The Christian Elephant in the Meccan Room: Dye, Tesei, and Shoemaker on the Date of the Qurʾān

Introduction

Much of the qurʾānic corpus is suffused with a selective adaptation of Christian traditions. This diagnosis applies to a panoply of cosmological and eschatological notions, to miscellaneous narratives, and to important aspects of qurʾānic diction.^[1] Moreover, at least sometimes the Qurʾān does not merely employ ideas that are traceable to the Christian tradition or talk about Christians but rather talks to Christians. A good example is Q 4:171, which urges the scripture-owners (ahl al-kitāb) not to “go too far in your religion (dīn),” an admonition that is then concretized by assertions about the status of Jesus (“the Messiah, Jesus son of Mary, was only a messenger of God, and his word that he cast unto Mary, and a spirit from him”), followed by a second-person warning not to “say ‘three’” (wa-lā taqūlū thalāthatun). It is obvious that this is a critique of Christian Trinitarianism and that the scripture-owners addressed here are Christians in particular.^[2] Christians, then, are not just a distant religious community far over the Qurʾān’s horizon.

Guillaume Dye, among others, has argued that the preceding observations are not easily reconciled with the current lack of evidence for organized Christian communities in the immediate milieu in which the Qurʾān’s genesis is supposed to have unfolded.^[3] It is this paucity of traces of organized Christianity in the Ḥijāz that constitutes the proverbial elephant in the (Meccan) room to which my title alludes, in an admittedly labored pun on Sūrah 105 (which is not meant to threaten any party to the scholarly debate here conducted with divine vengeance). The Christian elephant is also invoked in a recent article by Tommaso Tesei,^[4] and it figures prominently in Stephen Shoemaker’s 2022 monograph Creating the Qur’an: the “Christian void in the Qur’an’s traditional birthplace,” Shoemaker writes,

certainly makes it difficult to accept the standard narrative of the Qur’an’s origins entirely in Mecca and Yathrib during the lifetime of Muhammad. The cultural deprivations of the central Hijaz make it effectively impossible for a text so rich in Christian content, like the Qur’an, to arise strictly within the confines of this evidently Christ-barren milieu.^[5]

Incidentally, Shoemaker in particular adds a second, related aporia: he maintains that the pre-Islamic Ḥijāz was characterized by a far-reaching lack of literacy, which he considers to preclude that a sophisticated literary work like the qurʾānic corpus could have emerged there.^[6] Both Dye and Shoemaker contend, moreover, that the explanatory problem generated by the Christian elephant cannot be solved by an appeal to oral tradition. In Shoemaker’s words, the oral transmission of Christian lore “from individuals … who had travelled to Christian lands” cannot “sufficiently explain the deep familiarity with Christian tradition that the Qur’an demands from both its author(s) and audience.”^[7]

Even though not all members of the qurʾānic audience must necessarily have understood every subtle allusion in the text, Dye, Tesei, and Shoemaker are highlighting a genuine explanatory challenge for conventional accounts of the Qurʾān’s genesis. It should not be brushed aside too quickly. In the following, I shall nonetheless make two general points. First, I shall try to explain why the situation is in my view somewhat less aporetic than just portrayed, even if some loose ends do remain. Secondly, I will argue that a fair assessment of the state of the debate requires us to appreciate the equally real explanatory loose ends that are left by scholars like Dye and Shoemaker. In other words, we find ourselves having to choose between two paradigms that are both beset by difficulties, making it inadvisable to make up one’s mind based on exclusive attention to the deficits of one of the two competing outlooks in play.

Mollifying the Elephant

This section will raise a number of considerations that are apt to the impression of aporia that scholars like Dye and Shoemaker are generally at pains to create.^[8]

First, it can be objected that Dye, Tesei, and Shoemaker are being a tad too categorical. To be sure, the Islamic tradition does indeed portray Mecca as an environment that was “Christ-barren,” to employ Shoemaker’s delightful expression. Yet it is nonetheless worth recalling some potential indicants of a communal Christian presence in and around Mecca that have been excavated by Irfan Shahîd. These include most notably a regrettably succinct reference to a Christian cemetery around Mecca that is preserved in al-Azraqī’s (d. 865 CE) History of Mecca.^[9] It is difficult to dismiss such fragmentary pieces of information as fabrications, since they run so obviously counter to the dominant Islamic way of imagining the Meccan population on the eve of Islam as steeped in primitive idolatry. Perhaps, then, the later Islamic tradition did to some degree downplay the real degree of Christian presence in Mecca around c. 600 CE. Shoemaker poses the valid question what the ideological purpose of a conjectured erasure of Meccan Christianity might have been, given that the Islamic tradition was happy to reckon with Jews in Medina.^[10] But perhaps the hypothetical amnesia of Meccan Christianity was not an objective in itself but merely a knock-on effect of the strong tendency to imagine Mecca as a bastion of stone worship, with only a few monotheist loners vaguely groping around for a truth that would ultimately be supplied by Muḥammad. Hence, there may well have been more of a Christian presence at Mecca than Islamic sources are letting on – though clearly not a presence substantial enough, or mainstream enough, in order for Christian writers elsewhere to take note.^[11] Whether new archeological findings might lend further support to this conjecture will remain a matter for future research.

Even so, one may feel that the preceding line of thought is not sufficient to give us enough self-identifying Christians on the Meccan ground, as it were, in order for the Qurʾān to be possible. We should therefore consider, secondly, how proponents of a Ḥijāzī genesis of the Qurʾān might try to accommodate a relative, if not total, absence of communally organized Christianity from pre-qurʾānic Mecca. I share Shoemaker’s assessment that it is not auspicious to rest a lot of weight on the travels of individual Meccans, such as Muḥammad himself, whom one might suppose to have become acquainted with Christian ideas, doctrines, and language during trade journeys to the Fertile Crescent.^[12] One reason to be hesitant about such a scenario is the cogency of Patricia Crone’s argument that Meccan trade, if it existed, must have been of a much more modest scope and geographical range than some twentieth-century scholars were content to postulate.^[13] Accordingly, even if individual Meccans may well have travelled north and south, one ought to avoid hitching one’s confidence that enough Meccans might somehow have come by a sufficient knowledge of Christianity in order for a Ḥijāzī emergence of the Qurʾān to be possible to dubious assumptions about Mecca’s role as an international trade emporium. A second reason to remain hesitant about the supposition of Meccan “cultural tourism,” to use Shoemaker’s language,^[14] is that the crux is not merely to explain how Christian knowledge could have found its way to Mecca. Rather, the challenge is also to explain how enough members in Muḥammad’s audience could have possessed the requisite grasp of Christian narratives and notions in order to process the qurʾānic proclamations. For it is by no means the case that the Qurʾān comes across as introducing biblically based figures, narratives, and ideas to recipients who were as yet entirely unacquainted with them.^[15]

Dye gives due acknowledgment to another way in which proponents of a Ḥijāzī Qurʾān have tried to cope with the Christian elephant: by appealing to somewhat nebulous processes of oral tradition that might have linked the Qurʾān’s Ḥijāzī audience to Christian communities further afield, without resulting in the sort of substantial and organized Christian presence in the Ḥijāz for which we do not have positive evidence. One particular form that such oral transmission might have taken is Christian missionary preaching.^[16] Dye complains that an appeal to oral tradition may easily become an unfalsifiable get-out-of-jail card,^[17] and one can appreciate the salience of his protestation. Nonetheless, I do not think that the prospects for this type of rejoinder are as unpromising as Dye and Shoemaker make it out to be; an appeal to missionary activity might to some degree be capable of reigning in the Christian elephant. Thus, it remains a manifest possibility that Christian preachers, monks, and ascetics (perhaps associated with or setting out from al-Ḥīrah, Buṣrā, or Najrān) would have sought to convert the inhabitants of western Arabia, just as they did further north.^[18] This would have exposed the targets of such missionary endeavors to a certain number of Christian stories, to Christian imaginations of the Last Judgment and the afterlife, and to some basic Christian notions and concepts, such as the idea that Jesus Christ is the son of God.

There is, admittedly, no positive evidence at the current time that the Ḥijāz was indeed a significant target of Christian missionary efforts.^[19] The idea of Ḥijāzī exposure to Christian missionary preaching thus has the status of an auxiliary hypothesis: its function is to help defend a certain explanatory paradigm (namely, the notion that much of the Qurʾān can be placed against a pre-conquest Ḥijāzī background) against potentially falsifying objections.^[20] To put it in more figurative terms, the missionary hypothesis colors in a blank spot. Still, I would stress that given the fragmentary and problematic nature of the literary and other data that we are trying to make sense of, some imaginative filling-in of blank spaces, in the interest of creating a coherent historiographic picture, is inevitable for all parties to the debate. The best one can hope to do is to avoid multiplying such auxiliary hypotheses and to eschew flagrant inconsistencies between one’s various commitments.

If indeed there were missionary efforts of the sort I have just surmised to have existed, they cannot have been very successful, since we do not hear about them in Christian sources: Christian missionaries do not appear to have given rise to a Meccan or Ḥijāzī episcopate that attracted even fleeting mention by Christian authors further north.^[21] But the assumption that Ḥijāzī pagans had for some time been periodically harangued by Christian fire-and-brimstone preachers yet had simply not come round fits well with the fact that the qurʾānic opponents reject Muḥammad’s preaching about the final judgment as “ancient scribblings” (asāṭīr al-awwalīn).^[22] Christian missionary activity in the Ḥijāz would also give us a sufficient degree of awareness of Christianity as a serious ideological option in order to explain why the qurʾānic proclamations critique Christian belief and practices, and sometimes even address Christians directly.^[23]

It is true that one would not expect missionary exposure to yield a sophisticated understanding of the subtleties of Christian theology or ecclesiastical structures. Dye, Tesei, or Shoemaker are clearly inclined to consider this a problem for any attempt to make sense of the extent of qurʾānic awareness of Christian ideas, narratives, and phraseology.^[24] Despite everything conceded in my opening paragraph, however, I maintain that we must not exaggerate the specialist depth of qurʾānic acquaintances with Christian traditions. It is true that certain ideas for which there is ample Christian precedent are numerous and fundamental to the Qurʾān’s own theology. These include various cosmological and eschatological motifs, some stock arguments,^[25] and fragments of memorable phraseology and terminology. An example for the latter category is the expression rūḥ al-qudus (e. g., Q 16:102), which is manifestly descended from the Christian “Holy Spirit.” Yet it is doubtful that the Qurʾān anywhere conceives of the spirit as a person in a divine trinity.^[26] This illustrates quite nicely that the Qurʾān does not contain complex summaries of Christian theological positions. Indeed, in the case of Mary the apparent qurʾānic understanding that Christians consider her to be divine (Q 5:116) is not accurate.^[27] True, the Qurʾān does reflect awareness that Christians upheld the divinity of Jesus and casts Jesus himself as disavowing such a belief (Q 5:116–118). Yet this is hardly a very advanced piece of doctrinal information to have picked up. In general, I would therefore submit that the qurʾānic affinity with the Christian tradition is extensive rather than intensive (which is not meant to imply that the Qurʾān is theologically simplistic or to deny that the Qurʾān may be putting forward pointed alternatives to certain aspects of late antique Christian theology). Extensive rather than intensive acquaintance with Christianity fits a scenario of missionary exposure rather well.^[28]

It must of course be recognized that qurʾānic narrative is often so allusive that many scholars, including myself, feel or have felt compelled to assume some prior exposure to the stories in question on the part of (a significant subsection of) the qurʾānic audience. This general observation also applies to stories that must ultimately have reached the qurʾānic milieu from Christians, like the tale of the Sleepers of Ephesus (qurʾānically, the “Companions of the Cave”) narrated in Sūrah 18 or the accounts of the annunciation of John the Baptist and Jesus in Sūrah 19 (on which see excursus 1 below). Indeed, the qurʾānic retelling of the story of the Sleepers makes explicit reference to disputes surrounding this story in Christian sources, such as the number of the protagonists and the length of time that they spent miraculously asleep in their cave (Q 18:21–22.25–26).^[29] Yet stories, even stories with an embedded theological message, can certainly travel much more easily and further afield than more abstract doctrinal propositions that are only meaningful against the background of some level of theological training (or, perhaps, against the background of entrenched confessional affiliations). Moreover, the recounting of stunning miracles and dramatic divine interventions in the lives of specific individuals can be presumed to have suggested itself as an obvious missionary strategy, insofar as it translates the belief system of which the target audience is to be convinced into the universally comprehensible idiom of individual human fates and fortunes. Hence, I am not persuaded that the significant Christian imprint on many qurʾānic narratives is as such conclusive evidence that the qurʾānic milieu itself must have been heavily Christianized (as opposed to merely having been in some form of real historical contact with a Christianized milieu).

Apart from narratives, missionary exposure would also offer us a compelling channel of transmission for another important respect in which the Qurʾān shows a Christian imprint, namely, the numerous qurʾānic passages that enumerate various “signs” or indicants (āyāt) of God’s power and wisdom in the natural world.^[30] It is highly plausible that this qurʾānic “natural theology” builds on topoi of Christian proselytizing, where attempts to derive important theological propositions from everyday natural phenomena, accessible to the believer and unbeliever alike, would have made excellent sense.^[31] The eschatological resurrection is not (yet) an item of empirical observation, but God’s revivification of the earth by means of rain is (e. g., Q 30:50 or 41:39). And finally, missionary exposure would elegantly accommodate the predominant modalities of qurʾānic engagement with the biblical tradition at large, as described by Wilhelm Rudolph and Sidney Griffith: while the Qurʾān frequently paraphrases and echoes biblical or post-biblical phraseology, it only rarely offers precise citations of biblical (or post-biblical Jewish and Christian) material; and even where these do occur, they are invariably very brief.^[32] Thus, qurʾānic intertextuality has a hit-and-run character; we never see verbatim correspondence between an extended qurʾānic passage and some pre-qurʾānic biblical, Christian, or Jewish document of the sort that would indubitably indicate (at least to a historian who adopts a stance of methodical agnosticism) excerpting from a written source.

Against the preceding assessment, Tesei holds that the Qurʾān often comes across as the “product of a flourishing Christian center.”^[33] Yet this is not evidently true. Rather, I would submit that the Qurʾān is actually strikingly devoid of precise theological, heresiographical, and ritual Christian language: there are no clear qurʾānic equivalents for, say, “Dyophysite” or “Eucharist” (even if the “banquet table” sent down upon Jesus’s disciples in Q 5:112–115 has not implausibly been linked to the Last Supper) or for a “person” of the Trinity. Late antique Christianity was in important respects obsessed with ever more finely grained and, to an outsider, obscure doctrinal distinctions – e. g., if Jesus has two natures, a divine one and a human one, does he also have two wills, a divine one and a human one; and if so, what is the relationship between them? By contrast, the Qurʾān purposefully steers clear of such debates – partly because it rejects their very starting point that Jesus is more than just a human messenger, but also because the Qurʾān articulates a general criticism of gratuitous disagreement in religious matters (e. g., Q 19:37 and 43:65).^[34] As a result of this qurʾānic lack of interest in the subtleties of Christian theology, it is not clear from the Qurʾān alone how one would talk in Arabic about the two natures of Christ, or the three persons of the Trinity, or Mary’s status as a God-birther (theotokos),^[35] or the sacraments, or even about the tenet that God created the world from nothing, ex nihilo.^[36] In short, it would be extremely difficult to imagine a Miaphysite and a Dyophysite having a doctrinal altercation employing only the vocabulary of qurʾānic Arabic.^[37] Thus, the Qurʾān’s considerable degree of engagement with Christian traditions of a narrative, cosmological, and eschatological type is not concomitant with an equivalent degree of specific doctrinal, sacramental, and ecclesiastical references. This accords well with a scenario of Christian content being mediated by missionary exposure and/or other types of oral tradition.

As insinuated by the preceding reference to “missionary exposure and/or other types of oral tradition,” it seems quite possible to me that other forms of non-literate diffusion were as important in the Qurʾān’s historical context as Christian proselytizing. It would not be stretching credulity, for instance, to suppose that the qurʾānic audience – especially seeing that they inhabited a sanctuary reported to have attracted outside visitors – was in transregional contact with other Arabophone groups, located further south and further north, who had come into a Christian orbit and had converted.^[38] Such interaction could possibly have sufficed to equip the qurʾānic audience with the sort of basic grasp of foundational Christian notions, beliefs, and narratives that is required for the Meccan sūrahs’ Christian-infused content to be intelligible. This would dispense us from the need to posit an activity of Christian missionaries in the immediate vicinity of Mecca; the proselytizing that provides such an attractive explanation for certain features of the qurʾānic corpus, as just explained, might then be imagined to have gone on at arm’s length from Mecca, as it were. Intertribal transregional context of the sort just supposed is certainly corroborated by the circulation of pre-Islamic Arabic poetry across significant parts of the Arabian Peninsula and its northern fringe.^[39] For example, al-Nābighah al-Dhubyānī moved between al-Ḥīrah and the Ghassānids, as attested by his panegyric poetry,^[40] and one of the poems in his corpus refers to the Meccan Kaʿbah.^[41] The transregionally mobile character of literary forms and discourses would suggest quite naturally that similar mobility and seepage might have extended to religious ideas and practices as well.^[42] It bears stressing that such transregional cultural contact does not require us to suppose that a substantial number of Meccan traders journeyed deep into Byzantine territory and had significant cultural interactions there.

My third point in regard to the alleged impossibility of placing much of the qurʾānic corpus in a Ḥijāzī context is to reiterate a proposal made in an earlier publication: when we try to picture what sort of religious milieu the pre-Islamic Ḥijāz might have been, we should avoid thinking in terms of a tidy separation of fully paid-up, card-carrying Christians confronting pagans who rejected Christianity lock, stock, and barrel. Rather, we might envisage a fusion of Arabian cults with a certain number of Jewish and Christian concepts, narratives, and practices.^[43] Such syncretism, perhaps inspired by precisely the sort of transregional interaction with more explicitly Christian and Jewish groups just conjectured, would tally well with al-Azraqī’s famous report that there were pictures of Mary and Jesus in the pre-Islamic Kaʿbah, even if this account cannot of course be deemed to have an unproblematic claim to historicity.^[44] A syncretistic hypothesis would also fit eminently well with the latent monotheistic tendencies that scholars from Ibn Taymiyyah to Gerald Hawting and Patricia Crone have discerned in the beliefs of the Qurʾān’s “associating” opponents.^[45] These “associators” or mushrikūn clearly recognized the ultimate cosmic supremacy of a single creator deity, and they also seem to have deployed the notions of intercession and angels.^[46] Such a syncretistic scenario would certainly give us a baseline of audience familiarity with Judeo-Christian concepts and narratives of the sort that is arguably required in order for the qurʾānic proclamations to have made full sense to their original recipients. Most likely, such a syncretistic uptake and retooling of certain biblical and biblically based ideas and perhaps also practices would have been a response to increasing exposure to Christianity pressing in on the Ḥijāz. Indeed, quite possibly it was precisely the Ḥijāzī pagans’ ability to incorporate into their religious life a certain number of culturally prestigious Biblicist ideas, narratives, and practices that enabled them to withstand Christian attempts at missionary recruitment.

Decoupling the Qurʾān from Muḥammad?

The preceding section has sought to dispel the impression that the supposition of a Ḥijāzī genesis of much of the Qurʾān leads to a yawning aporetic chasm. I should nonetheless like to acknowledge that Dye, Tesei, and Shoemaker are putting their fingers on an explanatory problem that is not easily eliminated altogether, similar to the disconnect between the qurʾānic vision of nature and the ecological conditions in Mecca that has been set out by Patricia Crone.^[47] For instance, a recent article by Majied Robinson estimates that the size of the permanent population at Mecca, including children, wives, and slaves, was a mere 552 individuals, though he notes that the number could have been larger if Meccan society had a lower proportion of adult men than suggested by comparative demographic evidence.^[48] Based on this, Robinson holds that the Quraysh were decidedly minor players in pre-Islamic Arabia,^[49] “a small tribe whose status depended on the goodwill of powerful neighbours.”^[50] The small size of Mecca’s permanent population undoubtedly limits the degree of cultural dynamism and religious diversity we may imagine to have played out there.^[51] Of course, Mecca and ʿArafāt’s status as pilgrimage destinations may be assumed to have attracted external visitors (which does not require the maximalist scenario that Mecca and the ḥajj ritual around ʿArafāt had a properly pan-Arabian catchment area).^[52] In view of Robinson’s work, we may therefore want to conceive of pre-Islamic Mecca along lines once entertained in passing by Patricia Crone, as “deserted except for a small family of custodians maintained by pilgrims and other visitors.”^[53] Nonetheless, it remains quite debatable whether even a heavy emphasis on non-residential visitors provides us with the scale of cultural dynamism and diversity that one would be inclined to expect from the Qurʾān’s milieu of origin.

Yet instead of further pursuing how one might gently augment the plausibility of the Qurʾān’s traditional story of origin, let us now turn to the more radical solution that Dye, Tesei, and Shoemaker themselves propose for the Christian-elephant problem. Dye clearly signals that his preferred response is to date at least parts of the qurʾānic corpus after the death of Muḥammad and to decouple them from the latter’s Ḥijāzī context.^[54] The same basic idea is developed in a 2021 article by Tesei. Tesei studies a set of qurʾānic sūrahs consisting in the majority of the “early Meccan” sūrahs as defined by Theodor Nöldeke (though Tesei does not himself endorse or even acknowledge this appellation).^[55] Tesei agrees that the group of texts examined by him is coherent in style and themes and that it forms the earliest portion of the qurʾānic corpus.^[56] Where he does part company with the Nöldekian school is by suggesting that the rest of the Qurʾān, and even certain verses within the sūrahs discussed by Tesei, contain substantial material that postdates Muḥammad. At least prima facie, Tesei’s hypothetical subcorpus of early and authentically ‘Muḥammadan’ (my term, not Tesei’s) sūrahs features little evident Christian content, thus respecting the assumption that the latter would not be historically plausible in Muḥammad’s original Ḥijāzī context.^[57]

Shoemaker, too, operates with a similar model: like Dye and Tesei, he does not deny that the Qurʾān’s ultimate point of origin is the preaching of Muḥammad in western Arabia.^[58] Yet he also argues that Muḥammad’s proclamations underwent a protracted period of oral transmission before being committed to writing, probably in several regional streams and with a considerable degree of fluidity and malleability; according to Shoemaker, it was by and large only under the caliph ʿAbd al-Malik (reigned 685–705 CE) that the qurʾānic text as we have it was put together. Unlike Tesei, however, Shoemaker is not at all optimistic that we can confidently identify a kernel of qurʾānic material that was more or less verbatim delivered by Muḥammad: the Qurʾān as we have it is

the result of the constant, repeated recomposition of traditions that, while they may have their origin in Muhammad’s teaching, were subsequently reimagined, rewritten, and augmented during their transmission by his followers. Therefore, even traditions that possibly originated with Muhammad himself must be recognized in their present form as effectively new compositions produced on the basis of his ideas by his later followers in the very different circumstances of the newly conquered territories of the former Roman Near East and the Sasanian Empire.^[59]

As a result, Shoemaker is much more non-committal than Tesei in identifying specific qurʾānic passages that might contain elements of Muḥammad’s original preaching.

Dye, Tesei, and Shoemaker’s recognition that parts of the Qurʾān are rooted in a pre-conquest Arabian environment and may be associated with Muḥammad as a historical figure is noteworthy. It is one of the two main features that sets their theorizing apart from the older working hypothesis that was allusively suggested in the late 1970s by John Wansbrough.^[60] This divergent element of Dye, Tesei, and Shoemaker compared with Wansbrough is certainly well advised, since it means that they are not obliged to show how the figure of Muḥammad could have been invented from scratch. It furthermore enables them to recognize that the Qurʾān, in addition to its affinities with Christian and Jewish traditions, also has important continuities with the conceptual world of pre-Islamic Arabia. Perhaps the most striking such continuity consists in the fact that the Qurʾān recognizes animal sacrifice as a legitimate part of the cult of Allāh (e. g., Q 5:95, 108:2), thus going against a major trend of late antique religious history, the “end of sacrifice.”^[61] Another example for continuity with pre-Islamic Arabian discourse, as far as we can tell based on poetry and inscriptions, would be the names of three pagan Arabian deities in Q 53:19–20. Such observations do not directly refute the historical scenario developed by Dye, Tesei, and Shoemaker for the simple reason that animal sacrifice or pagan deities could quite convincingly be allocated to the original Ḥijāzī stratum of the Qurʾān, whether or not one additionally supposes that at least parts of this original Ḥijāzī layer have been preserved more or less verbatim.

One must squarely insist that what Dye, Tesei, and Shoemaker are proposing is a serious historiographical option that should not be shouted down as gratuitously sensationalist or self-evidently indefensible. There is, in fact, no reason why historical-critical scholars of the Qurʾān should not find it possible to accept and negotiate the existence of more than one feasible contextual framework for the textual data they are seeking to interpret. It is worth noting that instead of a stark dualism of two historical paradigms, often identified with terms such as “traditionalism” vs. “revisionism,” we are in fact confronted by three basic possibilities, which might be termed the “Muḥammadan,” “ʿUthmānic” or “early caliphal,” and “Marwānid” scenarios, respectively. According to the first scenario, the qurʾānic text is by and large the product of Muḥammad’s preaching in pre-conquest Mecca and Medina; according to the second one, at least some parts of the Qurʾān only reached their present form in the twenty-four years or so between the death of Muḥammad (traditionally dated to 632 CE) and the demise of the third caliph ʿUthmān (656 CE), widely associated with the promulgation of a written text of the Qurʾān; and according to the third option, significant portions of the qurʾānic text did not reach closure before the latter decades of the 600s.^[62] Especially the basic possibility of an ʿUthmānic or early caliphal model for the composition of parts of the Qurʾān is perhaps still insufficiently appreciated, despite Tesei’s article on the parallels between the Qurʾān and the war propaganda of the Byzantine emperor Heraclius, which leans towards a “mid-7^th-century redaction” of the Qurʾān rather than towards the Marwānid dating favored by Shoemaker.^[63]

As I have just remarked, it is in principle conceivable that the genesis of the Qurʾān might be satisfactorily explicable by more than one historical scenario. Or perhaps scholars will, at least for a certain time, need to keep in play all three of the above hypotheses, tentatively trying them out against different parts of the total set of data to be explained (the Qurʾān, archaeological and epigraphic evidence, the early Islamic tradition, non-Islamic texts). Specifically, it may be productive to shift the debate from generic disputes about the date of the Qurʾān to the most likely date of specific passages, on the understanding that some measure of mixing and matching could turn out to be entirely appropriate. For example, while I would continue to consider a Ḥijāzī context to be reasonably adequate for most of the Qurʾān, with regard to at least one qurʾānic statement I can appreciate the attractiveness of an early post-prophetic or early caliphal context. This is Q 3:7, an exceptional admission of the presence of ambiguity in the Qurʾān that would seem to reflect a nascent concern with offering guidance on how to exegetically come to terms with a closed scriptural corpus.^[64]

It stands to be expected that each of the three models up for debate will have specific explanatory benefits and costs. What the remainder of this article will do, therefore, is to draw attention to the explanatory costs or challenges of the basic proposal by Dye, Tesei, and Shoemaker that extended portions of the qurʾānic corpus postdate Muḥammad and hail not from the Ḥijāz but from the territories newly conquered by the early (or proto-) Muslims. By exploring this issue at some length, I am intending to counterbalance Shoemaker’s tendency to dismiss alternative paradigms of the Qurʾān’s emergence as “shackled to the traditional Islamic account of the Qur’an’s origins”^[65] – which I would understand to imply that they have nothing much going for them beyond gullibility, conformism, and a lack of scholarly pluck.

Testing the Dye-Tesei-Shoemaker Hypothesis

In my view, there are three principal difficulties with the hypothesis, or perhaps hypotheses, put forward by Dye, Tesei, and Shoemaker. None of these three difficulties is novel. However, I continue to be unsatisfied by the commentary, if any, that has been offered on these challenges by this article’s primary interlocutors. I conclude the section with further remarks that are largely specific to Tesei’s work, though they also have some bearing on Shoemaker.

First, as Fred Donner has highlighted some twenty-five years ago, the difficulty of pinpointing evidence of conquest-age concerns in the Qurʾān poses a problem for any attempt to date significant parts of the Islamic scripture after the beginning of the proto-Muslim conquests.^[66] For if the Qurʾān were a mid-Umayyad recomposition of earlier oral traditions, as posited by Shoemaker, it seems exceedingly plausible that we would be entitled to expect the resulting text to reflect both the invaders’ struggle against the Byzantines and the Sasanians as well as the various outbreaks of internal strife within their ranks. After all, at the time that Shoemaker proposes for the closure of the Qurʾān, its authors and redactors would presumably have been ensconced as the intellectual spearhead of a warrior elite who had every interest to explain, to themselves and to anyone else who cared to hear, why God had appointed them to be in charge over a vast tributary subject population (and perhaps also why God had elevated them over other groups within the same warrior elite). On this eminently reasonable supposition, however, one is bound to wonder why the Qurʾān is almost entirely bereft of any specific comments on these topics. Why, for example, is there no explicit qurʾānic endorsement of the proto-Muslims’ right to rule the Holy Land and the former dominions of the Byzantines and Sasanians? One cannot evade the difficulty by asserting that the hypothetical redactors of the Qurʾān were too shrewd to commit blatant anachronisms, for as Donner has observed, the hadith certainly does not shy away from anachronism^[67] – quite understandably so, since accurate predictions of future history would by no means be surprising if originating from a prophetic recipient of divine revelations.

For an impression of what an account of Muḥammad’s preaching that is palpably aligned with early post-qurʾānic preoccupations might have looked like, we may turn to the so-called Chronicle of Pseudo-Sebeos, an Armenian history with information up until 661 CE and hence probably composed in the 660s CE.^[68] According to this Armenian text, Muḥammad persuaded the “sons of Ishmael” that Abraham was their ancestor and that they were therefore entitled to seize the Holy Land as their rightful patrimony:

But now you are the sons of Abraham, and God is accomplishing his promise to Abraham and his seed for you. Love sincerely only the God of Abraham, and go and seize your land which God gave to your father Abraham. No one will be able to resist you in battle, because God is with you.^[69]

Notwithstanding the evident allusion to Genesis 15, the broad contours of this epitome of Muḥammad’s preaching may conceivably reflect the way Muḥammad was viewed by some of the early conquerors themselves. A conquest-age Qurʾān, one might expect, would be similarly concerned to leverage the figure of Abraham, who is after all one of the most prominent figures in the Qurʾān as we have it and is explicitly cast as the “father” and forerunner of the qurʾānic believers (e. g., Q 3:68, 22:78), to underwrite a comparable entitlement to Byzantine Palestine. Yet if we examine the Qurʾān in more detail, there is a remarkable dearth of material that could be taken to intimate such an Abrahamically grounded claim to the Holy Land. In fact, there is no explicit claim to the Holy Land at all, whatever the rationale. Instead, where qurʾānic incitements to militancy in “God’s path” are associated with territorial objectives, these are generally limited to the Meccan sanctuary (see, e. g., Q 2:191, 9:28).^[70] (For the sake of full disclosure, it should be noted that the preceding argument assumes that qurʾānic phrases like al-masjid al-ḥarām, “the sacred place of prostration,” or al-bayt al-ḥarām, “the sacred house,” refer to the Meccan Kaʿbah, which is equated with the “sacred house” at Q 5:97 and is also mentioned two verses before, in Q 5:95. Obviously, this presupposition could give rise to a whole subsidiary dispute.^[71])

There are, admittedly, some potential exceptions to the preceding diagnosis of a far-reaching absence of conquest-age concerns from the Qurʾān. One of the more compelling cases is Q 9:29, which commands the Qurʾān’s recipients to fight “those who were given the scripture” – i. e., Jews and Christians – until they pay tribute (al-jizyah). Even if this is hardly a smoking anachronistic gun, it is certainly a statement that would make excellent sense in a context in which the qurʾānic community had begun to subjugate the settled populations of the Fertile Crescent.^[72] As mentioned above, regarding Q 3:7 I continue to find myself attracted to an early post-prophetic dating, though I do not think that the alternative can be conclusively disproven. I would also not wish to prematurely dismiss Tommaso Tesei’s argument that the qurʾānic promise of immediate entry to paradise to those who die fighting “on God’s path” is dependent on the war propaganda of Heraclius. At least according to Tesei, Heraclius’s propaganda is unlikely to have reached Muḥammad’s Ḥijāzī milieu during the Islamic prophet’s lifetime and may therefore have “entered the Qurʾānic corpus at a late stage.”^[73] If, for the sake of the present argument, one were to grant Tesei’s conjecture,^[74] then passages like Q 2:154 and 3:169–171 – which could be viewed as presupposing an assimilation between death in battle and Christian notions of martyrdom – would also need to be dated after the death of Muḥammad.^[75] For good measure, one might additionally concede Q 2:102, which features a reference to Babylon and two angelic names (hārūt and mārūt) that ultimately stem from the Zoroastrian tradition: I would not presently want to rule out that the middle segment of this verse has a terminus post quem in the 630s, during the early stages of the conquest of the Sasanian empire.^[76] Yet even so, at the present state of the debate it looks as if even on a resolutely generous count we will come away at most with a fairly meagre smattering of qurʾānic verses that could be viewed as resonating with a post-conquest context. (As discussed in considerable detail in excursus 1 below, I am unpersuaded by Shoemaker and Dye’s argument that the Jesus-and-Mary pericope in Sūrah 19 reflects a Palestinian post-conquest milieu.^[77]) By contrast, Shoemaker’s Marwānid dating of the final redaction of the Qurʾān as a whole would naturally lead one to expect a much greater amount of such material. After all, Shoemaker explicitly assumes that the proto-Islamic community’s vague recollection of Muḥammad and his original preaching was subject to a dynamic process of oral retelling and recomposition in light of the circumstances and preoccupations of their later transmitters.^[78]

Shoemaker does make an attempt to parry the preceding objection: he claims that the Qurʾān is effectively “replete with anachronisms” by virtue of allegedly adjusting “the relationships between Muḥammad’s new religious community and Judaism and Christianity” in light of post-650 developments.^[79] Shoemaker here presupposes Fred Donner’s well-known claim that the proto-Muslim community had an ecumenical character and welcomed Jews and Christians among its members, and that Islam only developed into a separate communal identity in the first decades after Muḥammad’s death.^[80] Shoemaker also recognizes that there are “Qur’anic passages referring to Jews and Christians and their beliefs in a negative and polemical manner,”^[81] which is to say that they are not compatible with an ecumenical perspective and instead portray Muḥammad’s adherents as a separate religious community. Hence, Shoemaker reasons, the Qurʾān is out of step with the earliest stage of proto-Muslim identity formation, and this points to post-650 editing.

The problem here is that Donner’s hypothesis is by no means tantamount to an undisputable historical fact. Rather, as I have argued in a previous publication, at least the Medinan portions of the Qurʾān fairly consistently presuppose that Muḥammad’s followers form a community separate from Jews and Christians.^[82] Shoemaker in fact agrees that there is some explicitly non-ecumenical material in the Qurʾān, since, as we saw, he is positing a tension between the alleged fact of early Islamic ecumenicalism on the ground, on the one hand, and the presence of non-ecumenical, polemical statements in the Qurʾān, on the other, which latter he considers to be due to post-prophetic adjustments. However, the very presence of non-ecumenical material in the Qurʾān seriously destabilizes Donner’s ecumenical hypothesis, which is to a very significant extent based on the Qurʾān. If, in view of this, one rejects Donner’s hypothesis, as on balance I would, then the anachronistic contradiction that Shoemaker alleges between the supposed fact of early Islamic ecumenicalism and the Qurʾān simply vanishes.^[83]

To conclude the present point, one final respect in which the Qurʾān strikes me as lacking in conquest-age concerns is the extremely high proportion of material that is dedicated to polemicizing against an opposing group who are called “associators” (al-mushrikūn, alladhīna ashrakū). Unlike possibilities entertained in Hawting and Crone’s pioneering publications on the qurʾānic associators, I struggle to appreciate how they might be a plausible cipher for the Christians or Jews whom the conquerors encountered in Palestine and Iraq, given that the Qurʾān generally treats the associators as a group that is ostensibly distinct from the Christians or naṣārā and from the Jews (e. g., Q 5:82). Instead, qurʾānic polemics against shirk look like the sort of theme that would have been plausibly at home in Muḥammad’s original Ḥijāzī context. But how credible is it to suppose that the extremely considerable amount of such material was continuously transmitted and fleshed out, or even composed in the first place, during the period of conquest, when the main opponents of the qurʾānic community in religious terms would have been Christians, Jews, Zoroastrians, and Manichaeans?^[84]

The second of my three principal difficulties for Dye, Tesei, and Shoemaker is related to the previous one and might be termed the Qurʾān’s “interpretive recalcitrance,” by which I mean that in many respects the Qurʾān resists the interpretive construals that later Muslims endeavored to place on it. As Shoemaker stresses, oral traditions are constantly adapted by their transmitters, and material that is no longer relevant is either thoroughly modified or simply discarded.^[85] As regards the Qurʾān, however, many scholars have observed that it is often quite resistant to being harnessed to later Islamic views. For instance, it readily supports neither the claims of Umayyad loyalists nor of ʿAlid loyalists; it famously seems to mandate lashing for adultery (Q 24:2) rather than stoning, which is the canonical punishment for adultery committed by muḥṣan (adult, free, and Muslim) individuals in later Islamic law;^[86] Muslim exegetes insist that the phrase al-nabiyy al-ummī in Q 7:157–58 means “the illiterate prophet,” despite the fact that this is not at all confirmed by an examination of the qurʾānic use of the adjective ummī;^[87] according to Q 2:158, it is “no fault for you [plural]” to “circumambulate” al-Ṣafā and al-Marwa, whereas according to the standard Islamic ḥajj ritual it is mandatory to run back and forth between them;^[88] and the Qurʾān as we have it passes up numerous opportunities to inscribe into the very text of scripture concrete anecdotal details from the life of Muḥammad as relayed in the sīrah literature.^[89] Shoemaker’s view that the canonical version of the Qurʾān is a mid-Umayyad recomposition of vague and malleable oral traditions is not really capable, in my view, of accounting for this interpretive recalcitrance of the Qurʾān: if scripture was the eventual fallout of a protracted process of oral transmission, and if what was transmitted orally was subject to constant adjustments in light of the beliefs and practices of the transmitters, then significant disconnects between scripture and mainstream beliefs and practices at the time of the Qurʾān’s closure should not have arisen.^[90]

Nor should there really have been a general absence of qurʾānic or proto-qurʾānic material from early Islamic history, a phenomenon that is recognized by Shoemaker himself.^[91] For on Shoemaker’s theory, the final redaction of the Qurʾān under ʿAbd al-Malik formed the endpoint of a gradual process leading from the oral recollection of Muḥammad’s revelations to early written (but still open and fluid) collections to regional scriptural codices to, finally, an imperially sponsored standardized version of scripture.^[92] In this scenario, a precondition for the Qurʾān continuing to exist at each stage of the process is its being passed on by subsequent generations of transmitters, collectors, and redactors, whose accumulating work eventually culminated in a fixed scriptural canon. Yet the presupposition of continuous interest in and reworking of proto-qurʾānic material is difficult to square with the observation that the early legal tradition, for example, did not invariably conform to, and sometimes even appears to override, salient qurʾānic pronouncements. Similarly, the presupposition of continuous interest is not easy to reconcile with the fact that many of the biographical anecdotes that gradually sprang up around the figure of Muḥammad are only tenuously grounded in scripture. It is, therefore, mystifying how Shoemaker can maintain that

the composition of the Qur’an by al-Ḥajjāj and ʿAbd al-Malik … comports with one of the more bizarre features of the early Islamic tradition – that is, the almost complete absence of the Qur’an from the religious life of the Believers or Muslims for most of the first century of their existence.^[93]

For Shoemaker’s considered theory appears to be precisely not that al-Ḥajjāj and ʿAbd al-Malik invented the Qurʾān from scratch, producing a scripture ex nihilo. Rather, he conceives of ʿAbd al-Malik as supervising a “team of scholars” who “wove together and honed the various sacred traditions that had entered circulation among Muhammad’s followers during the seventh century.”^[94] ʿAbd al-Malik’s putative imperial scripture was accordingly rooted in earlier proto-scriptural traditions that must have enjoyed considerable circulation. But if that was the case, then the problem of the absence of qurʾānic or proto-qurʾānic material from early Islamic discourses remains as acute an explanatory difficulty for Shoemaker as for scholars who follow the Islamic tradition in attributing the canonical version of scripture to ʿUthmān.

As I have noted previously, a telling illustration of the general phenomenon of the Qurʾān’s interpretive recalcitrance is the fact that a verse from Sūrah 3, Q 3:96, speaks of the sanctuary that God has established “at Bakkah.”^[95] Later exegetes assume that bakkah must be referentially equivalent to makkah, Mecca. Whether or not this reading is true is not the point here. The point is rather that even the minuscule textual change that would be required to transform bakkah into makkah did not happen, forcing the exegetes to explain in various contorted ways why bakkah may in fact be identified with makkah. This faithful preservation of an extremely puzzling word over an alternative that would have been much more straightforwardly intelligible against the background of the reigning assumption that the qurʾānic corpus contained the ipsissima verba of Muḥammad as proclaimed in Mecca and Medina would be completely unexpected in the sort of oral tradition that Shoemaker surmises to have led up to ʿAbd al-Malik’s imperial canonization project. Instead, what really should have happened, at some point fairly early on in the oral transmission of proto-qurʾānic materials, is a tacit emendation of bakkah to makkah. A similar though somewhat less conclusive argument may be made for a number of cases where the Qurʾān’s consonantal skeleton would seem to diverge from certain basic principles of classical Arabic grammar (e. g., the use of al-ṣābirīn in the oblique case instead of the nominative al-ṣābirūn in Q 2:177): arguably, the more prolonged a process of textual transmission we posit, the higher the likelihood that such issues, too, would have been ironed out and normalized, thus obviating the need to tackle them by means of ingenious but sometimes forced ad hoc rationalizations.^[96]

To his credit, Shoemaker does appreciate that certain interpretive puzzles besetting the Qurʾān should simply not exist had the text been passed on via a continuous oral tradition, which ought to have ensured intelligibility.^[97] In his final chapter, he therefore has recourse to the auxiliary hypothesis that these parts of the Qurʾān predate Muḥammad and had already been written down when they were encountered by “Muhammad and his earliest followers.”^[98] “For whatever reason,” he writes, “Muhammad and his coterie of followers must have revered the words of these ancient writings, so much so that they eventually found their way into the canonical Qur’an.”^[99] However, this creates a further muddle that is nowhere solved. For as noted above, Shoemaker also argues at some length that it is inconceivable that the rudimentary sort of literacy that existed in the Ḥijāz might have been harnessed to transcribe the Qurʾān; yet his auxiliary hypothesis presupposes precisely what he has previously deemed to be inconceivable, namely, a written – and thus lexically stable – transmission of religious material in Arabic in pre-Islamic Arabia.^[100]

The third difficulty that I would like to raise in response to the work of Dye, Tesei, and Shoemaker is linguistic. All three share the view that the vast majority of qurʾānic references to and engagements with Christian traditions could only have originated in the post-conquest age and outside Muḥammad’s Ḥijāzī milieu.^[101] But then why did the hypothetical authors of these materials call the disciples of Christ al-ḥawāriyyūn, which is derived from Ethiopic ḥawārәyān, and did not for example label them rusul, which is the standard term in later Christian Arabic, or employ some Arabization of Syriac shlīḥē?^[102] Admittedly, the reason why the Qurʾān does not call the apostles rusul may very well be a deliberate attempt to distinguish them from proper prophets or messengers of God. However, this still fails to explain why the alternative to rusul that was selected is a word derived from Ethiopic. In general, the fact that some fundamental religious terms in the Qurʾān would seem to have their closest ancestors in Classical Ethiopic – including terms that figure in Christian-flavored materials, such as al-ḥawāriyyūn or al-shayṭān – sits well with the assumption that the associated traditions did at some point pass through western Arabia.^[103]

Finally, Tesei’s ambitious attempt to propose a precise catalogue of the qurʾānic sūrahs that could plausibly be viewed as the Qurʾān’s original Muḥammadan kernel gives rise to specific queries as well. As was noted above, Tesei’s corpus of sūrahs putatively going back to Muḥammad initially looks as if it is largely devoid of the sort of advanced Christian content that Tesei would deem contextually out of place in the pre-Islamic Ḥijāz.^[104] However, Tesei’s admirable willingness to enter into specific textual detail does make his theory vulnerable on several counts. For one, even the corpus of sūrahs discussed by Tesei contain eschatological statements of the sort that have significant parallels in the New Testament or in Syriac (and other) post-biblical Christian literature. These Christian-flavored eschatological motifs include the splitting or stripping away of the heaven (Q 69:16, 77:9, 78:19, 81:11, 82:1, 84:1), the eschatological earthquake (52:9, 56:4, 79:6–7, 99:1), the blast of the eschatological trumpet (Q 69:13, 74:8, 78:18), and God’s judging of the resurrected on the basis of written records (e. g., Q 81:10, 84:7–12).^[105] Phraseologically noteworthy, moreover, are the expressions yawmaʾidhin (Q 88:2.8, 89:23.25, 99:4.6, 100:11), corresponding (probably via Syriac) to Greek en ekeinē tē hēmera (Matt 7:22), and (yawm) al-dīn, “the (day of) judgment” (Q 82:9.15.17–18, 83:11, 95:7), which is evidently derived from Syriac yawmā d-dīnā.^[106]

The fact that all the qurʾānic references just provided stem from sūrahs belonging to Tesei’s Muḥammadan corpus illustrates that the Qurʾān’s eschatological imaginaire is too thoroughly pervaded by parallels to Christian literature in order to make it easily feasible to identify a sizeable amount of qurʾānic material that is devoid of significant Christian assonances: it is not easy to convey “apocalyptic fervor”^[107] of the sort that Tesei attributes to Muḥammad’s original preaching while complying with the far-reaching no-Christian-content-in-Mecca principle that Tesei espouses. Tesei is not unaware of this issue. For example, he appreciates that the qurʾānic expression al-sāʿah, referring to the “hour” of the eschatological judgment, corresponds to New Testament language,^[108] and he acknowledges that the term appears in one of the sūrahs included in his Muḥammadan nucleus, at Q 79:42. The way he proposes to resolve the difficulty is by maintaining that this verse is a later interpolation.^[109] However, one would be excused for feeling that this solution is somewhat ad hoc: apart from the fact that no considerations other than the occurrence of al-sāʿah are offered in support of the claim that Q 79:42 is secondary, the following verses 43–46 all contain pronouns referring back to al-sāʿah. (One might of course strike out Q 79:43–46, too.)

Shoemaker to some degree escapes the preceding quandary, since he does not believe it possible to identify which portions of the Qurʾān belong to the Islamic scripture’s original Ḥijāzī stratum. Nonetheless, his scenario is by no means entirely immune to the difficulty just raised. What, we may ask, was the message that Muḥammad preached? After all, Muḥammad must have had something to say that was sufficiently momentous in order to engender a religious movement that set off a dynamic of rapid and extensive conquests and eventually became Islam and produced the Qurʾān. Now, Shoemaker does seem to accept that Muḥammad’s original preaching had something to do with monotheism and eschatology.^[110] These are of course the two most prominent themes figuring throughout the qurʾānic corpus as a whole. Yet despite the variety of motifs and expressions in which these two general themes manifest themselves, it is a defensible claim that more often than not they are articulated in ways that exhibit concrete affinities with Christian or Jewish traditions. For instance, one of the primary ways in which the Qurʾān expresses its radical monotheism is by employing derivatives of the root sh-r-k, a terminological feature whose ultimate background is rabbinic.^[111] It is no easy task to imagine how Muḥammad might have enunciated a monotheistic and eschatological kerygma that was sufficiently untouched by Jewish and Christian notions and expressions in order to fit the rather minimal amount of engagement with the biblical tradition that Shoemaker would seem to deem possible in the pre-Islamic Ḥijāz. Of course, Shoemaker might retort that what derives from Muḥammad are only the general themes of monotheism and eschatology and none, or very few, of the specific ways in which these themes are voiced in the Qurʾān as we have it: Muḥammad preached that there was one God and that there would be an eschatological judgment, but he did so in terms that are largely unrecoverable. To my mind, though, this would be evasively, and indeed unsatisfactorily, vague. How precisely might a pre-Islamic resident of Mecca have spoken of the exclusive existence of one deity or of a day on which all humans will be resurrected and meet their reckoning without drawing, to an inevitably significant extent, on established Jewish and Christian language?

It is not straightforward, then, to identify a textual nucleus of the Qurʾān that is devoid of at least some important affinities with the biblical tradition, or – if one follows Shoemaker in rejecting the distinct identifiability of any sort of Muḥammadan nucleus in the Qurʾān – to move beyond the haziest outline of what Muḥammad might have preached. But if already the Qurʾān’s eschatology, natural theology, and radical monotheism are irreducibly conversant with specific Christian or Jewish notions, why should we be unduly surprised by, say, the further observation that the Qurʾān also narrates some pivotal scenes from the lives of Mary and Jesus? Why be minded to draw any rigid boundary between qurʾānic eschatology and monotheism, deemed sufficiently free of specific Christian assonances in order to have some grounding in a pre-Islamic Ḥijāzī context, and qurʾānic narrative, for which a Ḥijāzī context is rejected? And if that question seems difficult to answer, then Dye, Tesei, and Shoemaker’s deft concession of an original Ḥijāzī stratum in or underneath the Qurʾān, whether recoverable or not, ceases to be a viable theoretical position. This will, in turn, engender further complications, seeing that it is the notion of a Ḥijāzī basis to the Qurʾān that permits Dye, Tesei, and Shoemaker to sidestep thorny questions about the origin of, for example, the qurʾānic endorsement of animal sacrifice, as noted above.^[112]

Conclusion

The overall case that I have tried to make in the preceding can be restated in the form of four propositions, which will be followed by some additional observations:

1. There are undoubtedly some loose ends in the traditional scenario of the Qurʾān’s genesis. The lack of attestation for a Christian presence in the Ḥijāz constitutes such an anomaly, although I have tried to gesture towards ways in which the impression of an intractable aporia can be substantially alleviated.

2. It would be inappropriate to rule out a mid- or late-seventh-century dating of the Qurʾān, or of parts of the Qurʾān, on a priori grounds. As I have noted above, Q 3:7 might well postdate the activity of Muḥammad. The basmalah, too – the sūrah-opening formula “In the name of God, the truly Merciful” – could be a scribal addition that was incorporated into the Qurʾān only when the qurʾānic corpus was gathered together after Muḥammad’s death.^[113] At least a very early post-prophetic or early caliphal dating of certain qurʾānic passages would not be flagrantly incompatible with available radiocarbon datings of early qurʾānic manuscripts.

3. Despite the preceding two remarks, however, I contend that a late dating of substantial portions of the Qurʾān to the second half of the seventh century, especially to the Marwānid period, produces some very tangible explanatory loose ends. These are rarely confronted by those championing a mid- or late-seventh-century dating of the Islamic scripture.

4. Although Shoemaker would resist it, I find it entirely appropriate to expect a detailed analysis of the Islamic scripture attempting to assign specific qurʾānic material, themes, or phraseology (or at least precursor versions of specific parts of the Qurʾān) to the different stages of textual development that are posited by proponents of a mid- or late-seventh-century date of the Qurʾān. If indeed the composition of the Qurʾān spanned much of the seventh century, it really should be possible somehow to disaggregate the qurʾānic corpus as we have it into the legacy of a certain number of regional schools with characteristic theological and terminological features, just as the Pentateuch has been disaggregated into the literary output of different authors or schools (whether one conceives of these as having authored self-contained source documents or simply as being responsible for identifiable layers in the final product). As we saw above, Tesei has made a welcome and thought-provoking attempt to sketch a starting point for such an analysis. Nonetheless, I have endeavored to show that his model is beset by grave problems. While some of them are specific to his work, they do entail a general question mark over the possibility of reconstructing Muḥammad’s original Ḥijāzī kerygma in a manner that would not itself run afoul of the Christian-elephant dilemma that Dye, Tesei, and Shoemaker marshal against more traditional accounts of the Qurʾān’s genesis.

It is unfortunate that instead of attempts to wrap up loose ends and to flesh out general postulates, endorsements of a mid- to late-seventh-century dating of the Qurʾān sometimes succumb to the temptation of changing the topic of conversation, by averring or implying that a truly critical, sober, neutral, dogmatically uncommitted scholarly examination of the Qurʾān must end up debunking standard Muslim belief (a view presumably anchored in the conviction that genuinely critical scholarship will always end up debunking traditional belief). Thus, for Shoemaker to be “critical” seems to be primarily to display the fortitude of refusing to be “governed” by the Islamic tradition and to resist its “powerful influence” and “gravitational pull.”^[114] There is a sort of heroic iconoclasm here that bristles at the perceived demand to pay “obeisance to the Islamic tradition.”^[115] In the conclusion of Shoemaker’s book, this perspective becomes openly condescending: scholars who fail to agree with him must be in thrall to a “protectionist” discourse that “aims to shield the Qur’an from the rigors of historical-critical analysis.”^[116] Such pronouncements risk sounding like an attempt to compensate gaps in the historical argument by resorting to moralistic hectoring.

Of course, equivalent aspersions are often cast the other way: scholars who question important aspects of the standard Islamic narrative of origins have been tediously and inanely dismissed as somehow “anti-Islamic.” As a stimulating and thoughtful passage in Shoemaker’s introduction makes clear, he is writing from a position of protest against the position that religious studies must privilege insider perspectives.^[117] I find his criticism of this approach compelling, and I would squarely agree that it cannot be a criterion of valid historical scholarship on religion that its results must be acceptable to contemporary adherents of the religious tradition in question (which in practice risks meaning: to the most vociferous ones among what may in effect be a considerable range of contemporary adherents). I would add, however, that rejecting this latter approach does not entail its diametrical opposite that it should be considered a criterion of valid critical scholarship that it will be iconoclastic. In any case, scholars of all people really should find it possible to respect the good faith of those arriving at different conclusions and to try reasonably hard to give them a fair hearing rather than turning them into moronic straw(wo)men.^[118]

But there is more amiss in current research on early Islam than an occasional lack of disputational etiquette. In her recent survey of chronological theories in modern qurʾānic scholarship, Emmanuelle Stefanidis pays welcome attention to how chronologies of the Qurʾān construct what she calls a fabula, a narrative, around the textual data of the Qurʾān.^[119] Stefanidis shows that chronologies of the Qurʾān often come as a story of sorts, in which the textual evidence and a certain narrative contextualization thereof are rolled into one, such that the evidence confirms the narrative but is itself identified and presented in a way that presupposes the narrative. This, of course, chimes with the charge of circularity that has been levelled at attempts to reconstruct a relative chronology of the qurʾānic corpus.^[120] Even though I, for one, remain committed to the basic feasibility of a non-circular diachronic analysis of the Qurʾān,^[121] I would certainly concede that the writing of history has something in common with storytelling.

Yet I would also add that we, as scholars of the Qurʾān and early Islam, need to get better at transforming narratives into hypotheses that can become the subject of scholarly discussion, criticism, and falsification. This operation involves clearly identifying an explanandum – a certain data set (e. g., the Qurʾān) that requires historical explanation or interpretation – and then proposing a hypothesis that accounts for or makes sense of this explanandum. The hypothesis proposed should then be compared with alternative ones and shown to be simpler and more plausible than its competitors. Furthermore, scholars tabling a certain hypothesis ought to be proactively honest about evidence undermining the hypothesis at hand, which might either emerge from the original data set or from another one (e. g., from sources other than the Qurʾān). As we have seen on two occasions in the present article, one way in which such refractory data could be addressed is by devising auxiliary hypotheses – namely, hypotheses that are not designed to account for the explanandum at hand but rather to help defend the original hypothesis against objections, by insulating it from, or reconciling it with, inconvenient data. It is a hallmark of good scholarly practice to indicate very clearly whether a given proposition has the status of a hypothesis or of an auxiliary hypothesis, because at some point the weight of our auxiliary hypotheses might start to drag down the plausibility of our original hypothesis. Finally, the original hypotheses might give rise to corollaries (things that follow from it) or conjectures (things surmised to be true against the background of the hypothesis). Neither of these are strictly speaking grounds for preferring a hypothesis over its competitors; the value of a hypothesis resides in its ability to account for the original explanandum. But it might of course be the case that a given hypothesis accounts for more than one explanandum. This will give it a significant edge over alternatives, assuming that we are committed to keeping our scholarly constructs elegantly simple.

Despite an acute awareness of my own scholarly failings, I would lament the fact that what one all too often encounters in the study of the Qurʾān and early Islam are not hypotheses but narratives – comforting narratives, conspiratorial narratives, narratives with exciting shifts and transitions, but narratives just the same. I would submit that by endeavoring to be much clearer about the epistemological structure and status of our theorizing, we are more likely to minimize scholarship-as-storytelling and move beyond a confusing clash of self-sustaining but incompatible tales of early Islam. At the risk of appearing patronizing, this means that we need to keep at the forefront of our minds questions like the following: What am I trying to explain? Does it need explaining? Can it be explained more easily (more simply, more elegantly) in a different way? Is there opposing evidence? Does the latter necessitate an auxiliary hypothesis that significantly complicates my original explanatory posit or perhaps even contradicts it?

Excursus 1: Shoemaker and Dye on the Nativity Pericope in Sūrah 19

Shoemaker’s Creating the Qur’an is rather limited in its engagement with specific qurʾānic verses. It is therefore worthwhile examining an earlier article of his on the qurʾānic account of the nativity of Jesus in Q 19:22 ff. in order to appreciate how Shoemaker goes about trying to show that a particular section of the Qurʾān is rooted in a non-Ḥijāzī context and is highly probable to postdate Muḥammad.^[122] A recent publication by Guillaume Dye attempts to further consolidate Shoemaker’s argument.^[123] Nonetheless, the upshot of my assessment will be that Shoemaker and Dye do not succeed in demonstrating that Q 19:22 ff. draws on Palestinian local traditions that are likely to have entered into the Islamic orbit only in the wake of the Muslim conquest of Palestine.

Shoemaker begins by observing that the qurʾānic nativity scene in Sūrah 19 has its closest pre-qurʾānic parallel in an early Christian legend that forms part of the traditions around Mary’s dormition and assumption. In the version cited by Shoemaker, contained in the Ethiopic Liber Requiei, Joseph and Mary are travelling to Egypt in order to escape from Herod (cf. Matt 2:13–23); at a spring, the infant Jesus miraculously commands a palm tree to bend down in order to feed his parents.^[124] The Qurʾān, by contrast, positions what is recognizably a variant of the same palm tree miracle in the context of Jesus’s birth rather than during the Holy Family’s flight to Egypt (Q 19:23–26). Shoemaker contends that the only parallel for the Qurʾān’s past-partum recontextualization of the palm miracle is offered by traditions connected to the Palestinian Kathisma church, located between Jerusalem and Bethlehem. As Shoemaker explains, the Kathisma site was originally associated with a non-canonical version of the nativity (found in the Protevangelium of James) that situates the birth of Jesus in a remote cave on the road between Jerusalem and Bethlehem rather than at Bethlehem itself. But the Kathisma site also appears to have been linked with the palm tree miracle that Jesus allegedly performed en route to Egypt. Shoemaker concludes that the qurʾānic nativity scene must have been influenced by Palestinian local tradition, since it was only in connection with the Kathisma site that “the two early Christian traditions of Christ’s birth in a remote location and Mary’s encounter with the date palm and spring are brought together.”^[125] As I have already hinted, Shoemaker takes this diagnosis to imply that the nativity narrative in Sūrah 19 likely postdates the conquest of Palestine, which would have brought the proto-Muslim conquerors into close contact with the Kathisma site and its related narrative traditions.^[126]

Shoemaker’s hypothesis of a Palestinian genealogy of the qurʾānic nativity scene is appealingly concrete, and I have provisionally accepted it in an earlier publication, albeit without endorsing Shoemaker’s corollary of a post-conquest date for the nativity account in Q 19.^[127] However, a more detailed examination shows that the theory is ultimately speculative. I shall try to set the stage for this claim by means of three observations:

(1) While the Mary-and-Jesus account in Sūrah 19 is evidently rooted in quite a few salient Christian parallels, it also exhibits some important idiosyncrasies that ought to be given due attention. Most strikingly, the Qurʾān makes no reference to Joseph and depicts Mary as giving birth in solitude. Moreover, the qurʾānic nativity and palm miracle scene does not take place in transit, whether to Bethlehem or to Egypt. Rather, Mary is in a remote place because following her conception she has intentionally “withdrawn to a distant place” with Jesus (Q 19:22: fa-ḥamalathu fa-ntabadhat bihi makānan qaṣiyyā). These observations provide a strong prima facie indication that the Mary-and-Jesus narrative in Sūrah 19 is taking some liberty in reshaping extant Christian traditions. Hence, in making sense of the passage we ought to be wary of doing so primarily by seeking to pinpoint Christian precursor traditions that are presumed to be closely replicated in the Qurʾān.^[128]

(2) The hermeneutic stance just outlined is further buttressed by the observation that some core features of the qurʾānic nativity account exhibit unmistakable resonance with other narratives in Sūrah 19. Thus, Mary’s withdrawal in v. 22 (cited above; cf. also v. 16) corresponds to Abraham’s announcement that he will withdraw from his father and his people in v. 48 (wa-aʿtazilukum); and the sūrah’s initial narrative, on Zechariah, begins by portraying its protagonist as praying by himself (vv. 3 ff.) before leaving the sanctuary (al-miḥrāb)^[129] and facing his “people” (qawm; v. 11), just as Mary returns to her people (qawm) with her newborn child (v. 27). The three principal narratives of Sūrah 19 thus revolve around an alternation of solitude and communal confrontation, combined with the further motif of Zechariah, Mary, and Abraham receiving a God-given descendant.^[130] Note also that Q 19:52, which briefly alludes to the prophetic initiation of Moses, has another pious protagonist experience a solitary encounter with God. Another link between the Jesus-and-Mary pericope and the preceding story about the annunciation of the future birth of John (the Baptist) to his father Zechariah is thrown into relief by the three occurrences of the verb kallama (“to speak to someone”) in vv. 10, 26, and 29: Zechariah and Mary, two adults normally capable of speech, are silenced either by God (vv. 10–11: Zechariah’s sign that his request for a child will be fulfilled is his inability “to speak to” his people) or by the infant Jesus (v. 26: Mary is advised to maintain that she has vowed a silent fast to God, preventing her from “speaking to” anyone), whereas a newborn child not normally capable of speech is miraculously empowered to address first Mary (vv. 24–26) and then a general public casting aspersions on his mother’s chastity (vv. 30–33).^[131] In important respects, then, the Mary-and-Jesus narrative of Sūrah 19 is tailored to mesh with the thematic concerns of its literary environment. (This is, of course, entirely compatible with the plain fact that the passage is nonetheless informed by miscellaneous Christian motifs.^[132])

(3) Finally, and most importantly, the evidence presented by Shoemaker to the effect that it was in the context of the Kathisma site that the palm miracle became part of the nativity is emphatically circumstantial. Shoemaker shows that the Kathisma church was associated with a non-canonical nativity setting on the road to Bethlehem rather than at Bethlehem; he cites a passage from a pilgrimage guidebook authored between 560 and 570 by the so-called “Piacenza Pilgrim,” which links the Kathisma site with the miraculous appearance of a spring during Mary and Joseph’s flight to Egypt;^[133] and he argues that the palm miracle, which the Piacenza Pilgrim does not mention, must have been connected to the spring miracle, which he does mention.^[134] What Shoemaker does not produce, however, is an explicit report placing the palm-cum-spring miracle squarely in the context of the nativity. Barring further evidence, we thus lack an unequivocal Christian witness to the narrative fusion of the palm tree miracle and the nativity that is found in the Qurʾān. One may of course posit that such a fusion could or would naturally have happened as a result of the two events having attached themselves to one and the same Palestinian holy place. Yet explicit evidence corroborating this is not actually forthcoming. Rather, as has been recognized by Dye as well as Patricia Crone, it is only in the qurʾānic nativity scene that the fusion surfaces^[135] – and does so, as one may add, with a fair amount of standard Christian context being omitted or having been lost (namely, Mary being on the road to Bethlehem, or alternatively to Egypt, and being in the company of Joseph).

The evidence on the table therefore permits an alternative scenario. Perhaps what happened at the Kathisma site was merely the substitution of an earlier etiology centered around the Protevangelium’s non-canonical nativity account by another etiology centered on the palm tree miracle that supposedly occurred when Mary and Joseph rested en route to Egypt. Such a substitution could well have taken place without the palm tree miracle being absorbed into the nativity scene. If so, then the Qurʾān’s idiosyncratic placement of the palm miracle in the context of the nativity does not replicate a precursor narrative linked to the Kathisma site. Nor is there, in my view, a genuinely compelling reason to assume that the Qurʾān’s post-partum contextualization of the palm miracle must be a development of, or “variation”^[136] on, the fact that both the palm miracle and the nativity had been consecutively associated with one and the same Palestinian locale. Rather, the Qurʾān could in principle have independently drawn on the palm tree legend and transferred it to the context of the nativity, at least if we can identify an intelligible qurʾānic motive for such a transfer.^[137]

This motive would presumably have been that the transfer suited the thematic concerns of Sūrah 19, which, as we saw, exercise a tangible impact on the presentation and structure of the Jesus-and-Mary pericope. After all, it is almost certainly due to the palpable interest that Sūrah 19 takes in the solitude and isolation experienced by certain prominent figures at crucial moments in their lives that Joseph has been completely excised from the qurʾānic account of Jesus’s birth. It is therefore natural, and in my view also sufficient, to appeal to a similar explanation – namely, one centered on the theological preoccupations of Sūrah 19 – in order to explain why the Qurʾān draws the palm miracle into the scene of Jesus’s birth rather than placing it at a later moment in Mary’s life: the qurʾānic telescoping of Jesus’s delivery and the palm miracle into one and the same situation allows the qurʾānic nativity scene in Q 19:22 ff. to function both as the fulfilment of Mary’s exchange with God’s “spirit” in vv. 17–21, during which Mary learns that she is to have a son, and as a reassuring demonstration of God’s solicitude for those loyally devoted to him. To put it differently, the Qurʾān’s telescoped narrative, with its distinctive post-partum placement of the palm miracle, makes for a more effective follow-up to the preceding annunciation than a more faithfully Christian rendition of the story that one might hypothetically construct, in which the Qurʾān would have interposed between Jesus’s birth and the palm tree miracle an explanation of why Mary and her child needed to escape to Egypt on account of being persecuted by Herod, whom the magi had previously alerted to the birth of the messiah etc. None of this is pertinent to the point of the qurʾānic nativity account, and therefore quite understandably drops out, just as a skilled screenwriter adapting a novel will usually omit certain side plots and minor characters.

As intimated above, Shoemaker’s hypothesis about the Palestinian origin of Sūrah 19’s nativity pericope is also endorsed in a recent book chapter by Dye.^[138] Dye adds a further piece of circumstantial evidence in favor of a link between Q 19:16 ff. and the Palestinian Kathisma church. Before proceeding to this datum, however, we will need to review the interpretive crux that it is meant to resolve. After Mary and her newborn baby make their way back to her people in Q 19:27, Mary is notoriously addressed by her people as “sister of Aaron” in v. 28. The purpose of the address is most likely to serve as a contrast with Mary’s perceived promiscuity, just as the remainder of the verse evokes the respectability of Mary’s parents: “Your father was not a wicked man, and you mother was not unchaste.” But of course, from a biblical vantage point Mary is not the sister of Aaron and Moses. Does Q 19:28 perhaps indicate that the Qurʾān conflates Mary, the mother of Jesus, with Miriam, the biblical sister of Aaron, as a long line of older Western scholars have maintained?^[139] After all, both figures have identical names in Greek, Syriac, and Arabic, and Exod 15:20 even applies to Miriam the very same sobriquet, “sister of Aaron,” that Q 19:28 attaches to Mary.^[140]

Like quite a few recent scholars, Dye rejects the hypothesis of character conflation.^[141] He instead proposes that the link between Mary and Aaron is grounded in typology, a characteristic strategy by which Christian writers were wont to tie together the Old and New Testaments: Mary is describable by means of an epithet originally applicable to her namesake Miriam not because Mary is identical with Miriam but rather because Mary is prefigured and foreshadowed by Miriam, in the same way in which, say, Jonah’s emergence from the belly of the fish prefigures the resurrected Christ.^[142] Dye combines this line of interpretation (which is already proposed in a 1976 essay by Erwin Gräf^[143]) with an equally typological understanding of Q 3:35, where the mother of Mary is called “the wife of ʿImrān,” and presumably also of Q 66:12, where Mary is called “the daughter of ʿImrān.”^[144] ʿImrān is usually considered to be the Arabic name for the biblical Amram, the father of Moses and Aaron (Exod 6:18.20),^[145] though the Qurʾān does not actually name Moses’s mother or sister (Q 20:38–40, 28:7–11) and never directly mentions his biological father at all.^[146] Dye rules out that we may interpret the Qurʾān as saying that ʿImrān “is the name of Mary’s biological father, called Joachim in Christian sources (Prot 1–5), which are familiar to the author of the sura.”^[147] As Dye reiterates slightly later, our understanding of Q 3:35, 19:28, and 66:12 must not end up imputing to the Qurʾān “a confusion of any kind, since the author of sura 19 has an intimate knowledge of Christian traditions.”^[148]

Proceeding to his star exhibit, Dye then adduces a liturgical reading or lection “from the words of the prophet Jeremiah” that was used at the Kathisma church and is preserved, in a Georgian translation, in the codex Tbilisi A-144. This text, previously discussed by Gilles Dorival, ambiguously calls Aaron “the brother of Miriam/Mary,”^[149] employing a name that can, to the best of my understanding, refer to either of the two women, as in Greek.^[150] The expression cited is of course the exact inverse of the qurʾānic reference to Mary as the “sister of Aaron” at Q 19:28. To provide the briefest of summaries of the relevant passage from the Georgian Lection of Jeremiah, a prophecy ascribed to Jeremiah predicts that Aaron will recover the ark of the covenant from the hiding place where it was reportedly placed by Jeremiah, and it is in this context that “the priest Aaron” is glossed as the “brother of Miriam/Mary.” While this expression could in principle intend Aaron’s biological sister Miriam, a few lines later the ark of the covenant is equated with the “Holy Virgin Mary.” This might then suggest the understanding that earlier on Aaron has in fact been cast as the brother of Mary and not, or not just, of Miriam. Dye furthermore observes that a typological equivalence between Miriam, the sister of Aaron, and Mary, the mother of Jesus, is actually rare in ancient Christianity, although he acknowledges some further attestations for it.^[151] This relative rarity of a Miriam-Mary nexus leads Dye to posit that the “sister of Aaron” sobriquet in Q 19:28 indicates a link specifically to the liturgical traditions of the Kathisma church. Dye thus claims to have uncovered a second distinctive connection between Sūrah 19 and the Kathisma site, in addition to the qurʾānic telescoping of the nativity and the palm tree miracle that takes center stage in Shoemaker’s article.

There is more than one count on which one might query whether the similarity between Q 19:28 and Aaron’s designation as “the brother of Miriam/Mary” in Dye’s Georgian text is really sufficient to establish a direct connection between Sūrah 19 and the liturgy of the Kathisma church. To begin with, despite the complementarity obtaining between the Georgian manuscript’s “brother of Mary” and the qurʾānic “sister of Aaron,” in the Georgian Lection of Jeremiah the link between Mary and Miriam seems quite cursory and implicit. By contrast, other ancient attestations of a Miriam-Mary nexus are far less equivocal. Thus, Dorival – who also underlines the general scarcity of Greek texts propounding a Miriam-Mary typology – discusses two Greek attestations for a Miriam-Mary typology, including one by Gregory of Nyssa (On Virginity, chapter 19). Commenting on the designation of Miriam as “the sister of Aaron” in Exod 15:20, Gregory states explicitly that he considers Miriam to be a typological prefiguration of Mary the theotokos (i. e., the mother of Christ), since they likely shared the virtue of virginity.^[152] More recently, Nestor Kavvadas has quoted the Syriac translation of a homily by Severus of Antioch that displays a noteworthy leap from Miriam, the sister of Moses, to Mary the theotokos.^[153] Two places in Aphrahat’s Demonstrations also document a potential Miriam-Mary nexus, though neither of them is as strong as the passage from Gregory.^[154] Hence, notwithstanding the negative image of Miriam foregrounded by Dorival,^[155] the link between Miriam and Mary could in principle have been available on the margins of the wider Christian tradition rather than just via the Kathisma church. Moreover, some of the relevant Greek and Syriac prooftexts connect the two figures more overtly than the Georgian Lection of Jeremiah, and at least Gregory of Nyssa does so while engaging with Miriam’s biblical epithet “sister of Aaron” from Exod 15:20, which is literally identical with the title that Q 19:28 applies to Mary. Hence, if one is attracted to a typological interpretation of Q 19:28, the Georgian Lection of Jeremiah really is not the only, and certainly not the most compelling, intertext. This undermines the link with the Kathisma church in particular that is championed by Dye and Shoemaker.

At a more fundamental level, there are also reasons to question Dye’s reasons for resorting to a typological construal of Q 19:28’s “sister of Aaron” title in the first place. As we saw, Dye takes for granted that Q 19:28 could not possibly be casting Mary as being literally the sister of Aaron and that the way she is addressed in this verse should therefore be understood as the typologically motivated transference to her of a label properly applicable to her Old Testament namesake Miriam (to whom it is in fact applied at Exod 15:20). For Dye such a typological reading of the expression “sister of Aaron” goes hand in hand with an equally typological interpretation of Q 3:35’s reference to Mary’s mother as “the wife of ʿImrān”: since the predominant name of Mary’s father in the Christian tradition is Joachim, the qurʾānic ʿImrān cannot literally be Mary’s father.^[156] Yet there is reason to be uncomfortable with this categorical insistence that the Qurʾān must inevitably respect basic data of biblical history and genealogy. Certainly the two Medinan verses that describe Mary’s mother as “the wife of ʿImrān” (Q 3:35) and Mary herself as “the daughter of ʿImrān” (Q 66:12) do not look like anything other than literal family relationships, just as the references to “the wife of Noah,” “the wife of Lot,” and “the wife of Pharaoh” that figure in Q 66:10–11 – so immediately before the description of Mary as “the daughter of ʿImrān” in Q 66:12 – surely capture literal biological relationships.^[157] It seems contorted to evade the conclusion that ʿImrān is simply the qurʾānic name for Mary’s father, whatever he might be called in Christian sources. After all, the possibility of a radical reuse of biblical characters in the Qurʾān – meaning one that is incapable of being reconciled with the basic facts of the biblical text – is sufficiently borne out by the qurʾānic relocation of the figure of Hāmān from Achaemenid Babylonia to Pharaonic Egypt.^[158] There are further salient examples: at least some Meccan retellings of the Exodus suggest that the Israelites subsequently inherited Egypt rather than leaving Egypt for a Promised Land beyond the sea;^[159] the paternal link between Abraham and Ishmael does not seem to be present in the chronologically earliest qurʾānic references to these two figures;^[160] and Q 18:60–64 transfers to Moses an episode previously associated with Alexander the Great.^[161] Given all these examples, it strikes me as precarious to postulate that the Qurʾān must invariably conform to vital elements of biblical (or extra-biblical Christian) historiography even in the absence of explicit qurʾānic confirmation thereof.

Dye’s view that the qurʾānic characterizations of Mary as the “sister of Aaron” and the “daughter of ʿImrān” can only be true typologically rather than literally is therefore a significant hermeneutic decision: a straightforward reading of the phrases under discussion is disallowed as somehow underestimating the sophistication of the Islamic scripture. There are grounds to be worried about an interpretive circle here: it is assumed that only a maximally sophisticated reading of the Qurʾān, in line with the Christian tradition, will do; and the remarkable sophistication of the Qurʾān is then adduced as evidence against a Ḥijāzī context.^[162] I am not of course advocating an interpretation of the Qurʾān as crude or unsophisticated, nor do I take issue with an intertextually comparative reading of the Islamic scripture. But I am registering the concern that an interpretation of the Qurʾān as sophisticated in a very peculiar sense – namely, as nimbly versed in a wide array of biblical prooftexts and post-biblical traditions – may to some degree be primarily a reflection of the exceptional erudition of scholarly readers like Dye. At least to an extent, then, the Christian elephant is also the product of a hyper-erudite reading of the Qurʾān that prides itself on unearthing complex layers of biblical and Christian subtexts in the same way in which Jewish and Christian exegetes will take ludic delight in tying together seemingly unrelated scriptural prooftexts or events from sacred history.^[163]

What would a reading of Q 19:28 look like that is not driven by a priori assumptions about the Qurʾān’s unfailing conformity to the biblical tradition? One possibility that should not be dismissed too quickly is simply to take the verse at face value and accept that it is indeed casting the mother of Jesus as the sister of Aaron, making Jesus the biological nephew of Moses. Even if this may at first blush look like an appalling butchering of biblical sacred history, the hypothesis of a qurʾānic short-circuiting of Moses and Jesus would need to be refuted based on qurʾānic rather than biblical data. Unlike Christian sacred history, built on the idea that various Old Testament figures anticipate Jesus, the Qurʾān could in principle afford to be quite vague about the precise chronological position of Jesus in Israelite history. Perhaps, then, the onomastic identity of Miriam and Mary and an occasional typological nexus between them produced a certain degree of “permeability”^[164] that eventually lead to their amalgamation into a unitary character? Crucially, there is no reason why a confessionally non-partisan historian should feel obliged to describe such a hypothetical conflation of Mary and Miriam as a case of qurʾānic “confusion.”^[165]

Nonetheless, a consistent equation of Mary the mother of Jesus with Miriam the sister of Moses is difficult to sustain across the entire Qurʾān. This is so because there are, first, three Medinan verses – Q 2:87, 5:46, and 57:27 – that give the impression that Jesus is being cast as the final member of the Israelite sequence of messenger-prophets (rusul), who are said to have followed in the footsteps of Moses (Q 2:87; cf. 57:27) and in turn to have been followed by Jesus (Q 5:46; cf. 57:27).^[166] These verses evidently presuppose a significant genealogical distance between Moses and Jesus. Secondly, there is the Mary-and-Jesus pericope in the Medinan Sūrah 3. Meccan sūrahs recount how the infant Moses was tracked by his sister (who remains anonymous) after having been placed in an ark by his mother (Q 20:40, 28:11–12; cf. Exod 2). As others have observed before, it is difficult to reconcile this scene with Q 3:35–37, which portrays Mary as growing up in the Israelite temple (miḥrāb), similar to the description of her upbringing in the Protevangelium of James.^[167] The account of Mary’s birth in Sūrah 3 is certainly not redolent of the pre-Exodus setting that one would expect for a sibling of Moses.^[168]

To the evidence just presented one could respond by allowing for an internal qurʾānic development, by which an earlier Meccan presentation of Mary as the biological sister of Moses and Aaron in Q 19:28 gave way, in the Medinan proclamations, to an increasing genealogical distance between Moses and Jesus. In other words, we could try to read the presentation of Mary in Sūrahs 19 and 3 on the familiar model of an increasing qurʾānic acquaintance with biblical or biblically based traditions. Such a developmental approach does not strike me as entirely unviable. Yet if by the time of the Medinan Sūrah 3 it was recognized that Christians did not consider Mary to be identical with Miriam, why did Q 19:28’s formulation “sister of Aaron” not undergo revision, or at least attract a retrospective explanation forestalling the inference that Mary was a biological sibling of Moses and Aaron? This is certainly what other cases of secondary insertion or retrospective clarification in the Qurʾān would have led one to expect.^[169]

It is principally the difficulty just broached, and not the general premise that the Qurʾān could not be in conflict with biblical history, that ought to attract us to a non-literal construal of the “sister of Aaron” sobriquet (whereas with regard to the portrayal of Mary as the daughter of ʿImrān, where no equivalent argument applies, a literal reading must stand). Now, one way of achieving a non-literal interpretation of Q 19:28, thoroughly familiar by now, would be to see the application of the “sister of Aaron” epithet to Mary in Q 19:28 as grounded in a Miriam-Mary typology. However, even if we disregard Dorival and Dye’s assessment that a typological tie between Miriam and Mary is not frequent in ancient Christian literature,^[170] a typological understanding of Q 19:28 is not unproblematic. For one, the qurʾānic corpus is otherwise devoid of any conspicuous parallels or affinities between Mary and the anonymous sister of Moses who figures in Q 20:40 and 28:11–12, corresponding to the biblical Miriam: there is no evidence, apart from possibly Q 19:28 itself, that the Qurʾān discerned significant parallelism between the figures of Miriam and Mary that would explain why Q 19:28 might casually, without any further elaboration, transfer to Mary an epithet properly applicable to Miriam. There is, moreover, a very real question mark as to whether interpreters are entitled to consider Christian-style typologies to form a ready part of the Qurʾān’s standard repertoire for making sense of history.^[171] Given such impediments to a typological exegesis of Q 19:28, is there a way of interpreting the verse in a non-literal manner that does not require an appeal to typology as a stepping stone?^[172]

An obvious candidate for the explanatory vacancy just advertised is Suleiman Mourad’s proposal that the qurʾānic expression “sister of Aaron” (ukht hārūn) means something like “fellow tribeswoman of Aaron” rather than a literal sibling of Aaron,^[173] an interpretation that Mourad supports by surveying a host of non-literal qurʾānic occurrences of “brother” (akh) and “sister” (ukht).^[174] Now, it is true that late antique Christians did not generally attribute an Aaronide lineage to Mary.^[175] It may of course be that Q 19:28 is deliberately gainsaying the standard Christian idea that Mary was a descendant of David,^[176] but I am unconvinced that the question of Mary’s Davidic vs. Aaronide genealogy mattered much to the qurʾānic author(s) and immediate recipients.^[177] Perhaps, then, the point of calling Mary the “sister of Aaron” was not so much to contrast her alleged immorality with her status as a fellow tribeswoman of Aaron but rather with the chastity that would behoove a fellow sanctuary-attendant of Aaron – in other words, a colleague of Aaron. After all, Aaron was generally regarded as the first Israelite high priest and the ancestor of the priestly class (cf. Num 3).

It must be conceded that Aaron’s priestly status is not incontrovertibly reflected anywhere else in the Qurʾān. However, at least the temple or Israelite sanctuary (al-miḥrāb)^[178] does figure in connection with the qurʾānic Mary: following the Protevangelium of James, the Medinan verse Q 3:37 depicts her as being visited there by Zechariah, who is associated with the sanctuary already in Q 19:11 (cf. also Q 3:39). Indeed, one can speculate whether the “eastern place” to which Mary is said to have retired and where she is said to have concealed herself in Q 19:16–17 is perhaps an oblique reference to the Jerusalem temple, where the preceding Zechariah pericope is set.^[179] Given the association between the qurʾānic Mary and the temple (a link that is certain for Sūrah 3 and not indefensible for Sūrah 19) and given also the pivotal role of the temple in the Protevangelium’s account of Mary’s childhood, I find it persuasive to view Q 19:28 as corresponding to a functionally similar utterance found in chapters 13 and 15 of the Protevangelium, where both Joseph and the high priest accusing Mary of fornication contrast her presumed offense with the fact that she was “brought up in the Holy of Holies.”^[180] Q 19:28’s address of Mary, by her outraged people, as a “sister of Aaron” – meaning, perhaps, someone who is like Aaron and his priestly descendants closely related to the Israelite sanctuary – could quite aptly be read as contextually equivalent to the reproachful reminder in the Protevangelium that Mary was raised in the temple.

A different objection to Mourad’s interpretation of Q 19:28, whether in its original form (fellow tribeswoman of Aaron) or in the slightly revised version just proposed (fellow sanctuary-attendant of Aaron), emerges from the observation that all other non-literal qurʾānic occurrences of “brother” or “sister” share a basic connotation of contemporaneity, co-existence, or immediate contiguity. For example, Q 9:11 or 33:5’s “brothers in religion” are contemporaries sharing a certain religious orientation; when Q 15:47 calls the inhabitants of paradise “brothers,” this must mean that they co-exist in perfect amity; and where fellow tribesmen are styled as “brothers,” the link also appears to be one between contemporaries rather than between individuals separated by a considerable historical distance. The qurʾānic deployment of non-literal brotherhood does not therefore immediately fit an interpretation of Q 19:28 that takes for granted that Mary lived many generations after Aaron.^[181] Still, the rhetorical intent of the “sister of Aaron” sobriquet may be to position Mary in a relationship of what Claudia Rapp has called “notional equality”^[182] with regard to Aaron, by way of holding her up to the high standards of behavior deemed to be incumbent on any colleague or associate of such an illustrious figure as Aaron. Thus, on the interpretation presently entertained, the way in which Q 19:28 evokes the metaphor of sisterhood would certainly be distinctive compared with other figurative qurʾānic references to siblinghood, but it would not be downright irreconcilable with them.

In any case, regardless of whether one espouses the non-literal yet also non-typological reading of the “sister of Aaron” title just set out or instead prefers a typological one or even a literal one, I would reiterate two things. First, Dye’s case for a connection between Sūrah 19’s Mary-and-Jesus pericope and the Kathisma church via the Georgian Lection of Jeremiah is ultimately no more compelling than Shoemaker’s original argument. Secondly, the Medinan verses Q 3:35 and 66:12 unequivocally and literally identify Mary as the daughter of an individual named ʿImrān (who despite his onomastic similarity to the biblical Amram is nowhere in the Qurʾān linked to Moses), and in this regard the Qurʾān quite clearly parts ways with Mary’s Christian pedigree.^[183] Whatever we make of Q 19:28, the qurʾānic name of Mary’s father therefore drives home that biblical genealogy is not invariably a safe benchmark against which to interpret the Islamic scripture. The fact that at least one important genealogical disparity between the qurʾānic Mary and the Christian tradition thus continues to stand accords well with this article’s general argument that the qurʾānic proclamations more likely than not emerged in a milieu in which the presence of Christianity was somewhat diluted in comparison to regions like Palestine or Mesopotamia.

Excursus 2: Shoemaker on Oral Tradition

Despite having found much to critique in Shoemaker’s Creating the Qur’an, one of the book’s undoubtedly valuable features is its interdisciplinary breadth. Particularly stimulating is Shoemaker’s extensive engagement with memory science and the study of oral traditions in chapters 6 and 7.^[184] A proper assessment of these sections of his monograph will need to be undertaken elsewhere or by others. But two general remarks can appositely be made here.

First, there is the question of the implications of modern memory science for the controversial issue of the date and authenticity of poetry said to be pre-Islamic. In an endnote, Shoemaker asserts that “the collective findings of memory science and the study of oral cultures have indeed effectively proved the wholesale inauthenticity of this poetry as preserving the actual words of any pre-Islamic poets.” This, Shoemaker holds, entails that the burden of proof rests on the proponent of authenticity rather than the proponent of inauthenticity.^[185] Given my own extensive use of early Arabic poetry as a resource for shedding light on the cultural background of the Qurʾān,^[186] it may not come as a surprise that I disagree and would question Shoemaker’s opinion that modern memory science refutes the possibility of a near-verbatim transmission of early Arabic poetry. At least if we take our bearings from what Islamic-era sources tell us about its composition and transmission, these were not casual everyday activities comparable to the memory experiments conducted by Frederic Bartlett, which Shoemaker reviews in some detail.^[187] Rather, the production and preservation of poetry had become the preserve of trained specialists. A mnemonic culture that is sustained by skilled specialists is clearly more likely to secure accurate transmission than spontaneous hearsay among the general population.

In addition, two core features of early Arabic poetry in particular will have jointly acted as vital constraints on the memory of individual transmitters, namely, meter and rhyme: a given poem is characterized by the combination of a particular monorhyme and a particular meter, which impose significant limits on the substitution of a given phrase or verse by a variant (as do stock phrases and the general sequence of topics in qaṣīdah poems, which even in early poetry is at least to some degree subject to schematic patterning). Meter and rhyme would, for instance, have ruled out transferring entire verses from one poem to another that differs either in rhyme or meter. If one adds in an assumed ability to recall general meaning and imagery, a considerable degree of textual stability in the oral transmission of early Arabic poetry is well conceivable. There is of course a rich library of formulaic phrases and of more or less synonymous epithets that can be collated from the existing poetry, and where these are metrically equivalent, or can be made so, one cannot rule out that recitation involved a measure of improvised substitution. But the overall repository of stock phrases on which such improvised substitution must have drawn could still be authentically pre-Islamic and have been transmitted via a continuous practice of oral rendition. This succinct sketch of how orally transmitted poetry might have exhibited a significant degree of mnemonic stability due to the effects of combined constraints imposing limits on variability is loosely inspired by David C. Rubin.^[188] Parenthetically, while Rubin’s monograph is cited by Shoemaker,^[189] he does not engage with the latter’s argument in favor of the possibility of mnemonic stability in oral traditions. Despite my own lack of acquaintance with modern memory science, I accordingly wonder whether Shoemaker’s claim that modern memory science has “effectively proved the wholesale inauthenticity” of pre-Islamic poetry is not based on a self-servingly selective presentation of the literature.

The second comment to be made concerns the implications of the study of memory and oral traditions for the textual stability not of pre-Islamic poetry but of the Qurʾān. In this regard, Shoemaker should be credited with formulating a genuine insight: the interdisciplinary research marshalled by him does, in my view, roundly discredit a scenario in which the revelations promulgated by Muḥammad were meticulously preserved via an oral tradition before being secondarily committed to writing (whether after decades or after mere years or months). In other words, I share Shoemaker’s view that it is not likely that Muḥammad might have given a revelatory address (consisting in, say, a medium-length sūrah like Q 20), which was then spontaneously memorized by the audience present on the occasion and accurately recalled years later. For in the case of the Qurʾān, most of the factors and constraints that would have facilitated a reasonable degree of accuracy in the oral transmission of poetry are absent: the qurʾānic revelations being, apparently, a novel type of literature in Arabic, their earliest transmission could not yet have been the preserve of a group of specifically trained specialists. Moreover, the Qurʾān lacks meter, and the principles of qurʾānic rhyme are considerably less stringent than those governing poetry.

As noted earlier, Shoemaker’s own alternative proposal is an extended process of “constant, repeated recomposition,”^[190] in the course of which Muḥammad’s revelatory deposit was continuously reconstituted, modified, and expanded over several decades. This would obviously be one possible response to the preceding, though as I have argued throughout this article it creates considerable explanatory pressures in other regards. I am therefore attracted to a second possibility, which is to abandon the assumption of “the oral transmission of Muhammad’s teachings largely from memory for a period of at least two decades,” an idea that Shoemaker considers to be an integral feature of the “traditional Nöldekean-Schwallian narrative.”^[191] In my view, those opting to date much or most of the Qurʾān to the life of Muḥammad are well advised to recognize quite expressly that the composition, revision, and earliest transmission of the bulk of the Qurʾān was crucially reliant on writing. As I have written previously, at least apropos of complex texts like Sūrahs 2–5, which bear traces of distinctly scribal processes of textual revision and editing, it seems incontestable to me that these sūrahs could only have been assembled and stored in writing,^[192] even if the primary modality in which ordinary community members encountered them would nonetheless have been aural, that is, would have taken the form of hearing them – or excerpts from them – recited. At least some reliance on written storage and transmission is also abundantly likely for considerably shorter sūrahs like, say, Sūrah 37. It is really only for very brief pieces like Sūrahs 1, 105–106, and 112–114 that a scenario of exclusively oral transmission seems credible.

Significantly, the conjecture that qurʾānic sūrahs existed in writing already at the time of the Prophet does not deny that they are clearly optimized for oral delivery or that the qurʾānic proclamations present themselves as something to be recited and heard (e. g., Q 46:29, 72:1, 84:21) rather than as something to be silently perused. It also does not rule out that the written version of a qurʾānic sūrah might incorporate material that was first enunciated orally, perhaps even extempore. Both caveats are important in order to accommodate the fact that qurʾānic compositions employ techniques characteristic of oral composition, such as an extensive use of formulae and stock phrases as well as miscellaneous kinds of “oral typesetting” like serial vocatives.^[193] Nor does the hypothesis of written preservation from the time of Muḥammad onwards contradict the evident fact that based on and checked against a written stream of transmission, it is perfectly feasible for individuals faithfully to memorize the entire Qurʾān. Finally, it must of course be borne in mind that the earliest evidence we have for the written transmission of the Qurʾān points to a writing system that is in important respects underdetermined and therefore reliant on supplementary memorization, insofar as there is no routine usage of diacritics and vowels signs.

My suggestion that all but the shortest sūrahs were stored in writing already during the lifetime of Muḥammad is not unheard of: as Shoemaker observes, Angelika Neuwirth has raised similar ideas.^[194] The fact that this view is, in Shoemaker’s perception, “an outlier that is far from the mainstream”^[195] of scholars dating the Qurʾān to the life of Muḥammad is surely no reason to discount it as a viable scenario of the Qurʾān’s emergence, even though it is perhaps not advisable for anyone attracted to this approach to speak of the Qurʾān’s “oral composition” without due qualification.^[196] A considerable part of the problem here is the potential multivalence of the concept of orality, which could with some justification be taken to designate not only oral composition or oral transmission but also to characterize a text geared for oral recitation and styling itself accordingly (while nonetheless being transmitted, or even having been compiled, in writing). Yet appropriate distinctions should largely help to dispel the problem. The only reason why a proponent of an early dating of much of the Qurʾān might not be deemed to be entitled to the hypothesis of written composition and preservation would be Shoemaker’s claim that the pre-Islamic Ḥijāz was an essentially non-literate environment. However, on this issue the proverbial jury is, in my view, still out.^[197]