PARTS OF SUBSTANCES ARE SUBSTANCES: ANTIQUE AND LATE ANTIQUE CONTEXT

In our search for the roots of Avicenna's discussion of the definition of substance we will have to go back some distance in the history of philosophy. The starting point of the development of what was later to become Avicenna's definition of substance is undoubtedly Aristotle's Categories.

In Cat., Aristotle defines both primary and secondary substances as things “which are not-in-a-subject” (κοινὸν δὲ κατὰ πάσης οὐσίας τὸ μὴ ἐν ὑποκειμένῳ εἶναι – Cat. 3a7) or rather “not-in-a-subject at all” (ἐν ὑποκειμένῳ δὲ οὐδενί ἐστιν – Cat. 1a22, 1b3). Accidents, for their part, are in-a-subject. Being in-a-subject is articulated by Aristotle in the following manner: “By being ‘present in-a-subject’ I mean what is present not as a part in a whole, but being incapable of existence apart from the said subject” (ἐν ὑποκειμένῳ δὲ λέγω ὃ ἔν τινι μὴ ὡς μέρος ὑπάρχον ἀδύνατον χωρὶς εἶναι τοῦ ἐν ᾧ ἐστίν – Cat. 1a24-25, tr. based on Barnes).Footnote ⁸ It follows that if something is in its subject as a part of a composite, it is not in-a-subject par excellence, i.e. as accidents are. Does this mean that something that is a part of a composite is automatically a substance? Aristotle seems to give an affirmative answer: “The fact that the parts of substances appear to be present in the whole, as in-a-subject, should not make us apprehensive, lest we should have to admit that such parts are not substances: for in explaining the phrase ‘being present in-a-subject’, we stated that we meant ‘otherwise than as parts in a whole’” (tr. Barnes).Footnote ⁹ Does this apply only to organic parts of physical bodies or also to matter and form as parts of the hylomorphyic compound?Footnote ¹⁰ If one opts for the latter and believes that differentiae are in some sense forms one could conclude that differentiae are substances (as Alexander of Aphrodisias did).Footnote ¹¹ Aristotle touches upon this conclusion at Cat. 3a21. He assumes there that being not-in-a-subject is not a proprium of substances because it also applies to differentiae.Footnote ¹² One can draw a twofold conclusion from this passage: on the one hand, differentia is said to be not-in-a-subject – so we can seemingly conclude that it is substance. On the other hand, Aristotle explicitly contrasts differentiae to substances since he explains that being not-in-a-subject is not a proprium of either substance or differentia. Thus, Aristotle leaves unanswered the question as to whether we can automatically label conceptual parts of substances as substances: the case of differentiae (if they count as parts of substances) is highly ambiguous.

One of the first reactions to Aristotle's account of substance in Cat. that is known to us is found in Simplicius’ report concerning a mysterious Lucius.Footnote ¹³ Lucius’ aporia – reported by Simplicius – reveals that there was an interpretation of Aristotle's account of substance in the Categories during Antiquity that ascribed to him a view that some parts of substance – at least intelligible ones – are in their substance as in-a-subject and so ipso facto are not themselves substances. Lucius – according to Simplicius – pointed out that the “complements” of substance (τὰ συμπληρωτικὰ τῆς οὐσίας) should be thought of as parts of substance – for example, the whiteness and coldness that are parts of the substance of snow. If we accept Aristotle's definition of that which is not-in-a-subject, which allows parts of substances not to be in-a-subject, then two possibilities follow necessarily. Either (1) these complements are not-in-a-subject, in which case complements are substances; or (2) the definition of accidenthood is false and the condition “they are not parts of a subject” has to be omitted from it.Footnote ¹⁴ Since Lucius concludes that Aristotle must opt for the second possibility, he thinks that Aristotle would not agree with the first one.Footnote ¹⁵ Hence, in Lucius’ opinion Aristotle cannot agree that the aforementioned complements of substances are not-in-a-subject. For “whiteness” and “coldness” are qualities and therefore are good candidates for being accidents.

Following Lucius’ aporia Simplicius provides an answer given by PorphyryFootnote ¹⁶ where he quotes Porphyry's “In Cat. ad Gedalium”. Porphyry explicitly states there that the “complements” of substance are in their substance but not as in-a-subject – i.e. he chooses the first possibility offered by Lucius. The famous example of heat and fire comes to the forefront here: heat is in fire (ἡ θερμότης τῆς μὲν τοῦ πυρὸς) essentially (οὐσιωδῶς) and not as in-a-subject because it is its complement whereas in iron it is as in-a-subject.Footnote ¹⁷ Porphyry makes an important shift when proving that complements are not in-a-subject. He argues that only accidents that come to be and depart from their subject (γίνεται καὶ ἀπογίνεται ἐν τῷ σιδήρῳ ἄνευ τῆς τοῦ σιδήρου φθορᾶς) are in it.Footnote ¹⁸ In order to distinguish what is in-a-subject from what is not, Porphyry draws the main contrast between “substantial” and “accidental”. The former is not-in-a-subject, the latter is.Footnote ¹⁹ However, Porphyry still does not accept that everything that is substantial qualifies as substance. In his system, (1) what is in-a-subject is accident whereas (2) what is not-in-a-subject can be divided into (2a) substance and (2b) the complements of substance. If this is the case, then Porphyry introduces a third kind of entity alongside substances and accidents: “substantial complements”. Simplicius seems to grasp this difficulty in his reaction to Porphyry's opinion, although from the way he expresses himself it seems that he did not understand the argument.Footnote ²⁰ We shall come back to Simplicius’ solution to this problem later.

In In Cat., Porphyry discusses the Aristotelian passage at Cat. 3a21 where differentiae are said to have ‘being not-in-a-subject’ in common with substances. Here again, Porphyry argues that complements of substances are in their substances not as in-a-subject. However, his position retains the tension in the Aristotelian account at Cat. 3a21. On the one hand, Porphyry accepts the contrast between substances and differentiae. He further claims that differentiae are not substances according to the Aristotelian notion, but are rather just “substantial qualities” (ποιότης οὐσιώδης). That is, as might have been expected, he refuses to count differentiae as secondary substances. For, apparently, he concedes the hidden premise of Lucius’ aporia that qualities are not substances.Footnote ²¹ On the other hand, immediately in the next passage, Porphyry writes that differentiae are complements of substance and adds that the complements of substance are substances (τὰ συμπληρωτικὰ δὲ τῶν οὐσιῶν οὐσίαι).Footnote ²² As S. Strange correctly points out, Porphyry seems to be alluding here to Cat. 3a29.Footnote ²³ The first premise would be that the complements of substance are parts of substances; the second premise – that the parts of substances are substances (as at Cat. 3a29); the conclusion – that the complements of substances are substances. Again, it is not quite clear whether Porphyry makes the step that concludes that differentiae are substances, but his analysis provides some grounds for this interpretation.Footnote ²⁴ Still, the text as it stands seems rather to indicate a self-contradiction: differentiae are both “substantial qualities” and substances. This contradiction was to be removed in the later tradition, which opted for the second possibility alone.

It is important for the purposes of our later discussion to note Porphyry's emphasis, in this passage, that something's being an accident can be inferred from its being accidental. The substantial is contrasted to it as regards being not-in-a-subject. Heat for example can be removed from water without destroying its substance (ἀρθείσης γὰρ τῆς θερμότητος οὐ φθείρεται), so it is accidental for it. On the other hand, heat is substantial for fire because its removal does destroy fire's substance (ἀρθεῖσα γὰρ φθείρει τὸ πῦρ). Since Porphyry seeks to prove in this passage that differentiae are not-in-a-subject, the following theory can be reconstructed: if something is substantial it is not-in-a-subject, whereas if something is accidental, it is in-a-subject. Thus, Porphyry made a very, if not the most, important step towards the position that Avicenna would later come to attack. He introduced accidentality (i.e. extrinsicality) as a criterion for being in-a-subject or not. Only one step remains in the development of this line of thought: to remove the aforementioned contradiction by eliminating any contrast between differentiae and substances that can be suggested on grounds of Cat. 3a21.

This is exactly what later Neoplatonist Commentators would do. Simplicius reacts against Porphyry's solution to Lucius’ aporia and brings to its natural end what Porphyry began in the second section of the discussed passage from In Cat. Simplicius states that if parts of substances are substances, and complements – including differentiae – are parts of substances, we can conclude that differentiae of substances are substances – as noted earlier it was already Alexander who came to this conclusion.Footnote ²⁵

Ammonius takes up Porphyry's idea that differentiae have an indefinite status. In the case of heat in fire, we are dealing with something between substance and accident (ἀκριβῶς ἐν μέσῳ τῶν οὐσιῶν καὶ τῶν συμβεβηκότων), whereas such differentiae as “rational” and “mortal” are related to the substance (πρὸς τῇ οὐσίᾳ).Footnote ²⁶ Ammonius’ own approach to the solution of the problem is somewhat less messy than this, and anticipates that of Simplicius. Differentiae are parts of substance; they constitute it (συνιστᾶσιν). Aristotle says clearly that parts of substance are substances themselves. Ammonius emphasizes that Aristotle did not mean only physical parts at Cat. 3a29 but intelligible parts too and, among them, differentiae.Footnote ²⁷ As for Cat. 3a21, he solves the problem by saying that Aristotle is only contrasting sensible substances and differentiae, not substances and differentiae in general.Footnote ²⁸ The conclusion is simple and perspicuous: “Differentiae are substances in the strict sense” (αἵτινές εἰσι κυρίως οὐσίαι).Footnote ²⁹ Philoponus and Elias follow Ammonius in their Commentaries on Cat. 3a21, and acknowledge differentiae to be substances on the same grounds.Footnote ³⁰

With that, we have finished retracing the development of the definition of substance as applied to the case of substances’ parts. We started with Aristotle's twofold assumption: (1) that being in-a-subject excludes being a part of a composite, so that the parts of substances are not-in-a-subject; (2) that differentiae are not-in-a-subject. In the course of time, the Late Antique commentators on Aristotle developed Aristotle's account. If something is substantial, i.e. an intrinsic metaphysical part of a substance, it is substance itself – maybe because it is a form in this case, as Alexander claimed. If it is accidental it is an accident. Whether something is a substance or accident is to be inferred on the grounds of its current relation to that which it is predicated of. Thus, all the “complements” of substance, including differentiae, are substances because they are parts of substances. This fact suffices for inferring their being substance.Footnote ³¹

YAḤYĀ B. ʿADĪ AND IBN SUWĀR ON FIRE AND HEAT

The problem of the ontological status of the parts of substances emerges again in the writings of Yaḥyā b. ʿAdī. It is not quite clear why he came to speak about this problem.Footnote ³² The work most informative of his position on the issue in question is the forthcoming On Fire and Heat.Footnote ³³ In this treatise, Ibn ʿAdī discusses at length why heat's (ḥarāra) being a part of the definition of fire (nār) does not make it a substance (ǧawhar). While discussing this problem, Ibn ʿAdī reveals his understanding of the definition of substance and accident.

Ibn ʿAdī’s opponents argue as follows:

1. 1. Anything that cannot be removed from a substance in the imagination is its substantial part

2. 2. If something is a part of a substance, it is a substance itself

3. 3. Heat cannot be removed from the substance of fire in the imagination

4. 4. Thus, heat is a part of the substance of fire [from 1, 3]

5. 5. Thus, heat is a substance itself [from 2, 4]Footnote ³⁴

To this, Ibn ʿAdī reacts: “[Their saying] that everything that removes substance while being imagined as removed is a part of substance and substance as well – since the parts of substance are substances – this is a true premise (inna kull mā yarfaʿu bi-tawahhumihi Footnote ³⁵ murtafiʿan ǧawharan fa-huwa ǧuzʾ ǧawhar wa-ǧawhar ayḍan iḏā kānat aǧzāʾ al-ǧawhar ǧawāhir fa-inna hāḏihi al-muqaddima ṣādiqa).”Footnote ³⁶ It is clear that both Ibn ʿAdī and his opponents follow Porphyry's rule that everything intrinsic to a substance, i.e. substantial, is a substance. Ibn ʿAdī relies again on Porphyry's authority when he defines accident: “accident is that which comes to be and vanishes without its bearer being destroyed (al-ʿaraḍ huwa mā yakūnu wa-yabṭulu min ġayr fasād ḥāmilihi).”Footnote ³⁷ It is important to remember here that in the Categories’ definition of being in-a-subject, not the accident but the subject had to vanish in order that what is in-a-subject vanishes too (ἀδύνατον χωρὶς εἶναι τοῦ ἐν ᾧ ἐστίν):

Although these two definitions are compatible, one notes that Porphyry shifted the focus of the test from the substance onto the accident and Ibn ʿAdī follows him in this respect.Footnote ³⁸

If Ibn ʿAdī agrees with his opponents and the Porphyrian tradition that parts of substances are substances and that being accidental/substantial suffices to infer thing's being substance or accident, how can he disagree that heat is a substance within the substance of fire? Ibn ʿAdī’s answer is simple – fire is not a substance and subsequently heat cannot be labeled as “substantial” for fire and thereby itself be called “substance”. He reminds us of the definition of fire: “simple hot dry body (ǧism basīṭ ḥārr yābis)”. Ibn ʿAdī classifies fire as a “composition” (muǧtamiʿa) of substance and accidents, where the substance is “body” (ǧism) and heat and dryness are accidents.Footnote ³⁹ It is not fire that is the bearer (ḥāmil) of heat but exactly this “body”.Footnote ⁴⁰ On these grounds Ibn ʿAdī is able to reconcile his claim that heat is not a substance in fire with Porphyrian notions of substantiality and accidentality. So although the removing of heat removes also fire we cannot say that it is substantial for fire because fire is neither a substance itself nor is it a subject (mawḍūʿ) for heat. As for the “body” of fire, it is both substance and a subject for heat, but heat is not substantial for it and can be removed without this “body” being destroyed. Thus, heat is accidental for the “body” of fire although it is an integral part of fire as a whole. In this case nothing prevents heat's being an accident since it is something accidental for the substance that underlies it, i.e. the “body” of fire.Footnote ⁴¹

This solution seems to be rather ad hoc. One could possibly ask Ibn ʿAdī, how is he going to discern between the mentioned “compositions” and true substances? Why should not we suggest that “man” is also just a composition of the substance “living being” with the accident “rationality”? Parallel to Ibn ʿAdī’s analysis of fire, it is possible to take not “man” to be the “bearer” or subject for “rationality” but rather the “living being” that is also a part of his substance. As for the “living being” which is its true “bearer”, “rationality” is accidental for it. Thus, one could conclude that rationality is also an accident in “man”. Claiming that fire is an accident for the body of fire and not for the fire itself is not a solution. For the same can be said for every differentia, namely, that it always is accidental for its genus, in which case Ibn ʿAdī faces a danger that all differentiae are accidents.

However, the described analysis of fire and heat is not the only argument for heat's being an accident that Ibn ʿAdī produces in his treatise. Rather it is not a self-standing argument at all.Footnote ⁴² What Ibn ʿAdī seeks to prove is the fact that heat's inalienability from fire does not prevent its being an accident. However, it is because of heat's being an accident that we infer that fire is not a substance and not vice versa.

The true argument relies, not on Porphyrian notions of substantiality and accidentality, but rather on Ibn ʿAdī’s own understanding of the Categories’ account of substance and accident.Footnote ⁴³ One finds full definitions of substances and accidents that are based on the Aristotelian account of substance and accident in the Categories. Accident is defined as “that which exists in a thing not as its part so that its subsistence is impossible without the thing in which it is (al-mawǧūd fī šayʾ lā ka-ǧuzʾ minhu wa-laysa yumkinu an yakūna qiwāmuhu ġayr al-laḏī huwa fīhi)”. Substance is “that which is not in any subject at all (mā laysa huwa al-battata fī mawḍūʿ mā)”.Footnote ⁴⁴ As for the previous analysis, i.e. the Porphyrian approach to the establishing of substantiality and accidentality, it relied mostly on the “lā ka-ǧuzʾ minhū” part of this definition. However, the author of On Fire and Heat emphasizes on the part of the Aristotelian definition of substance that says that substance is not in any subject at all (al-battata fī mawḍūʿ mā).Footnote ⁴⁵ Ibn ʿAdī clearly understands this not only as “in any respect” but “in no possible case”. Substance, he says, has to be not in any subject whatsoever (fī mawḍūʿ min al-mawḍūʿāt al-battata).Footnote ⁴⁶ On these grounds, Ibn ʿAdī comes to the main claim of his treatise: one and the same entity that is described with the same definition cannot be both accident and substance (ġayr mumkin an yakūna maʿnā wāḥid – wa-huwa mā yuḥaddu bi-ḥadd wāḥid – ʿaraḍan wa-ǧawharan).Footnote ⁴⁷ This claim is supported by the principle of non-contradiction: it is not possible for opposites to be true at the same time (anna al-mutanāqiḍayn lā yumkinu an yaṣduqā maʿan).Footnote ⁴⁸ Why should we think that the “at the same time” is applicable to our case? Footnote ⁴⁹ For in every possible case we will deal with one and the same entity (maʿnā) or essence (ḏāt)Footnote ⁵⁰ – with something that is described by one and the same definition. Since it is well known that heat can be an accident in some cases, e.g. in iron (ḥadīd)Footnote ⁵¹ , and two opposite characteristics cannot be true of one and the same entity, it entails that heat – as an entity – is an accident in its essence and subsequently in the case of fire.

Although this argument is essentially new in the tradition it has not pushed Yaḥyā b. ʿAdī to retreat from the Porphyrian analysis of substantiality and accidentality. He concludes with the claim that the impossibility of “accidental substances” and “substantial accidents” follows from Aristotle's doctrine (rafaḍa Arisṭūṭālis min al-qisma iǧāb ǧawhar ʿaraḍī aw ʿaraḍ ǧawharī).Footnote ⁵² Unfortunately, Ibn ʿAdī breaks off his analysis of this point in this passage. However, if one combines his first semi-argument and the second argument from non-contradiction, it becomes clear how Ibn ʿAdī comes to his conclusion. It is simple in the case of the impossibility of accidental substances: if something is ever found to be accidental, it is always considered as an accident according to the Porphyrian system. The case of substantial accidents is more complex: if something is ever found to be in-a-subject, it has to be concluded that that thing, in its essence, is an accident. However, it is possible that in some particular case it happens to be an integral part of some composite entity (i.e. it cannot be removed without this entity being destroyed). In this case, it has to be concluded that this entity is not a substance – otherwise there would be something substantial that is an accident.

Thus Ibn ʿAdī has managed to combine the traditional Porphyrian approach to discerning substance and accident with his own idea, that being a substance or accident is something essential for an entity. He remains faithful to Porphyrian notions when he infers a thing's being an accident on grounds of its being accidental. This is how one establishes that heat is an accident in the case of iron. However, inalienability no longer suffices to conclude that something is a substance. Since being a substance or an accident is something essential for, say, some x, we must first investigate whether x has ever been an accident. If not, we can suggest that x is a substance qua being a substantial part of another substance (as in the case of “rationality” and “man”). However, if it was once accidental and an accident, we must conclude that it is also now an accident, being an inalienable part of the “composite” which is a combination of substance and accidents. This is how Ibn ʿAdī’s theory as it is found in his On Fire and Heat works. As we are about to see, it leads to a metaphysical problem that was noticed by Ibn ʿAdī’s opponents.

It is unknown what the positions of Ibn ʿAdī’s contemporary opponents and supporters were.Footnote ⁵³ However, his disciple Ḥasan b. Suwār b. Ḫammār wrote a whole treatise against his teacher on this topic. Although the treatise is lost, some of its arguments are preserved in margine of MS: Paris Bibliothèque nationale, ar. 2346 of Categories.Footnote ⁵⁴ This commentary shows that Ibn Suwār tried to retreat to the Porphyrian notions of substantiality/being substance and accidentality/being accident. In order to do this, he first once again emphasized the part of the definition of being in-a-subject that says that it is “not as a part of a subject”. Secondly, he tried to eliminate Ibn ʿAdī’s interpretation of “not in any subject at all” by suggesting a different linguistic reading.

Ibn Suwār's main goal is to prove that Ibn ʿAdī’s theory that substance has to be always substance, i.e. never in-a-subject (innahu laysa huwa fī šayʾ min al-mawḍūʿāt al-battata), is false.Footnote ⁵⁵ His first step is the argument that in this case all the differentiae (fuṣūl) would be accidents because they all are necessarily accidental to genera. Moreover, everything except prime matter would be an accident; for only prime matter is never in-a-subject.Footnote ⁵⁶ Since everything is actually accidental to at least prime matter, and since for Ibn ʿAdī nothing can be a “substantial accident” or an “accidental substance” at the same time it follows that there are no substances apart from prime matter. So everything that exists is a composite of prime matter and accidents, which Ibn Suwār assumes to be absurd.

Ibn Suwār's other mode of argumentation is mostly based on authorities. In order to support his position that parts of substances are inevitably substances – even if they are accidents in some other cases – he cannot help but mention that Ibn ʿAdī’s opinion finds no support among the Commentators (al-mufassirūn), nor in general among any Ancients (aḥad min al-mutaqaddimīn) and Moderns (aḥad min al-mutaʾaḫḫirīn).Footnote ⁵⁷ By contrast, Ibn Suwār quotes Ammonius, Porphyry and Simplicius in support of his own opinion. However, the most important authority remains Aristotle. It is he whom Ibn ʿAdī contradicts. Ibn Suwār draws on what is known to him as Aristotle's definition of substance and accident. Following the Porphyrian tradition he stresses the fact that “being in-a-subject” excludes being in-a-subject as its part. If substances are not-“in-a-subject” – Ibn Suwār claims – following the logical rule of negation, we have to define substance as “that which is not-in-a-subject unless it is its part… (laysa bi-mawǧud fī šayʾ illā ka-ǧuzʾ minhu…).”Footnote ⁵⁸ A little later, Ibn Suwār supports these considerations with quotes from Porphyry and Simplicius that have already been discussed above.Footnote ⁵⁹ On these grounds he agrees with Simplicius’ conclusion that everything that is a part of substance, including the so-called “constituents” (muqawwimāt), is substance. The very last argument is traditional – directing his reader to Cat. 3a29: “The parts of substances are substances (aǧzāʾ al-ǧawāhir ǧawāhir).”Footnote ⁶⁰

Thus, Ibn Suwār insists on remaining true to the Porphyrian approach to substantiality and accidentality. The main criterion is – as it was in the Commentators’ tradition – whether something is substantial or accidental. However, since that time Ibn ʿAdī created a new way of thinking – that substance has to be substance always, i.e. it cannot ever be in-a-subject or accidental. Moreover, Ibn ʿAdī relied in this argument on the authority of Aristotle who explicitly claimed that both primary and secondary substances have to be “not in-any-subject” at all (Arabic: al-battata fī mawḍūʿ mā; Greek: ἐν ὑποκειμένῳ δὲ οὐδενί).” In order to solve this difficulty Ibn Suwār appeals to a linguistic analysis of the Arabic expression “huwa allaḏī laysa al-battata fī mawḍūʿ mā”. He claims that the Arabic “mā” in the definition of substance does not mean “any subject” but rather “this particular subject” (mušār ilayhi wa-al-wāḥid bi-al-ʿadad or later al-mawḍūʿ al-fulānī).Footnote ⁶¹ From the grammatical point of view, Ibn Suwār is right and both readings are possible. However, if we address ourselves to the Greek original, the Aristotelian “ἐν ὑποκειμένῳ δὲ οὐδενί” is unambiguous. It is Ibn ʿAdī who has the correct understanding of the Aristotelian text and not Ibn Suwār. We can conclude that Ibn Suwār – being even more remote from the Greek original then Ibn ʿAdī wasFootnote ⁶² – misinterpreted the Aristotelian text. On the grounds of this misinterpretation, he concludes that Ibn ʿAdī’s approach to discerning the nature of substance and accident is wrong. Ibn Suwār is able to claim that one and the same thing can be one time substance and one time accident – as heat in fire and heat in iron.

The dispute “on fire and heat” among the two representatives of the Baghdad Peripatetic School provides two principal issues. Firstly, none of them questioned the Porphyrian approach that, from accidentality, inferred being-an-accident, and from substantiality, inferred being-a-substance. However, Yaḥyā b. ʿAdī claimed that substances have to remain substances always, i.e. they cannot ever be in-a-subject. This pushed him to the conclusion that one and the same thing cannot at one time be an accident and at another a substance; for being one or the other is something essential to them. Opposed to this, his disciple Ibn Suwār remained true to the Late Antique tradition in all respects. For him, the substantiality and accidentality of a thing only depends on its current circumstances. That's why one and the same thing can be at one time a substance – if it is a “complement” of some other substance – and at another an accident – if it is accidental for that substance.

It is hard to say what the fate of this dispute was in the Baghdad School after Ibn Suwār's treatise against Ibn ʿAdī. We cannot find this question explicitly discussed in the writings of the next generation. However, there is a sign that Ibn Suwār's opinion won out. Abū al-Faraǧ b. al-Ṭayyib – Ibn Suwār's disciple – apparently follows the traditional way of approaching substantiality and accidentality in his commentary on Categories. He does not show any knowledge of Ibn ʿAdī’s arguments and his position is compatible with those of Ibn Suwār and the Commentators. He claims that “differentiae of substances are both substances and accidents – they are accidents in relation (bi-al-qiyās ilā) to the genus because they are inherent in it (daḫīla ʿalayhi), and they are substance in relation to the species because they constitute it (kāna minhā tuqawwamu).”Footnote ⁶³ This position corresponds to Ibn Suwār's opinion that a thing's substantiality and accidentality is ambivalent. Ibn ʿAdī’s theory of substance and accident was lost for the Baghdad Peripatetic School. However, it was rapidly picked up by Avicenna in order to turn it against the Baghdad School itself.

AVICENNA ON THE DEFINITION OF SUBSTANCE AND ACCIDENT

Having looked at the approaches to the notions of substancehood and substantiality in Late Antiquity and the Baghdad Peripatetic School, we now turn to Avicenna and his reformist reaction against the previous tradition. As was noted at the beginning, one finds a discussion about the proper definition of substance and accident in al-Šifāʾ, al-Nafs I, 1. It immediately appears to be related to the discussion about the ontological status of the parts of substance. Avicenna remarks at the end of his brief analysis of this problem in al-Nafs that he discussed it in his “Logic” at length. Apart from this, Avicenna also touches upon the same problem in Ilāhiyyāt II, 1 – although the discussion there is very short – and it too closes with a reference to the “Logic”.Footnote ⁶⁴ In both cases Avicenna apparently means Kitāb al-Šifāʾ, al-Maqūlāt I, 3–6 and especially chapter I, 6 “On the falsehood of the saying that one thing can be both accident and substance from two [different] perspectives”.

One of the main goals of Avicenna's investigation of the definition of substance and accident in Maqūlāt I, 3–6 is a differentiation between accidental/substantial and accident/substance. This is supplied by a strong emphasis on the essential character of being substance or accident. As it becomes clear even from the title of Maqūlāt I, 6, the position that he wishes to refute is that something can at one time be an accident and at another time a substance (or even at the same time from different perspectives). This targeted position is identical with that of Ibn Suwār and Ibn al-Ṭayyib.

Avicenna first introduces the position that he is going to refute. It has three arguments in favor of the claim that one and the same thing can be both accident and substance:

1. (1) His opponent supposes that differentiae are substances (fuṣūl al-ǧawāhir ǧawāhir) and at the same time they are qualities (al-kayfiyyāt), i.e. a kind of accident (ʿaraḍ). On these grounds he draws the conclusion that one and the same thing can be both substance and accident (ka-anna al-šayʾ yakūnu ʿaraḍan wa-ǧawharan).Footnote ⁶⁵

2. (2) The second argument touches upon the status of form in matter. Form is in a material bearer (ḥāmil) not as its part. It is consequently an accident of matter. However, it is at the same time a part of the composite substance (al-ǧawhar al-murakkab i.e. the substance composed of form and matter). So, it is also substance because a part of substance is a substance (ǧuzʾ al-ǧawhar ǧawhar).Footnote ⁶⁶

3. (3) The third argument deals with the difference between whiteness (bayāḍ) and white (abyaḍ), the second being a composite of substance and whiteness (maǧmūʿ ǧawhar wa-bayāḍ). This composite is itself substance. Since whiteness is a part of white which is substance, it is substance itself. However, the same (bi-ʿaynihi) whiteness is known to be in some subject (like Socrates) not as its part. So, it is accident in relation to its subject and substance in relation to the composite “white”.Footnote ⁶⁷

A little later Avicenna explains the general methodological approach of his opponents: They distinguish substance according to three conditions: (a) it is in-a-subject as its part; (b) everything that is in-a-subject as its part is not an accident and (c) everything that is not an accident has to be substance.Footnote ⁶⁸

It is beyond any doubt that Avicenna's dialectical opponents represent the view of the Late Antique tradition welcomed by the Baghdad Peripatetics. In Ilāhiyyāt II, 1, Avicenna makes it clear that his discussion of the ontological status of the parts of substance depends on his knowledge of the same discussion in the Baghdad Peripatetic School:

It is not absolutely clear who in particular is being attacked and what treatises he had read on this problem. There is no doubt that Avicenna knew Yaḥyā b. ʿAdī’s position on the topic – since his own will be built on it. Moreover, in On Fire and Heat we have already seen the argument mentioned at Ilāhiyyāt II, 1 that heat is a substance in fire because it cannot be removed from it. Argument (2) stands close to the argumentation of Ibn Suwār and Avicenna pays much attention to proving that form's being in matter is not being in-a-subject.Footnote ⁷⁰ Argument (1) rather reminds one of the Late Antique considerations about substantial qualities: they are substantial and therefore substances according to Porphyry's rule but are still also qualities.Footnote ⁷¹ Avicenna's expression of the relativity of substancehood “bi-al-qiyās ilā” immediately reminds us of the same expression in Ibn al-Ṭayyib's commentary on Categories. We can thus conclude that Avicenna's discussion about the parts of substance was rooted in the Baghdad Peripatetic context, so that he directly or indirectly knew the main positions and arguments – and most importantly that of Ibn ʿAdī. So much for Avicenna's opponents; let us turn now to his own position.

Avicenna begins by defining both substance and accident. The reality of the essence (ḥaqīqat ḏātihi) of substance does not exist in-a-subject at all (lā fī mawḍūʿ al-battata), being not its part in such a way that it would not be able to exist being separated (mufāraqatuhu). Accident, on the contrary, inevitably is in something so that its quiddity (māhiyyatuhu) does not come into existence (lā taḥṣulu mawǧūda) unless there is a thing in which it inheres.Footnote ⁷² We can easily see that Avicenna lays a strong emphasis on the notions of essence and quiddity of a thing while he defines substance and accident. It is these terms that allow him to make the main observation of his argumentation:

Avicenna claims that something cannot be both accident and substance in different circumstances or under different relations because to be in need of any subject in order to exist or not is something that depends on its essence or quiddity and cannot vary. How do we know whether something has this need or not? Avicenna gives an answer to this question while dealing with the arguments of his opponents. The thing does not need to be in-a-subject if it is never in the subject (laysat fī šayʾ min ašyāʾ kamā yakūnu al-ʿaraḍ fī al-mawḍūʿ) – i.e. it is never an accident (lam yakun ʿaraḍan al-battata).Footnote ⁷⁴ Avicenna understands the “al-battata” from the definition of substance as “never”.Footnote ⁷⁵ If something is indeed never in-a-subject it can be acknowledged to be substance – i.e. to be not-in-a-subject in itself (fī nafsihi).Footnote ⁷⁶

Relying on these premises Avicenna solves the problem of the parts of substance. It is wrong to think that a thing's “being in-a-subject as its part” automatically renders it substance. It is still possible that it is a part of a composite (murakkab) where it can be an accident.Footnote ⁷⁷ How should we know then whether it is substance or not? We should investigate if it is ever in-a-subject. If not, it can be concluded that it is substance. If yes, it has to be admitted that this thing is accident – even if it is a part of its subject in this case and in thousand other cases.Footnote ⁷⁸ Thus, a differentia is recognized to be substance not because it is a part of some of its concrete subjects, but rather because it is never in-a-subject and always a part of it. As for the relation between differentiae and genus, it is like that of form and matter – i.e. not as accident in-a-subject.Footnote ⁷⁹

In the analysis and argumentation described so far Avicenna apparently follows Yaḥyā b. ʿAdī. Both of them refuse to conclude that something is substance only on the grounds that it is in-a-subject yet as its part. Something has to be always not-in-a-subject in order to be labeled as substance. Being a substance or an accident is something dependent on a thing's essence and is not relational. Being both a substance and an accident – either in one and the same case in relation to different circumstances, or at different times – is excluded by both authors. However, Avicenna does not only accept Ibn ʿAdī’s method to recognize substances. He goes further so that an important difference arises between the two philosophers in their theories of substantiality and accidentality.

It was established earlier that Ibn ʿAdī does not consider composites (murakkabāt) as substances. On these grounds he escaped from being forced to refute the whole Porphyrian method of distinguishing substance and accident. Ibn ʿAdī also maintained that there cannot be “substantial accidents” and “accidental substances”. However, we have seen that this could force him to the conclusion that nothing is a substance except prime matter (everything else being accidental for it). In order to escape this problem, the later Baghdad tradition rejected Ibn ʿAdī’s idea that an accident is always an accident. Their system allowed “substantial accidents” and “accidental substances” meaning by this “substantial to one subject and accidental to another”. Now, Avicenna takes Ibn ʿAdī’s path and assumes that accident is always accident, since it is its essential characteristic. How then does he escape the problem that everything would be an accident according to this theory? In order to do this, Avicenna refutes the whole approach of the preceding tradition – both Ibn ʿAdī and his opponents. He claims that being substantial/accidental and being substance/accident are not the same.

Avicenna establishes that ‘being a substance’ (ǧawhariyya) or ‘being an accident’ (ʿaraḍiyya) is not something relational (li-anna al-šayʾ bi-al-qiyās ilā šayʾ…), but rather they occur to a thing because of “itself” (li-annahu fī nafsihi ka-ḏālika).Footnote ⁸⁰ Thus, something is an accident if it is in-a-subject and needs this or some other subject in order to exist. Its coming into existence happened through the existence of its subject (ḏātuhu qad ḥaṣala mawǧūdan fī mawḍūʿ li-annahu mawǧūd fī hāḏā al-mawḍūʿ). It is an accidental accident (ʿaradī wa-ʿaraḍ) if it additionally is not a constituent (muqawwim) of its subject, i.e. it is not its subject's part.Footnote ⁸¹ The same works for substantial (ǧawharī). It only means that something is a part of an underlying subject, i.e. it is a constituent of it. However, something’s being substantial does not render it a substance. For being a substance means “being in its ‘self’ in no need at all of a subject” (fī nafsihi ġayr muftaqir ilā mawdūʿ al-battata). Being a substance is something non-relational whereas being substantial is relational.Footnote ⁸²

Thus, being substantial has a meaning different from being a substance and likewise being accidental is different from being an accident.Footnote ⁸³ Being substantial or accidental depends on a thing's current relation to its surroundings. It means that something is either added to a substance or is its integral part.Footnote ⁸⁴ By contrast, being a substance or an accident is non-relative and essential for a thing. It is its essence, i.e. the thing itself apart from its surroundings, that reveals whether the thing needs a bearer to come into existence or not. As for the notions of substantial and accidental they express whether the thing is added or not. Being a substance or an accident expresses an existential (in)dependence, i.e. does something need something else to come into existence or not – despite not being connected causally to it. Avicenna claims that the latter is an absolutely different story from the former. Our knowledge whether something is added to something or is its integral part does not reveal whether it is existentially independent or not.

Avicenna distinguishes between existential independence and cognitive separability. There is no difference between substance and accident at all on the level of epistemic analysis (fī al-wahm): both substance and accident can be conceptually separated from each other.Footnote ⁸⁵ However, on the level of extra-mental reality, a substance is independent whereas an accident cannot continue to exist being separated from it (lā yumkinu mufāraqatuhu li-mā huwa fīhi).Footnote ⁸⁶ On the grounds of this approach Avicenna creates a further – this time methodological – difference between substantial/accidental and substancehood/accidenthood. In the former case, it is conceptual analysis that provides us with information about whether something is substantial or accidental. If S can be separated conceptually from P, then P is accidental for it, as “black” can be conceptually removed from Ethiopian.Footnote ⁸⁷ On the contrary, if we cannot mentally entertain S without ascribing P to it, P is essential for it, as “figure” and “triangle”.Footnote ⁸⁸ However, in the analysis of existential dependence, epistemic analysis is claimed by Avicenna to be misleading. It does not matter whether one can conceive of an accident without its subject or not. Instead Avicenna suggests observing whether some P comes in a concrete case (muʿayyan) into existence only while some S underlies it. This means that the cause of the actual subsistence of P (not of its essence!) lies in S (ʿilla qiwāmihi hiya annahu fīhi).Footnote ⁸⁹ Although Avicenna says that one can conceive of Ethiopian with the attribute “black” and “black” without Ethiopian, black depends in its existence on Ethiopian when it comes to be in him. Hence the analysis levels of substantiality and substancehood are different ones. The one involves conceptual analysis in order to draw conclusions about the essential or non-essential character of an attribute. It builds the structures of essential dependence among the things. However, the other (substancehood/accidenthood) does not investigate what is essential and what is not. It reveals a different kind of ontological relation: something can come into existence only while something underlying exists. Although the essence of “blackness” does not anyhow depend on its subjects it needs them in order to come into existence. This latter is to be called “existential dependence”.Footnote ⁹⁰

Thus, in order to establish whether the thing under investigation is substance or accident we have to prove whether it is per se existentially independent. This is ascertained by discovering whether it is existentially independent in all possible cases. If it is existentially independent in every case known to us – i.e. it continues to exist even if its subject is removed – then it is independent per se and subsequently is a substance.Footnote ⁹¹ If it is found at least in one case to be dependent, then it is an accident in itself. This was also Ibn ʿAdī’s conclusion. However, now Avicenna differentiates between accidental and accident. This is precisely what allows him to escape Ibn ʿAdī’s dilemma, which entailed that all things are accidents because they are accidental, i.e. additional, to prime matter. The same worked for differentiae, which are similarly added to their genera. Avicenna disagrees; differentiae are not dependent on the genus or forms on matter. For genus is also dependent on differentiae and matter on forms.Footnote ⁹² Avicenna apparently thinks that mutual dependence rules out the sort of dependence involved in “being in-a-subject”. Since he has also drawn a distinction between accidental and accident, the conclusion follows: differentiae are accidental to genera but they are not accidents in them. Only if it happens for them to be once an accidental accident, are they accidents in their genera. Otherwise they are substances.Footnote ⁹³ The same works for matter and forms. Although forms are accidental to matter – for they are obviously additional to it – they are not in matter as in-a-subject and accordingly they are not accidents but substances.

Avicenna thus is committed to the view that there should be some substantial accidents and accidental substances. For instance, if fire is substance while being composed of some substance and the accident “heat”, then there are some substantial accidents. But are all the composed things (murakkabāt) substances in Avicenna? Rather not.Footnote ⁹⁴ Nevertheless it would suffice that composites according to Avicenna have some quiddity (māhiyya) or essence (ḏāt).Footnote ⁹⁵ For Avicenna states that “substantial” means the same as “essential” – i.e. being an integral part of an essence.Footnote ⁹⁶ Since being substantial or essential means nothing more than being integral or being added to something that has an essence, it follows that there are some substantial accidents. Namely, those accidents which are integral to the essences of the composites are indeed substantial accidents. What about accidental substances? Avicenna's strict distinction between accidental and accident is problematic in many respects because accidentality (like additionality) implies more or less the same meaning of ontological dependence as being an accident.Footnote ⁹⁷ It is consequently hard to find any accidental substances – cases when something is “added” yet is independent. Nevertheless, Avicenna explicitly states that there are accidental substances in Naǧāt Footnote ⁹⁸ and in Ilāhiyyāt III, 3.Footnote ⁹⁹ In both cases Avicenna resorts to the help of a rather artificial example that was also mentioned in the third argument of his opponents. Whiteness (bayāḍ) is an accident and is accidental. However, “white” (abyaḍ) – which means “that which possesses whiteness”Footnote ¹⁰⁰ – is substance yet also accidental for e.g. Zayd. The “white” is a kind of composite substance being composed of substance (Zayd?) and the accident “whiteness” while this substance is something accidental to Zayd. This theory is rather obscure. Avicenna probably means that “Zayd is being white” makes “the white” an ostensible substance, as when we say “This white [thing] which is Zayd”. Thereby Avicenna explicitly accepts accidental substances. “This white thing” is an ontologically independent subject being an accidental state of “Zayd”. Nevertheless Avicenna himself accepts that “accidental” and “accidenthood” are necessary concomitants of each other although they do not denote the same thing (wa-hāḏāni al-maʿnayāni wa-in talāzamā).Footnote ¹⁰¹ It is also clear why: the accidental substance “the white” is a composite of “whiteness” and “substance” and it always involves an accidental accident “whiteness” as its part. Again, although “the white” and “whiteness” are necessary concomitants they are different ontological entities with different characteristics: “the white” is an accidental substance whereas “whiteness” is an accidental accident. Does this mean that “accidental” and “accident” are only intensionally different, whereas extensionally the same? Avicenna's position on this is not clear enough here. In the case of whiteness it is the case. However, “accidental” could also include such cases as two bound substances like soul and body. They are “additional”, i.e. accidental to each other, but neither of them involves an accident. In this case we have two accidental substances but no accidents, so that the “accidental” and “accident” are also extensionally different.

To sum up: how can we decide whether some “B” which is predicated of “A” designates a substance or an accident according to Avicenna? First, we have to decide whether B is substantial or accidental to A, i.e. extrinsic or intrinsic. In order to do this, we check whether “A” can be conceived of without having “B”. If so, then we still have to check, whether “B” existentially depends on “A” or just accompanies it. If it is dependent on “A”, then “B” is an accidental accident. If “B” is independent though additional, it is an accidental substance. On the other hand, if “B” appears to be required for “A” the procedure is different. We have to check whether in any among the observed cases “B” appeared to be an accidental accident. If not, but has always been either an accidental substance or something substantial we can conclude that it is a substantial substance in this case. If, however, it has been an accidental accident at least once, then it is a substantial accident in this case. One could wonder, why not just check whether “B” depends on “A” as we did in the case when it was accidental (not required for “A”). Avicenna never clarifies this point, but we could suggest that, if “B” is substantial for “A” it will always seem that it depends on “A” – is it a substance or accident in itself. In order to escape this mistake – because this kind of dependence is something different from what Avicenna wants with “accident” – Avicenna suggest a different methodological route.

The most important theoretical consequence of this procedure is the differentiation between two kinds of metaphysical relation that two things may have to each other. (1) One is “being additional” or “being integral”. It corresponds to what we nowadays would call “essential grounding”. For it signifies the fact that an essence is grounded by its constituents (like “rationality” and “man”) and not grounded by the extrinsic attributes. This relation is checked by the test whether the conceptual negation of the predicate removes the subject. (2) As for “being dependent” or “independent”, it can be labeled as “existential grounding”; for it signifies that an attribute P is dependent in its existence on S – irrespective of whether P is essentially connected to S or not. The existential grounding relation is checked by statistical analysis: whether P's existence has ever been found dependent on the existence of an S.

Before we conclude, one remaining difficulty has to be solved. In Maqūlāt II, 1–2 Avicenna, following the Peripatetic tradition, tries to establish that neither “being” (mawǧūd) is a genus for all the categories, nor is “accident” (ʿaraḍ) a genus for the nine categories. In both cases his argumentation works in the same manner: he clarifies that neither “being” nor “accident” are constituents (muqawwim) of categories whereas genus has to be a constituent. Avicenna first shows that “being” is not the genus of categories – even if we agreed that it is used in them univocally (bi-al-tawāṭuʿ).Footnote ¹⁰² He proves it on the ground of his famous essence-existence distinction: if we imagine the quiddity of a triangle this does not entail that we imagine it being existent – whereas it does entail imagining it as a “figure” (lasta taḥtāǧu fī taṣawwurika māhiyyat al-muṯallaṯ an tataṣawwara annahu mawǧūd ka-mā taḫtāǧu an tataṣawwara annahu šakl). Thus, existence is not a constituent of a triangle's quiddity (fa mā kāna miṯla al-wuǧūd fa-laysa muqawwiman li-al-māhiyyatihi). “Being” is not something denoting an entity integral in the quiddity of a thing (fa-innahu ġayr dāll ʿalā maʿnā dāḫil fī māhiyyāt al-ašyāʾ). So, it cannot be its genus.Footnote ¹⁰³

The same works for “accident” in relation to the nine categories. It does not denote the natures (ṭabāʾiʿ) of these categories but rather a relation to what they inhere in (al-nisba ilā mā huwa fīhi). Avicenna explicitly states that being an accident is something accidental “ʿaraḍī” for the nine categories so that it is not constituent of their quiddities (amr ġayr muqawwim li-māhiyyātihā). Parallel to his proof that the existence of triangle is not obvious for us from the very notion of triangle, Avicenna says that neither is being accident obvious for the nine categories and had to be proven for each of them in the First Philosophy.Footnote ¹⁰⁴ So, “accident” is not a constituent and consequently not the genus of the nine categories.Footnote ¹⁰⁵

How can Avicenna assume that a thing is substance/accident in its essence (fī nafsihi or fī ḏātihi) if being an accident is something “accidental” for it? Can something be essential and accidental at the same time? Avicenna does not seem to see any inconsistency here but simply repeats that being in-a-subject is essential for accident, i.e. its essence entails it (ḏātahu taqtaḍī).Footnote ¹⁰⁶ However, this inconsistency can be solved if we investigate different meanings of “essential” in Avicenna as well as pay attention not only to what is parallel between the case of existence and accident but also to the differences between them.

The “essential” (ḏātī) can have several meanings in Avicenna. For us two of them are important now. First “essential” as it is used in Madḫal I, 5–6 where it means just “constituent” (muqawwim). It is in this respect that accident was called “accidental” – because it is not a constituent of essence. However, the essential necessity of accidenthood for accidents is understood in this article in a different way. Avicenna alludes to it when he defines both substantial and some accidental parts as: “that which occurs to the thing not through a cause external to it” (lā yakūnu lahu bi-ʿilla ḫāriǧa ʿan ḏātihī). Even something accidental can be labeled as essentially necessary in this sense of essential because the quiddity entails (muqtaḍiya) it.Footnote ¹⁰⁷ It is this sense of essential where existence and being an accident differ. The former is neither a constituent of a thing nor is it accidental yet essentially necessary. The latter is not a constituent either. However, it is essentially necessary because the very essence of a thing entails its being an accident. Hence Avicenna, speaking about the parallel between the relation of accident to the nine categories and that of being to quiddities, makes a restriction: “as regards not being integral to quiddity”.Footnote ¹⁰⁸ Thus, the parallel between the non-essentiality of existence and being an accident is valid only in the framework of the first meaning of essentiality, i.e. where essential is identical to constituent. By contrast, as for the second meaning of essential, existence is not essential whereas “being accident”, i.e. “being in need of a subject when existing” is essential.

Thus, the difficulty is solved. Avicenna's essentialist approach to defining the condition of being substance or accident remains valid. It is the essence of a thing that entails its being substance or accident – irrespective of what surrounds it. Avicenna's main insistence is that this happens due to the existential independence or dependence of a thing that is predicated of it per se.