Let’s talk survival bias

How many manuscript copies were produced of any given texts? How many were circulating at any given time? When were they lost, discarded or destroyed?  How often are the works penned on their pages cited in other texts by the same or different authors? These questions lie at the very heart of our epistemology and are usually subsumed under the term ‘survival bias’.

Works that survive longer obviously have a greater chance at being read, recited, reproduced, acknowledged. And it is today commonly assumed that the number of manuscript copies of a given work tells us something about its impact in its own time and on the succeeding traditions in which it inscribed itself—or was meant to do.

Samer Akkach, in his Letters of a Sufi Scholar: The Correspondence of ʿAbd Al-Ghanī Al-Nābulusī (1641-1731) [Leiden; Boston: Brill, 2010], gives an interesting statement that relates to the survival of Ibn Ṭūlūn manuscripts:

And by my life, this has been the habit of the people of our town, Dimashq al-Shām: they disregard the ʿulamāʾ and the people of virtue, and they antagonise, slander, and harm them. They abolish their sciences and perfections and allow their disrespectful fools and rejects to pester them. How many works of erudite men of knowledge, of men who grew up among them, have they disregarded and lost, neither respect-ing them nor taking note of their books and writings until they had all disappeared and perished?

And surely there was among them the best man of knowledge and the pride of all ḥadīth scholars, Ibn Ṭūlūn al-Ḥanafī, yet they disregarded him and lost his books and works, of which hardly any are now left; and those that are left are still in his own handwriting, since no-one cared to have them copied. And so they did to Ibn ʿAbd al-Hādī al-Ṣāliḥī: they lost his books and works, and they did the same to Raḍiyy al-Dīn al-Ghazzī… and to his son shaykh Badr al-Dīn al-Ghazzī and to his son shaykh Najm al-Dīn al-Ghazzī, the seal of the ḥadīth scholars (ḥuffāẓ). They disregarded them all and lost their numerous beneficial works […]. (23)

For more than a year now I have been pondering how we should understand al-Nābulusī’s rant. The context of this quote makes it clear that he situates himself and the resistance he faced within a long history of anti-intellectual sentiment in his hometown.

On the one hand, we could therefore take his complaints as a political statement which should serve to change public opinion for a more positive stance towards scholars. On the other hand, his assessment of the survival of Ibn Ṭūlūn’s corpus at least is corroborated by manuscript evidence: my own survey shows that more than three thirds of the roughly two-hundred surviving works remain only as autographs. But even so, this survey is a far shot from “hardly any are now left”. Thus, I would rather not credit this statement as a documentation of a historical reality.

What shall us concern here, however, is the other statement al-Nābulusī makes about Ibn Ṭūlūn’s written legacy: “no-one cared to have them copied”. What would that mean in the light of the introductory questions? Did Ibn Ṭūlūn become outdated in the century between his death and al-Nābulusī’s statement? Was al-Nābulusī even correct in this statement? And why, if so, did two-hundred works survive until today if no one cared to read and preserve them?

In “Women and Ḥadīth Transmission. Two Case Studies From Mamluk Damascus” [Studia Islamica 95 (2002): 71-94], Asma Sayeed writes that samāʿ recitation sessions that their “primary function … was to verify the accuracy of the text being read” and that “students would examine their own copies to ensure that these were identical to the text being read.” (80) Thus, an important author or ḥadīth transmitter could be identified through the number of copies made by their own students, if not later generations of students and copyists.

Ibn Ṭūlūn is credited with having taught uncountable numbers of students by his biographers. Nonetheless, apart from excerpts of some of his works, there is not much manuscript evidence of his samāʿ sessions. The overall survival of much of his corpus could be taken as an indicator that his general importance was still acknowledged but the more interesting question might be, what was actually lost by the time of al-Nabulusī.

Luckily, Ibn Ṭūlūn himself was not shy about recording such acts of transmission. His late biographical dictionary, Dhakhāʾir al-qaṣr fī tarājim nubalāʾ al-ʿaṣr is full of references to those, as I have alluded to with regard to his transmission of Thulāthiyyāt. He also documents the transmission of some of his own works that utilize or organize ḥadīth and akhbār. And it contains pertinent anecdotes such as his copying two texts by al-Ghazzālī and Ibn Sīnā both known as Risālat al-Ṭayr as well as the Muqaddimat ʿilm al-akhlāq by Abū ʿAbd Allāh Muḥammad b. Mūsā al-Irdibīlī from a fellow scholar’s collection who had reached Damascus in 942 (fol. 29b).

Thus, if we compare the number of biographees hearing a certain work with the number of surviving manuscript copies, we could get a better idea of the survival of his works and thereby somewhat reliable numbers to assess the survival bias related to his corpus more generally.

Overall, samāʿ transmission of Ibn Ṭūlūn’s publications is mentioned in fourteen biographies and refers to only twelve different works (numbers in parentheses give the number of mentions):

  1. al-Aḥādīth al-murawiyya fī al-basātīn al-nayrabiyya (1)
  2. al-Arbaʿīn ḥadīthan ʿan Arbaʿīn shaykhan fī Arbaʿīn bāban min al-ʿilm al-sharīf (3)
  3. al-Arbaʿīn / Arbaʿīn ḥadīthan min Arbaʿīn ḥadīthan mufarida bi-l-taṣnīf (3)
  4. al-Tuḥfa al-laṭīfa fī al-masāʾil al-mutaʿayyin (ʿalā al-shāfiʿiyya) fīhā taqlīd Abī Ḥanīfa (2)
  5. Fatḥ al-ʿalīm fī al-musalsalāt bi-ḥarf al-mīm (2)
  6. Manḥ al-jalīl fīmā warada fī maqām al-Khalīl (1)
  7. Munyat al-aṭfāl wa-bughyat al-rijāl (1)
  8. Qudrat al-jalīl fīmā warada fī maqām (al-)Khalīl (7)
  9. Risāla fī al-fīl (1)
  10. Sharḥ al-ṣuḍūr fīmā ruwiya fī al-fakhkh wa-l-ʿuṣfūr (6)
  11. Tuḥfat al-ḥabīb fīmā warada fī al-kathīb (4)

Interestingly, the same works are usually transmitted at the same places. Out of the 22 cases where a place is mentioned,  Barza is intricately connected to #8 (all seven transmissions) and #2 (both transmissions), whereas #10 is transmitted in al-Tall four times, #11 twice at Masjid al-Qadam, and only the ḥadīth collection #2 was transmitted also in Ṣāliḥiyya and, more specifically, at the Salīmiyya Mosque.

One possible explanation for the concentrations might be that all biographees participated in the same samāʿ session. In contrast to the place, the time of transmission is not frequently provided by Ibn Ṭūlūn. Altogether, only three dates are mentioned: one transmission of #8 is dated 9.6.949, one for #11 took place on 9.11.936, and the only act of transmission for #7 fell on 19.1.943. Yet, without further circumstantial information, it is not possible to determine whether these dates relate to the other respective transmissions as well.

If Sayeed’s assessment is correct, this short survey would account for at least 29 copies of Ibn Ṭūlūn works being produced, let alone those by other attendants not included in the biographical dictionary. I am doubtful that works like Qudrat al-jalīl fīmā warada fī maqām al-KhalīlTuḥfat al-ḥabīb fīmā warada fī al-kathīb or even Sharḥ al-ṣuḍūr fīmā ruwiya fī al-fakhkh wa-l-ʿuṣfūr were actually copied by every listener, simply because they were first and foremost devotional texts. The first two describe faculties of shrines and that is where they were mostly recited.

In any case, the surviving manuscript evidence is poor for all the works mentioned. In this respect, al-Nābulusī seems to have been correct. For brevity’s sake, I’ll give the numbers of surviving manuscript copies with autograph manuscripts being enumerated separately and underlined in another list:

  1. al-Aḥādīth al-murawiyya fī al-basātīn al-nayrabiyya (0/0)
  2. al-Arbaʿīn ḥadīthan ʿan Arbaʿīn shaykhan fī Arbaʿīn bāban min al-ʿilm al-sharīf (0/0)
  3. al-Arbaʿīn / Arbaʿīn ḥadīthan min Arbaʿīn ḥadīthan mufarida bi-l-taṣnīf (0/1)
  4. al-Tuḥfa al-laṭīfa fī al-masāʾil al-mutaʿayyin (ʿalā al-shāfiʿiyya) fīhā taqlīd Abī Ḥanīfa (0/0)
  5. Fatḥ al-ʿalīm fī al-musalsalāt bi-ḥarf al-mīm (0/0)
  6. Manḥ al-jalīl fīmā warada fī maqām al-Khalīl (0/0)
  7. Munyat al-aṭfāl wa-bughyat al-rijāl (0/0)
  8. Qudrat al-jalīl fīmā warada fī maqām (al-)Khalīl (0/0)
  9. Risāla fī al-fīl (2/1)
  10. Sharḥ al-ṣuḍūr fīmā ruwiya fī al-fakhkh wa-l-ʿuṣfūr (2/0)
  11. Tuḥfat al-ḥabīb fīmā warada fī al-kathīb (1/1)

Furthermore, the two copies for titles #9 and #10 were produced only around 1900. That leaves us with one surviving copy out of 29 alleged copies made. Today, it is found in Bursa, Hüseyin Çelebi MS 854/1, fols. 1b-7b. And even that copy contains only excerpts of the work.

It will require further research to understand fully why especially those of Ibn Ṭūlūn’s works disappeared that were embedded in the culture of oral/aural transmission, whereas his historiographical works fared much better in terms of preservation. Was just he sidelined in the succeeding traditions or does it bespeak greater historical shifts and changes in transmission and reading interests during the Ottoman period?



A whole Majmūʿa copied? MS Majāmīʿ 315

In March, I returned to the Dār al-Kutub in Cairo and spent a whole week with their collection. Ibn Ṭūlūn’s name pops up very often in the different collections of the Dār al-Kutub. The earliest acquisitions are certainly found in the Taymūriyya Library, whereas the general collection seems to have made a bulk order of Ibn Ṭūlūn autographs only by the early 1930s.

Aḥmad Taymūr did not only acquire “originals”, however; he also created copies. One modern way of bridging the gap between both was photography, and Taymūr made ample use of it to create facsimiles of autographs. At literally the same time, he also hired copyists to ensure he had personal copies of Ibn Ṭūlūn texts. One of the most intriguing one is MS Majāmīʿ 315 because it is not a copy of one text but rather of an entire multiple-text manuscript.

Oddly enough, MS Majāmīʿ 315 does not have a call number referring to his own collection. Yet, it does carry his stamp on the first and last page, one time in red and another time in blue. It contains 533 pages of 22,7 x 17,2 cm, each filled with 15 lines of text. The writing is larger and spaces between lines – as well as the margins – are much more generous than in Ibn Ṭūlūn autographs. Together, this would account for the volume’s thickness.

Headings are written in red ink and larger script, being aligned centrally in the page. The paper is of good quality and of white color, the very prominent watermark reading “Gouvernement Egyptien 1915-1917”. The dating of the copy is made even more precise by a note below a contents statement on a separate leave: 1336/1917-18 (the copyist is not named). The leave is of brown color. On the top of it is stated that all works (rasāʾil) in this manuscript were originally authored by Ibn Ṭūlūn. The page’s center is occupied by the contents statement, which includes eleven entries:

  1. al-ʿUqūd al-durriyya fī al-umarāʾ al-miṣriyya
  2. Nafaḥāt al-zahr fī dhawq ahl al-ʿaṣr
  3. Al-taʿrīf li-fann al-taṣḥīf
  4. Araj al-nasamāt fī aʿmār al-makhlūqāt
  5. al-Mulḥa fīmā warada fī aṣl al-subḥa
  6. al-Naḥla li-mā warada fī al-nakhla
  7. Ibtisām al-thughūr ʿammā qīla fī nafʿ al-zuhūr
  8. Laqsh al-ḥanak fīmā warada fī al-samak
  9. Risāla fī al-fīl [incomplete at the beginning]
  10. Sharḥ al-ṣudūr fīmā ruwiya fī al-fakhkh wa-l-ʿuṣfūr
  11. Tuḥfat al-aḥbāb fī manṭaq al-ṭayr wa-l-dawābb

Taymūr had this copy made from a manuscript containing exclusively Ibn Ṭūlūn autographs. It is now, and might have been back then, part of the library of the Baladiyya of Alexandria. Stephan Conermann records it in his biographical article of Ibn Ṭūlūn as MS Alex. Fun. 183. According to Conermann, it does contain 14 textual units in total, including all of those mentioned above. Even though it might seem like unnecessary repetition, I will reproduce the list for MS Alex. Fun. 183 here:

  1. Ijāza
  2. al-Naḥla li-mā warada fī al-nakhla
  3. ʿUnwān al-rasāʾil li-maʿrifat al-awāʾil [includ.: al-Arbaʿīn ḥadīthan fī ḍamn ʿunwān al-rasāʾil fī maʿrifat al-awāʾil]
  4. Irtiyāḥ al-khawāṭir fī maʿrifat al-awākhir
  5. Risāla fī al-fīl
  6. Laqs al-ḥanak fīmā qīla fī al-samak
  7. Sharḥ al-ṣudūr fīmā ruwiya fī al-fakhkh wa-l-ʿuṣfūr
  8. Ibtisām al-thughūr fīmā qīla fī nafʿ al-zuhūr
  9. Tuḥfat al-aḥbāb fī manṭaq al-ṭayr wa-l-dawābb
  10. Araj al-nasamāt fī aʿmār al-makhlūqāt
  11. al-Mulḥā fīmā warada fī al-subḥa
  12. Nafaḥāt al-zahr fī dhawq ahl al-ʿaṣr
  13. al-Taʿrīf fī fann al-taṣḥif
  14. al-ʿUqūd al-durriyya fī al-ʿumarāʾ al-miṣriyya

Two things should become immediately apparent in a comparison between both contents statements. First, what was taken out? Taymūr apparently shed his own recreation of the manuscript from all works related to ḥadīth transmission (in bold), including the ijāza (although this entry probably refers to Ibn Ṭūlūn’s original contents statement, which was a preferred place for people to document their ijāzas from him). While the two works, ʿunwān al-rasāʾil and Irtiyāḥ al-khawāṭir, were also copied, they were put in separate manuscripts. This obviously sets the remaining works in a very different framework than the original one.

Other ḥadīth works are kept, this knowledge is not declared void. But the genres dedicated to documenting the transmission itself seems not to have had any value for Taymūr. Secondly, this coincides with a restructuring of the manuscript, putting the once final text on rulers of Egypt first. Also, the longest work, Nafaḥāt al-zahr fī dhawq ahl al-ʿaṣr, is moved up to second place. Other works, discussing certain wildlife themes, have been brought into a more systematic form, beginning now with the general makhlūqāt and only later addressing specific kinds of flora and fauna: the date palm, flowers, fish, birds, and mounts or pack animals (I am uncertain, however, whether Ibn Ṭūlūn really talks about elephants or rather the Quranic Sura by that name).

Even though the multiple-text manuscript is not reproduced exactly as it was, MS Majāmīʿ 315 is exemplary in that it selects and reshapes the contents of only this one source manuscript.