When Was the Qur'an Written? From Recitation to Book

Between about 610 and 632, Muhammad recited to his followers in Mecca and Medina words that he and they understood as God’s speech, delivered through the angel Gabriel. The recitations were memorized, repeated in prayer and partly written down. Muslim tradition says that after his death they were gathered into one volume under Abū Bakr, and that around 650 the caliph ʿUthmān issued a standard text, sent copies to the main garrison cities and ordered the rest destroyed. Fragments of that text survive on parchment dated by radiocarbon to the seventh century, written in a script that recorded consonants but left the vowels, and often the dots that tell one letter from another, to the reader’s memory. Over the next three centuries scholars added the missing signs, sorted the permitted ways of reciting the text, and built the tools, such as the chronology of the suras and the theory of abrogation, through which it is still read. Majority

This chapter follows the Qur’an from recitation to standard book, and sets the tradition’s account beside the manuscripts. The general problem of the sources, and why the earliest biography of Muhammad is a century younger than him, is set out in §0; the life of the prophet is §2; the Dome of the Rock, the first great public display of Qur’anic text, belongs to §5. Whether the Qur’an is revealed is outside what historical method can test (§0). What a historian can date is when its words were written down, by whom the tradition says, and what the oldest copies contain.

Timeline

A recitation before a book

“Recite!”

The word qurʾān means “recitation” or “reading aloud”. Nöldeke and many later scholars connect it with the Syriac qeryānā, the scripture reading of the Christian liturgy.^[18] Majority Tradition places the first revelation in a cave on Mount Ḥirāʾ near Mecca, and takes it to be the opening of sura 96:

iqraʾ bi-smi rabbika lladhī khalaq; khalaqa l-insāna min ʿalaq; iqraʾ wa-rabbuka l-akram; alladhī ʿallama bi-l-qalam; ʿallama l-insāna mā lam yaʿlam

“Read! In the name of your Lord who created: He created man from a clinging form. Read! Your Lord is the Most Bountiful One who taught by [means of] the pen, who taught man what he did not know.” (Q 96:1–5, trans. Abdel Haleem)^[1]​^[19]

That this was the first revelation is the tradition’s account, not something the text itself says. Majority as tradition. Translations quoted in this chapter are M. A. S. Abdel Haleem’s (2004) unless stated; the Arabic follows the standard Cairo text.

The Qur’an talks about itself more than any other scripture of its time, and much of what it says concerns speech and memory. It tells the prophet not to hurry the words:

lā tuḥarrik bihi lisānaka li-taʿjala bih; inna ʿalaynā jamʿahu wa-qurʾānah; fa-idhā qaraʾnāhu fa-ttabiʿ qurʾānah; thumma inna ʿalaynā bayānah

“[Prophet], do not rush your tongue in an attempt to hasten [your memorization of] the Revelation: We shall make sure of its safe collection and recitation. When We have recited it, repeat the recitation and We shall make it clear.” (Q 75:16–19)^[20]

It answers critics who asked why it did not come down all at once: it was sent “in this way to strengthen your heart”, “in gradual revelation” (Q 25:32).^[2] And it promises that God will guard the “Reminder” he has sent down (Q 15:9).^[21] Consensus on what the verses say.

A book in heaven, a text on earth

The same Qur’an also calls itself a kitāb, a book or scripture, and places its original in heaven: “This is truly a glorious Qur’an [written] on a preserved Tablet” (Q 85:21–22).^[22] Angelika Neuwirth reads this double self-image, recitation on earth and book in heaven, as the mark of a text that grew in the liturgy of a community before it became a codex, and that claimed for itself the status Jews and Christians gave their written scriptures.^[23] Majority

A text recited over twenty years also responds to events and revises itself. One verse concedes that some revelations are “superseded or forgotten” and replaced “with something better or similar” (Q 2:106), another that the prophet will not forget “unless God wishes” (Q 87:6–7).^[24]​^[25] Later scholars would build a theory of abrogation on these verses (below). For the historian, they show that the Qur’an itself knows its text was not static while it was being delivered. Majority

What was written in Muhammad’s lifetime?

Muslim tradition says that scribes, Zayd ibn Thābit prominent among them, wrote down revelations as they came, on whatever was to hand, while the main medium remained memory.^[4] Majority that some writing happened. How much, and in what form, is unknown: no manuscript can be shown to come from Muhammad’s lifetime, and the earliest radiocarbon range, Birmingham’s, only reaches it at its early end. The Qur’an denies that Muhammad recited or wrote any scripture before it (Q 29:48), the basis of the later doctrine that he was unlettered.^[26] What the verse means about his ability to write is debated. Contested

The tradition’s account: Abū Bakr, ʿUthmān and the companions’ codices

Abū Bakr and the battle of al-Yamāma

The best-known account of the collection is preserved by al-Bukhārī (d. 870) on the authority of Zayd ibn Thābit. After many of the men who knew the Qur’an by heart were killed at the battle of al-Yamāma (c. 633), ʿUmar urged the caliph Abū Bakr to have it collected, and Abū Bakr gave the task to Zayd:

fa-tatabbaʿtu l-qurʾāna ajmaʿuhu mina l-ʿusubi wa-l-likhāfi wa-ṣudūri l-rijāl

“So I started looking for the Qur’an and collecting it from (what was written on) palm stalks, thin white stones and also from the men who knew it by heart.” (al-Bukhārī 4986)^[4]

Zayd says he found the last verses of sura 9 only with one man, Abū Khuzayma al-Anṣārī. The sheets passed from Abū Bakr to ʿUmar, and then to ʿUmar’s daughter Ḥafṣa, one of the prophet’s widows. Majority that this is the tradition’s account; Contested on its historicity.

ʿUthmān’s standard text

The second report, in the same collection, explains the standard text. During the campaigns in Armenia and Azerbaijan, the general Ḥudhayfa ibn al-Yamān was alarmed that soldiers from Syria and Iraq recited differently, and urged the caliph ʿUthmān (r. 644–656) to unite the community “before they differ about the Book as Jews and Christians did before”. ʿUthmān borrowed Ḥafṣa’s sheets and had Zayd and three men of Quraysh copy them, telling them that where they disagreed with Zayd they should write “in the dialect of Quraysh, the Qur’an was revealed in their tongue”. Then:

“ʿUthmān sent to every Muslim province one copy of what they had copied, and ordered that all the other Qur’anic materials, whether written in fragmentary manuscripts or whole copies, be burnt.” (al-Bukhārī 4987)^[6]

Later sources disagree on how many copies there were (four, five or seven), but agree on Kufa, Basra, Damascus and Medina.^[27] Majority The ʿUthmānic text is the one Muslims call the muṣḥaf ʿuthmānī, and its consonantal outline, the rasm, is the basis of every printed Qur’an today.^[7] Majority

Two famous manuscripts are shown as ʿUthmān’s own copies, one in Tashkent (the Samarkand Qur’an, said to be stained with his blood) and one at the Topkapı Palace in Istanbul. Palaeography and radiocarbon (for the Samarkand copy) date both to the eighth or ninth century; they are early and important copies, but not his.^[28] Majority

The companions’ codices

ʿUthmān’s order presupposes that other written versions existed, and the tradition names their owners. The most discussed are those of ʿAbd Allāh ibn Masʿūd, the leading Qur’an teacher of Kufa, and Ubayy ibn Kaʿb of Medina. The reports, gathered in works such as Ibn Abī Dāwūd’s Kitāb al-Maṣāḥif (“Book of the Codices”, tenth century) and collected by Arthur Jeffery in 1937, list hundreds of variant words and phrases.^[29] They also report differences of extent: Ibn Masʿūd’s codex lacked the opening sura and the last two (113–114), and Ubayy’s had two short extra suras, known as al-Khalʿ and al-Ḥafd, which read like prayers.^[29] Ibn Masʿūd is said to have resisted handing over his codex. Majority that the reports exist and are early; Contested on their detail. Muslim scholars have long explained the omissions as matters of what was written down, not of what was believed: the opening sura, for instance, was recited in every prayer. Consensus that this explanation exists.

The tradition also explains why variation was legitimate. In another report, ʿUmar hears a companion recite sura 25 differently from the way he learned it and drags him to the prophet, who approves both recitations and says: “This Qur’an has been revealed to be recited in seven different ways (sabʿat aḥruf), so recite of it whichever is easier for you.”^[30] What the seven aḥruf were was already disputed by medieval scholars; they are not the same as the later seven readings. Consensus

How old are these reports?

The collection reports reach al-Bukhārī through the Medinan scholar Ibn Shihāb al-Zuhrī (d. 742). Harald Motzki, analysing their chains of transmission, argued that they can be traced at least to al-Zuhrī and so to the early eighth century, much earlier than earlier Western sceptics had assumed.^[31] Majority Two older Western theses took the reports in opposite directions. John Burton (1977) argued that Muhammad himself had left a complete text, and that the stories of a later collection were invented by jurists who needed missing verses for their theory of abrogation.^[32] John Wansbrough, in the same year, argued that the text was not fixed until around 800 (see Chapter 0: How We Know).^[33] Neither view has survived the manuscript evidence below, which points to a fixed text well before 700. Majority

What the manuscripts show

A script that recorded consonants

Seventh-century Arabic was written in a script that distinguished fewer letter shapes than it had sounds. Several letters (b, t, th, n and y, for instance) shared one shape in the middle of a word, and short vowels were not written at all. Dots to tell the shapes apart already existed: PERF 558, a receipt for sheep written in Arabic and Greek in Egypt in April 643, is the oldest dated Arabic papyrus and uses them.^[5] Consensus But the earliest Qur’an copies use such dots sparingly, and add no vowel signs.^[13] Consensus

The consequence is that a written copy fixed the rasm, the consonantal skeleton, but not every reading of it: the same skeleton could be voiced as an active or a passive verb, a first or third person. The text lived in the combination of a written skeleton and a taught recitation. Consensus Muslim tradition credits the later signs to the Basran grammarian Abū l-Aswad al-Duʾalī (d. c. 688), who marked vowels with coloured dots; his students Naṣr ibn ʿĀṣim (d. 707–8) and Yaḥyā ibn Yaʿmar, said to have added the letter-dots at the request of the governor al-Ḥajjāj; and al-Khalīl ibn Aḥmad (d. c. 786–791), who replaced the coloured dots with the vowel signs still used.^[7] Majority that the signs came in stages; Contested on the names.

The script of the earliest copies is called ḥijāzī, after the region of Mecca and Medina: an upright, slanting hand on parchment in vertical format. Within a few decades it gave way to the larger, horizontal “Kufic” hands of the Umayyad period.^[34] Majority

The Birmingham leaves

In 2015 the University of Birmingham announced that two parchment leaves in its Mingana Collection (Mingana 1572a), written in ḥijāzī script and containing parts of suras 18 to 20, had been dated by the Oxford Radiocarbon Accelerator Unit to between 568 and 645 at 95.4% probability.^[3] Consensus on the measurement. The leaves had been bound with leaves of a later manuscript; the researcher Alba Fedeli recognised that they belong to the same codex as sixteen leaves in Paris (BnF Arabe 328c).^[3] Consensus

The result was widely reported as “the oldest Qur’an”. What it measures is the death of the animal whose skin became the parchment, so the writing is probably somewhat later, and the range, which begins before Muhammad’s prophecy, says nothing about which decade within it. François Déroche, the leading specialist in early Qur’anic manuscripts, expressed reservations about how much precision radiocarbon can give (Debates).^[34] What the leaves do show is a text in ḥijāzī script whose wording agrees with the standard text, on parchment from the first half of the seventh century. Majority

Tübingen, Berlin and Leiden

Birmingham was not alone. In 2014 the Tübingen University Library announced that its fragment Ma VI 165 had been dated by the ETH Zürich to 649–675.^[10] As part of the German–French project Coranica (2011–2014), Michael Marx and Tobias Jocham dated further fragments in Berlin and Leiden to the seventh century, including Berlin’s ms. or. fol. 4313 to 606–652, and compared the method against Arabic papyri and Syriac manuscripts of known date, which it dated correctly.^[11] Majority They also found archaic spellings, of names such as Dāwūd (David) and of common words, that later copyists modernized: the spelling of the Qur’an was still developing in the seventh and eighth centuries even where its words were not.^[11] Majority

The Codex Parisino-petropolitanus

The largest early copy is not radiocarbon-dated but studied page by page. Its 118 surviving leaves are divided between Paris (70, BnF Arabe 328a–b), St Petersburg (46), the Vatican and the Khalili Collection in London (one each). They preserve about 45% of the text, from sura 2 to sura 72, and were written by five scribes working at once.^[12] François Déroche, in a full study (2009), dated it on palaeographic grounds to the third quarter of the seventh century. Its wording is the standard rasm, with mainly orthographic differences from the Cairo text; its verse divisions and spellings vary between the scribes.^[13] Majority

The Ṣanʿāʾ palimpsest

In 1972 workers repairing the roof of the Great Mosque of Ṣanʿāʾ in Yemen found a mass of discarded parchment: thousands of fragments of Qur’an manuscripts, later sorted into several hundred codices. A German conservation project directed by Albrecht Noth, with Gerd-R. Puin as its local director in the early 1980s, worked on them.^[8] Consensus One codex, catalogued as DAM 01-27.1 and known as Ṣanʿāʾ 1, is a palimpsest: its parchment was scraped and rewritten. The upper text is the standard Qur’an. The lower text, readable under ultraviolet light and imaging, is not. Consensus

Behnam Sadeghi and Uwe Bergmann (2010) had a leaf of the codex that had been sold at auction in 1993 radiocarbon-dated: 578–669 at 95%, with a 75% probability of a date before 646.^[9] Sadeghi and Mohsen Goudarzi (2012) then edited the lower text of the leaves in Ṣanʿāʾ.^[35] They found that it agrees with the standard text in the great majority of its words, but differs in many small ways: words added or missing, different forms of a verb, transposed phrases, and suras in a different order. Its variants show little or no correspondence with those reported for the codices of Ibn Masʿūd or Ubayy. They concluded that it represents a companion codex of a type older than or independent of ʿUthmān’s, which they called C-1, and that the standard text is on the whole the better preserver of their common source.^[35] Majority

Not everyone reads it as a codex. Asma Hilali (2017) argued that both layers show signs of teaching and exercise, practice copies rather than a book meant to be read.^[36] Éléonore Cellard (2021), reconstructing many more leaves, concluded that they formed a complete codex produced by professional scribes.^[37] Contested What is not disputed is that someone in the seventh century copied a version of the Qur’an that differed from ʿUthmān’s, and that someone later erased it and wrote the standard text over it. Consensus

One written archetype

The most important recent argument is about spelling. Marijn van Putten (2019) examined how fourteen early manuscripts spell a single phrase, niʿmat allāh, “the grace of God”, whose second word can be written in two ways. Across the manuscripts, each occurrence is spelled the same way in the same place, even though the choice between spellings is arbitrary.^[38] Shared idiosyncrasies of this kind are what textual critics expect from copies of a single written exemplar, not from independent writings-down of a memorized text. Majority Together with the radiocarbon dates, this is the main reason most scholars now date the standard rasm to ʿUthmān’s reign, as the tradition says. Nicolai Sinai (2014), weighing the evidence for later editing, concluded that “the traditional dating of the standard rasm… to 650 or earlier ought to be our default view”, while conceding that rival versions were not displaced at once and that in the first decade after 632 the text may not yet have been fully fixed.^[7] Majority

From one skeleton to many readings

Because the rasm did not fix vowels and dots, people who agreed on the written text still recited it differently, and each city followed its own famous teachers. By the late ninth century the variety was large. The Baghdad scholar Ibn Mujāhid (d. 936), in his Kitāb al-Sabʿa (“Book of the Seven”), chose seven readers as authoritative: Nāfiʿ of Medina, Ibn Kathīr of Mecca, Abū ʿAmr of Basra, Ibn ʿĀmir of Damascus, and ʿĀṣim, Ḥamza and al-Kisāʾī of Kufa. Later scholars accepted three more, to make ten. Each reading is transmitted through two principal students.^[16]​^[39] Consensus A reading had to fit the ʿUthmānic skeleton, accord with Arabic grammar and come through a sound chain of teachers; readings outside the canon, including the companions’ variants, became shādhdh, “irregular”, acceptable for study but not for prayer. Majority

The differences between the readings are real but small: a vowel that makes “king” (malik) into “owner” (mālik) in sura 1:4, a pronoun, a verb’s person. Consensus Which reading a Muslim hears today depends largely on a twentieth-century printing decision. The Cairo edition of 1924, prepared under King Fuʾād, printed the reading of ʿĀṣim in the transmission of his student Ḥafṣ, and its text, with its verse numbering and its headings marking each sura as Meccan or Medinan, became the model for most printed Qur’ans.^[17] The transmission of Nāfiʿ by Warsh remains common in North and West Africa. Consensus

The Qur’an in its world

A text that assumes its hearers know the stories

The Qur’an names about two dozen prophets, most of them biblical, and refers to Adam, Noah, Abraham, Moses, David, Solomon, Jesus and Mary as figures its audience already knows. It rarely tells a story from beginning to end; it alludes, corrects and draws a lesson. Consensus Sidney Griffith has argued that there is no evidence of a written Arabic Bible before Islam, so that this knowledge must have circulated orally, in Arabic, among Arabs who had heard Jewish and Christian teaching.^[40] Majority Scholars such as Gabriel Said Reynolds read the Qur’an alongside the Bible and its Jewish and Christian interpretations, its “subtext”.^[41] Angelika Neuwirth reads it as a voice in the debates of late antiquity, answering the scriptures of others in its own terms.^[23] Majority on the parallels; Contested on the channels and on how much the Qur’an is answering rather than borrowing.

The Qur’an is aware of the charge that someone was teaching Muhammad: “We know very well that they say, ‘It is a man who teaches him,’ but the language of the person they allude to is foreign, while this revelation is in clear Arabic” (Q 16:103).^[42] It also challenges doubters to “produce a single sura like it” (Q 2:23).^[43] Consensus on what the verses say.

Mary, the cross and the Sleepers

Sura 19 is named after Mary and tells of the annunciation and the birth of Jesus; Mary withdraws “to a place to the east” and God sends his Spirit “in the form of a perfected man” (Q 19:16–17).^[44] Its narrative shares details with Christian texts outside the New Testament, such as the infant Jesus speaking in Mary’s defence (19:29–30).^[41] Majority On the crucifixion, the Qur’an answers both Jews and Christians: “They did not kill him, nor did they crucify him, though it was made to appear like that to them” (Q 4:157).^[45] What “made to appear like that to them” means has been debated by Muslim commentators since the first centuries; most held that another was crucified in his place. Consensus that the verse denies the crucifixion of Jesus as seen by his enemies; Contested on its exact sense.

Sura 18 tells of “the Companions in the Cave”, young believers who sleep in a cave for “three hundred years, … some added nine more” (Q 18:9–26).^[46] This is the Christian legend of the Seven Sleepers of Ephesus, who hid from persecution under the emperor Decius and woke under a Christian emperor; its earliest surviving version is a Syriac poem by Jacob of Serugh (d. 521). The Qur’an tells it without names or place, and draws from it a lesson about resurrection and the limits of human knowledge: “[Some] say, ‘The sleepers were three, and their dog made four’… Say [Prophet], ‘My Lord knows best how many they were’” (18:22).^[46] Consensus on the parallel.

Alexander, Gog and Magog

The same sura then tells of Dhū l-Qarnayn, “the two-horned one”, who travels to where the sun sets and where it rises, and builds a barrier of iron and molten metal against Gog and Magog (Q 18:83–98).^[47] Most scholars identify him with Alexander the Great, who was shown with ram’s horns on coins. Majority Kevin van Bladel argued that the passage retells, point for point, a Syriac Christian text, the Alexander Legend (Neṣḥānā), which the Syriac scholar Gerrit Reinink dated to about 630 and read as propaganda for the Byzantine emperor Heraclius; the Qur’an removes the pro-Byzantine elements and keeps the story of a God-guided ruler.^[48] Majority on the dependence. If the Syriac text is as late as 630, the Qur’anic passage cannot be early Meccan, as sura 18 is usually classed; either the sura grew later than tradition says, or the legend is older than its modern dating. Contested

The fringe: a Syriac Qur’an?

In 2000 a scholar writing as Christoph Luxenberg argued that the Qur’an was written in a mixed Syro-Aramaic and Arabic language, and that many obscure passages make sense when read as Syriac; on his reading, the “maidens of paradise” become white grapes.^[49] Specialists rejected the method: the Syriac scholar François de Blois called it the work of someone who knew “enough Syriac so as to be able to consult a dictionary”.^[50] Consensus among specialists that the thesis fails, though the Qur’an’s Syriac loanwords and Christian context are not in doubt.

Ordering the suras: Meccan, Medinan and their critics

The Qur’an’s 114 suras are not arranged by date: after the short opening prayer, they run roughly from longest to shortest. Consensus Muslim scholarship classed each sura as revealed in Mecca or Medina, using reports about the occasions of revelation, and the Cairo edition prints those labels. In 1860 Theodor Nöldeke’s Geschichte des Qorāns, drawing on that tradition and on style, divided the suras into three Meccan periods and a Medinan one: 48, 21, 21 and 24 suras. Short, rhymed, oath-filled suras came first; long, prosaic, legal ones last.^[18] Majority His scheme still shapes the field.

It has critics. Gabriel Said Reynolds argued in 2011 that the scheme rests on the very biographies whose reliability is in question, and that reading the Qur’an through a reconstructed life of Muhammad is circular.^[51] Defenders answer with internal evidence. Behnam Sadeghi showed that when passages are ordered by average verse length, other stylistic features change smoothly along the same sequence, which is what a real chronology would produce.^[52] Sinai holds that a relative chronology of this kind, independent of the biographies, broadly confirms the early/late distinction.^[53] Contested on precise dating; Majority that the short, highly rhymed suras are generally earlier.

Abrogation: a tool of later scholars

When one verse seemed to contradict another, early jurists asked which came later. The answer, naskh, “abrogation”, held that a later ruling could cancel an earlier one, building on Q 2:106: “Any revelation We cause to be superseded or forgotten, We replace with something better or similar.”^[24] The classic example is wine: verses that tolerate it, then warn against it, then forbid it.^[54] Majority The theory grew in stages. Scholars disagreed on how many verses were abrogated: some listed hundreds, the Egyptian polymath al-Suyūṭī (d. 1505) about twenty, and the Indian scholar Shāh Walī Allāh (d. 1762) only five.^[54] Consensus on the range. Some theorists also held that verses could be withdrawn from the text while their rulings remained, such as a verse prescribing stoning, which they said had been in the Qur’an but was no longer recited; Burton argued that such claims served jurists’ needs, not textual history.^[54] Contested For the history of the text, abrogation matters because it shows how much of the Qur’an’s later reading depended on a chronology that the text does not itself supply. Majority

The “Satanic verses”

Sura 53 names three goddesses of the Meccans and dismisses them: “Consider al-Lāt and al-ʿUzzā, and the third one, Manāt… these are nothing but names you have invented yourselves” (Q 53:19–23).^[55] Several early biographies and commentaries report that at first, after naming them, Muhammad recited two further lines:

tilka l-gharānīqu l-ʿulā, wa-inna shafāʿatahunna la-turtajā

“These are the exalted cranes, whose intercession is to be hoped for.”

The Meccans prostrated with the Muslims in delight; then Gabriel told Muhammad that Satan had cast the words onto his tongue, and Q 22:52 was revealed, which says that Satan insinuated something into the wishes of every prophet before him, “but God removes what Satan insinuates”.^[56]​^[57]​^[58] The story is in Ibn Saʿd (d. 845) and in al-Ṭabarī’s history and commentary (early tenth century), citing Ibn Isḥāq, but not in Ibn Hishām’s edition of Ibn Isḥāq’s biography. Consensus on where the reports appear.

Shahab Ahmed’s study of every early report found that in the first two centuries of Islam the story was near-universally accepted by those who transmitted it, as an episode in the prophet’s life that raised no special problem.^[59] Objections were raised from the tenth century; Ibn Ḥazm (d. 1064) called it a fabrication and, most influentially, Qāḍī ʿIyāḍ (d. 1149) rejected it, on the grounds that its chains of transmission were weak and that it contradicted the prophet’s protection from error (ʿiṣma); by the thirteenth century rejection was general, and it remains the orthodox view.^[59] Majority Modern historians are divided: W. Montgomery Watt argued that no Muslim would have invented so awkward a story, while more recent work leans towards reading it as a later narrative built around the Qur’anic verses.^[59] Contested The name “Satanic verses” is a modern English one; Salman Rushdie’s novel of 1988 and the crisis it caused are the subject of §12.

The Qur’an in public

In 691–692 the caliph ʿAbd al-Malik completed the Dome of the Rock in Jerusalem. Its mosaic band quotes Qur’anic verses about Jesus and the oneness of God, including Q 4:171, addressed to the “People of the Book”: “the Messiah, Jesus, son of Mary, was nothing more than a messenger of God… do not speak of a ‘Trinity’”.^[14] Some of its quotations differ from the standard text, and they were once taken as evidence that the text was still fluid. Estelle Whelan showed that the variations follow the conventions of inscriptions, which adapt quotations into a continuous statement, and argued that the inscription presupposes an already codified text.^[15] Majority The building, its date and its message belong to §5.

Under ʿAbd al-Malik’s governor of Iraq, al-Ḥajjāj ibn Yūsuf (d. 714), the reports gathered by Sinai describe a programme for the written Qur’an: counting its letters, words and verses, dividing it into sections for recitation, adding diacritics, sending codices to the towns and, in Medina, instituting public reading from a codex in the mosque. One report lists eleven places where al-Ḥajjāj “changed” the ʿUthmānic skeleton, mostly by a single letter, and a Christian polemicist of the early ninth century, al-Kindī, accused him of wholesale rewriting.^[7] Majority that he promoted and distributed the written text; Contested on whether he altered it. Sinai notes that if anyone had revised the text around 700, the passages that would most obviously have been corrected were left untouched.^[7]

The evidence

Primary sources

What is absent

• A manuscript from Muhammad’s lifetime. No surviving copy can be shown to have been written before 632; the earliest radiocarbon ranges reach his lifetime only at their early end. Consensus
• ʿUthmān’s own codices. None of the copies sent to the provinces is known to survive; the codices shown as his are later. Majority
• Vowels and full dotting. Seventh-century copies record the consonantal skeleton; the reading depended on teachers. Consensus
• A chronological arrangement. The suras were never put in the order of revelation; the Meccan and Medinan labels come from later scholarship. Consensus
• Commentary. No written commentary survives from the seventh century; the earliest extant full commentaries are of the eighth. Majority
• Seven canonical readings. Readers varied by city; the canon of seven dates from the tenth century. Consensus
• A printed standard. One reading became nearly universal only with the Cairo edition of 1924. Consensus

Debates

When was the text fixed?

Majority for ʿUthmān; Contested overall.

How reliable are the radiocarbon dates?

Supporters point out that the Coranica project tested the method on manuscripts and papyri of known date and found it accurate, and that several independent fragments cluster in the seventh century.^[11] Déroche cautioned that radiocarbon dates for some Umayyad manuscripts seem too early when set against their script and decoration, and that the method cannot give the precision the debates need.^[34] Stephen Shoemaker went further, arguing that radiocarbon results for early manuscripts have varied too widely between tests and laboratories to settle a question that turns on a few decades.^[60] Contested on precision; Consensus that the oldest fragments are seventh-century.

What is the Ṣanʿāʾ lower text?

A companion codex, independent of ʿUthmān’s and slightly inferior to it as a witness to their common source (Sadeghi and Goudarzi); a professional codex of a non-ʿUthmānic text (Cellard); or a school exercise (Hilali).^[35]​^[37]​^[36] All agree that its variants are mostly small and that its existence confirms the tradition that non-ʿUthmānic texts circulated. Contested

How did the Qur’an know Jewish and Christian traditions?

Through oral contact in Arabic, in Mecca, Medina and on trade routes (Griffith); through a shared culture of late antiquity in which the Qur’an is a participant, not a borrower (Neuwirth); or, on the more radical view, because some of its material took shape in contact with Syriac Christianity later or further north than the tradition places it, the implication some draw from the Alexander passage.^[40]​^[23]​^[48] Contested

Further reading