assurer de façon immédiate qu’on obtient
bien le même résultat avec di#érents représentants des rapports, mais dans la
seule mesure où l’on élimine une grandeur qui est toujours de même rang (par
exemple le conséquent du premier rapport), ce qui n’est pas le cas dans notre
exemple. Supposons en e#et que A / B = X / Y et C / D = Y / Z et que d’autre part
A / B = X / Y et C / D = Y / Z ; la proposition V-22 assure bien que X / Z = X / Z ;
tous les procédés où l’on élimine le conséquent du premier rapport fournissent
donc le même résultat; de même, tous les procédés où l’on élimine l’antécédent
fournissent le même résultat. La généralité introduite par Thābit, pour qui la
grandeur en commun n’est pas forcément le conséquent du premier rapport,
empêche une telle application immédiate. On peut néanmoins, en reprenant notre
exemple, démontrer que A / F = C / E de la façon suivante: on pose G telle que
A / B = G / C; puisque B / F = C / D, on aura donc A / F = G / D par application de
la proposition V-22; d’autre part, puisque A / B = D / E, on aura D / E = G / C; en
appliquant la proposition V-16, on aura donc C / E = G / D; et puisque les rapports
A / F et C / E sont égaux au même rapport G / D, ils sont égaux entre eux.
186 PASCAL CROZET
reste également définie et présentée comme telle: la séparation
d’un rapport d’un autre rapport sera ainsi leur liaison par
non-continuité. Un vocabulaire spécifique est enfin introduit:
la composition des rapports met en relation un composé
(mu’allaf) et des composantes (mu’allaf minhu).
Il s’agit ainsi surtout, pour le mathématicien, de préciser et
de mettre des mots sur une conception de la composition des
rapports qui, pour le moment tout du moins, reste fondamen-
talement euclidienne.
La multiplication des grandeurs

L’utilisation du terme d * arb (multiplier) pour désigner une
opération sur des grandeurs est chose courante chez les mathé-
maticiens arabes puisqu’on la rencontre aussi bien chez les
maîtres de Thābit, les Banū Mūsā,19 que chez ses successeurs
géomètres. Dans le cas de deux droites par exemple, le mot d * arb
a le même sens que le mot sat*h * (littéralement: plan), qui vient
rappeler une certaine orthodoxie euclidienne selon laquelle le
‘‘produit’’ d’une grandeur par une autre grandeur est constitué
du rectangle qu’elles entourent; l’équivalence des deux termes
et l’absence usuelle d’un tel rectangle, tant dans le texte que
dans les figures, montrent sans aucun doute une tendance à
l’arithmétisation du traitement des grandeurs.
Plus rares sont les tentatives de définir plus précisément ce
que pourrait être, dans un cas général, la multiplication de
grandeurs, ce qui est tout l’intérêt du passage que Thābit
consacre à ce sujet et que nous avons évoqué plus haut. Celui-ci
écrit en e#et:
Ce que j’entends en disant multiplier une grandeur par une grandeur est
prendre cette grandeur un nombre de fois KégalL au nombre des gran-
deurs dont l’assemblage constitue l’autre grandeur, et par lesquelles on
connaît sa mesure.20
19
Voir Roshdi Rashed, Les Mathématiques infinitésimales du IXe au XIe siècle.
Vol. I: Fondateurs et commentateurs (Londres, 1996).
20
THA
Si A et B sont de même genre, ce qui nous semble être la
situation privilégiée ici par le mathématicien (toutes les gran-
deurs dont il est question dans le traité sont du reste, comme il
est d’usage, représentées par des segments de droite et on peut
ramener tout rapport à un rapport de tels objets), on pourra
se donner une unité de grandeur U et poser A = aU et B = bU.
On aura alors, en respectant la règle d’homogénéité,21
AB = BA = ab(UU). Mieux encore, on peut définir la
division comme opération inverse, ce que Thābit fait implicite-
ment puisqu’il utilise plus loin la division d’une grandeur par
une autre.
Pour mieux apprécier le sens de cette définition aussi bien
que son rôle dans le traité de Thābit, il nous faut relever l’usage
que fera plus loin le mathématicien de la multiplication des
grandeurs. Celle-ci interviendra dans le troisième chapitre
exclusivement, dans la formulation de deux résultats non
démontrés en tant que tels par l’auteur mais issus directement
des Éléments: en premier lieu le fait que le rapport composé de
deux rapports est égal au rapport du produit des antécédents
au produit des conséquents, à savoir
A C AC
J ≠ ,
B D BD
et en second lieu l’équivalence.
A C
≠ k AD≠B C.
B D
Thābit aurait pu se satisfaire, comme avant lui les Banū Mūsā
ou après lui bien de ses successeurs géomètres, d’utiliser le mot
produit sans l’expliciter, dans un sens qui lui aurait permis de
rattacher ces deux résultats respectivement aux propositions
euclidiennes VI-23 et VI-16, où ils sont exprimés en termes
de rectangles entourés par des droites: rien en e#et n’eût été
changé dans ce chapitre, pas même son caractère algébrique. Si
Thābit prend soin de définir ainsi la multiplication des gran-
deurs, ce ne peut donc être qu’en relation avec la façon qu’il a
eu de traiter les grandeurs pour définir les rapports. Or l’emploi
du mot misāh * a dans la définition du rapport laisse penser que
la ‘‘dimension’’ des grandeurs (droite, plan ou solide) n’est pas
ici en jeu puisque, une fois posée une unité de grandeur U, on
pourra avoir:
21
Il nous semble que, par ‘‘prendre b fois la grandeur A’’, on doit comprendre:
prendre b fois l’aire AU.
188 PASCAL CROZET
A aU a
≠ ≠ .
B bU b
Ainsi nous semble s’éclairer, et de façon beaucoup plus précise,
à la fois la ‘‘manière numérique’’ de considérer les rapports de
grandeurs évoquée par al-Māhānı̄ à propos de Thābit, et la
façon qu’aura celui-ci de traiter les grandeurs comme des
nombres.22
UNE ÉTUDE COMBINATOIRE

Une fois posées ces définitions, le mathématicien va pouvoir
engager l’étude proprement dite de la composition des rap-
ports. Cette étude est tout entière contenue dans son second
chapitre, le troisième, d’égale importance, étant consacré aux
applications des résultats qui y seront dégagés. Contrairement
à ce dernier chapitre, dans lequel l’utilisation massive de la
multiplication et de la division des grandeurs distinguera de
façon nette la manière de Thābit du style de son prédécesseur
grec, le mathématicien va s’en tenir ici à une stricte orthodoxie
euclidienne, tant pour ce qui touche à la définition de la
composition des rapports, ce que nous avons déjà évoqué, que
pour ce qui relève des moyens utilisés pour mener son étude; ce
qui n’empêchera pas, on le verra, une originalité certaine dans
la façon d’appréhender le sujet.
Il faut remarquer à cet égard que la rédaction d’un tel traité
ne peut avoir été suscitée par une quelconque insatisfaction
quant à la manière dont les prédécesseurs du mathématicien
ont pu définir et appréhender la composition des rapports:
contrairement à al-Khayyām, Thābit met bien ici ses pas dans
ceux de ses devanciers grecs sans remettre en question la
légitimité de leur démarche. Il nous semble plutôt qu’il faille
rechercher les raisons d’une telle rédaction du côté de l’étude
de la figure-secteur, elle-même suscitée par le développement de
l’astronomie. Le traité sur la composition des rapports nous
semble, de ce point de vue, postérieur à celui sur la figure-
secteur; nous verrons en outre qu’il en reprend certains
résultats.
Mais rester dans le cadre légué par Euclide pour étudier la
composition des rapports ne pouvait qu’orienter l’examen
22
Dans un passage du traité, Thābit utilise en outre subrepticement le mot
nombre en lieu et place du mot grandeur (voir note 27).
THA
d’une telle notion vers la question suivante: que nous apprend
sur les grandeurs mises en jeu une relation de composition de
rapports? On comprend mieux dès lors le titre même de ce
chapitre: Sur la connaissance des grandeurs dont les rapports
sont composés les uns avec les autres. Si une telle relation de
composition met en jeu six grandeurs di#érentes A, B, C, D, E,
F:
A C E
≠ J ,
B D F
ses conséquences ne pourront qu’être des relations du même
type; et si certaines de ces grandeurs sont égales entre elles, on
pourra dans certains cas en déduire des relations d’un autre
genre (proportionnalité, égalité d’autres grandeurs, etc.). Le
projet de Thābit va donc être d’explorer de façon méthodique et
exhaustive toutes ces conséquences et, puisque l’on peut de
façon évidente interchanger des grandeurs dans la relation,
son travail va revêtir un aspect combinatoire qui constituera
tout l’intérêt de ce chapitre.
Le mathématicien va traiter successivement les sept cas
possibles où au moins trois grandeurs parmi les six sont a priori
di#érentes:
1. les six grandeurs sont di#érentes;
2. il y a cinq grandeurs di#érentes, l’une d’entre elles
apparaissant deux fois;
3. il y a quatre grandeurs di#érentes, l’une d’entre elles
apparaissant trois fois;
4. il y a quatre grandeurs di#érentes, deux d’entre elles
apparaissant deux fois;
5. il y a trois grandeurs di#érentes, l’une d’entre elles
apparaissant quatre fois;
6. il y a trois grandeurs di#érentes, l’une d’entre elles
apparaissant trois fois et une autre deux fois;
7. il y a trois grandeurs di#érentes, chacune d’elles appa-
raissant deux fois.
Qu’il ne s’intéresse pas au cas où il y a seulement deux
grandeurs di#érentes (ou une seule) semble justifié par la
première phrase du chapitre: ‘‘étant donné trois rapports
quelconques, ils traitent soit de six grandeurs, soit de cinq, soit
190 PASCAL CROZET
de quatre, soit de trois’’.23 Or à l’aide de deux grandeurs A et B,
on ne peut en e#et envisager que deux rapports di#érents
autres que le rapport d’identité et non trois, à savoir A / B et
B / A; il est facile de voir en outre qu’une relation de compo-
sition débouche alors soit sur une tautologie, soit sur l’égalité
de A et de B.
L’exploration de la première configuration, où les six gran-
deurs sont quelconques, avait déjà été réalisée dans son traité
sur la figure-secteur puisqu’elle permet de donner toutes les
formulations possibles du théorème de Ménélaüs.24 Thābit en
reprend ici l’analyse et la plupart des démonstrations, en
annonçant d’emblée qu’en partant de la relation
A C E
≠ J ,
B D F
on peut énoncer dix-sept autres relations de composition. On
voit que ces relations sont obtenues par permutation du sextu-
plet (A, B, C, D, E, F), mais toutes les permutations ne sont pas
admissibles puisque le rapport A / D, par exemple, ne peut être
composé de deux autres rapports où interviendraient les quatre
autres grandeurs. Les dix-sept relations valides qui sont ex-
posées sont toutes celles que l’on obtient en prenant comme
rapport composé (i.e. résultant de l’opération de composition)
le rapport de deux des six grandeurs de sorte que l’antécédent
soit d’un rang initial inférieur à celui du conséquent, ce qui
permettra ensuite au mathématicien d’ajouter les dix-huit
relations obtenues en prenant les rapports inverses.
Pour démontrer successivement la validité des dix-sept rela-
tions, Thābit utilise l’associativité et la commutativité de
l’opération de composition, mais aussi, nous verrons plus loin
de quelle manière, les relations démontrées antérieurement et,
pour deux cas des dix-sept cas, il introduit une quatrième
proportionnelle. L’ordre dans lequel il aborde ces dix-sept
relations est l’ordre lexicographique, selon les rangs initiaux
des grandeurs, interrompu une fois, comme il l’explique lui-
même, pour des raisons internes aux démonstrations.
Bien que Thābit n’explique pas dans ce traité comment il
parvient à ce nombre de dix-sept (dix-huit avec la relation de
23
24
Voir Bellosta, ‘‘Le traité de Thābit ibn Qurra sur la figure secteur’’, pp. 149
sqq.
THA
départ) et qu’il ne fasse apparaître aucune opération de dénom-
brement autre que l’énumération ordonnée des relations qu’il
se propose de démontrer, il nous semble que son analyse est
bien d’ordre combinatoire dans la mesure où ce qu’il démontre
n’est pas tant une relation particulière qu’une possibilité de
permutation entre les grandeurs. Expliquons-nous en repre-
nant la démonstration de la cinquième relation. Il s’agit de
démontrer que si
A C E
≠ J ,
B D F
alors
A B C
≠ J .
E F D
Or il a déjà démontré (il s’agit respectivement de la qua-
trième et de la première relation) que
A B C
≠ J ,
E D F
et que
A C E
≠ J .
B F D
Il su$t alors, dans cette dernière relation, de remplacer A
par A, B par E, C par B, D par D, E par C et F par F, autrement
dit substituer aux grandeurs de la relation d’origine les gran-
deurs qui occupent le même rang dans la quatrième relation,
pour obtenir le résultat recherché, à savoir
A B C
≠ J ,
E F D
cette substitution étant simplement exprimée par Thābit de la

façon suivante:
La première devient alors A, la deuxième E, la troisième B, la quatrième
D, la cinquième C et la sixième F.25
25
192 PASCAL CROZET
Autrement dit, en partant des permutations admissibles 1 et
4, il démontre que la permutation 5 est également admissible
en démontrant que 5 = 1 + 4.
On peut alors résumer ainsi la tâche que se propose le
mathématicien dans cette première partie du second chapitre:
dresser la liste, dans la mesure du possible en suivant l’ordre
lexicographique, des images du sextuplet (A, B, C, D, E, F) par
toutes les permutations admissibles, l’admissibilité étant dé-
montrée, lorsque c’est possible, en composant des permutations
admissibles.26
À l’issue de ces démonstrations et après avoir remarqué, au
moyen d’un corollaire obtenu en cours de route (‘‘pour tout
rapport composé de deux rapports, l’inverse de ce rapport-ci est
composé de l’inverse de ces deux rapports-là’’), que dix-huit
autres relations peuvent être obtenues par passage à l’inverse,
il résume l’ensemble de ses résultats dans un premier tableau:
septième sixième cinquième quatrième troisième deuxième première

origine F E D C B A
1 D E F C B A
2 F E D B C A
3 D E F B C A
0 0 0 0 0 D A
4 F C D B E A
5 D C F B E A
0 0 0 0 0 F A
0 0 0 0 0 C B
8 E F C A D B
9 C F E A D B
0 0 0 0 0 E B
10 E D C A F B
11 C D E A F B
6 E F B A D C
12 B F E A D C
0 0 0 0 0 E C
13 E D B A F C
14 B D E A F C
15 F C A B E D
26
En s’appuyant sur une interprétation similaire mettant en évidence la
composition des permutations, Sabine Rommevaux a donné le détail des dix-sept
démonstrations de Thābit telles qu’elles sont exposées non pas dans ce traité,
mais dans le traité connexe sur la figure-secteur (Sabine Koelblen, ‘‘Une pratique
de la composition des raisons dans un exercice de combinatoire’’, Revue d’histoire
des sciences, XLVII, 2 (1994), pp. 209–47; voir en particulier pp. 222–6).
THA
septième sixième cinquième quatrième troisième deuxième première
16 A C F B E D
0 0 0 0 0 F D
7 C D B A F E
17 B D C A F E
En transposant le commentaire qu’en donne le mathématicien,

si l’on note [1], [2], [3], [4], [5] et [6] les contenus respectifs des
six colonnes de droite pour une ligne donnée, on aura:
[1] [3] [5]
≠ J .
[2] [4] [6]
Il est remarquable que, par souci d’exhaustivité, Thābit tienne
à faire apparaître dans son tableau les rapports pour lesquels
on ne peut déduire aucune composition, comme par exemple le
rapport de A à D, le mathématicien plaçant alors un zéro dans
les colonnes adjacentes. Quant à la septième colonne, elle
indique le numéro de la démonstration.
Thābit introduit ensuite la notion la plus originale de son
étude, la notion de champ (h * ayyiz):
Il apparaît là clairement que, si nous partageons les six grandeurs en
deux groupes, que nous posons la première d’entre elles, la quatrième et
la sixième dans un groupe, et que nous posons les KgrandeursL restantes,
soit la deuxième, la troisième et la cinquième, dans un autre groupe,
alors le rapport de toute grandeur se trouvant dans l’un des deux
groupes à une grandeur de l’autre groupe, quelle qu’elle soit, est
composé des rapports des grandeurs restantes. Quant au rapport Kd’une
grandeurL à ce qui se trouve dans son groupe, il n’est pas composé des
rapports restants*. Que chacun des groupes soit appelé un champ, que le
groupe dans lequel se trouve le premier nombre27 soit appelé le premier
champ, et que l’autre groupe soit appelé le deuxième champ.28
Et, pour mieux marquer l’imagination de son lecteur, il propose
en outre le schéma suivant:
* Thābit ne démontre pas ce résultat dans ce traité. Mais, formulé autrement, il

avait fait l’objet d’une démonstration remarquable dans son traité sur la figure-
secteur (voir l’article d’Hélène Bellosta dans le précédent numéro).
27
Remarquons l’usage subreptice du mot nombre pour désigner une grandeur,
qui s’accorde avec le traitement numérique des grandeurs déjà évoqué.
28
194 PASCAL CROZET
Il s’agit ainsi d’une notion ensembliste avant la lettre, tant

dans sa définition que dans sa représentation, destinée avant
tout à assurer une certaine interchangeabilité des grandeurs
dans la relation de composition. On en trouvera une utilité
majeure dans le troisième chapitre lorsque, précisément, il
s’agira de se soustraire à l’aspect énumératif qui empreint celui
qui nous occupe pour donner des résultats qui fassent abstrac-
tion de la place des grandeurs. Mieux encore, cette notion
permettra de penser la relation de composition comme une
relation entre deux ensembles (par exemple {A, D, F} et {B, C,
E}) et ainsi de formuler des conclusions empreintes de la plus
grande généralité. Le résultat fondamental sur lequel le mathé-
maticien s’appuiera alors dans ses démonstrations, est que si
A C E
≠ J ,
B D F
alors
ADF = BCE,
autrement dit le produit des grandeurs du premier champ est
égal au produit des grandeurs du second, ce qu’il exprimera
dans ce troisième chapitre dans des termes voisins:
Ce qui vient du produit d’une des grandeurs qui sont dans le premier
champ par une autre grandeur parmi elles, et de ce que l’on a obtenu par
la troisième, est égal à ce qui vient du produit d’une des grandeurs qui
sont dans le deuxième champ, quelle qu’elle soit, par une autre grandeur
parmi elles, et de ce que l’on a obtenu par la troisième.29
L’interchangeabilité des grandeurs au sein d’un même champ
(et donc la notion même de champ) est alors le pendant de la
commutativité et de l’associativité de la multiplication des
grandeurs. Comme nous le verrons, le concept de champ permet-
tra en outre au mathématicien de distinguer facilement entre
les di#érents cas rencontrés, selon que les grandeurs considé-
rées dans ses problèmes appartiennent ou non au même champ.
Cette notion est bien ici le résultat d’une méthode proprement
29
Ibid., p. 224.
THA
combinatoire; elle ouvre la voie en particulier à un dénombre-
ment de toutes les relations de composition possibles en utili-
sant un produit d’arrangements, ce qui sera la voie explorée plus
* ūsı̄30: on a bien 36 = A33 A33.
tard par Nas*ı̄r al-Dı̄n al-T
Avant d’arriver au troisième chapitre, il reste cependant à
Thābit à poursuivre sa tâche pour épuiser toutes les consé-
quences d’une composition des rapports, à savoir envisager les
cas où deux grandeurs au moins parmi les six sont égales.
Si deux grandeurs parmi les six sont égales et si ces deux
grandeurs ne sont pas dans le même champ, il résultera de la
composition des rapports
A C E
≠ J ,
B D F
une relation de proportionnalité entre les quatre grandeurs
restantes: si par exemple A = C, les grandeurs D et B sont
proportionnelles aux grandeurs E et F. Il su$ra alors pour
Thābit de faire la démonstration pour un cas puis, en
s’appuyant sur le tableau précédent, de fournir un deuxième
tableau exhaustif recensant tous les cas possibles.
Les autres configurations sont traitées de manière similaire,
s’appuyant sur les précédentes pour aboutir à un tableau
exhaustif. Reproduisons à titre d’exemple le début du tableau
consacré au cas où deux grandeurs parmi les six sont égales
entre elles (nommées la première et la deuxième dans ce qui
suit), ainsi que deux autres parmi les quatre restantes (la
troisième et la quatrième). Plusieurs conclusions peuvent alors
être tirées de la composition des rapports
A C E
≠ J :
B D F
– soit les deux dernières grandeurs (la cinquième et la
sixième) sont égales (mention égalité dans la septième co-
lonne); par exemple si A = B et C = D, alors E = F;
– soit le rapport de la cinquième grandeur à la troisième est
égal au rapport de la troisième à la sixième (mention pro-
portion aux deux autres grandeurs); par exemple si A = B et
D E
C = E, alors ≠ ;
E F
30
Alexandre Pacha Caratheodory, Traité du quadrilatère attribué à
Nassiruddin-el-Toussy (Constantinople, 1891), pp. 12–14 du texte arabe, pp. 13–16
de la traduction française.
196 PASCAL CROZET
– soit le rapport de la cinquième à la première est égal au
rapport de la première à la sixième (mention proportion aux
deux premières grandeurs); par exemple si A = D et B = F,
E A
alors ≠ ;
A C
– soit le rapport de la cinquième à la sixième est égal au
rapport doublé de la première à la troisième (mention:
doublé par répétition); par exemple si A = D et B = C, alors
E A A
≠ J .
F B B
Quant à la huitième colonne du tableau, elle est destinée à
des renvois aux tableaux précédents selon les di#érents types
de conclusion.31
Les cas et Ce qui résulte, Les deux grandeurs Les deux autres Les deux
les items de l’égalité, de restantes grandeurs égales premières
par la proportion entre elles grandeurs égales
lesquels ou du entre elles
cela est doublement du
démontré rapport
huitième septième sixième cinquième quatrième troisième deuxième première
1 égalité F E D C B A
1 proportion aux F D E C B A
deux autres
1 égalité E D F C B A
1 égalité F C E D B A
1 proportion aux E C F D B A
deux autres
1 égalité D C F E B A
2 égalité F E D B C A
2 proportion aux F D E B C A
deux autres
2 égalité E D F B C A
2 égalité F B E D C A
2 proportion aux E B F D C A
deux autres
2 égalité D B F E C A
7 doublé par F E C B D A
répétition
13 doublé par F C E B D A
répétition
5 proportion aux E C F B D A
deux
premières
(. . .) (. . .) (. . .) (. . .) (. . .) (. . .) (. . .) (. . .)
31
Nous n’explicitons pas ces renvois, ce qui serait fastidieux et malcommode
puisque nous n’avons pas donné tous les tableaux, pour un intérêt relativement
limité.
THA
L’étude s’achève une fois transcrit le dernier tableau: avec
exhaustivité, Thābit a ainsi passé en revue toutes les configu-
rations possibles, et toutes les situations sont consignées dans
des tableaux auxquels on peut se référer. Seule la notion de
champ va toutefois lui être utile lorsqu’il s’agira d’envisager la
relation de composition dans le contexte de la résolution de
problèmes.
LES RAPPORTS COMPOSÉS, L’ALGÈBRE ET LES DONNÉES
Comme nous l’avons suggéré, le style du troisième chapitre
tranche profondément avec celui qui le précède même s’il en
constitue la suite naturelle. Annoncé comme traitant des
problèmes résolus à l’aide de la composition des rapports, il se
présente en réalité comme une succession de propositions
bâties selon un même modèle, si l’on excepte quelques-unes
d’entre elles qui constituent avant tout des lemmes pour les
suivantes: on suppose que certaines des grandeurs intervenant
dans une relation de composition de rapports sont connues
ainsi qu’éventuellement des produits, des sommes, des dif-
férences ou des rapports de grandeurs, et il s’agit de montrer
que d’autres objets (d’autres grandeurs, des produits ou des
rapports de grandeurs, etc.) sont eux aussi connus. La forme
prise par ces propositions fait ainsi de ce chapitre une sorte
d’extension du livre des Données d’Euclide qui serait entière-
ment consacrée à la composition des rapports. Du reste, comme
nous allons le voir, un certain nombre de résultats utilisés par
le mathématicien sont directement issus de l’ouvrage de son
prédécesseur, même si, nous le verrons également, il peut en
reprendre la justification pour l’occasion.
Le chapitre comprend vingt-deux propositions numérotées,
précédées d’une proposition préliminaire elle-même accompa-
gnée de corollaires; nous en dressons la liste dans notre annexe
no 1. Sur les trois manuscrits disponibles, le traité s’achève
brutalement au milieu de la proposition 22, qui nous semble
constituer un lemme pour une vingt-troisième proposition que
nous énonçons également. Le fait que la voie dans laquelle
s’était engagé Thābit dans cette démonstration ne semble pas
pouvoir aboutir pourrait bien être relié à l’inachèvement du
traité, tel du moins qu’il nous est parvenu (nous traitons en
détail de la proposition 22 dans notre annexe no 2).
La proposition préliminaire, dans son traitement, fait le lien
entre le style du second chapitre et celui du troisième: elle
198 PASCAL CROZET
comporte encore un aspect énumératif absent des propositions
suivantes et ouvre sur le résultat fondamental cité plus haut
concernant l’égalité du produit des grandeurs du premier
champ et du produit des grandeurs du second; elle débouche en
outre sur le dernier tableau récapitulatif du traité.
Proposition préliminaire: si, dans une relation de composition de rap-
ports, cinq grandeurs sur les six sont connues, alors la sixième
est connue.
Reprenons-en la démonstration, qui illustre sur un cas simple
(les autres propositions pourront faire preuve de beaucoup
plus de technicité) à la fois la manière algébrique de Thābit et
le rapport au livre des Données qui lui est lié. Le mathématicien
pose
A C E
≠ J
B D F
et suppose dans un premier temps que A, B, C, D et E sont

connues; il s’agit donc de montrer que F est connue. Il définit la
grandeur G en posant
D E
≠ ;
G F
on a alors
C C D C E
≠ J ≠ J
G D G D F
et par conséquent
C A
≠ .
G B
Il conclut alors que, puisque A, B et C sont connues, G est aussi

connue. Ce résultat pouvait être assuré par l’application des
propositions 1 et 2 des Données (le rapport qu’ont entre elles des
grandeurs données est donné; et: si une grandeur donnée a un
rapport donné avec une autre grandeur, celle-ci est donnée de
grandeur). Thābit, néanmoins, ne recourt pas à l’ouvrage de
son prédécesseur grec et justifie sa conclusion en traitant G à
la manière d’une inconnue dont il s’agit de découvrir la valeur.
Il écrit en e#et:
THA
si nous multiplions B par C et si nous divisons ce que l’on a obtenu par
A, ce qui vient de la division est connu, et c’est G, puisque les grandeurs
A, B, C et G sont proportionnelles.32
Autrement dit, G est connue puisque
G = (BC)÷A.
Dès lors, puisque pour Thābit le résultat d’opérations sur des

grandeurs connues est toujours connu, le traitement arithmé-
tique des grandeurs ne pouvait que vouer à l’inutilité des pans
entiers de l’ouvrage euclidien. De ce point de vue, peu de textes
soulignent avec autant de force que ce passage de Thābit
combien la démarche algébrique recouvre la démarche analy-
tique qui est le propre des Données; cette démarche algébrique,
que nous illustrerons sur un autre exemple, constitue en outre
l’un des caractères les plus marqués de ce traité.
À titre de comparaison et pour apprécier la distance prise
par le mathématicien arabe, reprenons la démonstration eucli-
dienne de la proposition 2 des Données. Supposons que A soit
donnée et que le rapport A / B soit donné. Euclide assure qu’on
peut trouver une grandeur C égale à A et un rapport C / D égal
au rapport A / B. Par permutation, on aura alors A / C égal à
B / D et donc B égal à D; la grandeur B est donc donnée
puisqu’on a trouvé son égale (par définition, une grandeur est
dite donnée de grandeur lorsqu’on peut trouver des grandeurs
qui lui sont égales). Si la démonstration euclidienne est bien
valide puisque s’accordant à la définition même d’une grandeur
donnée, on constate qu’elle est surtout très formelle et notam-
ment que, contrairement à la solution de Thābit, elle n’apprend
rien sur la grandeur B et n’o#re pas de voie vers sa détermina-
tion. On comprend dès lors que celui-ci ait pu la considérer
comme insatisfaisante et qu’il ait opté délibérément pour une
voie algébrique.
Mais revenons à la démonstration de la proposition prélimi-
naire de Thābit. Une fois assuré que G est connue, le mathé-
maticien écrit de même que
F = (GE)÷D,
et donc F est connue. Il passe ensuite, avec les mêmes moyens,

32
200 PASCAL CROZET
à la démonstration des cinq autres cas (selon les cinq gran-
deurs connues parmi les six) et, après avoir remarqué que la
séparation et la composition de deux rapports connus forment
dès lors des rapports connus, il résume ses résultats en présen-
tant le tableau suivant:
La sixième Les cinq grandeurs connues

grandeur
recherchée
sixième cinquième quatrième troisième deuxième première
F D E A C B
E C F B D A
D F C A E B
C E D B F A
B C A E F D
A D B F E C
Si l’on note comme précédemment [1], [2], [3], [4], [5] et [6] les
contenus respectifs des six colonnes pour une ligne donnée,
le commentaire récapitulatif du mathématicien tient en la
formule33:
[1][2]
[6] ≠
s [3] d
[4]
.
[5]
Il donne ensuite une démonstration alternative fondée sur
l’égalité
C E CE
J ≠
D F DF
A CE
(trois des termes de la proportion ≠ étant connus, le
B DF
quatrième l’est aussi, ce qui, moyennant discussion, permet
d’aboutir au résultat), puis il conclut ses préliminaires en
formulant l’égalité des produits des grandeurs des deux
champs.
Les vingt-deux propositions qui suivent, d’une complexité
croissante mais s’appuyant sur celles qui les précèdent, vont
33
Contrairement au reste de notre article, la barre horizontale indique ici une
division et non un rapport.
THA
présenter les mêmes caractéristiques: formulation bien souvent
non explicitée de la relation de composition (cette relation est
avant tout pensée comme une relation entre deux ensembles, à
savoir les deux champs, et non comme la composition de deux
rapports déterminés) et donc souci de la plus grande généralité
possible; influence des techniques algébriques; utilisation de la
notion de champ pour distinguer au besoin les di#érents cas
possibles en se soustrayant à une énumération fastidieuse.
Pour illustrer la façon dont ces trois éléments peuvent s’arti-
culer, nous allons prendre plusieurs exemples. Il faut noter au
préalable que, parmi les vingt-deux propositions, certaines sont
des lemmes techniques qui ne mettent pas en œuvre des
rapports composés et sortent donc d’une certaine façon du
cadre strict de cette étude: il s’agit pour l’essentiel (voir le cas
des propositions 11, 13, 18, 20 et 22) de propositions de nature
algébrique mettant en jeu des segments sur une même ligne
droite; nous renvoyons à notre annexe no 2 pour l’exemple
détaillé de la proposition 22.
Notre premier exemple, composé des propositions 1 et 2,
illustre à la fois l’usage de la notion de champ et le souci de
généralité manifesté par l’auteur.
Proposition 1: Si, pour six grandeurs, le rapport de deux d’entre elles est
composé de deux des rapports des grandeurs restantes et si
quatre d’entre elles sont connues, alors soit le produit soit le
rapport des deux grandeurs restantes est connu.
Supposons que les six grandeurs A, B, C, D, E et F soient toutes
liées par une relation de composition de rapports (quelle
qu’elle soit), et supposons que A, B, C et D soient connues. Les
grandeurs E et F sont soit dans deux champs di#érents, soit
dans le même champ.
Si elles sont dans deux champs di#érents, alors le rapport de
E à F est composé de deux des rapports des grandeurs A, B, C
et D. Ces deux rapports étant connus puisque ces quatre
grandeurs sont connues, le rapport de E à F est donc connu.
Si elles sont dans le même champ, alors les grandeurs qui
sont dans l’autre champ sont toutes connues, et la troisième
grandeur qui est dans le même champ que E et F est connue.
Puisque le produit EF est égal au produit des trois grandeurs
de l’autre champ divisé par cette troisième grandeur, ce produit
est connu.
202 PASCAL CROZET
Ce qu’il fallait démontrer.
Proposition 2: Si, pour six grandeurs, le rapport de deux d’entre elles est
composé de deux des rapports des grandeurs restantes, si quatre
d’entre elles sont connues et si la somme des deux grandeurs
restantes est connue, alors ces deux grandeurs restantes sont
connues.
Supposons que
A C E
≠ J
B D F
et que quatre grandeurs parmi les six soient connues. Les deux
grandeurs restantes sont soit issues de deux champs di#érents,
soit d’un même champ.
Si elles sont issues de champs di#érents, considérons qu’il
s’agit de A et B, ce qui ne nuit pas à la généralité. Alors,
A
d’après la proposition précédente, est connu et donc les deux
B
A+B A+B
rapports et sont connus; et puisque A + B est
B A
connu, A et B sont connus.
Si elles sont issues du même champ, considérons qu’il s’agit
de A et D. D’après la proposition précédente, le produit AD
est connu, et puisque A + D est connu, A et D sont connus.34
Bien qu’il ait fallu en cours de route distinguer deux cas en
faisant intervenir la notion de champ, la proposition 2 propose
donc un énoncé général qui n’en distingue aucun. On comprend
alors l’intérêt pour Thābit de ne pas faire disparaître la forme
d’énoncé proposée par les Données et d’en utiliser au contraire
le modèle pour concevoir son troisième chapitre: si l’algèbre
peut permettre, dans le cas de cette proposition, de donner les
expressions des deux grandeurs recherchées (ce que Thābit ne
fait pas du reste), ces expressions ne seraient pas uniques et
amèneraient à distinguer deux cas en nuisant à la simplicité
et à la généralité de l’énoncé. Autrement dit, la volonté de faire
apparaître la plus grande généralité possible, notamment en
n’explicitant pas la relation de composition, est ici plus forte
34
Remarquons que Thābit ne justifie pas sa dernière assertion, qui s’appuie sur
un résultat bien connu. Celui-ci peut être tiré de la proposition 85 des Données
(si deux droites comprennent un espace donné dans un angle donné et si leur
somme est donnée, chacune d’elles sera donnée). Mais on peut aussi en donner une
interprétation algébrique puisque A et D sont alors les solutions d’une équation
du second degré.
THA
que celle qui conduirait à la détermination e#ective des objets
recherchés.
Il reste cependant que l’influence de l’algèbre demeure dans
ce traité particulièrement prégnante. Reprenons par exemple,
pour conclure, la proposition 15.
Proposition 15: Si l’on a deux grandeurs, que le rapport de l’une d’elles
à l’autre est connu et composé de deux des rapports de quatre
autres grandeurs, que l’une des quatre grandeurs est connue, que
le rapport de deux grandeurs parmi les trois restantes, de l’une à
l’autre, est connu, et que le carré de l’une des ces deux grandeurs
et le carré de la troisième grandeur restante sont de somme
connue, alors chacune des trois grandeurs est connue.
A
Supposons donc le rapport connu et composé de deux
B
rapports construits à l’aide des grandeurs C, D, E et G, quels
que soient ces rapports; supposons en outre la grandeur C
D
connue, le rapport connu, et la somme D2 + G2 connue.35 Il
E
s’agit de montrer que D, E et G sont toutes trois connues.
Le mathématicien remarque pour commencer que, puisque la
D
grandeur C et le rapport sont connus, alors soit G est connue,
E
CG
soit le rapport est connu (par application de sa proposi-
D2
tion 10; voir notre annexe no 1).
Si la grandeur G est connue, alors D et E sont connus puisque
D
D + G2 et sont connus.
2
E
CG
Si le rapport est connu, on pose la grandeur H telle que
D2
C CG
= . Puisque ce dernier rapport est connu et que C est
H D2
C CG
connue, H est donc connue. Or on a = , donc
H HG
CG CG
= et par conséquent D2 = HG. Mais puisque
D 2 HG
D2 + G2 est connue, on a donc (HG) + G2 connue. Et,
conclut le mathématicien sans autre justification, puisque H
est connue, G est connue, d’où l’on déduit facilement que D et
E sont connues.
35
Thābit fait une partie de sa démonstration en indiquant que soit D2 + G2,
soit E2 + G2 est connu; par souci de simplicité, nous n’envisagerons que le
premier cas, ce qui ne nuit pas à la généralité.
204 PASCAL CROZET
On a donc vu apparaître, au cours de la démonstration, une
équation du second degré dont les coe$cients sont connus, et
dont Thābit assure dès lors en substance que la racine est
connue. Qu’un tel résultat puisse être obtenu par des voies
géométriques est une chose;36 il est certain cependant qu’une
appréhension algébrique du problème n’a pu que traverser la
pensée du mathématicien: outre le traitement arithmétique des
grandeurs qui est ici mis en œuvre, Thābit a traité et enseigné
ailleurs cette même équation en soulignant l’équivalence entre
voie algébrique et voie géométrique.37
CONCLUSION
Le traité que Thābit ibn Qurra consacre aux rapports composés
se révèle donc particulièrement riche sur plusieurs plans,
dépassant largement le cadre de la figure-secteur qui l’a fait
naître. Des thèmes très divers s’y entrecroisent en e#et:
concepts de nombre et de grandeur, concept de rapport, traite-
ment arithmétique des grandeurs, combinatoire et conception
ensembliste, volonté de donner des outils à l’analyse
géométrique et pour cela étendre les Données d’Euclide, mais
en même temps donner à l’algèbre une place de choix dans cette
analyse, ce qui conduira, d’une certaine façon, à sceller le
destin de l’ouvrage euclidien. Sans que soit fait état d’une
critique explicite quelconque tant des Éléments que des
Données, les inflexions apportées par Thābit aux conceptions
de son prédécesseur grec, tout autant que la manière par
laquelle il appréhende des objets anciens d’une manière nou-
velle nous semblent le signe d’une rationalité mathématique
renouvelée. Rassemblés en un seul traité, se trouvent là bien
des germes qui seront exploités par ses successeurs.
36
Le résultat peut être obtenu par exemple par application de la proposition 84
des Données.
37
Voir: Paul Luckey, ‘‘T I ābit b. Qurra über den geometrischen Rich-
tigkeitsnach weis der Auflösung der quadratischen Gleichungen’’, Berichte über
die Verhandlungen der Sächsischen Akademie der Wissenschaften zu Leipzig,
mathematisch-physikalische Klasse, 93 (1941): 93–114; Roshdi Rashed et Christian
Houzel, Recherche et enseignement des mathématiques au IXe siècle. Le recueil de
propositions géométriques de Na‘ı̄m ibn Mūsā, Les Cahiers du MIDEO 2
(Louvain / Paris, 2004).
THA
o
ANNEXE n 1
Propositions du troisième chapitre

Pour tenter de donner de ces propositions une présentation
synthétique tout en gardant la généralité souhaitée par
l’auteur, nous emploierons les notations suivantes:
– 5 (A, B, C, D, E, F) indique que les grandeurs A, B, C, D,
E, F sont toutes liées par une relation de composition de
A C E
rapport, quelle qu’elle soit (par exemple = J ou
B D F
D E C
= J ).
F A B
A A
– L’égalité = (C, D, E, F) indique que le rapport est
B B
composé de deux rapports des quatre grandeurs C, D, E et F,
A C E A C F
quels qu’ils soient (par exemple = J ou = J ).
B D F B E D
– Pour simplifier et dans les énoncés uniquement, l’expression
EF désignera la di#érence entre les grandeurs E et F, qu’il
s’agisse e#ectivement de EF si E>F ou qu’il s’agisse de
FE si F>E (nous préférons éviter le recours au concept de
valeur absolue, trop éloigné des mathématiques de l’époque).
Proposition 1: Si 5 (A, B, C, D, E, F) et si A, B, C et D sont

connues, alors ou bien le produit EF est connu, ou bien le
E
rapport est connu.
F
Proposition 2: Si 5 (A, B, C, D, E, F), si A, B, C et D sont
connues et si E + F est connue, alors E et F sont connues.
C
Proposition 3 (lemme): Si le produit AB et les rapports et
A
D
sont connus, alors le produit CD est connu.
B
E F
connues, si les rapports et sont connus et si la somme
G H
G + H est connue, alors E et F sont connues.
connues et si la di#érence EF est connue, alors E et F sont
connues.
E F
connues, si les rapports et sont connus et si la di#érence
G H
GH est connue, alors E et F sont connues.
206 PASCAL CROZET
A C E A2 C2 E2
Proposition 7: Si = J , alors 2 = 2 J 2.
B D F B D F
A
Proposition 8: Si = (C, D, E, F) est un rapport connu, si C est
B
D E D
connue et si les rapports , et sont connus, alors D, E et
E F F
F sont connues.
A
Proposition 9: Si = (C, D, E, G) est un rapport connu, si C est
B
C D E G I I K
connue et si les rapports , , et d’une part et , et
H I K L K L L
d’autre part sont connus, alors D, E et G sont connues.
A
B
D
connue et si le rapport est connu, alors ou bien F est
E
CF
connue, ou bien le rapport est connu.
D2
Proposition 11 (lemme): Si les deux grandeurs AB et BC sont
connues, si l’on partage AB en deux parties selon D, et si le
CDAD
rapport est connu, alors chacune des deux parties
BD2
AD et DB est connue.
A
Proposition 12: Si = (C, D, E, G) est un rapport connu, si C
B
D
est connue, si le rapport est connu et si D + G est connue,
E
alors D, E et G sont connues.
Proposition 13 (lemme): Si la grandeur AB est connue, si la
di#érence entre les grandeurs BC et CD est connue et si le
ABCD
rapport est connu, alors les grandeurs BC et CD
BC2
sont connues.
A
B
D
est connue, si le rapport est connu et si DG est connue,
E
A
B
D
est connue, si le rapport est connu et si D2 + G2 est connue,
E
THA
A
B
D
est connue, si le rapport est connu et si D2 G2 est connue,
E
Proposition 17 (lemme): En s’appuyant sur le schéma suivant, si
les grandeurs AB, BC et CD sont connues, et si le rapport
AECD
est connu, alors AE et BE sont connues.38
BEEC
Proposition 18 (lemme): En s’appuyant sur le schéma suivant, si

les grandeurs AB, BC et CD sont connues, et si le rapport
AECD
est connu, alors AE et BE sont connues.39
BEEC
A
B
connue, si DE est connue et si DF est connue, alors D,
E et F sont connues.
Proposition 20 (lemme): Si les grandeurs AB, BC et CD sont
connues (en s’appuyant sur l’un des schémas suivants), si on
ABBE
leur ajoute la grandeur DE, et si le rapport est
CEED
connu, alors AE, BE, CE et DE sont connues.
A
B
connue, et si les trois di#érences DE, DF et EF sont
connues, alors D, E et F sont connues.
38
Le point E est ici entre A et B. On peut alors démontrer que, quels que
soient CD et le rapport connu, le point E existe.
39
Les conditions sont les mêmes que précédemment, sauf pour le point E, qui
doit être entre B et C. On peut démontrer alors que, pour que le point E existe,
il faut que le rapport connu soit plus grand que le rapport de CD à
(AB + AC2AG), où G est le point tel que AG est la moyenne géométrique de
AB et AC; dans ce cas, il y a deux solutions pour E: l’une entre B et G, l’autre
entre G et C.
208 PASCAL CROZET
Proposition 22 (lemme): Si les grandeurs AB, BC et AC sont
connues (en s’appuyant sur l’un des trois schémas suivants),
CDDB
si l’on partage AB selon D, et si le rapport de est
AD2
connu, alors AD, DB et DC sont connues (voir l’annexe no 2
pour la démonstration, qui s’interrompt brutalement sur tous
les manuscrits disponibles).
La proposition 22 semble avoir été conçue pour servir de lemme

à la proposition suivante, absente du traité de Thābit.
A
Pseudo-proposition 23: Si = (C, D, E, F) est un rapport connu,
B
si C = D, si D + E est connue, et si l’une des di#érences
EF, DF ou ED est connue, alors D, E et F sont
connues.40
40
Notons que si C et D sont de deux champs di#érents, on a une
proportionnalité entre les grandeurs restantes qui avait été relevée dans le
E
deuxième chapitre, ce qui permet facilement de conclure puisque le rapport est
F
alors connu.
La proposition 22 permet de traiter le cas où C et D sont dans le même champ,
cas qui, lui, ne pouvait apparaître dans le deuxième chapitre. La relation de
EF
composition débouche alors en e#et sur le fait que le rapport est connu.
D2
Si c’est la di#érence entre E et D qui est connue, on voit immédiatement que E
et D sont connues puisque D + E est connue, et donc F est également connue.
Si c’est la di#érence entre E et F qui est connue, on peut supposer que F > E; en
e#et, que D + E soit connue entraîne ici que D + F est connue et donc les rôles
de E et F sont interchangeables dans l’énoncé. Posons dans ce cas A D = D,
D B = E et D C = F en plaçant les points A , B , C , D comme les points A, B, C,
D des deux premiers cas de la proposition 22. On a alors A B = D + E et
C D D B
B C = FE et donc A B et B C sont connus; d’autre part le rapport
A D 2
est connu; on peut donc appliquer la proposition 22: A D , D B et D C sont ainsi
connues et par conséquent D, E et F sont connues.
Si c’est la di#érence entre D et F qui est connue, supposons dans un premier
temps que F > D. Posons alors A D = D, D B = E et D C = F en plaçant les
points A , B , C , D comme les points A, B, C, D du troisième cas de la
proposition 22. On a alors A B = D + E et A C = FD et donc A B et A C sont
C D D B
connus; d’autre part le rapport est connu; on peut donc appliquer la
A D 2
proposition 22: A D , D B et D C sont donc connues et par conséquent D, E et F
sont connues. Si D > F, il faudrait utiliser un quatrième cas qui n’apparaît pas
THA
o
ANNEXE n 2
Démonstration de la proposition 22 du
troisième chapitre
Proposition: Si les deux grandeurs AB et BC ou les deux
grandeurs AB et AC sont connues [en s’appuyant sur l’un des
schémas suivants], si l’on partage AB selon D, et si le rapport
du produit de CD par DB au produit de AD par elle-même est
connu, alors chacune Kdes grandeursL AD, DB et DC est
connue.
Pour démontrer cette proposition, tâche qu’il laissera, nous

l’avons dit, inachevée, Thābit commence par définir le point E
en posant.
ED CDDB
= .
DA AD2
La place du point E contribue ici à distinguer le second cas du
premier selon que E est ou non entre A et B; dans le troisième
cas, où A est entre C et B, il est facile de voir que E est à
l’extérieur du segment AB.
ED EDDA
Puisque = , on aura donc CDDB = EDDA.
DA AD2
Thābit pose ensuite CF = AB. On aura alors
(CDDB) + (FDDA) = (EDDA) + (FDDA).
Mais (FDDA) + (EDDA) = EFDA, et donc:
(CDDB) + (FDDA) = EFDA.
dans la proposition 22, où le point C serait entre B et D et donc le point D défini
comme partageant la droite connue BC. Si notre hypothèse concernant la
pseudo-proposition 23 est bonne, cette absence suggère que non seulement la
démonstration de la proposition 22 est incomplète, mais encore que son énoncé
lui-même est incomplet.
210 PASCAL CROZET
Or, on a
(CDDB) + (FDDA) =
= (CDDB) + (CDDA) + (ABDA),
et
(CDDB) + (CDDA) = CDAB,
et donc:
EFDA = (CDAB) + (ABDA).
Il faut ensuite distinguer les trois cas de figures.
Dans les deux premiers cas, puisque CD + DA = CA,
l’égalité EFDA = (CDAB) + (ABDA) aboutit à:
EFDA = CAAB.
Puisque AC et AB sont connues, le produit EFDA est donc
connu.
EA ED
Or le rapport est connu puisque est connu (il su$t de
AD DA
ED
séparer le rapport dans le premier cas, ou de commencer par
DA
séparer le rapport inverse dans le deuxième cas). Le rapport
EFEA
étant connu ainsi que EFDA, on a donc EFEA
EFDA
connue. Or AF est connue puisque AF = AC + AB, donc EA et
EF sont de di#érence et de produit connus, elles sont donc
connues, d’où l’on déduit facilement que AD, DC et DB sont
connues.
Pour le troisième cas, Thābit reprend l’égalité
EFDA = (CDAB) + (ABDA),
en notant que l’on a alors
EFDA = AB(CD + AD),
le texte s’arrêtant sur cette remarque.
Il ne nous semble pas qu’en écrivant cette dernière égalité, on
puisse aboutir facilement à une conclusion. On peut toutefois
reprendre entièrement la démonstration du troisième cas de la
façon suivante.
Il a été démontré plus haut que, dans tous les cas, on a
CDDB = EDDA, ce qui donne, en ajoutant CDDA de
part et d’autre:
THA
CDAB = ECDA.
Or on a ici CD = AC + DA, donc:
ECDA = (ACAB) + (DAAB),
donc DA(ECAB) = ACAB, et donc le produit
DA(ECAB) est connu.
EA
Or le rapport est connu (par composition du rapport
DA
ED
connu ), donc le produit EA(ECAB) est connu.
DA
Mais la di#érence entre EA et (ECAB) est connue
puisqu’elle est égale à la di#érence entre AB et AC.
EA et (ECAB) sont donc de produit et de di#érence
connus; ils sont donc connus, donc DA est connue, et par
conséquent DB et DC sont eux aussi connus.
THE ARABIC ORIGINAL OF LIBER DE

COMPOSITIONE ALCHEMIAE
The Epistle of Maryānus, the Hermit and Philosopher, to

Prince Khālid ibn Yazı̄d
AHMAD Y. AL-HASSAN
The Liber de compositione alchemiae or the The Book of the

Composition of Alchemy is believed to have been the first book
on alchemy that was translated from Arabic into Latin. The
translator was the Englishman Robert of Chester who was one
of the earliest translators to flock to Spain to learn Arabic and
to translate some of the Arabic works. He completed his
translation on 11 February, 1144.
With the translation of this book, medieval Europe came to
be acquainted with alchemy for the first time. Thus Robert
writes in his preface to the translation: ‘‘Since what Alchymia
is, and what its composition is, your Latin world does not yet
know, I will explain in the present book’’.1
Alchemy remained something rather new to Europe until
more than a century later. Thus in 1267 Roger Bacon writes in
his Opus tertium (explaining to the pope the rightful role of the
sciences in the university curriculum and the interdependence
of all disciplines):
But there is another science which is about the generation of things
from the elements, and from all inanimate things, [. . .] of which we have
nothing in the books of Aristotle; nor do natural philosophers know of
these things, nor the whole Latin crowd of Latin writers. And since
this science is not known to the generality of students, it necessarily
follows that they are ignorant of all natural things that follow there
from. [. . .] And this science is called theoretical alchemy, which theorizes
about all inanimate things and about the generation of things from the
elements.2
1
Adam McLean, The Book of the Composition of Alchemy (Glasgow, 2002), p. 5.
2
Roger Bacon, quoted by John Maxson Stillman, The Story of Alchemy and
Early Chemistry (Dover, 1960), pp. 262–3.
214 AHMAD Y. AL-HASSAN
The Liber de compositione alchemiae acquired a prominent
place in the Latin alchemical literature. The names of
Maryānus (Morienus) and Khālid (Calid) became well known
to all alchemists in Europe. Their importance in alchemy
matched that of al-Rāzı̄ (Rhazes), Ibn Sı̄nā (Avicenna) and
Jābir (Geber).
A large number of Latin manuscripts have survived. These
were classified into several categories.3 Five contain the
original Latin text that was not altered by later editors. Two of
these un-edited manuscripts go back to the thirteenth century.
They are the Glasgow Hunterian Library MS 253, fols. 46r–53v,
and the Paris Bibliothèque Nationale MS Lat. 7156,
fols. 197–201v. They include the story, as told by Ghālib the
mawla (client) of Khālid, which relates how Khālid and
Maryānus (Morienus) came to meet each other. This is
followed by the dialogue between the two.
All the other numerous Latin manuscripts contain a revised
dialogue. Some contain a preface by Robert of Chester, and
some have an additional speech by Morienus. The various parts
were printed for the first time in 1559 in Paris. The printed
edition contains the preface of Robert of Chester, the speech
of Morienus, the revised account of Ghālib, and the revised
dialogue. The Latin title translates as: ‘‘Booklet of Morienus
Romanus, of old the Hermit of Jerusalem, on the Transfig-
uration of the Metals and the Whole of the Ancient
Philosophers’ Occult Arts, Never Before Published’’.4 The same
publisher issued a second edition in 1564. The text of the Paris
editions was printed in 1572 in a larger collection of alchemical
texts published in Basel.5 The Latin printed edition was trans-
lated into English, German and French.
The first English translation was done in the seventeenth
century and is contained in a manuscript in the British
Library, MS Sloane 3697. This is a translation of the Paris
Latin edition of 1564. This translation was first published by
3
Lee Stavenhagen, A Testament of Alchemy (Hanover, New Hampshire, 1974),
pp. 53–4, and appendix 1.
4
Morieni Romani, Quondam Eremitae Hierosolymitani, de transfiguratione
metallorum et occulta, summaque antiquorum philosophorum medicina, Libellus,
nusquam hactenus in lucem editus Paris, apud Gulielmum Guillard, in via
Iacobaea, sub diuae Barbarae signo, 1559.
5
Auriferae artis, quam chemia vocant, antiquissimi authores, sive Turba
Philosophorum, ed Petrus Pernam (Basilea, 1572), 2 vols.
LIBER DE COMPOSITIONE ALCHEMIAE 215
Holmyard in 1925 in its seventeenth century English.6 A recent
edition in current English was published by Adam McLean in
2002.7
In 1974 an English translation based on the oldest unrevised
Latin manuscripts was undertaken by Lee Stavenhagen
who published the Latin text on opposite pages to his
translation.8
As is customary with most historians of alchemy of the
nineteenth and twentieth centuries, such as Berthelot and
Ruska, doubts were cast on the old established knowledge
about the Latin translations of Arabic works. Thus the Latin
works of Jābir were considered by Berthelot to be authored by
a Latin Pseudo-Geber. The Morienus-Khālid dialogue did not
escape a similar kind of judgment. Ruska cast doubts about
Robert of Chester’s translation and on Khālid, Maryānus and
their dialogue, and he came out with the conclusion that the
whole Latin work was a compilation by an Italian Christian
cleric possibly as late as the fourteenth century. Other scholars
followed Ruska in this assumption.9
The curious thing is that Ruska knew about the existence of
several citations in Arabic alchemical literature extracted
from the Maryānus-Khālid dialogue, but this did not deter him
from coming out with his conclusion. The Italian compiler, he
assumed, should have known Arabic and he had interpolated
some Arabic citations. Ruska did not know as yet about the
existence of the complete Arabic texts.
Although Stavenhagen was also skeptical about Robert of
Chester and his Latin translation, yet he became convinced
that the work was ‘‘certainly a translation from Arabic’’. He
arrived at his conclusion after he saw Holmyard’s translation
of The Book of Knowledge Acquired Concerning the Cultivation
6
Eric Holmyard, ‘‘A Romance of Chemistry’’, a series of articles that appeared
in Chemistry and Industry, Part I, Jan. 23 (1925), pp. 75–7; Part II, Jan. 30, pp.
106–8; part III, March 13 (1925), pp. 272–6; Part IV, March 20 (1925), pp. 300–1;
Part V (printed IV by error), March 27 (1925), pp. 327–8. In this series of articles
Holmyard published the full text of the seventeenth century English translation
of Ye Booke of Alchimye (Sloane MS. 3697).
7
McLean, The Book of the Composition of Alchemy.
8
Stavenhagen, A Testament of Alchemy.
9
McLean, The Book of the Composition of Alchemy, p. 3; Julius Ruska,
Arabische Alchemisten (Wiesbaden, reprint, 1967), p. 48.
of Gold (al-‘ilm al-muktasab fı̄ zirā‘at al-dhahab) of Abū
al-Qāsim al-‘Irāqı̄,10 and by the mention of Maryānus and
Khālid in the commentary of Ibn Umayl on the Book of the
Silvery Water and Starry Earth (Kitāb al-Mā’ al-waraqı̄ wa-al-
ard* al-najmiyya).11 Stavenhagen did not know about the
existence of the complete Arabic manuscripts of the Maryānus-
Khālid dialogue.
In 1971 Sezgin indicated the existence of complete Arabic

manuscripts of the Maryānus-Khālid dialogue.12 Similarly in
1972 Manfred Ullmann’s Die Natur und Geheimwissenschaften
im Islam gave also similar information about the complete
manuscripts.13 Both furnished information about other Arabic
works that gave citations from the dialogue.
Thus the question of the Arabic origin of the dialogue was
settled. It was deemed necessary, however, to edit the Arabic
text and to translate it into English.
The present work aims at this. The writer had sought to
obtain copies of the two known Arabic manuscripts from the
libraries of Istanbul, and he was fortunate to receive help.14
These are Fatih 3227 (fols. 8b–18b) and S q ehit Ali Pasha 1749
(fols. 61a–74b). The writer was able also to secure copies
of several Arabic manuscripts that gave citations from
the Maryānus-Khālid dialogue. Appendix I gives a list of the
Arabic sources that were available for this study, and a list of
the manuscripts that were not available. It is believed that
more Arabic sources may appear in future.
The Arabic texts of Fatih and S q ehit Ali Pasha are similar to
each other with minor di#erences. The largest citation occurs
in Kitāb al-Shawāhid fı̄ al-h
* ajar al-wāh* id of Abū al-H
* asan
10
Abū al-Qāsim Muh* ammad ibn Ah * mad al-‘Irāqı̄, Kitāb al-‘ilm al-muktasab fı̄
zirā‘at al-dhahab; book of knowledge acquired concerning the cultivation of gold;
the Arabic text edited with a translation by E. J. Holmyard (Paris, 1923).
11
H. E. Stapleton and M. H. Husain, Three Arabic Treatises on Alchemy by
Muhammad Bin Umail, Arabic texts edited by M. Turab, Asiatic Society of
Bengal (Calcutta, 1933), pp. 54, 84.
12
Sezgin, Geschichte des arabischen Schrifftums, vol. IV (Leiden, 1971), pp. 111
and 126, and the Arabic updated version (Jeddah, 1986), pp. 163 and 188.
13
Manfred Ullmann, Die Natur und Geheimwissenshaften im Islam
(Leiden, 1972), pp. 192–3.
14
Professor Fuat Sezgin came up to help and the present writer is indebted to
him. He sent copies on CD-ROM which was an invaluable help in editing the
Arabic text. After Professor Sezgin’s quick support, IRCICA in Istanbul sent
another copy on microfilm.
al-H
* alabı̄, in BL MS add 23418. It was found that the text in this
MS has been revised so that it deviates to some extent from
that in Fatih and S q ehit Ali Pasha.
The Latin and the Arabic texts start with Ghālib’s account
and contain the dialogue. The Speech of Morienus is not part of
the Arabic text nor of the earliest un-revised Latin text that
was translated by Stavenhagen.
The last few pages of the Latin text are not found in the Fatih
and S q ehit manuscripts. Search will continue to find the
possible Arabic texts that correspond with these.
The account or prologue of Ghālib is reproduced in this
article in Arabic followed by an English translation of
this text, then follows Stavenhagen’s English translation of the
Latin text. This will enable the reader to examine the corre-
lation between the Arabic wording and the Latin text. The
important deviations between the Arabic and the Latin texts
are indicated. The endnotes indicate some of the distortions of
the Arabic names. There must have been errors in the Latin
translation due to some ambiguity in the Arabic text or to the
lack of understanding it. There is also sometimes a purposeful
editing while the translator was undertaking his work. These
will become apparent to the reader.
The Epistle of Maryānus, the Hermit and Philosopher, to Prince

Khālid ibn Yazı̄d
[A Testament of Alchemy]15
***
In the name of God the merciful and compassionate

[In the name of the Lord, holy and compassionate]16
***
[this is the story of how Khālid ibn Yazı̄d ibn Mu‘āwiya17 came into
possession of the spiritual riches handed down from Stephanos of
Alexandria to Morienus, the aged recluse, as is written in the book of
Ghālib, bondsman of Yazı̄d ibn Mu‘āwiya.18
Now Ghālib was Yazı̄d’s faithful servant,19 entrusted with all his
master’s possessions, and in time, it is said, became faithful servant
likewise to Yazı̄d’s son Khālid.]
***
15
Stavenhagen’s English translation of the Latin text follows our translation
of the Arabic text. Smaller font is used and all paragraphs are placed between
square brackets.
16
This is a translation of the Muslim verse that precedes the start of any text.
The word Lord is the ecclesiastic expression for God.
17
The names of Khālid, Yazı̄d and Mu‘āwiya were distorted in the various
Latin manuscripts. See Stavenhagen, A Testament of Alchemy, p. 2, footnotes 1
and 2.
18
This is a short introduction to the Latin text.
19
Ghālib was a mawlā and not a servant . . .
Ghālib the Mawlā of Khālid ibn Yazı̄d ibn Mu‘āwiya said that
the reason behind the Khālid’s accomplishment of the noble art
(al-s*an‘a al-karı̄ma) is that he went one day in a picnic to Dayr
Barrān in Damascus.20 He was fond of the art (al-s*an‘a),
fascinated by it so that he would not give preference to
anything else. He was constantly searching and experimenting
with it and enquiring about what might come in his way hoping
that he may arrive at it.
[Ghālib relates as follows how Khālid ibn Yazı̄d ibn Mu‘āwiya sought
out Morienus the Greek,21 who lived as a recluse in the mountains of
Jerusalem. One day, Khālid went abroad to a place called Dirmanam.22
He was assiduous in his quest for the Major Work,23 continually
inquiring after any and all whom he supposed to be privy to this
operation.]
***
On that day a man came to him and asked for permission to be

allowed to enter and see him. He was allowed in and he saluted
eloquently.
He said: I brought to the Amir a benefit that no body else had
matched.
[and on this occasion a certain man came to him and desired to speak
with him. Hearing of this, Khālid bid the man come before him. He
saluted Khālid, and Khālid returned his greeting.
The fellow then spoke thus to Khālid: ‘‘I dwell in the mountains of
Jerusalem, and I have come to you, O king,24 with delightful news. Never
has anyone before me given any king such cause to rejoice.’’]
20
Dayr Burrān is most probably Dayr Murrān in Damascus. It was on the
lower slopes of Jabal Qāsiyūn, overlooking the orchards of the Ghuta. It was a
large monastery, and around it was built a village and, one presumes, a residence
in which the caliphs could both entertain themselves and keep watch over their
capital. Dayr Murran often figured in poems of the time. The caliph Yazı̄d I
(father of Khālid) was staying there sometimes. Other caliphs and their
representatives visited or lived there on various occasions. (D. Sourdel, E.I.
under Dayr Murrān). Khālid, according to this text, used to stay sometimes at
Dayr Murran as well.
21
Rūmı̄ here denotes a person from the Greek Catholic church.
22
Dirmanam is a distortion of Dayr Murrān, see note above.
23
Major work is al-s*an‘a in Arabic. In the later Latin revised versions the
word Magistery was used.
24
Khālid was not a king because he did not become a caliph after his father
Yazı̄d. In the Arabic text he is called amı̄r or prince.
***
Khālid asked him: and what is the benefit that you have
brought with you?
[‘‘And what is this news?’’ asked Khālid.]
***
He said: I had learned that you are seeking the art (al-s*an‘a)
and are asking about it. (Ghālib said) that Khālid sat straight
and said: Yes. (The man) said: O Amı̄r, I live in Jerusalem and
I saw in it an ascetic called Maryānus (Marianus) al-Rāhib
(the Hermit) who has attained the art (s*an‘a). He donates
every year to Jerusalem a huge amount of money and gives the
poor and the needy.
[He replied: ‘‘I have heard many say that it is you who continually seek
after the operation which the philosophers call the Major Work. I will
bring you to the knowledge of it through a certain Romaean, who
lives as a recluse in the mountains of Jerusalem, but whose dwelling
place I well know. He sends large amounts of gold to Jerusalem every
year.’’]
***
Then Khālid said to him: If you are telling the truth I shall give
you what you will ask for, but if you have lied then you will
receive what you deserve.
[Khālid said to him: ‘‘If I find that you have told the truth, I will reward
you with whatever you may ask. But if you have lied, you may expect the
worst.’’]
***
The man said: I am content with this since you have treated me
with justice.
[‘‘Well,’’ the fellow replied, ‘‘so be it.’’]
***
Khālid was rejoiced and pleased with what the man had said.
He ordered for him a reward and a raiment and he promised
him good, according to Ghālib.
[Then Khālid rejoiced greatly and commanded that the man be rewarded
with gifts and raiment and much else, as he had promised him.]
***
Then he asked me to accompany him with a group of mawālı̄.

[And the king commanded me along with many other of his servants to
go with him.]
***
We traveled in realms rising up with some terrains and

descending with others. We stayed thus for several days in
search of that ascetic until we located him. We found that
he was an old man, weak, of good appearance and elegant
countenance, wearing a woolen robe and his skin showed as if
it was worn.25
[And so we set out. After wasting much time going from one place to
another, in hopes of chancing upon the recluse, we did indeed find him.
25
The word sunna has several meanings. It can mean picture or face. The
general meaning is that Maryānus had a wrinkled skin due to his old age.
He was tall of stature, though aged, and although lean, so noble of
countenance and visage that he was a marvel to behold. Yet he wore a
hair shirt, the marks of which were borne on his skin.]
***
We rejoiced at finding him and we treated him kindly and

persuaded him until we arrived and brought him to Khālid who
was greatly delighted to see him. We have never seen Khālid so
pleased at anything before.
[We rejoiced to have found him and spoke kindly to him, at last
persuading him with sweet words to relent, and brought him with us
back to our own country, there presenting him to King Khālid. Never
before had we seen the king so pleased by anything.]
***
Then he turned towards me and asked me about our journey in

the country and our return and I related to him what happened
with us from the beginning till the end.
[At last he turned to me and asked what had befallen us in going and
coming, and told him the story from beginning to end.]
***
Then he turned to the old man and asked him: what is your
name?
He answered: Maryānus al-Rūmı̄.
[Then the king regarded the aged man we had brought and wished to
know by what name he was called. The elder replied: ‘‘I am called
Morienus the Greek.’’]
***
Then Khālid asked: Since when have you been in this state?
[And Khālid asked: ‘‘How long is it now that you have dwelt as a recluse
in these mountains?’’]
***
He replied: Four years after the death of Hiraql.

[He replied: ‘‘I began my retreat four years after the death of King
Herakleios.’’]
***
Then Khālid said: Sit down Maryānus. He sat down and was
given a seat of honour. Khālid was pleased by his noble
appearance and politeness.
[Then the king bade Morienus be seated, and himself arose to give
Morienus a place of honor beside him, much pleased with his reserve,
modesty, and elegance.]
***
Then he said: O Maryānus, would it not have been kinder to

you if you were in a church or a monastery?
[The king said to him: ‘‘O Morienus, recluse though you be, would it not
be better that you live in the congregation of others, rather than alone
in the mountains?’’]
***
He said: May God guide the Prince. Good is from God and it
is in his hand to do what he wills. You are right, rest in that is
more and wandering causes more fatigue and tiredness; but a
farmer reaps what he sows, and I hope that good will result
from what I am in, if God wills. Man will not achieve rest
except by much toil.
[He said: ‘‘Perhaps, O king. But the virtues I look for are in God and in
his hand, who will do as he will. And while I grant that, as you have said,
life might be easier for me than in the mountains, still only he who sows
shall reap, and he must reap that which he has sown. Now I trust that I
have gained some little virtue of my own. A man cannot attain repose
except through labors of the spirit.’’]
***
Then Khālid said: Had this been said from the heart of
a believer. Then he said: O Maryānus I heard that you are a
virtuous and a devout person, therefore I desired to see you and
I have sent after you.
[Then the king said: ‘‘These things are true, if said from the heart by one
believing in God. O Morienus, I am pleased that you continue in your
faith. I wished to see you and therefore sent for you.’’]
***
Maryānus said to him: I am not unique. And among people

there are many like me. Death is awaiting each; it is harder on
bodies than their sins, and what follows death is longer, harder
and greater. And God is our aid.
[Morienus said to him: ‘‘You need not marvel at one such as me, a mere
son of the race of Adam. At best, I might only be somewhat comelier,
except that the passage of time has altered me. There are many like me
among men. And at the end is cruel death, than which no punishment is
worse; yet a harsher punishment awaits the spirit after death. But the
almighty Creator be our aid.’’]
***
Khālid said: God help us in dealing with him because he is a

cunning man despite his old age.
[The king replied: ‘‘Thus God may confound man, who is only scorned
the more, the more he is advanced in age.’’]
***
Then Khālid commanded that I take him to a part of the palace

and to bring to him a Christian man from among the elder
scientists to entertain him and talk to him so that he can feel at
home with him, which I did.
[Then the king commanded me to conduct Morienus to a dwelling near
the royal palace and to fetch one of the Christian elders who might
speak with him and comfort him with sweet words, and thus set his heart
at peace. I did so.]
***
Khālid used to visit him twice a day, to sit and chat with him
asking him about the various nations, past days, biography of
kings and the stories of the Greeks (the Byzantines). He told
him about the wonders of the people, their rule and their
a#airs, things that Khālid had never heard before. This caused
Maryānus to occupy a high place in the esteem of Khālid, more
than anybody has ever occupied before.
[and the king made it his custom to come twice every day to Morienus,
sitting down with him and speaking with him, but asking him nothing
concerning his magistery. The king often stayed long, and Morienus
confided greatly in him. Khālid inquired repeatedly about the customs
both royal and common of the Greeks, and about their times and
histories. Never at a loss for a reply, Morienus retold the marvels of
their deeds and discoursed expertly on their sciences, all matters such as
the king had never before heard. Nor had anyone before ever held such
a firm place in the king’s a#ection as Morienus soon came to hold.]
***
Then Khālid said to him: O Maryānus, I have pursued the art
(al-s*an‘a) for some time and searched about it and laboured in
it but I did not find anybody who can give me information or
guide me to it. And I ask you to enable me to know it and to
learn its treatment and you will have whatever you request,
and in addition you will be returned to your original place, and
you do not have to be afraid from me.
[Eventually Khālid addressed him: ‘‘O Morienus, know that I have long,
sought the Superior Work, but found none to counsel me in this matter.
Therefore I earnestly request that you prepare for me some portion of
your magistery. You shall have from me then whatever you may ask, and
I will see to it that you return to your own land, God willing. Nor need
you thenceforth have any fear of me.’’]
***
Maryānus said to him: I knew that you did not send after me
unless you had a need for me. As to what you said, O Prince,
that I do not have to be afraid from you, I have reached the
stage at which no body like me should be afraid except from
God. You have bestowed on me what is befitting for you, and I
have seen of your kindness, your sympathy, your benevolence,
your mercy and your love such that a person like me should not
hide anything of what you require. Added to this what I see of
your intelligence, your comprehension and the nobility of your
faith and your pursuit. God be praised.
[Morienus said to him: ‘‘O king, may God enrich you. Now I understand
that you have sent for me only out of great need. But I disregard the kind
assurance you added, namely that I should not fear you, inasmuch as I
have no need to fear anyone save God alone. You have approached me
as an equal in spirit, and now I see by your a#ection, excellence, and
discrimination that one such as I should have no reason to keep from
you anything of that which you seek, for you are indeed a man of good
intentions as well as deeds and most virtuous. Very well, you have
attained to your initiation and instruction simply and with the greatest
ease. May the Creator be praised!’’]
***
At this Khālid smiled and said: A person with whom compas-

sion is not e#ective will be harmed by crudeness. Haste is an
act of the devil.
[At this, King Khālid smiled, and then said: ‘‘The crudeness of haste
ensnares any man, unless he be ruled by patience. I am of the house of
Mu‘āwiya, and there is no strength save in great God most high.’’]
***
Maryānus said: I shall explain to you, and there is neither

might nor power but in God, the most high the supreme. May
God guide you to the better. Listen to this science (h * ikma) and
you will know what is needed and understand it and learn it
and will contemplate its inside and outside traits so that you
will become acquainted to it, if God, glory to him, wills.
This matter that you have requested cannot be attained by
any one by force and cannot be gained by violence and can only
be acquired from a scientist by kindness, a#ection and true
love. First it is a fortune from God, glory to him, who delivers
it to whom he chooses from among his creatures by supreme
power and causes him to learn it and discloses to him its
secrets; and this is one of the gifts of God, the high, who teaches
it to whom he loves from among his creatures and his subjects
and to those who are his guides and are submitting to him.
[Morienus then said: ‘‘O king, may God enrich you. Now attend to the
examination of this operation, and you will know it well and
understand. Consider it thoroughly from beginning to end, and you will
know all things that pertain to it, God willing. No one will be able to
perform or accomplish this thing which you have so long sought or
attain it by means of any knowledge unless it be through a#ection and
gentle humility, a perfect and true love. For this is something which God
gives into the sure keeping of his elected servants until such time as he
may prepare one to whom it may be handed on from among his secrets.
Thus it is only the gift of God, who chooses among his humble and
obedient servants those to whom he reveals it.’’]
***
Khālid then said: Surely there is no course of action except by

guidance from God, the most powerful and dignified.
[Khālid said to him: ‘‘Surely we know that nothing can be done without
the help and guidance of God, most high and eternal.’’]
***
Then Khālid said: Sit down, O Ghālib, and write what will take
place between me and him.
[Then King Khālid said to me: ‘‘O Ghālib, quickly now, sit down and
write all that we have said.’’]
***
Maryānus said: Know, O Khālid,26 that God, the high, has
created his servants weak from weakness27 and they cannot
hold back what he had advanced and they cannot advance
what he had held back, and they cannot know anything except
what God reveals to them, and they cannot understand
except what God gives them, and they cannot get except what
he has opened a way to it28 by his power. He made those whom
he has chosen from among his creatures seek the knowledge of
this mental gift that takes out its possessor from the hard toil
of this world and lead him to the riches of future life and its
delights. They continued transmitting its knowledge by inherit-
ance from one to the other until the science was eradicated and
its people had gone and the teachers could no longer be found.
From those genuine books that had remained there are
the books of the holy men and philosophers that were written
by our predecessors and were left as inheritance to those
successors whom God has willed to attain this art that was
described to be too elaborate and to be full of falsehoods. If they
have said too much and called things by other than their true
names and described them by symbols yet without doubt they
have explained them and clarified them and informed about
them by the art and by examples and allusions. They tried to
keep away the fools and to prevent darkness by intelligent
minds and true sayings and so they perplexed men of compre-
hension and reduced to nothing those who have no belief. They
signaled to men of science and comprehension and clarified and
explained. The wise should seek science and should not fall
short of it. Let him put his hope in God and desire from him that
he enthuses in him true guidance in all his a#airs, and to
bestow on him critical understanding, good handling, correct
interpretation and excellent compiling without deviation.
[Thereupon Morienus continued, ‘‘Almighty God in his power created
powerless servants who can neither undo what he has done nor advance
what he holds back, nor can they even know anything except what he
reveals to them or accomplish anything except what he grants to them.
Nor are they able even to possess anything except by the strength that
same God has conferred upon them, nor even govern their own spirits
except insofar and so long as he has ordained for them. And from
26
The Arabic text makes Maryānus call Khālid by his name without any
formality, whereas in the Latin text Morienus is addressing Khālid as ‘‘O, King’’.
27
28
among his servants, he chose to select certain ones to seek after the
knowledge he had established that rescues him who masters it from
the wretchedness of this world and assures him riches to come, God
willing. While those so chosen used to hand down this knowledge to
their own heirs, it was at last lost and its masters dispossessed of it when
none could be found any more who knew it. But of the books which set
forth the matter correctly there remained a few by the ancient seers who
went before us. They left their knowledge as a legacy to their successors,
whom God had chosen to become adepts according to the methods that
had been explained truthfully and forthrightly by their predecessors.
The ancients, however, did not refer to the matters pertaining to this
science by their proper names, speaking instead, as we truly know, in
circumlocutions, in order to confute fools in their evil intentions. This
they did by formulating their convictions and true sayings always in
parables, so that only those of great wisdom and resource would be able
to uncover their true meaning. Since the ancients thus disguised this
knowledge, those who would learn it must understand their maxims.
Nor may they draw back from this, but must fix their faith in God and
persist to the end that he bring them to this knowledge, improve; their
estate, and give them direct, unerring access to the methods the
science.]
***
Khālid said: O Maryānus, you have spoken and excelled, and

you preached and your message was well received, and gave
advice that caused relief, and it is not unfitting if somebody like
you with his scholarship, his age, his knowledge and his
judgment has promised to bring this into completion. Explain
to me what I am inquiring about and clarify it to me in an
unambiguous description that saves me from occupying my
mind and exerting it in this question. Is it from one thing or
from several?29
[King Khālid then said: ‘‘Now well taught and well spoken, O Morienus,
nor do I hold these precepts strange, coming as they do from a teacher
of such wisdom and years as yourself, who is willing that I should learn
this science. Therefore explain to me clearly that which I ask of you,
sparing me needless labor over this matter which I seek from you. Tell
me whether this operation is accomplished only by a single principle or
by several.’’]
29
This is the first question of the Morienus-Khālid dialogue.
Appendix I
Arabic texts of the Maryānus–Khālid dialogue
I. Available
(a) Complete manuscripts
1. MS Fatih 3227 (fols. 8b–18b)
2. S
q ehit Ali Pasha 1749 (fols. 61a–74b)
(b) Large citation
3. British Library MS add 23418, Abū al-H * asan al-H
* alabı̄,
al-Shawāhid fı̄ al-h * ajar al-wāh* id (fols. 123a–125b)
(c) Fragments
4. Al-‘Irāqı̄ al-Sı̄māwı̄, al-‘Ilm al-muktasab, British Library,
MS add 24016 (fols. 27; 28; 48)
5. Al-Gildakı̄, Nihāyat al-t*alab II, Berlin, MS 4184 (fol. 183)
6. Manuscript of Abdallah Yurki Hallaq, Aleppo (p. 180).
7. National Library of Medicine, Wahington, MS A-70
(fols. 53b–57b)
II. Existing but not available at the time of writing this paper
8. Khanji, Cairo, according to Kraus, Jābir ibn H * ayyān,
Textes choisis (Paris / Cairo, 1935), vol. I, pp. 182. Sezgin,
p. 126 (seems to be a complete one).
9. Hyderabad, Asafiya, according to Stapleton. See Sezgin,
p. 111.
10. Teheran, Khaniqah-i- Ni’matallah 145 (a fragment, 18b)
Sezgin, p. 126.
11. Leningrad University, MS Or. 1192, Sezgin, p. 126
12. Al-Sifr al-mubajjal, s. Siggel Katal. Gotha p. 65, see
Ullmann, p. 193, note 2.
13. Al-‘Irāqı̄ al-Sı̄māwı̄, K. al-Aqālı̄m al-sab‘a, s. Siggel
Katal. Gotha p. 25; see Ullman, p. 193, note 2.
14. Chester Beatty MS 5002 (fol. 55a). See Ullmann’s
Catalog, p. 172.
15. Personal Collection, see Kraus I, p. 187.
YU
zH* ANNA
z IBN SARA
z BIYU
z N: FURTHER STUDIES
INTO THE TRANSMISSION OF HIS WORKS
PETER E. PORMANN*
Manfredo Ullmann, magistro meo,
pro sua erga me liberalitate
necnon humanitate
grato animo dedicatum
Ibn Sarābiyūn is one of the last exponents of classical Syriac

medical writing, and one of the most influential authors for
the development of medical theory and practice in late ninth-
century Baghdad in particular, and for the Arabic medical
tradition in general. During the last thirty years, three import-
ant studies have been published regarding the life and work of
Ibn Sarābiyūn, each of which dealing with a di#erent aspect of
the transmission of this important author’s œuvre.1 Likewise,
during the last twenty-five years, a number of texts associated
with Ibn Sarābiyūn’s works have been edited, allowing us
today to shed new light on the relation between the original
Syriac and the numerous translations into Arabic, Latin and
* I would like to record my gratitude to Prof. Ch. Burnett, Dr E. Savage-Smith,

Prof. M. Ullmann, and the anonymous referee, who read an earlier draft of this
article and made invaluable comments. I am particularly indebted to Prof.
R. Kruk, Prof. A. van der Heide, Dr H. van de Velde and Prof. J. J. Witkam, who
assisted me in di#erent ways during my time in Leiden. This research was made
possible thanks to the Warden and Fellows of Merton College, Oxford, who
elected me to a Junior Research Fellowship. I finally wish to thank the sta# at
the Oriental Reading Room of the Bodleian Library, which has become my second
home over these last years, and especially the Keeper of the Oriental Collections,
L. Forbes, who has facilitated my research there tremendously.
1
M. Ullmann, ‘‘Yūh * annā ibn Sarābiyūn Untersuchungen zur Überlieferungs-
geschichte seiner Werke’’, Medizinhistorisches Journal, VI, 4 (1971): 278–96;
L. Richter-Bernburg, ‘‘Pseudo-T I ābit, Pseudo-Rāzı̄, Yuh* annā ibn Sarābiyūn’’, Der
Islam, 60 (1983): 48–77; and G. Troupeau, ‘‘Du syriaque au latin par l’inter-
médiaire de l’arabe: le Kunnāš de Yūh * annā ibn Sarābiyūn’’, Arabic Sciences and
Philosophy, 4, 2 (1994): 267–78. Later scholars find themselves often in the
situation of standing on the shoulders of giants; this is also my case: I am heavily
indebted to the work of Ullmann, Richter-Bernburg and Troupeau. But just as
they provide the foundation of this research, it is possible to go beyond them and
rectify some of their assumptions; this, of course, does not detract from the great
esteem one has for their work.
234 PETER E. PORMANN
Hebrew. Furthermore, through analysing and comparing a
number of manuscripts containing di#erent parts of Ibn
Sarābiyūn’s work which have not hitherto been considered
together, progress can be made towards answering the ques-
tion how Ibn Sarābiyūn was translated and used during the
medieval period.
The present contribution is divided into three parts. Firstly,
the question of Ibn Sarābiyūn’s biography and bibliography
will be tackled. It is surprising how much uncertainty still
exists today as to when this author lived and what he wrote.
Secondly, the most famous book by Ibn Sarābiyūn, the Small
Compendium on Medicine, and the question of its transmission
will be discussed. Thirdly, his Large Compendium will be the
focus of our attention; in this part, Arabic fragments will be
edited for the first time and used for a comparison of the Small
with the Large Compendium.
I
Very few things are known about the life of Ibn Sarābiyūn from
medieval bio-bibliographic sources. In the Fihrist of Ibn
al-Nadı̄m (written in 987 2 ) we are told:3
Yah* yā ibn Sarāfiyūn: All he wrote was Syriac. He lived at the beginning
of the [‘Abbāsid] dynasty. His two books on medicine were translated
into Arabic: The Large Compendium (kunnāš) of Yūh * annā,4 which
consists of twelve maqālas. The Small Compendium, [in] seven maqālas.
Al-Qift*ı̄ (d. 1248) gives a similar account, while Ibn Abı̄
Us*aybi‘a, provides us with more information. He mentions
di#erent translations of the Small Compendium, namely that
by Mūsā ibn Ibrāhı̄m al-H
* adı̄tIı̄, and that by Bar Bahlūl.5 The
latter quotes Ibn Sarābiyūn on a number of occasions in his
2
All dates, unless otherwise indicated, refer to the Common Era ().
3
Ibn al-Nadı̄m, Kitāb al-Fihrist, ed. G. Flügel, 2 vols. (Leipzig, 1871–2), p. 296,
l. 7–9; all translations, unless otherwise stated, are my own.
4
‘‘Yah* yā’’ and ‘’Yūh
* annā’’ are two variant forms of the same name ‘‘John’’
(Ioannes, ’I´ c, ).
5
Cf. M. Ullmann, Die Medizin im Islam, HO 1. Abt. Erg. 6.1 (Leiden etc.,
1970), p. 102; see also below pp. 239–40.
YU
zH* ANNA
z IBN SARA
z BIYU
zN 235
lexicon, and this is our only source for what meagre fragments
we have of the Syriac text.6
All these Arabic bio-bibliographers agree that Ibn Sarābiyūn
lived in the early period of the ‘Abbāsid dynasty, that is to say
in the second half of the eighth century. M. Ullmann and others
have argued that since Ibn Sarābiyūn quotes some later
authors such as H * unayn ibn Ish * āq (d. ca. 877) and Sābūr ibn
Sahl (d. 869), he must have lived in the second half of the ninth
century, and that the information contained in the bio-
bibliographical tradition is erroneous.7 Troupeau dismissed
this argument, saying that the later quotations could have been
interpolated by the Arabic translators, something Ullmann
thought was unlikely.8 There is another, stronger argument for
dating Ibn Sarābiyūn to second half of the ninth rather than
eighth century, which Ullmann was the first to invoke and
Troupeau did not discuss: Ibn Sarābiyūn revised (is*lāh* ) Sābūr
ibn Sahl’s Dispensatory, which means that he is later than an
author who certainly lived and wrote in the ninth century.9
Furthermore, H. Lehmann proposed 873 as a date for the
composition of the [Small] Compendium;10 yet although he
promised to substantiate this date, he has never done so in
6
For example under the entry ´
[Lexicon Syriacum auctore
Hassano Bar Bahlule, ed. R. Duval, 3 vols. (Paris, 1901), col. 1586, 3] we find the
following citation:
‘‘According to Ibn Serapion: ´

are thus small things resembling blisters
arising above the surface of the skin.’’
Yet, it is not even clear from which work by Ibn Sarābiyūn Bar Bahlūl quotes
here.
7
Cf. Ullmann, ‘‘Sarābiyūn’’, pp. 278–80, following e.g. L. Leclerc, Histoire de la
médecine arabe, 2 vols. (Paris, 1876), vol. 1, pp. 115–16.
8
Troupeau, ‘‘Du syriaque . . .’’, p. 269: ‘‘Or cet argument ne me paraît
absolument pas convaincant, car ces citations de médecins postérieurs à Ibn
Sérapion, ont très bien pu être introduites dans son Kunnāš par les traducteurs
du Xe siècle.’’; Ullmann, ‘‘Sarābiyūn’’, p. 279: ‘‘Diese Zitate kann man kaum als
spätere Interpolationen betrachten.’’
9
Ullmann, ‘‘Sarābiyūn’’, p. 279; cf. R. Degen, M. Ullmann, ‘‘Zum
Dispensatorium des Sābūr ibn Sahl’’, Welt des Orient, 7 (1973–4): 241–58, esp.
pp. 253–4, and O. Kahl, Sābūr ibn Sahl. Dispensatorium parvum (al-Aqrābādhı̄n
al-s*aghı̄r) (Leiden etc., 1994), pp. 17–18.
10
See Ullmann, Medizin, p. 102. The reference is to Lehmann’s review of
G. Sobhy’s The Book of Al Dakhîra (Cairo, 1928) in Orientalisch Literaturzeitung,
11 (1929): 869–70 [repr. in F. Sezgin, T I ābit ibn Qurra (d. 288 / 901): Texts and
Studies, Publications of the Institute for the History of Arabic-Islamic Science:
Islamic Medicine 32 (Frankfurt, 1996), pp. 277–9].
print. Apart from Lehmann’s unverifiable statement, the other
two reasons for dating Ibn Sarābiyūn to the ninth rather than
the eighth century are persuasive in their own right, and, a
fortiori, are convincing, when taken together: (1) the sheer
amount of quotations from ninth-century authors makes an
interpolation by a translator or redactor unlikely; and (2) if Ibn
Sarābiyūn revised Sābūr ibn Sahl’s work, he must a least be
contemporaneous.
Another conundrum which still occupies the minds of many
scholars, especially those concerned with medieval Latin
medicine, is that of ‘‘Serapion Iunior’’, a younger Ibn
Sarābiyūn, to be distinguished from a ‘‘Serapion Senior’’, the
Ibn Sarābiyūn discussed here and author of the Small and
Large Compendium. There are two aspects to this problem.
Firstly, we read in Arabic bio-bibliographical sources that a
Dāwūd ibn Sarābiyūn, who lived in the late eighth century, is
the brother of our Yūh * annā ibn Sarābiyūn.11 We have already
seen that this is impossible from a chronological point of view:
Yūh
* annā ibn Sarābiyūn lived in the ninth century. So what-
ever the historical truth about this Dāwūd ibn Sarābiyūn might
be, if he lived in the eighth century, he cannot be the brother of
Yūh
* annā ibn Sarābiyūn.12 The notion of two Ibn Sarābiyūn
brothers, Dāwūd and Yūh * annā, was erroneously combined
with another misunderstanding.
We have Gerard of Cremona’s Latin translation of Ibn
Sarābiyūn’s Small Compendium, the editio princeps of which is
entitled: Breviarium medicinae.13 But there was another Arabic
medical text translated into Latin around 1290 by the Jew
Abraham of Tortuso and commonly attributed to a Serapion,
namely the Liber aggregatus in medicinis simplicibus.14 Already
in the Middle Ages it became apparent that the Liber aggre-
gatus is more recent in date than the Breviarium medicinae,
11
Cf. Ullmann, ‘‘Sarābiyūn’’, p. 279; al-Qift*ı̄, Ta’rı̄h al-h
* ukamā’, ed. J. Lippert
(Leipzig, 1903), p. 431, 6–8; Ibn Abı̄ Us*aybi‘a, Kitāb ˘ ‘Uyūn al-anbā’ fı̄ *tabaqāt
al-at*ibbā’, ed. A. Müller, 2 vols. (Cairo and Königsberg, 1884), vol. 1, p. 109, 17–22
(quoted infra p. 239).
12
Ullmann provided a possible explanation for this error, cf. ibid.
13
(Venice, 1479); cf. W. Osler, Incunabula Medica: a Study of the Earliest
Printed Medical Books, 1467–1480 (Oxford, 1923), p. 110, no. 174; Ullmann,
‘‘Sarābiyūn’’, p. 282, n. 29 is incorrect.
14
Cf. P. Dilg, ‘‘The Liber aggregatus in medicinis simplicibus of Pseudo-
Serapion: An influential work of medical Arabism’’, in Ch. Burnett, A. Contadini
(eds.), Islam and the Italian Renaissance, Warburg Institute Colloquia 5 (London,
1999), pp. 221–31; cf. Ostler, ibid., no. 173.
YU
zH* ANNA
z IBN SARA
z BIYU
zN 237
and it was therefore attributed to a ‘‘Serapion Iunior’’ while
the author of the Breviarium medicinae became known as
the ‘‘Serapion Senior’’.15 M. Ullmann argued that this Liber
aggregatus de simplicibus medicinis could not have been
written before the end of the thirteenth century because of
similarities between it and the G
{ āmi‘ by Ibn al-Bayt*ar.16 In the
meantime, L. F. Aguirre De Cárcer edited Ibn Wāfid’s Kitāb
al-Adwiya al-mufrada (Book on Simple Drugs),17 which
J. C. Villaverde Amieva demonstrated is the Arabic original of
the Latin Liber aggregatus.18 Since Villaverde Amieva did this
in a review not readily available to scholars, at least in the
Anglo-Saxon world, I would like to rehearse his main argu-
ment. The prologue (as well as the rest) of Ibn Wāfid’s work is
identical to that contained in the Liber aggregatus as the
comparison of the first sentence in the two texts shows19:
Translation of the Arabic: Since I saw that the two books on simple
drugs by Dioscorides and Galen contain the knowledge [‘ilm] about
them [sc. the simple drugs] which one needs, and provide [all] the
information about them which one ought to have, except that they
require to be combined, for in each one of them there is a part of the
science about them [sc. the simple drugs] which is not found in the other
(for most of the time Dioscorides gives the form, outer appearance and
use of each drug, while Galen gives the substance, taste, quality,
potency and use), I made the task easier and took the labour upon me to
gather together [all] that is excellent in these two books and to organise
[the material according to] the utility in them.
15
Dilg, ‘‘The Liber aggregatus’’, p. 223.
16
Ullmann, Medizin, pp. 283–4.
17
Ibn Wāfid (m. 460 / 1067). Kitāb al-Adwiya al-mufrada (Libro de los
medicamentos simples), Edición, traducción, notas y glosarios, 2 vols. (Madrid,
1995); there is also a new edition by A. H* . Basaǧ (Beirut, 2000).
18
Aljamía, 9 (1997): 112–18; cf. Ullmann, Medizin, p. 273.
19
Ibid., pp. 114–15; sig. a 2 r (Venice, 1479); I changed the orthography of the
Latin and the Arabic slightly.
There can therefore be no doubt, as Villaverde Amieva states,
that the alleged Serapion Iunior, author of the Liber aggre-
gatus, is no other than Ibn Wāfid.20 Furthermore, a comparison
between the Latin translation and the edition by A. H * . Basaǧ
confirms this analysis. Likewise, whatever the historical
21
value of the information about Dāwūd ibn Sarābiyūn, he is not

the brother of the Yūh* annā ibn Sarābiyūn discussed here. The
distinction between a Younger and Elder Serapion is therefore
erroneous and should be abandoned. Yūh * annā ibn Sarābiyūn
lived in the second half of the ninth century and wrote two
works, the Small Compendium and the Large Compendium.
Before we come to these two works, it is necessary to deal
with another Ibn Sarābiyūn whom one might confuse with, or
relate to, the Yūh* annā ibn Sarābiyūn discussed here. He is
one Ibn Sarābiyūn ibn Ibrāhı̄m, the alleged author of a Kitāb
al-Fus*ūl al-muhimma fı̄ *tibb al-a’imma (Book of Important
Chapters on the Medicine of the Masters), contained in a
unique manuscript, Oxford, Bodleian Library, Oriental Collec-
tions, MS Hunt. 461, and described and partly edited by
20
Ibid., p. 113: ‘‘. . . el cotejo del prólogo . . . no dejaba lugar a dudas, y ponía
de manifiesto que la obra atribuida a Serapion es inequívocamente una tra-
ducción del Kitāb al-adwiya al-mufrada de Ibn Wāfid.’’
21
Unfortunately, this edition (see above n. 17), which is based on an unnamed
manuscript, is incomplete; cf. ibid., p. 3:
In fact, it breaks o# at the description of sorrel (rı̄bās; rumex L. and Var.):
Sorrel (rı̄bās): ’ǧt*yrš (acedera) in the vernacular [i.e. aljamiado, Spanish]. Ish
* āq
ibn ‘Imrān: sorrel is a plant having fresh shoots which are red and slightly green;
its leaves are big, broad, round and green; the taste of its shoots is pungent and
slightly sweet . . .
corresponding to [Lyons, 1525, fol. 157rb1–6]:
De Ribes. Ribes. Isaac eben Amaram. Ribes est planta habens capreolos recentes
rubeos ad uiriditatem tendentes et habet folia magna lata rotunda uiridia; et
habet grana quorum sapor est dulcis cum acetositate . . .
‘‘Mahamed eben ririfus’’, mentioned by Ullmann (Medizin, p. 283), is the
commentator of the Centriloquium, i.e. Ah * mad ibn Yūsuf ibn Ibrāhı̄m ibn al-Dāya
(d. 951 ); cf. GAL S I, p. 229. The similarity between the quotation by this Ibn
al-Dāya in the Liber aggregatus and Ibn al-Bayt*ar seems to be based upon the fact
that Ibn al-Bayt*ar uses Ibn al-Wāfid quite extensively.
YU
zH* ANNA
z IBN SARA
z BIYU
zN 239
M. Ullmann.22 Nothing is known about this author, but since
he quotes, among others, Ibn Sı̄nā (980–1037), he cannot have
lived earlier than the eleventh century, and he should not be
confused with Yūh* annā ibn Sarābiyūn.23
II
In the great fire of the Escorial in the year 1671, the only two
complete copies of Ibn Sarābiyūn’s Small Compendium were
lost, as well as two partial copies. Today, we are left with a
number of fragments, some of which are substantial, which
allow us to regain more than half of the Arabic version of this
work. Yet, the transmission of the Small Compendium is
problematic not only because of its fragmentary nature. It was
so popular that it was translated on a number of occasions by
di#erent people. Ibn Abı̄ Us*aybi‘a gives the following report
about the translations24:
Yūh
* annā ibn Sarābiyūn. All he wrote was in Syriac. His father
Sarābiyūn was a physician from Bāǧarmā.25 His [i.e. the father’s] two
sons became both outstanding physicians; they were: Yūh * annā and
Dāwūd; Yūh
* annā ibn Sarābiyūn wrote the following books: The Small
22
M. Ullmann, Rufus von Ephesos: Krankenjournale (Wiesbaden, 1978); the
manuscripts is described in a very detailed and comprehensive fashion in
E. Savage-Smith, Arabic Manuscripts on Medicine and Related Topics, vol. 1 of:
C. Wakefield (ed.), A New Catalogue of Arabic Manuscripts in the Bodleian
Library, Oxford (Oxford, 2004), forthcoming.
23
Ibid., p. 12; cf. Ullmann, ‘‘Sarābiyūn’’, p. 296; space does not permit to
discuss all the various spurious references, for instance, to ‘‘serapino’’
(a corruption from Greek c´ [cf. M. Ullmann, ‘‘Die arabische
Überlieferung der Schriften des Rufus von Ephesos’’, in Aufstieg und Niedergang
der römischen Welt, II 37.2, 1293–1349, 1301]) or ‘‘Saraphies’’ (denoting Serapion
of Alexandria [cf. Ch. Burnett, D. Pingree (eds.), The Liber Aristotilis or Hugo of
Santalla (London, 1997), p. 3]).
24
I, 109, 17–23; cf. Ullmann, ‘‘Sarābiyūn’’, pp. 280–1, and Troupeau, ‘‘Du
syriaque . . .’’, pp. 270–1.
25
According to Troupeau, ‘‘Du syriaque . . .’’, p. 268, the province called in
Syriac ; cf. EI2 s.v. Bād
I jarmā.
Compendium, the famous one, in seven maqālas, was translated by
[Mūsā ibn Ibrāhı̄m] al-H * adı̄tIı̄, the secretary for Abū al-H
* asan ibn Nafı̄s,
the physician, in the year 318 [ / 930 ]. This is a more idiomatic
translation than that by al-H * asan ibn al-Bahlūl [sic] al-Awānı̄
al-T
* ı̄rhānı̄. It was also translated by Abū Bišr Mattā.
In the following, I shall first review these three translators, and
then describe the extant Arabic manuscripts. Although I shall
rehearse some of the arguments found in the previous scholarly
work,26 it is nonetheless essential to look at the evidence again,
because new material has come to light that makes a fresh
interpretation possible.
Of the first translator mentioned by Ibn Abı̄ Us*aybi‘a, Mūsā
ibn Ibrāhı̄m al-H * adı̄tIı̄, we know nothing other than the infor-
mation given in this entry. He is also named in two manuscript
copies as the translator (see below). Al-H * asan ibn al-Bahlūl
al-Awānı̄ al-T* ı̄rhānı̄ (fl. 963), the second translator in Ibn
Sarābiyūn’s list, is none other than Bar Bahlūl, the famous
glossographer. He is mostly known for his Syriac lexicon
edited by R. Duval27 and is the author of a Book of Indications
written in Arabic28; furthermore, he translated Theophrastus’
Meteorology from Syriac into Arabic.29 Finally, Abū Bišr
Mattā (d. 940) translated, among other things, Aristotle’s
Poetics.30
As mentioned above, the Syriac original version of the Small
Compendium is nearly entirely lost for us.31 The Arabic manu-
scripts containing portions of the Small Compendium are the
following:
26
Cf. p. 233, n. 1.
27
Lexicon Syriacum, ed. Duval.
28
Kitāb al-Dalā’il, published in fascimile by F. Sezgin, Publications of the
Institute for the History of Arabic-Islamic Science, Series C, vol. 10 (Frankfurt,
1985), and also edited by Y. H * abbı̄ (Kuwait, 1987).
29
Cf. H. Daiber, ‘‘The Meteorology of Theophrastus in Syriac and Arabic
translation’’, in W. W. Fortenbaugh, D. Gutas (eds.), Theophrastus, His
Psychological, Doxographical, and Scientific Writing, Rutgers University Studies
in Classical Humanities 5 (New Brunswick and London, 1992), pp. 166–293;
H. Takahashi, ‘‘Syriac fragments of Theophrastean meterology and mineralogy’’,
in W. W. Fortenbaugh, G. Wöhrle (eds.), On the Opuscula of Theophrastus
(Stuttgart, 2002), pp. 191–224.
30
Ed. J. Tkatsch, Die arabische Übersetzung der Poetik des Aristoteles, 2 vols.
(Vienna and Leipzig, 1928 / 32); cf. W. Heinrichs, Arabische Dichtung und
griechische Poetik, Beiruter Texte und Studien 8 (Beirut, 1969), pp. 118–23;
G. Endress, art. Mattā b. Yūnus, Abū Bišr in EI2, 6, 844b.
31
See above p. 235, n. 6.
YU
zH* ANNA
z IBN SARA
z BIYU
zN 241
1. Madrid, Escorial, MS 818, fols. 129–135, containing book 1, ch.
1–3 (beginning); abbr. Ea.
2. Paris, BNF, MS 2918 (fonds arabe), fols. 155–170, containing
book 3, ch. 30–31, book 4, ch. 1–2 (beginning), 8 (end)–10
(beginning); abbr. P.
3. Madrid, Escorial, MS 852, fol. 39, containing book 4, ch. 23
(end); abbr. Eb.
4. Brussels, Bibliothèque Royale, MS 19891, containing books 5–7;
abbr. B.
5. Leiden, MS 2817 (Cod. 2070), containing book 7; abbr. L.
The manuscripts Ea and L contain information regarding the
translator. Eb and P do not provide any direct indication about
the translator since they lack both beginning and colophon. B
does have a title and colophon, but does not contain any
information about the translation.32 In the following, I shall
describe L in greater detail, because this description, not yet
undertaken in previous scholarship, yields a number of
interesting findings.
L is written on non-European paper, measuring 1822.5 cm2,
while the text takes up 1516 cm2, having approximately 14
lines per page. It is written in mostly unvocalised nash which
lacks most diacritical points33; rā’ is sometimes˘ distin-
guished from zā’ through a caron . There are rubrics for
titles, beginnings of recipes, section headings and so forth. On
the top of fol. 1a, there is a title in Greek: ` y c
y
´
´ c C
ć, and underneath in Arabic:
The third part of the kunnāš by Yūh * annā ibn Sarāpiyūn in the
translation of al-H
* adı̄tIı̄, the secretary.
Below this in Hebrew:
The third part of the book of Yūh
* annan ben Serāpiyōn On Medicine.
32
For a complete description of this manuscript cf. Ullmann, ‘‘Sarābiyūn’’,
pp. 283–4; the order of the folios (according to information provided by Pieter
Voorhoeve and reproduced by Ullmann) is as follows: 1–19, 56, 47–55, 20–29,
30–46, 57 #., the fifth maqāla starting on fol. 1b, the sixth on fol. 29a and the
seventh on fol. 46b.
33
This feature led de Goeje to say [Catalogus Codicum orientalium Bibliothecae
Academiae Lugduno-Batavae, vol. 5 (Leiden, 1873), p. 322]: ‘‘Codex antiquus
propter deficientiam punctorum passim lectu di$cilis est.’’
Underneath this, there is a recipe in Judaeo-Arabic, entitled:
( , retentive dressing).
The colophon on the last page (fol. 248a8 #.) says:
The end of the compilation from the Book of Medicine, written by Ibn
Sarābiyūn. Thanks be to God – for it is meet to thank Him – and His
blessings on Muh * ammad, His messenger and servant, and on the
righteous people.
Underneath we have an extremely interesting owner’s note in
Hebrew:
I received from Rabbi S { emū’ēl ben Rabbı̄ S

{ elōmō HamMe’ātı̄ three
guilders [?] of money; my pledge for him is this book, written in Arabic
script. In order that it [this book] be placed in the hands of the
aforementioned R. S { emū’ēl may the aforementioned three guilders [?]
come into my hand according to the law of the angel [?]. I signed my
name here in the month tišrı̄ in the year 73 according to the counting
from the creation [of the world; i.e. A.M.; corresponding to September
1312 ]: Yōsēf ben Rabbı̄ Benbeništı̄.
The actual text starts on fol. 1b:
In the name of God, the merciful, the compassionate. The seventh

maqāla, of the compendium (kunnāš) by Yūh
* annā ibn Sarābiyūn,
containing 37 chapters [bāb].
Then follows the lists of contents, starting as follows:
34
YU
zH* ANNA
z IBN SARA
z BIYU
zN 243
I What are the things that he who is resolved to compound remedies
correctly ought to know?
II What need is there for compound remedies?
III In which we discuss the general aims when adding simple drugs in
order to compose compound ones?
This information provides us with an interesting insight into

the transmission of Ibn Sarābiyūn’s Small Compendium.
Firstly, the seventh maqāla (section) is called ‘‘the third part
(ǧuz’)’’. This may suggest that the Small Compendium was
transmitted in three parts, the first possibly comprising maqāla
1–3, the second maqāla 4–6, and the third maqāla 7, which is
by far the longest, although there is the possibility that ǧuz’
(part) simply refers to the parts of a single copy. Furthermore,
according to the title, L contains the translation by Mūsā ibn
Ibrāhı̄m al-H
* adı̄tIı̄, the secretary (kātib) which, according to Ibn
Abı̄ Us*aybi‘a,35 was produced in the year 318 / 930. The Hebrew
owner’s note at the end is interesting in many respects. It
testifies to the involvement of the Meati family in medical
studies.36 But it also gives us a precious terminus ante for the
production of the manuscript: ‘‘in the month tišrı̄ of the year
73’’ refers to tišrı̄ 5073 of the Jewish calendar (Anno Mundi)
and roughly corresponds to September 1312 .37
Since L was translated by al-H * adı̄tIı̄, the question arises
whether B is based on the same translation. Because both
manuscripts contain the text of maqāla seven, we are in the
fortunate position to be able to compare the two. Lehmann had
already noticed that B sometimes di#ers from L and the Latin
translation by Gerard of Cremona (lat. G); for instance,
recipes are frequently introduced by the term (fabrication)
in B, while neither L nor lat. G have this term or its
Latin equivalent.38 Yet, generally speaking, both L and
B contain the same translations, as can be seen from the
35
See above pp. 239–40.
36
For S { emū’ēl ben Rabbı̄ S
{ elōmō HamMe’ātı̄ cf. A. Wasserstein, Galen’s
Commentary on the Hippocratic Treatise Airs, Waters, Places (Jerusalem, 1982),
esp. appendix II.
37
Assuming that the date refers to the (‘‘the small counting’’), which
only gives the centuries, not the millennia; cf. E. Mahler, Handbuch der jüdischen
Chronologie (Leipzig, 1916).
38
In a letter to Pieter Voorhoeve, dated 24 August 1952 and now kept together
with L, p. 5.
following passage from the beginning of chapter 15 on water-
purging drugs39:
Translation: The fifteenth chapter. Discussion about drugs that allevi-

ate wateriness and how to use them. One ought to use drugs alleviating
wateriness for people who have dropsy of the size of a waterskin [istisqā’
ziqqı̄]; women who su#er from white menstrual haemorrhage; people
who also have ulcers exuding pus; and people in whose bodily parts
there is flatulence, irritation and atrophy.
Plush-copper. Take one and a half mitIqāl of it with hydromel; one ought
to drink with it [thus B; L and lat. G: ‘‘shortly afterwards’’]43 vinegar,
lest one spit it out. Success is through God.
The di#erence between B and L, here, consists in B having
three additions which are in italics in the translation, namely
(‘‘and how to use them’’), (‘‘also’’) and
(‘‘Success is through God’’). These additions do not
occur in lat. G. In addition, in one instance in the short
passage quoted here, B reflects a textual tradition di#erent
from both L and lat. G (cf. n. 43); in other words: B has a
variant reading rather than a mere addition. One can speculate
whether these additions in B are those by Bar Bahlūl to which
Ea refers. We shall therefore now turn to this manuscript
which Troupeau has already described in part.46
39
The left-hand column contains lat. G. (fol. 76vb -25) and the right-hand
column the collated text of L (fol. 73b4) and B (fol. 85b -10).
40
] om. L.
41
] om. L et lat. G.
42
] B; L:
43
] B; L: ; cf. lat. G: ut odoret post parum acetum.
44
] om. L.
45
It seems that the lat. odoret (inhale) read a di#erent Arabic text ( ).
46
‘‘Du syriaque . . .’’, pp. 271–2.
YU
zH* ANNA
z IBN SARA
z BIYU
zN 245
Ea is a fragment, written in a somewhat sti# nash, which
contains the beginning of a copy of Ibn Sarābiyūn’s ˘ Small
Compendium. It is now bound together with other fragments of
medical texts such as the Kitāb al-H * āwı̄ by al-Rāzı̄, and the
Kitāb al-Malakı̄ by al-Maǧūsı̄. On the recto of the first folio of
47
the fragment (129a), we read the title of our work:48
The compendium (kunnāš) of Yūh * annā ibn Sarābiyūn, translated by

Mūsā ibn Ibrāhı̄m al-H
* adı̄tIı̄, with additions by Ibn al-Bahlūl.
Yet, as Troupeau already noted, the information provided on
the verso of this folio, containing the beginning of the actual
text, is at first glance contradictory49:
In the name of God, the merciful, the compassionate.

The first maqāla of Yūh * annā ibn Sarābiyūn’s Compendium (kunnāš),
translated by al-H * asan ibn Bahlūl al-T* ı̄rhānı̄.
Yūh* annā [ibn Sarābiyūn] said: Let us begin – with the help of God and
the benefit of His granting success – [to write] an abbreviated book on
the causes, indications, and treatment of diseases. The chapters (fus*ūl,
sg. fas*l) of this maqāla are thirty-two; the following are their headings
(ru’ūs):
The first on baldness, ophiasis and alopecia.
On the verso, Bar Bahlūl is mentioned as the translator, while

on the title recto, al-H
* adı̄tIı̄ is alleged to have translated this
text, and Bar Bahlūl is said to have contributed additions.
One might argue that the apparent contradiction could be
explained in the following terms: both the title and the begin-
ning of the treatise mean the same thing, namely that al-H * adı̄tIı̄
47
Cf. H. Derenbourg, H.-P.-J. Renaud, Les manuscrits arabes de l’Escurial,
vol. 2.2 (Paris, 1941), pp. 28–30.
48
Fol. 129a; cf. Troupeau, ‘‘Du syriaque . . .’’, p. 272.
49
Fol. 129b, Troupeau, ibid.
originally translated the treatise, and Bar Bahlūl made addi-
tions and corrections; on the verso, Bar Bahlūl is named as the
translator in a loose sense: he is the last editor, so to speak, of
the translation.50
We shall return to this problem later, but first of all, let us
consider the possible source for the additions alluded to on
the title page of Ea. In order to do so, we shall compare the
beginning of the first chapter in Ea and lat. G51:
Translation of the Arabic: The first chapter on baldness (s*ala‘ ), alopecia

(dā’ al-tIa‘lab) and the baldness (qara‘ ) called ophiasis (dā’ al-h
* ayya).
50
I shall propose a di#erent argument below, pp. 251–2.
51
Ea fols. 130b5–131a4; lat. G (fol. 2ra ult. #.).
YU
zH* ANNA
z IBN SARA
z BIYU
zN 247
This disease occurs for two reasons, as the excellent Galen taught us52:
Firstly because of lack of moisture nourishing the hair. Secondly, the
accumulation of unnatural and corrupting moistures that gather at the
roots of the hairs, even though baldness (s*ala‘ ) occurs because of lack of
moisture nourishing the hair, and for that there is no cure. This is
similar to plants and herbs which, once dried out, do not recuperate
because the soil is corrupted. The corruption of humours causes the
diseases called ophiasis and alopecia, but ophiasis is di$cult to cure,
more di$cult than alopecia. They both belong to the same category, if
you consider well what causes them, although their names are di#erent
with regard to form, since ophiasis has the appearance, on the head and
the rest of the body, of a snake shedding its skin; and by this means,
scaling occurs, which is similar to the skin shed by the snake. Alopecia
is called thus because it often occurs in foxes. If you consider well what
causes them, you will find them to be of the same category; yet they di#er
in action and in name: in action, since ophiasis is more di$cult to cure
than alopecia; or in name, since this is called ophiasis because of the
form of the snake and the shedding of its skin. Alopecia is called thus
because it occurs frequently in foxes.
As far as we can tell from this comparison, there are no

significant di#erences between the two texts. Yet there is one
feature which singles out the version contained in Ea. In Ea,
the term for ‘‘chapter’’ is fas*l, not bāb as in all the other
versions which have come down to us.53 So although Ea is
notably di#erent in one respect from the other versions, it does
not di#er greatly from lat. G on the level of content. We have
seen above that the additions in B (compared to the text in L)
did not have a Latin equivalent in lat. G. Since Ea follows lat.
G faithfully, then whatever additions are referred to on the
title page of Ea are not those contained in B. To put it
di#erently: when comparing B with L and lat. G, we found that
B has phrases not contained in lat. G. Since there is no such
divergence between Ea and lat. G, we can be confident that the
additions by Bar Bahlūl allegedly contained in Ea are not the
additions encountered in B.
We will come back to this first chapter of Ibn Sarābiyūn’s
work later, when discussing the similarities and di#erences
between the Small and Large Compendium. Before tackling the
problem of the di#erent translations of the Small Compendium
52
Galen, De compositione medicamentorum secundum locos (12.381, 11 Kühn), in
the chapter entitled:
` ’
ć
` ’
´ cc.
´ cc
`
53
In Eb, the term for ‘‘chapter’’ does not occur.
in greater detail, we need to examine the second Escorial
manuscript discussed by Troupeau, namely Eb. It consists
of only one leaf and is written in an Andalucian hand; it is
bound together with other medical writings by ‘Alı̄ ibn Rid * wān
and H* unayn ibn Ish * āq. The relatively short fragment is
54
interesting, since it allows us to compare it with the quotations

contained in the Kitāb al-Fāhir by Pseudo-Rāzı̄,55 which is
another important piece needed ˘ to solve the puzzle of the
di#erent translations.
Eb, fol. 39a, -7–39b6 Kitāb al-Fāh ir, ed. Koning, p. 122, 5 -ult.56
˘
Translation of Eb: If a hard swelling and stones occur, we can expect a

great danger. In this case, one ought to make an incision in the area
between the testicles down to the anus, and introduce into this same
place a tube so that the urine flows into it. Likewise, if retention of urine
occurs because of a blow, or something else is obstructing it [sc. the
urine], to the extent that the penis does not function and is erect so that
urine cannot flow through it, then we ought to leave it until the urine
flows constantly from it [sc. the penis], so that the wound may heal. For
it is necessary to use every method counteracting these causes from
which loss of life is feared, and to strive that the patient live, [even if it
be] in some dreadful state, and do not perish completely. This is the
end of the discussion on the diseases of the kidneys and the bladder
from Yūh * annā ibn Sarābiyūn’s book. Thanks be to God, Lord of the
universe.
54
Cf. Derenbourg, Renaud, Les manuscrits arabes de l’Escurial, 2.2, 59–60.
55
The Kitāb al-Fāh ir is not by al-Rāzı̄, as some scholars have often assumed.
L. Richter-Bernburg ˘(‘‘Pseudo-T I ābit, Pseudo-Rāzı̄, Yuh
* annā ibn Sarābiyūn’’) has
shown conclusively that this attribution is incorrect and should definitively be
abandoned; Troupeau (‘‘Du syriaque . . .’’, p. 276), who assumed that Kitāb
al-Fāh ir was composed by al-Rāzı̄, already compared the two versions.
56 ˘
P. de Koning, Traité sur le calcul dans les reins et dans la vessie (Leiden,
1896).
YU
zH* ANNA
z IBN SARA
z BIYU
zN 249
The two versions di#er considerably on the level of expression
and language employed. Small di#erences include the follow-
ing: in line 2, Eb has fa-yanbaġı̄ where Kitāb al-Fāhir has
fa-yaǧibu; line 4 šay’in āhara vs ġayri d ˘
I ālika; line 6 dā’iman
˘
minhu vs minhu dā’iman. Sometimes, Kitāb al-Fāhir departs
more substantially from Eb, as is the case in lines˘7–8 (line 9
not lending itself to comparison). Yet on the level of content,
there is hardly any di#erences. The question now arises
whether Kitāb al-Fāhir represents a di#erent version of Ibn
˘
Sarābiyūn’s Small Compendium, or whether the discrepancies
can be explained in terms of pseudo-Rāzı̄ having quoted his
source more loosely. U. Weisser has shown that al-Rāzı̄ in his
Kitāb al-H
* āwı̄ is not always faithful to his source; in not a few
cases, he strongly paraphrases rather than cites verbatim.57
The same could be true for the author of Kitāb al-Fāhir. The
˘
last sentence (lines 7–8) is quite di#erent in the two versions.
The Latin rendering by Gerard of Cremona (lat. G) runs as
follows:58
Si autem accidit illic apostema di$cile et times ex ipso timorem
magnum, tunc oportet ut administremus perforationem in loco qui est
inter testiculos et anum, et intromittas in locum ipsum cannulam, et
faciamus ut currat urina in ea. et similiter etiam quando accidit
expressio urinae ex percussione aut ex re alia quae sequitur ipsum donec
attrahatur cannula uirgae, et attrahatur donec non currat in ea urina.
tunc oportet ut perforetur, et dimittatur cannula donec sanetur illud
quod in circuitu est loci, et fit sursus urinae assiduae ex ea. Quod est
quia oportet ut administremus omni ingenio ea quae contraria sunt istis
causis ex quibus timetur perditio uiuorum et eligamus ut uiuat homo in
qualibet parte mali et non moriatur statim subito.
While Eb has wa-d I ālika annahu (this is because), Kitāb

al-Fāhir has the simpler wa-qad; lat. G sides with Eb: Quod est
quia.˘Likewise, for the rest of the sentence, lat. G follows the
diction of Eb much more closely than Kitāb al-Fāhir. On the
other hand, there are instances where Eb and Kitāb ˘ al-Fāhir
˘
together side against lat. G. In line 1, for example, both Eb and
Kitāb al-Fāhir have the juncture waramun *sa‘bun wa-h * as*an
˘
(a hard swelling and calculi), while lat. G only says apostema
57
U. Weisser, ‘‘Zitate aus De methodo medendi im H * āwı̄’’, in G. Endress,
R. Kruk (eds.), The Ancient Tradition in Christian and Islamic Hellenism
(Festschrift Drossaart Lulofs), (Leiden, 1997), pp. 279–318.
58
(Venice, 1525), fol. 19ra26 #.
difficile (a hard swelling), thus omitting to translate wa-h * as*an
(and calculi). This analysis may suggest two things: (1) the
author of Kitāb al-Fāhir had in front of him a version of the
Small Compendium which ˘ was closer to Eb than to the Arabic
source of lat. G; and (2) the author of Kitāb al-Fāhir took some
liberties when citing his source, sometimes paraphrasing ˘ it,
although generally being quite faithful. In that sense, Pseudo-
Rāzı̄’s attitude is in keeping with that of al-Rāzı̄ described by
U. Weisser.
Ullmann, Richter-Bernburg, and Troupeau all tackled
aspects of the question of what to make of the many quotations
in Rāzı̄’s Kitāb al-H* āwı̄ taken from Ibn Sarābiyūn.59 Ullmann,
followed by Richter-Bernburg,60 for instance, speculated
whether some of the quotations in Rāzı̄’s H * āwı̄ come from the
Large rather than the Small Compendium, since the quotations
in Rāzı̄’s H
* āwı̄ deviate from the text of the Small Compendium
contained in B.61 We can be sceptical as to whether Rāzı̄’s H * āwı̄
as Troupeau claimed,62 is closer to lat. G than e.g. P. We have
seen above that B, Ea, Eb are all, despite certain small
variations, quite similar to the presumed source of lat. G.
That this is also the case for P will be demonstrated in the
following.
P is yet another fragment of the Small Compendium, also
bound together with other medical texts.63 It contains a
number of errors, as Troupeau rightly notes.64 One chapter,
that on jaundice (IV 9), is entirely preserved in this
manuscript. It o#ers us the possibility to compare the Latin
translation with the Arabic contained in P and two di#erent
versions which occur in Rāzı̄’s H * āwı̄:
59
Kitāb al-H* āwı̄ fı̄ al-t*ibb, 1st ed., 23 vols. (Hyderabad, 1955–70); 2nd ed.
(Hyderabad, 1974 #.; repr. Beirut, 2000).
60
Ullmann, ‘‘Sarābiyūn’’, pp. 290–1, Richter-Bernburg, ‘‘Pseudo-T I ābit, Pseudo-
Rāzı̄, Yuh* annā ibn Sarābiyūn’’, pp. 74–7.
61
I will return to this point in the third part of the present contribution, when
discussing the relationship between the Small and the Large Compendium.
62
Troupeau, ‘‘Du syriaque . . .’’, p. 273: ‘‘Quant à la traduction latine de ce
fragment par Gérard de Crémone . . . elle semble plus proche de la citation
d’al-Rāzı̄ que du texte du manuscrit de Paris’’.
63
Cf. MacGuckin Le Baron de Slane, Bibliothèque Nationale. Département des
manuscrits. Catalogue des manuscrits arabes (Paris 1883–95), p. 522.
64
‘‘Du syriaque . . .’’, pp. 275–6.
YU
zH* ANNA
z IBN SARA
z BIYU
zN 251
lat. G P, fol. 165a8 Rāzı̄ H
* āwı̄ 7.157, ult. (a) Rāzı̄ H
* āwı̄ 7.168, 8 (b)
This synoptic table is illuminating in many respects. First of all

it shows that al-Rāzı̄ himself will change the text of the
quotation for no apparent reason. In line 1, both P and Rāzı̄
H
* āwı̄ (a) have birāz (faeces) while Rāzı̄ H * āwı̄ (b) has Itufl
(dregs), that is to say, he replaces a term with a synonym.
* āwı̄ (a) is similar to P on the level of content,
In line 2, Rāzı̄ H
but paraphrases rather than cites its source; for instance
the expression al-maǧārı̄ al-munaqqiyyatu li-al-marāri (the
passages which cleanse the bile) in P is changed into mas*abb
al-marār (the outlet of the bile) in Rāzı̄ H* āwı̄ (a), while in Rāzı̄
H
* āwı̄ (b) he changes the overall meaning. Line seven contains
another example of al-Rāzı̄’s changing the passage for no
apparent reason other than fondness for variation: P has
fa-laysa al-waǧa‘u fı̄ al-kabidi while al-Rāzı̄ says fa-lā ‘illata fı̄
al-kabidi ‘illatun (a) and fa-laysa bi-al-kabidi ‘illatun (b). The
second conclusion to be made when comparing the di#erent
versions is the fact that lat. G is closer to P than either
quotation in al-Rāzı̄. To pick just one example, in line 2, the
text in P fa-inna d I ālika yadullu ‘alā anna (this indicates that)
is translated in lat. G as tunc significat (then it indicates),
while the idea of ‘‘indicating’’ is absent from both versions
in al-Rāzı̄.
Let us now return to the original question: can we match Ibn
Abı̄ Us*aybi‘a’s account of the di#erent translators with the
conclusions drawn from the di#erent versions just discussed
Both Ea and L claim to be translated by al-H * adı̄tIı̄, although in
the case of the former, Bar Bahlūl seems to have revised
65
correxi.
66
In lines 7–8, this is the reading of the second edition, also reproduced in the
Beirut reprint (see above n. 59); the first edition has .
al-H* adı̄tIı̄’s version. L and B are for the most part identical,
although B sometimes has additions or variant readings di#er-
ent from both L and lat. G. In general, B, Ea, Eb, L, and P all
agree on the level of content with lat. G, although B, as
mentioned before, sometimes has small additions. There is one
fundamental di#erence between Ea on the one hand and B, L
and P on the other: the former refers to the chapters in the
book as fus*ūl (sg. fas*l), while the latter call them abwāb (sg.
bāb). P has numerous scribal errors, while the overall quality
of the other manuscripts is somewhat better. Furthermore,
Pseudo-Rāzı̄, the author of Kitāb al-Fāhir, and al-Rāzı̄ in his
Kitāb al-H * āwı̄ both quote from an Arabic ˘ version, but when
doing this, take certain liberties both at the level of expression
as well as content. Kitāb al-Fāhir can side with Eb against
lat. G. ˘
Ibn Abı̄ Us*aybi‘a said that there were three translations.
That by al-H * adı̄tIı̄ made in the year 930, was allegedly more
idiomatic than that by Bar Bahlūl.67 Ibn Abı̄ Us*aybi‘a also
mentions another translation by Abū Bišr Mattā. I would like
to propose the following argument. Let us assume that Ea
contains the translation by Bar Bahlūl, and not that by
al-H* adı̄tIı̄. Al-H * adı̄tIı̄ is mentioned as the translator only on the
title page (fol. 129a) of Ea, and not at the beginning of
the treatise itself (fol. 129b). Dr Álvarez-Millán68 confirmed
through autopsy what Troupeau69 already stated, namely that
the title on the recto and the beginning of the treatise on the
verso are by the same scribe. But there are inconsistencies
between the title (fol. 129a) and the beginning (fol. 129b): the
title, for instance, refers to Bar Bahlūl incorrectly as ‘‘Ibn
al-Bahlūl’’ while the beginning has the standard ‘‘Ibn Bahlūl’’.
This inconsistency could be explained in the following
terms: while the scribe copied the beginning of the treatise
from a reliable manuscript, he may have added the title under
his own steam or produced it from a di#erent, unreliable
source.
Additionally, if we accept that L contains the translation by
al-H* adı̄tIı̄ as indicated in the manuscript, we could explain the
di#erence in referring to ‘‘chapter’’ as fas*l and bāb respectively
67
See above pp. 239–40.
68
I would like to thank Dr A u lvarez-Millán for inspecting the manuscript
for me.
69
‘‘Du syriaque . . .’’, p. 272.
YU
zH* ANNA
z IBN SARA
z BIYU
zN 253
in the two manuscripts by the fact that L was translated by
someone else other than Ea. Furthermore, on the level of
content, both translations, that by Bar Bahlūl (Ea) and that by
al-H* adı̄tIı̄ (L), resemble each other quite closely, as can be seen
from the fact that they are similar to lat. G. There is, however,
another unsolved question: what translations did al-Rāzı̄ and
Pseudo-Rāzı̄ use In the case of al-Rāzı̄, it cannot have been
the version by al-H * adı̄tIı̄ for chronological reasons, as
Richter-Bernburg argued: al-Rāzı̄ died 925, and al-H * adı̄tIı̄ trans-
lated the Small Compendium in 930.70 On the other hand, we
know that Bar Bahlūl must have been active around the year
963, because Mārı̄ ibn Sulaymān writes in his book on the
Nestorian Patriarchs that Bar Bahlūl recommended ‘Abd zIšū‘
for the o$ce of the patriarch.71 That makes it highly unlikely
that al-Rāzı̄ used Bar Bahlūl’s version, and if we assume that
there were only the three mentioned by Ibn Abı̄ Us*aybi‘a, then
it follows that al-Rāzı̄ used that by Abū Bišr Mattā (d. 940).
Ullmann compared al-Rāzı̄ with B and concluded that, apart
from relatively small discrepancies, the two versions go back to
the same source, that is two say, represent the same transla-
tion.72 However, if B contains the same translation as L (with
some additions), if L consists of al-H * adı̄tIı̄’s translation and if
this translation was produced in 930 (with emphasis on the
‘‘ifs’’), then Ullmann must be incorrect for the chronological
reasons mentioned above.
In order to illustrate the similarities and di#erences between
the two texts, and in order to determine what source was used
by Ibn Sarābiyūn in the chapter ‘‘On Snakebites’’, I o#er the
following synoptic table:
70
Richter-Bernburg, ‘‘Pseudo-T
I ābit, Pseudo-Rāzı̄, Yuh
* annā ibn Sarābiyūn’’,
pp. 68–9.
71
H. Gismondi, Maris Amri Slibae De Patriarchis Nestorianorum Commentaria,
2 vols. (Rome, 1899), vol. 1, (txt.), 89 (tr.):
(Abū al-H
* asan ibn Bahlūl mentioned the bishop of Ma‘latIāyā [saying] that he is
apt for the o$ce of the patriarch.)
72
Ullmann, ‘‘Sarābiyūn’’, p. 290: ‘‘. . . da die Texte in der Hauptsache doch
übereinstimmen, sollte man annehmen, daß al-Rāzı̄ dieselbe Übersetzung
ausgeschrieben hat, die uns in der Brüsseler Handschrift erhalten ist.’’
1) lat. G fol. 52vb–5 2) PAeg. ii. 16,1–17,8 3) B fol. 17b -6 4) R<S 5) R<P73
The first column contains the Latin translation by Gerard, the

second the beginning of the chapter ‘‘On snakes’’ taken from
Paul of Aegina’s
´,74 the third the Arabic version
contained in B, and the forth al-Rāzı̄’s quotation from Ibn
Sarābiyūn (R<S) and the fifth al-Rāzı̄’s quotation from the
Arabic translation of Paul of Aegina (R<P). As we have seen
before and as Ullmann is right to insist, al-Rāzı̄ often changed
his source text, so that it is di$cult to determine whether the
di#erences are due to his altering the version from which he
quotes, or to his using a di#erent translation. We can surely say
that, on the level of content, R<S and B are quite similar, with
73
B has previously edited by Ullmann, ibid. p. 285, §§ 1–3; R<S is taken from
* āwı̄ 19.403, 12; R<P is taken from Rāzı̄ H
Rāzı̄ H * āwı̄ 19.390,7.
74
Ed. I. L. Heiberg, CMG 9.1–2, 2 vols. (Leipzig and Berlin, 1921/1924).
YU
zH* ANNA
z IBN SARA
z BIYU
zN 255
only slight omissions (e.g. lines 4–5) and variations. On the
level of language, things are more complicated.
Ibn Sarābiyūn used Paul of Aegina as his source in this
chapter, as in a number of other instances when dealing with
toxicology and gynaecology.75 I put R<P into the synoptic
table in order to be able to check the di#erences between R<S
and B against R<P, that is to say against quotations which
come from a completely di#erent translation. The question is:
are the di#erences between R<P and B greater than those
between R<S and B. The answer must be: only marginally. In
line 3, for instance, R<S uses the same verb as B for ‘‘to bite’’
(nahaša), while R<P employs a di#erent one (ladaġa), though
in the latter case the tense is di#erent (tanhašuhum vs.
nahašathum). On the other hand, in line 8 R<P sides with B
(tIumma for ‘‘there’’), while R<S has ba‘da d I ālika (similar line
14). In some cases, the fact that R<S and R<P agree against B
illustrates a certain linguistic preference of al-Rāzı̄; take, for
instance, the use of man instead of allad I ı̄ as relative pronoun
(line 3). Therefore, the comparison of R<S and B is inconclu-
sive as to whether the two are based on the same translation.
This means that, despite Ullmann’s cautious assumption that
R<S is based on the same source as B, it is more likely that
al-Rāzı̄ quoted from Abū Bišr Mattā’s translation.
Let me now draw some tentative conclusions about the
translations: Ea contains Bar Bahlūl’s translation of the Small
Compendium. L is based on al-H * adı̄tIı̄’s version, as is B,
although the latter sometimes has additions and variant read-
ings, not contained in lat. G. At this point, we cannot ascertain
on which translation P and Eb are based. Be that as it may, on
the level of content, all the manuscripts discussed here are
generally speaking very similar to each other as can be seen
from the fact that they rarely deviate from lat. G. Al-Rāzı̄
(Kitāb al-H* āwı̄) and Pseudo-Rāzı̄ (Kitāb al-Fāhir), when quot-
ing from Ibn Sarābiyūn, rephrase and paraphrase, ˘ which makes
it di$cult to determine from which translation their quotations
are taken. For chronological reasons, it seems likely that
al-Rāzı̄ quoted from the version by Abū Bišr Mattā, but for lack
of a critical edition and analysis of the Kitāb al-Fāhir, it is
˘
75
Cf. P. E. Pormann, The Oriental Tradition of Paul of Aegina’s
´,
Studies in Ancient Medicine 29 (Leiden, 2004), pp. 21–2.
impossible to determine with certainty which translation
Pseudo-Rāzı̄ used.
Ibn Abı̄ Us*aybi‘a said that the version by al-H * adı̄tIı̄ is ‘‘more
idiomatic’’ (ah * sanu ‘ibāratan) than that by Bar Bahlūl.
Troupeau was surprised by this statement and declared that Ea
is written in good Arabic.76 Yet, it appears that Ibn Abı̄
Us*aybi‘a did not criticise the grammatical quality of Bar
Bahlūl’s translation but rather its style. For instance, one
might qualify Bar Bahlūl’s use of ru’ūs (sg. ra’s) for ‘‘headings’’
as a syriacism (cf. 77 ); in fact, the use of ra’s for ‘‘chapter’’
is so unusual, that F.W. Zimmermann thought it odd.78 Bar

Bahlūl, in his glossary, also gives as the standard translation
for ´
the Arabic term ru’ūs.79
The Latin rendering by Gerard of Cremona (lat. G) was
translated into Hebrew by Mōšē ben Maz*liah * and is extant in
Oxford, Bodleian MS Mich. 207 (Ol. 569).80 Furthermore,
Andreas Alpago reworked lat. G, although his revision often
does not go much beyond orthography.81 Since his translation
is based on lat. G, which is extant, one should always use lat.
G when trying to recover the original Syriac or the Arabic
translation.
76
‘‘Du syriaque . . .’’, p. 274: ‘‘La langue du traducteur [sc. Bar Bahlūl] . . .
est grammaticalement correcte et . . . son expression . . . n’est pas du tout
mauvaise, . . .’’.
77
Cf. C. Brockelmann, Lexicon Syriacum, 2nd ed. (Halle, 1928), p. 728a s.v. n. 8.
78
F. W. Zimmermann, ‘‘The origins of the so-called Theology of Aristotle’’, in
J. Kraye et al. (eds.), Pseudo-Aristotle in the Middle Ages: The Theology and
Other Texts (London, 1986), pp. 110–240, p. 170: ‘‘As a translation of kephalaia,
ru’ūs would be a Graecism or Syriacism’’; but cf. P. E. Pormann, ‘‘The Alex-
andrian Summary (G { awāmi‘ ) of Galen’s On the Sects for Beginners: Commentary
or Abridgment’’, in P. A. Adamson et al. (eds.), Philosophy, Science and Exegesis
in Greek, Arabic and Latin Commentaries, 2 vols. (London, 2004) [forthcoming].
79
Lexicon Syriacum, ed. Duval, col. 1823, 15:
K´
according to Bar Serōšwai: chapters (rēšē), like the ´
of the
blessed Marc; chapters (ru’ūs).
80
Cf. Ullmann, ‘‘Sarābiyūn’’, p. 282; Troupeau, ‘‘Du syriaque . . .’’, p. 273;
A. Neubauer, Catalogue of the Hebrew Manuscripts in the Bodleian Library
(Oxford, 1886), no. 2087.
81
Cf. Troupeau, ibid.; this translation was printed posthumously as Serapionis
. . . Practica studiosis medicinae utilissima quam postremo Andreas Alpagus
Bellunensis . . . in latinum convertit: cujus translatio nunc primum exit in lucem
(Venice, 1550); for a more general appreciation of Alpago, see F. Lucchetta, Il
medico e filosofo bellunese Andrea Alpago († 1522), traduttore di Avicenna; profilo
biografico (Padua, 1964).
YU
zH* ANNA
z IBN SARA
z BIYU
zN 257
Before coming to the Large Compendium, it is necessary to
add a caveat. I called my conclusions ‘‘tentative’’ for at least
two reasons: first of all, because they are based on attributions
of translations in manuscripts and the bio-bibliographical
tradition. G. Endress, D. Gutas and M. Ullmann have amply
demonstrated that these attributions are often fictitious, and
that one should therefore be cautious not to take them at face
value.82 Second, we dealt for the most part with unpublished
texts here; once the fragments of the Small Compendium,
including the fragments in the Kitāb al-Fāhir, are edited, a new
˘
picture might emerge. This said, the conclusions drawn above
seem relatively probable and can serve as a hypotheses which
future scholarship will have to test.
III
We have seen that scholars speculated about the relationship
between the Small and the Large Compendium. Is the former
just an abbreviated version of the latter? Can the quotations
in al-Rāzı̄ which deviate from the Small Compendium be
explained as al-Rāzı̄ having employed the Large Compendium?
Until F. Sezgin published the Istanbul, Kütüphane Süley-
maniye, MS Aya Sofya 3716 in facsimile, scholars thought that
this manuscript contained the Large Compendium.83 In reality,
it is the medical aide-mémoire of a tenth-century Baghdad
physician called al-Kaskarı̄.84 Yet, within this work by al-
Kaskarı̄, there are a number of quotations from the Large
Compendium, which constitute the only fragments explicitly
attributed to this work. In the following, I shall present the
82
G. Endress, D. Gutas (eds.), A Greek and Arabic Lexicon (Leiden etc., 1992
#.), vol. 1, pp. 8*–9*; G. Endress, ‘‘Die griechisch-arabischen Übersetzungen und
die Sprache der arabischen Wissenschaften’’, in Symposium Graeco-Arabicum II
(Amsterdam, 1989), pp. 103–46, p. 110; reprinted in slightly modified form as:
‘‘8.7 Die Entwicklung der Fachsprache’’, in W. Fischer (ed.), Grundriß der
Arabischen Philologie, 3 vols. (Wiesbaden, 1982–92), vol. 3, p. 6. M. Ullmann
[Wörterbuch zu den griechisch-arabischen Übersetzungen des 9. Jahrhunderts
(Wiesbaden, 2002)] recently and brilliantly gave an excellent example of the
confusion as to who translated Galen’s On the Powers of Simple Drugs into
Arabic; cf. JRAS, 13.1 (April 2003): 105–7.
83
F. Sezgin, Book On Medicine. Kunnāš by Ya‘qūb al-Kaškarı̄, Publications of
the Institute for the History of Arabic-Islamic Science, Series C, vol. 17
(Frankfurt, 1985); cf. P. E. Pormann, ‘‘Theory and practice in the early hospitals
in Baghdad – Al-Kaškarı̄ On Rabies and Melancholy’’, Zeitschrift für Geschichte
der Arabisch-Islamischen Wissenschaften, 15 (2003): 197–248.
84
Pormann, ibid., pp. 199–202.
reader with an edition and translation of the opening of the
most important fragment, namely that on capillary disorders.85
This will lead to a comparison with a similar fragment from the
Small Compendium already discussed. Finally, I will address
some more global questions about Ibn Sarābiyūn and his work.
At the beginning of his aide-mémoire, al-Kaskarı̄ quotes an
entire chapter from Ibn Sarābiyūn’s Large Compendium:
Translation:
The first chapter of this book on the generation of hair over all [sā’ir]86
the body, from the Large Compendium by Ibn Sarāfiyūn, [comprising]
thirteen books [maqālas].
1. Since we aimed at discussing forms of harm befalling hair,
we ought first of all to investigate the cause which generates it
[hair].
85
Preserved at the beginning of the treatise on fol. 1a; the other quotations in
this manuscript are found on fols. 54b8; 55b7; 85a–7; 135a–6.
86
For this use of sā’ir in the sense of ‘‘all’’ cf. W. Wright, A Grammar of the
Arabic Language, 2 vols., 3rd ed. (Cambridge, 1896), vol. 1, p. 206D; R. Dozy,
Supplément aux dictionnaires arabes, 2 vols. (Leiden, 1881), vol. 1, p. 621a (s.v.),
E. W. Lane, Arabic-English Lexicon (London, 1863–93), p. 1282c (s.v.).
YU
zH* ANNA
z IBN SARA
z BIYU
zN 259
2. These causes for hair – like for all bodies subject to generation
and corruption – are four: the matter [material cause], the e#ects
[e#ective cause], the instrument [instrumental cause] and perfec-
tion [final cause]. Accordingly, we need to seek out these selfsame
[four causes] in the case of hair.
3. The matter of hair is a smoky vapour. There are two kinds of
vapour, as we learn from Aristotle87: one is dry and is the smoky
one, the other is wet and is the misty one. The dry vapour
corresponds to earth and fire, while the wet vapour corresponds to
water and air. Since the two [vapours] are generated by these two
[elements], hair does not consist of wet vapour, but of dry smoky
vapour. We can therefore say that hair is generated by warm and
dry matter.
4. The e#ective cause is the heat which kindles the vapour pushing
out [the hair] and brings it out.
5. The instrumental cause are the pores in the skin of the head and
the rest of the body, in which the smoky vapour gets stuck because
of its being [too] thick, so that is compacted, hardens and finally
becomes hair.
6. The final general cause is the cleansing of the body and the
purging of thick, vaporous superfluities. The specific cause is either
for the embellishment and beauty, as for instance the hair of the
beard; or for protection, as for instance in the case of the hair of
the eyelids or the eyebrows. Therefore, the growth of hair in warm
and dry bodies is strong, for the hair, in order to generate, requires
that the body possess warmth and dryness, so that in it [sc. the
body] smoky vapours be generated which are the matter of which
the hair is made.
There is a problem with the alleged number of maqālas of

which the Large Compendium consists. According to
al-Kaskarı̄, they are thirteen, but Ibn al-Nadı̄m only talks of
twelve.88 There is no obvious solution for this discrepancy,
apart from supposing a scribal error ( vs ) or
the possibility that a chapter was subdivided into two in
al-Kaskarı̄’s exemplar.
To this chapter on the causes of hair growth in the Large
Compendium, there is no precisely corresponding chapter in
the Small Compendium. The closest we get is the chapter edited
above from Ea on baldness, ophiasis and alopecia (bk 1, ch. 1).
In the extract from the Large Compendium, Ibn Sarābiyūn is
87
Aristotle, Meteorologica, 378a18: ´ ` ` ‘
’
´ c
c, ¢ ` ’
´ c
¢ć , ’
c
´ cf. id. De gen. et corrupt., 782b 8–20.
¢ ` ´ c,
88
See above p. 234.
quite concerned with causality. He first states that it is
necessary to know the causes of a disease in order to treat it
(1), and since each body has four causes, he will review them
also in the case of hair (2). These causes are, of course, the four
Aristotelian causes (material, e$cient, instrumental and
final). In §§ 3–6, Ibn Sarābiyūn discusses each one of them in
turn. When beginning to do so, the first authority invoked is
Aristotle, not Galen. Later on in the chapter, he mentions
Galen once, but only to say that the latter has taken a theory
from the former.89 In the Small Compendium we also find a
concern with causes: bk 1, ch. 1 opens with the statement:
‘‘This disease occurs for two reasons [li-sababayn], as the
excellent Galen taught us’’. But where the Large Compendium
follows Aristotelian ideas, the Small Compendium relies on
Galen. One might say that the former is more philosophical
than the latter.
This combination of Aristotelian ideas with Galenic medi-
cine is a specific feature of late antique Alexandria. In order to
illustrate this statement, I would like to quote just one instance
from its medical literature. Stephen of Athens, an author who
wrote a number of commentaries on Hippocratic works, is prob-
ably the same as a philosopher by the name of Stephen, having
lived in Alexandria in the 580s.90 Even if this Stephen cannot be
identified with absolute certainty, he clearly relies on, and
belongs to, the Alexandrian tradition.91 In his commentary on
the Aphorisms, he links Hippocratic theory to Aristotelian
philosophy92:
89
Fol. 2b6:
(This theory [sc. about why hair turns white] Galen has taken from Aristotle for
the latter mentions it [sc. this theory] in his books On Animals).
90
Cf. W. Wolska-Conus, ‘‘Stéphanos d’Athènes et Stéphanos d’Alexandrie.
Essai d’identification et de biographie’’, Revue des études byzantines, 47 (1989):
5–89; ead. ‘‘Les commentaires de Stéphanos d’Athènes au Prognostikon et aux
Aphorismes d’Hippocrate: de Galien à la pratique scolaire Alexandrine’’, Revue
des études byzantines, 50 (1992): 5–86.
91
W. Wolska-Conus, ‘‘Sources des commentaires de Stéphanos d’Athènes, et de
Théophile le Protospathaire aux Aphorismes d’Hippocrate’’, Revue des études
byzantines, 54 (1996): 5–66.
92
L. G. Westerink (ed.), Stephanus of Athens: Commentary on Hippocrates’
Aphorism, i: Sections I–II, Corpus Medicorum Graecorum 11.1.3.1 (Berlin, 1985),
p. 48, 28. The translation is that of Westerink with slight modifications.
YU
zH* ANNA
z IBN SARA
z BIYU
zN 261
’E
ćc ` ’ ´ ˛ ˛y

´˛
` ´ c ` ´ y
‘I ´ c
` ´ ´ ’ ´ . ` c
´ ´ c
´ cc´ ’ c
c

` ’´

,

ò ’
` ¢
`
´ ,
.
Also worthy of notice and admiration in this preface is Hippocrates’

technique and his way of reasoning. Everything which comes into being
has four generating causes: an e$cient, an instrumental, a material and
a final cause, etc.
Thus we can immediately see the close link between
the methods of describing natural processes in the Large
Compendium and in Stephen of Athens’s commentary: they
both resort to the four causes; yet Stephen is by no means the
only medical author to do so. A lot more could be said about
this topic, as well as the intertwining of medical theory and
philosophy, but space does not permit to explore it fully here.93
We can furthermore appreciate that the Large Compendium
di#ers from the Small Compendium; not only is it the case that
a whole chapter from the former has no exact equivalent in the
latter, but they diverge also in their theoretical orientation.
But does this initial comparison between the two texts allow us
to ascertain whether authors such as al-Rāzı̄, when quoting
from Ibn Sarābiyūn without reference to the exact work, used
the Small or the Large Compendium, especially, where there is
discrepancy between the quotation and the Latin translation of
the Small Compendium? In general, one has to be sceptical,
since there is always the possibility that the author citing Ibn
Sarābiyūn is responsible for the change or the addition. In the
case of al-Rāzı̄, this is quite likely, since he was notably
careless with his sources and did not recoil form manipulating
and rephrasing them. Moreover, al-Rāzı̄ did not have the
opportunity to put the final touch to his Kitāb al-H
* āwı̄, since it
was compiled from notes following his death.
93
Cf. J. Du#y, ‘‘Byzantine medicine in the sixth and seventh centuries: Aspects
of teaching and practice’’, in J. Scarborough (ed.), Symposium on Byzantine
Medicine, Dumbarton Oaks Papers 38 (1983), pp. 21–7, p. 22; M. Roueché,
‘‘Did medical students study philosophy in Alexandria?’’, Bulletin of the Institute
of Classical Studies, 43 (1999): 153–69; P. E. Pormann, ‘‘Jean le Grammarien et le
De sectis dans la littérature médicale d’Alexandrie’’, in I. Garofalo, A. Roselli
(eds.), Galenismo e medicina tarcoantica: fonti greche, latine e arabe (Napoli, 2003),
pp. 233–63, 251–2.
At the end of this article, I would like to take up and discuss
one other suggestion by H. Lehmann.94 Because of the exten-
sive use of Greek and Syriac idioms, Lehmann advanced the
hypothesis that Ibn Sarābiyūn might have been a Syro-Greek
(‘‘ein Syro-Grieche’’); by this he meant that Ibn Sarābiyūn was
of Greek origin, as one might also conjecture from the name of
his father, Serapion, which is that of an old family of Greek
doctors. Whether the name ‘‘Serapion’’ is a valid indication for
Ibn Sarābiyūn’s Greek origin remains doubtful. But the frag-
ment from the Large Compendium above points in the direction
of Alexandria. Ibn Sarābiyūn may or may not have known
Greek, but he was certainly influenced by Greek medical theory
as it was developed in Alexandria in late antiquity.
I have discussed the transmission of a medical text from
Syriac into Arabic (and later Latin and Hebrew). Some of the
questions posed by previous scholarship have been tentatively
answered, and other problems put into better focus. Yet, the
main task still remains in front of us: to edit the Arabic
fragments of Ibn Sarābiyūn’s Small and Large Compendium
and, on the basis of such an edition, to discuss Ibn Sarābiyūn’s
place in the development of medical theory and practice in the
ninth and tenth century. I am currently preparing such an
edition, which will hopefully facilitate further research into
this fascinating author.95
94
Contain in the letter already quoted (see above n. 38).
95
Peter E. Pormann, (The Small and Large
Compendium of Ibn Sarābiyūn) (Dār al-kutub al-‘ilmiyya: Beirut, 2006)
[in preparation].
PLATO’S REPUBLIC IN ARABIC
A NEWLY DISCOVERED PASSAGE
DAVID C. REISMAN
My aim here is to present an editio princeps of a newly

discovered Arabic translation of a very important passage
from Plato’s Republic (Book VI, 506d3–509b10) found in the
work entitled Kitāb fı̄ Masā’il al-umūr al-ilāhiyya (hereafter
Masā’il), penned by the somewhat obscure Neoplatonist Abū
H* āmid al-Isfizārı̄ (fl. mid-fourth / tenth c.). While an edition of
al-Isfizārı̄’s work has been published by Daniel Gimaret, the
manuscript he used (Ragıp Paşa 1463) lacked the literal trans-
lation of the Republic passage.1 The one other known exemplar
of the work, MS Z * āhiriyya 4871, dated slightly later than the
first, appears to be closer to the author’s original unedited
version; it contains the Republic passage.2 I am currently
preparing a new edition of the work on the basis of the
Z
* āhiriyya manuscript, but it seemed worthwhile to bring
the Republic passage to immediate attention. I do not attempt
here a thorough evaluation of al-Isfizārı̄’s philosophy as
presented in the Masā’il; this can be done successfully only
upon completion of the edition. However, to place the work in
its proper context, I do provide a brief overview of the trans-
mission of Plato’s Republic in Arabic; a discussion of new
information on al-Isfizārı̄’s Nachlass, and a description of the
1
Daniel Gimaret, ‘‘Un traité théologique du philosophe musulman Abū H * āmid
al-Isfizārı̄,’’ Mélanges de l’Université Saint-Joseph, 50.1 (1984): 210–52.
2
I thank Jamil Ragep for bringing the Z * āhiriyya manuscript to my attention in
Istanbul in 1998 and subsequently providing me with a photocopy of the relevant
folios. I thank Moujir M. Omari and Racha Omari for providing me with an
additional photocopy of the folios of the same manuscript, which helped clear up
some ambiguities in the first copy. I have access to the Ragıp Paşa manuscript
from the collection of microfilms of the late Franz Rosenthal (rad * iya Allāhu
‘anhu!). I also take the opportunity here to thank Dimitri Gutas, Peter Adamson,
Jon McGinnis and Asad Ahmed for their comments and suggestions on what
follows here as a whole. A. Hasnaoui made numerous corrections to the edition
and translation for which I am deeply grateful.
264 DAVID C. REISMAN
two exemplars of his Masā’il. The edited text of the passage
from the Republic is accompanied by a provisional translation.
I. PLATO’S REPUBLIC IN ARABIC

‘‘Complete translations of Platonic Dialogues, . . . according to
the information obtainable from Arabic Bibliographies, were
made very rarely. Not a single one of them has come down to
us, and the character of those quotations which we have before
us never seems, as far as we can now judge, to a#ord grounds
for the slightest probability that we are concerned with the
remains of a pure and complete text of a Platonic Dialogue;
therefore, a certain doubt may be entertained as to whether
the translations mentioned were verbal reproductions of
an unaltered Platonic wording.’’3 As is clear from Franz
Rosenthal’s carefully worded statement, it is commonly
accepted that no integral Arabic translation of any of Plato’s
dialogues was made during the more than two centuries of the
Graeco-Arabic translation movement. This general statement
may reasonably be extended to the case of Plato’s Republic.
What knowledge the authors working in Arabic had of the
Republic seems to have come in a piecemeal fashion from
summaries, abridgments, quotations or short references in
doxographies and commentaries.4 Thus, H * unayn ibn Ish * āq
(d. c. 260 / 873) tells us that of Galen’s Synopsis of the Platonic
Dialogues (˜
´ ´
, G { awāmi‘ kutub
Aflātu* n) he found (waǧadtu) only four of the eight books. Book
I covered Cratylus, Sophist, Politics, Parmenides, and Euthyde-
mus. Book II covered the first four books of the Republic (Kitāb
fı̄ al-Siyāsa). Book III covered ‘‘the remaining six books of the
Republic’’, along with Timaeus. Book IV covered Laws (ǧumal
ma‘ānı̄ al-itInatay ‘ašara maqāla fı̄ al-siyar). H* unayn states that
he translated the first three Books, thus including the synopsis
of the Republic in its entirety, for Abū G { a‘far Muh * ammad ibn
3
Franz Rosenthal, ‘‘On the knowledge of Plato’s philosophy in the Islamic
world, with addenda,’’ Islamic Culture, 14 and 15 (1940–1): 387–422, 396–8; see
p. 393.
4
See F. Klein-Franke, ‘‘Zur Überlieferung der platonischen Schriften im
Islam,’’ Israel Oriental Studies, 3 (1973): 120–39, esp. pp. 133f.
PLATO’S REPUBLIC IN ARABIC 265
Mūsā.5 Ibn al-Nadı̄m informs us in his Fihrist that H * unayn
wrote comment(s) (fassara) on Kitāb al-Siyāsa.6 Only stray
passages remain of the Arabic translation of Galen’s synopsis
of the Republic and, presumably, of H * unayn’s comment(s). The
evidence available to us for reconstructing these works is
problematic, in that there is little of it, and what there is raises
questions about the nature of their format.
T
I ābit ibn Qurra (d. c. 288 / 901) wrote a Risāla fı̄ H
* all rumūz.
Kitāb al-Siyāsa, according to the bibliography of his works
compiled by his great-great-grandson al-Muh * assin (d. 401 /
1010) as preserved in al-Zawzanı̄’s abridgment of al-Qift*ı̄’s
Ta’rı̄h al-h* ukamā’,7 as well as a Kitāb fı̄ al-Siyāsa which may
˘
also have drawn on some form of the Republic.8 According to
5
G. Bergsträsser, H * unain Ibn Ish * āq, Über die syrischen und arabischen
Galen-Übersetzungen, Abhandlungen für die Kunde des Morgenlandes (Deutschen
Morgenländischen Gesellschaft), XVII.2 (Leipzig, 1966), pp. 50–1 (Arabic), p. 41
(German). There is nothing unexpected to add here from Bergsträsser’s Neue
Materialien zu H * unain Ibn Ish* āq’s Galen-Bibliographie, Abhandlungen für die
Kunde des Morgenlandes (Deutschen Morgenländischen Gesellschaft), XIX.2
(Leipzig, 1932). See also V. Boudon, ‘‘Galien de Pergame,’’ in R. Goulet (ed.),
Dictionnaire des philosophes antiques (Paris, 1989), p. 459. In S * iwān al-h
* ikma (ed.
A. Badawı̄ [Tehran, 1974], p. 95) we find the specific title Kitāb G { awāmi‘ Kalām
Aflāt*ūn fı̄ Siyāsat al-mudun, which approaches the form in which al-Isfizārı̄ cites
the work, scil. Kitāb fı̄ Siyāsat al-mudun. This is not to be confused with the
Tadbı̄r al-mudun (Politics), found also in the S * iwān (ibid., p. 89). I digress
momentarily to note the peculiarly recurrent and widespread scribal error of
al-badan for al-mudun in both titles; found in al-Isfizārı̄’s Masā’il in another brief
reference (Ragıp Paşa 1463, 33v, l. 12), in S * iwān al-h* ikma, p. 95, n. 1, and in
al-‘Az mirı̄’s Kitāb al-Amad, ed. E. Rowson (New Haven, 1988), p. 84, n. 32. These
examples could be multiplied endlessly.
6
(Cairo, s.n.), p. 357. Dimitri Gutas informs me (email, 28 July 2003) that by
fassara Ibn al-Nadı̄m is indicating nothing other than H * unayn’s translation of the
Galen synopses and some of the evidence Gutas has collected in ‘‘Aspects of
literary form and genre in Arabic logical works,’’ in C. Burnett (ed.), Glosses and
Commentaries on Aristotelian Logical Texts, The Syriac, Arabic, and Medieval
Latin Traditions (London, 1993), p. 33, certainly suggest broader connotations.
However, the evidence from the lemma found in Ibn Buhtı̄šū‘ (see below) in
which H * unayn supplies a lexical comment after a literal quotation ˘ of a Republic
passage suggests, to my understanding, that here we might take Ibn al-Nadı̄m
literally. But given the speculative nature of this conclusion, I refer to H * unayn’s
work here throughout as ‘‘comment(s)’’.
7
Ed. J. Lippert (Leipzig, 1903), p. 120. See Dimitri Gutas’ comprehensive
‘‘Works of T I ābit b. Qurra,’’ in R. Rashed (ed.), Thābit b. Qurra, Science and
Philosophy in 9th Century Baghdad (London, forthcoming 2004) [I thank him for
providing me with a partial draft]. There Gutas numbers the work 116 and adds
the suggestion that it may have explained ‘‘the allegories or parables in Plato’s
Republic’’.
8
Gutas, no. 118; the work was originally in Syriac and then translated into
Arabic.
al-Mas‘ūdı̄, T I ābit’s son Sinān (d. 331 / 943) also wrote a work,
dedicated to colleagues in the court bureaucracy, on ethics
(ahlāq) and the divisions of souls, in which he included
˘
‘‘sparks’’ (luma‘ ) from Plato’s Republic (here, Kitāb fı̄
al-Siyāsa al-madaniyya).9 All of these works remain lost.
Al-Fārābı̄ (d. 339 / 951) had some knowledge of the Republic,
we must suppose, through H * unayn’s translation of Galen’s
synopsis or H * unayn’s comment(s), but his presentation is
not without problems. In his Kitāb al-G { am‘ bayna ra’yay
al-h* akı̄mayn, Fārābı̄ quotes Book II, 359a (the ‘‘non-just lies
between justice and injustice’’),10 and Book VIII, 546e–547a.11
But in the same work, Fārābı̄ refers to the Republic but
actually quotes Galen’s De Moribus,12 and, finally, concludes
9
Mas‘ūdı̄, Murūǧ al-d I ahab, ed. C. Barbier de Meynard and Pavet de Courteille
(Paris, 1861), vol. 1, p. 18; revised edition by Ch. Pellat (Beirut, 1966), vol. 1,
pp. 16–17. Mas‘ūdı̄ correctly describes the Republic as comprising ten books.
He also provides additional information on the contents of the work, which, after
the ‘‘sparks’’ from the Republic, covers the history of ‘‘kings and ministers’’,
historical reports (ah bār) all the way up to Caliph al-Mu‘tad * id (r. 892–902),
˘
‘‘discussing his friendship with him and his earlier days with him.’’ In a rare
moment of inattention, Rosenthal, ‘‘On the knowledge,’’ p. 417, attributed the
work in question to T I ābit himself. Dimitri Gutas (email communication, 6 July
2003) notes that T I ābit is better known as al-Mu‘tad * id’s boon companion, Sinān
being the confidant and physician of Caliph al-Muqtadir. This discrepancy,
however, is not enough to strengthen the perhaps unconscious association made
by Rosenthal. Gutas also suggests possible identifications of Sinān’s works from
al-Qift*ı̄’s bibliography in Ta’rı̄h al-h * ukamā’ (ed. Lippert, p. 197), including the
˘
Risāla fı̄ Ta’rı̄h mulūk al-suryāniyyı̄n and al-Rasā’il al-sult*āniyyāt wa-al-
ih wāniyyāt. The ˘description of the work by Mas‘ūdı̄ might be considered a little
˘
peculiar, in that it combined psychology with history, but the composition may
have been aimed at providing the kuttāb with a general knowledge of rulership.
10
Ed. F. M. Najjar as L’harmonie entre les opinions de Platon et d’Aristote, with
French translation by D. Mallet (Damas, 1999), p. 95; cf. the English translation
by Charles E. Butterworth, in Alfarabi, the Political Writings: ‘‘Selected
Aphorisms’’ and Other Texts (Ithaca, 2001), p. 141.
11
Ibid., p. 109.
12
Ibid., pp. 109–10. This is the quote translated by Rosenthal, The Classical
Heritage in Islam (London, 1992), p. 91 (and referred to by Mallet in the notes to
Najjar’s edition of the G { am‘, p. 173): ‘‘The young tree bends easily in every
possible direction whereas it is di$cult, and sometimes impossible, to transplant
a fully grown tree.’’ See De Moribus, ed. P. Kraus as Kitāb al-Ah lāq in Bulletin of
the Faculty of Arts of the Egyptian University, 5 (1939): 1–52, ˘esp. p. 31. Fārābı̄
tries to cover his bases by including in his reference Plato’s so-called al-Būlı̄t*ı̄
(although the manuscript readings on this point are very complicated). This is
to be compared to Fārābı̄’s concatenation of a Kitāb al-Būlı̄t*ı̄qı̄ and a Kitāb
al-Siyāsa by Aristotle in certain manuscripts of his Ih **sā’ al-‘ulūm; see Shlomo
Pines, ‘‘Aristotle’s Politics in Arabic philosophy,’’ Israel Oriental Studies, 5
(1975): 150–160, p. 150, n. 3a.
the work with an oblique reference to Book X, 614b #.13 There
is not enough evidence here to establish whether Fārābı̄ had
access to Galen’s synopsis, or H * unayn’s comment(s), nor
indeed, if he did, to establish the nature of those works,
particularly with regard to whether or not they preserved the
dialogue form.14
In the Epistles of the Ihwān al-S * afā’ we find a correct
reference to the ‘‘second book ˘ of the Republic (Kitāb al-Siyāsa)
for the story about the descent of Gyges into the chasm and
discovery of the magic ring, in a fairly close translation of that
narrative from the Republic, 359d #.15
Medieval Arabic medical writings contain an unusual
amount of references to Plato. In Abū Sa‘ı̄d ibn Buhtı̄šū‘’s (d. c.
450 / 1058) Risāla fı̄ al-T
* ibb wa-al-ah ˘
* dātI al-nafsāniyya, we find
a very important literal translation of a passage from the
Republic (here Kitāb al-Siyāsa al-madaniyya), Book III,
402e–403c, and accompanied by a gloss, attributed to H * unayn
and presumably from his comment(s), on what the Greeks
understood by the term ’´
(Ar. ‘išq).16 Ibn Buhtı̄šū‘ also
˘
13
Ibid., p. 159.
14
Much of Fārābı̄’s presentation of the books of Plato in his Falsafat Aflāt*un
is problematic; see his description of the Republic (Kitāb fı̄ al-Siyāsa) in the
edition by Walzer / Rosenthal, p. 20 and the translation by M. Mahdi, Alfarabi’s
Philosophy of Plato and Aristotle (Ithaca, New York, 1969), p. 65. Further
allusions to the Republic are to be had in his Tah **sı̄l al-sa‘āda, in ibid., pp. 35–6,
48. Shlomo Pines, ‘‘Aristotle’s Politics in Arabic philosophy,’’ detected in a
passage from only some manuscripts of Fārābı̄’s Ih **sā’ al-‘ulūm ‘‘political
doctrines . . . reminiscent of Plato rather than Aristotle.’’ This is slim pickings
indeed. Finally, I am not entirely convinced by the argument adduced by Bruce
Eastwood, ‘‘Al-Farabi on extramission, intromission, and the use of Platonic
visual theory,’’ Isis, 70 (1979): 423–5, that Fārābı̄’s treatment of vision in his
Mabādi’ ārā’ ahl al-madı̄na al-fād * ila, in which he makes the comparison between
light and vision and the action of the active intellect on the material intellect,
necessarily confirms that Fārābı̄ refers there to ‘‘the text of Plato’’ (i.e., Republic
507d–508d, incidentally part of the very passage found in al-Isfizārı̄’s quotation),
since the comparison had become such a common trope in Neoplatonic works
by Fārābı̄’s time, but more research obviously needs to be done on Fārābı̄’s
knowledge of Plato.
15
Rasā’il Ih wān al-S
* afā’, ed. Hayr al-Dı̄n Ziriklı̄ (Cairo, 1928), vol. 4, pp. 324–5;
cf. Rosenthal, ˘ ‘‘On the knowledge,’’˘ p. 397 (the reference here clears up
Rosenthal’s vaguer one).
16
Ed. F. Klein-Franke, Über die Heilung des Krankheiten der Seele und des
Körpers, Recherches, Nouvelle série, B. Orient chrétien, IV (Beyrouth, 1977), pp.
46–7 (Arabic), pp. 79–80 (German); to be used with the review by H. Biesterfeldt
in Bibliotheca Orientalis, 38 (1981): 772. Klein-Franke had earlier translated and
analyzed the Republic passage, along with a passage in the same work from the
Laws, in his ‘‘Zur Überlieferung,’’ pp. 129–32.
presents another passage from the Republic (Book X,
614b–615e) in his commentary on Pseudo-Galen’s Tah * rı̄m dafn
al-ah
* yā’. We must imagine in both cases that Ibn Buhtı̄šū‘
cites from Galen’s synopsis of the Republic or from H ˘
* unayn’s
comment(s). But only in the case of the first passage
17
(402e–403c) is the dialogue form of the Republic passage

preserved.18
The brief reference by Fārābı̄ to the passage in the Republic
dealing with justice is found in considerably expanded form
in Kitāb al-Sa‘āda wa-al-is‘ād, once thought to be by Abū
al-H
* asan ibn Abı̄ D
I arr (fl. fourth / tenth c.?), and now ascribed,
with insu$cient caution by most, to al-‘A z mirı̄ (d. 381 / 992).19
The work contains a lengthy epitome of portions of the
Republic (Kitāb al-Siyāsa), concentrated mostly in Book II and
which, while cast in dialogue form, does not reproduce ‘‘liter-
ally’’ Plato’s words. A. J. Arberry identified passages covering
Book I–II, 333–366, Book IX, 579, 585–7, and Book X, 613;
elsewhere in the work there is a summary of Book III, 416 #.20
There are additional quotations of Plato which, to my
knowledge, have yet to be identified.
Ibn Rušd’s commentary on the Republic, extant now only in
Hebrew, apparently also drew on Galen’s synopsis.21 If this is
the case, and there is no reason to question it, we would expect
Ibn Rušd’s work to give us a relatively good sense of the nature
of Galen’s synopsis. But, given the commentary style of Ibn
Rušd’s work, it is no surprise that the dialogue form of the
Republic is not preserved. Perhaps significantly, it lacks
17
Id., Über die Heilung, intro. pp. 8#. provides some evidence for this.
18
I have access only to Klein-Franke’s German translation of this passage in
his ‘‘Zur Überlieferung,’’ p. 132.
19
See Everett K. Rowson, A Muslim Philosopher on the Soul and its Fate:
al-‘Az mirı̄’s Kitāb al-Amad ‘alā l-abad (New Haven, 1988), pp. 15–17, who notes
M. Minovi’s confident attribution of the Sa‘āda to al-‘A z mirı̄ and the acceptance
of that attribution by F. Rosenthal, A. Ghorab and M. Arkoun.
20
Arberry discussed the work in ‘‘An Arabic treatise on politics,’’ Islamic
Quarterly, 2 (1955): 9–22, and edited and translated the lengthiest quotation in
‘‘Some Plato in an Arabic epitome,’’ Islamic Quarterly, 2 (1955): 86–99. See the
facsimile of a handcopy of the unique manuscript Chester Beatty 3702 by
M. Minovi in Al-Sa‘āda wa-al-is‘ād fı̄ al-sı̄ra al-insāniyya (Wiesbaden,
1336S { / 1957–8), pp. 233#., for the passages edited by Arberry.
21
See E. I. J. Rosenthal, Averroes’ Commentary on Plato’s Republic, University
of Cambridge Oriental Publications, 1 (Cambridge, 1966), introduction, p. 9. For
the medieval Hebrew and Latin translations, and modern English, German,
Italian and Spanish translations, see Hans Daiber, Bibliography of Islamic
Philosophy (Leiden, 1999), vol. 1, nos. 4558–4562; vol. 2, index, p. 438.
entirely (whether as lemma or gloss) the passage of concern to
us here.
This brief overview of knowledge of the Republic among
medieval Arabic authors is necessarily incomplete, but it
provides a good idea of the scanty evidence available to us for
establishing in what form precisely these philosophers and
physicians knew the Republic.22 This is an important consider-
ation. The quotation by Ibn Buhtı̄šū‘ might lead us to believe
˘ comment(s) of H
that either Galen’s synopsis or the * unayn may
have preserved the dialogue form of the work. In the case of
Galen’s synopsis, this is di$cult to accept at face value, given
that the only complete Galen synopsis of one of Plato’s works
preserved for us in Arabic, i.e. the Timaeus, does not.23 We can
say nothing comprehensive about H * unayn’s comment(s). A
newly discovered translation of a significant passage from the
Republic from al-Isfizārı̄’s Masā’il al-umūr al-ilāhiyya, edited
and translated below, does preserve the dialogue form. It seems
hardly credible that an Arabic translation of the Republic in its
entirety, and preserving the dialogue form, was in fact made.
The somewhat contradictory evidence concerning the nature of
Galen’s synopsis and H * unayn’s comment(s) should make us
22
I note in passing that Barhebraeus (d. 685 / 1286) cites a Syriac In Platonis
Rem Publicam in his Chron. Syr., according to Rainer Degen, ‘‘Galen im
Syrischen: Eine Übersicht über die syrische Überlieferung der Werke Galens,’’ in
V. Nutton (ed.), Galen: Problems and Prospects (London, 1981), pp. 131–66, p. 155
(no. 96), which Degen identifies as ‘‘Komm. zu Platons Phaedon (?).’’ The answer
to this riddle is to be found in G. Levi della Vida’s ‘‘Two fragments of Galen in
Arabic translation,’’ JAOS, 70 (1950): 182–7: other authors refer to Galen’s
synopsis of the Republic at this place in the biographical presentation [I thank
Hidemi Takahashi for clearing this up for me]. Finally, I note in passing the
rather large literature on the reference to Jesus, quoted from Galen’s synopsis of
the Republic by, among others, ‘Alı̄ ibn Yah * yā al-Nadı̄m, apud S
* iwān al-h
* ikma
(ed. Badawı̄, pp. 94–5): ‘‘ ‘Alı̄ ibn Yah * yā al-Nadı̄m stated in his book that he
compiled on history (ǧama‘ahu fı̄ al-ta’rı̄h ?) that Galen stated in his Book
{ awāmi‘ Kalām Aflāt*ūn fı̄ Siyāsat al-mudun˘ something that indicated that [Galen
G
came] after the time of Jesus.’’ R. Walzer’s Galen on Jews and Christians
(London, 1949) is best known here for this issue; a more recent, and very
thorough discussion is to be had from Stephen Gero’s ‘‘Galen on Christians, A
reappraisal of the Arabic evidence,’’ Orientalia Christiana Periodica, 56 (1990):
371–411.
23
Edited by Paul Kraus and Richard Walzer as Galeni Compendium Timaei
Platonis aliorumque dialogorum synopsis quae extant fragmenta, Corpus
Platonicum Medii Aevi, Corpus Platonicum, 1 (London, 1951). Regarding
preservation of dialogue form, the fact that Galen’s synopsis does not is little
indication here, however, given the nature of the Timaeus itself. The Kraus /
Walzer edition, Arabic, pp. 35 #., also contains some additional testamonia on the
survival of the Republic in Arabic.
wary about how much direct access each and every one of the
authors who quote the Republic had to either of these texts.
However, the one example that we might have been able to use
for comparison with the dialogue quotations from al-Isfizārı̄
and Ibn Buhtı̄šū‘ makes it di$cult to believe that quotations
that preserve ˘ the dialogue form came from Galen’s synopsis.
Ibn Rušd’s commentary made use of Galen’s synopsis and it
omits entirely the dialogue passages found in the works of
al-Isfizārı̄ and Ibn Buhtı̄šū‘.24 Given our relative ignorance
about H ˘
* unayn’s comment(s), we might envision other circum-
stances for the dialogue passages. I began the discussion above
with the assumption that, aside from Galen’s synopsis of the
Republic, any additional information on the text that Arabic
authors had access to probably derived from some sort of
doxography or compendium of political writings. This is the
theory presented by Shlomo Pines for any extant remains of
Aristotle’s Politics in Arabic.25 But if this is the case, we must
assume that such doxographies did not contain always mere
summaries of the Greek texts but, at least in some cases,
included almost verbatim translations.
Particularly good evidence for the often quite precise
quotations extant in such works was presented by Dimitri
Gutas with regard to Plato’s Symposion.26 Al-Isfizārı̄ introduces
his quotation of the Republic by stating twice that what he is
presenting is the ‘‘literal words’’ of Plato (bi-nafs alfāz*ihi)27;
this sort of confidence in his source suggests that al-Isfizārı̄ had
access to a translation of considerable authority. I would
hazard the guess that there was more of Plato’s Republic
circulating among medieval Arabic authors than the synopsis
24
See the ‘‘lacuna’’ in his commentary between Republic 505d and 507b in
Rosenthal’s translation, pp. 184–6; Ibn Rušd overlooks the end of the work
(Ibn Buhtı̄šū‘’s passage) in favor of general derogatory statements on Plato’s
methodology˘ (ibid., 250–1). However, this does not explain the absence of the
quotation that we find in al-Isfizārı̄’s work.
25
Pines, ‘‘Aristotle’s Politics.’’
26
‘‘Plato’s Symposion in the Arabic tradition,’’ Oriens, 31 (1988): 36–60,
drawing on, among other sources, Ibn Buhtı̄šū‘ and H * unayn ibn Ish
* āq’s Nawādir
al-falāsifa. See also his ringing endorsement˘ of Pines’ theory in ‘‘Galen’s Synopsis
of Plato’s Laws and Fārābı̄’s Talh ı̄s*,’’ in G. Endress and R. Kruk (eds.), The
Ancient Tradition in Christian and˘ Islamic Hellenism (Leiden, 1997), pp. 101–19,
p. 117, n. 25.
27
Ibn Buhtı̄šū‘’s phrase is no less authoritative: ‘‘the text of his words’’
˘ ed. Klein-Franke, loc. cit.; there is no equivocation here about the
(nas**s qawlihi),
‘‘sense’’ or ‘‘gist’’ of Plato’s work.
of Galen, some source which, while perhaps not a translation of
the entire work, most certainly contained sophisticated trans-
lations of the more important passages for Neoplatonism in
Islam.
II. AL-ISFIZA
z RIz AND HIS LEGACY
The unfortunate legacy of Abū H * āmid Ah* mad ibn Abı̄ Ish * āq
al-Isfizārı̄, an almost wholly unremembered scientist and
philosopher of fourth / tenth century Hurāsān, has been
compounded by the regular misidentification ˘ of him among
both medieval and modern scholars. Credit goes to Daniel
Gimaret for rehabilitating al-Isfizārı̄ by collecting what scant
information can be had on him from the medieval biogra-
phers.28 Gimaret came upon his name in an ‘‘enigmatic
passage’’ by the historian Ibn ‘Asākir who recounts a study
session in which one Abū al-‘Abbās, discoursing on the issue of
tawh * ı̄d, charged al-Isfizārı̄ and the philosopher al-Kindı̄ with
širk.29 Gimaret located three additional references to al-Isfizārı̄:
(1) in the bio-bibliographical collection known as S * iwān
al-h
* ikma; (2) in al-S { ahrastānı̄’s (d. 548 / 1153) Kitāb al-Milal
wa-al-nihal; and (3) a very brief, wholly uninformative biogra-
phy in al-Bayhaqı̄’s (d. 565 / 1165) Tatimmat S * iwān al-h
* ikma.
In the S * iwān al-h * ikma, al-Isfizārı̄ (only his nisba is given) is
present at a maǧlis of Abū G { a‘far al-Babūya along with Ibn
H
* ibbān, T * alh
* a and Abū Tammām.30 Gimaret reasonably
concluded that al-Isfizārı̄ ‘‘was a philosopher at the court of
Siǧistān around 955–960, in the company [of those men-
tioned]’’.31 Al-S { ahrastānı̄ lists one Abū H * āmid Ah * mad ibn
Muh * ammad al-Isfizārı̄ in his list of philosophers, where he
comes between T * alh* a ibn Muh * ammad al-Nasafı̄ and ‘Izsā ibn
28
‘‘Sur un passage énigmatique du Tabyı̄n d’Ibn ‘Asākir,’’ Studia Islamica, 47
(1978): 143–63.
29
Gimaret provided a French translation of the passage in full in ibid.,
pp. 143–6, and identified the true origin of the passage (the preface to Bazdawı̄’s
Kitāb Us*ūl al-dı̄n) which Ibn ‘Asākir put in the unknown Abū al-‘Abbās’ mouth.
30
S* iwān al-h
* ikma, ed. A. Badawı̄ (Tehran, 1974), p. 317.4–5. See also Joel L.
Kraemer, Philosophy in the Renaissance of Islam: Abū Sulaymān al-Sijistānı̄ and
his Circle (Leiden, 1986), p. 17 (with translation of the passage) and pp. 22f. (with
additional information on the other philosophers). Kraemer also notes the later
authors Ibn Sab‘ı̄n and al-S { ahrastānı̄, who appear to have copied from the S
* iwān.
31
Ibid., p. 157.
‘Alı̄ ibn ‘Izsā (d. 393 / 1003).32 The brief ‘‘biography’’ by al-
Bayhaqı̄ tells us only that Abū H * āmid Ah * mad ibn Ish * āq
al-Isfizārı̄ was a mathematician and philosopher and that
‘‘what he said in his writings is refined, without reproach and
unsullied by defect’’. Al-Bayhaqı̄ also includes ‘‘statements’’ he
attributes to al-Isfizārı̄, but which are likely of questionable
authority.33
The confusion surrounding the correct reading of the nisba
al-Isfizārı̄ was already noted by the S { āfi‘ı̄ biographer al-Subkı̄
(d. 771 / 1370) who, in his notice on Abū H * āmid al-Isfarā’inı̄
(d. 406 / 1016) in T * abaqāt al-šāfi‘iyya, says that he should not be
confused with the philosopher Abū H * āmid al-Isfizārı̄, and
recalls a reading of al-S { ahrastānı̄’s Milal where someone read
al-Isfarā’inı̄ instead of al-Isfizārı̄ in the list of philosophers.34 In
an ironic modern echo of al-Subkı̄’s report, Mawaldı̄, in his
edition of Avicenna’s Risāla fı̄ al-Zāwiya (more on which
below)35 identified him as the S { āfi‘ı̄ jurist Abū H * āmid Ah* mad
ibn Muh * ammad ibn Ah * mad al-Isfarā’inı̄, the very identification
al-Subkı̄ warned against.36 Suter misidentified the name he
read as ‘‘el-Isfarledi’’ as Abū H * ātim al-Muz*a#ar al-Isfizārı̄
(fl. mid-sixth / twelfth c.), the mathematician and contempo-
rary of ‘Umar Hayyām.37 Helmut Ritter, who first listed the
exemplar Ragıp˘ Paşa 1463 of al-Isfizārı̄’s Masā’il, calls him
al-Isqārı̄.38
The corruptions, Isfazārı̄>Isfarladı̄ or Isfarā’inı̄, can be
accounted for through orthographic irregularities. In multiple
32
Kitāb al-Milal wa-al-nihal, ed. Muh * ammad Sayyid Kı̄lānı̄ (Beirut, 1406 /
1987), vol. 2, p. 158; French tr. Jean Jolivet and Guy Monnot, Shahrastani, Livre
des religions et des sectes (Paris, 1986), vol. 2, p. 365 and n. 13. See also Kraemer,
Philosophy in the Renaissance of Islam, p. 90.
33
Gimaret, p. 159, rightly finds it of little use to reproduce them. Kraemer,
Philosophy in the Renaissance of Islam, summarizes al-Bayhaqı̄.
34
Apud Gimaret, p. 158.
35
Risāla fı̄ al-Zāwiya li-Ibn Sı̄nā, ed. M. Mawaldı̄ in Maǧallat Ma‘had
al-Mah*tūt*āt al-‘Arabiyya, 42 (1998): 33–93; reference here is to p. 56, n. 7.
36 ˘
On the jurist, see, for instance, Ziriklı̄, al-A‘lām (Beirut, 1979), vol. 1,
p. 211a.
37
Die Mathematiker und Astronomen der araber und ihre Werke, N 268.
Without reference to Suter, Gimaret noted that al-Muz*a#ar has his own entry in
al-Bayhaqı̄’s Tatimma. In his misreading of ‘‘Isfarledi’’ Suter was preceded by
L. Sédillot; see GAL, S I, p. 856.
38
H. Ritter, ‘‘Philologika III,’’ Der Islam, 28 (1929): 34–59, p. 46. His contents
list of the manuscript is emended below.
manuscripts, the aliph and rā’ of Isfizārı̄ are sometimes joined
and the zā’ is regularly undotted (thus , ).39
The confusion over the nisba is attended by disagreement
among the geographers on the pronunciation, and indeed
location, of Isfizār. Yāqūt reads Asfizār or Asfuzār. Sam‘ānı̄
pronounces it Isfizār. None agree on the location of the village
of Isfizār. Sam‘ānı̄ and Subkı̄ say it is a village between Harāt
and Siǧistān; Yāqūt says it is a village of Siǧistān en route to
Harāt. Is**tahrı̄ calls it a district in Hurāsān, in the region of
Harāt.40 ˘ ˘
Before concluding the dreary history of al-Isfizārı̄’s obscurity,
we must address the variations in his full name found in the few
brief references to him. To recount, al-Bayhaqı̄ has Abū H * āmid
Ah* mad ibn Ish * āq al-Isfizārı̄ and al-S
{ ahrastānı̄ has Abū H * āmid
Ah* mad ibn Muh * ammad al-Isfizārı̄. We must add to these what
is probably the correct reading from the title page of his
Masā’il in MS Ragıp Paşa 1463: Abū H * āmid Ah * mad ibn Abı̄
Ish
* āq al-Isfizārı̄.41 We can reconcile the minor variants in the
versions of al-Isfizārı̄’s name by suggesting that al-S { ahrastānı̄
provided Abū H * āmid’s father’s name ‘‘Muh * ammad’’ in place of
his kunya Abū Ish * āq and that al-Bayhaqı̄’s ‘‘Ish * āq’’ is a scribal
error that omitted ‘‘Abı̄’’ (this would not be surprising since
the ‘‘b.n.’’ of ‘‘Ibn’’ is often attached to the first name and
thus overlooked and the ‘‘Abı̄’’ may have been taken as ‘‘Ibn’’).
His full name might then be Abū H * āmid Ah * mad ibn Abı̄
Ish
* āq Muh * ammad al-Isfizārı̄. However, the addition of
Muh * ammad cannot be established confidently on the basis
of al-S{ ahrastānı̄’s reference alone.
The relative lack of knowledge about al-Isfizārı̄ is relieved by
the only later author whom al-Isfizārı̄ seems to have greatly
39
See for instance MSS Pertev Paşa 617, Istanbul University 4724, and
Nuruosmaniye 4894 of Avicenna’s Risāla fı̄ al-Zāwiya; the other exemplars of
Avicenna’s work, viz. Yale Beinecke Arabic Suppl. 51, Ayasofya 4829 and
Ayasofya 4849 have the correct rasm. Gimaret, pp. 158 and 160, provides
additional examples from other works.
40
Gimaret, p. 158. Kraemer, Philosophy in the Renaissance of Islam, p. 3, notes
the river Asfizār (Hārūd) that flowed into Lake Zarah in Sı̄stān / Siǧistān.
41
The other exemplar, Z * āhiriyya 4871, bungles it: Abū Ah * mad ibn Ish
* āq
al-Isfizārı̄ (fol. 135r, l. 2).
influenced (if only negatively): Avicenna.42 Avicenna refers to
al-Isfizārı̄ twice: once in his Letter to Ibn Zayla and another
time in his Risāla fı̄ al-Zāwiya. In his Letter to Ibn Zayla,
preserved in the Mubāh * atIāt,43 Avicenna responds to Ibn Zayla’s
charge of apparent inconsistency in his definitions of ‘‘practi-
cal’’ and ‘‘theoretical’’ philosophy as he outlined them in the
Prologue to the S { ifā’.44 In his response, Avicenna distinguishes
between the use of the term al-h * ikma al-‘amaliyya, practical
knowledge or wisdom, which is a disposition that produces
ethical actions, and the use of the term al-h * ikma al-‘amaliyya to
mean al-falsafa al-‘amaliyya, or the discipline that constructs
principles of ethical behavior. The latter is that part of philos-
ophy called ‘‘practical’’. The former is ‘‘one of the three ethical
virtues [along with courage and temperance]’’.45 Avicenna
notes in concluding his discussion that people err in their
understanding of this homonymous term, and he singles out
Abū H * āmid al-Isfizārı̄, who ‘‘made courage and temperance
equilibria [between two extremes] and made wisdom (h * ikma)
have no equilibrium (ġayr wasat*iyya)’’.46 He also notes that
Isfizārı̄ considered too much attention to obligatory actions
a vice.47
In his Risāla fı̄ al-Zāwiya, Avicenna provides a list of di#er-
ent views on the definition of ‘‘angle’’, one of which he calls the
‘‘new doctrine’’ (mad I hab muh * datI) and which argues that ‘‘an
angle is a certain quantity, but is neither a line nor a surface
nor a body; rather it is a species of quantity which is brought
about by two other species.’’ With his customary superior
air, Avicenna describes al-Isfizārı̄ as someone who ‘‘believes
himself to be filled with the foundations of philosophy because
42
There may be others influenced by al-Isfizārı̄, but I have found none to date.
There may be some evidence for a positive influence by al-Isfizārı̄ on Avicenna;
see the early philosophical treatment of the so-called Light Verse of the Quran by
al-Isfizārı̄ below in the text and translation. Avicenna would make this a
recurrent motif of his own philosophy.
43
An edited extract of the letter can be found in M. Bı̄dārfar’s edition,
al-Mubāh * atIāt (Qum, 1992), paragraphs 567–578. Discussion of the textual trans-
mission and recension process of this letter can be found in my The Making of
the Avicennan Tradition (Leiden, 2002), pp. 57–9 (manuscripts), 102–3 (brief
discussion of the unedited version of the letter), 198–200 (translation of the
introduction), 283–4 (edition of the introduction).
44
See al-S { ifā’, al-Madh al, p. 12, ll. 5–7, and p. 14, ll. 17–18.
45
The addition is from ˘ MS Ayasofya 4855, fol. 50r, l. 11.
46
MS Ayasofya 4855, fol. 50r, ll. 16–17.
47
Ed. Bı̄dārfar, p. 189, ll. 13–15. To Bı̄dārfar’s credit, he rightly identified the
Isfizārı̄ to whom Avicenna made reference, p. 189, note to par. 570.
of his autodidactic investigation (naz *aruhu fı̄hā bi-nafsihi) and
the self-imposed limitation [of his studies] to some of the books
by the ancients (awā’il).’’48
In his Masā’il al-Isfizārı̄ makes reference to three of his
books: Kitāb fı̄ al-H * ikma al-‘amaliyya;49 Kitāb fı̄ Ta‘arruf al-h* āl
ba‘d al-mawt; and Kitāb fı̄ H
50 * adatI al-‘ālam.51 We can tenta-
tively identify the first as the source of Avicenna’s reference in
his Letter to Ibn Zayla. The origin of Avicenna’s presentation
of al-Isfizārı̄’s views on the angle remains unknown. I hesitate
for now to give a complete characterization of al-Isfizārı̄’s
philosophy; this must come upon completion of the full edition
of his Masā’il. That he is linked to al-Kindı̄ in the quotation by
Ibn ‘Asākir, and to the circle of Abū Sulaymān al-Siǧistānı̄ in
the S
* iwān would suggest a generally Neoplatonist outlook, and
this may be confirmed by the text presented below.
We have one final piece of autobiographical information
from al-Isfizārı̄’s Masā’il: he concludes the work by noting that
he is imprisoned in Hwārazm.52 Given his association with the
court of the S * a#ārid˘ king Abū G
{ a‘far, we might imagine that
his imprisonment was associated with the death of that king in
352 / 963,53 but the fact that he was imprisoned in Hwārazm
might be evidence that he su#ered under the Afrı̄ġids ˘ or
perhaps lived long enough to witness the Ma’mūnid invasion in
385 / 995, but this might be far too late a date for a philosopher
who flourished in the middle part of the century.54
The importance of the quotation of the Republic found in
the Masā’il cannot be underestimated for the history of
Neoplatonism and medieval Arabic philosophy in general.55
Now that the evidence for the knowledge of this passage among
Arabic-writing philosophers is firmly established by the
example of al-Isfizārı̄’s Masā’il, I hope that others more
48
Risāla fı̄ al-Zāwiya li-Ibn Sı̄nā, ed. Mawaldı̄, p. 56. I thank Irina Luther for
translation of the first quotation, from her forthcoming study of Avicenna’s work.
49
Ragıp Paşa 1463, fol. 32r, l. 17; ed. Gimaret, q.v.
50
Ibid.; ed. Gimaret, q.v.
51
Ragıp Paşa 1463, fol. 38r, ult.; ed. Gimaret, q.v.
52
See Gimaret, Arabic text, p. 251, and his brief unanswered speculation on the
cause of the imprisonment, French intro., p. 210.
53
For the history of his reign, see Kraemer, Philosophy in the Renaissance of
Islam, pp. 6#.
54
See C. E. Bosworth, The New Islamic Dynasties (New York, 1996), pp. 178–80
for these dynasties.
55
See, briefly, Rowson, A Muslim Philosopher, pp. 267 and 293.
knowledgeable than I about the Graeco-Arabic translation
movement will assess it more thoroughly.
III. THE MANUSCRIPTS

There are two known exemplars of al-Isfizārı̄’s Kitāb fı̄ Masā’il
al-umūr al-ilāhiyya: Istanbul, Ragıp Paşa 1463, fols. 30r–47r;
and Damascus Z * āhiriyya 4871, fols. 134v–145r. The Damascus
manuscript has been described in great detail by Jamil Ragep
and E. S. Kennedy, who added a great deal more information on
the codex to the remarks of earlier scholars.56 For our purposes
we need only note that the Damascus codex as a whole is dated
between 550–558 / 1155–1163, though the copy of the Masā’il
itself was not dated by the anonymous Baġdādı̄ scribe.
The Ragıp Paşa codex has received somewhat less compre-
hensive treatment, though many of its works have been edited
or studied by other scholars.57 The codex can be divided into
56
Jamil Ragep and E. S. Kennedy, ‘‘A description of Z * āhiriyya (Damascus) MS
4871: A philosophical and scientific collection,’’ JHAS, 5 (1981): 85–108, with
references to the earlier, now largely irrelevant descriptions.
57
Contents and publication list is as follows (cf. Ritter, ‘‘Philologika III,’’
p. 46, n. 1). 1. fols. 1v–28r: Al-‘A z mirı̄, Kitāb al-I‘lām bi-manāqib al-Islām
(ed. A. A. Ghorab [Cairo, 1967], on the basis of this MS); 2. fols. 28v–29r:
[Anonymous], AmtIāl wa-ah bār h urāfiyya tah * tahā ma‘ānin falsafiyya ( = Ragep /
Kennedy, no. 23?); 3. ˘ 30r–47r: ˘ Al-Isfizārı̄, Masā’il al-Umūr al-ilāhiyya
(ed. Gimaret, ‘‘Un traité théologique’’; new edition forthcoming by Reisman); 4.
fols. 47r–48v: Ibn Suwār, Fı̄ anna Dalı̄l Yah * yā al-Nah* wı̄ ‘alā h * adatI al-‘ālam awlā
bi-al-qabūl min dalı̄l al-mutakallimı̄n as*lan (ed. Badawı̄, al-Aflāt*ūniyya
al-muh * datIa, 1955, pp. 243–7, presumably on the basis of this manuscript, though
this is not explicitly stated; tr. B. Lewin, ‘‘La notion de muh * datI dans le kalām et
dans la philosophie,’’ Orientalia Suecana, 3 (1954), 84–93 from this manuscript); 5.
fols. 49r–54r: ‘Abd Yašū‘ ibn al-Bahrı̄z, Kitāb H * udūd al-mant*iq (cf. Ragep /
Kennedy, pp. 103–4); 6. fols. 54v–57r: [Anonymous], Risāla fı̄ H * adat al-‘ālam
li-ba‘d
* al-muh * datIı̄n; 7. fols. 57r–59r: Miskawayh, Min Risāla fı̄ al-Lad Id
I āt
wa-al-ālām (ed. M. Arkoun, ‘‘Deux Épîtres,’’ BEO, 17 [1961–2], pp. 7–74, on the
basis of this manuscript); 8. 59v–60r: Miskawayh, Min Risālatihi fı̄ al-T * abı̄‘a
(ed. M. Arkoun, ‘‘Textes inédits,’’ AIsl., 5 (1963): 181–205, p. 196, on the basis of
this manuscript); 9. fol. 60r–v: Miskawayh, Min Risālatihi fı̄ G { awhar al-nafs
(id., pp. 197–8); 10. fols. 60v–61r: Miskawayh, Fı̄ al-Bah *It ‘anhā (id., pp. 198–9); 11.
fols. 61r–63v: Miskawayh, Min Kitāb al-‘Aql wa-al-ma‘qūl (ed. M. Arkoun, ‘‘De
l’intellect,’’ Arabica, 11 (1964): 80–7, on the basis of this manuscript); 12. fols.
63v–65r: Ibn Suwār, Maqāla fı̄ S * ifat al-raǧul al-faylasūf (tr. B. Lewin, ‘‘L’idéal
antique du philosophe dans la tradition arabe,’’ Lychnos (1955): 267–84, on the
basis of this manuscript; see also Kraemer, Humanism, 1986, pp. 126#.); 13. fols.
65v–68v: Abū al-Qāsim al-Kirmānı̄, Maqāla fı̄ S { arh* qawl rasūl Allāh . . .
sa-taftariqu ummatı̄ ‘alā ItalātI wa-sab‘ı̄n firqa (ed. S. Dedering, ‘‘Ein Kommentar,’’
Le Monde Oriental, 25 (1931): 35–43 on the basis of this MS; see also my
Avicennan Tradition, p. 171); 14. fols. 68v–86v: Miskawayh. Maqāla fı̄ al-Nafs
two parts, based on two colophons with dates separated by a
few months. In the colophon on leaf 89r, the scribe dates his
copying to Muh * arram 525 / December 1130.58 There the scribe
also identifies himself as S * adaqa ibn al-H* usayn ibn al-H * asan
ibn al-H
* addād, and we can be relatively certain that he is Abū
al-Faraǧ S
* adaqa ibn al-H * usayn ibn al-H* asan ibn Bahtiyār ibn
* addād al-Baġdādı̄, a minor historian who made ˘his living
al-H
as a copyist and whose own chief work is a d I ayl to the history
of Ibn al-Zāġūnı̄, itself a continuation of al-T * abarı̄. He died
in Baġdād in 573 / 1177.59 In the colophon on leaf 89r, Ibn
al-H
* addād states that he completed the copying of all the works
in the month he provides. This suggests that those works that
come after 89r were added later. A final colophon is found on
leaf 106r,60 dated Rabı̄‘ al-A z hir 525 / March 1131. Thus, it would
appear that Ibn al-H * addād˘ added additional works in the
intervening three months. Ownership notes on the title page
wa-al-‘aql wa-hiya ǧawāb sā’il sa’alahu ‘anhumā wa-h * all šukūk . . . fı̄ al-ǧawhar
al-bası̄t* al-qā’im bi-nafsihi (ed. Arkoun, ‘‘Deux Épîtres,’’ pp. 7–74, on the basis of
this manuscript); 15. fols. 87r–88v: [Proclus], Fı̄ ItIbāt al-s*uwar al-rūh * āniyya allatı̄
lā hayūlā lahā [Inst. theol., prop. 15–17, 21] (ed. / tr. G. Endress, Proclus Arabus
[Beirut, 1973], pp. 13f., 17f., 19#. [Arabic], 260#. [German] and Arkoun, ‘‘Textes
inédits,’’ pp. 200–2, attributing it to Miskawayh), French tr. B. Lewin, ‘‘Notes
sur un texte de Proclus en traduction arabe,’’ Orientalia Suecana, 4 (1955),
101–8; 16. fol. 89r: [Proclus], Mā al-Fas*l bayna al-dahr wa-al-zamān [Inst. theol.,
prop. 54] (ed. / tr. Endress, Proclus Arabus, p. 22 [Arabic], p. 271 [German], and
Arkoun, ‘‘Textes inédits,’’ pp. 202–3, attributing it to Miskawayh); 17. fol. 89v:
[Anonymous], Asmā’ al-h amr (prose and verse selections in a later hand); 18. fols.
90r–98r: Abū Bakr al-Rāzı̄,˘ Fı̄ Mā ba‘d al-t*abı̄‘a (ed. P. Kraus, Opera philosophica
[Cairo, 1936], pp. 113–34, on the basis of this manuscript); 19. fols. 98r–99v: Abū
Bakr al-Rāzı̄, Fı̄ Imārāt al-iqbāl wa-al-dawla (ed. P. Kraus, Opera philosophica
[Cairo, 1936], pp. 135–8, on the basis of this manuscript); 20. fols. 100r–106r: Ibn
Abı̄ al-Sarh * , Kitāb al-Rumūz (ed. S. M. Husayn, Maǧallat al-Maǧma‘ al-‘Ilmı̄
al-‘Arabı̄ [RAAD], 11 (1931): 641–55 on the basis of this manuscript; tr.
J. Bellamy, ‘‘The Kitāb ar-Rumūz of Ibn Abı̄ Sarh * ,’’ JAOS, 81 (1961): 224–46, on
the basis of the microfilm of this manuscript in Franz Rosenthal’s collection); 21.
fols. 106v–107r: [Anonymous], [Definitions of logical terms; philosophical questions
and answers].
58
Conclusion of [Proclus], Mā al-Fas*l bayna al-dahr wa-al-zamān. [Inst. theol.,
prop. 54].
59
See Ziriklı̄, al-A‘lām, vol. 3, p. 202. For details on his d I ayl and a discussion
of his name, see F. Rosenthal, A History of Muslim Historiography (Leiden, 1952),
p. 83, n. 1; along with C. E. Farah, The Dhayl in Medieval Arabic Historiography
(New Haven, 1967), p. 8; and James A. Bellamy, ‘‘The Kitāb ar-Rumūz of Ibn Abı̄
Sarh * ,’’ p. 227, who also records Ibn al-G { awzı̄’s suspicions about Ibn al-H * addād’s
heterodox beliefs.
60
Conclusion of Ibn Abı̄ Sarh * ’s Kitāb al-Rumūz.
include those by Ah * mad ibn Muh * ammad ibn Sālim al-S { āfi‘ı̄,61
undated; Muh * ammad ibn Abı̄ Bakr ibn Qiwām al-S { āfi‘ı̄, dated
814 / 1411–12; Ah * mad ibn ‘Abd Allāh, undated; and ‘Abd
al-Rah* mān al-G { ay‘ānı̄, dated 1080 / 1669–70.
Ibn al-H * addād was generally conscientious in copying the
works in his codex; in most instances, the texts are fully
pointed and often vocalized and he collated his copies, as is
evident from the corrective notes in the margins. However, in
the midst of the exemplar of al-Isfizārı̄’s Masā’il, at fol. 42r, the
style of the handwriting changes dramatically, from a careful,
pointed and vocalized text to a hurried unpointed scrawl,
though it is still clearly Ibn al-H * addād’s hand. The new style
includes the text from the middle of Question Twenty-One to
the conclusion of al-Isfizārı̄’s work (indeed it continues into the
next treatise in the codex). There may be some connection
between the hurried nature of the new style and the severe and
decidedly unskilled abridgment of the Republic passage that
Ibn al-H* addād’s copy displays. If Ibn al-H * addād is responsible
for this abridgment of our passage, we might conjecture that he
grew tired with the length of al-Isfizārı̄’s work, but it is equally
possible that the archetype of Ibn al-H * addād’s copy already
contained the abridged version.
This aspect raises the question of the relation between the
two exemplars of al-Isfizārı̄’s text. Both the Z * āhiriyya codex
and the Ragıp Paşa codex were copied in Baghdad in a
relatively close period of time. Ibn al-H * addād’s codex was
completed in 525 / 1130–1 and the anonymous scribe’s codex
now in the Z * āhiriyya was made over the course of the years
550–8 / 1155–1163. Also, the codices share at least two other
works besides al-Isfizārı̄’s Masā’il, including Ibn Bahrı̄z’s
H
* udūd al-mant*iq;62 and Proclus’ Fı̄ ItIbāt al-s*uwar (Inst. theol.,
Prop. 15, 17).63 Given the abridgment of the Republic passage in
Ibn al-H* addād’s codex and the copy dates of the two codices, it
61
The signature bears a striking resemblance to that of Ah * mad ibn Muh * ammad
ibn Sālim, known as Ibn S * as*ra, the qād
* ı̄ of Damascus who died in 723 / 1323; cf.
the specimen in Ziriklı̄, al-A‘lām, vol. 1, p. 222.
62
Z
* āhiriyya 4871, fols. 129r–142r, Ragep / Kennedy, no. 40; Ragıp Paşa 1463,
fols. 49r–54r (see preceding note, no. 5).
63
Z
* āhiriyya 4871, fol. 115r, Ragep / Kennedy, no. 34; Ragıp Paşa 1463, fols.
87r–88v (see preceding note, no. 15). But note that MS Z * āhiriyya 4871 also
contains Prop. 16, lacking in MS Ragıp Paşa 1463, and MS Ragıp Paşa 1463
also contains Prop. 21 and 54, lacking in MS Z * āhiriyya 4871; see Endress, Proclus,
Arabic pp. 13–22.
is impossible for either of the manuscripts to be a descendant of
the other but we might suppose that they shared a common
apograph, circulating in Baghdad and containing their over-
lapping works. Of course, if this can be established, we would
have to charge Ibn al-H * addād with the massive excision of the
Republic passage. More work is required here.
IV. THE EDITION AND TRANSLATION OF THE MASA

z’IL,
QUESTION TWENTY-TWO
Below I have edited and translated the entirety of al-Isfizārı̄’s

Question Twenty-Two in order that the Republic passage be
given the context al-Isfizārı̄ chose for it. The Z * āhiriyya manu-
script is the basis for this edition since it contains the Republic
passage in full. However, I have also included the abridged
version from the Ragıp Paşa exemplar in occasional left-side
columns for the purposes of establishing the nature of this
version (at turns paraphrastic and literal) and as correction to
Gimaret’s edition. I was able to work only from photocopies of
the Z* āhiriyya manuscript and this proved problematic in those
instances when words in the margins of the manuscript were
lost in the microfilming process. Where possible, the Ragıp
Paşa manuscript filled in these gaps, but in those cases where
it lacked the parallel text, editorial conjecture was the only
recourse. I do not register all of the conjectural emendations
made by Gimaret in his edition of the exemplar from Ragıp
Paşa in the apparatus of the edition (though I do add a few
more in the translation). Gimaret produced an admirable text
on the basis of a very problematic manuscript version and at
times his editorial conjectures show great resourcefulness;
unfortunately they are rarely correct, at least in comparison to
the more authoritative version in the Z * āhiriyya manuscript.
Finally, there is a great deal of information in the passage
edited and translated below that is worthy of commentary, but
for the purposes of this presentation, I have kept such glossing
to a bare minimum. The accompanying translation makes no
attempt to harmonize the Arabic and the Greek; rather it
renders the Arabic version of the Republic passage in as literal
a manner as possible so that those without knowledge of
Arabic may compare the Greek text.
List of Sigla
Z
* āhiriyya 4871
Ragıp Paşa 1463
Editorial correction
Editorial conjecture
— Omitted in manuscript
[ ] Enclose editorial seclusion in the text, and explanatory additions
in the translation
K L Enclose editorial conjecture
IV.A. Versions of the Masā’il, Question Twenty-Two

There are four di#erent versions (in two pairs) of Question
Twenty-Two in the two manuscripts of Isfizārı̄’s Masā’il. The
first pair is found in the Introduction of the Masā’il, where all
twenty-eight questions of the treatise are listed as a table of
contents (in both manuscripts); the second pair is found in the
Text of the Masā’il (again in both manuscripts).
1. Introduction of the Masā’il
Z
* āhiriyya 4871, fol. 141v.28–9 Ragıp Paşa 1463, fol. 30v.ult.–31r.2
(cf. Ragep / Kennedy, p. 106)
2. Text of the Masā’il
Z
* āhiriyya 4871, fol. 141v.28–9 Ragıp Paşa 1463, fol. 42r.23–4
All of the versions64 evince textual problems likely attribut-

able to scribal negligence. More interesting is the di#erent
64
Compare the text and translations of Gimaret: Arabic introduction version,
p. 216; Arabic text of the treatise version, p. 242; French translation of intro-
duction version, p. 213. Gimaret has reproduced the text of the treatise version
accurately from the manuscript, but it makes little sense.
formulation of the question as found in the Introduction in
relation to the formulation in the text of the Masā’il itself.
In the Introduction, the question (with variation in the two
manuscripts) is:
Z
* āhiriyya 4871 Ragıp Paşa 1463
Introduction Introduction
Why did we say that all things are Why did we say that we intellect
connected to [or: dependent on] all things only through God and
God and that we know them we know them through him, when
through Him, when we had said we had said elsewhere that God is
elsewhere that God is known known through the species of
through the species of created created beings?
beings?
The version in the text of the Masā’il recasts the question

significantly. Given the very corrupt nature of the last line of
this version in the Ragıp Paşa manuscript, I follow the
Z
* āhiriyya manuscript here:
Why did we say that we know all things only through God, He being the
cause for our knowing all that we know as well as the cause of the
existence of things, but it is necessity that leads us to knowledge of Him?
The first version, in the Introduction, sets forth an apparent
contradiction in which the two claims are: ‘‘we know every-
thing through God’’ and ‘‘we know God through the species of
His creation’’. This is an epistemological question; but, if we
follow the reading of the Z * āhiriyya manuscript, it begins with
an ontological assertion: ‘‘everything is connected / dependent
on God.’’ In the second version, in the text of the Masā’il, there
is a re-articulation of the ontological assertion: ‘‘everything is
connected / dependent on God’’ becomes ‘‘He is the cause of the
existence of things’’, and a substantive reformulation of the
epistemological claim: ‘‘we know Him through the species of
His creation’’ becomes ‘‘it is necessity that leads us to know
Him’’. Treated broadly, the versions in the Z * āhiriyya manu-
script are closer to one another in their elements than to the
Ragıp Paşa evidence. We find the following: (a) ontological
claim; (b) epistemological argument 1; (c) epistemological
argument 2. But there is a shu%ing of the order in the two
versions: (a)(b)(c) becomes (b)(a)(c), and (c) is radically
altered (‘‘by species’’ to ‘‘necessity’’). Excluding any other
evidence, the change may be attributable to al-Isfizārı̄ himself,
in that we might suppose that he drew up an initial list of
questions that he then refined in the course of his composition.
In comparing the versions in the Ragıp Paşa manuscript, I am
inclined to think that scribal error is responsible for the first
variation between the two manuscripts (na‘qiluhā was read for
ta‘alluquhā) since the pair ‘‘we intellect’’ and ‘‘we know’’ in
the Ragıp Paşa version does not su$ce as a distinction of claim;
and that this initial error led to the confusion the scribe had
with the second version of the question in the text itself.
However, any conclusions, be they related to the composition
process or to the textual transmission of al-Isfizārı̄’s work, will
require more investigation into both matters. It is su$cient
here to express the hope that such issues will be resolved as
editing and interpreting the text proceeds. The goal of the
present contribution is to make others aware of a passage from
al-Isfizārı̄’s work in which we find a hitherto unremarked
translation of a portion of Plato’s Republic.
IV.B. Edition of Masā’il, Question Twenty-Two
IV.C. Translation of Masā’il, Question Twenty-Two
Response.65 Because the first things that we know in
themselves through themselves, without demonstration or
syllogism, are [those] by means of which we know all of the
objects of knowledge and the sciences, such as that things
equal to one thing are themselves equal and that the whole is
greater than the part.66 So the intellect whose action is separ-
ate from every material substrate knows them only by means of
the light which is separate in itself from everything and which
penetrates everything by means of its divine power, just as
vision sees the colors presented to its action only by means of
a material light, which is sunlight (d * aw’). For the First Light is
nothing other than an existent that subsists by itself, lord of all
things, but distinct from all of them, and with a power that
penetrates all of them. As for this light that is the sunlight, it
is located (mawd * ū‘ ) in material things, like the sun, and its
rays extend from it to [all] regions.
‘‘Light’’, then, belongs to the class of homonym by name, like
calling, by way of similarity, the supports of a bed its ‘‘legs’’
and the haunches of the horse67 its ‘‘legs’’. So just as the thing
by means of which vision sees visible objects is called ‘‘light’’,
so too the thing by means of which one intellects the intelligi-
bles is called ‘‘light’’. The latter is the light and brightness
(d
* iyā’) of the Good and the former is the light and brightness of
the sun. The book of God, the Quran, articulates this as well:
‘‘God is the light of the heavens and the earth; His light can be
compared to the niche in which is a lamp’’.68 So let this issue be
explained by way of equality in the syllogism and what Galen
and a group of logicians call the syllogism that falls under the
category of relation.69 For this method is akin to anything that
compels the intellect to accept what it asserts. So it says
that light is in the world of intellect and the world of sense, and
the place of the intellect in relation to the intelligibles is like
65
See above for translations of the versions of the question.
66
Cf. al-‘Az mirı̄, Kitāb al-Amad, ed. / tr. E. Rowson, Muslim Philosopher,
pp. 94 / 95.
67
Gimaret, p. 242, al-ǧabal; MS Ragıp Paşa 1463 has an undotted rasm.
68
Quran 24:35. Avicenna will later make much use of this verse in similar
contexts; see Gutas, Avicenna, 164–5, 299–307.
69
For Galen’s ‘‘third species’’ of syllogism, the so-called ‘‘relational’’ syllogism;
see R. J. Hankinson, ‘‘Galen and the logic of relations,’’ in L. P. Schrenk (ed.),
Aristotle in Late Antiquity (Washington, D.C., 1994), pp. 57–75.
the place of sense in relation to the sensibles. Just as vision
cannot see visible objects on their own without the presence of
a third thing (I refer to the light that is sunlight), so too the
intellect does not intellect the intelligibles in the absence of a
third thing, scil. the light that is the Truth and the Divine Good
(blessed and high is He!). Similar to that is this individual
thing that is seen, whether it be man, ass70 or horse: seeing it as
an individual belongs to sight, while seeing it as a man, or an
ass or a horse, belongs to the intellect – I mean that knowing
each species belongs to the intellect. For the intellect is what
judges that Zayd and ‘Amr are one in terms of the common
universal form which gives them the intentions of ‘‘human’’.
Just as vision on its own cannot see this sensible individual
without the presence of the light that is the sunlight, but rather
by light does vision gain strength and the power to discern
what it sees, so too the intellect on its own does not ‘‘see’’ this
intellected species without the presence of the ‘‘light’’ that is
the Truth, but rather it is made capable by [the Truth] of
forming concepts of what it intellects. Just as the eye, when not
turned to the sunlight in the act of seeing, does not see and
becomes like the blind, so too the intellect, when not turned
toward the Truth, does not intellect and becomes lost71 and
confused.
The things involved in the senses are three: the agent, the
a#ected, and what is produced [by the act]. The agent [in this
case] is the objects of sense; the thing that is a#ected is the
organ of sense; and the thing produced is the act of sensing and
the act of imagination that is produced by it, which [the
intellect] assigns when it wants. So too in the case of the
intelligibles there is something that acts, something that is
a#ected, and something produced [thereby]. The agent [in this
case] is the First Intelligible and the First Truth and the First
Light and the First Good, which leads the intellect to the real
nature of things. What is produced is like the conceptual-
izations through the intellect, and the acquired intellect which,
70
I have corrected this (undotted in both manuscripts) on the basis of the next
line. Gimaret reads Itawran, bull, which is a nice conjecture; I assume an error of
nūran for the rasm.
71
Gimaret, p. 243, has the nice conjecture bāhitan for the undotted rasm in MS
Ragıp Paşa 1463; tā’ihan is pointed in MS Z
* āhiriyya 4871.
Aristotle maintains, continues outside of the body after the
body’s corruption.72
Just as this light that can be physically indicated is not only
the cause for vision seeing all visible things but also the first
thing seen by vision, after which [vision] then sees, through it,
the rest of the visible objects, so too the Light that cannot be
physically indicated in any way, i.e. that which is the Truth and
the Good (may His mention be exalted!) is not only the agent
cause of the intellect ‘‘seeing’’ all intelligibles but also the first
intelligible intellected by it, after which it intellects, by means
of it, the rest of the intelligibles.
In the same way that knowing the light that is sunlight is
something received by vision necessarily at the moment that it
sees visible objects, such that [vision] cannot see them without
seeing it first, equally, knowing the light which is the Good
[and] the Agent Truth (blessed be His name!) is something
received by the intellect necessarily at the moment that it
‘‘sees’’ the intelligibles, such that it cannot know them without
first knowing it. God, then, is the Agent Cause; the intellect
does not ‘‘see’’ what He sees, though it is by reason of Him that
the [intellect] ‘‘sees’’ all of the intelligibles. So not only do we
know God through what He intellects and produces, but also it
is through Him that we know the intelligibles.
Just as the light which is sunlight penetrates the regions of
the universe at the rise of the sun immediately and without
delay,73 neither cleaving to the bodies that it penetrates nor
mixing with them, so too the light which is the divine power
penetrates the entire universe without delay, neither mixing
with anything nor cleaving to any matter, on account of the
fact that the power of this light is infinite, and its divine power
is over [the world] in possessing and penetrating everything.
Despite the fact that we see the light that is the sunlight all
the time, we cannot easily define what it is and must be
satisfied with descriptions (rusūm) of it, like our saying that
sunlight is that by means of which one sees colors. It is equally
the case with the First Light that is the Truth, Existent,
Subsistent, Intelligible, whether intellected or not – I mean
that which is separate from all things in Itself. For the intellect
cannot easily define the nature of its substance, and so too a
72
De Anima III, 5.
73
Gimaret, p. 244, has read ziyān for zamān on the basis of a rather improbable
Persian etymology.
description that [the intellect can] understand may [have] to
su$ce, like our saying that the First Light is that by which the
intellect perceives all that it perceives, since [the First Light] is
the Agent of everything.
Just as the sun, which has light, is not simply the cause of
seeing what can be seen but is also the cause for the generation
and growth of the animals of the earth and the plants, so too
the First Light is not just the cause for the intellect perceiving
what it perceives but is also the cause of its existence and
the reason for the existence of all intelligibles. And because
the intelligibles are the reasons (asbāb) behind the sensibles –
for the universal [idea] ‘‘man’’ is the cause behind this physi-
cally indicated individual man – and the First Light is the
cause (‘illa) for all of the intelligibles, then undoubtedly it is
the cause of all intelligible and sensible things.
Now, in order that the researcher of this topic be well-trained
in this subject, I will recount the literal words of the Divine
Plato on the investigation into the knowledge of God, the Pure
Good. This is a literal quotation of what he said in his book on
Governing Cities [i.e. the Republic].74
Glaucon said: We will be satisfied by your describing for us
something about the good in the manner in which you related
the nature of righteousness75 and temperance (‘afāf) and the
other [virtues].
Socrates said: So I said to him, Oh friend!76 I too would be
satisfied completely with this. But I fear that I am unable, and
then, in satisfying myself with a desire and greed for it, I would
be telling a story to one forced to laugh at me. But, oh happy
ones, let us not talk for now about what the good is in itself. For
I see that arriving [at our goal] from the starting point of this
present motion is more than this time [allows]. However, of
that which we find to be an o#spring of this good and most
similar to it I would like to speak, if you yourselves also desire
this. If not, then let us drop [the matter].
74
The quotation here begins with Republic, 506d.
75
My conjecture of al-s*afa’, ‘‘purity of intention toward another’’ (cf. Lane, An
Arabic-English Lexicon, q.v.), for the manuscript’s al-s*ifāt (?), for the Greek

´ , is wholly conjecture, but al-‘adāla ( ´
), which we might prefer,
would entirely ignore the rasm.
76
*sāh
* i, for *sāh
* ib; see Wright, A Grammar of the Arabic Language, ii, 89A.
Adimantus77 said: Speak! For afterwards you will repay us
with the story of its father.
Then, after a little, [Socrates] said the following:78
Socrates said: These things we say are seen with the eyes and
we do not say that they are thought, while about the forms we
say that they are encompassed by thought but as for seen, then
no.
Glaucon said: You are completely correct in what you say.
Socrates said: With what do we see what we see?
Glaucon said: With sight.
Socrates said: So too we perceive audible things with hearing
and all sensible things with the other senses?
Glaucon said: Certainly.
Socrates said: So I said to him, Tell me, have you thought why
the creator of the senses made the power by which we see what
is visible and by which what is visible can be seen the
equivalent of the highest thing He created?
Glaucon said: I have not thought much about that.
Socrates said: Look for the reason79 for that in this manner:
Is it possible that hearing and the voice need a third thing
(ǧins) by means of which [the first] hears and the voice is
heard, and when the third is absent, hearing does not hear and
the voice is not heard?
Glaucon said: [They are in need of] nothing whatsoever.
Socrates said: I said to him, I reckon that many of the other
senses do not need [something else] – not to say that not one of
them needs something additional like this. But if not, is it
possible in your opinion that [one] needs something else? Tell
us what is that thing which you would say.
Glaucon said: In my opinion there is nothing I would say.
Socrates said: Tell us, do vision and the visible have need of
an additional power?
Glaucon said: How?
Socrates said: When vision is in the eye and what has it
wants to use it and there are things actually present, whenever
there is not present a third thing (ǧins) that is suitably and
77
It is perhaps significant that Adimantus is not named in this line of the
Greek. He is the brother of Glaucon; see Rep. 327c.
78
The translation resumes with 507b, thus omitting most of 506a–b.
79
tasabbab, Vth form imperative; see Dozy, Supplément aux dictionnaires
arabes, vol. 1, p. 622a.
properly disposed to such [use], you know that vision will see
nothing and that colors in this instance will not be seen.
Glaucon said: To what are you alluding with your words?
Socrates said: To the thing that you call ‘‘light’’.
Glaucon said: You speak only the truth.
Socrates said: The binding of the yoke80 between the sense of
vision and the faculty through which it sees the visible is in a
form (s*ūra) that is not slight and base, and is preferred over the
other bonds and pairings, since light is in no manner a mean
and base thing.
Socrates81 said: So which of the divine lights in the heavens
can you name as the cause (sabab) for that and make its lord
and agent, in order that its light and brightness will make our
vision see things best and will make the visible thing visible?
Glaucon said: Is it not the one that you yourself and the
others name? The fact is that you ask about none other than
the sun.
Socrates said: Tell me, Is the action of this vision in this
equivalent to the divine body?
Glaucon said: How?
Socrates said: Neither vision nor that through which it is,
i.e. what we call the eye, is the sun.
Glaucon said: Certainly not!
Socrates said: But it seems to me that it is more like the sun
than the other instruments of sense.
Glaucon said: By far.
Socrates said: And the power it also possesses by reason of
this is precisely like something that comes to it.
Glaucon said: Most certainly.
Socrates said: So tell me now, is it not also [the case that] the
sun is not vision, but perhaps it is the cause of vision once
vision has seen it?
Glaucon said: That is so.
Socrates said: So I said to him, What is meant now is that I
intended this precisely by my statement ‘‘the o#spring of the
Good’’, it being what the Good engendered in proportion to it
and like it. The Good in the intelligible world, in relation to the
intellect and the intelligibles, is [akin to] this o#spring in the
visible world, in relation to vision and the visible things.
80
qaran, pl. aqrān, rope binding two camels; see Hava, 602b.
81

y’, is omitted.
Glaucon’s response, ’A` ´ , ’´,

y y ’´
Glaucon said: How so? Reiterate it for me.
Socrates said: I said to him, You know that when one does
not turn the eyes to the bodies on whose colors the light and
brightness (d * aw’) of day falls, but rather to the bodies on which
the illumination (d * iyā’) of night falls, they turn dim and
become almost blind, as though no pure light is in them.82
Glaucon said: Certainly.
Socrates said: It is similar in the case of the living being and
that by means of which the soul intellects: when it is tied and
fixed in the place where truth and reality shine, it intellects
and knows it and appears to possess reason; but when it is tied
and fixed before a thing in darkness, that is, a thing subject to
generation and corruption, it [can only] form opinions and
wavers like a straying eye and goes on without reason, turning
its opinions up and down and becomes also like something
without reason.
Glaucon said: It seems to be [the case].
Socrates said: This thing that gives the truth to the things
apprehended by knowledge and that gives its power to the one
who possesses the knowledge is said to be the form (s*ūra) of the
good and, because it is the cause of knowledge and truth, there
will be conceived of it the same thing that is conceived by
knowledge. Now, since both of these together are part of the
fair and beautiful (I mean knowledge and truth), if you think
that there is another thing more fair and beautiful than these,
you will have thought correctly. As for knowledge and truth,
just as it is the case there that it is correct to believe that light
and vision are like the sun but it is incorrect to maintain that
they are the sun, so too it is correct here to believe that
[knowledge and truth] are like the good, but incorrect to
maintain that either one of them is the good. Rather, one
should still accord the benefit83 of the good a greater honor.
82
Note in this passage, ‘‘[pair of] eyes’’ is rendered by the dual attached
pronoun -humā, the feminine dual perfect ending -atā, and the feminine dual
nominal inflection ‘amyawayn (for the last we should expect ‘amya’ayn; see
Wright, i, 188D, rem. c).
83
We might expect malaka to translate ¢´, but what little of the rasm that
remains on the edge of the manuscript leaf in my photocopy suggests fā’ida.
I have not found a parallel use of this term. I find an instance of sāǧiyya to
translate ¢´ in Abū ‘UtImān ibn Ya‘qūb al-Dimašqı̄’s translation of Alexander of
Aphrodisias’ Fı̄ al-Istit*ā‘a (D25); see H. J. Ruland, Die arabischen Fassungen von
zwei Schriften des Alexander von Aphrodisias: U } ber die Vorsehung und U } ber das
liberum arbitrium (Diss. Saarbrucken, 1976), p. 205, l. 2 [I thank Dimitri Gutas for
providing me a copy of this work.]
Glaucon said: You have described a good and a beauty
without measure if it is what bestows knowledge and truth and
is above these and higher than them both in good and beauty.
For you do not say that it is pleasure.
Socrates said: That it is fair and beautiful is what I said. But
let us examine what is conceived about it and what is like it in
this manner specifically.
Glaucon said: How?
Socrates said: I imagine that you would say that the sun gives
visible things not only the power by which they are seen but
also their generation and growth and nourishment, though it is
not [itself] generated.
Glaucon said: How would it be generated?!
Socrates said: So too with the objects of knowledge: you
would not say that the good which they have is solely that they
be known, but also that their existence and being (ǧawhar) is
additionally from that [good], though the good is not [itself]
being, but rather transcends being by far in dignity and power.
[Al-Isfizārı̄ said:] So these are the words of the Divine Plato in

which he explains that the being of all sensible and intelligible
things is from God (blessed and exalted is He!) and by Him man
knows everything, and that since [He] fashions the being [of all
things], He himself is not being, just as the sun is what
generates all that comes to be, including animals and plants,
but is not itself generated and subject to change, though it is
composite and created.

ASP 2004 Vol.14 Issue.2

Uploaded by

Copyright:

Available Formats

You might also like

ASP 2004 Vol.14 Issue.2

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

ASP 2004 Vol.14 Issue.2

Uploaded by

Copyright:

Available Formats

Arabic Sciences and Philosophy, vol. 14 (2004) pp.

Thābit ibn Qurra and the Composition of Ratios

The Arabic Original of Liber de compositione alchemiae

The Liber de compositione alchemiae (The Book of the Composition of

Plato’s Republic in Arabic: a Newly Discovered Passage

Thābit ibn Qurra et la composition des rapports

L’original arabe du Liber de compositione alchemiae

Le Liber de compositione alchemiae (Le livre de la composition d’alchimie) est

La République de Platon en arabe: un texte nouvellement découvert

On croit généralement que le courant de traduction du grec vers l’arabe a

Dans l’introduction à un texte majeur consacré à la compo-

L’abord calculatoire de la composition des rapports dans des

DÉFINITIONS, EUCLIDIENNES ET NON-EUCLIDIENNES

m’avoir permis de franchir la barrière de la langue et de prendre connaissance de

La composition des rapports

La multiplication des grandeurs

UNE ÉTUDE COMBINATOIRE

cette substitution étant simplement exprimée par Thābit de la

septième sixième cinquième quatrième troisième deuxième première

En transposant le commentaire qu’en donne le mathématicien,

* Thābit ne démontre pas ce résultat dans ce traité. Mais, formulé autrement, il

Il s’agit ainsi d’une notion ensembliste avant la lettre, tant

huitième septième sixième cinquième quatrième troisième deuxième première

et suppose dans un premier temps que A, B, C, D et E sont

Il conclut alors que, puisque A, B et C sont connues, G est aussi

Dès lors, puisque pour Thābit le résultat d’opérations sur des

et donc F est connue. Il passe ensuite, avec les mêmes moyens,

La sixième Les cinq grandeurs connues

Propositions du troisième chapitre

Proposition 1: Si 5 (A, B, C, D, E, F) et si A, B, C et D sont

Proposition 18 (lemme): En s’appuyant sur le schéma suivant, si

La proposition 22 semble avoir été conçue pour servir de lemme

Pour démontrer cette proposition, tâche qu’il laissera, nous

THE ARABIC ORIGINAL OF LIBER DE

The Epistle of Maryānus, the Hermit and Philosopher, to

The Liber de compositione alchemiae or the The Book of the

In 1971 Sezgin indicated the existence of complete Arabic

The Epistle of Maryānus, the Hermit and Philosopher, to Prince

In the name of God the merciful and compassionate

On that day a man came to him and asked for permission to be

Then he asked me to accompany him with a group of mawālı̄.

We traveled in realms rising up with some terrains and

We rejoiced at finding him and we treated him kindly and

Then he turned towards me and asked me about our journey in

He replied: Four years after the death of Hiraql.

Then he said: O Maryānus, would it not have been kinder to

Maryānus said to him: I am not unique. And among people

Khālid said: God help us in dealing with him because he is a

Then Khālid commanded that I take him to a part of the palace

At this Khālid smiled and said: A person with whom compas-

Maryānus said: I shall explain to you, and there is neither

Khālid then said: Surely there is no course of action except by

Khālid said: O Maryānus, you have spoken and excelled, and

Ibn Sarābiyūn is one of the last exponents of classical Syriac

* I would like to record my gratitude to Prof. Ch. Burnett, Dr E. Savage-Smith,

‘‘According to Ibn Serapion: ´

value of the information about Dāwūd ibn Sarābiyūn, he is not

In fact, it breaks o# at the description of sorrel (rı̄bās; rumex L. and Var.):

‘‘According to Ibn Serapion: ´