When European philologists of the nineteenth and early twentieth centuries looked at the Muslim scripture, one question preoccupied them: Whence the Quran? The first Orientalists had little interest in what Muslims had to say about their text of revelation: they wanted to know where it came from, how it came to be, and the “pre-history” of the Quranic text.Footnote ¹ This type of study is undergoing what can legitimately be called a renaissance. After spending much of the last century adrift in the academic doldrums, studies of the Quran are now very active and much more wide-ranging, but the search for what we might call the pre-Islamic Quran is especially prominent.

This time around, however, one crucial element is drastically different. Scholars today are less inclined to accept without question what we must call the traditional or orthodox version of Islamic origins, that is, the general narrative presented in the biographies of Muhammad, the sīra literature (until recently mainly those of Ibn Isḥāq (d. 150/768) and al-Wāqidī (d. 207/823)), and validated by (a generally Sunni) consensus over the ages. Orientalists may not have believed that Muhammad’s message was genuinely divine, but they accepted the non-divine parts as genuine.

What is the problem with a tradition that has served Muslims well for centuries? Western scholars have expressed reservations about any number of things, chief among them the fact that all our main sources were written down long after events they purport to describe, by people with vested interests in certain issues. Muslims themselves were aware that much of the hadith literature could not be considered historical fact, and the sīra’s claim to truth is even more suspect. Hadith and the formation of Islamic law were the object of some revisionist scholarship, notably of Ignaz Goldziher (1850–1921) and Joseph Schacht (1902–69), but the Quran less so.

If nobody was doubting the standard account of how the Quranic text was stabilized during the caliphate of ʿUthmān, there was nonetheless a list of unresolved questions about the Quran and its revelation and canonization, as well as the form and content of the text itself. How to explain the biblical material in the Quran? Are Jewish influences the most evident? Or Christian? Or Arab? Where did the biblical references come from, and how could the Messenger’s audience have understood them, especially if one accepts the conventional Muslim view of the Hijaz as predominantly pagan? How do we explain the different versions of the Quran’s composition or, for that matter, the myriad accounts of passages being added or omitted? What is the relation between the oral and written Quran? Why do we have countless references to variant readings but hardly any manuscript evidence for them?

Scholars have always been aware that the foundations of the traditional narrative were less than certain, that there was often no reason to favour one version of events over another, and that it was hard to go beyond conjecture. But these nagging questions could not displace the weight of tradition. As Régis Blachère summed up, “Let us admit in return, however, that without the Tradition we would have to resign ourselves to knowing nothing about how such an astonishing religious book was created”.Footnote ²

It is easy to be critical of inertia in the field, but one should keep in mind that Quranic (and even Islamic) studies in the West was and remains very small, with just a handful of researchers working on very difficult material. It is hardly surprising that they clung to a model which worked reasonably well, for which sources were abundant, and most importantly, for which they could see no alternative. Furthermore, with so few participants, there was hardly room for dissenters or splinter groups. A few lonely voices protested the status quo with some vigour but little impact, let alone success.

The changes began in 1977, with the publication of two monographs that radically challenged the accepted narratives of Islamic beginnings. Hagarism, by Patricia Crone and Michael Cook, was an attempt to rewrite Islamic history using only non-Muslim sources. John Wansbrough’s Quranic Studies was an attempt to apply methods of biblical criticism to the Quran and its early exegeses. Both books saw Jewish influence as paramount, and both argued that the Quran was not compiled in the seventh-century Hijaz, but later and further north. As successful alternative models for early Islamic history, neither has garnered much acceptance. As spurs to alternative methods and approaches, both have proved hugely important; that few found their arguments persuasive was less important than the paths opened up for further investigations. In this respect, Cook, Crone, and Wansbrough demonstrated that it was OK to be wrong: even misguided efforts can make important contributions.

Hagarism and Quranic Studies are so frequently mentioned together that one might forget that they do not form a “sceptical” school. They are methodologically incompatible if not antithetical. Casting doubt on Islamic tradition can be fruitful, but is not a method in itself, and should not be confused with one. There certainly exists a revisionist school of thought, as exemplified in Amir-Moezzi and Dye, Le Coran des historiens Footnote ³ (2019), but even that work displays a broad range of opinion and cannot be reduced to a simple debunking of Muslim orthodoxy.

We are faced with the very real challenge of trying to make sense of or resolve the dilemmas posed by Islamic origins and of finding approaches that will enable us to do so. For all the renewed interest in the Quran, it has yet to lead to any substantial new consensus “on any aspect of the discipline, really”.Footnote ⁴ On the one hand this is frustrating; on the other it is evidence of a fascinating and challenging intellectual puzzle. The challenges are prominently on view in Stephen J. Shoemaker’s Creating the Qur’an: A Historical-Critical Study, which attempts to follow in the paths of Wansbrough, Crone and others, but ultimately resembles its illustrious predecessors only in being flawed and unconvincing.

The reader cannot but feel that Shoemaker, however he might object, has mistaken scepticism for a method. He covers a wide range of topics (historiography, manuscripts, carbon dating, geology, etc.), somewhat suspiciously arriving at the same conclusion in each case. The sources are far more complex and difficult than one would know from Shoemaker’s treatment. It is worthwhile, nonetheless, devoting some attention to the monograph. Our field is a small one, with very few people working on topics related to Quranic origins. Creating the Qur’an is published by a major university press and is available in open access. Its subject has a relatively broad appeal and at the same time lends itself to polemics and apologetics.

Shoemaker is unimpressed with the current state of the field, and among his stated aims is to counter “the ossified credence in the canonical Sunni narrative of the Qur’an’s composition – particularly as rearticulated by Nöldeke and Schwally – that has stultified progress in the academic study of the Qur’an’s origins for over a century now” (p. 13).Footnote ⁵

Shoemaker’s “canonical Sunni narrative” alludes to the account of the Quranic codex derived from two traditions reported by Ibn Shihāb al-Zuhrī (d. 742) and preserved in the Ṣaḥīḥ of al-Bukhārī (d. 870). In the first, Abū Bakr is persuaded by ʿUmar to order a written compilation of the revelation so that it may be safely preserved (some reciters having died in battle). In the second, ʿUthmān does the same but in more public fashion, in response to reports of varying recitations among the Muslims. Both collections, the first to preserve, the second to standardize, were led by the Medinan Zayd ibn Thābit. Once ʿUthman’s version is completed, copies are sent to various cities of the burgeoning empire, and the order given that all other existing copies are to be destroyed. This ʿUthmānic text is subsequently understood to be the basis for the written Quran as we know it today. Despite its widespread acceptance by most Muslims and non-Muslims alike, there are good reasons to be suspicious of this version of events (convenient early involvement of the first three caliphs, typical narrative techniques of repetition, the state of the Arabic alphabet at the time, etc.).

Shoemaker proposes a very different version of the Quran’s written canonization:

Should there be any doubt what he intends by “composition”, Shoemaker tells us he means neither “the mere compilation of textual material that has already been fixed into a certain form” nor “mere cosmetic adjustments to an already fixed text, such as adding textual divisions or diacritical marks”.

While he does not exclude the possibility that some of the Quran may have its origins in words of Muhammad uttered in early seventh-century Hijaz, what must have happened, in his view, is that as the Muslim soldiers conquered the vast swathes of neighbouring territory, they came into contact with other religious traditions and eventually incorporated some of their newfound lore into the Arabian “teachings” of Muhammad. The supposed Quranic milieu of the seventh-century Hijaz was largely illiterate, isolated, and had little or no contact with monotheist traditions so clearly represented in the Quran, and so the book as we know it must have taken definitive shape elsewhere, namely Syria and Iraq.

Shoemaker is highly critical of previous scholars’ faith in the traditional narrative of the Quran’s collection. He puts this down repeatedly to a “commitment to defending the Nöldekean-Schwallian/Sunni paradigm” (passim) without considering the reasons behind it. He assumes that such a position must be due to apologetics, the desire not to offend Muslims, or to intellectual incuriosity and weakness. (I would venture that it is due to the paucity of alternatives and the fact that most Orientalists have been better philologists than historians.)

He begins, accordingly, with a robust statement of where he stands vis-à-vis his predecessors in Quranic and Islamic studies. He stresses his adherence to the vision of an “historian of religion” (p. 3). In a brief sketch of Western Islamic studies, he laments “the outsize influence of Wilfred Cantrell [sic] Smith”. According to Shoemaker, “Smith was able to direct the training and influence the methodological approach of ‘many, if not the majority, of Islamicists who held (and continue to hold) positions in religious studies departments in North America’” (pp. 3–4). The only sources cited for the history of Islamic studies are those of Aaron Hughes, also a critic of “Wilfrid [sic] Cantwell Smith” and his claim “that no statement about a religion is valid unless it can be acknowledged by that religion’s believers”.Footnote ⁶ Smith’s dictum is indeed hard to accept without qualification, but one wonders if it is meant to apply to the study of scriptural origins, as opposed to the lived practice of a religious tradition. Both subjects fall under the rubric of “religious studies”, “history of religions” and so on, but they are fundamentally different areas of study. Shoemaker, however, makes no such distinction: any uncritical reiteration of standard Muslim belief about Quranic origins is a failure. Or this is what I assume, as he notes no difference. In any case, one would think that if the influence of W.C. Smith is indeed so pervasive, his name would be less frequently misspelled.

Moreover, it is unusual for someone to be so critical of an academic field and yet write a book whose arguments are taken almost entirely, though not always accurately, from secondary sources. An accessible account in English of the sources and opinions on scriptural origins would be desirable, but Creating the Qur’an is not that book.

The first two chapters treat the earliest sources on the Quran’s formation, as well as some Orientalist forays into the subject. This is the core of the book, from which all else proceeds, and thus is the focus of our review. Time and space do not permit a full inventory of all questionable claims and outright errors, which are substantial.

Shoemaker criticizes first of all the supposed “unanimity” of attribution to ʿUthmān of the first collection of the Quran. He is both right and wrong here. The sources themselves are not unanimous: they give myriad variant versions, and moreover, they speak unembarrassedly of competing codices alongside that of ʿUthmān. There is, however, a consensus (though not exactly unanimous) among believers that the consonantal “vulgate” dates to the reign of ʿUthmān, a consensus largely followed by scholars. Shoemaker has confused unanimity of sources with a later consensus. No one is saying that the sources are unanimous, thus he seems to be arguing against a claim that nobody is seriously making, and diverting attention from the more pressing question of how to understand the sources we do possess and asking why is there a preference for ʿUthmān.

Shoemaker here takes to task Alford T. Welch for his Encyclopaedia of Islam (2nd ed.) article on the Quran for referring to, in Welch’s words, “the unanimity with which an official text is attributed to ʿUthmān”:

Shoemaker has misrepresented Welch. As the surrounding sentences (and the context) make abundantly clear, Welch refers to the unanimity of the later consensus, not of the sources themselves, whose divergences and contradictions he has just documented. Welch makes the clear distinction between believing that the Quranic codex as we know it dates to the reign of ʿUthmān and accepting the ʿUthmānic account of al-Zuhrī in all its details; Shoemaker makes no such distinction and thus gives a gravely distorted version of his predecessor’s work.Footnote ⁷ The same is unfortunately true of his very brief and unsatisfactory treatment of Nöldeke/Schwally (pp. 20–22). The German orientalists’ case may not have aged well, but it is more complex and critical than one would know from Shoemaker’s remarks.

Chapter 1 then takes us through some passages in early historical and biographical sources touching on the collection of the Quran. Virtually all the significant material here originally appeared in the work of Alfred-Louis de Prémare (1930–2006), who argued against the “canonical” version of events, pointing out a number of inconsistencies and curiosities in the record.Footnote ⁸ Shoemaker sometimes misrepresents de Prémare, and when he attempts his own readings of the Arabic passages treated by the French scholar, the results are, at best, unreliable. De Prémare was not always convincing, but he differed from Shoemaker in two crucial respects: he engaged with the texts themselves quite closely, and he was more moderate in his sceptical claims (at least in the two main works of 2002 and 2004; he seems to have become a bit more rigid in later publications).

Shoemaker’s résumé of the relevant passages in Ibn Shabba’s (d. 878) Tārīkh al-madīna al-munawwara bears little resemblance to what one finds in the text. He begins with “Perhaps the most shocking aspect of Ibn Shabba’s assemblage of reports concerning the Qur’an’s production is the complete absence of any memory of Abū Bakr’s involvement in the process” (p. 42). Yet there it is: “the codices that Abū Bakr ordered Zayd to collect” (al-maṣāḥif allatī kāna Abū Bakr raḍiya allāh ʿanhu amara bi-jamʿihā zaydan).Footnote ⁹ Even de Prémare noted this.Footnote ¹⁰

Overall, what Shoemaker provides is an insufficient paraphrase of de Prémare (Aux origines du Coran, 74–5). On ʿUmar, de Prémare is correct that the pages have nothing to do with the actual collection of the Quran, and that they indicate a plurality of codices in ʿUmar’s day, although we know nothing more about them, i.e. whether they were necessarily complete, etc. Ibn Shabba’s reports focus almost solely on ʿUmar attempting to establish the reading of Quraysh as the preferable one, and in one case permitting the Companion Ubayy ibn Kaʿb to teach and recite but insisting that when the Quran was written, it should be according to the reading of ʿUmar and Zayd (qirāʾat ʿUmar wa-Zayd).Footnote ¹¹ But there is nothing here to justify Shoemaker’s “we learn that what [ʿUmar] was actually engaged in was not so much the initial compilation of the Quran as he was trying to establish the authority of one among several already collected versions” (p. 25).

Similarly, the section on ʿUthmān and the Quran focuses overwhelmingly on establishing the correct pronunciation and establishing the linguistic credentials of Zayd and others involved in the project, as well as the order to destroy other codices and well-known reports concerning the missing surahs and linguistic infelicities. I am not sure how literally we should take them, but all accounts refer to the creation of a new and revised version and not, as Shoemaker has it, the promotion of one already existing over others.

He then turns to the “Kitāb al-ṭabaqāt al-kabīr (The book of the major classes)” [sic] of Ibn Saʿd (d. 845). Shoemaker states that there is no mention of Abū Bakr or ʿUthmān collecting the Quran (pp. 26, 27). This is not the case. De Prémare made the mistaken claim, apparently unaware that the older editions of Ibn Saʿd are missing sections on the Medinans. In the more complete edition, both caliphs play their expected roles in the biography of Zayd ibn Thābit.Footnote ¹² At the very least this information calls into question Shoemaker’s conclusion:

On Sayf ibn ʿUmar (d. 796–797) and his Kitāb al-ridda wal-futūḥ (“The Book of the Wars of Apostasy and the Conquests”), Shoemaker’s errors are particularly egregious. This is one of the oldest accounts of the ʿUthmānic Quran, and it generally conforms to the “canonical” version known from al-Zuhrī’s reports in al-Bukhārī’s Ṣaḥīḥ. Shoemaker refers only to the differences between the “codices”, while Sayf’s account begins, as does al-Zuhrī’s version, with Ḥudhayfa noting differences and disputes over recitation. Sayf gives an interesting list of different codices (maṣāhif) in circulation, but the emphasis throughout is on establishing the correct recitation. Contra Shoemaker, there is no mention of “significant differences among these first codices” (p. 29). Sayf does speak of differences, but not in codices, only in recitations, and the word used for any variations is qaraʾa/qirāʾa, consistent with the majority of reports on the reasons for creating an official codex. ʿUthmān did not summon “representatives” from the various regions, as Shoemaker would have it; he gathered together those Companions of the Prophet who, as the text states, were present in Medina.

He also claims that “we do not learn whether ʿUthmān’s efforts met with any success or if his codex was received in these centers as a replacement for their local versions”. Yet Sayf’s text says in no uncertain terms that the effort was approved of, that everyone saw the merit of what he had done, and that they agreed on it and abandoned their other versions, with the exception of the Kufans who stuck with the reading of Ibn Masʿūd (wa-baththa-hā ilā l-amṣār fa-kullu l-nās ʿarafa faḍla dhālika ajmaʿū ʿalayhi wa-tarakū mā siwāhu illā mā kāna min ahli l-kūfa … Footnote ¹³ ). Without judging the reliability of Sayf’s account, let us at least understand what he says.

Shoemaker seems to miss the main point of Sayf’s account and misunderstand the text: the Muslims in the garrison cities learned the Quran at great remove from the Prophet himself, and so the report emphasizes the superior readings of those who learned the Quran from Muhammad in Medina, over those who learned it far away in Syria or Iraq. For example, the muṣḥaf of Miqdād was gathered in Syria, where people had not recited with the Prophet; Ibn Masʿūd, though based in Kufa, had recited with two Companions, then “collected the Quran in Kufa and wrote it in a codex” (jamaʿa l-qurʾān bil-kūfa wa-kataba muṣḥafan).Footnote ¹⁴ Thus ʿUthmān “had codices copied while he was in Medina along with those who had recited the Quran with the Prophet himself”. This is not, as Shoemaker would have it, “certain unspecified codices copied in Medina, presumably on the basis of yet another version of the text in use there”, but an effort to create an authoritative version resembling (supposedly) how Muhammad himself would have recited it.

Sayf has fa-ktataba l-maṣāḥif wa-huwa bil-madīna wa-fīhā lladhīna qaraʾū l-qurʾān ʿalā l-nabī. Following the thread and logic of the anecdote, I would understand this to mean that he had copies made that followed the reading of those who had learned from the Prophet. De Prémare thought it meant “recopying copies that already existed in Medina” (de Prémare, Aux origines du Coran, 83), but there is no reference to any specifically Medinan muṣḥaf, only to the recitation of Muhammad.

A major difficulty is that we do not know what is meant by “collect”. The Arabic jamaʿa sometimes means to collect or gather, but also “to memorize”, with respect to the Quran. The complexity of the issue has been well examined by Claude Gilliot, but receives virtually no attention from Shoemaker.Footnote ¹⁵

Our sense of what is meant by “collect” is shaped by the Zuhrī anecdotes, which describe a careful process of gathering material on which verses had been inscribed, as well as those remembered aurally, and of verification. Yet at the end, we read of the command that all other, non-ʿUthmānic, codices be destroyed. The obvious question, then, is how were those other codices compiled? Did each one require a similarly lengthy process of collecting bones and palm leaves, of interviews and authentication? We are in the realm of speculation here, but given the number of maṣāḥif mentioned in the sources, this seems unlikely.

The vast majority of reports about Quranic origins are concerned not with the gathering up of the revelation (the primary interest of many Western scholars) but with the establishment of the correct recitation. Reports of the former are in fact conspicuously rare. This gives one to think that people either no longer know how the Quran came to be written down in single codices, or they knew and took it for granted, finding it of little interest.Footnote ¹⁶ Sources such as Ibn Shabba and the others discussed here certainly describe a proliferation of codices.Footnote ¹⁷ Admittedly none of this is certain, but it seems plausible that the majority of what became the Quran was already well established in some written form at a very early date, probably during the lifetime of the Messenger. I suspect that the references to individual surahs (e.g. yā aṣḥāb sūrat al-baqara!) are a clue to this hidden history, but it is hard to say more.

So when Sayf tells us that the Companions outside Arabia “collected” the Quran, e.g. jamaʿa l-qurʾān bil-kūfa fa-kataba muṣḥafan … jamaʿa l-qurʾān bil-shām … (Sayf, 50), it is unlikely that they were starting from scratch (if, in any case, we are to accept the account at face value).

Our sources contain any number of accounts of things being added to or omitted from the ʿUthmānic Quran. Shoemaker strangely mentions very few. His arguments are based not on what the sources say, but on speculation about what they do not say: “Although such reports about the variations of these codices that have come down to us suggest only relatively minor differences from the canonical text, there is no reason to assume that this was in fact the case” (p. 33).

As Shoemaker warms to this theme, he again follows de Prémare, citing Arthur Jeffrey citing the exegete Abū Ḥayyān al-Gharnāṭī (d. 1344), who said that he had deliberately omitted “those variants where there is too wide a divergence from the standard text of ʿUthmān” (p. 33).Footnote ¹⁸ Shoemaker and de Prémare wish to see this as evidence for a vast variety of variants indicating substantial modification of the text. In fact the reference serves to support the traditional version of Quranic formation. Abū Ḥayyān states clearly that he refers only to readings at variance with the rasm, the skeletal text, although he uses the word sawād, in the sense of “black ink [of the text]”. The passage, which refers to a variant reading of Q 34: 14, is as follows:

There is nothing here to suggest anything other than conventional divergences from the consonantal skeleton of the ʿUthmānic text, which Muslim tradition has abundantly documented. Of course, it does not explicitly exclude major deviations of the kind envisaged by Shoemaker, but if you have read much of this material, I think you would find it hard to believe that major textual restructuring would be referred to as “deviating from the black ink”. Our exegetes were not so coy. Later Muslim scholarship has preserved abundant references to minor variations in the text and its recitation, often in painstaking if not mind-numbing detail. Are we to understand that either these accounts of minor variants are fabrications meant to lead us off the scent of the truly enormous changes that really occurred, or that all traces of those major revisions have been effaced, leaving only the minor ones? Shoemaker has nothing to say about this and seems unaware of the level of deception his theory demands.

In wanting to see evidence of significant changes in the Quran between proclamation and canonization, Shoemaker rightly notes that the distinction between different modes of divine communication was not initially clear. He is again following de Prémare (Les fondations de l’islam, 317–21), who discusses the word ḥadīth and its different usages, in the Quran and elsewhere, in an effort to demonstrate the ambiguity of the term and its relation to the Quran and its discourses (e.g. that the Quran is “the most beautiful ḥadīth” (Q 39: 23)). He also mentions the ḥadīth qudsī as well as some examples of variant readings and cases where Muhammad’s own reported words were very close to those of the Quran. De Prémare gives an interesting discussion of semantic range and potential ambiguity, but Shoemaker reduces it to a simple confusion between divine and prophetic authority. Citing Mohammed Ali Amir-Moezzi, Shoemaker writes: “Ibn Saʿd transmits a claim by [Salama ibn al-Jarmī] that he had collected ‘many qur’ans,’ from Muhammad, presumably meaning by this many of what the later tradition would regard as hadith” (p. 30). The term is qurʾānan kathīran, and I am not sure why we should presume this means “hadith”. Amir-Moezzi admits that the phrase may well mean to “collect passages of the Quran from the mouth of Muhammad”.Footnote ²⁰ This latter instance is perfectly consistent with early usage of the word qurʾān.

Shoemaker takes a reductive view: “So it is not at all clear what sort of distinctions would or even could have been made at the time of Abū Bakr or ʿUmar or ʿUthmān as to what should be collected and authorized as the divine revelation of the Qur’an, on the one hand, and the teachings of Muhammad, the hadith, on the other” (p. 30).

The distinctions are evident to anyone who has read much Quran and hadith: they represent totally different styles, content, and language. There could have been and most likely was some slippage between the two, as the handful of well-known examples indicates. Nonetheless, the distinction was ultimately pretty clear, and the number of confused cases remains very small. Fred Donner argued this case persuasively.Footnote ²¹ One might add that pioneering studies on the original meaning of qurʾān and on the nature of the ḥadīth qudsī were penned by William A. Graham, whose conclusions certainly diverged from Muslim tradition/orthodoxy, despite his being a prominent student of the nefarious W.C. Smith.Footnote ²²

Shoemaker claims that an ʿUthmānic canonization is “something that the history of religions repeatedly informs us is an extremely unlikely set of events” (p. 38). Piecing together his scattered remarks, we infer that because the Christian case followed a certain trajectory, the Islamic case should follow suit (e.g. 22, 31–32, 207–08). He notes that biblical scholars reject the attribution of the New Testament books to anyone really named Matthew, Mark, Luke, or John, and that it was not until well into the second century that early Christians began to think that they needed more precise information about Jesus’ teachings.

Furthermore, since certain evangelical Christians make claims that Jesus’ followers carried notebooks and thus preserved his words faithfully, Muslims no doubt did the same. The Islamic case must be assumed to follow the Christian model, unless proven otherwise. He then repeatedly tells us that for the historian of religion, it is impossible that the Quran’s form was fixed by the mid-seventh century:

I am not sure that anybody is claiming “a fully accurate remembrance” and, moreover, I have little doubt that the Islamic sources do reflect some of these same tendencies. The question is how and to what degree, and what do our sources tell us, directly or indirectly?Footnote ²³ The gospel experience is fine as a point of departure, but more consideration of the specifics is necessary for a conclusion.

Shoemaker takes the Christian example (he gives no other) as the template, and insists that the Islamic model conform. Of the abundant references to writing down the revelation in Muhammad’s lifetime, he is dismissive (“shockingly modest”, “based more on conviction and assertion than on proof and argument” (p. 21)). I am sure I am not alone in thinking that such references deserve more consideration, which one would expect from a work sub-titled A Historical-Critical Study.

Chapter 2 attempts a case for a Quranic canonization around the turn of the eighth century, at the impetus of ʿAbd al-Malik, with his governor al-Ḥajjāj playing a major role. Shoemaker again follows de Prémare closely but incautiously.

According to Shoemaker, scholars have long neglected the well-established reports attributing to ʿAbd al-Malik the “standardization” of the Quranic text as we have it today. It is true that recent decades have seen a greater interest in these reports, but the same could be said of almost anything in the field. (Moreover, the force of Shoemaker’s criticisms of current scholarship is weakened by his reliance on that very scholarship.)

In any case, the interventions during the reign of ʿAbd al-Malik are no secret: even Bell and Watt’s Introduction to the Qur’an mentions them. There are numerous references to the copying or production of a new version of the Quran, usually with reference to improved orthography, especially the addition of diacritics, but of other features as well. It is also mentioned that copies were made and sent out to the major cities, just as was allegedly done with the ʿUthmānic codex. It is by no means clear exactly what modifications were made, but all references are to the orthographic.

Shoemaker’s approach certainly grabs the attention:

To take the latter claim first, that manuscript evidence gives the lie to reports of “minor improvements”, Shoemaker’s reference here is to de Prémare’s discussion of ʿAbd al-Malik (Les fondations de l’islam, 294–6). The French scholar (in the pages cited) first of all admits that we don’t know of what ʿAbd al-Malik’s reforms consisted, nor do we know to whom should be attributed the development of diacritics and the like. He then makes the erroneous claim that the earliest quranic manuscripts lack all diacritical punctuation or vocalization. The relation to Shoemaker’s claim is unclear.

Shoemaker then argues that since we know that certain reforms said to have been carried out under al-Ḥajjāj were already present in pre-Ḥajjāj MSS and absent from some post-Ḥajjāj manuscripts (citing François Déroche), then the attribution of such reforms to the Umayyad governor cannot be correct: Ḥajjāj’s interventions then must have been much more substantial. Déroche is more circumspect in the passage referenced. He says that the Ḥajjāj project aimed to produce only an “improved muṣḥaf”, with mainly orthographic reforms. He notes that since some earlier manuscripts contained such reforms, the reports on the Ḥajjāj “second maṣāḥif project” must be taken critically, but nonetheless the steps described do “echo quite directly what has been observed in the manuscripts”.Footnote ²⁴ I do not see how this opinion can possibly “demonstrate unambiguously [the] falsehoods” claimed by Shoemaker. The manuscript studies indicate that various diacritics and vocalizations were employed, though not uniformly, through the Umayyad period, so the attempt to impose a standard system hardly precludes the pre-existence of some diacritics, and the fact that some later ones lack certain features is also hardly surprising. In any case, Shoemaker provides no examples, and the references he cites in support (Déroche, Éléonore Cellard) all refer uniquely to the orthographic and structural, and even state explicitly that the scope of changes is limited.Footnote ²⁵ In short: there is no evidence whatsoever in the manuscripts to indicate Shoemaker’s “significant interventions”.

As for actual reports of “much more significant interventions in the text”, Shoemaker is on similarly shaky ground. First, he declines to give us a full accounting, referring us for details to de Prémare and Omar Hamdan, leaving the work of describing the quranic interventions of Abd al-Malik and al-Ḥajjāj, the central focus of his thesis, to others.Footnote ²⁶ At this point the subtitle, A Historical-Critical Study, should have been rethought.

The first piece of supposed evidence for a Marwānid Quran is ʿAbd al-Malik’s statement, “In the month of Ramaḍān, jamaʿtu l-qurʾān”, which could mean “I collected” or “I memorized” it. It is highly unlikely that this statement is a declaration of a major state achievement of which all other traces have been effaced (and which, in any case, in Shoemaker’s view, was to be attributed to ʿUthmān, not ʿAbd al-Malik). It is much more likely a reminiscence of memorization.Footnote ²⁷

Shoemaker cites only one Muslim source that might appear to support his argument in any sort of way, the hadith found in Ṣaḥīḥ Muslim (and abbreviated in Bukhārī), in which al-Ḥajjāj is preaching from the pulpit (wa-huwa yakhṭubu ʿalà l-minbar):

According to Shoemaker, “al-Ḥajjāj appears to be addressing a group of scribes and scholars whom he has charged with composing what will be the new standard version of the Quran to replace the regional codices” (p. 49). A huge speculative leap is necessary to go from all other sources on Ḥajjāj’s reforms to such a statement. Moreover, what conclusions can one legitimately draw from such a puzzling passage?

How to understand allifū? To collect or compose, to order, would be the usual meanings, but such a choice would solve one mystery at the price of creating more. Nicolai Sinai’s alternative of “recite” makes more sense, but is not entirely satisfactory either.Footnote ²⁸ And what would it mean for Gabriel to allafa the Quran? The answer is hardly obvious. If al-Ḥajjāj was instructing “scribes and scholars” on making a new muṣḥaf, why do it from the pulpit? And why on earth would al-Ḥajjāj, of all people, give the surah order of Ibn Masʿūd’s codex, to which he was unambiguously opposed?

“One should not be confused here”, writes Shoemaker, “by the use of the word sūra, since this term has a broad meaning, even and especially in the Qur’an, where it simply designates a writing of some sort” (p. 49). This is an awkward paraphrase of de Prémare and Guillaume Dye, who proposed that the word in the Quran may not mean surah in the post-quranic sense, which is not at all the same as claiming “a broad meaning”.Footnote ²⁹

I do not know how to understand the passage, but Shoemaker is uninhibited by doubt: “Clearly, some sort of deliberate composition of the Qur’an under al-Ḥajjāj’s direction is in view in this report – presumably, a synthesis of the earlier regional codices”. The best one can say is that this is not impossible.

More to the point, Shoemaker states repeatedly that “numerous reports from the early Islamic tradition indicate that the changes to the Qur’anic text introduced at the direction of ʿAbd al-Malik and al-Ḥajjāj were in fact substantial” (p. 49), and yet this obscure passage is the only such evidence he brings, and it can hardly count as such. On the abundant sources we do possess, Shoemaker is silent.

All the more surprising, then, is that when there is an example of a writer describing major additions to the Vulgate, “a significant intervention”, Shoemaker stops short of mentioning it. He cites the Risālat al-Kindī, a Christian apologetic tract of the early ninth century. The author shows himself aware of many Islamic traditions regarding the scripture, including the interventions by al-Ḥajjāj. Shoemaker gives us al-Kindī’s words: “Then there was the intervention by al-Ḥajjāj ibn Yūsuf, who left no copy [muṣḥaf] that he did not acquire, and he removed many things from it” (p. 56). But he leaves out what follows, for al-Kindī’s text continues, “and he added things to it. They say that these included verses revealed concerning the Umayyads and the Abbasids, and they included names”.Footnote ³⁰ If Shoemaker feels that al-Kindī is a reliable source, why not mention this? Presumably because it is not the kind of intervention he is arguing for, viz. monotheistic lore added on to an Arabian substrate during the first generations’ occupation of Syria and Iraq. Excised references to Umayyads and Abbasids would recall certain Shiite claims, and there are many examples of alleged sectarian doctoring of the revelation, but Shoemaker cites none of them, either. The “significant interventions” he claims took place are confirmed nowhere – even in those cases that might suggest “significant intervention”, it is nothing like what he proposes.

The waters are further muddied by claiming support from other scholars. Most would now agree, I think, that the early history of the Quranic text and its recitation is more complicated than commonly thought, and contains more variety than any introductory textbook would admit. But this is very different from Shoemaker’s claim of major substantial revisions at the time of ʿAbd al-Malik.

A perusal of some of those “other scholars” indicates otherwise. Shoemaker frequently cites Chase Robinson in support of his Marwānid Quran position, but avoids a crucial sentence: “The scope of ʿAbd al-Malik’s Qur’anic project is impossible to measure, but we must envision it as one of editing and revising, rather than composing”.Footnote ³¹ Similarly, Fred Donner: “although we know relatively little about it, ʿAbd al-Malik ordered his governor in Iraq, Hajjaj, to prepare a re-edition of the Qur’an text that was provided for the first time with clear vowel and diacritical markings to ensure proper reading and recitation of the text”.Footnote ³² Nor does the work of François Déroche support Shoemaker’s claims. Most impressive perhaps is when he takes Jonathan Brockopp’s résumé of Wansbrough’s theories and states that Brockopp is arguing the same thing, even though Brockopp’s following sentence reads, “This thesis has been largely refuted …” (p. 210).Footnote ³³

The subsequent chapters take up various aspects of early Islamic and late antique history and historiography, such as the reliability of carbon dating, and the commercial significance of Mecca before Islam, as well as its natural environment, all drawing heavily on the work of Patricia Crone and others. Chapter 5, on literacy in the Hijaz, suffers from faulty analogies, bibliographic gaps, and a misunderstanding of “literacy” when applied to nomads and town-dwellers. Chapters 6–7 are interesting in that they treat “memory science”, which emphasizes the fallibility of human memory, with all indications that we are much worse at remembering than we think. The implication here is that there is no way that Muhammad’s proclamations could have been accurately preserved even for a short time, let alone for decades, before being written down. The only cases where accurate memorization can occur is with some degree of written support.

These chapters (6–7) do provide a good reminder that the traditional version of strictly oral preservation of the revelation is probably untenable, but they also point us in a direction quite different from Shoemaker’s arguments. If whatever passed for the Quran in Muhammad’s day and the following generation could not have been preserved accurately by mere oral transmission, we have a choice: we may consider it to have been subject to drastic changes and modifications until codified many decades later, or we might consider that there was indeed some writing going on and transmission was overall reliable. Which argument is better supported by available evidence? If the Quran varied so drastically in form and content over the decades, then why does all the evidence show, very consistently, relatively minor variations?Footnote ³⁴ How to explain the extent of the cover-up? Moreover, interesting as the 50 pages on memory science are, it would have been beneficial to have something about how this applies to the Quran, rather than, say, the detailed account of John Dean’s Watergate testimony. Here and elsewhere, Shoemaker remains at armchair’s length from the sources, not least the Quran itself, never explaining (or even vaguely suggesting) how the process he argues for would be manifest in the text.Footnote ³⁵

Similarly for ch. 8, “The qur’anic codex as process”, Shoemaker gives lengthy and not uninteresting examples from scholarship on early Christianity. I have little doubt that some of the material he cites here is of interest to Islamicists, but that would hardly be the end of the story: one still has to match theory, paradigm or example from other traditions with the available evidence for the Islamic case. Shoemaker proceeds directly from theory to conclusion, with the assumption that what happened in the Christian case must also be so for Islam. He skips the most difficult step of comparing evidence. The only references to the Quran in this chapter are two examples of parallel passages that could show signs of being derived from common sources (passages on the prophet Shuʿayb and descriptions of paradise in Q 55); these are the same examples given by Wansbrough, with much more substantial commentary, in Quranic Studies (and discussed more recently by Devin Stewart).Footnote ³⁶

This refusal to engage the sources is one of several frustrating aspects of Creating the Qur’an. (A more apposite title would have been Avoiding the Qur’an.) Two principal methodological weaknesses are evident. First, if one looks solely at the traditions involving the Quran, it is indeed hard not to be sceptical about the veracity of the traditional narrative of canonization. This scepticism is tempered but by no means eliminated by reading more widely, by getting a sense of how opinions are conveyed, of how countless other references indirectly support some version of tradition, and how the collection of the Quran is but one of many topics on which the early sources present contradictory versions of events. Shoemaker’s view, such as it is, is far too narrow for the material involved, and the energy and enthusiasm he brings to secondary sources could have been better applied to the primary texts.

Second, for all his zeal for iconoclasm and revisionism, it is Shoemaker who clings to traditional Muslim notions of oral transmission and late antique Arabia as an illiterate den of polytheism and idolatry. Current scholarship is suggesting that both literacy and monotheism were much more widespread than previously thought. It is still too early to say anything definitive, but if this turns out to be correct, it would resolve a number of issues, bringing us closer to understanding how a text like the Quran could have appeared in the Hijaz. (Shoemaker is highly critical of historical linguists’ standards of evidence, but they are higher than anything he provides himself.)

One should not rush to conclusions on the basis of preliminary findings, but if the Hijaz was more monotheist and more literate than previously thought, we would have little to revise, since we don’t know that much about pre-Islamic Arabia anyway. Shoemaker’s scenario, in contrast, requires revising our view of the Islamic period, for which we do have abundant sources that then must have been falsified. It is true that to posit greater literacy or widespread monotheism also requires that that the earliest sources left out or covered up something. But the early Muslims had a vested interest in playing up the differences between the Jāhiliyya and Islam. The degree of cover-up and revision demanded by Shoemaker is infinitely greater than that required by the assumption of greater literacy and more monotheism. Moreover, the latter view requires only that certain aspects of the past were not mentioned or downplayed; the Shoemaker scenario requires wholesale fabrication on a scale that is hard to consider, let alone accept. It involves far more than a few anecdotes about Abū Bakr and ʿUthmān.

This is pointedly clear in Shoemaker’s Conclusion, which has an epigraph from al-Suyūṭī’s Itqān: “The Qur’an was revealed in three places: Mecca, Medina, and Syria” (p. 259). As it follows an argument for a revelation somewhere north of Arabia, one might imagine that this is supposed to be the final nail in the coffin of the Hijāzī Quran, but Shoemaker adds only this:

The citation and comment are typical of the main problems with Creating the Qur’an. First, the citation means nothing like what he implies, which is obvious from the context. It is found in a section on where the Quran was revealed, and various reports state Mecca, Medina, or elsewhere – wherever Muhammad happened to be. Here, it says the Quran was revealed in Mecca, Medina, and al-Shām (“Syria”), which can refer to points north of the two holy cities.Footnote ³⁷ (Moreover, Shoemaker does not mention it, but the words are attributed to the Prophet, and in his view the Syrian-Mesopotamian additions came decades after the Messenger’s demise, so it is not even clear how the saying should fit his argument.)

Second, Shoemaker feels no need to explain how such clear evidence for his revisionist narrative could have gone unnoticed. Would it never be questioned by the thousands of scholars who read it? Has Shoemaker stumbled on something that generations of Orientalists and centuries of Muslim savants have missed? Such a major discovery on his part would need a bit more support and explanation. I am not saying this couldn’t happen, only that it would require more effort. As it stands, the plucking of single reports or sentences out of context does not suffice.

The topics treated in this book are interesting and important, for there remains much we do not know, and much that is frankly puzzling about what we think we know. It is all the more unfortunate that these questions receive here such unsatisfactory treatment. Those looking for an introduction to the “sceptical” or “revisionist” side of Quranic studies are better advised to consult the two lengthy articles by Guillaume Dye in the first volume of Amir-Moezzi and Dye (eds), Le Coran des historiens (2019): “Le corpus coranique: contexte et composition” and “Le corpus coranique: questions autour de sa canonisation”.Footnote ³⁸ In English, one may refer to Nicolai Sinai’s two-part article, “When did the consonantal skeleton of the Qur’an reach closure?”, Footnote ³⁹ which reaches conclusions quite opposed to those of Shoemaker, but nonetheless provides a more nuanced and even-handed account of the issues and opinions involved.