Arabic philosophy

curriculum. I describe the contents of the Shamsiyya to give some

idea of the range of topics covered by the average student of logic.

I then note certain elements of al-K¯atib¯ı’s treatment of the modal

syllogistic and examine its predecessors. I conclude by attempting a

broad characterization of later Arabic logic.

preliminary considerations

Whereas the study of medieval Western logic is now an established

field of research, contributing both to modern philosophy of logic

and to the intellectual history of the Middle Ages, the study of logic

in the precolonial Islamic world is still barely in its infancy. That fact

alone makes it difficult to write an introductory chapter on the field:

we are as yet unclear what contributions of the logicians writing in

Arabic are particularly noteworthy or novel. It is also a dangerous

temptation in this state of relative underdevelopment to cast an eye

too readily on the work of the Latin medievalist, and to import the

methods, assumptions, and even the historical template that have

worked so well in the cognate Western field.

This temptation must be resisted at all costs. There are many

important differences between the scholarly ideals and options of

the LatinWest and the MuslimEast; there are, also, many differences

between the various fortunes encountered by rigorous logical activity

in the two realms over the centuries.Aglance at the historiographical

preliminaries of Bochenski’s History of Formal Logic prompts the

following observations.4 First and foremost, Aristotle ceases by the

end of the twelfth century to be a significant coordinate for logicians

writing in Arabic – that place is filled by Avicenna. The centrality of

Avicenna’s idiosyncratic system in post-Avicennian logical writings

and the absence of Aristotelian logic in a narrowly textual sense

meant that Arabic texts dealing with Avicenna’s system were left

to one side by the medieval Latin translators. Instead, other, less

influential texts by Averroes and al-F ¯ar¯ab¯ı were translated, because

they did concentrate on Aristotle and spoke to thirteenth-century

Western logical concerns. Even at the outset, then, the insignificance

of Aristotle’s logical system in the Avicennian tradition worked to

Cambridge Companions Online © Cambridge University Press, 2006

Cambridge Companions Online © Cambridge University Press, 2006

Logic 249

distort Western appreciation of the relative importance of particular

logicians writing in Arabic.

A second difference is that the whole range of Aristotelian logical

texts were available in Arabic by about 900, and so the broad periodization

of medieval Latin logic into logica vetus and logica nova is

inappropriate as a way of periodizing logic written in Arabic;5 by the

time serious logical work began, the complete Organon was available.

Avicenna’s work marks the watershed for any helpful periodization.

Thirdly, Bochenski’s analysis of what preconceptions and historical

meanderings clutter the way to the proper study of medieval

Western logic (the collapse of acute logical study with the demise of

scholasticism, the ahistorical reductivism of post-Kantian logic, the

institutionalization of a psychologistic logic in neoscholasticism)6

do not apply to the study of the logic of medieval Muslim scholars –

even in the early twentieth century, it is clear that at least some

scholars were still in contact with the acute work of the thirteenth

century. There had been far less of a rupture in logical activity over

the intervening centuries. On the other hand, there have been postcolonial

efforts to find later Western logical achievements foreshadowed

in early Arabic logic, and this has damaged the prospects for

appraisal of the work by leading to a disproportionate focus onminor

traditions.7

Finally, only some of the characteristics Bochenski finds which

distinguish medieval Western logic from the logic of late antiquity

apply to the logic being written in Arabic at roughly the same time.8

It too is highly formal and metalogical in its treatment, and pedagogically

central; but no doctrine like supposition was developed,

and there seems to have been far less concern with antinomies. One

may say – nervously, given the current state of research – that Arabic

logic is somewhat closer to the logic of late antiquity in its concerns

and methods than medieval Latin logic. That said, one must guard

against an obvious alternative assumption, which is that Arabic logic

is by and large just one or other of the systems of late antiquity.9 We

already have enough control over Avicenna’s logic to know that is

false.

the tradition of logic and the madrasa

The average learned Muslim from the late thirteenth century on

acquired some logic as part of his intellectual arsenal. Very often,

Cambridge Companions Online © Cambridge University Press, 2006

Cambridge Companions Online © Cambridge University Press, 2006

250 tony street

that logic had been acquired from the lovely little textbook, the

Shamsiyya, perhaps the most studied logic textbook of all time.

The Shamsiyya belonged to and reflected a tradition which had its

own peculiar features. Further, the Shamsiyya had found a home in

the madrasa curriculum in the face of opposition from some quarters.

I examine aspects of each of these matters in the following

subsections.

The form and substance of the tradition

As a religious community, Islam had first come in contact with a

living logical tradition when the Muslim armies coming out of the

peninsula in the seventh century took control of the Fertile Crescent

where Christian communities had made logic an important part of

their studies. The logic they studied was a shortened version of what

had been taught in the Alexandrian curricula of the sixth century,

limited to the Categories, On Interpretation, and the assertoric syllogistic

of the Prior Analytics (which is to say, up to the end of the

seventh chapter of the first book).10 There are various reasons given

for why the Christians stopped at that point (al-F ¯ar¯ab¯ı was later to

write that it was due to a synodal decree, but there are good reasons

to doubt that was the reason),11 but probably it was because that

was the simplified basic logic they received. It was that basic logic

which was translated from Syriac into Arabic in the second half of

the eighth century.

The first Arabic translations of logic were executed, then, at the

beginning of the ‘Abb¯asid regime. No universally valid generalization

can be made about how the translations proceeded, but by and

large logical summaries were translated, then texts or fragments of

texts from the Organon, then commentaries on those texts of the

Organon; again, by and large, translations went from Greek to Arabic

by way of the Syriac. Even the earliest translations are, however,

an exception to that process; the earliest surviving text translated,

by Ibn al-Muqaffa‘ (d. 757) in the 750s, was indeed a summary, but

we know that it was made at roughly the same time as a translation

of the Topics.12 The early efforts gave way to better translations

by better-organized translators, and by the mid-ninth century

there were highly skilled translators working in the circle of al-Kind¯ı

(d. ca. 870) and, somewhat later, in the circle of H.

unayn ibn Ish. ¯aq

Cambridge Companions Online © Cambridge University Press, 2006

Cambridge Companions Online © Cambridge University Press, 2006

Logic 251

(d. 873). Though some of the logic texts translated were still descendants

of the Alexandrian summaries, more and more texts of the

Organon were translated and retranslated. The Organon texts were

ultimately furnishedwith commentaries by scholars of late antiquity

like Alexander, Ammonius, and Themistius.H.

unayn had particular

veneration for Galen, and many Galenic logical works now lost were

available to the Arabs.

It was not until the rise of the Baghdad Peripatetics, marked by

the activities of Abu¯ Bishr Matta¯ ibn Yu¯ nus (d. 940) and al-Fa¯ ra¯bı¯ (d.

950), that logicians started to focus closely on the Organon itself. Al-

F¯ar¯ab¯ı came to conceive his task as clearing awaymisinterpretations

of the Aristotelian text, many the result of the Syriac summaries, and

reviving true Peripatetic doctrine after a period of rupture. The Baghdad

Peripatetics who carried on the work of Abu¯ Bishr and al-Fa¯ ra¯bı¯

were able by about 1000 to produce a heavily annotated version of

the Organon which was accurate enough for close exegesis.13 The

Farabian tradition, both in Baghdad until the mid-twelfth century

and in Spain where it ended somewhat after the time of Averroes

(d. 1198), would always be concerned with this kind of exegesis.

At the same time the final version of the Arabic Organon was

being achieved, a young philosopher from Khur¯as¯an, Avicenna (d.

1037), had set about changing forever the course of Islamic philosophy.

Avicenna claimed for himself, by virtue of his Intuition (h.

ads),

the ability to judge the Peripatetic tradition, and to be in a position

to say what philosophical doctrines Aristotle should properly have

come to.14 One aspect of his philosophical activity particularly relevant

for present purposes led him to reinterpret sections of the Prior

Analytics, and it is his reinterpretation which served as the object

of study and debate among later scholars in his tradition rather than

the original Aristotelian system.

The leaders of this Heroic Age of Arabic logic, al-F ¯ar¯ab¯ı and

Avicenna, dealt with the books of the Organon one by one. But

after them, a major change in focus occurred in the tradition, here

described by Ibn Khaldu¯ n (d. 1406), the first (and still greatest) historian

of Arabic logic.

The later scholars came and changed the technical terms of logic; and they

appended to the investigation of the five universals its fruit, which is to

say the discussion of definitions and descriptions which they moved from

the Posterior Analytics; and they dropped the Categories because a logician

Cambridge Companions Online © Cambridge University Press, 2006

Cambridge Companions Online © Cambridge University Press, 2006

252 tony street

is only accidentally and not essentially interested in that book; and they

appended to On Interpretation the treatment of conversion (even if it had

been in the Topics in the texts of the ancients, it is nonetheless in some

respects among the things which follow on from the treatment of propositions).

Moreover, they treated the syllogistic with respect to its productivity

generally, not with respect to its matter. They dropped the investigation of

[the syllogistic]with respect to matter, which is to say, these five books: Posterior

Analytics, Topics, Rhetoric, Poetics, and Sophistical Fallacies (though

sometimes some of them give a brief outline of them). They have ignored

[these five books] as though they had never been, even though they are important

and relied upon in the discipline. Moreover, that part of [the discipline]

they have set down they have treated in a penetrating way; they look into it

in so far as it is a discipline in its own right, not in so far as it is an instrument

for the sciences. Treatment of [the subject as newly conceived] has

become lengthy and wide-ranging – the first to do that was Fakhr al-D¯ın al-

R¯az¯ı (d. 1210) . . . The books and ways of the ancients have been abandoned,

as though they had never been.15

By the time al-K¯atib¯ı wrote the Shamsiyya, at the height of the

Golden Age of Arabic logic, the discipline had changed from ranging

over all the subjects covered in the Organon to concentrating on narrowly

formal questions. And, as did all at his time, al-K¯atib¯ı wrote

on those questions as developed by Avicenna.

The study of logic

This is how the Shamsiyya acquired the form and focus it has. How

did it come to be studied in the madrasa – how, in short, did Islam

embrace a Greek-derived science at its educational heart? At the time

of the translation movement, some Muslim religious scholars were

open to the study of logic, some less so. One of the most commonly

cited examples of antipathy to logic in Muslim intellectual circles is

that expressed by a grammarian, Abu¯ Sa‘ı¯d al-Sı¯ra¯ fı¯, in a debate that

took place in the 930swith one of Baghdad’s leading Peripatetics, the

Abu¯ Bishr Matta¯ mentioned above. The debate touched on a number

of complex issues, but most significantly for present purposes

it serves to show how doubtful scholars were about claims for the

utility of Greek logic. Al-Sı¯ra¯ fı¯ taunted Abu¯ Bishr about his trust in

logic: “The world remains after Aristotle’s logic as it was before his

logic . . . you can dispense with the ideas of the Greeks as well as

you can dispense with the language of the Greeks.”16

Cambridge Companions Online © Cambridge University Press, 2006

Cambridge Companions Online © Cambridge University Press, 2006

Logic 253

Al-Fa¯ ra¯bı¯, spurred by Abu¯ Bishr’s humiliation, set about showing

how logic complemented the Islamic sciences. This he did by showing

that logic underpinned and guaranteed the arguments deployed

in theology and law. In his book The Short Treatise on Reasoning

in the Way of the Theologians, “he interpreted the arguments of the

theologians and the analogies of the jurists as logical syllogisms in

accordance with the doctrines of the ancients.”17 In The Short Treatise,

we find analyses of the paradigmatic argument, of the argument

used by Muslim theologians called “reasoning from the seen to the

unseen,” and of the “juristic argument” itself.18 This programmatic

defence of logic was adopted by an important Muslim jurist, Abu¯

H.

¯amid al-Ghaz¯ al¯ı (d. 1111), who prefaced his most famous juristic

digest with a short treatise on logic, saying that knowledge of logic

was indispensable for a proper control of jurisprudence.19 Logic was

thereafter widely recognized as a necessary part of a scholar’s training.

The way that recognition had been won led subsequent writers

of logic manuals to consecrate at least a part of their manuals to the

reduction of juristic argument forms to the syllogism, a reflex carried

over from this early time when Muslim scholars contested the

place of logic in Islamic society.

The Farabian strategy succeeded in making logic a commonly

studied discipline, but dissenting voices could still be heard. A

famous condemnation of logic, issued by Ibn al-S. al ¯ah (d. 1245), lets

us see how pious doubts arose about the discipline.

As far as logic is concerned, it is a means of access to philosophy. Now

the access to something bad is also bad. Preoccupation with the study and

teaching of logic has not been permitted by the Lawgiver. The use of the

terminology of logic in the investigation of religious law is despicable and

one of the recently introduced follies. Thank God, the laws of religion are not

in need of logic. Everything a logician says about definition and apodictic

proof is complete nonsense. God has made it dispensable for those who

have common sense, and it is even more dispensable for the specialists in

the speculative branches of jurisprudence.20

But as a formal system, logic is unobjectionable, and an influential

jurist and famous hater of logic, Ibn Taymiyya (d. 1328), was prepared

to concede that point, albeit rather mockingly:

The validity of the formof the syllogismis irrefutable . . . But it must bemaintained

that the numerous figures they have elaborated and the conditions

Cambridge Companions Online © Cambridge University Press, 2006

Cambridge Companions Online © Cambridge University Press, 2006

254 tony street

they have stipulated for their validity are useless, tedious, and prolix. These

resemble the flesh of a camel found on the summit of a mountain; the mountain

is not easy to climb, nor the flesh plump enough to make it worth the

hauling.21

contents of the shamsiyya

There is much in the structure of the Shamsiyya that is familiar to

anyone who has looked at medieval Latin treatises on logic, particularly

the division of the three main parts of the treatise (parts 2, 3,

and 4) into Terms, Propositions, and Syllogism.22 There are other

deep similarities, especially the way logic is defined, and the way

the treatment of universals is developed. Much of the material would

work perfectly well as an introduction to logic in the medievalWestern

tradition; it is, like its Latin counterparts, ultimately descended

from the Aristotelian texts.

1. Introduction

(a) What logic is

(b) Subject matter of logic

2. Terms

(a) Utterances

(b) Individual concepts

(c) On universals and particulars

(d) Definition

3. Propositions

(a) Introduction: definition of proposition and its parts

(b) On the categorical proposition

i. Parts thereof, and kinds

ii. On quantification

iii. On equipollence and existence

iv. On modal propositions

(c) On the hypothetical proposition

(d) On rules governing propositions

i. Contradiction

ii. Simple conversion

iii. Obversion

iv. Equivalences among hypothetical propositions

4. Syllogisms

(a) Definition of “syllogism,” and its kinds

(b) Syllogisms with mixed premises

(c) On conjunctive syllogisms with hypothetical propositions

Cambridge Companions Online © Cambridge University Press, 2006

Cambridge Companions Online © Cambridge University Press, 2006

Logic 255

(d) On repetitive syllogisms

(e) Further matters to do with syllogistic

i. Linked syllogisms

ii. Reduction

iii. Induction

iv. Example

5. Conclusion

(a) Syllogistic matter

(b) On the parts of the sciences

Still, some points in the Shamsiyya would not serve to introduce

material treated by its Latin contemporaries, most especially the

division of the syllogistic, the division of different kinds of discourse,

and the modal logic.

The division of the syllogistic in nearly all works belonging to

the dominant Avicennian tradition reproduces a division set out by

Avicenna and claimed by him to be one of his own innovations.

According to what we ourselves have verified, syllogistic divides into two,

conjunctive (iqtira¯nı¯) and repetitive (istithana¯ ’ı¯). The conjunctive is that in

which there occurs no explicit statement [in the premises] of the contradictory

or affirmation of the proposition in which we have the conclusion;

rather, the conclusion is only there in potentiality, as in the example we

have given. As for the repetitive, it is that in which [the conclusion or its

contradictory] occurs explicitly [in the premises].23

The division probably relates to epistemic questions to do with perfecting

syllogisms, and ramifies through to the analysis of the proof

by reduction to an impossibility, both topics too complex to be

explored here.Averroes criticized it as a division, and he and al-F ¯ar¯ab¯ı

used an alternative division common in the Latin tradition.24

Another strange aspect of the Shamsiyya relative to its Latin contemporaries

is its section devoted to syllogistic matter (al-ma¯dda

al-qiya¯ siyya). This was referred to above in the quote taken from

Ibn Khaldu¯ n, who said that the later logicians “dropped the investigation

of [the syllogistic] with respect to matter, which is to say,

these five books: Posterior Analytics, Topics, Rhetoric, Poetics, and

Sophistical Fallacies (though sometimes some of them give a brief

outline of them)” – the brief outline became a way to deal with the

later parts of the Alexandrian arrangement of the Organon. The syllogism

was taken to be what formally underlies all arguments, but

Cambridge Companions Online © Cambridge University Press, 2006

Cambridge Companions Online © Cambridge University Press, 2006

256 tony street

propositions making up those arguments may differ; these differentiated

stretches of discourse manipulated by the syllogism in turn

determine what kind of argument is produced, be it demonstrative,

dialectical, rhetorical, poetical, or sophistical. There was dispute

about the criteria to differentiate syllogistic matter, but in the

Avicennian tradition to which the Shamsiyya belongs, the criteria

were epistemological.25

the modal syllogistic

With that we come to the modal syllogistic, perhaps the most foreign

of the Shamsiyya’s doctrines. It is not obvious from the table of

contents given above just how much of the Shamsiyya is given over

to the modals; one needs to realize that the sections on contradiction,

conversion, obversion, and syllogismswithmixed premises are all on

modal logic, amounting to over eight pages of the twenty-eight and

a half pages of Sprenger’s printing. In other words, nearly one-third

of an introductory treatise is given to modal logic. Throughout its

treatment of the modals, a distinction between the dha¯ tı¯ and was. fı¯

readings of the proposition crops up again and again. Whence comes

this distinction, unknown in the West?

To answer this question, we need to return to the Aristotelian system.

Let us consider the following famous problem. Aristotle wants

this syllogism to be valid:

(1) Every c is b, every b is necessarily a, therefore every c is

necessarily a.

but he wants to reject the next syllogism:

(2) Every c is necessarily b, every b is a, therefore every c is

necessarily a.

The “necessarily” in the first, valid, syllogism, seems to belong to