I will be quoting from renowned Islamic scholar and Mariologist Stephen J. Shoemaker’s Creating the Qur’an: A Historical-Critical Study, published by University of California Press, Oakland, California 2022, Chapter 3, Radiocarbon Dating and the Origins of the Qur’an. Shoemaker demonstrates why carbon dating is inexact and cannot decisively prove that the oldest extant Quranic MSS, e.g., Birmingham Quran, Sana’a etc., date from before the very end of the seventh century AD. All emphasis will be mine.
As we noted at the end of the preceding chapter, the radiocarbon dating of certain early manuscripts of the Qur’an has become something of a flashpoint in recent studies of the Qur’an.1 For those who wish to maintain the accuracy of the traditional Sunni account of the Qur’an’s composition, as well as its contrived scholarly offspring, the Nöldekean-Schwallian paradigm, some radiocarbon analyses of these early manuscripts could appear to validate their convictions. Yet at the same time, repeated attempts to date these same early codices have yielded drastically different results in some cases, seeming to indicate that something is not working quite right with this method of dating, at least for parchments from the early medieval Near East. Nevertheless, scholars committed to the traditional narrative of the Qur’an’s origins have fervently upheld the accuracy of those studies favoring their position, while searching out reasons to impugn the results that do not.
The scientific luster of these results can and often does beguile scholars (particularly when the results support their presuppositions), even as it seems increasingly clear that this method is not entirely accurate for dating early Qur’ans, at least not within a range narrower than a century or two. Indeed, scholars of the Dead Sea Scrolls faced very similar difficulties when trying to radiocarbon date documents from that collection. A deeper look into the ins and outs of radiocarbon dating can help us to understand why, for the time being at least, radiocarbon dating has not proved a reliable method for determining the date of the Qur’an’s formation. While these methods of scientific analysis are welcome and useful for the contribution that they can bring, they nevertheless have so far failed to deliver any sort of “silver bullet” that can instantly resolve the many complex issues surrounding the early history of the Qur’an. Instead, it seems that for the time being we must continue the hard work of historical-critical analysis, alongside the data from radiocarbon analysis, in order to understand the history of the Qur’an’s composition and canonization. (Pp. 70-71)
The answer would appear to lie in the fact that the original Qur’anic text of the Sanaa manuscript’s erased lower writing is a nonstandard version of the Qur’an that deviates regularly from the received version now identified as the “ʿUthmānic” Qur’an. As such, it is an extremely rare, although not unique, witness to the diverse ways with which the Qur’an continued to circulate still at the end of the seventh century. Efforts have been made to identify the manuscript’s original Qur’an with one of the early Companion codices as described by the later tradition, without much success. Instead, what we have in the undertext of Sanaa 01–27–1 is a witness to a different, early version of the Qur’an.25 Only once the “ʿUthmānic” text had achieved dominance was it erased and replaced with the canonical version of the Qur’an in the middle of the eighth century. Thus, as Déroche concludes, it would appear that noncanonical versions of the Qur’an were still being produced as late as 700 CE and were only eliminated eventually through the efforts of ʿAbd al-Malik and al-Ḥajjāj to establish a particular version of the Qur’an as canonical.26 (P. 77)
… Therefore, given the current state of our knowledge, it remains the fact that the form of the text as written onto the Tübingen parchments corresponds with other Qur’ans from the early eighth century, which bear the hallmarks of production under imperial auspices.40
A similar early Qur’an, whose “discovery” was heralded online around the same time as the Tübingen manuscript, has emerged from the Mingana collection at the University of Birmingham. The manuscript had been in the university library’s collection for almost a century at the time (Mingana Islamic Arabic 1572a), overlooked until radiocarbon analysis of a folio, done at Oxford University, dated its parchment to sometime between 568 and 645 CE with a probability of 95.4 percent (1456 BP ± 21; 578–646 CE with IntCal 20).41 As with the Tübingen manuscript, following the announcement of these results, the press and online media quickly went into a frenzy over this “oldest” witness to the Qur’an, beguiled once again by the supposedly rock-solid scientific evidence of the radiocarbon dating. The only problem is that if we strictly follow the radiocarbon dating, the parchment seems a bit too early for the tradition of an ʿUthmānic collection, if not also for Muhammad’s authorship, at least in the case of earlier dates within the range of possibilities. As Gabriel Reynolds notes, the very early results from the radiocarbon dating of this manuscript would in fact seem to confirm that early datings of folios from the Sanaa manuscript are not, as some have suggested, the result of a “botched job,” but are instead relatively accurate datings of the parchments used in this codex.42 The so-called “Birmingham Qur’an” consists in fact of just two leaves from an ancient manuscript that have been bound together with seven leaves from another manuscript. Yet Fedeli, who “discovered” the Birmingham manuscript, has also identified sixteen folios in the Bibliothèque nationale de France from the same early manuscript (MS BnF328c). There are thus eighteen total folios from this early manuscript, and their analysis forms a major part of Fedeli’s dissertation, which convincingly demonstrates that the text written on the parchment seems to be significantly more recent. Indeed, despite being credited with discovering the world’s oldest Qur’an in the press, Fedeli has from the start insisted that this witness to the Qur’an should not necessarily be dated as early as this particular radiocarbon analysis might suggest.43 (P. 80)
Given the state of the Qur’anic text as it was copied onto this manuscript, it seems extremely unlikely that this Qur’an could possibly date to the time indicated by the radiocarbon analysis of the parchment. If we insisted on such a date, between 568 and 645 CE, then we must revise the traditional narrative of the Qur’an’s origins not to a later date, but in the opposite direction, concluding that it took place much earlier than the Nöldekean-Schwallian paradigm would have it. In such a case we must assume that a highly advanced and technical practice of writing was in place well before ʿUthmān (who began to reign in 644 CE), but this is extremely unlikely, as we will see in chapter 5. Likewise, such a dating requires the circulation of even older exemplars that could have been copied by the individual who produced this particular manuscript. This manuscript simply is not a first draft of the Qur’an for reasons that Fedeli has amply demonstrated, and so we cannot imagine that this might somehow be one of ʿUthmān’s initial codices. Yet the possibility of a pre-ʿUthmānic Qur’an leads us in another direction and to another possibility—namely, that the Qur’an, or at least some significant parts of it, is in fact pre-Muhammad. In fact, Reynolds suggested as much following the announcement of the Birmingham Qur’an’s radiocarbon dating.47
As Reynolds briefly remarks, there are many elements of the Qur’anic text that early Islamic scholarship simply could not understand, which is rather puzzling if the text had a continuous transmission from Muhammad through the early community. This is particularly so in instances where the meaning of certain words and their vocalization is largely unknown—indeed, sometimes the Qur’an itself does not seem to fully understand some of its own declarations.48 James Bellamy convincingly demonstrated in several articles that these passages indicate, at least in some cases, that “there was no oral tradition stemming directly from the prophet strong enough to overcome all the uncertainties inherent in the writing system.”49 One could readily understand such ignorance if in fact the Qur’an—at least in some parts—were a much older text that predated Muhammad and his new religious movement, written using language that the members of that movement did not always comprehend. Both Michael Cook and Patricia Crone (as noted already above) have suggested this hypothesis in their more recent works, and it is one that we will return to in the final chapter of this book, when we come to consider the context implied by the contents of the Qur’an.50 (Pp. 81-82)
If one is still clinging to some hope that we might be able to find a way out of the messiness of this method, well, things are about to get even more complicated. I know of two instances in which early Islamic documents with known dates were subjected to radiocarbon analysis, and the results were not at all reassuring. Déroche had samples from two dated Qur’ans analyzed by the Lyon lab: one with a known date indicating its production in 1020 CE and the other in 907 CE. The radiocarbon dating of the first Qur’an came in at 1130 BP ± 30 years, or between 871 and 986 CE with a 95 percent probability (774–994 CE with IntCal 20). “The most probable dates,” Déroche further reports, “arranged in decreasing order of probability were 937, 895 and 785AD. The closest result, that is to say 937 AD, is separated by eighty-three years from the date provided by the colophon.” Even if we use the upper limit of the date range—that is to say, 986 CE—the difference still amounts to thirty-four years, around a third of a century.61 For the second Qur’an, the radiocarbon date was determined at 1205 BP ± 30, with a calibrated date of between 716 and 891 CE (704–941 CE with IntCal 20). Déroche identifies the most probable dates, “once again in decreasing order of probability: 791, 806 and 780AD. The most probable result, 791AD, is 116 years earlier than the actual date.”62 It is true, however, that in this case the uppermost date is reasonably close to the actual year in which the Qur’an was copied. Nevertheless, absent this specific information regarding its production, we would be very much at sea in dating this Qur’an, and it is certainly quite possible that the parchment used for this codex was a century or so older than the text itself. Or it may be that, again, for whatever reason, something is not working with our calibration of historical C-14 levels.
Fred Donner has also performed similar tests of this method and its accuracy. Although the results have not yet been published, Donner revealed them publicly during the question and answer session at the presidential address for the International Association of Qur’anic Studies in November 2018. Professor Donner was kind enough to allow me to relate the gist of his findings in advance of their pending formal publication.63 He took samples from an undated papyrus, which, based on content, he is quite sure dates to early in the seventh century. He sent samples to two labs. The first one returned a dating in the early 800s CE. The second lab, Oxford, gave a result of 650–700 CE, which is closer to the suspected date, but still a little too late. In light of these results, he sent samples from two dated papyrus letters to the Oxford lab, without revealing that he already knew the dates. One letter was dated to 715 CE, and the other to 860 CE; both samples came back with dates around 780 CE, much too late for the former, while indicating use of an eighty-year-old papyrus in the case of the later. In both instances the radiocarbon date was an altogether inaccurate indicator of the age of the texts in questions, beyond a general dating to the eighth or ninth centuries. And here, once again, we also see dramatically different results obtained from different labs for the same artifact. If this is a method whose results are truly scientific, it seems that the results should be able to be reproduced and replicated with regularity. (86-87)
How any refinements in radiocarbon dating for this region will impact our calibration of the raw radiocarbon dates for the parchments from Sanaa and anywhere else in southern Arabia remains unknown and will likely continue to remain unknown for some time to come. Therefore, it would be prudent to abandon any efforts to assign dates to these materials on the basis of radiocarbon measurements with any greater precision than a century or so for the foreseeable future. Just to give an idea of how much the differing radiocarbon levels from the southern hemisphere that would affect Sanaa seasonally could have an impact on calibration of radiocarbon age to dates CE, we give the dates calculated for each object according to the Northern and Southern hemispheric datasets with a 95 percent probability side by side (see table 4).85 In some cases, the differences are relatively minimal, yet in others they are significant. In general, one will note, the concentrations of 14C in the Southern Hemisphere yield later datings, as is the trend of this calibration dataset. In some cases, the difference is only a couple of decades; in others, dating with the data from the Southern Hemisphere could change the date of an object by a century. We would therefore be wise, I think, to use such data with greater caution than some scholars have hastily proposed and resist the temptation to misuse the method of radiocarbon analysis in attempting to date an object with greater precision than the method can presently provide. Clearly, these are all early manuscripts from the beginnings of Islam: radiocarbon dating affirms this, which we already knew. But what it cannot do, at least not as of yet, is date the text of the Qur’an in these manuscripts with any precision to a time before the very end of the seventh century, at the earliest. (Pp. 92-93)
Therefore, while radiocarbon dating adds an important new tool for studying the early manuscripts of the Qur’an, it must be used with caution, fully acknowledging its limitations and in conjunction with other methods of historical analysis. To invoke the results of radiocarbon dating as if it were the only data that matters is intellectually irresponsible and should be avoided, particularly since we have seen just how complex and often uncertain the process still remains. Indeed, Yasin Dutton similarly observes that a clear tendency can be observed in the results that leans toward dating manuscripts much earlier than otherwise seems to be likely, and he accordingly concludes as follows: “while the technique is broadly useful, it cannot be expected to yield the accuracy of dating that would be important.”91 And so, it seems, we ourselves are left to conclude that, despite the sensational claims of a few scholars, which have been amplified by the internet, the radiocarbon dating of a number of early Qur’anic manuscripts does not prove the historical accuracy of the Nöldekean-Schwallian paradigm. On the contrary, the convergence of all the presently available evidence—radiocarbon and historical— is not at all incompatible with the Qur’an’s composition into its present form only around the turn of the eighth century under the direction of ʿAbd al-Malik and al-Ḥajjāj; indeed, it would seem to favor this conclusion. (Pp. 94-95)
table 4. Calibrated Dating of Folios from Early Qur’anic Manuscripts in Sana according to Hemispheric Differences
Folio Northern Hemisphere (IntCal20) Southern Hemisphere (SHCal20)
Stanford ’07 (Arizona) 583–670 CE 602–774 CE
01-25-1 fol. 22 (Lyon) 554–645 CE 578–661 CE
01-27-1 fol. 2 (Lyon) 543–643 CE 576–660 CE
01-27-1 fol. 2 (Zürich) 598–649 CE 605–669 CE
01-27-1 fol. 11 (Lyon) 434–603 CE 529–643 CE
01-27-1 fol. 11 (Zürich) 605–660 CE 645–680 or 751–767 CE
01-27-1 fol. 13 (Oxford) 599–655 CE 607–680 or 750–768 CE
01-29-1 fol. 8 (Lyon) 436–640 CE 541–644 CE
01-29-1 fol. 8 (Zürich) 641–669 CE 654–772 CE
01-25-1 fol. 22 (Lyon) 554–645 CE 578–661 CE
01-29-1 fol. 13 (Lyon) 598–665 CE 636–773 CE
01-29-1 fol. 13 (Zürich) 605–662 CE 647–680 or 750–768 CE
Using the data from Marx and Jocham 2019, 216 rather than Robin 2015b, 65, since the former is more recent and gives more precision. (Ibid., p. 92)
FURTHER READING
Answering Islam Blog 