In our previous blog post, we argued that al-Tabari (d. 310/923) had to hand an extensive written collection consisting of sets of well-written notes.

In this post, we demonstrate that the data we have collected helps us begin to visualise the notebooks that served as al-Tabari’s ‘data carriers’.1 Scholars have often mined al-Tabari’s citations for evidence of earlier works. They have given the lion’s share of their attention to reputed authors such as Mujahid and Sayf b. ʿUmar who lived two or more generations before al-Tabari. By contrast, our starting point is his direct informants and the material they passed on to al-Tabari. In some cases, they combined material from many different sources.

Tree Maps

We visualise al-Tabari’s notebooks using a tree map that was created from a subset of the isnads introduced by the phrases ‘he told me’/’he told us’ (haddathana/haddathani) – specifically, isnads with at least three transmitters after al-Tabari. The tree map displays each person in an isnad as a rectangle; the size of the rectangle is proportional to the number of times the person is cited. Our main tree map contains four layers of rectangles: al-Tabari himself (dark pink), his direct informants (dark orange), their informants (light orange) and the latters’ informants (yellow). For example, in the first notebook, Bishr b. Muʿadh’s (d. 245/859–60) main source is Yazid b. Zariʿ (d. 182/798-9?). All but twenty of the 2,387 chains that contain three persons after al-Tabari run thus: Bishr b. Muʿadh < Yazid b. Zariʿ < Saʿid.

It is important to stress that our tree maps do not represent the entirety of the data set, only the subset that fits their parameters. A tree map focused on isnads containing three transmitters will exclude all isnads that are shorter than three transmitters. Similarly, a tree map visualising isnads of at least five transmitters will exclude isnads with fewer than five, resulting in a smaller subset and a different tree map. Historians interested in the earliest layers of transmission may find the longer isnad tree maps particularly interesting, as they highlight chunks of material that seem to have passed through multiple generations to al-Tabari’s direct informants and on to him. But they should remember that the data set was created to prioritise al-Tabari’s direct informants, not transmitters five generations removed from them.

The images included here are excerpts from a larger set of tree maps that we will publish with our data set in a forthcoming print publication. Bibliographical information is collated from publications by Gilliot, Athari and Lucas,2 as well as from Ibn Nadim’s Fihrist and the relevant Encyclopaedia of Islam articles. We have also consulted other sources.3

Six Sets of Notes

What follows is a series of visualisations of six possible – even likely – collections of notes used by al-Tabari.

Notebook 1: Bishr b. Muʿadh (d. 245/859–60), cited 2,418 times, filtered to 2,387 chains of at least three transmitters.

Image 1: The possible structure of al-Tabari’s notes on his lessons with Bishr b. Muʿadh in Basra.

Al-Tabari relies on Bishr b. Muʿadh chiefly for reports going back to an authority he calls Saʿid, transmitted via Yazid b. Zariʿ. Ibn al-Nadim mentions in his bio-bibliographical Fihrist a Kitab Saʿid b. Bashir ʿan Qatada within a list of compilations of Quran commentary (tasmiyat al-kutub al-musannafa fi tafsir al-Qurʾan). Qatada b. Diʿama (d. ca. 117/735) was a ‘successor’, that is, he belonged to the generation of Muslims after the Prophet’s death. He was famed for his memory and his knowledge of genealogies, lexicography, historical traditions, Quranic exegesis, variant Quran readings (qiraʾāt) and Hadith. As for Saʿid, Muhammad b. Saʿd (d. 230/845) has an entry for Abu ʿAbd al-Rahman Saʿid b. Bashir al-Azdi, who, he says, died in 170 (786 or 787 CE), at the start of Harun al-Rashid’s caliphate; he also mentions that Saʿid b. Bashir was from Basra but settled in Damascus. He quotes several reports from Qatada through Saʿid b. Bashir.

If we look at the isnads that continue after Saʿid, we indeed find that nearly all go back to Qatada. But can we be sure that the Saʿīd who passed on material from Qatada was it Saʿid b. Bashir? We cannot. Al-Tabari received reports from Qatada through several routes, including via a transmitter he names explicitly as Saʿīd b. Bashīr and another he calls Saʿīd b. Abī ʿUrūba, who may or may not be the same person. It is perhaps significant that when he refers specifically to Saʿid b. Bashir as a source for Qatada, his direct informant was not Bishr b. Muʿadh but rather a variety of other people (as many as nine). This means that we cannot be certain that there is only one Saʿīd who reported from Qatada or that al-Tabari received material from Saʿid b. Bashir through Bishr b. Muʿadh .4 This is where further work on named entities would be particularly useful. We can say, however, that whatever the origin of its contents, al-Tabari’s Bishr b. Muʿadh notebook contained a lot of material from Qatada.

Al-Tabari used this notebook mainly for the Jāmiʿ al-bayān, but he also picked it up for the Taʾrikh and a few times for the Tahdhib. The citations from it in the Taʾrikh all occur early in the book, with the last (at milestone 984) being found in al-Tabari’s reporting on the sixth year of the hijra. The sixth year is when the Prophet attempted to return to Mecca to make the lesser pilgrimage (ʿumra) and ended up concluding the accord of al-Hudaybiyya (March 628). Al-Tabari cites Bishr as one of several sources on the events.

Notebook 2: Muhammad b. Humayd al-Razi (d. 248/862), cited 2,879 times, filtered to 2,312 chains of at least three transmitters.

Image 2: The possible structure of al-Tabari’s notes on his lessons with Muhammad b. Humayd in Rayy.

In terms of the frequency of citation, the notes al-Tabari obtained from Muhammad b. Humayd were by far the most important for the former’s works. Given the large number of quotations from these notes, they may have occupied numerous books or booklets. In biographical literature, Muhammad b. Humayd is well-known as a transmitter of Ibn Ishaq’s Sira through the intermediary Salama b. al-Fadl, and it is likely that the material he transmitted from Ibn Ishaq was bound together. In this case, the difference between such notebooks versus a ‘book’ obtained through an ijāza may have been slight. The data rightly challenges us to think harder not just about ‘actual books’ but also about the concept of notes and notebooks.

In writing the Taʾrikh, our data would say that al-Tabari relied almost exclusively on the notes from Muhammad b. Humayd for Ibn Ishaq’s Sira.5 But for his other works he drew also on other notebooks containing material from the Sira, especially those of Abu Kurayb (d. 248/862), Yaʿqub b. Ibrāhim al-Dawraqi (d. 252/866), and Sufyan b. Wakiʿ (d. 247/861), as well as at least seven others.

Given the large number of citations represented in this tree map, we might imagine that the material al-Tabari obtained from Jarir b. ʿAbd al-Hamid, Mihran b. Abi ʿUmar and Abu Tamila Yahya b. Saʿid would likewise have been bound. But how these notebooks would have been stored within al-Tabari’s own shelving system is another question.

Nearly all material transmitted by Mihran from ‘Sufyan’ is quoted in the Tafsir from Sura 41 (‘Fussilat’) onwards; this is most likely a collection of the exegetical opinions of Sufyan al-Thawri (d. 161/778). Al-Tabari also cites other direct informants for Sufyan al-Thawri’s opinions, including Yunus b. ʿAbd al-Aʿla (d. 264/877), Abu Kurayb (d. 248/862), and al-Harith b. Muhammad (d. 282/895), as well as the chains Muhammad b. Humayd < Salama b. al-Fadl < Sufyan and Muhammad b. Humayd < Harun < Sufyan.

It would be worthwhile to compare the fragmentary Quran commentary attributed to Sufyan6 with al-Tabari’s Tafsir. We have not done this. But we do have a machine-readable tafsir text attributed to Sufyan al-Thawri in the OpenITI corpus (under identifier Shia002462; 29,140 words),7 which has twelve short alignments with al-Tabari’s text; the longest of these measures sixty-one words.8

Notebook 3: Al-Muthanna b. Ibrahim al-Amuli (d. unknown), cited 2,259 times, filtered to 1,973 chains of at least three people.

Image 3: The possible structure of al-Tabari’s notes from his lessons with al-Muthanna in Rayy (possibly Amul also).

The notes that al-Tabari assembled from al-Muthanna b. Ibrahim al-Amuli are quite diverse, with the largest pieces deriving from transmitters named Ishaq b. al-Hajjaj and Abu Salih ʿAbd Allah.

The latter appears in Ibn al-Nadim’s Fihrist as Abu Salih ʿAbd Allah b. Muhammad b. Yazdad b. Suwayd (d. 223/837-8). Ibn al-Nadim calls him an eloquent scribe (ahad al-kuttab al-bulaghaʾ) and attributes to him a Kitab al-Taʾrikh and a book of epistles (Kitab rasaʾilihi). Al-Tabari refers to him in the Tafsir and the Tahdhib as ‘the scribe of al-Layth’ (katib al-Layth); al-Layth appears within the tree map, too. This seems to refer to the famous traditionist and faqih al-Layth b. Saʿd (d. 175/791).9 Abu Salih ʿAbd Allah also transmitted from Muʿawiya b. Salih. In our isnads, Muʿawiya’s reporting goes back to Ibn ʿAbbas through ʿAli b. Abi Talha.

Edward Zychowicz-Coghill has recently collected witnesses to a Taʾrikh by al-Layth on the basis of a recension by Yunus b. Bukayr that is quoted in five other texts. He argues that al-Tabari did not draw on al-Layth’s Taʾrikh.10 However, our data reveals additional evidence in the Tafsir that contradicts his contention. Al-Layth is in fact cited in the Taʾrikh, but through other isnads (and not, according to our data, through Yunus b. Bukayr).

Al-Layth’s situation is complex, and readers will be forgiven if they find it a bit hard to follow. For us, the most important feature of it is that it allows us to discern how works that Ibn al-Nadim may have recognised as written works were broken down before al-Tabari’s lifetime, absorbed into subsequent writings, and then reabsorbed into al-Tabari’s book. This process seems to have carried memories of earlier works and their conservators (e.g. the reference to katib al-Layth) but without either citing the works themselves or making an effort to keep them intact. That is why reconstructions are so tricky.

The case of al-Tabari and, indeed, text reuse data in general, confirm what Beatrice Gruendler has recently argued, which is that in the third/ninth century (as well as beyond) the authorial agency of scholars such as al-Tabari was often exerted through acts of ‘memorizing, collecting, arranging, commenting, and transmitting of pre-existing sources’. As she writes, ‘This dealing with extant text made a scholar no less of an author in the eyes of his contemporaries, as he provided a repository of the knowledge that they needed.’ Gruendler distinguishes between different types of authorial agency. In poetry, authorship was recognised in the creation of a text, and a ‘basic difference already existed between a “ripped off” verbatim formulation and a borrowed reformulated motif’. In the case of poetry, ‘word-by-word’ copying was frowned on, though ‘a degree of reformulation changed the discourse from ownership to literary inspiration’.11

If we step back, we can understand these two forms of agency as part of a wider epistemology, which attributed greater prestige to adaptation than to simple, perfect reproduction. The reproduction of a work – as a katib – did not initiate one into the genealogy of a master, whereas a student’s transmission of a teacher’s work (as a rawi or otherwise) incorporated the student’s persona into the text’s lineage.

This hierarchy of prestige can be detected in al-Tabari’s citations. Consider the following examples, which indicate al-Tabari’s efforts to track specific pieces.

Notebooks 4 and 5: Ismaʿil b. ʿAbd al-Rahman al-Suddi (d. ca. 127/745) via Musa b. Harun al-Hamdani (d. unknown) and Muhammad b. al-Husayn (d. 261/874), respectively.

Image 4: The possible structure of al-Tabari’s notes from his lessons with Musa b. Harun al-Hamdani (d. unknown) and Muhammad b. al-Husayn (d. 261/874).

The direct informants here are the obscure Musa b. Harun and Muhammad b. al-Husayn al-Baghdadi, a Hadith scholar from Nisa in modern-day Turkmenistan). Both transmit through different intermediaries (ʿAmr vs. Ahmad) from Asbat b. Nasr al-Hamdani, who in turn transmits from al-Suddi, a mawla who was a popular preacher in Kufa. In both cases, al-Suddi’s ‘box’ (that is, the material he provided) is neatly nested inside Asbat’s. However, the material from al-Suddī accounts for less than half of the three-person isnads from either Musa b. Harun (who starts a total of 508 chains) and or Muhammad b. al-Husayn (who starts 528).

Examining these citations, G. H. A. Juynbol questioned the degree to which the material attributed to al-Suddi could be used to understand his views: ‘In al-Tabari’s Tafsir, countless exegetical remarks ascribed to al-Suddi can be found and could conceivably be brought together in a volume. Whether such a compilation would allow conclusions as to a certain bias or predilection on his part, if any, has as yet to be established.’ Ibn al-Nadīm does indeed mention a Kitab Tafsir al-Suddi within the section, mentioned above, that lists compilations of Quranic commentary (tasmiyat al-kutub al-musannafat fi tafsir al-Qurʾan).

Notebook 6: ʿAbd Allah b. al-ʿAbbas (d. ca. 68/687–8) via Muhammad b. Saʿd (d. 276/889).

Image 5: The possible structure of al-Tabari’s notes on information received through a family isnad from Muhammad b. Saʿd (d. 276/889) in Baghdad.

Al-Tabari met Muhammad b. Saʿd b. Muhammad b. al-Hasan b. ʿAtiyya b. Saʿd b. Junada al-ʿAwfi in Baghdad. The information he received from the latter originates with ʿAbd Allah b. al-ʿAbbas (d. ca. 68/687–8), a member of the first generation of Muslims who is frequently called the father of Quranic exegesis. The full chain, in which the description of family relations reflects the respective quoted speaker’s perspective, runs as follows: Muhammad b. Saʿd < abi [‘my father’] < ʿammi [‘my paternal uncle’, i.e. Muhammad b. Saʿd’s great-uncle] < abihi [‘his father’, i.e. the great-uncle’s father] < Ibn ʿAbbas. Franz Rosenthal described al-Tabari’s material from Muhammad b. Saʿd as an ‘unpublished’ book and ‘the heirloom of a particular family’ and noted that ‘a family isnad as complete as this is most remarkable’.12

It should be noted that although generating our data was challenging, once generated it is relatively easy to read. Filtering the data in the Transmission Chains table to order the isnads alphabetically by the name in position 5, for example, reveals numerous identical isnads running back to Ibn ʿAbbas. The interpretation of these isnads will, of course, remain a subject of debate.

The Limits of Our Argument

Scholars such as Lawrence Conrad, Andrew Rippin, Ella Landau-Tasseron and Arietta Papaconstantinou have rightly raised questions about efforts to reconstruct lost sources.13 There is an understandable concern in the field about the risks associated with mining later sources for evidence of earlier ones.

Our work is a comparatively modest effort. Our primary focus is on what al-Tabari had to hand, not on whether any given chunk of material can in fact be traced back to Umayyad times. We make no judgement, for example, about whether the Ibn ʿAbbas notebook actually goes back to the man himself. Indeed, we are sceptical that the notebooks can be reconstructed. Instead, they can be described, given that we possess only extracts from them.

As for the earlier layers, we are also generally sceptical about the possibility of reconstructing ‘actual books’ from earlier time periods on the basis of quotations and citations found in works that survive today. Our scepticism arises from the fact that authors treated their notes as reusable materials. Now that technology can aid us in new ways, further work on assessing alignments will be worthwhile, but we should be searching for more nuanced ways to describe what we find. What was the nature of these earlier works? Corpora of quotations can be collected and described and sketches of the early written tradition developed on the basis of this evidence, as Kevin Jaques is doing for Ibn Ishaq and as Zychowicz-Coghill has done for al-Layth b. Saʿd.

Our main bone of contention with efforts at reconstruction that mine later sources for earlier works is that they often give little or no attention or agency to the persons we call al-Tabari’s direct informants. Al-Tabari drew on a material reality that had its own form and organisation. Broadly speaking, our work reflects the field’s rising attention to the materiality of texts, although we do not work with manuscripts. Rather, we direct our attention to where the material has vanished but direct citations of it remain.

In our next post, we compare al-Tabari’s books and investigate how different notebooks fed into different works.

  1. Beatrice Gruendler, The Rise of the Arabic Book (Cambridge, MA: Harvard University Press, 2020), 67. Maxim Romanov has made this argument for al-Dhahabi’s (d. 748/1348) Taʾrikh al-Islam; see e.g. Romanov, ‘Observations of a Medieval Quantitative Historian?’, Der Islam 94/2 (2017), 462–95, []{.ul}, at 482ff. 

  2. Claude Gilliot, ‘La formation intellectuelle de Tabari (224/5–310/839–923)’, Journal asiatique 276 (1988), 203–44; Akram Atharī, Muʿjam shuyūkh al-Ṭabarī (Amman and Cairo: n.p., 2005); and Scott C. Lucas, Selections from the Comprehensive Exposition of the Interpretations of the Verses of the Qurʾān (Cambridge: Royal Aal Al-Bayt Institute for Islamic Thought and Islamic Texts Society, 2017), esp. ii, 467–82 (‘Appendix B: Ṭabarī’s Teachers’). 

  3. E.g. Muhammad S. H. Hallaq, Rijal Tafsir al-Tabari, jarhan wa-taʿdilan: Min tahqiq Jamiʿ al-bayan ʿan taʾwil ay al-Qurʾan li-Ahmad Shakir wa-Mahmud Shakir (Beirut: Dar Ibn Hazm, 1999). 

  4. Cf. Athari, Muʿjam shuyukh al-Tabari, 159. 

  5. R. Kevin Jaques has assembled a list of persons upon whom al-Tabari relied for Ibn Ishaq’s Sira, plus locations within the Taʾrikh and Tafsir

  6. Sufyan al-Thawri, Tafsir al-Qurʾan al-karim, ed. Imtiyaz ʿAli ʿArshi (Rampur: n.p., 1965). 

  7. This work was published as Tafsir al-Thawri, edited by ‘a committee (lajna) of ʿulamaʾ’, in 1403 AH [1982–3 CE]. 

  8. The alignments are available at []{.ul} (included in the data set associated to these blogs). 

  9. Al-Tabari refers to al-Layth b. Saʿd (also as al-Layth b. Saʿd al-Misri), but in our data set this name never appears in the al-Muthanna < ʿAbd Allah chains, which refer only to ‘al-Layth’. Al-Tabari may well have maintained in his citations the different surface forms of names that he had recorded in his notes. Al-Khatib al-Baghdadi (d. 463/1071) describes an ʿAbd Allah b. Saliḥ b. Muhammad b. Muslim Abu Salih as an Egyptian who was the katib al-Layth b. Saʿd and who accompanied the latter to Baghdad. The names here do not quite match (ʿAbd Allah b. Muhammad versus ʿAbd Allah b. Salih). See al-Khatib al-Baghdadi, Taʾrikh Baghdad, OpenITI, Shamela0000736-ara2.mARkdown, search term: كاتب الليث. 

  10. Edward Zychowicz-Coghill, The First Arabic Annals: Fragments of Umayyad History (Berlin: de Gruyter, 2021), 3. 

  11. Gruendler, Rise of the Arabic Book, 60–2. 

  12. Franz Rosenthal, general introduction to al-Tabari, The History of al-Ṭabarī, i: General Introduction and From the Creation to the Flood, trans. Franz Rosenthal (Albany: State University of New York Press, 1989), 6 and 215, n. 337. 

  13. Lawrence Conrad, ‘Recovering Lost Texts: Some Methodological Issues’, Journal of the American Oriental Society 113/2 (1993), 258–63; Andrew Rippin, ‘Tafsir Ibn ʿAbbās and Criteria for Dating Early Tafsir Texts’, Jerusalem Studies in Arabic and Islam 18 (1994), 38–83; Ella Landau-Tasseron, ‘On the Reconstruction of Lost Sources’, al-Qanṭara 25 (2004), 45–91; Arietta Papaconstantinou, review of Robert Hoyland, Theophilus of Edessa’s Chronicle and the Circulation of Historical Knowledge in Late Antiquity and Early Islam, Le Muséon 126 (2013), 459–65.