Between Qum and Qayrawān: Unearthing early Shii ḥadı̄th sources

Abstract In this article, I develop and test a new methodology of unearthing early Shii ḥadı̄th sources that served as the basis for the later collections of the fourth/tenth century. This method, besides answering the question of historicity, enables us to understand the dissemination of texts across times and regions. As a case-study, I examine what is alleged to have been the first Shii legal ḥadı̄th collection, a work attributed to ʿUbaydullāh b. ʿAlī al-Ḥalabī (d. c. 148/765). By comparing the reports transmitted on the authority of al-Ḥalabī in the Twelver ḥadı̄th compendium originating in Qum, al-Kulaynī's al-Kāfī, and an Ismaili legal ḥadı̄th composition, al-Qāḍī al-Nuʿmān's al-Īḍāḥ, composed in Qayrawān, I demonstrate that both works trace their material to an earlier Kūfan source of the second/eighth century, with each work drawing on the same material independently. A cross-regional textual analysis of later ḥadı̄th compendia, in this case composed by contemporaneous scholars, residing in different regions, affiliated to dissimilar religious persuasions, reveals the transmission of identical material; this finding contributes to our understanding of both geographical transmission of early sources and compositional arrangements of the later ḥadı̄th compendia.


Introduction
In academic research to date, limited attention has been paid to the origins and development of Shii ḥadī th. It stands distinct from its Sunni counterpart in three different respects: theological extension, geographical location, and mode of transmission. 1 Regarding theological extension, the cosmic role assigned to the Imams in Shii theology facilitated the theorization of their religious authority; and hence the reports that recorded their sayings and practices also qualified as ḥadī th. The reports attributed to the Imams, Shiites assert, enjoy the same standing and force as those that have been attributed to the Prophet, 2 though the former, as a mark of distinction, is occasionally referred to as akhbār (reports). 3 In terms of geography, the Imams continued living in Medina but their followership largely consisted of Kūfans. The Shii ḥadī th literature, therefore, in case of its rightful attribution to the Imams, is a confluence of Medinese legal tradition, in which the Imams participated, and Kūfan legal thought, in which their companions operated. The transmission of material in early Shii ḥadı̄th collections is believed to have been predominantly through written records: a ḥadı̄th's isnād, therefore, reflects, at least at some stage, the transmission chain of a written document. 4  For the purposes of this article, I have classified Shii ḥadı̄th collections into "early" and "later" sources. By the former, I refer to those collections which are believed to have been composed during the times of the Imams, i.e. before the end of lesser occultation (260/874). The collections of this period include, but are not limited to, uṣūl (foundational collections), jawāmiʿ (comprehensive collections), nawādir (anthologies of miscellaneous reports), muṣannafāt (thematically arranged collections), mubawwabs (topically arranged collections), among others. A clear distinction between these genres is yet to be made, for often they are used inconsistently and interchangeably, referring to early Shii ḥadı̄th or ḥadı̄th-based works. It is evident, however, that not all of these sources enjoyed the same status as uṣūl in serving as the primary source for the early ḥadı̄th material. They are all grouped together, it should be noted, because they share the characteristic of being composed before the end of lesser occultation. By later, I refer to the larger organized collections of the fourth/tenth and fifth/eleventh centuries. These collections, in the Twelver context, collectively came to be known as al-uṣūl al-arbaʿa as key sources and been absorbed by the more developed, refined, elaborate and thematically arranged larger collections, fell into disuse or were lost. Unearthing these early sources and examining their relationship with the later ḥadı̄th compendia is the primary concern of my study.
Given the absence of contemporaneous, consistent, and independent early sources, it is highly unlikely that we can be certain of whether there was a real historical referent (i.e. whether the reports recorded actual events) for the material found in the later ḥadī th compendia. However, a credible layer and historical kernel of early ḥadī th material preserved in the later Shii ḥadı̄th collections of the fourth/tenth century can, I propose, still be uncovered. Al-Īḍāh and al-Kāfī could contribute significantly to our understanding of that early material, but an appropriate methodology with which to analyse them has not yet been devised. Here, I argue that a cross-regional textual analysis of these two earliest surviving larger ḥadı̄th collections unearths a layer of early sources accessed by both the authors independently of each other. Cross-regional textual analysis entails conducting a comparative study of a set of reports preserved in the later collections composed in two distant geographical locations by contemporaneous authors adhering to distinct religious persuasions. As a case study, I will cross-examine the reports of ʿUbaydullāh b. ʿAlī al-Ḥalabī cited in al-Īḍāh and al-Kāfī in a quest to unearth its earliest layers. Such cross-regional textual analysis offers exciting possibilities for tracing the origins and dissemination of early texts across times and regions. For the purpose of the present article, I will analyse the result of this investigation in three areas: historicity of al-Ḥalabī's collection; its incorporation into al-Īḍāh and al-Kāfī and their compositional arrangements; and the question of its authorship.
The geography and geographical transmission of early sources lie at the forefront of this project, for they contribute to examining not only the origins of early sources but also their amalgamation and absorption in the later larger collections. Haider and Sadeghi have argued for the consideration of regionalism and geographical associations of the transmitters in the study of ḥadī th transmission. Sadeghi highlights the importance of geographic clustering of narratives, vocabulary, syntactic structures and legal positions in dating ḥadı̄th. 5 Haider, on the other hand, concentrates on identifying the regional associations of the transmitters of reports in order to reconstruct, and thereby date, the religious practices of a specific region. 6  the contribution of regionalism to our understanding of early ḥadı̄th material and therefore remain extremely pertinent to my project because of its direct concern with the geographical movements of early Shii ḥadī th texts. My study, however, employs regionalism to examine the dissemination of early sources across regions and how it informs our understanding of the historicity of early sources, on the one hand, and their reception and treatment, on the other, in the regions in which they travelled. My study demonstrates the usefulness of cross-regional textual analysis in four ways. First, it independently attests to the historicity of the titles, otherwise thought to have become extinct, recorded in the Twelver bio-bibliographical works of the fifth/eleventh century. 7 Second, it enhances the credibility of the reports incorporated in the later, larger, thematically arranged ḥadī th compendia. Their contemporaneous compilation coupled with the authors' geographical distance renders any possibility of collusion or forging of material highly unlikely. Third, it identifies the trajectory, travel history, and transmission network of the early sources. Fourth, it offers the opportunity to examine the intellectual connections not only between two later ḥadī th compendia but also between them and their shared sources: what dictated their choices, arrangements, and adjustments in their respective collections? In this respect, my conclusions are in broad agreement with those of Motzki and Schoeler in relation to the Sunni ḥadī th corpus: that is, the bulk of ḥadī th material (including forgeries) has a history before the surviving works, and earlier credible layers of material can be excavated from the later, fourth/tenth century, ḥadī th collections. 8 In search of early Shii ḥadı̄th sources: approaches and methodologies The pioneering studies of Goldziher and Schacht concerning the historicity of Muslim tradition shaped the academic discourse on the dating and attribution of ḥadī th works throughout the twentieth century. 9 The next generation of scholars in relation to the credibility of the corpus of ḥadī th were found at two ends of a spectrum: "sceptical" to "sanguine", or "revisionist" to "traditionist". 10 Their fundamental concerns were: is dating and reconstructing Islamic traditions possible? Is an isnād (chain of transmission) a useful tool for the dating of early sources? Should the isnāds be trusted as reliable documentary evidence? And can the text (matn) and its stylistic structure help us determine its earliest date of circulation? In order to engage critically with these questions, several methodological approaches were designed to examine the historicity of the Muslim traditions. Motzki has summarized them into four major approaches: isnād criticism, matn criticism, isnād cum matn/ matn cum isnād analysis, and examining the dating of the collections where traditions appear. 11 Shii ḥadı̄th, it should be noted, does not necessarily face the challenges posed to Sunni ḥadī th tradition, nor is it an ideal ground for testing the approaches designed to investigate the historicity of the latter. This is because it comes predominantly from Imams al-Bāqir (d. 114/733) and al-Ṣādiq (d. 148/765) who belonged to an intellectual milieu which had just witnessed the emergence of the written transmission of ḥadī th. In other words, contrary to Sunni ḥadī th, which purports to extend back to the time of the Prophet or Companions, the bulk of Shii ḥadī th is a production of the first half of the second/eighth century. This feature, along with other peculiarities of Shii ḥadī th tradition discussed in the introduction, I argue, demands a completely different approach that could address the issues with which it has historically grappled.
Modarressi's Tradition and Survival is by far the most extensive study on Shii literary activities of the first two centuries of Islam. His laborious work neatly fits into the larger project of reconstructing early Islamic works initiated by Abbott, Aʿẓamī, and Sezgin, sharing precisely the same concerns, addressing exactly the same questions, and using a similar methodological approach. 12 Modarressi's scholarship centres around the idea that the earliest sources of ḥadī th were recorded re-ordered the sequence. Motzki argues that the method of isnād cum matn/matn cum isnād analysis is more reliable than the other approaches which are either "inaccurate" or "less sound". He acknowledges that his method is a "revival" of the project initiated by Jan Hendrik Kramers and Joseph van Ess (see Motzki, "Dating Muslim traditions", 250 and the sources cited there). It should be noted that the method of isnād cum matn analysis was reconstructed by both Motzki and Schoeler independently of each other at about the same time. This is based on the latter's self-assertion in Gregor Schoeler, in writings and were accessible to the fourth/tenth-century scholars of Qum and Baghdad who faithfully incorporated them into their larger collections after extracting and classifying their material into thematically arranged chapters. This seemingly organic development is believed to have been so smooth that the early ḥadī th corpus, with a careful deconstruction of isnāds, could possibly be reconstructed. These isnāds, Modarressi posits, "predominantly represented authors' chains of transmission to those earlier records rather than oral transmission of individual quotations". 13 To ascertain whether a later collection has drawn its material from earlier written sources, he proposes cross-verifying the isnāds of the ḥadı̄th with the transmission lines of books recorded in bio-bibliographical dictionaries.
In his view, the correspondence between both the chains (i.e. chains of report transmission and chains of book transmission) helps us "ascertain whether a later work quotes directly from an earlier source". 14 For instance, al-Kulaynī (d. 329/941) cites a total of 504 reports on the authority of Ismāʿīl b. Abī Ziyād al-Sakūnī with the following recurring isnād: The isnāds of al-Kāfī and the transmission lines of al-Najāshī and al-Ṭūsī illustrate that Ibrāhīm b. Hāshim is the common link responsible for the transmission of al-Sakūnī's collection(s). 19 The correspondence between al-Kulaynī's isnāds to al-Sakūnī's reports and al-Najāshī and al-Ṭūsī's transmission lines to al-Sakūnī's collection(s) indicates that al-Kāfī's citations most probably originate from al-Sakūnī's collection(s). Modarressi maintains that his method is based on "concrete evidence" that takes the data of bio-bibliographical dictionariesa resource not available to Sunni authorsinto account to examine the origins of the early sources. 20 This approach faces several methodological challenges. First, it presupposes that the isnāds recounted in bio-bibliographical dictionaries are independent attestations for the genuine transmission of a book and they have not been lifted from the isnāds of the reports to advance the idea that the transmission of ḥadı̄th has taken place through the medium of writing. The biographical dictionaries reveal that duplicating and synthesizing isnāds were not uncommon practices. Ibn Buṭṭa (d. c. 330/942), for instance, is accused of blending chains of individual reports into transmission lines of books (kāna . . . yuʿalliqu al-asānīd bi al-ijāzāt) in an attempt to demonstrate that Shii ḥadī th is, essentially, transmitted through the medium of writing. 21 Second, the sceptics consider isnāds to be the most vulnerable component of a ḥadī th. But even a sound isnād does not necessarily indicate the veracity of a ḥadı̄th, for it is quite possible that an astute forger will deploy a sound isnād for a bogus text. 22 The same applies to bio-bibliographical transmission lines: they are not immune to the challenges posed to isnāds. Though some recent studies have carefully reconstructed the sources of existing bio-bibliographical dictionaries by tracing the citations supposed to have been preserved in the latter, it is evident that, methodologically, such reconstructions, until supported by independent attestations, do not contribute to investigating the historicity of the sources in question. 23 Third, the approach of cross-referencing isnāds does not take into account the redactions of any specific early collection that is not listed in the bio-bibliographical dictionaries. The book of Ḥarīz, for instance, is reported to have been transmitted by Ḥammād b. ʿĪsā, as illustrated by the three isnāds of al-Ṭūsī and two isnāds of al-Najāshī. 24 bio-bibliographical dictionaries do not speak about another possible recension of Ḥarīz's book, that which is transmitted by Yāsīn al-Ḍarīr and was accessible to al-Kulaynī through his teachers. 25 Fourth, this approach takes as its starting point the idea that Shii ḥadī th were transmitted through the medium of writing and thereby engages in what Stewart calls "educated guesswork" by assigning ḥadī th that may have been transmitted orally to certain works that match its content. 26 Another approach that also attempts to trace the origins of early Shii ḥadī th sources through the existing pool of literature is increasingly coming to be known as "bio-bibliographical analysis (taḥlīl-i fihristī)". Al-Sayyid Aḥmad al-Madadī al-Musawī (b. 1951), a leading scholar of the Shiite seminary of Qum and the chief advocate of this approach, postulates that Shii ḥadı̄th, since its very early stages, has been transmitted through the medium of writing. The early writings were then fully incorporated in the later larger collections. He shares this premise with Modarressi. The process of "authentication" of Shii ḥadı̄th, therefore, requires, he adds, bio-bibliographical analysis of the isnāds that identifies the source from which a set of reports has been transmitted rather than the conventional approach of biographical (rijālī) assessment which evaluates the trustworthiness of individual transmitters. The bio-bibliographical analysis will result, Madadī argues, in mass authentication of the reports if: (a) the source text is identified; (b) its attribution to an early author is established; and (c) its faithful transmission to the next generation of scholars is ascertained. The primary aim of this approach, it emerges, is to establish the authoritativeness (ḥujjiya) of the early ḥadī th sources, as opposed to evaluating individual isolated reports, and hence serves the legal, not historical, interest of a jurist. In other words, the supposed beneficiary of this analysis is fiqh and not the study of history. Though Madadī's approach appears to be more rigorous, as it entails several layers of biographical and bio-bibliographical examination, how it substantially differs from Modarressi's method is an open question that merits further investigation. 27 Due to the methodological challenges such isnād-based reconstruction projects face, I propose an alternative approach that undertakes the task of identifying independent attestation as to the genuine existence of the sources in question. This is ascertained through "cross-regional textual analysis"from where it derives its nameof the later larger ḥadı̄th collections.

Cross-regional textual analysis
At the outset, it should be made clear that my approach, like Modarressi's, works on the basis that the Shii ḥadī th corpus was, in the main, recorded and transmitted through writing rather than orally. 28 The early rudimentary collections of the second/eighth century furnished the collectors of the later thematically arranged works (muṣannafāt) with some first-hand written sources. Building on this premise, my proposed method attempts to trace the trajectory of those early sources through a rigorous cross-regional textual analysis of the later works that have drawn their material from them.
The following three considerations form the nucleus of this method: 1. The later collections under analysis should, for optimum results, be contemporaneous. This is particularly important because, if they were not contemporaneous, the possibility of direct access to an early source by the later of the two non-contemporaneous secondary collections could always be contested. That is, it is possible (perhaps likely), that the citations in the later collection are simply drawn from those in the earlier one. Consider the case of Ibn Bābawayh (d. 381/991) who had access to the ḥadī th compendia of al-Kulaynī. One cannot reject the possibility that Ibn Bābawayh relied (if in only few instances) on the citations of al-Kulaynī, rather than citing the original source directly. 29 2. The later collections should have been compiled in different regions to eliminate the possibility of them having consulted the same (physical) copy of the text; if they consulted the same copy, then they would fail to provide independent attestation as to the original text's existence. 3. The force of the conclusion is augmented by entertaining a third supportive consideration: the religious persuasions of the authors. The religious affiliation of the author with a particular set of doctrines involves, it is assumed, accepting or rejecting texts that are rejected or accepted (respectively) by their opponents. In cases when both parties preserve and cite an identical text without any distortion or interpolation, the chances of it being forged are substantially reduced.
The case study presented in this article will demonstrate that a cross-regional textual analysis of the later thematically arranged collections which factors in the aforementioned three considerations helps us determine the historicity of early sources. In addition to investigating the historicity of early sources, it enhances our understanding of the intellectual connections and the emerging traditions as they developed and spread out in different regions. Cross-regional textual analysis also underscores the importance of geography in the transmission of knowledge and how, textually, we can uncover geographical distributions and functions of early Shii literary activities. In what follows, I test this method to unearth the earliest layers of ʿUbaydullāh b. ʿAlī al-Ḥalabī's collection and demonstrate the ways in which it deepens our understanding of this early, arguably earliest, Shii ḥadī th source: its historicity, travel history and isnād networks; its incorporation into later larger ḥadī th collections and the latter's compositional arrangements; and its authorship. among rather distant Shii communities. 32 In a similar correspondence, the Shii community of Rass is reported to have sought al-Murtaḍā's opinion on whether, given their inability to deduce law, it was appropriate for them to consult, for their religious practices, a "foundational text (kitāb aṣl) such as Kitāb al-Ḥalabī ". 33 These exchanges highlight the wider appeal of al-Ḥalabī's work; they also indicate that it continued being copied and circulated in the fifth/eleventh century, especially considering the fact that other, similar, early sources had ceased to exist by this period. Ibn Ṭāwūs (d. 664/1265), based on the references made in two of his works, is arguably the last Shii scholar believed to have had access to al-Ḥalabī's collection. 34 The juxtaposition of Kitāb al-Ḥalabī with the likes of some mature and established compositions of the fourth/tenth and fifth/eleventh centuries such as Risālat al-muqniʿa, Risālat Ibn Bābawayh, Kitāb Shalmaghānī, and al-Kāf ī illustrates its extensive popularity despite the fact that its content had already been subsumed by the very texts with which it was equated. It is worth noting that Kitāb al-Ḥalabī is introduced as a kitāb aṣl (source text) vis à vis Risālat al-muqniʿa and Risālat Ibn Bābawayh, both characterized as kitāb muṣannaf (composition, usually a legal composition), and al-Kāf ī, characterized as kitāb riwāya (ḥadī th collection). Al-Najāshī referred to al-Ḥalabī's text as al-kitāb al-mansūb ilayhi (a book attributed to al-Ḥalabī), whereas al-Ṭūsī introduced it as kitāb muṣannaf maʿmūl ʿalayhi (a composition that is widely used). 35 Al-Qāḍī al-Nuʿmān, on the other hand, consistently cites the work with the title Jāmiʿ al-Ḥalabī ("al-Ḥalabī's collection"). It appears that al-Ḥalabī's work did not bear any specific title and, therefore, different scholars assigned different titles, mainly in adjectival form, based on its early origins (aṣl), thematic arrangement (muṣannaf) and comprehensiveness ( jāmiʿ). In reference to the content of the book, it is difficult to ascertain what exactly it entailed, but based on the citations recorded in the later collections, it can be assumed that the work contained legal issues in the form of ḥadı̄th related on the authority of al-Ṣādiq. In other words, it appears to be a legal ḥadī th collection rather than a treatise of fiqh or a handbook of legal opinions.
The paucity of sources does not allow us to determine whether al-Ḥalabī composed this work in Medina where his Imam lived, or whether it was a result of his interaction with the latter during his sojourn in Kūfa. Nonetheless, based on the multiple isnāds illustrating the networks through which al-Ḥalabī's collection was disseminated, it is safe to conclude that it was Kūfa, typical of any early Shii work, from where the book made its way to Qum, Baghdād, Silvan, Rass, and Ḥilla. The reports transmitted on the authority of al-Ḥalabī were also known to North African Ismaili dāʿīs in Qayrawān. Reporting the distinguished status of his teacher and the extent of his scholarly activities, the senior Ismaili dāʾī Ibn al-Haytham (b. c. 273-77/886-87) reports: And whatever I may forget, I shall never forget the dāʿī of Malūsa, the shaykh of the community and their legal authority, Aflaḥ b. Hārūn al-ʿIbānī. He combined his activity as a dāʿī with the sciences of the religious law, and he reached back to the time of Abū Maʿshar and al-Ḥulwānī and transmitted on their authority from al-Ḥalabī. 36 Notwithstanding the anecdotal nature of this report, it offers a valuable piece of information about the accessibility of al-Ḥalabī's collection to the late third/ninth and early fourth/tenth-century Ismaili dāʿīs in North Africa. Given the fact that al-Qāḍī al-Nuʿmān (d. 363/974) was not only a contemporary of Ibn al-Haytham but also a junior colleague in charge of the Fatimid collections, in his capacity as a librarian between 322-334/934-946, it is conceivable that he also had access to this work.
In the second half of the fourth/tenth century, Abū Ghālib al-Zurārī (d. 368/ 978) reports that his family collection contained Kitāb ʿUbaydillāh b. ʿAlī al-Ḥalabī. 37 It was also known to Ibn al-Nadīm (d. 385/990) who lists it among the popular legal works of Shiite scholars. 38 Its popularity in Qum can be gauged by Ibn Bābawayh's (d. 380/991) reception of it from three of his teachers. The collection continued to receive attention in Baghdad in the fifth/eleventh century. Al-Najāshī states that he had several isnāds for the transmission of this collection but, restricted by his commitment to brevity, he offers only one isnād. In contrast, al-Ṭūsī listed all four of his transmission lines. 39 Careful scrutiny of these extensive bundles of isnāds reveals that they all converge at a single common link, i.e. Ḥammād b. ʿUthmān (d. 190/806). Since the collection did not survive the vagaries of time, one has to trace its content and reconstruct it through cross-regional textual analysis of the later ḥadī th collections.

Al-Ḥalabī's collection in al-Īḍāḥ
The nisba al-Ḥalabī appears 103 times in the extant fragment of al-Īḍāḥ. It draws reports from two titles ascribed to al-Ḥalabī: Jāmiʿ al-Ḥalabī and Kitāb  al-Ḥalabī al-maʿrūf bi al-masāʾil (henceforth Kitāb al-masāʾil). 41 Diverting from his method of quoting complete isnāds for the sources that he cited, al-Qāḍī al-Nuʿmān does not follow the same practice for these two titles, nor does he provide the full name of their author(s). Jāmiʿ al-Ḥalabī is cited 51 times in al-Īḍāḥ whereas Kitāb al-masāʾil is quoted in 52 instances (Table 1).

Al-Ḥalabī's collection in al-Kāfī
Contrary to al-Qāḍī al-Nuʿmān and his own fellow Twelver traditionists, al-Kulaynī does not cite his sources, but rather adopts the style, prevalent in Sunni ḥadī th tradition, of rendering complete isnād for every single report. Given this limitation, it is difficult to ascertain whether al-Kulaynī had direct access to al-Ḥalabī's collection. Nonetheless, its content, judging from a significant number of reports cited on the authority of al-Ḥalabī, appears to have been available to him in Qum. The statistical data obtained by examining major Shii ḥadı̄th compendia, collectively known as "the that the latter remains the most important source for unearthing the earliest layers of al-Ḥalabī's collection. 45

Analysis
Having introduced the isnād networks and travel history of al-Ḥalabī's collection, I now turn to conduct a cross-regional textual analysis of its reports cited in al-Īḍāḥ and al-Kāfī. Such an analysis, I will illustrate, contributes to our understanding of both its early origins and later dissemination in Qum and Qayrawān. It not only allows us to unearth the earliest layers of al-Ḥalabī's collection, but also helps us gain insight into how its content was received, processed, and arranged in the later larger ḥadī th compendia. In what follows, I demonstrate the utility of this analysis in three areas: historicity of al-Ḥalabī's collection; its incorporation into al-Īḍāh and al-Kāfī and their compositional arrangements; and the question of its authorship.

Historicity
This part may be misread as an attempt to establish the authenticity of early Shii sources on which the later collections relied. This is not the objective of my study. The process of authentication requires the availability of various The reader should not assume that al-Ḥalabī's reports were transmitted via a non-al-Ḥalabī route by Ibn Bābawayh. This is due to the author's convention of citing isnād. The recurring isnāds are cited not in the body of the text, but rather in a dedicated section appended to the book. 48 See n. 47. The same is partially true in respect to Tahdhīb al-aḥkām and al-Istibṣār mentioned in the third and fourth rows of Table 2.
redactions of an early source reaching back to the author (who may not even be the individual to whom the reports are attributed): these redactions are not immediately available to researchers. The application of cross-regional textual analysis only suggests that it is safe to assume that the sources of the later collections could be traced historically at least one generation earlier, if not more.
Below I attempt to unearth al-Ḥalabī's collection by cross-examining its citations recorded in al-Kāfī and al-Īḍāḥ that meet all three aforementioned conditions. First, though al-Qāḍī al-Nuʿmān outlived al-Kulaynī by more than four decades, al-Īḍāḥ, his first legal work, is believed to have been composed at the very beginning of his scholarly career, between 315-320/927-932, a period that roughly coincides with al-Kāfī's compilation. 49 Second, as regards the geographical locations of their authors, al-Īḍāḥ was composed in Qayrawān, whereas al-Kāfī, judging based on the authorities from whom al-Kulaynī transmitted most of his reports, was compiled in Qum. 50 Third, al-Qāḍī al-Nuʿmān offered his services to Fatimid Imam-Caliphs under whose patronage he composed al-Īḍāḥ, whereas al-Kulaynī was raised, trained, and studied in the Twelver intellectual milieus of Rayy, Qum, and Baghdad. Though the early Shii ḥadī th is justifiably considered a shared legacy of both Ismailis and Twelvers (they do, after all, share the same lines of Imams from ʿAlī to al-Ṣādiq), the possibility of differences in selection, arrangement, and interpretation of the reports should not be underestimated. The fulfilment of these three conditions, I argue, advances my hypothesis that the sources of the later, larger ḥadı̄th collections date back at least a generation earlier, if not more. Table 3 14  3  3  3  21  7   Total  23  5  3  6  53  13 52 By obscurity I mean these 13 reports do not seem to offer verbatim citations of ḥadı̄th, but rather resemble edicts or editorial statements of the author.
recurring isnād leading to al-Ḥalabī. These self-same citations suggest that al-Ḥalabī's collection(s) existed some generations earlier than al-Kulaynī and al-Qāḍī al-Nuʿmān. The most likely explanation for its provenance and early circulation is that the work should have been compiled in Kūfa in the second/eighth century before it was transmitted to Qum via transmitters such as Ibrāhīm b. Hāshim 53 and to Qayrawān via early Ismaili dāʿīs. It was then incorporated and absorbed in the larger ḥadī th collections compiled in these regions.
Reading al-Īḍāḥ gives the impression that al-Qāḍī al-Nuʿmān had numerous early sources of ḥadī th at his disposal. Given his role as a librarian of the Fatimid khizānat al-kutub (library, lit. treasure house of books), it is conceivable that he had access to a redaction of al-Ḥalabī's collection that was present in North Africa, perhaps through Ibn al-Haytham and Aflaḥ b. Hārūn al-ʿIbānī, the dāʿī of Malūsa. 54 On the other hand, reading al-Kāfī gives the impression that al-Kulaynī, albeit through mediation of other intermediatory sources as will be demonstrated below, had access to the reports transmitted on the authority of al-Ḥalabī. Bearing in mind the aforementioned three considerations, we also know that their access to this early source should have been independent of each other. There seems no plausible explanation for the concurrence of these identical renditions except that both al-Kulaynī and al-Qāḍī al-Nuʿmān had access to works that contained the earliest layers of the source in question. These 23 reports can then be seen as representing that earliest layer of al-Ḥalabī's collection.

Composition and compositional arrangements
Reading al-Kāfī alongside al-Īḍāḥ enables us not only to investigate the historicity of their shared sourcesin our case al-Ḥalabī's collection-but also to analyse their own composition and compositional arrangements. The claim that the fourth/tenth-century Shii ḥadī th collections were composed directly from the early sources of the mid-second/eighth century is untenable. There were a number of intermediatory texts compiled between them. These texts, it is reported, were larger and more structured compared to their predecessors but not as large or thematically organized as their successors of the fourth/tenth century. . See al-Najāshī, Rījal, 58-60. It is unclear, though, whether these were independent books or, simply, chapters of a single large collection. It should also be noted that these texts were occasionally referred to as uṣūl in its broader sense, i.e. works that were composed during the time of the Imams. See Muḥammad b. ʿAlī b. Shahrāshūb, Kitāb maʿālim al-ʿulamāʾ fī fihrist kutub al-Shīʿa wa asmāʾ al-muṣannifīn The sources of al-Kāfī, in reference to al-Ḥalabī's collection, as shown below, were comprised of these intermediatory texts. The second and third columns of Table 3 illustrate a small, yet significant, number of five reports 56 from ʿUbaydullāh al-Ḥalabī and three reports 57 of a certain al-Ḥalabī 58 that are found in other Twelver ḥadī th compendia. The tracing of identical reports in other collections highlights two points: first, al-Ḥalabī's collection also served as a source for ḥadī th compendia of Qum and Baghdad; second, al-Kulaynī, unlike al-Qāḍī al-Nuʿmān, did not fully extract all the reports of al-Ḥalabī's collection in al-Kāfī but rather appears to have relied on other sources for similar content. Why did al-Kulaynī choose to quote similar content from a different authority despite the distinguished status of al-Ḥalabī and the unmatched reputation of his collection? Do we know if al-Ḥalabī's collection was accessible to him and his contemporaries in Qum? What does that tell us about the sources of al-Kāfī?
The answers to these questions might lie in al-Kulaynī's reliance on nawādir works for the compilation of al-Kāfī. These sources are presumed to have incorporated the content of early foundational collections (uṣūl) without proper thematic organization (hence the name nawādir). Unlike Ibn Bābawayh and al-Ṭūsī, two distinguished members of the scholarly networks of Qum and Baghdad respectively, al-Kulaynī was an "outsider". He hailed from Rayy, studied in Qum and taught in Baghdad where he resided towards the end of his life. It is, therefore, quite conceivable that he might not have had direct access to the uṣūl that were available to more well-established Qummī scholars. 59 A cursory glance at the isnāds of al-Kāf ī reveals that his sources were primarily nawādir (anthologies of miscellaneous reports) and muṣannafāt (thematically arranged collections) composed by third/ ninth-century Qummī scholars. It is no exaggeration that at least half of al-Kāf ī's reports, and probably more, are based on three sources: al-Nawādir of Ibrāhīm b. Hāshim (d. Al-Kulaynī's reliance on an unusually broad range of these nawādir partly explains why al-Kāfī is stylistically different from the other three Twelver ḥadī th compendia, namely al-Faqīh, Tahdhīb al-aḥkām, and al-Istibṣār.
The reconstruction of al-Ḥalabī's collection from the reports of al-Kāfī is, then, based on the grounds that al-Kulaynī's access to it should have been via intermediary sources, i.e. nawādir compiled by his Qummī predecessors. This can also be gleaned from Table 2 which demonstrates that Ibn Abī ʿUmayr relates from Ḥammād a total of 523 reports, 446 of which contain a recurring chain of transmission: Ibn Abī ʿUmayr → Ḥammād → al-Ḥalabī. One can surmise, invoking Modarressi's hypothesis, that al-Kulaynī's citations are based, albeit through his nawādir sources, on al-Ḥalabī's collection. The isnāds of al-Kāfī indicate that the author had access to al-Ḥalabī's reports through the following three chains: There is not sufficient internal or external evidence to support the claim that al-Kulaynī had direct access to early uṣūl of the mid-second/eighth century. I argue that it is due, rather, to his use of nawādir works that the reports of a single aṣl is transmitted via different isnāds in al-Kāfī. For instance, both the Nawādir of Ibrāhīm b. Hāshim and the Nawādir of Aḥmad b. Muḥammad b. ʿĪsā extracted reports from al-Ḥalabī's collection. Naturally, the isnāds of al-Kāfī, in respect to citing the reports attributed to al-Ḥalabī, will differ depending on the nawādir al-Kulaynī chose to extract a particular report. Notwithstanding this disintegration, the force of my conclusion remains intact, for nawādir works are seen as intermediatory works between early uṣūl works and later ḥadı̄th compendia. In other words, had nawādir survived, their reports would have directly attested against the citations of al-Īḍāḥ. In their absence, our second-best choice is their successor: al-Kāfī.
Verbatim citations of a significant number of reports, 53 in total, could not be traced in al-Kāfī or any other Twelver ḥadī th collection. The legal opinions described in these reports, however, are traced, though they are attributed to other Imams and worded differently. Simply put, the reports cited in al-Īḍāḥ, in these instances, are not alien to Shii legal thought. Whilst these reports may not prove helpful in reconstructing al-Ḥalabī's collection, they do help us gain a better understanding of the authors' selection processes. Their differences, then, could be explained by taking into account the fact that the epicentres of Twelver Shii ḥadı̄th in the fourth/tenth century, Qum and Baghdad, provided Twelver scholars with a host of early Kūfan sources that were not necessarily available to al-Qāḍī al-Nuʿmān in Qayrawān. In these instances, Twelver scholars, it could be argued, opted to cite similar reports from other uṣūl. Until it is corroborated by more substantial evidence, this hypothesis remains, at best, tentative.
A rigorous cross-examination of al-Ḥalabī's reports cited in al-Kāfī and al-Īḍāḥ also reveals that the latter contains far more reports than the former. In the chapter of al-ṣalāt in al-Kāfī, for instance, al-Kulaynī cites only 46 reports from al-Ḥalabī, compared to 104 reports in a rather incomplete portion of the same chapter in the extant fragment of al-Īḍāḥ. 64 This reflects al-Qāḍī al-Nuʿmān's extensive use of al-Ḥalabī's collection, so much so that it is safe to assume that he incorporated all its reports in his voluminous al-Īḍāḥ. On the other hand, one could also argue that al-Qāḍī al-Nuʿmān endeavours to offer an exhaustive list of reports in each section, whereas al-Kulaynī appears to be content with citing a representative example relevant to a given chapter. This comparative analysis that reads al-Kāfī through the lens of al-Īḍāḥ facilitates a new understanding of the former's engagement with intermediatory sources that were obscured or forgotten with the emergence of larger thematically arranged ḥadı̄th compendia.

Authorship
One of the issues that cross-regional textual analysis attempts to address is the question of authorship. Here, I am not interested in assessing the veracity of attribution so much as in examining how the dissemination of texts across regions, their absorption into larger collections, and the intellectual vibrancy of the regions to which they travel result in differences in authorship attribution.
The fourth column of Table 3 illustrates two points: first, six reports attributed to al-Ḥalabī in al-Īḍāḥ are identical with those cited on the authority of Muḥammad al-Ḥalabī in Twelver ḥadī th compendia; second, three of these reports are attributed to Jāmiʿ and the other three to Kitāb al-masāʾil. In reference to Twelver ḥadı̄th compendia, all six reports are exclusively found in Tahdhīb al-aḥkām and al-Istibṣār. Al-Ṭūsī relates these reports on the authority of Muḥammad al-Ḥalabī via ʿAbdullāh b. Muskān. Did al-Ṭūsī mistake ʿUbaydullāh for Muḥammad? Did Baghdadī scholars have access to Muḥammad al-Halabī's collection, which was not available to their Qummī counterparts? Did the two brothers record identical reports in their independent collections, leading to different attributions based on the sources al-Qāḍī al-Nuʿmān and al-Ṭūsī consulted? What do these discrepancies tell us about the authorship of Jāmiʿ al-Ḥalabī and Kitāb al-masāʾil? And how do we make sense of the attributions of identical reports to two different titles in two different regions? These are critical questions with which cross-regional textual analysis attempts to engage.
Madelung, rather reluctantly, proposes that Jāmiʿ al-Ḥalabī and Kitāb al-masāʾil were either variant versions of a single text or two different sections of the Kitāb ascribed to ʿUbaydullāh in Twelver sources. 65 Kitāb al-masāʾil, according to Modarressi, though different in style, was "part of the larger version of [ʿUbaydullāh b. ʿAlī] al-Ḥalabī's Kitāb". 66 The assumption that these two works are different versions or sections of a single larger collection is not supported by the treatment they receive in al-Īḍāḥ. The mention of both titles, in several instances with a conjunction, attests to the fact that al-Qāḍī al-Nuʿmān considered them to be two separate works. 67 Furthermore, contrary to his consistent pattern of using pronouns ( fīhi or fīhā) for the same titles consulted for a previous report, al-Nuʿmān cites the full titles, one after the other, of these two works. 68 Therefore, it is safe to conclude that al-Qāḍī al-Nuʿmān treated them as two separate works. Furthermore, there is no mention of Kitāb al-masāʾil under the entries of ʿUbaydullāh b. ʿAlī al-Ḥalabī in any of the extant Twelver bio-bibliographical dictionaries. Lastly, the styles of the two texts are significantly different: the question-and-answer format of the reports attributed to Kitāb al-masāʾil is not to be found in Jāmiʿ al-Ḥalabī. 69 In reference to their authorship, a closer cross-examination of their content cited in al-Īḍāḥ with that recorded in the Twelver sources suggests that the latter treated them as part of one single collection of Muḥammad al-Halabī. This collection is presumed to be Kitāb mubawwab fī al-ḥalāl wa al-ḥarām as introduced by al-Najāshī in his introduction of Muḥammad al-Halabī. 70 It can then be argued that the supposed work of Muḥammad al-Halabī was known with two separate titles in Qayrawān and Baghdad, namely Kitāb al-masāʿil and Kitāb mubawwab fī al-ḥalāl wa al-ḥarām respectively. One can, thus, surmise that Jāmiʿ was ʿUbaydullāh's work and Kitāb al-masāʿil was his brother Muḥammad's. However, numerous instances of overlapping reports do not allow us to form a conclusive opinion on their authorship, particularly when al-Qāḍī al-Nuʿmān appears to have attributed both collections to a single author. 71 The cross-regional textual analysis of al-Īḍāḥ and Twelver ḥadī th sources thus enables us to engage with questions concerning the authorship of early sources. As shown above, there are clear discrepancies in these attributions. The sources of al-Īḍāḥ, I argue, reflect earlier layers of Kūfan sources than those which can be found in its Qummī and Baghdadī counterparts. The sources of the latter, it is observed, were refined and processed in the then intellectually vibrant Twelver ḥadī th tradition. Their content was debated, selected, and appropriated before it could qualify to be cited in a given collection. Such scholarly engagement also indicates that Qummī and Baghdadī scholars enjoyed access to a greater variety of sources.

Summary
Several inferences may be drawn from Table 3. First, the striking resemblance of more than 25 per cent of the reports cited in al-Īḍāḥ and al-Kāfī, having discounted the obscured reports, evidently suggests the mutual provenance of their sources. If one adds 14 identical reports cited in Twelver ḥadı̄th compendia on the authority of ʿUbaydullāh al-Ḥalabī or other Ḥalabīs to the 23 reports in al-Kāfī, this resemblance occurs in 41 per cent of all surviving reports. Second, whereas al-Qāḍī al-Nuʿmān treated Jāmīʿ al-Ḥalabī and Kitāb al-masāʾil as two different works, al-Kulaynī's isnāds make no distinction between them. The same applies to other Twelver ḥadī th compendia. Third, al-Kulaynī's sources appear to be wider than al-Qāḍī al-Nuʿmān; while only the latter appears to have had direct access to the uṣūl, the former relied on more voluminous intermediatory collections (such as nawādir) compiled by Qummī scholars.
The preceding investigation has produced promising results. The statistical data obtained through cross-regional textual analysis of al-Kāfī and al-Īḍāḥ enabled us to trace the historicity and, to an extent, the contents of an early Kūfan ḥadı̄th source with a fair degree of accuracy. It has enhanced our understanding of the composition and compositional strategies of the later collections besides addressing the question of authorship. My findings, thus, complement Kohlberg's assessment that "a detailed study of [al-Qāḍī] al-Nuʿmān's works" might "shed further light on Shii tradition as a whole". 72 Al-Īḍāḥ's contribution to understanding the dissemination of early texts across times and regions, therefore, remains critical.

Conclusion
The primary focus of this article centred around developing a methodology that could help us investigate the historicity and geographical transmission of early Shii ḥadı̄th sources. Two assumptions formed the basis of my hypothesis. First, Shii ḥadī th, at the behest of Imams al-Bāqir and al-Ṣādiq, originated in the first half of the second/eighth century, which roughly coincides with the Sunni writing of ḥadī th. Second, contrary to Sunni ḥadı̄th tradition, which arguably relied on oral transmission, Shii ḥadī th, since its inception, was transmitted through the medium of writing. I have argued that these two features of Shii ḥadı̄th merit different treatment and that the tools designed to evaluate Sunni ḥadı̄th tradition do not appear to be sufficiently effective.
After having examined the existing literature on the origins, circulation, and methodological challenges of early Shii ḥadī th sources, I proposed a new method that traces the layers of early sources with a higher degree of accuracy. The historicity of a source, this method proposes, is better assessed by conducting a cross-regional textual analysis of the later ḥadī th compendia that purport to have faithfully transmitted its content. I argued that a cross-regional textual analysis of ḥadī th compendia that contain identical material but are composed by contemporaneous authors with distinct religious persuasions in distant locations 72 Kohlberg, "Introduction", 179. indicates the mutual provenance of their sources. My study has demonstrated that such a rigorous analysis, besides enabling us to unearth the earliest layer of Shii ḥadī th sources, helps trace the trajectory of dissemination of texts across times and regions. I have tested this method on al-Kāfī and al-Īḍāḥ in an attempt to excavate the earliest layer of al-Ḥalabī's collection, arguably the earliest Shii legal ḥadī th source and, in turn, to examine the geographical movements and intellectual exchange between Kūfa, Qum, and Qayrawān.
The preliminary testing conducted to investigate the historicity of other early sources has produced similar results. The methodology employed here can be applied to a number of early Shii collections in order to produce a more complete picture of the early sources of Shii ḥadī th tradition. 73 I am well aware of the limitations of my proposed method. First, the three stringent measures suggested, for a holistic assessment, in this analysis are not immediately available in all cases. Most of the later Shii ḥadī th collections are composed by non-contemporaneous Twelver scholars of Qum and Baghdad. Second, al-Īḍāḥ survives only as a small fragment that contains citations from a meagre 21 early sources. A complete manuscript of al-Īḍāḥ, if ever found, would greatly enrich our understanding of early Shii ḥadī th tradition. Third, the data obtained through cross-regional textual analysis does not always result in the identification of an overwhelming amount of identical material. In excavating the earliest layers of al-Ḥalabī's collection from the later sources, I was able to discover 41 per cent of identical material. The difference, though minor, of the remaining 59 per cent seems to have resulted from the nature of its dissemination across time and regions.
Notwithstanding the practical limitations caused by the scarcity of early material, my hypothesis, on a rather optimistic note, could also be tested on surviving Zaydi, Ismaili, and even Sunni ḥadī th sources. The Shii doctrinal and legal thought emerged in the same scholarly milieu in which Sunni thought flourished, sharing the same concerns, operating within the same intellectual framework, and consulting similar sources. The cross-regional textual analysis of later Kūfan and Medinese ḥadı̄th collections compiled by proto-Sunni and proto-Shii transmitters may help excavate the earliest layer of Islamic thought of the late first/seventh and early second/eighth centuries. This is a very ambitious project, and the efficacy of the method proposed here remains to be seen. It is hoped that the method outlined in this article will serve as a useful point of departure for future work on cross-regional and inter-and intra-sectarian modes of transmission in Islamic literature more broadly. The cross-regional textual analysis of the reports cited in al-Īḍāḥ and Amālī indicate that they shared a common source dating back to an earlier period.