Textual History of the Qur'an

This article examines the transmission history of the Quran. The perfect preservation of the Quran is an article of faith for most schools and sects of Islam and figures highly in the beliefs of Muslims around the divine nature of their religion. The orthodox Islamic scholars argue that the Qur'an today is identical to that received by Prophet Muhammad. This contention however is challenged both by parts of the Islamic tradition itself and the findings of modern scholarship.

Before Caliph Uthman standardised the Quranic consonantal text (QCT) around 650 CE, large numbers of variants later documented by Muslim scholars were read by various companions of Muhammad, often differing in whole words and phrases. Academic experts have found material support for such reports in some of the oldest Quran manuscripts. There are also hadith reports that substantial numbers of verses had already been lost. The fear of permanently losing verses is said to have motivated the initial collection of the Qur'an under Caliph Abu Bakr. Notwithstanding a number of scribal errors during the initial copying process, Uthman was essentially successful in stabilising the QCT, or rasm. However, due to limitations in the early stage of Arabic orthography in use at that time, a wide variety of oral readings (qira'at) within this standardised rasm was possible and continued to occur until the oral readings too were stabilised over centuries and orthography developed to more fully document them. Tens of thousands of variants are attributed to readers of the first couple of centuries, which include Muhammad's companions and early reciters, within and outside the standard rasm, besides the ten canonical readings and their transmitters. The vast majority of recitation and printed Qurans in use today are based on the transmission of Hafs from the reading of 'Asim.

Introduction

The Qur'an is claimed to be a revelation from Allah to Prophet Muhammad through the Angel Gabriel. It was revealed to Muhammad in stages, taking 23 years to reach its completion.

The textual and oral transmission history of the Quran are interconnected. As discussed in this article, standardisation first occured in the written consonantal skeleton. This acted as a constraint on the variant oral readings (qira'at), which too eventually became standardised and written down in stages. This was accomplished by Muslims over a period of many centuries. Muslims would argue that the Qur'an was preserved by Allah as he had promised.^[2]

Earliest Collection of the Quran

Unfortunately, at no time during Muhammad's life did he ever order the Quran to be compiled into a single book.^[3]. Many Qurra' (“reciters”) had died before the Quran had been compiled in written form^[4], so Abu Bakr asked Zaid bin Thabit to collect it into a book.^[5] Zaid was reluctant, for Muhammad had never ordered such an action to be taken.^[3] He stated, "I started looking for the Qur'an and collecting it from (what was written on) palm-leaf stalks, thin white stones, and also from the men who knew it by heart, till I found the last [two] verse[s] of Surat at-Tauba (repentance) with Abi Khuzaima al-Ansari, and I did not find it with anybody other than him."^[6]

Sahih Bukhari contains contradictory material around the collection and preservation of the text of the Qur'an:

Muhammad's Own Recollection of the Verses

While even today there are many memorizers (huffaz) of the complete Qur'an, the earliest Muslims did not have the benefit of choosing a standard qira'at (reading) and standard written Qur'an complete with diacritics in book form to help them or their teachers in the learning process.

The Qur'an itself records that Muhammad himself had forgotten portions of the Qur'an^[7][8] Muhammad may also have had a somewhat flexible approach to variant readings, typical of oral performance traditions - see the Qira'at section later in this article.

Hadith also exist to the point that Muhammad himself forgot parts of the Qur'an and needed his followers to remind him:

In the below hadith it seems Muhammad's companions also forgot passages of the Qur'an:

Companion Codices and the Uthmanic Standard

Caliph Uthman Standardises the Rasm and Burns the Other Texts

Multiple sources report that the third caliph Uthman was concerned because there were clear differences in the recitation of the Qur'an among the people of the Sham (modern day Israel/Palestine, Jordan, Lebanon and Syria) and the people of Iraq. The differences were so great Uthman and his companions feared future dispute about the true Qur'an and its contents. So Uthman asked Hafsa for her copy so that a committee could write a single version of the rasm (an early stage of Arabic orthography, often called the Qur'anic consonantal text (QCT), which lacked diacritics such as short vowel signs and with scarce use of dotting to distinguish certain consonants). Uthman then sent out his official Quranic codex to a small number of important cities and ordered that all other copies and fragments be burned. This occurred around 650 CE. During the prior 20 years since Muhammad's death, and for some time afterwards, thousands of variants read by the companions which often did not fit this rasm were in circulation, as documented in hadiths and works such as Ibn Abi Dawud's Kitab al Masahif.^[9]

Narrated Anas bin Malik:

Disagreements on the Qur'an

Sahih hadiths and tafsirs record a great many variations of recitation among the sahaba (companions) for many different verses.

• An example involving multiple companions is found in two consecutive verses. Bukhari and Muslim record that ibn 'Abbas and Sa'id b. Jubair respectively added the word "servicable" to describe the boats in Quran 18:79, and added the words, "the boy was an unbeliever" to Quran 18:80.^[10][11] Al Tabari's tafsir for these verses include reports that ibn Mas'ud too narrated them this way.

• Another interesting example, recorded in a sahih hadith that appears in many collections, concerns a variant reading of verse Quran 2:238. It was given by Aisha, according to whom in this verse it says 'the middle prayer and the Asr Prayer', as she heard Muhammad reciting it.^[12] As such, her version of the verse combines what was, according to another hadith, the pre-abrogated version of the verse, which mentions the asr prayer, and post-abrogation version, which says the middle prayer.^[13]. What purpose would be served by an abrogation to replace one specific word with another that more ambiguously indicates the same prayer (according to most scholars) is a mystery.

• An interesting example of companions still reciting a verse allowing a practice that was supposed to have been abrogated concerns the controversial topic of nikaah al-mut'ah (temporary marriage), which Muhammad banned in his final years according to sahih hadiths (some use other hadiths to argue that instead Umar did so after his death), and verse 4:24, which says:

...Then as to those whom you profit by, give them their dowries as appointed; and there is no blame on you about what you mutually agree after what is appointed; surely Allah is Knowing, Wise.

Quran 4:24

Al-Tabari's tafsir for verse 4:24 includes narrations saying that ibn 'Abbas, ubayy ibn Ka'b, and Sa'id ibn Jubayr (others too in other tafsirs) included the words 'until a prescribed period' ('ila ajal musamma') after the words 'whom you profit by'.

Some Muslim scholars sought to explain the reported differences in the mushafs (codices) of the companions as merely being their own exegetical glosses. Such an explanation may be possible in some instances, but certainly not in others such as when pronouns or grammatical forms are changed or words are reported to have been simply omitted, for example in Quran 112:1 where Ibn Mas'ud and Ubayy omitted the word "Say" (qul)^[14][15], or ibn Mas'ud's omission of the entire verse Quran 94:6.^[16][17] Another explanation was that these were variations in the revelation ("ahruf", discussed in a section below), which encounters some of the same problems, as well as the issue of their sheer quantity and the difficulty of explaining the purpose of the less clear or specific wordings of the same sentences in the Uthmanic Qur'an.

Many other examples of such variations among the sahaba are discussed in another online article^[18] and in the next few sections below.

Qur'an of Ibn Mas'ud

It was widely reported that Abdullah Ibn Mas'ud's Qur'anic text omitted surah al-Fatiha and the mu'awwithatayni (surahs 113 and 114).^[19] Ibn Mas'ud's denial that the last two surahs were part of the Qur'an is also recorded in Sahih Bukhari.^[20]

When we come to the rest of the Qur'an, we find that there were numerous differences of reading between the texts of Zaid and Ibn Mas'ud. The records in Ibn Abu Dawud's Kitab al-Masahif fill up no less than nineteen pages^[21] and, from all the sources available, one can trace no less than 101 variants in the Suratul-Baqarah alone.^[22]

The following are just a few of the differences in illustration of the nature of the variations between the texts:

• Sahih Bukhari 6:60:468 and Sahih Muslim 4:1799 both record that Ibn Mas'ud's followers were adamant that he and Muhammad had read Quran 92:3 with the words, By the male and the female. rather than And by Him Who created male and female.

• Quran 2:275 begins with the words Allathiina yaakuluunar-ribaa laa yaquumuuna, meaning "those who devour usury will not stand". Ibn Mas'ud's text had the same introduction but after the last word there was added the expression yawmal qiyaamati, that is, they would not be able to stand on the "Day of Resurrection".

The variant is mentioned in Abu Ubaid's Kitab Fadhail al-Qur'an.^[23][24][25] The variant was also recorded in the codex of Talha ibn Musarrif, a secondary codex dependent on Ibn Mas'ud's text, Talha likewise being based at Kufa in Iraq where Ibn Mas'ud was based as governor and where his codex was widely followed.^[26]

• Quran 5:89, in the standard text, contains the exhortation fasiyaamu thalaathati ayyaamin, meaning "fast for three days". Ibn Mas'ud's text had, after the last word, the adjective mutataabi'aatin, meaning three "successive" days.^[27]

This variant derives from at-Tabari and was also mentioned by Abu Ubaid and al Zamakhshari.^[28] This variant reading was, significantly, found in Ubayy ibn Ka'b's text as well^[29] and in the texts of Ibn 'Abbas^[30] and Ibn Mas'ud's pupil Ar-Rabi ibn Khuthaim.^[31] Ibn Mas'ud's reading was used by Hanafi scholars to rule that the fasting must be on successive days while Shafi scholars said this was not necessary. This and other Ibn Mas'ud variant readings used in Hanafi jurisprudence (such as restricting hand amputation to right hands) are discussed by Dr Ramon Harvey.^[32]

Pre-eminent status of Ibn Mas'ud as a reciter of the Qur'an

Muhammad ordered Muslims to learn the Qur'an from four individuals and the first of them was Abdullah bin Mas'ud.^[33] So, according to Muhammad, Ibn Mas'ud was an authority on the Qur'an.

Ibn Mas'ud swore that he knew all the surahs of the Qur'an, saying "By Allah other than Whom none has the right to be worshipped! There is no Sura revealed in Allah's Book but I know at what place it was revealed; and there is no verse revealed in Allah's Book but I know about whom it was revealed. And if I know that there is somebody who knows Allah's Book better than I, and he is at a place that camels can reach, I would go to him".^[34]

After Muhammad's choice of Abdullah bin Mas'ud, he was followed by Salim, the freed slave of Abu Hudhaifa, Mu'adh bin Jabal and Ubai bin Ka'b.^[33]It is notable that we do not find any mention of Zayd Bin Thabit who was ultimately entrusted by Abu Bakr with the task of collecting the Qur'an and later or alternatively as part of Uthman's Committee.

Ibn Mas'ud's Disagreement with Uthman

The Qur'an that Ibn Mas'ud had was known and agreed upon by many Muslims in Kufa. When Uthman ordered that all codices must be destroyed and that only Zayd's codex is to be preserved, the reaction of Abdallah ibn Mas'ud was defensive.

"I have not led them [the people of Kufa] astray. There is no verse in the Book of Allah that I do not know where it was revealed and why it was revealed, and if I knew anyone more learned in the Book of Allah and I could be conveyed there, I would set out to him".^[35]

A hadith graded sahih appears in al-Tirmidhi's collection. After mentioning Uthman's attempts to enforce a codice (mushaf), we read:

Similar comments from Ibn Mas'ud are recorded elsewhere.^[36][37]

Dr. Ramon Harvey writes, "The sources point to no less than a century of Kufan resistance to the imposition of a canonised Qur’anic text, with Ibn Masʿud’s variant readings openly used in ritual prayer and even taught as the dominant tradition."^[32]

Qur'an of Ubayy bin Ka'b

Ubayy ibn Ka'b, was another one of the four which were singled-out by Muhammad,^[33] and was considered the best reciter of the Qur'an.^[38] He was known as Sayidul Qura' (The Master of Reciters). Umar the Caliph also agreed that Ubayy was the best reciter, even though he rejected some of what he recited, and said that Ubayy refused to change his recitation.^[39] As detailed in another section below, Ubayy had 116 surahs in his codex, two more than the Uthmanic Qur'an.

Some examples where Ubayy agreed with Ibn Mas'ud and disagreed with Zayd include the following:

• In Quran 22:78 the standard reading says "the faith of your father Abraham (is yours). He hath named you Muslims". In Ubayy's reading, the somewhat ambiguous "He" (huwa) is replaced with "Allah".^[40][41]

• In the standard reading of Quran 39:3 the polytheists say "We worship them" (na'buduhum), whereas Ubayy read them saying "We worship you (plural)" (na'budukum).^[42][43]

• There are a number of cases where whole clauses differed in his text. In Quran 5:45, where the standard text reads wa katabnaa 'alayhim fiiha, meaning "and We inscribed therein for them (the Jews)", the reading of Ubayy ibn Ka'b was wa anzalallaahu alaa banii isra'iila fiiha, meaning "and Allah sent down therein to the Children of Israel."^[44][45] This is similar to the lower text of the Sanaa 1 palimpsest, which reads here "wa katabnaa alaa banii isra'iila".^[46]

Qur'an of Ibn 'Abbas

Among the many examples of different readings attributed to Ibn 'Abbas are the following (further to those already mentioned above).

• Sahih Muslim^[47] and Sahih Bukhari^[48] record that ibn 'Abbas read verse Quran 26:214 with the additional words, "and thy group of selected people among them".

• Ibn 'Abbas is widely reported in al Tabari's tafsir to have said that "ascertain welcome" (tasta'nisu) in Quran 24:27 was a scribal error, and instead should say "ask permission" (tasta'dhinu), a subtly different meaning in Arabic. This narration was also reported elsewhere and classed sahih by al-Hakim, Dhahabi and ibn Hajar^[49]. Ubayy and ibn Mas'ud (the latter with different word order) are also reported in al-Tabari's tafsir to have read tasta'nusu. The ibn Mas'ud wording (except singular instead of plural) is also found in the lower text of the Sana'a palmpsest.^[50]

• He also regarded "Your Lord has commanded" (waqada rabbuka) in Quran 17:23 to be a scribal error which should have read "And your Lord has advised" (wawassa rabbuka)^[51][52], as was the reading of Ubayy, ibn Mas'ud and others according to al-Tabari.^[53]

Extant Early Manuscripts

A significant number of early Hijazi manuscript fragments have been radio-carbon dated to the first Islamic century, covering the majority of the Qur'an between them. All but one of those discovered so far have been of the Uthmanic text type (the exception being the lower text of the Ṣan'ā' 1 manuscript, which is a palimpsest (codex which was washed and written over) found in Sana'a, Yemen, and which contains variants not found in any of the accepted readings of the Qur'an.^[55]). However, these manuscripts are not identical. Every early manuscript falls into a small number of regional families (identified by variants in their rasm, or consonantal text), and each moreover contains non-canonical variants in dotting and lettering that can often be traced back to those reported of the Companions.^[56] The Ṣan'ā' 1 manuscript is especially known to have this feature.^[57][58][59]

Michael Cook's stemmatic analysis of the above mentioned regional variants reported in the Uthmanic copies has shown that they form a tree relationship^[60]. By analysing orthographic idiosyncrasies common to known manuscripts of the Uthmanic text type, Dr. Marijn van Putten has given further proof that they all must descend from a single archetype, generally assumed to be that of Uthman.^[61]

The rasm is only part of the story of the textual transmission of the Qur'an. In the earliest Quran manuscripts (and, we can assume, in the fragments originally collected by Zayd), homographic Arabic consonants were only sparsely dotted to distinguish them, which Adam Bursi has found was the typical scribal practice at that time even for Arabic poetry.^[62] They also lacked any short vowels or other diacritics. When vocalised manuscripts with vowel and other diacritics start to appear, many (mostly non-canonical) readings are found to be imposed upon the rasm.^[63]

Scribal errors in Uthman's codices

Academics generally believe that the above mentioned regional variants were scribal errors made when the original copies of Uthman's consonantal text were produced. These feature also in the canonical readings (qira'at) of those regions, which were required to keep within the scope of the Uthmanic text. The strongest tradition holds that four copies were made, one each for Medina in the Hijaz, Syria (Hims, or less likely, Damascus^[64]), Basra and Kufa in modern day Iraq. As mentioned above, Michael Cook identified that these 40 or so variants form a stemmatic relationship that indicates a written copying process between the four codices.^[60] His list was based on al-Dani's work (d. 444 AH) and can also be read online in a paper by van Putten.^[65] Compiling a similar but improved list of the regional variants widely attested by Muslim scholars, Hythem Sidky reconstructed the same stemma as found by Cook for what must have been four regional exemplar codices.^[64] This is mainly derived due to the twelve variants shared by Syria and Medina to the exclusion of Basra and Kufa, fifteen isolated Syrian variants and three isolated Kufan variants.^[66] Sidky also found an "excellent agreement" between these reports and the earliest manuscripts. In additon, Sidky was able to reproduce the stemmatic result by means of a phylogenetic analysis of these regional differences within the earliest manuscripts (except that Kufa did not achieve a separate node since only one early Kufan manuscript is available). Sidky also found that "a comparison of literary reports against the earliest manuscripts reveals that knowledge of the regional variants does not date back to the time of canonization but was accumulated over time through careful scrutiny of regional muṣḥafs". This indicates that the Uthmanic committee were unaware or did not share information about these differences. He has also commented separately on this topic that further reasons for believing them to be scribal errors are that they are so few in number in what were obviously and reportedly intended to be identical copies, and that they are so insignificant, looking like typical scribal errors that occur in later copying, especially compared to the kinds of more meaningful variants found in companion readings (see earlier section on these above).^[67]

Lost Verses and Surahs from the Qur'an

Professor Yasin Dutton briefly lists several examples in Islamic literature of differences "involving the addition or omission of substantial amounts of material, such as an āya or even a sūra".^[68] These examples and one or two others are explored in this section.^[69]

The lost verse on stoning

There are claims in the hadith that certain verses are missing. For example the 'stoning verse' for adultery. The present day Qur’an does not contain the penalty of Rajm (stoning) for married adulterers, which abrogated the previous penalty. Rather, the Qur'an now extant assigns whipping as the punishment for adultery:

The lost verse of Rajm (stoning) was originally found in Surah al-Ahzab^[70]. According to hadiths recorded in Al-Suyuti's Itqan the lost verse read, "The fornicators among the married men (ash-shaikh) and married women (ash-shaikhah), stone them as an exemplary punishment from Allah, and Allah is Mighty and Wise,", or alternatively, "A married man or woman should be stoned, without hesitation, for having given in to lust." ^[71]

This verse, along with verses regarding adult suckling, were written on a piece of paper and were lost when a sheep or goat ate them.^[72] The loss of the stoning verse is confirmed by Caliph Umar in sahih hadith in which this verse is said to have been included in the book "sent down" to Muhammad, "the Book of Allah".^[73] In another sahih hadith appearing in many collections^[74], Muhammad says he will judge a married woman who committed adultery with an unmarried man by "the Book of Allah" (meaning the Qur'an^[75]) and orders the woman to be stoned and the man to receive 100 lashes. Before becoming lost, the verse on adult suckling had already been abrogated and replaced with a watered down version. Evidently it was not very popular, and was resisted by some of Muhammad's wives.^[76]

Islamic scholars typically explain the loss of the stoning verse as a type of abrogation where the verse is no longer recited but the ruling still applies. Al-Suyuti in his Itqan gives various hadiths in which Muhammad and the Muslim community felt uneasy about writing down, and possibly even reciting such a harsh verse, having witnessed its implementation.^[77] It seems that as a result even the recitation of the exact wording for this verse was lost over time. It is unclear how this is compatible with preservation by calling it abrogation even though the ruling remains, particularly when it involves such a serious topic as a death penalty.

Most of Surah al-Ahzab was lost

It seems that the verse on stoning was not the only one to disappear from surah al-Ahzab.

With sahih isnads appearing in many hadith collections via ‘Aasim ibn Bahdalah, from Zirr, we have this hadith:

Islamqa.info, the popular fatwah website accepts the hadith and that the verses were lost on the authority of the scholars. Its isnad was graded by al-Tabari and al-Albani as sahih, even more emphatically by ibn Hazm, “sahih, as clear as the sun” (إسناده صحيح كالشمس), and hasan (good) by ibn Kathir and ibn Hajar.

Corroborating evidence is given by Qurtubi at the beginning of his tafsir for Surah al-Ahzab. He records this recollection by 'A'isha, although the chain includes ibn Lahee'ah, who many consider weak for having an unreliable memory:

Surah al-Hafd and Surah al-khal'

We know that, whereas Ibn Mas'ud omitted three surahs (al-Fatihah, 113 and 114) from his Qur'an mushaf (codex), Ubayy ibn Ka'b had 116 surahs in his, including two extra short surahs, al-Hafd (the Haste) and al-Khal' (the Separation), which he placed between what are surahs 103 and 104 in Uthman's Qur'an^[80].

In form they are du'as (supplications, prayers), much like Al-Fatihah placed at the beginning of the Qur'an, and surahs 113 and 114.

According to some early sources^[85], ibn Mas'ud too included Khal' and Hafd in his Qur'an mushaf (codex)^[86], as also did Ibn 'Abbas in his mushaf, though its presence in Ubayy's is much more widely documented. One hadith records that Uthman recited them as supplications ^[87], as did Umaya bin Abdullah and Umar according to Al-Suyuti.^[84] One hadith says that these were du'as given by the angel Jibreel to Muhammad.^[88] Al-Suyuti quotes another scholar saying that Surah al-Khal' and Surah al-Hafd were removed from the Qur'an and are now used as du'as.^[89][90]

It doesn't seem that there was agreement among the Muslims on whether these were just du'as or parts of the Qur'an, particularly given that such an important figure as Ubayy ibn Ka'b (possibly ibn Mas'ud and ibn 'Abbas too) recorded them in his widely reported Qur'an codex.

Nor does it seem there was complete agreement on whether other surahs that resemble du'as belonged in the written Qur'an given that ibn Mas'ud left out of his mushaf Surahs Al-Fatihah, and 113 and 114 (called Al-Mu'awwidhatan), as mentioned above. Al-Qurtubi's tafsir contains a narration from ibn-Mas'ud that he omitted Al-Fatihah for brevity^[91], and there was a theory to explain his omission of surahs 113 and 114^[19]. The canonical Qira'at (recitations of the Qur'an) of the Kufan readers include all 3 surahs yet have ibn Mas'ud in some chains of their isnads. However, it is known that the canonical reader Hamza made use of ibn Mas'ud's readings (via his teacher, al A'mash) only insofar as they complied with the Uthmanic rasm.^[92]

Sometimes the 14th century scholar Muhammad Abdul Azim al-Zurqani is quoted, who suggested that the companions who included Al-Hafd and Al-Khal' in their Qur'an mushafs were merely noting them down as du'as alongside the Qur'an, and that this had led to the confusion over whether they were considered Qur'anic. But it is a very unlikely theory that Ubayy ibn Ka'b (and possibly other companions) who recorded these surahs in his mushaf would allow such a misunderstanding to occur. We have two independent lists saying that Ubayy ibn Ka'b sequenced these two du'as between what are now surahs 103 and 104.^[80]

Some argue that Ubayy ibn Ka'b must have had no issue with the two extra surahs being left out because there is a hadith recorded 9 centuries after Muhammad, which says that Uthman had Ubayy dictate the text for Zaid to write down, with refinements by Sa’id bin al-‘Aas^[93]. Such evidence is very late, as well as contradicting sahih hadiths about Zaid's collection process.

The Missing Surah with the Two Valleys

Abu Musa al-Ash'ari, one of the early authorities on the Qur'an text and a companion of Muhammad, claimed a surah which resembled at-Tawba (also known as Bara'at) in length and severity was forgotten and lost, but included a passage on the greed of man, which is not in today's Qur'an. Various narrations have slightly differing wording for this lost passage, which is consistent with it being insufficiently remembered.

Ibn Abbas was likewise unsure whether it was part of the Qur'an or not:

Ubai said that it was considered as a saying from the Qur'an for a while during Muhammad's lifetime. At best, it could be claimed to be an example of a type of abrogation where the verses are lost. Why the verse would be abrogated is unexplained.

Al-Suyuti records the recollection by Abu Waqid al-Laithii of the occasion when the lost passage about the valleys was revealed. He says that Muhammad claimed it as a revelation from Allah, just like when he received other revelations.^[94]

Lost verses from Surah at-Tawba

Surah at-Tawba (also known as al Bara'at) was originally equal to the length of al-Baqara according to narrations recorded by al-Suyuti (best known for his Tafsir al-Jalalayn) in The Itqan^[95] and Tafsir al-Qurtubi^[96]. In a Hasan hadith in the collection of Tirmidhi, Uthman is narrated as saying that they didn't know whether or not Surah at-Tawba was part of Surah al-Anfal, and Muhammad died without making it clear, so they were placed together.^[97]

The "Bring a surah like it challenge"

As detailed in the sections above, there were non-Qur'anic surahs and verses that sounded very much like those of the Qur'an. Surah al-Hafd and Surah al-khal', and the verses about Adam and the valleys sounded so Qur'anic that they were at one time believed to be so by speakers of 7th century Arabic, Sahabah no less. Theories that these were once part of the Qur'an and later abrogated, or that Al-Hafd and Al-Khal' were du'as given to Muhammad by Jibril do not give a reason why they may have been abrogated or why confusion was allowed to arise about the status of the latter two when they were recorded in the mushaf of Ubayy (and perhaps other companions).

The Qira'at (Variant Oral Readings of the Qur'an)

As mentioned above, numerous possible oral readings of the Qur'an can be and were imposed upon the Uthmanic rasm (an early stage of Arabic orthography, which lacked diacritics such as short vowel signs and with scarce use of dotting to distinguish certain consonants). Today we have seven or ten canonical qira'at, which are slightly different early oral recitations or readings of the Qur'an by famous readers. There were once many more qira'at, from which twenty-five were described by Abu 'Ubayd al-Qasim ibn Sallam two centuries after Muhammad's death, and restricted to seven after three centuries following a work by Abu Bakr Ibn Mujahid (d.936 CE). A further three qira'at were added to the canonical seven many centuries later by ibn al-Jazari (d.1429 CE) - those of Abu Jafar, Ya'qub and Khalaf. These three were popular since the time of the seven^[98], and provide additional variants^[99]. Some scholars regarded them as having a somewhat less reliable transmission status than the seven.^[100] Ibn al Jazari lamented that the masses only accepted the seven readings chosen by ibn Mujahid.^[101]

Al Zarkashi (d.1392 CE) argued that even the differences in the canonical readings are mutawatir (mass transmitted), despite each only having one or a small number of single chains of transmission between Muhammad and the eponymous reader, because the inhabitants in the cities in which they were popular also heard them. Professor Shady Nasser finds it hard to accept al Zarkashi's argument since in that case "variants within one Eponymous Reading should not have existed", as well as due to the presence of multiple popular readers in each city.^[102] As noted above, most canonical readings are not found in early vocalised manuscripts. Ibn al-Jazari subsequently obtained a fatwa (from ibn al Subki) declaring that all 10 readings were fully mutawatir, though later he changed his mind.^[103]

Each of the Qira'at has two canonical transmissions (riwayat) named after its transmitters, one of which is the basis for any particular text (mushaf) of the Qur'an. For example, the mushaf used mainly in North Africa is based on the riwayah of Warsh from Nafi (the reading of Nafi transmitted by Warsh). As Nasser explains, the two-Rawi canon for ibn Mujahid's choice of seven readings was effectively canonized due to the popularity both of a simplified student Qira'at manual by al-Dani (d.1053 CE; who in another more complicated work documents many more transmissions), and a poetic form of this manual by al-Shatibi (d.1388 CE).^[104] The canonical transmitters all differ in their readings, even when they transmit from the same reader. This two transmitter system was expanded when ibn al Jazari canonized the three readers after the seven mentioned above, giving twenty canonical transmitters for the ten readers in total.

Relationship between Qira'at and Ahruf

The legitimacy of variant oral readings is derived from some hadith narrations that the Qur'an was revealed to Muhammad in seven ahruf. The word ahruf literally means words or letters, but is commonly translated as modes of recitation. The nature of these ahruf generated a wide range of theories, some more plausible than others.^[106] A popular, though problematic theory was that these were dialects of seven Arab tribes, and only one, that of the Quraysh was retained by Uthman. However, most variants among the canonical readings are not of a dialect nature^[107]. It also makes little sense of Sahih Bukhari 9:93:640 in which Muhammad appealed to ahruf when two companions who were both of the Qureshi tribe disagreed on a reading. A more tenable view is that the ahruf represent variant readings at certain points in the Quran.

A related question on which scholars differed was whether or not all the ahruf were preserved. One group including ibn Hazm (d.1064 CE) believed that all seven ahruf were accomodated by the Uthmanic rasm (consonantal skeleton), finding it unimaginable that anything would be omitted.^[108]. Al-Tabari argued that only one harf was preserved by Uthman (which he interpreted to mean the dialect of the Quraysh), while Ibn al Jazari said the view of most scholars is that only as many of the ahruf as the Uthmanic rasm accommodated were preserved^[109]. Indeed, this latter is more viable theologically, for the non-Uthmanic companion readings must be fraudulent under the first view, and problems with the second view include those mentioned above.

As part of the majority view reported by ibn al Jazari, the Uthmanic codex was based on the harf of the "final review" or final revealed version of the Quran^[110]. However, there were around 40 scribal errors in the official copies of the Uthmanic text (see below).^[111] Canonical qira'at were required to comply with this range rather than an entirely unified text. Indeed, in some cases they even strayed beyond these boundaries.^[112]

Some scholars believe that the seven ahruf story was invented at an early stage to accommodate the proliferation of variant readings. Given that the Qur'an and hadith reveal that Muhammad would forget entire verses, another plausible theory would be that he was inconsistent in his recitation and used it as a convenient excuse^[113], as well as a way of avoiding disputes between his followers^[114] by saying that there were up to seven valid ways to recite a verse.

Differences in the Qira'at

Muslims are commonly told that the differences between the Qira'at can be explained away as styles of pronunciation or dialect and spelling rules (called uṣūl, rules that apply to the entire reading). Yet there is another category, farsh, of individual differences, which also includes changes in wording. In some cases the variations added or ommitted words, or are completely different words or contradict each other in meaning. The Corpus Coranicum database^[115] and the nquran website^[116] can be used as neutral online sources for verifying the existence of such variations in the Qira'at. The Bridges translation can be selected on quran.com and highlights words with canonical variants, listing them in English with their readers as footnotes. An interesting example is given below, and more of them are listed in the next section about the popular Hafs and Warsh transmissions.

Altogether, there are more than 1000 words with variants among the canonical readings of the Quran^[117], about two percent of the total. Some are regarded as dialect differences, while others including vowel differences affect grammar and meaning. Around 300 involve different consonantal dotting, generally changing attached pronouns or producing a different root word. In addition, there are around 40 variants arising from the regional Uthmanic codices (see below), in a few cases adding or omitting insignificant words. It is common for a word to have more than two variants, with no obvious intention in so much variety.^[118]

In Quran 18:86, Dhu'l Qarnayn finds the sun setting in a muddy spring, according to the Qira'at used by today's most popular transmissions of the Qur'an. However, in around half of the various Qira'at the sun intead sets in a warm spring. The latter variant is even used in some English translations. It is easy to see how the corruption arose (whichever one is the variant). The arabic word حَمِئَة (hami'atin - muddy) sounds very similar to the completely different word حَامِيَة (hamiyatin - warm). Al-Tabari records in his tafseer for this verse the differing opinions on whether the sun sets in muddy or warm water.

The reading of ibn Amir, which is one of those qira'at containing hamiyah instead of hami'ah, is still used in some parts of Yemen, and used to be more widespread.^[119]. In written form this difference is not just a matter of vowel marks. Even the consonantal text with dots is different, though in the original Uthmanic orthography they may have looked the same due to the very limited use consonantal dotting and different uses of alif at that time. A scan of a printed Qur'an containing the mushaf of Hisham's transmission from ibn Amir's reading can even be read online and it can be seen that حَامِيَة (warm) is used in verse 18:86^[120].

For further discussion, see the section Origin of the Qira'at Variants further below.

Differences in the Hafs and Warsh Texts

Apart from other earlier variant readings, and those of al-Duri from Abu Amr still used in Sudan, and of Hisham from ibn Amir still used in parts of Yemen, there are two different readings of the Qur'an currently widespread in printed text (mushaf), named after their respective 2nd-century transmitters: Hafs from Asim (one of the Kufan readers) and Warsh from Nafi (of Medina).

The Hafs reading is the more common and used in most areas of the Islamic world. Warsh is used mainly in West and North-West Africa as well as by the Zaydiya in Yemen. Here are some of the differences.

Differences Between Other Canonical Readings

There are many more differences between other transmissions besides those of Hafs and Warsh. All are available in printed form and in the online sources mentioned above. The following are a few examples, mostly of conflicting variants in quoted dialogue incidents.

The standard Islamic view is that every variant within the canonical qira'at (readings) were recited by Muhammad, and the canonical readers made choices from among the authentic variants passed down to them. The claim is that even when the variants are completely different words or when words are added or omitted, that these are all divinely revealed alternatives. This doesn't address variants that contradict each other such as the examples above, nor explain superfluous variants (examples of which are even more common^[118][124]). Nor does it explain why there would be so many authentic variants that just so happened to be accommodated by the Uthmanic orthography or sound similar. As mentioned above, over 1000 out of c.77,000 words in the Qur'an have canonical variants. If such a large number of variants all pre-date the Uthmanic rasm standard and are authentic, that would imply that many thousands more had also been uttered by Muhammad that did not go on to pass this filter. It seems doubtful that the rasm standardisation would have been successful had it requried so many words to be discarded by the early Muslim communities, well beyond the degree of variation reported even of companion codices.

A more extensive study of differences between the Hafs and Warsh transmissions and comparisons with Qur'an manuscripts can be read online^[125]. Further studies of dialogue variants and superfluous variants are also available.^[126][127]

Origin of the Qira'at Variants

The Uthmanic codex was written in a rasm, which is a "defective" Arabic script, meaning that there was inconsistent use of alifs, no markings for short vowels and sparse (if any) dots that were in later times used to distinguish different but identical looking consonants.

Professor Shady Nasser shows that at the time when ibn Mujahid wrote his Kitab al Sab'ah selecting the 7 eponymous readings that later became canonical, adherence of readings to the Uthmanic rasm and good Arabic grammar were already important criteria ^[128], but ibn Mujahid restricted his selection to just 7 by choosing the consensus readings from each of Mecca, Medina, Basra, Damascus and the 3 most popular readers from Kufah, where the legacy of Ibn Mas'ud's (now banned) reading meant that there was no dominant Uthmanic reading in that city.^[129].

Nasser further shows that grammarians such as al-Farra^[130], and scholars such as al-Tabari readily criticised variants in these same readings shortly before they were canonized^[131] (as did al-Zamakhshari 200 years afterwards)^[132]). Even ibn Mujahid said variants now considered canonical were wrong.^[133] After ibn Mujahid's book, a genre of literature arose that "indicates the rising need to provide grammatical and syntactic proofs in order to back up the arguments necessary to assess the superiority of one reading over another." ^[134]. Ibn Mujahid's decision to select just 7 readings drew frequent criticism after its publication^[135]. The consensus notion that all variants in these 7 were divinely preserved in a chain back to the Prophet himself only came about later, by which time there was of course no room for arguments and reasoning to try to prove the superiority of one variant over another.^[136] As Nasser writes, "The problem that caused heated discussion for centuries afterwards was the origin and transmission of the eponymous Readings; were these Readings transmitted through tawātur or single chains of transmission? Are there Readings better than others or are they equally divine?"^[137].

Dr Marijn Van Putten has shown that while the canonical readings largely comply with the Uthmanic rasm, more specifically they also each closely comply with the regional variants of that rasm, which were sent out to the major intellectual centres of early Islam and contained a small number of copying mistakes. So, the Kufan readings closely correspond to the variants found in the rasm of the codex given to that city and so on.^[138][139]

This would be an extraordinary coincidence if the variants are entirely due to oral transmissions going back to the recitations of Muhammad (though certainly the general agreement between readings where the rasm is ambiguous demonstrates that there was also oral transmission ^[140]). Instead, the regional correspondence of rasm and oral reading variants is easily explained if the readings were adapted to fit the codices given to those regions. By analysing the reported variants between regional codices, modern scholarship has confirmed that they form a stemma (textual tree relationship), suggesting that those particular variants did not originate in oral transmission.^[141]

If qira'at variants could sometimes arise from the rasm, we should also expect this to occur even in places where the rasm did not vary. Munther Younes highlights a particularly interesting example among the hundreds known.^[142] In Quran 4:94 we have the canonical variants fa-tabayyanū or fa-tathabbatū. In this case, the variant root words do not share even a single consonant in common (bāʼ-yāʼ-nūn versus thāʼ-bāʼ-tāʼ), but nevertheless both variants fit the defective script of the Uthmanic rasm, which lacked dots and vowels. Other examples of variants with the same rasm include Quran 6:57, where 4 of the canonical 7 qira'at have yaqḍi l-ḥaqqa "He judges the truth" rather than yaquṣṣu l-ḥaqqa "He declares the truth"^[143] and Quran 10:30 where two readers have tatlū (recounts, recites), whereas the other five have tablū (tests) ^[144]. In these examples the similarity between the variant readings is graphic (how the rasm looks) rather than phonic (how they sound).

The variants in the seven eponymous readings are only rarely evident in the earliest manuscripts. They contain the above mentioned regional differences, but otherwise tend to reflect dotting and lettering traceable to the more substantial variant readings of the Companions and not necessarily of the seven.^[56]

The Qurra' Community

In a detailed monograph on Ibn Mujahid's canonization of the seven readings, Nasser shows that written notes played a significant role in transmission of the readings in the 2nd century. Despite their best efforts, some canonical readers and their transmitters were said to have doubts about their (often unique) readings. Abu 'Amr, al Kisa'i, Nafi, and the transmitters of 'Asim (Hafs and Shu'ba) are all reported "retracting a reading and adopting a new one" in some cases. Shu'ba "became skeptical" of his teacher 'Asim's reading of a certain word and adopted another, and said he "did not memorize" how certain words were read. In one instance Ibn Dhakwan, the transmitter of Ibn Amir's reading, found one reading for a word in his book/notebook, and recalled something different in his memory. When the detailed recitation of a word was unknown, "the Qurrāʾ resorted to qiyās (analogy)", as too did Ibn Mujahid when documenting the readings as he often faced conflicting or missing information. There were also cases of transmitters misattributing variants to the wrong eponymous reader (some transmitted more than one reading), and readers adapting to what they regarded as flawed parts of the Uthmanic rasm.^[145]

In one summary he writes, "The multiple readings reported on behalf of the same Eponymous Reader or Canonical Rāwī, were not only due to transmission errors, inaccuracies, the 'flexibility' of the consonantal rasm, and the existance of a depository of different, yet acceptable traditions from the previous generations of Qurʾān masters. These readings were also generated because Qurʾān Readers occasionally modified and changed their readings over time, retracted certain readings, corrected others, and struggled to remember how precisely some variants were performed."^[146]

While ibn Mujahid only gave formal isnads from himself to the Eponymous readers (whose readings he documented partially based on written notes^[147]), he gave some biographical sketches of the transmissions between the Prophet and these readers, the chains of which generally had at least 4 or 5 links though occasionally 3.^[148] Most of the seven main readers and their canonical transmitters did not escape criticism for their reliability in hadith and/or their Qur'an recitations in at least some biographical sources^[149].

Changes to the spoken Arabic dialect of the Qur'an

In a number of papers and forthcoming book,^[150] van Putten has identified that the Quranic Consonantal Text (QCT) reveals certain features about the Hijazi dialect in which it was originally uttered. These do not match the dialects found in the canonical qira'at, nor for this reason the orthography of Qur'ans published today. Evidence from internal rhyme (when the traditional qira'at recitations and later orthography break the rhyming structure of a passage) is particularly helpful in this regard, supplemented by ancient epigraphic (inscription) evidence, including transliterations of the Arabic of that region and time into other languages. Van Putten and Stokes have found that the QCT "possessed a functional but reduced case system, in which cases marked by long vowels were retained, whereas those marked by short vowels were mostly lost".^[151] Van Putten has also found, in line with the accounts of early Muslim linguists, that the Hijazi dialect spoken by Muhammad had lost the use of the hamza except for word-final ā.^[152] He has also found that nunation at the end of feminine nouns ending with -at was not present in this dialect.^[153]

False Report about Al Hajjaj and the Uthmanic Qur'an

Al-Hajjaj Ibn Yusuf Al-Thakafi, who lived in the years AD 660-714, was a teacher of the Arabic language in the city of Taif. Then he joined the military and became the most powerful person during the reign of Caliph Abd al-Malik Ibn Marawan and after him his son al-Waleed Ibn Abd al-Malik.

One report claims that the Uthmanic Qur'an was changed by Al-Hajjaj, inserting 11 small changes into the text and sending them out to the main cities. Less dramatic reports exist of him making reforms to orthography, some of which have been disproven by dated manuscripts, but this is the only claim that touches on the underlying reading of the text.

However, this report is not considered credible by academics for a number of reasons, including the fact that all extant manuscripts (except for the Ṣan'ā' 1 palimpsest lower text) can be traced to a single archetype, as explained above. Moreover, Sadeghi and Bergmann have shown that the Basran author of this report about al-Hajjaj had simply mistaken some errors in a particular manuscript as being the Uthmanic standard and compared it with the manuscript of al-Hajjaj.^[154]

See also

• Quran Variants - Articles and Resources page

References