I sometimes weary of arguing carbon-14 with Islamic apologists, so today I am letting Guillaume Dye do it for me. Hérésies, 66f:
C14 dating problems
One would hope that a definitive solution could be provided by the C14 dating of the oldest material evidence of the Qur'anic text - even if it come only from fragments, longer or shorter. Behnam Sadeghi has procured a C14 dating of a palimpsest from Ṣan'ā' (usually cited by the name DAM 01-27.1 but which Sadeghi named, inexplicably, "Ṣan'ā'1"). By his findings, the probabilities that the parchment belongs to the period between 614 and 656 are of 68%; they are 95% for the period between 578 and 669; otherwise, the probabilities that the parchment predates 671 are 99%, they are 91.8% for the date 655; 75.1% for 645, and 56.2% for 635 (the probability becomes less than 50% for 632).
[Footnote 33: B. Sadeghi & U. Bergmann, "The Codex of a Companion of the Prophet and the Qur’ān of the Prophet", Arabica 57.4 (2010), 343-436; 348, 353. The C14 gives the date when the animal died, whose skin was used to make the parchment. By default, one admits that the interval between the killing of the beast and the copy of the manuscript can not be very long (months or years).]
The scriptio inferior does not represent the text of the 'Uṯmānic Vulgate (unlike the scriptio superior): there are some differences in the use of grammatical persons, certain suffixes, some expressions, and in the order of suras. Nevertheless, it presents a recognizable version of the Qur'anic text we can read today. If these figures are reliable, they would support the traditional version, since we have evidence that later than the 660s [CE], a substantial portion of the rasm had reached a form close to the current Koran. Other C14 datings seem to support this thesis:
[Footnote 34: Cf. http://www.islamic-awareness.org/Quran/Text/Mss/radio.html, which will be read more for the references than for the analysis of the data. A recent announcement (July 21 2014) reported a new C14 dating Koranic fragments today preserved in the library of the University of Leiden (http://www.news.leiden.edu/news-2014/oldest-koranfragments.html). The palaeography suggests a date between 770 and 830, but according to C14 dating, these fragments should be located between 650 and 715, 652 even appearing as the more likely date (but - what the article does not mention - the date 747, although in principle less plausible, is not excluded). Similar discordance (that should be taken seriously) divides the paleography and the radiocarbon in the case of the Tübingen manuscript (MA VI 165), located between 649 and 675 by the C14, and which seems rather to date from the middle of the eighth century according to paleography.]
One might think that the case is closed - but things are actually more
complicated. While radiocarbon dating is a valuable tool (this is not to reject scientific methods, which are very useful), it does not end there unless it meets at present formidable problems when it comes to date the Qur'anic manuscripts. Too much confidence in the C14 dating therefore seems unreasonable.
[Footnote 35: On the precautions and necessary-conditions in the use of C14-dating results, cf. R. E. Taylor, Radiocarbon Dating. An Archaeological Perspective (Orlando: Academic Press, 1987), 15-38, 105-146. It appears that the conditions that would legitimize an extreme confidence in radiocarbon dates are not met, at present, for old Qur'anic manuscripts.]
Two examples (not exhaustive) will suffice for this case [36: F. Déroche, Qur’ans of the Ummayads. A First Overview (Leyden: Brill, 2014), 11-14]. The first concerns the "Coran de la nourrice". We know that this manuscript preserved in the mosque of Kairouan was copied in 1020. However, the C14 analysis gives a date set, with a 95% probability, between 871 and 986, the most likely dates, in descending order, being 937, 895 and 785 (sic). The gap between the highest date (986) and the known date is only thirty-four years, which seems encouraging - but such a span, if it were required to determine whether a[nother] manuscript dates from the time either Sufyanid or Marwanid, would hardly help us. As for the date considered most probable (937), it is anterior by eighty-three years from that of the copy of the manuscript.
Another example: two other folios of the manuscript studied by Sadeghi were dated, for the one, between 543 and 643, and for the other between 433 and 599 - which poses a serious problem. [note 37 - It is therefore difficult to understand why [Nicolai] Sinai writes:
Since Déroche does not supply further details [GD : il en donne pourtant], it seems preferable on the time being to rely on Sadeghi and Bergmann’s results, although further testing is probably called for (« Part I », loc. cit., p. 276, n. 21). If the C14 dating had here the expected reliability, the most logical solution would be to locate the manuscript between 578 and 599, sole period common to the three datings. This is obviously absurd, and that the dating of Sadeghi and Bergmann appears less aberrant than others does not mean it is right (other recent analyzes provide 595-658, 566-657 and 430-611: the problem thus remains). It is not an achievement on which to base the history of the manuscript transmission of the Quran in the first century of the Hegira.]
I will not try to explain these anomalies (calibration problems, contamination at early date, eg from a carbon-based ink?) but the consequence is that the radiocarbon dates of ancient Koranic manuscripts must be taken with caution, even when they do not give aberrant results. It is thus very possible that the scriptio inferior in DAM 01-27.1 dates to the Marwanid period.
Dye goes on in p. 68 to trash similar claims for the Emesan codex "Parisino-petropolitanus", formerly known as BNF 328. I would apply the same argument to all that hype about the Birmingham Qur'an we saw here last July; that was obviously Marwanid. Had Dr Dye published this part of the essay then? I wish I'd had it, so I could have used it.