ASP 2008 Vol.18 Issue.1

Arabic Sciences and Philosophy, vol. 18 (2008) pp.
1–18
doi:10.1017/S0957423908000453 2008 Cambridge University Press
MESURER LE CONTINU, DANS LA TRADITION

ARABE DES LIVRES V ET X DES ÉLÉMENTS
MAROUANE BEN MILED

École nationale d’Ingénieurs de Tunis, Lamsin, P.O. BOX 37 – 1002
Tunis-Belvedere, Tunisie
Email: marouane.benmiled@fss.rnu.tn
Abstract: In order to find positive solutions for third-degree equations, which he did
not know how to solve for roots, ‘Umar al-Khayyām proceeds by the intersections of
conic sections. The representation of an algebraic equation by a geometrical curve
is made possible by the choices of units of measure for lengths, surfaces, and
volumes. These units allow a numerical quantity to be associated with a geometrical
magnitude. Is there a trace of this unit in the mathematicians to whom al-Khayyām
refers directly in his Algebra? How does this unit enable the measurement of
quantities and rational and irrational relations? We find answers to these questions
in the commentaries to Books V and X of the Elements.
Résumé: Afin de trouver les solutions positives des équations du troisième degré
qu’il ne sait pas résoudre par radicaux, ‘Umar al-Khayyām procède par intersections
de sections coniques. La représentation d’une équation algébrique par une courbe
géométrique est rendue possible par les choix d’unités de mesure pour les longueurs,
les surfaces et les volumes. Ces unités permettent d’associer une quantité numérique
à une grandeur géométrique. Existe-t-il une trace de cette unité chez les mathé-
maticiens auxquels se réfère directement al-Khayyām dans son Algèbre? Comment
cette unité permet-elle de mesurer les quantités et les rapports rationnels et
irrationnels? Nous trouvons des réponses à ces questions dans les commentaires aux
Livres V et X des Éléments.
Dans son Algèbre, ‘Umar al-Khayyām (1048–1131) résout les équa-

tions algébriques des trois premiers degrés. Pour les deux premiers
degrés, il procède par radicaux, comme les algébristes savaient le
faire depuis al-Khwārizmı̄. Pour le troisième degré, il donne un
classement de toutes les équations et il construit par intersections de
coniques les racines positives de celles qu’il ne sait pas résoudre par
radicaux.1
L’association d’une courbe conique à une équation, c’est-à-dire la
représentation d’une équation algébrique par une courbe géo-
métrique, est rendue possible par les choix d’unités de mesure pour
les longueurs, les surfaces et les volumes. Ces unités permettent
d’associer une quantité numérique à une grandeur géométrique. Ce
1
R. Rashed et B. Vahabzadeh, Al-Khayyām mathématicien (Paris, 1999).
Downloaded from https://www.cambridge.org/core. Tsinghua University, on 13 Apr 2018 at 07:53:48, subject to the Cambridge Core terms of use, available at
https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0957423908000453
2 MAROUANE BEN MILED
faisant, la frontière qui existe entre les nombres, objets arithmé-

tiques ou algébriques, et les grandeurs continues, objets géo-
métriques, disparaı̂t dans le cadre d’une algèbre géométrique.2
Bien que l’œuvre d’al-Khayyām soit aujourd’hui bien connue,3 il
arrive encore qu’on attribue aux mathématiciens du XVIIe siècle
l’utilisation d’unités de mesure permettant l’association d’une
courbe géométrique à une équation algébrique:4
Par le mélange des méthodes algébriques et géométriques, Pierre de Fermat
et René Descartes créèrent, au XVIIe siècle, la géométrie des coordonnées.
Ils n’utilisaient pas cependant cette dénomination ni l’idée des axes. Le but
commun qu’ils poursuivaient était de lier les équations algébriques avec les
courbes et les surfaces en se limitant aux valeurs positives. Leurs ‘‘systèmes
de coordonnées’’ leur permettaient une représentation concrète de la
position relative des points de la courbe les uns par rapport aux autres, qui
ne variait pas en fonction d’une transformation de coordonnées.5
Afin d’en reconstituer la tradition, on peut chercher la trace d’une
telle unité de mesure chez les prédécesseurs d’al-Khayyām. Roshdi
Rashed l’a trouvée chez les géomètres infinitésimalistes des IXe–XIe
siècles. Je décrirai ici comment un courant issu des lectures arabes
du Livre X des Éléments d’Euclide a également contribué à l’élabo-
ration de ce concept. Puis, partant de la corrélation qui existe entre
les notions de rapport et de mesure, je me tournerai à nouveau vers
des commentaires au Livre X des Éléments, pour montrer de quelle
manière l’intérêt pour les questions de rapport et de proportionna-
lité, visible notamment dans les commentaires arabes au Livre V des
Éléments, a accompagné des travaux qui se situent à l’intersection de
l’algèbre et de la géométrie.6 Pour finir, je traiterai de la situation des
2
R. Rashed, ‘‘L’algèbre’’, dans R. Rashed (éd.), Histoire des sciences arabes, 3 vol. (Paris,
1997), vol. II, pp. 31–54, voir pp. 41–5.
3
Nous devons à F. Woepcke la première édition et traduction de l’Algèbre (L’Algèbre
d’Omar al-Khayyāmı̄ [Paris, 1851]). Plus récemment, l’Algèbre, ainsi que le Traité sur la
division du quart de cercle ont été édités dans R. Rashed et A. Djebbar, L’œuvre algébrique
d’al-Khayyām (Alep, 1981). Ce travail a été repris à la faveur d’un nouveau manuscrit et
complété par l’édition du Commentaire d’al-Khayyām sur les E u léments d’Euclide, dans
Rashed et Vahabzadeh, Al-Khayyām mathématicien, qui est désormais l’édition de référence
pour les travaux mathématiques de ‘Umar al-Khayyām.
4
Il faut noter que, contrairement à Descartes, al-Khayyām est resté fidèle à la règle
d’homogénéité. Voir à ce sujet R. Rashed, ‘‘La Géométrie de Descartes et la distinction entre
courbes géométriques et courbes mécaniques’’, dans J. Biard et R. Rashed (éd.), Descartes et
le Moyen Âge, E u tudes de philosophie médiévale LXXV (Paris, 1997), pp. 1–26, aux pp. 15–16.
5
E. Knobloch, ‘‘Algèbre et géométrie’’ (Traduction de V. Halleux), dans M. Blay et R.
Halleux (éd.), La science classique, XVIe–XVIIe siècle, dictionnaire critique (Paris, 1998),
pp. 673–91, à la p. 683.
6
A
v partir de l’éclairage de commentaires arabes du Livre X des Éléments, je tenterai
d’apporter des réponses à des questions soulevées lors de discussions avec Bijan
Vahabzadeh, sur la place de l’algèbre dans les commentaires arabes du Livre V. J’utiliserai
largement son édition et sa traduction des commentaires d’al-Māhānı̄ et d’al-Khayyām au
Livre V des Éléments d’Euclide (B. Vahabzadeh, ‘‘Trois commentaires arabes sur les
concepts de rapport & de proportionnalité’’, Thèse, Université Paris 7-Denis Diderot, 1997–
1998).
MESURER LE CONTINU 3
grandeurs continues vis-à-vis des nombres chez al-Khayyām, ses
prédécesseurs ainsi que ses contemporains.
I. L’UNITE
u DE MESURE CHEZ AL-KHAYYA
zM
Afin d’étudier les équations du troisième degré qu’il ne sait pas

résoudre par radicaux, al-Khayyām se ramène à des égalités entre
des expressions algébriques représentables par des courbes coniques;
l’intersection de deux coniques fournit alors la solution de l’équa-
tion. Pour ce faire, al-Khayyām est amené à élaborer des unités de
mesure pour les lignes, les surfaces et les volumes. Voici ce qu’en dit
Roshdi Rashed:
Il s’agit donc d’élaborer une théorie des équations de degré c3 et d’étendre
la théorie des équations des deux premiers degrés pour y inclure les
équations cubiques. Mais ici al-Khayyām rencontre un premier obstacle, qui
s’avère bénéfique. Pour les équations des deux premiers degrés, al-
Khwārizmı̄ [. . .] avait donné des solutions par radicaux qu’il avait justifiées
géométriquement. [. . .] Mais, pour les équations cubiques on ne connaissait
pas encore la solution par radicaux. [. . .]
u laborer une théorie des équations de degré c3, c’est d’abord enrichir les
E
termes primitifs de l’algèbre hérités des prédécesseurs algébristes, pour être
en mesure d’énumérer toutes les équations et de fonder un calcul géo-
métrique susceptible de mener à leur solution. Al-Khwārizmı̄ et ses succes-
seurs ont fourni des termes comme l’inconnue, son carré, les puissances
supérieures, les expressions polynomiales et les opérations arithmétiques
sur celles-ci, l’équation, . . . Sur ce plan, al-Khayyām n’avait rien à ajouter.
Mais toute la question demeure, d’attribuer à certains de ces termes un
sens géométrique, soit dans le plan, soit dans l’espace. Le concept fonda-
mental qui rend possible cet acte, et sur lequel insiste al-Khayyām, est
celui d’unité de mesure qui, s’il est convenablement défini en rapport
avec celui de dimension, permet l’application de la géométrie à l’algèbre,
et ainsi l’élaboration de la première théorie géométrique des équations
algébriques. [. . .]
Al-Khayyām se pose d’emblée la question des dimensions que l’on est en
droit d’utiliser en géométrie. Il écrit: ‘‘Et si l’algébriste emploie le carré-
carré dans des problèmes de géométrie, c’est métaphoriquement, et non pas
proprement, étant donné qu’il est impossible que le carré-carré fasse partie
des grandeurs. Ce qui se trouve dans les grandeurs, c’est d’abord une seule
dimension, c’est-à-dire la racine, ou, rapporté à son carré, le côté. Puis, les
deux dimensions, c’est-à-dire la surface – le carré (māl, x2 ) dans les
grandeurs est donc la surface carrée. Enfin, les trois dimensions, c’est-à-dire
le corps – le cube dans les grandeurs est le solide limité par six carrés’’. C’est
donc cette notion de grandeur qu’il faudrait asseoir sur celle d’unité de
mesure. L’idée d’al-Khayyām est claire. Il écrit: ‘‘Et toutes les fois que dans
ce traité nous dirons: un nombre est égal à une surface, nous entendons par
le nombre un quadrilatère à angles droits, dont l’un des deux côtés est un et
l’autre une droite égale en mesure au nombre donné’’. Il poursuit: ‘‘Et
toutes les fois que nous disons: un nombre est égal à un solide, nous
entendons par ‘nombre’ un parallélépipède rectangle, dont la base est le

carré de l’unité et dont la hauteur est égale au nombre donné’’.
Cette conception présuppose [. . .] une définition de l’unité linéaire, d’une
unité conventionnelle de mesure, de l’unité plane qui est le carré de cette
dernière, et de l’unité solide, solide dont la base est une unité plane et la
hauteur une unité linéaire. Dans chaque dimension, le nombre est la
multiplicité de l’unité pour chaque dimension.7
On retrouvera la même démarche au XIIe siècle dans l’œuvre de
Sharaf al-Dı̄n al-T
* ūsı̄, successeur le plus illustre d’al-Khayyām. Voici
ce qu’en dit Christian Houzel:
Sharaf al-Dı̄n al-T * ūsı̄ utilise, comme Khayyām, une unité de longueur E
pour pouvoir passer d’une espèce de grandeur à une autre et n’écrire que des
équations homogènes. Mais il est plus explicite sur ce point en introduisant
un vocabulaire adapté: il parle de racine linéaire (al- jidhr al-khat**tı̄, côté
d’un carré), de racine plane (al-jidhr al-sat*h * ı̄, surface LE où L est la
racine linéaire et E l’unité linéaire), de carré plan (al-māl al-sat*h
* ı̄), de carré
solide (al-māl al-mujassam, solide SE où S est le carré plan) et de racine
solide (al-jidhr al-jismı̄, solide PE où P est la racine plane).8
II. L’UNITE
u DE MESURE CHEZ LES PRE
u DE
u CESSEURS D’AL-KHAYYA
zM
On peut rechercher la trace d’une telle unité de mesure chez les

prédécesseurs d’al-Khayyām. Roshdi Rashed l’a trouvée chez les
géomètres infinitésimalistes des IXe–XIe siècles:
Ce concept d’unité de mesure, lié à celui de dimension, est bien présent chez
les prédécesseurs d’al-Khayyām lorsqu’ils traitent de calcul géométrique
[. . .]. Voici ce qu’écrivent les Banū Mūsā: ‘‘L’unité plane par laquelle on
mesure la surface est une surface de longueur un, de largeur un, et à angles
droits. L’unité solide par laquelle on mesure un solide est un solide de
longueur un, de largeur un, et de profondeur un, dont les surfaces s’élèvent
les unes sur les autres suivant des angles droits’’. Ibn al-Haytham s’exprime
dans des termes semblables.9
Il reste à savoir si cette unité de mesure existe chez les algébristes
auxquels se réfère al-Khayyām. Il nous donne, lui-même, à plusieurs
reprises,10 des informations sur la tradition dans laquelle s’inscrivent
ses travaux sur les équations cubiques:11
7
Rashed et Vahabzadeh, Al-Khayyām mathématicien, pp. 7–9.
8
Ch. Houzel, ‘‘Sharaf al-Dı̄n al-T * ūsı̄ et le polygone de Newton’’, Arabic Sciences and
Philosophy, 5.2 (1995): 239–62, à la p. 244. Christian Houzel se réfère dans cet article à
Sharaf al-Dı̄n al-T * ūsı̄. Œuvres mathématiques. Algèbre et géométrie au XIIe siècle, Texte
établi et traduit par Roshdi Rashed, Collection ‘‘Sciences et philosophie arabes – textes et
études’’, 2 vol. (Paris, 1986), vol. I, p. CXXXVI.
9
Rashed et Vahabzadeh, Al-Khayyām mathématicien, p. 8.
10
Rashed et Vahabzadeh, Al-Khayyām mathématicien, pp. 116 et 254–6.
11
Roshdi Rashed a attiré l’attention à plusieurs reprises sur ce témoignage d’al-Khayyām
(Al-Khayyām mathématicien, p. 7; ‘‘L’algèbre’’, p. 42, n. 26 et 27).
Quant aux Anciens, aucun de leurs propos sur ces propositions ne nous est
parvenu; peut-être, après les avoir recherchées et examinées, ne les ont-ils
pas saisies; ou peut-être leur recherche ne les a-t-elle pas obligés à les
examiner; peut-être enfin, rien de ce qu’ils en ont dit n’a-t-il été traduit dans
notre langue. Quant aux Modernes, c’est al-Māhānı̄ qui parmi eux se trouva
amené à analyser par l’algèbre le lemme qu’Archimède a utilisé, le consi-
dérant comme admis, dans la proposition quatre du deuxième livre de son
ouvrage sur La sphère et le cylindre. Il est alors parvenu à des cubes, des
carrés et des nombres en une équation qu’il ne réussit pas à résoudre après
y avoir longtemps réfléchi; il trancha donc en jugeant que c’était impossible,
jusqu’à ce que parût Abū Ja‘far al-Khāzin qui résolut l’équation par les
sections coniques.12
D’après ce passage, et d’autres du même texte, il apparaı̂t que dans
un premier temps, les prédécesseurs arabes d’al-Khayyām ont traduit
des problèmes issus de la géométrie grecque en équations cubiques
et qu’au moins un de ces problèmes ait trouvé sa solution par
intersection de sections coniques. Al-Khayyām nous donne les noms
d’al-Māhānı̄, d’al-Khāzin, d’Abū Nas*r b. ‘Irāq, d’Abū al-Jūd; al-
Māhānı̄ n’ayant pas trouvé de solutions à l’équation qu’il a établie.
Ces travaux s’inscrivent dans une tradition algébrique connue grâce
aux travaux de Roshdi Rashed. Celle-ci commence avec la fixation
des termes de l’algèbre par al-Khwārizmı̄ et ses successeurs: les
entiers naturels, les inconnues et leurs puissances carrées, cubiques
et éventuellement de degrés supérieurs, liés entre eux dans des
expressions faisant intervenir une égalité, le tout définissant une
équation algébrique. Al-Khwārizmı̄ classe les équations des deux
premiers degrés et donne des algorithmes algébriques pour les
résoudre (par radicaux). Il démontre ses algorithmes par des tech-
niques géométriques. Puis, Thābit b. Qurra traduit les problèmes
algébriques dans un langage géométrique euclidien afin d’asseoir les
démonstrations d’al-Khwārizmı̄ sur des bases solides (c’est-à-dire
axiomatiques). Dans le même temps (IXe–Xe s.), al-Māhānı̄ puis, Ibn
‘Is*ma, al-Khāzin, al-Ahwāzı̄, pour ne citer qu’eux, traduisent dans le
langage de l’algèbre des problèmes quadratiques ou cubiques issus
notamment du Livre X des Éléments d’Euclide ou de La sphère et le
cylindre d’Archimède.13 Au sein de cette articulation, qui lie algèbre
et géométrie, se trouve un emplacement naturel pour une unité de
mesure des grandeurs. Nous allons voir qu’elle y occupe e#ective-
ment une place centrale.
Si les textes d’al-Māhānı̄ (IXe s.) et d’al-Khāzin (Xe s.), sur la
traduction algébrique de problèmes solides issus de la géométrie
grecque, sont perdus, leurs textes concernant la mise en équations
12
Traité d’algèbre, dans Rashed et Vahabzadeh, Al-Khayyām mathématicien, p. 116.
13
Voir M. Ben Miled, Opérer sur le Continu, traditions arabes du Livre X des Éléments
d’Euclide, avec l’édition et la traduction du commentaire d’Abū ‘Abdi Allāh Muh * ammad b.
‘Izsā al-Māhānı̄, Préface de Roshdi Rashed (Carthage, 2005); Rashed, ‘‘L’algèbre’’.
algébriques des problèmes quadratiques du Livre X des Éléments

nous sont parvenus. A v cette occasion, les grandeurs rationnelles et
irrationnelles, définies dans le Livre X, sont interprétées numérique-
ment, généralisées aux quantités irrationnelles définies par radicaux
carrés, cubiques et d’ordres supérieurs, et soumises aux calculs
algébriques. Dans les définitions des quantités rationnelles et irra-
tionnelles (ou sourdes) apparaı̂t l’unité de mesure dont le rôle est de
permettre le passage de la géométrie à l’algèbre. On lit dans le
commentaire d’al-Khāzin au Livre X des Éléments d’Euclide:
Les quantités homogènes sont mesurées par une quantité (miqdār) qui
leur est homogène; comme les lignes par une ligne, les plans par un plan et
les corps par un corps. La quantité posée pour étalonner (li-al-‘iyār) et
mesurer (li-al-taqdı̄r) est rationnelle, et les quantités qui sont mesurées par
elle sont rationnelles car elle est un et par son unité elle les dénombre une
fois ou plusieurs fois [. . .]. Comme la longueur du corps qui est mesurée par
une longueur donnée comme l’empan, la coudée ou la brasse; sa surface qui
est mesurée par le carré qui est un par un de l’empan, de la coudée ou de
la brasse; et sa profondeur qui est mesurée par le cube qui est un par un
puis par un [. . .]. Et tout ce que cette quantité à mesurer (al-miqdār
al-mikyāl) mesure par une partie de ses parties, sa moitié ou son tiers ou son
quart, ou par des parties de ses parties, comme ses deux tiers ou ses deux
cinquièmes ou ses trois cinquièmes, est aussi rationnel. En général, toute
quantité (miqdār) dont le rapport à cet étalon (mi‘yār) est le rapport d’un
nombre à un nombre est rationnel. Et ce qui existe d’autre que ce que nous
avons cité, si on l’y réfère, est dit sourd.14
On retrouve les mêmes conceptions des unités de longueur et de
surface, et les mêmes définitions de la rationalité et de l’irrationalité,
dans le commentaire d’Ibn ‘Is*ma (Xe s.), antérieur à celui d’al-Khāzin
et connu de ce dernier.15
Nous disons que la ligne rationnelle est la donnée posée comme étalon
pour mesurer toutes les lignes, comme l’empan, la coudée et la brasse et
tout ce qui est posé comme étalon des quantités est rationnel. De même, tout
ce qui est compté par le rationnel est rationnel et tout ce qui en est des
parties, comme les deux tiers, le cinquième et les trois cinquièmes. En
général, tout ce dont le rapport à lui (le rationnel) est comme le rapport
d’un nombre à un nombre, est une ligne rationnelle qui lui est commensu-
rable en longueur. Il en est ainsi des surfaces qui peuvent être rationnelles
ou sourdes de plusieurs façons. Et l’étalon de la surface est le carré un par
un.
Dans les commentaires d’Ibn ‘Is*ma et d’al-Khāzin au Livre X des
Éléments, les unités de mesure de longueur et de surface sont établies
pour définir, sous une forme algébrique, les grandeurs rationnelles et
14
Al-Khāzin, Tafsı̄r *sadr al-maqāla al-‘āshira min Kitāb Uqlı̄dis, Tunis, MS BN 16167, fol.
65v.
15
Al-Khāzin se réfère au texte d’Ibn ‘Is*ma (Tunis, MS BN 16167, fol. 10r ).
irrationnelles du Livre X. Les segments de droites, ainsi que toutes
les grandeurs, sont représentés par leur représentation algébrique.
Les unités de longueur et de surface (et de volume, chez al-Khāzin
seulement), ainsi que les quantités qu’elles mesurent, ou qu’une ou
plusieurs de leurs parties mesurent, sont dites rationnelles, les autres
sont dites irrationnelles ou sourdes. L’unité linéaire est la traduction
algébrique de la droite rationnelle proposée, fixée par Euclide dans le
Livre X, qui permet de définir les droites et les surfaces rationnelles
et irrationnelles à partir des notions de commensurabilité et
d’incommensurabilité (Éléments, Définitions X.1–4). L’unité plane
est une généralisation de l’unité linéaire. Généralisation heureuse
car, outre ses conséquences algébriques, elle se marie parfaitement
avec la géométrie du Livre X.16 C’est e#ectivement grâce aux unités
de longueur et de surface qu’Ibn ‘Is*ma et al-Khāzin parviennent à
traduire en équations algébriques du second degré les problèmes
quadratiques du Livre X. On peut supposer que c’est grâce à ces
mêmes unités, ainsi que sur l’unité de volume introduite aussi dans
son commentaire du Livre X, qu’al-Khāzin fait reposer la traduction
en équations du troisième degré du Lemme d’Archimède (éventuelle-
ment d’autres problèmes solides), et qu’il le résout par inter-
sections de coniques. Ainsi, le concept d’unité de mesure, lié à celui
de dimension, permettant d’associer des équations à des courbes
coniques, a existé dans la tradition algébrique à laquelle se réfère
al-Khayyām. Il reste à établir si ces unités de longueur, de surface et
de volume permettent de mesurer toutes les quantités continues,
rationnelles ou irrationnelles.
III. MESURER LES GRANDEURS
Dans le commentaire d’al-Khāzin, les unités de mesure étant posées

pour chacune des trois premières dimensions, la notion euclidienne
de commensurabilité est remplacée par celle de ‘‘mesurabilité’’ par
l’unité ou l’une de ses parties. Pour al-Khāzin, mesurer veut dire
dénombrer. Autrement dit, mesurer par un nombre entier, sans reste.
Une grandeur en mesure une autre si elles sont toutes deux homo-
gènes, et si la première ajoutée à elle-même un certain nombre de fois
est identique à la seconde. Ainsi, les quantités rationnelles sont
mesurées par l’unité, ou par une partie de l’unité et les quantités
sourdes ne peuvent être mesurées à partir de l’unité ou de ses parties,
dans le sens ici entendu par mesurer.
Cette notion de mesure est étroitement liée à la possibilité
d’exprimer ou de prononcer sans radicaux la valeur numérique d’une
grandeur. Al-Khāzin explique au sujet des quantités sourdes:
16
Ben Miled, Opérer sur le Continu, pp. 115–44.
Et ce qui existe d’autre que ce que nous avons cité, si on l’y réfère,17 est dit
sourd. J’entends qu’il n’est pas possible de le prononcer, sauf par un radical,
comme on dirait racine de trois ou racine de cinq. Nous avons émis
e#ectivement la condition qu’il y soit référé car il existe parmi ces quantités
sourdes, certaines qui sont prononçables si on les réfère les unes aux autres.
Comme racine de cinq qui est le tiers de racine de quarante-cinq. L’un est
ainsi trois et l’autre un. Et comme racine de un et quart, qui est la moitié de
racine de cinq. Mais elles sont irrationnelles en référence à la quantité qui
a été posée comme étalon (mi‘yār) [. . .], j’entends celle par laquelle ont été
appelées ces racines sourdes.18
Ainsi, deux quantités irrationnelles, donc ‘‘non prononçables’’, peu-
vent être prononçables l’une relativement à l’autre. Une quantité est
prononçable en référence à une autre si leur rapport est celui de deux
entiers. La notion de mesure ainsi définie permet de mesurer les
quantités rationnelles à partir de l’unité de mesure, elle permet aussi
de mesurer une quantité irrationnelle à partir d’une autre quantité
irrationnelle qui lui est commensurable. Mais l’unité de mesure ne
permet pas de mesurer les quantités irrationnelles.
Al-Ahwāzı̄ (Xe s.; postérieur à al-Khāzin), s’exprime dans des
termes similaires pour définir la commensurabilité en longueur de
deux segments de droites:
La commensurabilité en longueur est que le rapport d’une des deux lignes à
l’autre soit comme le rapport d’un nombre à un nombre. Et si tu veux, tu dis
que la ligne la plus petite est une partie exprimable de la plus grande et la
plus grande ligne un multiple de la plus petite. Et si tu veux, tu dis que la
plus petite dénombre la plus grande des fois exprimables.19
Il est à noter qu’avec al-Ahwāzı̄, ‘‘dénombrer’’ ne s’entend pas
compter un nombre entier de fois, mais une quantité rationnelle de
fois, ce qui ne change rien quant au fond, à savoir que l’unité ne
mesure pas les quantités irrationnelles.
Lorsque l’unité de mesure est fixée, et qu’elle prend la même valeur
que le un de l’arithmétique, la possibilité d’exprimer une quantité
numérique fournit un critère intrinsèque à cette grandeur pour
décider de sa rationalité. Ainsi les entiers et les fractions sont
rationnels car exprimables (sans radicaux), les autres quantités sont
irrationnelles ou sourdes. Ce critère avait déjà servi de définition à
al-Māhānı̄: dans son Explication du dixième livre de l’ouvrage
d’Euclide, il avait définit les droites rationnelles et irrationnelles à
partir de la possibilité d’exprimer ou de prononcer leurs valeurs
numériques:
17
Si l’on se réfère aux quantités rationnelles.
18
Tunis, MS BN 16167, fol. 65v.
19
Traduit à partir de [Téhéran, MS Daniskada i-adab, Imām Jum‘a, g284 / 4] et de [Le
Caire, MS Dār al-Kutub, 4528k / 1], qui appartiennent à deux traditions manuscrites
di#érentes.
Les droites, du point de vue de l’expression, n’ont que deux sens. Soit elles
sont rationnelles, comme lorsqu’on dit dix, douze, trois et demi, six et un
tiers et leurs analogues dont on exprime la grandeur et prononce la
quantité. Soit, elles ne sont pas rationnelles et s’appellent sourdes; celles
dont on ne peut pas exprimer la grandeur ni prononcer la quantité. Comme
les racines des nombres qui ne sont pas carrés, tels dix, quinze, vingt et les
côtés des nombres qui ne sont pas des cubes et les autres bords et extrémités
des nombres,20 ou ceux obtenus par composition ou soustraction, ou
composition avec une rationnelle, ou dont on soustrait une rationnelle, ou
qui est soustrait d’une rationnelle, et les espèces de composition et de
soustraction semblable à cela.21
Cette conception numérique des grandeurs permet aux commenta-
teurs du Livre X une investigation quantitative du Continu, basée
sur des équations et des calculs algébriques.
Il demeure que ni l’unité de mesure, ni même ses parties, ne
permettent de mesurer les quantités irrationnelles; ce qui est pour-
tant nécessaire, sans doute pour justifier les calculs algébriques sur
les quantités rationnelles et irrationnelles des commentaires au
Livre X, mais plus encore pour déterminer une solution d’une
équation du troisième degré à partir de l’intersection de deux
coniques représentées dans le plan, ou pour fonder les travaux
des infinitésimalistes arabes qui s’inscrivent dans la tradition
archimédienne.22
En tentant de mesurer à partir de l’unité de mesure, dans le sens de
dénombrer, toutes les quantités, y compris irrationnelles, on arrive
naturellement à l’algorithme suivant:
Si à partir d’une grandeur U rationnelle, éventuellement prise
comme unité de mesure, on cherche à mesurer une grandeur G qui lui
est homogène et plus grande, soit elle la mesure (c’est-à-dire qu’elle
la dénombre un nombre entier de fois) et alors la grandeur G est
rationnelle, soit elle ne la mesure pas. Dans ce cas, si le reste R est
une partie de l’unité, c’est-à-dire si R mesure U, alors R mesure aussi
G, on peut notamment en conclure que G est rationnelle. Mais si R
n’est pas une partie de U, alors on considère le reste R obtenu en
cherchant à mesurer U avec R. Si on réitère l’opération précédente
en cherchant à mesurer R avec R , et ainsi de suite et qu’on obtienne
finalement un reste qui mesure le reste précédent, alors la grandeur
G est rationnelle. Sinon le procédé ne se termine jamais, et, à moins
qu’une périodicité n’apparaisse, seul le mathématicien immortel peut
20
‘‘Les autres bords et extrémités des nombres’’ ne sont, sans doute, pas les racines
n-ièmes des entiers, avec n>3, mais très probablement les racines d’ordres 2n3m, avec n, m
entiers naturels (Ben Miled, Opérer sur le Continu, p. 41).
21
M. Ben Miled, ‘‘Les commentaires d’al-Māhānı̄ et d’un anonyme du Livre X des
Éléments d’Euclide’’, Arabic Sciences and Philosophy, 9.1 (1999): 89–156, à la p. 122.
22
R. Rashed, Les mathématiques infinitésimales du IXe au XIe siècle, 4 vol. (Londres, 1993–
2002).
espérer conclure que la grandeur G est irrationnelle; mais . . . pas

avant la fin de sa vie!
Ce procédé appelé anthyphérèse, issu de la Proposition X.2 des
Éléments, est utilisé dans les mathématiques arabes pour manipuler
les rapports de grandeurs.
IV. DU LIEN ENTRE LES NOTIONS DE MESURE ET DE RAPPORT CHEZ

AL-KHAYYAz M ET SES PRE
u DE
u CESSEURS
Dès le début de son commentaire au Livre X, al-Khāzin introduit en

même temps les notions de mesure et de rapport:
Les classes de la quantité continue (al-kammiyya al-muttas*ila) – que sont
la ligne, le plan, le corps, le lieu et le temps – sont appelées des grandeurs
(a‘z
*ām). Les grandeurs, lorsqu’elles sont rapportées les unes aux autres
et que certaines sont mesurées par d’autres, sont des quantités mesurées
(maqādı̄r). Le rapport et la mesure se trouvent dans les quantités
homogènes.23
Puis, en même temps qu’il définit les quantités rationnelles et
irrationnelles à partir des unités de mesure, al-Khāzin, de même que
les autres auteurs de commentaires algébriques du Livre X de la
même période, établit la proposition suivante, issue de Éléments X.5:
‘‘En général, toute quantité dont le rapport à cet étalon (l’unité de
mesure) est le rapport d’un nombre à un nombre est rationnel’’.24 La
réciproque est aussi énoncée. Cette proposition et sa réciproque
mettent en relief le lien étroit qui existe entre les notions de mesure
et de rapport. La définition du rapport, par anthyphérèse, donnée par
les algébristes arabes, et que l’on trouve dans le Traité sur le rapport
d’al-Māhānı̄, est la suivante:
Le rapport entre chaque paire de grandeurs homogènes – et aussi entre
chaque paire de nombres – est l’état qui survient à chacune des deux
lorsqu’elle mesure son conjoint ou que son conjoint la mesure. Et ce qui
survient quant à cela consiste en trois espèces: ou bien la plus petite,
lorsqu’elle mesure la plus grande, l’épuise par la mesure et il ne reste rien
d’elle; – ou bien elle ne l’épuise pas, mais il reste de la plus grande un
reste plus petit que la plus petite tel, que lorsque la plus petite est
mesurée par lui, il l’épuise par la mesure; ou il reste d’elle [i.e. de la plus
petite] un reste plus petit que le premier reste tel, que lorsque le premier
[reste] est mesuré par lui, il l’épuise; ou il reste de lui [i.e. du premier
reste] un reste tel, que lorsque le deuxième [reste] est mesurée par lui et
que l’on fait cela continuellement, on aboutit à un reste qui épuise le
reste qui le précède; – ou bien si l’on fait cela continuellement, on
n’aboutit pas à un reste qui épuise le reste qui le précède. Les deux
premières espèces d’entre les espèces de la mesure surviennent dans
23
Tunis, MS BN 16167, fol. 65v.
24
Ibid.
les nombres et les grandeurs, et la troisième dans les grandeurs seule-
ment.25
A
v partir de cette définition anthyphérétique, le rapport de deux
grandeurs, même incommensurables, peut être perçu comme une
quantité numérique. Al-Māhānı̄ écrit:
[. . .] la science du rapport des grandeurs et de leur proportionnalité selon
les règles est une chose sur laquelle on pourra s’arrêter en partant de la
connaissance du rapport selon la voie numérique, puis des propositions qui
sont dans le début du dixième Livre.26
En e#et, au début du Livre X des Éléments, la Proposition X.2 établit
que:
Si, de deux grandeurs inégales proposées la plus petite étant retranchée de
la plus grande de façon réitérée et en alternance, le dernier reste ne mesure
jamais le [reste] précédent, les grandeurs seront incommensurables.27
Si, au bout d’un nombre (fini) d’étapes, le dernier reste mesure le
reste précédent, alors les grandeurs sont commensurables et le
dernier reste est la plus grande commune mesure des deux grandeurs.
C’est l’objet de la Proposition X.3. Elle est utilisée dans la Proposi-
tion X.5 pour démontrer que deux grandeurs commensurables ont le
rapport d’un nombre (entier) à un nombre (entier).
Plus bas, al-Māhānı̄ définit la proportion comme l’identité de deux
rapports définis par anthyphérèse:
De la proportion: les grandeurs dont le rapport est le même sont celles qui
sont telles, que lorsque la première et la troisième sont mesurées par la
deuxième et la quatrième – ou inversement – le nombre de fois de leur
mesure est égal; et que s’il reste d’elles deux grandeurs plus petites que les
deux plus petites, et que les deux plus petites soient mesurées par elles, le
nombre de fois de cette mesure est aussi égal; et ainsi indéfiniment.28
Si, comme le fait al-Māhānı̄, on définit le rapport par anthyphérèse,
on considère la mesure quantitative du rapport et on définit la
proportion comme l’identité de deux rapports, on parvient alors à
une mesure théorique des quantités sourdes à partir de l’unité de
25
B. Vahabzadeh, ‘‘Al-Māhānı̄’s Commentary on the concept of ratio’’, Arabic Sciences
and Philosophy, 12.1 (2002): 9–52; ‘‘Le commentaire d’al-Māhānı̄ sur le concept de rapport’’,
dans A. Hasnawi (éd.), Actes du Colloque de la Sihspai (Carthage), Beı̈t al-Hikma, 28 nov–3
déc. 2000, à paraı̂tre. Comme pour le rapport, les définitions d’al-Māhānı̄ de la proportion-
nalité, de la proportion et du rapport plus grand reposent toutes sur le principe
d’anthyphérèse (voir Vahabzadeh, ‘‘Trois commentaires arabes’’, I, pp. XXV–XXVII, II, pp.
3–5).
26
Vahabzadeh, ‘‘Al-Māhānı̄’s Commentary on the concept of ratio’’; ‘‘Le commentaire
d’al-Māhānı̄ sur le concept de rapport’’.
27
Euclide, Les Éléments, Traduction du texte de Heiberg et commentaires par Bernard
Vitrac, 4 vol., I (Livres I–IV ) avec une introduction de Maurice Caveing (1990), II (Livres V–
IX ) (1994), III (Livre X ) (1998), IV (Livres XI–XIII ) (Paris, 1900–2001), vol. III, p. 94.
28
Vahabzadeh, ‘‘Al-Māhānı̄’s Commentary on the concept of ratio’’; ‘‘Le commentaire
d’al-Māhānı̄ sur le concept de rapport’’.
mesure. Il su$t pour cela de considérer que le rapport de deux

grandeurs est identique au rapport d’une quantité à l’unité. Cette
quantité est alors la mesure du rapport; elle est rationnelle si les
deux grandeurs sont commensurables, et irrationnelles sinon.29 Ou,
de façon équivalente, on peut considérer que le rapport de deux
grandeurs est identique au rapport de l’unité à une quantité. C’est ce
que fait al-Khayyām dans le troisième Livre de son Commentaire sur
les postulats d’Euclide. Après avoir défini le rapport de façon simi-
laire à al-Māhānı̄, il propose de considérer ‘‘la grandeur de ce
rapport en tant qu’il est rapport entre grandeurs’’;30 la ‘‘grandeur du
rapport’’ étant une mesure quantitative du rapport. Il écrit:
Que l’on fixe à l’avance une grandeur en tant qu’unité – i.e. on se donne une
grandeur et on l’appelle unité – et l’on rapporte à elle les autres grandeurs.
Car il y a inévitablement dans chaque mesure une chose fixée en tant
qu’unité, et le reste rapporté à elle par la voie du nombre.
Et si le rapport entre grandeurs n’est pas numérique, on rapporte son
carré au carré de l’unité; ou bien le carré de son carré, ou le carré du carré
de son carré, et ainsi de suite à l’infini; ou bien on laisse ce rapport inconnu
en tant que mesure, puisqu’il n’y a absolument aucun moyen d’en saisir la
quantité rapportée à cette unité fixée.
Et je ne dis pas que le rapport quantitatif doit nécessairement être mesuré
pour être connu. Je dis au contraire qu’il n’y a aucun doute que chaque
rapport soit entre grandeurs, de sorte qu’il est possible de supposer une
grandeur de même genre en tant qu’unité, telle que le rapport de cette unité
supposée à une autre grandeur intelligible sera alors égal à ce rapport
donné. Et cette grandeur ne doit pas, pour être intelligible, être nécessaire-
ment intelligible dans les choses sensibles; en raison de notre incapacité à
comprendre une règle de l’art qui permettrait de la déterminer. Et souvent
ce rapport est inconnu du point de vue du nombre, <mais> connu du point
de vue de la géométrie. Mais il n’en résultera pour nous aucun mal après
que nous avons réalisé que le rapport entre grandeurs est associé à une
chose numérique, ou dans la potentialité du nombre.31
Ainsi, la définition anthyphérétique du rapport permet de mesurer
à partir de l’unité les quantités sourdes de même que les rapports
29
Ramener le rapport de deux grandeurs au rapport d’une grandeur à l’unité de telle
façon que les quatre grandeurs soient proportionnelles, n’est pas une idée nouvelle. Dans la
traduction arabe par al-Dimashqı̄ (Xe siècle) du commentaire de Pappus au Livre X des
Éléments, on peut lire: ‘‘Les grandeurs commensurables ont toutes pour rapport l’une à
l’autre, le rapport d’un nombre à un nombre. Et dans certaines situations, si l’on fait le
rapport avec la grandeur posée définie (la rationnelle proposée du Livre X), nous arrivons à
la di#érence qui existe entre les rationnelles et les sourdes’’. (The Commentary of Pappus on
Book X of Euclid’s Elements, éd. G. Junge et W. Thomson [Cambridge, Mass, 1930; réimpr.
New York, 1968], p. 197. Je traduis en français.) En revanche, ce qui est nouveau, c’est
d’utiliser cette proportionnalité afin d’obtenir une mesure du rapport de deux grandeurs
incommensurables.
30
Commentaire d’al-Khayyām sur les E u léments d’Euclide, dans Rashed et Vahabzadeh,
Al-Khayyām mathématicien, p. 340.
31
Ibid., p. 372–4. Sur la ‘‘potentialité du nombre’’, voir infra, la note 20, ainsi que le texte
qui s’y rapporte.
de grandeurs incommensurables. Ceux-ci sont introduits dans le
domaine des nombres. Les calculs arithmétiques e#ectués dessus s’en
trouvent justifiés, de même que les représentations géométriques des
équations et de leurs solutions.
V. NOUVELLES CONCEPTIONS DU NOMBRE ET DE L’UNITE

u
Mesurer les grandeurs à partir d’unités de mesure, et concevoir de

façon quantitative les rapports de grandeurs incommensurables,
amènent nécessairement à revoir les conceptions de nombre et
d’unité. Chacun sait qu’Euclide définit le nombre comme la multitude
composée d’unités, l’unité étant ‘‘ce selon quoi chacune des choses
existantes est dite une’’.32 De plus, on sait qu’une telle unité n’est pas
divisible. Les grandeurs étant divisibles à l’infini, les conceptions
euclidiennes de nombre et d’unité ne peuvent s’y appliquer. Pappus
soulève ainsi ce problème dans son commentaire au Livre X, traduit
par al-Dimashqı̄:
Les nombres découlent et augmentent d’une chose qui est la plus petite et
elles s’écoulent à l’infini; alors que les grandeurs, au contraire, j’entends
qu’elles commencent du tout fini et elles s’écoulent dans la division à
l’infini.33
Le problème est double, diviser l’unité et concevoir numériquement
les rapports de grandeurs incommensurables.34 Les conceptions
euclidiennes de nombre et d’unité ne peuvent donc pas s’appliquer
aux objets manipulés par al-Māhānı̄, al-Khāzin ou al-Khayyām,
c’est-à-dire aux grandeurs continues et à leurs mesures numériques.
Cette question de nature philosophique sort du cadre des mathéma-
tiques, nous dit al-Khayyām, ce qui n’interdit pas de considérer une
unité divisible et de rattacher aux ‘‘nombres’’ les quantités sourdes
ainsi que les rapports de grandeurs, même incommensurables:
Quant à étudier si le rapport entre grandeurs comprend le nombre dans son
essence, ou bien s’il est inséparable du nombre, ou bien s’il est joint au
nombre d’une manière extrinsèque à son essence à cause d’une autre chose,
ou bien s’il est joint au nombre à cause d’une chose inséparable de son
essence sans que l’on ait besoin d’une assertion extrinsèque: c’est là une
étude philosophique à laquelle le géomètre n’a absolument pas à se
32
Définitions VII.1 et 2, trad. Vitrac.
33
The Commentary of Pappus on Book X, éd. Junge et Thomson, §3.
34
Les deux questions de la divisibilité infinie, et de la preuve de l’existence de grandeurs
incommensurables par le procédé anthyphérétique de soustractions alternées, sont confu-
sément mêlées dans le commentaire de Pappus au Livre X, afin de caractériser le Continu à
partir des Propositions X.1 et 2. J’explique dans Opérer sur le Continu, pp. 102 et suiv., en
me basant sur un travail de Maurice Caveing, comment l’imprécision de Pappus peut se
justifier historiquement et mathématiquement (M. Caveing, ‘‘Quelques remarques sur le
traitement du continu dans les Éléments d’Euclide et la Physique d’Aristote’’, dans J.
Dieudonné, M. Loi et R. Thom [éd.], Penser les mathématiques, Textes préparés et annotés
par F. Guénard et G. Lelièvre [Paris, 1982], pp. 145–66, aux pp. 153 et suiv.).
consacrer. Mais il faut que l’on sache que lorsque l’on parle ici de la
composition du rapport, c’est en tant qu’il est joint aux notions de nombre
et d’unité ou en puissance ou en acte.35
Le rapport est donc une notion numérique. Bien que cette quantité
numérique ne soit pas toujours accessible e#ectivement (‘‘en acte’’),
elle l’est ‘‘en puissance’’.
Pour isoler la quantité numérique associée au rapport, al-Khayyām
égalise ce dernier au rapport de l’unité posée à une grandeur:
Nous supposons l’unité, et nous posons son rapport à la grandeur G égal au
rapport de A à B. (Et l’on ne devra pas concevoir la grandeur G en tant
qu’elle est une ligne, une surface, un solide ou un temps. Il faudra au
contraire la concevoir en tant qu’elle est abstraite dans l’intellect de ces
caractères adjoints, et en tant qu’elle est rattachée au nombre: non en tant
que nombre absolu véritable car le rapport entre A à B pourrait ne pas être
numérique, de sorte que l’on ne pourra pas trouver deux nombres selon leur
rapport. – Et ceux qui font des calculs, je veux dire ceux qui font des
mesures, parlent souvent de la moitié de l’unité, de son tiers, ainsi que des
autres parties, bien que l’unité soit indivisible. Mais ils signifient par là non
pas une unité absolue véritable dont seraient composés les nombres vérita-
bles: au contraire, ils signifient par là une unité donnée qui, selon eux, est
divisible. Et ils traitent ensuite à leur gré les grandeurs suivant cette unité
divisible, et suivant les nombres qui en sont composés; et ils parlent souvent
de la racine de cinq, de la racine de la racine de dix; ainsi des autres choses
qui abordent dans le cours de leurs dialogues, et dans leurs constructions et
leurs mesures. Mais ils signifient par là uniquement un cinq composé
d’unités divisibles, comme nous l’avons mentionné. Il faudra donc que l’on
sache que cette unité est celle qui est divisible, et que la grandeur G doit
être considérée comme un nombre, comme nous l’avons mentionné, quelque
grandeur qu’elle soit. – Et quand nous disons: nous posons le rapport de
l’unité à la grandeur G égal au rapport de A à B, nous ne voulons pas dire par
là qu’il est de notre pouvoir de façonner cette notion dans toutes les
grandeurs, i.e. que nous faisons ce que nous disons par une règle de l’art.
Nous voulons au contraire dire par là qu’il n’est pas impossible qu’elle
existe dans l’intellect. Et notre incapacité à faire cela n’indique pas que la
chose est impossible en soi.36 Comprends donc ces notions).37
35
Commentaire d’al-Khayyām sur les E u léments, dans Rashed et Vahabzadeh, Al-Khayyām
mathématicien, p. 374.
36
Ce passage est à rapprocher d’un autre, extrait d’un texte d’Ibn al-Haytham, où est
démontrée la possibilité de carrer le cercle. Dans la démonstration, Ibn al-Haytham est
amené à considérer le rapport d’une lunule à un cercle. A v partir de ce rapport et de la
quadrature de la lunule, Ibn al-Haytham déduit la possibilité de carrer le cercle. Nous
savons aujourd’hui que le rapport qu’il considère est transcendant. Pour cette raison la
démonstration d’Ibn al-Haytham n’est pas e#ective. Il s’en explique de la façon suivante:
‘‘Les vérités des notions intelligibles n’ont pas besoin d’être trouvées et déterminées en acte
par l’homme, mais si la démonstration établit la possibilité de la notion, alors cette notion
devient vraie, que l’homme la détermine en acte ou ne la détermine pas’’ (Sur la quadrature
du cercle, éd. Rashed, dans Les mathématiques infinitésimales, II, pp. 96–7).
37
Commentaire d’al-Khayyām sur les E u léments, dans Rashed et Vahabzadeh, Al-Khayyām
mathématicien, pp. 378–80.
Pour al-Khayyām, le rapport de deux grandeurs, même incommensu-
rables, se ramène au rapport de l’unité à une grandeur ‘‘abstraite’’.
Celle-ci peut être considérée comme un nombre, mais un nombre
particulier, construit à partir d’une unité divisible qui permet de
‘‘composer’’ tant les quantités rationnelles que les quantités irration-
nelles. Ainsi, longueurs, surfaces et volumes peuvent être égalés à
des nombres. Il demeure qu’une grandeur ne pouvant avoir de
quatrième dimension, un nombre ne peut être interprété géo-
métriquement comme un carré-carré. Al-Khayyām écrit: ‘‘Et si
l’algébriste emploie le carré-carré dans des problèmes de géométrie,
c’est métaphoriquement, et non pas proprement, étant donné qu’il est
impossible que le carré-carré fasse partie des grandeurs’’.38 Les
puissances de l’inconnue strictement supérieures à trois demeurent
donc du ressort d’une algèbre dénuée d’interprétation géométrique;
contemporain d’al-Khayyām, al-Karajı̄ pratiquait une telle algèbre.
VI. APERC
q U DES NOTIONS DE RAPPORT ET DE NOMBRE CHEZ
AL-KARAJIz
Il vient, de l’examen qui précède, que la conception numérique du

rapport, que l’on trouve chez al-Khayyām, s’inscrit dans cette tra-
dition qui repose sur la lecture algébrique de textes géométriques
grecs, notamment des Livres V et X des Éléments d’Euclide et de La
sphère et le cylindre d’Archimède. Parallèlement à cette première
tradition s’en est développée une seconde – issue entre autres des
Arithmétiques de Diophante et des travaux d’al-Khwārizmı̄ et d’Abū
Kāmil – qui a donné naissance à l’école d’al-Karajı̄, en permettant
l’application de l’arithmétique à l’algèbre. L’objectif d’al-Karajı̄ était
d’établir les règles du calcul algébrique pour opérer sur les poly-
nômes en ‘‘la chose’’ et son inverse (c’est-à-dire les expressions
polynomiales en x ou en 1 / x), ainsi que sur les expressions monômes
(ou simples) et composées connues (c’est-à-dire les quantités numé-
riques, y compris irrationnelles).39 Ce projet, qui a abouti avec
al-Samaw’al al-Maghribı̄,40 est également redevable à la tradition des
lectures algébriques du Livre X des Éléments.41
Chez ces algébristes, si la définition euclidienne des nombres de-
meure pour caractériser les entiers, l’approche opératoire du Continu
38
Traité d’algèbre, dans Rashed et Vahabzadeh, Al-Khayyām mathématicien, p. 122.
39
R. Rashed, ‘‘Al-Karajı̄’’, Dictionary of Scientific Biography, vol. VII (New York, 1973),
p. 240–6; repris dans Entre arithmétique et algèbre, pp. 31–42, aux pp. 32–7.
40
Al-Bāhir en algèbre, Eu dition, notes et introduction par Salah Ahmad et Roshdi Rashed
(Damas, 1972); R. Rashed, ‘‘Les recommencements de l’algèbre aux XIe et XIIe siècles’’, dans
J. E. Murdoch et E. D. Sylla (éd.), The Cultural Context of Medieval Learning (Dordrecht,
1975), p. 33–60; repris dans Entre arithmétique et algèbre, pp. 43–70.
41
M. Ben Miled, ‘‘Les quantités irrationnelles dans l’œuvre d’al-Karajı̄’’, dans R.
Morelon et A. Hasnawi (éd.), De Zénon d’Élée à Poincaré, Les Cahiers du Mideo, no 1
(Paris, 2004), pp. 27–54.; et Opérer sur le Continu, p. 185 et suiv.; p. 217 et suiv.
nécessite de considérer numériquement les quantités rationnelles et

irrationnelles. Dans al-Badı̄‘,42 al-Karajı̄ définit comme des nombres
les lignes droites rationnelles et irrationnelles introduites par Euclide
dans le Livre X. Il les partage en quantités simples et en quantités
composées. Les quantités simples sont en nombre infini, elles forment
les di#érents ordres que sont les racines n-ièmes des entiers:
Je te montre la transposition de ces dénominations (les grandeurs ration-
nelles et irrationnelles définies dans le Livre X) aux nombres et je les
augmente [. . .]. Je dis que les quantités simples sont infinies. La première est
rationnelle dans l’absolu comme cinq; la deuxième est la rationnelle en
puissance, comme racine de dix; la troisième est connue par rapport à son
cube, comme l’arête de vingt; la quatrième est la médiale, elle est connue par
rapport à son carré-carré, comme la racine de la racine de dix; la cinquième
est le côté du carré-cube; ensuite [vient] le côté du cubo-cube. Et ainsi elles
se divisent à l’infini.43
Puis, en composant par additions et di#érences les quantités simples,
al-Karajı̄ obtient une infinité de quantités composées.44 Al-Karajı̄
appelle choses ou encore racines, non seulement les inconnues des
équations, mais aussi des variables servant à décrire les opérations
algébriques sur les inconnues et les connues. La chose peut prendre
des valeurs rationnelles ou irrationnelles. Ces valeurs ont le statut
de nombre. Dans al-Fakhrı̄, il définit la racine comme le nom donné à
un type de nombres: ‘‘Sache que la racine est un nom pour chaque
nombre multiplié par lui-même’’.45 Plus loin, en classant les quantités
rationnelles et irrationnelles suivant qu’elles sont simples ou com-
posées, il les qualifie de nombres: ‘‘Le nombre se divise en deux
classes, simple et composée’’.46 Il introduit une définition calcula-
toire de la valeur du rapport qui s’oppose à celle d’al-Khayyām tout
en l’éclairant par endroit. La chose (shay’), traditionnellement
l’inconnue des algébristes, devient alors une variable représentant la
valeur du rapport:
Le rapport de toute grandeur (miqdār) à une autre grandeur, est une chose
qui, lorsqu’elle est multipliée par ce à quoi on rapporte, donne ce que l’on
rapporte. Et cette règle est celle que l’on a donnée dans le chapitre de la
division (qisma) car la division et le rapport, dans ce sujet, sont mêmes.47
42
Rashed, ‘‘Les recommencements de l’algèbre’’, p. 44–9.
43
Al-Karajı̄, L’algèbre al-Badı̄‘, Manuscrit de la Bibliothèque vaticane Barberini
Orientale, no. 36, 1, édité par Adel Anbouba (Beyrouth, 1964), p. 29, l. 16 et sqq.
44
Il faut rapprocher cette définition des quantités rationnelles et irrationnelles de celle
d’al-Māhānı̄.
45
Al-Karajı̄, Al-Fakhrı̄, dans A. S. Sa‘ı̄dān, Tārı̄kh ‘ilm al-jabr fı̄ al-‘ālam al-‘arabı̄, 2 vol.
(Koweit, 1986), vol. I, pp. 95–308, vol. I, p. 98 (je traduis).
46
Ibid., p. 101.
47
Ibid., p. 115. Plus bas, al-Karajı̄ précise que la division di#ère du rapport: [. . .] car
quatre relativement à vingt est di#érent de vingt relativement à quatre. Vingt relativement
à quatre induit le sens de la division: c’est le quintuple; tandis que quatre relativement à
vingt induit le sens du rapport: c’est le cinquième. Et entre cinq et le cinquième il y a une
Ainsi, le rapport A / B est identifié à la solution de l’équation
A = xB. Le nombre n’est donc pas ici une multiplicité d’unité mais
une quantité numérique rationnelle ou irrationnelle. L’objectif à
atteindre étant l’établissement d’un calcul algébrique sur ces quan-
tités connues (ainsi que sur les inconnues), la distinction importante
ne se situe plus entre les quantités rationnelles et les quantités
irrationnelles, mais entre les quantités simples (ou monômes) et les
quantités composées (ou polynômes). Le nombre est défini dans un
cadre où les di#érentes puissances négatives et positives de
l’inconnue sont établies et tous les produits, que nous noterions
aujourd’hui xn · xm = xn + m, avec n et m des entiers relatifs, étudiés.
L’unité appartient alors à un ‘‘rang’’, celui de x à la puissance 0: ‘‘La
partie du māl (carré algébrique) par le māl est un’’.48 Les nombres
sont à leur tour ramenés à ce ‘‘rang’’: ‘‘Et toute chose, si elle est
divisée par une chose de même genre (i.e. qui lui est homogène), ce
qui en est obtenu est un nombre. Comme la division des māl par les
māl, et des cubes par les cubes’’.49
Ces passages extraits d’al-Badı̄‘ et d’al-Fakhrı̄ nous indiquent que
les notions de rapport et de nombre ont aussi une histoire chez les
algébristes de l’école d’al-Karajı̄.50 Qu’elles y sont liées aux calculs
algébriques (formels) sur des variables représentant des inconnues
ou des expressions polynomiales connues. Dans cette tradition, la
notion de nombre renvoie à celle de rang.51
grande di#érence’’. Dans sa thèse, Bijan Vahabzadeh soulève le problème suivant: ‘‘Il y a un
point précis qui nous a posé un problème particulièrement délicat et que nous devons
mentionner: c’est que dans le Livre III de l’opuscule d’al-Khayyām, chaque fois que l’unité
intervient dans une proportion, la position qu’elle a dans la proportion nous semble
problématique. Par exemple, le rapport de A à B étant donné, al-Khayyām fixe une
grandeur U en tant qu’unité, et prend la valeur G telle que le rapport de U à G soit égal au
rapport de A à B. Il dit par la suite que G est le rapport de A à B. Or cela nous semble
problématique, puisque la fonction de la grandeur G, qu’il considère comme un nombre, est
justement d’exprimer une mesure (ici celle du rapport de A à B), et qu’une mesure consiste
à savoir ‘‘combien de fois’’ une quantité contient l’unité, ce qui est justement exprimé par le
rapport de la quantité à l’unité et non pas le contraire.’’ (‘‘Trois commentaires arabes’’, pp.
VI–VII.) La question soulevée ici est importante, elle concerne la valeur même d’un rapport
de grandeurs incommensurables. La di#érence entre les notions de rapport et de division,
qu’al-Karajı̄ précise ici, constitue un élément de réponse à ce problème.
48
(1 / x2 ) · x2 = 1 (al-Karajı̄, Al-Fakhrı̄, I, p. 101).
49
Al-Karajı̄, Al-Fakhrı̄, I, p. 112; Ben Miled, Opérer sur le Continu pp. 185 et suiv.; 217 et
suiv. Cela pourrait apporter un éclairage au passage d’al-Khayyām, cité p. 13: ‘‘Le rapport
entre grandeurs est associé à une chose numérique, ou dans la potentialité du nombre’’.
C’est-à-dire que le rapport entre grandeurs appartiendrait à la puissance, i.e. au rang, de
l’unité et des nombres. Ce passage d’al-Khayyām s’explique également de la façon suivante:
le rapport de deux grandeurs est ‘‘joint aux notions de nombre et d’unité ou en puissance’’,
cas où les grandeurs sont incommensurables, ‘‘ou en acte’’, cas de la commensurabilité;
comme le dit al-Khayyām lui-même, dans le passage cité en p. 14.
50
Cette tradition remonte également à Sinān b. al-Fath * (voir R. Rashed, ‘‘L’idée de
l’algèbre selon al-Khwarizmi’’, Fundamenta Scientiae, 4 [1983]: 87-100; repris dans Entre
arithmétique et algèbre, pp. 17–29, à la p. 21, n. 11; Ben Miled, Opérer sur le Continu, p. 262,
n. 8).
51
Sur l’évolution de la notion de ‘‘rang’’ à celle de degré, voir Ben Miled, ‘‘Les quantités
irrationnelles dans l’œuvre d’al-Karajı̄’’ et Opérer sur le Continu, pp. 208–13.
Il demeure qu’une étude complète et comparative des di#érentes

approches du traitement numérique du Continu dans les mathéma-
tiques arabes, reste encore à mener. Elle devrait prendre en compte
non seulement les points de vue géométrico-algébrique d’al-Khayyām
et arithmético-algébrique d’al-Karajı̄, mais aussi ceux des géomètres
infinitésimalistes des IXe–XIe siècles,52 ou encore ceux des mathé-
maticiens qui, se situant en aval d’al-Karajı̄, donnèrent des approxi-
mations numériques des nombres irrationnels et inventèrent les
fractions décimales.53
52
Voir Rashed, Les mathématiques infinitésimales.
53
R. Rashed, ‘‘Analyse combinatoire, analyse numérique, analyse diophantienne et
théorie des nombres’’, dans Histoire des sciences arabes, II, pp. 55–91.
Arabic Sciences and Philosophy, vol. 18 (2008) pp. 19–58
AL-FAz RA
z BIz’S LOST TREATISE ON CHANGING
BEINGS AND THE POSSIBILITY OF A
DEMONSTRATION OF THE ETERNITY OF THE
WORLD*
MARWAN RASHED
École Normale Supérieure, 45 rue d’Ulm, 75230 PARIS Cedex 05, France
Email: marwan.rashed@ens.fr
Abstract: This article proposes a reconstitution of the philosophical tenor of

al-Fārābı̄’s lost treatise On Changing Beings (Fı̄ al-Mawǧūdāt al-mutaġayyira). It is
shown that this work is not only a response to book VI of John Philoponus’ Contra
Aristotelem, but that its real issues can only be grasped in the context of the author’s
metaphysical system. Although, for al-Fārābı̄, genuine demonstrations proceed from
the cause to the caused, thus following the order of being, it will be explained how
he also admits a strictly physical proof of the simple fact, independently of its cause,
and that the physical demonstration of the eternity of the world pertains to this type
of proof. This physical proof is specifically directed against the Kindian doctrine of
creation.
Résumé: Cet article se propose de reconstituer la teneur philosophique du traité

perdu Sur les êtres changeants (Fı̄ al-Mawǧūdāt al-mutaġayyira) d’al-Fārābı̄. On
montre que cette œuvre n’est pas seulement une réponse au livre VI du Contra
Aristotelem de Jean Philopon, mais que ses enjeux véritables ne se laissent saisir que
dans le cadre du système métaphysique de l’auteur. Si, pour al-Fārābı̄, la démons-
tration véritable va de la cause au causé, suivant ainsi l’ordre de l’être, on
expliquera comment il admet également une preuve strictement physique du simple
fait, indépendamment de sa cause, preuve dont relève la démonstration physique de
l’éternité du monde. Cette preuve physique est spécifiquement dirigée contre la
doctrine kindienne de la création.
Al-Fārābı̄’s treatise On Changing Beings (Fı̄ al-Mawǧūdāt al-

mutaġayyira), now lost, is attested by the ancient bibliographers and
alluded to by later scholars. The material for the reconstruction is
more or less known since Steinschneider’s book on al-Fārābı̄,1 and it
leaves no place to doubt as to al-Fārābı̄’s general concern: he
elaborated on the arguments of Physics VIII 1 for the eternity of
*I would like to thank Dr Peter Adamson who invited me to present the first stage of this
research at the Conference ‘‘The Age of al-Fārābı̄’’ (London, 2006), and to whom my
English text owes many improvements.
1
Cf. M. Steinschneider, Al-Fārābı̄ (Alpharabius). Des arabischen Philosophen Leben und
Schriften etc. (Saint Petersburg, 1869), pp. 119–23.
20 MARWAN RASHED
motion. But nobody, so far as I know, has attempted to determine in

a more precise way how the treatise bears upon Aristotle’s argu-
ments.2 Was it intended to be an independent proof for the eternity of
the world? Or only an important step in a more complex one? Or a
reflexion on the physical and cosmological aspects of the transmis-
sion of motion? Or a mere refutation of certain attacks directed
against Aristotle? Hence my present purpose is not so much to give
a philological account of all the testimonies than to suggest what
may have been the treatise’s internal organization and, above all, its
significance for our understanding of al-Fārābı̄’s views on Aristotle’s
cosmology.
It is well known that Philoponus’ argument against Physics VIII 1,
as expressed in the sixth Book of his Contra Aristotelem, was one of
al-Fārābı̄’s targets.3 It is perhaps less well known – and does not seem
to have been emphasized by the specialists – that Books IV and VIII
of Galen’s De demonstratione may help to explain certain peculiar
features of al-Fārābı̄’s strategy.4 This case, however, is not as clear as
the situation with Philoponus, because Galen is surely not the
immediate adversary al-Fārābı̄ aims at rebuking. His function, I
would say by way of anticipation, is rather to remind al-Fārābı̄ of the
extreme di$culty of demonstrating the eternity of the world, if one is
really cautious to commit neither a petitio principii nor an appeal to
the senses or to the testimony of mankind.
These two scholars constitute, so to say, the ‘‘Greek frame’’ of
al-Fārābı̄’s reflexion. But in any philosophical debate, contemporary
issues are always at stakes. What was the philosophical present of
al-Fārābı̄? Certainly, the insistance of the mutakallimūn on the
non-eternity of the world constitutes one of its major elements. But
the most peculiar feature of this debate is perhaps the fact that two
scholars who a priori would have sided against the theologians,
namely al-Kindı̄ and T I ābit ibn Qurra, stood in fact half way between
the two camps: al-Kindı̄, because he extended Aristotle’s finitism
beyond its Aristotelian limits, namely to time, which was for Aristo-
tle a case of potential infinity;5 and T I ābit ibn Qurra, because he did
not only deny the potentiality of Aristotle’s infinity, but went so far
as to admit every form (temporal, mathematical, substantial) of
2
Some aspects have been dealt with by J. Puig Montada, ‘‘Zur Bewegungsdefinition im
VIII. Buch der Physik’’, in G. Endress and J. A. Aertsen (eds.), Averroes and the Aristotelian
Tradition (Leiden / Boston / Köln, 1999), pp. 145–59.
3
See Puig Montada, ‘‘Zur Bewegungsdefinition im VIII. Buch der Physik’’, pp. 146–50 and
H. A. Davidson, Proofs for Eternity, Creation the Existence of God in Medieval Islamic and
Jewish Philosophy (New York / Oxford, 1987), pp. 43–4.
4
On al-Fārābı̄’s knowledge of Galen’s De demonstratione, see below, p. 24.
5
See the collection of treatises on this theme in R. Rashed and J. Jolivet, Œuvres
philosophiques et scientifiques d’al-Kindı̄, vol. 2: Métaphysique et cosmologie (Leiden /
Boston / Köln, 1998).
AL-FA
z RA
z BIz’S ON CHANGING BEINGS 21
actual infinity.6 Thus, and that is to my mind an important point, to
argue for time’s infinity without admitting actual infinity was not, in
al-Fārābı̄’s days, a small matter, nor one that would sollicit broad
agreement even within the narrow philosophical tradition. Indeed
al-Fārābı̄’s deafening silence regarding al-Kindı̄ may be due, at least
partially, to the burden the latter put on the former’s shoulder by
arguing so carefully against the infinity of time.
I. SOME UNCONCLUSIVE ARGUMENTS FOR THE ETERNITY OF THE

WORLD
Despite the fact that he relied extensively on al-Fārābı̄’s writings,

Maimonides sided with the theologians on the question of the world’s
eternity. He agreed with Galen’s claim that the issue did not admit of
demonstration, and this fact was su$cient, according to him, for
declaring true the litteral meaning of the Holy Scriptures: the world
as a whole had a beginning.7 Still, Maimonides delivers us with an
account of numerous arguments for the eternity of the world. He
first presents four physical ‘‘methods’’ (t*uruq) he attributes to
‘‘Aristotle’’ himself. He then declares that those ‘‘who came after
him (viz Aristotle) have elicited from his philosophy some methods,
by which they establish the eternity of the world with respect to God,
praised be His name’’.8
Al-Fārābı̄’s influence on Maimonides is well-known and it is hardly
possible that the latter excluded the former from the group of
Aristotle’s followers.9 Maimonides’ division of the proofs between
‘‘natural’’ and ‘‘theological’’ thus raises a problem. Unless one
demonstration is more concise, elegant, or general than another,
there is no reason to multiply proofs showing one and the same
conclusion. Unfortunately, Maimonides gives us here no hint as to
why Aristotle’s followers added proofs to those of the Master. What
is the meaning of this clear-cut division between natural (and
ancient) proofs on the one side, theological (and modern) proofs on
the other? Which role, if any, did al-Fārābı̄ play in the history of the
arguments for the eternity of the world? These are all problems
6
See the edition and English translation in A. Sabra, ‘‘Thābit ibn Qurra on the infinite
and other puzzles. Edition and translation of his discussions with Ibn Usayyid’’, Zeitschrift
für geschichte der arabisch-islamischen Wissenschaften, 11 (1997): 1–33. I shall publish soon
an improved edition of this important text.
7
See Mūsā Ibn Maymūn, Dalālat al-H * ā’irı̄n, ed. H. Atay (Ankara, 1974), p. 350 sqq. On
the fact that Maimonides does not believe in the eternity of the world but uses it as a
postulate, see R. Rashed, ‘‘Philosophie et mathématiques selon Maı̈monide: le modèle
andalou de rencontre philosophique’’, in T. Lévy and R. Rashed (eds.), Maïmonide
philosophe et savant (1138–1204) (Leuven, 2004), pp. 253–73, esp. pp. 260–1.
8
Ibn Maymūn, Dalālat, p. 311, ll. 6–7.
9
In his famous letter to Samuel Ibn Tibbon, Maimonides clearly alludes to the fact that
he considers al-Fārābı̄ second to nobody, except perhaps Aristotle. See S. Munk, Mélanges
de philosophie juive et arabe (Paris, 1955), p. 344.
22 MARWAN RASHED
raised by the treatise On Changing Beings. For even if al-Fārābı̄

seems to have agreed with the arguments of Physics VIII 1 – to which
his treatise was devoted –, we still have to determine the exact nature
of this agreement. Did he hold that the arguments were correct and
su$cient to prove the eternity of the world? Correct but insu$cient?
Incorrect but, once amended, su$cient? Irretrievably incorrect and
insu$cient? The answer to these questions will enlarge our under-
standing of the relationship between al-Fārābı̄ and the Aristotelian
tradition.
Towards adressing these issues, in the first section I begin with a
brief presentation of some of the physical methods appearing in
Maimonides’ Guide, and suggest how they might seem to be insu$-
cient.10 In the second section I turn to the arguments actually dealt
with in Physics VIII 1 according to al-Fārābı̄, and to the interpreta-
tion he presumably gave of them in On Changing Beings. Since they
seem unconvincing as well – at least in light of Galen’s criticisms –,
I consider more widely in the third section some tenets of al-Fārābı̄’s
philosophical system, in order to determine what his full account of
the eternity of the world may have been.
1. Argument from the nature of matter

The argument from the nature of matter – matter must be already
there if one is to explain every change such as generation11 – is
problematic for two reasons. First, it takes for granted what the
theologians, and al-Kindı̄ along with them, precisely deny, i.e. the
impossibility of the coming into existence of something from nothing.
Thus, even if this argument could be considered as scientifically
valid, its polemical e$cacy was feeble. Secondly, and perhaps more
decisively, this argument is not su$cient to assess the validity of
Aristotle’s position as against that of the Timaeus. The eternity of
the world viewed as this three-dimensional enmattered space is a
common good of the whole Greek philosophical tradition.12 What
distinguishes Aristotle’s doctrine from his master’s is that according
to the former, it is always this world which has existed, with all its
celestial individuals and biological species. Abū Bakr al-Rāzı̄ may
serve to illustrate the point, for he explicitly argued for the eternity
of matter while adhering to a Platonic cosmogony. We can extend
10
I shall here rely on Davidson, Proofs for Eternity, pp. 12–30.
11
12
As the author of the G { am‘ himself puts it: cf. Abū Nas*r al-Fārābı̄, L’harmonie entre les
opinions de Platon et d’Aristote, texte et traduction, F. M. Najjar et D. Mallet (Damascus,
1999), §58, p. 135, ll. 7–8 Najjar. Mallet, ibid., p. 134, translates: ‘‘En e#et, tous les discours
des savants des autres doctrines et des autres confessions ne montrent, dans le détail, que
l’éternité de la matière <dans le passé> et sa permanence <dans le futur>’’.
AL-FA
z RA
these remarks to the argument from the vacuum, since it ends with
the a$rmation of matter, and hence of eternity.13
2. Argument from the concept of possibility

A more serious candidate is the argument from the concept of possi-
bility.14 The possibility of a thing resides in the actual existence of
some other thing. Otherwise, the possible thing would be either al-
ready existent (and hence be necessary, on pain of infinite regress), or
incapable of existing, hence never existing and presumably not poss-
ible after all, but rather ‘‘impossible’’. This argument, however, begs
the question, since it leaves out any ‘‘theological’’ concept of possi-
bility, i.e. possibility as the thing’s existence in God’s mind. It will thus
fail to persuade the adversary. If one is to admit the thesis of creation
ex nihilo, then one can admit on the same ground the possibility of
anything subsequently created by God out of nothing. Or, more simply,
even if one does not allow the existence of miracles in the world
history, one cannot rule out the eventuality of a divine creation of the
world, followed by never ending physical processes such as genera-
tions and heavenly revolutions. According to this line of thought, God
would only create the world and give it its first impulse. We find
something of this sort in the writings of John Damascene.15
3. Argument from the nature of the celestial spheres

That leaves us with the argument from the nature of the celestial
spheres,16 which appears inconclusive for similar reasons. To infer
the absence of corruption from the absence of contrariety in the
cyclical motion is an assumption which can resist neither one of the
following objections. First, it does not rule out a distinction between
the temporary (the Cosmos as we know it) and the permanent (the
period of disorder plus the period of order, Timaeus’ scheme): even if
we admit that the three-dimensional space is eternal, it is possible
that the universe was previously in a pre-cosmic state, with no
sublunar elements and no celestial spheres.
A second objection was made by Galen himself and, more interest-
ingly, it was quoted by al-Fārābı̄. In his commentary on De caelo I 3,
270b 4–5, Averroes writes the following:17
13
Cf. Davidson, Proofs for Eternity, pp. 14–15.
14
15
Joannes Damascenus, Expositio fidei, ed. B. Kotter (Berlin, 1973); [ = Patristische Texte
und Studien 12], Section 8 (pp. 18–31), p. 21, ll. 69–70 in part.
16
17
Averrois Cordubensis commentum magnum super libro De celo et mundo Aristotelis, ex
recognitione Francis James Carmody † in lucem edidit Rüdiger Arnzen, 2 vols. (Leuven,
2003), t. I, pp. 44–5, textus 22, ll. 83–91.
24 MARWAN RASHED
Then, <Aristotle> said: ‘‘and reason confirms the sight and the sight,
reason’’ etc. As to such propositions, Abū Nas*r said that in their case,
conviction is very near to certain truth. And since Galen thought that
nobody can know whether the world is eternal except through these
propositions which take their origin in sensation and in the testimony of the
past, he said in his book ‘‘On his own opinions’’ that he had no knowledge
about whether the world was created or eternal a parte ante. And it is clear,
from what he says in his book which he has called ‘‘Demonstration’’, that on
the subject of the world’s eternity, he does not use types of propositions
di#erent from these.18
It is obvious that Averroes owes his quotation of Galen’s De
demonstratione to the passage of al-Fārābı̄ he is presently dwelling
upon. We can reconstruct the following process. While commenting
on Aristotle’s sentence, al-Fārābı̄ remarked that the type of proposi-
tions employed here by Aristotle was not absolutely demonstrative.
Hence his probable remark: Galen was not entirely mistaken in
objecting that every physical proposition on the eternity of the world
was a priori condemned by the fact that in this matter, we must
necessarily rely on sense-perception and the Ancients’ testimony. To
say that the celestial realm was ever so is to accept as true some
ancient astronomical observations transmitted from generation to
generation. But everybody will agree that observation is not free
from error, nor historical transmission absent from distortions.
We are here confronted with an exegetical problem, one we can
fortunately solve thanks to Abū Bakr al-Rāzı̄’s quotations in his
Doubts against Galen. If Galen, in his De demonstratione, makes use
of such a kind of ‘‘nearly apodictic’’ propositions that we find in
Aristotle’s De caelo, does this entail that he attempted in this work
to prove some sort of cosmic eternity? Though it is what al-Rāzı̄
wants us to believe that he did, such a reading must be rejected on
the basis of two important testimonies on De demonstratione IV,
namely a self-reference in On Marasmus (Marc. VII, 671) and a
discussion in Philoponus’ Contra Proclum, 599.17#. Both texts show
that Galen argued against the sole claim that everything that has
been generated must perish.19 Galen’s position is that we cannot
prove whether the world has been generated or not and that we
cannot prove whether it will perish or not. But although there is no
18
Deinde dixit: ‘‘Et ratio testatur visui et visus rationi’’ etc. Tales propositiones in eis
dixit Albunacir quod fides est propinquissima veritati certe; et cum Galienus estimavit quod
nullus potest scire mundum esse eternum nisi per has propositiones quarum origo est a
sensu et testimonio vetustatis, dixit in suo libro quem posuit in eis que credidit quod nullum
certum habebat de mundi utrum esset novus aut antiquus; et manifestum est quod ipse non
utitur in antiquitate mundi nisi talibus propositionibus ex verbis suis in libro suo quem
appellavit Demonstrationem.
19
In collaboration with Riccardo Chiaradonna, I am presently preparing a new edition,
with a commentary, of the extant fragments of Galen’s De demonstratione. We shall give
there a full account of these fragments.
AL-FA
z RA
hint whatsoever which would make the world’s generation more
probable than its eternity a parte ante, the cosmic stability seems to
suggest that it will not pass away. As al-Rāzı̄ puts it in what is likely
to be an exact quotation:
Galen categorically a$rmed, in the fourth book of his treatise On demon-
stration, that the world does not pass away, saying: ‘‘if the world were
subject to corruption, neither the bodies which are in it would stay in the
unique state they have, nor the distances between them, nor the quantities,
nor the movements, and it would be necessary for the sea’s water that was
before us to pass away. But none of these things leaves its state nor changes.
For astronomers have observed them during many thousands of years.
Necessarily, then, the world is not subject to decay; therefore, it does not
undergo corruption’’.20
It seems likely that the astronomers’ observations mentioned in this
text are precisely what al-Fārābı̄ alludes to in his commentary on
Aristotle’s De caelo. Still, Galen made use of ‘‘this kind of proposi-
tions’’ not to demonstrate the eternity of the world a parte ante, but
the mere likelihood of its eternity a parte post.
We can now return to al-Fārābı̄’s thesis. The first thing to be noted
is that al-Fārābı̄ appears as an acute reader of Galen. For he accepts
the non-cogency of such propositions. The only disagreement with
Galen is that al-Fārābı̄ sides with Aristotle in thinking that we can
apply these ‘‘nearly certain’’ propositions to the question of past
eternity, whereas Galen limits their use to future eternity. Given
that, we must be very careful when interpreting al-Fārābı̄’s propin-
quissima veritati certe, ‘‘very near to certain truth’’. For according to
someone as deeply convinced of the necessity of logical rigor as was
al-Fārābı̄, to be ‘‘very near to the truth’’ is not the same as being true.
Nor, a fortiori, is it the same as being apodictic. Admittedly, al-Fārābı̄
allowed some non apodictic proposition to belong to science. But
they never belong to, so to say, the ‘‘most scientific’’ part of science.
They are always didactic, propaedeutic, etc.21 Is al-Fārābı̄ likely to
have accepted these propositions in a demonstration of the eternity
of the world? Obviously not. The issue was much too sensitive and
important for him be satisfied by a mere approximation of the truth.
However, in another passage dealing with the eternity of the world
in the context of Galen’s agnosticism, al-Fārābı̄ claimed that the
eternity of the celestial sphere, contrarily to that of the sublunar
world, was perfectly demonstrable. Along the lines of Top. I 11, if
perhaps in a more elaborate way, al-Fārābı̄ begins his discussion of
20
Cf. Muh * ammad ibn Zakariyā al-Rāzı̄, Kitāb al-S { ukūk ‘alā G
{ ālı̄nūs (Dubitationes in
Galenum), ed. M. L. ‘Abd al-Ghanı̄ (Cairo, 2005), p. 44.
21
See, for example, al-Fārābı̄, Kitāb al-Hat*āba, in Deux ouvrages inédits sur la Rhétorique,
ed. J. Langhade and M. Grignaschi (Beirut, ˘ 1986), pp. 59–61 (Arabic text) and pp. 58–60
(French translation).
26 MARWAN RASHED
what a problem is by stressing the numerous aspects by which it can

be considered as particularly important. Roughly speaking, these
aspects can lie in the elevated matter of the question under consider-
ation, in its di$culty, or in its significance for the common people. A
very good example of such a problem consists in the interrogation ‘‘is
the world eternal a parte ante or not?’’, since it contains in itself each
aspect which contributes to make a problem a significant one:
And some of the things about which the philosophers’ opinions are divergent
are of important significance. And their importance and magnificence arise
either from their nobility in themselves, or from the nobility of the things
which are known in virtue of them, or because it is most useful for the people
to have a knowledge of them; or their importance arises from the di$culty of
attaining their causes, or from the di$culty of the path leading to the
discovery of their demonstrations. For example, our saying ‘‘is the world
eternal a parte ante (azalı̄) or not?’’ is among the things about which
philosophers diverge, and it is important in virtue of the fact that the essence
of the quaesitum is in itself noble – since it is the world as a whole; and to this
is conjoined the nobility of the thing one is led to by this knowledge. Indeed,
its knowledge is the path leading to divine science. Furthermore, attaining
the causes of its eternity a parte ante, if one is to prove that it is eternal a
parte ante, is di$cult, and attaining the causes of its incipience, if one is to
prove that it is incipient, is di$cult as well. Furthermore, the knowledge
people have of it is of important usefulness for them. Moreover, if one
happens to be in error in such cases, then that will be a cause of error in very
many things, and if one happens to hold a correct view thereon, that will be
a cause of attaining the correct view about very many things.22
In the lines immediately following, al-Fārābı̄ gives further examples
of such ‘‘problems’’. It is to be noted that in each case they pertain to
the theoretical part of philosophy, and not to ethics. The first
example, already quoted, is the first part of the first Kantian
antinomy. Al-Fārābı̄ then mentions its second part. Then follows
something akin to the second Antinomy, and ultimately some prob-
lems belonging to the question of time and modality:
To this kind pertain our sayings: ‘‘is the world finite or infinite?’’; ‘‘is body
divisible ad infinitum?’’; ‘‘is it possible for a thing that its existence be
possible without this thing being existent at all, neither in the past nor in
the future?’’; ‘‘is there a thing for which it is possible by virtue of its nature
to be non-existent without non-existence occurring to it either in the past or
in the future?’’; ‘‘is it possible for what has never ceased to exist in the past
to pass-away in the future?’’; ‘‘is it possible for what will never cease to exist
in the future not to have existed in the past?’’.23
22
Cf. al-Fārābı̄, Kitāb al-G{ adal, in al-Mant*iq ‘inda al-Fārābı̄, ed. R. al-‘Aǧam, 3 vols.
(Beirut, 1985–6), vol. 3, pp. 80–1. On this passage, see also G. Vajda, ‘‘A v propos d’une
citation non identifiée d’al-Fārābı̄ dans le ‘Guide des Egarés’ ’’, Journal Asiatique, 253
(1965): 43–50.
23
Al-Fārābı̄, Kitāb al-G
{ adal, p. 81 al-‘Aǧam.
AL-FA
z RA
The fact that each one of the problems quoted belongs to theoretical
philosophy shows that al-Fārābı̄ does not here consider dialectic as
exclusively concerned with ethical discussion: it can also deal with
scientific problems. The question that arises is of course the follow-
ing: is dialectic, by itself, able to prove one horn or the other of the
problem? Before answering it, al-Fārābı̄ describes in which sense,
and to what extent, the question of the eternity of ‘‘the world’’ is
semantically ambiguous:
Such things deserve to be examined and scrutinized at length, and also that
one dedicate one’s e#orts to apply dialectic to them. And that is Aristotle’s
purpose when he says: ‘‘and regarding those things about which we have no
argument, in reason of24 their importance, it is in our opinion that our
saying ‘in virtue of what?’ about them is di$cult, such as our saying: ‘is the
world eternal a parte ante or not?’ ’’ (Top. I 11, 104b 14–16 cf. 505.15–16
Badawı̄). For this example he proposes is very dialectical under one aspect,
since when we say ‘‘is the world eternal a parte ante or not?’’, insofar as we
employ this wording, it is not possible at all that we produce a certain
syllogism, neither of the fact that it is eternal a parte ante nor of the fact
that it is not eternal a parte ante. For our word ‘‘the world’’ is an ambiguous
word and, moreover, taken as indefinite. So, if the world is taken in its
entirety in such a way, <it will be found to have> many parts, one of which
is clearly not eternal a parte ante, another such that it is possible to produce
about it a syllogism showing that it is eternal a parte ante, and another of
unclear status. Thus, when we take the world in its entirety, it is sometimes
eternity a parte ante which is imagined, and sometimes incipience, so that
we always produce opposed syllogisms. The only way then is to examine, for
each of its parts, whether it is eternal a parte ante or not, and in how many
ways a thing can be eternal a parte ante, and in how many ways it is said to
be not eternal. This is the method leading to the production of its
demonstration, whereas according to the first method, it is not possible to
produce its demonstration, the syllogisms produced being opposed syllo-
gisms in each case.25
From this text, one could gain the impression that it is possible for
dialectic to prove the eternity or the non-eternity of the world. But a
final paragraph indicates that for al-Fārābı̄, there was a distinction
to be made between a dialectical discussion of the question – where
the word ‘‘the world’’ is taken as an indefinite whole, so that
inferential arguments can be produced in favour of both sides of the
problem – and a scientific (demonstrative) one, where the term is
split into its di#erent meanings (or, which amounts to the same, the
object into its di#erent parts):
And for this reason, since Galen the doctor did not follow the demonstrative
method, on this quaesitum in particular, he thought that there was no
24
Correcting aw into id I , in conformity with Badawı̄’s edition (Mant*iq Arist*ū [Cairo,
1980], vol. 2, p. 505, l. 15).
25
Al-Fārābı̄, Kitāb al-G
{ adal, pp. 81–2 al-‘Aǧam.
28 MARWAN RASHED
demonstration of it, that the demonstrations pertaining to it were counter-

balancing each other, and that it was among the things subject to doubt. It
is for this reason that Aristotle has considered these quaesita as most proper
to dialectic, since the discussion about them, when they are taken in such
ways, can be neither settled nor decided.26
Al-Fārābı̄, if I am right, accuses Galen of not having understood that
the dialectical discussion of the eternity problem, even though
technically possible and theoretically justified, was not the only
possible one. For scientific demonstration (probably by dividing and
examining separately the di#erent parts of the world) has a role to
play, and in fact it alone is able to settle the issue. Unfortunately,
there is not the slightest hint, in al-Fārābı̄’s text, regarding the
method to be employed in demonstrating the eternity of some part
(presumably the celestial one, or a part of it) of the world. Let us also
note that in this context, the mention of some part of undetermined
status is puzzling: is there in the world some cosmological place,
intermediate between the two others (say, for example, the sphere of
the Moon), in whose case suspension of judgment is required? Or is
some psychic principle in the chain of generations intended here?
Al-Fārābı̄ remains silent on this point.
This doctrine is confirmed by an excursus in the commentary on De
interpretatione, where al-Fārābı̄ explains the scientific usefulness of
metathetic propositions by the following example:
It is, therefore, wrong to say that this section is useless. If it were indeed
useless, how should we know whether a quaesitum on, say, motion, should
run: ‘is motion beginningless or incipient’, or: ‘is motion beginningless or
not beginningless’? Similarly with the question whether the world is
beginningless or incipient. Is this or the question whether the world is
beginningless or not beginningless the quaesitum? Would such a question be
a quaesitum even if unquantified? Or is no question a quaesitum unless
quantified like this: ‘Is every movement eternal, or is not every movement
eternal?’— ‘Is every world eternal, or is not every world eternal?’.27
The remark on quantification must be explained against the back-
ground of al-Fārābı̄’s understanding of the Topics. One of Galen’s
errors lies in the fact that he did not quantify his quaesitum, so that
his whole approach of the question the eternity of ‘‘the world’’ was
unsound. But that is not all. The present text shows also that even if
al-Fārābı̄ is ready to accept that in a way, ‘‘the world’’ is not
beginningless (ġayr azalı̄), since at least some of its parts must have
a temporal beginning, we have still to distinguish between ‘not
26
Kitāb al-G
{ adal, p. 82 al-‘Aǧam.
27
See Alfarabi’s Commentary on Aristotle’s ` ¢ ´ (De interpretatione), ed. W.
Kutsch and S. Marrow (Beirut, 1986), p. 222, ll. 2–8. The translation is borrowed from F. W.
Zimmermann, Al-Fārābı̄’s Commentary and Short Treatise on Aristotle’s De Interpretatione
(Cambridge, 1991), p. 217.
AL-FA
z RA
beginningless’ and ‘incipient’ (h* āditI). From the non-beginningless of
‘‘the world’’ as a whole, it does not follow that the world as a whole
is incipient, insofar as something h * āditI may be eternally h
* āditI.
II. THE TREATISE ON CHANGING BEINGS
We are left with the arguments from motion and time, which formed
the core of the treatise On Changing Beings. The crucial text for the
argument from motion is Phys. VIII 1, 251a 8–28, where Aristotle
intends to prove that there always has been and always will be
motion. The articulation of this passage is clear, and I shall begin
with a brief account of it.
• (1) Aristotle starts by recalling his definition of motion as it
appeared at Phys. III 1, 201a 10–11: motion is the actualization
of the movable qua movable. This is taken by Aristotle to
imply the existence of the movable. But the movable itself
must either have been generated after not having existed, or
be eternal a parte ante.
• (1.1) Let us first suppose that the movable has been gener-
ated. This fact presupposes the presence of some generative
motions previous to our supposed ‘‘first’’ motion. A
contradiction.
• (1.2.1) On the other hand, if the movable has not been
generated but was always existing at rest, that implies the
existence of some motion to prevent the movable from under-
going some (actual) motion. Hence, the allegedly first motion
was preceded by this motion. A contradiction.
• (1.2.2) Moreover, a motion will be required in order to
counter-act the eternal motion which was an impediment to
the movable’s motion, so as to permit its actual motion.
Another contradiction.
• At this point, Aristotle has proved by reductio that no motion
can take place without being preceded by another motion.
• (2) He then adds two subsidiary (cf.
` ` ´ ) argu-
ments from the nature of time.
• (2.1) The first one is that time presupposes motion; but
everyone, with the sole exception of Plato, admits that time
always has been and always will be. Therefore, the same is
true of motion.
• (2.2) The second argument – I take it, with MS E and the
Arabic translation (cf. p. 810: wa-ayd * an), that the text read ´ ;
˜’ at 251b 19 – is that the ‘‘now’’ is a mean (´ )
and not
between the past and the future. Therefore, there is no first
and no last ‘‘now’’. Hence, time is eternal in both directions,
and consequently motion.
30 MARWAN RASHED
There are two apparent flaws in this demonstration. The first one is
that the opponent is not likely to admit (with 1.1.) that the gener-
ation of the movable temporally precedes its motion.28 It is not
absurd to suppose that generation and motion take place simul-
taneously, and destruction and rest as well. The second one bears
upon time. For after all, it is Aristotle’s assumption that time is the
number of motion (cf. 2.1).29 One may very well suppose, without
contradicting oneself, that there was an infinite time of rest before
the first motion occurring in the world. To these two problems, we
must add a di$culty appearing in (2.2). Just as we can imagine a line
starting at one point and drawn to infinity, we can imagine time
starting at one ‘‘now’’ and going then always further. There is no
obvious reason which accounts for why every ‘‘now’’ should be
considered as a mean between a past and a future. From this point of
view, the correlation between time and motion does not help us, at
least prima facie: a motion starts at one ‘‘now’’ and then goes on.
Why then refrain from thinking that the world as a whole started its
existence and motion at one ‘‘now’’ (the first ‘‘now’’, so to say) and
then went on until today?
Three authors, Ibn Bāǧǧa, Ibn Rušd and Ibn Maymūn, all born
in Muslim Spain, have preserved some material from al-Fārābı̄’s
On Changing Beings. The testimonies – for testimonies they are
rather than fragments – can be divided into three groups. The texts
of the first group, which are by far the most numerous, are
concerned with the idea that there is a motion prior to every
motion. The second group contains two texts dealing with time.
The third group contains only one testimony, which is said to
come ‘‘from the beginning’’ of al-Fārābı̄’s treatise, where he is said
to have held the thesis that every motion is by definition
continuous.
It is obvious that there is a connexion between the first two groups
and the two parts of Aristotle’s demonstration. It is perhaps even
more obvious that there is a connexion between them and Philo-
ponus’ criticisms in Contra Aristotelem VI. For Simplicius’ refutation
is so extensive that we can gain through it a pretty clear idea of
Philoponus’ arguments: Philoponus first refuted the fact that the
moved thing is temporally prior to its motion, and then what he
understood as three Aristotelian arguments in favour of the eternity
of time.30 One of them, disproving time’s infinity, is al-Fārābı̄’s
obvious target in the second group of texts.
28
This reproach was already made by Philoponus. Cf. Simplicius, In Phys. 1133.16–1134.29
Diels (translated in Philoponus, Against Aristotle, on the Eternity of the World, transl. by C.
Wildberg (London, 1987), pp. 125–6 [ = fr. 109]).
29
Cf. Simplicius, In Phys. 1166.32–1167.16, translated by Wildberg, p. 139 ( = fr. 125).
30
Cf. Philoponus, Against Aristotle, transl. Wildberg, p. 133.
AL-FA
z RA
1. First group of testimonies: there is a motion prior to every motion
Averroes, in his commentary on Physics VIII 1, informs us that he
had first followed al-Fārābı̄’s interpretation, according to whom
Aristotle in this chapter aimed at proving that before every motion,
there was a motion. Because, according to him, this interpretation
was not entirely satisfying, Averroes reflected continuously on the
problem, until the true meaning of the text at last appeared to him:
Aristotle’s sole concern here was to prove the eternity of the whole
motion, i.e. the motion of the celestial sphere. Prima facie, the
di#erence seems rather tenuous, all the more so since al-Fārābı̄’s
ultimate goal, as we shall see, was to infer from the eternal succes-
sion of finite motions the existence of an everlasting continuous
motion. The first text where Averroes describes al-Fārābı̄’s position,
in a somewhat general way, is the following:
I say that the exposition which I have just presented is the one which is
understood at first sight, and this is what al-Fārābı̄ has understood, to judge
from what he has said in his book On Changing Beings. And Avicenna, as
well as Avempace the Andalusian, have understood the same thing, namely
that Aristotle’s intention in the first chapter of this book was to explain that
before every change, there is a change, and that change never ceases
according to the genus – so as to proceed from that to the explanation of the
fact that there is a first eternal motion – either one or many – which
contains everything. But there is some doubt on this issue. This is why
al-Fārābı̄ in his book On Changing Beings attempted to deal extensively
with this question, examining in how many ways it is possible to imagine
that before every motion, there is a motion, what can be true in this and
what not. For this reason, his inquiry into this is intricate.31
After alluding to his own attitude towards al-Fārābı̄’s interpreta-
tion – from acceptance to reservation –, Averroes reassesses his
point, which, paradoxically enough, found its first expression in
Philoponus’ criticisms: it is false, even for Aristotle, that prior to
every motion, there is a motion, since this is not true of the celestial
motion:
For Aristotle attempts to explain (1) whether the first motion containing
the world (or the first motions, if they are more than one) is created in such
a way that before, there was no motion at all; motion would then be created
out of nothing according to the genus, and everything would have begun to
move after nothing was previously moving; (2) or whether the first motion
which contains everything, present in the first thing moved (or in the first
things moved, if motions are more than one) is eternal, having never ceased
31
Cf. Aristotelis De Physico Auditu cum Averrois Cordubensis variis in eosdem
commentariis (Venetiis apud Iunctas, 1562), fol. 339A–B. See also ibid., 345C sqq. and Ibn
Rušd, Kitāb al-Samā‘ al-t*abı̄‘ı̄ (Hyderabad, 1365 H.), pp. 110–11 (also edited by J. Puig in the
Corpus Commentariorum Averrois in Aristotelem, Epitome in Physicorum libri [Madrid,
1983], pp. 134–5).
32 MARWAN RASHED
and never going to cease, being, as he said, ‘‘a sort of life for the things
existing by nature’’ (250b 14–15). And if he has proposed the explanation of
the motion’s above mentioned definition, it was in order to show that motion
is such <i.e. eternal> and not, contrarily to what al-Fārābı̄ and others have
thought, in order to explain that before every motion, there is a motion. The
inquiry he is here engaged in applies universally to the whole world; it is not
true, according to Aristotle’s opinion, that before the motion containing the
whole, there may be a motion, or that before the change which is the first
change, there may be a change. Therefore, their opinion was false. For this
opinion is particular: it is something true of the particular motions which
are contained by the universal motion, insofar as that <property> follows
from the universal motion and not because it exists in them primarily and
essentially. That for it <i.e. the first motion> it is impossible <i.e. to have a
motion before itself>, shall appear clearly later. Therefore, he could not
have assumed the eternal continuation present in these motions acciden-
tally as a sign of the eternal continuation in the first motion itself (or in the
first motions), to say nothing of the fact that it has not yet been explained
here that they32 are so accidentally – but this is manifest and will appear
more clearly later.33
Averroes nowhere appears to be more explicit on al-Fārābı̄’s pos-
ition. However, the fact that the latter examined as fully as possible
the di#erent types of relation between a mover and a moved is
transparent from Ibn Bāǧǧa’s quotations.34 It is obvious that Ibn
Bāǧǧa, in his commentary on the two last books of the Physics,
borrows many distinctions from al-Fārābı̄’s treatise, which he men-
tions by its title half a dozen of times. Al-Fārābı̄’s aim was very
probably to show that in each one of the distinguished cases of
motion, there was a motion prior to the motion considered. The
method employed is what he himself calls a perfect induction.35
Now, this way of arguing is obviously not a demonstration of the
eternity of the world. It is the demonstration that according to some
Aristotelian principles, the world is eternal. That is, if we dismiss the
idea of a demiurgic God who creates being and motion at the same
instant, if we accept only natural processes deprived of every kind of
volition, then Aristotle was correct in maintaining that there is a
motion prior to every motion. Philoponus held that it is false, even for
Aristotle, that there is a motion prior to every motion. One must be
very careful about this distinction: At this stage, Philoponus proved
not that the world had a beginning, but that according to the
principles of Aristotelian dynamics, it might have had a beginning.
Of course, this first result prepared the way for an assessment of the
32
reading eos for ens Junt.
33
Averroes, In Phys. 339C–F.
34
See the full list in Puig Montada, ‘‘Zur Bewegungsdefinition im VIII. Buch der
Physik’’, p. 151, n. 25.
35
See the valuable account in J. Lameer, Al-Fārābı̄ and Aristotelian Syllogistics. Greek
Theory and Islamic Practice (Leiden / New York / Köln, 1994), pp. 133–75.
AL-FA
z RA
Christian version of creation. But it did not constitute, as such, this
assessment. Al-Fārābı̄, symetrically, did not prove the eternity of the
world, but the fact that on some Aristotelian assumptions pertaining
to physics, motion had to be viewed as taking place eternally. Thus,
we must interpret al-Fārābı̄’s argument as a mere counter-argument
directed against Philoponus’ attack.
2. Second group of testimonies: the unreality of time

What now about the second tenet of Aristotle’s proof, time? In his
Epitome of the Metaphysics, Averroes mentions al-Fārābı̄’s On
Changing Beings in the following terms:
If time is generated, there will be a determinate ‘‘now’’, before which there
was no past time; but it is impossible to imagine a ‘‘now’’ determined in
actuality and present which is not preceded by a past, all the more to have
the representation of it when time is imagined according to its very essence.
Error can occur in this only if we are to imagine time with the help of what
imits it, namely the line. For the line, inasmuch as it has a position and
exists in actuality, is necessarily finite [. . .]. When, then, we represent time
to ourselves in this way, as if it were a straight line, it is impossible for it not
to be finite. This kind of deceit belongs to the Deceitful Topoi, under the
topos of transposition and substitution (tah * ta mawd* i‘i al-nuqlati wa-al-
ibdāl). Abū Nas*r has dealt at length with this notion in the Changing
Beings.36
The straight line, so al-Fārābı̄, is only an image of time. This image
is imperfect because the relation between the parts and the whole is
not identical in both cases. In the case of the line, all the parts exist
in actuality; in the case of time, no part exists in actuality. The
impossibility of an infinite line is only due to the actual existence of
all its parts, as soon as one recognizes the impossibility of an actual
infinity. Since time does not share this distinctive feature, the
impossibility of an infinite time has not been demonstrated.
It is very probable that al-Fārābı̄ himself, in On Changing Beings,
described the case as a particular type of sophism. For one passage of
his Deceitful Topoi (Fı̄ al-Amkina al-muġlit*a) elaborates on ‘‘trans-
position and substitution’’ as a sophistic category, itself divided into
many subspecies:
Among these topoi, there is also the transposition (al-nuqla) into that which
can be substituted (yubdala) to the thing and play its role: either word, or
similar, or universal, or particular, or prior, or posterior, or conjoined, or
things opposed, or imagination we have of the thing in the soul, or its
sensible tokens.37
36
Ibn Rušd, Kitāb Mā ba‘d al-t*abı̄‘a, ed. Hyderabad, pp. 128–9.
37
See al-Fārābı̄, al-Amkina al-muġlit*a, ed. R. Al-‘Aǧam, in al-Mant*iq ‘inda al-Fārābı̄,
vol. 2, p. 160.
34 MARWAN RASHED
In the course of his discussion, al-Fārābı̄ gives some examples of

these ‘‘imaginations’’ that we may substitute to the thing itself:
As to the imaginations of the things in the soul, they are deceitful insofar as
many things, at many times, can be represented only under the form of
something else. To these things pertains what cannot be, or can hardly be,
represented under its proper form, such as [our representation of what is
before the world – for it immediately occurs to our souls an infinite time
before it – and] our representation of what there is outside the world – for
it immediately occurs in our souls either an infinite void, or an infinite
body.38
The example of time, which I have put between brackets, must be
interpolated. Otherwise, we would have nothing but a statement of
Philoponus’ and al-Kindı̄’s creational creed,39 so that the whole
project of al-Fārābı̄ would become unintelligible. This textual con-
jecture receives some confirmation from a parallel passage in the
Qiyās, where the problematic illustration is not to be found:
For example, we know by demonstration that the sun is bigger than the
earth. But such as we see it, we imagine that its size is one foot on one foot,
so that what we imagine about the sun is opposed to what we know about
it. And similarly,40 we know by demonstration that there is no infinite body
or infinite line outside the world. But when41 we imagine the world’s sphere,
it immediately occurs to our soul either an infinite void, or an infinite body
– so that what we imagine to be outside of the world is contradicts what we
know.42
There is not mention in the Qiyās of what there was ‘‘before the
world’’. I take it for legitimate, therefore, to excise this illustration in
the first passage. Averroes’ allusion in the Epitome of the Meta-
physics argues for the opposite idea: time being eternal in both
direction, it is meaningless to speak of a ‘‘before’’ of the world.
What, then, is exactly the status of past and future things? Ibn
Maymūn tells us that they were ‘‘imaginative and not real’’:
38
See al-Fārābı̄, al-Amkina al-muġlit*a, p. 161.
39
See Philoponus, In Phys. 613.19–27, who also alludes to the ´ of the
´ in
the same context and Kindı̄, First Philosophy, ed. Rashed-Jolivet, p. 21, ll. 13–18. For an
English translation (less accurate than the French one, ibid., p. 20), see A. I. Ivry,
Al-Kindı̄’s Metaphysics (Albany, 1974), p. 63: ‘‘. . . whatever has no matter and is not joined
to matter is not represented in the soul at all, and we do not think that it is a
representation. We acknowledge it only because it is a necessity to a$rm it, as when we say
that outside the body of the universe there is neither void nor plenum i.e., neither emptiness
nor body. This statement is not represented in the soul, for ‘neither void nor plenum’ is
something which the sense has not apprehended, and it is not attached to a sense so that it
could have an image in the soul, or be believed to have an image. It is something which
only the intellect necessarily perceives, in accordance with the premises which will be set
forth’’.
40
Reading ka-d I ālika instead of li-d
I ālika.
41
Reading lammā instead of kamā.
42
Al-Fārābı̄, S
{ arh * al-Qiyās, ed. M. Danesh Pajuh, in al-Mant*iqiyyāt li-al-Fārābı̄, 3 vols.
(Qom, 1408–10H), vol. 2, p. 451.
AL-FA
z RA
They are things exclusively imaginative and not real <wahmiyya lā wuǧūdiy-
ya>. Abū Nas*r al-Fārābı̄ made this contention and discovered the topoi of
imagination in all its particular cases, as will clearly and manifestly appear
to you, if you reflect without bias on his book known as On Changing
Beings.43
This discussion is interesting not only because it shows how al-
Fārābı̄ answered Philoponus’ and al-Kindı̄’s temporal finitism. It may
also attest al-Fārābı̄’s sensitiveness as regard to the fundamental
problem of time: time being intrinsically tied to motion, the eternity
of time presupposes the eternity of motion. At Physics VIII 1,
Aristotle as interpreted by Philoponus developed three arguments in
favour of the eternity of time.44 He sees the first in the presence of a
‘‘before’’ and an ‘‘after’’ with respect to any event. The second in the
fact that all Physicists, except Plato, have assumed time to be
eternal. The third in the fact that time includes the ‘‘now’’ in its
definition; but the ‘‘now’’ is nothing else than a mean between two
periods; there can be, therefore, neither a first ‘‘now’’ nor a last one.
We can of course dismiss the second argument, for it is merely
confirmative. I shall come back later to the first and third arguments.
For the moment, I shall confine myself to two di$culties, the first
arising from the relation between time and motion, the second from
the relation between time and the ‘‘now’’.
The first di$culty is this: if time is nothing but something our
intellect apprehends in motion, then there is no sense to speak of a
temporal ‘‘before’’ of the universe: there was no motion before the
universe, because there is simply no ‘‘was’’ correctly refering to such
an ahistorical period. Time is not to be thought of independently of
the world taken as a set of things in motion. Time is an epiphenom-
enon arising from a phenomenon of the world, namely motion. There
is something naive, and un-Aristotelian, in arguing for the impossi-
bility of a time before motion and concluding from there to the
necessity of the motion’s eternity. For the supposition of a time
before motion is absurd.45
The second di$culty lies in the danger of vicious circularity in
arguing for the relation between time and the ‘‘now’’. When we say
that the ‘‘now’’ is a limit between two periods of time, isn’t it the case
that we define the ‘‘now’’ by time? But time is notoriously defined by
43
Ibn Maymūn, Dalālat, 221,3–5.
44
Cf. above, p. 30.
45
One might argue that time’s eternity is deduced from an internal experience, akin to
our intuition of the first principles. After being internally convinced by the necessity of
time being eternal, and once we have realised that time is a concomitant of motion, we
conclude that motion too is eternal. But this seems to be in contradiction with Physics IV,
where it was shown not that time presupposes motion, but that our perception of time
presupposes our perception of motion (cf. Phys. 219a 3–4: ‘´ ` ´ ’´ `
´ ). It is impossible, then, to distinguish between a subjective and an objective aspect
of time. Time is the entirely subjective apprehension of an entirely objective phenomenon.
36 MARWAN RASHED
the ‘‘now’’. Furthermore, even if it is not a vicious circle, we have to

explain why it is not a petitio principii. For if time and the ‘‘now’’ are
not to be dissociated, it is plain that in using the ‘‘now’’ in order to
demonstrate something about time, we may be suspected to make use
of our conception of time in order to prove our conception of time.
This is precisely the reproach Philoponus addresses to Aristotle.
It is evident that al-Fārābı̄ was aware of these di$culties, which
were certainly not new when he wrote his treatise.46 A concrete hint
can be find in a passage of his treatise On Deceitful Topoi, where he
explains the error of the ‘‘definitions’’ which take as a definiens some
part of the representation we have of the definiendum. By way of
example, al-Fārābı̄ mentions ‘‘the one who has defined motion as the
body’s passage and time as the period that the motion counts’’.47
These two cases must be understood against their Aristotelian
background. In the case of time, we should note that Aristotle has
not defined time by a period of time, but with reference to anteriority
and posteriority: ‘‘time is the number of motion according to the
before and the after’’. Hence, al-Fārābı̄ implicitly rejects Galen’s
(and others’) criticisms against Aristotle’s definition of time as
implying circularity.
I conclude this section by pointing out that the sole references to
the treatise On Changing Beings, in the context of a discussion of
time, are dialectical: al-Fārābı̄ attempted to explain why temporal
infinity must not be counted as a case of actual infinity, hence his
defense of the possibility that time be eternal. Here again, we find a
counter-argument opposed to Philoponus, but not a real – i.e. positive –
demonstration of the eternity of some part of the world (namely, the
heavens).
3. Some further remarks on the demonstrability of eternity

At this stage, our initial task seems to be complete. According to the
transmitted testimonia, it is plain that al-Fārābı̄ refuted two argu-
ments of Philoponus. But no more than that. In particular, there is no
evidence that al-Fārābı̄ tried to give a positive proof of the eternity of
the heavens. When we try to account for this curious absence, the
first explanation which presents itself to us is the general structure
46
It is the core of Galen’s arguments against Aristotle’s definition of time in De
demonstratione VIII.
47
Al-Fārābı̄, al-Amkina al-muġlit*a, p. 153.6–7 al-‘Aǧam: man h * adda al-h * arakata bi-annahā
zawālu al-ǧismi wa-al-zamāna muddatan ta‘udduhā al-h * arakatu. It is noteworthy that the
first definition ultimately goes back, through Abū ‘Alı̄ al-G { ubbā’ı̄, to al-Kindı̄ (see R. M.
Frank, Beings and Their Attributes. The Teaching of the Basrian School of the Mu‘tazila in
the Classical Period [Albany, 1978], p. 117, n. 30), whereas the second appears exactly under
this form in al-Kindı̄’s Epistle of definitions (see the edition in ‘A. al-A‘sam, al-Mus**talah *u
al-falsafı̄ ‘inda al-‘Arab [Cairo, 1989], p. 192: al-zamānu muddatun ta‘udduhā al-h * arakatu,
ġayru Itābitati al-aǧzā’i). On al-Fārābı̄’s critical appraisal of al-Kindı̄, see below, p. 51.
AL-FA
z RA
of al-Fārābı̄’s system, i.e. the relationship between physics and
metaphysics. Let us first say a word about this possible explanation.
In the Great Book of Music, after a classification of sciences
according to the way they bear upon their principles, al-Fārābı̄ deals
with analysis and synthesis.48 The former proceeds from the last
consequences to the principles, the latter from the principles to the
last consequences. In some cases, synthesis follows the order of
existence: the principle of explanation, there, coincides with the
principle of existence. That this idea is of paramount importance for
al-Fārābı̄’s cosmology is confirmed by two texts. First, in the Philos-
ophy of Plato, al-Fārābı̄ identifies the whole philosophical method
with division (qisma) and composition (tarkı̄b, also the Arabic word
for ‘‘synthesis’’), which is a radicalization of what we find in his
Greek predecessors.49 Secondly, the Attainment of Happiness devotes
some lengthy pages to explaining how, after having regressed up-
ward to some first principle, we then use this principle in our way
downward to prove some facts which were unknown until then.50 To
anticipate, I think that the motion’s eternity pertains to this cat-
egory: taking motion as a fact, we go ‘‘upward’’ and prove the
existence of some mover; but the existence of this mover will explain
in turn the fact that motion is eternal.
Thus, one may hazard that for al-Fārābı̄, the assumption of
nature’s existence in II 151 is a condition for establishing by analysis
the eternity of motion in VIII 1 and the existence of the Prime Mover,
but that the assumption of the Prime Mover is necessary for
demonstrating by synthesis the reality of nature. The first move
seems relatively uncontroversial. It consists in assuming that the
analytic progression of the Physics is not limited to book VIII, but
begins much earlier in the treatise. As to the second move, it is
grounded on the assumption that the full justification of the ‘‘natu-
ralism’’ of book II can be given only after the regressive method has
reached its highest point (the Prime Mover). Until then, the ‘‘evi-
dence’’ of natural motion, as it is a$rmed in Physics II 1, is to be
considered as only probable. The ‘‘synthesis’’ will start with a
consideration of the Prime Mover and then prove some facts about
the natural world. It is only when we examine God’s attributes such
48
See Kitāb al-Mūsı̄qā, ed. Gh. ‘A. Khashaba and M. A. el-Hefny (Cairo, 1967), pp. 185–8.
49
See al-Fārābı̄, Falsafat Aflāt*ūn, ed. F. Rosenthal and R. Walzer (London, 1943), p. 15, l.
18–p. 16, l. 1; transl. M. Mahdi in Alfarabi. Philosophy of Plato and Aristotle, revised edition
(Cornell, 2001).
50
See al-Fārābı̄, Kitāb Tah
**sı̄l al-sa‘āda, ed. G
{. A
z l Yāsı̄n, in al-A‘māl al-falsafiyya (Beirut,
1981), vol. 1, pp. 124–9. English translation in Mahdi, Alfarabi. Philosophy of Plato and
Aristotle, pp. 16–19. It is very interesting to note that according to this text, the role of
synthesis is not only to reconsider the data discovered by analysis, but to use them in order
to discover new facts. The whole method of analysis-synthesis, then, far from being a mere
didactic way of exposition, has a deep heuristic value.
51
See below, p. 43.
38 MARWAN RASHED
as unicity, simplicity, etc. that we shall demonstrate some of the

nature’s features – which, of course, will not contradict the truths of
the Physics.
Briefly put, we shall demonstrate that divine emanation entails the
di#erent kinds of motion we know and, pari passu, that our basic
conception of nature as expressed in Physics II was correct.52 That
may be what al-Fārābı̄ wanted to emphasize in the numerous pas-
sages where he alludes to the fact that after having ascended to the
highest part of physics and reached some beings which pertains to
metaphysics, we have then first to theorize them in the frame of the
metaphysical discipline and, subsequently, to go downward and
examine again, under this new perspective, physical objects. In the
Book of Letters, he says that in this second move, we have to examine
the physical objects ‘‘insofar as these things <i.e. the metaphysical,
non-categorial, beings> are their causes’’.53 It is plain that the new
element is the idea of a hierarchy between beings. Here one might
appeal to the idea that the celestial beings, being ‘‘nearer’’ to the
First Principle, must have stronger unitary features – such as
temporal, material and kinetic continuity – than the sublunary
items. This move appears also, albeit briefly, in the summary of what
al-Fārābı̄ knows as the two last books of the Metaphysics. According
to him, book XI ( = ) is devoted to ‘‘the principle of substance and
being as a whole, as well as the assessment of its essence and of
the fact that it is essentially knowing the truth of the essence’’.54 It
is also devoted to ‘‘the separate beings which are after it and to
the modality of the hierarchy of beings out of it’’.55 As to book XII
( = M), it is concerned with ‘‘the principles of the physical things
and of the mathematical things’’,56 i.e. two kinds of being which have
helped us to prove the existence of the First Principle and which will
be now ontologically justified as being caused by this Principle. I
shall not engage here in the distinctions which ought to be made
between two apparently di#erent models, one where analysis corre-
sponds to physics and synthesis to metaphysics, and the other where
both analysis and synthesis belong to metaphysics. Such a task
would need a thorough discussion of the summary of the Metaphysics
52
I agree with H. Zghal, ‘‘Métaphysique et science politique: les intelligibles volontaires
dans le Tah **sı̄l al-sa‘āda d’al-Fārābı̄’’, Arabic Sciences and Philosophy, 8 (1998): 169–94,
p. 176, n. 16, that the first chapters of the A z rā’ Ahl al-Madı̄na al-Fād
* ila and some other
‘‘political’’ works of al-Fārābı̄ are examples of this divine knowledge of beings.
53
Cf. al-Fārābı̄, Kitāb al-H* urūf, ed. M. Mahdi, 2nd edition (Beirut, 1990), p. 29, ll. 17–21.
54
Cf. F. Dieterici, Alfārābı̄’s Philosophische Abhandlungen aus Londoner, Leidener und
Berliner Handschriften herausgegeben (Leiden, 1890), p. 38. German translation: id.,
Alfārābı̄’s Philosophische Abhandlungen aus dem arabischen übersetzt (Leiden, 1892), p. 60:
fı̄ mabda’i al-ǧawhari wa-al-wuǧūdi kullahu wa-itIbāti huwiyyatahu wa-annahu ‘ālimun
bi-al-d
I āti h
* aqqa al-d I āt.
55
Ibid.: wa-fı̄ kayfiyyati tartı̄bi wuǧūdi al-mawǧūdāti ‘anhu.
56
Ibid.: fı̄ mabādi’ al-t*abı̄‘iyyāti wa-al-ta‘lı̄miyyāt.
AL-FA
z RA
and of the Philosophy of Aristotle which lies beyond the scope of the
present discussion. However, it is important to note that in the first
case at least, the ‘‘metaphysics’’ intended by al-Fārābı̄ is explicitly
not that of Aristotle. Al-Fārābı̄ is aware that we find hardly any
‘‘synthetic physics’’ in the genuinely Aristotelian corpus, except
perhaps in Book M as interpreted in the summary. Thérèse-Anne
Druart has collected a great deal of evidence showing that al-Fārābı̄
was conscious of this insu$ciency, from his point of view, of
Aristotle’s writings.57 It would seem that he conceived of his philo-
sophical task as to supply the incomplete system with its lacking
part.58
57
See in particular her article ‘‘Al-Fārābı̄ and emanationism’’, in J. Wippel (ed.), Studies
in Medieval Philosophy (Washington, D.C., 1987), pp. 23–43. The author, however, does not
establish any connexion between al-Fārābı̄’s dissatisfaction with Aristotelian metaphysics
and his interest for analysis and synthesis as such. It may be suggested that al-Fārābı̄ was
more ready to accept the completeness of Aristotle’s Metaphysics – Book M being its
synthetic part – than that of its physics. The ‘‘metaphysics’’ we lack is in fact an
emanationist cosmology, obviously not furnished by Aristotle’s De caelo.
58
To the texts discussed by Druart, I would like to add an unnoticed reference in
Averroes’ Great Commentary on the Physics (181C–E, ad Phys. IV 11, 219a 30–b 3), where we
find attributed to al-Fārābı̄ a recurrent criticism of some Aristotelian procedure. Al-Fārābı̄
is said to have repeatedly (multotiens) objected to Aristotle for his simultaneous use of (i)
the ‘‘method’’ of ‘‘finding something through some other thing found’’ ( via . . . aliquid
invenire per aliud inventum ) and (ii) ‘‘the method of suppression’’ ( via ablationis). Averroes
does not agree with the Second Master and sides with the First: true, the first method,
taken alone, is deficient ( diminuta = nāqis*a); but if it is associated with the second, then,
contrary to al-Fārābı̄’s opinion, we have the correct proof we were in search of. For
example, the recognition of time as ‘‘the number of motion according to prior and
posterior’’, which proceeds along this double path, is correct. What, now, was Aristotle’s
method with which al-Fārābı̄ repeatedly expressed his dissatisfaction? I think that the
remote background is furnished by A.Po. I 5, where we find associated the twin concepts of
¢
´ and ’˜ in the discussion of essential predication. In the Arabic translation
of Abū Bišr Mattā ibn Yūnus, these words are rendered by wuǧida / wuǧūd and irtafa‘a /
irtifā‘, and it seems very likely that we have here the Arabic words lying behind the Latin
invenire and ablatio. But I do not think that what is intended here is exactly the doctrine
set forth by Aristotle in A.Po. It is much more reminiscent of criticisms adressed by
al-Fārābı̄ to the theologians (cf. al-Fārābı̄, Kitāb al-Qiyās, ed. R. Al-‘Aǧam, in al-Mant*iq
‘inda al-Fārābı̄, vol. 2, pp. 49 #.; on this text, see Lameer, Al-Fārābı̄ and Aristotelian
Syllogistics, pp. 204–32, in part pp. 219–27). The first of these is directed against their habit
of spelling out what property is conjoined with some other in every hyparctic subject minus
the one under consideration, to conclude that it belongs also to the one under
consideration. Al-Fārābı̄ objects that every inductive procedure, in order to be valid, must
be complete. The second criticism is focused on their use of suppression. Because of their
nominal conception of a ‘‘cause’’ ( ‘illa), the theologians contend that if the suppression of
the cause from a subject implies the suppression of some attribute, then the presence of the
cause in it implies the presence of the attribute. But as al-Fārābı̄ notes, according to an
Aristotelian understanding of the ‘‘cause’’, we would rather have to conclude that the
presence of the attribute implies the presence of its cause. Interestingly enough, al-Fārābı̄
goes on to remark that the simultaneous use of both methods is redundant. Now, it is
obvious that al-Fārābı̄ cannot have in mind, when dealing with Aristotle, exactly what he
objected to the theologians. It is more likely that he felt some dissatisfaction even with the
use of the amended methods such as he explains them. The reason is obvious: neither
method can go beyond the level of the mere fact. They are unable, by themselves, to give us
the cause of the phenomenon under consideration. In other words, they are typically
Aristotelian in so far as they allow analysis to proceed, but they have to be supplemented
by a synthetic demonstration.
40 MARWAN RASHED
The synthetic proof would run as follows: the eternal beings are
not in time in the sense that their being is not strictly included in the
whole of time, but they are in time in the sense that their permanence
is coextensive with time. In the Opinions of the Inhabitants of the
Virtuous City, al-Fārābı̄ insisted from the outset on the First’s
eternity, understood as an infinite time.59 And in the Political
Regime, he begins by stating the doctrine that ‘‘when (matā) the
First exists according to the existence which belongs to Him, then
it follows necessarily that all other natural existents exist out of
Him . . .’’.60 The fact that al-Fārābı̄ establishes a temporal connexion
between God’s and the creatures’ existence is striking. One may
object that matā need not to have temporal connotations here and
could just introduce the first half of a conditional. But the temporal
connexion appears clearly one page later, where al-Fārābı̄ goes so far
as to a$rm: ‘‘and for this reason, the existence of what comes out of
Him is not at all temporally posterior to Him (ġayra muta’ahhirin
‘anhu bi-al-zamān), even if it is posterior to Him in every˘ other ˘
acception of posteriority’’.61 Thus, even if God is not circumscribed by
time, He exists at every time. Even if He creates time, God exists at
every ‘‘now’’.
If God, accordingly, is temporally eternal, i.e. eternally temporal,62
we easily understand why the first result of emanation should be
eternal. The argument is the following. Actual infinity is to be
excluded from the chain of causation, i.e. at every moment, the
chain between any event and its first e$cient cause must be finite,
otherwise there would be no event at all. On the other hand,
temporal eternity, or even eternal regularity a parte ante, is no
explanation of the fact that something exists here and now.
Already Alexander of Aphrodisias, in his De providentia, had briefly
explained away the domino e#ect, which was more than an elusive
temptation of Aristotle’s biology.63 I have tried elsewhere to show
that in the latter’s well-known motto ‘‘A man begets a man and the
sun’’, Alexander attempted to reduce the role of man and emphasize
that of the astral principle.64 All this leads to the following alterna-
tive: either (i) every event is a sort of miracle directly produced by
God’s particular volitions – the mutakallimūn’s solution, with vari-
ants – or (ii) there must be at least one physical cause of every-
thing in the world, which is itself under God’s direct influence.
59
See the passage quoted below, n. 62.
60
Al-Fārābı̄, Kitāb al-Siyāsat al-madaniyya, ed. F. M. Najjar, 2nd edition (Beirut, 1993),
p. 47, ll. 11–12.
61
Ibid., p. 48, ll. 19–20.
62
As it is a$rmed in the A z rā’, p. 27 Karam-Chlala-Jaussen: fa-li-hād
I ā huwa azalı̄, dā’im
al-wuǧūd bi-ǧawharihi wa-d I ātihi.
63
See De providentia, 93.8–95.16 Ruland.
64
See Essentialisme. Alexandre d’Aphrodise entre logique, physique et cosmologie (Berlin /
New York, 2007), pp. 278–85.
AL-FA
z RA
The fundamental character of this alternative has been noted by
Maimonides. After having presented in the Guide numerous argu-
ments for the eternity of the world, he rejects this thesis because it
leads to the negation of miracles.65 To understand this assertion, one
must in fact address the problem the other way round. Philosophers
in the Aristotelian and Neoplatonist tradition, because of their
conception of emanation, are sceptical about all forms of perturba-
tion a#ecting the uniformity and continuity of God’s way of acting on
the world, in particular miracles. It is because of some conception
pertaining to God’s simplicity, permanence and changelessness that
miracles, according to them, are better avoided. The case of motion
is only a particular instance of a more general claim regarding
uniform causality. Therefore, whatever may have been stated in the
Changing Beings, it is clear that the physical proof as such was
incomplete. It is true under the condition that we accept that motion
is not a perpetual, or a least indefinitely repeated, miracle occurring
in the world. But in this case, provided that the item discovered
by analysis is eternal, unchanging, e$cient and simple, we can
legitimately infer that the world is eternal.
Things, however, are not so simple. For it is plain that in the
Philosophy of Aristotle, al-Fārābı̄ holds both that Physics VIII con-
forms itself to some sort of regressive pattern leading to the First
mover and that it proves the eternity of motion.66 Let us quote the
relevant passage:
(A = chap. 1–3) Then he investigated, among other things, what the quid-
dity of motion implies for the motion’s succession in time, in terms of infinity.
(B = chap. 4–5) Then he gave many rules in the case of bodies, rules
which are entailed by their motion and by the principles moving them: the
moved bodies around us are moved by other bodies which are next to and in
contact with them, and these again by others which are next to and in
contact with them, and these again by others next to and in contact with
them. The bodies which move other bodies are near to them in their
positions, or in contact with them, successive to one another; but that does
not reach infinity in number.
(C) And after he has laid down in what precedes (i = Bk VII, chap. 2) in
how many ways and according to how many types the natural body moves
by its nature another body, and having said (ii = chap. 7) that the last body
which moves some moved body successive to it is also moved, but only
according to local motion at the exclusion of the other types, (iii = chap. 8)
that its local motion is not straight but circular, (iv, cf. chap. 6) that it
moves around all other moved natural bodies, and (v, cf. chap. 9) that it is
impossible that there be behind this another body moving it; and since he
65
Ibn Maymūn, Dalālat, pp. 350–1.
66
Here, al-Fārābı̄ develops and systematizes ideas set forth by Greek commentators. See
D. Morrison, ‘‘Philoponus and Simplicius on tekmeriodic proof’’, in Eckhardt Kessler (ed.),
Method and Order in Renaissance Philosophy of Nature. The Aristotle Commentary Tradition
(Ashgate, 1998), pp. 1–22.
42 MARWAN RASHED
had previously shown that it is impossible that there be an infinite body;

therefore, it follows from that that there is some finite body which moves all
the natural bodies and that the last thing which is contained by this body is
moved circularly around the others.
(D = chap. 6?) Then, he investigated this body which is moved circularly:
is it moved without mover, or has it a mover? And he showed that it has a
mover.
(E) (i = chap. 10) Then he investigated the principles moving the bodies
moved circularly by nature: are they bodies or not, are they incorporeal
essences, even if they are in matter and body? After having carefully
investigated this point, it appeared to him that what gives these bodies
which are at the extremity their circular motion is some being, which can be
neither a nature, nor natural, neither body nor in a body. It is never and in
no way in matter. (ii) We must examine it in another investigation and
examination, di#erent from the investigation and examination proper to
natural philosophy.67
This summary reveals how al-Fārābı̄ reshapes the content of Physics
VII and VIII in order to underline the basic analytic character of the
argument. It would take me too long to enter into all the details here,
but we can note the partial suppression of chap. 6, one of the most
important and di$cult of Book VIII. To put it roughly, this chapter
relies on the arguments for the eternity of motion as they appear in
the first chapter,68 in order to prove the continuity of motion, hence
its unicity,69 hence the existence and unicity of the mover.70 That
means that for Aristotle, the proof for the eternity of motion is an
important step in the general demonstration of the unicity of the
Prime Mover. But if we turn to al-Fārābı̄’s text, we see that
continuity does not appear at all, whereas eternity is mentioned only
at the beginning of the presentation and seems to play no role
afterwards. Al-Fārābı̄ presents it rather as if it were a result in itself,
perhaps important for our comprehension of the world, but with no
bearing upon the demonstration that follows. As a corollary, we can
note that the introduction of the circularity of motion in al-Fārābı̄,
compared to what we find in the Physics, is feeble – which is not very
surprising, since it was for Aristotle a direct consequence of the
continuity of motion. To sum up, al-Fārābı̄ neglects all that pertains
to the ‘‘horizontal’’ axis of Aristotle’s proof, i.e. time’s infinity, and
concentrates exclusively on the ‘‘vertical’’ one, i.e. the necessarily
finite regression from e#ect to cause at every instant. We can even say
that for al-Fārābı̄, the demonstration of the Prime Mover would be
exactly the same even if the world was temporally finite.71
67
Falsafat Arist*ūt*ālı̄s, pp. 96–7.
68
Cf. in particular Phys. VIII 6, 259a 15.
69
Ibid., 259a 16–18.
70
Ibid., 259a 18–20.
71
Whether by influence or coincidence, al-Fārābı̄ adopts here a strategy already used by
Alexander in Quaestio I 1. Alexander puts forward what he baldly calls a ‘‘showing’’ ( ˜ )
AL-FA
z RA
I think that the two peculiarities of al-Fārābı̄’s summary of Physics
VIII, i.e. its stress on the regressive character of the whole book and
its presentation of chapters 1–3 as if they bore no relation to the rest
of the book, are two tenets of the same fundamental interpretation:
physics as such is incomplete, and we have to wait until the
‘‘metaphysical turn’’ in order to be able to demonstrate, by way of
synthesis, the world’s eternity. But that does not imply that the
demonstration of eternity as a mere fact is incomplete.
Let us explain this. It is plain that according to al-Fārābı̄, book
VIII represents the achievement of the whole Physics. In the
previous books of the Physics, Aristotle spelled out motion and its
concomitants, so that when starting our reading of book VIII, we
already know that motion exists, how it must be conceived, and its
relation to place and time. Thus, we can safely aim at establishing
something on the ground that there is sublunar motion, and that
each sublunar motion (i) is continuous and (ii) has a beginning and
an end.
Of course, the knowledge of ‘‘nature’’ we acquire through Physics
II is (dialectically) well-grounded.72 But can it be considered as
‘‘certain knowledge’’? In the Conditions of Certitude, al-Fārābı̄ men-
tions, among the six criteria that must be met in order for ‘‘absolute
certitude’’ to obtain, the fact that the knower must know that the
proposition p he knows is true, that it is impossible that p not be true
and that there is no time at which p can be false.73 But even if I
believe that there is no direct divine agency in the world and the fact
that there is no direct divine agency in the world is true, how can I
know that this proposition is true, that it is impossible for it not to be
true and that there is no time, past or future, at which it may be false?
How to convince Plato, Galen, Philoponus, the Mutakallimūn, that it
is not a petitio principii? Of course, Aristotle wrote at Physics II 1:
What nature is, then, and the meaning of the terms ‘‘by nature’’ and
‘‘according to nature’’, has been stated. That nature exists, it would be
absurd to try to prove; for it is obvious that there are many things of this
kind, and to prove what is obvious by what is not is the mark of a man who
is unable to distinguish what is self-evident from what is not. (This state of
mind is clearly possible. A man blind from birth might reason about
colours.) Presumably therefore such persons must be talking about words
without any thought to correspond.74
of the existence of the First Mover, which, because of its analytical character ( ’
’´ ), he considers not to be a demonstration ( ’
´ ). Interestingly enough,
Alexander, like al-Fārābı̄ after him, proceeds from sublunar motions to celestial motion to
the First Mover.
72
See Falsafat Arist*ūt*ālı̄s, p. 93.
73
On this text, see D. Black, ‘‘Knowledge ( ‘ilm) and certitude ( yaqı̄n) in al-Fārābı̄’s
Epistemology’’, Arabic Sciences and Philosophy, 16 (2006): 11–45.
74
Phys. II 1, 193a 1–9 (ROTA translation).
44 MARWAN RASHED
But it is evident that in al-Fārābı̄’s time, there were people talking

about words, but with thoughts to correspond – these thoughts being
the entire conceptual edifice of kalām – who were convinced that
nature was, if not a mere flatus vocis, in any case no more than a
designation of God’s habits. And this point was crucial in the
demonstration of the eternity of the world from the succession of
motions. Unlike Aristotle, al-Fārābı̄ cannot have considered as
scientifically su$cient merely to a$rm the existence of nature (i.e.
motion) as we know it.
To conclude this section, we have to distinguish between two kinds
of proof for the world’s eternity. The real proof is taken from the
general cosmology of al-Fārābı̄, and is as such ‘‘metaphysical’’ (in
the sense of the Book of Letters), or synthetic: it derives the heavens’
eternity from the First Principle’s eternity, simplicity, etc. But this
does not preclude that there may be some physical (or analytical)
proof of the fact that the heavens are eternal.75 A consideration of
time, motion, place, would give us the analytical necessity of time’s
beginningless. The rather clumsy allusion of the Philosophy of
Aristotle would point to this second proof.
4. A physical demonstration of the eternity of the world?
(a) On time’s definition

The last part of the present paper is more speculative. I would like to
use some fleeting allusions scattered in the Arabic tradition in order
to reconstruct al-Fārābı̄’s physical proof of the eternity of motion. I
think that this proof may stem from the considerations about the
instant in Aristotle’s Physics. As already said, Aristotle presented as
an argument in favour of time’s infinity the fact that by its very
nature, the instant is an intermediary, or a mean, between the past
and the future. Thus, it is by definition impossible that there be a first
instant. Therefore, time is infinite. But the phrase ‘‘by definition’’ in
the previous sentence is ambiguous. If we can argue that a first
instant is physically absurd, Aristotle is of course correct. But if the
sole absurdity is that it contradicts our verbal definition of an
instant, itself grounded on time’s infinity, then, as Philoponus
remarked, it is a blatant petitio principii.
What about al-Fārābı̄? Time, for him, seems to be perfectly intel-
ligible in terms of being coextensive (int*ibāq), a concept which
75
The distinction here is one between demonstrative proofs, which show the dioti by
giving a causal middle term, and proofs that merely show the hoti. But what is the causal
middle term involved in the proof for eternity? I assume that it is God’s per se attributes of
eternity and e$ciency. We would have something like this: ‘God is an eternal cause; every
eternal cause has an eternal e#ect; therefore God has an eternal e#ect (the world)’. I am
grateful to Peter Adamson for helping me to formulate this point.
AL-FA
z RA
would be perfectly applicable even if time were finite. Let us quote a
passage appearing in the Book of Letters:
The particle ‘‘when’’ is used as a question about the relation between what
occurs and the time determined, known, and coextensive with it, as well
as a question about the two extremities of this time which are coextensive
with the two extremities of the existence of what occurs – be it a body or
bodiless –, once it has been at rest or moved, or in something at rest or
moved. No being requires a time in which its existence takes its shape,
nor <a time the existence of which would be> the cause of the existence of
any being whatsoever (1). For time is something (2) which necessarily
surpervenes upon motion. It is a plurality which is counted by the intellect
in such a way that by means of it the [intellect] estimates and determines the
existence of what is either moved or at rest.76
(1) aw <ilā zamānin yakūnu wuǧūdūhu> sababan li-wuǧūdi mawǧūdin as*lan added by
me (cf. H* urūf 83.18: fa-inna sababa wuǧūdi al-zamān huwa an etc. and Falsafat Arist*ūt*ālı̄s
96.3: laysa yuh * tāǧu ilayhi [i.e.: al-zamān] fı̄ wuǧūdi mawgūdin as*lan): aw <li-yakūna>
sababan li-wuǧūdi mawǧūdin as*lan added by Mahdi 0 (2) šay’un my proposition: matā MS,
Mahdi.
These lines, despite two textual corruptions, are clear: al-Fārābı̄

explains, as in the other texts referred to in my apparatus, that time
76
Alfarabi’s Book of Letters, ed. M. Mahdi, 2nd edition (Beirut, 1986), 62.2–10. This text
seems to me interpolated. For immediately after having dealt with the letter an, taken to
express existential perdurance in general, the text as we have it describes in a short
paragraph the letter matā, ‘‘when’’. And immediately afterwards, it devotes some two pages
to the categories, including the ‘‘when’’. From the outset, such a disposition strikes me as
arbitrary and redundant. It is arbitrary because ‘‘when’’ is the only category dealt with
before the passage about the categories. It is redundant because ‘‘when’’ will appear also at
the beginning of the treatment of the categories. Moreover, this paragraph on ‘‘when’’
breaks the natural progression of the text. It would have been perfectly understandable if
the author had dealt first with being in general and, immediately afterwards, with
categorial being. Al-Fārābı̄ is not Heidegger, there is no reason for him to distinguish the
‘‘when’’ from all other secondary categories. My suspicion of interpolation is strengthened
by some problematic elements in the sentence immediately following (I give first a litteral
translation of the Arabic such as it appears in the single manuscript used by Muhsin
Mahdi): ‘‘Its case is not similar to that of place, since the kinds of bodies necessarily
require the places in / about the things he has enumerated previously’’. It seems to me that
the transmitted Arabic cannot be construed. Three problems arise. First, the meaning of the
preposition fı̄ is not clear, whatever be the sense retained ( in or about, i.e. in the case of). A
second di$culty, the subject of the verb ah **sā, ‘‘he (?) has enumerated’’. Since it cannot be
the ‘‘intellect’’ mentioned before, who has ‘‘enumerated’’ something? Aristotle? Al-Fārābı̄?
Someone else? Thirdly, the reference of the words min qabl, ‘‘previously’’ is puzzling. In the
Book of Letters such as we have it, there is only one paragraph before the text we are
dealing with, and it makes no mention of the question of place or of things in place. The
only way of emending this sentence, then, is to add something like ‘‘as he / we said’’
immediately before ‘‘in / about the things’’ and perhaps to change allatı̄ ah **sāhā (‘‘which he
has enumerated’’) into allatı̄ ah **saynāhā (‘‘which we have enumerated’’, cf. A z rā’ ahl
al-madı̄na al-fād * ila, chap. 11, p. 51: wa-hād I ihi al-mawǧūdāt allatı̄ ah
**saynāhā) or allatı̄
ih
**sā’uhā, (‘‘the enumeration of which <is>’’) so as to allow the subject (either grammatical
or ad sensum) to be al-Fārābı̄. Both corrections are paleographically slight: ‘‘Its case is not
similar to that of place, since the kinds of bodies necessarily require the places, <as we
have said> in the case of the things that we have enumerated previously’’. In both cases
(subject Aristotle or al-Fārābı̄), the reference to the previous passage makes no sense in the
Book of Letters.
46 MARWAN RASHED
is coextensive with the existence of the thing, but that time is not
necessary in order for the thing to exist. It is supervenient upon
motion and, above all, it is only for our intellect that it is so.
Al-Fārābı̄, or Aristotle according to al-Fārābı̄, would have opposed,
to the objective existence of place, the subjective character of
being-in-time. In confirmation, we may note that against Alexander,
and in agreement with Themistius, al-Fārābı̄ considered that the
celestial sphere itself has a place.77
If, however, there were no time, but only motion, without an
intellect to ‘‘count’’ motion,78 how could we refute Philoponus?
Suppose that we were to cut o# a temporal slice of cosmic history, a
year for example, or a day. Physical motions will have a beginning
and an end, exactly like in a Christian cosmology. Even if we agree
with Aristotle that there is no first smallest interval of motion (for
their can always be some smaller interval), there will be a time
point after which motion takes place. There is no contradiction in
considering exactly a year, or exactly a day. The contradiction, if
any, a#ects dynamics (the cause of the first motion), not kinematics
(the description of the first motion). And dynamics, as already
said, reflects only our presuppositions about nature, the thesis,
for example, that there is no divine miracle taking place in the
world.79
77
Cf. Averroes, In Phys. 141K#. (for al-Fārābı̄: 142B–C). For a discussion of this problem,
see my ‘‘Alexandre d’Aphrodise et la Magna Quaestio. Rôle et indépendance des scholies
dans la tradition byzantine du corpus aristotélicien’’, Les Études classiques, 63 (1995): 295–
351 [now reprinted in M. Rashed, L’héritage aristotélicien. Textes inédits de l’Antiquité
(Paris, 2007), pp. 85–141].
78
The idea of interpreting place and time in terms of being coextensive (int*ibāq) is
interesting. True, we find the word a century later, in the treatise On Place of al-H * asan ibn
al-HaytIam (cf. R. Rashed, Les mathématiques infinitésimales du IXe au XIe siècle, vol. 4: Ibn
al-Haytham, Méthodes géométriques, transformations ponctuelles et philosophie des
mathématiques [London, 2002], pp. 666–85, in part. pp. 675–7). It is however also used by
al-Fārābı̄, with a di#erent connotation. Cf. Book of Letters §39, p. 83: ‘‘ ‘when’ is posterior
to ‘where’, for the cause of the existence of time is that the body is translated in some
‘where’, so that time does occur then. Time is coextensive with ( yant*abiqu) the thing and is
related to it because of its being coextensive with its existence. This relation is similar to
that one, – I mean to the thing’s relation to its place’’. This passage is translated into
French by Ph. Vallat, Farabi et l’École d’Alexandrie. Des prémisses de la connaissance à la
philosophie politique (Paris, 2004), pp. 374–5, who is certainly correct in emending nisbata
into sababa and yanfa‘ilu into yantaqilu. However, his reading (p. 374, n. 7) yuh * datIa (‘‘is
produced’’) instead of yah * dutIa (‘‘does occur’’) is questionable, since in the presentation of
the same doctrine, al-Fārābı̄ describes time as being ‘‘supervenient upon motion’’ (H * urūf
62.6–7: ‘ārid* un . . . ‘an al-h
* araka, cf. Falsafat Arist*ūt*ālı̄s 96.2–3) – and obviously, the primary
function of modal distinction is to allot some sort of ontological status to supervenience
and / or concomitance apart from e$ciency. Perhaps more importantly, his discussion (ibid.)
and translation of yant*abiqu by ‘‘être approprié’’ or ‘‘être placé au même degré d’être’’ are
philosophically misleading. Al-Fārābı̄ surely has in mind the coincidence of two extended
magnitudes.
79
Peter Adamson draws my attention on a similar thought experiment in the text On
Metaphysics ascribed to al-Rāzı̄. See Abū Bakr al-Rāzı̄, Rasā’il falsafiyya, ed. P. Kraus
(Cairo, 1939), pp. 128–9.
AL-FA
z RA
(b) A possible solution
We have just spoken of a ‘‘temporal slice of cosmic history’’. As a
model for a Christian world, this image seems prima facie adequate.
But this is an illusion. For it does not take into account the peculiar
character of the first instant of the world, that is of creation properly
understood. Something more than simple ‘‘being’’ happens at the
first instant: namely, an instantaneous leap from not-being into
being, for which God alone can be responsible. The first instant of
being is not only a first instant of being, it is an instantaneous
transformation of non-being into being.
Thus, I presume that al-Fārābı̄ presented his creationist adversar-
ies with the following dilemma: in the temporal segment OM, where
nothing temporally precedes the point O, you must either say that O
belongs to the continuum OM like any other temporal point of it or
that it has some special status, due to the fact that some transition
from non-being to being individualizes it (this last thesis, as we shall
see, constitutes the tacit model of al-Fārābı̄’s adversaries, and in
particular of al-Kindı̄’s doctrine of creation). The first assumption
means that O is only potential, like any temporal point: the ˜ ,
according to Aristotelianism, is like a point in a continuous motion:
it is simple, not made double by some state of rest.80 If the temporal
point O is to be distinguished from the other points of the continuum
OM, it is because there is some kind of break occurring at it. But this
is contrary to the hypothesis that there was nothing before it.
Therefore, we must consider the existence of O as merely potential,
according to Physics VIII 8, 263a 11–29 and Aristotle’s admitting
80
That al-Fārābı̄ accepted this conception of motion is testified by a quotation in Ibn
Bāǧǧa (ed. M. Ziyādeh, S { urūh
* āt al-Samā‘ al-t*abı̄‘ı̄ li-Ibn Bāǧǧah al-Andalusı̄ [Beirut, 1978],
p. 226): ‘‘If we suppose some distance AB equal to the distance BC and we suppose D
moving from A and E moving from C with motions of equal velocity, they will arrive
simultaneously at B. Let us suppose that D returns to A and that E moves continuously.
Then, E will reach A before D reaching A. For it is plain that when D reached B, it was at
B during some time, equal to the time by which E precedes D. For without rest, there would
be no reason for D to reach the end after E, since the distance is the same and since the
motions are of equal velocity. It is clear that the mobile which came back uses the point
<B> as a beginning and as an end, at two instants, whereas the one which moved
continuously was at B in a single instant. And that gives Aristotle the solution to Zeno’s
aporia known as the dichotomy from the point of view of the thing itself, not in a verbal
way. And he declared that what he has said in the sixth Book was only verbal and not from
the point of view of the thing itself. And that is what Abū Nas*r made use of in his Book On
the Demonstrative Discourse. For Aristotle has admitted in the sixth Book only that there is
an infinite number of divisions in the continuum, as well as in the time. For he divides the
finite inasmuch as it is infinite in a finite <time>, but inasmuch as it is infinite. However,
the division of the infinite is impossible altogether. And he admits in the finite line, in an
absolute way, an infinite number of halves. It is for this reason that the aporia was
dismissed; but when he considered the thing as it is in itself, he proved that there are
absolutely no divisions in the continuum, neither finite in number nor infinite, for when it
possesses divisions, then it is not a single continuum, but numerous continua. Divisions are
in it only potentially, and they are necessarily infinite in number, according to what he
proved in the sixth Book, i.e. that the continuum is what is divided into a continuum’’.
48 MARWAN RASHED
there that his former answer (i.e. in the sixth Book) was not entirely
adequate: there is no ‘‘creation’’ at O, only existence, as at every
‘‘normal’’ point of OM. The only escape would be to posit some kind
of temporal atomism, according to which every instant is distinct
from every other, and may be said to correspond to some divine
‘‘pulsation’’, really subsistent on its own (i.e. not potentially embed-
ded in some continuum).81
Needless to say, the question of God’s acting on the world has
already had a long and rich history by the time of al-Fārābı̄. To
mention only some (eternalist) Platonists, Proclus is in deep dis-
agreement with how Plotinus seems to equate eternity and being as
such. According to him, a hypostazied eternity (’´ ) must be
located at a higher place than the intelligible realm and is only
participated by the eternal Ideas.82 The Plotiniana Arabica are in
agreement with this Greek tradition in its decision to locate the first
principles – the One according to Plotinus, certain transcendent
principles according to Proclus – beyond eternity.83 The Arabic
Plotinus holds two significant and mutually connected thesis with
regard to the creation of the world. First, the Creator is beyond
eternity and stands to it as a cause to its e#ect;84 secondly, creation
is something which happens ‘‘all at once’’ and ‘‘in no time’’ (duf‘atan
wāh* idatan, bi-lā-zamān).85 The two theses are not independent from
one another because the Creator’s being beyond time prevents His
act of creation from needing some period of time in order to be
fulfilled. It is much more natural, in this case, to hold that God’s act
of creation takes place in no time and all at once. There is a
necessary gap between God’s atemporality and the world’s extended
81
Accordingly, the ‘‘atom’’ of time will be defined as the basic component of God’s acting
upon the world. The world is nothing but the series of all the ‘‘pulsations’’ of God’s creative
activity. There is no basic di#erence between the world’s duration and a continued creation,
occurring at instants so near to one another that they produce, if not continuity, at least
the cinematographic illusion of it. The most famous (and disputed) example of such a
doctrine is that of Descartes. For such an interpretation, see in particular J. Wahl, Du rôle
de l’instant dans la philosophie de Descartes (Paris, 1920) and M. Gueroult, Descartes selon
l’ordre des raisons, 2 vols. (Paris, 1968), vol. 1, pp. 280–1 in particular. For a long reply to
this view, see J.-M. Beyssade, La philosophie première de Descartes (Paris, 1979).
82
On this debate, see C. d’Ancona Costa, ‘‘Esse quod est supra eternitatem. La cause
première, l’être et l’éternité dans le Liber de causis et dans ses sources’’, in Recherches sur le
Liber de causis (Paris, 1995), pp. 53–72, in part. pp. 55–63.
83
For more details, see d’Ancona Costa, ibid., pp. 63 sqq.
84
See the edition of the text in ‘A. Badawi, Aflāt*ūn ‘inda al-‘Arab, 3rd edition (Kuwayt,
1977), p. 114 (‘‘you must remove from your imagination every generation in time if you want
to know how the true, abiding, noble beings were originated from the first originator,
because they are only generated from Him atemporally, and between the origination and
the originator, and the making and the maker, there is no intermediary at all’’) and p. 130
(‘‘the first cause is standing still, resting in itself, and is not in eternity or time or in place,
but rather eternity and time and place and the rest of the things are only supported and
fixed through it’’). I borrow both English translations from P. Adamson, The Arabic
Plotinus. A Philosophical Study of the Theology of Aristotle (London, 2002), p. 143.
85
Cf. Badawi, Aflāt*ūn ‘inda al-‘Arab, pp. 31, 41, 70, 114.
AL-FA
z RA
eternity, which seems more readily preserved by supposing an
instantaneous creation than by submitting God to the world’s way of
being in time (i.e. to eternity).
Al-Kindı̄, in a passage of his epistle On the Quantity of Aristotle’s
Books very likely known to al-Fārābı̄, had underlined the fact that
divine creation takes no time. Only ‘‘unbelievers’’ (kāfirūna) may be
tempted to assign a period of time to it. Let us translate this
important text:
Then <Aristotle> said – in reason of the negation of the creation of the
heavens which is rooted in the heart of the unbelievers, because of what
they believe about the period of time pertaining to their creation in reason
of the analogy they draw with human acts, since according to them, the
most considerable work needs the longest period of human work, so that
according to them, the most considerable sensible realization takes place in
the longest time – that God, may He be praised, does not need a period of
time for His creation, in reason of what he made clear, since he established
‘‘it’’ out of ‘‘not it’’; so that the one whose ability reached such a point as
to produce bodies out of no bodies and to extract being out of not-being, he
does not need, since he has the power of producing out of no matter, to
produce in time. For since the human act is impossible without matter, the
act of the one who does not need matter in order to produce what he
produces does not need time.86
It is clear that al-Kindı̄ here rejects the idea that the divine act of
creation is some continuous process. Divine creation must be under-
stood as something temporally unextended, exactly like a geometrical
point. An instantaneous creation is the direct consequence of God’s
instantaneous volition, associated with His omnipotence. To deny this
philosophical creed amounts to no less than siding with the unbeliev-
ers. Unfortunately, we do not know how al-Kindı̄ managed to recon-
cile this doctrine with the Aristotelian claim of time’s continuity,
ultimately grounded on the isomorphy of time, length and motion.87
According to al-Kindı̄’s typology, many Aristotelian philosophers
are certainly ‘‘unbelievers’’. There is a strong tendency, exemplified
by Ammonius, but already attested in Aristotle, to interpret God’s
behaviour as some sort of action on the world.88 It is probable that
al-Fārābı̄ accepted this label. But what is action? Here again, we are
86
Rasā’il al-Kindı̄ al-falsafiyya, ed. M. A. Abū Rı̄da (Cairo, 1950), vol. I, p. 375. On this
epistle, see J. Jolivet, ‘‘L’épı̂tre Sur la quantité des livres d’Aristote, par al-Kindı̄ (une
lecture)’’, in R. Morelon and A. Hasnawi (eds.), De Zénon d’Élée à Poincaré, Recueil
d’études en hommage à Roshdi Rashed (Leuven / Paris, 2004), pp. 665–83 (for our passage,
see pp. 674–6). See also P. Adamson, Al-Kindı̄ (Oxford, 2007), chapter 3, pp. 63–4 in
particular.
87
On al-Kindı̄ on time, see J. Jolivet, ‘‘Al-Kindı̄, vues sur le temps’’, Arabic Sciences and
Philosophy, 3, (1993): 55–75.
88
See On generation and corruption, I 6, esp. 323a 11–34. For a summary of Ammonius
views, see K. Verrycken, ‘‘The Metaphysics of Ammonius Son of Hermeias’’, in R. Sorabji,
Aristotle transformed. The Ancient Commentators and Their Influence (London, 1990),
pp. 199–231.
50 MARWAN RASHED
confronted with a discussion engaged by Iamblichean neoplatonism

against Plotinus. Whereas Plotinus took action and passion to
belong to the general category of motion and assimilated motion to
some punctual activity, his adversaries did their best to maintain
the separation between extended and as such partially unrealized
motion and pure activity, which is an instantaneous and perfect
state.89 At an exegetical level, this discussion is rooted in the
di$culties, already faced by Alexander, of reconciling two seemingly
opposite statements of Aristotle’s Physics: whereas at Physics I 3,
186a 11–16, Aristotle charges Melissus for not having recognized the
possibility of a change ‘‘all at once’’, he emphatically asserts in Book
VI (see esp. VI 4, 234b 10–20) that every change must be progressive.
Even if the sources are not as clear as we may wish, it seems that
Alexander sided with Physics VI, i.e. held every motion to be
continuous. This at least is the thesis we find attributed to him by
some anonymous commentators mentioned by Averroes,90 and this
would allow us to understand better Iamblichus’ strategy in Simpli-
cius’ commentary on the Categories.91 For the idea expressed here
that even when some change takes place all at once, like the
illumination of some space, it is actually produced by some motion
(the rising of the sun),92 is more or less the same as Alexander’s
compromise. It is very probable, then, that in order to reply to
Plotinus, his neoplatonist successors made use of some Peripatetic
distinctions.
That continuity was a crucial feature of motion also for al-Fārābı̄
is attested by an unparalleled testimony in Averroes’ commentary on
Physics VIII 3, where we are told that al-Fārābı̄ insisted on it at the
beginning of his lost book:93
And since things are as we have said, it is conspicuous that what al-Fārābı̄
has a$rmed at the beginning of his book On Changing Beings, i.e. that in
the definition of motion in general appears continuity, is not true.94
Unfortunately, Averroes does not spell out the reason why al-Fārābı̄
wanted to make explicit this relation between motion and continuity.
It would have been all the more interesting, since we find no hint in
this direction in Aristotle’s first chapter. It is possible that al-Fārābı̄
89
See below, n. 92.
90
See Averroes, In Phys. 265L-M: ‘‘et ideo expositores . . . dicunt quod Alexander exponit
quod omnis transmutatio est in tempore, sed quaedam latet sensum’’. On this problem, see
also R.W. Sharples, Theophrastus of Eresus. Sources for his Life, Writings, Thought and
Influence, Commentary on Volume 3.1, Sources on Physics (Texts 137–223) (Leiden / Boston /
Köln, 1998), pp. 79–82.
91
See Simplicius, In Cat. 299.1 sqq.
92
See in part. 308.29–32.
93
Averroes, In Phys. 360E.
94
Et cum ita sit sicut narrauimus, bene apparet quod illud quod dixit Alfarabius in
principio sui libri De entibus transmutabilibus, scilicet quod de definitione motus in
uniuersali apparet continuatio, est sermo non verus.
AL-FA
z RA
took the reference to Empedocles’ alternating phases of motion and
rest as an implicit target, Aristotle’s idea being a contrario that it is
impossible for motion to be interrupted by periods, or even instants,
of rest. If rest does occur, then some other motion is required in order
to bring about another motion. In other words, the sequence motion–
rest–motion implies at least three (di#erent) motions: the two under
consideration, and a third one to initiate the second.
I confess that the texts are not as explicit as one might wish.
However, I think that there is at least one text in the genuine corpus
which testifies that al-Fārābı̄, in the Changing Beings, was implicitly
criticizing al-Kindı̄’s creationism. It is Averroes again, in the sen-
tence immediately following the one we have just translated, who
gives us the decisive clue. After criticizing al-Fārābı̄ for his assump-
tion that continuity primarily belongs to motion,95 he adds:96
And similarly what al-Fārābı̄ says in the Categories about the definition of
action and passion is not true. For it is plain from the definition put forward
there that every action and passion is continuous [. . .].97
Let us open al-Fārābı̄’s Kitāb al-Maqūlāt. The stress laid by him on
the notion of continuity appears already in the opening words of the
section devoted to passion. Al-Fārābı̄ writes that ‘‘passion is the
passage of the substance from an item to an item and its change from
a thing to a thing; and what proceeds between the two things
according to a continuous path (‘alā al-ittis*āl), is subject to pas-
sion’’.98 In the course of his discussion, al-Fārābı̄ insists heavily on
the same idea. Let us translate this important text:
And this process can take place from a quality to a quality, as in the case of
the body’s passage from blackness to whiteness, which is a whitening, and
its passage from coldness to heatness, which is a heating. When it undergoes
passion, what was in it at the beginning withdraws from it little by little,
and what it is proceeding to takes place in it, little by little and part by part,
according to a continuous path, until the process breaks o# and stops. And
at each time of its passion, the body is at some indeterminate part of what
takes place in it and at some indeterminate part of what withdraws from it.
For what is heated proceeds to heat little by little, according to a
continuous path, one part of heat after another, whereas the parts of
coldness withdraw from it one after another. However, we cannot deter-
mine, as long as the process persists, which part of heat took place in the
body, neither what amount of them took place in it, nor which part of
coldness ceased neither which amount of them. For every time you wish to
determine a part of heat that took place in it or a part of coldness that
95
See above, p. 50.
96
Averroes, In Phys. 360F.
97
Et similiter illud quod dicit Alfarabius in Praedicamentis de definitione eius quos est
agere et pati non est verum. Ex illa enim definitione apparet quod omnis actio et passio est
continua . . .
98
Al-Fārābı̄, Kitāb al-Maqūlāt, ed. R. al-‘Aǧam (Beirut, 1985), p. 113.
52 MARWAN RASHED
ceased, or some amount of them, you find that it is no more at this part nor
at this amount. And this happens until the process reaches its end and stops.
Only then can you determine which part took place and what amount what
was produced in it.99
Al-Fārābı̄ could not be more explicit. Every passion is a continuous
process. Even if the subject, at each instant of the process, is in some
state, the nature of a continuous process makes it impossible to
determine its state. The same, of course, must be said for action.
Immediately after these lines, al-Fārābı̄ says that ‘‘there is no
di#erence between our saying ‘it is subject to passion’, ‘it changes’
and ‘it moves’ ’’.100 That leads him to a$rm that the species of
passion are identical with the species of motion as Aristotle spells
them out in the Categories (chap. 14, 15a 13–14):
The species of this genus are the species of motion, i.e. generation,
corruption, augmentation, diminution, alteration and translation.
Generation is the passage from non-body to the production of body, or
from non-substance to the production of substance’’. Corruption is the
passage from body to the production of non-body, or from substance to
the production of non-substance, like the generation and construction of
the house little by little, thing after thing and part by part according to a
continuous path, until the house is realized [. . .].101
There is not the slightest allusion, in the Categories, to the continuity
of passion. But the example of the ‘‘house’’ reveals what al-Fārābı̄
has in mind: the doctrine of Physics VI 6, where Aristotle a$rms that
all changes – which are to be identified with the motions of the
Categories, cf. Physics III 1, 200b 33–34 – must be continuous,
illustrating this by the example of the house.102
Al-Fārābı̄ thus adopts a theory of his own, which implies a
reorganization of the Aristotelian table of the categories. Superfi-
cially, he may seem to mirror the ambiguity, in Arabic philosophical
terminology, between al-fi‘l in the sense of ’ ´ (bi-al
fi‘l = ’ ´◊ ) and al-fi‘l in the sense of
˜ (al-fi‘l wa-al-
infi‘āl = `
˜ `
´ ), and even to side with Plotinus in the
old debate of the commentators.103 But that is not the whole story: for
against Plotinus and with Alexander (and neoplatonic authors who
followed him), al-Fārābı̄ insists on the fact that every motion is in
fact continuous. Concerning the terminological question, al-Fārābı̄
is second to nobody in noting possible discrepancies between usual
99
Ibid., pp. 113–14.
100
Ibid., p. 114.
101
Ibid., p. 114.
102
See R. Sorabji, Time, Creation and the Continuum (London, 1983), p. 409#.
103
See R. Chiaradonna, Sostanza movimento analogia. Plotino critico di Aristotele (Napoli,
2002), pp. 147–225 and R. Sorabji, The Philosophy of the Commentators 200–600 AD, A
Sourcebook (London, 2004), vol. 3, Logic and Metaphysics, pp. 98–102 (with further
bibliography).
AL-FA
z RA
language – even usual translators’ language – and philosophical
significations.
I think that if al-Fārābı̄ decided to reorganize Aristotle’s cat-
egories and insisted on the fact that continuity should be considered
as the prominent feature of action and passion, it is probably because
he was willing to rule out the possibility of an action all at once. In
other words, al-Fārābı̄ aimed at purifying Aristotelianism from a
heterogeneous concept introduced in order to explain the existence
of the world: that of divine creation, or ibdā‘. We have already seen
its central position in the text of al-Kindı̄ translated above. This is
confirmed and explained in al-Imtā‘ wa-al-Mu’ānasa, where Abū
Sulaymān al-Siǧistānı̄ is asked about the number of species of
‘‘motion’’.104 He first answers by listing, like Aristotle and al-Fārābı̄,
the six items of the Categories: translation, generation, corruption,
augmentation, diminution, alteration. He then adds:
Al-Kindı̄ said: and there is here another motion, i.e. the motion of creation
(h
* arakatu al-ibdā‘ ). Note that there is between it and the motion of
generation a di#erence, because creation does not take place out of a
substrate, whereas the motion of generation occurs by the corruption of
some substance which appeared before it. For this reason, it was said that
generation is the passage from some low state to some high state.105
Two features of al-Kindı̄’s description, when compared with al-
Fārābı̄’s text, are striking. The first one, of course, is that he felt the
need to add a new kind of motion to the Aristotelian list.106 We know
from the text of the Epistle On the Quantity of Aristotle’s Books that
God’s first creative act happened all at once in no time, which
contradicts the main attribute of motion according to al-Fārābı̄.
The second feature is that when describing generation, al-Kindı̄
insists on the fact that a substance must come out of a substance,
whereas al-Fārābı̄ a$rms that ‘‘generation is the passage . . . from
non-substance to substance’’. This necessity, felt by al-Kindı̄, of
distinguishing two types of generation was already felt in Patristic
circles of late Antiquity. I have thus drawn attention in a previous
paper to a passage in Pseudo-Dionysius taken over by Maximus
Confessor where real ´ (where something is created out of
nothing) is opposed to Aristotelian elementary transformation
(
´ ), where for example fire comes to be out of air.107 Al-Fārābı̄’s
104
See Abū H * ayyān al-Tawh
* ı̄dı̄, al-Imtā‘ wa-al-mu’ānasa, ed. A. Amı̄n and A. Al-Zayn
(Cairo, n. d.), vol. 3, p. 133.
105
Ibid.
106
See also al-Kindı̄, Epistle on Definitions, p. 190 al-A‘sam: ‘‘the creation is the
manifestation of the thing out of non-being’’ ( al-ibdā‘u huwa iz*hāru al-šay’i ‘an laysa).
107
See M. Rashed, ‘‘La classification des lignes simples selon Proclus et sa transmission
au monde islamique’’, in C. d’Ancona Costa and G. Serra (eds.), Aristotele e Alessandro di
Afrodisia nella tradizione araba (Padova, 2002), pp. 257–79, now in L’héritage aristotélicien,
pp. 303–25, pp. 317–20.
54 MARWAN RASHED
Aristotelian puritanism may well be then a response to such ideas

upheld by al-Kindı̄. We do not have to posit another type of motion.
Al-Fārābı̄’s God, unlike al-Kindı̄’s, falls under the jurisdiction of the
Aristotelian continuum.108
CONCLUSION
In conclusion, I would like to return to the passage from

Maimonides’ Guide of the Perplexed. The Moderns, he tells us, have
adduced new proofs, from God’s nature, to the stock of physical
arguments explicitly recognized by Aristotle. It seems to me that this
remark, important as it is, misrepresents the whole movement of
al-Fārābı̄’s philosophy. The goal was not to pile up arguments, but to
understand the twofold articulation of a unique demonstration: first
‘‘upwards’’ from the e#ect to the cause, and then ‘‘downwards’’ from
the cause to the e#ect. Only thus, is it possible to reassess, against all
the criticisms, Aristotle’s claim of the eternity of the world. And the
interpretation of Physics VIII 1 as ascending from partial motions to
the eternal motion of the celestial sphere is part and parcel of this
general progression. But at another level – this one being perhaps
more speculative –, I think that there was a physical proof of eternity
according to al-Fārābı̄, i.e. of the simple fact that motion is eternal.
This proof, specifically directed against al-Kindı̄, was based on the
fact that every action is a continuous motion, i.e. that the idea of an
instantaneous creation of the world is self-contradictory. There is no
clear evidence that this physical proof was put forward in the
treatise On Changing Beings. But it was, so to say, latent in the
108
Let us note that this kind of considerations on the continuum is perhaps not
altogether absent from the ‘‘synthetic’’ demonstration of the eternity of the world. Since
creation seems to be an action of some sort and action, according to al-Fārābı̄, is to be
assimilated to the bestowal of motion, which is always continuous, we cannot conceive of
divine creation as an action in the usual sense. For it is impossible for divine creation to
take place in a finite interval of time, because its length would be arbitrary, therefore
contradictory to God’s omnipotence. Indeed, for each time-interval we wish to assign to the
act of creation, it is legitimate to think that God could have achieved exactly the same
result in half the time. And so on ad infinitum. Nor can we suppose that God has created
something in a very small time interval, since this interval would be ‘‘very small’’ only for
us. As Leibniz accurately notes, it is as if God had wished to hide beyond the threshold of
human perception the constraint imposed on Him to create everything in some period of
time (see G.W. Leibniz, Pacidius Philalethi [in Philosophische Schriften VI.3 (Berlin, 1980)],
p. 560, ll. 21–24: ‘‘. . . perinde esset ac si Deus incongruitates quasdam, quas in natura
scilicet evitare non poterat, tegere tantum nobis ac dissimulare voluisset, transferendo
scilicet illas in minutiora rerum, ubi animadverti non possint’’). Therefore, since an instant
is per se nothing, emanation must be punctually eternal (or eternally punctual). Emanation
implies eternity, it is not conditioned by it. Eternity is not something already given when
emanation comes to the fore. Even if this continuity somehow pluralizes divine unity, it is
an analytic condition of God’s e#usion out of Himself. Averroes was perfectly right in
suggesting a close relation between the beginning of the treatise On Changing Beings and
the definition of action and passion in al-Fārābı̄’s Categories.
AL-FA
z RA
refutation of Philoponus’ attacks against the Aristotelian notion of
an instant as a mean between the past and the future.
Appendix: the Kitāb al-G

{ am‘ and the 9th Question
In the Kitāb al-G

{ am‘, al-Fārābı̄ seems to favour some kind of
temporal creationism. I translate the most representative
passage:
And what has led <the people according to whom Aristotle thinks that the
world is eternal> to this opinion is also what he has mentioned in the book
On the Heavens and the World, i.e. that the All does not have a temporal
beginning. According to that, they have the opinion that he upholds the
eternity of the world. But this is not so, since he has already shown, in this
book and in other physical and theological books as well, that time is
nothing else but the number of the celestial sphere’s motion and that it
proceeds from it. And what proceeds from a thing does not comprehend this
thing. The meaning of his saying that the world does not have a temporal
beginning is that the world is not generated little by little with its parts, in
the way plants or animals, for example, are generated. For what is
generated little by little with its parts is such that its parts are prior to one
another according to time. But time proceeds from the celestial sphere’s
motion, so that it is absurd to attribute to the latter’s own proceeding a
temporal beginning. Hence it is plain that <the celestial sphere> comes to
be by its creation ex nihilo from the part of the Creator – may He be
magnified –, occurring all at once, without time, and that it is from its
motion that time proceeds.109
The author explains that it is an error to ascribe to Aristotle the view
that the world is eternal. Aristotle has only shown, in the De caelo,
that its beginning occurs all at once. It is not a natural, continuous
process, but a timeless initiation. The underlying idea is of course
that the temporal length between today and the first instant of the
world is finite. It consists in a finite amount of time units.110 Such a
view is an expression of the di$culty we have stressed above: if time
proceeds from motion, it appears hopeless to try to settle the question
of the being-in-time of the world: if the world is eternal, so is time;
if the world is non-eternal, so is time. Nothing can allow us to
infer something about the eternity of the world from our mere
apprehension of time.
However, the solution of the G { am‘ deeply contradicts some
obviously genuine passages of al-Fārābı̄ on the question of eternity a
parte ante. We have seen above how he rebuked Galen in his
109
G{ am‘, §55, pp. 128–9 Najjar-Mallet.
110
It goes without saying that if this text adequately represents al-Fārābı̄’s view on the
question, the whole enterprise of the treatise On Changing Beings must be interpreted as
merely exegetical (exactly as if one of our contemporaries commented upon the same
passage of the Physics).
56 MARWAN RASHED
commentary on the Topics. My entire interpretation of the Treatise

On Changing Beings rested on the impossibility that any motion be
punctual. That was the crucial argument which, according to me, led
to the conclusion that the celestial motion is eternal. Given this
evidence, it seems impossible that al-Fārābı̄ would have implied,
even in an exoteric piece of writing, that the period of time during
which there have been some celestial bodies is finite and, above all,
that some action can occur ‘‘all at once in no time’’. But this is
exactly what the author of the G { am‘ does. Even if al-Fārābı̄ had tried
to present as harmonious the views of Plato and Aristotle on the
subject,111 he would have said, rather, that the creation of the world
is taking place perpetually, and not that divine creation, being
instantaneous, implies the non eternity of the world, which is an
obvious non sequitur.
The text of the G { am‘, however, is not the only one to ascribe to
al-Fārābı̄ the denial of the infinity of time. The answer to the ninth
Question seems also to entail temporal finitism. Let us translate this
text:
– He was asked about the world: is it subject to generation and corruption,
or not? And if it is subject to generation and corruption, are its generation
and corruption similar to the other bodies’ generation and corruption? Or is
it of some other kind, and if so how? – He said: generation, in fact, is a
composition of some sort or similar to composition; and the corruption is a
dissolution of some sort or similar to dissolution. And if one were to say,
instead of ‘‘composition’’ and ‘‘dissolution’’, ‘‘association’’ and ‘‘dissocia-
tion’’, this would be licit as well. For everything, then, whose composition
takes place out of more numerous parts, the time of its composition is
longer; and everything whose dissolution takes place out of more numerous
parts, the time of its dissolution is longer. And for each one of these
processes, when the parts of the thing are less numerous, the time required
by its composition and dissolution is shorter. The least number of things
which can be a#ected by composition and dissolution is two, since the thing
which is one is not subject to composition and dissolution. Moreover,
composition and dissolution can happen only in time; and time has a
beginning, which is the pure ‘‘now’’; and the beginning of the thing is
di#erent from the thing. Then, the composition and the dissolution which
happen to two things take place in the pure ‘‘now’’ and those which happen
to things more numerous than two take place in time. And the length and
shortness of this time is in accordance with the multitude and the paucity
of these things. The things in the world, such as animals, plants etc., are
always composed out of things more numerous than two. Therefore, their
generation and corruption due to the multitude which is in their parts and
elements take place in time. But the whole world, according to the truth, is
composed out of two things, its proper matter and form. Thus, its generation
takes place all at once without time, according to what we have shown, and
111
A project which he did not share, see M. Mahdi’s introduction in Alfarabi: Philosophy
of Plato and Aristotle, pp. 3–10.
AL-FA
z RA
similarly its corruption. It is clear, then, that everything that is subject to
generation must be subject to corruption. And we have shown that the
world in its entirety is subject to generation and corruption and its
generation and corruption do not take place in time. But the parts of the
world are subject to generation and corruption and their generation and
corruption take place in time. God, may He be praised and exalted, is the
one who has given the world its existence, He is the true One, Creator of the
universe, una#ected by generation and corruption.112
It seems very unlikely that this text is a faithful testimony of
al-Fārābı̄’s views on the eternity of the world. This is not just because
it is introduced by the verb qāla, ‘‘he said’’, which reflects a
redactor’s intervention.113 The whole argument seems un-Farabian:
its crypto-atomism, as well as its willingness to ascribe to the world
as a whole one form and one matter, are probably too scholastic to be
attributed to the Second Master.114 Moreover, the argument that the
time required by a process of composition / dissolution is propor-
tional to the number of parts is problematic, to say the least: for
which reason isn’t it possible for a plurality of associations linking
together the parts of some physical object to be created or destroyed
simultaneously? However, it is interesting to note that its sphere of
argumentation more or less coincides with the passage already
quoted from the G { am‘. In both cases, the author tries to explain that
the fact that the world is not generated like other beings does not
amount to saying that it is eternal, but only that its ‘‘punctual’’
generation bears some very peculiar features: for both authors, it
occurs ‘‘all at once’’, i.e. ‘‘not in time’’.115 The solution given in the
Question appears to be nothing but a (scholastic) elaboration on the
doctrine expressed in the G { am‘.
J. Lameer has supposed that the G { am‘ and the Questions were not
genuine.116 I cannot enter here into the many details of his argumen-
tation, which I find entirely convincing.117 I shall only stress that the
two cases are rather di#erent, according to Lameer. Whereas he
considers that the G { am‘ has nothing to do with al-Fārābı̄, he takes
the Questions to include a great deal of Farabian material compiled
by some later scholar or group of scholars. Our own reflections fit
pretty well with this reconstruction. In all likelihood, the compiler of
the answer to Question 9, by drawing from the G { am‘, thought that he
was excerpting an authentic work of al-Fārābı̄. In doing so, he sided
112
Al-Fārābı̄, G
{ awābāt li-Masā’il su’ila ‘anhā, ed. G
{. A
z l Yāsı̄n in: al-Fārābı̄, Al-A‘māl
al-falsafiyya (Beirut, 1992), pp. 313–69, pp. 317–19.
113
Cf. Lameer, Al-Fārābı̄ and Aristotelian Syllogistics, p. 26. This was also the conviction
of R. Walzer, ‘‘The rise of Islamic philosophy’’, Oriens, 3 (1950): 1–19, esp. p. 16.
114
It appears in the Liber de causis. Cf. ed. Badawı̄, §17, p. 19.
115
On the neoplatonizing antecedents, see Mallet-Najjar, L’harmonie, p. 177, n. 4.
116
Cf. Lameer, Al-Fārābı̄ and Aristotelian Syllogistics, pp. 25–39.
117
But see above, n. 110.
58 MARWAN RASHED
with one of the four ancient lists of al-Fārābı̄’s writings,118 with some
manuscripts still at our disposal (but not the most authoritative
witness from Diyār Bakr)119 and, above all, with Avicenna, who in
his Answers to al-Bı̄rūnı̄’s Questions unequivocally presents as
belonging to ‘‘Abū Nas*r al-Fārābı̄’s Kitāb al-G { am‘ bayna ra’yay
al-H
* akı̄mayn’’120 a thesis actually to be found in the existing
treatise.121
In conclusion, I must express serious doubts as to the authenticity
of the G{ am‘. If taken as genuine, its doctrine would contradict not
only what appears to have been the very subject of the Changing
Beings, but also two important passages from al-Fārābı̄’s logical
corpus. True, my position may seem weakened by the fact that I am
obliged to excise a line in the summary of the Sophistical Refutations
which a$rms (in a very Kindian way) the non eternity of the
world – but I think that apart from doctrinal considerations, the
close parallel in the Qiyās speaks in favour of that choice.
118
The title appears only in the list transmitted by al-Qift*ı̄, under the slightly di#erent
form (Fı̄ ittifāqi ārā’i Arist*ūt*ālı̄sa wa-Aflāt*ūn); in Ibn Abı̄ Us*aybi‘a, we find Fı̄ ittifāqi ārā’i
Abuqrāt*a wa-Aflāt*ūn (see Steinschneider, p. 133 and the list p. 217 [no 49]). The title is
absent from the list preserved in MS Esc. 884 (Casiri 879, ed. in Steinschneider, pp. 214–20)
and from the list of Ibn al-Murah h im I shall publish in the next issue of this Journal.
119
See Mallet-Najjar, L’harmonie, ˘ ˘ p. 45.
120
Cf. Al-Biruni and Ibn Sina, Al-As’ilah wa’l-Ajwibah (Questions and Answers ), ed.
S. H. Nasr and M. Mohaghegh (Tehran, 1974), p. 40, ll. 12–13.
121
Cf. G
{ am‘, §§35–41.
AL-FA
z RA
z BIz’S KITA
zB AL-H zF AND HIS ANALYSIS
* URU
OF THE SENSES OF BEING
STEPHEN MENN 1
Department of Philosophy, McGill University, 855 Sherbrooke St. W.,
Montreal, QC H3A 2T7, Canada
Email: stephen.menn@mcgill.ca
for Fritz Zimmermann
Abstract: Al-Fārābı̄, in the Kitāb al-H * urūf, is apparently the first person to
maintain that existence, in one of its senses, is a second-order concept [ma‘qūl
thānı̄]. As he interprets Metaphysics 7, ‘‘being’’ [mawjūd] has two meanings,
second-order ‘‘being as truth’’ (including existence as well as propositional truth),
and first-order ‘‘being as divided into the categories.’’ The paronymous form of the
Arabic word ‘‘mawjūd’’ suggests that things exist through some existence [wujūd]
distinct from their essences: for al-Kindı̄, God is such a wujūd of all things. Against
this, al-Fārābı̄ argues that existence as divided into the categories is real but
identical with the essence of the existing thing, and that existence as truth is
extrinsic to the essence but non-real (being merely the fact that some concept is
instantiated). The H * urūf tries to reconstruct the logical syntax of syncategorematic
or transcendental concepts such as being, which are often expressed in misleading
grammatical forms. Al-Fārābı̄ thinks that Greek more appropriately expressed many
such concepts, including being, by particles rather than nouns or verbs; he takes
Metaphysics to be discussing the meanings of such particles (comparable to the
logical constants of an ideal language), and he takes these concepts to demarcate
the domain of metaphysics. This explains how al-Fārābı̄’s title can mean both ‘‘Book
of Particles’’ and ‘‘Aristotle’s Metaphysics.’’
Résumé: Al-Fārābı̄, dans le Kitāb al-H

* urūf, semble être le premier philosophe à
soutenir que l’existence, en l’un de ses sens, est un concept de second ordre (ma‘qūl
thānı̄). Selon son interprétation de Métaphysique 7, ‘‘être’’ (mawjūd) a deux
significations, l’une, l’‘‘être comme vrai’’ (qui inclut à la fois existence et vérité
propositionnelle), de second ordre, l’autre, l’‘‘être en tant que divisé selon les
catégories’’, de premier ordre. La forme paronymique du terme arabe ‘‘mawjūd’’
suggère que les choses existent par une certaine existence (wujūd) distincte de leur
essence: pour al-Kindı̄, Dieu est ce wujūd de toutes les choses. A v rebours, al-Fārābı̄
soutient que l’existence en tant que divisée selon les catégories est réelle, mais
1
I would like to thank Peter Adamson, Amos Bertolacci, Thérèse Druart, Carlos
Fraenkel, Yoav Meyrav, Marwan Rashed, and Robert Wisnovsky for comments on earlier
versions of this material. Charles Butterworth gave helpful advice on the state of the text of
the Kitāb al-H* urūf. It was through conversations with Fritz Zimmermann, and through
reading his magnificent introduction to his Al-Farabi’s Commentary and Short Treatise on
Aristotle’s De Interpretatione (Oxford, 1981), that I first began to understand what Fārābı̄
might be about. (However, it will be clear that I want to connect Fārābı̄’s logic with his
metaphysics in a way Zimmermann does not; but this is because I am talking about the
H
* urūf, not about the De Interpretatione commentary.)
60 STEPHEN MENN
identique à l’essence de la chose existante, tandis que l’existence comme vrai, si elle
est extrinsèque à l’essence, est toutefois non-réelle (puisqu’elle revient simplement
au fait que quelque concept soit instancié). Le Kitāb al-H* urūf s’e#orce de reconstru-
ire la syntaxe logique des concepts syncatégorématiques ou transcendantaux tels
que l’être, souvent exprimés sous des formes grammaticales fourvoyantes. Al-Fārābı̄
considère que, de manière plus appropriée, le grec exprimait de tels concepts, y
compris l’être, au moyen de particules plutôt que de noms ou de verbes; il tient
Métaphysique pour une discussion de la signification de telles particules (com-
parables aux constantes logiques d’un langage idéal) et ces concepts pour
assurant la délimitation du domaine de la métaphysique. Cela explique qu’on puisse
entendre, sous le titre choisi par al-Fārābı̄, aussi bien ‘‘Livre des Particules’’ que
‘‘Métaphysique d’Aristote’’.
I. THE PROGRAM OF THE KITAzB AL-H

* URUzF
Almost forty years after its publication, Fārābı̄’s Kitāb al-H * urūf
remains an intriguing puzzle.2 It has not been integrated into general
accounts of the history of Arabic philosophy, let alone philosophy
more broadly, and while the work has been studied from various
aspects, its overall aims and achievement have not really been
clarified. I am writing a monograph which I hope will help situate the
H
* urūf within Fārābı̄’s philosophical work and in its broader context,
but I want here, abstracting from much scholarly detail about the
text and about its relations to Fārābı̄’s other works and to earlier and
later writers, to try to extract what seems to me to be one funda-
mental point that Fārābı̄ is making, which seems not to have been
mentioned in the scholarship, and which I think can give us a new
focus in reading the text. The H * urūf is an extraordinarily di$cult
work to classify – some of it is talking about the categories, some
about the origin of language, some about the relations between
philosophy and religion, some about the kinds of questions asked in
science and the other ‘‘syllogistic arts,’’ and Fārābı̄ never makes
explicit what these discussions have to do with each other – and
di#erent scholars have approached it starting from these di#erent
particular discussions. The majority of the scholarship has concen-
trated on the discussions of language and of philosophy and religion
in Part Two. Other scholars have concentrated on the accounts of
the categories and of the scope of logic in Part One, connecting them
with Fārābı̄’s writings related to the Organon, notably his com-
mentary on the De Interpretatione.3 But some of the central concerns
2
I cite the first and only edition, Abū Nas*r al-Fārābı̄, Kitāb al-H
* urūf, ed. Muhsin Mahdi
(Beirut, 1969; the copyright date is given as 1970; there is an unaltered 1990 reprint). I will
cite it by Mahdi’s part and paragraph numbers (e.g. ‘‘I,89’’), followed by page and line
numbers if I am quoting specific lines. I will cite Mahdi’s (Arabic) introduction by ‘‘Mahdi’’
and page number.
3
Besides some discussion in Zimmermann ( Al-Farabi’s Commentary and Short Treatise)
and in Philippe Vallat, Farabi et l’école d’Alexandrie (Paris, 2004), see in this direction
Shukri Abed, Aristotelian Logic and the Arabic Language in Alfārābı̄ (Albany, 1991) and
Stéphane Diebler, ‘‘Catégories, conversation et philosophie chez al-Fārābı̄,’’ in O. Bruun
FA
z RA
z BIz’S KITAzB AL-H
* URUzF 61
of the Kitāb al-H * urūf, although they might be called logical in a
broad sense, belong squarely to metaphysics: they deal, in particular,
with the concept of being, which Fārābı̄ in On the Aims of the
Metaphysics describes as the subject [mawd * ū‘] of first philosophy.
Indeed, I will try to show how Fārābı̄ might have understood the
H
* urūf as a whole as a contribution to metaphysics, and as restating
in a new context some of the central contributions of Aristotle’s
Metaphysics (and I will say a little about what the new context might
be, that would make such a restatement necessary); and so thinking
through Fārābı̄’s project in the H * urūf should also shed some light on
how he read the Metaphysics. Certainly Fārābı̄’s approach to meta-
physical topics in the H * urūf is heavily linguistic, beginning from the
terms used to signify metaphysical concepts, and distinguishing their
di#erent meanings. But he has good reasons for adopting this
approach in order to overcome what he sees as deep-seated meta-
physical confusions, and to disentangle the logical syntax of the
concepts. It is here, and in particular with the logical syntax of the
concept of being, that the H * urūf makes what seems to me its most
distinctive contribution.
Fārābı̄ is apparently the first person ever to have said that the
concept of existence, at least in one basic sense of existence, is a
second-order concept, that is, a concept that applies to concepts
rather than directly to objects – an insight most often credited to
Frege.4 Furthermore, Fārābı̄, like Frege and Russell, works out this
idea in the context of imagining a logically ideal language, i.e. a
language in which the grammatical form of our expressions would
make perspicuous the logical form of the judgments they express, or
at least in imagining a language that would correct discrepancies
between grammatical and logical form that arise in the particular
natural language in which he is writing. Still further: Fārābı̄’s chief
interest is not in the ordinary subject- and predicate-terms (what
Frege would call object-words and concept-words) of a logically ideal
language, but rather in what we would now call the logical constants
and L. Corti (eds.), Les catégories et leur histoire (Paris, 2005), pp. 275–305; see now also
Thérèse-Anne Druart, ‘‘Al-Fārābı̄, the Categories, Metaphysics, and the Book of Letters,’’
forthcoming in Medioevo, 33 (2007).
4
Nicholas Rescher many years ago (‘‘A ninth-century Arabic logician on: Is existence a
predicate?’’, Journal of the History of Ideas, 21 [1960]: 428–30) called attention to Fārābı̄’s
interest in the logical structure of assertions of existence. Unfortunately, this was before
the Kitāb al-H
* urūf was published, and Rescher had at his disposal only the very abbreviated
and not very deep discussion in Fārābı̄’s Risāla fı̄ jawāb masā’il su’ila ‘anhā (ed. F.
Dieterici in Alfārābı̄’s Philosophische Abhandlungen [Leiden, 1890], pp. 84–103), #16 (p. 90),
which we can in retrospect see to be a quick summary of a point developed at length in the
H
* urūf. Also unfortunately Rescher followed Goichon in attributing the spurious Fus*ūs*
al-H* ikam, and thus an Avicennian essence-existence distinction, to Fārābı̄. Nonetheless,
Rescher deserves credit for scenting that there was something important going on on this
issue in Fārābı̄, and that it somehow grew out of Posterior Analytics II; on which more
below.
62 STEPHEN MENN
of the language, or in the concepts that would be signified in a

logically ideal language by such logical constants, including the
concept of existence. Indeed, Fārābı̄ seems to demarcate metaphysics
precisely as dealing with those concepts that would be signified in a
logically ideal language by logical constants. He seems to have been
the first person to have taken anything remotely like this approach
to metaphysics. Nonetheless, Fārābı̄ does not see himself as an
innovator, but as a transmitter of Greek scientific philosophy to the
Arabic-speaking Muslim world. He sees the investigation of logical
syntax, and of logical constants, as an Aristotelian program, and he
works out his account of existence, in particular, in trying to restate
what he thinks are basic insights of Aristotle’s Metaphysics that have
been overlooked by his own Arabic-speaking contemporaries; it is in
attempting to reconstruct the syntax of Aristotle’s Greek that he is
led to describe how di#erent kinds of judgments of being would be
expressed in an ideal language, or in something closer to the ideal
than Arabic. This is a very peculiar way of reading Aristotle’s
Metaphysics, but it is one that made sense in Fārābı̄’s context, and
that allowed him to make important philosophical progress. In my
monograph I will try to locate the H * urūf both within Fārābı̄’s larger
project and within the reception-history of the Metaphysics. Here I
want to say just enough about the overall structure of the H * urūf and
about its relationship to the Metaphysics to locate Fārābı̄’s account
of being within the particular project of the H * urūf; I will then turn to
his analysis of the senses of being, concentrating on the senses of
existence (that is, of 1-place being, expressed in sentences of the form
‘‘X is’’ rather than ‘‘S is P’’), showing how he reconstructs an
Aristotelian doctrine of being from (above all) Metaphysics 7 and
Posterior Analytics II, and what he thinks its main lessons are. In
particular, I will try to bring out the place of the second-order
concept of existence among the di#erent senses of being that Fārābı̄
recognizes, and the reasons why he thinks it was important for
Aristotle, and is even more important in Fārābı̄’s own time and place,
to distinguish such a second-order sense.
The idea that the Kitāb al-H * urūf stands in some special relation-
ship to Aristotle’s Metaphysics is certainly not new. Mahdi in his
(heroic) editio princeps added to the title ‘‘Alfarabi’s Book of
Letters’’ the subtitle ‘‘Commentary on Aristotle’s Metaphysics.’’ Now
the H* urūf is not in any straightforward sense a commentary on
anything, and Mahdi weakened his case by supporting it on an
argument from exclusion (the H * urūf devotes considerable energy to
the categories, Aristotle discusses the categories in the Categories
and the Metaphysics, the H * urūf is not a commentary on the
Categories, therefore it is a commentary on the Metaphysics, Mahdi
pp. 30–4). Nonetheless, given that Fārābı̄ himself refers to Aristotle’s
FA
z RA
* URUzF 63
Metaphysics, in On the Aims of the Metaphysics, as ‘‘al-kitāb al-
mawsūm bi-al-h * urūf,’’ the book known by letters (because of the
letter-names of the individual books), it seems very unlikely that he
himself would write a Kitāb al-H * urūf without intending to refer to
Aristotle’s treatise.5 And that there is at least some overlap in
contents with the Metaphysics (described in more detail below) is
obvious on inspection. While his text is not a commentary on the
Metaphysics, it is a reasonable conjecture that Fārābı̄ sees it as
restating what he takes to be the main scientific contributions of
Metaphysics, that his treatise will do for his Arabic-speaking Muslim
audience what Aristotle’s Metaphysics originally did for its original
Greek audience.6 I think this conjecture is in fact correct, but it has
to overcome some obvious di$culties. The most obvious di$culty is
that the three Parts [abwāb] into which Mahdi (following the major
breaks in the text) has divided the treatise do not at first sight seem
to have much to do either with each other or with the Metaphysics;
and, worse, the contents of two of these parts suggest an entirely
di#erent interpretation of the phrase ‘‘kitāb al-h * urūf.’’
I’ll try to set out the di$culty by sketching the topics of the
di#erent parts, starting with Part Two, which has so far received the
lion’s share of the attention. That part is an account of the rise of
demonstrative science within a given linguistic and religious com-
munity, placed in the context of a schematic history of the develop-
ment of all the arts in such a community, concentrating on what
Fārābı̄ calls the ‘‘syllogistic arts,’’ that is, arts whose exercise
depends essentially on reasoning, such as rhetoric, dialectic, sophis-
tic, and demonstrative science. Fārābı̄ is especially interested in the
language of these arts, and especially of demonstrative science. He
talks about the origin of language as such; but language as it
naturally arises is not well suited to being the vehicle of demonstra-
tive science, because it is chiefly devoted to naming and describing
the objects of immediate practical interest to human beings, which
are not the main objects of theoretical interest. But natural language
5
The passage from On the Aims of the Metaphysics is in Dieterici, Alfārābı̄’s
Philosophische Abhandlungen, p. 34. Mahdi pp. 34–7 discusses the di#erent witnesses to the
title, which show minor variations, and sometimes conflate the book with the Kitāb
al-Alfāz
*. The manuscript itself has no title at the beginning, and says at the end ‘‘this is the
end of Abū Nas*r al-Fārābı̄’s Risālat al-H
* urūf’’ (III,251, p. 226,21). Fārābı̄’s On the Aims is
itself in one manuscript given the same title, Risālat al-H * urūf (see Mahdi p. 36). Ibn
al-Nadı̄m’s Fihrist ([Cairo, 1929], p. 352) cites Aristotle’s Metaphysics as Kitāb al-H * urūf,
adding that it is ordered according to the order of the h * urūf (i.e. letters of the alphabet) of
the Greeks, and refers to the individual books as h * urūf.
6
By contrast with the On the Aims of the Metaphysics, a much shorter and much easier
book which is about Aristotle’s Metaphysics (as the Kitāb al-H * urūf is not), and much of
which does not resemble the Metaphysics at all – notably in containing very little argument,
little even of doctrinal statement, mostly just indications of topic. However, the first part of
On the Aims might be described as Fārābı̄’s reworking of a small part of the Metaphysics,
namely the meta-metaphysical chapter E1.
64 STEPHEN MENN
can be extended in two main ways to provide the terminology of the

syllogistic arts. First, terms of ordinary language can be extended
metaphorically to new meanings, and can then be frozen for use as
technical terms; and this process of metaphorical extension and
freezing can be repeated ad libitum. Second, new terms can be formed
from old ones by regular grammatical processes of morphological
derivation, and again, this type of extension can be repeated and
combined with the first type.
Besides discussing the development of their arts and their lan-
guage within a single community, Fārābı̄ also discusses the trans-
mission of the syllogistic arts, especially demonstrative science,
across linguistic and religious boundaries. Most of the scholarly
attention has gone to his discussion of transmission across religious
boundaries. Religion for Fārābı̄ is a kind of imitation or practical
enactment of a philosophy (either a demonstrative philosophy or
one resting on sub-demonstrative reasoning); and di$culties will
arise notably when demonstrative philosophy is transmitted into an
already constituted religious community, whose adherents may be
reluctant to admit that their religious claims as merely imitations of
conclusions of demonstrative philosophy, and so they may resist the
importation of philosophy. But Fārābı̄ is at least equally interested
in problems of transmission across linguistic boundaries. Suppose
you are translating a scientific text into a language that does not
already have a technical vocabulary. In translating the technical
terms of the source text, derived from the metaphorical extension
and technical freezing of words in the source-language, you will be
confronted with a choice whether to create a technical vocabulary in
the target-language by imitating the metaphors of the source-
language in the target-language, where they may be much less
natural, or whether instead to create a new technical vocabulary out
of metaphors which are more natural to the target language; either
way, there is a risk of misunderstanding. Fārābı̄ talks about these
problems in general, but also makes it clear that he is especially
concerned with the transmission of demonstrative philosophy and its
vocabulary from Greek into (Syriac and thus) Arabic.
Thus far Part Two. This does not sound much like anything in the
Metaphysics, although it might vaguely recall the account of the rise
of theoretical science in Metaphysics A1–2 (but it is not clear that
Fārābı̄ knew these chapters, which are missing in the extant Arabic
translations of the Metaphysics).7 Parts One and Three seem very
7
On the complicated history of the di#erent Arabic translations of di#erent books of the
Metaphysics, see Amos Bertolacci, ‘‘On the Arabic translations of Aristotle’s Metaphysics,’’
Arabic Sciences and Philosophy, 15 (2005): 241–75, and the studies Bertolacci cites
(including, on A, earlier work of Bertolacci and of Cecilia Martini). When I cite the
Metaphysics in Arabic, I will refer to the lemmata in Averroes, Tafsı̄r Mā Ba‘d at*-T
* abı̄‘at,
ed. M. Bouyges, 4 vols. (Beirut, 1938–52).
FA
z RA
* URUzF 65
di#erent from Part Two (although not so di#erent from each other).
These parts are organized around lists of (allegedly) equivocal terms:
in each case Fārābı̄ lists the di#erent meanings of the term, starting
with an ordinary-language meaning and then explains how it is
metaphorically extended to a series of further meanings, especially
technical meanings in the di#erent arts and especially in philosophy.
The terms treated in Part One include many of the names of the
categories, or interrogative particles from which names of categories
are derived, such as ‘‘kam’’ = ‘‘how much?’’ and ‘‘kayfa’’ = ‘‘how?’’
or ‘‘qualis?’’; he also discusses such terms as ‘‘being,’’ ‘‘essence,’’
‘‘accident,’’ ‘‘from,’’ and ‘‘because.’’ Part Three applies the same
method to consider the di#erent kinds of scientific question or
investigation described in Posterior Analytics II, whether it is, what
it is, that it is, why it is; in each case Fārābı̄ distinguishes di#erent
meanings of ‘‘whether’’ or ‘‘what’’ or ‘‘that’’ or ‘‘why,’’ and, in
particular, di#erent meanings that they have in di#erent syllogistic
arts. Clearly if Parts One and Three of the H * urūf resemble any part
of the Metaphysics, it is , and indeed there is heavy overlap
between the lists of equivocal terms investigated in H * urūf Part One
and in .
Kitāb al-H* urūf Part One also discusses the relations between
primitive terms (‘‘terms of first imposition’’) and the terms derived
or ‘‘paronymous’’ [mushtaqq] from them. A proper noun like
‘‘Socrates,’’ a common noun like ‘‘horse,’’ and also an abstract
accidental term like ‘‘whiteness’’ are all primitive terms; by contrast,
the concrete accidental term ‘‘white’’ is paronymous or derived from
‘‘whiteness,’’ not necessarily in the sense that it arises later in the
history of the language than ‘‘whiteness’’ (although this is more
plausible in Arabic than it is in the English example), but in the
sense that something is called white because there is whiteness in it.
While Aristotle applies this distinction between paronymous and
non-paronymous terms only to nouns, Fārābı̄ thinks that all verbs
(by which, as is standard in Arabic grammar, he means finite verbs)
are paronymous from their mas*dars (a mas*dar is a nomen actionis,
comparable to a Greek infinitive but handled morphologically and
syntactically like any other noun). The grammatical form of a
paronymous term suggests that something is X by having an X-ness
present in it (or V’s by having an action of V-ing in it). But Fārābı̄
seems particularly interested in cases where the grammatical form
of a term misleadingly fails to track its logical form, the cases
that, according to Aristotle’s On Sophistical Refutations, can give
rise to sophisms of ˜ ˜
´
– usually translated ‘‘sophisms
of figure of speech,’’ but it might be better to say ‘‘sophisms
of grammatical form,’’ i.e. sophisms which arise because the
66 STEPHEN MENN
grammatical form of an expression fails to correspond to its logical

form, and which can be solved only by diagnosing this discrepancy.8
Given this sketch of what Fārābı̄ does in the three parts of the
Kitāb al-H
* urūf, we can begin to see how the three parts might fit
together, and what they might have to do with the Metaphysics,
specifically with Metaphysics . The parts do not seem to fit in the
transmitted order One-Two-Three; rather, Part Two reads as pro-
legomena to Parts One and Three.9 Part Two shows us that technical
vocabulary arises from a series of metaphorical extensions (and
morphological derivations); the same term can have meanings in
di#erent arts that di#er from each other and from its ordinary-
language meanings (or from the meanings of its ordinary-language
cognates, since it may not itself occur in ordinary language).
Confusions are thus likely to arise. Aristotle found it necessary to
write for his students or for the readers of his other works, to avoid
misunderstanding of his scientific vocabulary by distinguishing the
di#erent meanings of each term and explaining how the scientific
meanings have arisen from more ordinary meanings; and it is that
much more necessary for Fārābı̄ to do something similar, since the
8
On the appearances created by the grammatical form of paronymous expressions, and
instances in which they may mislead, see notably H * urūf I,20–21, I,26, I,36 and I,84; the
last-mentioned text is discussed below. Aristotle discusses sophisms of ˜ ˜
´
in
On Sophistical Refutations 4, 166b10–19, and at length in chapter 22. Categories 5, 3b13–16
says that species- and genus-terms like ‘‘man’’ and animal ‘‘seem in accordance with the
˜ ˜ ´’’ to signify some this, but in fact do not; comparison with the
Sophistical Refutations texts makes it clear that he means their grammatical resemblance to
proper nouns gives rise to fallacies of ˜ ˜
´
, some of which he discusses in
Sophistical Refutations 22. Fārābı̄ is here picking up on an important Aristotelian theme,
although developing it in new directions. On Fārābı̄ and Aristotle on paronymy, see
especially the important discussion in Zimmermann pp. xxiv–xli. Zimmermann also discusses
ways in which Fārābı̄ draws on the grammatical tradition, and considerations that may
have made Fārābı̄ more likely than Aristotle to believe that paronymous terms are actually
derived from earlier non-paronymous terms. (At p. xxxvii n. 3, commenting on our
Categories passage on how we can be misled by the grammatical form of ‘‘man,’’
Zimmermann says, ‘‘The shapes of words can of course only be held to be misleading in
particular cases if they are held to be significant in general; though I fail to see how the
shapes of Greek nouns can be supposed to be significant in general. There is no way of
telling that, say, ’´
is a substantive and
´ an adjective short of knowing what
these words mean.’’ But Aristotle is thinking of the fact that a common substantive term
such as ’´ , like a proper noun and unlike adjectives such as
´ , is not inflected
for gender and so never becomes paronymous. The word
´ , which Aristotle takes to be
the basic non-paronymous form, becomes paronymously
´ or
´ when said of a
masculine or feminine subject [and homonymously
´ when said of a concrete neuter
subject], as ´ , which he takes to be the basic non-paronymous form, becomes
paronymously ´ or ´ when said of a masculine or neuter subject [and
homonymously ´ when said of a concrete feminine subject]; but a woman is
’´
, not ’´ . On all this see my ‘‘Metaphysics, dialectic, and the Categories,’’
Revue de Métaphysique et de Morale, 100 [1995]: 311–37.)
9
On the order of the parts, see Mahdi’s discussion, pp. 40–3. The transmitted beginning
of Part One is fragmentary; Mahdi suggests that what is now Part One fell out of some
ancestor of the extant manuscript, and that what survived of it was put back in the wrong
place. Mahdi conjectures that the original order may have been Two-Three-One; I think
Two-One-Three is more likely.
FA
z RA
* URUzF 67
transmission of technical vocabulary from Greek into Arabic has
created even more opportunities for confusion.
I think this is indeed how Fārābı̄ is thinking, but it is not a full
solution. On this account it seems that and the Kitāb al-H * urūf
would be carrying out a rather elementary and propaedeutic task of
philosophical education; and they would seem to have no more
connection with metaphysics than with any other branch of philos-
ophy or science. Furthermore, this account fails to explain a funda-
mental fact about the Kitāb al-H * urūf, namely that the majority of the
terms discussed in Parts One and Three are grammatically particles,
rather than nouns or verbs. ‘‘H * urūf’’ (sg. h* arf), besides meaning
‘‘letters,’’ can also mean grammatical ‘‘particles,’’ where this has to
be taken in a somewhat broader sense than is customary in Greek
grammar, to mean any word that can neither be declined like a noun
nor conjugated like a verb, covering a range of short uninflected
words, pronouns and prepositions and adverbs and conjunctions and
the like (Fārābı̄ describes h
* urūf in this sense in his Kitāb al-Alfāz* ).10
It seems very unlikely to be a coincidence that most of Parts One and
Three of the Kitāb al-H * urūf are about h * urūf in this sense. But how
can the same book be Kitāb al-H * urūf both in the sense of filling the
role of Aristotle’s Metaphysics, and in the sense of being about
grammatical particles? Is this just a bizarre pun?
To see what is going on, we need to say something about Fārābı̄’s
attitude toward Greek philosophy. Fārābı̄ is unusual among
medieval philosophers in being interested in Greek philosophy not
just as a doctrine or as a discipline that he can practice, but as a
historical artifact. He is very interested in who the Greek philos-
ophers were, in who they were writing for and why and under what
political and religious circumstances, and in how their writing
follows the contours of the Greek language. He does not trust Greek
philosophy as it is presented to him by the Arabic translation-
literature, and is constantly trying to second-guess the translators
and to reconstruct what lies behind the veil of the translations.
Unfortunately, he has very little evidence to go on. His comments on
the Greek language in the Kitāb al-H * urūf make it all too plain that
he did not know Greek. His evidence about the language seems to
come from metalinguistic remarks in the translated texts (whether of
Aristotle or of later writers, some of them drawing on the Greek
grammatical tradition), from scattered comments by Arabic writers
or personal acquaintances (probably Christians with at least frag-
mentary knowledge of Greek as a sacred tongue), and from his own
10
There is an elaborate classification of h
* urūf, Kitāb al-Alfāz* al-musta‘mala fı̄ al-mant*iq,
ed. Muhsin Mahdi (Beirut, 1968), pp. 42–56: pronouns, articles, prepositions, ‘‘adverbs’’
[h
* awāshı̄ – the most interesting class, including most of the h * urūf discussed in the Kitāb
al-H* urūf, with many subspecies] and conjunctions.
68 STEPHEN MENN
reconstruction of the di$culties that the translators must have

faced. Now Fārābı̄, as we have seen, is very concerned with cases
where the grammatical form of an expression misleadingly fails to
correspond to its logical form, leading to the danger of sophisms of
˜ ˜
´
. When a language is stretched beyond its usual
expressive capacities such discrepancies are likely to occur, and the
Arabic translations of Greek philosophical texts, and the Arabic
philosophical texts written in imitation of the language of the
translations, are full of such awkward and potentially misleading
expressions. Fārābı̄, applying a methodological principle of charity,
seems to assume that in each case the Greek original was free of such
infelicities. This means that, in practice, his reconstructed Greek
serves as an ideal logical language, i.e. a language in which gram-
matical form always tracks logical form. (He is not theoretically
committed to the view that Greek is an ideal logical language, and
would deny it if asked, but this is what he assumes in practice.) Now,
where grammatical form tracks logical form, a non-paronymous
noun will signify either a substance or a being in some accidental
category, and a paronymous noun or a verb will signify that such a
being is present in or attributed to some underlying subject. But
metaphysics, as Fārābı̄ understands it, isn’t about things in the
categories (H * urūf I,11–17); it’s about the categories themselves
(especially substance) and about syncategorematic or transcenden-
tal concepts such as being, unity, essence, cause, and also God (who
also, for Fārābı̄, does not fall under any category).11 And, where
grammatical form tracks logical form, these concepts would be
signified, not by non-paronymous nouns, not by paronymous nouns
or verbs, but by particles. Particles in an ideal logical language thus
correspond roughly to what we would call logical constants. And it
is clear that Fārābı̄ thinks that at least some important metaphysical
notions were expressed in Greek by particles (or by terms morpho-
logically derived from particles). Most strikingly, the fragmentarily
transmitted beginning of Kitāb al-H * urūf Part One says that the
words ‘‘un’’ and ‘‘ūn’’ in Greek were particles of a$rmation, com-
parable to ‘‘inna’’ in Arabic (thus something like a Fregean
assertion-stroke), with ‘‘ūn’’ more strongly emphatic, and that the
Greeks use ‘‘ūn’’ to signify God by contrast with all other beings,
which they call rather ‘‘un’’ (I,1, p. 61,10-13). (This is presumably a
11
On the Aims stresses the non-material character of the objects of metaphysics
(non-material either by being separate immaterial substances, or by being universal
attributes which apply both to material and to immaterial things), the H * urūf their
non-categorial character. But this is not as great a di#erence as it might seem, since for the
H
* urūf ‘‘each of the categories [i.e. anything in any of the categories] . . . is predicated of
some sensible [and thus material] ´
’’ (I,6, p. 64,2–4). Diebler and Vallat seem to me to
exaggerate the extent to which the Kitāb al-H * urūf is about categories; its interest in the
categories points to what is beyond them.
FA
z RA
* URUzF 69
distorted account of the words ’´ and ’´ , nominative singular
neuter and masculine of the participle of the verb ‘‘to be’’; it is
perfectly true that Christians refer to God as ¢ ’´ , ‘‘he who is,’’
following Exodus 3:14
’ ´
’ ¢ ’´ , although the form ’´ is not
peculiar to God and neither ’´ nor ’´ is a particle of a$rmation.
Fārābı̄ probably did not know any languages that distinguish a
masculine from a neuter gender, so this explanation of the di#erence
is unlikely to have occurred to him.)
More generally, it seems likely that Fārābı̄ thought that the Greek
original of Metaphysics was devoted specifically to distinguishing
the many meanings of particles, rather than of nouns or verbs; this
would explain why Parts One and Three of the Kitāb al-H * urūf, whose
function corresponds to that of as he understood it, also concen-
trate on particles. But it is important to draw a distinction. Fārābı̄ is
very unlikely to have thought that every lexical item that heads a
chapter of was a particle; he surely knew that, for instance, the
Greek word for ‘‘cause’’ (heading 2) was a noun, and indeed some
of the lexical items heading sections of Kitāb al-H * urūf Part One
(such as ‘‘jawhar’’ = ‘‘substance’’ and ‘‘ ‘arad * ’’ = accident) are also
nouns.12 However, in such cases, he is likely to have thought that the
many meanings of the noun in question track the many meanings of
some particle, whether the noun is morphologically derived from the
particle or not. And in at least some cases this is in fact correct. Thus
the many meanings of ‘‘cause’’ discussed in 2 correspond to
di#erent meanings of ´ + accusative, ‘‘because’’ [ = lima], which is
certainly a particle for Fārābı̄ or for the Arab grammarians (the
correspondence between senses of ‘‘cause’’ [sabab] and ‘‘because’’
[lima] is made explicit at Physics II,7 198a14–16).13 Presumably also
the di#erent meanings of ’´ [ = jawhar] in 8 correspond to
di#erent meanings of the question ´
’ , ‘‘what is it?’’, and while
‘‘´’’ in Greek is declinable, the Arabic equivalent ‘‘mā’’ is an
uninflected particle. In fact, peering through the veil of the transla-
tions, Fārābı̄ would have had good inductive reason to believe that
every chapter of was about the many meanings of some particle.
Thus besides 8 on ’´, the chapters 13 and 14 on the categories
of quantity and quality are headed by terms derived from interroga-
tives which in Arabic are indeclinable particles [13, ´ , becomes
kamiyya; 14, headed by ´ but switching to ´ , becomes
kayfiyya]. Other chapters are even more explicitly about particles,
12
The full list of nouns treated is: jawhar = substance, ‘arad * = accident, nisba and id
* āfa
(‘‘relation’’ in a wider and a narrower sense), dhāt = essence, shay’ = thing,
mawjūd = being.
13
The Arabic of this passage is at Arist*ūt*ālı̄s, Al-T
* abı̄‘a, ed. ‘Abdurrah
* mān Badawı̄, 2
vols. (Cairo, 1964–5), vol. 1, p. 136. Metaphysics 2 also discusses the final cause under the
heads of
¢´
= min ajli and ‘´ = li-with-subjunctive (1013a32–b3). Fārābı̄ discusses min
ajli in Kitāb al-H
* urūf I,106.
70 STEPHEN MENN
18 about ´ + accusative [ = bi-dhāt or li-dhāt], 24 about

’
[ = min]; 23 is about ’´

but says in its last sentence (1023a23–5)
that the meanings of ’´

correspond to those of ’´

˜’, and
the Arabic, lacking a verb for ‘‘have,’’ makes the whole chapter
about the meanings of the preposition li- (corresponding roughly to
the Greek dative), with the final sentence saying that the meanings
of li- and of fı̄ [ =
’ ] correspond.14 11 is on prior and posterior,
which become particles in the Arabic, qabla [ = ´
] and ba‘da
[ = ¢´
]. Furthermore, 1 implies that the senses of ’´ corre-
spond to those of ¢´
= minhu; 22 1022b32–3 says that privation is
said in as many ways as the -privative, which becomes a separate
particle in the Arabic (Bouyges prints bi-lā but the text is uncertain);
and while the Arabic of 15 takes as its header mud * āf in place of ´
, the text still uses the preposition ilā = ´ enough that a reader
would probably conclude that the senses of mud * āf correspond to the
senses of ilā.15 Of course di#erent languages will express di#erent
concepts by means of particles, and Fārābı̄ would have reason to
suspect that the Greek original uses particles, either for the header-
term or for something whose meanings correspond to the meanings of
the header-term, even in those cases where the Arabic translation
does not; in any case, he thinks that these categorial (for 8, 13–15)
and transcendental or syncategorematic concepts should be ex-
pressed by particles, and that, given the inevitable infelicities of the
translation-process, the Greek original would have expressed them
in this logically perspicuous way more often than the translation
does.
From this perspective we can see why a discussion of the many
meanings of the terms covered in or in Parts One and Three of the
Kitāb al-H * urūf would belong specifically to metaphysics, and also
why the semantic discussion would not be merely an elementary
propaedeutic. Each syncategorematic etc. term (each logical con-
stant) has its own peculiar logical syntax, which we will be tempted
to assimilate to the logical syntax of ordinary categorematic terms,
thus engendering metaphysical confusions and sophisms of ˜
˜
´
, especially when the grammatical form of the term is that
of an ordinary non-paronymous or paronymous noun. We will need to
sort out these confusions and to disentangle the correct logical
14
Averroes’ commentary here (654,9–13) explicitly refers to fı̄ and li or lahu as h * urūf, and
he takes this up again in his commentary on the beginning of 24 (657,2–5), which also
refers to min =
’ , the subject of 24, as a h * arf; noted already by Mahdi, p. 33. (Warning:
the Arabic of this section of Averroes is lost, and was reconstructed by Bouyges from the
Hebrew and Latin translations.)
15
This conclusion would be further supported by the Categories chapter on mud * āf
(Mant*iq Arist*ū, ed. ‘Abdurrah * mān Badawı̄, 3 vols. [Cairo, 1948–52], vol. 1, pp. 48–54). Fārābı̄
in H
* urūf I,370–55 has a complicated and interesting discussion of nisba (relation in a broad
sense) and id * āfa (specifically the category of relation), again closely connected with the
meanings of di#erent particles, which I will discuss in the monograph.
FA
z RA
* URUzF 71
syntax, not only in teaching our freshmen, but also in guarding
ourselves against persistent temptations of thought, and in replying
to eminent past and contemporary philosophers. Fārābı̄ would see
himself as modeling himself here on Aristotle’s corrections of the
confusions of earlier Greek philosophers, perhaps Plato or (if Fārābı̄
cannot accept the possibility of conflict between Aristotle and Plato)
at any rate Parmenides;16 Fārābı̄ will aim a similar critique at earlier
Arabic writers.17
II. EXPRESSIONS FOR BEING IN GREEK AND ARABIC
However, the single most important term for Fārābı̄’s project in

Kitāb al-H
* urūf Part One, and surely also in Metaphysics , is ‘‘being’’
[mawjūd]. What does this have to do with particles? One answer can
be taken from the discussion in Kitāb al-H * urūf Part Three of the
question from Posterior Analytics II
’ ’´
= hal mawjūdun, which
asks for the existence [wujūd] of a thing (III,228–49), the many
16
Whether Fārābı̄ is willing to accept the possibility of conflict between Plato and
Aristotle depends in part on the authenticity of the Kitāb al-Jam‘ bayna ra’yay
al-H* akı̄mayn, which has been contested by Joep Lameer, Al-Fārābı̄ and Aristotelian
Syllogistics (Leiden, 1994), pp. 30–9. Here I leave both the authenticity issue, and the larger
issue of Fārābı̄’s evaluation of Plato, open.
17
Amos Bertolacci (per litteras) suggests interpreting the title Kitāb al-H * urūf simply as
‘‘book of particles,’’ and not taking the treatise as a whole as standing in any special
relation to the Metaphysics, although he grants the undeniable relation between Kitāb
al-H* urūf Part One and Metaphysics . It will be clear that I agree that the title means,
among other things, ‘‘book of particles.’’ But, as noted above, it seems very unlikely that
Fārābı̄ would have used this phrase without intending a reference to the Metaphysics as
well; and it should now be clear why a book on particles, treated the way that the Kitāb
al-H* urūf treats them, would stand in a special relation to metaphysics, and indeed to the
Metaphysics, read in a way that makes central. The Kitāb al-H * urūf (by contrast with the
Kitāb al-Alfāz* ) is concerned not primarily with the particles as expressions (and not with
particles in any one language) but rather with what they signify, and it is interested mainly
in their scientific and specifically metaphysical significata. This brings it close to
Metaphysics , whose aim, according to Fārābı̄ in On the Aims of the Metaphysics, is the
‘‘di#erentiation of what is signified by each of the expressions that signify the
subject-matters of this science and the species and attributes of its subjects’’ (Dieterici, p.
35) – thus it is not about the expressions (particles for the H * urūf, their grammatical form
unspecified here) but about their significata, and specifically about those that fall under
‘‘this science,’’ i.e. metaphysics. But the relation of the Kitāb al-H * urūf to the Metaphysics is
not simply to : when Fārābı̄ interprets and reworks in the H * urūf, he reads very much
as part of the larger Metaphysics, and the H * urūf also has sections corresponding to parts of
the Metaphysics beyond , including (as we will see below) Z17, which draws on the
Posterior Analytics to provide a crucial clarification about substance. Many of the particles
discussed in the H * urūf have both a logical sense, signifying a second intention, and a
‘‘real’’ scientific sense; the scientific sense will be specifically metaphysical (because
extra-categorial – both the logical sense and the metaphysical sense will fall outside the
categories, but for opposite reasons), and the main interest of the H * urūf will be in the
metaphysical sense, describing the logical sense chiefly to ward o# conflations between it
and the metaphysical sense. We will see below how this works in the crucial case of
expressions for being. There are of course some particles which have only logical senses,
such as ‘‘and,’’ and the H * urūf (unlike the Alfāz* ) generally does not bother to mention these
expressions at all.
72 STEPHEN MENN
meanings of the question hal mawjūdun may be taken either as many

meanings of mawjūd or as many meanings of the particle hal, and we
can say that the meanings of mawjūd track the meanings of hal, in
the same way that the meanings of ‘‘cause’’ track those of ‘‘because.’’
(‘‘There are some particles that, when they are connected with a
thing, signify that what is sought is the knowledge of its wujūd, not
the knowledge of its quantity or time or place, like our saying ‘hal,’ ’’
Kitāb al-Alfāz * p. 47,13–14.) However, there is another and more
direct connection between being and particles. Fārābı̄ thinks that
the Greek word for ‘‘being,’’ which is pronounced something like
‘‘astı̄n’’ and is at least sometimes translated into Arabic as ‘‘maw-
jūd,’’ is a particle – at any rate, he says that it is not a verb, his
description of its functions seems to imply that it cannot be a noun
either, and the only remaining option is that it is a particle. This
may seem strange, and it needs some qualification and some
context-setting.
Fārābı̄ thinks that Greek has several words that express being. In
particular, it has past-tense and future-tense copulas, corresponding
roughly to Arabic ‘‘kāna’’ and ‘‘yakūnu,’’ and these are of course
verbs. However, ‘‘in all the other languages [i.e. other than Arabic],
such as Persian and Syriac and Sogdian, there is an expression
[lafz
*a] which they use to signify all things without specifying one
thing as opposed to another thing [i.e. it is a 1-place predicate which
is true of any subject whatever], and they also use it to signify the
connection [ribāt*] between the predicate [khabar] and what it is
predicated of: this is what connects the predicate [mah * mūl] with the
subject when the predicate is a noun or when they want the predicate
to be connected with the subject simpliciter without any mention of
time’’ (I,82, p. 111,4–8): Fārābı̄ assumes that this holds also in Greek.
By contrast, in Arabic, although the propositions ‘‘X was Y’’ and ‘‘X
will be Y’’ are most naturally expressed using the verbal copula
KWN, ‘‘X kāna Y’’ or ‘‘X yakūnu Y,’’18 the present-tense or (more
precisely) tenseless proposition ‘‘X is Y’’ is most naturally expressed
simply by ‘‘X Y,’’ e.g. ‘‘Socrates is wise’’ = ‘‘Suqrāt*u h
* akı̄mun,’’ with
no verb or any other word to connect the subject-noun and the
predicate-noun. The only case where it is natural in a tenseless
sentence to use a third term to connect the subject-noun and the
predicate-noun is where the subject and predicate are both definite:
thus if I tried to express ‘‘Socrates is the sage’’ by ‘‘Suqrāt*u
al-h
* akı̄mu,’’ this would most naturally be understood not as the
sentence ‘‘Socrates is the sage’’ but as the noun phrase ‘‘Socrates
the sage,’’ ‘‘the wise Socrates’’; and so, to make it clear that I mean
the sentence, I use the ‘‘pronoun of separation’’ ‘‘huwa’’ or ‘‘hiya,’’
18
As we will see, Fārābı̄ prefers a di#erent and less natural way of expressing these
propositions in Arabic (see discussion in Zimmermann, pp. xliv–xlv and cxxxi–cxxxiv).
FA
z RA
* URUzF 73
‘‘Suqrāt*u, huwa al-h * akı̄mu,’’ literally ‘‘Socrates, he the sage,’’ i.e.
‘‘Socrates, he is the sage,’’ i.e. ‘‘Socrates is the sage.’’ So in a
present-tense or (rather) tenseless sentence, either the copula is not
expressed or it is expressed by a pronoun; there is no natural way to
express the copula of such a sentence by a verb.
Here, for Fārābı̄, Arabic shows a mix of good and bad features, i.e.
of ways in which the grammatical form of an expression corresponds
to and reveals its logical form, and ways in which the grammatical
form is misleading about the logical form. The distinction between
tensed and tenseless propositions is of great logical importance, and
it is a good thing that Arabic expresses this distinction grammati-
cally. (It is good, in particular, that propositions asserting essential
predicates, ‘‘Socrates is [a] man’’ or ‘‘man is [an] animal,’’ can be
expressed without tense-marking: any tense-marker, even of the
present tense, would suggest that the predicate might previously
have failed to hold of the subject, or might cease to hold of the
subject, and therefore that it is accidental and not essential to the
subject.) Fārābı̄ assumes that Greek must be at least as logically
perspicuous as Arabic, and it never crosses his mind that Greek
might have failed to express grammatically the distinction between
tensed and tenseless propositions.19 However, it is also logically
important to express the copula (the copula-term is ‘‘necessary in the
theoretical sciences and in the art of logic,’’ I,83 p. 112,3; one reason
would be to distinguish the subject and predicate sides of an
assertion, and to distinguish a subject-predicate sentence from a
subject-attribute noun phrase), and Arabic is deficient in not usually
expressing the copula in tenseless sentences. Fārābı̄ assumes that
Greek, like Persian and Sogdian, avoids this peculiar deficiency of
Arabic: thus Greek will express both tensed and tenseless copulas,
and will express them by di#erent (although morphologically
related) words. The tenseless copula, ‘‘hast’’ in Persian and ‘‘astı̄’’ in
Sogdian, is in Greek something like ‘‘astı̄n’’ (I,82 p. 111,11). ‘‘Astı̄n’’
cannot be a verb, since as Aristotle says, ‘‘a verb is what consignifies
time, no part of it signifying separately’’ (De Interpretatione 16b6–7).
It also seems that it cannot be a noun, since if it is ungrammatical to
predicate one noun of another by saying simply ‘‘X Y,’’ it will still be
ungrammatical if I insert a third noun between them; if the copula is
itself a noun, an infinite regress of nouns will be needed to connect
the subject with the nominal predicate. Thus ‘‘astı̄n’’ must be a
19
In fact Greek, like Arabic, can use a ‘‘nominal,’’ i.e. verbless, sentence to express a
predication without expressing any tense, e.g. ‘‘ ¢ ´ ´ .’’ However, in Greek it is
plausible to suppose that this is merely an elliptical expression for ‘‘ ¢ ´ ´

’ ,’’ whereas in Arabic this is impossible, both because the predicate complement of the
copula-verb KWN is in the accusative while the predicate of a nominal sentence is in the
nominative, and because the copula-verb must be either in the perfect, signifying the past,
or in the imperfect, signifying the future.
74 STEPHEN MENN
particle.20 For this reason, although ‘‘astı̄n’’ is a perfectly reasonable

Arabic rendering of ‘‘ ’´
’’ and we might just retransliterate it as
‘‘ ’´
,’’ I will continue to write it ‘‘astı̄n,’’ to distinguish the particle
in Fārābian ideal reconstructed Greek from the third-person singular
present indicative verb in historical classical Greek. Fārābı̄ of course
grants that Greek will also have past-tense and future-tenses
copulas, which will consignify time and will therefore be verbs,
paronymous from ‘‘astı̄n.’’21 And all of these words, ‘‘astı̄n’’ and the
verbs derived from it, will also have 1-place uses to express tensed
or tenseless existence. ‘‘X astı̄n’’ without a complement will mean
‘‘X exists’’ (not marked for tense), and in that sense ‘‘astı̄n’’ will
‘‘signify all things without specifying one thing as opposed to
another thing.’’
Fārābı̄ says (I,83–66, pp. 112–15) that the translators have had a
di$cult time with ‘‘astı̄n,’’ since there is no word in Arabic that does
exactly what ‘‘astı̄n’’ does in Greek. This crucial piece of meta-
physical vocabulary is thus an example of the kind of di$culties of
transmission described in Part Two (problems of discrepancy be-
tween grammatical and logical form will be involved, and more
surprisingly also problems of metaphor), which help to explain why
something like Metaphysics is all the more necessary when philos-
ophy has been transmitted from one linguistic community to another,
and thus when its vocabulary has been through a process of trans-
lation. The translators (di#erent translators, and presumably also
the same translators in di#erent contexts) have chosen di#erent
Arabic words to stand for ‘‘astı̄n,’’ each with some advantages and
20
With all this compare Fārābı̄’s commentary on De Interpretatione 16b22–5, Sharh *
al-Fārābı̄ li-Kitāb Arist*ūt*ālı̄s fı̄ al-‘Ibāra, ed. Wilhelm Kutsch and Stanley Marrow (Beirut,
1960), pp. 43–8. There are a number of tensions between this work and the Kitāb al-H * urūf,
arising notably from Fārābı̄’s lack of interest in metaphysical uses of mawjūd and related
terms in the context of the De Interpretatione. But here too Fārābı̄ argues that terms such
as mawjūd or astı̄n or un cannot, as far as their logical form is concerned, be either strictly
nouns (on pain of a regress of copulas), or strictly verbs (because they do not consignify
time); he concludes that Aristotle here calls them verbs in an extended sense, as signifying
or consignifying subject-predicate connection although not time. Fārābı̄ is much less
interested in particles here than in the Kitāb al-H * urūf, and says that Aristotle is not
interested in particles in the De Interpretatione (48,5–7). Nonetheless, following things that
Aristotle says (in the here distorting Arabic translation) at 20a3–5 and 21b5–7, Fārābı̄
speaks of ‘‘the particle yūjadu’’ at 129,6 and ‘‘the particle wujūd’’ at 165,23 (although these
are of course not grammatically particles in Arabic, nor would yūjadu be a particle in any
language if taken strictly as tensed). I hope to come back to the comparison between the
Kitāb al-H * urūf and the De Interpretatione commentary in my monograph.
21
On these ‘‘hyparctic verbs’’ [kalim wujūdiyya] in non-Arabic languages, comparable to
kāna and yakūnu in Arabic, see H * urūf I,82, and also Sharh * li-Kitāb Arist*ūt*ālı̄s fı̄ al-‘Ibāra
46,13–20. But in the latter text astı̄n is paronymous from the basic infinitive [mas*dar] of the
hyparctic verbs, whereas the H * urūf text denies that it is paronymous or derived from any
mas*dar (presumably it is itself the mas*dar from which the hyparctic verbs are derived).
Perhaps in the De Interpretatione commentary the idea is that astı̄n is a tenseless but
paronymous participle, like mawjūd. Of course a language, even Greek, might work like
this grammatically, but if so, from the point of view of the H * urūf, it is failing to perfectly
reflect logical form.
FA
z RA
* URUzF 75
disadvantages; and thus, although Fārābı̄ does not say so explicitly,
he must do some detective work to determine when it is in fact the
same word ‘‘astı̄n’’ (or others morphologically derived from it)
behind the di#erent terms of the Arabic translations.
One strategy, which Fārābı̄ himself generally follows, is to trans-
late ‘‘astı̄n’’ and the verbs derived from it by words derived from the
triliteral WJD, whose basic meaning is ‘‘to find.’’ When words from
this root are used to express the concept of being, they are being
metaphorically extended from their basic use in natural language;
also they will not have the same relationships of morphological
derivation as the corresponding Greek terms (in particular, ‘‘astı̄n’’
itself will be translated by a paronymous term), and presumably the
grammatical relationships in Arabic will be more remote from the
logical relationships than are the grammatical relationships in
Greek. These are disadvantages of this strategy; the main advantage
seems to be that it is possible to render the tensed and tenseless
copulas by words from the same root, and also to make the same
terms serve in both 1-place and 2-place uses. Thus for ‘‘X was Y,’’
instead of saying ‘‘X kāna Y,’’ we say ‘‘X wujida Y,’’ literally ‘‘X has
been found [to be] Y’’; for ‘‘X will be Y,’’ instead of saying ‘‘X yakūnu
Y,’’ we say ‘‘X yūjadu Y,’’ literally ‘‘X will be found [to be] Y’’; and
for ‘‘X is Y’’ without any expression of tense, instead of saying simply
‘‘X Y’’ without a third term, we say ‘‘X mawjūdun Y,’’ literally ‘‘X
found22 Y,’’ i.e. ‘‘X [is] found [to be] Y,’’ i.e. ‘‘X is Y.’’ Likewise, and
in much more natural Arabic, we can say ‘‘X wujida,’’ ‘‘X has been
found,’’ for ‘‘X was, i.e. existed [or was present, e.g. in a place]’’; ‘‘X
yūjadu,’’ ‘‘X will be found,’’ for ‘‘X will be, i.e. will exist’’; and ‘‘X
mawjūdun,’’ ‘‘X [is] found,’’ for ‘‘X is, i.e. exists’’ without any
expression of tense. The main disadvantage to this mode of expres-
sion that Fārābı̄ notes is that ‘‘mawjūd,’’ used to translate ‘‘astı̄n,’’ is
grammatically paronymous, and thus gives rise to the appearance
that something is mawjūd through a wujūd, found through a finding
or existent through an existence, just as things are white through a
whiteness:
the expression ‘‘mawjūd’’ is, in its first imposition in Arabic, paronymous,
and every paronymous term by its construction gives the impression that
there is in what it signifies an implicit subject and, in this subject, the
meaning [ma‘nā] of the mas*dar from which [the term] was derived [i.e. as
‘‘white’’ implies, without explicitly mentioning, a subject in which white-
ness is present]. For this reason the expression ‘‘mawjūd’’ has given the
impression that there is in every thing a meaning / entity [ma‘nā]23 in an
22
Where ‘‘found’’ is the passive participle of ‘‘find,’’ not the preterite active.
23
Fārābı̄’s use of ‘‘ma‘nā’’ (literally ‘‘meaning’’) here is influenced by its use in kalām for
‘‘entity,’’ especially in contexts where the entity would be named by an abstract noun or
mas*dar: thus I ask whether Zayd, or whether God, is living through a ma‘nā, i.e. through
some entity, life, which is present in him.
76 STEPHEN MENN
implicit subject, and that this meaning / entity is what is signified by the
expression ‘‘wujūd’’: so it gave the impression that wujūd is in an implicit
subject, and wujūd was understood as being like an accident in a subject.
(I,84, p. 113,9–14)24
To avoid this misleading appearance, Fārābı̄ says (I,86, pp. 114–15,
and cp. I,83 p. 112,8–19), some of the translators preferred instead to
translate astı̄n and its cognates by means of a vocabulary derived
from the word ‘‘huwa,’’ not in its basic use as a pronoun meaning
‘‘he,’’ but in its use as a ‘‘pronoun of separation’’ serving as a copula
in sentences like ‘‘Suqrāt*u, huwa al-h * akı̄mu,’’ ‘‘Socrates, he [is] the
sage,’’ i.e. ‘‘Socrates is the sage.’’25 Indeed, in 7, which Fārābı̄
thinks is about the many meanings of ‘‘astı̄n’’ (in fact Aristotle
introduces the chapter by speaking about the ways in which ` ’´ is
said, but then goes back and forth freely between meanings of ’´ and
meanings of ’´
or
˜’), the translator gives the heading as
‘‘huwiyya,’’ and then renders Aristotle’s 2-place uses of ’´
and

˜’ sometimes by ‘‘huwa,’’ sometimes by ‘‘yakūnu,’’ and quite often
by silence, leaving the Arabic reader to guess that it is the same
Greek word in each case. Fārābı̄ says that the reason many writers
refused to use the vocabulary of ‘‘huwa’’ and ‘‘huwiyya’’ is that
‘‘huwiyya’’ is not good Arabic (I,86, p. 114,15–20); we might add that
it also does not have tensed cognates, and that it cannot be used in
1-place contexts for ‘‘X exists’’ (indeed, the translator of 7 seems to
avoid all 1-place constructions – in the case of ‘‘the not-white is,’’
1017a18–19, by leaving out the whole phrase). Fārābı̄ himself says
that you can use whichever expression you like, as long as you are
aware of their misleading grammatical form and take care not to be
led astray by it (I,86, pp. 114,20–115,12).
III. THE TWO MAIN SENSES OF ‘‘MAWJU

z D’’ AND THE
CORRESPONDING SENSES OF ‘‘WUJU z D’’
Fārābı̄’s treatment of ‘‘mawjūd’’ in Kitāb al-H

* urūf Part One is his
counterpart to Metaphysics 7, and draws heavily on 7. One
obvious di#erence from 7 is that Fārābı̄ asks not only about the
24
Fārābı̄ adds that wujūd, since the word means literally ‘‘finding,’’ might be imagined to
depend on a human finder. This seems a bit silly – who would really make such a mistake? –
but the fact is that Kindı̄ says in On First Philosophy that ‘‘human wujūd’’ consists of two
wujūd’s (19,4), namely the wujūd of the senses and the wujūd of the intellect. He means
‘‘two modes of cognition,’’ and by ‘‘mawjūd’’ in this context he means ‘‘cognized’’ or
‘‘cognizable’’ by a human being; but he does not seem to distinguish sharply between saying
that there are two kinds of cognizable object, with two corresponding modes of cognition,
and saying that there are two kinds of existing thing, with two corresponding modes of
existence. (I cite Kindı̄ from R. Rashed and J. Jolivet, Œuvres philosophiques et scientifiques
d’al-Kindı̄, vol. 2: Métaphysique et cosmologie [Leiden, 1998].)
25
Fārābı̄ gives the example ‘‘hādhā huwa Zaydun,’’ ‘‘this is Zayd,’’ and says that it is
‘‘very unlikely in Arabic that they are using [‘huwa’] here as a pronoun’’ (I,83, p. 112,12–
13), presumably because if so it would duplicate the demonstrative pronoun ‘‘hādhā.’’
FA
z RA
* URUzF 77
senses of ‘‘mawjūd’’ but also about the corresponding senses of
‘‘wujūd.’’ Indeed, one of Fārābı̄’s main concerns is to ask, for each
meaning of ‘‘mawjūd,’’ whether the things that are mawjūd are
mawjūd through a wujūd really distinct from their essences, as the
things that are white are white through a whiteness really distinct
from their essences. The fact that ‘‘mawjūd’’ in Arabic is grammati-
cally paronymous creates the appearance that this is so, but Fārābı̄
thinks that this is a basic metaphysical error, and he wants to
eliminate this error by an examination of each sense of ‘‘mawjūd.’’
Since the word ‘‘astı̄n’’ in Greek is not paronymous, Aristotle cannot
have faced precisely the same problem. Nonetheless, Fārābı̄ thinks
that an interpretive exposition of some of what Aristotle says about
the di#erent senses of ‘‘astı̄n’’ in 7 is the best way to solve the
problem about whether things are mawjūd through a wujūd really
distinct from their essences. So Fārābı̄ seems to think that Aristotle
too was concerned to eliminate the same metaphysical error, presum-
ably because some earlier Greek philosophers (perhaps Plato or
Parmenides) had fallen into that error. So the error can be made even
independently of the misleading grammatical structures in which
Arabic expresses the concept of being, although presumably the
grammar of Arabic makes the error all the more tempting.
On Fārābı̄’s analysis, the error of thinking that things are mawjūd
through a wujūd really distinct from their essences can arise, not
only from being misled by the paronymous form of ‘‘mawjūd,’’ but
also from confusing di#erent senses of ‘‘mawjūd,’’ and this is why we
can eliminate it by distinguishing those senses. Aristotle in 7
distinguishes four senses of being, namely being per accidens, being
per se (the sense of being that is divided into the ten categories
– since it is not said of them univocally, it falls into ten sub-senses),
being as truth, and being as actuality and potentiality (thus this last
sense falls into two sub-senses). Fārābı̄, however, concentrates
overwhelmingly on only two of these senses, being as the true and
being per se (he first distinguishes being as having a quiddity outside
the soul from the sense of being which signifies the categories, I,88,
but then reduces these to a single sense, I,90);26 and his accounts of
26
See below on the relation between being as having a quiddity and being as signifying
the categories. For ‘‘the true’’ as a sense of being Fārābı̄ prefers *sādiq, whereas the
translator of Metaphysics has h * aqq; but Fārābı̄ uses h
* aqq at Principles of the Opinions of
the People of the Perfect City (Al-Farabi on the Perfect State, ed., tr., comm., Richard Walzer
[Oxford, 1985]), p. 75. Fārābı̄ discusses actuality and potentiality at H * urūf I,93–8, and these
are of course important notions for him, but he takes them as divisions of being-as-
having-a-quiddity rather than as independent senses of being. The distinction between
existing in potentiality and in actuality or perfection will resurface in a brief but important
passage on God’s manner of existing, H * urūf III,240, p. 218,18–21. But it is clear, and
noteworthy, that Fārābı̄ takes the fundamental structure of Metaphysics 7 as a distinction
between only two senses of being. Fārābı̄, and Arabic writers generally, treat actuality and
potentiality as attributes of being, analogous to unity and multiplicity, rather than as senses
of being, and thus treat Metaphysics as akin to I rather than to E or Z. This is interesting
78 STEPHEN MENN
both of these are highly expansive and interpretive. Although 7 is

Aristotle’s ‘‘o$cial’’ statement of the di#erent senses of being, it is
highly compressed and often leaves it unclear how the di#erent
senses are supposed to be related or distinguished – the account of
being as the true, especially, is extremely compressed and leaves
many interpretive questions open. One particular source of interpre-
tive di$culty is that Aristotle does not make fully clear whether each
of these senses of being is supposed to apply only in 1-place contexts
(‘‘X is’’), only in 2-place contexts (‘‘X is Y’’), or both, and, if in both,
how its uses in the two kinds of context are related. All the examples
he gives of being as the true are 2-place, and modern commentators
have often thought that being as the true is for Aristotle a sense only
of 2-place being; by contrast, if being per se is said non-univocally of
the di#erent categories, in one way of substances, in another way
of qualities, and so on, it seems that it should be primarily a sense
of 1-place being, a quasi-genus (which would be a genus if it were
univocal) of which substance, quality, and so on are the
quasi-species.
Fārābı̄ clearly does not distinguish being-as-truth from being-as-
divided-into-the-categories in this way: while he is well aware of the
distinction between 1-place and 2-place being,27 he thinks that both
of the main senses of being from 7 apply both in 2-place and in
1-place contexts. (Fārābı̄ apparently thinks that, for any sense of
being, ‘‘Y is’’ is true in that sense i#, for some X, ‘‘X is Y’’ is true in
that sense, and I think this is probably right as an interpretation of
Aristotle.) But the 1-place contexts are the most important for
Fārābı̄’s argument, because it is especially here that the question
arises whether what is mawjūd is so through a wujūd really distinct
from its essence. It is also in the case of 1-place being-as-the-true
where Fārābı̄’s formulations are most distinctive.
Fārābı̄’s formulas for the two main meanings of being are at first
sight unenlightening, and it needs some work to bring out the
contrast he is intending to draw. He paraphrases being per se as
‘‘being circumscribed [munh * āz] by some quiddity outside the soul,
whether it has been represented [tus*awwira] in the soul or has not
been represented’’ (I,88, p. 116,7), and being-as-the-true as ‘‘being
outside the soul and being by itself [bi-‘aynihi] as it is in the soul’’
(I,88, p. 116,5). Fārābı̄ does not intend this as a distinction between
1-place being on the one hand and 2-place being on the other.
and rather surprising, and there is an important contrast with Thomas Aquinas; I will
discuss all this in the monograph.
27
Thus ‘‘there is an expression which they use to signify all things without specifying
one thing as opposed to another thing, and they also use it to signify the connection
between the predicate and what it is predicated of,’’ I,82 p. 111,4–6, cited above; Fārābı̄ will
often consider sentences illustrating 1-place and 2-place being separately, but in a way that
brings out analogies between the two cases.
FA
z RA
* URUzF 79
Perhaps the paradigm cases of being per se are 1-place – to be
‘‘circumscribed by some quiddity outside the soul’’ is a 1-place
predicate, applying to beings in the categories (propositions do not
have quiddities), and the word for ‘‘represent’’ (the fifth form of
S
* WR, mas*dar ‘‘tas*awwur’’) is technical for conceiving an object, by
contrast with judging that (the second form of S * DQ, mas*dar
‘‘tas*dı̄q’’) – but Fārābı̄ clearly intends being per se to apply to 2-place
cases as well (notably III,228-31, discussed below, treats 1-place and
2-place cases in parallel: Aristotle gives both 1-place and 2-place
examples of being per se in Posterior Analytics I,4). Likewise, the
paradigm cases of being as the true may be 2-place, but Fārābı̄ makes
clear here that he wants to include 1-place cases as well: to quote
more fully, being ‘‘can be said of every judgment such that what is
grasped [mafhūm] by it is by itself [bi-‘aynihi] outside the soul as it is
grasped, and in general of everything represented [mutas*awwar] and
imagined in the soul and every intelligible / thought [ma‘qūl]28 which
is outside the soul and is by itself [bi-‘aynihi] as it is in the soul’’ (I,88,
p. 116,3–6). In I,91 he gives the void as an example of something that
is not true, because it is not in itself outside the soul (p. 118,4–8), and
in III,228, discussing the question ‘‘does the void exist [hal al-khalā’u
mawjūdun]?’’ and other examples of the
’ ’´
question of Posterior
Analytics II, he says ‘‘the meaning of the question is whether what is
grasped in the soul through the expression is outside the soul or not,
i.e., whether what of it is in the soul is true or not, for the meaning
of truth is that what is represented in the soul is by itself outside the
soul, and the meaning of being and of truth here is one and the same’’
(pp. 213,23–214,3).29 Thus in 1-place cases to say that X is true is to
say that the concept of X is instantiated outside the soul.
If the di#erence between being per se and being-as-the-true is not
between 1-place and 2-place being, we might think from Fārābı̄’s
glosses on the terms in I,88 that the main di#erence is that being-
as-the-true only applies to something once it has been represented in
28
In Fārābı̄’s usage, ‘‘ma‘qūl,’’ though literally ‘‘understood’’ or ‘‘intelligible,’’ refers to
things in the soul rather than to the objects which the soul understands, following the
translators’ use of ‘‘ma‘qūl’’ for ´ in the De Interpretatione and in the standard
description of the Categories as about ‘‘words signifying things by means of thoughts.’’ Thus
when De Interpretatione 16a3–4 says that spoken words are symbols of ´ (athār in
the Arabic) in the soul, Fārābı̄ paraphrases this by ‘‘ma‘qūlāt,’’ Sharh * li-Kitāb Arist*ūt*ālı̄s fı̄
al-‘Ibāra 24,13–20; Aristotle does use ´ = ma‘qūl at 16a10. Ma‘qūlāt are thus not
restricted to intellectual as opposed to sensitive or imaginative representations: Fārābı̄ is
capable of distinguishing ma‘qūlāt ma‘qūlāt, intelligible intelligibles, from ma‘qūlāt
mah * sūsāt, sensible intelligibles (Kitāb al-H
* urūf I,6 p. 64,4–5).
29
Likewise in III,247, ‘‘dialectic does not rise, in the meanings of mawjūd, above its
ordinary[-language] meanings: and thus by our saying ‘is man mawjūd?’ [hal al-insān
mawjūd] must be understood the meaning ‘is man one of the mawjūdāt which are in the
world?’, like what is said about the heaven, ‘it is mawjūd’ [innahā mawjūda], or about
the earth, ‘it is mawjūd,’ and all these come back to ‘that they are true’ [innahā *sādiqa]’’
(p. 223,13–16). Thus when the dialectician says that the earth is mawjūd, he means that it is
true. I will come back to this text below.
80 STEPHEN MENN
the soul, whereas being per se applies to anything ‘‘whether it has

been represented or whether it has not been represented’’ (I,89,
pp. 116,23–117,1): being-as-the-true would thus be the narrower sense,
and this is indeed what Fārābı̄ says in I,91. However, it becomes clear
from the discussion in Part Three that there is a much deeper sense
in which being-as-the-true is broader than being per se, because it
applies even to things which do not have quiddities outside the soul,
notably negations and privations (and presumably per accidens
unities such as white man): ‘‘it is not the case that everything which
is grasped through some expression, such that what is understood by
it is also outside the soul, also has an essence, e.g. the meaning of
privation: for it is a meaning which is grasped, and it is outside the
soul as it is understood [to be], but it neither is nor has an essence’’
(III,240, p. 218,12–15). Accordingly, Fārābı̄ says in III,229 that it’s
only once we’ve established that X has being-as-truth that we can go
on to ask whether it has being-as-having-a-quiddity-outside-the-soul,
where the latter is narrower.30 Thus we can say that X is mawjūd in
the sense of the true when we do not yet know what category X falls
under, or even whether it falls under any category at all (since X may
turn out to be a negation or privation or a per accidens unity), and
thus the meaning of ‘‘mawjūd,’’ in the sense of the true, must apply
univocally to all categories and even to negations. Likewise we can
say ‘‘X mawjūdun Y,’’ in the sense that the judgment is true, when we
do not know what category Y falls under, or whether it falls under
any category at all, and the meaning will be the same for all
categories and even for negations.
Fārābı̄ is here picking up on a point that Aristotle makes in 7,
that being-as-truth applies equally to negative and a$rmative propo-
sitions: ‘‘being [
˜’] and ‘is’ also signify that [something is] true,
and not-being that [it is] not true but false, equally in a$rmations
and in denials, e.g. that Socrates is musical [’´
´ ´ ]
because this is true, or that Socrates is not white [’´
´ ’

´ ], because that is true’’ (1017a31–4). That is: while ‘‘Socrates is
not white’’ most obviously asserts a not-being (Socrates’ not-being-
white), Aristotle insists that it also asserts a being (Socrates’
being-not-white),31 although only in the minimal sense of being that
applies equally to all propositions, not in the stronger sense that he
calls being per se. But as far as we can tell from 7, this univocal
minimal being-as-truth applies only in 2-place contexts, whereas
Fārābı̄ is applying it also in 1-place contexts: the not-white (or
30
This two-stage procedure, beginning by investigating being-as-truth and proceeding to
investigate being per se, is supposed to apply both in 1-place cases (on which I am
concentrating here) and in 2-place cases. I will return to this passage and to this parallel
below.
31
Thus Aristotle preposes ‘‘ ’´

’’ in ‘‘ ’´

´ ’
´ ’’ to make it clear that
‘‘ ’´

’’ does not stand under the negation sign.
FA
z RA
* URUzF 81
privation in general) exists, in the sense of being outside the soul
as it is represented within the soul, just as much as the white does.
By contrast, being in the sense of having a quiddity outside the soul
is said per prius et posterius (I,92), primarily of substances and
derivatively of things in the other categories (indeed it is not said
univocally even of all substances, but primarily of the first causes
and derivatively of what exists through them), and does not apply at
all to negations.
Fārābı̄ says in I,89 that ‘‘the wujūd of what is true is a relation of
the intelligibles to what is outside the soul’’ (p. 117,4–5). That is: for
something to exist (have 1-place being) in the sense that it is outside
the soul as it is in the soul, is for some concept to be instantiated. So
when I say that the thing exists in this sense, I am predicating
existence, not of an external thing, but of a concept, and saying that
there is some thing of which that concept holds. (Presumably, if I say
‘‘X mawjūdun’’ tenselessly, I mean that there is something of which
that concept holds at same time, not necessarily at the present.) Thus
wujūd in this sense is what Fārābı̄ calls a ‘‘second intelligible’’ or
‘‘second intention’’ [ma‘qūl thānı̄]. Fārābı̄ is apparently the inventor
of this expression (taken up by Avicenna and Averroes from him, and
by the Latins from them), which he has introduced near the begin-
ning of H * urūf Part I as transmitted (I,7).32 A second intention is a
concept applying to concepts, so something that is predicated of
thoughts or ‘‘intelligibles’’ in the soul rather than directly of
external things. Being-a-predicate, for Fārābı̄, is a basic example of a
second intention, and being-truly-predicated-of-some-external-thing
is a second intention derived from that basic second intention. Thus
what Fārābı̄ is saying about (1-place) mawjūd in the sense of the
true, and the corresponding sense of wujūd, is close to what Frege
32
Fārābı̄ has apparently modeled the phrase on the grammarians’ notion of a ‘‘term of
second imposition,’’ i.e. a metalinguistic term: thus in H * urūf I,8, responding to the objection
that positing second intentions will lead to an infinite regress, he compares the term
‘‘accusative,’’ which can be put into the accusative without an objectionable regress. On
the rather complicated background of the notions of first and second imposition, see
Zimmermann esp. pp. xxx–xxxv. Fārābı̄ introduces the domain of second intentions in order
to demarcate the realm of logic as opposed to the other philosophical sciences (which deal
with first intentions falling under the categories) and to metaphysics (which deals with
things not falling under the categories), and also implicitly as opposed to grammar (which
deals with words rather than with their meanings; on the background of the H * urūf in the
debate between the philosophers and the grammarians, see Mahdi pp. 44–9, Zimmermann
pp. cxviii–cxxix, and A. Elamrani-Jamal, Logique aristotélicienne et grammaire arabe [Paris,
1983]). In place of Fārābı̄’s ‘‘ma‘qūl’’ Avicenna usually says ‘‘ma‘nā,’’ but at least once he
says that the subject-matter of logic is al-ma‘ānı̄ al-ma‘qūla al-thāniya which depend on
al-ma‘ānı̄ al-ma‘qūla al-ūlā, The Metaphysics of the Healing (ed. and tr. M.E. Marmura
[Provo, 2005]), 7,16–7. ‘‘Intention,’’ or ‘‘intentio’’ in Latin, is usually thought of as
equivalent to Arabic ‘‘ma‘nā’’ rather than to ‘‘ma‘qūl,’’ but there is no di#erence in
meaning: Avicenna simply decided to substitute the term ‘‘ma‘nā’’ for Fārābı̄’s ‘‘ma‘qūl.’’
And rightly so: ‘‘ma‘qūl’’ is a very peculiar word to use in this sense, and its meaning
becomes clear only against the background of the uses of ‘‘´ ’’ in Aristotle and his
commentators.
82 STEPHEN MENN
means by saying that existence, in the sense that is symbolized by the

existential quantifier, is a second-order concept (although Fārābı̄
has a psychological notion of concept, not the Fregean notion of a
concept as a function whose values are the true and the false). He is
apparently the first person to say anything like this. (Aristotle does
say in Metaphysics E4 that being-as-truth is an a#ection in the soul,
but Aristotle at least in that text is thinking exclusively of 2-place
being-as-truth, and not of existence in any sense. It is as far as I know
an innovation of Fārābı̄’s to make being-as-truth an all-inclusive
univocal 1-place predicate, although he will have been started in this
direction by passages like Metaphysics Z4 1030a25–26, where some,
speaking ˜ , say even of not-being that it is, not that it is
simpliciter but that it is not-being, and this applies to other non-
substances too.) However, Fārābı̄ di#ers from moderns who speak of
existence as a second-order concept or second-order predicate in that
he does not distinguish sharply between 1-place and 2-place contexts,
and is willing to lump 1-place and 2-place being-as-truth under the
same sense of being. Thus while I think it is fair to say that what
Fārābı̄ calls 1-place being-as-truth is what we symbolize by the
existential quantifier, and that he means the same thing by calling it
a second intention that we mean by calling it a second-order
predicate, it is not the existential aspect of it that he thinks makes it
second-order.33 Fārābı̄ also di#ers from many moderns in that he does
not think that ‘‘exists’’ is always a second-order concept. If the
concept of X is instantiated, we can ask whether X also has a
quiddity, falling under some one of categories, and, if so, we can say
‘‘X exists’’ in a further sense, the sense of having-a-quiddity-outside-
the-soul; and existence here is a first-order concept, since it is
predicated of the external instance of the concept, not of the concept
in the soul. And when I say that X is mawjūd in the sense of having
a quiddity outside the soul, the wujūd which I am predicating of X is
just that quiddity.
As I have said above, Fārābı̄’s rewrite of 7 di#ers from the
original in that he is concerned not only with the di#erent senses of
‘‘mawjūd’’ but also with the corresponding senses of ‘‘wujūd’’;
indeed, it seems that one main reason he wants to distinguish
between di#erent senses of ‘‘mawjūd’’ is in order to distinguish
corresponding senses of ‘‘wujūd,’’ and he wants to do this in order to
33
If we want to find a modern parallel to Fārābı̄’s second-order 2-place being-as-truth, let
it be the Fregean function with two arguments, an object and a concept, whose value is the
true if the object falls under the concept, and the false otherwise. Frege would not be
especially interested in this second-order concept (why not just apply the first-order concept
to the object, rather than applying this awkward second-order concept to the object and the
first-order concept?), but he couldn’t deny its legitimacy. But of course he would not think
of this concept and the second-order concept symbolized by the existential quantifier as
giving anything like the same sense of being; to assimilate them in this way would be to
undercut what is distinctive about the existential quantifier.
FA
z RA
* URUzF 83
eliminate the appearance, strengthened in Arabic by the paronymous
form of the word ‘‘mawjūd’’ but possible even in Greek, that a thing
is mawjūd through a wujūd other than its essence. The opponent
must be asked what he means by ‘‘mawjūd’’: mawjūd as having a
quiddity outside the soul, mawjūd signifying the categories, or
mawjūd as the true? If he means mawjūd as having-a-quiddity-
outside-the-soul, then the corresponding wujūd is just that quiddity.
If the mawjūd is something divisible, something that could be spelled
out in a definition, then according to Fārābı̄ we can say that the
wujūd is the articulated quiddity, so that man is a mawjūd and the
wujūd through which he is mawjūd is rational animal (I,89, follow-
ing the Aristotelian usage in which a definition gives the ’´ or the

˜’ or the ´ ˜’
˜’ of a thing). We can also say of each of the parts
that would be mentioned in the definition, the genus and di#erentia
or form and matter, that it is the (at least partial) wujūd of the thing,
and Fārābı̄ says that this applies especially to the di#erentia (again
I,89, again following Aristotle, e.g. Metaphysics 8 1017b17–22 for the
parts as well as ‘‘the ´ ˜’
˜’, whose ´ is a definition,’’ as the
’´ of the thing; Metaphysics Z12 says that the ultimate di#erentia
of a thing is its ’´). The wujūd corresponding to the sense of
mawjūd signifying the categories is just the relevant category (so for
Socrates it is substance): this is a genus, and so reduces to the wujūd
of what has a quiddity outside the soul, and Fārābı̄ uses this to justify
subsuming mawjūd-signifying-the-categories under mawjūd-as-
having-a-quiddity-outside-the-soul (I,89–90; thus typically he speaks
of only two senses of mawjūd, having-a-quiddity-outside-the-soul and
the true).34 The distinction between a mawjūd in this sense and its
wujūd can be no greater than that between a thing and an essential
part through which the thing exists; in the case of a simple indivis-
ible mawjūd there is no distinction at all. Wujūd in this sense is
something real, but not univocal to things in di#erent categories and
perhaps not even to things in the same category, and it is not
extrinsic to the essence of the mawjūd that possesses it: rather, it just
34
The identification of the wujūd corresponding to mawjūd in this sense with the
category depends on the assumption that mawjūd in this sense is univocal to all things
within a given category; Fārābı̄ wavers on this. The point of speaking of ‘‘mawjūd as
immediately signifying the categories’’ or the like is that it is a sense that immediately falls
into ten senses according to what it is predicated of: it is a quasi-genus of the ten
categories, and if it were said univocally of them it would be a genus and would be included
in the essences of all the categories and of all their species, but in fact when it is said of a
quality it is not really distinct from quality, when it is said of a quantity it is not really
distinct from quantity, and so on. However, Fārābı̄ considers the possibility that being may
have more senses than that. I,88 prefers the view that ‘‘mawjūd’’ is said equivocally of the
di#erent summa genera, and then said univocally within each genus, but also admits the
possibility that it is also said equivocally of the di#erent things under each summum genus;
and I,92 says that, within the category of substance, the primary and causally independent
substances are more properly mawjūd than the substances that depend on them. I will
return to the I,92 passage below.
84 STEPHEN MENN
is that essence, or a part of that essence. By contrast, if the question

is raised concerning mawjūd as the true, this will be mawjūd through
a wujūd which is univocal and extrinsic to the essence (the propo-
sition ‘‘X exists,’’ in the sense of ‘‘the concept of X is instantiated,’’
is not analytically true), but which is not real, since it is merely a
second intention. Only by confusing these two senses of mawjūd
could we conclude that there is a wujūd which is real, predicated
univocally of all things, and extrinsic to their essence.
IV. METAPHYSICS AND POSTERIOR ANALYTICS II
Fārābı̄’s account of the senses of ‘‘mawjūd’’ in Kitāb al-H * urūf Part

One is clever and o#ers a plausible way of dividing the senses so as
to support his conclusion that in no sense of ‘‘mawjūd’’ does a thing
exist through a real wujūd really distinct from its essence. But the
real depth, and even the real justification, for Fārābı̄’s way of
dividing the senses of ‘‘mawjūd,’’ comes only when the passage is
read in conjunction with Kitāb al-H * urūf Part Three. (I have of course
already cited Part Three in making the point that being in the sense
of having a quiddity outside the soul is narrower than being as the
true.)
Kitāb al-H * urūf Part Three is methodologically uniform with Part
One, and quite possibly continuous with it, in that it too is examining
the many senses of particles (or of transcendental or syncategore-
matic notions which would in an ideal language be signified by
particles), distinguishing their senses in the di#erent syllogistic arts
from each other and from ordinary-language uses. However, while
Kitāb al-H * urūf Part One heavily overlaps with Metaphysics in the
list of terms it investigates, and may be described as Fārābı̄’s version
of , Part Three has no real overlap with (or with any other part
of the Metaphysics), and instead follows (with some expansion) the
list of questions or topics of scientific investigation from Posterior
Analytics II.35 But Fārābı̄ is doing something deliberate and inter-
esting in including Part Three in the Kitāb al-H * urūf. It is not simply
that Posterior Analytics II organizes some of its discussion under the
headings of particles,
’ ’´
and ´
’ and ¢´ and ´ , and that
Fārābı̄ therefore decides to combine it with Metaphysics in a
catalogue of particles. Rather, Fārābı̄ is claiming that the discus-
sions of Posterior Analytics II have metaphysical significance, and
that we can give greater scientific depth to the discussions of
Metaphysics by filling them out from Posterior Analytics II; and he
35
The only overlap with comes in the discussion of lima = ` ´, but even this is closer
to Posterior Analytics II than to 2. See below for the relation of Fārābı̄’s list of questions
in H
* urūf Part Three to the lists current in the Kindı̄ circle, which explains much of Fārābı̄’s
divergence from the list in Posterior Analytics II.
FA
z RA
* URUzF 85
thinks that and Posterior Analytics II work together to eliminate
deep errors that arise when we fail to discern the distinctive
scientific meaning of the particles.
The Posterior Analytics is read by Arabic philosophers, not simply
as analyzing science or as prescribing how to give an exposition of
scientific results, but as laying down a method for demonstrating (in
Book I) and defining (in Book II). Book II gives, in particular, the
scientific method for investigating
’ ’´
and ´
’ , that is,
existence and essence. Also, as Fārābı̄ takes it, this book distin-
guishes the scientific from the dialectical method of investigating
these topics. Furthermore and more surprisingly and interestingly,
Fārābı̄ claims that there are distinctive scientific-as-opposed-to-
dialectical senses of essence and existence, or di#erent senses of the
particles ‘‘hal’’ and ‘‘mā’’ in the di#erent syllogistic arts. Thus he
thinks that Posterior Analytics II is laying down not only a distinc-
tive scientific methodology but also a distinctive scientific ontology,
a scientific understanding of essence and existence, which will add
scientific depth to , and which together with will give a crucial
intellectual substructure for the rest of the Metaphysics.
The most distinctive contribution of Posterior Analytics II,1-10 to
the understanding of the four scientific questions, and the key to its
account of definition in particular, is the claim that the investigation
´
’ stands to the investigation
’ ’´
as the investigation ´
stands to the investigation ¢´. That is: to say, scientifically, what X
is, is to give the cause of the fact that X is. Thus, for example, to say
what a lunar eclipse is is the same as to give the cause of the fact that
there is a lunar eclipse, namely that it occurs due to the interposition
of the earth between the moon and the sun; this statement of the
cause of lunar eclipses will be incorporated into the scientific
definition of lunar eclipse as (say) ‘‘darkening of the moon at
opposition due to the interposition of the earth between the moon
and the sun.’’ One important consequence, for Aristotle, is that we
cannot scientifically investigate what a lunar eclipse is unless we
have first established (whether by sense-perception or by inference)
that there are lunar eclipses,36 just as we cannot scientifically
investigate why the earth is spherical until we have first established
that it is spherical. Now this might seem to threaten a vicious circle:
we cannot investigate what X is until we have established that X is,
but how can we hope to establish that X is if we don’t yet know what
it is that we’re trying to establish, i.e., if we don’t yet know what X is?
Aristotle breaks this circle by saying that, in order to investigate
36
Presumably he isn’t saying that we can’t investigate unless we know that there is a
lunar eclipse right now; but we must know at least that there are potentially or habitually
lunar eclipses (and we are not going to get very far investigating their causes unless there
were eclipses in the past whose circumstances were recorded and which can therefore be
investigated to determine their causes).
86 STEPHEN MENN
whether X is, we have to begin from a preliminary account of what X

is, but that this preliminary account, unlike the scientific account of
what X is, does not cite the causes of the fact that X is. Thus in the
case of lunar eclipse, the preliminary account might be ‘‘darkening
of the moon at opposition.’’ That preliminary account is enough for
us to investigate whether there are lunar eclipses or not, i.e., whether
the moon is sometimes darkened at opposition. If it is, then we can
investigate why the moon is darkened at opposition; and if we
discover (say) that the moon is darkened at opposition because the
earth, when it comes directly between the moon and the sun, blocks
the light of the sun from reaching the moon and being reflected o# its
surface, then we can incorporate this causal discovery into a scien-
tific definition of lunar eclipse as ‘‘darkening of the moon at
opposition due to the interposition of the earth between the moon
and the sun’’ or the like.
Fārābı̄ in Kitāb al-H
* urūf III,215–44 explains all this faithfully and
intelligently (for the quiddity of a thing, asked for in the scientific
use of the interrogative particles mā or mādhā [‘‘what?’’], as a cause
of the wujūd of the thing, so coinciding with what would would be
asked for by the particles lima or li-mādhā or bi-mādhā [‘‘why’’ or
‘‘through what’’], see III,216–17; for the eclipse example, and the
preliminary and final accounts of what an eclipse is, see III,244). And
in trying to draw ontological implications out of Posterior Analytics
II, and to make ontology scientific by this means, he is clearly
inspired by Aristotle’s Metaphysics Z17, where Aristotle proposes to
reframe the investigation ‘‘what is X?’’ as a causal investigation,
taking what X is as the cause of that X is, as the Posterior Analytics
recommends (Z17 recycles some of the examples from Posterior
Analytics II – eclipse, thunder, man). To state that X is in a way that
makes it amenable to causal investigation, we must rewrite it with a
2-place ‘‘is,’’ ‘‘S is P,’’ so that we can investigate why P belongs to S
(thus rewrite ‘‘eclipse is’’ as ‘‘darkening-at-opposition belongs to the
moon’’). Fārābı̄ takes up this point, using Aristotle’s examples of
2-place rewriting from Metaphysics Z17, at H * urūf III,244 pp. 221,13–
222,2. And Fārābı̄ is here picking up the genuinely Aristotelian thesis
that scientific methodology rather than dialectic, the Posterior
Analytics rather than the Topics, is the right starting-point for
investigating essences.
However, Fārābı̄ also wants to connect this account of how we
investigate
’ ’´
and ´
’ with what he has said in Part One
about the di#erent senses of being, and with an account of the
di#erent meanings of the particles in the di#erent syllogistic arts. By
Aristotle’s rules the dialectician never investigates ‘‘why’’ questions,
and so the dialectical definitions (the dialectician’s answers to
‘‘what’’ questions) do not include the cause: to know, for instance,
FA
z RA
* URUzF 87
that a lunar eclipse is not merely a darkening of the moon at
opposition, but a darkening of the moon at opposition due to the
interposition of the earth between moon and sun, depends on a
precise knowledge of this particular object, whereas the dialectician
has a less precise ability that extends to all objects in general. Thus
it is reasonable to say that the preliminary definitions from which the
scientist begins are dialectical definitions, answering ´
’ / mā
huwa in the sense which the dialectician attaches to this question
and to the interrogative particle mā = ´, rather than in the special
scientific sense of the question and the particle. (A ‘‘definition’’ in
this loose sense expresses a ‘‘quiddity’’ in a correspondingly loose
sense, the sense which justifies the loose claim in I,89 that everything
that has being as truth also has some quiddity; when III,240 contra-
dicts this, and says that privation does not have a quiddity, it is using
the scientific sense in which only species of the categories have
quiddities.) So it seems that the procedure of investigation which
Posterior Analytics lays down for the 1-place questions, whether X is
and what X is, will have three stages: first a preliminary dialectical
investigation of what X is (or an investigation of this question in the
dialectical sense of mā), then an investigation of whether X is, and
then (assuming the answer is yes) a properly scientific investigation
of what X is (or an investigation of this question in the distinctively
scientific sense of mā).
However, it will have to be more complicated than this for Fārābı̄.
As we have seen, Fārābı̄ identifies the 1-place question hal X
mawjūdun, in the sense of being as the true, with the question
whether there are any X’s, i.e. whether the concept of X is instanti-
ated. We might naturally identify this with the question
’ ’´
= hal
mawjūdun of Posterior Analytics II, but there is a problem: Posterior
Analytics II says that the cause of the fact that X is is what X is, but
Metaphysics E4 says that being-as-truth has no cause except ‘‘some
a#ection of thought’’ (1027b34–1028a1). Fārābı̄ solves this by bring-
ing his version of the 7 distinction between two senses of being into
Posterior Analytics II as well. The crucial texts are in H * urūf III,228–
31, on the uses of hal (or hal mawjūdun) in the sciences. III,228–9 are
concerned with 1-places uses, hal X mawjūdun, and distinguish two
versions of this question, which can and should be asked in sequence.
The initial question, e.g., whether void exists, is asking whether the
concept of X in the soul corresponds to something outside the soul
(III,228, esp. p. 213,22–3); then, if the answer is yes, we can ask hal X
mawjūdun in a second sense, meaning ‘‘whether this thing has
something by which it is constituted and which is in it’’ (III,229 p.
214,6–7); if the answer to this second question is also yes, we can go
on to ask ‘‘what is it’’ or ‘‘what is its wujūd,’’ asking for that by
which this thing is constituted, which will be a cause of its existing
88 STEPHEN MENN
in the second sense (III,229 p. 214,9–12). (III,230–31 describe the

analogous series of questions in the 2-place case: first hal X mawjū-
dun Y asking whether the subject and predicate are combined
outside the soul as they are in the soul; then, if yes, hal X mawjūdun
Y asking whether the wujūd or constitution or quiddity of X
necessitates that it is Y; then, if yes, the inquiry into the cause of X’s
being and being Y.) Fārābı̄ says explicitly that the first sense of hal
X mawjūdun is asking about being as truth (III,228 p. 213,23–214,3),
and indeed it fits precisely with the account of being as truth in
H* urūf Part One. And it is likewise obvious that the second sense of
hal X mawjūdun is asking about being as having a quiddity outside
the soul as described in H * urūf Part One (‘‘that by which this thing is
constituted,’’ which is its wujūd in the second sense, is what is
signified by its definition, III,229 p. 214,9–11).
Thus Fārābı̄ has split the Posterior Analytics II investigation
’
’´
X into two stages corresponding to the two senses of ’´
from
7 as he interprets them. After giving a ‘‘definition’’ which unpacks
the content of the concept of X, we will first ask whether X has
being-as-truth, i.e. whether the concept is instantiated; this is a
univocal concept, and the question can be asked and answered even
if we do not know what category X falls under, or whether it falls
under any category at all. Then, if the answer is yes, we will go on to
ask whether X has being in the second sense, i.e. whether it has a
quiddity. This is no longer a question about the concept of X, but
about an X really existing outside the soul, asking whether it is X
through some simple positive quiddity (rather than through a
privation or through the combination of two quiddities). This ques-
tion is not univocal, since e.g. substantial quiddities and qualitative
quiddities are not univocally quiddities; rather, there is a question
whether X is X through a substantial quiddity, another question
whether X is X through a qualitative quiddity, and so on. Then, if the
answer to one of these questions is yes, we will go on to ask for the
cause of X’s being, in the sense of being as divided into the
categories, and this cause will be this quiddity. The claim of
Metaphysics E4, that being-as-truth has no cause outside the soul,
can be reconciled with the procedure of Posterior Analytics II by
saying that the quiddity is the cause, not of being-as-truth, but only
of being-as-divided-into-the-categories: the Posterior Analytics ques-
tion
’ ’´
is not simply about being-as-truth, but rather has one
sense for each sense of being.
Thus the Posterior Analytics, as Fārābı̄ reads it, envisages not a
three-stage but a four-stage process. In a 1-place case, there will be
first an investigation of what X is in the dialectical sense of mā, then
an investigation of whether X is in the sense of being-as-truth, then
FA
z RA
* URUzF 89
an investigation of whether X is in the sense of having-a-quiddity-
outside-the-soul, then an investigation of what X is in the scientific
sense of mā. (Analogously in the 2-place case as described in H * urūf
III,230–31, once we have dialectical definitions of X and Y, we
investigate whether X is Y in the sense of being-as-truth, then
whether X is Y per se, then why X is Y: this is an analogy and more
than an analogy, since we investigate 1-place cases of being by
unpacking them as 2-place cases, e.g. investigating whether and
what lunar eclipse is by investigating whether and why darkened-at-
opposition belongs to moon, according to H * urūf III,244.) And since
Fārābı̄ says that dialectic discusses only the ordinary-life sense of
mawjūd, and that this means that something is true, i.e. (in 1-place
cases) that it exists outside the soul (III,247, p. 223,13–21), we can say
that the four stages investigate first what X is in the dialectical
sense, then whether X is in the dialectical sense, then whether X is in
the scientific sense, then what X is in the scientific sense. (Fārābı̄
cites here Alexander of Aphrodisias, who in his Topics commentary
[53,2–10] had noted a dispute about whether in questions like ‘‘do the
gods exist?’’ the predicate is a genus or an accident. As Fārābı̄ notes
[III,246, p. 223,9–12], Alexander himself classifies these as questions
of accident; Fārābı̄ implies that Alexander is right to do so, since in
a Topics commentary he is discussing not scientific questions but
only the questions that arise in dialectical encounters, where we ask
whether X exists only in the dialectical sense.)
Dialectic and science thus proceed in opposite directions, dialectic
from mā to hal or from essence (to the extent that it can be expressed
in a dialectical definition) to existence, while science, picking up
where dialectic leaves o#, proceeds from hal to mā or from existence
to essence. Since science understands the essence of a thing as the
cause of its existence, which we can grasp only after we have grasped
the fact of its existence, the scientific philosopher will not imagine an
essence which does not in itself possess existence, or an existence
which needs to be added to the thing from without, as an accident.
Fārābı̄’s implication is that only someone who confuses the scientific
with the dialectical meanings of hal and mā – or, rather, someone
who is familiar only with the dialectical meanings, and who tries to
do philosophy by representing features of dialectical investigation as
features of reality – will be tempted to regard existence as something
real and really distinct from the essence of the thing to which it
applies. When people think they can abstract away from the exist-
ence of X to grasp the essence of X, the subject of which existence is
predicated, but which is of itself indi#erent to existence and non-
existence, they are not in fact grasping any real thing: they may
formulate a definition of X, independently of whether X exists, but
that definition will be merely a nominal definition, spelling out what
90 STEPHEN MENN
is implicit in the concept of X or what the word ‘‘X’’ means, and not
what the thing X is: ‘‘someone who knows what man, or anything
else, is must first know that it is: for no one knows what a
non-existent thing is, but rather [they know] what the formula or the
name signifies, when I say goatstag; but what goatstag is, it is
impossible to know’’ (Posterior Analytics II,7 92b4–8). If we are really
talking about X, it is not something indi#erent to existence and
non-existence, and so existence cannot be a$rmed of it without
tautology. There must be some way to say ‘‘X exists’’ without
tautology (there must be, in particular, some instances in which ‘‘X
exists’’ can be meaningful but false), but here the subject of the
judgment is not any real thing outside the mind, but rather the
concept of X in the mind, which is what the word ‘‘X’’ immediately
signifies, and the predicate is a$rming of this concept in the mind
that there is something outside the mind that corresponds to it.
There is thus a good Aristotelian genealogy behind Fārābı̄’s Fregean
conclusion about the logical syntax of existence – un-Fregean, of
course, in its understanding of concepts as things in the mind.
V. FA
z RA
z BIz AND HIS OPPONENTS ON BEING AND GOD
If Fārābı̄ takes such care to avert the error of supposing that things
exist through a real wujūd really distinct from their essences, he
must have had real opponents, people who really fell into that error
(as Fārābı̄ saw it). But who?
One possible target is the thesis of many Mu‘tazilite mutakallimūn
that what is non-existent is a thing [al-ma‘dūmu shay’un]: ‘‘when
God wishes to create a thing, he says to it ‘be!’ and it is’’ (Q. 2:117
etc.), so there must be some not-yet-existent thing which God is
addressing (and the results are di#erent when God addresses his
command to a not-yet-existent human or to a not-yet-existent horse),
and that same thing is first in the state of non-existence [‘adam] and
then in the state of existence [wujūd]. The Mu‘tazilites are certainly
a possible target: Fārābı̄ is at pains in the Kitāb al-H * urūf to
distinguish philosophy, which can be demonstrative, from kalām,
which is at best dialectical, and he might welcome the opportunity to
distinguish the philosophical from the kalām analysis of wujūd. But
the philosophers are by no means all agreed on the analysis of wujūd,
and Fārābı̄ will have had opponents within the philosophers’ camp as
well. When Aristotle draws distinctions in senses of being, he too is
trying to avert errors into which he thinks earlier philosophers fell
by not distinguishing: doubtless the Megarians, but in the first
instance Plato and Parmenides, who (as Aristotle represents them)
posit a single being-itself, such that for anything else to exist is for it,
not existent by its essence, to exist by participating in being-itself.
(Parmenides, unlike Plato, drew the conclusion that nothing else
FA
z RA
* URUzF 91
does exist.)37 And the position that Aristotle is opposing had not died
out between his time and Fārābı̄’s, and indeed it was sometimes
treated as Aristotelian. Thus Proclus’ teacher Syrianus in comment-
ing on the Metaphysics and explaining the sense in which metaphys-
ics is a science of being qua being, says that ‘‘beings, qua beings,
proceed from being, which is not this being, but being-itself’’ (In
Metaphysica 45,29–30); the Liber de Causis says that ‘‘all things have
being [huwiyya] on account of the first being, all living things are
moved per se on account of the first life, and all intellectual things
have knowledge on account of the first intellect’’ (92,2–4),38 reflect-
ing Proclus Elements of Theology Proposition 101, where ˜ is prior
to all intelligent things, life to all living things, and being, most
universally, to all things which are or participate in being. Al-Kindı̄,
drawing on the Qur’ānic name of God al-h * aqq but also on Meta-
physics 1 993b23–31, where if X is the cause to Y of its being true, X
is truer than Y, so that the principles of eternal things are the truest
of all things, and where in general the truth of a thing is proportional
to its being, says that ‘‘the cause of the existence [wujūd] and
a$rmation [thabāt] of everything is the Truth [al-h * aqq], since every-
thing that has a being [inniyya] has a truth / reality [h* aqı̄qa]; so the
Truth necessarily exists [mawjūd] for beings [inniyyāt] which exist.
And the most noble part of philosophy and its highest in degree is
first philosophy, I mean the knowledge of the first Truth which is the
cause of every Truth’’ (9,12–14).39 These texts thus seem to be
37
For Aristotle’s critique of the failure of Parmenides (and others following him) to
distinguish the senses of being, see Physics I,2–3 and Metaphysics N2.
38
I cite the Liber de causis from O. Bardenhewer, Die pseudo-aristotelische Schrift Ueber
das reine Gute bekannt unter dem Namen Liber de causis (Freiburg, 1882). The idea that the
more universal things are the higher causes, with being-itself [huwiyya or inniyya] as the
most universal, is crucial to the book, and is first stated right near the beginning, 59,3–60,3.
39
On Kindı̄’s intentions in this passage, and its Aristotelian and neo-Platonic
background, see Cristina d’Ancona, ‘‘Al-Kindi on the subject-matter of the First Philosophy.
Direct and indirect sources of Falsafa al-ūlā, Chapter one,’’ in J.A. Aertsen and A. Speer
(eds.), Was ist Philosophie im Mittelalter? (Berlin and New York, 1998), pp. 841–55. In fact,
for Kindı̄’s project in On First Philosophy, ‘‘being’’ is less important than ‘‘one’’ as the
predicate that God possesses primarily, and other things only derivatively and improperly.
Fārābı̄ seems to center the H * urūf on the concept of being and, surprisingly, does not
include unity among the equivocal transcendental or syncategorematic terms that he
investigates (although it is in his model, Metaphysics 6); so it might seem that if he is
setting out to refute Kindı̄, he is refuting him on a relatively minor point and letting the
major one pass. However, Fārābı̄ gives the missing treatment of unity (in fact at much
greater length than his treatment of being) in his Kitāb al-Wāh * id wa-al-wah* da (edited by
Muhsin Mahdi [Casablanca, 1989]): if the Kitāb al-Wāh * id is not actually a missing part of
the fragmentarily transmitted Kitāb al-H * urūf (as Mahdi suggested), it is at any rate a
necessary complement, needed to complete the argument of the H * urūf – perhaps it was
originally intended as part of the H * urūf but got too long and was spun o# as a separate
treatise. And while the Kitāb al-Wāh * id is not overtly polemical (indeed, at first sight it
looks rather boringly expository), it can be shown that it is in fact drawing the conceptual
distinctions between senses of unity needed to overthrow the Kindian project (and that
Averroes correctly appreciated this, and treated it as continuous with the H * urūf). I will
discuss this in detail in my monograph.
92 STEPHEN MENN
licensing, in the name of Aristotelian first philosophy, an inference

from the many things-that-are to a single first being-itself in which
they participate in order to exist (or a truth-itself in which they
participate in order to be true and thus to exist).
It seems likely, then, that Fārābı̄ wishes to restate what he sees as
the correct Aristotelian scientific understanding of being, and dis-
tinction of its several senses, against Kindı̄ and his circle, including
the authors / adaptors of such texts as the Liber de Causis, whom he
wishes to expel from the history of scientific philosophy.40 This
likelihood is strengthened by the fact that the Kindians, like Fārābı̄
in the H* urūf, make heavy use of particles and of names derived from
particles in order to signify metaphysical notions including being.
The Kindians use, of course, inniyya (or anniyya) and huwiyya and
māhiyya (sometimes mā’iyya) and the standard category-names, but
also several terms which did not survive into the standard Arabic
philosophical lexicon: from huwa the peculiar tahawwı̄ (apparently a
fifth-form mas*dar, ‘‘being caused to be,’’ Kindı̄ On First Philosophy
41,5); from laysa ‘‘is not’’ (grammatically a defective verb, but
classified by Fārābı̄ as a particle, Alfāz* 45,11–12) the nouns lays
‘‘not-being’’ and the back-formation ays ‘‘being’’;41 and the abstracts
from interrogative particles haliyya = existence, ayyiyya = specific
di#erentia (from ayy ‘‘which’’), and limiyya = cause (from lima
‘‘why’’ = ` ´ ). In the prologue to On First Philosophy Kindı̄
distinguishes four scientific questions, hal, mā, ayy and lima, asking
respectively for the existence [inniyya], the genus, the specific
di#erentia, and the final cause (11,4–9). S.M. Stern has shown that
this list and its variants (the list is very common and almost standard
in the Kindı̄ school, with a-mawjūd sometimes substituted for
hal and kayfa for ayy), which are accompanied by some quite
un-Aristotelian interpretations of what the questions are asking
40
Fārābı̄ in On the Appearance of Philosophy (in Ibn Abı̄ Us*aybi‘a, ‘Uyūn al-anbā’ fı̄
*tabaqāt al-at*ibbā’, ed. A. Müller [Cairo and Königsberg, 1882–4], vol. 2, pp. 134–5; English
translation in Franz Rosenthal, The Classical Heritage in Islam [London, 1975], pp. 50–1)
keeps complete silence on Kindı̄ (as well as on Abū Bakr al-Rāzı̄), representing himself, a
Muslim student of Christian teachers, as the person who brought scientific philosophy into
Islam. But of course he knew of Kindı̄’s existence, and he names him in one text that I am
aware of, the Kitāb Ih **sā’ al-ı̄qā‘āt, which attacks Kindı̄ for wrongly tracing certain rhythms
in Arabic music back to the Greek philosophers. There is a summary of the treatise in
Amnon Shiloah, ‘‘Traités de musique dans le Ms. 1705 de Manisa,’’ Israel Oriental Studies, 1
(1971): 305–7; extracts were subsequently published by Muhsin Mahdi, in Nus*ūs* falsafiyya
muhdāt ilā al-Duktūr Ibrāhı̄m Madkūr, ed. ‘Uthmān Amı̄n (Cairo, 1976), pp. 75–8. I am
grateful to Badr el-Fekkak and Ahmed Hasnaoui for bringing this text to my attention and
sending me copies of these publications.
41
The word ‘‘ays’’ is apparently Kindı̄’s invention based on an etymology of laysa as
lā + aysa, but the etymology is correct, and Kindı̄ is likely to be influenced by the Syriac
ı̄th, which is (I am told) not a verb but a particle or noun working like the Hebrew yesh
(‘‘yesh of X’’ = ‘‘existence of X’’ = ‘‘X exists’’); the Arabic, Syriac and Hebrew words are
apparently all cognate.
FA
z RA
* URUzF 93
for,42 are taken not directly from the list of four scientific questions
in the Posterior Analytics, which Kindı̄ and his circle did not have
access to, but rather from a list of questions in the sixth-century
Alexandrian prolegomena to philosophy (surviving in the versions of
Elias and David),
’ ’´
, ´
’ , ¢ ˜´ ´
’ and ` ´
’ .43
Fārābı̄ in Kitāb al-H* urūf Part Three is taking up this Kindian list
of particles and their meanings and correcting it from Posterior
Analytics II; in particular, he wants to distinguish the scientific from
the dialectical meanings of these particles, and to set out what he
sees as the Aristotelian scientific approach to essence and existence,
founded on the analogy ´
’ :
’ ’´
::´ ::¢´ . Fārābı̄ is accepting
from the Kindians that particles can signify metaphysical things and
are indeed the grammatical type best suited to signifying metaphysi-
cal things, but it is crucial for him to distinguish the metaphysical
sense, in which (for instance) astı̄n and its Arabic equivalents signify
a first intention transcending the categories, from the logical sense
in which they signify a mere second intention. And, in the process, he
wants to block an unscientific metaphysics, and an over-hasty
solution to the problem of the ontological and theological descrip-
tions of metaphysics, that turns on saying that the wujūd in which all
things participate is something both real (like wujūd from mawjūd as
having an essence) and extrinsic to the things (like wujūd as truth),
namely God.
At this point it might be asked how di#erent Fārābı̄’s own
metaphysics really is. After all, he opens the Principles of the
Opinions of the People of the Perfect City by saying that ‘‘the first
being [mawjūd] is the first cause of being [wujūd] to all other beings
[mawjūdāt]’’ (p. 56), which sounds close to the formulations we have
seen from Syrianus through Kindı̄. (One might even cite this as
evidence that the ontotheology of the Perfect City is merely ‘‘exo-
teric,’’ by contrast with the scientific and non-theological metaphys-
ics of the H * urūf.) However, it can be shown (although I will not do
this in detail here) that the Perfect City is remarkably scrupulous in
applying the terminological and conceptual clarifications of the
42
Sometimes mā asks for the genus and ayy for the di#erentia; sometimes mā asks for the
whole genus-di#erentia definition and kayfa for the ’´ . Lima is always taken to ask for
the final cause; sometimes mā (asking for the genus) is associated with the material cause
and kayfa (asking for the di#erentia) with the formal cause, sometimes mā (asking for the
definition) with the formal cause, hal with the material cause and kayfa with the e$cient
cause.
43
See Stern in A. Altmann and S.M. Stern, Isaac Israeli (Oxford, 1958), pp. 10–23. In the
Alexandrian list of questions, ¢ ˜´ ´
’ might be translated either as ayy or as kayfa; by
` ´
’ the texts do indeed mean the purpose. The Alexandrians list these as
introductory questions which one should ask about philosophy itself, before beginning its
study proper, but this is clearly a general scheme which is being applied to the case of
philosophy. These four questions are all questions that should be asked, in series, about a
term X; the Posterior Analytics’ distinction, and analogy, between the 1-place questions
’
’´

and ´
’ and the 2-place questions ¢´ and ´ has been lost.
94 STEPHEN MENN
H
* urūf, and can in fact be seen as carrying out a metaphysical
program from the H * urūf. It is perfectly Aristotelian to investigate the
cause of the existence of a thing, and while in a sense the essence of
X is the cause of the existence of X, that essence, as expressed in the
scientific definition of X, may contain references to things outside X,
as the definition of lunar eclipse contains a reference to the earth as
the cause blocking the sun’s light from the moon. Any science takes
for granted (as given by the senses or by hypothesis) the existence of
its primitives (e.g. for geometry, point, straight line, circle) but
investigates and tries to prove the existence of everything else that
falls under its domain (e.g. regular pentagons) by deriving their
existence from that of the primitives; and, at least on Fārābı̄’s
understanding of Aristotle (supportable by Metaphysics E1 1025b2–
18), what is treated as primitive in the particular sciences is no
longer primitive from the point of view of the universal science of
metaphysics, and its essence and existence can be investigated. If X
turns out not to be a primitive, but rather is caused to exist by Y, in
such a way that the scientific definition of X contains a reference to
Y, then Y will be a cause of wujūd to X, and Y will be mawjūd in a
stronger sense than X is. Thus H * urūf I,92 says that mawjūd in the
sense of having a quiddity outside the soul is said per prius et
posterius; and
what is more perfect with respect to its quiddity and independent of other
things in order for its quiddity to be realized, while other things require this
category in order for their quiddity to be realized and to be understood, this
is more deserving than the other things that it should be, and should be said
to be, mawjūd. Next, what in this category [i.e. substance] requires a
di#erentia or genus in this category in order for its quiddity to be realized
is more deficient with respect to quiddity than the thing in this category
which is a cause of the realization of its quiddity. And among the things in
this category which are causes by which the quiddity is realized, some are
more perfect than others in respect of quiddity and more deserving to be
called mawjūd. And one continues in this way to ascend in this category to
the more and more perfect in respect to quiddity until one reaches what in
it is most perfect in respect to quiddity, such that there does not exist
anything in this category more perfect than it, whether this is one or more
than one. And this one thing, or these things, are more deserving than other
things to be said to be mawjūd. And if something is discovered outside all
these categories [as God is for Fārābı̄] which is the ur-cause [musabbib] of
the quiddity’s being realized, which is prior to anything in this category,
this is the cause in the quiddity of the other things in this category, and
what is in this category is the cause in the quiddity of the other categories.
Thus the beings, where what is meant by ‘‘being’’ is what has a quiddity
outside the soul, are ordered by this order. (I,92, pp. 118,16–119,8)
Thus Fārābı̄ takes the Aristotelian ontological dependence of acci-
dents on substances (which has the result that being is said primarily
FA
z RA
* URUzF 95
of substances and derivatively of accidents) as corresponding, in the
universal science of being, to the causal-definitional dependence of
eclipse on moon and sun and earth in astronomy; and he thinks that
we can trace such causal-definitional dependence back further, not
just from accidents to substances, but also from ordinary substances
to an ultimate simple, or ultimate simples, and that only these will be
mawjūd (in the sense of having a quiddity) in the primary sense. And,
as we have seen from H * urūf I,89, wujūd corresponding to mawjūd as
having-a-quiddity can be distinguished from its bearer if the mawjūd
is a complex and the wujūd is the articulated quiddity expressed in
the definition or some special part of that articulated quiddity, but in
the case of a simple the mawjūd and its wujūd are absolutely
identical.
Fārābı̄ takes this point up again in H * urūf Part Three, with
application specifically to metaphysics or divine science and to God.
We can ask whether God exists in the sense of being-as-truth, just as
we can ask whether man exists in this sense, i.e. whether the
concepts are instantiated, but in the case of man we can further ask
whether (unlike the not-white) he also exists in the sense of having
a real positive quiddity by which he is constituted, which would be
his wujūd; but since God has no cause by which he is constituted, and
cannot be scientifically defined, does it follow that God (like the
not-white) does not exist in this second sense (so III,239)? Fārābı̄
answers that God too exists in the second sense, not in that he has a
quiddity or wujūd, but in that he is a quiddity or wujūd: even if we
begin with a complex, we must ultimately reach something simple
which is its own wujūd, on pain of an infinite regress (III,240–41). We
may of course, ask whether God exists in the sense of asking whether
there is some cause through which he is constituted, but in this
sense, the answer is no (III,243). Fārābı̄ says that we must start by
investigating in this way, in order to clear away imperfections from
God, and thus to enable a grasp of God’s positive wujūd: ‘‘what exists
[mawjūd] absolutely [‘alā al-it*lāq] is the mawjūd which is not related
to anything at all. And what exists absolutely is the mawjūd whose
wujūd is only through itself and not through anything else other
than it; and our saying about it ‘does it exist [hal huwa mawjūdun]?’
is in this sense. And what is sought / investigated here is the contrary
of what is sought / investigated in our saying ‘does man exist?’: for
what is sought / investigated in our saying ‘does man exist?’ is
whether man has a constitution through something else or not,
whereas here what is sought / investigated in our saying ‘does it
exist?’ is whether it is something whose constitution is through itself
and not through something other than itself, and whether its wujūd
is a wujūd that does not require, in order for it to be mawjūd through
it, anything else which is in any way other than itself’’ (III,242, pp.
96 STEPHEN MENN
219,17–220,2). Fārābı̄ adds the analogous point about investigating

2-place being with respect to God, that asking e.g. ‘‘is he [hal huwa
mawjūdun] intellect’’ means not asking whether there is some middle
term that connects him with intellect, but ‘‘whether the wujūd
through which his constitution is not through something other than
he is that he is intellect . . . and whether his essence [dhāt] is that he
is intellect’’ (III,242 p. 220,2–4); likewise, to ask ‘‘is he [hal huwa
mawjūdun] an agent or cause of the wujūd of what is other than
himself’’ (p. 220,5) is not to ask whether he acts through some
attribute or some act superadded to his essence, but rather whether
his very wujūd or quiddity necessitates that he is the agent or cause
of the existence of other things.
This is clearly a charter for the theology of the Perfect City, but it
is also clearly founded on the theory of the meanings of mawjūd and
wujūd in H * urūf Part One. And when Fārābı̄ says that we investigate
the essence and other predicates of simple things by investigating
whether they are simple things, denying everything that would entail
composition, and seeing what remains, he is ultimately following
Aristotle’s Metaphysics: ‘‘there is no investigation or teaching in the
case of simples, but rather a di#erent mode of investigating such
things’’ (Z17 1041b9–11); ‘‘those things which are just a being [¢´

˜’ ] and an actuality, about such things there is no being
deceived, but rather either thinking them or not; but the ´
’ is
investigated about them, whether they are such or not’’ (10
1051b30–33) – that is, the only way to ‘‘investigate’’ such a simple
substance, beginning with a nominal definition like ‘‘the mover of
the daily motion,’’ is to investigate whether it is such a simple
substance, and, if the answer is yes, the investigation ceases. What
results, at least as Fārābı̄ develops it, is an ontotheology, but it is an
ontotheology very di#erent from that of Syrianus and Proclus and
the Liber de Causis, on which God (or for Syrianus and Proclus the
second principle, the
¢` ’´
) is a universal form of being, causing
being to all things immediately, and therefore being prior to the less
universal causes which communicate more particular attributes to
particular ranges of beings. For Fārābı̄ God is the immediate cause
only of the intelligence of the outermost heaven; he is not universally
related to all beings equally, but is the end-result of a step-by-step
retracing of things, via the heavens, back to their first causes. God is
his own wujūd, but if there is any sense in which he is the wujūd of
other things, it is only in the sense in which earth might be called the
wujūd of lunar eclipse because it is mentioned in the causal defini-
tion of lunar eclipse, and has nothing to do with the existence shared
by all things, which for Fārābı̄ can only be the second intention
being-as-truth.
FA
z RA
* URUzF 97
Avicenna, of course, is deeply influenced by Fārābı̄’s understand-
ing of Aristotle’s Metaphysics and of metaphysics as a science. But he
may also have taken more from the tradition of Proclus and the Kindı̄
circle than Fārābı̄ was willing to. This is a very complicated issue.
But what is certain is that Averroes, reading Avicenna and reading
the Kitāb al-H* urūf, thought that Avicenna was guilty of the same
errors that Fārābı̄ had already exposed, and that only a confusion
between mawjūd in the sense that is divided into the categories and
mawjūd in the sense of the true, into which Avicenna was led by his
reliance on systematically misleading Arabic translations, could
have led him to conclude that the wujūd of things other than God is
something real and univocal and extrinsic to the essences of the
things. Averroes makes deep use of the H * urūf, and of the closely
related Kitāb al-Wāh * id wa-al-wah
* da, in his Epitome of the Metaphys-
ics, Long Commentary on the Metaphysics, and Tahāfut al-Tahāfut,
especially against Avicenna; and in so doing he helped ensure a
future for Fārābı̄’s analysis of the senses of being, and for the critical
reflection on the concept of being which Fārābı̄ thought was the
necessary first step in any scientific metaphysics. I will try to explore
this history, in Averroes and in Thomas Aquinas and others who are
concerned to respond to Averroes and thus (whether they know it or
not) to Fārābı̄, in my monograph in progress, as well as exploring
more deeply the connections and tensions between the Kitāb al-
H
* urūf and Fārābı̄’s other writings. I hope to have said enough here
to show that the H * urūf was intrinsically very important, as a
contribution to metaphysics and specifically ontology, as well as to
the reception of Aristotle’s Metaphysics (recentered on as well as
the more obvious ) and more generally to the retrospective imagi-
nation of Greek culture and language and to the elaboration of the
idea of an ideal language and of the tensions between logical and
grammatical form. I think it can also be shown that the H * urūf was
not a dead end, but rather opens a path to Fārābı̄’s other writings
and to those of his Arabic and Latin successors.44
44
For a first sketch of what Averroes and Thomas did with the H
* urūf, see my review of
Anthony Kenny, Aquinas on Being (Oxford, 2002), in Philosophical Review, 115.3 (July
2006): 391–5.
AVICENNA’S THEORY OF PRIMARY MIXTURE
ABRAHAM D. STONE
Department of Philosophy, University of California, Santa Cruz, Cowell
Academic Services, 1156 High St., Santa Cruz, CA 95064, USA
Email: abestone@ucsc.edu
Abstract: Ancient Peripatetics and Neoplatonists had great di$culty coming up

with a consistent, interpretatively reasonable, and empirically adequate Aristote-
lian theory of complete mixture or complexion. I explain some of the main problems,
with special attention to authors with whom Avicenna was familiar. I then show
how Avicenna used a new doctrine of the occultness of substantial form (whose
roots are found in Alfarabi) to address these problems. The result was in some
respects an improvement, but it also gave rise to a new set of problems, which were
later to prove fateful in the history of early modern philosophy.
Résumé: Les anciens Péripatéticiens et les Néoplatoniciens ont éprouvé de grandes

di$cultés à mettre sur pied une théorie aristotélicienne du mélange total, ou de la
complexion, qui soit à la fois consistante, herméneutiquement raisonnable et
empiriquement adéquate. J’explique quelques-uns des problèmes principaux, en
accordant une attention spéciale aux auteurs familiers d’Avicenne. Je montre
ensuite comment Avicenne utilise une nouvelle doctrine du caractère occulte de la
forme substantielle (dont les racines sont à rechercher chez Alfarabi), pour traiter
ces questions. Le résultat atteint représentait, sous certains aspects, un progrès,
mais il donnait aussi naissance à une nouvelle série de di$cultés, qui scelleront son
destin dans l’histoire de la première philosophie moderne.
Avicenna’s views on mixture have attracted a certain amount of

attention, both in the Middle Ages and more recently, for several
reasons. First, the nature of mixture is a fundamental issue in
Aristotelian / Neoplatonic physics, as well as in Galenic medicine:
according to both of those closely related traditions, most if not
all bodies in the universe are mixtures of di#erent ingredients
– ultimately, of the elements. Second, as had become evident already
in antiquity, the issue is not only fundamental but very di$cult.
The later ancient accounts all have significant problems, whether
conceptual, empirical, interpretative or (most commonly) all three
at once. Third, Avicenna’s approach to the problem, whatever
exactly the details of it, is clearly quite di#erent from those of his
100 ABRAHAM D. STONE
predecessors or successors. Hence in catalogs of views on mixture,

Avicenna tends to appear as a singleton.1
I would also like to add a fourth reason for interest, which has been
less appreciated: namely, that Avicenna’s innovation in this area
depends upon his radical moves in others, especially on the relation-
ship between substantial form and sensible quality. In particular,
Avicenna seems to have been the first to maintain clearly that no
sensible characteristic is ever essential to a substance – i.e., that the
true di#erentiae of substances are always occult. Since this latter
view was vastly influential on later thought, we should welcome any
light that can be thrown on Avicenna’s understanding of it, and on
his motivation for adopting it. I will suggest that problems about
mixture played a major role.2
In general, mixtures can be divided into two types: primary
mixtures, in which the ingredients are simple (elemental) bodies,
and secondary mixtures, in which the ingredients themselves are
composite. I will focus on primary mixture here; I hope to take up
Avicenna’s views on secondary mixture (and higher levels of
composition in general) on another occasion.
I. TERMINOLOGICAL PRELIMINARIES
At least two Greek words are naturally translated as ‘‘mixture’’:

´ (or ˜), from the verb ´ , and
˜ , from
´ .
The Arabic translators associated these two verbs, respectively, with
the roots kh-l-t* and m-z-j. Hence ´, with rare exceptions, is
translated by a term such as khalı̄t* or ikhtilāt*,
˜ by mizāj or
1
See A. Maier, ‘‘Die Struktur der materiellen Substanz,’’ in An der Grenze von Scholastik
und Naturwissenschaft (Rome, 1952); F.A.J. de Haas, ‘‘Mixture in Philoponus: An encounter
with a third kind of potentiality,’’ in J.M.M.H. Thijssen and H.A.G. Braakhuis (eds.), The
Commentary Tradition on Aristotle’s De generatione et corruptione: Ancient, Medieval and
Early Modern (Turnhout, 1999), pp. 21–46; and the earlier sources cited by each.
Below I will cite Plato’s and Aristotle’s works, using standard abbreviations, from the
Oxford Classical Texts editions, except De generatione et corruptione, ed. Harold H. Joachim
(Oxford, 1922). Page citations from Galen’s works are given in the form xK / y / z, where x is
the page number in the appropriate volume of Claudii Galeni Opera omnia, ed. D.C.G. Kühn
(Leipzig, 1821–33), y is the page and line numbers in the Greek edition I have used, and z is
the page and line numbers in the Arabic version I have used, if any. Abbreviations and
details of the editions used will be given at the first citation of a given work.
CAG abbreviates Commentaria in Aristotelem Graeca.
Avicenna’s Shifā’ is cited as follows. Sh. M. = al-Shifā’: al-Mant*iq, Sh. T * . = al-Shifā’:
al-T
* abı̄‘iyyāt, Sh. Il. = al-Shifā’: al-Ilāhiyyāt, all ed. Ibrahim Madkour et al. (Cairo, 1950–).
In the case of the T * abı̄‘iyyāt and the Mant*iq, I cite the part (fān) by a short version of its
name, e.g. Simā‘ = al-Simā‘ al-t*abı̄‘ı̄, Samā’ = al-Samā’ wa-al-‘ālam.
2
The only other author I am aware of who emphasizes this connection is H. Eichner, in
the exhaustive ‘‘introductory study’’ to her edition of Averroes’ Middle Commentary on the
GC (Averroes’ Mittlerer Kommentar zu Aristoteles’ De generatione et corruptione [Paderborn,
2005]), pp. 162–87. (Below I cite this work as ‘‘Einleitende Studie,’’ as opposed to the actual
text of Averroes’ commentary, which is separately paginated, and which I cite as M.C. in
GC.)
AVICENNA’S THEORY OF PRIMARY MIXTURE 101
imtizāj, and so forth. Latin translation traditions are less clear:
although forms of ´ / kh-l-t* correspond fairly reliably to forms
of miscere (or commiscere, etc.), the translation of
´ / m-z-j
is much less stable. When a distinct term is used at all, it is per-
haps most often complectere, but others – especially temperare and
confundere – are also common. Here I will reserve ‘‘mixture,’’ and
forms of the verb ‘‘to mix,’’ for forms of ´ / kh-l-t*, and translate
forms of
´ / m-z-j using ugly, but unambiguous, quasi-
neologisms such as ‘‘complexion’’ and ‘‘complect.’’3
As for the meaning of these terms: in GC 1.10, Aristotle contrasts
‘‘composition’’ (´ = tarkı̄b), in which, small, possibly invis-
ible, pieces of the ingredients remain in their original state, with
‘‘mixture,’’ in which the small pieces, having first been composed (or
‘‘apposed,’’ ´ ), are altered and unified by mutual action
and passion, such that the result is homoeomerous: every part is of
the same kind as the whole.4 Although ´ is the main term used in
this discussion,
˜ and related terms are used several times,
apparently as equivalent to ´: see especially 328a8–9, 12. In two
other passages, however, these terms are related di#erently: at Top.
4.2.122b25–31, ‘‘mixture’’ is the genus of which ‘‘complexion’’ is a
species – namely, to all appearances, the very species which is called
both ‘‘mixture’’ and ‘‘complexion’’ in GC 1.10; similarly, at De sensu
3.440a31–b17, mixture is divided into two species: one in which the
smallest parts of the ingredients are merely ‘‘apposed,’’ and another
in which they are ‘‘totally’’ ( ¢´ ´ ◊ ´ ) mixed.5 Later
Greek authors derived their standard terminology from the latter
two passages: namely, ‘‘mixture’’ as the genus, with ‘‘complexion’’
and ‘‘apposition’’ as its species.6 The definition of ‘‘complexion,’’
3
‘‘Temper’’ would actually be preferable, since complectere properly translates ´

(and confundere ˜). But it is not really a usable English translation.
4
GC 1.10.328a5–15, b14–22. For ´ see 328a33. It is di$cult to determine an
Arabic equivalent to this term, but it appears that ◊˜ ’ ’´ at GC 1.10.327b34 was
translated as yaqa‘ ba‘d* uhā ilā jānib ba‘d
* (see M.C. in GC, c. 85 [ad 327b31–28a5], 70,17–18,
and see the Arabo-Hebrew translation, La Traduzione arabo-ebraica del De generatione et
corruptione di Aristotele, ed. A. Tessier [Rome, 1984]: yasumu qez*atam ‘al z*ad qez*atam).
5
Unfortunately the Arabic version of the De sensu, like that of GC 1.10, does not survive.
6
See Alexander, De mixtione 13.228,27–36 (cited from R.B. Todd, Alexander of
Aphrodisias on Stoic Physics: A Study of the De mixtione with Preliminary Essays, Text,
Translation and Commentary [Leiden, 1976]) and Philoponus, In GC (CAG 14.2, ed. M.
Hayduck [1901]), 1.2, 22,23–7, 1.9, 187,22–5, and see also Plotinus, Enn. 2.7.1.4–8 (cited from
Plotini Opera, ed. Paul Henry and Hans-Rudolf Schwyzer, 3 vols., Oxford Classical Texts
[1964–83]) and Galen, De temperamentis libri iii, ed. G. Helmreich (Stuttgart, 1969), 1.9.562–
3K / 34,5–16 and 564K / 35,1–2. The De mixtione was apparently unknown in Arabic,
although Alexander may well have said something similar in his GC commentary, the
surviving fragment of which is, in fact, preserved only in Arabic (see the English
translation, Alexander of Aphrodisias: On Aristotle On Coming-to-Be and Perishing 2.2–5, tr.
Emma Gannagé [London, 2005]). Philoponus In GC was available, although no Arabic
version is extant; Avicenna mentions it by title in his correspondence with al-Bı̄rūnı̄ and
clearly implies that he has read it. See Al-As’ila wa-al-ajwiba, ed. S.H. Nas*r and M.
Moh * aqqeq (Tehran, 1974), 13,7–9, and see the discussion in D. Gutas, Avicenna and the
however, tends to be based on GC 1.10 – i.e., on what Aristotle gives

there as the definition of ‘‘mixture.’’7
Avicenna’s terminology is based on this later ancient usage – in
fact, he explicitly declares his preference for it, while mentioning the
other possibility.8 More importantly for present purposes, he holds
that primary composition always results in a homoeomer, or in
other words is always complexion, always mizāj.9 And the problems
he faces are continuous with those which bothered his Greek
predecessors with respect to
˜ .
II. THE ANCIENT PROBLEMS OF COMPLEXION
It is important to realize that there was not just one such problem,
but rather a whole nest of interrelated ones. All had to do with the
relationship between: (1) the bodies or substances of the elements;
(2) the primary qualities (hot, cold, moist, and dry); (3) the unified,
homoeomerous body of the complex; and (4) the properties of the
complex (which, in addition to qualities of medium heat / coldness
and medium moisture / dryness, might include secondary sensible
qualities, such as colors or odors, as well as faculties such as mag-
netism or dormitivity10 ). This multi-way relationship was multiply
problematic.
First, there is the distinction between (1) and (2) – if, indeed, there
is any such distinction to be made. This issue particularly exercises
Galen, because Hippocratic texts sometimes speak of bodies as
complexions of ‘‘powers’’ ( ´ ) – among which are the primary
qualities – without mentioning any species of body in which they
inhere,11 and this was taken literally by some of Galen’s opponents.12
For Aristotle, however, the elements are fire, air, water, and earth,
Aristotelian Tradition: Introduction to Reading Avicenna’s Philosophical Works (Leiden,
1988), pp. 289–90. (I do not understand Eichner’s skepticism about this, ‘‘Einleitende
Studie,’’ pp. 184–5.)
7
See GC 1.10.328b22 and cf. Alexander, De mixt. 14.231,10–12; Galen, De el. 9.490–91K /
138,1–14 / 115,5–116,2.
8
See Sh. T* ., Kawn / fasād, 6, 126,18–127,3, and see Eichner, ‘‘Einleitende Studie,’’ pp. 163–4,
who reaches a similar conclusion. At 127,11, corresponding to GC 1.10.327b24, Avicenna
actually ‘‘corrects’’ Aristotle’s ` ´ (which, as is clear from Averroes, M.C. in GC
c. 84 [70,2], was translated in the standard way as al-mukhtalat*ayn) to al-mumtazajāt
– unless, indeed, this is a sign that Avicenna used a di#erent translation.
9
Sh. T
* ., Af‘āl / infi‘ālāt, 2.2, 266,4–6.
10
The dormitivity of opium, later made famous by Molière, is discussed by Galen at De
temp. 1.7.585–6K / 48,20–49,16. Magnetism (and electricity) are discussed at De el. 14.507–
8K / 156,10–11 / 138,7 and De nat. fac. (in Scripta minora, vol. 3, ed. G. Helmreich [Leipzig,
1893]), 2.3.85K / 162,18–20, 2.7.106K / 178,5–7. Both of these examples are important for
Avicenna, and we will return to them below.
11
See e.g. De prisc. med. (De l’ancienne médecine, ed. J. Jouanna [Paris, 1990]), 4.137,6–8,
9.144,5–7, and see the discussion in Hippocrates: On Ancient Medicine, ed. M.J. Schiefsky
[Leiden, 2005], pp. 252–4, 275, 288.
12
Athenaeus of Attalia and his followers: see De el. 6, and see De Lacey’s discussion,
pp. 185–7.
which he describes as species of body and of substance, while hot,
cold, moist and dry are species of quality,13 which as such can never
be found without substance.14 Thus, Galen assures us that the
Hippocratic way of speaking merely involves naming the elements by
the qualities which they have in the extreme; there cannot, strictly
speaking, ever be qualities without underlying bodies.15
Given the distinctness of (1) and (2), however, there arises a
di$cult problem about the relationship between them. There must be
some relationship, since Aristotle derives the number of elements
from the permissible combinations of the primary qualities.16
Alexander gives what seems to be the standard Peripatetic under-
standing of this: an element – for example, fire – consists of extreme
primary qualities – in this case, heat and dryness – in matter.
Extreme heat and dryness are the di#erentiae of fire, and hence its
essence (` ˜’ ◊˜ ` ´ ), but its being includes also the
matter.17 For Galen, similarly, the ultimate elements are composed of
matter and qualities;18 when extreme heat comes into matter, then
‘‘that body will have become an element.’’19 But, although this agrees
well with Aristotle’s statements about the elements in particular,
and also with his general statements to the e#ect that corporeal
substances are di#erentiated by sensible qualities,20 it nevertheless
runs into two serious di$culties. First, it makes sensible substances
into mere collections of accidents in matter, contradicting both
(what was taken as) Aristotle’s definition of accident (that it is in its
subject not as a part21 ) and his doctrine of the priority of substance
– in particular, that a substance cannot be made up out of acci-
dents.22 Second, if the being of, e.g., fire, consists of extreme heat,
extreme dryness and matter, then what accounts for fire’s corporeity,
i.e. for it’s being essentially a species of body?23 In response to the
13
See e.g. Metaph. 3.5.1001b32–2a2, 7.2.1028b8–11, 5.14.1020b8–12; Cat. 8.9a28–31; GC
1.3.319a15–17. At GC 2.3.330a 30–b3, Aristotle refers to the primary qualities as ‘‘elements,’’
but nevertheless makes a clear distinction between them and the bodies of fire, air, water
and earth. A.L. Peck, in his introduction to the Loeb edition of PA ([1983], pp. 30–31),
suggests that Aristotle uses ´ to refer to the elements at 2.1.646a15 (an interpretation
echoed by De Lacey in the introduction to De el., p. 187). For alternative understandings of
this passage, see below, pp. 105 and 112.
14
See Cat. 5.2b3–6; GC 1.5.320b24–5; 1.10.327b20–22. And see, following this, Galen, De el.
6.474K / 120,7–8 / 94,3; 9.479K / 124,19–22 / 100,4–5, and Alexander, De mixt. 13.228,13–16.
15
De temp. 1.6.542K / 21,18–19; 1.7.552K / 27,17–22; De el. 6.457K / 102,1–7 / 70,13–71,4; see
also De el. 7.476–7K / 122,4–14 / 96,6–97,2.
16
See again GC 2.3.330a30–b7.
17
De mixt. 13.229,30–230,5 (and see GC 2.1.329a10–13, 24–35; Ph. 1.6.189a32–b1).
18
De el. 8.480K / 126,8–9 / 101,10.
19
De el. 9.481–2K / 128,6–7 / 104,3–4; see also, similarly, 6.469–70K / 114,16–20 / 86,4–6.
20
Cael. 3.4.302b30–3a3; GC 1.1.314b17–20, 1.3.318b16–18.
21
Cat. 2.1a24–5.
22
Ph. 1.6.189a33–4. See also Metaph. 7.1.1028a31–3; Top. 6.6.144a23–7.
23
In all fairness, Alexander and Galen may, like the Stoics before them and like Philoponus
later on, not make a distinction between matter and (qualitiless) body (see GC 1.5.320b14–17).
first problem, later Neoplatonic commentators accepted one version

or another of Porphyry’s theory of ‘‘substantial quality’’ – i.e. that
heat in fire, for example, though indeed a quality, is not an accident,
precisely because it is essential to and ‘‘completive’’ of a substance.24
Responses to the second problem were far more complicated, in part
because Aristotle in some places defines body as a species of
quantity,25 and in part because of Plato’s view of the elements as
composed of triangles – a view which he defends by appeal to the fact
that they are species of body,26 and which Aristotle attacks.27 There
is no room to go into the details here.28 Important for our purposes is
only that, in some way or other, although the elements are di#eren-
tiated by extreme primary qualities, their being or substance as a
whole involves, in addition to these qualities and matter, also
corporeity, hence bulk and three-dimensional extension.
It is in this context that we should understand the dispute between
the Stoics and the Peripatetics over the relationship between (1) and
(3) above. As both Galen and Alexander report it, the Stoics
maintain that complexion is according to the entire substance of the
ingredients, which, given that complexion is by definition complete
and total mixture, means that, in this case, di#erent bodies are
completely extended through one another. But Aristotle says that
bodies cannot extend through each other,29 and so the Peripatetics
hold, in contrast, that complete and total mixture is of qualities only,
not of substances, i.e. bodies.30 This Peripatetic doctrine (with which
Galen also agrees) is then to be understood as follows. The primary
qualities not only di#erentiate the elements; they also are or confer
the power to change the qualities of another substance in such a way
24
For extensive discussion of the background and variants of this view, see F.A.J. de
Haas, John Philoponus’ New Definition of Prime Matter: Aspects of its Background in
Neoplatonism and the Ancient Commentary Tradition (Leiden, 1997), pp. 180–250, although I
do not agree with everything he says there. See also my own brief discussion, ‘‘Simplicius
and Avicenna on the essential corporeity of material substance,’’ in R. Wisnovsky (ed.),
Aspects of Avicenna (Princeton, 2001), pp. 87–8.
25
Cat. 6.4b24; Cael. 1.1.268a6–8; Metaph. 5.6.1016b27–8.
26
Tim. 53c4–d6.
27
Cael. 3.1, 4.2.
28
For some discussion, see de Haas, New Definition, ch. 2, and again my ‘‘Simplicius and
Avicenna.’’ However, I must correct one claim which I make there (91), namely that Simplicius
never speaks of qualitiless body as ‘‘second matter’’ or ‘‘second subject.’’ He does, in fact, say
the latter – not in the passages from the Physics and Categories commentaries on which I based
myself there, but elsewhere, in his De Caelo commentary (CAG 7, ed. I.L. Heiberg [1894], 3.1,
565,2–3; 3.2, 599,4–5). I must now admit that I find this point of Simplicius’ view obscure.
29
GC 1.5.321a 8–9, b15–16.
30
See Galen De el. 9.489K / 136,15–18 / 113,14–114,3: the disagreement is how ’ ¢´

´ `
´
˜ ’ ¢´ = tamtazaj al-ashyā’ allatı̄ tamtazaj: whether ˜
´ ˜ [sc. ’ ’ ´ ’˜ ] = bi-‘amal kayfiyātihā faqat* ba‘d* fı̄ ba‘d
* , or ˜

˜ ’´ ’ ’ ´ ’˜ = bi-tatafarrad jawāhiruhā ba‘d* fı̄ ba‘d
* , wa-mudākhila
ba‘d
* li-ba‘d* . (Note that H * unayn and I disagree slightly with De Lacey’s translation of the
Greek here.) See also De nat. fac. 1.2.5K / 104,11–15, and see Enn. 2.7, where Plotinus deals
with the same controversy, though without naming the disputants.
as to change its species. It follows that when bodies of di#erent
elements are apposed, there will be a mutual interaction by which
one or both may be changed in species. When the qualities of one
body overpower those of the other, the weaker is assimilated in
quality to the stronger,31 and this results in augmentation (growth in
magnitude) of the stronger body (16.238,17–20). The growing body
can be said to grow ‘‘everywhere’’ ( ´ ) (236,26–7), even though
bodies cannot flow through each other, so that corporeal substance is
added to only one place, from the outside (where the weaker body
once was), because what counts as growing is the (increasing)
portion of matter in which its form – i.e., its complement of primary
qualities – is present (235,34–236,14).32 If, on the other hand, the
qualities, i.e. ´ , of the two bodies are evenly matched, then
they mutually su#er a change in quality, to the point where both take
on a new, medium quality, and the result is complexion. Here, again,
the complexion happens ‘‘everywhere’’ – meaning that the form, i.e.
quality, of the new homoeomer is everywhere the same, even though
the original corporeal substances are not literally extended through-
out the same space.33
This relatively straightforward view also provides straightforward
interpretations of some key Aristotelian texts. The discussion in GC
1.10 can, after substituting ‘‘complexion’’ for Aristotle’s term ‘‘mix-
ture,’’ be read literally: the ingredients must be capable of mutual
action and passion (328a17–23); where one ingredient predominates
we get augmentation rather than ‘‘mixture,’’ i.e., complexion (ll.
23–8); when the ´ of the two are equal, ‘‘each changes to the
conquering [complex quality] from its own nature, and does not
become the other, but rather [something] intermediate and common’’
(ll. 28–31); ‘‘mixture,’’ i.e. complexion, is ‘‘the unification of the
ingredients via their alteration’’ (328b22). Furthermore, Aristotle’s
statement that the ingredients in the ‘‘mixture’’ (i.e., complex)
‘‘neither remain ’ ´ ◊ . . . nor are either one or both corrupted:
for their ´ is preserved’’ (327b 29–31) can now be understood
to mean that, while the elements are not actually present throughout
the mixture, their ´ , i.e. qualities, remain. Finally, the passage
at PA 2.1.646a 12–15, where Aristotle says, ‘‘one might posit that
primary [composition] is out of those [things] which are called by
some ‘elements,’ such as earth, air, water, and fire,’’ and then adds:
‘‘but it would be better perhaps to say: out of ´ ,’’ can be
understood to mean that it is, strictly speaking, elemental ´ ,
31
Alexander, De mixt. 13.230,7–13.
32
All this is based on Aristotle, GC 1.5.321b10–322a4. Galen’s view appears to be similar:
see De nat. fac. 1.7, 1.11, 2.3.
33
See Alexander De mixt. 14.230,30–34; 15.231,15–16; and see again Galen, De el. 9.490K /
138,1–14 / 115,5–116,10 and Philoponus, In GC 1.2, 22,23–5.
i.e. qualities, that are mixed in the complex, rather than the bodies of
the elements themselves.
This Peripatetic view, however, has two big disadvantages. The
first concerns the sense in which the elements remain and are
preserved in the complex. A superficial reading of the slogan, that
the qualities are mixed while the bodies or substances are not, might
be that small particles of water, fire, etc. remain in place, while their
qualities spread out and combine. If the ingredients are later separ-
ated out, each particle would take back its own qualities.34 But, of
course, this cannot be correct, if it is the sensible qualities them-
selves that, when added to matter, make a given body into water or
fire. A true homoeomer is not a mere apposition of small parts which
appears homoeomerous (to those of us who don’t have the eyes of
Lynceus); if its body has really been ‘‘unified,’’ no particles of the
elements can remain.35 As Alexander explains, it is not numerically
the same water which goes into the complex that emerges from it
upon separation; rather, separation involves each ingredient coming
out of every part of the complex.36 But how, then, can the Peripatet-
ics understand the relationship between (1) and (3): how is a
complex a type of mixture, of which the elements are parts, rather
than a new substance which has come into being with the perishing
of its ‘‘ingredients’’? Yet Aristotle defines the elements as bodies
which ‘‘cannot be divided into [parts] which di#er in form,’’37
implying that composites can be, and he also insists that mixture
(i.e., complexion) is not the same as generation and corruption.38
Alexander clearly worries about this problem,39 but it becomes
much worse given the Porphyrean scruples about the categories
mentioned above. If substance is strictly prior to accidents, then we
must carefully distinguish ordinary change in quality (alteration)
from change in substantial quality (generation and corruption).40
34
For example, one might get that impression from Galen’s description of a substance
like blood or milk: it is not strictly one ( ¢` ’
˜ = wāh* idan bi-al-h
* aqı̄qa), but composed
out of di#erent and opposite parts; while they are complected, they make it medium
(´ = mutawassat*) in quality, but, once separated out (
´ = idhā tamayyazā)
each shows its own idea or nature ( ’´ = *tabı̄‘a) (De el. 11.495–6K / 142,17–23 / 121,4–122,3).
35
For the eyes of Lynceus, see GC 1.10.328a13–15.
36
De mixt. 15.231,24–9. The last part is based on Aristotle’s own argument at GC
2.7.334a31–5 (on which see further below, p. 111).
37
Aristotle, Cael. 3.3.302a15–18, and see Galen’s definitions: ‘‘the first and simplest /
separate [¢ ´ = mufrada] by nature, which can’t be analyzed / partitioned
[ ˜ = tutajazza’] into anything else’’ (De el. 1.414–15K / 58,2–3 / 12,8–9); ‘‘the
smallest, first, and simplest parts’’ (8.480K / 126,11–12 / 102,1–2).
38
GC 1.2.317a23–7.
39
He tries to explain, I think rather unconvincingly, why the separation of the
ingredients doesn’t, on his view, involve ` ´
` ´ (De mixt. 15.231,17–
22).
40
‘‘Alteration’’ strictly speaking refers to the former: ‘‘motion in quality . . . not the
quality which is in the substance [’ ˛˜ ’´
◊ = fı̄ al-jawhar] (for the di#erentia is also a
quality), but the a#ective [quality] [`
´ = al-infi‘āl]’’ (Ph. 5.2.226a26–9). See also
But the change in quality which is supposed to occur in complexion
cannot be either: not the latter because, as we have just seen,
complexion must be distinguished from generation and corruption;
not the former because a mere alteration could never change the
primary qualities. Substances in general can receive the contraries of
qualities they have received, but not the contraries of qualities in
which they ‘‘substantiate’’ – as fire, for example ‘‘substantiates in’’
heat.41 Or as Philoponus puts it: it could not receive a lesser degree
of heat and still remain fire, because fire as such (` ˜ ˛¢̃ ˜ ) is the
extremely hot.42 This di$culty is reflected in Alexander’s terminol-
ogy: whereas Galen consistently uses ‘‘alteration’’ for the change
involved – or, more precisely,
’ ¢´ ` ’´ ’ ´43 –
Alexander uses simply ‘‘change’’ ( ´ ).44 Among the later
Neoplatonists, it gave rise to intricate arguments as to whether and
how the (substantial) forms and / or qualities of the elements remain
in the complex.45
So much for the first disadvantage mentioned above. The second is
less flashy, but perhaps more serious, because it concerns the
empirical adequacy of the whole theory. It has to do with the
relationship between (3) and (4). The problem is that composite
substances are not generally well-characterized simply by their
position on a two-dimensional field defined by the hot-cold and
dry-moist axes. On the contrary: every composite has secondary
sensible qualities, such as color and odor, and not all such qualities
seem to be accidental.46 Moreover, some if not all composite sub-
stances have other, non-sensible ´ . Galen is most sensitive to
this worry, and it is easy to see why from the two examples mentioned
above: the dormitivity of opium and the attractivity of magnets. As a
physician he was, of course, very interested in the faculties of drugs,
and attraction is crucial to his anatomical theories: he argues
GC 1.5.320a12–14; Cat. 14.15a13–33. Despite what de Haas says (New Definition, 138), GC
1.2.317a22–7 seems to be about the same issue. Cf. Philoponus, In GC 1.2, 42,12–17; and see
also 2.4, 231,28–232,6, and Gannagé, Alexander on GC 2.2–5, 63.
41
Porphyry In Cat. (CAG 4.1, ed. A. Busse [1887]), 99,6–10; cf., similarly, Isagoge (also in
CAG 4.1) 9,16–18 / zIsāghūjı̄, li-Furfurı̄yūs al-s*ūrı̄, naql Abı̄ Uthmān al-Dimashqı̄, ed. Ahmed
Fouad al-Ahwani (Cairo, 1952), 78; Simplicius, In Cat. (CAG 8, ed. C. Kalbfleisch [1907]), 5,
98,13–19.
42
In GC 2.7, 271,32.
43
De nat. fac. 1.2.4K / 103,20–21.
44
See, e.g., De mixt. 14.231,11, where ` ˜ corresponds to Aristotle’s
’ ´ (GC 1.10.328b22).
45
See especially Simplicius In Cael. 3.3, 601–2; 3.7, 659–61; Philoponus In GC 2.7; and see
de Haas, ‘‘Mixture in Philoponus,’’ for further sources.
46
Philoponus lists the sweetness of honey and the whiteness of white lead and snow as
substantial qualities (see De aeternitate mundi contra Proclum [Leipzig, 1899] 11.5.424,24–5;
for snow and white lead see Aristotle, EN 1.4.1096b23 and Plotinus, Enn. 2.6.1.20–22, 31–33).
De Haas, New Definition, pp. 167 n. 10, 231, claims that ‘‘strictly speaking’’ these are only
inseparable accidents. But although Simplicius apparently agrees (see In Cat. 5, 98,16–17), I
see no sign that Philoponus does.
vigorously that bodily organs have their own proper attractive,

repulsive, assimilative, and retentive faculties, against those (such
as Asclepiades and Erasistratus) who explained such phenomena
mechanically.47 Galen even says (perhaps exaggerating) that every
being has an attractive ´ by which it attracts what is appro-
priate to it,48 and applies this to the organs of the body, as well as
(following Hippocratic texts) the nutrition of plants and the action
of drugs.49 The magnet (¢ ‘H ´
´ = al-h
* ajar al-maghnat*ı̄s) is
mentioned as an obvious example of such an attractive faculty.50
But how can any of this be explained in terms of the primary
qualities or any medium between them? It will not be easy.51 And yet
Galen, and anyone else who follows the above Peripatetic view,
seems committed to the claim that the medium quality is the
(substantial) form of a homoeomerous complex.52 And ought we not
to understand, if we know what the form of some substance is, why
certain qualities or faculties are consequent to that?53 Thus we have
the seed that will eventually flower into the problem of occult
qualities or faculties – faculties which cannot be explained by the
known essential characteristics, or primary qualities, of bodies (and
note that the most infamous examples, dormitivity and attraction,
47
The entire De nat. fac. is devoted to this argument. For the attractive ( ¢

´ ) and
repulsive (or ‘‘propulsive’’:
´ ) faculties, see De nat. fac. 3.6.160K / 216,22–4, and see
Avicenna, al-Qānūn fı̄ al-t*ibb, ed. Sa‘ı̄d al-Lah *h
* am (Beirut, 1994), 1.1.6.1.3, p. 133. For the
mechanism of Galen’s opponents – their view that biological processes are ‘‘steered only by
material impulses’’ – see De nat. fac. 2.3.80K / 159,11–12.
48
De el. 14.507K / 156,10 / 138,6–7; De nat. fac. 1.14.55K / 141,5–8. See also De nat. fac.
1.15.60K / 145,3–6.
49
De nat. fac. 1.12.29–30K / 122,9–12, 1.14.53K / 139,26–140,4; De el. 14.507K / 156,1–6 /
137,10–138,1, citing Hippocrates, De nat. hom. 6.3 (La nature de l’homme, ed. Jacques
Jouanna [Berlin, 2002], 180,10–15 / Kitāb Buqrāt* fı̄ *tabı̄‘at al-insān, ed. J.N. Mattock and
M.C. Lyons [Cambridge, 1968], 9,13–15).
50
De el. 14.507–8K / 156,10–11 / 138,7; De nat. fac. 1.14.44–5K / 133,16–20. Although
Avicenna does not, to my knowledge, anywhere repeat Galen’s wide claim about there being
an attractive faculty in every being ( ¢
´ ◊ ˜ ’´
= fı̄ kull wāh
* id min al-ashyā’
al-mawjūda), he does echo all the particular points about plants, organs, drugs, and
magnets: for plants and organs, see Sh. T * ., Nabāt, 1, 3,5–7; for drugs and magnets, see
further below.
51
See De temp. 1.7.586K / 49,4–6, against those who take it for granted that the
dormitivity of opium is explained by the moisture and cold of its complexion: this kind of
thing is not ‘‘simply and easily known,’’ but rather ‘‘requires a vast amount of
investigation’’ ( ´ ˜ ´ ´ ).
52
See explicitly Quod animi mores corporis temperamenta sequuntur 3.774K / 37,16–20 /
13,17–20 (Greek edition: in Scripta minora, vol. 3, ed. G. Helmreich; Arabic edition: Galens
Traktat ‘‘Daß die Kräfte der Seele den Mischungen des Körpers folgen’’ in arabischer
Übersetzung, ed. H.H. Biesterfeldt [Wiesbaden, 1973]). (The view is attributed there to
Aristotle, but Galen agrees: see 5.785K / 46,17–23 / 20,14–17; 787K / 48,3–4 / 21,13–14).
53
Hence the full end of the above cited passage is: ‘‘requires a vast amount of
investigation, and perhaps will not be discovered, if one does not first know how to
understand the complexion of damp and dry and cold and hot.’’ See also De nat. fac.
3.7.167K / 221,24–22,2, where Erasistratus is ridiculed for not accepting that all ’ ´
that exist in the parts of the body are due solely to the complexion of the primary qualities.
And see Philoponus, In GC 2.4, 232,6–12.
get in on the ground floor). But although the problem begins in
Galen, the occultation is due to Avicenna.
III. AVICENNA’S NEW SOLUTION
The above problems are laid out in texts which were well-known to
Avicenna (for, even if he did not have the De mixtione, he would have
found everything in Galen and Philoponus, and most likely in
Alexander’s own In GC). But he approached them with a new tool in
hand – a tool which may have been, if not forged just for this purpose,
at least adopted by Avicenna for it: namely, a new picture of the
relationship between substance and quality.
The roots of the new doctrine are to be found already in Alfarabi.
To gauge the magnitude of the change, recall that Porphyry says fire
cannot receive the contrary of heat because it ‘‘substantiates in’’
heat. Now contrast Alfarabi’s explanation of how corporeal things
like fire are imperfect, in comparison with the immaterial intellects:
in order to achieve its e#ects, fire is in need of ‘‘an organ [āla]
external to its essence,’’ something merely ‘‘consequent upon that in
which fire substantiates,’’ namely heat.54 Elsewhere he explains his
view more clearly, and distinguishes between di#erent ranks of
corporeal substance, as well. Celestial bodies, he says, cannot act on
other things ‘‘unless they acquire another being external to their
substance and to the things in which they substantiate,’’ meaning ‘‘a
quantum or a quale or some other [thing] from among the remaining
categories.’’55 Sublunar substances are more imperfect still: they
cannot even achieve their own perfections or entelechies (kamālāt)
without ‘‘other beings external to their substance from among the
other remaining categories’’ – for example, magnitude, figure, place,
hardness, color, heat, or cold.56 The main motivation here seems to
be metaphysical: Porphyry’s fudge of ‘‘substantial quality’’ is now
rejected; a firm line is drawn between substances and members of the
other categories, such that there cannot be anything intermediate.
Alfarabi does not, however, emphasize the radicality of his break
with his predecessors in this regard, or seem overly interested in
either the problems or the advantages of the new position. The major
problem is the di$culty of explaining just what are the ‘‘things’’ in
which substances ‘‘substantiate.’’ But Alfarabi, to my knowledge,
does not address that issue. As for advantages, the only one which
54
Al-Farabi on the Perfect State: Abū Nas*r al-Fārābı̄’s Mabādi’ ārā’ ahl al-madı̄na
al-fād
* ila, ed. R. Walzer (Oxford, 1985), 2, §1, 92,10–12; 10, §9, 174,8.
55
Al-Fārābı̄’s The Political Regime (Al-siyāsa al-madaniya, also known as the Treatise on
the Principles of Beings), ed. F.M. Najjar (Beirut, 1964), 53,15–54,1.
56
Ibid., 66,9–12. See also Alfarabi’s paraphrase of the Isagoge, where heat is classed as an
inseparable accident (rather than a di#erentia) of fire (D.M. Dunlop, ‘‘Alfarabi’s Eisagoge,’’
Islamic Quarterly, 3 [1956]: 117–138, p. 125,12).
clearly attracts him is the extra systematicity he is able to introduce

into his cosmic scale of perfection – always something near and dear
to his heart. The possibilities for resolving problems about mixture
remain unexplored. Alfarabi does attach great importance to the
hierarchy, within the sublunar world, of more and less composite
‘‘mixtures’’ or ‘‘complexions’’ – he makes no clear distinction be-
tween the terms57 – but he is characteristically abstract and hazy
about the details. How, for example, is a human being supposed to
result from the ‘‘mixture’’ of a lower animal with – what?58 More-
over, the few details he does give are not encouraging. On the one
hand he says that the ingredients (akhlāt*) from which something,
e.g. a drug, is compounded, are its matter, whereas its form (s*ı̄gha or
*sūra) is ‘‘the faculty by which it acts’’ (al-qūwa bi-hā yaf‘al fi‘lahu).59
In cases like this, he says – that is, in the case of some artificial
substances (such as wine and drugs) and of ‘‘most’’ natural ones –
the form is not sensible; only its e#ects (af‘āl) are observed.60 This
perhaps suggests a way of dealing with the problem of secondary
qualities and faculties (it certainly puts him in a better position that
Galen). But it also implies that some natural substances do have
sensible forms, and if these include the elements, as seems plausible,
many if not all of the problems about primary mixture will remain the
same as before.
In any case, Avicenna’s view is unequivocal.61 The di#erence
between substance and accident is a di#erence in mode of being,
hence a fortiori a di#erence in essential character, i.e. in species:
It is impossible for there to be a single thing whose quiddity is so deficient
in being that there is any thing at all in which it exists as a thing in a
subject, and whose quiddity nevertheless does not require that there be any
thing whatsoever in which it exists as a thing in a subject.62
It follows that no term at all could apply univocally both to accidents
and to the essential characteristics of substances: in particular,
the elements cannot be di#erentiated by heat, cold, or any other
sensible quality. As we have seen, Avicenna is following Alfarabi in
maintaining this. But he is far more emphatic, and far more vocifer-
ous in denouncing the opposite view.63 More importantly, he says
57
See A z rā’ 8, §1, 134,17–136,1: ikhtilāt* al-ashyā’ . . . wa-imtizājātuhā 136,1: tilka
al-imtizājāt; §2, 138,3–4: al-ikhtilāt* wa-al-imtizājāt; §3, 140,7 (and throughout the preceding
paragraph): al-ikhtilāt*.
58
See again 8, §3.
59
Ih**sā’ al-‘ulūm, ed. Osman Amine (Cairo, 1968), 4, p. 115.
60
Ibid., pp. 114–15.
61
For an earlier discussion, with some di#erences in emphasis, see my ‘‘Simplicius and
Avicenna,’’ pp. 86–8; see also my article ‘‘Avicenna,’’ forthcoming in A. Balestra, G.
Segalerba and H. Gutschmidt (eds.), Substantia: sic et non (Basel: Schwabe-Verlag).
62
Sh. M., Maqūlāt, 1.6, 46,16–19.
63
‘‘This is false and impossible, and these examples are all false’’ (Sh. M., Maqūlāt, 1.6,
46,8); ‘‘this is a great error’’ (Sh. Il. 2.1, 58,14–15).
something about what the essential characteristics of substances in
general, and of the elements in particular, actually are. In the case of
a simple body such as water, he explains, the ‘‘nature,’’ i.e. ‘‘that
from which there proceeds [yas*dur] the motion or alteration, and
likewise the rest and quiescence, which are generated from its
essence’’ is identical to the ‘‘form’’ or ‘‘quiddity.’’64 Thus although
the form or quiddity of water, in which it substantiates, is nameless
and non-sensible, we do know it qua nature, i.e. as a faculty (qūwa =
´ ) from which sensible e#ects, such as cold, moisture, and
weight, ‘‘proceed,’’ in the absence of impediment. The form can
therefore ‘‘borrow’’ its name from the names of those naturally
concomitant e#ects.65
As indicated by the talk about modes of being and their relative
deficiency, Avicenna, like Alfarabi, is partly motivated by the desire
to assign a categorical status to the di#erentiae while avoiding
embarrassing entities of ambiguous ontological rank. To make his
solution work he needs to say, and does say, a lot more about the
status of the true di#erentiae themselves. For our purposes, however,
what is important is not the ontological payo#, but rather the new
light cast on the relationship between the bodies of the elements and
the primary qualities.
We can best approach this by thinking of Avicenna as interpreting,
first and foremost, not any specific passage in Aristotle, but rather
the Peripatetic doctrine reported by Galen and Alexander: that only
qualities are complected, and not, as the Stoics maintained, bodies or
substances. Since Galen and Alexander themselves hold that the
extreme primary qualities are constitutive of the elements, this
doctrine meant for them that there are no longer bodies of the
elements present in the complex at all; only their qualities or ´
are still present, insofar as the new qualities of the complex are
medium between those extremes. Hence the definition of complexion
as ‘‘the unification of the ingredients via their alteration’’ means
that a single unified body emerges out of the small apposed bodies of
the ingredients – not, indeed, because one body flows or extends
through another, but because the constitutive qualities, by which the
bodies of the ingredients were di#erentiated from one another,
change and become uniform. For Avicenna, however, there are no
such things as constitutive qualities. That only qualities, not bodies,
64
Sh. T
* ., Simā‘, 1.6, 34,8–9, 11 (where the understanding of the term ‘‘nature’’ derives, of
course, from Aristotle’s definition at Ph. 2.1.192b21–3). In the case of ‘‘composite’’ bodies,
on the other hand, the nature, while still a part of the form, is not to be simply identified
with it (Sh. T * ., Simā‘, 1.6, 35,7–10). But ‘‘composite’’ here probably does not include
homoeomers; apparently he has in mind bodies which, in addition to a nature, have also on
or more types of soul. See Sh. T * ., Simā‘, 1.5, 30,7–10; Qānūn 1.1.6.1.1, p. 130 (and cf. Galen,
De nat. fac. 1.1.1K / 101,1–8).
65
Sh. T* ., Simā‘, 1.6, 34,11–35,2; Kawn / fasād, 6, 131,6–10 (and cf. Galen, De nat. fac.
3.3.149K / 208,22–4).
are complected, must therefore mean something di#erent. In what is,

in e#ect, Avicenna’s version of GC 1.10.327b23–31, there are again
two possible outcomes – augmentation and complexion – when ‘‘the
elemental bodies are in contact’’ and ‘‘act upon one another.’’66 But
these are now described as follows:
Either one dominates the other, and changes it to its [own] substance, and
there is generation of the species of the dominant one, and corruption of the
dominated; or neither achieves domination over the other to the point of
changing its substance, but it changes its quality to a point at which the
action and passion stabilize, and there arises in it a uniform quality, which
is called the complex [mizāj]; and this combination is called complexion
[imtijāz].67
Aristotle’s definition of complexion, in other words, now means that
there is no truly unified substance: the ingredients are united merely
by alteration, while their substances remain unchanged. So while the
qualities are complected, the original corporeal substances – the
small elemental bodies – are still in place, numerically the same as
before.
If, moreover, a true essential characteristic is a qūwa, i.e. ´ ,
we also get a new, surprising understanding of the statement that the
ingredients are not corrupted because ‘‘their ´ is preserved.’’
For Alexander and Galen, ´ refers there to the primary
qualities, which remain behind when the substances of the elements
are no longer present. For Avicenna it is the opposite, as he makes
clear in a passage where, unusually, he explicitly emphasizes his
dependence on Aristotle’s text:
Then the First Teacher said, after that: ‘‘for the ingredients are preserved
in qūwa,’’ and he said, ‘‘but their qūwa is preserved,’’ and he meant: [they
are preserved] in the active qūwa which is the form, and he did not mean: in
the qūwa with respect to passions, which belongs to matter in its essence.
For the man wanted to refer only to something which belongs to them along
with their not being corrupted. But that can only be if the qūwa which is
their essential form remains.68
PA 2.1.646a12–15 can also be interpreted in the same spirit.
The interpretation may have been helped along by the fact that
both ’ ´ and ˜ are standardly translated using the same
forms of f-‘-l. In the above quote, for example, the phrase al-qūwa
al-fi‘liyya is surely supposed to translate ¢
` ´ . Simi-
larly in the continuation, where Avicenna distinguishes between the
nature of an elemental body (which as we know is also its form) and
a ‘‘fi‘l which proceeds from it,’’ which is a quality or other accident.
66
Sh. T* ., Kawn / fasād, 6, 126,11–12.
67
Ibid., 126,15–127,1; see also Sh. T * ., Simā‘, 1.9, 50,16–51,8.
68
Sh. T* ., Kawn / fasād, 6, 127,11–15; see also 7, 138,16–17; and see Eichner, ‘‘Einleitende
Studie,’’ pp. 165–6.
Since the word ’ ´ ◊ in Aristotle’s phrase ’ ´
◊ ` ¢ ´
’´
˜ ´ ’ ’´ (327b24–5) would no doubt have been
translated bi-al-fi‘l,69 it seems possible that Avicenna understood the
phrase to mean, in e#ect: being di#erent in sensible quality from (the
complex) that is generated out of them.70
In any case, even with all the help it can get from ambiguities of
translation, the theory that small elemental bodies are still present
in the complex is at first sight di$cult to attribute to Aristotle, since,
in both GC 1.10 and 2.7, he seems at pains to deny precisely that.
Hence Simplicius and Philoponus, in defending (or imagining a
defense of) a very similar view, maintain (or imagine its proponents
maintaining), against Aristotle, that flesh and bone are not truly
homoeomerous.71 The same thinking is evident in Maier’s treatment
of Avicenna. After accurately describing his understanding of
qūwa = ´ and of ¢´ ’ ´ , she goes on to say that
the resulting view was unacceptable to his successors because ‘‘a
mixtum in the strict sense is supposed to be a homogeneous stu#,
whose smallest parts are eiusdem rationis with the whole,’’ so that
Avicenna’s theory could only mean that multiple elemental forms are
present in every part of the mixture.72 Her own reason for declaring
that latter conclusion unacceptable is perhaps not so convincing.73
But we can supply a stronger reason: since the elements are corpo-
real substances, i.e. their substances are bodies, Maier’s unaccept-
able conclusion is precisely the view that all Aristotelians and
Neoplatonists reject: namely, the Stoic view of complexion as one
body extending through another.
Maier’s argument, so strengthened, is irresistible. But, whatever
his Latin successors may have thought, what it implies for Avicenna
himself is that he somehow holds both that small particles of the
elements remain in the complex and that a mixtum, or rather a
complectum, is a ‘‘homogeneous stu#.’’74 How can he reconcile these
69
See the Arabo-Hebrew translation, ed. Tessier: be-fo‘al.
70
However, a di#erent reading is suggested by Sh. T * ., Kawn / fasād, 6, 131,17–132,1.
71
See Philoponus In GC 2.7, 269,25–270,5; Simplicius In Cael. 3.7, 660,18–661,14.
72
‘‘Struktur,’’ pp. 25–8.
73
She says that ‘‘the essence of elemental form consists in the fact that it immediately and
exclusively informs prime matter,’’ so that ‘‘matter can, indeed, take on di#erent elemental
forms one after another, but not simultaneously’’ ( ibid., p. 27). The ‘‘immediately’’ would be
outright denied by Avicenna (and by others who accept some version of corporeal form or
‘‘unqualified body’’ as an intermediate subject), and the ‘‘exclusively’’ – which is the heart
of the matter – has no obvious source in Aristotle, so far as I know.
74
Although Maier is aware of the Kawn / fasād texts, which she cites from manuscript,
she may nevertheless have been misled here because she approaches Avicenna by way of the
later Latin tradition, which, however, was mostly dependent on Averroes’ (accurate but
incomplete) report of Avicenna’s position (In Cael., 3 c. 67, 635,115–39 [227rb–va]), and to a
lesser extent on Avicenna’s brief remarks in Sh. T * ., Simā‘, 1.6. See Maier, ‘‘Struktur,’’ pp.
23, 93 n. 15; S. van Riet, ‘‘Le De generatione et corruptione d’Avicenne dans la tradition
latine,’’ in Thijssen and Braakhuis (eds.), The Commentary Tradition, pp. 69–77; and see
Eichner, ‘‘Einleitende Studie,’’ pp. 139–45.
two? The basic answer is that, as we have already seen, he has a

metaphysical distinction not available to Philoponus and Simplicius.
Since, from their point of view, the elements are constituted by
sensible qualities, a complex in which bodies of the elements remain
would be, as Simplicius puts it, like a cloak made of di#erent colored
threads: it might seem to have a uniform quality from far o#, but if we
could only examine it more closely – if we had the eyes of Lynceus, or
the tactile equivalent – we would observe its non-uniform micro-
structure. But for Avicenna this is no longer the case. The complex,
he can freely admit, is not homoeomerous in substance. Of course it
isn’t, since, on his understanding, that would be the Stoic view, that
there is complexion of the bodies of the elements. According to us
(Avicennan) Peripatetics, on the other hand, only qualities are
complected – and the result is what we should expect, namely
perfectly uniform medium qualities. But only these uniform acciden-
tal qualities, not the di#ering substantial forms, are sensible. So in
the case of a compound unified by complexion, ‘‘sense sees it as
homoeomerous.’’75 Lynceus, in other words, could no more discover a
lack of uniformity here than can we.
This puts Avicenna on the right side of Aristotle’s arguments in
GC 1.10. More di$cult to understand is how he can be reconciled
with the text of 2.7. There Aristotle objects to a view (which he
ascribes to Empedocles) according to which a homoeomerous sub-
stance like flesh consists of ‘‘preserved elements apposed to one
another in small parts,’’ on the grounds that, in that case, ‘‘fire and
water will not be generated out of just any part of flesh’’; rather, ‘‘as
stone and brick [are generated] from a wall, each out of a di#erent
place and part.’’76 Not only does Avicenna’s position seem to have
exactly this implication, but he actually relies on that feature of it in
chapter 7 of the Kawn / fasād, which is devoted to a polemic against
his Baghdadi opponents, who maintain that the ingredients in
complexion ‘‘divest themselves of their forms . . . and then put on a
single form, so that they come to have a single matter and a single
form.’’77 Against this, Avicenna points out that complexes are not
uniformly a#ected by external influences: that exposure to flame, for
example (in an alembic), typically results in some parts of them
going up as vapor and others remaining behind. Relying (as always)
on a principle of su$cient reason, he concludes that there must
already have been some di#erence in ‘‘aptitude’’ between di#erent
parts of the complex before the heating began. A further argument,
the details of which are not important for our purposes, then
75
Sh. T
* ., Af‘āl / infi‘ālāt, 2.2, 266,4–5.
76
GC 2.7.334a 34–b2.
77
Sh. T
* ., Kawn / fasād, 7, 133,5–7. For this identification of the (unnamed) ‘‘modern’’
opponents attacked here, see Eichner, ‘‘Einleitende Studie,’’ pp. 170–2.
establishes that this di#erence must be due to a distinction
(tamāyuz) in substantial form – that is, a distinction of bodies.78 The
complex qua complex, in other words, is not a unified body, but
rather a collection of bodies united by cohesion and by common
sensible qualities. So we return to the doctrine we saw above – but
now apparently via the claim, directly against Aristotle, that the
disintegration of the complex will be just like taking stone and brick
out of a wall.
Since, however, Avicenna emphasizes his agreement with Aristotle
on just this point, he must understand GC 2.7 di#erently. Most likely
he begins with the fact that the chapter actually poses a dilemma:
both those (like Empedocles) who do not allow the elements to
be generated out of one another and those (like Aristotle) who do
have trouble explaining the generation of homoeomers (334a18–23).
Empedocles, as we have seen, cannot explain how each ingredient
can re-emerge out of any part of the complex. But Aristotle has his
own explaining to do. If none of the elements remain in the complex,
why isn’t the result simply a complete absence of primary quality
and / or elemental form – i.e., nothing but (prime) matter (334b4–6)?
As in 1.10, then, the dilemma is this: if the elements are preserved,
there is no true homoeomery; but if they are not, we get corruption,
rather than complexion. The solution, moreover, sounds similar:
homoeomerous bodies come to be from ‘‘the contraries [i.e., the
primary qualities] or the elements being mixed’’; the elements can
then re-emerge because those same contraries exist in the complex
‘‘somehow potentially [ ´ ]’’ – but, Aristotle adds, ‘‘not in such
a way as the matter’’ (334b16–19). Now recall that Avicenna, in his
interpretation of 1.10, contrasts ‘‘the qūwa . . . which belongs to
matter’’ with ‘‘the active qūwa which is the form,’’ and claims that
Aristotle’s statement about the ingredients’ remaining in qūwa refers
to the latter.79 So on Avicenna’s reading, Aristotle escapes his own
horn of the dilemma precisely via the doctrine that the substances of
the elements are preserved. Why doesn’t he, then, land straight on
the other horn? Presumably because he, unlike Empedocles, does
allow the elements to be transformed into one another. Fire or water
can come out of any part of the flesh, but only in an extreme case in
which fire or water might come to be out of, e.g., pure earth – a case
in which, perhaps, the entire piece of flesh might come to be one or
the other. Meanwhile, Avicenna emphasizes, in more normal cases,
in which the ingredients are merely separated, rather than trans-
formed, the last thing we expect is for just any ingredient to emerge
from just any part of the complex. In fact, he turns the tables against
his opponents, pointing out that, if that were the case, we could find
78
For the full argument see Sh. T * ., Kawn / fasād, 7, 134,2–135,4.
79
Sh. T
* ., Kawn / fasād, 6, 127,12–13 (and see also, again, Ph. 5.2.226a26–9).
a piece of flesh which, when heated in the alembic, would entirely

evaporate, or one which would entirely turn to sediment.80
Given these interpretative moves, Avicenna can reconcile his
position with Aristotle while avoiding the first disadvantage which
we noted in the Peripatetic view espoused by Alexander and Galen.
Because the primary qualities are no longer thought of as constitu-
tive, the elements can easily lose these qualities, in whole or in part,
without being corrupted.81 Avicenna is well aware of the crucial
metaphysical move which allows him this solution, denied to his
predecessors: ‘‘the commentators,’’ he explains, are confused be-
cause they fail to distinguish between ‘‘forms and accidents,’’ i.e.
between ‘‘the natural forms of these [elemental] bodies and their
qualities’’:
And since they thought that these qualities, all or some of them, are the
forms of these bodies . . . the one of them who followed the best path [i.e.,
Philoponus] said: the qualities are preserved, but thwarted in force, whereas
the bodies are [only] potentially pure.82
He then goes on to refute this view, and several variants, using the
powerful new metaphysical tool he has taken from Alfarabi.
But what of the second disadvantage in the Peripatetic position
– namely, the problem of explaining secondary qualities and facul-
ties? Clearly the situation is changed dramatically by the demotion
of the primary qualities. The change is not all for the better, however.
One immediate result has nothing to do with complexion, but
rather with properties of the elements themselves. If fire, for
example, is constituted by heat and dryness, and earth by cold and
80
Sh. T
* ., Kawn / fasād, 7, 135,2–3.
81
Hence de Haas’s initial framing of the problem in terms of change of constitutive
qualities (‘‘Mixture in Philoponus,’’ pp. 28–9) would already be unacceptable to Avicenna.
Like Maier, de Haas (and / or his source, Zaborella) may have been misled by Averroes. The
latter treats as an absurd consequence of Avicenna’s view, that fire might be fully present
in the complete absence of heat and dryness ( In Cael., 3 c. 67, 635,132–6 [227rb]) – implying
that Avicenna still does give some constitutive role to the primary qualities. But, precisely
because Avicenna does not give any qualities that role, he would see no absurdity in the
case Averroes describes: he takes ice, for example, to be water in which the sensible quality
of moisture is completely absent (Sh. T * ., Kawn / fasād, 6, 130,14–16; Af‘āl / infi‘ālāt, 2.1,
255,12). (Cf. Galen’s treatment of ice as water which – like all water – is moist in the
extreme, but which lacks the softness normally associated with moisture: De temp. 2.3.598K /
12–18. Philoponus, on the other hand, seems indecisive on this issue: see In GC 1.7, 147,4–9.
The source for all of these views is Aristotle, Metaph. 8.2.1043a9–10.)
82
Sh. T* ., Kawn / fasād, 6, 127,18–128,4. For Philoponus, see In GC 2.7, 271,25–272,10.
‘‘Thwarted in force’’ (maksūr al-sawrā thus translates

´ (271,5–6). (Here I am
in disagreement with Eichner, ‘‘Einleitende Studie,’’ pp. 185–6, n. 137. No doubt this is an
odd translation of

´ , but Philoponus’ use of the term is itself odd and invites
loose translation – cf., e.g., Kupreeva’s decision to translate it as ‘‘inhibited’’ (Philoponus
on Aristotle’s ‘‘Coming-to-Be and Perishing 2.5–11’’ [Ithaca, 2005], p. 62). That the Latin
translator rendered Avicenna’s term with fractus rather than castigatus does not seem
relevant, nor would it have posed much of a barrier to a Latin reader familiar with
Philoponus in identifying the origin of the phrase.)
dryness, why is it that fire tends upwards and earth downwards?
What connection is there between levity and heat, or gravity and
cold? Similarly, why should the transition from cold and moist water
to hot and moist air result in a change in volume, i.e. in quantity?
This problem is now solved, in a manner of speaking, by the fact that
heat and dryness have no closer – and therefore no more distant – a
connection with the true di#erentia of fire than do levity and bulk:
Every one of the elements has a form by which it is what it is, and
consequent to that substantial form are entelechies [kamālāt] of the
category [bāb] of quale, and of the category of quantum, and of the category
of where. And there is proper to each one of them heat or cold . . . and
dryness or moisture . . . and a natural measure of quantity, and natural
motion and natural rest.83
This ‘‘solution,’’ however, alleviates the mystery as to why, say,
water is heavy only by generating an equally intractable mystery as
to why it is moist. Both are results of the same nameless occult
faculty.
As for complex substances, notice, first of all, that we have not yet
accounted for their existence, on Avicenna’s view. For Alexander or
Galen, a new set of primary qualities automatically constitutes a new
substance; the question is only how a substance so constituted could
give rise to new secondary qualities and faculties. But for Avicenna
the complexion, i.e. the medium tangible quality itself, is merely an
accident. If the question is how that tangible quality can give rise to,
for example, a color, then the answer is that it certainly cannot,
‘‘because the complexion is a tangible quality, and color is not
tangible.’’84 Thus it seems impossible to understand the origin, not
just of secondary qualities and faculties, but of complex substances as
such. If complexions are accidental qualities, where will we ever get
a substantial forms other than those of the elements? The answer, to
make a long story short, is (a limited form of) occasionalism:
What must be said about all this is one thing, and that is that the compound
body is adapted, by its complexion, to receive a disposition, or a form, or a
proper faculty, and this emanates to it from the giver of forms and of
faculties, and no other. And their emanation from it is due to its liberality,
and because it does not fail whatever is worthy and adapted.85
Some complexes do not receive any such added emanation at all,
and some receive only another accidental property – for example, a
color. But some also receive a new substantial form: the form of
83
Sh. T
* ., Kawn / fasād, 6, 129,15–130,1.
84
Sh. T
* ., Af‘āl / infi‘ālāt, 2.1, 254,9.
85
Sh. T* ., Af‘āl / infi‘ālāt, 2.1, 256,9–11. The ‘‘giver of forms’’ here is God (l. 14);
elsewhere, however, as in Alfarabi, it seems to be the Active Intellect (Sh. Il. 9.5, 410,14–16,
411,9).
a homoeomerous complex substance.86 The new substantial form

then enters into each small elemental particle, such that each
becomes, for example, an actual particle of fire which is also flesh;
consequently, the entire composite body becomes a single body of
flesh.87
Now, in some sense, this doctrine is obviously a disaster for the
explanation of secondary qualities and faculties of complex sub-
stances. Galen and Alexander give us a clear explanation of how a
new substance could result from a change in the primary qualities,
and some (albeit faint) hope of explaining all its properties on that
basis. Avicenna, on the other hand, leaves us with nothing but
unspecified interactions between nameless and unknown faculties,
which, in some unknown and unpredictable way, call down new
forms and qualities from the heavens. As he puts it: ‘‘there is no way
to grasp the correspondence between particular complexions and the
faculties and dispositions which are consequent to them.’’88 On the
other hand, there is some gain in that those previously mysterious
qualities and faculties are now no more mysterious than anything
else. The apparent problem about them arose, as Avicenna explains,
from people’s natural tendency to wonder at what is unusual:
The majority of them are not concerned to know why fire can burn a great
city in a single hour, or why cold dries [i.e., freezes] water, but they are
concerned to know why a magnet attracts iron. . . . So that if one were to ask
why cold does this they would not know, and would say: because that is its
nature . . . and similarly with respect to fire. . . . And the discerning among
them . . . says: because the matter of fire has acquired a form which
essentially has this e#ect [fi‘l]. . . . But then he is not content, similarly, to
say of the magnet: because the complexion is the cause of the composite’s
acquiring a faculty which by its essence and nature attracts iron.89
The truly wise will, of course, content themselves with the latter
answer. And similarly in the case of drugs – those that act, not ‘‘by
their qualities,’’ i.e. by their complexion of elements, but ‘‘by their
substance,’’ i.e. by the new, consequent specific substantial form:90
here, too, Avicenna would advise exactly the answer infamously
given by Molière’s bachelor to the question about opium: quia est in
eo virtus dormitiva.91
86
For all these possible outcomes, see Sh. T * ., Af‘āl / infi‘ālāt, 2.2, 261,4–13.
87
Sh. T* ., Kawn / fasād, 7, 136,13–16. (This consequence of Avicenna’s view is put in the
mouth of his Baghdadi opponents as an objection, but he ends up accepting it.)
88
Sh. T
* ., Af‘āl / infi‘ālāt, 2.1, 255,7–8.
89
Ibid., 255,11–256,6.
90
Ibid., 2.2, 262,1–2. Cf. Galen’s description of the di#erence between food and drugs: in
the case of food we need the ‘‘substance’’ ( i.e., corporeal bulk) of what we ingest; in the
case of a drug, we need its qualities, and ingest the substance only because the qualities can
never be found on their own (De el. 6.474K / 120,3–9 / 93,5–94,4).
91
Le malade imaginaire, in Œuvres complètes (Paris, 1962), vol. 2, p. 848.
As that last quote suggests, the long-term e#ects of Avicenna’s
choices here was quite significant.92 The doctrine specifically about
complexion was, as Maier correctly points out, never popular with
his successors – in part, perhaps, as I have indicated, because they
had no accurate report of it. But the new view on the relationship
between substance and accident was extremely influential. Hence
Thomas Aquinas, in the De ente et essentia – an early work, parts of
which are virtually transcribed from the Shifā’ – explains that the
proper di#erentiae of angels are ‘‘hidden’’ (occulte) from us, and then
adds: ‘‘For even in sensible things the essential di#erentiae them-
selves are unknown, so that they are [instead] signified by accidental
di#erentiae which arise from the essential ones, as a cause is
signified by its e#ects.’’93 In his much later GC commentary, he takes
exactly the same position, and applies it to the primary qualities:
Substantial di#erentiae, because they are unknown, are manifested by
accidental di#erentiae. And thus we often use accidental di#erentiae in
place of substantial ones. And in this way the Philosopher says here that
hot and cold are the substantial forms of fire and earth. For heat and cold,
since they are proper passions of those bodies, are proper e#ects of their
substantial forms.94
Moreover, although Thomas is generally known as an opponent of
Avicenna’s occasionalism, he takes over intact the key part of it:
namely, that sublunar causes can only dispose the matter to receive
substantial forms, while the actual source of those forms is the
angels and celestial bodies.95 If it is true, as I have suggested, that
Avicenna adopted these metaphysical positions in part because of
their usefulness in his theory of complexions, then that despised and
misunderstood theory had the most far-reaching possible conse-
quences. For, as is well known, it was precisely the rejection of these
seemingly idle and unknowable substantial forms, along with their
attendant, non-explanatory, occult faculties, that would eventually
drive the most important developments of early modern philosophy.
92
Note that ´ and qūwa are often rendered in Latin as virtus: see, e.g., Burgundio
of Pisa’s translation of GC 1.10.327b31 (Aristoteles Latinus, vol. 9.1, ed. J. Judycka [Leiden,
1986]), and the translation of Sh. T * ., Simā‘, 1.6, 35,8 (Liber primus naturalium: tractatus
primus: De causis et principiis naturalium, ed. S. Van Riet [Leiden, 1992]). Although I have
not located the passage, I consider it likely that Molière is quoting (perhaps indirectly)
from the Qānūn.
93
De ente et essentia, c. 5, tom. 43 of S. Thomae Aquinatis Doctoris Angelici Opera omnia,
iussu impensaque Leonis XIII P.M. edita (Rome, 1976), p. 379a, ll. 75–80.
94
Sententia super libros De generatione et corruptione c. 3 lect. 8, ad 1.3.318b14, tom. 3 of
Opera omnia (Rome, 1886), p. 293a.
95
See Quaestiones disputatae de potentia Dei q. 5 a. 1c.; Summa theologiae I q. 115 a. 3 ad
2.
SOME QUESTIONS REGARDING AVICENNA’S

THEORY OF THE TEMPORAL ORIGINATION OF THE
HUMAN RATIONAL SOUL
MICHAEL E. MARMURA
Department of Near and Middle Eastern Civilizations, University of
Toronto, 4 Bancroft Avenue, Toronto, Ontario, Canada M5S 1C1
Email: emarmura@cogeco.ca
Abstract: In Avicenna’s expositions of his theory of the temporal origination of the

human rational soul, its h * udūth, one meets di$culties in understanding of what he
actually means. Some of the expressions used are left unexplained and one has to
extract their meaning from discussions given in a di#erent context. There are also
ambiguities in his use of such terms as al-‘aql al-kulliyy (the universal intellect) and
al-nafs al-kulliyya (the universal soul). Although in one place he makes it clear that
these expressions refer to concepts that exist only in the mind, distinguishing them
from ‘aql al-kull (the intellect of ‘the whole [universal]’) and nafs al-kull (the soul of
‘the whole [universel]’), the distinction is not uniformly observed. In a number of
his works the term ‘‘universal’’ is used to refer to both the celestial intellect and the
celestial soul. There is also an ambiguity in his statements about the role the
celestial soul plays in the emanation of the human rational soul. In some discussions
he seems to hold that the rational human soul emanates from both the celestial
intellect and the celestial soul. The Metaphysics of al-Shifā’ (The Healing) suggests
a resolution of this ambiguity. At the same time, there are statements in this work
that are left unexplained and one has to look for their explanation in other books of
The Healing. Thus questions do arise regarding the details of Avicenna’s theory of
the temporal origination of the human rational soul. His general exposition of his
theory, however, remains comprehensible, its pivotal position within his entire
philosophical system clear.
Résumé: Dans les exposés avicenniens de sa théorie de l’origine temporelle de l’âme

humaine rationnelle, son h * udūth, on a parfois du mal à comprendre ce que l’auteur
veut vraiment dire. Quelques-unes des expressions que celui-ci utilise sont laissées
sans explication, et l’on doit en extraire le sens à partir de discussions tirées d’un
contexte di#érent. On trouve aussi des ambiguı̈tés dans son utilisation de termes
comme al-‘aql al-kulliyy (l’intellect universel) et al-nafs al-kulliyya (l’âme uni-
verselle). Même si, dans un endroit, il rend manifeste que ces expressions se réfèrent
à des concepts n’ayant d’existence que dans l’esprit, en les distinguant de ‘aql al-kull
(l’intellect du ‘tout [universel]’) et du nafs al-kull (l’âme du ‘tout [universel]’), cette
distinction n’est pas observée de façon constante. En e#et, dans plusieurs de ses
ouvrages, le terme ‘‘universel’’ est utilisé pour référer aussi bien à l’intellect céleste
qu’à l’âme céleste. Il existe aussi une ambiguı̈té dans les énoncés avicenniens au
sujet du rôle joué par l’âme universelle dans l’émanation de l’âme humaine
rationnelle. Dans quelques discussions, il semble soutenir que l’âme humaine
rationnelle émane aussi bien de l’intellect céleste que de l’âme céleste. La Métaphy-
sique d’al-Shifā’ (La Guérison) suggère une résolution de cette ambiguı̈té. En même
122 MICHAEL E. MARMURA
temps, quelques énoncés de cet ouvrage sont laissés sans explication, de sorte que
l’on doit en chercher l’éclaircissement dans d’autres livres de la Guérison. On voit
donc que des questions se posent véritablement au sujet des détails de la théorie
avicennienne de l’origine temporelle de l’âme rationnelle humaine. L’exposé général
de sa théorie reste cependant compréhensible, aussi la position clé de celle-ci à
l’intérieur de son système philosophique reste-t-elle claire.
Avicenna (Ibn Sı̄nā) (d. 1037) is noted for his theory of the temporal
coming into existence of the human rational soul; the soul’s h * udūth.
The basic elements of his theory can be expressed as follows:
An individual rational human soul comes into existence the
moment a human body, suitable for its reception and for becoming its
instrument, is formed. It comes into existence as an emanation
necessitated by the celestial intellects.1 The particular composition
of the body’s humors and the time in which the soul is created,
individuate it. Thereafter, it is further individuated as it acquires its
own unique dispositions. It thus becomes an individual rational soul
di#erent from each and every other similarly created individual
human rational soul. It retains its individuality throughout its
earthly existence and beyond. Although associated with a particular
body, the soul is not imprinted in it. It remains immaterial and
incorruptible. With the corruption of the body, it continues to exist
in the hereafter, rewarded or punished according to its performance
during its earthly existence.2 Before its temporal origination, it
1
Ibn Sı̄nā (Avicenna), Al-Shifā’ [The Healing]: al-T * abı̄‘iyyāt [Physics] (6); al-Nafs
[Psychology], ed. G. C. Anawati and S. Zayid (Cairo, 1975), p. 207, line 4 where it is stated,
‘‘the preparedness of the bodies necessitate that the existence of the soul would emanate for
them from the separate causes.’’ The ‘‘separate causes’’ are primarily the celestial intellects.
It is normally (and in our view correctly) interpreted that the emanation is from the Active
Intellect, the last of the series of intellects emanating from God. Now as we shall indicate
in Part III, the term, ‘aql al-kull, literally ‘‘the intellect of ‘the whole’,’’ is used in two
senses. It is used collectively to refer to the totality of the celestial intellects which would
include the Active Intellect. It is used to refer to the first intellect emanating from God.
With the celestial intellects, all emanations ultimately proceed from this first intellect,
through the mediation of successive intellects, the last of which is the Active Intellect.
Avicenna, however, does not always specify that the emanation of the human rational soul
is from the Active Intellect, perhaps understandably, since this intellect itself is an
emanation from the higher intellects, and because (as we will be indicating) there are other
celestial activities that ‘‘help’’ this emanation. The reference to this work will be
abbreviated, Psychology.
2
A representative statement of Avicenna’s views on reward and punishment in the
hereafter (with some variations in other works) is given in his Risāla Ad *h
* awiyya fı̄
al-Ma‘ād [An Ad * hawite Treatise on the Afterlife]. Adh * awite refers to its dedication to a
certain Abū Bakr Ibn Muh * ammad on the feast of sacrifice, al-Ad *h
* ā.
According to this treatise, after the death of the body, immortal souls consist of those
that are perfect and purified, and those that are not. Perfect souls are those that fully
realize that they have intellectual perfection. Some, in addition to this, have purified
themselves from bodily dispositions. These find eternal bliss in the hereafter, contemplating
the celestial intelligences and God. Some that realize that they are capable of intellectual
perfection, but had been sullied in their earthly existence by animal dispositions, do not
THE HUMAN RATIONAL SOUL 123
existed neither as one of a multitude of individual rational souls,
nor as one soul that by acquiring a mundane existence becomes
fragmented into many individual souls.
As an emanation from the celestial realm, the soul’s creation
becomes part of Avicenna’s emanative cosmology. Once its place in
his cosmology is grasped, its pivotal position in Avicenna’s entire
philosophy becomes evident. It connects his metaphysics with his
psychology. It forms an essential part of his moral philosophy, his
eschatology, his theory of prophecy and its attendant political
philosophy.
Avicenna’s expositions of this theory are supported by closely
reasoned arguments. These, within his system and its premises, are
compelling. At the same time, however, there are aspects of the
theory that he does not explain. Some of the expressions he uses are
not properly defined. One also encounters shifts in di#erent works in
his use of certain terms relating to his cosmology. These shifts are
not fully addressed. There are, moreover, questions arising from his
emanative theory for which Avicenna does not provide explicit
answers. It is these aspects of his expositions that we will be
discussing.
Before doing so, however, there is a need for further discussion of
the theory as such. For this, we will turn to his major philosophical
work, the voluminous Healing (al-Shifā’).
II
In The Healing, the theory of the temporal origination of the human

rational soul is discussed in two of its volumes: in the Psychology
(al-Nafs)3 and in the Metaphysics (al-Ilāhiyyāt).4 In the Metaphysics
attain immediate bliss, but eventually achieve it when in the hereafter the traces of animal
dispositions are gradually erased. Souls that had been aware of having intellectual
perfection, but did not seek it – more than this, had disowned and opposed it – live in
eternal torment, ever seeking, but never attaining, true bliss. Then there are deficient souls
to whom it never occurred that they have intellectual perfection and deficient souls of
imbeciles and children, incapable of awareness of such perfection. Each of these groups
remain in the life to come without absolute happiness and without absolute misery. Ibn
Sı̄nā, [Al-Risāla] al-Ad
*h* awiyya fı̄ al-Ma‘ād, ed. H. ‘A z *sı̄ (Beirut, 1984), Ch. VII, pp. 145–58,
particularly, p. 152 #. This reference will be abbreviated Risāla in the notes.
This edition has much to commend it, except that it takes no account of the earlier,
scholarly Italian edition with an introduction, Italian translation and notes by Francesca
Lucchetta: Avicenna, Epistola Sulla Vita Futura, Testo Arabo, Traduzione, Introduzione e
Note (Padova, 1969). This edition will be referred to as Lucchetta.
Both editions are based in part on the edition of S. Dunya (Cairo, 1949). Subsequently,
however, each have used additional manuscripts unknown to either: Lucchetta, a Leiden
and a Manchester manuscript and has consulted the Latin translation of the Italian
physician Alpago (d. 1522); ‘A z *sı̄ a Berlin and an Istanbul manuscript.
If there are significant di#erences between the two editions, we will note them.
Otherwise, the references will be to ‘A z *sı̄’s edition.
3
The main argument for the rational soul’s temporal origination is in Psychology, Bk. V,
Ch. 3, pp. 197–201, the argument begins on p. 198, line 9 to the end.
it forms part of Avicenna’s theory of emanation. Although there is no

explicit discussion of emanation in the arguments for the soul’s
temporal origination in the Psychology,5 the theory of emanation is
assumed in these arguments. Thus, for example, we have the state-
ment, ‘‘if the matter of a body suitable to become an instrument of
the soul and a kingdom for it is temporally created, the separate
causes (al-‘ilal al-mufāriqa) bring into temporal existence the par-
ticular soul or this occurs by means of them.’’6 Now, ‘‘the separate
causes,’’ are none other than the celestial entities in Avicenna’s
triadic emanative scheme.7 Without cognizance of this scheme,
Avicenna’s theory of the temporal origination of the human soul
cannot be properly understood. A brief review of this scheme is hence
in order.8
The world emanates necessarily from God, the Existent Necessary
in Himself, as a consequence of His self-knowledge. This self-
knowledge entails knowing Himself as the cause of all other ex-
istents. His act of self-knowledge is eternal and its consequence is the
existence of an eternal world. The first9 necessitated e#ect of His
self-knowledge is an eternal first intellect. This intellect encounters
three facts of existence, God as an existent necessary in Himself, its
own existence as necessitated by God and its own existence that in
itself is only possible. It undergoes eternally three acts of knowledge:
knowledge of God as necessary in Himself, knowledge of itself as
necessitated by God and knowledge of itself as being possible. These
three acts result in the necessary emanation of three things: another
intellect, the soul of the outermost sphere of the heavens, and the
body of that sphere. The second intellect undergoes a similar act of
knowledge resulting respectively in the emanation from it of another
The preceding and succeeding chapters, the second and fourth, are closely related to the
third. Chapter 2 argues at length to prove that the rational soul is not ‘‘imprinted’’ in the
body and Chapter 4 o#ers a number of arguments to prove the soul’s immateriality and
incorruptibility and concludes with an argument in refutation of transmigration.
4
Avicenna, The Metaphysics of the Healing: A Parallel English-Arabic Text, translated
and introduced and annotated by Michael E. Marmura (Provo, Utah, 2005). The Arabic text
is basically that of the 1960 Cairo edition. The main discussions relevant to the theory of
the rational soul’s temporal creation are in Bk. IX, Chs. 3–5, and Ch. 7, on the hereafter. Bk
X, Ch. 1, is also relevant as it sums up Avicenna’s emanative system.
This reference will be abbreviated Metaphysics in the notes.
5
This is probably in part because the Psychology is part of the Physics – emanation
belongs to Metaphysics.
6
Psychology, p. 203, lines 6–7.
7
As indicated in note 1, above, these are primarily, the celestial intellects. Still, as we
shall indicate, there is an ambiguity in Avicenna’s statements regarding the role of the
celestial soul / souls in the emanation of the human rational soul.
8
This has been discussed in detail by a number of scholars. See in particular Herbert A.
Davidson’s Alfarabi, Avicenna and Averroes on Intellect: Their Cosmologies. Theories of the
Active Intellect. And Theories of Human Intellect (New York and Oxford, 1992), Ch. 4,
pp. 74–126.
9
‘‘The first,’’ in an ontological, not a temporal order.
intellect, another soul and another body, the body of the fixed stars.
This triadic process is repeated by every successive intellect. The last
of the intellects is the active intellect, from which the world of
generation and corruption emanates.
At this point, however, the process is reversed. What is now first
emanated is prime matter, which, however, must have form. It
acquires the form of at least one of the four basic elements, earth,
water, air and fire. As these elements combine in various ratios, new
forms emanate on them from the Active Intellect (the giver of forms),
resulting in a hierarchy of existents that constitute various kinds of
inanimate beings. Further combinations induce the reception of the
forms of organic beings, of plants and animals. It is the varied ways
in which the animal bodily humors, their amzija (singular, mizāj),
combine that determines the emanation of the forms that constitute
the di#erent animal species, and ultimately of the human rational
soul. In his al-Ishārāt wa al-Tanbı̄hāt (Pointers and Remarks)
Avicenna observes:
Look at the wisdom of the Creator. He began and created principles.10 He
then created from them ‘‘mixtures’’ (amzija) and prepared each ‘‘mixture’’
for a [distinct] species. He rendered the ‘‘mixtures’’ remote from [having]
equilibrium (al-i‘tidāl) so as to make the species remote from perfection. He
made the closest to [attaining] a possible equilibrium, the mizāj of the
human, so that the rational soul would use [the human body] as a
habitation.11
In Metaphysics Bk. X, Ch. 3, Avicenna states that the ‘‘individual
who is a prophet is not one whose like recurs in every period. For, the
matter receptive of a perfection like his occurs only in few bodily
compositions (amzija)’’.12 This indicates that the mizāj determines
the strengths or weaknesses of the rational souls in various humans.
Brief as this statement is, it relates his theory of the temporal origin
of the rational soul to his political philosophy, which maintains that
the majority of humans are incapable of attaining philosophical
knowledge. In this, Avicenna adopts the essentials of the political
philosophy of his predecessor al-Fārābı̄ (d. 950).13
10
As al-T* ūsı̄ explains it in his Commentary (p. 317 in note below), these are the four
elements, earth, water, air and fire.
11
Ibn Sı̄nā (Avicenna), al-Ishārāt wa al-Tanbı̄hāt ma‘a Sharh * Nas*ı̄r al-Dı̄n al-T
* ūsı̄ [with
the Commentary of Nas*ı̄r al-Dı̄n al-T * ūsı̄], ed. S. Dunya, 4 vols. (Cairo, 1960), vol. 2, pp. 316–
18. The use of the term istawkara, ‘‘to take as a nest’’ or ‘‘as an animal habitation’’ is
simply a manner of speaking, not to be taken literally. It does not entail the soul’s
‘‘dwelling’’ in the physical body, or being imprinted in it.
12
Metaphysics, p. 367, par. 1.
13
See M. E. Marmura, ‘‘The Islamic philosophers’ conception of Islam,’’ in R. G.
Hovannisian and S. Vryonis Jr. (eds.), Islam’s Understanding of Itself (Malibu, California,
1983), pp. 87–102, Part III, pp. 93–102. This article is reprinted in M.E. Marmura, Probing in
Islamic Philosophy (Binghamton, New York, 2004), pp. 391–408, Part III, pp. 397–408.
III
Turning to the problematic aspects of Avicenna’s theory of the soul’s

origination, we will begin with Psychology, Bk. V, Ch. 3. Here
Avicenna o#ers a proof for the human rational soul’s temporal
origination and discusses the related question of its individuation.
To show that the soul came to exist at a moment in time, he begins by
arguing that it could not have had a previous existence as an eternal
entity. For then it would have existed either as one of a multitude of
individual souls, or as one soul, one entity. Neither alternative, he
then argues, is possible:
The rational souls that come into temporal existence, though
numerous, are one ‘‘in species and meaning (al-naw‘ wa al-ma‘nā).’’
The Multiplicity of things takes place either due to ‘‘quiddity and
form (al-māhiyya wa al-s*ūra) or in relation to the elements and
matter that are multiple and by which they become multiple’’. There
is nothing in their quiddity and form that would individuate them. To
be individuated they require an association with a material body.
Since in their supposed prior existence as individuals there is no
body to individuate them, they could not have existed as individual
souls.14
If, on the other hand, we suppose that they had a prior existence as
one soul, this too is impossible. For, then, if two bodies came to exist,
two souls would come to exist in two bodies. In other words, the one
soul, which is not a spatial entity, would become potentially divisible
in relation to two spatial bodies. A non-spatial entity cannot in such
a circumstance be divisible. And if the one soul came to exist in more
than one human, then an obvious absurdity would ensue. For then
this would mean that knowledge in the one individual would be
shared by others.15 As Avicenna later on the discussion puts it, ‘‘if it
were one and in relation to many, it would have knowledge or
ignorance in all [individuals], [so that] what is in the soul of ‘Amr,
would not be hidden from Zayd.’’16
The chapter raises and discusses the question of the soul’s indi-
viduation. Avicenna admits throughout the chapter that some of the
individuating factors may not be known to us. In the course of
answering an objection to his theory, namely, that he cannot hold, as
he does, that the soul retains its individuality after the corruption of
the body, since there would be no body to individuate it, he indicates
some of the ‘‘initial’’ factors for its individuation:
After the separation of the souls from bodies, the souls would have been
brought into existence, each one of them an essence necessarily singled out
14
This is the gist of the argument in Psychology, p. 198, line 9–p. 199, line 3.
15
Ibid., p. 199, lines 4–12.
16
Ibid., p. 200, lines 10–11.
through the di#erences of their material that had been [associated with
them], through the di#erences of the times of their temporal creation, and
the di#erences of the dispositions17 they necessarily have according to their
di#erent bodies.18
We refer to the di#erences in materials and times as the ‘‘initial’’
factor because they are bodily. They are ‘‘bodily,’’ because they
either directly refer to the bodily temperament or the time this bodily
temperament is formed, inducing the reception of the soul. Avicenna,
as we shall indicate shortly, also speaks of things that happen
‘‘thereafter,’’ where the dispositions that separate one soul from
another, thereby individuating them, are no longer ‘‘bodily.’’ In a
statement that sums up some of the things he says about individua-
tion, the distinction between the initial individuating factors and the
subsequent ones is blurred. It is in this passage that we encounter
some of the expressions that are not fully explained. He writes:
No doubt that due to something (bi-amrin mā) [the souls] became individu-
ated and that that ‘‘something’’ in the human soul does not consist in its
being imprinted in matter. For the falseness of holding [this view] has been
shown. Rather, this is one of the dispositions (hay’a min al-hay’āt),19 one of
the powers (quwwa min al-quwā), one of the spiritual accidents (‘arad * min
al-a‘rād
* al-rūh
* āniyya) or a number of these that in combination individuate
it, even if we are ignorant of them.20
What Avicenna is referring to when he speaks of ‘‘dispositions’’ is
not made clear, but at least in what follows we can get some sense of
his meaning. It is his statement about the ‘‘spiritual accidents’’ that
is in need of explanation. Before turning to this expression, it is best
to quote what comes after the above statement, as it sheds some light
on what he means by ‘‘dispositions’’ and ‘‘powers.’’
We are certain that it is possible that once the soul comes into existence
with the coming into existence of some temperament (mizāj) that there
would temporally come to exist for it thereafter [my italics] a disposition (an
tah
* duth lahā hay’a ba‘dahu) in the rational acts and a#ections that are on
the whole distinct from the dispositions similar to it in another, in the way
the two temperaments in the two bodies are distinct. [It is also possible],
that the acquired disposition called the ‘‘intellect in act’’ would have some
* addin mā)21 that di#erentiates it from other souls,
[sort of] definition (‘alā h
17
There seems to be an ambiguity here – it is not clear whether this (as it seems) refers
to the initial causes of individuation, the subsequent causes, or both.
18
Ibid., p. 200, lines 4–6.
19
The term hay’a in this context is di$cult to translate. Its nuance is that of a general
state of a#airs characterizing a thing. Afnan, in his Lexicon, under hay’at, translates it aptly
as either ‘‘state’’ or as ‘‘disposition.’’ See Soheil Afnan, A Philosophical Lexicon in Persian
and Arabic (Beirut, 1968), p. 325.
20
Ibid., p. 200, lines 17–20.
21
That is, not a definition in the strict sense that applies to the universal quiddity, but
not to an individual. See Avicenna’s detailed discussion of this in Metaphysics, V, Ch. 8, pp.
186–9, particularly p. 188. par. 8 #., where Avicenna shows how an individual like Socrates,
and that there would occur to it an awareness of its individual essence,22

this awareness being some disposition in it, proper to it, not belonging to
another. It is possible that there will occur in it from the direction of the
bodily powers also a special disposition, this disposition being related to the
moral dispositions or identical with them. [It is also possible] that they
would have other properties, hidden from us, that necessarily adhere to the
souls with their temporal origination and thereafter, in the way the similar
individual [instances] of the bodies of [di#erent] species adhere to them,
whereby [the species] are di#erentiated by them as long as [the individual
instances], exist.23 The souls would likewise be di#erentiated by properties
in them, regardless of whether there are bodies or no bodies, and regardless
of whether or not we know these states, or know only some of them.24
In this passage, one must distinguish between the initial cause of
individuation, namely, bodily temperament and the time in which the
rational soul comes to be and what happens ‘‘thereafter,’’ namely,
the development of dispositions, such as having ‘‘the intellect in act’’
and that of self awareness, that are no longer directly dependant for
their individuation on bodies, but are unique to each soul, making it
an entity separate from every other rational soul.
To return to the expression, ‘‘one of the spiritual accidents,’’ what
does Avicenna mean by it? This is not immediately clear. We have
two terms here, ‘‘spiritual’’ and ‘‘accidents.’’ To begin with the term
‘‘spiritual,’’ what naturally comes to mind is that this refers to what
is not physical, to what in religious terms is ‘‘holy.’’ Avicenna
himself uses it in this sense, at least once in the Psychology, when he
refers to the prophet’s mind as ‘‘the holy spirit’’ (al-rūh
* al-qudsiyya)
in its revelatory prophetic function.25 Again in a di#erent treatise,
Avicenna is critical of the Christians who uphold the doctrine of
bodily resurrection, yet maintain that rewards and punishment in
the hereafter are not physical, but spiritual.26 Is this then, the
cannot be defined, but known only, through direct observational reference and, by
implication, a veridical transmitted report of such a direct observation.
22
This is a reference to the soul’s immateriality and relates to the famous passage in
Psychology, Bk. I, Ch. 1, p. 13, lines 9–10, of the example of the flying man,’’ where one can have
a direct awareness of one’s own existence, totally independent of one’s body and its faculties.
23
That is, as long as there are individual instances of the species in the terrestrial world
that are in the constant flow of generation and corruption. It is these continuous individual
instances of the species that di#erentiate one species from another.
24
Ibid., p. 201, lines 1–10.
25
Ibid., Bk. V, Ch. 6, p. 219, line 17.
26
That Avicenna is only too well aware of the use of the term ‘‘spiritual,’’ as being none
bodily is seen in Risāla Ad *h
* awiyya – in his criticism of the Christians who believe in bodily
resurrection, but maintain that reward or punishment in the hereafter are not bodily.
Avicenna writes: ‘‘If reward and punishment are spiritual, what purpose would there be for
the resurrection of the body? Moreover, what is this spiritual reward and spiritual
punishment and how is it portrayed to [the masses] so that they would desire and fear [it]?
No! Nothing of this is portrayed to them other than their being in the hereafter like angels.
If it were portrayed to them regarding the matter of spirituality other than this, they would
go astray in understanding it, [thereby] understanding something other than what is said to
them. However, what the masses imagine of the a#air of the angels – even though they do
non-physical sense in which Avicenna is using the term? This seems
unlikely, particularly when we turn to the term ‘‘accidents,’’ which
suggests that this would be an initial individuating factor, not
something that happens ‘‘thereafter.’’
When discussing the soul’s initial individuation, Avicenna states
that human souls become individuated as single souls within their
species ‘‘through states that are not necessary for them inasmuch as
they are souls.’’27 In other words, the individuating causes, that is,
what we termed the ‘‘initial’’ ones, are not intrinsic to the souls, but
are accidental occurrences. He adds that ‘‘the accidents that attach
[to the souls] follow a beginning that is necessarily temporal because
it follows a cause that occurs to some but not others.’’28 Can what is
spiritual in the non-physical sense be an accident extrinsic to
Avicenna’s immaterial souls? Surely, this can hardly be the case, and
hence one has to search for another meaning of ‘‘spiritual.’’
The Psychology provides us with another meaning, though in a
di#erent context. In Book II, Ch. 2, but more fully in Bk. V, Ch. 8,
entitled, ‘‘Expounding the [Bodily] organs that belong to the Soul.’’
He states that ‘‘the first thing that carries the bodily faculties
belonging to the soul (al-quwā al-nafsāniyya al-badaniyya) that
passes through the [internal bodily] outlets is a subtle spiritual body
(jism lat*ı̄f rūh
* aniyy) and that this body is spirit (wa inna dhālika
al-jism huwa al-rūh * ).’’29 Avicenna is primarily concerned in this
chapter is with the connection of the soul to the bodily organs and
the connection of the bodily organs to each other. Spirit, the subtle
body, acts as the substratum for the vegetative and animal soul and
as the intermediary that allows the immaterial rational soul to act on
the material body. A main point that Avicenna makes is that,
through the mediation of this subtle body, this spirit, the rational
soul’s first attachment is to the heart. This is the principle organ
from which the other powers of the organs emanates through the
mediation of the spirit.
not dare say it – is that the angels are miserable, having neither pleasure nor rest, neither
eating, drinking, nor marrying, and that they praise and worship [God] night and day, and
in the end are not rewarded for it.’’ Risāla, p. 13, line 9–p. 14, line 2.
27
Psychology, p. 199, lines 9–10.
28
Ibid., lines 10–11.
29
Ibid., p. 232, lines 11–12. Avicenna seems to be in agreement with the distinction
between spirit and soul articulated by the physician, translator and philosopher, Qust*ā Ibn
Lūqā (d. 900). Thus H.Z. U } lken in his edition of a number of Avicenna’s treatises, Ibn Sina
Risāleleri, 2 (Ankara, 1953): 83–94, not inappropriately includes Qust*ā Ibn Lūqā’s treatise
on the di#erence between spirit and soul where Ibn Lūqā writes: ‘‘The first di#erence
between the two is that spirit is a body, while the soul is not a body; that spirit is contained
in the body, whereas the soul is not; that when the spirit leaves the body, it ceases to exist,
whereas the soul’s acts on the body cease, but it itself does not cease to exist; that the soul
changes the body and bestows life on it through the mediation of the spirit, whereas spirit
does this without mediation.’’ Kitāb al-Farq Bayn al-Rūh * wa al-Nafs, Ta’lı̄f Qust*ā ibn Lūqā
al-Yūnānı̄ [The Book on the Difference between Spirit and soul, authored by Qust*ā Ibn Lūqā
the Greek], in Risāleleri, 2, p. 93.
The primary concern of the chapter is physiology as it relates to

psychology. It is not concerned with the question of individuation as
such. Nonetheless, it brings it to mind indirectly. This is when the
temperament, the mizāj, of this spirit, this rūh
* is briefly discussed.
[The spirit] has a special temperament (lahu mizāj makhs*ūs* ) and its
temperament also di#ers according to the need for the di#erence it under-
goes by means of which it becomes a carrier for di#erent faculties. For the
temperament with which it is angry is not suitable for the temperament with
which it desires and senses. Nor is the very temperament suitable for seeing
the one suitable for the motive spirit.30
The point of this quotation is that it indirectly relates the spirit to
what we have termed the initial stage of individuation, as its concern
is with bodily temperaments. It has little to do with the subsequent
development of individuating dispositions such as ‘‘self-awareness,’’
where the rational soul has direct awareness of itself, entirely
independent of any bodily faculty. In the Psychology, the distinction
between the initial and subsequent stages of individuation is made,
as it were, in passing. It is easily missed. Moreover, as we have noted
in commenting on a key passage,31 the distinction is blurred.
IV
Another di$culty one encounters in understanding Avicenna’s

theory of the soul’s temporal origination is what appears to be a lack
of consistency in his use of certain terms. These terms refer to what
is meant by al-‘aql al-kulliyy, ‘‘the universal intellect,’’ al-nafs
al-kulliyya, ‘‘the universal soul,’’ and by ‘aql al-kull, literally, ‘‘the
intellect of ‘whole’ ’’ and nafs al-kull, ‘‘the soul of ‘the whole’.’’32
In his treatise on definitions, Kitāb al-H * udūd,33 Avicenna defines
these terms. He makes it clear that the expression ‘‘universal
34
30
Ibid., p. 232, line 15–p. 233, line 21.
31
See note 28, above.
32
Risāla, pp. 114–21.
33
Avicenne, Livre des Définitions [Kitāb al-H * udūd], édité, traduit et annoté par A-M.
Goichon (Cairo, 1963). This is a revised version of her initial 1933 edition. The early edition
was based on the two manuscripts then available to Goichon – the revised edition is based
on a selection taken from 15 manuscripts. She uses as its base a University of Istanbul MS.
(A Y 4711) dated 579 / 1183, known to be the earliest of the dated manuscripts. In her
French translation, Goichon follows this basic edition. (References will be made to the Arabic
Text of Goichon’s 1963 edition of Kitāb al-H * udūd, abbreviated in the notes as H * udūd). An
English translation, based on Goichon’s 1963 edition of the Arabic text, with commentary, is
included in Kiki Kennedy-Day, Books of Definition in Islamic Philosophy (London and New
York, 2003), pp. 28–114 (of the translation); pp. 115–159 (of commentary). The best known of
earlier printings of the collection of nine Avicennan treatises, Tis‘ Rasā’il, that include the
treatise on definitions is the 1908 Cairo printing. The substance of the definition of the terms
mentioned above is also found in Avicenna’s letter to al-Jūzjānı̄, included in Ajwiba ‘an
‘Ashr Masā’il [Answers to Ten Questions], in Risāleleri, 2 p. 78 #.
34
H* udūd, pp. 14–16, sections 26 and 27.
intellect’’ is the meaning ‘‘said of,’’ (al-maqul ‘alā)35 the numerically
many intellects that belong to individual people. This ‘‘universal
intellect’’ has no independent subsistence, that is, no extra mental
existence. It exists ‘‘only in conception.’’36 In other words, he is not
using the term to refer to the celestial intellects in his triadic
emanative system. He is speaking of the universal concept humans
have, which for Avicenna exists only in the mind.
The same applies to the expression, ‘‘universal soul.’’ It ‘‘is the
meaning said of many that are numerically di#erent, in answer [to
the question] ‘what is it?’ where each [‘of the many’] is a soul
belonging specifically to an individual.’’37 The ‘‘individual,’’ the
shakhs*, again, is a human. Avicenna does not elaborate on this, but
it is clear, by the parallel of this to the ‘‘universal intellect,’’ that he
is speaking of a universal meaning, predicable of all human souls – a
universal meaning that exists only in conception, not in external
reality.
Turning to the definition of ‘aql al-kull, and its parallel nafs
al-kull. Al-kull, ‘‘the whole,’’ refers to the universe. ‘Aql al-kull can
refer to two things. The first is that it designates ‘‘the totality of the
entities / essences that are denuded from matter in all respects
(jumlat al-dhawāt al-mujarrada ‘an al-mādda min jamı̄‘ al-jihāt),’’38
terminating with the intellect that acts in / on human souls (al-‘aql
al-fa‘‘āl fı̄ al-anfus al-insāniyya).39 The second is that it refers to the
first intellect that emanates from God, the intellect of the outermost
sphere. In a parallel fashion, nafs al-kull refers either to the totality
of the celestial souls, or to the soul of the outermost sphere, which
moves that sphere, whose movement encompasses the movement of
35
Sometimes this expression is used as an equivalent of ‘‘predicated of.’’
36
Ibid., p. 14, section 26, lines 1–3.
37
Ibid., p. 15, section 27, lines 1–3, or, in Kennedy-Day’s translation, ‘‘the universal soul
is understood as referring to di#erent individuals, in answering the question, ‘What is it?’
For each of these individuals, a particular soul belongs to an individual.’’ Kennedy-Day,
Books of Definition, p. 104, section 27, lines 2–4.
38
Ibid., p. 14, section 26, lines 6–7. The reference here is to the celestial intellects that
terminate with the Active intellect. The text also adds that these ‘‘are not moved, either
essentially or accidentally, but only through desire.’’ This is according to the manuscript
used by Goichon as the base. The Bodleian (Oxford) manuscript given in the notes reads
tuh* arrik, ‘‘that move,’’ that is, that cause movement, not tatah * arrak, ‘‘are moved.’’ The
Oxford reading indicates that the celestial intellects, as the objects of desire, are the
movers. Regarding the above definition of ‘aql al-kull, an ambiguity lingers in Avicenna’s
writings; it is not always entirely clear whether (using this definition) references to ‘aql
al-kull apply to all the celestial intellects taken together or to each and every one taken
singly, or whether taking them singly is included in the term applying to them collectively.
One also notes that in the H * udūd (par. 24) there is a general and a specific definition of
‘‘active intellect.’’ The general definition covers the more specific meaning, namely, the
Active Intellect, that by its illumination actualizes the material intellect. The more general
meaning is with respect to its simply being an intellect. It is a formal substance that is in
itself denuded from matter, not denuded by some other thing.
39
the entire celestial spheres. Each sphere has in addition its own
motion. In each of the emanated triads, the sphere’s motion is caused
by the soul’s desire for the intellect of the triad.
The substance of the definition of the terms mentioned above is
also found in Avicenna’s letter to al-Jūzjānı̄,40 but with a di#erence.
This di#erence is in the definition of al-Kulliyy. After defining it as in
Kitāb al-H
* udūd, where it is a universal that exists only in the mind,
he tells us that in addition ‘‘it is said of one thing in existence
that is related to many or to ‘a whole’’’ (wa yuqāl li-shay’ wāh * id fı̄
al-wujūd yunsab ilā kathı̄rı̄n aw ilā kull). In other words, al-kulliyy is
not used exclusively to refer to the universal concept that exists only
in the mind, but can refer to ‘‘one thing in existence that is related
to many or to ‘a whole’.’’ Hence, we meet instances in Avicenna’s
writings belonging to di#erent periods where al-kulliyy refers to a
celestial intellect or a celestial soul.
In one of Avicenna’s early treatises on the soul, al-Mabh * ath ‘an
al-Quwā al-Nafsāniyya (Investigating the Faculties of the Soul),41
probably the treatise Avicenna refers to in one of his later works,42 as
having been written ‘‘at the beginning of my life, forty years ago,’’43
the Active Intellect, is referred to as al-‘aql al-kulliyy.44 In his Risāla
al-Ad*h
* awiyya, where the indications are that it is not an early work45
we meet what appears to be an identification of al-nafs al-kulliyy[a]
and al-‘aql al-kulliyy with nafs al-kull and ‘aql al-kull. In this
treatise, after a detailed exposition of theories of transmigration,
Avicenna argues that these theories rest on two false premises,
namely, that the rational soul is a material form and that souls before
joining a particular body had a previous existence. He then sums up
his own theory of the rational soul’s temporal origination as follows:
Therefore the truth is that the soul is created with the temporal coming to
be of the bodily temperament (al-mizāj al-badaniyy). For the bodily tem-
perament is a cause for the body’s becoming receptive – [either] from the
universal soul, or (aw) from the universal intellect, or from one of the
40
This is in Ajwiba ‘an ‘Ashr Masā’il [Answers to Ten Questions], Risāleleri, 2; p. 78 #.
41
Included in F. Ahwani’s Ah * wāl al-Nafs (Cairo, 1952), a copy of the 1876 edition by S.
Landauer in ‘‘Die Psychologie des Ibn Sina,’’ Zeitschrift der Deutschen Morganlandoschen
Geselschaft, 29 (1876): 335–418 (Arabic text, pp. 339–72).
42
Risāla fı̄ al-Kalām ‘an al-Nafs al-Nāt*iqa (An Epistle, a Discourse on the Rational Soul),
Included in Ahwani’s Ah * wāl al-Nafs, pp. 193–9.
43
Ibid., p. 199, line 5.
44
Laundauer, ‘‘Die Psychologie des Ibn Sina,’’ p. 371; p. 177 in Ahwani.
45
Yahya Michot has discovered a manuscript of Avicenna’s youthful work written to the
vizier Abū Sa‘d, published as Ibn Sı̄nā: Lettre au vizir Abū Sa‘d (Beirut, 2000). The edition
of the text with its French translation constitutes a significant contribution to Avicennan
studies. Michot, however, argues that the Risāla Ad *h
* awiyya is also an early work,
dedicated to the same vizier, Abū Sa‘d. Michot’s argument certainly has its points, but is
not entirely convincing. There are indications in content and style strongly suggestive that
it belongs to a later period. Michot has not really explained why, in the extant manuscripts,
the dedication is never to the above mentioned vizier, but to Abū Bakr Ibn Muh * ammad.
separate causes – of the substance of the soul through which the species of
that body is perfected.46
In the above, the phrase, ‘‘[either] from the universal soul, or from
the universal intellect, or from one of the separate causes’’ is a
translation (partly a paraphrase because of the di$culty of translat-
ing al-kulliyyayn) of min al-nafs aw al-‘aql al-kulliyayn, aw sabab min
al-asbāb al-mufāriqa. Al-kulliyayn is normally the dual of al-kulliyy.
If this is the case here, then clearly, al-kulliyy is used to refer to nafs
al-kull and ‘aql al-kull. Avicenna could have been using the dual
form as a ‘‘short hand’’ for indicating the dual of nafs al-kull and ‘aql
al-kull. But in view of what he said in the letter to al-Jūzjānı̄, this
seems unlikely. Taking al-kulliyayn as the dual of al-kulliyy accords
with what he says in the letter al Juzjānı̄.
The above translation is based on ‘A z *sı̄’s edition. The passage in
Lucchetta’s edition referring to the universal soul and universal
intellect, translates : ‘‘For the bodily temperament is the cause of the
body’s becoming receptive from the universal soul and (wa) the
universal intellect (min al-nafs wa al-‘aql al-kulliyyayn),47 or from
one of the separate causes, of the substance of the soul [. . .]’’.48
In both editions we note a number of things in common. One is the
order of mentioning al-nafs, before al-‘aql. One also notes that in
‘A
z *sı̄’s edition the emanation of the human rational soul comes from
either the ‘‘celestial soul,’’ or the ‘‘celestial intellect.’’ As we shall
see in Part V below, the ambiguity in the relation of nafs al-kull to
the individual human soul persists. It also persists in Lucchetta’s
edition where the emanation of the rational soul comes from both the
celestial nafs and the celestial ‘aql, or from one of the separate
causes. The suggestion here is that celestial nafs and the celestial
‘aql are conjoined as an alternative cause of the soul’s emanation.
In one place in the treatise Fı̄ Ithbāt al-Nubuwwāt (On the Proofs
of Prophecies),49 the two are conjoined. It is said that the acquired
46
Risāla, p. 123, lines 12–13.
47
Here again, we have the expression al-kulliyyan, as in ‘A z *sı̄’s edition.
48
Lucchetta, Epistola Sulla Vita Futura, p. 129, lines 4–6.
49
Ibn Sı̄nā, Fı̄ Ithbāt al-Nubuwwāt, edited and introduced by Michael E. Marmura
(Beirut, 1968). References to this work will be abbreviated, Ithbāt.
Doubts about the attribution of this treatise to Avicenna were raised (not unreasonably)
by Davidson in his Alfarabi, Avicenna and Averroes, on Intellect, p. 87, n. 56 and p. 91, n. 74
continued on p. 92.
The question of its authenticity was discussed in the English and Arabic introductions to
the edition and reason were given for the belief that it is authentic. The question was raised
primarily because the Introductions at that time stated that none of the standard medieval
Arabic sources listing Avicenna’s works included it. In the more detailed Arabic
Introduction the earliest of the authors mentioned was al-Bayhaqı̄ (d. 1170). In the
biographical account of Avicenna al-Bayhaqı̄ makes no mention of the treatise. In another
part of his Tatimmat S * iwān al-H
* ikma (Lahore, 1351 / 1932), a part overlooked by the editor,
a further listing of Avicenna’s works is given. This part includes on p. 178, line 17, the title
Risāla fı̄ Ithbāt al-Nubuwwa (A Treatise on the Proof of Prophecy). The term ‘‘prophecy’’ in
intellect in humans derives from ‘‘ ‘the universal intellect (al-‘aql

al-kulliyy)’, ‘the universal soul (al-nafs al-kulliyy50 )’, and ‘the soul of
the world (nafs al-‘ālam)’.’’51 In the succeeding paragraph, the Active
Intellect from which the primary intelligibles are received by the
human rational soul is referred to as al-‘aql al-fa‘‘āl al-kulliyy.52
Again, in another place, where the context indicates that what is
being referred to is ‘aql al-kull, the first intellect emanating from
God, this is called al-‘aql al-kulliyy.53 Such di#erent uses of these
terms without proper explanation is not peculiar to this treatise.54
We also encounter, in a treatise attributed to Avicenna, mention of
a habitual use of the term kulliyy to refer to both the celestial
intellect and the celestial soul. Thus we find the statement, ‘‘the
human soul derives from a divine emanation that connects with it
and a$xes it with its nature, [the source of the emanation] being in
itself a substance; this is among the things it has become a habit to
name ‘the universal intellect’ and ‘the universal soul’ (wa huwa
mimmā jarat al-‘āda bi-tasmiyatihā al-‘aql al-kulliyy wa al-nafs
al-kulliyy).’’55 [Unless bi-tasmiyatihā is a scribal or printing error, its
feminine ending would refer to al-‘āda].
the singular is given in five of nine manuscripts that give a title to the treatise. On p. 64,
line 13, of a more recent version of al-Bayhaqı̄’s work, edited by D.R.F. al-‘Ajam (Beirut,
1994) that utilized a manuscript in the American University of Beirut, the above title is also
given, this time in the biographical account of Avicenna.
In raising the question of the treatise authenticity its prose has to be taken into account.
It begins in the bah * thı̄, ‘‘investigatory’’ peripatetic style characteristic of much of
Avicenna’s writings. The language is detached and analytic. It continues to be so in the
exegetical part of the treatise until we reach paragraph 38 on p. 56. The earlier somewhat
pedantic prose now acquires a higher literary quality and becomes more personal. It is
di$cult to read ‘‘fa-mā halaka man halaka . . .’’ and imagine that it is not Avicenna who is
speaking. The same applies to subsequent paragraphs, notably, the somewhat Dantesque
paragraph 40.
50
One notes that the Rampur manuscript used in the edition has the ungrammatical
al-‘aql al-kull and al-nafs al-kull, possibly a corruption of ‘aql al-kull and nafs al-kull
(Ithbāt, p. 44). This is also found in the second Rampur manuscript, not used in the edition,
which seems to be a copy of the one used.
51
52
Ibid., p. 44, par. 8, second line. Here we encounter the ambiguity mentioned in note 38
above, namely, whether ‘aql al-kull, in its first definition refers to all the celestial intellects
collectively, or also to each singly.
53
Ibid., p. 52, par. 29. Again the Rampur manuscript has it as al-‘aql al-kull.
54
What is perhaps peculiar is their appearance in one and the same treatise, but as
indicated above, they appear in di#erent contexts. At the beginning of the treatise we are
told that it is a ‘‘summary’’ of a conversation. Perhaps its ambiguities are partly due to this.
55
This treatise is included in a collection of works attributed to Avicenna, first published
in Majmū‘ Rasā’il al-Shaykh al-Ra’ı̄s (Hayderabād, 1353 / 1935), Risāla fı̄ al-Sa‘āda wa
al-H* ujaj al-‘Ashra ‘alā anna al-Nafs al-Insāniyya Jawhar (A Treatise on Happiness and the
Ten Arguments that the Human Soul is a Substance ). It consists of 22 pages The printing is
based on two manuscripts, but these are not identified. The quotation is from p. 13.
This work in Gohlman’s ‘‘Bibliography of the works of Ibn Sina’’, by an unknown
medieval Arabic source, is entitled ‘‘An essay on the Attainment of Happiness, which is
known as the Ten Arguments ( Maqāla fı̄ Tah **sı̄l al-Sa‘āda, wa Tu‘raf bi-al-H
* ujaj al-‘Ashr).’’
It is listed as number 32. W.E. Gohlman, The Life of Ibn Sina (Albany, New York, 1973),
p. 100 (Arabic text), p. 101 (translation).
What is really ambiguous is the role of the celestial soul / souls
play in the temporal creation of the human rational soul. As we turn
to the Metaphysics of the Healing, we find significant statements that
can be interpreted as o#ering a resolution to this ambiguity.
The emanation of the celestial intellect terminates with ‘‘the last of

them which is close to us.’’ From it ‘‘there emanates with the
participation of the celestial movement something having the con-
figuration of the forms of the lower world by way of receptivity to
action.’’56 Clearly, this intellect ‘‘which is close to us,’’ is none other
than the Active Intellect. We note that the emanation occurs in
participation with the celestial movements, a point to which we will
shortly return. As indicated earlier, with the Active Intellect the
emanation process is reversed. The first thing to come into existence
is matter, which cannot exist without form. It thus acquires the form
of the four basic elements, and, as these combine, formed matter
acquires, from the Active Intellect, the giver for forms, higher forms,
leading ultimately to bodily humoral combinations that induce the
emanation of the human rational soul.
We also notice in the above quotation a reminder that the Active
Intellect is the last of the celestial intellects. As such, its existence,
like theirs, is in itself only possible, necessitated by its proximate
celestial intellect, and ultimately by the Necessary Existent, God.
This raises the question of why Avicenna does not invoke the three
acts of cognition to explain the emanation of existents in the
terrestrial world. It seems implicit in his system that a triad of
existents would still continue, this time, however, only as a triad of
kinds, not of individuals.57
Avicenna’s answer to this is included in his answer to a larger
question, namely, why does the emanative process of celestial intel-
lects stop with the Active Intellect. His answers relates to the
question of the role the celestial souls play in the emanation of
terrestrial existents, as it puts this role in its proper cosmological
setting. After a discussion of the process of celestial emanation with
the idea that the triadic contemplative acts produce celestial triads,
Avicenna answers the larger question mentioned above as follows:
It does not necessarily follow that this idea should proceed ad infinitum
such that under each separate [intellect] there would be a separate
[intellect]. For we say: If the existence of multiplicity proceeds necessarily
56
Metaphysics, Bk. IX, Ch. 5, p. 335, par. 3, lines 12–15.
57
That is, of species of bodies, souls and intellects, namely, the individual human
rational souls that remain one ‘‘in species and meaning.’’ Psychology, Bk. V, Ch. 3, p. 198,
line 11.
from the intellects, this is due to the multiplicity of meanings in them. This
statement of ours, however, is not convertible whereby every intellect
having this multiplicity would have to have, as a necessary consequence of
its multiplicity, these e#ects. Nor are these intellects identical in species, so
that the consequences of their meanings are identical.58
Avicenna thus acknowledges that the Active Intellect also has the
triadic ideas. But this does not mean that what emanates from it must
be a triads of individuals. Otherwise, the process would proceed
indefinitely and will not explain the existence of the terrestrial
world. For Avicenna with his triadic emanate scheme is also engaged
in giving a philosophical basis for his own modified Ptolemaic
astronomy. But what about the idea that the triad can continue in
the world of generation and corruption as a triad of kinds? This,
however, would violate Avicenna’s idea that from the one only one
proceeds. To be sure, this principle applies above all to the Necessary
Existent. But it also applies to the emanated intellects, where from
each of the three acts of contemplation, only one thing emanates.59
To invoke the cognitive principle that applies to the emanation of
celestial entities to the Active Intellect would violate this principle
since the emanation of the terrestrial world entails pluralities of
individual existents.
We will now turn to Avicenna’s first paragraph of Ch. 5 of
Metaphysics Bk. IX:
Once the celestial spheres’ number become complete, the existence of the
elements follow necessarily thereafter. This is because the elemental bodies
are generable and corruptible. Thus, their proximate causes must be things
that receive a species of change and motion; [and it must be the case] that
that which is a pure intellect is not alone a cause for their existence.60
It is the reference to the ‘‘proximate causes’’ that ‘‘must be things
that receive a species of change and motion,’’ that brings us to role
the celestial souls play in the temporal origination of the human
soul. These souls are the causes for the eternal motions of the
spheres. The soul of the outermost sphere moves the sphere through
its desire of the intellect of that sphere. This is the pervasive motion
that contains the motions of all the subsequent spheres. Each of
these also has its own motion, each motion in the triad being caused
by the desire of each soul in each triad for the intellect of that triad.
All these diverse motions interact, causing a complexity of causes
a#ecting the changes and transformations in the terrestrial world.
These changes, as indicated earlier, lead to the formation of the
bodily humoral composition that induces the reception of the human
58
Metaphysics, Bk. IX, Ch. 4, p. 331, par. 12, lines 22–30.
59
Ibid., p. 330, par. 11, lines 23–26.
60
Ibid., p. 334, par. 1, lines 12–14.
rational soul. The role of the outermost celestial soul and that of
each of the successive celestial souls is that of a preparatory,
indispensable aid for the emanation of the human rational soul from
the celestial intellect / intellects. Cognizance of this role should
clear some of the ambiguities we have met in Avicenna’s statements
about the source of this emanation.
The Metaphysics, however, makes other statements about the
celestial soul / souls that remain unexplained. After a lengthy expo-
sition showing how the di#erences in the celestial spheres and their
motions cause their counterparts in terrestrial bodies, acting as aids
for the emanation of specific characteristics of terrestrial bodies,
Avicenna states the following:
And just as the common and proper nature there [in the celestial realm] are
principles or aids for the nature of proper and common [things] here,
similarly, what necessarily accompanies the proper and common things
there by way of the various changing relations that take place in them by
reason of motion is a principle for the change of states and transformation
here. Similarly, the mixing of their relations there is either a cause or an aid
for the mixing of the relations of these elements [here].61
Thus far there is nothing startling in what is being said, except for
the statement about the mixing of ‘‘relations there is either a cause
or an aid for the relations of these elements here.’’ That these are ‘‘an
aid’’ is not di$cult to understand. The alternative, of their being a
cause, however, requires an explanation which is not given. This is
followed with the statement:
The celestial bodies also have an influence on the bodies of this world
through the qualities proper to them, some of which flow to this world. Their
souls also have an influence on the souls of this world (wa li-anfusihā ta’thı̄r
* an fı̄ anfus hādhā al-‘ālam).62 Through these ideas it becomes known
ayd
that the nature that governs these bodies in terms of perfections and form
comes to be from the soul dispersed in the heavens or through its aid.63
Again, the use of the term ‘‘cause’’ calls for explanation. We have
two statements that are not clear. The first is the manner in which
some of the qualities of the celestial bodies influence the terrestrial
bodies. In what sense do they ‘‘flow to this world?’’ Is this some kind
of emanation?
The second is the statement that the celestial souls have an
influence on the souls of this world. The souls in this world include
human souls. How do the celestial souls ‘‘influence’’ them? Is it
merely as an aid to their being emanations, or is there something
61
Ibid., p. 336, par. 7, lines 33-p. 337, line 1.
62
The translation of this sentence in the original typescript is missed in the present
edition.
63
more to this influence? The nuance of Avicenna’s statement suggests

that there is something more.
For this ‘‘something more,’’ one has to go back to the Psychology,
where the influence of the celestial souls on human souls is dis-
cussed.64 One must recall here that the celestial souls, unlike the
celestial intellects that have knowledge of particulars only ‘‘in a
universal way,’’65 have knowledge of particulars in their particular-
ity. This knowledge is sometimes transmitted to ordinary people in
their dreams in the form of images. These images symbolize knowl-
edge the celestial souls have, but which with ordinary human beings
requires proper interpretation. Prophets, however, are favored to
receive such knowledge, which includes knowledge of future contin-
gents, in their waking hours. This represents one type of prophetic
revelation involving the prophet’s imaginative faculty.66
The questions raised in this paper should not obscure the place of
the human rational soul within Avicenna’s cosmological view. There
are, as we have tried to indicate, some uncertainties about what he
precisely means in the exposition of his theory of the soul’s temporal
origination. The main outline of the theory, however, remains
comprehensible and its centrality to his entire philosophical system
clear.
64
Psychology, Bk. 4, Ch. 2, p. 154 #.
65
See the author’s, ‘‘Some aspects of Avicenna’s theory of God’s knowledge of
particulars,’’ Journal of the American Oriental Society, LXXXII, 3 (1962): 299–331, reprinted
in Probing in Islamic Philosophy, pp. 71–95.
66
A higher form of prophetic revelation is the intellectual, received by the prophet’s
rational faculty from the celestial intellects. This is intellectual, conceptual knowledge. It is
received by the prophet’s rational faculty directly as an intuition. Some of this knowledge
can descend from the prophet’s rational faculty to his imaginative faculty, transformed into
images that imitate it and symbolize it. Psychology, Bk. 4, Ch. 6, pp. 219–20.
IN MEMORIAM
Muhsin Sayyid Mahdi (1926–2007)
Muhsin Sayyid Mahdi, James Richard Jewett Emeritus Professor

of Arabic at Harvard University, died on July 9 in Brookline,
Massachusetts, after a long series of illnesses.
Universally acclaimed as the doyen of medieval Arabic and Islamic
philosophy, Professor Mahdi was born June 21, 1926 in the Shiite
pilgrimage city of Kerbala, Iraq, of a father who practiced medicine
according to the principles of Galen. He pursued his elementary and
early secondary school studies there, but went to Baghdad for the
last two years of secondary school. Awarded an Iraqi government
scholarship to study business administration at the American
University in Beirut, he found himself so attracted to philosophy
while there that he fulfilled the requirements for a major in both
subjects. After a year as a lecturer in economics at the University of
Baghdad (1947–1948), Mahdi won another scholarship – this one to
study economics at the University of Chicago.
Not long after his arrival at Chicago in 1948, he began to study
with Arnold Bergsträsser, Yves Simon, and, above all, Nabia Abbott
and Leo Strauss. Economics gave way to philosophy, especially to
the recovery of the history of Arabic and Islamic philosophy, and
Mahdi entered the Committee on Social Thought – having as class-
mates both Seth Benardete and Allan Bloom, the latter becoming a
life-long friend. He finished his Ph.D. studies in 1954, submitting a
brilliant dissertation that was published shortly afterwards as Ibn
Khaldun’s Philosophy of History: A Study in the Philosophical
Foundation of the Science of Culture. After an interlude as a visiting
lecturer in the Seminar für Wissenschaftliche Politik and in the
Orientalisches Seminar at the University of Freiburg im Breisgau,
Mahdi returned to Baghdad and took up positions in the Law College
and the Faculty of Arts and Sciences at the University of Baghdad.
In 1957, he accepted a position as assistant professor in the
Department of Oriental Languages and Civilizations at the Univer-
sity of Chicago where he remained, rising to the rank of full
professor, until 1969. He resisted numerous o#ers from di#erent
institutions during this time, but finally accepted Harvard Univer-
sity’s o#er of the James Richard Jewett Professorship in Arabic. At
Harvard from 1969 until his retirement in 1996, he served as director
140 IN MEMORIAM
of the Center for Middle Eastern Studies and also as chair of the
Department of Near Eastern Languages and Civilizations.
Professor Mahdi conducted postdoctoral study and research at the
University of Paris in addition to the University of Freiburg im
Breisgau. He was a Rockefeller Foundation Research Fellow and a
Fulbright research scholar in Morocco. He held visiting professor-
ships at the University of Freiburg im Breisgau, the American
University in Cairo, the Central Institute of Islamic Research in
Pakistan, the University of California-Los Angeles, and the Univer-
sity of Bordeaux. Long a member of the Advisory Council for the
Department of Near Eastern Studies at Princeton University, he was
a founding member and president of the Société Internationale pour
l’Histoire des Sciences et de la Philosophie Arabes et Islamiques
(SIHSPAI), as well as founding member and Board member of the
Middle East Studies Association.
He served on the editorial boards of Arabic Sciences and Philos-
ophy: A Historical Journal, Journal of Near Eastern Studies, Ham-
dard Islamicus, and Interpretation: A Journal of Political Philosophy.
He served as president of the American Research Center in Egypt
and held the distinction of having been the first corresponding
member of the Cairo Academy of Arabic Language. In later years, he
spent a great deal of time in Paris, where he lectured at the Institut
du Monde Arabe, participated in seminars, and was a familiar and
beloved figure in the cafés and bookshops frequented by intellectuals
from all over the Muslim world, many of them former students. Less
than a month before his death, he was awarded an honorary
doctorate by the American University in Cairo.
Thoroughly versed in ancient Greek, medieval Jewish and Chris-
tian philosophy, as well as modern Western political philosophy,
Muhsin Mahdi acquired an incomparable command of the Arabic
language in its rich and varied historical and geographical manifes-
tations. He grounded himself in the well-established methods of
critical editions of manuscripts developed by European scholars so
that he could establish the same rigorous standards for research in
Arabic and Islamic philosophy. Early in his career, he searched for
and found long lost manuscripts wherever his travels took him and
then graciously shared them with fellow scholars. He is especially
known for the recovery, edition, translation, and interpretation of
many of Alfarabi’s writings. Indeed, building on Leo Strauss’s early
insights, he showed clearly in his 2001 Alfarabi and the Foundation
of Islamic Political Philosophy how Alfarabi fundamentally altered
the Arabic-Islamic tradition.
A demanding and inspiring teacher, Mahdi emphasized meticulous
analysis and interpretation of philosophical texts in Arabic. He
worked closely with students from the Middle East, North Africa,
IN MEMORIAM 141
Europe, and the United States, several of whom met to honor him at
the time of his 65th birthday and then published a collection of essays
in his honor, The Political Aspects of Islamic Philosophy. With Ralph
Lerner of the University of Chicago and the late Fr. Ernest Fortin
of Boston College, he co-edited the famous Medieval Political
Philosophy: A Sourcebook containing selections in translation from
Arabic, Hebrew, and Latin. He is equally famous for his critical
edition of the 1001 Nights, especially for proving that they consist
only of 282 nights and for his painstaking account of how 18th and
19th century Orientalist scholars falsely expanded the collection. In
April 2005, the Center for Middle Eastern Studies at Harvard
University organized a conference in Professor Mahdi’s honor on
precisely this theme: The Arabian Nights, Eastern and Western
Vantage Points.
Charismatic and charming, with a ready and hearty laugh, Muhsin
Mahdi maintained the stance of a true philosopher in an era marked
by conflict between and among the three Abrahamic faiths. Even
though he sometimes wrote on modern and contemporary political
thinkers, he staunchly resisted political engagement. All the same,
he was terribly a#ected by the destruction of his native land
beginning in 2003.
Professor Mahdi is survived by his wife, Sarah Roche-Mahdi; two
daughters, Fatima and Nadia, from a previous marriage to Cynthia
Risner; and two stepdaughters, Rachel and Rebekah Gerstein. He is
also survived by his first wife, Louise Carus Mahdi.
CHARLES E. BUTTERWORTH
cebworth@gvpt.umd.edu
Select Bbibliography
Ibn Khaldun’s Philosophy of History: A Study in the Philosophic Foundation
of the Science of Culture (London: George Allen and Unwin / New York:
MacMillan, 1957; reprint Chicago: University of Chicago Press, Phoenix
Books, 1964, 1971, and 1974; Persian translation: Felsefeye tārikhe Ibn
Khaldūn, trans. Majı̄d Mas‘ūdı̄, Teheran: BTNK, 1973).
Die Geistigen und Sozialen Wandlungen in Nahen Osten (Freiburg:
Rombach, 1961).
Abū Nas*r al-Fārābı̄, Falsafat Arist*ūt*ālı̄s, Critical edition of Arabic text
(Beirut: Dar Majallat Shi‘r, 1961).
Alfarabi’s Philosophy of Plato and Aristotle, English translation of Arabic
text (New York: MacMillan, 1962; reprint Ithaca, NY: Cornell University
Press, 1969, 1974, and 2002).
Medieval Political Philosophy: A Sourcebook, Co-editor with Ralph Lerner
and Ernest Fortin (New York: MacMillan, 1963; reprint Ithaca, NY:
Cornell University Press, 1967, 1973, and 1984).
142 IN MEMORIAM
Abū Nas*r al-Fārābı̄, Kitāb al-Milla wa-Nus*ūs* Ukhrā, Critical edition of

Arabic text (Beirut: Dar el-Machreq, 1968).
Abū Nas*r al-Fārābı̄, al-Alfāz
* al-Musta‘mala fı̄ al-Mant*iq, Critical edition of
Arabic text (Beirut: Dar el-Machreq, 1968).
Abū Nas*r al-Fārābı̄, Kitāb al-H
* urūf, Critical edition of Arabic text (Beirut:
Dar el-Machreq, 1969).
The Political Orientation of Islamic Philosophy (Washington, DC: George-
town University Press, 1982).
Alf Layla wa Layla, Critical edition of Arabic text (Leiden: E. J. Brill. Parts
1 & 2, Arabic Text, 1984. Part 3, Introduction and Analyses, 1994); reprint
(Leiden: E. J. Brill, Paper-back edition of Arabic text, 1984).
Abū Nas*r al-Fārābı̄, Kitāb al-Wāh
* id wa al-Wah * da, Critical edition of Arabic
text (Casablanca: Toubkal, 1990).
Alfarabi and the Foundation of Islamic Political Philosophy: Essays in
Interpretation (Chicago: University of Chicago Press, 2001).

ASP 2008 Vol.18 Issue.1

Transféré par

Informations du document

Titre original

Copyright

Formats disponibles

Partager ce document

Partager ou intégrer le document

Options de partage

Avez-vous trouvé ce document utile ?

Ce contenu est-il inapproprié ?

Droits d'auteur :

Formats disponibles

ASP 2008 Vol.18 Issue.1

Transféré par

Droits d'auteur :

Formats disponibles

Arabic Sciences and Philosophy, vol. 18 (2008) pp.

MESURER LE CONTINU, DANS LA TRADITION

MAROUANE BEN MILED

Dans son Algèbre, ‘Umar al-Khayyām (1048–1131) résout les équa-

faisant, la frontière qui existe entre les nombres, objets arithmé-

Afin d’étudier les équations du troisième degré qu’il ne sait pas

entendons par ‘nombre’ un parallélépipède rectangle, dont la base est le

On peut rechercher la trace d’une telle unité de mesure chez les

algébriques des problèmes quadratiques du Livre X des Éléments

III. MESURER LES GRANDEURS

Dans le commentaire d’al-Khāzin, les unités de mesure étant posées

espérer conclure que la grandeur G est irrationnelle; mais . . . pas

IV. DU LIEN ENTRE LES NOTIONS DE MESURE ET DE RAPPORT CHEZ

Dès le début de son commentaire au Livre X, al-Khāzin introduit en

mesure. Il su$t pour cela de considérer que le rapport de deux

V. NOUVELLES CONCEPTIONS DU NOMBRE ET DE L’UNITE

Mesurer les grandeurs à partir d’unités de mesure, et concevoir de

Il vient, de l’examen qui précède, que la conception numérique du

nécessite de considérer numériquement les quantités rationnelles et

Il demeure qu’une étude complète et comparative des di#érentes

Abstract: This article proposes a reconstitution of the philosophical tenor of

Résumé: Cet article se propose de reconstituer la teneur philosophique du traité

Al-Fārābı̄’s treatise On Changing Beings (Fı̄ al-Mawǧūdāt al-

motion. But nobody, so far as I know, has attempted to determine in

I. SOME UNCONCLUSIVE ARGUMENTS FOR THE ETERNITY OF THE

Despite the fact that he relied extensively on al-Fārābı̄’s writings,

raised by the treatise On Changing Beings. For even if al-Fārābı̄

1. Argument from the nature of matter

2. Argument from the concept of possibility

3. Argument from the nature of the celestial spheres

what a problem is by stressing the numerous aspects by which it can

demonstration of it, that the demonstrations pertaining to it were counter-

II. THE TREATISE ON CHANGING BEINGS

2. Second group of testimonies: the unreality of time

In the course of his discussion, al-Fārābı̄ gives some examples of

the ‘‘now’’. Furthermore, even if it is not a vicious circle, we have to

3. Some further remarks on the demonstrability of eternity

as unicity, simplicity, etc. that we shall demonstrate some of the

had previously shown that it is impossible that there be an infinite body;

But it is evident that in al-Fārābı̄’s time, there were people talking

4. A physical demonstration of the eternity of the world?

(a) On time’s definition

These lines, despite two textual corruptions, are clear: al-Fārābı̄

confronted with a discussion engaged by Iamblichean neoplatonism

Aristotelian puritanism may well be then a response to such ideas

In conclusion, I would like to return to the passage from

Appendix: the Kitāb al-G

In the Kitāb al-G

commentary on the Topics. My entire interpretation of the Treatise

for Fritz Zimmermann

Résumé: Al-Fārābı̄, dans le Kitāb al-H

I. THE PROGRAM OF THE KITAzB AL-H

of the language, or in the concepts that would be signified in a

can be extended in two main ways to provide the terminology of the

grammatical form of an expression fails to correspond to its logical

reconstruction of the di$culties that the translators must have

18 about ´ + accusative [ = bi-dhāt or li-dhāt], 24 about

II. EXPRESSIONS FOR BEING IN GREEK AND ARABIC

However, the single most important term for Fārābı̄’s project in

meanings of the question hal mawjūdun may be taken either as many

particle.20 For this reason, although ‘‘astı̄n’’ is a perfectly reasonable

III. THE TWO MAIN SENSES OF ‘‘MAWJU

Fārābı̄’s treatment of ‘‘mawjūd’’ in Kitāb al-H

18 about ´ + accusative [ = bi-dhāt or li-dhāt], 24 about

Abū Nasr al-Fārābı̄, Kitāb al-Milla wa-Nusūs* Ukhrā, Critical edition of