AVERROES’ “EPISTLE ON DIVINE KNOWLEDGE” AS A DIALECTICAL WORK: BETWEEN FORBIDDEN INTERPRETATION AND PHILOSOPHICAL TRAINING

Abstract Averroes’ “Epistle on Divine Knowledge” presents four different dialogues on two textual levels. These dialogues, the syllogistic structure of the arguments in them, and their use of contradictories indicate that the “Epistle on Divine Knowledge” is structured nearly entirely in accordance with the descriptions of dialectic we find in Averroes’ commentaries on Aristotle's Topica. Accordingly, Averroes’ solution to the question of how God can have universal knowledge of particular things is a dialectical account of the distinction between Divine and human knowledge. Moreover, at a crucial point in the “Epistle on Divine Knowledge” Averroes refers to Aristotle, Metaphysics Β, which he considers to a dialectical exposition of questions on metaphysics. This reference suggests that Averroes sees the “Epistle on Divine Knowledge” as a kind of dialectical inquiry aimed at answering questions that arise at the outset of studying metaphysics. So, while it is possible to view the “Epistle on Divine Knowledge” as a dialectical interpretation of Quran 67:14, its primary purpose is to introduce its readers to metaphysical speculation. Thus it does not violate Averroes’ legal prohibition given in the Decisive Treatise against declaring dialectical interpretations in books available to the general public.


INTRODUCTION
In his Decisive Treatise, Averroes decrees that interpretations (al-taʾwīlāt), of the Law, "ought not to be declared to the multitude (al-ǧamhūr) nor established in rhetorical or dialectical books." 1 Shortly thereafter, Averroes goes so far as to associate declaring "interpretations to those not adept in them" with heresy (al-kufr) on the grounds that it leads to damnation (halāk) in this world and the next. 2 Indeed, so against public dialectic is Averroes that the ideal state Averroes describes in his Commentary on Plato's Republic is one without public dialectic or dialecticians. 3Still, Averroes himself employs dialectical methods not only in scientific works intended for an audience adept in such argumentation, but also in more general works such as Tahāfut al-tahāfut and even in the Decisive Treatise itself.Dialectic is also present throughout Averroes' "Epistle on Divine Knowledge," 4 and 1 Averroes, Decisive Treatise and Epistle Dedicatory, trans.Charles Butterworth (Provo, Utah: Brigham Young University Press, 2001), p. 26, para.45.On the legal form of the Decisive Treatise as a fatwā, see Daniel Heller-Roazen "Philosophy before the Law: Averroes's Decisive Treatise," Critical Inquiry, vol.32 (2006), p. 412-442. 2 Averroes, Decisive Treatise, p. 27, para.47. 3 Yehuda Halper, "Expelling Dialectics from the Ideal State: Making the World Safe for Philosophy in Averroes' Commentary on Plato's Republic," in Alexander Orwin (ed.), Plato's Republic in the Islamic Context: New Perspectives on Averroes' Commentary (University of Rochester Press, 2022), p. 69-86. 4The first editor of this work, Marcus Joseph Müller, gave this treatise the Arabic title, Ḍamīma, "Appendix" in Philosophie und Theologie von Averroes (Munich, 1859).The scribe of the manuscript used by Müller referred to the text as "The question which Abū al-Walīd (may God be pleased with him) mentioned in the Decisive Treatise" (preserved as a subtitle in Müller's edition, p. 128: ‫في‬ ‫الوليد‬ ‫ٔبو‬ ‫ا‬ ‫ذكرها‬ ‫التي‬ ‫المسئلة‬ ‫عنه‬ ‫الله‬ ‫رضى‬ ‫المقال‬ ‫.)فصل‬ Muhsin Mahdi notes that this "is not a formal title and does not form part of the work as written or dictated by Averroes; it is a scribe's explanation."Moreover, Mahdi notes that this work is specifically addressed to "one of his companions," and argues that this companion is in fact the Almohad Caliph Abū Yaʿqūb Yūsuf.Accordingly, he says, that this treatise "was not meant to have a title: it is an epistle dedicatory."See Muhsin Mahdi, "Averroes on Divine Law and Human Wisdom," in Joseph Cropsey (ed.), Ancients and Moderns: Essays on the Tradition of Political Philosophy in Honor of Leo Strauss, (New York: Basic Books, 1984), 114-131, esp.117-118.Charles Butterworth takes up Mahdi's suggestion and gives the work the title "Epistle Dedicatory" in his edition and translation, Averroes, Decisive Treatise and Epistle Dedicatory.The two Hebrew translations of this work, one by Ṭodros Ṭodrosi of Arles and the other anonymous, give their own titles to the work.Ṭodros' title appears as "Treatise on Eternal Knowledge" ‫הקדמון(‬ ‫במדע‬ ‫מאמר‬ in two manuscripts, and ‫בקדום‬ ‫מדעת‬ … ‫מאמר‬ in another manuscript).The anonymous translation, which survives in only one manuscript gives as a title "Epistle on the indeed, Averroes ends the short work by quoting Quran 67:14 ( ‫َق‬ ‫ل‬ ‫,)خَ‬ which can be translated "Does he (God) not know, he who created, since he is perspicacious and informed?" 5 In either case, Averroes suggests that the entire "Epistle on Divine Knowledge" and its question about God's knowledge of generated things is in a sense an interpretation of this verse.The "Epistle on Divine Knowledge," then, is a dialectical interpretation of the Quran and as such would seem to be explicitly prohibited from being written down and presented to the multitude according to Averroes' Decisive Treatise.Does Averroes' "Epistle on Divine Knowledge" go against the legal ruling Averroes laid down in the Decisive Treatise?
It is, of course, possible to answer this question using Averroes' own justification for discussing the connection between wisdom and Law and interpretation, viz.that such issues and questions have gained a status of being widely held among people (šuhra … ʿinda al-nās). 6This, indeed, would explain Averroes' use of dialectical arguments in the Decisive Treatise, and perhaps in the Exposition and Incoherence as well.Yet, while these works use some dialectical arguments, they are not thoroughly dialectical in the way of the "Epistle on Divine Knowledge," which we shall see is structured nearly entirely according to the descriptions of dialectic we find in Averroes' commentaries on Aristotle's Topica.That is, the "Epistle on Divine Knowledge" is fundamentally dialectical in a way we do not see in Averroes' other writings and so we may ask why he wrote in this way here and why, moreover, writing such a thoroughly dialectical work is permitted?
Before answering this question, we shall examine the dialectical character of the "Epistle on Divine Knowledge" in light of Averroes' own descriptions of dialectic in his Short and Middle Commentaries on Aristo-Meaning of the Doubt attendant on the Eternal's Knowledge (May He be Exalted)" ‫ית(‬ ‫הקדמון‬ ‫בידיעת‬ ‫הקורה‬ ‫הספק‬ ‫בענין‬ … ‫.)אגרת‬ See Silvia Di Donato, "La tradizione ebraica dell'opuscolo di Averroè sulla scienza divina," in Irene Kajon, Luise Valente, and Francesca Gorgoni (ed.), Philosophical Translations in Late Antiquity and in the Middle Ages.In Memory of Mauro Zonta (Rome: Aracne, 2022), p. 161 and 164, and the discussion on p. 149-150.Both Hebrew translations use the term "Eternal" to mean Divine.In a course I attended on Averroes' Decisive Treatise at the University of Chicago taught by Joel Kraemer and Ralph Lerner in 2003, Prof. Kraemer suggested using the title, "Treatise on Divine Knowledge," relying, as I recall, on Ṭodros' title.Here I have adopted the title, "Epistle on Divine Knowledge," in an attempt to combine these approaches.tle's Topica.Even though the Middle Commentary was probably written after the "Epistle on Divine Knowledge" it is Averroes' most detailed work on dialectic and probably presents his views best, even if they were in less developed form at the time he wrote the "Epistle on Divine Knowledge."Then we shall follow Averroes' comparison of solving the difficulty of the "Epistle on Divine Knowledge" to untying a knot to its source in Aristotle's Metaphysics Β and examine what Averroes has to say about dialectic in his Middle Commentary there.This will allow us to suggest an explanation of the role of dialectic in Averroes' "Epistle on Divine Knowledge" that is consistent with Averroes' philosophical project.

DIALECTIC IN THE "EPISTLE ON DIVINE KNOWLEDGE"
One cannot escape the dialogical structure of the "Epistle on Divine Knowledge," which presents four different dialogues on two textual levels.First, there is an apparent frame dialogue between the author, viz.Averroes, and an unnamed interlocutor, whom some have supposed to be the Caliph Abū Yaʿqūb Yūsuf. 7Nested within that frame, are three other short dialogues.One between a first person plural "us" and someone known as "the adversary" (al-ḫaṣm), another between the first-person plural "us" and the mutakallimūn (para.5), 8 and the third betweem "us" and al-Ġazālī (para.6-7).Each of these dialogues is between two people and each includes a questioner and a respondent.So, even though the term al-ḫaṣm is more frequently used in the context of rhetoric, it is clear that there is no audience here and the adversarial contexts of dialogues 2 and 3 are dialectical, rather than rhetorical. 9Moreover, all four 7 This is suggested by Muhsin Mahdi in "Averroes on Divine Law and Human Wisdom," p. 118-119.This suggestion is repeated by Charles Butterworth in Decisive Treatise, p. xl-xli. 8In fact, para.5 is careful to use the passive voice and the sense that one of the interlocutors is "us" is supplied from context, including from the fact that the "us" ‫عندنا‬ shows up again at the opening of paragraph 7, despite the use of the passive in para.6. 9 Glossarium graeco-arabicum lists ḫaṣm as a frequent translation of ἀντίδικος in the rhetoric and the verbal form, ḫaṣama as translating ἀμφισβητέω in the rhetoric.Note that both Hebrew translations of the "Epistle on Divine Knowledge" render ḫaṣm by baʿal rib (though Ṭodros Ṭodrosi adds the definite article).See Silvia Di Donato, "La tradizione ebraica dell'opuscolo di Averroè sulla scienza divina," p. 141-169.While the Hebrew term, baʿal rib, is often used in the context of rhetoric, it also appears in Samuel ben Judah of Marseilles' Hebrew translation of Averroes' Short Commentary on Aristotle's Topica.See Averroes, Short Commentary on Aristotle's Logical Organon: Topica.Trans.Jacob ben Makhir Ibn Tibbon, revised by Samuel ben Judah of Marseilles, ed.Yehuda Halper (Mahadurot: Modular Hebrew Digitally Ren-dialogues concern a single "doubt," šakk, about God's eternal knowledge of created, i. e., generating things.
A dialectical problem is an inquiry that leads either to choice and avoidance or to truth and cognizance … About [this problem] either people's opinions go any way, or the opinions of the many are opposite those of the wise, or the opinions of the wise are opposite those of the many, or each (sc.wise and many) go opposite with themselves.
Al-Damašqī apparently translates Aristotle's "dialectical problem" (πρόβλημα διαλεκτικὸν) as al-masʾala al-manṭiqiyya, "logical questioning," though ʿAbd al-Raḥman Badawī points to a note above the line that reads al-maḥāwiriyya al-ǧadiliyya, meaning something like "dialectical pivots." 10 Neither reading is clearly relevant for our purposes.Yet, in his Middle Commentary, Averroes restates what is apparently the same passage as follows.

‫ما‬ ‫شك‬ ‫يلحقه‬ ‫بل‬ ‫لمشهور‬ ‫بحسب‬ ‫بنفسه‬ ‫صدقه‬ ‫معلوما‬ ‫يكن‬ ‫لم‬ ‫ما‬ ‫فهو‬ ‫الجدلى‬ ‫المطلوب‬ ‫المشهور.‬ ‫في‬
The object of dialectical inquiry is that whose truth is not known in itself according to what is widely-held, but that which is attended by doubt with respect to what is widely-held. 11erroes goes on to give examples related to choice, such as whether or under what circumstances wealth or poverty is to be preferred, and examples related to truth and knowledge, such as whether the world is eternal or created, both favorite examples of Aristotle's.Averroes also make special mention of doubts that occur to believers regarding what is widely-held in their religions.Averroes, then, takes "doubt," šakk, to be central to dialectic, even though it does not play so clearly prominent a role in Aristotle's text, even in Arabic translation."Doubt" is clearly a framing subject of the "Epistle on Divine Knowledge."Indeed the word šakk appears 13 times in the short, eleven paragraph text.Moreover, the inquiry of the "Epistle on Divine Knowledge" is centered around this doubt.Indeed, the text seems to be divided fairly evenly into two parts: (1) The determination of the doubt (taqrīr hāḏa al-šakk, paragraphs 1-5) and (2) The solution to the doubt (ḥall hāḏa al-šakk , paragraphs 6-11).Even if we follow Charles Butterworth's division of the text into three parts (in addition to what he sets as introductory and concluding paragraphs), 12 we take paragraphs 8-10 as "Consequences" to the solution of the doubt in paragraphs 6-7.That is to say, the treatise is clearly an inquiry into "that which is attended by doubt." The doubt in question, whether and how God knows created things, is moreover one about something "whose truth is not known in itself according to what is widely-held."What makes something "widely-held" (mašhūr)?This concept is loosely connected to Aristotle's notion of ἔνδοξα as developed in the Nicomachean Ethics and Topica.In the Topica Aristotle connects it to what all or some people believe, especially the wise. 13Averroes follows Aristotle in this in both his Middle Commentary (para.21) and his Short Commentary (para.13), while giving a more systematic breakdown into the kinds of wise people (scientists, experienced doctors, etc.) who might hold different opinions.This is significant because, as Averroes notes, if all believe something, there is no doubt and so, no need for dialectical methods.In the case of the doubt about God's knowledge, we know that some who might be considered wise, viz. the mutakallimūn, have opinions about it which are patently wrong. 14Accordingly, the doubt about God's knowledge of created things is not known in itself according to what is widely-held.
Moreover, according to Averroes in both the Short and Middle Commentaries on the Topica, the contradictory or opposite of something well-known is also well-known. 15That is to say, if the view of the mutakallimūn is well-known, then so is its contradictory.This is a further indication that this doubt is "according to what is widely-held."Accordingly, it is clear that the discussion of the "Epistle on Divine Knowledge" in general is a dialectical inquiry, as Averroes understands 12 That is, the division he employs in Averroes, Decisive Treatise and Epistle Dedicatory. 13Aristotle, Topica 100b21-23.Cf.Ethica Nicomachea 1145b5. 14"Epistle on Divine Knowledge," para. 5. 15 Short Commentary para.13 and Middle Commentary para.21.

it.
Averroes' dialectical approach can be felt further in the way he structures opposing arguments in the determination of the doubt, in a form that is readily translatable to syllogisms.In the Prior Analytics and elsewhere Aristotle generally introduces syllogistic premises with the Greek εἰ, meaning "if," and signals the conclusion with the particle ἄρα.These terms come into Arabic as ʾin and lazima ʾanna respectively.These terms appear with some frequency in Averroes' determination of the doubt, suggesting that he is putting the arguments in syllogistic form.This is further tied in with the art of dialectic, as Averroes says at the opening of his Middle Commentary on the Topics:

‫مقدمات‬ ‫من‬ ‫نعمل‬ ‫ٔن‬ ‫ا‬ ‫سائلين‬ ‫كنا‬ ‫ٕذا‬ ‫ا‬ ‫بها‬ ‫نقدر‬ ‫التي‬ ‫الصناعة‬ ‫بالجملة‬ ‫هي‬ ‫الصناعة‬ ‫هذه‬ ‫يروم‬ ‫كلى‬ ‫وضع‬ ‫كل‬ ‫حفظ‬ ‫وعلى‬ ‫حفظه,‬ ‫المجيب‬ ‫يتضمن‬ ‫وضع‬ ‫كل‬ ‫ٕبطال‬ ‫ا‬ ‫على‬ ‫قياسا‬ ‫مشهورة‬ ‫مجيبين.‬ ‫كنا‬ ‫ٕذا‬ ‫ا‬ ‫ٕبطاله‬ ‫ا‬ ‫السائل‬
This art is in general the art through which we are able, when we are questioners, to construct a syllogism out of well-known premises in order to refute any thesis which the respondent has committed himself to defendor to defend any universal thesis which a questioner strives to refute, when we are respondents. 16at is, according to Averroes -and here is following al-Fārābī's reading of Aristotle's opening line of Aristotle's Topics 17 -the dialectician should be able to argue both sides of a (universal) thesis using syllogisms built out of well-known premises.This is, in fact, what we find in paragraph 3 of the "Epistle on Divine Knowledge."Averroes presents two contradictory theses followed by arguments in the form of a syllogism.The theses are: (T) Created things in God's knowledge are the same before they exist as they are after they exist.
(¬ T) Created things in God's knowledge are not the same before they exist as they are after they exist.
This formulation sounds somewhat awkward because it takes as its subject the created things as objects of God's knowledge.Averroes then 16 Middle Commentary, para. 1. 17  takes the negative thesis (¬ T) as a premise (beginning with ʾin) and draws the conclusion (lazima ʾanna) that eternal knowledge changes (mutaġayyiran) in response to creation.This argument assumes that a change in the object of God's knowledge is a change in God's eternal knowledge itself.This assumption is not controversial and so could be accepted as a universal well-known premise according to Averroes' conditions for dialectic.Thus we can restate this argument as a syllogism: The objects of God's knowledge are subject to change (i.e., ¬ T) The objects of God's knowledge are part of eternal knowledge (Some) eternal knowledge is subject to change The conclusion that (any) eternal knowledge is subject to change is, according to Averroes "absurd" (mustaḥīl). 18Accordingly, this syllogism is brought by Averroes to refute thesis ¬ T.
When examining the contradictory thesis, T, viz.created things in God's knowledge are the same before they exist as they are after they exist, Averroes constructs a literary dialogue with an unnamed adversary to interrogate the question of whether created things are in themselves the same after they are created or different.The adversary admits (salima) that they are not the same, and thereby is led to admit that the knowledge of created things changes when those things are created.This admission is equivalent to ¬ T and the adversary has thus been led into accepting both T and ¬ T, i. e., into a contradiction.
From this Averroes concludes, "One of two things is obligatory; either eternal knowledge differs in itself, or generated things are not known to it" (para.3).These two are not proper contradictories.Yet they do follow from another set of contradictories: (S) God knows created things.(¬ S) God does not know created things.
If S, then we are faced with T or ¬ T, which are either absurd or selfcontradictory according to Averroes.Yet ¬ S is also "absurd" (mustaḥīl), according to Averroes, though he does not say why -and indeed is famously blamed for holding precisely this position. 19In any case, Aver-18 Glossarium graeco-arabicum lists this word as a possible translation of ἄτοπος.Note that the parenthetical additions of "some" or "any" here are not in Averroes' text, but are not inconsistent with his argument. 19This controversy may be based somewhat on Averroes' statement in his Short Commentary on Aristotle's Parva naturalia that separate intellect know only universals, not particulars.See Averroes, Compendia librorum Aristotelis qui Parva naturalia vocantur, ed.H. Blumberg (Cambridge, Mass.: The Medieval Academy of America, roes is clearly pursuing a dialectical approach, via thesis and its contradictory and arguing each on the basis of syllogism.
This approach continues in the dialogues Averroes creates with the mutakallimūn and with al-Ġazālī in paragraphs 5 and 6 respectively.The mutakallimūn hold thesis T, but deny that God's knowledge changes when the things change.Averroes points out to the imagined interlocutors that this is not consistent with what knowledge of something that changes is.Note that he does this, too, by setting up contradictory theses, viz.
(V) When things come into existence, a change occurs, viz.coming from nothing into existence.
(¬ V) When things come into existence, a change does not occur.
Those who hold ¬ V, he says, "are being contentious" (kabirū), 20 while V must imply that God knows the change in that which comes into existence, thereby raising the questions of T and ¬ T, what Averroes calls "the previous doubt."Again, we see a thesis and its contradictory with arguments to rule out the possibilities, i. e., dialectical argumentation.
The solution of this doubt begins with another mini-dialogue.This time with al-Ġazālī.This dialogue does not identify an answerer or a respondent.Moreover, Averroes focuses on al-Ġazālī's meaning, maʿnāhu, rather than on his actual statement (qawl).According to Averroes, al-Ġazālī claimed (zaʿama) that knowledge and what is known are related (anna al-ʿilm wa-al-maʿalūm min al-muḍāfa)."Just as one of two related things may change and the other related thing not change in itself, so it would seem to occur in the case of God's knowledge, may he be glorified).That is, they change in themselves but his knowledge … does not change." 21Averroes refutes this view by appeal to the proper understanding of the category of relation.Averroes notes that the subject (mawḍūʿ) of the relation need not change along with a change in the object of the relation, but the relation (al-ʾiḍāfa) itself does actually 1972), p. 74-75.Averroes says in Metaphysics Λ that divine providence is applied to the species only, not to individuals.See Averroes, Tafsīr mā baʿd aṭ-ṭabīʿat, ed.Maurice Bouyges (Beirut, Imprimerie Catholique, 1938-42), vol.3, p. 1607 (C.38.r), cf.p. 1707-1708 (C.51.ii).See also Richard Taylor, "Averroes' Epistemology and its Critique by Aquinas," in R. E. Houser (ed.), Medieval Masters: Essays in Memory of Msgr.E. A. Synan (Houston, Tex., 1999), p. 147-177. 20Aristotle also frequently dismisses certain arguers as "contentious" (ἐριστικός)see, e. g., De sophisticis elenchis 172a8-9, though it is not entirely clear that this is equivalent to the Arabic here.Still, Glossarium graeco-arabicum lists al-mukābara as a translation of ἐριστικόν in Themustius' Commentary on Aristotle's De anima (https://glossga.bbaw.de/glossary.php@id=189716.html). 21Butterworth trans.modified.change.Averroes' example is a column on Zayd's right at one point that is on his left at another, without Zayd moving.Zayd has not changed, even though the relation "to the right of Zayd" has changed to "to the left of Zayd." Averroes does not here deny that knowledge is a relation. 22Indeed, when Averroes discusses relation (al-ʾiḍāfa) in some detail in the context of what he identifies as topos 27 in his Middle Commentary on the Topics (para.165), he frequently uses knowledge (al-ʿilm) as an example.In this he follows Aristotle who also employed ἐπιστήμη as an example in the parallel passage in his Topica, at 125a.Averroes identifies a kind of relation that is determined by prepositions, such as li, and notes that sometimes things related in this way can convert such that when A is related to B, B is also related to A. An example of this, says Averroes in a section preserved only in the 14th c.Hebrew translation of Qalonimos ben Qalonimos, is knowledge and what is known. 23Moreover, notes Averroes there in a section preserved in the Arabic, knowledge is an example of something that can be said by a syllogism 24 of that which is known and of the soul that knows.Knowledge exists in the soul and in the things which are known and which are outside of the soul.If it should happen that an inquiry is into the soul, then the knowledge will necessarily exist in the thing which is known.
well as in the soul of the knower.This kind of knowledge is fundamentally distinct from the knowledge one gains when looking into one's own soul. 25l-Ġazālī's mistake was not only that he did not know how to argue properly about relation, it was also because he made "a syllogism between what is not seen and what is witnessed." 26That is he made a syllogistic inference about God's knowledge based on his own, human knowledge.This syllogism meant that he combined the two kinds of knowledge that Averroes mentioned in his discussion of topos 27, knowledge by syllogism of things outside the soul and knowledge of what is in the soul.In fact, al-Ġazālī applied what he knew from his own soul's knowledge to a kind of Knowledge that is distinctly outside of his soul, viz.God's knowledge.Al-Ġazālī's big problem, then, was that he did not know how to make topical arguments of things in relation to one another.This may have been due to the fact that al-Ġazālī never had the opportunity to read Averroes' Middle Commentary on Aristotle's Topica.
In responding to al-Ġazālī and in solving the initial doubt, Averroes employs an argumentation technique he recommends throughout his commentaries on the Topica: he takes the opposite, not of the proposition T, but of one of its terms, in this case God's knowledge. 27Al-Ġazālī themselves in themselves and in the soul of the knower.Should an existing thing change, the knowledge would also change.When it comes to the soul's knowledge of soul, then there is no possibility of a par between the knowledge of the knower and the knowledge in the thing known.Indeed, since they are identical, the soul is the cause of the knowledge of the soul.That God knows himself would follow from the Arabic translation of Aristotle's Metaphysics Λ; Averroes, Averroes, Tafsīr mā baʿd aṭ-ṭabīʿat, vol.3, p. 1692 (T.51.r).See also Averroes' comments on p. 1700-1701 (C.51.r-s).Cf.Steven Harvey, "Notes on Maimonides' Formulations of Principle K," Iyyun: The Jerusalem Philosophical Quarterly, vol.68 (2020), p. 233-244.See also Averroes, Tahafot al-tahafot.L'incohérence de l'incohérence, ed.Maurice Bouyges (Beirut, Dār al-Mašriq, 3rd ed., 1992), p. 459, 22. Accordingly, he is the cause of his own knowledge.This may be at the heart of what Averroes has in mind in the "Epistle on Divine Knowledge" when he distinguishes between generated knowledge which is caused by the existing things and eternal knowledge, i. e., divine knowledge, which is the cause of those things and which God knows through knowing himself (para.7).This argument is highly conjectural.Note that Di Giovanni argues that God's knowledge of himself can not be productive of his own knowledge.According to him, the arguments in the "Epistle on Divine Knowledge" and Metaphysics Λ are not consistent."Philosophy Incarnate," p. 152-155. 26Butterworth trans.modified, p. 41.Butterworth translates the term qiyās here "analogy."Readers who consider that what is meant here is not a proper syllogism are invited to replace "syllogism" and "syllogistic inference" with "analogy" throughout this paragraph.Cf. n. 24 above.and others had assumed that God's knowledge is like any other knowledge, but Averroes argues that it is in a different state (al-ḥāl … ḫilāf…).The state of human knowledge is dependent on, or the cause of the existing things which it knows.Consequently it changes when they change.If God's knowledge is in a different state, then it is not dependent on the existing things which it knows.Averroes, however, goes beyond what one could infer from taking an opposite view and says that God's knowledge is the cause of the created things, or that the created things are dependent on God's knowledge.This fundamentally different kind of knowledge does not change even when the existing things change.This allows him to adopt proposition T, since created things in God's knowledge are unchanging, even if the created things in themselves are subject to change.Eternal knowledge is not affected by any changes in the created things, since it is prior to them in causation and independent of them.This effectively solves the doubt, in Averroes' view.
It should be clear by now that the "Epistle on Divine Knowledge" is a thoroughly dialectical work, and follows the criteria for dialectic that Averroes himself finds in his commentaries on Aristotle's Topica.I have discussed in some detail the dialectical character of all of the internal dialogues.The frame dialogue, which is not adversarial, would seem to suggest that the dialectic between Averroes and the unnamed addressee is for the sake of practice and learning, as outlined at the opening of the Middle Commentary on the Topica and at the very end.Thus, e. g. in his comments on Topica VIII.4,Averroes refers to those whose intention is training in this art and determining the thing sought which is spoken of with regard to the demonstrative science, not those whose intention is contention. 28

‫البرهانى‬ ‫العلم‬ ‫نحو‬ ‫فيه‬ ‫يتكلمون‬ ‫الذى‬ ‫المطلوب‬ ‫وتوطئة‬ ‫الصناعة‬ ‫بهذا‬ ‫الارتياض‬ ‫غرضهم‬ ‫الذين‬ ‫الغلبة.‬ ‫غرضهم‬ ‫الذين‬ ‫لا‬
Earlier, at the opening of Book VIII, Averroes notes, "the philosopher and the dialectical person share in the inquiry into discovering the topos." 29If the topos here is something like Divine Knowledge as causal knowledge, then the entire "Epistle on Divine Knowledge" could be an exercise in coming to discover that.It is thus possible to see the work as a dialectical work aimed at coming to the basis of an argument about 27  Divine Knowledge with an eye to demonstrative sciences, i. e., as on the way to philosophy proper.Also, at the very end of the Middle Commentary Averroes interprets Aristotle's statement, "One should not engage in dialectic with everyone, nor should one exercise with one who just happens to be there." 30verroes takes this to mean that one should avoid using dialectical arguments with the "dialectical person" (al-insān al-ǧadalī) whose intention is training (al-ʾirtiyyāḍ). 31This would seem to indicate that dialectical arguments ought to be taken up with those who are not dialectical people, but people training in dialectical arguments in order to gain proficiency in demonstrative science, i. e., with potential philosophers.

METAPHYSICS AND THE EPISTLE ON DIVINE KNOWLEDGE
In fact, Averroes hints at an even more specific intended readership for the dialectical arguments of the "Epistle on Divine Knowledge" at the end of paragraph 2. There Averroes says, "For one who does not know the knot will not be able to untie it" ‫الحل(‬ ‫على‬ ‫يقدر‬ ‫لم‬ ‫الربط‬ ‫يعرف‬ ‫لم‬ ‫من‬ ‫ٕنه‬ ‫.)فا‬This would appear to be a restatement of what Aristotle says at Metaphysics Β, 995a29-30: "It is not possible to untie a knot about which you are ignorant" (λύειν δ' οὐκ ἔστιν ἀγνοοῦντας τὸν δεσμόν).Usṭāṯ's translation of Metaphysics Β, which was the one Averroes used at least when composing the Long Commentary on the Metaphysics, 32 renders this line as follows: ‫الرباط‬ ‫جهل‬ ‫من‬ ‫يحل‬ ‫ان‬ ‫يقدر‬ ‫.ولا‬ 33 The similarity between this line and the one at the end of Averroes' "Epistle on Divine Knowledge," para.2, is quite clear.Indeed, the two are so similar that there is virtually no room for doubting that the "Epistle on Divine Knowledge" is referring to Metaphysics Β.
What does Averroes mean to convey by this reference?Well, to my mind, it is a rather clear signal to any reader familiar with Metaphysics Β that the inquiry presented in the "Epistle on Divine Knowledge" is intended to correspond to the kind of inquiry Aristotle describes in Metaphysics Β. Recall that Metaphysics Β is the book in which Aristotle presents a series of ἀπορίαι which must be addressed before beginning the search for knowledge (ἐπιστήμη).Now Averroes, in both the Middle and Long Commentaries on the Metaphysics, identifies the process of addressing these ἀπορίαι as dialectic.I bring here what he says in the Middle Commentary, since it is probably chronologically closer to the writing of the "Epistle on Divine Knowledge" (and the approach in the Long Commentary does not differ significantly for our purposes).Since the Middle Commentary is not extant in Arabic, 34 I bring it only in the 14th century Hebrew translation of Qalonimos ben Qalonimos of Arles.There we find the following: We must first examine the deep questions that are mentioned in this science which we seek … Indeed, this is necessary because the first thing those who want to grasp knowledge of things and their principles do is make a strong inquiry into the dialectical statements that are doubtful from among the deep questions in that genus. 35

‫החכמה‬ ‫בזאת‬ ‫יזכרו‬ ‫אשר‬ ‫העמוקות‬ ‫בשאלות‬ ‫ראשונה‬ ‫לחקור‬ ‫מוכרחים‬ ‫אנו‬ ‫להשיג‬ ‫הרוצים‬ ‫פעל‬ ‫שהתחלת‬ ‫לפי‬ ‫מחוייב‬ ‫זה‬ ‫היה‬ ‫ואמנם‬ … ‫הנה‬ ‫המבוקשת‬ ‫המספקים‬ ‫הנצוחיים‬ ‫המאמרים‬ ‫החקירה‬ ‫חוזק‬ ‫הוא‬ ‫והתחלותם‬ ‫הדברים‬ ‫ידיעת‬ ‫הסוג…‬ ‫באותו‬ ‫אשר‬ ‫העמוקות‬ ‫מהשאלות‬
The ἀπορίαι have apparently become deep questions.Indeed, it seems to me that the question-answer format of addressing these issues, as we find, for example, in Metaphysics Β, played a large part in Averroes' association of these questions with dialectic.The doubt associated with the questions is, no doubt, another factor in Averroes' decision to connect the ἀπορίαι with dialectic.
Averroes continues, For in as much as the doubter is unable to understand some of the deep The bind, or knot, is thus associated with two chief components of dialectic, doubt and the stance between two opposite propositions.The doubter who is caught in this bind is accordingly in the predicament of dialectic, as Averroes understands it.
Averroes goes on: One who is in doubt about something cannot resolve his doubt with something from within the genus of statements which necessarily led him into the bind on that matter, i. e., the dialectical statements, but rather with another genus of statements, i. e., demonstrative statements. 37חייבו‬ ‫אשר‬ ‫המאמרים‬ ‫מסוג‬ ‫הוא‬ ‫בדבר‬ ‫ספקו‬ ‫שיתיר‬ ‫אפשר‬ ‫אי‬ ‫בדבר‬ ‫המסופק‬ ‫המאמרים‬ ‫והם‬ ‫אחר‬ ‫מסוג‬ ‫אבל‬ ‫הנצוחיים,‬ ‫המאמרים‬ ‫והם‬ ‫הענין‬ ‫באותו‬ ‫הקשר‬ ‫המופתיים.‬ Clearly, the resolution of the doubts raised through questioning and dialectic is through demonstration, rather than through dialectic.In other words, true solutions to metaphysical questions are through demonstrations, not through dialectic.
Nevertheless, Averroes gives us the following syllogism accounting for why dialectic is useful at the beginning stages of studying metaphysics.
If grasping the truth about these deep questions is resolving the bind that occurs to the understanding with inquiry about them and if this resolving occurs after the bind, it necessarily follows that before inquiring into them, you should first inquire into the statements that are similar in understanding to the bind.These are the dialectical statements.This is one reason it is necessary to precede deep questions with a dialectical inquiry.This syllogism is clearly intended to show that although demonstrations are preferable, we ought to begin with dialectical statements before proceeding to demonstrations.Yet, what kind of syllogism is this?Clearly it is of the first figure: If a is b and b is c, then a is c.As such it is 36 Averroes, Commento medio, ed.Mauro Zonta, vol.2, t. 1, p. 9. 37 Averroes, Commento medio, ed.Mauro Zonta, vol.2, t. 1, p. 9. 38 Averroes, Commento medio, ed.Mauro Zonta, vol.2, t. 1, p. 9.
valid.Yet examination of the first premise, viz.that grasping the truth about these deep questions is resolving the bind, makes clear that this is not a demonstrative premise.Indeed, the notion that resolving questions is grasping the truth does not completely conform to what Averroes had just said in the previous sentence, viz.that demonstration is the proper way to the truth.While demonstration could be in answer to questions, it need not be.Rather it would seem to be the case that what Averroes has in mind here is dialectic, especially in light of the conclusion.That dialectical resolution of doubt is grasping the truth is at best a dialectical premise, accepted by dialecticians, but not by those of the demonstrative class.That is, this syllogism is a dialectical syllogism.
Why does Averroes employ a dialectical syllogism to argue for the importance of dialectic?Let me suggest a dialectical answer.Either the reader recognizes it as a dialectical syllogism or not.If he recognizes it as dialectical and is familiar with demonstrations, then he does not need to work too much on the questions and answers in Metaphysics Β, but can skim them over or skip them and then move on to demonstrations.If not, then he must learn them and thoroughly familiarize himself with the kinds of syllogisms before he can move on to do demonstration proper.This kind of dialectical syllogism, then, performs a didactic function; it works with and encourages students who have not thoroughly understood the content of the Posterior Analytics tradition, while also indicating to those who do understand the syllogism that this is a dialectical, not demonstrative argument. 39

CONCLUSION
In the first part of this paper, I argued that the arguments Averroes employs in the work that came to be known as the "Epistle on Divine Knowledge" are dialectical and can be understood according to the description of dialectical arguments in Aristotle's Topica, as interpreted by Averroes in his Middle Commentary on the Topica.In the second part of the paper, I argued that Averroes uses a literary allusion to associate the "Epistle on Divine Knowledge" with Metaphysics Β, and the arguments of the "Epistle on Divine Knowledge" with the kind of dialectical 39 An anonymous reader suggests that Averroes employed a dialectical syllogism here because demonstration about God's knowledge is not possible and dialectic is the best that can be achieved.This may be the case, but Averroes is far from arguing in the "Epistle on Divine Knowledge" either that demonstration about this issue is not possible or that the argument he gives here is the best that can be achieved.
arguments we find there.Now it is also clear that the subject of the Epistle, God's knowledge of created things, is metaphysical, and indeed discussed by Averroes in his Middle and Long Commentaries on Metaphysics Λ. Averroes's discussion in those places is quite well-known and it is clear that his solution to the problem of God's knowledge of particulars is roughly the same in all places: God and God's knowledge are one and the cause of those particulars, and so his knowledge is of a different kind. 40Whatever Averroes' approach to Metaphysics Λ, it is clear that his approach to the question in the "Epistle on Divine Knowledge" is dialectical.It is dialectical, I suggest, in the way that Averroes sees Metaphysics Β as dialectical, viz. it is of an introductory kind, meant to be supplanted by demonstrations at a later point.
This use of dialectic is exactly parallel to the use of dialectic in education we find in Averroes' Commentary on Plato's Republic.As I have argued elsewhere, while Averroes generally removes dialectic and dialectical arguments from his version of the ideal city described in the Republic, dialectics is incorporated into the education of the guardians, i. e., of the potential philosophers. 41This is due to its educational value, a point which Averroes also emphasizes at the beginning of the Middle Commentary on the Topica.I believe it is clear that the "Epistle on Divine Knowledge" too plays a didactic role.It is a short dialectical solution to a problem that is treated at greater length and with better preparation in the commentaries on Metaphysics Λ.As Averroes notes at the 40  beginning of paragraph 6, the proper discussion would be long (ṭawīlan) and so what he presents here is the point (al-nuqṭa) at which this will be resolved, i. e., not the full demonstration of the resolution.In this case, the "Epistle on Divine Knowledge" is a work of didactic dialectic, meant for training potential philosophers.
Let me add as a kind of afterward that I do not think the Caliph Abū Yaʿqūb Yūsuf is the addressee of this letter, since he is not a potential philosopher.Averroes does not name the addressee of the "Epistle on Divine Knowledge," but only praises his good mind (ḏihn) and noble nature (ṭabʿ) which he says are greater (kaṯīran) than those who have pursued these sciences.Averroes then continues to say that the addressee's theoretical reflection (naẓr) has culminated in the doubt with which the "Epistle on Divine Knowledge" is concerned.Averroes refers to the addressee in the second person plural in the Epistle, which can indicate formality and respect of the kind expected in a literary treatise.In the Decisive Treatise, Averroes refers to the addressee of the Epistle as "one of our friends," and while the Arabic ṣāḥib can also mean "lord" or "master" it is more frequent in its use as "friend" or "fellow traveler." Accordingly, I do not see enough here to justify the statement that "the formula of address gives the reader to understand that the one addressed is a prince in high political office, and strongly suggests that he is Averroes' friend and patron the Almohade ruler Abū Yaʿqūb Yūsuf … for whose benefit Averroes had embarked on his commentaries on Aristotle more than a decade earlier." 42Rather, why not assume Averroes' praise for the addressee to be a genuine compliment to his abilities?Why not assume the addressee to be a student of philosophy, who is sharp, intellectually gifted, and somewhat scientifically advanced?Perhaps, indeed, he has attained the level of the student of Metaphysics Β, as Averroes' literary allusion would suggest, and he has encountered 42 Mahdi, "Averroes on Divine Law and Human Wisdom," p. 118-119.Sarah Stroumsa, Andalus and Sefarad: On Philosophy and its History in Islamic Spain (Princeton University Press, 2019), p. 134-144, calls into question the extent to which the commentaries were in fact commissioned by Abū Yaʿqūb Yūsuf, especially in light of the fact that Averroes was most likely already far into his commentary writing project before his legendary meeting with the Caliph.Di Giovanni argues that the argument of the "Epistle on Divine Knowledge" is intended to interpret the Almohad doctrine of the homonomy of knowledge between God and man in an Aristotelian manner that could encourage readers to pursue philosophy and metaphysics further.Still his view is that this work is directed toward general thinkers in Andalusia living under Almohad rule and perhaps even some immersed in theology.He does not mention the Caliph as the possible addressee.See Di Giovanni, "Philosophy Incarnate," p. 156-162.questions but is not adept enough at metaphysical demonstration to resolve them.This work would help such a person, without fully explaining all demonstrations, and at the same time steer the reader into further metaphysical speculation.