Materials for the History of the Text of the Koran

Arthur Jeffery

CRITICAL INVESTIGATION OF THE TEXT of the Koran is a study which is still in its infancy. Within the fold of Islam it seems never to have attracted much attention. The growth of the Qurra‘is evidence that there was some interest in the question in the early days of Islam1 but with the fixing of the text ne varietur by the Wazirs Ibn Muqla and Ibn ‘Isa in 322 A.H. at the insistence and with the help of the savant Ibn Mujahid (d. 324)2, and the examples made of Ibn Miqsam (d. 362) and the unfortunate Ibn Shanabudh (d. 328) who persisted in making use of the old readings after this fixing of the text3, such interest as there was seems to have come to an end. Variant readings within the limits of the seven systems4 that were admitted as canonical by the decision of Ibn Mujahid naturally continued to be studied by a limited group of scholars, and the readings of the other uncanonical Readers occasionally received attention, more particularly the systems of the Ten5 and the Fourteen6, who were nearest to canonical position, though at times others also were included7. No definite attempt, however, was made to construct any type of critical text of the Koran8, and for the most part textual studies were confined to questions of orthography (rasm) and pause (waqf). Thus the older variants, even though they were known to be represented in some of the older codices, for the most part survived only in the works of two classes of savants, firstly certain exegetes who were interested in the theological implications of such variants, and secondly the philologers who quoted them as grammatical or lexical examples.

Originally published as the Introduction (pp. 1-18) and pp. 20-24, 114-16, 182-84 of Materials for the History of the Text of the Qur’ān: The Old Codices, edited by Arthur Jeffery (Leiden: E.J. Brill, 1937).

It is thus that in the Koran commentaries of az-Zamakhshari (d. 538),9 of Abu Hayyan of Andalus (d. 745),10 and the more recent Yemenite writer ash-Shawkani (d. 1250)11 who seems to have used some good old sources no longer available to Western scholars, we find recorded a goodly number of old variants representing a different type of consonantal text from that officially known as the ‘Uthmanic text, and in the philological works of such writers as al-‘Ukbari (d. 616)—the blind philologer of Baghdad12, Ibn Khalawaih (d. 370)13—the savant of the Hamdanid Court of Saif ad-Dawla at Aleppo, and the even more famous Ibn Jinni (d. 392),14 a not inconsiderable amount of such material has been preserved, which in some cases, indeed, proves to be one source from which it came to the exegetes.

To apply this material to a critical investigation of the text of the Koran seems never to have occupied the attention of any Muslim writer. In the Itqan15, as-Suyuti’s great compendium of Muslim Koranic science, we have recorded a great deal that concerns matters of the Muslim Massora, matters of considerable interest for the history of the exegesis of the Koran, but very little that bears on the investigation of the text.

Nor has the subject attracted much attention in the West. Nöldeke opened it up in 1860 in the first edition of his Geschichte des Qorans, and Goldziher drew attention to its importance in the first lecture of his Richtungen,16 but it received no systematic treatment until Bergsträsser undertook his Geschichte des Qorantexts17 as the third part of the revised edition of Nöldeke’s work, and with characteristic thoroughness began to work down to bedrock on the subject. It is an extraordinary thing that we still have no critical text of the Koran for common use. Flügel’s edition which has been so widely used and so often reprinted, is really a very poor text, for it neither represents any one pure type of Oriental text tradition, nor is the eclectic text he prints formed on any ascertainable scientific basis. Some of the Kazan lithographs18 make an attempt at giving the seven canonical systems on the margin, but only very incompletely. The same is true of the curious Teheran lithograph of 1323, which prints parts of the text in Kufic script (with interlinear naskhi) and parts in ordinary script, with a selection of the seven on the margins. The best text so far available is the Egyptian standard edition of 1342 (1923)19of which there are several later prints. This edition attempts to present a pure type of text according to one tradition of the Kufan school as represented by Hafs ‘an ‘Asim, though unfortunately some corruptions have crept in owing to the use by its editors of younger authorities on the Kufan tradition instead of going back to older and better sources.20

The orthodox Muslim theory of the text is well known. According to this theory the Prophet arranged to have the revelations written down immediately they were revealed and used to collate once every year with the Angel Gabriel the material that had thus far been revealed. In the last year of his life they so collated it twice21. When the Prophet died the text of the Koran was thus already fixed, and all the material gathered in an orderly fashion though it had not yet been written out, at least not in book form. Under the caliphate of Abu Bakr took place the writing of it out in a first official recension. Later, in the caliphate of ‘Uthman it was discovered that all sorts of dialectal peculiarities had crept into the recitation of the text, so ‘Uthman formed a committee, borrowed from Hafsa the copy made by Abu Bakr, and on its basis had a standard codex written out in the pure dialect of Quraish. Copies of this were made and sent to the chief centers of the Muslim empire where they became metropolitan codices, and all other codices that had been formed were ordered to be burned. This was the second recension and all modern editions produced in the East are supposed to be exact reproductions of the text (though not of the form) of this ‘Uthmanic recension.22

Very little examination is needed to reveal the fact that this account is largely fictitious. Nothing is more certain than that when the Prophet died there was no collected, arranged, collated body of revelations. Recent research by Dr. Bell of Edinburgh and Prof. Torrey of Yale has suggested that there is internal evidence in the Koran itself that the Prophet kept in his own care a considerable mass of revelation material belonging to various periods of his activity, some of it in revised and some of it in unrevised form, and that this material was to form the basis of the Kitab he wished to give his community before he died. Death, however, overtook him before anything was done about the matter. If this is so we are at a loss to know what became of this material, which obviously would have been the community’s most precious legacy.23 The earliest strata of tradition available to us make it quite certain that there was no Koran left ready as a heritage for the community. The Prophet had proclaimed his messages orally, and, except in the latter period of his ministry, whether they were recorded or not was often a matter of chance. Some pieces of revelation material seem to have been used liturgically and so probably would have been written. Some pieces he himself caused to be written down in permanent form as they were of a definite legislative character.24 Besides these there were numerous portions, generally small pieces, though sometimes pieces of considerable extent, that were in the possession of different members of the community, either memorized or written down on scraps of writing material that happened to be handy. Certain individuals among the early Muslims, perhaps even a little before the Prophet’s death, had specialized in collecting or memorizing this revelation material. They and their successors became known as the Qurra’—the Reciters, later the Readers, who constituted as it were the depository of revelation. Tradition says that it was the slaughter of a great number of these at the Battle of Yamama in 12 A.H. that caused interest to be aroused in getting all the revelation material set down in permanent written form, lest with the passing away of the Qurra’ much of it should be lost.25

That Abu Bakr was one of those who collected revelation material was doubtless true. He may possibly have inherited material that the Prophet had stored away in preparation for the Kitab. That he ever made an official recension as the orthodox theory demands is exceedingly doubtful. His collection would have been a purely private affair, just as quite a number of other Companions of the Prophet had made personal collections as private affairs. It was after the death of the Prophet that these collections became important. We have well-known stories of how ‘Ali, Salim, Abu Musa and others had collections, and there are traditions which give lists of those who had commenced making collections or memorizing during the lifetime of the Prophet. As no two of these lists agree with one another to any great extent one is driven to conclude that while it was known that such collections were made there was no accurate information, save with regard to a few names, as to who made them.26 Orthodox theory, even to the present day, has insisted that the word jama’a “to collect” used in these traditions means nothing more than “to memorize” and so does not imply that the collection was made in written form. As, however, ‘Ali brought along what he had collected on the back of his camel, as some of the collections had come to have independent names, and as ‘Uthman, after sending out his official copies to the metropolitan cities, had to order all other copies to be burned, there cannot be the slightest doubt that there were written collections.

What we find in early Islam, as a matter of fact, is only what we might have expected to find. Different members of the community who were interested began to collect in written form so much as they could gather of the revelation material that had been proclaimed by the Prophet. Later, with the gradual expansion of the Muslim empire, some of these collections began to acquire notoriety as they came to be in some sort authoritative in different centers. Naturally it would be those collections that could claim some completeness that would attain to this position of eminence. Thus we read that the people of Homs and Damascus followed the codex of Miqdad b. al-Aswad27, the Kufans that of Ibn Mas‘ud, the Basrans that of Abu Musa al-Ash’ari, and the Syrians in general that of Ubai’ b. Ka’b (Ibn al-Athir, Kamil, III, 86). Here we have the beginning of metropolitan codices, each great center following that collection, or perhaps we may say that type of text, which had local fame.

Now when we come to the accounts of ‘Uthman’s recension, it quickly becomes clear that his work was no mere matter of removing dialectal peculiarities in reading, but was a necessary stroke of policy to establish a standard text for the whole empire. Apparently there were wide divergences between the collections that had been digested into codices in the great metropolitan centers of Medina, Mecca, Basra, Kufa and Damascus, and for political reasons if for no other it was imperative to have one standard codex accepted all over the empire. ‘Uthman’s solution was to canonize the Medinan codex28 and order all others to be destroyed. It is very significant that the Qurra’ were violently opposed to ‘Uthman because of this act,29 and there is evidence that for quite a while the Muslims in Kufa were divided into two factions, those who accepted the ‘Uthmanic text, and those who stood by Ibn Mas’ud, who had refused to give up his codex to be burned.30

There can be little doubt that the text canonized by ‘Uthman was only one among several types of text in existence at the time.31 To canonize the Medinan text was doubtless the natural thing to do, since in spite of the fact that Kufa early came to have the reputation of being par excellence the center of Koranic studies, the prestige of Medina, the Prophet’s own city, must at that time have been enormous, and the living tradition would doubtless have been most abundant there. We may even say that a priori the Medinan text had all the chances in its favor of being the best text available. Nevertheless it is a question of the utmost importance for any study of the history of the Koranic text, whether we can glean any information as to the rival types of text that were suppressed in the interests of ‘Uthman’s standard edition.

In the works of the exegetes and the philologers we not infrequently come across variant readings that have been preserved from one or other of these displaced codices. Sometimes the reference is merely to a “codex of the Sahaba” or “a certain old codex” or “in certain of the codices” or “in the former text.” At times it is to one of the cities—“a codex of Basra,” “a codex of Homs,” “a codex of Ahl al-Aliya” (Baghawi II, 52). Sometimes it is to a codex in the possession of some particular person, as “a codex belonging to al-Hajjaj” (Khal. 122; Gin. 60), or “a codex belonging to the grandfather of Malik b. Anas” (Muqni 120), or a codex used by Abu Hanifa (see Massignon’s al-Hallaj, I, 243 n. 5), or one of Hammad b. az-Zibriqan (Khal. 55; Muzhir II, 187). Mostly, however, the references are to the well-known old codices of Ibn Mas‘ud, Ubai’ b. Ka‘b, etc., which were known to go back to the time before the canonization by ‘Uthman of one standard type of text.

The amount of material preserved in this way is, of course, relatively small, but it is remarkable that any at all has been preserved. With the general acceptance of a standard text other types of text, even when they escaped the flames, would gradually cease being transmitted from sheer lack of interest in them. Such readings from them as would be remembered and quoted among the learned would be only the relatively few readings that had some theological or philological interest, so that the great mass of variants would early disappear. Moreover, even with regard to such variants as did survive there were definite efforts at suppression in the interests of orthodoxy. One may refer, for instance, to the case of the great Baghdad scholar Ibn Shanabudh (245-328), who was admitted to be an eminent Koranic authority, but who was forced to make public recantation of his use of readings from the old codices.

Ibn Shanabudh’s was not the only case, and such treatment of famous scholars32 was not encouraging to the study of the variants from the pre ‘Uthmanic period. That orthodoxy continued to exert this same pressure against uncanonical variants is revealed to us from many hints from the period subsequent to Ibn Shanabudh. For example, Abu Hayyan, Bahr VII, 268, referring to a notorious textual variant, expressly says that in his work, though it is perhaps the richest in uncanonical variants that we have, he does not mention those variants where there is too wide a divergence from the standard text of ‘Uthman. In other words, when we have assembled all the variants from these earlier codices that can be gleaned from the works of the exegetes and philologers, we have only such readings as were useful for purposes of Tafsir and were considered to be sufficiently near orthodoxy to be allowed to survive.33

Modern Muslim savants almost invariably set aside the variants recorded from the old codices on the ground that they are Tafsir, or as we should say, explanatory glosses on the ‘Uthmanic text, and they roundly condemn such ancient scholars as Ibn Khalawaih and Ibn Jinni for not knowing the difference between Qira’at and Tafsir. It is clear, however, that only such qira’at as were of the kind that could be used for tafsir had any likelihood of being preserved.


In the fourth Islamic century there were three books written on this question of the old codices which had some influence on later studies. These were the works already mentioned of Ibn al-Anbari, Ibn Ashta, and Ibn Abi Dawud. In each case the book was entitled Kitab al-Masahif and in each case the work, while dealing with the ‘Uthmanic text, its collection, orthography, and the general Massoretic details with regard to it, dealt also with what was known of the old codices which it had replaced. The most famous of the three was that of Ibn al-Anbari (d. 328), a work which was doubtless composed before the canonization by Ibn Mujahid of the Seven Readers. The work is lost but from the use made of it by later writers such as-Suyuti,34 one gathers that it contained a certain amount of Tafsir as well as information as to the readings from the old codices. The work of Ibn Ashta (d. 360) seems to have been of somewhat similar scope. He was a pupil of Ibn Mujahid and wrote a special work al-Mufid on the subject of the uncanonical variants,35 besides this work on the codices which was also used by as-Suyuti.36 The only work of this kind that has survived, however, is that of Ibn Abi Dawud (d. 316) which, unfortunately, seems to have been the narrowest in scope of them all.

‘Abdallah b. Sulaiman b. al-Ash’ath Abu Bakr b. Abi Dawud37 as-Sijistani was born in 230 A.H. the son of the Imam Abu Dawud whose collection ranks third among the canonical collections of Hadith. He was born in Sijistan but his father took him early on his travels and he is said to have visited Khorasan, Isfahan, Fars, Basra, Baghdad, Kufa, Medina, Mecca, Damascus, Egypt, al-Jazira and ath-Thughur. In every place where there were scholars his father set him to learn from them, so that he may be said to have been the pupil of most of the great savants of his day.38There is a story that when he came to Kufa he had only one dirham which he spent on thirty bushels of broad beans. Each day he ate a bushel of the beans and by the time they were finished he had mastered a thousand traditions (or some say 30,000) from the Kufan teacher Abu Sa’id al-Ashajj.

His chief fame all his lifetime was as a tradionist. There is a story that he returned to Sijistan in the days of ‘Amr b. al-Laith and some of his fellow townsmen gathered together to request him to recite to them Hadith that he had learned on his journeyings. He refused on the ground that he had no book, but they retorted, “What need has the son of Abu Dawud of books?” So he submitted with good grace and dictated a great number of traditions from memory. When he got back to Baghdad he found that the story had preceded him and the Baghdadis were saying that he had fooled the innocents of Sijistan. But when they hired scribes to go to Sijistan and bring back copies of what Ibn Abu Dawud had dictated there, they found that on comparing them with the authorities in Baghdad they could find only six mistakes in all that he had dictated from memory.

In Koranic studies he was a pupil of Abu Khallad Sulaiman b. Khallad (d. 262), Abu Zaid ‘Umar b. Shabba (d.262),Yunus b. Habib (d.267), Musa b. Hizam at-Tirmidhi (c. 260), and Ya‘qub b. Sufyan (d. 277), and was one of the teachers of Ibn Mujahid (d. 324) and an-Naqqash (d. 351). He wrote a number of works on Koranic subjects. In the Fihrist, pp. 232, 233 we find mentioned:

A book of Tafsir(see also Fihrist 34 [11]; Dhahabi, II, 80; al-Khatib, IX,


Kitab an-Nasikh wa’l-Mansukh (see Fihrist 37 [25]; Dhahabi, II, 80).

Kitab Nazm al-Koran.

Kitab Fada‘il al-Koran.

Kitab Shari’at at-Tafsir.

Kitab Shari‘at al-Maqari’.

Dhahabi also mentions a book called al-Quran, which probably means his Kitab al-Masahif,39 which is also sometimes called, though with less justice, Kitab Ikhtilaf al-Masahif. Al-Khatib mentions a book on qira‘at which may refer to the Masahif-book or may be another work, for Abu ’l-Mahasin in an-nujum az-Zahira (Eg. ed. III, 222) mentions him as a writer on qira’at.

There are a number of traditions going back to him that are not pleasing to orthodoxy and so there was put into circulation the legend that his father had branded him as a liar, and therefore no attention is to be paid to material that is dependent on his authority. This, of course, is tendential, and the biographers usually regard him as trustworthy, the Mughni even noting that his father’s branding him as a liar was over something other than Hadith.40 To the last he seems to have held the respect of his townspeople for there is a pleasing story of how when he was old and blind he used to come and sit on the mimbar while his son Abu Ma’mar would sit on the step below him with the book. From his book the son would mention the particular Hadith and then from memory the old man would go on reciting to the people.

Of his Kitab al-Masahif there are three manuscripts known, one in the Zahiriya Library at Damascus (Hadith, No. 407), one in the Egyptian State Library (Qira’at, No. 504), and one in my own possession. Both these latter, however, are copies of the Zahiriya MS, so that we are really dependent on the one manuscript for establishing the text.

The number of actual variants given in this text is very small and obviously represents only those that happened to be found in his particular collection of traditions; Most of the vatiants he notes are also to be found in other Koranic works. His chief importance is that he brings before us so many codices of which we have no mention as such in any other source at present available. The codices of Ibn Mas‘ud, Ubai b. Ka’b, Hafsa, Anas and others are mentioned in numerous other sources, but though we find numerous references to shadhdh readings of such early authorities as ‘Ubaid b. ‘Umair, ‘Ikrima, al-A‘mash, Sa’id b. Jubair and others, we did not know of actual codices of theirs, though in some cases we strongly suspected their existence. An interpolation in the text might seem at the first glance to be seeking to avoid the implications of this fact by making Ibn Abi Dawud say that he uses the word mushaf (codex) in the sense of harf or qira’a (reading) so that the variants he quotes need not be regarded as coming from actual written codices. There can be little doubt, however, that when he speaks of the mushaf of So and So he really means a written codex. In the case of some of the codices he mentions we have, of course, ample evidence from other sources of their independent existence, and in the case of some others the nature of the variants quoted strongly suggests that they must have been derived from written codices.

There are a few other old codices mentioned in other works which are not given by Ibn Abu Dawud. Adding them to his lists in the interests of completeness we can draw up the following scheme of the old codices.

(a) Primary Codices:

Salim (d. 12)

‘Umar (d. 23)

Ubai’ b.Ka‘b (d. 29)

Ibn Mas’ud (d. 33)

‘Ali (d. 40)

Abu Musa al-Ash’ari (d. 44)

Hafsa (d. 45)

Zaid b. Thabit (d. 48)

‘A’isha (d. 58)

Umm Salama (d. 59)

‘Abdallah b. ‘Amr (d. 65)

Ibn ‘Abbas (d. 68)

Ibn az-Zubair (d. 73)

Ubaid b. ‘Umair (d. 74)

Anas b. Malik (d. 91 )

(b) Secondary Codices:


It is of course obvious that all the information we can gather regarding the text of these early codices is of the utmost importance for the textual criticism of the Koran. This in the absence of any direct manuscript evidence 41 gives us our sole witness to the types of text which ‘Uthman’s standard text superseded. It is possible, as we have already seen, that in choosing the Medinan text tradition for canonization ‘Uthman chose the best of the texts available. We can never know this for certain the one way or the other unless the unexpected happens and we recover some considerable portion of one of the rival texts. A collection of the variants still surviving from the old codices is our sole means of forming any judgment as to the type of text they presented.

The question arises, of course, as to the authenticity of the readings ascribed to these old codices. In some cases it must be confessed there is a suspicion of readings later invented by the grammarians and theologians being fathered on these early authorities in order to gain the prestige of their name. This suspicion is perhaps strongest in the case of distinctively Shi‘a readings that are attributed to Ibn Mas‘ud, and in readings attributed to the wives of the Prophet. It is also felt in regard to some of the readings attributed to Ibn ‘Abbas, who as the “Ubermensch des tafsir” (Goldziher, Richtungen, 65) tended to get his authority quoted for any and every matter connected with Koranic studies. On the whole, however, one may feel confident that the majority of readings quoted from any Reader really go back to early authority.

The more difficult question is that of defective transmission. Occasionally in reading the Commentaries one finds a reading that is commonly known as coming from a certain early Reader attributed to quite another source. Where authorities can be weighed it is generally possible to decide which attribution is correct, but in cases where a variant is quoted by only one source which is otherwise known for the carelessness of its citation of authorities, one can never be sure that that particular variant is correctly attributed to the Reader given. A similar problem of accurate transmission naturally attaches to the variants themselves. Being uncanonical variants there was none of the meticulous care taken over their transmission such as we find for the canonical readings, and we not infrequently have various forms of the variant attributed to the same Reader in different sources. In such cases nothing can be done but to give them all in the hope that further information may enable us to decide between them. Some of the variants in the form in which they have survived to us seem linguistically impossible, and in certain cases this has been noted in the source which quotes the variant. The defect is doubtless due to faulty transmission, and it is possible that some scholar may even now spot where the corruption lies and restore us the original reading.

Bergsträsser in his preliminary collection of the uncanonical readings of Ibn Mas‘ud and Ubai’42 made an attempt to estimate the value of these two texts as compared with the ‘Uthmanic text. With the increase of material one feels less inclined to venture on such a judgment of value. It is true that in some cases the uncanonical variants from these old codices may be interpreted as improvements on the ‘Uthmanic text, as e.g., bῑma [regarding] instead of bimithli mā [like, as] in II, 137/131 may have been suggested by motives of piety: or expansions thereof as in II, 275/276 where the added yaumu‘l-Qiyama [the day of resurrection] may be regarded as an explanatory inflation. In such cases the ‘Uthmanic text would seem to be the more primitive text which the other types assume as their basis. But on the other hand there are equally many cases where the facts point the other way. For instance in II, 9/8 the ‘Uthmanic yakhadi’ūna [endeavor to deceive] may be regarded as an attempt to soften the idea of deceiving Allah which is suggested by the alternative reading yakhda‘ūna [deceive]; or allāhi [to God] in II, 196/192 may have been set for theological reasons instead of lilbayti [to the house of God] or the present form of II, 240/241 may be taken as an expansion of the simpler form given in the other codices. Bergsträsser drew attention to the number of cases where the variant in the old codices was merely a synonym for the word in the text but the cases are about evenly balanced for the simpler word being in the ‘Uthmanic text or in the variant.

Remembering that we have in our hands only a very small portion of the variants from these codices, and that what we have consists in the main only of such variants as were not too unorthodox, we may take the following collections as the base for our further investigation into the earliest stage in the formation of the text of the Koran.

The material which follows is taken from the writer’s collections made with a view to a critical text of the Koran. They will of course appear in their place in the apparatus criticus to that text when it appears, but the assembling of them here under the individual names was essential that scholars might be able to deal critically with the evidence of each codex as a whole. The main sources from which the variants have been drawn are:

Abu Hayyan, Al-Bahr al-Muhitt, 8 vols., Cairo 1328.

Alusi, Ruh al-Ma‘ani fi Tafsir al-Koran wa Sab’ al-Mathani, 30 vols., Cairo,

n. d.

Baghawi, Ma’alim at-Tanzil, 7 vols., Cairo 1332 (On margin of the Tafsir


Baidawi, Anwar at-Tanzil wa Asrar at-Ta‘wil, 5 pts., Cairo 1330.

Balawi, Kitab Alif Ba; 2 vols., Cairo 1287.

Banna’, Ithaf Fudala’ al-Bahar fi’l-Qira‘at al-Arba’ata ‘ashar, Cairo 1317.

Fakhr ad Din ar-Razi, Mafatih al-Ghaib, 8 vols, Cairo 1327.

Farra’, Kitab Ma‘ani al-Koran. Ms. Stambul, Nuru Osmaniya 459.

Ibn al-Anbari, Kitab al-Insaf, ed. Gotthold Weil, Leiden 1913.

Ibn Hisham, Mughni al-Labib, 2 pts., Cairo 1347.

————, Tahdhib at-Tawadih, 2 pts., Cairo 1329.

Ibn Jinni, Nichtkanonische Koranlesarten im Muhtasab des Ibn Ginni, von G.

Bergsträsser, Munchen 1933.

Ibn Khalawaih, Ibn Halawaihs Sammlung nichtkanonischer Koranlesarten,

herausgegeben von G. Bergsträsser, Stambul 1934.

Ibn Manzur, Lisan al-‘Arab, 20 vols., Cairo 1307.

Ibn Ya‘ish, Commentary to the Mufassal, ed. Jahn, 2 vols., Leipzig 1882.

Khafaji, ‘Inayat al-Qadi wa Kifayat ar-Radi, 8 vols., Cairo 1283.

Marandi, Qurrat ‘Ain al-Qurra’, Ms. Escorial 1337.

Muttaqi al-Hindi, Kanz al-‘Ummal, 4 vols., Hyderabad 1312.

Nasafi, Madarik at-Tanzil wa Haqa’iq at-Ta‘wil, 4 vols., Cairo 1333.

Nisaburi, Ghara’ib al-Koran (on the margin of Tafsir at-Tabari).

Qunawi, Hashiaala l-Baidawi, 7 vols., Stambul 1285.

Qurtubi, Al-Jami‘li Ahkam al-Koran, 2 vols. (all so far published), Cairo 1935.

Shawkani, Fath al-Qadir, 5 vols., Cairo 1349.

Sibawaih, Le Livre de Sibawaih, ed. Derenbourg, 2 vols., Paris 1889.

Suyuti, Al-Itqan fi ‘Ulum al-Koran, ed. Sprenger, Calcutta 1857

Suyuti, Ad-Durr al-Manthur fi ’t-Tafsir al-Ma‘thur, 6 vols., Cairo 1314.

Suyuti, Al-Muzhir, 2 vols., Cairo 1282.

Tabari, Jami’ al-Bayan fi Tafsir al-Koran, 30 vols., Cairo 1330.

Tabarsi, Majma‘al-Bayan fi ‘Ulum al-Koran, 2 vols., Teheran, 1304.

‘Ukbari, Imla’ fi ‘l-I’rab wa ‘l-Qira’at fi Jami‘al-Koran, 2 pts., Cairo 1321.

‘Ukbari, I‘rab al-Qira’at ash-Shadhdha, MS Mingana Islamic Arabic 1649.

Zamakhshari, Al-Kashshaf, ed. Nassau Lees, Calcutta 1861.


‘Abdallah b. Mas‘ud (sometimes quoted in the sources as ‘Abd Allah and sometimes as Ibn Umm ‘Abd)43 was a Companion and one of the early Muslims who could boast that he had joined the faith earlier than ‘Umar. As a youth he had herded cattle for ‘Uqba b. Abi Mu’ait and so was sometimes referred to contemptuously as the Hudhali slave (Tabari, Annales, 1, 2812). When he became a Muslim he attached himself to the Prophet and became his personal servant. He went on the Hijra to Abyssinia and also to Medina and was present at both Badr and Uhud. It was his boast that he had learned some seventy suras directly from the mouth of the Prophet, and tradition has it that he was one of the first to teach Koran reading (Ibn Sa’d, III, i, 107). He seems not to have been a great success when tried in an official capacity, but at Kufa, to which the caliph sent him, he became famous as a traditionist and as an authority on the Koran. Tradition tells that he was one of the four to whom Muhammad advised his community to turn for instruction in the Koran.44 It was doubtless his close personal contact with the Prophet over so many years that gave such prestige to his opinions on Sunna and Koran.

We have no information as to when he began to make his codex. Apparently he began to collect material during the lifetime of the Prophet and worked it up into codex form when he was established at Kufa and was looked to as the authority on Koranic matters. At any rate we find his codex in use there and followed by the Kufans before the official recension was made by ‘Uthman. When ‘Uthman sent to Kufa the official copy of his standard text with orders that all other texts should be burned, Ibn Mas‘ud refused to give up his copy, being indignant that the text established by a young upstart like Zaid b. Thabit should be given preference to his, since he had been a Muslim while Zaid was still in the loins of an unbeliever.45 There seems to have been considerable difference of opinion in Kufa over this question of the codex, some accepting the new text sent by ‘Uthman, but a great many continuing to hold by the codex of Ibn Mas‘ud,46 which by that time had come to be regarded as the Kufan text. The strength of the position of his codex in Kufa is well illustrated by the number of secondary codices of which some information has come down to us and which followed the text of Ibn Mas’ud. It was from its vogue in Kufa that his codex came to be favored by Shi‘a circles, though one is not disposed to accept as genuine all the Shi’a readings that are attributed to his codex, nor indeed those found in Sunni sources in favor of Ahl al-Bait.

It was well known in the early days of Islam that one peculiarity of Ibn Mas‘ud’s codex was that it did not contain suras I, CXIII and CXIV, i.e. the Fatiha, which is an opening prayer to the book, and the Mu’awwidhatani with which it ends.47 Modern scholarship on quite other grounds holds that these were not origianlly part of the Koran but are of the nature of liturgical additions. That Ibn Mas’ud knew of these passages as used liturgically is evident from the fact that we have preserved to us notes of words in which he differed from the customary way of reading them.

A second peculiarity equally well known was that the order of suras in his recension differed considerably from that of ‘Uthman’s recension. Two lists giving this sura order have been preserved to us, which do not, however, entirely agree with one another. The earlier is that given by Ibn an-Nadim (377)48 in the Fihrist p. 26 (ed. Flügel) on the authority of al-Fadl b. Shadhan (d. before 280), which runs as follows:

2, 4, 3, 7,6, 5, 10,49 9, 16, 11, 12, 17, 21, 23, 26, 37, 33, 28,24,8, 19,29, 30, 36, 25, 22, 13, 34, 35, 14, 47, 31, 39, (40 bis 46), 40, 43, 41, 46, 45, 44, 48, 57, 59, 32, 50, 65, 49, 67, 64, 63, 62, 61, 72, 71, 58, 60, 66, 55, 53, 51, 52,50 54, 69, 56, 68, 79, 70, 74, 73, 83, 80, 76, 75, 77, 78, 81, 82, 88, 87, 92, 89, 85, 84, 96, 90, 93, 94, 86, 100, 107, 101, 98, 91, 95, 104, 105, 106, 102, 97, 103, 110, 108, 109, 111, 112.

The suras missing here are 1, 15, 18, 20, 27, 42, 99, 113, 114. That suras 1, 113, 114 were omitted in his codex we have already seen, but as variants from all the others omitted here are found quoted from him, the material of which they are composed must have been in his codex. Indeed they are all to be found in the list of his suras given in the Itqan. When we examine these missing suras we discover that 15 is the last in the ALR [Alif, Lam, Ra] series; 18 comes immediately before the KHY’S [Kaf Ha, Ya, Ain, Sad] sura (19) and is suspected to have had some connection therewith (Goossens in Der Islam XIII, 211); 20 is the sole TH [Ta, Ha] sura; 27 is the TS [Ta, Sin] sura which breaks in between two TSM [Ta, Sin, Mim] suras; 42 is the HM ‘SK [Ha, Mim Ain, Sin, Ḳaf] sura which breaks into the HM [Ḥa, Mim] suras, so that one may suspect that there is something behind their omission in the Fihrist. Yet in view of the fact that the missing suras are in the list in the Itqan, and the Fihrist itself expressly says that it reckoned 110 suras whereas there are only 105 in the list, the probability is that the list as we have it has been defectively written.

The second list is in the Itqan of as-Suyuti (ed. Calcutta, p. 151), quoting from Ibn Ashta a statement going back to Jarir b. ‘Abd al-Hamid (d. 188), who related traditions from al-A‘mash and others of Ibn Mas’ud’s school.51 This list runs:

2, 4, 3, 7, 6, 5, 10, 9, 16, 11, 12, 18, 17, 21, 20, 23, 26, 37, 33, 22, 28, 27, 24, 8, 19, 29, 30, 36, 25, 15, 13, 34, 35, 14, 38, 47, 31, 39, 40, 43, 41, 42, 46, 45, 44, 48, 59, 32, 65, 68, 49, 67, 64, 63, 62, 61, 72, 71, 58, 60, 66, 55, 53, 52, 51, 54, 56, 79, 70, 74, 73, 83, 80, 76, 77, 75, 78, 81, 82, 88, 87, 92, 89, 85, 84, 96, 90, 93, 86, 100, 107, 101, 98, 91, 95, 104, 105, 106, 102, 97, 99, 103, 110,108,109,111, 112, 94.

Here we find missing besides the expected 1, 113, 114, the suras 50, 57, 69, for whose omission no reason can be suggested save that they may have dropped out by scribal error. Well known variants are quoted from each of them and they are all in the list in the Fihrist. The two lists correspond sufficiently closely for us to supply the missing members of the one from the other, and we may treat them as variants of a common tradition as to the sura order in Ibn Mas’ud’s codex.

The value of this tradition is another matter.52 It is not a priori likely that the arrangement of material in any of the rival codices would have followed the same combination into suras as in the text established for ‘Uthman by Zaid b. Thabit. In the accounts of that official recension we find bits of material coming in and the committee considering the most appropriate place to put them, and it is against all probability that the composite suras made up of bits of Meccan and bits of Medinan material, of very different date and provenance, would have been fitted in exactly the same way by different collectors. Neither is it likely that the different collectors would have chosen the same titles for the suras. The traditions as to the sura order, in the case of this and of other of the old codices, come from persons who were familiar with the ‘Uthmanic sura order, but knew that the material was differently disposed in the other codices, and so constructed a sura list to express the difference.53

The variant readings which follow are necessarily arranged according to the order of the present official text. Sometimes in the sources the variant is expressly said to come from the codex of Ibn Mas‘ud. More often it is merely given as a reading (harf or qira’a)of Ibn Mas‘ud. Occasionally also readings are given as coming from the Companions of Ibn Mas’ud, but as these obviously represent the tradition as to his text they are included here. In view of the great importance of the readings of Ibn Mas‘ud and Ubai, all readings from them that survive are included in the lists even where they do not depend on a different consonantal text from that of ‘Uthman. It has also seemed worth while to note the places where they are specially recorded as supporting the textus receptus.


Ubai b. Ka’b was one of the Ansar who after the Prophet’s coming to Medina served as his secretary.54 He is said to have been the one who wrote out the treaty with the people of Jerusalem (Ibn ‘Asakir II, 329). He was one of those who specialized in the collection of revelation material and figures among the four to whom Muhammad is said to have advised his community to turn for Koran instruction. In some respects his authority on Koranic matters was even greater than that of Ibn Mas’ud. He was known as Sayyid al-Qurra’ [“the Chief of the Readers of the Koran”], the Prophet is said to have referred to him as aqra’a ummati [“he who made the Muslim community recite”] and to have been commanded by Allah to hear Ubai recite to him portions of revelation, which probably means that Ubai was the repository of certain material of a legislative character which the Prophet would have him read over to him from time to time.

We have no knowledge of when his codex was made, but we do know that before the appearance of the ‘Uthmanic standard text his codex had already come into vogue in Syria. Ibn Abi Dawud ... has a story of how some Syrians made a codex and came to Medina to check it over with Ubai, and though at that time the standard text was in use, no one dared to dispute the peculiar readings that were derived from Ubai. He seems to have had an important part in the actual work of producing the canonical text for ‘Uthman at Medina. His name appears in these stories in various connections but the whole account is too confused to enable us to understand precisely what his relation to the standard text was.55

His codex is definitely stated to have been among those destroyed by ‘Uthman. Its sura order was reported to have differed from that of ‘Uthman’s, and as in the case of Ibn Mas‘ud’s codex, we have two lists of his sura order. According to the Fihrist, p. 27, his order was—

1, 2, 4, 3, 6, 7, 5, 10, 8, 9, 11, 19, 26, 22, 12, 18, 16, 33, 17, 39, 45, 20, 21, 24, 23, 40, 13, 28, 27, 37, 34, 38, 36, 15, 42, 30, 43, 41, 14, 35, 48, 47, 57, 58, 25, 32, 71, 46, 50, 55, 56, 72, 53, 68, 69, 59, 60, 77, 78, 76, 75, 81, 79, 80, 83, 84, 95, 96, 49, 63, 62, 65, 89, 67, 92, 82, 91, 85, 86, 87, 88, 64, 98, 61, 93, 94, 101, 102, al-Khal’, al-Hafd, 104, 99, 100, 105, 107, 108, 97, 109, 110, 111, 106, 112, 113, 114.

In this list are missing suras 29, 31, 44, 51, 66, 70, 73, 74, 90, 103, but we have two extra suras, al-Khal’ and al-Hafd. As, however, we actually know of variants from him in all of these save 103, the probability is that the material of them formed part of his codex.

The other list is in the Itqan 150, 151, which gives the order—

1,2,4,3,6,7,5, 10,8,9, 1, 19 26,22, 12, 15, 16, 33, 17,39,20, 21, 24, 23, 34, 29, 40, 13, 28, 27, 37, 38, 36, 15, 42, 30, 57, 48, 47, 41, 46, 50, 55, 56, 72, 53, 70, 73, 74, 44, 31, 45, 52, 51, 68, 69, 59, 60, 77, 78, 75, 81, 65, 79, 64, 80, 83, 84, 95, 96, 49, 63, 62, 66, 89, 90, 92, 82, 91, 86, 87, 88, 61, 98, 93, 94, 101, 102, 103, 104, 99, 100, 105, 106, 107, 108, 97, 109, 110, 111, 112, 113, 114.

From this are missing suras 18, 25, 32, 35, 43, 54, 58, 67, 71, 76, 85, but all these save 54 are to be found in the list of the Fihrist, and we know of variants from 54. It is evident that we cannot place any reliance on the lists, which as in the case of the lists for Ibn Mas’ud’s codex, must be regarded as later formations not based on the original codex.

His codex seems not to have been the source of any secondary codices, though it would seem to have been copied, and if we are to believe the Fihrist, a copy of it was still extant in the time of Ibn Shadhan, i.e., in the middle of the third Islamic century. There is also a story of how Ibn ‘Abbas presented a man with a codex written according to the qira’a of Ubai (Durr IV, 170).

There are some tales about the survival of Ubai’s codex, but it is clear that it perished early, for there is the account in Ibn Abi Dawud, p. 25 of how some people from ‘Iraq came to Ubai’s son Muhammad and asked to consult his father’s codex, but Muhammad had to tell them that was impossible as the codex had been seized by ‘Uthman. Bergsträsser was inclined to think the readings from Ubai’s codex less significant than those of Ibn Mas‘ud, but the truth seems to be that his codex, not having the support of a great metropolitan center like Kufa, left permanent record of less of its peculiar readings than is the case of the codex of Ibn Mas’ud.

It is remarkable how often his variants agree with those of Ibn Mas‘ud against the ‘Uthmanic codex. One suspects that sometimes there has been a confusion in the tradition, and that readings of the one have been attributed to the other. This is certainly so when we find a single source attributing to Ubai a reading that is known as a peculiarity of Ibn Mas‘ud, and it is curious that al-Marandi’s Qurat’Ain al-.Qurra’, which is our richest source for Ubai’s readings, attributes to him a great many readings, which are found elsewhere recorded only for Ibn Mas’ud.

In the MS of Ibn Abi Dawud only four readings from Ubai’s codex are listed, but as he is quoted in the Commentaries for so large a number one suspects that some leaves were missing in this place in the original from which the Zahiriya MS was copied.


There is persistent tradition among the Shi‘as that ‘Ali b. Abi Talib was the first after the death of the Prophet to make a collection of the material of the Koran, and even Sunni sources know that he prepared a codex of his own. The most widely accepted form of the story is that after the Prophet’s death, while the Companions were busy about electing a successor, ‘Ali shut himself up in his house and made a vow that he would not put on this outdoor cloak until he had made an assemblage of the Koranic material into a codex. This caused some little comment as he did not come out to pay homage to Abu Bakr the newly elected caliph, but ‘Ali explained his oath, and when the work was finished he packed it up on the back of his camel and brought it to the Companions saying “here is the Koran that I have assembled.”56

There are many variations of the story. Some said that it was only six months after the Prophet’s death that ‘Ali set about making a recension.57 Others say that he sat down and in three days wrote it all out from memory and arranged it in the order in which it was revealed.58 A more interesting embellishment is that when the Prophet was about to die he summoned ‘Ali and told him where the material for the Koran was hidden in a secret place behind his couch, and bade him take it from thence and edit it.59

Although the common story is that ‘Ali’s codex had the suras arranged in some sort of chronological oder (Itqan, 145), quite different arrangement is given by al-Ya‘qubi (Historiae II, 152ff.) according to whom ‘Ali arranged the suras in seven groups,60 each group beginning with one of the seven long suras and called by its name. The schema is:

I. 2, 12, 29, 30, 31, 41, 51, 76, 32, 79, 81, 82, 84, 81, 98 Al-Baqara. 886 verses, sixteen suras.

II. 3, 11, 12, 15, 33, 44, 55, 69, 70, 80, 91, 97, 99, 104, 105, 106. Al-‘Imran. 886 verses, fifteen suras.

III. 4, 16, 23, 36, 42, 56, 67, 74, 107, 111, 112, 103, 101, 85, 95, 27. An-Nisa’. 886 verses, seventeen suras.

IV. 5, 10, 19, 26, 43, 49, 50, 54, 60, 86, 90, 94, 100, 108, 109. Al-Ma‘ida. 886 verses, fifteen suras.

V. 6, 17, 21, 25, 28, 40, 58, 59, 62, 63, 68, 71, 72, 77, 93, 102 Al-An’am. 886 verses, sixteen suras.

VI. 7, 14, 18, 24, 38, 39, 45, 47, 57, 73, 75, 78, 88, 92, 110. Al-A’raf. 886 verses, sixteen suras.

VII. 8, 9, 20, 35, 37, 46, 48, 52, 53, 61, 64, 65, 83, 113, 114. Al-Anfal. 886 verses, sixteen suras.

This makes only 109 suras actually recorded, those missing being 1, 13, 34, 66, and 96. Unfortunately, no reliance can be placed on it for it is obviously dependent on the sura divisions of the ‘Uthmanic text, which ‘Ali’s codex was hardly likely to follow, and of course it contradicts the other tradition that he arranged the material chronologically. This tradition of chronological arrangement is incidentally supported by the fact that there lingered for long the knowledge that in ‘Ali’s codex the first suras were 96, 74, 68, 73, 111, 81 (Itqan, 145). In any case the above list is not accurate, for division I which is said to contain 16 suras contains only 15, division II which is said to have 15 actually has 16, division III said to contain 17 has only 16, and division III said to contnin 16 has only 15.

When ‘Uthman made his official recension ‘Ali seems to have warmly supported it, saying that had he been in ‘Uthman’s position he would have done the same thing. It would appear that he gave up his own codex in favour of the new edition and it was probably burned at that time. Had it survived it is quite certain that the Shi’as would have adopted it as their standard codex, whereas in Shi‘a hands we find only copies of the ‘Uthmanic text even when they are said to have been written by ‘Ali or one of this sons,61 and the one pre-‘Uthmanic codex whose readings seem to have been favored by the Shi‘as is that of Ibn Mas’ud.62

Even when in later literature we have references to the codex of ‘Ali, as when Ibn Sirin (d. 110) is said to have written to Medina for some information regarding it, or when ath-Tha’labi in his Tafsir (Sprenger, Leben III, xliv) notes that in ‘Ali’s codex sura II had 286 verses, or when Ibn an-Nadim, Fihrist 28, tells us that a copy lacking a few leaves was preserved in the ‘Alid family for generations, the probability is that the reference is to a copy of the ‘Uthmanic text made by or for ‘Ali rather than to his own pre ‘Uthmanic text.

Consequently, we have to bear in mind that all uncanonical variants quoted from ‘Ali, while they may go back to variant readings that he remembered were in his own recension of the Koran, may on the other hand be merely his interpretation of the ‘Uthmanic text.

Ibn Abi Dawud lists ‘Ali’s codex, apparently meaning his noncanonical codex, but quotes only one reading from it.

You can support the site and the Armed Forces of Ukraine by following the link to Buy Me a Coffee.

If you find an error or have any questions, please email us at admin@erenow.org. Thank you!