The ʿUthmānic Codex: Understanding how the Qur’an was Preserved

The history of the preservation of the Qur’an has been studied and written about by Muslim scholars in extensive detail. It is a scripture that is intimately familiar to the heart and tongues of countless believers, memorized word-for-word by Muslims across the globe and continuously recited in daily prayer. The Qur’an is the Divine speech of Allah revealed to the Prophet Muhammad ﷺ through the angel Jibrīl (Gabriel), and its preservation has been assured by Allah Most Exalted Himself. Allah says, “Verily, We have revealed the Qur’an and We will surely protect it.” This is in contrast to previous scriptures revealed by Allah, such as the Tawrāt and Injīl, whose preservation was entrusted to the People of the Book: “To them was entrusted the protection of Allah’s Book.” The timeless preservation of the Qur’an has even been regarded by many scholars as one facet of its miraculous nature.

It is important for Muslims to understand precisely what is meant by the preservation of the Qur’an. The Qur’an was preserved through both mass memorization and written recording. The written preservation of the Qur’an occurred in stages, the final stage of which was commissioned by the third caliph, ʿUthmān ibn ʿAffān (d. 35 AH/656 CE). Through a process unlike any other in human history, the words and verses of the Qur’an populated the pages of manuscripts around the world with the same clarity with which they were etched in the hearts and minds of the companions of the Prophet ﷺ.

It is that process that the present article aims to explain, providing an overview of the history behind the ʿUthmānic codex and how it was compiled. The discussion will draw upon primary sources in the Islamic tradition (including historical works and Hadith literature) along with analysis of classical and contemporary Muslim scholarship on this topic. It will also evaluate a number of theses that have emerged in Western academia with respect to the history of the Qur’anic text including research related to Qur’anic manuscripts.

The present article is the sequel to the authors’ previous article discussing the variant readings of the Qur’an that existed during the time of the Prophet’s companions. To briefly recap the context provided there: during the lifetime of the Prophet Muhammad ﷺ, the Qur’an was revealed piecemeal over a period of 23 years. As reported by numerous companions, the Prophet ﷺ himself reviewed the Qur’an annually with angel Jibrīl (Gabriel) every Ramadan, and twice during the final year of his life. Whenever Allah revealed verses of the Qur’an, the Prophet ﷺ would recite them aloud to the companions who would memorize them, and he would summon scribes to write the verses down. A total of 65 companions reportedly served as scribes. The evidence indicates a large number of companions had memorized the Qur’an in its entirety during the Prophet’s lifetime or shortly after. The Qur’an’s description of itself as both recitation and book alludes to its own oral and textual preservation. Although all of the Qur’an was written down during the lifetime of the Prophet ﷺ (according to the strongest view), the written fragments were not gathered into an official unified compilation, although some individual companions had written personal codices.

Following the death of the Prophet Muhammad ﷺ in 11 AH/632 CE and under the caliphate of Abū Bakr (d. 13 AH), the first official compilation of the Qurʿan was undertaken in the year 12 AH. Abū Bakr was persuaded by ʿUmar ibn al-Khaṭṭāb (d. 23 AH) who emphasized the importance of compiling the Qur’an after a large number of its reciters were martyred during the Battle of al-Yamāmah. Zayd ibn Thābit was appointed to the task by Abū Bakr, and he compiled the Qur’an “from date-palm tree bark, parchment, thin white stones, and the hearts of men.” Imam Layth ibn Saʿd (d. 175 AH) said, “The first to compile the Qur’an was Abū Bakr and Zayd transcribed it. The people would come to Zayd ibn Thābit and he would not write a single verse except with two witnesses.” Zayd did not transcribe verses from memory despite the fact that he and other companions had them memorized. Nor did he simply transcribe the Qur’an from existing written copies. Rather, he followed a meticulous process ensuring that the memory and writing of every verse was backed by direct testimony. Al-Sakhāwī (d. 643 AH) explained that the requirement for ‘two witnesses’ meant two people who possessed it in writing and could testify to having written down the verse from the Prophet ﷺ precisely as they had learned it. Thus, the compilation process for each verse in the Qur’an was backed by the combined attestation of written materials, memorization, and direct testimony.

While scribes had written the verses of the Qur’an on different materials during the time of the Prophet ﷺ, they had not been compiled into a single text. Indeed, it would not have made sense to do this during the lifetime of the Prophet ﷺ, when verses were still being revealed and added to various chapters while other verses were abrogated, and the predominantly unlettered community of Muslims could rely on direct oral transmission from the Prophet ﷺ himself. It was only after the revelation had ceased that such a task was necessary. The result of Abū Bakr’s project was that the written text of the Qur’an was gathered into a single compilation; this process took less than a year (between the Battle of al-Yamāmah in 12 AH and Abū Bakr’s death in 13 AH). The Qur’an was written on parchment, a material derived from the untanned skins of goats, sheeps, and calves cut into sheets. Written sheets of parchment bound together are termed a codex, the ancestor of the modern book. Similarly, the Arabic word muṣḥaf (pl. maṣāḥif) linguistically refers to a text that consists of written pages between two covers, and has come to refer exclusively to written copies of the Qur’an. 

The early Muslim community witnessed a rapid expansion of the Islamic empire through a dazzling series of victories against the existing superpowers of the world, the Byzantine and Persian empires. During the reign of ʿUmar ibn al-Khaṭṭāb and subsequently ʿUthmān ibn ʿAffān, the Islamic empire came to extend from Northern Africa to Central Asia to the frontiers of the Indian subcontinent. The vast numbers of new Muslims were in need of those who could teach them the Qur’an. To this end, the companions appointed individuals well versed in the recitation of the Qur’an to travel to distant lands to teach it.

Many senior companions themselves left Madīnah and took up residence in distant cities to serve as teachers, including ʿAbdullāh ibn Masʿūd in Kūfah, Abū Mūsa al-Ashʿarī in Baṣra, Muʿādh ibn Jabal in Palestine, Abū al-Dardāʾ in Damascus, and ʿUbādah ibn al-Ṣāmit in Ḥumṣ, while Zayd ibn Thābit and Ubayy ibn Kaʿb remained in Madīnah. Numerous students in these cities learned the recitations of the Qur’an from these senior companions. Muslims in these distant lands copied manuscripts of the Qur’an based on the codices and readings of companions who resided among them, as scholars like Ibn ʿAṭṭīyah (d. 541 AH) mention. However, the multiplicity of Qur’anic readings taught by different companions soon became a source of confusion. While the companions understood that the Qur’an had been revealed according to seven aḥruf and therefore could be recited with a certain degree of variation permitted by the Prophet ﷺ himself, this concept was unfamiliar to new Muslims in the newly conquered lands of Islam. Naturally, disputes arose whereby some rejected and repudiated the readings of others. The illustrious companion Ḥudhayfah ibn al-Yamān (d. 36 AH) raised this concern to ʿUthmān.

Ḥudhayfah ibn al-Yamān came to ʿUthmān at the time when the people of Shām and the people of ʿIrāq were waging war to conquer Armenia and Azerbaijan.

Ḥudhayfah was afraid of their [the people of Shām and Irāq] differences in the recitation of the Qur’an, so he said to ʿUthmān, “O chief of the Believers! Save this nation before they differ about the Book [Qur’an] as Jews and the Christians did before.”

So ʿUthmān sent a message to Ḥafṣah saying, “Send us the manuscripts of the Qur’an so that we may compile the Qur’anic materials in perfect copies and return the manuscripts to you.” Ḥafṣah sent it to ʿUthmān.

ʿUthmān then ordered Zayd ibn Thābit, ʿAbdullāh ibn al-Zubayr, Saʿīd ibn al-ʿĀṣ and ʿAbd al-Raḥmān ibn al-Ḥārith ibn Hishām to rewrite the manuscripts in perfect copies. ʿUthmān said to the three Qurayshī men, “In case you disagree with Zayd ibn Thābit on any point in the Qur’an, then write it in the dialect of Quraysh, the Qur’an was revealed in their tongue.”

They did so, and when they had written many copies, ʿUthmān returned the original manuscripts to Ḥafṣah.

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.

This report and others indicate that, as Muslims from various regions joined the military campaigns in Armenia and Azerbaijan, those who had learned the Qur’an from different companions disputed with one another over their divergent recitations. This narrative from Ḥudhayfah provides an important insight into why the codex of ʿUthmān differed from the compilation of Abū Bakr. 

First, the reasons for undertaking the projects differed. The compilation of Abū Bakr was meant to record the Qur’an in its entirety to ensure that verses of the Qur’an would not be lost with the death of those who had memorized them. The ʿUthmānic codex, meanwhile, was meant to unite the entire Muslim nation upon a single text to eliminate confusion caused by Muslims reciting different variant readings of the Qur’an. Therefore, the use of the compilations differed as well. During the time of Abū Bakr, the compilation was simply kept in safekeeping while Muslims continued to read according to the way they had been taught and according to the personal copies of the Qur’an that they had in their possession. Adhering to the ʿUthmānic codex, however, was compulsory, and copies were sent to major cities in the Islamic empire. Any written copies of the Qur’an that did not conform to the ʿUthmānic codex were burned or corrected. These are the fundamental differences between the two compilations.

In his description of the ʿUthmānic compilation, al-Baghawī (d. 516 AH) explains that in seeking to unify the Muslims upon a single text, ʿUthmān compiled the Qur’an according to one ḥarf (mode of reading) to reduce differences (this point will be discussed in greater detail later). Al-Baghawī writes:

Then the companions of the Messenger of Allah ﷺ used to recite the Qur’an after him according to the seven aḥruf that the Messenger of Allah ﷺ taught them to recite by the permission of Allah. This continued until there were (disputes over) differences between the reciters in the time of ʿUthmān and the matter became grave and people from the different regions wrote to ʿUthmān, beseeching him by Allah to unite people before the situation deteriorated. Ḥudhayfah ibn al-Yamān came to ʿUthmān from the battle in Armenia and advised him concerning that. So ʿUthmān gathered the Muhājirīn and Anṣār and consulted them on the matter of compiling the Qur’an in codices according to one mode of reading (ḥarf) in order to diminish the differences and unite people. They agreed with his opinion and encouraged him to do it and deemed it to be the most cautious course of action with respect to the Qur’an. Thus, ʿUthmān then requested Ḥafṣah to send him the ṣuḥuf (written copy compiled during Abū Bakr’s era) in order to transcribe the maṣāḥif (written codices). She sent it to him and so he instructed Zayd ibn Thābit and the three Qurayshī committee members and they transcribed from it the codices and he sent them to the different regions.

Similarly, ʿUthmān’s goal of unifying the people is also evident in other narrations. In Akhbār al-Madīnah, ʻUmar ibn Shabbah (d. 262 AH) narrated that ʿUthmān said to the companions and the people of Madīnah, “It has reached me that some of you say that my reading is better than your reading, and this almost constitutes disbelief. And if you differ today, those after you will differ even more.” The people asked, “What is your opinion?” ʿUthmān replied, “That I should gather the people upon one muṣḥaf so that there be no division (furqah) nor disagreement (ikhtilāf).” The people replied, “Excellent is your opinion.” This is also taken by some scholars as an indication that the idea of compiling a standard muṣḥaf for the community was one that ʿUthmān had contemplated even prior to the news brought by Ḥudhayfah about the disputes amongst Muslims on the frontiers, which simply further highlighted the need that ʿUthmān had foreseen.

From the foregoing discussion, a clear picture emerges of the context and circumstances that motivated the compilation of the ʿUthmānic codex. As for when this took place, we know from early historical sources that ʿUthmān’s reign was during 23–35 AH, the military campaign in Armenia likely took place in 24–25 AH, and Ḥudhayfah was appointed there in 24 AH, following which he returned to Madinah and advised ʿUthmān regarding compiling the muṣḥaf. This places the likely date of the ʿUthmānic compilation around 25 AH (645 CE). This is supported by Ibn Ḥajar al-ʿAsqalānī (d. 852 AH), who argues against other historians and scholars who favored the later date of 30 AH (such as Ibn al-Athīr, Ibn Khaldūn and Ibn al-Jazarī). Ibn Ḥajar adduces as supporting evidence a report from ʿUthmān wherein he admonished the people in a sermon, saying, “O People, it has only been thirteen [or fifteen] years since your Prophet ﷺ and you are disputing concerning the Qur’an!” On the basis of these evidences, many contemporary researchers have also agreed with Ibn Ḥajar’s assessment. Omar Hamdan suggests that the ʿUthmānic compilation project likely took place in stages, the first in 25 AH involving the actual compilation process and sending maṣāḥif specifically to Iraq and Syria as they were the regions involved in the disputes reported by Ḥudhayfah, while the second stage occurred in 30 AH and involved sending maṣāḥif to several other Muslim lands (e.g., Makkah, Bahrain, Yemen) that were not involved in the campaign in Armenia and Azerbaijan. While the suggestion of different stages is plausible in and of itself, one must take note of considerations from stemmatics (discussed below) in describing which maṣāḥif were sent in each stage.

A. The committee members

The task of assembling the ʿUthmānic codex was a major undertaking for which the Caliph ʿUthmān appointed a committee; having multiple members ensured a greater familiarity with multiple readings. 

ʿUthmān called Zayd ibn Thābit, ʿAbdullāh ibn al-Zubayr, Saʿīd ibn al-ʿĀs and ʿAbd al-Raḥmān ibn al-Ḥārith ibn Hishām, and then they wrote the manuscripts of the Qur’an in several codices (maṣāḥif). ʿUthmān said to the three Qurayshī men, “If you differ with Zayd ibn Thābit on any point of the Qur’an, then write it in the dialect (lisān) of Quraysh, as the Qur’an was revealed in their language.” So they acted accordingly.

From this report, we note that the core group of the committee comprised Zayd ibn Thābit (d. 45 AH), who was from the Anṣār (Madīnan Muslims), and three Muslims from the Quraysh. This report also raises the important question of what it means for the Qur’an to have been written according to, and revealed in, the dialect of Quraysh; how we reconcile this with the hadith of the Qur’an being revealed according to seven aḥruf; and whether the ʿUthmānic codex included more than one ḥarf. All of this will be discussed below.

Zayd ibn Thābit was a natural choice to lead the committee, having been the pre-eminent scribe of the Prophet (ﷺ). Moreover, he had memorized the entire Qur’an during the lifetime of the Prophet ﷺ and was the one responsible for compiling it under Abū Bakr. Some scholars mention that Zayd was selected in part because he attended the final review with the Prophet ﷺ. Al-Baghawī (d. 516 AH) states:

It is said that Zayd ibn Thābit attended the final review in which it was clarified what was abrogated and what remained. Abū ʿAbd al-Raḥmān al-Sulamī (d. 74 AH) said, “Zayd recited the Qur’an twice to the Prophet ﷺ during the year in which he passed away, and this recitation is called the qirāʾah of Zayd because he transcribed it for the Prophet ﷺ and recited it to him and witnessed al-ʿarḍah al-akhīrah (the final review), and he taught its recitation to people until he passed away. That is why Abū Bakr and ʿUmar relied upon him in its compilation and ʿUthmān appointed him in charge of writing the maṣāḥif—may Allah be pleased with them all.”

As noted in our previous article, there is some conflicting evidence concerning the matter of a specific companion attending the final review. Moreover, when Abū Bakr mentioned the qualities of Zayd when selecting him for the first compilation, he did not make any mention of the final review. It suffices to state that Zayd had not only memorized the entire Qur’an during the lifetime of the Prophet ﷺ, but would regularly transcribe it and recite it to him up until he passed away. Given his experience, expertise and knowledge, Zayd’s credentials were well-established among the companions.

Abū ʿAmr al-Dānī (d. 444 AH) explains that the significance of having Qurayshī members on the committee was to ensure that the written codex conformed to the dialect of the Quraysh. The members of the committee were all young at the time of  the ʿUthmānic compilation in 25 AH. Zayd ibn Thābit (d. 45 AH) was 36 years old, ʿAbdullāh ibn al-Zubayr (d. 73 AH) and Saʿīd ibn al-ʿĀṣ (d. 58 AH) were both 24 years old, and ʿAbd al-Raḥmān ibn al-Ḥārith ibn Hishām (d. 43 AH) was 23 years old. Saʿīd ibn al-ʿĀṣ (d. 59 AH) had returned from the frontiers along with Ḥudhayfah and was likely also familiar with the nature of the disputes causing confusion amongst the Muslims. He was also known for his proficiency in the Arabic language. One report notes:

ʿUthmān asked, “Who is the best scribe (man aktab al-nās)?” The people replied, “The scribe of the Prophet ﷺ, Zayd ibn Thābit.” He asked, “So who has the best Arabic (fa ayyu-nāsi aʿrab?)?” They replied, “Saʿīd ibn al-ʿĀṣ.” ʿUthmān said, “Let Saʿīd dictate and Zayd transcribe.”

A report traced back to Muḥammad ibn Sirīn (d. 110 AH) mentions that ʿUthmān gathered a total of 12 men. This indicates that in addition to the four core committee members, there were eight others who perhaps joined the committee later, or possibly played a secondary role in reviewing the manuscript compiled by the core members and transcribing additional copies of the muṣḥaf. The names of these additional members can be adduced from multiple other narrations scattered across disparate sources. These included six more companions (Ubayy ibn Kaʿb (d. 30 AH), ʿAbdullāh ibn ʿUmar (d. 73 AH), Anas ibn Mālik (d. 93 AH), ʿAbdullāh ibn ʿAbbās (d. 68 AH), ʿAbdullāh ibn ʿAmr ibn al-ʿĀṣ (d. 63 AH), Abū al-Dardāʾ (d. 32 AH)) and two tābiʿīn (Mālik ibn Abī ʿĀmir—the grandfather of Imam Mālik—and Kathīr ibn Aflaḥ). 

It should be noted that Ubayy ibn Kaʿb’s inclusion is subject to some debate. Imam al-Dhahabī cites narrations suggesting that Ubayy passed away during the reign of ʿUmar in 19 AH or 22 AH. However, Ibn Sirīn’s report mentions Ubayy by name, and this is supported by another more detailed account on the authority of Hānī al-Barbarī, the mawlā of ʿUthmān, which states that ʿUthmān sent Hānī to Ubayy with some words from three verses, the spelling of which Ubayy revised. Thus, evidence supports the view of al-Wāqidī who said, “And we have heard those who said that [Ubayy] died in the year 30 AH during the khilāfah of ʿUthmān, and that is the most well established opinion in our view, and that is consistent with ʿUthmān instructing him to compile the muṣḥaf.”

The formation of the committee thus ensured that the compilation process occurred based on direct learning from the Prophet ﷺ as attested by memorization, in addition to scribal experience and native eloquence in the Arabic language.

B. Did the ʿUthmānic text encompass the variant readings?

We now return to one of the most important questions that Muslim scholars have discussed concerning the ʿUthmānic muṣḥaf: did it include the variant readings of the seven aḥruf? Recall that during the lifetime of the Prophet Muhammad ﷺ, the Qur’an was recited in different ways which the companions had learned from the Prophet ﷺ. Ibn ʿAbbās narrated that the Prophet ﷺ said: “Jibrīl recited the Qur’an to me in one ḥarf. Then I requested him [to read it in another ḥarf] and continued asking him to recite in other aḥruf until he ultimately recited it in seven aḥruf.”In another narration, the Prophet ﷺ says, “‘O Jibrīl! I have been sent to an illiterate nation among whom are the elderly woman, the old man, the boy and the girl, and the man who cannot read a book at all.’ He said: ‘O Muhammad! Indeed the Qur’an was revealed in seven aḥruf (i.e., seven different ways of reciting).’”

Does the ʿUthmānic codex encompass the seven aḥruf? There are several factors to consider when answering this question. The first concerns the meaning of the seven aḥruf, a topic discussed in The Origin of the Variant Readings of the Qur’an. One must keep in mind that the scholars mentioned below did not share the same definition of what constitutes the aḥruf. Nonetheless, most of their views coalesce around a similar notion: the aḥruf entail different ways of reciting the Qur’an. Related to the same topic is the question of what it means for the text to encompass these different ways of varying. As well, there is also the question of the meaning of the aḥruf being a rukhṣah (concession) and its implications. Finally, there is the fact that the text of the ʿUthmānic codex was written with the Arabic letters lacking in consonantal dotting (naqṭ al-iʿjām) and with a complete absence of vowelization markings (naqṭ al-iʿrāb). As a result, a variety of readings can be accommodated by the consonantal skeleton of the text. The question is therefore whether the readings accommodated by the consonantal skeleton of the ʿUthmānic codices include only one ḥarf or some (or all) of the variation of the seven aḥruf.

The first opinion is that the ʿUthmānic codex was compiled according to only one ḥarf. This opinion was held by Ibn Jarīr al-Ṭabarī (d. 310 AH), Abū Jaʿfar al-Ṭaḥāwī (d. 310 AH), al-Naḥḥās (d. 338 AH),  Ibn ʿAbd al-Barr (d. 463 AH), al-Baghawī (d. 516 AH), al-Abyārī (d. 616 AH), Ibn Taymiyyah (d. 728 AH), Ibn al-Qayyim (d. 751 AH), and many other scholars. The principal evidence cited by many scholars in support of this view is ʿUthmān’s aforementioned statement to the committee: “If you differ with Zayd ibn Thābit on any point of the Qur’an, then write it in the dialect (lisān) of Quraysh, as the Qur’an was revealed in their language.” Initially, reciting according to seven aḥruf was a concession intended to make it easy for the unlettered Arab tribes to learn the Qur’an. However, after Islam spread to distant lands, the additional readings became a source of confusion and hence were no longer required. Moreover, as discussed earlier, the primary motivation for the ʿUthmānic muṣḥaf itself was the fact that Muslims were disputing concerning the different readings. This was the very concern brought forth by Ḥudhayfah. It follows that the ʿUthmānic muṣḥaf would aim to gather people upon one reading to reduce the confusion.

The second opinion is the opposite: that the ʿUthmānic codex contains all seven aḥruf, and not a single ḥarf is left out. This opinion was most famously espoused by Qaḍī Abū Bakr al-Bāqillānī (d. 403 AH), but was also held by Abū ʿAmr al-Dānī (d. 444 AH), Ibn Ḥazm (d. 456 AH), and ʿAlam al-Dīn al-Sakhāwī (d. 643 AH). The underlying logic is rather straightforward—how could ʿUthmān or any companion decide of their own accord to leave out of the Qur’an something that Allah had revealed to the Prophet Muhammad ﷺ? Al-Bāqillānī argues that the ʿUthmānic codex only left out those readings that had been abrogated, or readings that included commentary alongside the Qur’anic verses, or readings that were incorrectly attributed to the Prophet Muhammad ﷺ, which he believes were the real source of dispute that Ḥudhayfah encountered between the Muslims of Syria and Iraq. The modern-era Azharī scholar Muḥammad ʿAbd al-Aẓīm al-Zarqānī (d. 1367 AH/1948 CE) and other scholars held this view as well. The contemporary researcher al-Ḥasānīn follows al-Bāqillānī in this conclusion and argues that there is no evidence that the dispensation of the seven aḥruf was limited to a specific time period, nor is there any evidence that ʿUthmān and his committee dispensed with the other aḥruf. Similarly, ʿAbd al-Qayyūm al-Sindhī argues that the very existence of different qirāʾāt today necessitates that the variation of the aḥruf remains. After all, if the variation found in the seven aḥruf had been completely eliminated, then where did all the variation in the different qirāʾāt come from? Moreover, why did the ʿUthmānic maṣāḥif contain minor differences (discussed below) if it was only compiled according to one ḥarf?

If we pause to evaluate these two opinions, we may note that both have some merits. The first opinion fits the historical context that led to the ʿUthmānic muṣḥaf, while the second opinion connects the qirāʾāt to the aḥruf and explains how variation in recitation remained within the Muslim ummah. As for al-Bāqillānī’s and Ibn Ḥazm’s objection that the companions could not leave out readings that were revealed, the authentic narrations and historical evidence indicate that this is precisely what they did, and this was sufficiently explained by Al-Ṭabarī and al-Ṭaḥawī by noting that the companions understood that reciting according to the seven aḥruf was a concession and not an obligation. We have numerous authentic variant readings from the Prophet’s companions that are not found in the ʿUthmānic muṣḥaf and there is no evidence to suggest that they were abrogated. Much to the contrary, many companions continued to recite those readings after the passing of the Prophet Muhammad ﷺ. If the ʿUthmānic muṣḥaf included all the variant readings of the seven aḥruf then why would those readings be excluded? This demonstrates that the second position is not accurate.

There is an obvious middle ground between these two positions which leads to the third opinion, often described as the majority position, that the ʿUthmānic muṣḥaf contains some of the variation of the seven aḥruf. The qirāʾāt that are recited until today reflect the variation which remained from the seven aḥruf.  Ibn al-Jazarī (d. 833 AH) writes:

As for whether ʿUthmānic codices encompass all the seven aḥruf then this is a major topic… the position taken by the majority of the scholars from the earlier and later generations and the Imams of the Muslims is that these codices encompass that which the consonantal skeleton (rasm) can accommodate from the seven aḥruf. 

Elsewhere, Ibn al-Jazarī writes that the ten widely transmitted qirāʾāt recited throughout the Muslim world are part of the seven aḥruf. Ibn Ḥajar (d. 852 AH) explained that what remained after the ʿUthmānic compilation were the differences from the other aḥruf that could still be accommodated by the skeletal text of the ʿUthmānic codices—so what remained, in other words, were “some of the differences of the aḥruf, not all of them.”Ibn Ḥajar cites Abū al-ʿAbbās ibn ʿAmmār al-Mahdawī (d. 440 AH) who states, “The most correct position which is upheld by the experts is that what is recited now are some of [the differences] of the seven ḥurūf which were permitted to be recited and not all of them.” The remnants of the aḥruf are thus found amongst the various qirāʾāt recited today, while those aḥruf that did not conform to the ʿUthmānic codex were abandoned.Al-Mahdawī also states:

Verily these qirāʾāt that we recite constitute a portion of the seven aḥruf, upon which the Qur’an was revealed, which are still practiced because they conform to the ʿUthmānic muṣḥaf that the ummah had consensus upon. And the remainder of the ḥurūf were abandoned due to their differing from the written script of the muṣḥaf. This is because it is not obligatory upon us to recite with all the seven ḥurūf in which the Qur’an was revealed and because the Prophet ﷺ permitted for us to recite some of them to the exclusion of others based on the Divine statement, “Recite that which is easy from it” (Qur’an 73:20).

This middle position effectively reconciles the preceding first and second opinions. On one hand, the reason for the ʿUthmānic muṣḥaf was to reduce the differing and disputes amongst the Muslims by gathering them upon a standard text agreed upon by the companions of the Prophet (ﷺ). It brought about unity and solidarity in the ummah. Therefore, the first group of scholars are largely correct in that ʿUthmān sought to gather the Muslims upon a text that was primarily based upon one ḥarf. However, what needs to be noted is that some reading variations remained (from the seven aḥruf) that could still be recited because they conformed to the ʿUthmānic text, and these continued to be recited in certain regions of the Muslim world, becoming associated with the names of famous reciters to become the eponymous qirāʾāt that we know today.

The most precise characterization of this complexity is found in the writings of Makkī ibn Abī Ṭālib (d. 437 AH):

Those readings were no longer in practice which differed from the script of the (ʿUthmānic) muṣḥaf from the seven aḥruf in which the Qur’an was revealed, based upon the unanimous consensus on the script of the muṣḥaf. Thus, the muṣḥaf was written based on one ḥarf, and its script accommodates more than one ḥarf as it was void of dotting and vowels.

Christopher Melchert explains this view by stating that the consonantal skeleton is written with one ḥarf in mind, but the diacritical and vowelization marks preserve elements of the other six aḥruf as well. The contemporary Qur’anic scholar Ghānim Qaddūrī al-Ḥamad notes that there is no contradiction between stating that the ʿUthmānic muṣḥaf was written according to one ḥarf but that it accommodates more than one ḥarf, and cites the above quote by Makkī. The written text was naturally dictated and transcribed with one reading in mind, despite the script being able to accommodate more than one reading. Thus, it reduced the differences but did not eliminate them altogether. Essentially, “it unified the script without unifying the pronunciation.” Therefore, the middle position is best seen as a clarification of the first view with a more accurate characterization.

There is a subtlety that appears to have escaped the attention of many contemporary researchers on the topic of whether the ʿUthmānic muṣḥaf includes the seven aḥruf. While some include themselves in the third category (that the ʿUthmānic text accommodates some of the variation of the seven aḥruf), their views actually align more closely with the second category since they believe that whatever was left out of the ʿUthmānic muṣḥaf was not a valid reading since it was either not an authentic reading or was abrogated by the final review of the Prophet ﷺ with Zayd, or was Qur’an commentary. This doctrine that the Prophet’s final review (al-ʿarḍah al-akhīrah) abrogated other readings (aḥruf) is something that one finds commonly mentioned by scholars on the subject. However, other scholars argue that this claim lacks any definitive proof, which is a requirement to substantiate a claim of abrogation. Yaḥyā Aḥmad Jalāl and Halā Nāyif al-Mashāqbeh contend that the former view is based on the unsubstantiated assumption that leaving out some authentic revealed readings (qirāʾāt) would somehow undermine the preservation of the Qur’an. However, this is not how many classical scholars of Islam understood the matter. Confronted with the reality that there are authentic readings from the companions absent from the ʿUthmānic muṣḥaf, rather than engaging in conjecture regarding abrogation, it is far simpler to conclude that these are merely readings that go back to the concession of the seven aḥruf and hence were not included in the ʿUthmānic codex. This has been discussed in detail in the preceding article, The Origin of the Variant Readings of the Qur’an (see section entitled, “How do we view the variants reported from companions?”).

The most accurate position on this matter, then, is that the ʿUthmānic muṣḥaf (and by consequence the widely-accepted qirāʾāt) contained some of the variation found in the seven aḥruf, but not all of it. Indeed, it excluded those variant readings that entailed addition, subtraction, or substitution to the consonantal skeleton of the script (the rasm). On the other hand, those variant readings remained which related to pronunciation, alternate vowelization, and minor consonantal shifts that could be accommodated by the ʿUthmānic script, and it is from these that the qirāʾāt emerged, tracing back to readings the companions learned from the Prophet Muhammad ﷺ.

C. Write in the dialect of Quraysh

ʿUthmān instructed the committee, “If you differ with Zayd ibn Thābit on any point of the Qur’an, then write it in the dialect (lisān) of Quraysh, as the Qur’an was revealed in their language.” This has been used by many scholars as evidence that the ʿUthmānic muṣḥaf was written according to one ḥarf. Two questions arise from this statement.

The first question: given the hadith of the seven aḥruf, wasn’t the Qur’an revealed in multiple dialects and not just that of Quraysh? With respect to this question, some scholars add two important qualifiers to ʿUthmān’s statement: first, that the Qur’an was initially revealed according to the dialect of the Quraysh, according to one ḥarf, before permission was subsequently granted for it to be recited according to more than one ḥarf; and second, that the Qur’an contains many words that belong to Arabic dialects besides that of the Quraysh, meaning that most of it was revealed according to the dialect of the Quraysh.

The second question: what did ʿUthmān mean when he ordered that the Qur’an be written in the dialect of the Quraysh? Al-Jaʿfarī argues that in cases where ʿUthmān’s committee differed over the writing of words, ʿUthmān instructed them to write it (uktubūhū) according to the Qurayshī dialect, but he did not mandate that they only recite the Qur’an according to that dialect, thus leaving the potential for people to still recite according to other variant readings in those instances where the script accommodates different readings. 

To use a simple example cited by Ibn ʿAbd al-Barr, the Quraysh (and other Ḥijāzī dialects) tended not to pronounce the glottal stop (hamzah ء) in many words. Thus for example, the word muʾmin (believer) can be pronounced with a long vowel mūmin (as it is still recited according to most Madīnan and Baṣran reciters); meanwhile, the dialect of Tamīm was known for its use of the hamzah and is reflected in the readings of the Kūfan and Syrian reciters. The Arabic script مومن has been recited in both ways (with most modern day maṣāḥif printed according to Ḥafs ʿan Āṣim, adding the diacritical hamzah accordingly—مؤمن). Both readings have been authentically transmitted by the companions from whom the reciters learned, and consequently this represents an example of variance from the seven aḥruf that can still be recited due to its conformity with the script of the ʿUthmānic muṣḥaf. In some instances, the spelling between different readings differed as in the case of the word al-tābūt التابوت in Qur’an 2:248, which can be spelled and pronounced al-tābūh التابوه as it was by Zayd. ʿUthmān, however, preferred the first spelling and pronunciation because it accorded with the Qurayshī dialect. The actions of ʿUthmān and his committee in selecting a preferred reading from among various valid readings can be used as proof and precedent for the practice of ikhtiyār (selection) which was later practiced by the famous reciters. 

Moreover, ʿUthmān had precedence from other companions in this regard. When ʿUmar ibn al-Khaṭṭāb heard that ʿAbdullāh ibn Masʿūd was teaching the Qur’an according to his native dialect of Hudhayl to people in Kūfah, ʿUmar sent him a letter asking him to teach in the dialect of Quraysh, and not in Hudhayl, for the Qur’an was revealed in the dialect of Quraysh. Ibn ʿAbd al-Barr (d. 463 AH) notes that ʿUmar’s instruction can be understood as ikhtiyār, not as suggesting that Ibn Masʿūd’s reading was impermissible. Rather, ʿUmar was simply favoring the Qurayshī reading as preferable due to it being the original reading in which the Qur’an was recited, and the reading which was native to the Prophet Muhammad ﷺ himself. As Abū Shāmah notes, the effort required for non-Arabs to learn the Qur’an in Arabic is equivalent between dialects, hence it was preferable for non-Arabs to learn according to the dialect of Quraysh.

D. The method of producing the ʿUthmānic codex

If we return to the narrations concerning the ʿUthmānic compilation, we may note multiple indications that the process consisted of not only transcription and dictation but several stages of review and verification. The narration in Ṣaḥīḥ al-Bukhārī makes clear that ʿUthmān requested the compilation of Abū Bakr to use in the process of transcribing the new codex. Recall that the manuscript of Abū Bakr had been verified through an extremely meticulous process, not simply relying on the companions who had memorized the Qur’an in its entirety, but also ensuring each verse had two witnesses who had written it down in the presence of the Prophet ﷺ in a manner that matched the way the companions had memorized it. With Abū Bakr’s passing, the compilation passed into the hands of ʿUmar ibn al-Khaṭṭāb, and then, after his passing, into the hands of his daughter Ḥafṣah bint ʿUmar, the widow of the Prophet Muhammad (ﷺ).

ʿUthmān sent a message to Ḥafṣah saying, “Send us the manuscripts of the Qur’an so that we may copy the Qur’anic materials in perfect copies and return the manuscripts to you.” Ḥafṣah sent it to ʿUthmān. ʿUthmān then instructed Zayd ibn Thābit, ʿAbdullāh ibn Al-Zubayr, Saʿīd ibn Al-ʿĀṣ and ʿAbd al-Raḥmān ibn al-Ḥārith ibn Hishām to rewrite the manuscripts in perfect copies.

From this narration, one might be inclined to state, as many scholars did, that ʿUthmān’s compilation was assembled by simply copying the manuscript compiled by Abū Bakr. However, Muḥammad Muṣṭafā al-Aʿẓamī argues that combining the above account with other narrations paints a more complete picture in which ʿUthmān’s compilation was assembled independently of Abū Bakr’s compilation, which was only used during the verification process after ʿUthmān’s committee had finished transcribing the muṣḥaf. The first narration that al-Aʿẓamī cites in support of this narrative comes from Muṣʿab ibn Saʿd:

ʿUthmān delivered a sermon and said, “The people have diverged in their recitations, and I am determined that whoever holds any verses dictated by the Prophet ﷺ himself must bring them to me.” So the people brought their verses, written on parchment and bones and leaves, and anyone contributing to this pile was first questioned by ʿUthmān, “Did you learn these verses [i.e., take this dictation] directly from the Prophet ﷺ himself?” All contributors answered under oath, and all the collected material was individually labeled and then handed to Zayd ibn Thābit.

The second narration al-Aʿẓamī cites comes from Mālik ibn Abī ʿĀmir (one of the committee members and the grandfather of Imam Mālik ibn Anas), who states:

I was among those upon whom the muṣḥaf was dictated [from the written sources], and if any controversies arose concerning a particular verse they would say, “Where is the scribe [of this parchment]? Precisely how did the Prophet ﷺ teach him this verse?” And they would resume scribing, leaving that portion blank and sending for the man in question to clarify his scribing.

These narrations indicate that the muṣḥaf of ʿUthmān was prepared autonomously and gathered through an independent process before with the compilation of Abū Bakr. One can further strengthen al-Aʿẓāmī’s argument by noting that ʿUthmān’s instructions to the committee on what to do if they differ concerning a verse only make sense if they were conducting an independent compilation process. If they were simply copying the text of Abū Bakr’s compilation, then there would be no occasion for them to differ in the first place. Moreover, the narrations indicating that Zayd transcribed while Saʿīd  dictated also suggest a process more involved than mere copying. Finally, we have an account which, in spite of its weaknesses, mentions that ʿUthmān requested Ḥafṣah’s copy after the compilation by Zayd was completed, which was then reviewed and compared with the ʿUthmānic codex and confirmed to be in agreement.

The question that emerges is why ʿUthmān would undertake this exhaustive process when he could have readily copied the manuscripts already compiled by Abū Bakr. Al-Aʿẓamī offers the following reasoning:

One may wonder why Caliph ʿUthmān took the trouble to compile an autonomous copy when the end product was to be compared with the [compilation of Abū Bakr] anyway. The likeliest reason is a symbolic one. A decade earlier thousands of Companions, engaged in the battles against apostasy in al-Yamamah and elsewhere, were unable to participate in the Ṣuḥuf’s compilation. In drawing from a larger pool of written materials, ʿUthmān’s independent copy provided these surviving Companions with an opportunity to partake of this momentous endeavor.

In the above account no inconsistencies were found between the [compilation of Abū Bakr] and the independent muṣḥaf and from this two broad conclusions emerge: first, the Qur’anic text was thoroughly stable from the earliest days and not (as some allege) fluid and volatile until the third century; and second, the methods involved in compilation during both reigns were meticulous and accurate.

If al-Aʿẓamī’s conclusion is correct, one must note that rather than merely having a ‘symbolic’ reason for conducting an independent compilation, ʿUthmān also had a very practical and tangible reason: reducing differences in the ummah’s readings entailed elimination of some of the variant readings from the seven aḥruf. Since the goal of Abū Bakr’s compilation was simply to preserve the text of the Qur’an, arbitrating between variant readings and different dialects was never a stated goal or component of the process. It is possible that Abū Bakr’s compilation included a mix between a greater number of readings from different dialects and other types of variants from the seven aḥruf. Therefore, the advantage of repeating the collection process and independently reviewing every verse from written sources in addition to memory afforded the committee the opportunity to affirm with the highest degree of confidence and certainty that the reading they selected for the writing of the muṣḥaf was the reading taught and recited by the Prophet Muhammad ﷺ himself.

At this juncture, it is useful to clarify a set of narrations which have been the source of some confusion concerning the detection of a “missing” verse during the compilation process. During the compilation of Abū Bakr, while seeking two witnesses for each passage, Zayd ibn Thābit came across a passage for which he found only one witness. Zayd reports, “Abū Bakr sent for me, so I collected the Qur’an till I found the last part of Sūrah al-Tawbah (verses 9:128–129) with Abū Khuzaymah Al-Anṣārī and did not find it with anybody else.” This narration, found in Ṣaḥīḥ al-Bukhārī, explicitly notes that this took place during the first compilation, and that the witness  was Abū Khuzaymah. Again, one must clarify that this does not mean that Abū Khuzaymah was the only one who knew this verse, as all the companions who had memorized the Qur’an knew it. However, they were seeking a witness who had documented the verse in the presence of the Prophet Muhammad (ﷺ). The very reason they knew exactly which verse was missing was because they had memorized it.

Another narration, also from Zayd ibn Thābit and related in Ṣaḥīḥ al-Bukhārī, mentions the same incident but with the name Khuzaymah al-Anṣarī:

So I started locating Qur’anic material and collecting it from parchments, scapulae, leaf-stalks of date palms, and from the memories of men (who knew it by heart). I found with Khuzaymah al-Anṣārī two verses of Sūrah al-Tawbah which I had not found with anybody else [he cites Qur’an 9:128–129].

Such a minor difference in the name (Abū Khuzaymah vs Khuzaymah) would typically not raise any questions, except that we have a third narration, also in Ṣaḥīḥ al-Bukhārī, about an identical incident involving a different verse with Khuzaymah.

When we wrote the Holy Qur’an, I missed one of the verses of Sūrah al-Aḥzāb which I used to hear Allah’s Messenger ﷺ reciting. Then we searched for it and found it with Khuzaymah ibn Thābit al-Anṣārī. The verse was “Among the Believers are men who have been true to their Covenant with Allah, Of them, some have fulfilled Their obligations to Allah [i.e., they have been killed in Allah’s cause], And some of them are [still] waiting” (33:23) So we wrote this in its place in the Qur’an.

Notice that this narration does not explicitly mention whether this occurred during Abū Bakr’s or ʿUthmān’s compilation. There are other narrations of this incident in other sources that use either name for each incident. Al-Bāqillānī argues that these narrations may be dismissed as contradictory and conflicting with more reliable evidence, or may be reinterpreted in various ways. Ibn Ḥajar al-ʿAsqalānī (d. 852 AH) and others reconciled the narrations by stating that the first incident took place during the compilation of Abū Bakr, and that it was the verse of Sūrah al-Tawbah that was found with Abū Khuzaymah. Meanwhile, the second incident took place during the compilation of ʿUthmān with the verse of Sūrah al-Aḥzāb, which was found with Dhū al-Shahādatayn Khuzaymah ibn Thābit al-Anṣārī, a different person from Abū Khuzaymah ibn Aws ibn Yazīd ibn Aṣram. Al-Aʿẓamī uses this as evidence for his argument that ʿUthmān repeated the process of summoning witnesses for each verse, otherwise there is no reason they would not be able to find the verse from al-Aḥzāb if the compilation of Abū Bakr was in front of them. Al-Jaʿfarī suggests it is not inconceivable that the written parchment of one verse may have gone missing during the time interval between the two compilations, which exceeded a decade.

On the other hand, we have scholars like Ibn Kathīr (d. 773 AH) and others who held that both incidents happened during the time of Abū Bakr. While Muḥammad Ḥasan Jabal agrees with Ibn Ḥajar’s analysis that the incidents described by Zayd refer to two different people, he also concurs with Ibn Kathīr’s opinion that both occurred during the compilation of Abū Bakr. 

Ghānim Qaddūrī al-Ḥamad adduces additional evidence suggesting that the incident of the lost verse happened only during the time of Abū Bakr, but  also that both incidents are in fact one incident that happened with the same person. He cites the following narration from the introduction to Kitāb al-Mabānī, in which Zayd describes the compilation during the time of Abū Bakr:

I completed one review and I noticed I was missing this verse [33:23] so I asked the Muhājirīn and the Anṣār and I did not find it [in written form] with any of them, although I knew the verse and the Prophet ﷺ had dictated it to me, however I disliked to establish it until someone else testified alongside me. And then I obtained it from Khuzaymah ibn Thābit al-Anṣārī whom the Prophet ﷺ had made his witness equal to that of two witnesses. So I wrote the verse and then I conducted another review. And I found that I was missing two verses although I knew them [9:127–128]. So I inquired about them from Muhājirīn and the Anṣār and I did not find them with any of them except with Khuzaymah ibn Thābit al-Anṣarī whom the Prophet ﷺ had endorsed his witness. So I wrote them at the end of Barāʾa [Sūrah al-Tawbah].

Ghānim Qaddūrī al-Ḥamad states that this account alleviates some of the confusion found in other sources. He also notes that “the similarity between the two names (i.e., Abū Khuzaymah and Khuzaymah) and their mention in different narrations using the very same phrasing indicates that they are both names referring to the very same companion and that is Khuzaymah ibn Thābit al-Anṣārī.” This is the simplest and easiest explanation for what would otherwise seem to be a rather striking coincidence of the exact same circumstances involving two different individuals with almost identical names. Nonetheless, other ways of reconciling the narrations, such as that of Ibn Ḥajar, were also accepted by many scholars. Finally, it should be mentioned that the verse(s) in question have been attested in the earliest manuscripts, and it is not reported that a single companion codex omitted them.

There is of course a remarkable story behind the significance of Khuzaymah ibn Thābit being the witness for this verse. According to the narration in Ṣaḥīḥ al-Bukhārī, the Prophet Muhammad ﷺ regarded Khuzaymah’s witness as equal to that of two witnesses, earning him the title of Dhū al-Shahādatayn (Possessor of Dual Testimony). The story behind this begins with the Prophet ﷺ purchasing a horse from a bedouin (in other narrations identified as Sawāʾa ibn al-Ḥārith al-Muḥāribī). After agreeing upon the price and the purchase, the Prophet ﷺ asked the bedouin to accompany him to retrieve the payment. While on their way, people saw the horse and, unaware that it had already been sold, started bargaining with the bedouin for it. The bedouin took advantage of the situation and told the Prophet ﷺ, “If you want this horse, then buy it [by bidding higher], otherwise, I will sell it.” The Prophet ﷺ asked, “Have I not already purchased it from you?” The bedouin asked the Prophet ﷺ to produce a witness. Khuzaymah ibn Thābit testified on his behalf, whereupon the Prophet ﷺ asked him how he could testify when he wasn’t present. Khuzaymah replied that “Because I believe in you (as a Prophet) and know that you do not speak except with that which is truth.” Thereupon, the Prophet ﷺ made the testimony of Khuzaymah equal to that of two witnesses. This was a Prophetic endorsement of the purity of Khuzaymah’s faith and character in hastening to testify to the truthfulness of the Messenger of Allah (ﷺ). And it is said that the Prophet ﷺ still chose to return the horse to the bedouin, undoubtedly as a gesture of kindness, integrity, and generosity. 

This narration underscores the Divine will behind the miraculous turn of events that caused the one person about whom the Prophet ﷺ had made this declaration to be the same person with whom Zayd found the written copy of the ‘missing’ verse(s).

Since the time of ʿUthmān until today, the 114 chapters of the Qur’an have always been written in maṣāḥif in the same sequence, beginning with al-Fātiḥah and ending with al-Nās. Was this order mandated by Divine guidance and taught by the Prophet ﷺ himself (referred to as tawqīfī), or did the companions deduce it themselves through their own inference (ijtihād)? This is a well-known difference of opinion debated in classical and contemporary works of Qur’anic studies. The first view, attributed to many scholars, is that the present order of the sūrahs is based on Divine instruction. Abū Jaʿfar al-Naḥḥās (d. 338 AH) states, “The order of the sūrahs is from Allah and His Messenger ﷺ, without any involvement of anyone else.” The evidence cited in support of this position includes the fact that the companions of the Prophet ﷺ unanimously agreed upon this order of the chapters. Moreover, the Prophet ﷺ recited the Qur’an every year to Jibrīl, which logically requires that it was recited in a particular order. Abū Bakr ibn al-Anbārī (d. 328 AH) states, “Jibrīl would teach the Prophet ﷺ the location of each verse and sūrah, so the order of the sūrahs is akin to the order of verses and letters, all of it being from the Prophet Muhammad (ﷺ).” Al-Baghawī writes, “It is established that the work of the companions was to assemble the Qur’an into one compilation, not to determine its sequence, for verily the Qur’an is written in al-Lawḥ al-Maḥfūẓ according to its sequence in our maṣāḥif.” It has also been narrated from Aws ibn Ḥudhayfah al-Thaqafī (d. 59 AH) that he asked the companions how they used to divide the Qur’an when they completed its recitation and they replied, “Into three sūrahs, then five, seven, nine, 11, 13, and then the mufaṣṣal sūrahs.” This totals 48 chapters before the mufaṣṣal chapters, which start either with al-Ḥujurāt (chapter 49) or Qāf (chapter 50). This indicates that there was a fixed order to the chapters known to the companions in the time of the Prophet (ﷺ). 

The second opinion is that the sequence of the sūrahs was based on the ijtihād of the companions of the Prophet ﷺ. Ibn Juzayy al-Kalbī (d. 741 AH) writes, “The sequence of the sūrahs according to their present arrangement is based on the action of ʿUthmān and Zayd ibn Thābit and those who wrote the musḥaf with them. And it has been said that it is based on the action of the Messenger of Allah ﷺ, however that is weak and refuted by the relevant narrations.” Commenting on a narration in Ṣaḥīḥ Muslim where the Prophet ﷺ recited Sūrah al-Nisāʾ prior to Sūrah Āl ʿImrān, Qāḍī ʿIyāḍ writes, “This is an evidence for those who say that the sequence of sūrahs is based on the ijtihād of the Muslims when they wrote the muṣḥaf and was not defined by the Prophet ﷺ but rather he left it to the ummah after him, and this is the opinion of the majority of scholars, and the opinion of Imam Mālik and the preferred view of al-Qāḍī Abū Bakr al-Bāqillānī.” Of course, this narration is not decisive evidence as most scholars do not regard it obligatory to recite the sūrahs in prayer in order. Furthermore, it should be noted that Imam Mālik did state that the companions made their decision based on what they heard from the Prophet ﷺ, which seems to bring the two opinions closer together. 

Some scholars, like Ibn ʿAṭiyyah, held that some of the sūrahs were arranged according to ijtihād but not all of them. The first and foremost evidence used to support this opinion is the fact that it is reported that some of the companions’ codices had the sūrahs arranged according to a different order. However, these were not official public codices but rather personal compilations comprising sūrahs these companions learned from the Prophet (ﷺ). Ibn Baṭṭāl notes also that these companion codices were compiled before the revelation of the Qur’an was concluded and its sūrah sequence was finalized.

Another critically important piece of evidence for those who suggest that the chapter sequence was, at least in part, based on the ijtihād of the companions, is a narration whose authenticity is disputed in which Ibn ʿAbbās asked ʿUthmān why he placed Sūrah al-Tawbah after Sūrah al-Anfāl in the Qur’an without writing the basmalah between them, and ʿUthmān replied that it was not clear whether they were part of the same chapter and the Prophet ﷺ passed away before clarifying the matter. This narration led scholars like al-Ṣuyūṭī, al-Bayhaqī, and Ibn al-ʿArabī to consider these two sūrahs to be the exception, while all the remaining sūrahs in the Qur’an were ordered based on the sequence learned from the Prophet (ﷺ). 

However, this narration has been subjected to extensive critique from the perspective of both the chain of transmission (isnād) and the content (matn). This hadith is only narrated by ʿAwf ibn Abī Jamīlah (d. 146  AH) from Yazīd al-Fārisī. The religious beliefs of ʿAwf have been described as Qadarī and Shiʿī, which introduces the potential for sectarian bias in narrations related to ʿUthmān’s role in the preservation of the Qur’an. More importantly, Yazīd al-Fārisī is listed as a weak narrator by Imam al-Bukhārī himself. On the basis of the problems in the chain of transmission and the content of the narration, it was rejected as baseless by the hadith scholar Aḥmad Shākir, although this view has been contested by ʿAbdullāh Judayʿ who views the narration to be authentic. With respect to the content of the narration, it states that the Prophet ﷺ passed away without clarifying this matter, and yet al-Tawbah was revealed 15 months before he passed away which renders this unlikely, particularly given the constant teaching and daily ritual recitation of the Qur’an. Moreover, ʿAlī ibn Abī Ṭālib gives us a different explanation for the absence of the basmalah at the beginning of al-Tawbah: this is how the sūrah was revealed since the basmalah is a declaration of mercy, while the opening of al-Tawbah is a declaration of war. A final piece of evidence comes from manuscript research. The Ṣanʿāʾ palimpsest lower text (discussed in detail below), which likely predates the ʿUthmānic codification, also places Sūrah al-Tawbah after Sūrah al-Anfāl without a basmalah in between, thus making it unlikely that this was a new decision of the ʿUthmānic committee.

Of course, all Muslim scholars agree that the order of the sūrahs in the Qur’an since the ʿUthmānic muṣḥaf is binding and obligatory to follow. Imam Rabīʿah (d. 136 AH) was asked why the Qur’an begins with the Madīnan chapters al-Baqarah and Āl ʿImrān when the Makkan revelations preceded them. He replied, “They were placed at the beginning because the Qur’an was arranged according to knowledge possessed by those who compiled it and those who were with them, and they had unanimous consensus upon that knowledge, so this [the ordering of the sūrahs] is a final matter that cannot be questioned.” Indeed, that the sequence of sūrahs in the ʿUthmānic muṣḥaf was unanimously followed by the early Muslim community, with people completely abandoning the alternative arrangements of various companions’ personal codices, arguably indicates widespread agreement on the singular authority of the sequence of the ʿUthmānic codex. This would not have been the case if the ordering of the chapters was a matter of opinion. 

The position that the current sequence of chapters in the muṣḥaf has been Divinely determined appears to be the strongest view on the strength of all the evidence discussed. Also relevant to this discussion is the wealth of scholarly literature surrounding the intricate connections (tanāsub) between adjoining sūrahs as well as the thematic structure (naẓm) and continuity of the Qur’an from one chapter to the next. The former includes not only tafsīr literature but entire works devoted to the topic such as al-Burhān fī tanāsub suwar al-Qurʾān by Ibn al-Zubayr al-Gharnāṭī (d. 708 AH), Naẓm al-durar fī tanāsub al-ayāt wal-suwar by al-Biqāʿī (d. 885 AH), Tanāsuq al-durar fī tanāsub al-suwar by al-Suyūṭī (d. 911 AH), and Jawāhir al-bayān fī tanāsub suwar al-Qurʾān by ʿAbdullāh Ṣiddīq al-Ghumārī (d. 1413 AH). The vast multitude of intertextual connections uncovered by such astute insights is seen by many scholars as effectively settling the debate on this matter.

When the compilation of the ʿUthmānic codex was complete, ʿUthmān sent an official copy to each of a number of major cities within the Muslim world and instructed that other unofficial copies of the Qur’an be burned.

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.

When presented without context, particularly by the unlearned, this act of burning takes on almost dark undertones, as though it somehow undermines the preservation of the Qur’anic text. However, for those familiar with all the facts analyzed in the preceding discussions, ʿUthmān’s instruction is entirely expected and was in fact an integral part of the history of the preservation of the Qur’anic text. The whole point of the ʿUthmānic project was to unify the Muslim ummah on one muṣḥaf that would eliminate confusion caused by the proliferation of different aḥruf taught by various companions in different lands. Had ʿUthmān not undertaken the burning of copies of the Qur’an which differed from the official copies, and thus not eliminated a source of the disputes, the entire exercise would have been futile. There also would have remained the possibility of individual Muslims reciting from copies that contained scribal errors or contained the commentary of companions alongside verses of the Qur’an. Given that it would be quite unfeasible to check and correct each and every personal codex to eliminate the possibility of error, the logical solution was to simply eliminate those codices that had not been copied from the ʿUthmānic archetypes.

Perhaps this episode generates such confusion because book burning incidents in Western history are typically seen as a manifestation of condemnation or desecration. For instance, after renouncing Arianism and converting to Catholicism, the Visigothic King Reccared I (d. 601 CE) ordered all Arian books to be collected and burned. However, in the Islamic tradition, one of the respectful ways to dispose of any text containing verses of the Qur’an is in fact through burning it. Ibn Baṭṭāl writes, “And ʿUthmān’s instruction to burn the manuscripts and copies of the Qur’an after his compilation demonstrates that it is permissible to burn books that contain the names of Allah [or Qur’anic verses] and that is a form of respect towards it and protection from it being stepped upon or discarded on the ground.” Some copies of the Qur’an in the time of ʿUthmān were eliminated in other ways including erasing or tearing.

Moreover, the historical evidence indicates that elimination was not the only outcome for other copies. Muḥammad Ḥasan Jabal and Muḥammad Muṣṭafā al-Aʿẓamī both note that another possible outcome was that manuscripts could be reviewed and corrected according to the ʿUthmānic muṣḥaf. They cite a narration from ʿAbd al-Aʿlā al-Kilābī that mentions that he visited the home of Abū Mūsā al-Ashʿarī, who was with Ḥudhayfah and ʿAbdullāh ibn Masʿūd, and there was a copy of the Qur’an that ʿUthmān had sent along with the instruction to correct their own copies accordingly (wa amarahum an yuqīmū maṣāḥifahum ʿalayh).

The compilation of Abū Bakr which remained in the possession of Ḥafṣah was not erased or burned by ʿUthmān. However, the compilation was later requested by Marwān ibn al-Ḥakam (at that time the governor of Madīnah). Ḥafṣah refused. However, when she passed away in 41 AH/665 CE, Marwān retrieved it and had it destroyed. Marwān stated, “I only did that because whatever was in it was already written and preserved in the [ʿUthmānic] muṣḥaf. And I feared that time would pass and people would start to have misgivings about this [Ḥafṣah’s] ṣuḥuf or would say, “It contained something that has not been written.”

The result of the ʿUthmānic codification is that Muslims of all provinces began adhering to a single agreed-upon text in their recitation of the Qur’an, abandoning those readings that did not conform to it. Oddly, Shady Hekmat Nasser refers to the codification process undertaken by ʿUthmān as a “failure” because it “was not able to produce a single unified Reading of the Qur’an.” However, as previously discussed, there is no evidence that ʿUthmān mandated for Muslims to recite in only one qirāʾah. Rather, Muslims were allowed to continue to recite the way they had been taught so long as their readings conformed to the authorized text. While it is true that differences remained in the qirāʾāt, there was now a common reference point to ensure that readings were sound. So long as a person was reciting according to the way they had learned from the companions of the Prophet ﷺ, and the reading conformed to the codex, their reading was to be considered valid. This removed the problem of Muslims rejecting the validity of other Muslims’ readings.

Nasser’s assessment could not be any further from reality; the very fact that every non-ʿUthmānic copy of the Qur’an on the face of the earth vanished with scarcely a trace is the greatest testament to the success of the ʿUthmānic project. Shortly after the ʿUthmānic project in 25 AH, Muslims everywhere around the world were reciting the exact same text, and this has continued up until the present, more than 14 centuries later. The reality of the matter is that the ʿUthmānic project is an astonishingly successful act the likes of which human history has rarely witnessed.

Al-Aʿẓamī concludes:

The unanimous agreement to dispose of (or amend) all earlier copies made ʿUthmān’s script and spelling the new standard; from then on every Muslim learning the Qur’an had to conform with the ʿUthmāni text. Where a person’s previous schooling was at odds with this text, he was not granted leave to recite or teach in that divergent manner. So what could such a person do? Attending an official reciter’s circle was the simplest solution, to learn the Book in accordance with the conditions laid and thereby regain the privileges of teaching and recitation. ʿUthmān’s unparalleled success in this regard is proof positive that his actions echoed the voice of the community.

Thus, ʿUthmān’s instruction to burn or amend Qur’anic manuscripts that differed from the official codices reflected the community’s agreement upon the ʿUthmānic codex. Rather than undermining the text’s preservation, this historical event decisively secured it.

The success of the ʿUthmānic project was amplified by the widespread approval of the companions of the Prophet Muhammad ﷺ. ʿAlī ibn Abī Ṭālib said, “Do not speak of ʿUthmān except with what is good, for I swear by Allah that he did not do what he did with the maṣāḥif except with complete approval from the rest of us.” In another report he said, “May Allah have mercy on ʿUthmān! Were I in charge, I would have done as he did.” Muṣʿab ibn Saʿd (d. 103 AH), the son of Saʿd ibn Abī Waqqāṣ, said, “I was present among the people when ʿUthmān burned the other maṣāḥif and they were pleased with his action, and no one opposed it.” Indeed, the unanimous consensus (ijmāʿ) of the Prophetic companions upon the ʿUthmānic codex renders it as indubitable and incontestable as any other fundamental component of the religion. Abū Jaʿfar al-Ṭaḥāwī (d. 321 AH) comments:

Zayd ibn Thābit, the Prophet’s scribe, wrote the codex for ʿUthmān and the companions of the Prophet ﷺ agreed upon that, thus it became ijmāʿ (unanimous consensus). And transmission by consensus is a decisive proof (ḥujjah) the likes through which Islam has been transmitted to us such that we are able to learn its teachings and such that we learn the number of prayers and the like.

Essentially, what al-Ṭaḥāwī is stating is that everything we know about Islam was transmitted to us through the very same community that came to collective agreement upon the ʿUthmānic codex. Epistemologically, this is the greatest proof of the veracity of the text. This generation is the same one that learned the Qur’an directly from the Prophet ﷺ and transmitted to us all of the teachings of our faith. Their collective agreement is therefore unassailable and binding.

Nonetheless, there are a number of narrations that require clarification due to mentioning displeasure or disagreement on the part of ʿAbdullāh ibn Masʿūd (d. 32 AH). These narrations have been examined and analyzed in great detail by a number of scholars. The credentials of Ibn Masʿūd with respect to knowledge and recitation of the Qur’an are well known. Ibn Masʿūd was the first Muslim to ever recite the Qur’an in public. His recitation of the Qur’an was praised by the Prophet ﷺ himself for being “as fresh as how it was revealed.” And he is one of the four from whom the Prophet ﷺ told others to learn the Qur’an. This being the case, what was the reason for his opposition and what exactly was he opposed to?

In his work al-Maṣāḥif, Ibn Abī Dāwūd (d. 316 AH) devoted a chapter to the subject, entitling it, “The displeasure of Ibn Masʿūd towards that,” meaning the ʿUthmānic project. Careful review of the narrations in question demonstrates that Ibn Masʿūd’s displeasure was on two counts: (1) he was excluded from the committee that compiled the ʿUthmānic codex and (2) he did not wish to give up his personal muṣḥaf to be destroyed nor to abandon the reading to which he was most accustomed.

As for the first reason, Ibn Masʿūd felt unfairly excluded from the codification process itself. He was certainly one of the most learned companions of the Prophet ﷺ with respect to the Qur’an, and so he was incensed at the fact that the project was overseen instead by Zayd ibn Thābit, almost 20 years his junior. In addition, ʿAbdullāh ibn Masʿūd was from the Muhājirīn, those who migrated with the Prophet  ﷺ from Makkah to Madīnah, having experienced persecution in Makkah and having sacrificed everything out of their commitment to Islam. Meanwhile Zayd ibn Thābit was a Madīnan Muslim; i.e., of the Anṣār. Being upset about the matter, Ibn Masʿūd made some uncharacteristically harsh statements. Al-Zuhrī (d. 124 AH) reports that ʿAbdullāh ibn Masʿūd disliked Zayd ibn Thābit copying the maṣāḥif, and he said, “O Muslims! [How is it that] I am removed from recording the muṣḥaf and it is overseen by a man, by Allah, when I accepted Islam he was but in the loins of a disbelieving man”—referring to Zayd ibn Thābit. 

The companions of the Prophet ﷺ were human beings and experienced human emotions. Abū Bakr Al-Anbārī (d. 328 AH) was of the opinion that Ibn Masʿūd said this in a moment of anger and later recanted his statement after his anger subsided. Al-Anbārī writes:

The objection made by ʿAbdullāh ibn Masʿūd arose in a state of anger and it is not to be acted upon nor taken from him. There is no doubt that once his anger passed, he was satisfied with the excellence of the decision of ʿUthmān and the companions of the Messenger of Allah ﷺ and concurred with their agreement and abandoned his opposition to them.

Moreover, as al-Anbārī also notes, the virtue for which Zayd was selected was that he had memorized the entirety of the Qur’an during the lifetime of the Prophet ﷺ, while Ibn Masʿūd learned seventy some sūrahs from the Prophet ﷺ and the remainder after the Prophet ﷺ passed away. As mentioned earlier, Zayd ibn Thābit had also undertaken the compilation during the time of Abū Bakr and therefore his expertise and experience was well-established. That, along with his availability and young age, qualified him to lead the ʿUthmānic project. Abū al-ʿAbbās al-Qurṭubī (d. 656 AH) also points out that the ʿUthmānic committee was tasked with writing the Qur’an according to the dialect of Quraysh, while Ibn Masʿūd was from Hudhayl and was known to recite in the dialect of Hudhayl. Another important factor was simply logistical; as al-Dhahabī (d. 748 AH) pointed out, Ibn Masʿūd was in Kūfah, not Madīnah, when ʿUthmān undertook the codification process. It would have not been ideal to delay the entire project for Ibn Masʿūd to travel to Madīnah, particularly when the matter had already resulted in conflict, as identified by Ḥudhayfah.

As for the second reason for Ibn Masʿūd’s opposition, it was not because he objected to the reading of the ʿUthmānic codex itself but rather because he wished to continue reciting according to his own reading, and he did not wish to surrender his muṣḥaf to be destroyed. As Abū Isḥāq al-Shāṭibī (d. 790 AH) astutely observed:

No one disagreed in this matter except ʿAbdullāh ibn Masʿūd. He refused to discard the reading that he possessed which differed from the ʿUthmānic codices. He said, “O People of Iraq! O People of Kūfah! Conceal the maṣāḥif which you have in your possession and withhold them. For verily Allah states, ‘And whoever withholds anything will come forth with what he has withheld on the Day of Judgment.’ So meet Allah with the maṣāḥif.” So consider carefully Ibn Masʿūd’s words because he did not disagree with compiling the ʿUthmānic codex but rather he disagreed concerning another matter.

What al-Shāṭibī has noted is that Ibn Masʿūd never expressed disapproval of the ʿUthmānic muṣḥaf’s compilation, nor did he deny the authenticity of the reading upon which it was compiled. In fact, we do not have a single statement from him criticizing a reading that is found in the ʿUthmānic codex. Instead, what appears to be the case is that Ibn Masʿūd did not wish to surrender his personal copy of the Qur’an to be burnt. He did not wish to abandon those readings that could not be accommodated by the ʿUthmānic text, as he had learned them from the Prophet ﷺ and taught them for decades. In other words, Ibn Masʿūd refused the obligation to adhere to the ʿUthmānic text, but did not otherwise deny the validity of the compilation or its reading. In support of this view, one may note that Ibn Masʿūd stated, “Whose qirāʾāh is it that you command me to recite? Verily, I recited directly to Allah’s Messenger ﷺ more than 70 chapters of the Qur’an and the companions of Allah’s Messenger certainly know that I am the most knowledgeable of them when it comes to the Book of Allah. And if I knew of anyone more knowledgeable than me, I would travel to him.” Another point of evidence is that Ibn Masʿūd never expressed disagreement with Abū Bakr’s compilation since it was never imposed upon him.

Furthermore, one may note that many scholars were of the view that Ibn Masʿūd later recanted his position. Ibn Abī Dāwūd himself alludes to this by titling a chapter, “Ibn Masʿūd’s [subsequent] approval  of ʿUthmān’s compiling the maṣāḥif,” in which he includes a conciliatory narration from Ibn Masʿūd affirming the validity of readings according to the seven aḥruf. The historian Ibn ʿAsākir (d. 571 AH) said, “It was narrated that Ibn Masʿūd later agreed and adhered to and approved of ʿUthmān’s decision and revised his previous position.” Al-Dhahabī (d. 748 AH) said: “It was reported that Ibn Masʿūd agreed and followed ʿUthmān.” In Faḍāʾil al-Qurʾān, Ibn Kathīr (d. 774 AH) said it was reported that Ibn Masʿūd recanted. However, if one adopts the view of al-Shāṭibī, then since Ibn Masʿūd never objected to the correctness of the ʿUthmānic codex in the first place, there was nothing to recant. It is also possible that Ibn Masʿūd was not initially pleased with the obligation of adhering to a codex that didn’t favor his reading,  but after his anger passed, he recognized the importance of the ʿUthmānic project.

Ibn Masʿūd’s massive influence in Kūfah slowed the transition to the ʿUthmānic codex. The early scholar al-Aʿmash (d. 148 AH) is reported to have said, “I reached Kūfah, and the reading of Zayd was not with them except as the reading of ʿAbdullāh is with you today: no one recited it except for one or two people.” Other companions expressed some hesitation in changing their reading as well. When ʿAlqamah ibn Qayṣ (d. 62 AH) traveled to Greater Syria, he met the companion Abū al-Dardāʾ, and the latter asked ʿAlqamah about how Ibn Masʿūd recited Sūrah al-Layl. ʿAlqamah responded that Ibn Masʿūd recited verse 3 as “wal-dhakari wal-unthā” (and by the male and the female) instead of “wa mā khalaqa al-dhakara wal-unthā” (and by that which created the male and the female), whereupon Abū al-Dardāʾ testified that he had learned the verse in the same way from the Prophet ﷺ. People tried their best to convince him to change his reading. Yet, he did not wish to do so. Understandably, the way the Prophet  ﷺ had taught him to recite the verse was something dear to him that he did not want to forsake. At the same time, Abū al-Dardāʾ did not approve of the staunch opposition shown by Ibn Masʿūd, saying, “We used to consider ʿAbdullāh [Ibn Masʿūd] to be soft-hearted, so why would he behave impulsively with the rulers?” Another companion, Abū Mūsā al-Ashʿarī, sought to compromise. When the ʿUthmānic muṣḥaf was sent to him, he said, “Whatever you find additional in my muṣḥaf, do not remove it, and whatever is missing, then write it.”

From the foregoing discussion it should be evident that none of these reports undermine the unanimous consensus of the companions upon the authenticity and correctness of the ʿUthmānic codex. Instead, it shows us that a few companions wished to continue reading as they had before and, quite understandably, did not wish to forsake the readings they had learned directly from the Prophet ﷺ.

Due to the success of the ʿUthmānic codex, all other readings eventually became practically extinct. They would continue to exist only as scattered narrations in the literary sources of the Islamic tradition, in the knowledge of specialists in qirāʾāt, and in some manuscript sources (see discussion on the Ṣanʿāʾ palimpsest below).

The committee of ʿUthmān did not transcribe only one codex. Rather, several copies were made and these were distributed to different lands within the Islamic empire. Anas ibn Mālik narrated, “When they had copied the manuscripts, ʿUthmān sent one muṣḥaf from those maṣāḥif that they had copied to every province.” This was directly related to the goal behind the entire project—to unify the Muslim ummah on one muṣḥaf in order to eliminate the disputes that had emerged between Muslims reciting according to different aḥruf. The success of the project was therefore contingent upon all Muslims following the official copies of the ʿUthmānic text sent to them.

A. The number of regional codices and their destinations

How many official codices were written and to which regions were they sent? The minimum number agreed upon by scholars is four; one remained in Madīnah while the rest were sent to Basra, Kūfah, and Shām (Greater Syria.) Abū ʿAmr al-Dānī (d. 444 AH) writes, “The majority of scholars are of the view that when ʿUthmān wrote the muṣḥaf, he made four copies and sent one to each region, so he sent one to Kūfah, one to Basra, one to Shām, and kept one with him.” A number of scholars, including al-Jaʿbarī (d. 732 AH) and Ibn al-Jazarī (d. 833 AH) also distinguished between the original codex that ʿUthmān kept for himself and the copy that ʿUthmān made for Madīnah. Other regions are mentioned by Abū Ḥātim al-Sijistānī (d. 255 AH), “When ʿUthmān compiled the Qur’an and wrote the maṣāḥif, he transcribed seven codices. And he sent one to Makkah, one to Shām, one to Yemen, one to Bahrain, one to Basra, one to Kūfah, and he kept one in Madīnah.” Ibn Kathīr (d. 773 AH) also states the same but substitutes Egypt for Bahrain. Seven is the number supported by Makkī ibn Abī Ṭālib as well. Ibn al-Jazarī himself prefers the opinion of eight, taking the seven mentioned by Abū Ḥātim and adding ʿUthmān’s personal codex. Al-Yaʿqūbī (d. 292 AH) does not include the personal codex but mentions all the aforementioned regions and adds al-Jazīrah, for a total of nine. These accounts may be reconciled by suggesting that the codices were dispatched in stages. The initial copies dispatched were indeed four including Basra, Kūfah, and Shām alongside the one kept in Madīnah, and these regions were prioritized because they were the regions where Muslims had specifically experienced conflict over their differing readings, as Ḥudhayfah had told ʿUthmān. Subsequently, the other Muslim lands mentioned acquired their own official copies.

Note that some later sources mention that ʿUthmān sent a reciter with each muṣḥaf. Al-Jaʿbarī (d. 732 AH) cites an account without an isnād which states: “ʿUthmān instructed Zayd ibn Thābit to recite according to the Madīnan codex, he sent ʿAbdullāh ibn al-Sāʾib with the Makkī codex, Mughīrah ibn Abī Shihāb with the Syrian codex, Abū ʿAbd al-Raḥmān al-Sulamī with the Kūfan codex, and ʿĀmir ibn ʿAbd Qays with the Basran codex.” However, this report is not found in earlier extant sources and it is not necessary to assume that this took place as there were already eminent companions of the Prophet ﷺ in each of those cities teaching the Qur’an.

B. The degree of similarity between the regional codices

How carefully and precisely did the scribes produce these regional codices? The answer is that manuscript evidence has proven that the copying process was extremely precise. We can see this with respect to spelling conventions (orthography). Sometimes the same word is spelled differently in different places in the Qur’an, and in some cases those orthographic idiosyncrasies are replicated across manuscripts. The contemporary researcher Marijn van Putten conducted an evaluation of the spelling of the word niʿmat (blessing) in the Qur’an. This word  occurs 23 times in the Qur’an and is sometimes spelled نعمت and sometimes نعمة, a variation that is commonly encountered with other feminine nouns as well (e.g., كلمة/كلمت ,رحمة/رحمت, etc.). One would expect such trivial spelling variations to exhibit no pattern, and yet examining 14 early Qur’anic manuscripts (predominantly from the first or second century hijrī), Van Putten demonstrated that there is remarkable consistency and agreement between manuscripts across نعمة/نعمت spellings. Thus, for example, 3:103, 5:11, 14:28, 14:34, etc. are always spelled نعمت in all the manuscripts, while 2:211, 5:7, 14:6, 16:18, etc. are always spelled نعمة. Van Putten writes:

Such orthographic idiosyncrasies allow us to show that the Qur’anic manuscripts go back to a single written archetype from which all of these documents were copied. If two manuscripts do not descend from copies of a single archetype, we would not expect the same spelling to occur in the exact same location time and time again. However, this is exactly what we find: highly idiosyncratic spellings occur again and again in the same spelling in the exact same location across all early Qur’anic manuscripts. Such variation can only be the result of precise written transmission.

Thus, these ʿUthmānic codices must have been produced through a very high fidelity process with meticulous review to allow for such patterns to be replicated in the subsequent generations of maṣāḥif. We have reports from the earliest Muslim community that take the spelling conventions in the ʿUthmānic muṣḥaf to be authoritative. Imam Mālik (d. 179 AH) was asked whether it was acceptable for people to write the maṣāḥif according to the spelling conventions of their time, and replied that they should write it according to the original spelling. Some scholars clarified this by noting that there is flexibility allowed in orthographic conventions on certain matters (such as soft vowels) but not others.

Note that one must be careful to distinguish between two different but related questions. The first is whether it is required to adhere to the rasm (orthography) of the ʿUthmānic text. The second is whether the rasm is tawqīfī (based on Prophetic instruction) or isṭilāḥī (based on the customary spelling practices and orthographic conventions adopted by the Prophet’s companions). Ghānim Qaddūrī al-Ḥamad notes that confusion between these two matters resulted in the claim that the rasm ʿUthmānī is tawqīfī. We do encounter this view mentioned by scholars as early as Abū al-Faḍl al-Rāzī (d. 454 AH). Nonetheless, the stronger view is that the rasm was isṭilāḥī; i.e., that the companions merely adopted the spelling practices to which they were accustomed, and this accounts for some degree of variation. 

If the rasm is isṭilāḥī, is it still a requirement to follow it? On the one hand, the majority of scholars said that following the rasm remains an obligation. On the other hand, Abū Bakr al-Bāqillānī (d. 403 AH) argued that following the rasm of the ʿUthmānic codex is not a requirement. Some scholars sought to adopt a middle position, stating that while it is not binding upon all people to follow the rasm, scholars of Islam should nevertheless endeavor to study and preserve knowledge of the ʿUthmānic rasm. Indeed, we do find Qur’anic manuscripts that did not adopt the rasm of the ʿUthmānic codex, such as the muṣḥaf of Ibn al-Bawwāb (d. 413 AH) written in 391 AH, among others. However, the strongest position is the majority position: adhering to the rasm is an obligation due to the consensus of the companions upon it and the intimate connection between the rasm and the permitted reading traditions (qirāʾāt).

C. The degree of difference between the regional codices

Notwithstanding the incredible degree of consistency between copies, we also know that, in very particular locations, the regional ʿUthmānic codices were not completely identical. The regional codices contained a small number of very slight differences that Muslim scholars devoted themselves to studying and enumerating. Some of these differences consisted of variations in spelling convention with no difference in pronunciation (e.g., الصرط or الصراط), while others related to different readings that existed among the companions of the Prophet ﷺ due to the different aḥruf. Narrations listing these differences were compiled by the famous reciters themselves, including Kitāb ikhtilāf maṣāḥif al-Shām wal-Ḥijāz wal-ʿIrāq by Ibn ʿĀmir (d. 118 AH), Kitāb ikhtilāf maṣāḥif  ahl al-Madīnah wa ahl al-Kūfah wa ahl al-Baṣrah by al-Kisāʾī (d. 189 AH), Kitāb Ikhtilāf al-Maṣāḥif by Khalaf ibn Hishām al-Bazzār (d. 229 AH). Another work entitled Kitāb ikhtilāf ahl al-Kūfah wal-Baṣrah wal-Shām fī al-maṣāḥif was compiled by the grammarian al-Farrāʾ (d. 207 AH). While the aforementioned works are no longer extant, the data concerning the variants has been detailed in extant works including Abū ʿUbayd al-Qāsim ibn Sallām’s (d. 224 AH) Faḍāʾil al-Qurʾān, Ibn Abī Dāwūd’s (d. 316 AH) Kitāb al-maṣāḥif, Abū Bakr al-Anbārī’s (d. 328 AH) Marsūm al-khaṭṭ, Abū al-ʿAbbās al-Mahdawī’s (d. 440 AH) Hijāʾ maṣāḥif al-amṣār, Abū ʿAmr al-Dānī’s (d. 444 AH) al-Muqniʿ, and al-Andarābī’s (d. 470 AH) al-Īḍāḥ fī al-qirāʾāt. As shall be discussed below, these works contain explicit statements affirming that these variants are all valid Qur’anic readings authentically transmitted from the Prophet Muhammad ﷺ. In other words, they already existed as Qur’anic readings prior to the transcription of the codices.

For the four codices unanimously agreed to be ʿUthmānic, al-Dānī lists a total of 36 points of difference, an astronomically small quantity in a text exceeding 320,000 letters. This fraction is a hundredth of a percent (0.01%), which means the regional codices are 99.99% identical. The companions of the Prophet ﷺ and the early reciters of the Qur’an were fully aware of these differences. Abū ʿUbayd reports that the differences between the Syrian codex and the ʿIrāqī codex were enumerated by the companion Abū al-Dardāʾ (d. 35 AH) himself. Abū ʿUbayd points out that the differences between the regional codices almost all amount to single-letter differences, such as an additional و (for example, و يقول in 5:53 in the Kūfan and Basran codices versus يقول in the Madīnan and Syrian codices), or the presence of an ا (for example, او ان in 40:26 in the Kūfan and Basran codices versus و ان in the Madīnan and Syrian codices). Aside from such single-letter differences, the only other difference among the four codices is the two-letter word هو in 57:24 (فإن الله هو الغني الحميد) in the Kūfan and Basran codices, which is not present in the Madīnan and Syrian codices. After enumerating these differences, Abū ʿUbayd states, “And as for all of these variants, then as I informed you, it is not permissible for anyone to deny or reject any of them, for they are all considered by us to be the Divine speech of Allah, and as such, the prayer of one who recites them is sound.”

These differences are not simply a matter of historical interest, as Muslims have continued to recite all of them in the form of the qirāʾāt. Thus, 5:53 is recited without the و by the Madīnan reciters (Nāfiʿ and Abū Jaʿfar), Makkan reciter (Ibn Kathīr), and Syrian reciter (Ibn ʿĀmir) as per their codices, while it is recited with the و by the Basran reciters (Abū ʿAmr and Yaʿqūb) and Kūfan reciters (Ḥamzah, ʿĀṣim, al-Kisāʾī, Khalaf) as per the Basran and Kūfan codices. That the variants mentioned in the codices are also present in the qirāʾāt confirms that these differences are attested in oral transmission from the Prophet ﷺ himself, a topic analyzed in greater detail below.

D. Why did the regional codices differ?

If the goal of the ʿUthmānic compilation was to unify the ummah upon one muṣḥaf and to eliminate the confusion and conflict surrounding reading with different aḥruf, then why would ʿUthmān and his committee include a small number of differences in the regional codices? And what was the origin of these differences? Muslim scholars have clarified that these variants related to authentically transmitted readings of the Qur’an from the Prophet ﷺ. We have already noted Abū ʿUbayd’s statement that all these variants are kalām Allāh (Divine speech). Likewise, Abū Bakr al-Udfuwī (d. 388 AH) and Makkī ibn Abī Ṭālib (d. 437 AH) mention that scholars affirmed that the variants between the regional codices belong to the differences of the seven aḥruf, and that the companions “relied on each reading (ḥarf) they heard the Messenger of Allah ﷺ recite in two ways, so in some codices they wrote one way and in other codices they wrote another, so that both ways would be available to the Muslims.” Therefore, all the regional variants are authentic readings from the Prophet (ﷺ). Ibn Idrīs (d. c. 400 AH) writes concerning the isolated variant of the Syrian codex which omits the initial و in 2:116:

And both readings are correct, and this [variant] was not a result of negligence from the companions (may Allah be pleased with them) as some ignoramuses may presume. But rather they wrote both ways deliberately because Allah revealed the verse both ways, so they wrote it with a waw in some of the codices and omitted it from the other, in order to serve as notification [of those two readings]. And this explanation applies to every case of addition or omission of letters in the codices.

The author of Kitāb al-Mabānī (approx. 425 AH) writes:

And the evidence that these variants between the codices were written correctly and with certainty, deliberately and intentionally, and to preserve both readings for the Muslims which the Messenger of Allah ﷺ recited on two different occasions at different times, and that those instances of addition, subtraction, or substitution in the text are not due to forgetfulness (sahw) of the transmitter nor omission by a negligent scribe, [the evidence] is that the readings are all sound and eloquent, and each variant is corroborated [in its linguistic acceptability] by decisive proof from the truth for its preponderance.

The author then proceeds to cite various linguistic explanations for the suitability of both readings in the case of the variants. Abū ʿAmr al-Dānī (d. 444 AH) also states that this was deliberately done by the committee of ʿUthmān in order to accommodate certain variants. Al-Dānī writes:

If a questioner were to ask concerning the reason that led to the variants comprising additional letters in the script of the maṣāḥif, I say: the reason for that according to us is as follows: When ʿUthmān compiled the Qur’an in the maṣāḥif, he wrote them according to one script based upon the dialect of the Quraysh, leaving out anything else that was not authentic or established, out of consideration for the ummah and caution for the people of this religion. And it was clear to him that these different readings [i.e., the variants between the ʿUthmānic codices] are from Allah and were revealed as such, and were heard from the Prophet ﷺ as such. He knew that collecting them together in one codex would not be possible except by repeating the same word twice, and evidently, to write it in that manner would entail confusing and altering the written script.  So he distributed them between the codices, so these additional letters are present in some and omitted in others so that the ummah could preserve them as they were revealed by Allah, and as they were heard from the Messenger of Allah ﷺ, so this is the reason why they are written differently between the codices.

In his comments in 2:285, al-Sakhāwī (d. 643 AH) also explains the spelling difference between manuscripts as a way of deliberately accommodating both readings. 

Note that some scholars opined that the motivation to include these particular readings within the text was because they were already being recited by Muslims in different regions, based on authentic oral transmission of the companions from the Prophet ﷺ, and could be easily accommodated with only slight modification. Abū Bakr al-Udfuwī (d. 388 AH) considers it possible that the scribes included these particular variants because they were already being taught to Muslims by the companions who lived in these regions. Al-Mahdawī (d. 440 AH) affirms the same view. He indicates that these variants (1) were divinely revealed and (2) corresponded to the oral transmissions already prevalent within each Muslim community. He writes, “And these variants were affirmed in the manuscripts that were written and distributed to the various Muslim provinces by ʿUthmān, along with the early Muslim community who agreed with him, due to their knowledge that they belong to that upon which the Qur’an was revealed. So they were affirmed so that every community could recite according to their transmission.”

Similarly, Al-Zarqānī explains that in general, the muṣḥaf was written without vowelization and diacritical marks to accommodate multiple readings, and where multiple readings could not be accommodated by the skeletal text, they were deliberately placed in different codices. He further argues that the underlying logic of placing different readings in different codices is to present them as equally valid; any other option, such as placing one reading in the margins and the other in the text, would be suboptimal because it creates the impression of a preferred reading. 

On the other hand, Ghānim Qaddūrī al-Ḥamad doubts that this was done with the specific goal of accommodating additional readings; rather, the goal of the committee was to write the Qur’an according to one ḥarf (i.e., one reading). However, in some cases due to the prevalence of two readings that were extremely similar, both equally well-established from the Prophet ﷺ, they were regarded as equivalent and basically taken as belonging to the same ḥarf. Likewise, Al-Aʿẓamī writes:

...these variations are inconsequential to the meaning of each verse and bear no alteration to the semantics whatsoever. But they cannot be attributed to carelessness. Zayd ibn Thābit, in each case finding both readings to be authentic and of equal status, retained them in different copies. The inclusion of both side by side would only have wrought confusion; alternatively, placing one of them in the margin would imply a lesser degree of authenticity. By placing them in different copies he accommodated them on equal terms. The modern approach to textual criticism requires that, when variations arise between two manuscripts of equal status, the editor cites one of the two in the core text while the deviations are consigned to footnotes. This method is unjust however, as it demotes the value of the second copy. Zayd’s scheme is much the fairer; by preparing multiple copies he sidesteps any implications that this or that reading is superior, giving each variant its just due.

In summary, then, Muslim scholars agree upon the authenticity of these variants between the regional codices, but they differ on the reason or motivation for their inclusion. Some scholars consider these variants a deliberate attempt to accommodate all or part of the variation of the aḥruf (relating to the earlier difference of opinion on that topic), while others consider these to be cases where readings were equally well established, which precluded giving priority to one over the other. One might imagine a third scenario in which such miniscule differences emerged during the transcription process because the reciter recalled another authentic qirāʾah which differed from the one used in the manuscript being copied. Thus, an existing oral precedent influenced the scribe during the writing process to include a slightly different reading (oral interference). The companions would not consider a small number of such slight differences to undermine the goal of unifying the Muslims (provided that they were all authentically established from the Prophet ﷺ), and hence during verification and review of the regional codices, both readings would be accepted. This will be evaluated in further detail below.

E. The regional codices and stemmatics

Examination of the data provided by Abū ʿAmr al-Dānī (d. 444 AH) concerning the differences between the regional codices of Syria, Madīnah, Kūfah and Baṣrah reveals that they form what is essentially a perfect stemma, or family tree showing the relationship between the codices. This observation was initially made by the German orientalist Theodor Nöldeke, and later updated and refined by the historian Michael Cook. Stemmatics analyzes variants of a text in order to determine which text was the ‘parent’ from which another copy was made. In order to understand the argument, let us first provide some basic examples to illustrate the general concept underlying the application of stemmatics. Assume that we find three copies of a book, which have small variations in the following sentence:

Book 1: “There was a gray car outside the house.”

Book 2: “There was a gray cat outside the house.”

Book 3: “There was a gray cat outside the house chasing a mouse.”

In this example, Books 2 and 3 agree on a cat outside, while Book 1 says it was a car. We know that Book 3 must have been copied from Book 2, but it would not have been copied from Book 1.

To further elucidate the matter, consider the similarities and differences between the following squares, imagining that each represents a codex (“S” for Syrian, “M” for Madīnan, “B” for Basran, and “K” for Kūfan).

Figure 1: Squares with variants forming a stemma.

Imagine that you were told that one of these squares was the original, and every time it was copied a few squares were changed from blue to yellow, or from yellow to blue. How would you determine which square was the original? Notice that the squares only differ with respect to the position of yellow squares, the coordinates of which are identified by the row numbers and column letters. Notice furthermore that S has two isolated variants shared by none of the others (G1 and D7) and K has three isolated variants shared by none of the others (I3, I7, and A5), while M and B have no isolated variants. Every yellow square on M is also present in S, and every yellow square in B is also present in K, while B and K share no common yellow squares with S and M. In other words, S never agrees with B and K against M, B never agrees with S against M or M against K, and K never agrees with M or S against B.

This is, in some loose sense, analogous to what Cook observes with the regional codices:

• Syrian muṣḥaf: 16 places where it is different from all the other three maṣāḥif, and 13 places where it agrees with the Madanī muṣḥaf against the Basran and Kūfan muṣḥaf. It never agrees with the Basran and Kūfan maṣāḥif against the Madanī muṣḥaf.
• Madanī muṣḥaf: No isolated readings (i.e., readings that are not found in any of the other maṣāḥif). It has only 13 places where it agrees with the Syrian muṣḥaf against the Basran and Kūfan maṣāḥif.
• Basran muṣḥaf: Never agrees with the Syrian muṣḥaf against the Madanī muṣḥaf. One isolated reading (23:85–89) is present which is not found in any other muṣḥaf. Cook agrees with a narration describing this as a later addition after the time of ʿUthmān since it doesn't fit the stemmas.
• Kūfan muṣḥaf: six isolated readings that are not found in any other muṣḥaf. Otherwise, it always agrees with the Basran muṣḥaf against the Madanī and Syrian maṣāḥif.

Based on these differences, Cook argues that we can reconstruct the relationships between the maṣāḥif, knowing that the Syrian muṣḥaf is more closely related to the Madanī muṣḥaf than to the Basran and Kūfan maṣāḥif, which means the Syrian muṣḥaf was copied from the Madanī muṣḥaf or vice versa. Similarly, the Madanī muṣḥaf is more closely related to the Basran muṣḥaf than it is to the Kūfan muṣḥaf, because the Kūfan muṣḥaf has six places that are isolated readings. So therefore the Basran muṣḥaf is between the Madanī muṣḥaf and the Kūfan muṣḥaf.

Based on this argument, Cook presents four possible stemmas:

Figure 2: Possible stemmas according to Michael Cook

Let us consider the merit of Cook’s argument from stemmatics. Does it provide convincing evidence that one codex must have been copied from another codex? The answer is affirmative, since otherwise the fact that they form a perfect stemma becomes a sheer coincidence (or an elaborate ruse). Had the codices been written independently of one another and ʿUthmān told the committee to put certain specific variants in the Syrian codex, and certain specific variants in the Madīnan codex, and so on for the Basran and Kūfan codices, then one would expect each codex to have its own isolated variants (mufradāt) or to share some variants with each of the other codices. There would be some cases where the Basran and Syrian codices agree against the Madīnan, or cases where the Kūfan and Madīnan agree against the Basran. Rather than forming a stemma, the picture would look something like this:

Figure 3: Squares with variants exhibiting no possible stemma.

However, where Cook errs in the matter is his claim that the existence of a stemma rules out the possibility of invention (i.e., the possibility that these variants were deliberately introduced into the codices). This is imprecise. The existence of a stemma rules out the possibility of synchronous invention, where codices are being written simultaneously and variants are being dispersed between them during the writing process. However, it does not exclude the possibility of sequential invention, where one codex is copied from another, one at a time, and during the copying process the scribe is deliberately instructed to make a few specific changes to the original at particular places in order to accommodate other readings. In the course of synchronous invention, there would be no generation of a stemma while in the course of sequential invention, it is impossible for the copying process not to produce a stemma. Note that a detailed evaluation of various hypothetical scenarios will be provided in the subsequent section including a refutation of the allegation of scribal errors.

The subject of stemmatics and the regional codices was recently revisited by Hythem Sidky, who extended the analysis to examine over fifty Qur’anic manuscripts for the very same regional variants reported in literary sources. Likewise, analytical studies of the regional variants in smaller subsets of existing Qur’anic manuscripts have also been performed by Muḥammad Muṣṭafā Al-Aʿẓamī, Ala Vahidnia, and Mohammad Said Mitwally Ibrahim Alrahawan. Many of these manuscripts have been dated to the earliest period of Islamic history. The results of these recent studies are remarkable. The very same textual variants reported in the Islamic literary sources are found distributed in the manuscripts according to the very same patterns. The manuscript evidence thus confirms the historicity of the ʿUthmānic codices and the accuracy of the traditional Islamic sources in documenting regional variants.

For instance, the Codex Parisino-petropolitanus (CPP) demonstrates agreement with the variants reported for the Syrian codex. CPP dates to the first century hijrī and was recovered from the Mosque of ʿAmr ibn al-ʿĀṣ in Fustat, Egypt. Today, most of the codex is located in the Bibliothèque nationale de France (national library of France), in Paris. The manuscript is written in ḥijāzī script.

Figure 4. CPP (Arabe 328a) showing verse 3:184 written according to the Syrian ʿUthmānic codex with wa bi-zubur as opposed to wa al-zubur. This is also exactly how the word is recited according to the qirāʾah of the Syrian reader Ibn ʿĀmir. Image from Bibliothèque nationale de France, annotation added.

The famous muṣḥaf known as the “Qur’an of ʿUthmān” in Tashkent, Uzbekistan (also known as the Samarqand codex), demonstrates concordance with the regional variants for the codex that ʿUthmān sent to Kufa. This manuscript dates to the late second century hijrī, i.e., roughly one century after the ʿUthmānic project. It was previously located in St. Petersburg, Russia, and in 1905 a facsimile copy was produced by S. Pisarev. It is written in kūfic script.

Figure 5. Pisarev facsimile from the Samarqand codex showing verse 36:35 written as ʿamilat (عملت) instead of ʿamilat-hu (عملته). This is also exactly how the word is recited by the Kūfan reciters Ḥamzah, al-Kisāʾī, Khalaf, and Shuʿbah’s transmission of ʿĀṣim. Image from CorpusCoranicum.

By similar analysis of regional variants, one may note for instance the Madīnan regional variants in the famous Topkapi Qur’an located in the Topkapi Palace museum in Istanbul, Turkey.

F. The allegation of scribal errors

With respect to the allegation that the ʿUthmānic codices contained scribal errors, there are two separate issues that need to be addressed: (1) the claim that variants between the regional codices are scribal errors; and (2) the claim that certain narrations from the companions indicate that there are scribal errors.

Typically in textual criticism, when textual variants are encountered in a manuscript tradition, the general assumption is that the original reading has been corrupted. However, there are a number of clear reasons why this assumption cannot be lazily applied to the textual variants of the ʿUthmānic regional codices. First, it ignores the existence of the qirāʾāt, the established reading traditions. The historical evidence inarguably establishes that different ways of reciting the Qur’an preceded the existence of the ʿUthmānic codex. If a textual variant could have already existed in the reading traditions, then its inclusion in the ʿUthmānic codex cannot be automatically attributed to scribal error. The codices were not produced exclusively through a copying process without any oral interference. We are informed in the literary sources that during the transcription process, one person would be designated to dictate and recite, while the other person would transcribe. For instance, ʿUthmān said, “Let Saʿīd dictate and Zayd transcribe.” If the scribe was only reading a text in front of him, there would be no role for the reciter. On the other hand, if the reciter was the only one reading the text and the scribe was writing exclusively based on what he heard from the reciter, then we would not obtain the pattern of orthographic idiosyncrasies discussed earlier which point to a single textual archetype from which other copies descend. The synchrony of oral transmission alongside textual transmission created a highly effective filtering mechanism whereby a textual variant could only be transmitted if it conformed to a pre-existing oral reading.

Secondly, the maximum number of isolated variants is approximately 15, which is found in the Syrian codex. If these were assumed to be transcription errors during the process of copying the Madīnan codex, this quantity is astronomically smaller than expected, and the remaining codices have even fewer variants. By way of comparison, in the text of the four gospels (roughly similar in length to the Qur’an), the Codex Sinaiticus and Codex Vaticanus contain 3036 textual variants, including entire omitted phrases and verses. If we ignore omissions of words, phrases and verses in the Codex Sinaiticus and pretend for a moment that it predominantly has single-letter textual variants like the Syrian codex, then we still end up with the astoundingly discrepant rate of one variant per 21 words in the gospels, and one variant per 5100 words in the Qur’an. The only scientifically plausible account is to judge the underlying mechanism generating textual variants in either scenario to be categorically different; for the Codex Sinaiticus the mechanism includes scribal errors and for the ʿUthmānic codices it is the reading traditions. 

Furthermore, examining the variants of the ʿUthmānic regional codices demonstrates that the variants are not randomly distributed through the codex. Of the 15 variants in the Syrian codex, seven are located exclusively in Sūrah al-Anʿām and Sūrah al-Aʿrāf. This requires us to assume extremely high fidelity copying throughout vast swathes of the Qur’an with sudden inexplicable sporadic lapses, including during the transcription phase as well as subsequent review and verification. Either way, it is far simpler to conclude that these were specific instances where oral readings influenced the inclusion of specific textual variants during the transcription process. This is particularly true if we accept evidence from the tradition concerning the deliberate inclusion of the alif in 23:87 and 23:89, which confirms that the early Muslim community was concerned about such minutiae with respect to correspondence between reading traditions and manuscripts. As such, we should consider minute alterations similar to 23:87 and 23:89 to be the result of scribes including existing oral readings rather than errors.

In the narration concerning the process of writing the ʿUthmānic codex we are informed about one word in the entire Qur’an on which the committee differed. Ibn Shihāb al-Zuhrī (d. 124 AH) reported, “They differed on that occasion concerning whether to write Al-Tābūt (التابوت) or Aa-Tābuh (التابوه). The Quraysh said ‘Al-Tābūt’ while Zayd said ‘Al-Tābūh.’ Their disagreement was brought to ʿUthmān, so he said, ‘Write it as Al-Tābūt, for it was revealed in the tongue of the Quraysh.’” Interestingly, among the textual variants of the regional codices we note that for 5:54, the Syrian and Madīnan codices have the unassimilated Qurayshī form of the word (يرتدد), which matches 2:217, while the Kūfan and Basran codices have the assimilated form as per the Tamīmī dialect (يرتد). Given that these are precisely the types of dialectical variations for which the committee was responsible to seek ʿUthmān’s approval, it is more rational to presume the deliberate inclusion of a known variant from the reading traditions than the committee overlooking the same exact word twice (either during writing the Syrian and Madīnan codices or during writing the Kūfan and Basran codices) in 5:54 but never in 2:217. 

Thirdly, the nature of the textual variants encountered in the regional codices strongly argues in favor of these variants representing pre-existing oral readings. Absent from the ʿUthmānic codices are any typical instances of transcription errors like parablepsis secondary to homoeoteleuton, haplography, or dittography. All of the textual variants of the regional codices result in syntactically and semantically viable readings. In cases where there is a perceptible impact on the meaning, both readings bring about mutually complementary and linguistically sound interpretations. What fortuitous circumstances would bring about the massively improbable coincidence of a scribal error adding or omitting a م in the one location where it is semantically appropriate to have both readings based on different preceding passages?

Fourthly, the reception of the ʿUthmānic codices was overwhelmingly successful and quickly became the basis upon which all qirāʾāt were recited and other maṣāḥif were transcribed. As previously mentioned, the early Muslim community was aware of these textual variants, and indeed catalogued them extensively as evidenced by the numerous early works and narrations on the subject. By all accounts, the Prophet’s companions and their succeeding generations had no qualms reciting and teaching the Qur’an according to these codices with regional variants. Not a single narration records hesitation over reciting a letter determined to be a textual variant. None of these sources declare these particular variants between the regional codices to be the result of scribal errors in Qur’anic manuscripts, even though they did discuss the concept of scribal errors extensively, as we shall observe in the next section.

Abū Bakr al-Udfuwī (d. 388 AH) is one of the early scholars to repudiate the notion that the regional variants could be the result of scribal errors. Al-Udfuwī writes:

The understanding of occurrences of addition or subtraction in the maṣāḥif is that they are written according to the reading of those companions sent to each of the provinces. And what indicates this is that the Qurrāʾ (famous reciters) attribute their readings to the senior companion of their region. And these variants, in their addition or omission, were recited during the time of the Prophet Muhammad ﷺ. And were that not the case, they could not be written in some codices and omitted from others. And it is not possible to imagine that they could be errors from the scribe, for verily Allah has preserved it (i.e., the Qur’an). And the proof of that is that when ʿAlī began his reign as the caliph, he did not change anything from it but rather endorsed the action of ʿUthmān. And they used to dislike even dotting the maṣāḥif out of fear of adding to it. So how could they add letters (or variants) and have such additions approved... [al-Udfuwī then gives examples illustrating the early Muslims’ abundant caution in adding any markings of any kind, including chapter markings, verse markings, and consonantal diacritical markings.] So it is not possible with such caution that these letters could arise except via explicit proof (naṣṣ) or knowledge, and they could not occur by the error (wahm) of scribes.

One may observe that al-Udfuwī mentions both the first and fourth points of consideration identified above. That is, he mentions the existence of established reading traditions with these regional variants, and he mentions the community’s reception of these variants, particularly the example of ʿAlī ibn Abī Ṭālib. Moreover, he contrasts the reception of these regional variants with the early Muslim community’s reticent attitude towards even the most trivial manipulations of the text.

Fifthly, one may address the question of purpose. Why include only a small number of seemingly trivial variants? There are a number of historically plausible scenarios that would explain the small number of textual variants that we see. Not all of these scenarios require that the matter had been discussed and deliberated upon by ʿUthmān and the committee prior to the initiation of the compilation project. The stated goal of the project was to reduce the differing and conflict over readings taking place among Muslims by uniting them upon one muṣḥaf. It is plausible that during the transcription process, the participation of different members from the extended committee of 12 members would have influenced the decision about the inclusion of a minute number of variants in some codices as opposed to others—a decision which would have ultimately been reviewed and approved by ʿUthmān prior to sending out the codices. One aspect of the extensive review and verification process of very minor changes is highlighted in the report of Hānī al-Barbarī. He reports:

I was with ʿUthmān when the committee was reviewing the maṣāḥif. So he sent me to Ubayy ibn Kaʿb with the scapula of a sheep upon which was written (لم يتسن) and (لا تبديل للخلق) and (فأمهل الكافرين). So Ubayy asked for something with which to write and revised the spellings, such that he wrote (لخلق) removing one of the two ل, and he wrote (فمهل), and he wrote (لم يتسنه) adding a (ه).

The above examples of single-letter changes were carefully considered and discussed, and these particular examples do not show up as textual variants among the regional codices. However, in the above example, the role of Ubayy ibn Kaʿb is limited to correcting some spellings on specifically requested words. Given that the people of Syria were more familiar with the reading of Ubayy ibn Kaʿb, it is not unlikely that ʿUthmān would ensure that the Syrian codex received additional participation from Ubayy ibn Kaʿb.  

Recall that scholars like al-Udfuwī and al-Mahdawī suggested that the regional variants were intended to accommodate readings of a particular region. Recall also that some scholars suggested that the ʿUthmānic committee initially included only four core members before expanding to include a total of 12 members. This also makes sense of the differing narrations concerning Ṣaʿīd ibn al-ʿĀṣ dictating versus Ubayy ibn Kaʿb dictating and Ṣaʿīd reviewing. The involvement of companions like Anas ibn Mālik and Ubayy ibn Kaʿb in the Madīnan codex, and particularly Ubayy ibn Kaʿb in the Syrian codex, suggests a plausible scenario for the minor variation encountered in the regional codices.

The sixth point, though somewhat technical, appears to empirically exclude the possibility of attributing the regional variants to scribal error. It draws upon the previous discussion of stemmatics and combines it with a consideration related to the mutashābihāt (passages in the Qur’an that resemble one another). Notice first that several of the textual variants appear to recapitulate passages found elsewhere in the Qur’an. Thus, the Syrian/Madīnan variant for 5:54 matches 2:217, the Syrian/Madīnan/Basran variant for 46:15 matches 29:8, the Kūfan/Basran 57:24 matches 60:6, the Madīnan/Basran/Kūfan 40:21 matches 35:44 and 30:9, the Syrian 8:67 matches 9:113 while Madīnan/Basran/Kūfan 8:67 matches 3:161, the Syrian 6:32 matches 12:109 and 16:30, the Syrian 3:184 matches 35:25, the Kūfan/Basran 43:71 matches 41:31, the Syrian 7:141 matches 14:6 while the Madīnan/Basran/Kūfan 7:141 matches 2:50 and 20:80, and the Syrian/Madīnan/Basran 6:63 matches 10:22. This predilection of textual variants for mutashābihāt verses appears to have gone unnoticed in prior studies of these variants. Yet, it can be no accident that mutashābihāt passages are so highly represented among the textual variants.

What follows is an argument by way of reductio ad absurdum. Those who consider these to be scribal errors would have to concede that they are not instances of misreading the text being copied but rather a case of memory interference known as assimilation of parallels which would arise from recollection of a similar passage. But on that assumption, the error would dissolve the original difference between the two passages. Thus, if the Qur’an originally contains a passage that says A1 and another similar passage that says A2, a memory-based error would result in two passages that say A1 or two passages that say A2. Overall, memory-based errors homogenize differences between mutashābihāt passages, resulting in fewer and fewer differences. Projecting backwards, the direction of heterogeneity allows one to identify the original text. We may exclude those examples for which both regional variants match other Qur’anic passages, since it is not possible to identify directionality in such cases. Memory-based errors should accumulate as one descends from parent to child in the stemma. The direction of increasing homogeneity should also be the direction of transcription within the stemma. Yet, this leads to conflicting directions violating the stemma (see diagram), demonstrating that these regional variants cannot be attributed to error.

Figure 6. The presumed directionality of copying from parent codex to descendent codex based on the variants arising through the process of assimilation of parallels. Given that the inferred directions are mutually contradictory and that these contradict the relationships deduced from stemmatics, it can be reasonably concluded that the regional variants were deliberately included to conform to an existing reading transmitted from the Prophet ﷺ by a companion and thus do not represent errors.

In other words, on the claim of scribal errors, either the existence of the stemma becomes a massive coincidence or the predilection of regional variants for mutashābihāt becomes a massive coincidence. Notice of course that the Qur’an contains multitudes of mutashābihāt which feature no variants in the regional codices, including some which are quite extensive (e.g., 2:57–59 and 7:160–162). These would be the passages most ‘vulnerable’ to assimilation of parallels, and yet they are free of textual variants. On the other hand, the mutashābihāt which feature textual variants are comparatively much easier to sort out and generally the least confusing to huffāẓ (those who have memorized the Qur’an). All of this lends further credence to the argument that these regional variants are deliberate inclusions of pre-existing reading traditions and nothing in the ʿUthmānic codices is attributable to scribal error.

We may now turn our attention to a particular set of narrations recorded in the Islamic tradition which appear to attribute scribal errors to the ʿUthmānic codex. Scholars have analyzed these narrations with respect to their authenticity as well as their potential implications and interpretations.

The first matter of concern is a statement of ʿUthmān ibn ʿAffān himself. A number of narrations state that when the ʿUthmānic codex was complete, he examined it and said, “You have done an excellent and outstanding job. I see that it contains some anomalies (laḥn), however the Arabs will recite it in the correct manner.” 

First, it should be noted that the chains of transmission for this narration are disconnected and consequently it was deemed inauthentic by many scholars. Abū Bakr ibn al-Anbārī (d. 328 AH) writes:

The narrations from ʿUthmān concerning that cannot be used as evidence because they are disconnected and do not trace back to him. And logically it is not possible that ʿUthmān, the leader of the Muslim ummah and the ruler of the people and their role model, would unify the people upon the muṣḥaf that will serve as the exemplar, and after noticing that there are discrepancies or mistakes in the script he would choose not to correct them.

Al-Anbārī continues by pointing out that this conflicts with the narration of Hānī al-Barbarī, mentioned above, which shows that ʿUthmān went to the trouble of sending specific words to Ubayy ibn Kaʿb to ascertain their precise spellings. Makkī ibn Abī Ṭālib simply points out that the companions of the Prophet ﷺ agreed on the authenticity of what is between the two covers of the muṣḥaf, and it is impossible that they would have agreed upon error. Abū ʿAmr al-Dānī (d. 444 AH), like al-Anbārī, notes that this narration is not authentic and that its apparent meaning negates the possibility that ʿUthmān could have said it. ʿUthmān and the noble companions completed an exhaustive and extensive project to compile the muṣḥaf. To simply just leave mistakes in the text after that would render the entire project pointless. He concludes that it is not possible for anyone to say this nor permissible for anyone to believe it. Similar arguments were presented by Ibn Taymiyyah (d. 728 AH). 

Moreover, scholars pointed out that even if one were to accept the narration as authentic, the word laḥn does not necessarily mean “mistake.” It can also refer to dialectical differences, among a variety of other meanings.Abū al-Ḥusayn ibn al-Munādī (d. 336 AH) was of the opinion that laḥn refers to spelling variants in the rasm where the spelling appears slightly different from the expected pronunciation, which should not be an issue since the recitation of the Qur’an is taught orally. Al-Dānī (d. 444 AH) himself provided a similar interpretation and offered examples of Qur’anic words that are written with an extra alif (لأاذبحنه ), waw (سأوريكم ), or yā (نبأى). It is possible for the uninitiated to mispronounce such words if they have not learned from a teacher how to recite the Qur’an, again underscoring the importance of the oral transmission. There are much more striking examples in English such as words like neighbor, queue, knife, laughter, and asthma, where the pronunciation differs considerably from what would be expected given the spelling. This being the case, the interpretation of the statement of ʿUthmān poses no challenges.

A few statements regarding scribal errors have also been related from ʿAbdullāh ibn ʿAbbās, which have been studied in detail by many scholars. The most famous of these statements concerns the word despair (ييأس) in verse 13:31: “The scribe wrote it that way while he was sleepy.” With respect to the authenticity of this statement, the discussion revolves around the transmitter Jarīr ibn Ḥāzim, who sometimes erred in what he relayed from memory. Abū ʿUbayd recorded the narration of Ibn ʿAbbās without including this statement, instead including only the part of the narration stating that Ibn ʿAbbās read it differently. Imam Aḥmad ibn Ḥanbal appears to have accepted the authenticity of the narration, while it was rejected by others. Ibn Ḥajar al-ʿAsqalānī (d. 852 AH), however, was not convinced of the report’s inauthenticity, and he stated that it was more appropriate to find a suitable interpretation for the report. Ibn Taymiyyah (d. 728 AH) explained that some of the early Muslims denied certain readings with which they were unfamiliar, mentioning 13:31 as an example. He then states, “And this is a mistake known through unanimous consensus and mass (mutawātir) transmission. In spite of that, due to the fact that such a reading had not been established for them by mass transmission, they did not disbelieve [by denying it] even though it would constitute disbelief for someone for whom proof of its mass transmission has been established.” In other words, it was only during that initial period in Islamic history, before the widespread recognition of certain readings  authentically transmitted from the Prophet ﷺ, that some people could be excused for not knowing them. This is similar to how certain Islamic rulings were not known to a select few companions, but later became widely transmitted.

Note that Ibn ʿAbbās’s reading of this verse was also shared by ʿAlī ibn Abī Ṭālib, ʿIkrimah, Mujāhid, Ibn Abī Mulaykah, ʿAlī ibn Badīmah, Shahr ibn Ḥawshab, and others. This establishes that the reading with which Ibn ʿAbbās was most familiar was also widespread. The companions lived in an era where people commonly wrote their own personal copies of the Qur’an, and so the companions would regularly encounter and correct instances of human error in recitation and in writing. The use of the word yayʾas was likely unfamiliar to some of them since it has been described as belonging to, among others, the dialect of Hawāzin, as mentioned by al-Qāsim ibn Maʿan (d. 175 AH). There is no difficulty in affirming that Ibn ʿAbbās, due to his unfamiliarity with this reading of 13:31, assumed it was only found in a copy in which the scribe had erred. Once it became clear to him that this was a reading taught by the Prophet ﷺ, he accepted it. This is evident from the fact that Ibn ʿAbbās himself explained the word yayʾas found in the ʿUthmānic codex to mean yaʿlam (knowledge), as reported in an authentic transmission through the tafsīr of ʿAlī ibn Abī Ṭalḥah. These same principles can be used to account for other narrations from Ibn ʿAbbās or other companions that seem to imply disagreement with the reading found in the ʿUthmānic codex.

For instance, ʿUrwah ibn al-Zubayr asked ʿĀʾishah about three verses—4:162, 5:69 and 20:63—which each contain a word in an unexpected grammatical tense, to which she replied, “O my nephew, this is the doing of the scribes who erred in the transcription.” However, scholars of tafsīr noted that all of these examples have grammatical explanations in their support, either because they are justified by the dialect of other tribes or because the syntax of the verse supports a reading that justifies the shift in grammatical tense from that which is expected to that which is observed. Thus, the statement of ʿĀʾishah is either inauthentic, misunderstood, or relates to a personal misjudgment due to lack of familiarity with the reading found in the ʿUthmānic codex.

Abū Layth al-Samarqandī (d. 375 AH) was one of those who rejected such narrations because they imply a logical impossibility, namely that the companions of the Prophet ﷺ would be aware of such errors in the ʿUthmānic codex and simply not bother to make any corrections. He wrote:

And some ignoramuses (juhhāl) state this is an error from the scribe when he wrote the exemplar codex [he then mentions their citation of the aforementioned statements of ʿUthmān and ʿAʾishah]. However, scholars consider this to be farfetched and the narrations are not established from ʿUthmān nor from ʿĀʾishah, may Allah be pleased with them both, because the companions of the Messenger ﷺ were the protectors of the religion and the role models in law and rulings, so it is inconceivable that they would leave in the book of Allah errors that others would correct. 

Similar arguments have been made by Abū Isḥāq al-Zajjāj (d. 311 AH) and al-Zamakhsharī (d. 538 AH), with the latter pointing out that only someone unfamiliar with the works of Arabic grammar and their schools of thought could accept the proposition that these are scribal errors.

In commenting on verse 4:162, Ibn Jarīr al-Ṭabarī persuasively demonstrated the impossibility of ascribing a scribal error to the reading found in the ʿUthmānic codex by observing that the reading found in the codex of Ubayy is identical to that found in the ʿUthmānic codex. He comments, “So had that been a scribal error, it would necessitate that all the others codices should read differently than ours in which the scribe erred. And the fact that our muṣḥaf and the muṣḥaf of Ubayy agree on that [reading] demonstrates that what is found in our muṣḥaf is correct and not an error.” Note that this is an early example of a type of ‘proto-stemmatic’ consideration, the likes of which modern historians utilize to construct an archetype of a text and identify the original manuscript or reading. Al-Ṭabarī then continues with an additional historical argument: “Furthermore, had that been a scribal error, the companions of the Prophet ﷺ, from whom the Qur’an is taken, would not have continued to teach Muslims according to an error, but rather would have corrected it with their tongues (i.e., recitation) and would have instructed the ummah according to the correct reading. And in the very fact that Muslims have unanimously transmitted that reading exactly as it is written in the [ʿUthmānic] codex is the clearest evidence of its veracity and correctness.”

Based on the foregoing considerations, the strongest position is that this statement from ʿĀʾishah simply indicates an instance where a companion was unfamiliar with the reading found in the ʿUthmānic codex, just like the statements from Ibn ʿAbbās discussed earlier. Mansour Hamad Eidi notes that this should not be surprising, as it is only logical that certain qirāʾāt would be unfamiliar to some companions. It is difficult for a person today to master and remember the various qirāʾāt, let alone for a person in the earliest period of Islamic history when writing was less common and individuals had their own personal codices. This is precisely what illustrates the immeasurable benefit brought about by the ʿUthmānic codex. It unified all Muslims across the ummah upon the same text and provided them all with a clear reference point for the recitation of the Qur’an.

Earlier we discussed how Qur’anic manuscripts corroborate very precise details mentioned in the Islamic traditional sources, such as the regional variants  distributed in manuscripts in the same patterns described by the traditions. Perhaps one of the most fascinating case studies in the correlation between traditional sources and manuscripts is the Ṣanʿāʾ palimpsest. A palimpsest is a manuscript where the original writing (termed the lower text) was erased (by scraping or washing) and written over (the upper text).  The residue of the ink from the erased text, however, induces a subtle color change in the parchment. With careful examination and with the aid of ultraviolet light, the writing of the lower text can be identified.

In 1973, during the restoration of the roof of the Ṣanʿāʾ mosque in Yemen, a storage area between the roof and ceiling was found containing thousands of ancient documents, including the Ṣanʿāʾ palimpsest. This palimpsest represents a very early manuscript, radiocarbon-dated to before 51 AH/671 CE with 99% probability and before 41 AH/661 CE with 95% probability. In other words, this manuscript dates very closely to the time of the ʿUthmānic compilation (25 AH/645 CE). The Ṣanʿāʾ palimpsest lower text was a copy of the Qur’an that was erased and written over, and the time interval between the lower and upper text is very short (approximately 10–50 years). The upper text is consistent with the ʿUthmānic codex. The lower text likewise is consistent with the ʿUthmānic codex in the content of its verses and their sequences; it does not contain any verses absent from the ʿUthmānic codex nor does it feature verses in an order contrary to the ʿUthmānic codex. However, the lower text has some important differences with respect to chapter order and variant readings. In these differences, the lower text demonstrates an incredible degree of consistency with what Islamic traditional sources have described for the personal codices of companions like ʿAbdullāh ibn Masʿūd and Ubayy ibn Kaʿb. 

With respect to chapter order for instance, the lower text of the Ṣanʿāʾ palimpsest has Sūrah Yūnus after Sūrah al-Māʾidah, Sūrah al-Kahf after Sūrah Yūsuf, Sūrah al-Jumuʿah after Sūrah al-Munāfiqūn, and Sūrah al-Qiyāmah after Sūrah al-Insān, identical to the sequences reported for Ibn Masʿūd and Ubayy ibn Kaʿb. With respect to variant readings, there are numerous parallels as well. For instance, in 19:23, Ubayy ibn Kaʿb is reported to have read falammā ajāʾahā instead of fa-ajāʾahā; we find the lower text here is consistent with the reading of Ubayy. In 20:128, Ibn Masʿūd recited a-wa-lam yahdī instead of a-fa-lam yahdī; we find the lower text is consistent with the reading of Ibn Masʿūd. In 63:10, Ibn Masʿūd and Ubayy reportedly read fa-ataṣaddaqu instead of fa-aṣ-ṣaddaqa; the lower text is consistent with their reading. These readings are designated as shādhdh (anomalous) readings by Muslim scholars since they differ from the ʿUthmānic codex and are no longer recited. A discussion of such shādhdh readings has already been provided in our previous article.

In all of its differences, the lower text exhibits precisely the same kind of minor variant readings that Islamic sources state existed in the early period of Islam under the license of reciting the Qur’an according to seven aḥruf. There were many valid ways of reciting the Qur’an which differed in the small ways described in the traditions and witnessed in the manuscripts. When ʿUthmān compiled the muṣḥaf, it was not necessary for him to include every reading. This is the reason we find some non-ʿUthmānic readings in the Ṣanʿāʾ Palimpsest. This shows that certain readings were widespread and remained popular shortly after the standardization of the text until they eventually became extinct.

Prior to the ʿUthmānic codex, there was no standardization, and Muslims recited according to the reading they had been taught. After the ʿUthmānic compilation, readers had to adhere precisely to the ʿUthmānic codex, and earlier manuscripts were erased and re-written or otherwise destroyed. Previously, certain orientalists and revisionist Western scholars such as John Wansbrough claimed that the companion codices were fictions conjured up by Muslims, or that the traditional Muslim narrative concerning the ʿUthmānic codex was wrong. The arrival of such manuscript evidence confirming even the most minute details of the Islamic traditions surrounding the ʿUthmānic compilation and the companion codices has conclusively falsified such claims.

At a time when Muslims were small in number and intensely persecuted in Makkah, and the future of Islam seemed uncertain, Allah revealed Sūrah al-Ḥijr containing therein a promise: “Verily, We have revealed the Qur’an and We will surely protect it” (Qur’an 15:9). The Prophet ﷺ used to hasten to repeat the words of revelation lest they be forgotten, but Allah assured him, “It is certainly upon Us to enable its gathering and its recitation” (Qur’an 75:17). 

By any human calculation, this promise of preservation would have seemed impossible to fulfill at the time. This was especially the case when numerous readings proliferated in Madīnah, a development necessitated by the diversity of Arabic dialects as well as differences in linguistic capabilities and literacy within the early Muslim community. The more one ponders the incredible diversity of readings at that time and the geographical span of the early Muslim empire, the more it becomes incredible to fathom how the entire ummah came to be united upon a single text of the Qur’an, which Muslims have continued to recite in every age and every region. All of this took place in an era that preceded the printing press, computers, and the internet. 

Fakhr al-Dīn al-Rāzī (d. 606 AH) points out that the phenomenon of the Qur’an’s preservation is unparalleled. Every other ancient scripture has been subject to some degree of textual alteration and corruption. In spite of its foes who would spare no effort to attempt to undermine it, the Qur’an has stood the test of time, which serves as one of “the greatest miracles.” Moreover, this Divine promise constitutes a prophecy concerning the unseen, which is yet another miracle, and one that had already stood the test of time for six centuries when al-Rāzī made his remarks. Today, after 14 centuries, the Divine promise remains fulfilled and even more vividly attested.

The accumulation of manuscript evidence corroborating precise details preserved in the literary sources of the Islamic tradition has further testified to a fact Muslims have always known, that the promise of Allah has been fulfilled. As Abū Shāmah al-Maqdisī (d. 665 AH) comments, the muṣḥaf contains precisely what was revealed by Allah, taught by the Prophet Muhammad ﷺ, and written by the companions. The preservation of the Qur’an is a historical and lived reality, experienced by every Muslim who continues to recite the exact same text today that was assembled by the companions of the Prophet under the supervision of the Caliph ʿUthmān, as it was recited to them by the Prophet Muhammad ﷺ, and as it was revealed through the Angel Jibrīl by Allah, Glorified and Exalted.

ʿUthmān said, “During the time of the Prophet, chapters with numerous verses would be revealed and so whenever something was revealed, he would summon one of the scribes and instruct them, ‘Place these verses in the sūrah which mentions this or that.’ And when a single verse was revealed he would say, ‘Place this verse in the sūrah which mentions this or that.’ Al-Anfāl was one of the first sūrahs revealed in Madīnah, and al-Barāʾah was one of the final portions of the Qur’an revealed, and their stories were similar so I thought that it was part of it [i.e., that they belonged to the same chapter]. Then the Messenger of Allah ﷺ passed away before this was clarified, so it is for this reason that I placed them together and did not write the basmalah between them and placed them with the long sūrahs.” Jāmiʿ al-Tirmidhī, no. 3086; Musnad Aḥmad, nos. 399 and 499.

to their Ancestral ʿUthmānī Codices,” al-Burhān: Journal of Qurʾān and Sunnah Studies 5, no. 2 (2021): 1–47.