Which Koran? (Part II)

The Significance of Koranic Variation
by Ibn Warraq
(March 2008)



The variants between Korans (as set out last month) are not trifling, and are, in fact, of great significance. The problem is to work out what significance, and this proves to be no easy matter. For a flat-footed fundamentalist like Maududi, the admittance of any variant whether in the extant printed Korans available in the Islamic world or in the manuscripts like the Samarqand Quran or those recorded in the Hadith, commentaries and grammars is, of course, devastating. Variants constitute an irrefutable, knock-down argument against his absurdly rigid position (as quoted in part I), a position not held by all Muslim scholars, however. I believe their significance lies in a wider context, in their profound implication for the sources of the rise of Islam, for the forging of Islamic identity, for the genesis of the Koran itself, for Islamic Jurisprudence, for the so-called oral tradition, and for the history of the Arabic language and orthography. I shall leave these implications for later.

Even simply on their own terms, variants do result in significant differences in meaning which in turn have consequences for Islamic practice, ritual and belief. Thus, the variants in the printed Korans are not trivial. As an example of a variant reading on the level of vocalisation though not of the underlying graphic shape (or, in Arabic, rasm), there are the last two verses of Sura LXXXV: 21-22: (21) bal huwa qur’änun majïdunfï lawhim mahfüzun or mahfuhzin. The last syllable is in doubt. The Hafs Koran has, as we saw earlier, mahfuhãin, the genitive, giving the meaning, “It is a glorious Koran on a preserved tablet.” This is a reference to the fundamental Muslim doctrine of the Preserved Tablet. But the Warsh transmission has the nominative ending -un, and we get “It is a glorious Koran preserved on a tablet.” Did the doctrine arise out of the reading, or did the doctrine influence the choice of the reading?

In Sura III verse 13, there is much ambiguity as the exact reference of the pronoun is not clear:

Bell: “You have already had a sign in two parties which met, one fighting in the way of Allah, another unbelieving, who saw them with their eyes twice as many as they were ….”

Yusuf Ali: “There has already been for you a sign in the two armies that met (in combat): One was fighting in the cause of God, the other Resisting God; these saw with their own eyes Twice their number.”

Arberry: “There has already been a sign for you in the two companies that encountered, one company fighting in the way of God and another unbelieving; they saw them twice the like of them, as the eye sees ….”

This verse is a said to be a reference to the miracle of the battle of Badr, when Muslims putatively defeated forces twice their own number. However, this interpretation is much easier if we read the verb as saying “you saw them” tarawnahum, as in the Warsh reading, and not yarawnahum (they saw them) as in the Hafs reading. Warsh gives us a miracle, Hafs gives us a confusion of pronouns.

Ignaz Goldziher, one of the creators of modern Islamic Studies, showed how Hadith, Muslim Tradition, reflected “the social, political and religious ideals of transmitters themselves and of the societies or groups they served as spokesman. By Sunna was to be understood, not the inherited instruction of the Prophet, but the ius consuetudinis of a group or party, large or small. By hadith is meant the vehicle of that sunna, a report, verbal or written, conveying a description of the relevant practice, opinion or custom approved by the disseminators of the report.”[1] Influenced by Goldziher’s work, Joseph Schacht elaborated a thesis that “rather than spreading out from an original centre at Medina, Islamic Law originated in the provinces. Reference of the Sunna to the Prophet was the end rather than the beginning of a process. Its purpose was to verify some local legal viewpoint. In other words, the Sunna differed and was differently defined from region to region.”[2]

Of course, the conclusions of scholars like Goldziher and Schacht are equally applicable to Koranic variants, many of which are known to us through hadith, rather than extant Koranic manuscripts. In other words, the variants reflect the ideology, as Burton shows, of groups that wish to argue for their own viewpoint, to establish a ruling, to settle conflict of sources. For example the rite of Tawäf, going round the two hills of Safä and Marwa during Hajj, Pilgrimage, are considered obligatory by certain Muslim jurists despite a certain ambiguity in Surah II.158, which is interpreted by some to mean that the Tawäf was optional. Others still also regarded the Tawäf as optional but this time the view “was explicitly derived from the variant reading of II.158 transmitted in the muæùaf of `Abdullah Ibn Mas‘üd.”[3]

Burton argues that when practice was at variance with the Koran, the partisans of the practice appealed to the Sunna of the Prophet, their opponents “improve the wording of the Qur’än, inserting a word and appealing to the authority of a Companion of the Prophet , from whom not merely a variant reading, but a variant Qur’än had apparently been transmitted. The alleged variant reading unmistakably proceeded from one of two rival and competing interpretations. To that extent the reading arose at a secondary stage.”[4]

There is a similar exchange of argument and counter-argument concerning the penalty for breach of oaths [Surah V.89], a three days’ fast, ending as before with an appeal to a variant reading from Ibn Mas‘üd. Al-Shäfi‘ï argued that the Koran did not stipulate if the fast should be consecutive, hence Muslims were free to choose consecutive or separate days. Ùanïfs argued that the fast should be consecutive, as a variant reading of Ibn Mas‘üd indicates. The same variant reading was attributed to Ubayy.[5] Ubayy also had a very significant variant reading of Surah IV.24 concerning the Muslim Law on marriage; only with his interpolation does IV.24 “sanction the doctrine of mut’a, or temporary marriage, rejection of which was elsewhere being propounded on the basis of information from a third Companion of the Prophet as a part of the Sunna. Evidently the Qur’än, in the form of the Ubayy reading, is playing the role of a counter-sunna, rather, a counter-exegesis, the function of the Ubayy interpolation to gloss and bring out the full meaning of the root of samta`tum, mt`.”[6]

As Al-Suyütï put it, “The differences in the readings indicate the differences in the legal rulings.”[7] Thus we have two opposing doctrines – the invalidation of the ritual purity [wudü’] and the contrary doctrine – depending on how we read a certain word in IV.43 and V.9 as lämastum or lamastumügel have the “defective” writing, with the long vowel after the letter läm indicated by a dagger alif, only the Flügel has the plene alif. Similarly, we have two opposing doctrines depending on how we read II.222 –yathurna or yattahirna -concerning the permissibility of sexual intercourse with a menstruating woman at the expiry of her period but before she has cleansed herself. [8]

Finally we have the example of V.6, as al- Suyütï says, “The verse was revealed to sanction two distinct legal doctrines: arjulakum – enjoined the washing of the feet, ajulikum -permitted the wiping of the feet.”[9] Herbert Berg summarises the larger significance of these two interpretations, “Al-Tabarï adduces 47 hadiths which seek to clarify the expression wa-arjulakum ila al-ka`bayn (and your feet to the ankles) of Quran V:6. The first 27 hadiths read the passage as arjulakum (accusative); the other 20 hadiths read the passage as arjulikum (genitive)….Goldziher would see in these two sharply divided sets of hadiths the vestiges of a later debate within the Muslim community about the proper form of wudü’ (ablution) that has been projected back to the earlier generations of Muslims. Schacht might trace this ablution debate in other texts to determine the relative chronology and the provenance of the hadiths. He might also, along with Juynboll, seek a common link to help date the debate. ‘Ikrima is a candidate since he appears in five hadiths, though the isnads form more of a spider pattern. Wansbrough would abandon such use of the isnad except to note that their presence implies that the 47 hadiths reached their final form after 200 A.H.[9 Century, C.E.] Moreover, the hadiths are primarily halakhic and masoretic: they contain pronouncements from the Prophet, his Companions and their Successors and have recourse to variant readings and grammatical explanations. Their presence implies a relatively late date as well.”[10]

Burton following Al- Suyütï argues that “the majority of variant readings came to be regarded as little more than exegeses that had gradually crept into the texts transmitted from the Companions.”[11]  While the latter observation may well explain some of the mechanism of how the Companion texts came into being, I would go further and doubt the very existence of Codices belonging to the Companions; they have been conjured up by exegetical hadiths. In other words, the question of variants leads inexorably to the questions of the authenticity of hadiths relating them.


In his translation of the Koran, the British convert to Islam, Marmaduke Pickthall [1930] had the scholarly courtesy to tell us that the copy of the Koran [mushaf] that he had used was a lithograph copy of that written by al-Hajj Muhammad Shakarzadeh at the command of Sultan Mahmud of Turkey in 1246 A.H. [circa 1830 C.E.].[12] It does not tell us enough, however. We still do not know which Koran, which manuscript, the scribe Al-Hajj had relied on. The situation is even worse with other translators of the Koran. George Sale [1734] in his Note to the Reader of his translation [13] tells us, “As I have had no opportunity of consulting public libraries, the manuscripts of which I have made use throughout the whole work have been such as I had in my own study,….” But he does not specify which manuscripts he had in his possession.

J. M. Rodwell [1861][14] used Gustav Flügel`s edition. Flügel published the Arabic text of the Koran in 1834, and a concordance in 1842. We no longer know on which Arabic manuscripts Flügel depended for his published text, but when Jeffery and Mendelsohn examined the orthography of the Samarqand Quran Codex, a ninth century C.E. work produced in Iraq, they found something astonishing : “The most striking fact in this list [of verse endings] is the number of coincidences of verse endings in the Codex  with those adopted by Flügel in his text ….Since we are entirely in the dark as to the source from which Flügel drew his verse divisions, these coincidences are significant. Flügel’s verse endings agree with none of the known systems whose tradition has come down to us, nor with any that we have been able to trace in the Masoretic literature under the section Ru’üs al-Ayy, and it has been generally assumed that he selected his verse endings on an arbitrary system of his own. The number of agreements between his system and that followed in this Codex, however, suggest that he may have been following the system of some MS in his possession which may have followed some divergent Oriental tradition. It must be admitted, however, that the table Shebunin [Russian scholar who studied the original manuscript in St. Petersburg in 1891] constructs of the divergences between the Samarqand Codex and the Flugel text in the matter of verse endings, is equally long and imposing, so that it is obvious that the question of Flügel’s system of verse division awaits further elucidation.”[15]

At any rate, Flügel’s edition remained the standard one for reference for all of the nineteenth century.

While E. H. Palmer [1880],[16] and N. J. Dawood [1956][17] do not indicate which Arabic text they were using, Yusuf Ali [1934] says he mainly used the “Egyptian edition published under the authority of the King of Egypt” for his numbering of the verses; there is no indication if he used the same edition for the translation itself.[18]

A. J. Arberry [1964],in the introduction[19] to his translation, makes the extraordinary claim worthy of an Islamic fundamentalist,” ….[T]he Koran as printed in the twentieth century is identical with the Koran as authorized by ‘Uthmän  more than 1300 years ago.”[20] One wonders how Arberry knows that the present printed Koran [The Egyptian version of 1342 A.H.?] is identical to the so-called ‘Uthmänic one; did he look at and compare dated manuscripts that can be said to be genuinely ‘Uthmänic? No wonder Arberry does not feel obliged to reveal which Arabic text he used, let alone which manuscript.

Régis Blachère in his French version[21] used the Arabic text of the Cairo edition of 1342 A.H./1923 C.E.[22] However, the Cairo edition is not based on a comparison of  manuscripts but a comparison of readings in written sources such as hadiths, Koranic commentaries, lexica, and so on, but ultimately derived from the reading of Hafs (805) from ‘Äsim (744), with a reliance on an Oral Tradition about the orthography of the Koran. Again, manuscripts do not seem to have played a significant role at arriving at a Koranic text. I shall come back to the 1342 Cairo text later.

Admittedly, some of the above translations were meant for a general public, but so was Gideons` Bible, and yet the latter gives the list of previous translations consulted, and the original texts used; for Hebrew the celebrated R. Kittel edition of Biblia Hebraica was referred to, and for the Greek, the 23rd edition of Nestle Greek New Testament.[23] As for the Biblia Hebraica itself, in their forward to the New Edition [1977], Wilhelm Rudolph and Karl Elliger write, “There is no need to defend the use of the Leningrad Codex B 19 A (L) as the basis of the Hebrew Bible, whatever one may think of its relationship to the Ben Asher text. …In any event, L is still `the oldest dated manuscript of the complete Hebrew Bible ‘ [dated 1009  or 1008 C.E.].[24]  If we consult the Greek New Testament edited by F. H. A. Scrivener [1903] we are informed on the title page that the text utilized is the one established by Stephanus in 1550 C.E. with variants from Bezae, Elzevir, Lachmann, Tischendorf, Tregelles, and Westcott -Hort.[25] Here we learn which texts were examined for translations, and if we go to the texts themselves we are immediately apprised of the manuscript used.

The situation is different and, at present, far more difficult in the world of Islamic Studies.  A Western scholar simply does not have a complete or comprehensive catalogue of all the extant Koranic manuscripts round the world at his disposal. Many collections remain uncatalogued such as the Damascus Korans of Istanbul.[26] There are also many private collections not inventoried or are inaccessible to scholars. There are scattered references to Koranic manuscripts in various articles in the two editions of the Encyclopaedia of Islam, such as those by Bernhard Moritz in the article Arabia, sub-section Arabic Writing in EI Ist edn., or Dominique-Sourdel`s  Khatt in the 2nd edn., but no comprehensive treatment of the subject. As Déroche remarks, “The bulk of the material, manuscripts without illumination or in more ordinary hands of later periods, have not even been examined or catalogued in spite of their importance for the study of a wide range of subjects, from popular piety to the diffusion of the book in the Islamic lands.”[27] Déroche’s own article in the Encyclopaedia of the Qur’än [EQ, henceforth]is perhaps the first of its kind, but Déroche also seems unaware of the significance of the variants, since he minimizes them. He writes,” …most of the manuscripts currently known are very close to the canonical text.” And yet adds immediately afterwards the observation, “Some fragments of Hijäzï codices found in San‘ä’  are said to include some textual variants which were not recorded by later literature , and to offer an order of the suras differing from the arrangements of both the canonical text and the codices of Ibn Mas‘üd  and Ubayy.”[28] There is no further discussion of the San‘ä’ finds. It is clear Déroche is not interested in variants and what their wider meaning might be.  Déroche’s fellow contributor to the EQ, Fred Leemhuis, on the other hand, thinks there are variants  and they are important, “Although the concept of the ‘Uthmänic rasm suggests a uniform and invariable text, such uniformity is not presented by most of the oldest extant codices. Considerable variation is found especially in connection with long a and words which in later classical Arabic orthography required a hamza. Even the word qur’an is found spelled as qrn (e.g. in Q 50.1 of the St .Petersburg fragment as reproduced in E. Rezwan, Fruhe Abschriften, 120-1).[29] In addition to their value for study of the Qur’än’s textual history such evidential examples are important for the history of Arabic orthography.”[30]

Even if they have access to the necessary catalogues, it is not certain that infidel researchers will be allowed to examine Koranic manuscripts with their skeptical, profane eyes. Then there is the additional problem of the dating of Koranic manuscripts; polemics and prejudice have penetrated this field as well. Presumably no revisionist who follows Wansbrough in his argument that the Koran was not put into its final form until the 9th Century C.E. would accept an early date for any complete Koranic manuscript. There are indeed some leaves, folios and Koranic inscriptions that have been dated to the Eighth Century C.E. or earlier  but no complete Korans that can be dated with confidence to earlier than the Ninth Century C. E.

But even if these difficulties are resolved, one has the impression that Western scholars, on the whole, are simply not interested in examining Koranic manuscripts for the sake of variants, to see what they might teach us about the history of that text, about the history of Arabic orthography, and about the history and nature of the Arabic language.  Most scholars have uncritically accepted the Islamic version of the history of the text, and even believe, as Arberry does, that “the Koran as printed in the twentieth century is identical with the Koran as authorized by ‘Uthmän more than 1300 years ago.”

Werner Diem [born 1944] presumed to write the history of Arabic orthography without, astonishingly enough, looking at a single manuscript! He complacently announced, “Koranic manuscripts, however, have not been looked through, because they generally go back to a time after ‘Uthmän, and because they do not preserve the old orthography as faithfully as the readers did.”[31] How can he know that Koranic manuscripts did not preserve the old orthography without looking at the manuscripts?  And how does one establish, without circularity, what the “old orthography” is in the first place? Brockett would answer, “by consulting the oral Tradition.” Brockett justifies taking the 1342 Cairo text as the basis for comparison with other printed texts that he wished to discuss and examine by pointing to “its clarity and faultless accuracy.” How does he know it is accurate? Where is the original ‘Uthmänic text to which it can compared for its accuracy? The Egyptian scholars responsible for the 1342 Cairo relied on the Oral Tradition about the Orthography of the Koran. Brockett then adds an unclear, even obscure note:

“Unlike the actual written Tradition of manuscript-copies, which had been exposed to an on-going effect over fourteen centuries, and in various locations, this oral Tradition about the graphic form [of the Koran] had begun to be preserved in writing since about the early third century A.H….Moreover, the record of this oral Tradition about the orthography of the Qur’än over the two and a quarter centuries is carefully documented in these written works, implying that the exposure to these centuries had no effect either. For the Egyptian scholars, therefore, the Tradition about the graphic form of the Qur’än stretched right back to the times of the third caliph. The effect of time was, if possible, even less after the writing down of this oral Tradition, and so that the written sources used by the Egyptian scholars date from the 5th century A.H. and later does not diminish their justification in using them. Whatever free rein had existed would have been well before even the first writing down.”[32]

One thing is clear: “No Manuscripts please, we are Koranic scholars!” A host of questions leap to mind. What is an “on-going effect”?  “The actual written Tradition of manuscript-copies”? There is also a naïve faith in oral Tradition. All oral traditions are inherently unstable: you cannot rely upon oral tradition to scientifically reconstruct the events at the dawn of Islam. The chances are that the material transmitted will have undergone a considerable amount of change: people’s memories – the most fragile of human faculties – may have failed them, and their prejudices, even fears of being accused of impiety, will have affected, distorted, or altered the contents of what was being transmitted. Finally, all the thousands of variants that we do have, have also come putatively from oral Tradition, later collected and written down; in other words, oral Tradition can lead to alternative texts to that of the 1342 Cairo Text. How do we choose from among them? In fact, as Gerd- R. Puin has argued, “the existence of variant readings indicates that neither the oral tradition nor the [textual] context were strong enough to rule out the emergence of alternative readings.”

The status that the 1342 Cairo text has acquired as the textus receptus has had unfortunate consequences. Here is how Arabist and linguist Pierre Larcher expresses his regrets:

“In theory, all Arabist linguists know (or should know) that the Koranic text, such as we know it today, is not ne varietur. The tradition of “seven canonical readings,” laid down in the Xth century, is, as we have just suggested, all that remains of a variation, which was much more widespread  and lasted much longer  which wants to pass the thesis, ideologically more convenient than historically confirmed, off as true of an ‘Uthmanic recension’. But, in practice, even this ‘residual’ variation is not linguistically exploited. The exclusive citation of the Cairo edition (
Hafs ‘an ‘Asim, i.e., reading of ‘Asim transmitted by Hafs), recommended, when it is not imposed, by so many journals, has ended by conferring on the Koranic text an untouchability that historically it never had!  A pity, even if the objective assigned by Rudi Paret [34] to the study of qira’ät (“to put to good use the known and still unknown variants with a view to studying the ancient Arabic dialects and, in general, with a view to preparing a historical grammar of Arabic”) seems today excessively ambitious. Nonetheless, the simple collation of the Cairo edition with the Western version (N. Africa, W. Africa) of Warsh ‘an Nafi’ (reading of Nafi’ as transmitted by Warsh) is always fruitful from the linguistic point of view. To give one example: while in the Eastern Koran there are five occurrences of salam, with a short a, of which four occur in combination with the verb ‘alqŒ (IV. 90 and 91; XVI. 28 and 87), in the Western Koran, there are six: the latter reads in fact salam in IV. 94 (where the word is equally combined with ‘alqa) while the former reads salŒm (with long  a) Which suggests  1) that salam and salŒm  are two variants of one and the same word  and 2) that the collocation ‘alqŒ al-sala(Œ)m, has everywhere the sense, not of  “offering peace”, as Masson would have it and with which her translation is sweetened, but really of “offering his submission”, and 3) allows us to hypothesize on the way that the three concepts  of “submission”, “protection / preservation” and “peace” are connected to one another, and subsumed under the root slm. “Peace” is understood negatively as “preserving (the war)” and “protection” as a result of “submission”.”[35]

There is also a worrying tendency to interpret all manuscripts, inscriptions and coins by the standard of the 1342 Cairo Text. For instance, surely it is scientifically unsound to look at a Koranic manuscript and then judge a particular spelling or writing of a word as “incorrect”, or a “scribal error” with the yardstick of the 1342 Cairo text, that is to prejudge the issue. Perhaps the manuscript records a more ancient spelling or an entirely different word or text, and may have some significance which we cannot dismiss a priori as a mere scribal error in the way Jeffery and Mendelsohn do in their otherwise very valuable study of the Samarqand Quran Codex.[36] For example, Jeffery and Mendelsohn note that at Sura II.119 the Samarqand Quran has, intriguingly, some word ending in sara, but this is dismissed as a mistake since there is no such word in the 1342 Cairo text. Potentially significant orthographic variants are similarly brushed away, as at II.171, II.172, III.78, III.88, III.167, III.17, XX.47, and so on. And yet these two scholars note that the Samarqand Quran, “Where it deviates it presents numerous points of interest, so that a detailed comparison is of a certain importance.”[37]

But the most egregious scientific sin committed by Jeffery and Mendelsohn is repeated by Francois Déroche and Sergio Noseda in their facsimiles of the Korans from the British Library in London and the Bibliothèque Nationale in Paris.[38] None of these three Koranic manuscripts is pointed or vowelled. Déroche and Noseda present the original manuscript facsimile on one side, and on the opposite side for comparison they reproduce the 1342 Cairo text with its full panoply of fatahs, shaddas, dots and dagger alifs. Again, for the scientific study of the manuscript the reproduction of the 1342 Cairo text is premature, it only prejudges the issue. In fact, no scholar, with the noble exceptions of Adrian Brockett and Gerd R-Puin, has ever considered how to represent an unpointed manuscript, short of drawing the basic shapes (rasm), so that proper scientific remarks can be directed to it. Here is how Brockett explains:

“Distinctions between
Qur’än readings can be fine and are sometimes a matter of subtle differences in the archaic orthography of the Qur’än, so in order to write about them in English, it is necessary to have a precise system of transliteration. Since, moreover, the vocal form of the Qur’än was not originally indicated in writing, it is useful to have a system which can highlight, where necessary, which elements are vocal and which are graphic. (The term ‘vocal form’, with respect to the Qur’än, is used throughout to signify the consonantal skeleton fully fleshed out with diacritical marks, vowels, and so on. The term ‘graphic form’ refers to the bare consonantal skeleton.)”

Gerd Puin, the German scholar most closely involved with the of the 16000 sheets or parchments  of Koranic fragments discovered in
San‘ä’, Yemen, has uncovered even more variants in the rasm that are not found in the mammoth work of eight volumes, Mu‘jam al-qirä’ät  al-qur’äniyyah,[39] edited in Kuwait recently. This dictionary lists over ten thousand variants, of which about a thousand are variants of or deviations in the rasm. In just 83 sheets of Koranic fragments written in the Hijäzï or Mä’il style, tentatively dated on stylistic grounds to the early 8th century,  Puin discovered at least 5000 deviations in the rasm, never recorded before, not even in the Seven, Ten or Fourteen Readings tolerated by orthodoxy. The Hijäzï Korans show differences in the sytem of counting of verses from the two dozen or so schools of counting; even the sequence of suras is often at variance with not only the Standard Egyptian edition but with the sequence of suras in the Korans of Ibn Mas‘üd and Ubayy. These deviations cannot be dismissed as mere scribal errors (lapsus calami) since the so-called errors are repeated with the same word several times in several fragments studied by Puin. Thus, as Puin emphasizes, it makes common philological sense to look for a rationale. The recurrent deviations from the Standard Egyptian text must be taken seriously, and cannot be swept under the carpet, and attributed to scribal inadequacy.[40]

One of Puin’s conclusions is that though there was an oral tradition, (otherwise the Koranic text could not have been read), there were deliberate changes in the oral tradition of the reading of the Koran. Thus this oral tradition was not very stable or elaborate – changes must have occurred as can be seen in the variant orthography to be found in the San‘ä’ manuscripts. Puin suggests that the long “a” sound could be rendered by the Arab letter yä’, and originally the name in the present Koran that is read as “Ibrähïm” must have been read “Abrähäm“. In other words, at some stage that fact that the long “ä” was rendered with the yä’ was forgotten – hence the so-called oral tradition was not strong or even non-existent.[41]

It is clear that many hundreds of variants, though not all, were invented by Muslim grammarians, philologists, and exegetes of the 3rd and 4th Muslim centuries to explain all sorts of obscurities of the Koran, whether of sense or reference, Koranic grammatical aberrations,[42] or even more seriously, for doctrinal reasons to defend some particular theological position.[43] A kind of ethics of variants had developed by the 9th century A.D., according to which only variants that were not too far from Islamic orthodoxy or doctrines, or not too ungrammatical were to be accepted and preserved. Hence, if there had been startling deviations or variants they would have been suppressed. Thus the variants that do remain are not always very significant. But we need to make a distinction between the variants fabricated by the Muslim exegetes, and the variants to be found in the rasm in manuscripts such as those examined by Puin. The sheer number of variants in the orthography in manuscripts dated as early as 715 A.D. seem to cast doubt on the traditional account of the compilation of the Koran. The San‘ä’ fragments seem to suggest that even in the 8 th century A.D., there was no definitive text of the Koran.

Andrew Rippin has drawn conclusions similar to Puin’s. Referring to the San‘ä’  manuscripts, Rippin writes, “the text contains variant readings of a minor nature that suggest to some scholars that the idea of an oral tradition running  parallel  to the written one cannot be given historical credence. What we may have evidence of is the interpretative nature of the detailed annotations that were added to the text later: that is, that the current text is the product of reflection upon a primitive written text and not upon the parallel transmission of an oral text as the Muslim tradition has suggested.”[44]

Rippin goes on to discuss Sura XXI.4 and 112. Should the two verses begin with the imperative “Say!”, [in Arabic:qul] thus indicating that God is the speaker, or should the word be read as “He said” [qäla]?  What do the printed Korans say? Much depends on the answer to these deceptively trivial questions. Before quoting Rippin in full, here is a rapid survey of some of the translations and Arabic texts.


              ARABIC TEXT:


1.  Saudi: XXI:4: qäla  with plene alif: translated as “He (Muhammad, pbuh)[45] said …”

              XXI.112  qala [to be read as qäla] with defective -dagger-alif  translated as “He (Muhammad, pbuh) said…”


2. Muhammad Ali: XXI:4: qala [to be read as qäla] with defective -dagger-alif translated as “He said:…”

                         XXII:112: qala [to be read as qäla] with defective -dagger-alif translated as “He said:…”


3. Yusuf Ali: XXI:4:  qala [to be read as qäla] with defective -dagger-alif translated as “Say [sic, strictly speaking it should of course be translated, ” He said …” ],:…”

                XXI:112:  qala [to be read as qäla] with defective -dagger-alif translated as “Say [sic]:…”     


4. Istanbul: XXI:4:      qäla with plene alif

                 XXI:112:  qäla with plene alif


5. Iranian: XXI:4       qäla with plene alif

               XXI:112   qäla with plene alif


6. Taj: XXI:4: qala [to be read as qäla]with defective -dagger-alif translated as [in Urdu] “He said …”

          XXI:112: qala [to be read as qäla] with defective -dagger-alif translated as [in Urdu ] “The Prophet said …”


7. Warsh I: XXI:4:  qul  [Say]

                XXI:112:  qul


8. Warsh II: XXI:4:  qul

                 XXI:112:  qul


9. Lebanese: XXI:4:  qäla   with plene alif translated as    “Say” [ sic ]

                  XXI:112: qala [to be read as qäla] with defective -dagger-alif translated as  “Say”[sic]


10. Qalun: XXI: 4:  qul

               XXI:112:  qul     


11. Flügel: XXI:4:  qäla with plene alif

               XXI:112:   qäla with plene alif




 George Sale  XXI:4:        Say

                   XXI:112:     Say


 M. Pickthall   XXI :4:      He saith

                   XXI:112:    He saith 


R. Blachère     XXI:4:   (Notre Apotre ) a dit  = (Our Apostle )[46] said

                    XXI:112:  Dis =Say


A. J. Arberry    XXI:4:     He says

                     XXI:112:  He said



M. Kasimirski   XXI:4:     Dis = Say

                    XXI:112:  Mon Seigneur dit =My Lord Says


D. Masson      XXI:4:    Il a dit =He said

                    XXI:112:   Dis =Say.


N. J. Dawood  XXI:4:        Say

                    XXI :112:  Say


E. H. Palmer    XXI:4:        Say

                    XXI :112:    Say


R. Bell            XXI:4:        Say

                    XXI:112:     Say


M. Henning    XXI:4:        Sprich [German] = Say

                   XXI:112:     Sprich [German] = Say


Here is how Rippin analyzes the significance of this particular variant:

“The very last verse (112) of sura 21 starts “He said [qäla], ‘My Lord, judge according to the truth. Our Lord is the All-Merciful’ “. The reference to “My Lord”  and “Our Lord” in the text indicates that the subject of  “He said ” cannot be God but is the reciter of the Qur’än, in the first place understood to be Muhammad. Such a passage, in fact, falls into a common form of Qur’änic speech found in passages normally prefaced by the imperative “Say!” (qul).The significant point here is that in the text of the Qur’än, the word here translated as “He said” is, in fact, more easily read as “Say!” due to the absence of the long “a” marker (something which commonly happens in the Qur’än, to be sure, but the word qäla is spelled this way only twice – the other occasion being in Qur’än 21:4 and that occurs in some of the traditions of the writing of the text). In the early Sana manuscripts, the absence of the long “a” in the word qäla is a marker of an entire set of early texts. But why should it be that this particular passage should be read in the way that it is? It really should read “Say!” to be parallel to the rest of the text. This opens the possibility that there was a time when the Qur’än was understood not as the word of God (as with “Say!”) but the word of Muhammad as the speaking prophet. It would appear that in the process of editing the text, most passages were transformed from “He said” to “Say!” in both interpretation and writing with the exception of these two passages in Sura 21 which were not changed. This could have occurred only because somebody was working on the basis of the written text in the absence of a parallel oral tradition”.[47]

One cannot, I think, continue to maintain that variants are trivial and have no bearing on the meaning or that they are of no great significance. Several very important theses have emerged from the above discussions.

1.      Variants have always been acknowledged: e.g. Bukhärï , Abï Däwüd, al- Suyütï.

2.      Manuscript variants show that the Koran, like any other text, has a history, a history different from the Traditional Islamic account of the Koran’s compilation.

3.       Those variants that were invoked served many purposes:

i)                    In August Fischer’s opinion, Koranic textual variants “for the most part (Fischer’s emphasis) consist of no more than attempts at emendation made by philologically trained Koran specialists on difficult passages in the Uthmanic redaction”.[48]

  ii)     Polemical, see A. Rippin, Qur’än 7:40,  Until the camel Passes through the Eye of the Needle, Arabica, Tome xxvii, Fasc .2, pp. 107-113 “Variants such as those for Surah 7:40 were created when polemically–based pressures on the exegetes were the strongest and the attitudes towards the Qur’änic text less confining.” p.113.

       iii)     Doctrinal. The variants reflect the ideology, as Burton shows, of groups that wish to argue for their own viewpoint, to establish a legal ruling, to settle conflict of sources.

4.      The existence of variants casts doubt on the existence of an oral tradition. Skepticism of an oral tradition has been expressed by Fritz Krenkow,[49] A. Rippin, C. Luxenberg, Gerd –R. Puin and G. Lüling. The latter wrote, “It has long since been proven that there was in principle no oral tradition at all, either for Old Arabic Poetry or for the Koran, as now this book goes on to demonstrate by its reconstruction of the editorially reworked Christian hymnody in the Koran as well as of many (on the level of writing) reworked Old Arabic classical poems.” [50]

5.      This thesis leads to the conclusion that the redactor or redactors of the Koran was or were working on the basis of the written text in the absence of a parallel oral tradition.

6.      The so-called Seven Readings of the Koran should not be taken too literally since seven has a symbolic value derived perhaps from ancient Babylonian times with their notion of the seven stars and planets. The Koran itself talks of the seven heavens (XVII.44), seven gates to hell (XV.44), seven oceans (XXXI.27), and there is also the motif of seven in the story of Joseph.[51]

7.      The story of the collection of the Koran under ‘Uthmän is perhaps only a calque [52]on the story of the destruction of the heretical writings of Arius on the orders of Constantine as recounted in Socrates and Sozemenus. Socrates quotes this letter from Constantine to the bishops and the people, “Since Arius has imitated wicked and impious persons, it is just that he should undergo the like ignominy. Wherefore as Porphyry, that enemy of piety, for having composed licentious treatises against religion, found a suitable recompense, and such as thenceforth branded him with infamy, overwhelming him with deserved reproach, his impious writings also having been destroyed; so now it seems fit both that Arius and such as hold his sentiments should be denominated Porphyrians, that they may take their appellation from those whose conduct they have imitated. And in addition to this, if any treatise composed by Arius should be discovered, let it be consigned to the flames, in order that not only his depraved doctrine may be suppressed but also that no memorial of him may be by any means left. This therefore I decree, that if anyone shall be detected in concealing a book compiled by Arius, and shall not instantly bring it forward and burn it, the penalty for this offense shall be death; for immediately after conviction the criminal shall suffer capital punishment. May God preserve you! ” [53]


Sozomen tells us, “The emperor punished Arius with exile, and dispatched edicts to the bishops and people of every country, denouncing him and his adherents as ungodly, and commanding that their books should be destroyed, …” [54]

Under Theodosius II, the writings of Nestorius were also burned. Here is how Gibbon puts it, “After a residence at Antioch of four years, the hand of Theodosius subscribed an edict which ranked him [Nestorius] with Simon the magician, proscribed his opinions and followers, and condemned his writings to the flames, and banished his person first to Petra in Arabia, and at length to Oasis, one of the islands of the Libyan desert.”[55]

8.      None of the above theses lends credibility to the Traditional Islamic understanding of the Koran, its origins, its compilation and its redaction.


Perhaps it is time to start taking variants and Koranic manuscripts seriously.

[1] John Burton The Collection of the Qurän Cambridge: Cambridge University Press , 1977, p.5

[2] Ibid., p.6

[3] Ibid., p.31

[4] Burton , op.cit., pp.31-32.

[5] Burton, op.cit., pp.34-35.

[6] Burton , op.cit., p.36.

[7] Jaläl al Dïn `Abd al –Raùmän b.Abï Bakr Al-Suyütï,  Itqän fï ‘ulüm al-Qurän, 2 vols., in 1 , Ùalabï, Cairo , 1935 / 1354, pt 1, p.82 , quoted by Burton op.cit., p.36.

[8] Burton op. cit,

[9] Al- Suyütï, op. cit, quoted by Burton op .cit., p.37

[10] Herbert Berg The Development of Exegesis in Early Islam. Richmond, Surrey ,U.K: Curzon Press.2000 p.221

[11] Burton , op. cit., p.39

[12] M. Pickthall .The Meaning of the Glorious Koran, London: George Allen and Unwin Ltd., 1930, p. viii.

[13] George Sale, The Koran  London: Frederick Warne and Company [Circa 1890] [Ist edn. 1734] p. ix

[14] J. M. Rodwell. The Koran London: J. M. Dent & Sons Ltd., 1921 [Ist edn. 1861] Preface p.16

[15] A. Jeffery and I. Medelsohn . The Orthography of the Samarkand Quran Codex .Journal of the American Oriental Society ,62 (1942 ) pp.180-181, Chapter 7.5

[16] E. H. Palmer The Koran Oxford: Oxford University Press , 1949 [ Ist edn. 1880 ].

[17] N. J. Dawood. The Koran  Harmondsworth [U.K]: Penguin Books, 1990 [Ist edn.1956].

[18] A. Yusuf  Ali. The Holy Koran. Lahore [Pakistan]: Shaikh Muhammad Ashraf  1938 [Ist Edn.1934], p.iv

[19] Arberry in his short introduction seems to have uncritically swallowed whole every single Islamic dogma on the Koran, from it being a revelation to its untranslatability.  See A. J. Arberry.The Koran Interpreted  Oxford: Oxford University Press, 1964, Introduction, p.ix-xiii.

[20] A. J. Arberry  The Koran Interpreted  Oxford: Oxford University Press, 1964, Introduction, p.ix.

[21] R, Blachère Le Coran, Paris: G. P. Maisonneuve & Cie, 1949 . p. XII

[22] Cairo Edition: Blachère, Jeffery give 1342 / 1923 as the date of publication;  Jeffery and Mendelsohn [1942], however, give 1344 / 1925; R. S. Humphreys gives 1347/ 1928; G. Bowering and Brockett give1924. Would Post-Modernists say all the dates are valid ?!

[Blachère., op. cit. p. xii; A. Jeffery .Materials for the History of the Text of the Quran, Leiden: E. J. Brill, 1937; Jeffery/Mendelsohn, The Orthography of the Samarkand Quran Codex, Journal of the American Oriental Society,  3  (1942)  p.177 footnote 5; A. Brockett. Studies in Two Transmissions of the Q uran,  St. Andrews, Scotland: Doctorate Thesis, 1984; R. S. Humphreys, Islamic History, Princeton: Princeton University Press, 1991, p. 21. G. Bowering, Chronology and the Quran, in the Encyclopaedia of the Quran, Vol. 1, Brill: Leiden , 2001, p. 334]

[23] The Holy Bible Placed by The Gideons, La Habra, Ca.: The Lockman Foundation, 1977, pp. xx-xxii.

[24] Biblia Hebraica Stuttgartensia, K. Elliger and W. Rudolph, edd.  New Edition, Stuttgart: Bibelgesellschaft , 1967/77 p. XII

[25] Greek New Testament  Ed. F. H. A. Scrivener, New York: H. Holt & Co. 1903

[26] EI 2 edn [New Edn] s.v. Khatt.

[27] F. Déroche, Manuscripts of the Qurän , in Encyclopaedia of the Qurän, Vol. Three: J-O, ed. J. D. McAuliffe, Brill: Leiden-Boston, 2003, pp. 255

[28] Ibid., p. 257 right hand column .

[29] E. A. Rezwan, Frühe Abschriften des Korans, in J. A. Petrosjan et al. (eds.), Von Bagdad bis Isfahan Buchmalarei und Schriftkunst des Vorderen Orients (8.-18.Jh) aus dem Institut fur Orientalisti , St. Petersburg, Lugano 1995, 117-125

[30]  F. Leemhuis, Codices of the Qurän in Encyclopaedia of the Qurän, Vol. One: A-D, ed. J. D. McAuliffe, Brill: Leiden-Boston, 2001, p.350


[31] W. Diem, Untersuchungen zur fruhen Geschichte der arabischen Orthographie. Teile I -IV, Orientalia Vol.48-50, 52 (1979-81, 1983), Teil I  p. 211, translated and quoted by Gerd R-Puin, Variant Readings of the Koran, Chapter 8.3 in present volume, footnote 19.


[32] A. Brockett. Studies in Two Transmisions of the Qurän  St. Andrews, Scotland: Doctorate Thesis, 1984, pp. 9-10

[33] Gerd R-Puin. Variant Readings of the Koran, Chapter 8.3 in present volume.

[34] V. “QirŒ<a” in EI²

[35] P. Larcher  Coran et Theorie Linguistique de l`enunciation, in Arabica, XLVII (2000), pp. 443-444; Chapter 5.4 in present volume

[36] A. Jeffery and I. Mendelsohn. The Orthography of the Samarqand Quran Codex. JAOS ,3 (1942) 175-94 passim, Chapter 7.5 in present volume.

[37] Ibid., p.182.

[38] F. Déroche and S. Noja Noseda (eds.), Sources de la transmission du texte coranique I. Les manuscrits du style higazi vol.i. Lemanuscrit arabe 328 (a) de la Bibliothèque nationale de France, Lesa 1998; vol.ii Le manuscrit or.2165 (f. 1 a 61) de la British Library. Lesa  2001.

[39]  ‘Abd al-‘ Äl Sälim Makram (wa-) Aùmad Mukhtär ‘Umar (I‘däd ) : Mu‘jam al-qiräÆät al-Qur’äniyyah , ma‘a  maqadimmah fi l-qiräät wa-ashar al-qurrä  I-VIII. Al-Kuwayt: Dhät as-Saläsil 1402 –1405 /1982-1985

[40] Gerd R-Puin. Neue Wege der Koranforschung: II.Über die Bedeutung der ältesten Koranfragmente aus Sanaa (Jemen) für die Orthographiegeschichte des Korans, Universität des Saarlandes Magazin  Forschung, 1 (1999), 37-40.

[41]  Ibid . p.40

[42]  See A. Rippin, Qur’än 21 :95: A Ban is Upon any Town  JSS ( 24 ) 1979, 43-53,

      “… , the variants still show traces of their original intention: to explain away grammatical and lexical

      difficulties. While obviously this is not true of all variant readings in the Qur’än, many variants being too slight to alleviate any problem, in Sura 21: 95 and in many others the exegetical nature of Qur’änic variants is apparent.”  p.53 .

[43]   See A. Rippin, Qur’än 7: 40,   Until the camel Passes through the Eye of the Needle, Arabica, Tome xxvii, Fasc 2 , pp 107-113  “Variants such as those for Surah 7: 40 were created when polemically –based pressures on the exegetes were the strongest and the attitudes towards the Qur’änic text less confining.” p.113 .

[44] A. Rippin. Muslims, Their Religious Beliefs and Practices  London: Routledge, 2nd Edition  2001, p.30

[45] Brackets in original .

[46] Brackets in original.

[47] A. Rippin. Muslims, Their Religious Beliefs and Practices London: Routledge, 2nd Edition  2001, pp.30-31

[48] A. Fischer in Der Islam XXVIII (1948), 5f.n.4 quoted by R. Paret, Kirä`a, E.I.2nd Edn.

[49] F. Krenkow. The use of writing for the preservation of Ancient Arabic poetry, in: A Volume of Oriental Studie , presented to E. G. Browne on his 60th Birthday, edd. T. W. Arnold and R. A. Nicholson. Cambridge, 1922, pp. 261-268

[50]  Günter Lüling  A Challenge to Islam for Reformation  Delhi: Motilal Banarsidass Publishers, 2003, pp. XLI-XLII

[51] A. Rippin. Numbers and Enumeration in E.Q. vol.3 p.552.

[52] I think it was Lawrence I. Conrad who first suggested such a thesis. At a conference at the University of Mainz, Germany, in 2002,  a paper by Conrad was read out – he was unable to attend at the last moment because of a car accident – in which he puts forward this idea. I have not seen the paper in written form since that conference so I cannot say if he elaborates on this and explains what he meant. I do not know if he had the acts of Constantine and Theodosius in mind, they are my proposals.

[53] Socrates Church History From A.D.305-439.   trans. A. C. Zenos, in A Select Library of Nicene and Post-Nicene Fathers of the Christian Church   edd. P. Schaff and H. Wace, vol.II  W.E.Eerdmans, Michigan, 1997. Book I , Chpater 8.p.14.

[54] Sozomen. The Ecclesiastical History  trans. Chester D. Hartranft  in A Select Library of Nicene and Post-Nicene Fathers of the Christian Church .  edd. P.Schaff and H.Wace , vol.II  W. E. Eerdmans, Michigan, 1997. Book I. ch.XXI  p.255.

[55] E. Gibbon,   The Decline and Fall of the Roman Empire,  Modern Library, New York   No date,  Vol. II ,Ch.XLVII, p.825. Gibbon’s source was  The Imperial Letters in the Acts of the Synod of Ephesus ( Concil.tom.iii , pp.1730-1735 )

To comment on this article, please click

To help New English Review continue to publish important scholarship such as this, please click here.

If you have enjoyed this article by Ibn Warraq and would like to read more, please click here.


Leave a Reply

Your email address will not be published. Required fields are marked *

New English Review Press is a priceless cultural institution.
                              — Bruce Bawer

Order here or wherever books are sold.

The perfect gift for the history lover in your life. Order on Amazon US, Amazon UK or wherever books are sold.

Order on Amazon, Amazon UK, or wherever books are sold.

Order on Amazon, Amazon UK or wherever books are sold.

Order on Amazon or Amazon UK or wherever books are sold

Order at Amazon, Amazon UK, or wherever books are sold. 

Order at Amazon US, Amazon UK or wherever books are sold.

Available at Amazon US, Amazon UK or wherever books are sold.

Send this to a friend