The collection and canonization of the QurʾĀn

Authored by: Herbert Berg

Routledge Handbook on Early Islam

Print publication date:  August  2017
Online publication date:  August  2017

Print ISBN: 9781138821187
eBook ISBN: 9781315743462
Adobe ISBN:




Except in the case of some New Religious Movements, there usually exists a significant gap of decades, even centuries, between the time the words were first spoken, recited, sung, composed, and/or written and the time they became encapsulated within a text that was (mostly) unalterable and recognized as “scripture.” That is, the processes of compilation and canonization are often chronologically distant from the origin of the material. The traditional accounts in Islam and even some theories put forth by scholars skeptical of those accounts place the origins, compilation, and canonization in close temporal proximity. Yet, an examination of these accounts and theories shows that the situation in early Islam was more complex, and more skeptical theories suggest those accounts are not just inaccurate, but were fabricated and circulated to mask the true processes that led to the Qurʾān.

 Add to shortlist  Cite

The collection and canonization of the QurʾĀn

Except in the case of some New Religious Movements, there usually exists a significant gap of decades, even centuries, between the time the words were first spoken, recited, sung, composed, and/or written and the time they became encapsulated within a text that was (mostly) unalterable and recognized as “scripture.” That is, the processes of compilation and canonization are often chronologically distant from the origin of the material. The traditional accounts in Islam and even some theories put forth by scholars skeptical of those accounts place the origins, compilation, and canonization in close temporal proximity. Yet, an examination of these accounts and theories shows that the situation in early Islam was more complex, and more skeptical theories suggest those accounts are not just inaccurate, but were fabricated and circulated to mask the true processes that led to the Qurʾān.

Traditional accounts of collection and canonization

Later Muslim theology, in identifying the Qurʾān with the “preserved tablet” (lawḥ maḥfūẓ) and “mother of the book” (umm al-kitāb) (Q 85:21–22 and 43:4, respectively), could be said to maintain that the canonization of the Qurʾān occurred with this original, eternal, heavenly archetype. The first revelation of the Qurʾān as given in the sīra, despite being a recitation (lit. qurʾān), did not speak of itself as a scripture nor was it understood as such by Muḥammad. On the one hand, the gradual and piecemeal revelation for over two decades also initially seems somewhat incompatible with the theological view. On the other hand, ḥadīths state that Gabriel used to meet Muḥammad every night of Ramadan to teach or review the Qurʾān with him (see, for example, al-Bukhārī n.d.: no. 6). A similar divine supervision could be inferred from Q 75:16–17: “Do not move your tongue to hasten it, for indeed it is upon Us is its collection and its recitation” and from Q 13:39 “God effaces what He wills or establishes, and with Him is the Mother of the Book” (see also Q 2:160; 16:101). Taken together, these verses and accounts suggest divine editorial control, and perhaps even redaction – though the Qurʾān contains no explicit command to put the revelations into writing or a specific order. Towards the end of his life, “the Prophet said, ‘Every year Gabriel used to revise the Qurʾān with me once only, but this year he has done so twice’” (al-Bukhārī n.d.: nos. 3623, 3624). Muḥammad understood this to portend his death, but it also suggests divine supervision, this time over a complete and final version (the textus receptus ne varietur) – though not in the form of a written codex. If this was the case, after Muḥammad’s death all that was required was to convert the oral canon into a written one by writing down all the revealed materials and placing them in the order God had mandated via Gabriel, collecting them from what was preserved on parchments, shoulder blades, leaf stalks of the date-palm, pieces of leather, and flat stones, and in “the hearts of men” (al-Bukhārī n.d.: nos. 4679, 7191; see also al-Suyūṭī 2011: 140–141).

Yet the collection and canonization process(es) seem a little more complex (and confused) according to the various traditional accounts of Abū Bakr’s, ʿUmar’s, and ʿUthmān’s roles in the collection of the Qurʾān. The divine archetype and the physical codex are distinct (in their production, if not in their content). The first collection of the Qur’an, according to the most popular version, was prompted by the battle of Yamāma (11/632), and ʿUmar became concerned about the number of men who had died – men who had memorized the Qurʾān and whose death might thus mean the loss of some part of the Qurʾān that they alone might know. The first caliph, Abū Bakr, acquiesced and set Zayd b. Thābit the task of collecting the materials from the aforementioned varied sources. The sheets (ṣuḥūf) upon which he wrote this “first” collection passed to ʿUmar upon his predecessor’s death, and to the second caliph’s daughter, Ḥafṣa, 1 upon his death (al-Bukhārī n.d.: no. 7191). W. Montgomery Watt has pointed out several problems and discrepancies in this account, including: (1) the assumption that there was really no attempt to order the revelations prior to Yamāma; (2) the existence of conflicting traditions variously identifying ʿUmar or Abū Bakr as having come up with the idea; (3) the accounts that make ʿUmar the first collector of the Qurʾān (and others ʿAlī) (see Ibn Abī Dāwūd n.d.: 16); (4) the fact that few of those who died at Yamāma would have known much of the Qurʾān by heart; (5) the odd lack of authority accorded to this caliphal collection; and (6) the even odder passing of this official copy to ʿUmar’s daughter and not his successor, ʿUthmān (Watt and Bell 1970: 40–42; for a rebuttal of these objections, see Jones 1983: 237–238). In any case, the exercise was repeated under the third caliph when differences in how the Qurʾān was being read or recited came to a head between troops drawn from Syria and those from Iraq during expeditions to Armenia and Azerbayjan. When ʿUthmān became aware of the disputes, he once again called upon Zayd and other prominent Companions to collect the Qurʾān. In cases of dispute, the Qurashī dialect was to be preferred. Ḥafṣa’s ṣuḥūf were consulted, five (or seven) copies were made, and distributed to the major cities of Islam – “rival” copies, both ṣaḥīfas (individual leaves) and maṣḥafs (codices), were to be burned and subsequent copies based on this ʿUthmānic codex (al-Suyūṭī 2011: 141–144). The compilation became the textus ne varietur, and so the canonization process was also complete. As Watt maintains, “it is certain that the book still in our hands is essentially the ʿUthmānic Qurʾān” (Watt and Bell 1970: 44). But even based on traditional Muslim sources, the process of canonization was far more complex, both in terms of compilation and canonization of a ne varietur text.

First, there are reports of an earlier collection made by Muḥammad himself. Ibn Shabba (d. 262/876) preserves two traditions that support the production of a complete codex (lit. ṣuhūf) during Muḥammad’s lifetime and other reports suggest Muḥammad employed scribes to record the Qurʾān. ʿUthmān is said to have had ʿĀʾisha bt. Abū Bakr send him the parchment on which the Qurʾān was transcribed by the Prophet when God revealed it to Gabriel, and he revealed it to Muḥammad (Ibn Shabba 1979: III 997–998). Al-Suyūṭī comes to a similar though not so specific conclusion: the whole of the Qurʾān was written down during the life of Muḥammad, “but it was not collected all in the same place nor had its sūrahs been arranged” (2011: 137). Second, and perhaps more compelling, are the many reports of the pre-ʿUthmānic codices, particularly those of ʿAbd Allāh b. Masʿūd (d. 32/652–653) – who in one report is said to have been present when Gabriel checked Muḥammad’s recitation of the Qurʾān just before latter’s death (Ibn Abī Shayba 1989: VII 204) – Ubayy b. Kaʿb (d. 19/640 or 22/643), and Abū Mūsā al-Ashʿarī (d. 42/662). Various ḥadīths argue that the ʿUthmānic not the Masʿūdic codex was first, and others the reverse (see for example, Ibn Ḥanbal 1895: I 362). In a tradition that seems to reflect an attempt to resolve this dispute:

A man complained to the Prophet (pbuh), “ʿAbd Allāh b. Masʿūd taught me to recite a surah; Zayd taught me to recite it and so did Ubayy b. Kaʿb. Their recitations differed. Which recitation do I adopt?” The Prophet (pbuh) was silent. ʿAlī who was at his side said, “Everyone should recite just as he was taught. Each of the recitations is valid (ḥasan) and beautiful (jamīl).”

(al-Ṭabarī 1992: I 36 no. 14)

Regardless of whether this occurred or is merely a projection of a later dispute going back to the time of Muḥammad, it is clear that multiple readings of the Qurʾān existed, and that some felt Ibn Masʿūd’s version to be more complete, correct, and authentic, and they supported his refusal to have his copy destroyed during the promulgation of the ʿUthmānic codex. Ibn Masʿūd’s and Ubayy’s codices differed not only in the order of the surahs, but also in content, with the former omitting Q 1, 113, and 114, and the latter including two additional short prayer-like surahs similar to these three. Sean Anthony (forthcoming) argues that, based on legal and ḥadīth literature, the two surahs absent from the ʿUthmānic codex remained part of the oral, ritual canon, though in the written canon they remained liminal and disputed. 2 Abū Mūsā al-Ashʿarī is recorded as remembering two verses (or surahs) that are not in ʿUthmān’s codex (Muslim n.d.: no. 1050). Later Muslims would reject these doubts, hoping or stating that the verses in the surahs within the ʿUthmānic codex were divinely ordained (al-Suyūṭī 2011: 144–147), as was the arrangement of the surahs themselves. That is to say, all are identical to the final recited version approved by Gabriel (Ibn Saʿd 1957: II 195; al-Suyūṭī 2011: 148). Here too Watt concurs largely with the Qurʾān’s completeness: “Whatever view is taken of the collection and compilation of the Qurʾān, the possibility remains that parts of it may have been lost. … There is no reason, however, to think that anything of importance has gone astray … with perhaps minor exceptions, we have the whole of what was revealed to Muḥammad” (Watt and Bell 1970: 56). 3 Behnam Sadeghi and Uwe Bergmann, however, based on the recovered lower text of the Ṣanʿāʾ 1 palimpsest, make a strong case for the existence of non-ʿUthmānic Companion codices (2010).

Alternative interpretations: Muḥammad as compiler

John Burton argues that the accounts of the collection of the Qurʾān “are a mass of confusions, contradictions and inconsistencies” (1977: 225). For instance, Abū Bakr, ʿUmar, and ʿUthmān, the first three successors to Muḥammad, are identified with having instigated the collection of the Qurʾān. Attempts at harmonizing these disparate accounts involved claiming Abū Bakr to be the instigator and ʿUmar to be the completer of the project, ʿUmar to be an instigator and ʿUthmān also, or all three to be involved, with the first two caliphs making the first collection, and ʿUthmān making a final version. Other Western scholars have suggested that ʿUthmān was solely responsible, but his later unpopularity or his connection to the later unpopular Umayyad dynasty caused earlier Muslims to project the task of collecting the Qurʾān onto his two predecessors. Burton points out that the one thing that all these accounts share, however, is that they consistently exclude Muḥammad’s participation in the collection of the materials of the Qurʾān.

Why would Muḥammad need to be excluded? According to Burton, “a safe and certain recension of the valid revelations was unthinkable” while Muḥammad was alive, since the abrogation (naskh) and withdrawal of any passage was a “daily possibility” (1977: 232). Naskh might involve a new revelation that results in an earlier revelation no longer having legal force nor appearing in the Qurʾān, or it might have a new revelation that removes the older revelations’ legal force, though both still appear in the Qurʾān. Even though it suggests that the Qurʾān is incomplete, the former occurrence is not problematic since Q 2:106 itself suggests that God can cause a verse to be forgotten and replaced by a better verse. He points out a third possibility, however, made by the infamous “stoning verse.” Q 24:2 states that the punishment for adultery is flogging, but several well-attested ḥadīths insist (via ʿUmar) that the punishment is stoning – based on a passage revealed to Muḥammad, but somehow not present in the Qurʾān. In other words, there was a new revelation which overturned an earlier one, yet only the earlier one appears in the Qurʾān. This third form is problematic for it suggests that the Qurʾān as preserved is incomplete. Burton concludes that “the Muslims simply could not afford to be found or to find themselves in possession of a Qurʾān document that had been edited, checked and promulgated by its Prophet-recipient” (1977: 232). Others (Abū Bakr, ʿUmar, ʿUthmān, Zayd b. Thābit, etc.) must therefore be the collectors of the document. The conflicting reports’ true purpose, therefore, is to obscure the true collector and collator of the Qurʾānic revelations: Muḥammad.

In the 1930s, Richard Bell had also suggested that Muḥammad had a very significant role in constructing the Qurʾān. By looking at shifts in the grammatical construction, rhyming scheme, or content of a passage he concluded that almost all surahs originally consisted of numerous separate passages (with only rare unified compositions of any great length) and could be dissected into their component parts. Bell then argued that the Qurʾān was in written form when the redactors such as Zayd b. Thābit started their work; he entertained the possibility that the materials could have been written by Muḥammad himself, but thought it more likely that they were written down by others at his dictation (Bell 1937: I vi). Q 6:7 suggests that the idea of producing a book on papyrus was current and had at least entered Muḥammad’s mind. In addition, ṣūḥuf (that is, separate, unbound sheets used for writing) are connected with Qurʾānic revelation in Q 20:133, 53:37, 80:13, 87:18, and 98:2. Nor, when the accusation is made that the Qurʾān is written down (Q 25:6–7), does it deny the charge.

Bell’s framework for the proto-canonization process is more intriguing. Using the development of ideas within the Qurʾān and references to historical events as guides, Bell suggested that the composition of the Qurʾān fell into three main periods. The early period (from which only fragments survive) consisted primarily of lists of “signs” and exhortations to worship God. The Qurʾān period covered the latter part of Muḥammad’s time in Mecca, and the first year or so in Medina. During this period Muḥammad sought to produce a qurʾān – understood to be an Arabic “recitation” of the gist of previous revelations. The Book period, beginning near the end of the second year in Medina, is characterized by Muḥammad’s efforts at producing a kitāb or “book” that would be an independent scripture akin to the Torah and Gospels (Bell 1937: I vi). This task was cut short by Muḥammad’s death, and so the canonization process was left to his successors, Abū Bakr, ʿUmar, and ʿUthmān. Interestingly, though she would dispute many of Bell’s conclusions, Angelika Neuwirth also speaks of three modes or phases of the Qurʾān: the “oral Qurʾān” as it was first communicated by Muḥammad to the first listeners, the kitāb as excerpts from the heavenly book particularly in Medina, and the Qurʾān as codex, when it became the scripture of a community after the death of Muḥammad (Neuwirth 2006: 145–155). Neither Bell’s nor Neuwirth’s hypothesis is incompatible with the later compilation under the first three caliphs.

Changes to the canon

Whether one holds that the collection of the Qurʾān was divinely guided, completed under the early caliphs, or a product of Muḥammad’s own efforts as both Burton and Bell suggest, the process of canonization took more time and only ended (albeit arguably) with Ibn Mujāhid (d. 324/936). The ʿUthmanic codex or, as François Déroche and others prefer, the ʿUthmānic rasm (that is, the skeleton script without vowel markers or diacritics to distinguish the identically shaped consonants) would necessarily undergo changes to emend its defective script. Initially, a codex served mostly as an “elaborate mnemonic device” (Watt and Bell 1970: 47), but as the community rapidly expanded the extensive memorization could no longer be presupposed. The script needed to be improved with the addition of vocalization and diacritics to distinguish homographs if significantly divergent qirāʾāt (readings) were to be prevented. The ʿUthmānic rasm was by no mean yet ne varietur; whether the consonantal skeleton of the ʿUthmānic codex was open to change was another matter.

During the time of the Umayyad caliph ʿAbd al-Malik (r. 685–705) the issue of defective script seems to have been addressed. Additions to the ʿUthmānic rasm were likely far more gradual (based on manuscript evidence), but the traditional accounts focus on two figures: ʿUbayd Allāh b. Ziyād (d. 67/686) and especially al-Ḥajjāj b. Yūsuf (d. 95/714). ʿUbayd Allāh b. Ziyād, an Umayyad who became the governor of Iraq in 56/675, is credited with introducing 2,000 letters into the muṣḥaf, primarily alifs and the lengthened alifs for qālū and kānū (Ibn Abī Dāwūd n.d.: 129). Although doubts have been raised about this tradition, Omar Hamdan has argued for its authenticity (2010: 796–800). Al-Ḥajjāj b. Yūsuf, the governor of Iraq from 75/694 to 95/713, was responsible for more sweeping orthographical reforms – not so much to end political disputes, but rather to burnish the image of the Umayyads. Moreover, “the main cause for the diversity of the Qurʾanic readings lies not in the written text, but rather in its orality” (Hamdan 2010: 799). Ibn Abī Dāwūd present a tradition that states al-Ḥajjāj changed the muṣḥaf in eleven verses (n.d.: 130). This tradition does not reflect the scope of what Hamdan describes as the “Maṣāhif Project” (2006: 135–174; 2010).

For about 40 years until the time of ʿAbd al-Malik b. Marwān, people recited the codex of ʿUthmān. Then, faulty readings multiplied and spread in Iraq. Al-Ḥajjāj b. Yūsuf al-Thaqafī turned to his secretaries and asked them to put some signs on these ambiguous letters. It is said that Naṣr b. ʿAsim carried out this operation, placing single or doubled dots in different places. For a time, we therefore read the text with its diacritization. However, despite the use of dots, faulty readings still occurred. It was then that vocalization (iʿjām) was created. One now reads by following the system of dots and vowels. When careful attention was not paid to a word by not complying with this system, faulty readings occurred. Means to reduce them were sought, and no other way was found than to stick to the instructions from the experts in oral recitation.

(Ibn Khallikān 1994: II 32)

Under the instruction or with the consent of ʿAbd al-Malik, al-Ḥajjāj instructed a number of scribes to insert diacritics so consonants could be distinguished and to add vocalization. According to Hamdan the project occurred between 84/703 and 85/704, employed the private muṣḥaf of ʿUthmān, and involved the counting of all the consonants, words, and verses of the Qurʾān to ensure completeness and integrity (2010: 801, 806, 809–815). That this was not an entirely novel concept is attested by the presence of diacritical marks to distinguish homographs in at least one papyrus and one inscription dated 22/643 and 24/645, respectively, though they appear less (but were not absent) in Qurʾānic manuscripts (Déroche 2014: 135–136; Hamdan 2006: 146–148). Once it was complete, al-Ḥajjaj followed the ʿUthmānic model for standardization and had conflicting Qurʾānic recensions destroyed, 4 starting in Kufa, where Ibn Masʿūd’s muṣḥaf remained in use, and the new codices were sent to the major cities of the Umayyad empire. ʿAbd al-ʿAzīz, the governor of Egypt (r. 60/680–85/704), rejected the copy sent to Egypt, favoring the ʿUthmānic codex. “The goal of al-Ḥajjāj was not only a standardization of the rasm, 5 but also of the oral tradition of the qirāʾāt” (Hamdan 2010: 825). In achieving the latter, he was less successful.

In fact, the various readings or recitations, qirāʾāt, continued to develop unabated until Ibn Mujāhid. With the support of vizier Ibn Muqla (d. 328/940), he sought to enforce greater uniformity, or rather stop the increasing multiplicity and complexity (Melchert 2000: 18, 22), by limiting the acceptable readings. In this, he was building upon the work of earlier grammarians and readers (Shah 2004). He argued that seven qirāʾāt, those of seven 8th-century scholars, were equally valid. Each conformed to the ʿUthmānic rasm, was grammatically correct, and enjoyed the consensus of the community. These seven represented readings of different districts (Medina, Mecca, Damascus, Basra, and three from Kufa – though not Ibn Masʿūd’s, which had been prevalent in Kufa).

This “oral canonization” – or more precisely, “limited agreement and manageability” (Melchert 2000: 22; see also Shah 2004: 94) – was only possible because the text, that is, the rasm, had already been fixed. Differing but equally valid qirāʾāt have been supported using ḥadīths such as the report that ʿUmar heard a surah being recited differently than he had heard Muḥammad recite it. He dragged the reciter before Muḥammad, and the latter had the man and ʿUmar recite the surah. He then affirmed both as having been revealed that way, for “This Qurʾān was sent down in seven (different) ways [aḥruf], so recite from it whatever is easy for you” (Mālik b. Anas n.d.: no. 477). 6 According to Viviane Comerro, initially this ḥadīth permitted the liturgical recitation of ancient qirāʾāt – especially that of Ibn Masʿūd – because the Companions and Successors themselves had done so, even though these qirāʾāt did not conform to the ‘Uthmānic rasm. Although a minority position, it was accepted but eventually fell victim to Ibn Mujāhid’s reform (Comerro 2013), ironically using the same sets of ḥadiths. 7 Christopher Melchert, however, sees no justification for a connection between aḥruf and qirāʾāt. 8

Later this list of seven was expanded to ten and then fourteen, each also becoming canonical (Melchert 2008). That Ibn Mujāhid’s work became so prominent and that he could have Ibn Shannabūdh (d. 328/939) 9 punished for having used the version of Ibn Masʿūd in a liturgical context, confirms the establishment of a ne varietur textual norm accepted officially at the end of the 3rd/9th century – the culmination of the canonization process. Only minor, superficial changes took place afterwards, from the division of the text (e.g., into fifths or tenths, takhmīs and taʿshīr respectively) and the use of different scripts (e.g., ḥijāzī v. Kufic), different sizes and shapes (i.e., oblong v. vertical), and ornamentation and illumination, to the production of lavish editions for the elite (Déroche 2014: 140–141; Hamdan 2006: 816–821).

Revisionist accounts

Although the likes of Burton and Bell may seem radical in the sense that they dismiss a large number of traditional accounts of the collection and compilation of the Qurʾān as largely inaccurate or even fabricated, they do, however, largely accept the basic timeline of those processes as given in tradition. In fact, they emphasize the role of Muḥammad in the process. Scholars termed “revisionists” are by and large far more skeptical and suggest not only a new timeline but sometimes even a new milieu for the origin of Qurʾānic materials, their compilation, and eventual canonization. This skepticism is born out of belief that “the sources on the rise of Islam are of questionable historical value” (Crone 1987: 203) and that “what we know of the seventh-century Ḥijāz is the product of intense literary activity” (Wansbrough 1987: 14–15). This skepticism has led some to conclude “we do not know and probably never can know what really happened; all we can know is what later people believed happened” (Rippin 1985: 157).

Although skepticism about the compilation of the Qurʾān predated John Wansbrough (see below), it was his suggestion the Qur ʾān was “the product of an organic development from originally independent traditions during a long period of transmission” (Wansbrough 1977: 47) that was particularly radical. These independent pericopes or “prophetical logia” originated separately with communities within a Judeo-Christian sectarian milieu (Wansbrough 1977: 50). This hypothesis breaks the connection between Muḥammad and the Qurʾān, though the biography of the former, the “Arabian prophet” (to use Wansbrough’s term), represents a historicization of the logia via exegesis. As for the canonization process, Wansbrough suggested that the style of the Qurʾān belies the claim to a single author or editor but rather seems to be “the product of an organic development from originally independent traditions during a long period of transmission” (1977: 47). Moreover, “it is of course neither possible, nor necessary, to maintain that the material of the canon did not, in some form, exist prior to that period of intensive literary activity,” but the ne varietur text occurred only “towards the end of the second century” (1977: 44). 10 Wansbrough’s revision of Islamic history and in particular the late date for the canonization of the Qurʾān has been challenged on many fronts. For example, Harald Motzki (2001) and Gregor Schoeler (2010) defend the ḥadīths that underpin the traditional accounts, and Behnam Sadeghi and Hohsen Goudarzi (2012) seek to show the Ṣanʿāʾ 1 manuscript diverges from and predates the ʿUthmanic codex, proving that the aforementioned revision must be wrong. 11

Günter Lüling and Christoph Luxenberg also revision the origins of the material that makes up the Qurʾān. The former argues that as much as a third of the Qurʾān – the shorter, poetic surahs –originated as pre-Islamic, Christian hymns. The incoherence of some Qurʾānic passages required later editing and so misinterpretation took place, aided greatly by the initially defective Arabic script in which these passages were recorded (Lüling 1974). Luxenberg goes further by focusing on problematic Qurʾānic terms and passages, seeking to find explanations in Aramaic. This could involve examining exegetical and lexicographical works for Aramaic readings, searching for Syro-Aramaic homonyms that might better explain the passages, or even seeking possible Aramaic roots by altering the diacritical points and vocalizations (since neither appeared in the earliest written versions of the Qurʾān). Given this approach, it is perhaps unsurprising that Luxenberg often finds those Aramaic forerunners and so he concludes that the Arabic Qur’an, or much of it, was excerpted from a Syriac canonical and/or proto-scriptural Urtext. He suggests that Mecca was an Aramean settlement in which an Aramaic-Arabic hybrid was spoken. Later, Arabic-speaking exegetes and philologists unfamiliar with or unaware of this hybrid had only the written Qurʾān’s defective script to work with, which was standardized only in the second half of the 8th century (Luxenberg 2000). Luxenberg has convinced very few scholars (for an exception, see Chapter 18 in this volume). 12

Not only do some scholars revision the compilation of the Qurʾān, but some the canonization. Paul Casanova’s argument that the Qurʾān was not closed with ʿUthmān but remained fairly fluid until ʿAbd Mālik and his governor al-Ḥajjāj standardized and canonized it (1911–1924: 103–142) has more recently been championed by Alfred-Louis de Prémare and Stephen Shoemaker (de Prémare 2002: 278–306; de Prémare 2010: 189–221; Shoemaker 2012: 146–58). Chase Robinson summarizes some of the reasons to doubt the traditional account (Robinson 2005: 101–104). A state-sponsored distribution of a single official version as early as the reign of ʿUthmān seems atypical of the centuries-long processes that led to scriptures in antiquity and late antiquity, especially considering that it is said to have gone from an oral text and one written on shoulder blades and stones to a canon on vellum or papyrus. Moreover, ʿUthmān’s rule was contentious and fairly short, and was a polity “that lacked many rudimentary instruments of coercion and made no systematic attempt to project images of its one authority – no coins, little public buildings or inscriptions” (Robinson 2005: 102). ʿAbd al-Malik had the motivation and means to impose such standardization, however. In addition, there are Qurʾānic texts datable to the caliph’s period that depart from the ʿUthmanic codex: for example, a letter by Ḥasan al-Baṣrī (d. 728) in which he cites a “Qurʾānic passage” not in the Qurʾān and the deviations from the Qurʾān in the inscriptions on the Dome of the Rock. David Powers’ study of the passages in the Qurʾān and sīra about Zayd b. Ḥāritha, Muḥammad’s one-time adopted son whose ex-wife he married, also lends support to this hypothesis. Powers argues that certain passages in the Qurʾān were revised and others were added. These were not merely minor variations or a misreading of the unvoweled and non-diacriticized ʿUthmānic rasm. Rather they were significant revisions and editions of the passages in question that were both theologically and politically motivated. The implication is that the Qurʾān remained open and fluid for three-quarters of a century between the death of the Prophet and the caliphate of ʿAbd al-Malik (Powers 2009).

This “Marwānid hypothesis” has of course been strongly challenged by, for example, Sadeghi and Bergmann (2010: 343–435). Given the evidence of early manuscripts of the Qurʾān (Sadeghi and Bergman 2010; Sadeeghi and Goudarzi 2012), Nicholas Sinai discusses the possibility of the “emergent canon model” in which the Qurʾānic text may have achieved recognizable form by 660 but thereafter was reworked until canonized by ʿAbd al-Malik. The epigraphic evidence, the reports on al-Ḥajjāj, and Christian sources used by de Prémare are argued to be unconvincing (Sinai 21014a). In favor of the closure of the Qurʾān instead in the mid-7th century, Sinai adduces the unanimous ascription of the standard rasm to ʿUthmān and the burning of rival codices (acknowledged even by his opponents) and the aforementioned text-critical arguments of Sadeghi, Goudarzi, and Bergmann. He points out that archaic Qurʾānic grammatical features were not made to conform to later usage and that no narrative contextualization crept into the Qurʾān despite its prevalence prior to the year 700. Sinai also expands on Fred Donner’s observation that in the Qurʾān “we find not a single reference to events, personalities, groups or issues that clearly belong to periods after the time of Muḥammad” (Donner 1998: 49), which makes sense only if the standard rasm is early. Although Sinai recognizes that his conclusion is not unassailable, given the preponderance of evidence in favor of the traditional dating of the standard rasm, that is, the ʿUthmānic rasm, it ought to be the default position (2014b). Even Patricia Crone, who is usually considered a revisionist because of her early work, accepts the evidence of the carbon-dated manuscripts and so abandoned the mid-Umayyad codification of the Qurʾān. She concludes that there is no longer a good reason to doubt the ʿUthmānic codification; that is to say, the Qurʾān existed when the tradition said it did (2016: xiii).


Whatever else one thinks of John Wansbrough’s revisionist theories, he alerted scholars of early Islam that there is a marked difference between the origins of materials that become scripture and the canonization of those materials so that they become scripture. The confusion of those two processes in the case of Islam is not surprising. In Christianity and Judaism, material thought to be revelations from God were written decades or centuries later and certainly only canonized at least a century and a half later or more in the case of Christianity. And it was carried out by the Church. But the Qurʾān speaks of itself as God’s speech and both it and tradition could be read as implying that it was “divinely compiled and canonized.”

There seems little doubt that the traditional account is problematic or at least the processes of compilation and canonization were somewhat more complex than they initially appear. Scholarship on these processes has not (yet?) come to a complete consensus that accounts for the contradictions and complexity. Scholars who address these issues cannot even agree on what constitutes a methodology or evidence. Thus, Burton dismisses the Islamic tradition, but so does Wansbrough – yet their hypotheses could hardly be more different. Bell, in putting so much of the compilation at the feet of Muḥammad, may seem quite radical, but not that different from the theological tradition that had God directing the canonization process or at least the editorial part of the process via Gabriel. And while the Marwānid hypothesis has some appeal, it raises more issues than it solves. This debate about the compilation and canonization is likely to continue for the foreseeable future, for it is a long way from the first qurʾān or kitāb to a fully vocalized and diacriticized ne varietur text that comes in fourteen readings. That is as it should be; the centrality of the Qurʾān’s birth and growth lies at the very heart of the study of early Islam.


Ruqayya Khan (2014) suggests that Ḥafṣa had a significant editorial role in the compilation of the Qurʾān. Sean W. Anthony and Catherine L. Bronson (2016) raise strong objections to that claim. See also note 4 below.

Anthony is following Angelika Neuwirth’s assertion of two modes of simultaneous publication and codification: textual and ritual:

In view of the fact that the creation of the Qurʾān’s scriptural corpus was completed within a singularly short space of time, and the authoritative codification and dissemination of the entire text … followed just as swiftly, it is easy to lose sight of a second parallel process: the emergence of the an oral canon which was tangible within live recitation and whose Sitz im Leben was the community’s service, the ritual … with its central prayer rite, the ṣalāt.

(Neuwirth 2014: 141; see Anthony forthcoming)

For an in-depth analysis of the historical sources about the compilation and transmission of the Qurʾān, see Viviane Comerro (2012). The two key sources for the reports on the compilation of the Qurʾān are the Ṣaḥīḥ of al-Bukhārī (d. 256/870) and the Tārīkh of al-Ṭabarī (d. 310/923), though there are at least six other competing narratives. She concludes that these reports reflect what the community idealized the Qurʾān to be, but not what really happened (Comerro 2012: 197). Using his isnād-cum-matn methodology, Motzki also attempts to undermine the Western skepticism about reports about the collection of the Qurʾān (particularly that of John Wansbrough and John Burton, for which see below). He concludes:

We are not able to prove that the accounts on the history of the Qurʾān go back to eyewitnesses of the events which are alleged to have occurred. We cannot be sure that things really happened as is presented in the traditions. However, Muslim accounts are much earlier and thus much nearer to the time to the alleged events than hitherto assumed in Western scholarship. Admittedly, these accounts contain some details which seem to be implausible or, to put it more cautiously, await explanation, but the Western views which claim to replace them by more plausible and historically more reliable accounts are obviously far from what they make themselves out to be.

(Motzki 2001:31)

Gregor Schoeler also challenges the hypotheses of Burton and Wansbrough. He argues that the traditional reports that much of the Qurʾān was written down in some form at the time of Muḥammad’s death are “not unbelievable” (2010: 782). As for the compilation of the Qurʾān under Abū Bakr and/or ʿUmar and the official edition under ʿUthmān, the reports are fairly consistent with each other and with how other texts were treated. Schoeler does not regard each tradition as authentic, yet does “believe in something like the authenticity of the overall picture that tradition conveys of the codification of the Qurʾān” (2010: 788). That is, “the compilation and redaction of the Qurʾān under ʿUthmān (unanimously supported by tradition) is, if not proven, then at least extremely probable” (Schoeler 2010: 789).

As for the muṣḥaf of Ḥafṣa, it had been destroyed much earlier by Marwān (d. 65/685), who managed to obtain it from her brother after she had died. (Ibn Abī Dāwūd n.d.: 32). He feared that it might diverge from the ʿUthmānic codex. Anthony and Bronson discuss the possibility that it was destroyed even earlier by ʿUthmān, but neither may be the case; “the diverse accounts of Ḥafṣāh’s codex serve as a literary means to add testimonies to the veracity and success of ʿUthmān’s project” (2016: 113).

The process of improving the original defective orthography was certainly complete by the end of the 9th century, but the reports may be suspect since the early ʿAbbāsid caliphs sought to erase or take credit for Umayyad achievements.

For some of the various interpretations of aḥruf, see al-Ṭabarī’s discussion (1992: I 53–55).

Comerro adds:

[T]‌he study of traditions informs us on some crucial elements of the history of the text: the plasticity of its composition and oral transmission; the antiquity of its writing; the fixation of a model written under ʿUthmān; its gradual canonization; the preservation of textual variants as a reflection of the original oral diversity and then the philologists’ interest; the parallel theologizing of the history of transmission.

(Comerro 2013; see also Comerro 2012).

Melchert argues:

Western scholars have also asserted that Ibn Mujāhid’s choice of seven acceptable readings was related to the hadith report that the Qurʾan had been revealed in seven aḥruf …. Yet al-Ṭabarī interprets it as referring to seven recensions of which only one had been preserved, the other six irretrievably lost in ʿUthmān’s codification. … He thought the seven aḥruf had nothing to do with the qirāʾāt. A little later, Ibn Ḥibbān (d. 354/965) would write of thirty-five to forty different explanations for the hadith report of seven aḥruf. … Al-Suyūtī quotes half a dozen authorities against identifying the Seven Readings with the seven aḥruf of the hadith report.

(Melchert 2000: 19)

Elsewhere Melchert asserts that “premodern tradition of Qurʾan scholarship unanimously denies any equation of known readings with the seven aḥruf” (2008: 83). Even al-Suyūṭī found about 40 different understandings of the seven aḥruf (2011: 104–115).

Mālik b. Anas (d. 179/795) found it “necessary to declare that the ruler had a duty to prevent the sale or recitation of the versou of Ibn Masʿūd” (Cook 2000: 121) and al-Ḥajjāj likewise “would often threaten to kill the people of Kufa should they not cease following the reading of Ibn Masʿūd” (Hamdan 2010: 799).

On the basis of an example of exegetes unanimously misunderstanding a passage of the Qurʾān and an example of a discontinuity between Qurʾānic legislation and Islamic law, Crone argues that the traditional account of the Qurʾān’s origins and canonization seems very unlikely (1994).

For a concise defense of the revisionist position, see Koren and Nevo (1991).

Claude Gilliot is also an exception. He points out that Arabia had extensive interactions with the nearby Aramaic, Jewish, and Christian cultures, and suggests the Aramaic trail set by Lüling and Luxenberg may yet lead to the pre-Qurʾānic lectionary (Gilliot 2010: 164).


Anthony, S.W. Forthcoming. Two ‘Lost’ Sūras of the Qurʾān: Sūrat al-Khalʿ and Sūrat al-Ḥafd between Textual and Ritual Canon (1st–3rd/7th–9th Centuries). Jerusalem Studies in Arabic and Islam .
Anthony, S.W. and Bronson, C.L. 2016. Did Ḥafṣah edit the Qurʾān? A Response with Notes on the Codices of the Prophet’s Wives. Journal of the International Qurʾanic Studies Association 1:93–125.
Bell, R. 1937. The Qurʾān: Translated, with a Critical Re-arrangement of the Surahs . 2 vols. Edinburgh: T&T Clark.
Bell, R. 1953. Introduction to the Qurʾān . Edinburgh: Edinburgh University Press.
al-Bukhārī, M.b.I. n.d. Ṣaḥīḥ al-Bukhārī. Khan, M.M. trans.
Burton, J. 1977. The Collection of the Qurʾān . Cambridge: Cambridge University Press.
Casanova, P. 1911–1924. Mohammed et la fin du monde: étude critique sure l’Islam primitive . Paris: P. Gauthier.
Comerro, V. 2012. Les traditions sur la composition du muṣḥaf de ʿUthmān . Beirut: Orient-Institut Beirut.
Comerro, V. 2013. The Traditions of the Composition of ʿUthmān’s muṣḥaf. International Qurʾanic Studies Association blog.
Cook, M. 2000. The Koran: A Very Short Introduction . Oxford: Oxford University Press.
Crone, P. 1987. Meccan Trade and the Rise of Islam . Princeton: Princeton University Press.
Crone, P. 1994. Two Legal Problems Bearing on the Early History of the Qurʾān. Jerusalem Studies in Arabic and Islam 18:1–37.
Crone, P. 2016. The Qurʾānic Pagans and Related Matters: Collected Studies in Three Volumes . Volume 1. Siurua, H. ed. Leiden: Brill.
Déroche, F. 2006. Written Transmission. In: Rippin, A. ed. The Blackwell Companion to the Qurʾān . Malden, MA: Blackwell Publishing, 172–186.
Déroche, F. 2014. Qurʾans of the Umayyads: A First Overview . Leiden: Brill.
Donner, F.M. 1998. Narratives of Islamic Origins: The Beginnings of Islamic Historical Writing . Princeton: Darwin Press.
Gilliot, C. 2010. On the Origins of the Informants of the Prophet. In: Ohlig, K.-H. and Puin, G.-R. eds. The Hidden Origins of Islam . Amherst, MA: Prometheus Books, 153–187.
Hamdan, O. 2006, Studien zur Kanonisierung des Korantextes: al-Ḥasan al-Baṣrīs Beiträge zur Geschichte des Korans . Wiesbaden: Harrassowitz.
Hamdan, O. 2010. The Second Maṣāḥif-Project: A Step towards the Canonization of the Qur’anic Text. In: Neuwirth, A. , Sinai, N. , and Marx, M. eds. The Qurʾān in Context: Historical and Literary Investigations into the Qurʾānic Milieu . Leiden: Brill, 795–835.
Ibn Abī Dāwūd al-Sijistānī . n.d. ʿA.A. Kitāb al-maṣāḥif . Beirut: Dār al-kutub al-ʿilmīya.
Ibn Abī Shayba, ʿA.A.b.M. 1989. Al-Muṣannaf fi al-ḥadīth wa-al-āthār . Al-Lahham, S. ed. 8 vols. Beirut: Dār al-fikr.
Ibn Ḥanbal, A.b.M. 1895. Al-Musnad . 6 vols. Cairo: al-Maṭbaʿa al-maymanīya.
Ibn Khallikān, A.ʿA.Ḥ.b.M. 1994. Wafayāt al-aʿyān wa-anbāʾ abnāʾ al-zaman . ʿ Abbās, I. ed. 8 vols. Beirut : Dār ṣādir.
Ibn Saʿd, M. 1957. Kitāb al-ṭabaqāt al-kubrā . 8 vols. Beirut: Dār ṣādir.
Ibn Shabba, ʿU. 1979. Taʾrīkh al-Madina al-munawwara . Shaltūt, F.M. ed. 4 vols. Mecca: Dār al-turāth.
Jones, A. 1983. The Qurʾān – II. In: Beeston, A.F.L. , Johnstone, T.M. , Serjeant, R.B. , and Smith, G.R. eds. Arabic Literature to the End of the Umayyad Period . Cambridge: Cambridge University Press, 228–245.
Khan, Ruqayya . 2014. Did a Woman Edit the Qurʾān? Hafṣa [sic] and Her Famed ‘Codex.’ Journal of the American Academy of Religion 84:174–216.
Koren, J. and Nevo, Y.D. 1991. Methodological Approaches to Islamic Studies. Der Islam 68:87–107.
Lüling, G. 1974. Über den Ur-Quran: Ansätze z. Rekonstruktion vorislam, christl. Strophenlieder Quran . Erlangen: Lüling.
Luxenberg, C. 2000. Die syro-aramäische Lesart des Koran: Ein Beitrag zur Entschlüsselung des Koransprache . Berlin: Das Arabische Buch.
Mālik b. Anas . n.d. Muwaṭṭā Mālik.
Melchert, C. 2000. Ibn Mujāhid and the Establishment of Seven Qurʾanic Readings. Studia Islamica 91:5–22.
Melchert, C. 2008. The Relation of the Ten Readings to One Another. Journal of Qurʾanic Studies 10:73–87.
Motzki, H. 2001. The Collection of the Qurʾān: A Reconsideration of Western Views in Light of Recent Methodological Developments. Der Islam 78:1–34.
Muslim b. al-Ḥajjāj . n.d. Ṣaḥīḥ Muslim. Siddiqui, A.H. trans.
Neuwirth, A. 2006. Structure and the Emergence of Community. In: Rippin, A. ed. The Blackwell Companion to the Qurʾān . Malden, MA: Blackwell Publishing, 140–158.
Neuwirth, A. 2014. Scripture, Poetry, and the Making of a Community: Reading the Qurʾān as a Literary Text . Oxford: Oxford University Press in association with the Institute of Ismaili Studies.
Powers, D.S. 2009. Muhammad is Not the Father of Any of Your Men: The Making of the Last Prophet . Philadelphia: University of Pennsylvania Press.
de Prémare, A.-L. 2002. Les fondations de l’Islam: entre écriture et histoire . Paris: Éditions du Seuil.
de Prémare, A.-L. 2010. ʿAbd al-Malik b. Marwān and the Processes of the Qurʾān’s Composition. In: Ohlig, K.-H. and Puin, G.-R. eds. The Hidden Origins of Islam . Amherst, MA: Prometheus Books, 189–221.
Rippin, A. 1985. Literary Analysis of Qurʾān, tafsīr, and sīra: the Methodologies of John Wansbrough. In: Martin, R.C. ed. Approaches to Islam in Religious Studies . Tucson: The University of Arizona Press, 151–163.
Robinson, C.F. 2005. ʿAbd al-Malik . Oxford: Oneworld.
Sadeghi, B. and Bergmann, U. 2010. The Codex of a Companion of the Prophet and the Qurʾān of the Prophet. Arabica 57:343–436.
Sadeghi, B. and Goudarzi, M. 2012. Ṣanʿāʾ 1 and the Origins of the Qurʾān. Der Islam 87:1–129.
Schoeler, G. 2010. The Codification of the Qurʾān: A Comment on the Hypothesis of Burton and Wansbrough. In: Neuwirth, A. , Sinai, N. , and Marx, M. eds. The Qurʾān in Context: Historical and Literary Investigations into the Qurʾānic Milieu . Leiden: Brill, 779–794.
Shah, M. 2004. The Early Arabic Grammarians’ Contributions to the Collection and Authentication of Qurʾanic Readings: The Prelude to Ibn Mujāhid’s Kitāb al-Sabʿa . Journal of Qurʾanic Studies 6:72–102.
Shoemaker, S.J. 2012. The Death of a Prophet: The End of Muhammad’s Life and the Beginnings of Islam . Philadelphia: University of Pennsylvania Press.
Sinai, N. 2014a. When Did the Consonantal Skeleton of the Quran Reach Closure? Part I. Bulletin of the School of Oriental and African Studies 77:273–292.
Sinai, N. 2014b. When Did the Consonantal Skeleton of the Quran Reach Closure? Part II. Bulletin of the School of Oriental and African Studies 77:509–521.
al-Suyūṭī, J.D. 2011. The Perfect Guide to the Sciences of the Qurʾān , vol. 1: Al-Itqān fī ʿUlūm al-Qurʾān . Algar, H. , Schub, M. , and Abdel Ḥaleem, A. trans. Reading: Garnet Publishing.
al-Ṭabarī, A.J.M.b.J. 1992. Tafsīr al-Ṭabarī al-musammā Jāmiʿ al-bayān fī taʾwīl al-Qurʾān . 12 vols. Beirut: Dār al-kutub al-ʿilmīya.
Wansbrough, J. 1977. Quranic Studies: Sources and Methods of Scriptural Interpretation . Oxford: Oxford University Press.
Wansbrough, J. 1987. Res Ispa Loquitor: History and Mimesis . Jerusalem: The Israel Academy of Sciences and Humanities.
Watt, W.M. and Bell, R. 1970. Introduction to the Qurʾān . Edinburgh: Edinburgh University Press.
Search for more...
Back to top

Use of cookies on this website

We are using cookies to provide statistics that help us give you the best experience of our site. You can find out more in our Privacy Policy. By continuing to use the site you are agreeing to our use of cookies.