Evidence in favour of a mid-seventh-century closure of the Quran
I now turn to arguments that can be adduced in support of the proposition that the Quran's standard rasm reached closure by c. 650. The results of Sadeghi and Bergmann's radio carbon dating of the Ṣanʿā' palimpsest are certainly relevant here but, as pointed out above, they do not as such preclude the possibility that the full standard rasm of the Quran might only have emerged in the second half of the seventh century. Furthermore, the radio carbon dating of Quranic manuscripts is still in its infancy and has been known to produce anomalies,Footnote 2 so additional, albeit less straightforwardly scientific, considerations are certainly not superfluous.
Unanimous ascription of the standard rasm to ʿUthmān
The emergent canon model entails that traditions about ʿUthmān's promulgation of a standardized consonantal skeleton of the Quran can only have started to circulate after the assumed closure of the standard rasm, i.e. after 700.Footnote 3 Yet the authoritative consonantal skeleton of the Quran is unanimously traced back to ʿUthmān, not just by the Sunni tradition but also by other Islamic groups such as the Khārijites and the Shiites. Any attempt to reconcile these two things is faced with a double challenge. First, is it historically credible to suppose that a pan-Islamic consensus about the canonical version of the Quranic text could have formed at a time when the Islamic community had already spread across large swathes of territory from Spain to Iran and had split into several mutually hostile groups? It appears unlikely that a caliph like ʿAbd al-Malik could have coerced his various adversaries, some of whom did not hesitate to take up arms, to adopt his version of scripture plus its attribution to ʿUthmān. A defender of the emergent canon model might respond by suggesting that most Muslims were so swiftly won over to the newly arrived Marwanid canon that even dissident groups like the proto-Shiites could not escape the pull exerted by the majority text and the legend of origins that went with it. I would concede that such a picture, although perhaps surprising,Footnote 4 is not downright impossible, especially if one accepts that the primary medium in which the Quran was present to early Muslims would have been the recitation of brief passages from memory and for ritual purposes, rather than a systematic and frequent consultation of complete manuscripts.
Yet even if the challenge of painting a credible historical picture for the spread of the standard rasm is met (and the preceding remarks only intimate a possible starting point), a second query looms: how is it that the literary tradition displays no palpable vestiges of the true origin of the standard rasm? If the final redaction of the Quran had only taken place around 700 or later, rather than under ʿUthmān, should we not expect some echo of this to survive at least in Shii or Khārijī sources, which are not beholden to the mainstream Sunni view of early Islamic history?Footnote 5 The argument is one from silence, to be sure, and it hinges on the crucial premise that the Islamic literary tradition can be trusted to preserve the full range of people's views, at least about major public events, that was extant at the beginning of the eighth century.Footnote 6 Is such confidence justified? After all, the fact that the Exodus and the Israelite conquest of Canaan would have been highly significant public events has not prevented Biblical scholars from raising very weighty doubts that they ever occurred. Given the diversity of political and doctrinal viewpoints expressed in early Islamic literature, one may reasonably insist that the Islamic historical tradition is of a different kind from the ancient Israelite one: it does not necessarily give the impression of having gone through some bottleneck in the first half of the eighth century that was sufficiently narrow to explain the obliteration of virtually all traces of how the canonical rasm of the Quran really originated and spread. Nevertheless, the assumption that the Islamic literary tradition does record the entire spectrum of opinions about major public events that was extant in, say, 710, when people would presumably still have been aware that the so-called ʿUthmānic text was a recent arrival, certainly remains debatable.Footnote 7 Generally speaking, the higher the salvation-historical significance of an event, the less it seems impossible that the majority view of that event could have completely drowned out contrary perspectives.
It is therefore worthwhile to present and flesh out a train of thought briefly sketched by Gregor Schoeler, which in my view considerably strengthens the unanimity argument.Footnote 8 Schoeler observes that ʿUthmān's enforcement of a standardized consonantal skeleton of the Quran comes across as deeply controversial in our sources. For instance, al-Ṭabarī and al-Balādhurī have him defend himself against the charge of having reduced the Quran to a single book (viz., a single recension)Footnote 9 or even of having “burnt the book of God” (viz., rival recensions thereof),Footnote 10 and Sayf ibn ʿUmar transmits a speech by ʿAlī defending ʿUthmān against the invective “burner of the codices” (ḥarrāq al-maṣāḥif).Footnote 11 The Khārijite tradition, too, condemns him for the burning of Quranic codices.Footnote 12 Schoeler concedes that statements vindicating ʿUthmān's measures are likely to be apologetic fabrications. But the very fact that people took the trouble to fabricate them would presuppose that the accusation which they are designed to dispel – namely, that ʿUthmān had “burnt the book of God” – was very much in the air. In other words, we are dealing with a genuine controversy: it is highly unlikely that the hypothetical originators of the legend describing ʿUthmān's promulgation of a standard text of the Quran would at the same time also have circulated accounts defending ʿUthmān's actions against accusations – and rather harsh ones at that – which had not yet been voiced.Footnote 13 But if we are dealing with a genuine controversy about ʿUthmān's standardization, the latter has a very good claim to being a historical fact rather than an Umayyad fantasy.
As so often, different stories could be constructed around the literary data. For instance, assuming that the belief in ʿUthmān's promulgation of the canonical rasm only emerged in the second half of the seventh century or later, the above traditions might be the fallout of common curiosity: at some point, everybody had come to believe that ʿUthmān had standardized the Quran, and it was also known that he had been unpopular towards the end of his reign (he had been murdered, after all). Against this background, people may have speculated with which particular wrongs his enemies could have reproached him, and may have surmised that having “burnt the book of God” would surely have featured on the list. Umayyad loyalists might then have reacted with traditions designed to clear ʿUthmān of the charge. Or perhaps, after the legend of ʿUthmān's standardization had been successfully launched, people might have increasingly realized that there was something inherently problematic about the burning of rival codices (as these were, after all, copies of the Quran), which then inspired the addition of an appropriate passage to ʿUthmān's apology. However, in the absence of concrete textual support such storylines seem unnecessarily convoluted. What is by far the easiest explanation is surely to suppose that ʿUthmān did indeed officially endorse a recension of the Quran and destroy competing copies, and that this measure caused him to be vilified in certain quarters as the “burner of the codices”.
There are nonetheless two things that bear emphasizing. First, the fact that ʿUthmān propagated a standardized version of the Quran does not as such require that the Quranic standard rasm did not undergo any reshaping afterwards. Second, and more importantly, prudence requires us to suspend judgement on everything that goes beyond the rather limited factual core identified above: whether ʿUthmān's measures were an attempt to suppress quarrels about the correct reading of the Quran, whether it was Ḥudhayfa who brought the matter to his attention, and whether the recension endorsed by him was a faithful transcript of leaves that were in the possession of Ḥafṣa and had been compiled during the reign of Abū Bakr cannot be reliably ascertained. For all we know, the full narrative about the promulgation of the ʿUthmānic text could be teeming with later expansions, accretions and embellishments. This possibility is augmented by the fact that al-Zuhrī, the common link of the traditions about ʿUthmān's standardization initiative, may legitimately be suspected of having been susceptible to the exigencies of Umayyad “state expediency” (Goldziher).Footnote 14
Text-critical arguments
As pointed out in Part I, Sadeghi has underscored that the fact that the lower writing of Ṣanʿā' 1 was erased in order to make room for the standard version of the Quran does not require that the text-type attested by the lower layer (C-1) predates the standard rasm. He then presents evidence suggesting that the common prototype from which the two recensions are descended is actually preserved more faithfully by the canonical rasm than by C-1. His most impressive argument to this effect consists of a detailed study of the major non-orthographic rasm variants contained in the lower text of nine pages of Ṣanʿā' 1.Footnote 15 According to Sadeghi, these are generally more likely to have arisen from the canonical rasm, or from a prototype that corresponded to it more closely than to C-1, than vice versa. This assessment turns on the assumption that copyists of early Quranic manuscripts, who Sadeghi argues were working from dictation, would have been more likely to drop brief textual segments than to add them, unless an addition can be accounted for as originating from an inadvertent assimilation of the verse in question to a similar or neighbouring one.Footnote 16 Thus, when confronted with two variant readings, XY and X, for a given verse, and excluding the existence of parallel or nearby verses displaying the wording XY, Sadeghi would ceteris paribus deem a development XY > X to be somewhat more likely than one leading from X to XY (and call XY an “irreducible plus”).Footnote 17 This assumption is based on the fact that omissions may be attributed to straightforward scribal mistakes, whereas additions not caused by accidental assimilation are more likely to have been deliberate;Footnote 18 and while Sadeghi does not rule out that conscious expansions may have occurred, he does posit that simple errors would have been more frequent. Against this background, Sadeghi has found that of the fourteen major pluses displayed by the standard rasm in comparison with C-1 at least three are irreducible.Footnote 19 By contrast, none of the nine major pluses displayed by C-1 in comparison with the standard rasm are irreducible, meaning that if the standard rasm is assumed to be primary, the wording of C-1 could have arisen from the standard text by means of widespread accidents of transmission, rather than as a result of less frequent kinds of alteration. This, according to Sadeghi, creates a certain presumption that the wording of the standard rasm is older than that of the lower layer of the palimpsest.Footnote 20
Sadeghi's study, a full appraisal of which is beyond the scope of this article, certainly marks a major advance: for the first time, sophisticated methods of textual criticism have been applied to the Quran, thus opening up promising avenues for future research, irrespective of whether an analysis of the remainder of the palimpsest's lower writing will bear out the trend detected by Sadeghi.Footnote 21 His tentative conclusion that the wording of the standard rasm seems to be older than that of C-1 also coheres well with the conclusion of the preceding section. Nevertheless, a few issues that deserve further discussion may be flagged up. Above all, it is important to state how much more probable an evolution XY > X has to be than its reverse in order to make us confident that the fact that manuscript A exhibits a given surplus of “irreducible pluses” over manuscript B really indicates that A is older. One must also note, as do Sadeghi and Goudarzi, that different parts of a manuscript could belong to different textual families whose value would have to be assessed separatelyFootnote 22 (a Biblical case in point being the Codex Alexandrinus). Finally, even if a default presumption in favour of the standard rasm being better than that of the lower layer of Ṣanʿā' 1 were established, this obviously does not exclude (and is not taken to exclude by Sadeghi) that particular passages could have undergone subsequent revision and expansion, although this possibility is likely to shrink further – or be confirmed – as the remainder of Ṣanʿā' 1's lower writing is published.
Internal features of the Quran (i): lack of fit with post-650 Islamic history
While the preceding sections have focused on extra-Quranic literary sources and on manuscript evidence, the strength of the emergent canon model can also be probed by examining internal characteristics of the standard rasm itself. Such an inquiry proceeds counterfactually, i.e. it involves the thought experiment of trying to hypothesize what kind of document we might have expected the presumed editors of the Quran to have produced if they had been active until around 700, and then checking the result against the kind of text with which we are in fact confronted.
Perhaps the most popular argument of this kind is Fred Donner's observation that in the Quran “we find not a single reference to events, personalities, groups or issues that clearly belong to periods after the time of Muḥammad – ʿAbbāsids, Umayyads, Zubayrids, ʿAlids, the dispute over free will, the dispute over tax revenues and conversion, tribal rivalries, conquests etc”.Footnote 23 Shoemaker has attempted to parry this line of reasoning by contending that the Quran's lack of allusions to later Islamic history “may simply reflect the fact that the Quran is generally not a predictive text”.Footnote 24 As a result, the early Muslims who potentially continued to shape the Quranic corpus after Muḥammad's death may not have been tempted to insert into it vaticinations of later events, and the absence of such foretellings does not prove that no posthumous editing occurred. Indeed, given the general scarcity of names and dates in the Quran, it is far from obvious that we would be entitled to expect an explicit mention of, say, the counter-caliph ʿAbdallāh ibn al-Zubayr. On the other hand, the question at stake is not so much whether the Quran contains or does not contain anachronisms in the strict sense but whether we can detect in it concerns that are best understood as those of editors active in the second half of the seventh century rather than those of the Meccan and Medinan Urgemeinde. If the Quranic rasm did not reach closure until c. 700, it does seem odd that it should nowhere engage with the major developments that defined Islamic history between 630 and 700, in particular the unprecedented speed with which an alliance of “barbarian” tribes from the fringes of the Byzantine and Sasanian empires established themselves as the masters of an immense territory, and the bitter disputes and civil wars that soon wreaked havoc on the unity of the conquerors.
Shoemaker also demurs that following Donner's logic, “one could similarly make the argument that the Christian Gospel according to John, which does not assign any predictions to Jesus beyond his own lifespan (or a few days thereafter), must accurately reflect his life and teaching and date to sometime before 60 ce”,Footnote 25 whereas the majority of scholars would of course date the text three or four decades later. Yet even the Gospel of John occasionally gives away the time of its composition. For example, the story of Jesus' healing of a man who had been blind from birth concludes with the statement that “the Jews had already agreed that anyone who confessed Jesus to be the Messiah would be put out of the synagogue” (John 9:22). As Bart Ehrman comments, “we know that there was no official policy against accepting Jesus (or anyone else) as messiah during his lifetime. On the other hand, some Jewish synagogues evidently did begin to exclude members who believed in Jesus' messiahship toward the end of the first century. So the story … reflects the experience of the later community that stood behind the Fourth Gospel”.Footnote 26 Hence, the argument that if the Quran had been an open text until the second half of the seventh century then, like other ancient writings, it somehow ought to reflect the historical context from which it supposedly emerged (albeit not necessarily by virtue of explicit name-dropping) still stands. As long as scholars have not managed to demonstrate that certain Quranic passages – and preferably, passages with a distinct stylistic and terminological profile! – are only intelligible, or best intelligible, when placed in a post-conquest context, a dating of the standard rasm to before 650 seems heuristically preferable.
If one shifts the burden of proof, however, it is not evident that the Quran's lack of palpable fit with post-prophetic Islamic history, while excluding a major later reshaping of the text, also rules out minor additions and modifications: there is nothing “out of period” about the story of Jesus and the woman taken in adultery (John 7:53–8:11), often considered to be a later addition to the Gospel of John due to its absence in early New Testament manuscripts. Of course the argument from unanimity discussed in the previous section raises grave questions as to how any changes that are assumed to have been made to the Quranic rasm as late as ʿAbd al-Malik could have been consensually adopted by dissident groups such as the proto-Shiites. In order to strengthen further the position that the standard rasm of the Quran had largely stabilized by the middle of the seventh century the next two sections will examine two further internal characteristics of the Quranic corpus. (It must be emphasized that the conclusion of early stabilization does not pertain to orthographic matters, such as the spelling of long ā, which continued to evolve for much longer.Footnote 27)
Internal features of the Quran (ii): absence of narrative framing
The Quranic texts clearly presuppose an individual messenger figure,Footnote 28 yet are notoriously unforthcoming with specific details about him. This is why Islamic exegesis has found the technique of biographical contextualization to be such an indispensable hermeneutic tool: isolated Quranic segments are clarified by inserting them into a narrative – often one that seems to have been tailor-made for its exegetical function – describing a particular situation from the life of Muḥammad.Footnote 29 Nonetheless, in spite of their supreme interpretative utility, no such contextualizing narratives have seeped into the actual text of the Quran,Footnote 30 again suggesting an early date of closure.
It is true that Shoemaker would date the emergence of narratives about Muḥammad that incorporate Quranic quotations to after 700.Footnote 31 In my view, however, it has by now become reasonably certain that already towards the end of the seventh century narratives about Muḥammad containing Quranic elements were in circulation. An episode that has been particularly thoroughly scrutinized is the story about Muḥammad's first revelation on Mount Ḥirā', significant parts of which, including a Quranic quotation, Andreas Görke and Gregor Schoeler have been able to trace back to the Medinese traditionist ʿUrwa ibn al-Zubayr (d. 93/711–12 or 94/712–13).Footnote 32 Another biographical episode integrating Quranic quotations is the well-known story of the ʿĀ'isha scandal, which Görke and Schoeler likewise ascribe to ʿUrwa.Footnote 33
The situation, then, is as follows: the early Islamic reception of the Quran displays a trend towards the biographical narrativization of Quranic material; and by 700, suitable narrative material about Muḥammad containing scriptural quotations had come into existence, while Muḥammad had also become an important political symbol (in 685–6, an Iranian governor of ʿAbdallāh ibn al-Zubayr minted the first existing coins mentioning Muḥammad,Footnote 34 and from 691–2 on ʿAbd al-Malik's coinage as well as the inscriptions of the Dome of the Rock invoke Muḥammad as a foundational religious figureFootnote 35). How, then, do we explain the fact that the Islamic tradition so scrupulously managed to keep apart the utterances of its prophet and the exegetical narratives with which these utterances were so much more readily understandable? As comparative evidence from the Hebrew Bible demonstrates, collections of prophetic logia seem to display a natural tendency to attract legends about the life and times of the respective prophet.Footnote 36 It is therefore indicative of an early stabilization of the Quran that such legendary accretions could apparently no longer be incorporated into the Islamic scripture and needed to be outsourced to a separate body of literature. For example, neither the Quran nor the material in the early chapters of the book of Isaiah suggest that the prophet in question was a miracle worker; in the case of Isaiah, however, it is Isaiah 38 that turns him into one, whereas in the case of Muḥammad this happens in the ḥadīth.Footnote 37
From all of this one takes away the consistent impression that the text of the Islamic scripture must have set rather early. One might object that the Quran's hypothetical Marwanid redactors could have shied away from incorporating full-blown biographical narratives because they did not wish to alter the general character of the text as a collection of prophetic utterances. But the Quran lacks even the most editorially minimalist techniques of biographical contextualization, such as the insertion of superscriptions tying specific scriptural passages to certain events in Muḥammad's life (see Isaiah 1:1, Jeremiah 1:1–3, and the various Psalmic superscriptions associating the following text with the life of David).Footnote 38 The fact that the Quranic corpus as we have it is remarkably uncontaminated not only by fully-fledged sīra narratives but also by such minor redactional accretions is most easily accounted for by a mid-seventh-century date for the standard rasm's closure.
Internal features of the Quran (iii): lack of linguistic normalization
A third characteristic of the Quran that suggests fairly rapid stabilization are its various “rough edges”.Footnote 39 These include, for example, archaic grammatical features that were not brought into line with later usage, such as employment of an in the sense of “lest” (e.g. Q 4:176, 16:15).Footnote 40 Arguably the most striking example is provided by a handful of passages that violate basic rules of case agreement in classical Arabic (Q 2:177, 4:162, 5:69 and 20:63).Footnote 41 These verses appear to have given umbrage already to early Muslims: it is reported that ʿUthmān himself, when presented with the copies of the Quran that he had ordered to be produced, found incorrect expressions in them, but gave the command for no changes to be made “because the Arabs will change them with their tongues” (thus advocating a solution along the lines of the masoretic Qre-Ktiv distinctionFootnote 42); and Muḥammad's wife ʿĀ'isha is said to have commented that such verses are “the work of the scribes; they wrote it out wrongly”.Footnote 43 Bergsträsser is surely right that traditions which enjoin Muslims to improve on the Quranic text as they recite it, or openly recognize that God's word is contaminated by typos, are likely to be earlyFootnote 44 and may well date to before 700, thus indicating that people did not have to await the advent of Sībawayh to notice the above problems. Yet neither al-Ḥajjāj nor anyone else ever seems to have tried to correct the rasm of these verses.
A similar case is Q 3:96, according to which “the first house [of worship] that was founded for humankind is that in bakka”. Islamic exegetes generally propose that the expression bakka is a variant for makka, i.e. Mecca, yet are forced to construct rather intricate derivations of the expression bakka in order to make the point. What would have been by far the simplest way of remedying the problem, and presumably one on which all major Islamic confessions should have been able to agree – namely, to change a bā' into a mīm – was apparently no longer practicable by the time people started paying serious attention to the verse.Footnote 45 In the Hebrew Bible, too, anomalous or obscure expressions and place names have frequently been retained, but there they have often given rise to interpretative glosses inserted into the text,Footnote 46 which are conspicuously absent from the Quran.Footnote 47
The textual shock freezing indicated by examples such as the above is most easily explained as the early Islamic community's reaction to the death of the charismatic messenger who had been the sole person endowed with the authority to publish and modify divine revelations. In any case, it stands to reason that if al-Ḥajjāj or anybody else had revised the Quranic text around 700, even on a very minor scale, passages such as the above should have been the first to be emended, which would merely have required exchanging single letters and might therefore have occurred almost instinctively.
Conclusions
One must admit that the argument at the end of the preceding section is not unassailable. First, can we be sure that there might not have been more passages of the sort presented above that were corrected (such as Q 48:24, where we get makka instead of bakka)? Second, even if it is conceded that most of the Quranic corpus must have been extant before al-Ḥajjāj and that the latter did not correct existing portions of the text, does this rule out the insertion of new material? In other words, can we rule out that the Quran might have constituted a literary corpus which was textually stable yet could still be added to? Such objections obviously raise a burden-of-proof dilemma: should we require proponents of the conventional view that the standard rasm of the Quran had become fixed by 650 to produce conclusive proof of the absence of later additions, or should we instead require scholars insisting on the possibility of later additions to prove that such additions do in fact exist? I would submit that the latter position is more reasonable: if the only swans we have ever encountered are white ones, it is the proponent of the existence of black swans whom we may legitimately expect to argue his case. Similarly, as long as no Quranic passages with a distinct stylistic and terminological profile have been compellingly placed in a late seventh-century context, the traditional dating of the standard rasm (excepting certain orthographical features) to 650 or earlier ought to be our default view.Footnote 48
Assuming a mid-seventh-century dating of the standard rasm, can we go back any further? Even if one were fully to underwrite the hypothesis that the standard text of the Quran preserves the common textual prototype of the two more faithfully than C-1, and to ascribe, on a probability of 3:1, a terminus ante quem of 646 to Ṣanʿā' 1,Footnote 49 this prototype may only have reached closure in the late 630s. Thus, it seems wise to concede that during the first decade or so after Muḥammad's traditional year of death, the latter's literary legacy may not yet have been fully fixed.Footnote 50
There are nevertheless good reasons to believe that the arrangement of verses in most sūras does go back to the lifetime of Muḥammad. Sadeghi and Goudarzi have underscored the general convergence between the variants exhibited by the lower writing of the Ṣanʿā' palimpsest and what the qirā'āt literature tells us about the spectrum of variance characterizing the earliest stages of the Quran's transmission.Footnote 51 On the basis of the literary sources alone, we may also accept that different recensions disagreed about the inclusion of a handful of short sūras (Q 1, 113, 114, as well as Sūrat al-Khalʿ and Sūrat al-Ḥafd).Footnote 52 But neither C-1 nor the literary sources reveal any genuine disagreement about the contents of each sūra,Footnote 53 thus lending credence to the view that at least the majority of them were extant by Muḥammad's death. It bears repeating that this does not rule out the possibility that existing texts could to some degree have been expanded, reshaped and updated during the first post-prophetic decade. Ultimately, it is above all the rigorous literary analysis of each individual sūra that can determine whether there are reasons for suspecting that it may have undergone early post-prophetic alteration or expansion.
