1.1. The Epistemological Context

The logical-epistemological system of classical kalām, first of all, is not proposition-based. Although theologians often speak of assertions (e.g. qawl) as being true or false, at the most basic level their reasoning concentrates on entities (shayʾ) and notions, rather than propositions as such. The primary object of enquiry X is typically an entity, a property, or a state of affairs, and the aim of enquiry is to predicate X of either of the following two judgements (ḥukm):

1. a. Affirmation (ithbāt). The affirmation of X means that X exists, or is a true state of affairs.

2. b. Negation (nafy). The negation of X means that X does not exist, or is impossible (muḥāl) of existence, or is an untrue state of affairs.

If it is true – that is, if it corresponds (muṭābiq) to reality – and if it is arrived at through sound means, each particular judgement of this sort will constitute an individual knowledge-item (ʿilm). If the judgement is untrue – that is, if it does not correspond to reality – it will be an individual item of falsehood (jahl), that is, an erroneous conviction.

For example, theologians will refer to each of the following as a knowledge-item. (As classical kalām did not develop a formal style of formulating statements, the judgement assigned to X can be expressed in various ways depending on the context, including by using verbs or verbal nouns, e.g. ‘we affirm’ [nuthbitu], ‘is affirmed’ [thābit], or ‘must be affirmed’ [yajibu, or yanbaghī ithbātuhu].)

1. a. The affirmation of a Creator (ithbāt al-ṣāniʿ) (an entity);

2. b. The negation of a second pre-eternal being (often referred to simply as nafy al-thānī);

3. c. The affirmation of the attribute ‘pre-eternal’ for the Creator (ithbāt ṣifat al-qidam) (a property);

4. d. The affirmation of the property of ‘being conditional upon the presence of life’ for each of knowledge, capacity and volition;

5. e. The negation of the co-location of any two atoms (what we may describe as a state of affairs).

If a particular judgement does not correspond to a true state of affairs (e.g. the affirmation of a second pre-eternal being), it will be a falsehood.

Knowledge-items divide into two categories. First, immediate (ḍarūrī) knowledge-items are obtained by the knower without prior cogitation, effort or choice. These include, most importantly, self-evident (badīhī) knowledge-items, such as my knowledge that I exist and that a thing cannot be, at once, living and dead, or pre-eternal and created, and knowledge-items gained through sensory perception (idrāk, ḥiss). Second, discursive (naẓarī), or acquired (muktasab), knowledge-items are drawn from other knowledge-items that are already present in the knower's mind through the process of reflection (naẓar), or cogitation (fikr). If it reflects soundly, the mind must rely on proper evidence (dalīl), be it either rational (ʿaqlī) or scriptural (samʿī), and by this means arrive at a new knowledge-item, referred to as ‘what is evidenced’ (madlūl, madlūl ʿalayhi). Except where otherwise indicated, ‘evidence’ in the present article will refer to rational, to the exclusion of scriptural, evidence.

Both the evidence and what is evidenced are found, strictly speaking, not in the mind, but in the external world of objects, though they may be represented in mental knowledge-items. A case in point is a piece of writing, which evidences the existence of a writer. The evidence, just like that which is evidenced, can be either an entity, a property, a state of affairs, or a combination of these. A piece of writing, for instance, consists of ink (entity) arranged in specific shapes on a surface (state of affairs). The evidentiality (dalāla, wajh al-dalāla) of the entity, property or state of affairs that serves as evidence is effectively an indicator (ʿalāma) that is found within that item of evidence. It could, for example, be some kind of pointer (ishāra), or an effect (athar); and, as such, it will underpin the relation (taʿalluq) between the evidence and what is evidenced. By virtue of my recognition of this relation, my knowledge of the evidence becomes an antecedent (aṣl) from which I arrive at a consequent (farʿ) knowledge-item.Footnote ⁷

The process of inference (istidlāl) occurs when, motivated to enquire on a given problem, one reflects on one or more potential items of relevant evidence, with the aim of hitting upon (intizāʿ, istinbāṭ) the exact evidentiality of one would-be item of evidence. Theologians speak of different ‘methods of inference’ (ṭuruq al-istidlāl, ṭuruq al-istishhād), which are procedures or forms (rasm) of reasoning –essentially valid forms of argument – that allow us to identify and apply evidentialities.Footnote ⁸

1.2. The Argument from Ignorance

The question we now need to consider is this: What view should I take concerning X if I investigate sufficiently in search of relevant evidence either to affirm or to negate X, but identify none? According to some theologians, X in this case can be neither affirmed nor negated. This view would be an assertion of epistemic possibility (tajwīz) concerning the existence (or truth) of X. For all we know, it would be possible (yaṣiḥḥu) that X exists (or is true), and also possible that X does not exist (or is false). X is then said to be ‘possible of existence’ (jāʾiz, mumkin). This view, effectively a suspension of judgement, would constitute only a supposition (ẓann), rather than a knowledge-item, concerning X.

Another opinion is that from the absence of evidence (intifāʾ al-dalīl) to affirm X, we must infer that X should be negated, i.e. that X does not exist, or is false. In other words, it would be false to assert that X is possible of existence (tajwīz, yaṣiḥḥu ithbātuhu).Footnote ⁹ (No Muslim theologian ever maintained the reverse, namely that the absence of evidence to negate X can, by itself, justify the affirmation of X, a view that goes against the grain of Islamic theology, though it occasionally features in other contexts.) In this case, the fact of the absence of evidence for X will itself be referred to as ‘evidence’ (dalīl); and the argument that starts from this evidence and arrives at the negating of X will be treated by its proponents as a valid method of inference (istidlāl).

An argument premised on this principle (that the absence of evidence for X indicates the absence of X) is what we would term (pejoratively, I should add) an ‘argument from ignorance’.Footnote ¹⁰ As we shall see, this form of argument was accepted as valid by many, probably most, Muʿtazilīs and by some classical Ashʿarīs. The only extant substantial defences and most applications, both explicit and implicit, of this argument occur in Muʿtazilī sources. And it appears that the rate of occurrence of applications of the argument from ignorance in the works of the early-sixth/twelfth-century Muʿtazilī al-Malāḥimī is considerably higher than it in earlier texts, especially those of the Bahshamī sub-branch of Baṣran Muʿtazilism, including ʿAbd al-Jabbār. Both al-Malāḥimī and Abū al-Ḥusayn al-Baṣrī confirm this by stating that they use the argument ‘in many places’.Footnote ¹¹

Kalām theologians refer to the argument from ignorance in various ways that indicate the absence of evidence, often using the expression ‘intifāʾ’ (the reflexive of ‘nafy’; hence, ‘negated-ness’, or ‘absence’). The Muʿtazilī theologian Taqī al-Dīn al-ʿUjālī (d. late 6th/12th or early 7th/13th c.) refers to it as ‘evidence from the absence of evidence’ (dalālat nafy al-dalāla).Footnote ¹² As evidence is meant to be real and present outside the mind, the argument is also described as starting from the ‘non-existence’ (ʿadam) of evidence, rather than merely from our unawareness of any. Al-Juwaynī, for instance, refers to this principle as ‘inference from the non-existence [of evidence]’ (al-istidlāl bi-al-ʿadam).Footnote ¹³ Al-Rāzī refers to it as ‘inferring negation from the non-existence of evidence for affirmation’ (al-istidlāl bi-ʿadam dalīl al-thubūt ʿalā al-nafy).Footnote ¹⁴

Aside from these labels, the principle is articulated in a wide range of formulations, of which the following are representative examples that appear in the works of both proponents and opponents of the argument:

1. A. The absence of both self-evidence and evidence constitutes evidence for negation (intifāʾ al-ḍarūra wa-al-dalīl dalīl ʿalā al-nafy).Footnote ¹⁵

2. B. That for which there are no means to affirm must be negated (mā lā sabīl/ṭarīq ilā ithbātihi fa-al-wājib nafyuhu, or lā yaṣiḥḥu).Footnote ¹⁶

A and B are equivalent, since ‘means’ (sabīl, ṭarīq) refers to the various ways in which immediate and discursive classes of knowledge are obtained, including sense perception and evidence-based inference.

1. C. The absence of evidence for a thing indicates the absence of that thing (intifāʾ al-dalīl ʿalā al-shayʾ yadullu ʿalā intifāʾihi).Footnote ¹⁷

2. D. That for which there is no evidence must be negated (mā lā dalīl ʿalayhi yajibu nafyuhu).Footnote ¹⁸

3. E. The evidence for the non-existence of X is the non-existence of evidence for it (dalīl ʿadam [kadhā] ʿadam dalīlihi).Footnote ¹⁹

In the last three examples, the reference to immediate knowledge is omitted because it is either implicit or irrelevant, if the thing in question is not knowable through sensory perception.

All these formulations appear to start from ‘means’ – that is, sensory perceptions and evidence – that are actually present to us. So, it seems: If, presently, X is not known immediately, and if there is no evidence for X, then X must be negated. This presupposes that there is no distinction between presently-unknown objects (for which there are presently no means to know) and absolutely-unknowable objects (for which there cannot possibly be any means to know). All unknowns are necessarily unknowable, and all knowables are necessarily known.

Such an extreme view, of course, would be indefensible, and is unlikely to be what the theologians had in mind. This is confirmed in the following, more precise formulation given by al-Malāḥimī in his extensive defence of the argument from ignorance (to be treated in detail further below):

1. F. That for which there cannot possibly be any means to know whatsoever must be negated (mā lā yaṣiḥḥu an yakūna ilā al-ʿilm bihi ṭarīq aṣlan yajibu nafyuhu).Footnote ²⁰

This formulation introduces a distinction between what is circumstantially unknown but potentially knowable, and what is absolutely unknowable. A certain thing X, it would seem, may exist without us having access to any means to know it, not because such means are absolutely and necessarily absent, but because they are not readily available to us. The main cause for this circumstantial absence of the means to know X appears to be the presence of an impediment (māniʿ) preventing us from knowing X.

The distinction seems, in the first instance, to violate the argument from ignorance. For if the fact that X is not known immediately or through evidence does not directly entail the absence of X, the question then becomes: How can we tell whether a hypothetical thing X is absolutely unknowable and hence non-existent, or only circumstantially unknown and potentially knowable and hence possible of existence? Al-Malāḥimī does not answer this question. But I would suggest that the distinction only affects knowledge gained through sensory perception, as it is the only means of knowledge that can be prevented by impediments, such as the presence of an object that obstructs vision or the minuteness or great distance of the object of vision. The types of impediments that affect perception, the Muʿtazila maintain, are all known to us (maʿqūl).Footnote ²¹ By contrast to sensory perception, the process of mental reflection cannot possibly be hindered, in a sound mind, by any impediments.Footnote ²² It follows that if X is not knowable through sensory perception, and if there is no evidence for X, then we can be certain that X must be negated. So in the light of the more precise formulation F, I propose the following interpretation of the classical-kalām argument from ignorance:

For the Muʿtazila, therefore, if X is knowable through perception, I might be unable to perceive it because of a recognisable impediment that is affecting my perception of X. However, if X is only knowable through mental reflection starting from evidence for X, then such evidence, as we shall see, must be actually present.

The contexts in which the argument from ignorance is applied, in both theology and jurisprudence, are numerous and diverse. The following are some representative theological examples:

1. a. There is no evidence that God is generated in time; therefore, He is not generated in time.Footnote ²³

2. b. There is no evidence for a second pre-eternal being; therefore, a second pre-eternal being must be negated (nafy al-thānī).Footnote ²⁴ (There is evidence to affirm the first pre-eternal being.)

3. c. There is no evidence to affirm any classes (jins) of accidents (or any classes of attributes for physical objects) other than those known to us; therefore, no further classes of accidents other than those known to us are conceivable.Footnote ²⁵

4. d. There is no evidence that God has attributes other than those known to us on the basis of evidence; therefore, God has no attributes that are unknown to us.Footnote ²⁶

5. e. There is no evidence of the existence of any entitative determinants (maʿnā) in God's nature; therefore, there are no entitative determinants in His nature.Footnote ²⁷

In some cases, the argument from ignorance is the only argument put forward to prove the view in question. In other cases, it is listed among multiple arguments for the same view.

When the argument from ignorance came to be defined as a distinct form of argument and how the debate surrounding it first developed remain unclear. The earliest extant substantial discussions in kalām, to my knowledge, are dated to the late fourth/tenth century. And already at that stage it was viewed, including by its exponents, as a special, and unavoidably suspicious, form of argument. For unlike other forms of argument it started, not from real evidence, but from the absence of evidence, and hence appeared to produce something (a judgement, and hence a knowledge-item) out of nothing. No one, it seems, considered this principle self-evident; so it required proof. To this end, two proofs were adduced. The first, deployed by the Muʿtazila, will be termed the ‘argument from the unknowns’, and is discussed next. The second, encountered in some classical Ashʿarī sources, will be termed the ‘all-or-none argument’, and will be discussed in § 2.1 below.

All extant defences of the argument from ignorance, of which I am aware, occur in kalām texts rather unexpectedly as interpolations in places where this form of argument is applied in relation to a particular theological problem. In the following exploration of two versions of the argument from the unknowns, we shall concentrate on three substantial passages that occur (a) in the volume on the createdness of the Qurʾān in ʿAbd al-Jabbār's Mughnī;Footnote ²⁸ (b) in the discussion of God's knowledge in Abū al-Ḥusayn's Taṣaffuḥ al-adilla;Footnote ²⁹ and (c) in the discussion of God's attributes in al-Malāḥimī's Muʿtamad.Footnote ³⁰

1.3. ʿAbd al-Jabbār's Argument from the Unknowns

Muʿtazilī sources substantiate the principle, ‘If there is no evidence for X then X must be negated’, by arguing that refraining from the negating of X ‘will lead to jahālāt’ (yuʾaddī ilā al-jahālāt, or ilā jahālāt).Footnote ³¹ In most cases the argument is given without further elaboration, and even in the three passages just mentioned the concept of the jahālāt and how these can be entailed are presented quite ambiguously. So let us start by defining two different, but related, senses of the term ‘jahālāt’.

In most occurrences in this context, the expression ‘jahālāt’ occurs as the plural of ‘jahl’, as opposed to ʿilm (pl. ʿulūm) understood in the sense of ‘a knowledge-item’.Footnote ³² Two definitions of ‘jahl’ are given in the sources.Footnote ³³ The first (henceforth jahālāt ₁), referred to as ‘plain ignorance’ (jahl basīṭ), is the absence of any knowledge of X (by which the classical theologians actually mean the absence of a judgement of affirmation or negation on X; the mere conception of X does not constitute knowledge). The second (henceforth jahālāt ₂), referred to as ‘compound ignorance’ (jahl murakkab), refers to erroneous conviction, a misbelief, concerning X, i.e. a firmly-held conviction that does not correspond to reality (iʿtiqād jāzim ghayr muṭābiq). A misbelief, or falsehood, may be based on either uncritical imitation (taqlīd) or specious evidence (shubha).Footnote ³⁴

So, in broad outline, what I am referring to as the ‘argument from the unknowns (jahālāt)’ proceeds as follows. If, despite the absence of evidence for X₁ we assert that X₁ is possible, we will have to concede that any number of other things, properties or states of affairs (X₂, X₃, … X_n) for which no evidence is at hand are likewise possible. It will follow that there could be countless objects (X₂, X₃, … X_n) that are present but unknown, indeed unknowable, to us. The would-be fact of our ignorance of each of these unknowns is an ‘instance of ignorance’ (jahālāt ₁).

This implication may seem harmless enough. However, it leads to a further implication: for by conceding having jahālāt ₁, it is argued that we must then concede having jahālāt ₂. In other words, by conceding the possibility of countless unknowns (X₂, X₃, … X_n) we must also concede the possibility that many of our knowledge-items (Y₁, Y₂, … Y_n) are in fact little more than misbeliefs (jahālāt ₂).

The sources establish this implication using two distinct strands of the argument from the unknowns. The earlier strand appears to have been used by the Bahshamī Muʿtazila and is associated in the sources with ʿAbd al-Jabbār, though it is unclear whether he was the first to articulate it. The second strand is put forth approximately a century later by al-Malāḥimī, whose Muʿtamad contains, to my knowledge, the most extensive extant defence of the argument from ignorance in a classical theological text. Al-Malāḥimī's version of the argument from the unknowns may have been first introduced by Abū al-Ḥusayn al-Baṣrī, whose sub-branch of Baṣran Muʿtazilism he followed.Footnote ³⁵

In the Mughnī, ʿAbd al-Jabbār argues that conceding the possibility (tajwīz) of what is neither known immediately nor attested by evidence will compel us to concede the possibility of unknowable classes of entitative determinants (maʿnā), i.e. accidents.Footnote ³⁶ These will be jahālāt ₁. To concede that unknowable accidents are possible, he goes on to argue, ‘will lead to jahālāt’ – that is, jahālāt ₂ – in the sense that it will render a good deal of our knowledge into misbeliefs. To substantiate this claim, he specifies major areas of knowledge that will be affected should we postulate the possibility of such accidents. His most detailed discussion appears in a passage in a book entitled Sharḥ al-jumal wa-al-ʿuqūd, which is cited by al-Malāḥimī.Footnote ³⁷

First, if it is plausible for unknown accidents to exist (a case of jahālāt ₁), then they could act as causes (ʿilla, muʾaththir).Footnote ³⁸ This will undermine our knowledge of cause-and-effect relationships (which might then become a case of jahālāt ₂), to the extent that we will be unable to identify with certainty the cause for any property or occurrence. For the true cause could in fact be an accident that is entirely unknowable to us, rather than the accident that we had identified, through some means or other, as the cause. For instance, the two accidents – the true cause and the apparent cause – could be merely concomitant.

Second, we will lose our knowledge of the relations of contrariety between different accidents (another case of jahālāt ₂).Footnote ³⁹ As contrary accidents cannot co-exist in the same substratum, an accident (say, whiteness or dryness) that inheres in a certain substratum will cease to exist if and when a contrary (ḍidd) accident (blackness or moistness) comes into being in the same substratum and replaces it. However, this principle will become conjectural if the possibility of unknowable things is conceded. For we will then have to concede that the passing away of the prior accident may be due to an unknown accident, rather than to the subsequent accident known to us. It will even become plausible for supposedly contrary accidents to co-exist in the same substratum, even though this does not normally occur.

Third, we will no longer be certain of the relation between the act and its agent, and hence unable to attach praise or blame for any acts.Footnote ⁴⁰ For it will be possible that the act is engendered not by the agent's capacity, but by an unknown thing – which, of course, will in turn demolish all the ethical and theodicean doctrines of the Muʿtazila. This third implication involves a multitude of misbeliefs (jahālāt ₂).

Such implications, ʿAbd al-Jabbār goes on to argue, will render unsound the principles which underpin evidence (uṣūl al-adilla). And if what we deem evidence (dalīl) can no longer indicate what it is presumed to evidence, then the same will be true of expressions (ʿibāra),Footnote ⁴¹ with the effect that we will lose our confidence concerning the objective realities (ḥaqīqa) that we take ‘nouns’ (ism) to signify. We could consequently claim, for instance, that the words of the Qurʾān mean whatever we want them to mean. The connection that ʿAbd al-Jabbār makes between evidence and nouns seems to be that both are considered objective indicators of specific realities (ḥaqīqa), and that it is by virtue of this property that they provide access to knowledge of the realities indicated.

ʿAbd al-Jabbār concludes that ‘feigning ignorance’ (tajāhul) of the realities of nouns and entitative determinants in this manner must be rejected as absurd and disingenuous. He does not explain why, but he would expect us to look for an answer in his treatment of cognate strands of scepticism elsewhere in the book, a task that goes beyond our present scope.Footnote ⁴² In any case, since those who reject the argument from ignorance against whom ʿAbd al-Jabbār here disputes are theologians rather than sceptics, the spectre of extreme scepticism that he invokes may seem an apt point to end the argument.

1.4. Al-Malāḥimī's Argument from the Unknowns

Al-Malāḥimī, the pre-eminent propounder of the argument from ignorance, cites ʿAbd al-Jabbār's argument from the unknowns, which postulates unknown classes of entitative determinants, including the three major implications given as cases in point. However, his source apparently does not go on to illustrate how these implications impact the epistemic status of evidence and language. In his account of ʿAbd al-Jabbār's argument, therefore, the scepticism that follows from the notion that our world could be haunted by things that are totally unknowable to us is extreme, but not necessarily global. It threatens to undermine major doctrinal cornerstones of theology, demolishing many of our certainties, but it will only subvert knowledge of various classes of objects, rather than knowledge per se. So in al-Malāḥimī's view, a respondent could simply accept the three foregoing implications and concede that we lack knowledge of certain classes of objects, without becoming liable to the charge of being stubbornly irrational (mukābir). For since, according to ʿAbd al-Jabbār himself, our knowledge of the relations between causes and effects, contrary accidents, and agents and acts, is discursive and acquired (muktasab), rather than self-evident and immediate (ḍarūrī), a sound-minded person (ʿāqil) could plausibly deny any or all of these relations.Footnote ⁴³ The weakness of this version of the argument from the unknowns, according to al-Malāḥimī, is that it fails to deliver a conclusive knockout, but presupposes the opponent's commitment to several specifically Muʿtazilī doctrines. It is these doctrines that depend on, and presuppose, the soundness of the argument from ignorance, not the other way around.

So let us turn to al-Malāḥimī's own defence of the argument from ignorance, the most substantial in an extant source, and historically, as we shall see further below, the most important. He includes this defence in one of the many places in the Muʿtamad in which the argument from ignorance is applied. The subject is introduced as follows:

The wording suggests that al-Malāḥimī considered himself the first to establish this principle in a ‘conclusive manner’ (bayān shāfī). Yet his defence may have been borrowed from Abū al-Ḥusayn al-Baṣrī who too writes that he applies this principle ‘in many places’.Footnote ⁴⁵ The increased rate of application of the argument from ignorance, and probably the mounting criticism voiced against it as a consequence, may have given one of these two theologians impetus to offer a more robust defence of the argument than was afforded by ʿAbd al-Jabbār. In view of the absence of more substantive evidence to the contrary, we must treat al-Malāḥimī's defence as essentially his own.

Like his predecessors, al-Malāḥimī proves the argument from ignorance by appealing to the argument from the unknowns. He writes:

In contrast to ʿAbd al-Jabbār's version of the argument from the unknowns, al-Malāḥimī presents his own version as a tighter and epistemologically more systematic basis for the argument from ignorance, since it demonstrates that to reject the latter will vitiate (afsada) absolutely all knowledge, specifically immediate knowledge and discursive knowledge.

First, the view that unknowable things could be in existence will undermine confidence (thiqa) in self-evident knowledge, because it will force us to concede that our perception (the only way in which we gain immediate knowledge of the existence of things) could suffer from impediments (māniʿ) that, whilst being entirely unknowable to us, prevent our perception of some objects and distort our perception of others. There hence could be major errors in the knowledge we obtain through sensory perception.Footnote ⁴⁸ Four illustrative examples are given:

1. 1. A mustard seed could, in reality, be as big as a mountain, though it appears miniscule because most of its parts are imperceptible to us.

2. 2. Each human being could, in reality, have multiple heads and eyes, though only one head and two eyes are perceptible to us.

3. 3. In our vicinity, we could presently be surrounded by elephants, armies and loud noises that are entirely imperceptible to us.

4. 4. Hazards could be present in our path – mountains and walls that we would collide with, wells that we would fall into, and fires that we would walk into – though none of these things are in any way perceptible to us. We should, therefore, walk as cautiously as the blind.

So the consequence of rejecting the argument from ignorance is not only that there could be entities in existence that we are entirely unaware of (jahālāt ₁), but moreover that we could be misperceiving those entities that we do perceive, such that objects would have parts or attributes that are imperceptible and unknowable to us (jahālāt ₂). The world could be a very different place from what we believe it to be.

Next, al-Malāḥimī turns to discursive knowledge, arguing that the possibility of the mind suffering from insurmountable epistemological handicaps, unbeknownst to us, implies that for each bit of evidence adduced to affirm X, there could be conclusive counter-evidence (shubha, also referred to as dalīl) negating X, which are present but entirely unknowable to us. Likewise, for each bit of evidence adduced to negate X, there could be conclusive counter-evidence affirming X, which are present but unknowable to us. We would subsequently lose confidence in all discursive knowledge; for no matter the quality and quantity of evidence that we can adduce for any view that we hold, we can never be certain of the absence of counter-evidence that, if it were to become available to us, would force us to give up that view.Footnote ⁴⁹

In short, al-Malāḥimī argues that by rejecting the argument from ignorance, we do not only admit to being ignorant of any entities, properties or states of affairs that are unknowable to us, but must furthermore concede that we can only have presumptions and conjectures (ẓann), to the exclusion of knowledge-items, whether immediate or discursive. He writes,

Al-Malāḥimī is careful to emphasise that in his version of the argument from the unknowns, the implication is global scepticism, a consequence that no kalām theologian will concede. As we have seen, in his account of ʿAbd al-Jabbār's version of the argument, the implication is only that certain doctrines will collapse.

In both ʿAbd al-Jabbār's and al-Malāḥimī's versions, the argument from the unknowns is peculiar in that it is a reductio ad absurdum that proves a metaphysical point on the basis of purely epistemological grounds. The possibility of existence of an unknown thing is shown to be absurd, not metaphysically, but only in so far as it leads to scepticism: If an unknowable thing X₁ is possible, then X₂, X₃, … X_n will be equally possible (jahālāt ₁), and as a result our supposed knowledge-items Y₁, Y₂, … Y_n will turn out to be mere presumptions (jahālāt ₂), a consequence that would be undesirable and detrimental to us in various ways. One obvious counter-argument is that there might be a separate set of criteria through which the truth of Y₁, Y₂, … Y_n is guaranteed, but which is unaffected by the possibility or otherwise of X₁, X₂, … X_n. The availability of such a set of criteria would preclude the catastrophic implications that supposedly follow from the postulating of X₁. (This, as we shall see, is a point that al-Rāzī makes in response to al-Malāḥimī.Footnote ⁵¹) However, in the absence of such criteria, the argument from the unknowns would be incomplete without an adequate explanation of why the espousal of scepticism is absurd. What the argument from the unknowns, in both its versions, manages to achieve is to illustrate that rejecting the argument from ignorance must be made at the expense of a sacrifice so great that no kalām theologian would entertain it.

Having made his case, al-Malāḥimī considers the following five possible counter-arguments (discussed here not in their original order). As we shall see, some feature in sources critical of the argument from ignorance.

Objection 1. Before demonstrating the absurdity of affirming the possibility of existence of unknowable things, we are able to hypothesise some unknowable things without in any way undermining our knowledge, whether immediate or discursive. If our knowledge is not undermined by the hypothesising of certain unknowable things, it should not be undermined by the hypothesising of an indefinite number of further unknowable things. Al-Malāḥimī responds by explaining that postulating any unknowable thing is absurd, not inherently and immediately, but indirectly on account of its implying jahālāt ₂, an implication that is discernible only through reflection.Footnote ⁵²

Objection 2. Arguing from the absence of evidence for X to the negation of X is not more reasonable and compelling (awlā) than arguing from the absence of evidence for the negation of X to the affirmation of X. As with his reply to the previous objection, al-Malāḥimī here responds that in the former case X is impossible, not intrinsically, but on account of its consequences. In contrast, no absurdities will follow from affirming the possibility of X when no evidence is known to negate it.Footnote ⁵³ On the face of it, this view seems problematic, for the same hypothetical thing X could, at once, have no evidence either to affirm or to negate it. Hence:

1. A. There is no evidence for the existence of unicorns.

2. B. Therefore, unicorns do not exist.

However,

1. C. There is no evidence for the non-existence of unicorns.

2. D. Therefore, unicorns are possible.

The existence of unicorns cannot, at once, be both impossible (muḥāl) and possible (jāʾiz). The problem is resolved only if C is made false by A, which constitutes evidence for the non-existence of unicorns. By contrast, C does not constitute evidence for anything, and hence does not affect the verity of A. In the final analysis, the existence of unicorns is impossible.

Objection 3. Exponents of the argument from ignorance do affirm things for which no evidence is present. Theologians affirm the existence of a pre-eternal Creator, who precedes the existence of everything else, although they also hold that the only way to prove His existence is to argue from the existence of His creatures. Therefore, there was a time in the past, before God created anything, when no evidence of His existence was present. If it is true of one existent entity under one set of circumstances (aḥwāl) that no evidence exists for it, the same should be conceivable of other entities and under different circumstances.Footnote ⁵⁴ Al-Malāḥimī responds to this objection by explaining that the argument from ignorance only applies to ‘what there cannot possibly be any means to know whatsoever’, and as such is absolutely and eternally unknowable (as already discussed above). The case of God before creation is no exception: for since He has always been capable of creating things, the existence of evidence for His existence has always been possible (yaṣiḥḥu).

The argument from the unknowns, as already explained, is made from the standpoint of the knower, and starts from the grave epistemological consequences that the absence of evidence will have on him. It is not argued that the absence of evidence for an existent entity is metaphysically impossible. Therefore, no evidence is necessarily existent in the absence of a human knower. Presumably, al-Malāḥimī would explain that as soon as a human knower comes into being, God will be under an ethical obligation to make Himself known, including by creating evidence of Himself, and that it is by virtue of this duty that such evidence become necessarily existent.Footnote ⁵⁵

Objection 4. An exponent of the argument from ignorance will also concede the possibility of Gabriel being present in our vicinity at this very moment. Although his presence or absence is unknowable to us, no absurdities will follow from this instance of ignorance.Footnote ⁵⁶ This objection appears to presuppose the more basic view that Gabriel and other angels exist, which, al-Malāḥimī will argue, is known discursively, partly on the basis of scriptural evidence. It rather targets the component of his argument from the unknowns that concerns immediate knowledge obtained through sensory perception. How is Gabriel different from the hypothetical elephants and mountains that, whilst present within the scope of our sensory perception, are nonetheless imperceptible to us?

Al-Malāḥimī responds to the objection by claiming that Gabriel's presence in our vicinity is not inherently unknowable to us. It is not something for which there are no means (ṭarīq) for us to know. For, as a body (jism), Gabriel must be visible (marʾī), and is actually seen by other angels.Footnote ⁵⁷ Humans, however, are unable to see him because of an impediment (māniʿ) that is apprehensible (maʿqūl) to us, namely the subtlety (riqqa) of his body (in the sense that it consists of thinly-dispersed, as opposed to densely-agglomerated, atoms). For this reason, we must suspend judgement (tawaqqafnā) on whether or not Gabriel is now present in our vicinity.

There are two key assertions here. First, Gabriel is perceptible, so his presence in a certain location is in principle knowable to those near him. This, however, seems beside the point; for the pertinent question is not whether Gabriel is perceptible to other angels, but whether he is perceptible to us humans. Whether the hypothetical elephants (to use al-Malāḥimī's example) that are imperceptible to us are perceptible to one another is immaterial. Second, we know the obstacle that prevents us from perceiving Gabriel. So although I do not know whether or not he is present in my vicinity, I at least know why I do not have this knowledge, just as I know why I can only guess whether Zayd is hiding behind this wall or that wall. Al-Malāḥimī intends to contrast this with his aforementioned view that there cannot be any impediments that are unknowable to us, which would prevent us from perceiving objects in our vicinity. But if we suffer from an irremediable perceptual handicap that prevents us from perceiving subtle-bodied beings, could there not be such beings, other than Gabriel, in our vicinity? Al-Malāḥimī would probably answer this challenge by referring to the basic existential distinction between Gabriel, whose existence is supported by evidence, and hypothetical beings, whose existence is supported by none.

Objection 5. What exactly is intended by ‘means’ (ṭarīq) in the description ‘what is unknowable by any means’?Footnote ⁵⁸ If only the evidence of reason (ʿaql) are intended, then we must deny the reality both of things that can only be affirmed by revealed evidence, such as the details of the hereafter, and of ‘things unknown’ (ghuyūb), which are known exclusively to God. If, however, both rational and revealed evidence are intended, then we will not be entitled to negate anything simply because no rational evidence is at hand, since it may still be affirmed by revealed evidence.

In response, al-Malāḥimī distinguishes between what is knowable through rational evidence alone and what is knowable through revealed evidence alone. Only rational evidence is intended with respect to the former class of objects of knowledge. Otherwise, if in the absence of rational evidence for X, where X can only be affirmed through rational evidence, we nonetheless assert that X is possible (and hence may be affirmed through revealed evidence), then all our immediate and discursive knowledge will be vitiated, as was argued earlier. In consequence, we will have no means to prove the veracity of revelation, as it must be founded on rational grounds.

With respect to what is knowable by the evidence of revelation alone, only that type of evidence is intended. Al-Malāḥimī identifies the scope of ‘what is knowable through revealed evidence’ by two broad parameters: (a) it excludes notions on which the soundness of revealed evidence depends (e.g. that God exists, and that the world is created); and (b) it includes only notions that the mind recognises as possible (jāʾiz). If X fulfils these two conditions, one must then look for revealed evidence for X. If no revealed evidence is known, then X must be negated.

Al-Malāḥimī excludes the detailed occurrences of the hereafter as irrelevant to the problem at hand. These occurrences, he maintains, are presently non-existent, but will be knowable in the future through sensory perception. The upshot of this view and of his response to the third objection is that the argument from ignorance applies only to the present reality, and not to past or future realities.

Al-Malāḥimī goes on to make a further, rather problematic exclusion. He writes:

It is not entirely clear what al-Malāḥimī understands by the Qurʾānic expression ‘ghayb’, nor what he means by saying that the means to know such things ‘elude us’. If ‘ghayb’ refers to physical objects and places present in a supernal realm (and in kalām, of course, all things other than God are physical), such as heaven and hell, then presumably he would explain our inability to perceive them as due to certain impediments, which are essentially the same as those that prevent me from seeing the planet Pluto or Antarctica from where I am sitting. It is hard to see how any other understanding of ‘ghayb’ could be countenanced without seriously violating the argument from ignorance.

To conclude our treatment of Muʿtazilī sources, it appears that school members were motivated in their espousal of the argument from ignorance by a complex combination of commitments of theirs – ontological, epistemological, soteriological, theodicean and dialectical – a comprehensive exploration of which goes well beyond the scope of the present article. Chief among these is arguably the notion that humans are under an obligation to arrive at ‘perfect knowledge’ of God (kamāl al-maʿrifa), including knowledge of all His attributes, and that human salvation depends on fulfilling this obligation. For the Muʿtazila, the attainment of this knowledge is intrinsically obligatory, and hence neither made obligatory nor can be made non-obligatory by God's command. So it is a duty on Him to provide humans with the means necessary to fulfil this duty of theirs; for otherwise, it would lie beyond their capacity to do so (taklīf mā lā yuṭāq). Opponents of this view, including al-Rāzī, maintain that humans are not obligated to attain perfect knowledge of God, but only to know His attributes to an extent that allows them to ascertain the truth of the prophecy of Muḥammad.Footnote ⁶⁰ A further consideration motivating Muʿtazilī defences of the argument from ignorance is that by creating unknown things in this world, God, as we have seen, would not only undermine our knowledge, but He moreover would be performing an arguably useless act (ʿabath). On both grounds, this creative act would be evil and contrary to His justice and wisdom.

2.1. Investigation and Disjunction (al-sabr wa-al-taqsīm)

The first discussion of the argument from ignorance in the Shāmil occurs early in the book in the general section on the methods of inference (ṭuruq al-adilla), where it is included in the appraisal of the argument by elimination, or the so-called method of ‘investigation and disjunction’ (al-sabr wa-al-taqsīm).Footnote ⁶¹ In broad outline, this method involves, first, an investigation to identify all possible solutions to a given query, then proceeding to eliminate as many of these divisions as possible; if one division remains un-eliminated, it will be affirmed.Footnote ⁶² Al-Juwaynī makes a distinction between two types of this method, according to whether or not it is immediately evident that the disjunction is exhaustive (ḥāṣir), defined as one in which all possible divisions are accounted for (inḥiṣār). If the disjunction is dichotomous, and set out on the basis of a simple affirmation and negation, it will be self-evident that a third division is inconceivable. For example, a thing is either created in time or not created in time (i.e. pre-eternal); if we establish that it is not created in time, it necessarily follows that it must be pre-eternal. If, however, the disjunction is ‘unrestricted’ (mustarsil), in the sense that it does not consist of a clear-cut affirmation and negation, whether or not it is exhaustive will not be as obvious. This, non-dichotomous type of disjunction is often encountered when theologians seek to identify a cause (ʿilla, mūjib). A case in point is the following disjunction constructed to identify the cause of the attribute ‘knowing’ in a knower: the cause could be either the entity itself (nafsuhu), or an entitative determinant, or an external agent, or the absence of a condition, or the absence of an entitative determinant; if any four divisions are eliminated, the fifth is then affirmed. Until further evidence is provided to demonstrate the exhaustiveness of the disjunction, one cannot exclude the possibility that a further division, which may be the correct division (al-qism al-ṣaḥīḥ), is unwittingly omitted.Footnote ⁶³ The Muʿtazila respond to this objection, according to al-Juwaynī, by appealing to the argument from ignorance: ‘The evidence for negating [the unknown division] in question is the absence of evidence affirming it’.Footnote ⁶⁴ To this, al-Juwaynī replies:

In a rather casual manner, al-Juwaynī then objects to the basic notion that the very absence (ʿadam) of evidence can in itself be construed as evidence. Al-Anṣārī (d. 512/1118) writes in his paraphrase, ‘On what grounds have you transformed the absence of your knowledge into knowledge of non-existence (jaʿaltum ʿadam ʿilmikum ʿilman bi-al-ʿadam)!’.Footnote ⁶⁶ Al-Juwaynī goes on, rather half-heartedly, to disprove this notion by arguing that to count the absence of evidence for affirmation as evidence for negation is not more compelling than counting the absence of evidence for negation as evidence for affirmation (the second of the aforementioned objections considered by al-Malāḥimī). He does not elaborate, but will revisit this objection elsewhere in the Shāmil.

Although al-Juwaynī calls into question the second, non-dichotomous type of investigation and disjunction, he then attempts to salvage it. He lists three ways that could justify the application of this method of reasoning, which, we are told, are collated from different places in al-Bāqillānī's works.Footnote ⁶⁷

The first is that a valid non-dichotomous disjunction can be constructed as, or analysed to, a series of consecutive dichotomous disjunctions, each involving a simple choice between affirmation and negation, such that no possible divisions are omitted at any stage. For instance, it can be argued that the thing that engenders a certain property (ḥukm) must be either an entitative determinant or not an entitative determinant; if it is not an entitative determinant, then it must be either the thing itself or something other than the thing itself; if it is something other than the thing itself, then it must be either a necessitating cause or a voluntary agent; and so forth.

The second way is what we may call the ‘all-or-none’ argument, which, al-Juwaynī indicates, the Muʿtazila fail to appeal to in this context.Footnote ⁶⁸ He illustrates this principle, which is closely associated with the classical kalām notion of particularisation (takhṣīṣ), with the following example. If an opponent claims that a thing has an attribute that is known neither immediately nor discursively, the possibility of existence of that attribute will be no different than that of an infinite number of other unknown attributes that we may hypothesise. Since all such unknown hypothetical attributes, including the attribute affirmed by the opponent, will be equivalent with respect to both their possibility of existence and the absence of supporting evidence, we must either affirm that an infinite number of unknown attributes actually exist in the thing in question, which is absurd, or deny the existence of any. Al-Rāzī, as we shall see a little later, considers this principle to exhibit the same confusion between epistemology and ontology from which the argument from ignorance suffers. What is even more problematic here is that the all-or-none principle effectively also validates the argument from ignorance, which al-Juwaynī has just rejected.

Third, al-Juwaynī reports that in some places al-Bāqillānī maintains that ‘the absence [of evidence] may itself serve as evidence’ (al-ʿadam qad yakūnu dalīlan). Without elaborating, he avers that this line of argument (maslak, ṭarīqa) is valid and non-equivalent to the argument from ignorance used by the Muʿtazila, although it is less ‘sound’ than the foregoing two ways. For it is susceptible to the objection that the absence of evidence could be due to one's unawareness (ghafla) of evidence.Footnote ⁶⁹ It is quite perplexing that al-Juwaynī here attempts to differentiate between al-Bāqillānī's view and the Muʿtazilī argument from ignorance. This appears a rather abortive attempt to distance his esteemed predecessor from what he presents as a Muʿtazilī principle. We shall return to al-Bāqillānī shortly.

So, in the logical-epistemological part of the Shāmil, al-Juwaynī expresses reservations concerning the method of investigation and disjunction, but nonetheless attempts to salvage it. No such attempt is made in the much later Burhān, where the subject is discussed in two places. Here, only the dichotomous disjunction, which rests on a simple affirmation and negation, is accepted as sound, to the exclusion of ‘unrestricted’, non-dichotomous disjunctions, ‘most instances of which are unsound (bāṭil)’.Footnote ⁷⁰ The grounds of this assessment are still the same:

Disjunctions of this unrestricted sort can only yield conjectural conclusions (ẓann). These, al-Juwaynī adds, may be acceptable in jurisprudence, but not in the rational disciplines (maʿqūlāt), particularly theology, which demand certainty (qaṭʿ) and knowledge (ʿilm).Footnote ⁷² This objection, in the Burhān, to the application of unrestricted disjunction not only presupposes an unqualified rejection of the argument from ignorance – a significant development from his position in the Shāmil – but also exhibits al-Juwaynī's commitment to carry this rejection through to its logical implications.

2.2. Two Senses of ‘Possibility’ (jawāz)

The second discussion of the argument from ignorance in the Shāmil appears much later in the book as a digression in the chapter that interrogates the Muʿtazilī doctrine of generation (tawallud), which occurs in the larger discussion of the creation of human acts. This chapter occurs in a part of the book that is now lost. However, a summary is available in the extant abridgement of the Shāmil.Footnote ⁷³ Parts of the discussion are also reproduced verbatim by al-Juwaynī's student Abū al-Qāsim al-Anṣārī in his theological compendium the Ghunya, and transmit crucial details left out in the abridged recension.Footnote ⁷⁴

At one point in the discussion of generation, al-Juwaynī attempts, in ad hominem manner (ilzām), to force his Muʿtazilī adversaries to concede that the generator (muwallid) could be an accident that does not subsist in any substratum (maḥall), a notion that immediately raises the question whether a third type of primary created entities, other than atoms and accidents, is possible.Footnote ⁷⁵ The challenge leads to the following possible response:

This appeal to the argument from ignorance invites a dedicated response, focusing on the problem whether there could be classes of accidents, particularly accidents of colour, other than those known to us. Al-Juwaynī reports two earlier views. First, according to most Muʿtazila and some Ashʿarīs, accident classes are both finite in number (muntahiya) and determinate (munḥaṣira, maḥṣūra). Second, according to ‘many kalām theologians’, accident classes are indeterminate in number.Footnote ⁷⁷ The reasoning behind the latter view is that in the absence of evidence either to affirm or to negate them, further classes of accidents are possible (jāʾiz); and if they are possible in themselves, they will fall within the sphere of God's creative capacity (maqdūr), for only impossible (muḥāl) things fall outside this sphere.

The problem with the latter position, as al-Juwaynī observes, is that the expression ‘possible’ has a double meaning.Footnote ⁷⁸ Asserting that a class of entities is possible (tajwīz) could be an assertion of either ‘epistemic possibility’ or ‘metaphysical possibility’. By exploring this distinction, al-Juwaynī champions a third view on the question whether unknown accident classes are possible.

Epistemic, or subjective, possibility, firstly, is an admission of doubt, or conjecture (tajwīz shakk, tajwīz tashkīk, taraddud). If, for instance, I say, ‘It is possible (jāʾiz, yajūzu) that Zayd is at home’, I will be admitting that I do not know whether Zayd is at home and acknowledging the possibility that the contrary may be true. As al-Juwaynī explains, affirming this type of possibility in relation to X does not constitute a knowledge-item about X, but is ‘found within oneself’ (yajiduhu al-shakhṣ min nafsihi), which means that it expresses a fact about the individual, namely that he is unable to affirm or to negate X given the evidence available to him. Therefore, such a judgement ‘can be negated’. For instance, if I get to know where Zayd is, I will no longer have uncertainty about whether or not he is at home. I will ‘negate’ the judgement of possibility I had made earlier.

Metaphysical, or objective, possibility, on the other hand, is contingency (tajwīz imkān). To say that a certain state of affairs is possible in this sense implies two facts. First, it is not necessary (wājib), since it is only one among multiple states of affairs that in themselves are all equally possible. A case in point is that the created beings that actually exist at the present moment and their arrangements (hayʾa) and situations (waḍʿ) are contingent, and that different beings and arrangements are equally conceivable. In Ashʿarī occasionalism, there is no metaphysical necessity in the existence, number, shape, properties, or spatial arrangement of any objects in the world. Second, a possible thing is not impossible (muḥāl), and as such falls within the sphere of God's capacity (maqdūr). For instance, an object must be either moving or motionless; a third contrary is inconceivable. And although two atoms have an infinite number of possible spatial arrangements, one in particular is inconceivable, namely that the two be located in exactly the same place. In contrast to epistemic possibility, a judgement of metaphysical possibility, according to al-Juwaynī, constitutes a knowledge-item and as such can never be negated, since possible things can become neither necessary nor impossible. As knowledge-items, assertions of metaphysical possibility can be obtained either immediately or discursively. Metaphysical possibility, hence, is an objective and unalterable truth pertaining to things themselves and, in contrast to epistemic possibility, has no dependence on the individual who makes the judgement.

In the light of this distinction, al-Juwaynī resolves the ambiguity in his predecessors' application of the expression ‘possible’, which led them to maintain that accident classes are infinite in number. He argues that our mind affirms the possibility of existence of accident classes other than those known to us only in the subjective, epistemic, rather than the metaphysical, sense of possibility. So in the absence of evidence for the existence of further accident classes, we should infer neither their impossibility, nor their infinite number, but must suspend judgement (tawaqquf) altogether.Footnote ⁷⁹ He indicates that some of his predecessors likewise suspended judgement on this problem, yet this position of theirs was borne out of pious caution (iqtiṣād, hence a ‘God-knows-best’ attitude), rather than proper investigation and understanding.Footnote ⁸⁰ This suggests that al-Juwaynī considered himself the first to substantiate this view.

2.3. Criticism of the Earlier Theologians' Stance towards Philosophy

He then returns to the problem of the hypothetical third type of primary created being, another major case on which the argument from ignorance is brought to bear. In a key passage reproduced by al-Anṣārī (quoted in full in the present subsection and the next), he begins by highlighting his predecessors' neglect of this problem before taking issue with the general epistemological principle that explains this neglect. He writes that in contrast to the problem of unknown accident classes, on which earlier theologians had much to say,

The two problems mentioned at the beginning of the passage (¶¶ 1–2) are analogous in that the former concerns the possibility of an unknown class of primary created entities other than atoms and accidents, while the latter concerns the possibility of unknown classes of accidents. Where they differ, according to al-Juwaynī, is that the third class of primary created entities is not a merely hypothetical notion postulated to test other theological theses, but in fact is embodied in the teachings (madhhab) of ‘all’ the philosophers, who affirm the existence of immaterial intellects and rational souls that are neither space-occupying (unlike atoms) nor subsistent in space-occupying things (unlike accidents) (¶ 3). Unknown classes of accidents, by contrast, feature in Muʿtazilī and Ashʿarī discussions mainly as a hypothetical notion. Al-Juwaynī criticises earlier theologians for neglecting to give due attention to, and to disprove, the notion of a third class of primary created entities, even though it is represented in a heretical doctrine (¶ 1).

Al-Juwaynī then makes a connection between the earlier theologians' neglect of the philosophical doctrine of immaterial entities and their application of the argument from ignorance to the problem of a third class of primary created entities (¶ 4). The exact nature of the connection implied, however, is not stated clearly. It appears that al-Juwaynī considers both the earlier theologians' neglect of this doctrine and their application of this form of argument in this context to be simply symptomatic of their wider unawareness of the philosophers' teachings, i.e. both the doctrines and their supporting evidence and arguments. He furthermore indicates that the argument from ignorance is not only an invalid form of reasoning (manhaj) that ‘does not lead to truth’, i.e. knowledge (ʿilm), but also, considered through the prism of dialectical strategy, an inapt stance to take in respect of a heretical doctrine. Manifesting, in some cases, as a refusal to engage with notions that appear to be unsupported by evidence (as defined in the theologians' own theory of evidence), the argument from ignorance may even itself have the effect of promoting an evasive and negligent dialectical strategy, a self-contented and insular attitude and hence a broader failure by theologians to address newly-arising heresies.

Al-Juwaynī appears to favour the following alternative strategy for dealing with a hypothetical notion X for which no evidence is available. One must first reflect in search of evidence negating X. If no such evidence is identified, one must withhold judgement on whether or not X is possible. This is the point that al-Juwaynī arrives at in regard of the question whether unknown accidents are possible. However, if such evidence is identified, then it may be put forth to negate X. The task of setting out such evidence becomes dialectically imperative when X is affirmed by actual adversaries, not only because their view might constitute a dangerous heresy that ought to be confuted, but moreover, it seems, because it is likely to be backed by evidence claims. Al-Juwaynī, accordingly, does not suspend judgement on the philosophers' theory of immaterial entities, but proceeds, a little later, to disprove it using a positive argument. I shall examine this argument, which is of no relevance to our present article, in a forthcoming study on the development of theological views on body-soul dualism.

2.4. Al-Bāqillānī Revisited, and Criticised

The passage cited in the previous subsection continues as follows:

To substantiate his verdict that the argument from ignorance ‘does not lead to truth’ (¶ 4), al-Juwaynī refers us (¶ 5) to his discussion of the methods of inference much earlier in the Shāmil, which we have already examined (§ 2.1 above). But he also provides further criticism, which shows that his views on the subject appear to have developed significantly by the time he wrote the chapter on generation, which occurs in the last third of the book. Al-Juwaynī revisits, and concentrates on, al-Bāqillānī's view that ‘the absence [of evidence] may itself serve as evidence’ (al-nafy yadullu), which earlier he had attempted, quite unconvincingly, to distance from the argument from ignorance, which he associates solely with the Muʿtazila. Here, al-Bāqillānī's view is equated with the argument from ignorance.

Al-Juwaynī refers us to al-Bāqillānī's work on the principles of jurisprudence al-Taqrīb wa-al-irshād, where, we are told, he claims that the absence of both immediate knowledge and evidence may constitute evidence for negation.Footnote ⁸⁵ Al-Bāqillānī wrote three versions of the Taqrīb – long (al-kabīr), medium (al-awsaṭ), and short (al-aṣghar) – a portion of the last of which is extant.Footnote ⁸⁶ We also have an abridgement of the Taqrīb, probably of the long version, entitled Talkhīṣ al-Taqrīb, which al-Juwaynī prepared at an earlier stage in his career (and hence is unrepresentative of his developed and more independently-minded thought embodied in the Shāmil).Footnote ⁸⁷ In his section on the definition of evidence (maʿnā al-dalīl wa-ḥaqīqatuhu), al-Bāqillānī defines ‘evidence’ as ‘whatever’ (kull amr) may serve as the object of sound reflection, through which it becomes possible to acquire a knowledge-item that is not already known, whether the evidence ‘be existent or non-existent (maʿdūm), pre-eternal or created’.Footnote ⁸⁸ The expression ‘amr’ is used intentionally instead of the term ‘thing’ (shayʾ) to avoid implying that a thing could be non-existent, a notion that the Ashʿarīs reject. Al-Juwaynī's abridgement adds that ‘you might hear some ignorant people (jahala) expressing the view that using non-existence as evidence (al-istidlāl bi-al-ʿadam) is inadmissible’, and confirms that ‘it is indeed admissible to use non-existence as evidence’.Footnote ⁸⁹ Al-Bāqillānī writes that the cases in which the absence of evidence (ʿadam al-adilla) will itself be counted as evidence are too many to enumerate; so he only gives the following three examples. First, the failure of a false prophet to perform miracles serves as evidence of his fraudulence. Second, the absence of evidence of a legal ruling indicates that no ruling should be affirmed. Third, the absence of evidence for classes of attributes (ajnās min al-ṣifāt, by which mainly classes of accidents are intended here) other than those known to us serves as evidence for their non-existence.Footnote ⁹⁰ This last view contradicts the aforementioned position that al-Juwaynī attributes to al-Bāqillānī in the Shāmil, namely that it is possible that there be further, unknown classes of accidents. That al-Bāqillānī may have changed his mind on this particular question is suggested much later by al-Jurjānī (d. 838/1434).Footnote ⁹¹ All the same, what is perfectly clear is that, in the Taqrīb, al-Bāqillānī endorses the argument from ignorance and applies it to both theological and juristic problems, so much so that al-Juwaynī's abridgement of the text describes those who reject this principle as ‘ignorant’ (an insult that may have been inserted by al-Juwaynī himself).

Let us return to the above-quoted passage from the Shāmil, transmitted by al-Anṣārī. After directing us to the discussion of the argument from ignorance in the Taqrīb (¶¶ 5–6), al-Juwaynī criticises al-Bāqillānī for failing to apply this principle consistently (¶ 7). We are referred to an argument that al-Bāqillānī levels against the Muʿtazila, which is not transmitted in any of his extant works, but is summarised elsewhere in the Shāmil. Among the arguments reportedly deployed in some Muʿtazilī sources to prove that pre-eternal beings other than God are impossible is the following argument from ignorance, in which the all-or-none principle is invoked. There is no evidence for the existence of pre-eternal beings other than God. So if it is possible that pre-eternal beings other than God exist, there can be multiple views (aqwāl) on exactly how many such beings there actually are (and it might be relevant here that there are various actual doctrines affirming multiple pre-eternal beings, some affirming two, others three, and yet others more). None of these views will be more compelling than (awlā min, yatarajjaḥu ʿalā) the others, as they all equally lack the support of evidence. So these equally-compelling views will contradict (tataʿāraḍu) each other, and will hence cancel each other out (tatasāqaṭu). The conclusion is that the existence of no eternal beings other than God can be affirmed. The argument was reportedly also used by ‘many’ earlier Ashʿarīs, including Ibn Fūrak (d. 406/1015).Footnote ⁹²

Some Muʿtazilīs did indeed use the argument from ignorance to prove the impossibility of a second pre-eternal being. Al-Malāḥimī, for instance, considers different types of evidence in detail to illustrate the absence of any evidence for a second pre-eternal being, and concludes that such a being is impossible.Footnote ⁹³ However, he grounds this argument explicitly in the argument from the unknowns, rather than the all-or-none argument. He even reports that ʿAbd al-Jabbār considered the argument from ignorance inapplicable to the problem of a second pre-eternal being, precisely because postulating such a being, in his view, does not entail any jahālāt.Footnote ⁹⁴ So it appears that al-Juwaynī reproduced the version of the argument used by Ibn Fūrak and other Ashʿarīs, and erroneously ascribed it to the Muʿtazila as well. This, it should be noted, is a strange error, considering that he had earlier observed that the Muʿtazila do not employ the all-or-none argument.Footnote ⁹⁵

Al-Juwaynī remarks that the foregoing proof for the non-existence of a second pre-eternal being is considered unsound by critical investigators (muḥaqqiqūn), since it argues from the mere absence of evidence. They disprove it by arguing that just as there is no evidence for the existence of a second pre-eternal being, there is also no evidence for the non-existence of such a being (the second objection responded to by al-Malāḥimī). So if the absence of evidence for its existence implies its non-existence, the absence of evidence for its non-existence should likewise imply its existence. Given the contrariety (muʿāraḍa) of these two outcomes, they will cancel each other.Footnote ⁹⁶ We are not told here who the ‘critical investigators’ in question are. However, this is the same objection that, in the above passage (¶ 7), al-Juwaynī attributes to al-Bāqillānī in order to illustrate that although he considers the argument from ignorance valid and applies it to some problems, he nonetheless dismisses it as invalid in others. Whether al-Bāqillānī's views might have simply evolved from one position to the other is unclear.

To recap, it appears that when he begins to write the Shāmil al-Juwaynī attributes the principle that ‘the absence of evidence serves as evidence’ to the Muʿtazila, and rejects it as unsound. He attempts to distinguish between this principle and the notion, advocated by al-Bāqillānī and some other pre-Juwaynian Ashʿarīs, that ‘non-existence may serve as evidence’, and endorses the all-or-none argument in which al-Bāqillānī grounds this principle (§ 2.1 above). Later in the Shāmil, al-Juwaynī develops a relatively more consistent opposition to the argument from ignorance: he equates the earlier Ashʿarī principle to that of the Muʿtazila, he seems to reject the all-or-none argument, and he ceases to defend al-Bāqillānī and instead accuses him of inconsistency for applying the argument from ignorance in some cases and refusing to do so in others.

2.5. The Principle of Conversion (inʿikās)

The position a theologian takes on the validity of the argument from ignorance does not only affect his treatment of views that he opposes, but has implications on his own views as well. If the absence of evidence for X entails that X must be negated, the same rule should apply to cases in which evidence is adduced for X, but then proves unsound. Whence the principle that for E to be evidence for X, it must satisfy two conditions: (i) concurrence (iṭṭirād), i.e. that the presence of E indicates the affirmation of X; and (ii) convertibility (inʿikās), i.e. that the absence of E indicates the negation of X.

Al-Juwaynī discusses conversion in a separate section in the Shāmil following his treatment of the argument from ignorance in the section on the methods of inference (discussed in § 2.1 above). He reports that some ‘stupid’ jurists maintained that rational evidence must be convertible, a view that he rejects as a ‘dreadful error’.Footnote ⁹⁷ This error, he writes, arises from a confusion between evidences (dalīl) and causes (ʿilla), which is a symptom of the tendency to define ‘evidence’ in the strictest sense as something that has an extra-mental presence.Footnote ⁹⁸ Causes and effects are indeed convertible: if cause C exists, then necessarily its effect X exists, and if C does not exist, then necessarily X does not exist.Footnote ⁹⁹ This is explained as follows:

In contrast to causes, evidence does not necessitate, or engender (aththara), what is evidenced. Nor is it engendered by it. Indeed, a thing can exist without the presence of any evidence for it.Footnote ¹⁰¹ And a thing can be attested by multiple evidences, one or more of which may prove unsound though the thing continues to be attested by the remaining evidences.Footnote ¹⁰²

In the later Burhān, we encounter major transformations in al-Juwaynī's epistemology, which mostly lie beyond our present purview. However, one innovation introduced in his treatment of the conversion of evidence is pertinent and deserving of brief mention, namely that he internalises evidence, reducing it to a mental process. ‘Evidence’ (dalīl), he declares, refers to the sound and deductive exercise of mental reflection, through which new knowledge-items are obtained.Footnote ¹⁰³ It is nothing other than ‘the establishing of a consequent on the basis of an antecedent’ (bināʾ maṭlūb ʿalā muqaddam).Footnote ¹⁰⁴ It is not a thing (shayʾ) that bears a connection (mutaʿalliq) to what is evidenced, or that serves as a means (wasīla) leading to it.Footnote ¹⁰⁵ There is, hence, no real and intrinsic connection between the evidence and what is evidenced (taʿalluq al-dalīl al-ʿaqlī bi-madlūlihi lā ḥaqīqata lahu).Footnote ¹⁰⁶ Al-Juwaynī by this equates evidence with the mental process of inference, consequently denying that it has, in the first place, an extra-mental presence and that it bears an objective connection (taʿalluq) of one sort or another with what is evidenced.

This unconventional definition of evidence met with opposition among some of his immediate Ashʿarī successors. Abū ʿAbdallāh al-Māzarī (d. 536/1141), the Maghribi commentator on the Burhān, remarks that al-Juwaynī's understanding of evidence ‘differs from the view accepted by most people’, and reasserts the standard view:

Yet the tide was turning, due in large part to al-Juwaynī's destructive criticism of his predecessors' logic and epistemology, which paved the way to al-Ghazālī's (d. 505/1111) championing of Aristotelian logic in jurisprudence, leading ultimately to its introduction into the closely-associated field of theology. Al-Ghazālī himself, however, has little to say specifically on the argument from ignorance. After all, the subject, as was mentioned at the start of the present article, was not discussed in the manuals of the Aristotelian logical tradition, to which he contributed.

3.1. Establishing the Absence of Evidence

Al-Rāzī explains that kalām theologians establish the minor, particular premise, ‘There is no evidence for X’, in either of two ways. They may list all the evidence produced by any exponents of X, and then argue that it is ‘unsound or weak’. Alternatively, they may undertake an exhaustive search (ḥaṣr) for all grounds (wujūh) on which X could conceivably be evidenced, and then declare that no evidence exist because none were found (wijdān). The former method, al-Rāzī remarks, is more apt; for since in both cases, one will ultimately argue from the failure to identify any evidence, it is better to declare the absence of evidence at the outset than to burden oneself with the requirement of having to undertake an exhaustive search for grounds of evidence on behalf of the exponents of X. Al-Rāzī seems to be making a dialectical point here: namely, that if a given doctrine is refuted on the basis of the absence of evidence adduced for it, the burden of proof should lie on the exponent of the doctrine in question, rather than on the respondent.

Proceeding to query the premise, al-Rāzī observes that the statement, ‘There is no evidence for X’, can have either of two interpretations: that the arguer does not know any evidence, or that the thing itself (fī nafs al-amr) lacks any evidence attesting to it.Footnote ¹¹⁹ In a reductio ad absurdum, he then reveals both interpretations as untenable. According to the former, the argument will be as follows: ‘We do not know any evidence for X; and all that for which we do not know any evidence must be negated’. However, the major premise, that the individual's mere ignorance of a thing constitutes evidence for its non-existence, is clearly absurd; for it implies: (i) that all ordinary people (ʿawāmm) are justified in denying the existence of things for which they do not know any evidence; (ii) that the atheists' denial of the existence and oneness of God, prophecy and the hereafter constitutes knowledge, because they do not know any evidence to affirm these doctrines; and (iii) that the more ignorant the individual is of evidence, the more knowledgeable he will be!

According to the latter interpretation of the minor premise, the argument will be as follows: ‘For X, no evidence exists; and all that for which no evidence exists must be negated’. Echoing points made by al-Juwaynī, or considered and rejected by al-Malāḥimī, al-Rāzī objects that the first premise in this argument cannot be inferred simply by disproving all the evidence adduced by the exponents of X. For (i) there can be further evidence that the exponents are unaware of. Moreover, (ii) even if no evidence remains unaccounted for at present, we may obtain evidence in the future, particularly from revelation. Without revealed evidence, a myriad of things that are unknowable by reason alone will remain unbeknownst to us, such as the number of the heavens, the description of heaven and hell, the nature of reward and punishment in the hereafter, and the presence of angels and jinn in our vicinity. The obvious difficulty that the latter point appears to run into is that Muslim theologians expect no further revelations in the future. What al-Rāzī seems to be arguing here, however, is that certain things had been known only to God, and became known to humans only after God imparted some knowledge to them, and that it is perfectly conceivable that other things are still known only to God and will never become known to humans until and unless He imparts further knowledge to them. Whether this may never happen, or may happen in the hereafter, is beside the point.

Confirming his view that there are likely to be in existence things that are entirely unknown to us, since we have not encountered (lam naqif ʿalā) any evidence for them, are some of al-Rāzī's comments, in his Great Commentary on the Qurʾān, on the expression ‘the things unknown’ (al-ghayb) in Q. 2.3. This expression, he explains, denotes things that are ‘absent to our sensory perception’ (ghāʾib ʿan al-ḥāssa), which divide into things for which there is evidence, and things for which there is no evidence. The latter are known only to God.Footnote ¹²⁰

3.2. Criticism of the Argument from the Unknowns

Next, al-Rāzī turns to the major, universal premise in the argument from ignorance, ‘All that for which there is no evidence must be negated’. He provides a summary account of two arguments that are given in support of this premise, without naming specific theologians.Footnote ¹²¹ The first is no other than al-Malāḥimī's above-discussed argument from the unknowns, for which al-Rāzī most probably draws on his Muʿtamad, or possibly on Abū al-Ḥusayn al-Baṣrī (considering, as mentioned earlier, the possibility that the latter may have been the source for al-Malāḥimī's version of the argument).

Al-Malāḥimī first maintains that if unknowable things could be in existence, then we must concede that there could be imperceptible objects in our vicinity, that we could be misperceiving objects that we do perceive, and consequently that we cannot trust any of the immediate (ḍarūrī) knowledge we obtain through perception.Footnote ¹²² Al-Rāzī queries the claim that we know that we are not standing next to imperceptible mountains and do not have multiple imperceptible heads (two examples used by al-Malāḥimī). What is this knowledge based on?

If, first of all, it is premised on (mutawaqqif ʿalā) the principle that all that for which there is no evidence must be negated, then two implications will follow.Footnote ¹²³ First, the line of reasoning will be circular: All that for which there is no evidence must be negated, because otherwise there could be imperceptible objects in my vicinity; however, there are no imperceptible objects in my vicinity, because all that for which there is no evidence must be negated. Second, my knowledge of the non-existence of imperceptible objects in my vicinity will be discursive, rather than immediate, since it will be attained by means of an argument (mustafād min dalīl). Therefore, undermining the soundness of these knowledge-items will not affect the soundness of immediate knowledge per se, contra al-Malāḥimī's argument. If, however, my assertion that no imperceptible objects exist in my vicinity is not premised on the principle that all that for which there is no evidence must be negated, then the falsity of the latter principle will not affect the soundness of my assertion. Either way, al-Rāzī concludes, the claim that it would be impossible for any imperceptible things to exist in our vicinity, and that it would be absurd to think otherwise, remains an unfounded axiom.

He then turns to al-Malāḥimī's second claim, namely that if it is possible for unknowable things to be in existence, then we will have no trust in our discursive knowledge, because for each item of evidence-based knowledge that we have there could be counter-evidence unbeknownst to us.Footnote ¹²⁴ Al-Rāzī's responds as follows:

Calling attention to his promotion of Aristotelian logic in kalām, al-Rāzī reiterates the theory that certainty is attained by means of a demonstrative syllogism, which he outlines in the previous section (§ 1.6), on ‘the methods of inference’. Other than the validity of the syllogism and the truth of the premises, no further conditions are required for the certainty of the conclusion, such as proof of the impossibility of counter-evidence. Logic, hence, offers the criteria through which the integrity of our present discursive knowledge-items is guaranteed, such that even the affirmation of jahālāt ₁ will not imply jahālāt ₂ (see our earlier discussion of this problem, pp. 186–7 above).

3.3. Criticism of the All-or-None Argument

Having responded to the Muʿtazilī argument from the unknowns, al-Rāzī in the same section (§ 1.7.1) attends to the second argument given in support of the premise, ‘All that for which there is no evidence must be negated’.Footnote ¹²⁶ The argument goes as follows: That for which there is no evidence is infinite in number; therefore, if one thing for which there is no evidence is affirmed, an infinite number of things for which there is no evidence must likewise be affirmed, which is absurd. This is none other than the aforementioned ‘all-or-none’ principle that al-Juwaynī uses in the Shāmil, problematically, to salvage the non-dichotomous form of the method of investigation and disjunction.Footnote ¹²⁷ Almost certainly with al-Juwaynī in mind, al-Rāzī reveals this principle for what it is: not an alternative to the argument from ignorance, but rather an alternative theory on which to base the argument from ignorance.

Al-Rāzī analyses the all-or-none argument as an analogy (qiyās), in which the original case (aṣl) is ‘the actual existence of an infinite number of things’, and the secondary case (farʿ) is ‘the existence of that for which no evidence is available’. The judgement (ḥukm) that applies to the original case and must then extend to the secondary case is ‘being impossible’ (imtināʿ). Al-Rāzī dismisses the analogy as erroneous on the grounds that there is a clear difference (fāriq) between the two cases, namely that the grounds (ʿilla) of the judgement in the original case (the impossibility of an actual infinite) do not apply to the secondary case, because of which the judgement cannot be extended to the latter. He also notes that another tactic to undermine the analogy is to deny the judgement attached to the original case, though it is unlikely that he would actually argue for the conceivability of an actual infinite.Footnote ¹²⁸

The all-or-none principle is returned to a little later in a dedicated section (§ 1.7.5.1). We are told that when kalām theologians,

In other words, if all quantities are on a par in their intrinsic likelihood, and in the absence of an external factor to tip the balance for any particular quantity, and hence to ‘particularise’ it, then either all quantities must be affirmed, or all must be negated. This principle is applied in almost any context in which no evidence is found to indicate that a specific quantity is preponderant over other quantities, and it works both ways: either to negate quantities altogether (as when it serves as a basis for the argument from ignorance), or occasionally to affirm an infinite quantity. Al-Rāzī gives examples for each. For instance, while there is evidence for the existence of one pre-eternal being, to affirm a second pre-eternal being is ‘not preponderant over’ the affirmation of a third, or a fourth, etc.; so one must affirm either an infinite number of pre-eternal beings, which is absurd, or none beyond the one.Footnote ¹³⁰ Also, if it is proved that God has knowledge, then He either knows only one thing (the minimum quantity of knowledge required to be a knower) or absolutely all things; if it is then argued that God knows more than one thing (e.g. the two knowledge-items that He exists, and that the world exists), it will follow that He must know absolutely all things.

Citing two reasons, al-Rāzī dismisses the all-or-none principle as ‘false and extremely feeble (rakīka jiddan)’. The second is the same as the objection directed at the analogy, just discussed. The first echoes a foregoing criticism of the argument from ignorance, which highlights the earlier theologians' confusion between epistemological claims and metaphysical ones. To say, ‘No quantity is preponderant over another quantity’, can have either of two interpretations: either, ‘I do not have any evidence that indicates that one or more quantities are preponderant over other quantities’, or, ‘No quantity is in itself preponderant over other quantities’. The former proposition, al-Rāzī argues, may be true, but it does not directly entail the latter, the argument from ignorance having already been shown to be unsound. The latter proposition implies that nothing can particularise (murajjiḥ, mukhaṣṣiṣ) one quantity to the exclusion of other quantities, such that that quantity becomes differentiated (mumtāz) from the rest, without us knowing that particulariser. Al-Rāzī rejects this as a claim (daʿwā) that has never been substantiated by any evidence, and counters that it is indeed perfectly conceivable that the actual quantity in question be, say, two or three, or for occurrences to take place at a certain point in time rather than another, without us knowing why.Footnote ¹³¹

Al-Rāzī's refutation of the principle that ‘the absence of evidence can serve as evidence of absence’ is central to his criticism of the other ‘weak methods’ of earlier theology, especially analogical arguments discussed in § 1.7.2 of Nihāyat al-ʿuqūl. This is most evident when he discusses the method of investigation and disjunction, presented here as one of the main ways to determine the cause (ʿilla) in the ‘original case’ of an analogical argument. Reproducing al-Juwaynī's argument, he writes:

As al-Juwaynī does in the Burhān, al-Rāzī makes no attempt to salvage the complex, non-dichotomous type of the method of investigation and disjunction.

3.4. Muʿtazilī Reception: The Case of Taqī al-Dīn al-ʿUjālī

Although the later reception of al-Rāzī's criticism of the argument from ignorance goes beyond the scope of our present study, exception must be made for one source that reveals its immediate and remarkable impact on Muʿtazilism. The source in question is the book titled al-Kāmil fī al-istiqṣāʾ fī-mā balaghanā min kalām al-qudamāʾ, authored by a certain Taqī al-Dīn al-ʿUjālī, a sixth/twelfth- or seventh/thirteenth-century figure whose precise identity has been surrounded by a fair amount of confusion.Footnote ¹³³ The author is evidently a follower of Abū al-Ḥusayn al-Baṣrī and his school-member al-Malāḥimī, to whom he refers as ‘our shaykhs’, and his criticism frequently targets the Bahshamī branch of Muʿtazilism, including the founding figure Abū Hāshim al-Jubbāʾī (d. 321/933) and ʿAbd al-Jabbār.Footnote ¹³⁴

The Kāmil borrows liberally from al-Rāzī's Nihāyat al-ʿuqūl and al-Malāḥimī's Muʿtamad, although the former source is nowhere identified by name. That the Nihāya is a source for the Kāmil, and not vice versa, is unquestionable. For al-Rāzī's style of writing and argumentation is readily identifiable in the passages shared between the two texts; and it would moreover be highly uncharacteristic of him, in his mature and independently-minded works, to paraphrase extensively and uncritically even from the most authoritative sources, not to mention an obscure and possibly contemporaneous Muʿtazilī text. This point is corroborated by an array of specific textual evidence. For instance, as I show below, the Kāmil overlaps in some places with the Nihāya and in other places with the Muʿtamad, but never with both at once; had the Kāmil been a source for the Nihāya, there would have been evidence in the latter of material incorporated from the Muʿtamad into the former. In view of its dependence on the Nihāya, the Kāmil must have an approximate terminus post quem of 570/1175.

Following his treatment of God's attributes, the author of the Kāmil dedicates a whole chapter to the argument from ignorance. The chapter is headed, ‘On whether or not it is possible that God has an attribute other than the attributes affirmed by’ the Muʿtazila, this being one of the main problems in which the argument is deployed and the problem most relevant to the chapter's context in the book.Footnote ¹³⁵ Al-ʿUjālī begins by stating briefly that Abū Hāshim al-Jubbāʾī and ʿAbd al-Jabbār maintained the impossibility of unknown divine attributes, adducing the argument from ignorance, and he then lists several other particular problems in which the argument is applied.Footnote ¹³⁶ This discussion takes up approximately one tenth of the chapter.

The rest of the chapter consists of a general discussion of the argument from ignorance; hence the alternative chapter title that appears in a table of contents included in the author's preface, namely ‘On that that for which there is no evidence must be negated’.Footnote ¹³⁷ Contrary to what we would expect of a follower of Abū al-Ḥusayn and al-Malāḥimī, the author proceeds, not to defend the argument from ignorance, but to provide a comprehensive exposition (taḥrīr) of the opposing view, that the absence of evidence to affirm or to negate X indicates the possibility (imkān) of X and that one thus ought to suspend judgement on X (tawaqquf).Footnote ¹³⁸

This exposition is based largely on § 1.7.1 of al-Rāzī's Nihāya, with some material added from al-Malāḥimī's Muʿtamad. It begins by summarising the defences of the argument from ignorance, as follows:

1. 1. It paraphrases al-Rāzī's formulation of the argument from ignorance as a syllogism.Footnote ¹³⁹

2. 2. It summarises al-Rāzī's account of the ways in which defenders of the argument establish the first premise.Footnote ¹⁴⁰

3. 3. It then lists three proofs for the second premise, as follows:

   1. 3.1. It paraphrases al-Rāzī's summary of al-Malāḥimī's argument from the unknowns.Footnote ¹⁴¹

   2. 3.2. It reproduces al-Malāḥimī's summary of ʿAbd al-Jabbār's argument from the unknowns in Sharḥ al-jumal wa-al-ʿuqūd.Footnote ¹⁴²

   3. 3.3. It reproduces al-Rāzī's account of the all-or-none argument.Footnote ¹⁴³

The author obviously found al-Rāzī's presentation of the defenders' position so neat and accurate that he reproduced it, including the summary of al-Malāḥimī's defence, even though he had the latter's Muʿtamad at his disposal. The latter serves only as a source for ʿAbd al-Jabbār's defence.

Then follows a refutation of the argument from ignorance, which the author divides into three main parts. Here is the broad outline:

1. 4. He provides the following counter-evidence (nuqūḍ):

   1. 4.1. We must accept the possibility of some things for which no evidence is available, such as the presence of an angel in our vicinity and the existence of God before He created evidence of Himself. The cases mentioned are none other than those addressed in al-Malāḥimī's Muʿtamad.Footnote ¹⁴⁴

   2. 4.2. He counters the view (articulated by al-Malāḥimī) that the argument from ignorance only applies to things that are absolutely unknowable, rather than to those that are knowable and only circumstantially unknown. He argues that it should accordingly be conceded that we may be suffering from impediments affecting our knowledge of reality, which are presently unknown to us, but may become known in the future.Footnote ¹⁴⁵

2. 5. He then challenges the defenders of the argument from ignorance to substantiate (taṣḥīḥ) each of its two premises:

   1. 5.1. He paraphrases al-Rāzī's objections to the first premise, including the objection that although we may presently lack evidence for X, such evidence may be provided by revelation in the future.Footnote ¹⁴⁶

   2. 5.2. He paraphrases al-Malāḥimī's defence against a similar objection.Footnote ¹⁴⁷

   3. 5.3. In response to al-Malāḥimī's defence, he develops al-Rāzī's objection.Footnote ¹⁴⁸

   4. 5.4. He paraphrases al-Rāzī's refutation of al-Malāḥimī's argument from the unknowns for the second premise.Footnote ¹⁴⁹

   5. 5.5. Replicating an objection put forth by al-Rāzī against al-Malāḥimī, he briefly argues that ʿAbd al-Jabbār's argument from the unknowns is circular.Footnote ¹⁵⁰

   6. 5.6. He paraphrases al-Rāzī's refutation of the all-or-none principle.Footnote ¹⁵¹

3. 6. He paraphrases al-Rāzī's argument that if the absence of evidence for affirmation constitutes evidence for negation then the absence of evidence for negation must constitute evidence for affirmation.Footnote ¹⁵²

Only minor original contributions are made by the author of the Kāmil, particularly in 4.2, 5.3 and 5.5. The rest is collated from al-Rāzī's Nihāya and to a much lesser extent from al-Malāḥimī's Muʿtamad. Although the discussion is presented as a summary of an earlier debate, the author clearly opposes the argument from ignorance, as attested by the facts that he develops the case against it and that his treatment relies much more heavily, in both structure and content, on the Nihāya than it does on the Muʿtamad. That the author names only Abū Hāshim al-Jubbāʾī and ʿAbd al-Jabbār to the exclusion of al-Malāḥimī, the earlier and major co-follower of Abū al-Ḥusayn whose views are comprehensively demolished, may betray a desire to associate the argument from ignorance with the Bahshamīs.

The impact of al-Rāzī's refutation of this principle of classical kalām epistemology on a late-sixth/twelfth- or early seventh/thirteenth-century follower of al-Malāḥimī, the chief defender of this principle, confirms its definitiveness. Whether any later theologians rose to the task of advancing a serious response to al-Rāzī is highly doubtful, but remains to be seen.