Clearly defining the place of prophetic hadiths in the epistemology of Sunni Islam has proven extremely difficult. On the one hand, Sunni hadith scholars and legal theorists elaborated two parallel but contrasting scales for describing their certainty that a hadith represented the authentic words or deeds of the Prophet. On the other hand, these Muslim scholars employed hadiths in a wide range of scholarly discourses and homiletics with seeming disregard for both of these epistemological rankings. The scale developed by legal theorists and adopted into Sunni Islam in the late fourth/tenth and early fifth/eleventh centuries has been well studied. (1) But what about the epistemological scale of the formative Partisans of Hadith (ahl al-hadith), the original "Sunni" (ahl al-sunna wa-l-jama'a) scholars, who preceded this adoption? What did al-Shafi'i (d. 204/820) or Ibn Hanbal (d. 241/855) mean when they said that a hadith was "sound" (sahih)? (2) Did they mean that they believed that the Prophet had actually said that statement, or that he probably had, or did they only mean that it was indicative of his normative precedent? When al-Bukhari (d. 256/870) or al-Tirmidhi (d. 279/892) declared a hadith lo be sound or "fair" (hasan), how did those terms reflect their opinion on the historical truth of the hadith in question? If a sahih hadith was an authenticated report of the Prophet, how could scholars so regularly state that one hadith was "sounder" (asahh) than another? (3) How do we translate the historical vision of early Muslim scholars into terms that are comprehensible in modern Western thought? (4)
In this article, I contend that ahl al-hadith did not view the historical reliability of hadiths through the epistemological lens of later Sunni legal theorists. Rather, they conceived of sound hadiths as providing what I will define as historical certainty. Despite their open obsession with the authentication of hadiths through the isnad, they frequently employed hadiths that did not live up to the sahih rating because the scholars felt that circumstances corroborated their reliability. These ahl al-hadith scholars also used hadiths they knew were unreliable because of the overpowering charisma and utility of words phrased in the prophetic idiom, turning to a variety of euphemistic devices to reconcile this practice with their stated commitment to textual authenticity.
The term ahl al-hadith, which I translate as Partisans of Hadith, is certainty mercurial. The extent to which the term was actually used by those scholars later considered to belong to that school, and the extent to which that school represented a consistent approach to Islamic law and dogma, are both serious questions. As Melchert, Spectorsky, and Lucas have shown, there was real diversity within the "traditionist jurisprudent" school. (5) Diversity, however, does not preclude overarching commonality, and discussing this school by one name or another is necessary for any coherent discussion of thought in the early Islamic period. In this article, I treat as falling under the general ahl al-hadith umbrella those scholars who prioritized the derivation of norms from texts (nusus) above consistency in legal analogy, selecting these proof texts through the emerging science of hadith criticism (jarh wa-ta'dil). This article will include al-Shafi'i for example, in the arena of the Partisans of Hadith due to his close and symbiotic scholarly relationship with prominent traditionist jurists like Ibn Hanbal and the influence of his works on other prominent Partisans of Hadith (including his landmark work in developing hadith criticism). (6) Of course, Ibn Hanbal and al-Shafi'i differed in their legal thought, and I shall certainly distinguish between them. But here I contend that they shared comparable views on the epistemological and historical value of hadiths.
THE PROBLEMATIC PRISM OP LEGAL THEORISTS
We risk two pitfalls when thinking about how the early Sunnis viewed the historical reliability of hadiths. On the one hand, we might assume a common-sense approach to historical reports (a pedestrian one not complicated by historiographical debate), which seeks a clear "yes or no" answer to the question "did an event happen or not?" On the other hand, we risk misunderstanding the framework of classical Sunni epistemology, which has often been understood to consign almost all discussion of historical reports to a fog of probability.
As laid out by Wael Hallaq and others, classical Muslim legal theorists held that the vast preponderance of the hadith corpus consisted of ahad hadiths, namely, reports transmitted by a limited number of chains of transmission. Even when transmitted by sahih isnads, these hadiths were therefore only probably authentic statements of the Prophet, according to the epistemological worldview of these legal theorists. (7) To achieve certainty about the authenticity of any report emanating from the past, legal theorists required massive corroboration (tawatur) seldom if ever attained in the hadith tradition.
Hallaq and others have noted that "[c]ertainty concerning the details of human behavior was considered unattainable" by Muslim scholars. (8) This admission of ambiguity informed the legal theorists' approaches to the epistemological rating of historical reports and facilitated the remarkably diverse range of opinions in Islamic substantive law. In reading Hallaq's argument, however, we must remember that these Muslim legal theorists of the fifth/eleventh century were speaking in the language of pre-Islamic, Near Eastern epistemology. (9) Although it is not clear if the Islamicate epistemological bifurcation of certainty ('ilm, qat', yaqin) and probability (zann) was directly derived from Near Eastern antecedents, the Sunni tradition did inherit the heritage (and abiding disputes) of Aristotelian thought, Stoic logic, and the skepticism of the early Platonic Academy. (10)
While these Hellenistic traditions did not share a uniform division of certainty and probability, they did consider what we today would call epistemological certainty to be a rarity. In his dialogue on the nature of the gods, Cicero replies to the criticism of frustrated fellow Romans of the leisure class that Academic philosophers obnoxiously refuse to regard anything as certain. Cicero explains that even true perceptions often contain an element of falsehood that we are unable to detect. "It follows," he says, "that we can attain only to a number of probable truths, which although they cannot be proved as certainties, yet may appear so clear and convincing that a wise man may well adopt them as a rule of life." (11) Detractors of the skeptical Academic tradition had not understood that, in the post-Aristotelian philosophical world, epistemological certainty was a rara avis in everyday life and not at all to be expected. (12)
In the universe of Aristotelian proofs, certainty was knowledge "that could not be otherwise." (13) It was the product of demonstration (apodeixis), a deduction based on premises that are certain and thus producing conclusions that are certain. Beyond this narrow scope of discourse, however, humans subsist in a realm of probability: knowledge of things that are "for the most part" (eikos), that could be otherwise. Dialectical argumentation uses the same mechanics as demonstration (the syllogism and its various forms), but its premises and thus its results are only probable. In rhetoric, both the examples that the speaker employs and the deductions that he makes on the basis of implicit premises shared by the audience (an enthymeme) generally rely on "things that can for the most part be otherwise" and are thus seldom certain. (14) The daily lives and discourses of scholars and laymen, whether in speeches or debate, are built on and most often limited to probability. This is equally true for knowledge of the past and the investigation of history.
In our own discourse we must also remember that epistemological certainty is not the "certainty" that we mean when we speak about "historical reliability" or "certainty" in our daily lives. Certainty in our daily discourse is the "common sense" certainty of Thomas Reid and the commanding "probability" of Hume, not the epistemological certainty of Descartes. (15) In discussions of past events, as Voltaire declared, "historical truths are nothing but probabilities." (16) That does not, however, mean that we treat all knowledge of the past as merely probable in the conventional sense of the word (e.g., "It is probable that I will go to the store").
It is on this point that the work of Hallaq and others on the epistemology of Muslim legal theorists can be misleading. It might lead us into thinking that the epistemological probability with which Muslim scholars viewed ahad hadiths meant that they believed that these hadiths were only "probably" true in our conventional sense of the word, and that they harbored effective doubts about the reliability of these hadiths. If this were the case, however, and Muslim scholars all believed that the hadiths that they cited in their writings were never more than "probably" the words of Muhammad, why would they find such evidence compelling in their discussion of law and dogma?
Certainly, Hallaq emphasizes the consensus of Sunni scholars that these "probabilistic" ahad hadiths were considered probable enough to carry probative weight in discussions of law and even ritual. (17) But at an intuitive level, how could members of a scholarly community consistently argue, debate, and convince one another with a type of evidence if they all treated it as only being probably true? (18) As Robert Summers puts forth clearly in the case of Anglo-American law, key to the legitimacy of any legal system is the people's belief that it in general corresponds with factual truth, both in its promised results and evidentiary standards. (19) Furthermore, why...