Where did the Quran come from?

Summary

• The Quran's written skeleton was fixed early. The consonantal text — the bare letters, before anyone added the dots and vowel marks that tell you how to read them — was almost certainly standardised by about 650 CE, within twenty years of Muhammad's death.¹
• Three independent lines of evidence converge on that date. Radiocarbon puts several early parchments in the seventh century.² Fourteen early manuscripts reproduce the same arbitrary spelling choices in the same places, which only happens if they all descend from one written master copy.¹ And the text itself contains nothing about the Arab conquests, the civil wars, or the fights over the caliphate that dominated everything Muslims wrote after 650.³
• The late-canonisation theories that popular Christian apologetics leans on are dead, and some of them were killed by their own authors. John Wansbrough dated the Quran's closure to about 800 CE; Marijn van Putten writes that "this position in its most extreme form is to be discarded."¹ Patricia Crone and Michael Cook's Hagarism (1977) is the single most-quoted revisionist book in Christian apologetics. Cook said in 2021, "I think we got an awful lot of it wrong."⁴ Crone wrote in her last book that "there is no longer any good reason to doubt that ʿUthmān set up a commission that produced a Qurʾān."⁵
• A Christian apologist who cites Hagarism without the retraction is misleading his readers. So is one who repeats Christoph Luxenberg's claim that the Quran's paradise maidens are really white grapes. That claim rests on re-reading a dozen unrelated verses at once, and the specialists who checked it — including scholars with no stake in defending Islam — rejected it.⁶
• Günter Lüling and Luxenberg are what people half-remember as "the earlier documents the Quran was copied from." Lüling argued that about a third of the Quran is a rewritten pre-Islamic Christian hymnbook; Luxenberg argued it is a misread Syriac church lectionary. Neither thesis survived review. The real and much better-supported question — how much biblical and Christian material the Quran draws on — is a different question.
• The live scholarly disagreement is fifty years wide, not two hundred. Stephen Shoemaker argues the text was still being shaped until about 700, under the caliph Abd al-Malik.⁷ Nicolai Sinai and van Putten argue for about 650, under Uthman. Everyone in that argument agrees there was a single written archetype; they disagree about whose it was.⁷
• What the tradition itself reports is messier than the popular Muslim claim. On al-Bukhari's own account, a verse of surah 33 went missing from the first collection and had to be recovered from one man's copy, the end of surah 9 was found with only one man, the committee was told to resolve disputes in favour of the Quraysh dialect, and Uthman ordered every other copy burnt.⁸ One of those four points has since been undercut by the manuscript evidence, and this article says so where it comes up: the Sanaa palimpsest carries the last verses of surah 9 in their expected place, which is why Sadeghi writes that "one should not read too much into the report" about Khuzayma.²

Verdict.  The mainstream academic dating of the Quran is early, and the honest reading of the evidence supports it.  A written standard existed by about 650 CE, and the text we read today descends from it with its consonantal skeleton largely intact.  The popular Christian-apologetic case for a late, invented, or borrowed Quran rests on a body of work the field has examined and largely rejected, and it should be dropped.  But an early text is not a divine text.  Dating tells you when words were written down.  It does not tell you who was speaking, and it cannot by itself settle whether a book is revelation.  Muslims and Christians both overclaim here:  Muslims when they treat an early date as proof of divine authorship, Christians when they treat a book's antiquity as something to be argued away.

• Where this article's own case is weakest. Four places. First, the radiocarbon evidence is much shakier than either side usually admits — one folio of the Sanaa palimpsest was measured by four laboratories and came back with four incompatible dates, one of them a century before Muhammad was born.⁷ Second, the palaeographers who study the handwriting keep dating these manuscripts later than the carbon dates do, and the specialist who studied the Birmingham manuscript most closely said so in print.⁹ Third, "early" applies to the consonantal skeleton only. The vowels and most of the consonant dots came later, and Sinai says flatly that reconstructing a single original Quran may not be possible at all.¹⁰ Fourth, everything between Muhammad's death in 632 and Uthman's standardisation around 650 is known only from reports written down more than a century later, and this article does not pretend otherwise.

The claim being tested

Four separate propositions usually get blurred into one.  They need to be pulled apart, because the evidence for them is very different in kind and in strength.

1. A man named Muhammad recited these words in Arabia between about 610 and 632.
2. Those recitations were gathered and written into a single standard codex under the third caliph, Uthman, around 650.
3. That codex is substantially the text Muslims read today.
4. The words originated with God and were delivered to Muhammad by the angel Gabriel.

Proposition 4 is the religious claim, and it is not a historical question.  No amount of manuscript work can confirm or refute it.  This article is about 1, 2 and 3, and about the counter-claim heard in Christian apologetics:  that the Quran is a late compilation, or a reworking of earlier Christian documents, or a book that took shape long after the man it is attributed to.

The Muslim claim, stated in Muslim terms, is that the Quran is kalam Allah — the speech of God — recited by Muhammad over about twenty-two years, memorised by his companions, written down piecemeal in his lifetime, gathered into sheets under Abu Bakr, standardised into a single codex under Uthman, and transmitted since then without loss.  The Quran itself says God will guard it:  Quran 15:9, "Indeed, it is We who sent down the message [i.e., the Qur'ān], and indeed, We will be its guardian."¹¹  Note that the bracketed words are the translator's, not the Arabic;  the Arabic word is al-dhikr, "the reminder."  Muslim commentators have read that as the Quran for a very long time, and that reading is reasonable — but the verse does not name a mechanism, a scope, or an end point.

Whether the text has been preserved since Uthman is a separate question with its own article:  has the Quran been perfectly preserved?  Whether its stories come from earlier Jewish and Christian books is a third question:  does the Quran contain material taken from earlier books?  This article stops at the point where the standard text exists.

The strongest case for the Muslim claim

A believing Muslim should recognise this section.  It is the case at full strength, before any evaluation.

The tradition's own account, in its own words

The story is told inside Sahih al-Bukhari, in the chapter titled bab jam' al-Qur'an, "the collection of the Quran."  Al-Bukhari (d. 256/870) prints no grading line on his own material, because inclusion in his collection is itself the claim being made.

Stage one: the scraps.  Zayd ibn Thabit was Muhammad's scribe.  Bukhari 4986 has him list what he collected from:  palm stalks, thin white stones, and the chests of men — al-'usub wa-l-likhaf wa-suduri l-rijal.¹²  A parallel report at Bukhari 7191 adds leather and hides to the list.

Stage two: Abu Bakr's sheets.  After the battle of Yamama in 12/633 killed many of the men who had memorised the Quran, Umar pressed Abu Bakr to have the text collected.  Zayd's account is candid about the objection he raised twice:  "How can you do something which Allah's Apostle did not do?"¹²  The resulting sheets passed to Umar and then to his daughter Hafsa, one of Muhammad's widows.

Stage three: Uthman's standard edition, about 30/650.  Bukhari 4987 gives the reason and the method.  Hudhayfa ibn al-Yaman came back from the Armenian and Azerbaijani front alarmed at how differently the Syrian and Iraqi troops were reciting, and told the caliph:  "O chief of the Believers!  Save this nation before they differ about the Book (Qur'an) as Jews and the Christians did before."  Uthman borrowed Hafsa's sheets, appointed a committee of four — Zayd plus three Qurashis — told them that where they disagreed with Zayd they should write the word in the dialect of Quraysh, sent one copy to each province, "and ordered that all the other Qur'anic materials, whether written in fragmentary manuscripts or whole copies, be burnt."¹³

That is the traditional account.  It is not a claim that the text fell from the sky.  It is a claim about an editorial process with named participants, a stated motive, and a documented outcome.

What modern evidence has done for that account

Here is the part that surprises people who expect scholarship to be hostile.  Over the last two decades the physical and linguistic evidence has moved toward the traditional dating, not away from it.

The single written archetype.  Marijn van Putten, a linguist at Leiden, published the strongest single argument in the field in 2019, and it is open access.  Quranic spelling contains choices that carry no information.  Feminine nouns in the construct state are usually written with a final ha', but about 22 per cent of the time with a ta' instead.  For the word ni'mah, "grace," the split is nearly even:  twelve one way, eleven the other.  There is no rule.  It is scribal whim.¹

Van Putten checked all twenty-three occurrences across fourteen early manuscripts.  Every manuscript spells each occurrence the same idiosyncratic way, with disagreement at only two locations in the whole set.  His conclusion:

There is only one possible explanation for the strong agreement across the many different Quran manuscripts with the two possible spellings of niʿmat: there must have been a single written archetype from which all Quranic manuscripts of the Uthmanic text type are descended.¹

And the second half, which matters just as much:

It is exactly these kinds of orthographic idiosyncrasies that would be impossible to reproduce through a process of writing down from dictation.¹

He replicated the pattern with three other quirks:  rahmat and la'nat behave the same way;  jaza'u in construct is spelled one way except in exactly five places;  al-mala'u is spelled one way except in exactly four.¹  This is a stemmatic argument, of the kind classicists use to reconstruct lost manuscripts.  It does not appeal to tradition at all.  His dating conclusion:  "The manuscripts examined in this study are sufficiently early that a codification of the Uthmanic text type is perfectly consistent with an attribution to its traditional source: ʿUthmān b. ʿAffān."¹

In his 2022 book he states the position for the whole field:  "While the manuscript evidence cannot preclude with absolute certainty an earlier redaction, a time much later than his reign is now a virtual impossibility."¹⁴

The regional codices form a family tree.  Michael Cook showed in 2004 that the consonantal variants medieval Muslim scholars reported for the Medina, Damascus, Kufa and Basra master copies form a clean stemma with no cross-contamination — Damascus shares readings with Medina but never with Kufa or Basra.  Medieval Muslim scholars had no concept of stemmatics, so the pattern cannot have been manufactured.  Yasin Dutton then found the same Syrian variant set in actual manuscripts.¹

The Sanaa palimpsest vindicates the tradition's variant reports.  In 1973 restoration work on the Great Mosque of Sanaa opened a cavity holding around 15,000 parchment fragments.  Among them was a Quran written over an erased Quran.  When Behnam Sadeghi and Mohsen Goudarzi finally published the erased lower layer in 2012, it turned out to be a genuine non-Uthmanic codex — the only one known.  Their verdict:  it is "direct documentary evidence for the reality of the non-ʿUthmānic text types that are usually referred to as 'Companion codices,'" and its variants "are of the same kinds as those reported for those codices."²  In other words, the ninth-century Muslim literature listing what Ibn Mas'ud and Ubayy read differently was reporting something real, not inventing it.

Sadeghi's stronger finding cuts the same way.  Comparing the palimpsest with the standard text, he concluded that the standard text is the better witness to what Muhammad recited:  "textual criticism suggests that the standard version is the most faithful representation, among the known codices, of the Qurʾān as recited by the Prophet."¹⁵  That is a more conservative conclusion than classical Muslim scholarship itself reached, since the classical tradition put the Companion codices on a par with Uthman's.

The suras were already assembled before Uthman.  This is the finding that most directly contradicts the older Western picture.  Sadeghi and Bergmann:  "The most secure conclusion of the present study is that the sequences of verses and sentences were fixed already in the Prophetic prototype."¹⁵  The palimpsest has a different chapter order but the same verses in the same order inside each chapter — and Sadeghi points out that "even a careful memorizer who reproduces the words exactly is prone to getting the order of the verses wrong."²  The one true oddity in the standard Quran, the missing basmala at the head of surah 9, is already there in the palimpsest, so it was not Uthman's committee's invention.²

The text is silent about everything that happened after 632.  Fred Donner's argument from Narratives of Islamic Origins is a counterfactual:  if the Quran had crystallised slowly in Iraq over two centuries, it would look different.  The hadith literature is saturated with the political fights of the first two Islamic centuries — obedience to the imam, what to do with a tyrant, what to do when two men claim the caliphate.  The Quran has essentially nothing.  "The discrepancy between Qur'an and hadith on the question of political leadership is striking, and suggests strongly that the two bodies of material are not the product of a common 'sectarian milieu.'"³  His summary in the later book:  the Quran "did coalesce very early in the history of Muhammad's community — within no more than three decades of Muhammad's death," because "meticulous study of the text by generations of scholars has failed to turn up any plausible hint of anachronistic references to important events in the life of the later community."¹⁶

The rough edges were never smoothed.  Sinai's list is short and pointed.  Quran 2:177, 4:162, 5:69 and 20:63 break classical case agreement.  Quran 3:96 spells the sanctuary bakka where everyone expects makka — a one-letter fix that nobody made.  If anyone had revised the text around 700, those are the first places a reviser would have touched.¹⁷

The tradition preserves attacks on Uthman that nobody would invent.  Gregor Schoeler's argument, which Sinai calls his strongest:  the sources preserve accusations that Uthman "burnt the book of God" and called him harraq al-masahif, "burner of the codices," and they preserve defences answering those accusations.  "The very fact that people took the trouble to fabricate them would presuppose that the accusation which they are designed to dispel … was very much in the air."¹⁷  An invented event does not generate a century of hostile polemic and apologetic reply.

One author, on statistical grounds.  Behnam Sadeghi ran the most serious stylometric test the Quran has had:  four independent style markers — mean verse length, and three non-overlapping lists of 28, 114 and 3,693 morphemes — measured across a proposed chronological ordering of the text.  All four vary smoothly across seven phases.  His conclusion on authorship:

One would conclude that style backs the hypothesis of one author.  For the sake of argument, suppose there were two authors ... Then one would have to say that the style of A moved along a trajectory towards that of B, or that B picked up where A stopped, with respect to not only verse length, but also frequencies of the most common morphemes and frequencies of uncommon words.  It is much easier to imagine a single author.¹⁸

Patricia Crone changed her mind, in print, at the end of her life.  In the preface to her collected studies, written shortly before her death in 2015, the most famous sceptic in the field wrote:

To me, this was a breakthrough.  The palimpsest undoubtedly came from a complete Qurʾān.  There was a complete Qurʾān by the second half of the seventh century.  It was not identical with the one we have today in every detail, but the variants do not change the fact that it is the same book. … But for all that, we have a hard fact: the Qurʾān existed by the time when the tradition says it existed.  There is no longer any good reason to doubt that ʿUthmān set up a commission that produced a Qurʾān.⁵

The revisionist chronologies, stated at full strength

Popular Christian apologetics draws on a body of work known as revisionism.  The label covers at least four separable claims, and running them together is the commonest error in this whole subject.

An article that cites a revisionist for the fourth claim is on solid ground.  An article that cites the same author for the second is usually citing something he or his colleagues have retracted.  What follows states each position from its author's own book.

John Wansbrough (1928–2002): canonisation around 800

Wansbrough taught at the School of Oriental and African Studies in London and applied Old Testament form and redaction criticism to Islam's sources.  In Quranic Studies (Oxford, 1977) he argued that the Quran is composite, that "the period required for its achievement was rather more than a single generation," and that the canon closed only at the beginning of the third Islamic century.¹⁹  His reasoning was an argument from the silence of the literature:  Islamic law was not derived from scripture until the ninth century, the early creed known as Fiqh Akbar I never cites the Quran, and the whole apparatus of variant-reading scholarship appears only in the ninth century.

Derivation of law from scripture (halakhic exegesis) was a phenomenon of the third/ninth century, and while the obvious inference is admittedly an argumentum e silentio, the chronology of the source material demands that it be mentioned. … It is of course neither possible, nor necessary, to maintain that the material of the canon did not, in some form, exist prior to that period of intensive literary activity, but establishment of a standard text such as is implied by the ʿUthmanic recension traditions can hardly have been earlier.¹⁹

On the Uthmanic story itself he offered a fork:  either the suppression of rival texts was "so instantly and universally successful that no trace of serious opposition remained, or … the story was a fiction designed to serve another purpose."¹⁹  And he located the whole process not in Arabia but in Mesopotamia, in a "sectarian milieu" of arguing Jewish and Christian groups.¹⁹

His strongest argument is real and nobody disputes it.  The chronological gap is enormous.  The earliest connected biography of Muhammad is Ibn Hisham's recension (d. 218/833) of Ibn Ishaq (d. 150/767).  The hadith canons are late ninth century.  The exegetical literature is ninth century.  Wansbrough's move was to say that a literature this late, this internally contradictory, and this obviously shaped by later doctrinal need cannot license conclusions about the seventh century at all.

Crone and Cook, Hagarism (1977)

Patricia Crone (1945–2015) and Michael Cook (b. 1940) published Hagarism: The Making of the Islamic World with Cambridge University Press in 1977.  Their method is stated in the preface:  "This is a book written by infidels for infidels, and it is based on what from any Muslim perspective must appear an inordinate regard for the testimony of infidel sources."²⁰  Their opening move is the sentence Christian apologists quote most:

There is no hard evidence for the existence of the Koran in any form before the last decade of the seventh century, and the tradition which places this rather opaque revelation in its historical context is not attested before the middle of the eighth. … The only way out of the dilemma is thus to step outside the Islamic tradition altogether and start again.²⁰

Starting again produced a reconstruction in which Muhammad preached Jewish messianism with Umar as the messiah, the movement was a Judeo-Arab alliance aimed at the Holy Land, the hijra was an emigration out of Arabia rather than from Mecca to Medina, and the original sanctuary was in north-west Arabia rather than at Mecca.²⁰

Günter Lüling (1928–2014): a Christian hymnbook under the Quran

This is one of the two theories people half-remember as "fragments of earlier documents the Quran was copied from."

Lüling's Über den Urqur'an (1974), later expanded in English as A Challenge to Islam for Reformation (2003), argued that about one third of the present Quran has a pre-Islamic Christian text underneath it — specifically a strophic hymnal of the early sixth century, anti-Trinitarian in doctrine, composed by Christians in a Christian Mecca with a Christian Ka'ba and a Christian Quraysh.  He held that the Quran contains four distinct layers of language, that the transmitted text is the last of several editorial revisions, and that the original Islamic movement was a move away from Hellenistic Trinitarian Christianity back to an older Semitic tribal religion.²¹

His method is the interesting part.  Because early Arabic was written without dots or vowel marks, the bare letters underdetermine the reading.  Lüling proposed eight editorial techniques by which, he said, a Christian hymn could be converted into Islamic prose:  re-vowelling, re-pointing consonants, slightly deforming the letters, adding or dropping words, exploiting orthographic variation, exploiting words whose Hebrew, Syriac and Arabic senses differ, assigning meanings by loose association, and simply ignoring syntax.  Restoring strophic form and rhyme was then supposed to recover the original.²¹

Christoph Luxenberg: a Syriac lectionary

The pen name belongs to a scholar who has never published under his own.  Die syro-aramäische Lesart des Koran (2000), in English as The Syro-Aramaic Reading of the Koran (2007), argues that the language of the Quran is not Arabic at all but a mixed Aramaic-Arabic idiom, and that the transmitted reading of many passages is a later misreading of a Syriac original.  In his own summary:

Contrary to the earlier assumption of a dialect of Arabic spoken in Mecca, the present study has shown that, insofar as the Arabic tradition has identified the language of the Koran with that of the Quraysh, the inhabitants of Mecca, this language must instead have been an Aramaic-Arabic hybrid language. … This would lead one to assume that Mecca was originally an Aramean settlement.²²

He denies the oral tradition outright:  "The unsuspected extent of the misreading that has come to light in connection with numerous Koran passages raises the question of the authenticity of the previously alleged oral Arabic tradition. … On the contrary, this necessitates the assumption from the beginning of a text transmitted in writing."²²

The lectionary claim is in the name.  Qur'an, he argues, is the Syro-Aramaic qeryana, a church lectionary — a book of readings for the liturgy.²²  And the single most-repeated item in popular apologetics comes from here.  Quran 44:54 and 52:20 promise the blessed hur 'in, traditionally "wide-eyed maidens."  Luxenberg re-lexicalises the phrase through Syriac as white, crystal-clear grapes.  His own words are "White" (grapes): "jewels" (or rather) "crystals," not the "white raisins" of the newspapers.²²

Nevo and Koren, and the argument from documentary silence

Yehuda Nevo (d. 1992) was an archaeologist who surveyed Arabic rock inscriptions in the Negev;  Judith Koren completed Crossroads to Islam (2003) after his death.  Their four conclusions:  the Arabs took the Byzantine east without major battles because Byzantium had already withdrawn;  they were pagan at the takeover and the elite then adopted a simple monotheism that became Islam over 100 to 150 years;  Muhammad "is not a historical figure, and his official biography is a product of the age in which it was written";  and "The Qur'an is a late compilation; it was not canonized until the end of the 2nd century A.H. or perhaps early in the 3rd."²³

The original contribution is the epigraphy.  Nevo sorted hundreds of Negev rock inscriptions — mostly private prayers written by ordinary people — into classes by creed.  The earliest class, from about 40–60 AH onward, he called "Indeterminate Monotheism":  monotheistic prayers with no distinguishing marks of any known religion, and no Muhammad.²³  Muhammad appears later, and recognisably Islamic formulae later still.  Their defence of the argument from silence is explicit:  "if one is trying to ascertain whether the event took place at all, the situation is different.  One unshakeable contemporary account would settle the question, and the fact that there is none, if demonstrable, becomes 'positive' evidence."²³

The factual core of that observation is correct and worth stating plainly:  Muhammad's name does not appear on a coin, in a state document, or on a monument before the 660s at the earliest and the 690s in state use.²⁴

The Inarah circle: Muhammad as a title for Jesus

Karl-Heinz Ohlig, Volker Popp and Gerd-R. Puin form the core of a Saarbrücken-based group whose English-language flagship is The Hidden Origins of Islam (2010).  Ohlig starts from the ordinary source-critical point — reports written two hundred years after the fact "should be taken into consideration only with great reservations" — and then goes much further.  He argues that the era beginning in 622 was originally an Arabian-Christian reckoning of time later converted to a Muslim meaning, that the Umayyad and even the early Abbasid rulers were Christians, and that the idea of Muhammad was detached from its original connection to Jesus only in the second Islamic century.²⁵

Luxenberg supplies the philological core in the same volume.  The Dome of the Rock inscription of 72 AH contains muhammadun 'abdu llahi wa-rasuluhu.  Luxenberg reads muhammad not as a name but as a gerundive participle — "the praised one" — and takes the clause as a Christological doxology about Jesus, on the grounds that the surrounding inscription is a chain of quotations about Christ.²⁵

Their strongest datum is genuine.  The 691–92 Dome of the Rock inscription really is the earliest securely dated occurrence of muhammad, and it really does sit in an aggressively anti-Trinitarian, Jesus-focused text.  Hoyland's edition reproduces the whole chain:  Quran 4:171 and 4:172 on the Messiah as a messenger and not a son, Quran 19:33–36 on Jesus, and Quran 3:19.²⁴  Any account of the monument has to explain why a caliph proclaiming a new religion filled his great building with polemic about Christ.

Stephen Shoemaker: closure under Abd al-Malik, about 700

Shoemaker is the serious current case.  He is a professor of religious studies at Oregon and publishes with Pennsylvania and California — the only figure on this list who is fully inside the academic conversation.  Creating the Qur'an (2022) is open access.  His conclusion:

The bewildering confusion and complexity of the early Islamic memory of the Qur'an's formation, as we saw in the first two chapters, only reaches some level of clarity once we recognize ʿAbd al-Malik as the primary agent responsible for producing and enforcing the canonical textus receptus of the Qur'an.  Under his supervision, a team of scholars wove together and honed the various sacred traditions that had entered circulation among Muhammad's followers during the seventh century, creating a new imperial Qur'an that was imposed across the caliphate, displacing its antecedents in the process, often by force.⁷

The argument has five legs.  The collection reports are mutually incompatible, which is itself evidence that no memory of the event survived.  There is an independent and well-attested Islamic tradition crediting the standardisation to Abd al-Malik's viceroy al-Hajjaj — "In stark contrast to the disarray of reports concerning earlier individuals, the canonical Qur'an's composition during ʿAbd al-Malik's rule (685–705) seems highly credible."⁷  Radiocarbon dating is unreliable.  The Hijaz was too small, too poor, too illiterate and too free of organised Christianity to have produced a text this dense with Christian content.  And modern memory science shows that oral communities constantly and unconsciously revise what they transmit.

Shoemaker is candid about the book he is building on.  The Death of a Prophet (2012) opens by conceding that "The imperfections of Hagarism should not lead us to discount completely the important insights that both this study and its approach have to offer," and complains that Crone and Cook noticed a striking convergence in the non-Muslim sources and then did nothing with it beyond the remark "The convergence is impressive."²⁶

His strongest leg is the fourth, which Sinai has nicknamed the Christian elephant in the Meccan room.  Sinai — who argues against him — concedes the problem is real:  "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."²⁷

What happened to those theories

This is the part popular apologetics leaves out.

Wansbrough was refuted by manuscripts he did not have

Van Putten's 2019 paper opens by naming him:

This was famously challenged by Wansbrough (1977), who posited that the collection of the canonical text can only have taken place as late as two or three centuries after the date traditionally assigned to it.  With the discovery and description of many early Quranic manuscripts that are clearly earlier than the late date proposed by Wansbrough, it is now clear that this position in its most extreme form is to be discarded.¹

Sinai calls Wansbrough's dating "by now an easily vanquished straw man."²⁸  Sadeghi's aside in 2012 is sharper, and it is aimed squarely at people who use manuscripts selectively:  "Those who discuss both Wansbrough's theories and early manuscripts do not draw the elementary inference that the latter refutes the former: they are interested in the manuscripts only because they believe they refute traditional views."²

What survives of Wansbrough is his genre analysis — the observation that the Muslim biographical literature is salvation history, a genre with its own conventions, not a chronicle.  That contribution is real and is still used.  His dating is not, and citing it as live scholarship is a factual error.  Wansbrough himself was hostile to Hagarism, on the ground that Christian and Jewish apocalyptic texts are themselves salvation history and cannot be used as a neutral control.²⁹

Crone and Cook disowned Hagarism, on the record, in their own voices

Michael Cook, interviewed by Intisar Rabb for the Islamic Law Blog on 26 January 2021 and asked why he would not write Hagarism now:

The short answer is, you know, I think we got an awful lot of it wrong.  But, nevertheless, the thinking that I did back then has been kind of useful to me quite often down the years …

Why not now?  Because I think we got a lot wrong.  That's easy. … And I think I said something like, I like to think over the years you get a bit wiser.  So that's why not now.⁴

Patricia Crone, writing for openDemocracy on 31 August 2006:

There is no doubt that Mohammed existed, occasional attempts to deny it notwithstanding.  His neighbours in Byzantine Syria got to hear of him within two years of his death at the latest; a Greek text written during the Arab invasion of Syria between 632 and 634 mentions that "a false prophet has appeared among the Saracens" and dismisses him as an impostor on the ground that prophets do not come "with sword and chariot".³⁰

And on the Quran:

Most importantly, we can be reasonably sure that the Qur'an is a collection of utterances that he made in the belief that they had been revealed to him by God. … But that he uttered all or most of them is difficult to doubt.  Those who deny the existence of an Arabian prophet dispute it, of course, but it causes too many problems with later evidence, and indeed with the Qur'an itself, for the attempt to be persuasive.³⁰

Ten years later she added the argument that finally moved her off the late-dating position.  The earliest Muslim commentators plainly do not understand some legally important Quranic words.  She had once read that as evidence the text was elevated to scripture late.  She reversed it:

If they all reproduce the same text without understanding what it means, the obvious inference is that they are dealing with scripture that was already old in their time, not that it was elevated to scriptural status later.  This is my position now.⁵

This matters for how Christians argue.  Hagarism is quoted constantly in apologetics aimed at Muslims, almost always without the retraction.  Quoting a book whose surviving author has said "I think we got an awful lot of it wrong" is not evidence.  It is the same move Christians rightly complain about when a critic quotes a New Testament scholar's discarded early position as though it were his settled view.  Drop it.

What Crone did not retract should be reported with equal care, because it is a live problem and it is not in Islam's favour.  She never withdrew the Mecca argument:  "Not a single source outside Arabia mentions Mecca before the conquests, and not one displays any sign of recognition or tells us what was known about it when it appears in the sources thereafter."  She added that the Quran describes its opponents as farmers of wheat, grapes and olives — Mediterranean staples that Mecca cannot grow.³⁰  Her mature position is therefore:  Muhammad certainly existed, the Quran is substantially his, the tradition is broadly right in outline — but the geography is a problem.  That is what should be attributed to her.

Lüling's thesis was rejected even by the volume convened in his honour

Lüling submitted his doctorate at Erlangen in 1970 and received the highest mark.  Within a fortnight the Habilitation it entitled him to was withdrawn.  He submitted the full collection of reconstructed texts as a Habilitationsschrift in 1973;  the Faculty of Philology voted it down in February 1974;  he litigated for years and was finally dismissed from the university in July 1978.²¹  That chronology comes from Lüling himself, relayed by an English exposition of his work, and its most damaging element — that the Munich Arabist Anton Spitaler engineered the outcome — is his own charge, not a documented finding.  Part of it is independently corroborated:  Zellentin cites a January 1970 memorandum in which Spitaler, refereeing a job application, spent three pages attacking the grammar of Lüling's reconstructions, and the same De Gruyter volume carries a separate chapter on the legal history.³¹  That is a story about academic politics, and it deserves sympathy.  It is not an argument for the thesis.

The thesis itself has been tested.  In 2019 Georges Tamer convened a De Gruyter volume specifically to give Lüling a fair hearing.  Holger Zellentin's contribution is a full-length assessment of Lüling's reading of surah 96.  His opening:

I will thus not hesitate to dismiss the main ideas put forward by Dr. Lüling almost in their entirety, and invite scholars as well as the broader public to spend their energy on more mature approaches.³¹

Zellentin is careful to credit what Lüling got right — "The extent to which Lüling's methodology pre-empted important developments in the field stands in tragic contrast to the dismissive ways in which his work is widely (and, in my view, justly) treated."³¹  But his verdict on the whole genre is the one to keep:

Any attempt to reconstruct a text behind the text, in the absence of any objective tools such as manuscript variants, is by necessity circular.  In the best of cases, such as that of Younes, the results of speculative philology are interesting; in the worst ones, such as those of Lüling and Luxenberg, they are vexing.³¹

That sentence names the real problem.  If you may re-vowel, re-point, add and drop letters, and switch between three languages' lexicons at will, you can reconstruct almost any text under almost any other.  Without manuscript evidence for the postulated original, there is no way to check the result.

Luxenberg's philology did not survive review

Four specialist reviews appeared:  François de Blois, Claude Gilliot, Simon Hopkins and Federico Corriente.  Devin Stewart, who is more sympathetic to emendation than most, summarises them:  "De Blois is most caustic, Hopkins harsh but more restrained, Corriente critical but somewhat sympathetic, and Gilliot the most sympathetic of the lot."³²

De Blois's objection is grammatical and elementary.  Luxenberg reads a final alif in certain Quranic words as an Aramaic determinate-state ending rather than an Arabic accusative.  Applied to the sentence at issue, the result "is neither classical nor modern Arabic, but simply wrong."  His broader complaint:  once the mixed-language premise is granted, "the author evidently thinks it possible to take any Arabic word that vaguely resembles something in Syriac and to determine its meaning not from the Arabic but from the Syriac lexicon."  His verdict:  "His book is not a work of scholarship but of dilettantism."³³

Walid Saleh's chapter in The Qur'an in Context is the fullest published answer, and it comes from a scholar of Quranic exegesis, not an apologist.

The work has fundamental flaws in its methodology, indeed if one could call such a hash of a work methodical in any way, in itself reason not to bother reviewing it.  The book, however, has generated such excitement—never before has a work on Qurʾanic studies been featured in major European and American newspapers and magazines—that a refutation of its faulty premises has to be undertaken.⁶

Luxenberg's chief warrant is that al-Tabari repeatedly writes that "the people of exegesis are in discord" about a verse, which supposedly shows the Arabic was already opaque.  Saleh's answer:  "Thus to come across the sentence 'the people of exegesis are not agreed on the meaning of this verse' in aṭ-Ṭabarī's Jāmiʿ al-bayān is like finding fish in the sea."⁶

On the grapes specifically, three arguments settle it.  First, the maidens of paradise are not something the exegetical tradition is confused about, so the warrant for a fresh reading is absent in this case.⁶  Second, to make the grape reading work Luxenberg must re-read the ordinary words around the phrase — including the verbs for "marry" and "be purified."  Saleh counts the cost:  "He has to do this in at least thirteen instances, rendering the Qurʾanic text absolutely unstable."⁶  Stewart, working independently, reaches the same objection:  the fact that Luxenberg "must emend so many individual items in texts that are not adjacent or connected by any physical pattern calls the entire complex of emendations into question."³²  Third, and decisively, the Quran itself supplies a synonym.  Quran 52:24 gives ghilman where other verses give wildan for the youths of paradise:

the Achilles' heel of Luxenberg's analysis is Q 52:24, where a synonym, ghilmān for wildān, is given. … if the consonantal Qurʾanic text is giving a synonym to a word then it must mean that that word, wildān, was understood at the time of the fixing of the codex (if not at the time of its writing) to mean what the synonym means, i.e. youths.⁶

Stewart adds the philological point that kills the reading on its own terms:  Luxenberg thinks it odd to describe beautiful women's eyes as "white," but that is not what the Arabic says.  Hawar "refers to a very particular beauty trait, a strong contrast between the whites of the eyes and the dark irises and pupils."³²

Saleh's overall verdict:  "Luxenberg's method is oracular, not philological. … Luxenberg's work is in the final analysis a Christian polemical tract."⁶

Crone, who had every professional reason to welcome a radical thesis, wrote that Lüling's and Luxenberg's books "are open to so many scholarly objections (notably amateurism in Luxenberg's case) that they cannot be said to have done the field much good."³⁰

Stewart is worth reading for what he grants.  He accepts one of Luxenberg's emendations — at Quran 37:78 and its parallels, reading wa-barakna, "We blessed them for posterity," instead of wa-tarakna — noting it requires no change to the written letters at all, and that Jacob Barth proposed it in 1916.³²  Emendation of the Quran is a legitimate scholarly tool.  Luxenberg's use of it is not.

The Inarah circle is a parallel conversation, not part of the field

There is no substantial engagement with Ohlig, Popp and Puin in the peer-reviewed Quranic-studies literature.  Their members publish mainly with Hans Schiler and Prometheus rather than with university presses.  The three pillars each fail independently.  Reading muhammad as a title has to survive the Arab-Sasanian and Arab-Byzantine coin legends, where the phrase sits inside a governmental profession of faith.  The Christian-Umayyad claim requires that the entire seventh- and eighth-century Syriac, Greek and Armenian record — which distinguishes the new rulers from Christians constantly and often bitterly — be misread.  And the philology carrying the whole structure is Luxenberg's.

The one item worth keeping from the volume is the plain fact about the Dome of the Rock's date and its Christological content, which can be sourced from any standard edition without adopting Ohlig's reading of it.

Puin's 1999 expectation, and the study that followed

In January 1999 Toby Lester's Atlantic article introduced English-speaking readers to Gerd-R. Puin, the German palaeographer who had worked on the Sanaa fragments.  Puin's remarks are genuine and are quoted accurately here:

My idea is that the Koran is a kind of cocktail of texts that were not all understood even at the time of Muhammad.  Many of them may even be a hundred years older than Islam itself.  Even within the Islamic traditions there is a huge body of contradictory information, including a significant Christian substrate; one can derive a whole Islamic anti-history from them if one wants.³⁴

So many Muslims have this belief that everything between the two covers of the Koran is just God's unaltered word. … The only way to break through this wall is to prove that the Koran has a history too.  The Sana'a fragments will help us to do this.³⁴

Twenty-seven years later, that is still an expectation, not a finding.  Puin never published the systematic study he described.  Sadeghi and Goudarzi did, in 2012, and their results run the other way:  the lower text carbon-dates before 671, its verses and their order inside the suras are, with two exceptions Sadeghi names, those of the standard text, its variants match the kinds reported in the Muslim variant literature, and the standard text is the better witness to the common prototype.  On Puin's own output they note:  "G. Puin's scholarly output on the Ṣanʿāʾ collection consists of three publications in which he says very little about the manuscripts and does not discuss the palimpsest."²  His working assumption, they add, was "that the standard reading is a corruption in every case in which there is some other reading in any manuscript."²

An article that quotes the 1999 interview without the 2012 study is reporting a twenty-seven-year-old prediction as though it were a result.

The evidence that actually dates the text

Three kinds of evidence bear on when the Quran became a fixed book:  physical dating of the parchment, the handwriting on it, and the internal features of the text itself.  Each has real limits, and both apologetic camps routinely overstate the first.

Radiocarbon: real numbers, real error bars, and a real problem

Radiocarbon dates the animal whose skin became the parchment.  It does not date the ink.  A scribe could write on stock decades old;  a parchment could also be scraped and reused, which is exactly what happened to the Sanaa palimpsest.

The Sanaa palimpsest.  The most-quoted measurement is from the Arizona AMS laboratory on a folio auctioned in London and studied at Stanford in 2007.  The raw figure is 1,407 ± 36 years BP, calibrating to 614–656 CE at 68 per cent confidence and 578–669 CE at 95 per cent.¹⁵  Sadeghi then computed one-sided probabilities:  the parchment is older than 646 with probability 75.1 per cent, older than 655.5 with 91.8 per cent, older than 660.5 with 95.5 per cent.  His conclusion:  "The probability that the parchment is older than AD 646 is 75.1 %, or a three-to-one likelihood."¹⁵  In the 2012 edition he states the same result as:  the parchment dates "to the period before AD 671 with a probability of 99% (before 661 with the probability of 95.5%, and before 646 with a probability of 75%)."²

Now the honest part.  Nine radiocarbon measurements have been published on folios of that one codex, from five laboratories, and they do not cohere.⁷

Folio 13 alone was measured four times, at four laboratories, and the four ranges do not overlap in any single year:  Lyon put it at 388–535, Kiel at 430–495 or 530–610, Zürich at 590–650 and Oxford at 595–658.  Shoemaker's comment deserves to be quoted rather than paraphrased:  "these assays have yielded significantly different results for the very same parchment leaf!  Such results hardly settle the matter but instead only confirm that, again, something clearly is not working correctly with this method."⁷  Déroche's judgement on the impossibly early Lyon results, in his own words rather than through Shoemaker's paraphrase, is that they are "results which simply cannot be accepted," and his suggested cause — that "the problem may lie with the conditions (arid or semi-arid climate) under which the cattle, the hides of which were later turned into parchment, was raised" — is speculation, as he does not conceal.³⁵

But the incoherence must not be oversold either, and Déroche is the reason.  On the same page he reports a controlled test that the apologetic use of "radiocarbon does not work" always leaves out.  He took a sample from each of three disputed Qurans, cut each sample in two, and sent one half of each to a different laboratory.  The two laboratories agreed closely on all three:  677–858 against 676–869, 671–773 against 650–764, and 684–867 against 672–853.  The method reproduces when it is tested that way.  His conclusion is a caution about weight, not about validity:  "although recent publications seem overconfident in their reliance on the C14 method, the last word should stay with the philologist, the historian or the palaeographer."³⁵

Christian apologetics on the Sanaa numbers is usually wrong in a specific way.  The figures "543–643" and "433–599" circulate as though they were the date of the palimpsest, or as though they measured the gap between its two layers.  They are two of the nine measurements above — both from Lyon, the laboratory whose third result on the same codex returned 388–535 CE, before Muhammad was born.  Give the whole table or none of it.  And no radiocarbon measurement dates the upper text at all;  the upper layer is dated by handwriting alone, so the gap between the two layers is simply unmeasured.²

Birmingham.  In July 2015 the University of Birmingham announced that "Radiocarbon analysis has dated the parchment on which the text is written to the period between AD 568 and 645 with 95.4% accuracy.  The test was carried out in a laboratory at the University of Oxford."³⁶  The raw figure is 1,456 ± 21 BP;  recalibrated with the current IntCal20 curve the range is 578–646 CE, which is the figure Corpus Coranicum now publishes.⁷

The press release went further than the evidence, and that overreach is the source of most subsequent confusion.  David Thomas was quoted saying the text "must have been in a form that is very close to the form of the Qur'an read today, supporting the view that the text has undergone little or no alteration."  Moving from a parchment date to a text date is precisely the step specialists reject.

Palaeography keeps coming out later than the carbon

Alba Fedeli wrote the primary study of the Birmingham leaves, and she is not a hostile witness.  Her conclusion from the handwriting is that the scribe was copying from a written exemplar and doing it well, which itself implies elapsed time:

The mechanism of copying from an exemplar implies consequently that the work could not have been executed very early, as the written exemplar requires a period of time for producing the exemplar and also the establishment of a mechanism of copying from an authoritative text.  Moreover, the regular and coherent presence of a blank line between two sūras seems to be interpreted as a sign of a later practice, as it was introduced and established during the so-called second maṣāḥif project accomplished in the period between 84-85 AH (703-705 CE), whose main initiator was al-Ḥajjāj (d. 95/713) as interpreted by Hamdan.⁹

She then says it outright:  her palaeographic hypothesis "disagrees with the results of the carbon 14 analysis, which reveals a high probability of an earlier date."⁹  She lists other mismatches:  one Sanaa manuscript carbon-dated 657–690 has "clear palaeographical features from 710-715 CE";  a group of giant Qurans Déroche dated to the second half of the eighth century came back 640–765;  a manuscript with an endowment note reading 907 CE was dated by Lyon to 716–891.⁹

Déroche, the field's leading codicologist, is equally cautious in the other direction.  He rejects the early carbon date for the Sanaa palimpsest on palaeographic grounds:  the carbon results "are not conclusive since the comparative material suggests in this case a date in the second half of the first/seventh century."³⁵  His best-documented control case is a manuscript called the Qur'an of the Nurse, whose colophon and endowment deed date it firmly to 410/1020 and whose radiocarbon result was 871–986 CE at 95 per cent — eighty-three years too early.³⁵

Shoemaker draws the fork bluntly.  If the parchment date is the text date, the Birmingham leaves land partly before Muhammad's mission:  "either the Qur'an is largely pre-Muhammadan, a possibility that by no means should be completely excluded, or we must accept the fact that radiocarbon dating cannot provide us with a basis for dating the texts of the earliest Qur'ans with much precision at all."⁷

The honest position is Fedeli's.  The parchment is old.  The writing on it is probably a generation or two younger.  Neither fact rescues or wrecks anybody's theory, and a Christian apologist cannot run "Birmingham proves the Quran predates Muhammad" and "the carbon dates are unreliable" in the same lecture.

What the Sanaa palimpsest actually contains

This is the only surviving non-Uthmanic Quran, so its contents matter more than its date.

It is a real codex, not a notebook.  Its lower text runs continuously through whole suras — part of surah 8, all of surah 9 with illegible gaps, then surah 19.  Sinai:  "Such a sequential succession of verses and suras is not what one would expect of a disparate selection of quranic material copied out as a scribal exercise, but is entirely consonant with the process of producing a complete scriptural codex (muṣḥaf)."³⁷

Its chapter order is different.  The eleven surviving transitions give runs like 11 → 8 → 9 → 19, 12 → 18, 15 → 25, 34 → 13, 63 → 62 → 89 → 90.  Some match the order reported for Ubayy ibn Ka'b's codex, some Ibn Mas'ud's, some neither.²

Its wording differs in hundreds of places.  On the four London folios alone Sadeghi and Bergmann counted 35 differences in minor elements — particles, pronouns, prefixes — and 25 cases where a whole word or phrase is different.  Eighteen of those 25 sound like the standard reading:  same root, different conjugation, singular for plural, active for passive.¹⁵

Two of those differences touch law, and Christians should not pretend otherwise, nor Muslims deny it.  Both are at Quran 2:196, the pilgrimage verse.  The standard text says "do not shave your heads until the offering reaches its destination";  Sanaa 1 omits "your heads."  The standard text lists the expiation as "fasting, or alms, or an offering";  Sanaa 1 has only "fasting or an offering."  One of three legal options is missing, and Sadeghi notes that neither version has a parallel elsewhere in the Quran, so it cannot be explained as a copyist harmonising to a familiar phrase.²

But nothing theological changes.  The substitutions that carry semantic weight are of this kind:  "on account of their wrongdoing" for "on account of their unbelief" at Quran 2:88;  "those who prosper" for "those who are guided" at Quran 9:18;  "the fire" for "hell" at Quran 9:73.³⁷  No variant adds or removes a doctrine.  A Christian apologist who says the palimpsest shows the Quran once taught something different can be refuted from the published edition in five minutes.

And two features cut hard against the "rewritten Quran" narrative.  Sadeghi's finding is that "C-1 has the same verses and the same order of verses as the standard Qur'ān," and he draws the inference himself:  "even a careful memorizer who reproduces the words exactly is prone to getting the order of the verses wrong."²  He names two exceptions and the article names them too:  the three-word verses Quran 20:31 and 20:32 stand in reverse order, a transposition the literary sources also report for Ubayy ibn Ka'b's codex, and Quran 9:85 is missing, which he explains as a scribe's eye skipping between two identical endings.²  And the missing basmala at the head of surah 9, the one formatting anomaly in the whole standard Quran, is already present in the palimpsest, so it was not introduced by Uthman's committee.²

The internal evidence: what the text does not say

Sinai's two 2014 papers in the Bulletin of the School of Oriental and African Studies are the best treatment of the dating question in English, and their structure is worth noting:  he argues against the strongest rival hypothesis rather than asserting the tradition.

He names the rival the emergent canon model — the view that the Quran, having reached a recognisable form by 660, continued to be reworked until about 700 — and its supporters:  Chase Robinson, Alfred-Louis de Prémare, David Powers and Stephen Shoemaker.  Part I examines the evidence for a late closure and finds each piece inconclusive.  The Dome of the Rock's divergent quotations fit either dating, because inscriptions adapt quotations to context.  The dossier on al-Hajjaj indicates real activity, but the reported measures fit dissemination and suppression of rival copies rather than composition.²⁸

Part II gives the positive case:  the unanimity of Sunni, Shi'a and Kharijite traditions in tracing the standard text to Uthman;  the preserved accusations against Uthman;  the absence of any post-632 content;  the absence of biographical framing that prophetic collections normally attract;  and the unrepaired grammatical anomalies.  His conclusion is carefully hedged, and the hedge should be quoted with it:

as long as no Quranic passages with a distinct stylistic and terminological profile have been compellingly placed in a late seventh-century context, the traditional dating of the standard rasm (excepting certain orthographical features) to 650 or earlier ought to be our default view.¹⁷

He is explicit that this is a burden-of-proof argument, not a proof:  "if the only swans we have ever encountered are white ones, it is the proponent of the existence of black swans whom we may legitimately expect to argue his case."¹⁷  And he concedes ground.  Orthography is excluded from the claim;  spelling kept evolving for a long time.  Uthman's measures did not work immediately, and Ibn Mas'ud's recension was still in use in Kufa at the end of the seventh century.  Even on the earliest reading of the palimpsest, "it seems wise to concede that during the first decade or so after Muḥammad's traditional year of death, the latter's literary legacy may not yet have been fully fixed."¹⁷

The stylometry, and what it can and cannot show

Sadeghi's 2011 Arabica study is the most serious quantitative test the Quran has had.  His logic is worth understanding because it is unusual:

given a corpus of texts, it would appear impossible to rearrange it to make many independent markers of style vary gradually, as opposed to in a jagged fashion.  This is because as soon as one rearranges a text to make one marker of style vary smoothly, that will affect the behaviors of all the other markers, interfering with their smoothness.  Finding an arrangement that makes all variations smooth would be quite extraordinary.¹⁸

He tested a very detailed chronology proposed by the Iranian engineer and politician Mehdi Bazargan, who split the 114 suras into 194 blocks.  Four independent markers — mean verse length and three non-overlapping morpheme lists — vary smoothly across a sequence of seven phases.  His two results that matter here are that the text has one author, and that the ordering is independent of the Muslim literary sources:  "this essay answers the above three questions … without any recourse to the statements of early Muslims about the history of the Qurʾān."¹⁸

His own caveats are extensive and this article will not hide them.  The internal order of the Medinan half is not confirmed:  "its second half, consisting of Groups 12-22, which happen to correspond to Medina in the traditional reckoning, remains largely unconfirmed."¹⁸  Two of the six boundaries rest on three markers rather than four.  The corpus is small — about 78,000 word-tokens against the 4.7 million of the Dickens corpus he benchmarks against.  Style cannot establish the direction of time;  that comes from meaning.  And legal material, concentrated in the late groups, may carry its own stylistic signature that has not been separated from chronology.¹⁸

Emmanuelle Stefanidis's assessment is the fair one, and it cuts against over-claiming in either direction:  "Sadeghi's assertion that stylometry firmly proves the single authorship of the Muslim scripture is tempered by the Qur'an's limited size and the lack of a parallel textual corpus to which it can be meaningfully compared. … Stylometry cannot produce a chronological order; it can only put one to the test."³⁸

The documentary record, including what it does not show

The dated, non-literary evidence is thinner than either side would like, and it is worth setting out in order because it is the one body of evidence nobody can accuse of being retrojected.

• AH 22 / 643. A bilingual Greek-Arabic receipt for sixty-five sheep, Papyrus Erzherzog Rainer 558, uses the Hijri era in an ordinary administrative document eleven years after Muhammad's traditional death. Its Greek side calls the conquerors magaritai, a calque of Arabic muhajirun, "emigrants."³⁹
• AH 66–67 / 685–87. The first coins reading bism Allah Muhammad rasul Allah — "In the name of God, Muhammad is the messenger of God" — are Arab-Sasanian drachms minted at Bishapur. Hoyland identifies the issuing governor as a man serving the rival caliphate of Ibn al-Zubayr, and concludes: "If this reconstruction is correct, it would mean that the earliest attested Islamic profession comes from an opposition party."²⁴
• AH 72 / 691–92. The Dome of the Rock. "Muhammad is the messenger of God" appears six times. Five distinct Quranic passages are quoted, in wording that matches the received text, in a chain of polemic about Christ.²⁴
• AH 77 / 696–97 onward. Abd al-Malik's reform coinage carries Quran 9:33 and the whole of Quran 112 as standard legends.²⁴

And here is the concession that belongs in this section rather than buried in a footnote.  Sean Anthony, whose book is the standing academic defence of Muhammad's historicity, states it plainly:  "Although none of the seventh-century non-Muslim sources yet know of the Qurʾan — they refer rather to Muḥammad's teachings, his traditions, or his laws rather than a scriptural corpus."³⁹  The earliest non-Muslim writers to name the Quran as a book are eighth-century:  John of Damascus and the Syriac Disputation of Bet Hale.  Seventh-century outsiders knew there was a prophet and knew he had teachings.  They do not describe a scripture.

That is a genuine gap, and Shoemaker's case is built in it.  Anthony himself draws the opposite inference from the same fact — that the Arabic scripture's traces in coins, papyri and inscriptions precede outsiders' awareness of it, which he takes as showing the Quran is the earliest artifact of the movement rather than a later one.  Both readings are available.  What the silence is not, on either reading, is evidence that no such text existed:  it is an argument from the silence of hostile outsiders who had no reason to catalogue their enemies' books.

The strongest Muslim responses

"Western scholarship has confirmed the traditional account"

This is half true and routinely oversold, usually by quoting van Putten or Motzki without their qualifications.  What the field grants:  an Uthmanic standardisation happened;  a single written archetype existed;  the Quran is early;  Muhammad existed.  What it does not grant:  perfect preservation, the historicity of the collection narrative in detail, or an unbroken oral transmission of the vocalised text.

Take Harald Motzki's work as the test case, because it is the strongest Muslim-friendly result in the literature.  Motzki applied isnad-cum-matn analysis — comparing the chains of transmission and the wording of every surviving version of a report — to the collection traditions.  He found twenty-two transmission lines across eight collections, all intersecting at Ibn Shihab al-Zuhri (d. 124/742), with the textual groupings matching the chain branches.  That is a real result and it pushes the reports back about a century earlier than Ignaz Goldziher's or Joseph Schacht's scepticism allowed.  But Motzki's own statement of what he proved is:  "We are not able to prove that the accounts on the history of the Qurʾān go back to eyewitnesses of the events."⁴⁰  Ninety years is much better than two hundred and fifty.  It is not contemporaneous documentation.

Al-Azami on the two-witness rule and the burning

Muhammad Mustafa al-Azami (d. 2017), Professor Emeritus at King Saud University with a Cambridge doctorate, wrote the most substantial modern Muslim defence of Quranic textual integrity in English.  Three of his arguments deserve a straight answer.

First, the collection was documentary, not casual.  Ibn Hajar's reading of Abu Bakr's instruction, which al-Azami adopts, is that Zayd was to sit at the mosque entrance and record a verse only if it came with two witnesses testifying that it had been written in the Prophet's presence.  This is a real argument and not a slogan:  a doubled documentary criterion is a serious procedure by any standard, ancient or modern.  The honest reply is that the rule is reported by Ibn Hajar (d. 852/1449), eight centuries after the event, and that Bukhari's own account of Zayd's search — palm stalks, stones, and the memories of men, with the end of surah 9 found with a single individual — does not describe a two-witness procedure being applied.

Second, the burning was less dramatic than critics make it.  Al-Azami notes that Anas's report of burning "represents only one possible scenario out of many," that other narratives have the copies torn or the ink erased, and that Jami' at-Tirmidhi's version of the same report — carrying the same chain from al-Zuhri, and graded hasan sahih by al-Tirmidhi himself — omits the burning clause entirely.⁴¹  That last point is accurate and worth conceding.  A critic who makes the bonfire the centre of the story should know that an early independent line of the same report does not have it.  The Arabic in Bukhari 4987 is nonetheless unambiguous:  an yuhraqa, the passive of haraqa, the ordinary word for setting something on fire.

Third, the bonfire proves consent.  Al-Azami's argument:  "This scene alone, of the Prophet's Companions in their thousands eyeing the bonfire as old Qur'anic fragments were tossed in, is a powerful testimony that they all assented to the purity of the Mushaf's text."⁴¹  Against this stands the tradition's own record of dissent.  Ibn Mas'ud — whom Bukhari 4999 has the Prophet name first among the four men to learn the Quran from — refused to surrender his codex and told the Iraqis to hide theirs.⁴²  Ibn Abi Dawud preserves the Egyptian delegation's grievance list against Uthman, whose first item is "We resent that he erased the Book of God Almighty."

Where al-Azami's argument becomes unfalsifiable, and this should be said plainly.  His rule for new manuscript discoveries:

If any researcher now brings forth a parchment with wording that is significantly different from ʿUthmān's Muṣḥaf, one whose consonantal backbone does not settle into the Muṣḥaf's text, it cannot be accepted as Qurʾān.  This is because it cannot share 'equal status' with the Muṣḥaf, which was compiled from first-hand documents backed by personal testimony.⁴¹

This is coherent on its own premises, and a Muslim may reasonably answer that it is the right rule for a community defining its own canon — canon is not textual criticism.  But it is not an evidential argument, and it must not be presented as one.  Any find that agrees confirms preservation;  any find that disagrees is by definition not Quran.  Christians should notice that this is exactly the move they object to when a Mormon apologist makes it about the Book of Mormon.

Neuwirth's literary argument for the traditional scenario

Angelika Neuwirth, who directs the Corpus Coranicum project in Berlin, accepts the Uthmanic scenario on literary grounds while refusing to call it proven:

The traditional scenario of the ʿuthmānic redaction, the hypothesis that the remnants of the Prophet's recitations were collected soon after his death to form the corpus we have before us, is thus plausible though not possible to prove.⁴³

Her reason is elegant.  The sura order "does not follow any logical let alone theological principle, yet the division into suras, most of them evidently genuine literary units, is maintained."  That is at once a conservative and a theologically disinterested arrangement, which suggests a redaction carried out without elaborate planning — and before the later doctrinal picture of the Prophet had formed.⁴³  A committee shaping scripture for an eighth-century agenda would have produced something tidier.

Common objections and responses

"Wansbrough proved the Quran is a ninth-century compilation"

He argued it;  he did not prove it, and the manuscripts have since refuted it.  Sinai calls the position "an easily vanquished straw man";  van Putten says it "is to be discarded."  Any presentation that still runs this line is working from the state of the question before 2012.  Retire this claim.

"Hagarism showed Islam was invented in the eighth century"

Both authors have disowned the central thesis.  Cook:  "I think we got an awful lot of it wrong."  Crone:  "There is no doubt that Mohammed existed."  Quoting the 1977 book without the 2006 and 2021 retractions is not scholarship, and a Muslim who knows the literature will say so, correctly.  Retire this claim.

"The Birmingham manuscript is older than Muhammad, so the Quran predates him"

The parchment carbon-dates to 578–646 CE at 95 per cent, a range that includes Muhammad's entire career.  Radiocarbon dates the animal, not the ink, and Déroche documents a manuscript with a firm colophon of 1020 CE that carbon-dated to 871–986.³⁵  Nothing here shows the Quran predates Muhammad.  Note also that this claim and the "radiocarbon is unreliable" claim cannot both be true, and Christian apologetics has been heard making both.  Retire this claim.

"The houris are really white raisins"

Answered in full above.  In short:  Saleh counts at least thirteen other passages that have to be re-read to make one phrase work;  the exegetical tradition is not confused about the maidens, so the warrant for the emendation is missing;  and Quran 52:24 supplies its own synonym, ghilman for wildan, which shows the words were understood as "youths" when the codex was fixed.  Retire this claim.

"The Sanaa palimpsest contains a completely different Quran"

It contains a recognisably different edition of the same book.  Different chapter order;  hundreds of wording differences, mostly particles and synonyms;  two small legal differences at Quran 2:196;  the same verses in the same order inside the suras, with two exceptions Sadeghi names;  no doctrinal change.  Sadeghi, a secular academic with no stake in defending Islam, concluded that the standard text is the better witness to what Muhammad recited.  A Christian apologetic that leans on the palimpsest is leaning on something that mostly supports the other side.

"Uthman burned the evidence, so nobody can know what the original said"

Half true, and the false half matters.  The tradition reports the burning and reports Uthman being attacked for it, which is one reason scholars think it happened.  But suppression did not work:  Ibn Mas'ud's recension was still used in Kufa at the end of the seventh century, al-Hajjaj was still trying to suppress it around 700, copies derived from it were reported in the tenth century, and the Sanaa palimpsest is a physical survivor of a non-Uthmanic text type.³⁵  Muslim scholarship itself preserves a large catalogue of the variants:  Arthur Jeffery's 1937 Materials collects them across fifteen "primary" and thirteen "secondary" codices.⁴⁴  Michael Cook's stemma and van Putten's spellings both reach behind the surviving manuscripts to a common ancestor.  The honest claim is that standardisation was attempted, resisted, and slow — not that the evidence was destroyed.

"Muhammad does not appear in the record before about 690, so he is an invention"

The documentary observation is correct and should be granted.  The inference is not.  Crone, who spent her career pressing the documentary sources, rejected it by name:  "few would accept the implied premise that history has to be reconstructed on the sole basis of documentary evidence."³⁰  The name does appear earlier in literary sources outside Islam — a Syriac flyleaf note of about 637 and Thomas the Presbyter's chronicle of about 640 both name him — and the question of Muhammad's existence has its own article:  did Muhammad exist?

"The Quran is a cocktail of texts, many older than Islam"

That is Puin's 1999 phrase, offered in a magazine interview and never developed into a study.  Sadeghi's stylometry is the published, quantitative, testable argument against it, and it finds one author.  Citing Puin as though he represented the field, while ignoring the peer-reviewed work that answers him, is exactly the double standard this series has committed not to commit.

"An early text proves the Quran is from God"

This is the Muslim-side version of the same overreach, and it deserves the same answer.  An early date establishes that words were fixed in writing soon after they were spoken.  It says nothing about where they came from.  The Book of Mormon has an unusually good textual history — a printer's manuscript, an 1830 first edition, a documented dictation — and no Christian thinks that settles anything about its origin.  The same logic runs the other way.  Whether the Quran is revelation is decided by other tests entirely:  its accuracy about the world, its accuracy about the earlier scriptures it claims to confirm, and the prophetic credentials of the man who recited it.  Those are handled in does the Quran contain scientific miracles?, does the Quran describe Christianity accurately?, and does Muhammad's life look like a prophet's?.

"Radiocarbon is worthless, so we can ignore the early dates"

Also wrong.  The measurements do cluster in the seventh century, and Sidky notes that the field's assessment of the method is far more favourable than its loudest critics allow.⁴⁵  The correct statement is narrower:  radiocarbon on parchment gives ranges of a century or more, dates the animal rather than the writing, and has produced irreconcilable results on the same leaf.  It is a constraint, not a clock.

"Daniel Brubaker documented thousands of textual changes in early Qurans"

He documented thousands of corrections — physical erasures, insertions and overwriting in early manuscripts.  That is a real contribution, and it does refute the popular claim that not one dot has ever changed.  But Hythem Sidky's review works through all twenty of Brubaker's showcase examples and assigns most to ordinary scribal error:  eye-skip, harmonisation to a parallel verse, dropped words that leave ungrammatical Arabic.  The last is decisive, because a scribe does not deliberately introduce a grammatical error.  Sidky's verdict:  "the vast majority of examples in the book are best explained through simple scribal error."⁴⁵  He also names the structural flaw:  treating corrections that move a manuscript toward the 1924 Cairo edition as evidence of a text evolving toward a standard is "a teleological argument for an end goal that did not exist at the time."⁴⁵

What evidence would change the conclusion?

What would strengthen the case for a late Quran.

• A Quranic passage with a distinct style and vocabulary that can be securely tied to a late-seventh-century situation. This is Sinai's own stated condition, and nobody has met it.
• A dated manuscript, inscription or papyrus from before 690 quoting the Quran in a form materially different from the received text. The Dome of the Rock's quotations diverge in trivial ways; something larger would be decisive.
• A second non-Uthmanic codex, from a different region, with a text as divergent as Sanaa 1 but diverging in different places. One outlier is a witness; two independent outliers would be a pattern.
• Manuscript evidence for a pre-Quranic Arabic Christian text that maps onto Quranic passages. This is what Lüling and Luxenberg needed and never had, and it is what would move their theses from speculation to argument.
• Confirmation that al-Hajjaj's "second masahif project" of 84–85 AH involved composition rather than dissemination. The dossier on him is real; its interpretation is not settled.

What would strengthen the case for an early Quran.

• A critical edition. There still is none. Donner's point stands as the single most useful fact for showing that both apologetic camps are overclaiming: every scholarly study of a Quranic passage made in the last century, including his own, "can only be considered provisional — pending the arrival at last of a critical edition of the text."⁴⁶
• The Documenta Coranica facsimile and full reconstruction of the Sanaa lower text, in preparation at Corpus Coranicum, which will supersede both current editions.
• More radiocarbon on the same codices from multiple laboratories, published with raw BP figures rather than pre-calibrated ranges — so that the laboratory disagreements can be diagnosed instead of averaged away.
• Van Putten's archetype method applied to a wider set of idiosyncrasies and a wider set of manuscripts. It is currently the strongest argument in the field and it has not been refuted.

What would not change anything either way.  More radiocarbon on parchment alone, quoted without palaeography.  More arguments from the silence of outsiders.  And any further round of re-pointing the consonantal text to produce a hidden original, in the absence of a manuscript of that original.

Sources