David S. Margoliouth

If uniformity of text be required in a sacred book, the Hebrew Bible would seem to fulfill the demand better than either the Greek New Testament or the Arabic Koran. The varieties which have been collected from manuscripts of the first are almost negligible; important differences are found either in alternate copies of the same documents which are incorporated in the Old Testament, or in ancient versions, the use of which for textual criticism is hazardous. In the criticism of the New Testament the ancient versions play an important part; but the manuscripts also are far from uniform, and in some parts exhibit widely differing recensions. Until January of this year no ancient version of the Koran had been introduced into the criticism of that book; Dr. Mingana, who has discovered a Syriac version of high antiquity, and described it in the Bulletin of the John Rylands Library, is the pioneer in this matter. The same scholar in his Leaves from Three Ancient Korans (Cambridge, 1914) called attention to noteworthy variants in old manuscripts. But orthodox Islam does not assume uniformity of text for its sacred book; it admits seven or even ten canonical recensions, differing ordinarily, but not always, in minutiae; and in addition to these there are a great number of uncanonical readings, attested by distinguished personages. Ahmad b. Musa b. Mujahid (ob. 324 A.H.) composed nine works embodying the readings of different authorities, one of them the Prophet himself! These by no means exhausted his activities in the collection of various readings. We may endeavor to classify these varieties and account for their existence.

The Koran (ii. 100) assumes that it is perpetuated partly in the memory and partly in writing; and asserts that Allah at times commits texts to oblivion or causes them to be erased, to substitute something better or not inferior. This process was regarded by some of the Prophet's contemporaries as clear evidence of imposture (xvi. 103); leaving this matter alone, we may at least notice that nescit vox missa reverti; a text might be officially removed, yet survive, owing to those in possession of it being ignorant of the abrogation or neglecting it. Hence among the Various Readings which are quoted some may actually represent an earlier or a later form of the same revelation. Thus in v. 91 the ordinary text prescribes a fast of three days for impecunious persons who wish to compensate for perjury: Tabari (ob. 310 A.H.) notes authorities for the assertion that Ubayy b. Ka'b and 'Abdullah b. Mas'ud added the word successive, making the penance much more severe. He adds that as the word is "not found in our copies," we cannot build anything upon it; the analogy of compensation for failure to fast in Ramadan (ii. 181) indicates that the days need not be successive; still it would be safer to make them so. Shafi'i (ob. 204 A.H.) seems to leave it to the individual Moslem to choose the reading which he prefers.1 It is a conceivable view that the word successive might have been added or omitted by the Prophet himself.

The fact that the revelations might be abrogated is likely to have seriously affected the importance attached to the Koran in the Prophet's time; he had the reputation of being at the mercy of each speaker (ix. 61). The story told by Bokhari about iv. 972illustrates the effect of this quality on the Koran. A text had been revealed asserting that the believers who stayed at home were not the equals of those who went out to fight. A blind man complained that the latter course was impossible for him. A revelation came adding the words except those who suffer from some infirmity.

When with the Prophet's death revelation ceased, such texts as had been preserved acquired vast importance; they were all that could be known of the will of God. Since, if the tradition is to be believed, there was no official copy in existence, those who claimed the monopoly of portions might aspire to be dictators of the community: a far safer plan than that tried by those who claimed to be prophets. We may well believe that the measure taken by the third caliph, of issuing an official edition and ordering all unofficial copies to be burned, was a political necessity. That this act brought about an insurrection wherein he was murdered is the most probable explanation of the first civil war of Islam. Dr. Mingana3 has called attention to a tradition that anothcr official edition was produced by the famous (or notorious) Hajjaj b. Yusuf (ob. 95 A.H.) near the end of the first century; and to this there may be a reference in an Abbasid manifesto of the year 284 A.H wherein the Umayyads are charged with "altering the Book of Allah,"4 though no attempt is there made to substantiate the charge. According to another authority (as will be seen) what this personage did was to introduce punctuation. What both supposed recensions imply is that there was variety - or at any rate something to be altered - before they were made. Some variants might remain in oral transmission after they had been officially condemned.

If the collectors of the Koran had to trust for portions of it to oral tradition, it is unlikely that the standard of accuracy was sufficiently high to ensure uniformity. In the article "Parallel Passages in the Koran" (Moslem World, July 1925) Mr. E. E. Elder shows that there are noteworthy variations in the different versions of the same narrative which the Koran contains. Even when it quotes itself, the quotations are not always what we should call accurate. An example may be taken from iv. 139, He has sent down unto you in the Book that when ye hear the signs of Allah discredited and ridiculed, ye shalt not sit with them until they plunge into another topic. The reference would seem to be to vi. 67, When thou seest those who plunge into our signs, turn aside from them until they plunge into another topic. Clearly the former is a loose paraphrase rather than a quotation, since the differences are many and serious. We need not credit the earliest transmitters of the Koran with greater accuracy. For a long time there was uncertainty as to what was Koran and what was not. Verses of poets were at times cited in the pulpit as the Word of Allah.5 There are occasions when the inaccuracy of those who cite it is astounding. The caliph Mansur, when in his controversy with an 'Alid pretender he wished to prove that an uncle could be called a father, cited xii. 38: I (Joseph) followed the sect of my fathers, Abraham and Ishmael and Isaac and Jacob;6 the argument depends on the name Ishmael, which is not found in the text! The proof-passage intended by Mansur is ii. 127, where Jacob's sons say to him, We shall worship the God of thy fathers, Abraham and Ishmael and Isaac; where, as Mansur observes, "he began with him (the uncle) in preference to the immediate parent." Neither Mubarrad nor Ibn Khaldun, who produce the letter with the quotation from xii. 38, notices the mistake; Tabari7 omits the wrong quotation, but apparently cannot find the right one. An even more extraordinary case is that of Bokhari, who in the first section of his Kitab al-Manaqib mentions an occasion in consequence of which "there was revealed Unless ye contract relationship between me and you." No such revelation is found in the Koran. In justice to the commentators it must be mentioned that they are puzzled by this statement of the foremost traditionalist. Their expedients, however, are rather desperate: one is the suggestion that such a text had been revealed, but was afterward abrogated; another that Bokhari is giving the sense of some passage in lieu of citing it verbally, procedure for which the example of the Prophet's encomiast Hassan b. Thabit is cited, who however had the excuse of verse.8

It is unlikely that those from whose memories portions of the Koran were collected were more accurate than Mansur and Bokhari, and the elimination of oral tradition by the reduction of the whole to writing was a great step toward uniformity. Only the script chosen for the Koran left much to be desired. The script in use in pre-Islamic Arabia is clear, efficient, and beautiful; the signs for the twenty-nine consonants are distinct, and a vertical line separates word from word. The script chosen for the Koran is a modification of the North Semitic alphabet of twenty-two consonants, not however in the direction of increase but in that of decrease; whence the same sign stands for sounds which bear no resemblance to each other. In a work on textual corruption by an author whose death-date is 382 A.H.9 the necessity of learning the Koran from teachers is illustrated by the case of one Hamzah who afterward became the author of a canonical recension; he started reading without such aid and read the opening words of Surah II, "That Book la zaita fihi (no oil in it)," where he should have read la raiba (no doubt). He got, in consequence, the title al-Zayyat, "the dealer in oil." There was no difference in the signs representing these two words. According to this author the word used for corruption of the text means properly employment of manuscript in lieu of oral instruction, which was a necessity. He quotes verses which prove that the reading of such a text unaided was a difficult performance. One is by the poet Tammam (ob. 228 A.H.):10

When they are fettered they march along, but when they are unfettered they cannot get away

i.e., when the words are given diacritic points they can be read, but not otherwise. Another is by an earlier author, Ru'bah (ob. 145 A.H.), who however refers not to any Muslim writing but to the Christian Gospel11:

As though it were some doctors' Gospel, whose punctuator makes clear that which his pen has written thereon with ink; when some reader spells it out under his breath, the diacritic points bring out the words intended, and the distinguishing circles or tattooing reveal the contents to the intelligence of one who takes them in, unless indeed he has to have it translated.

This reference to the two systems of Syriac punctuation is of great interest; moreover the poem can be dated with fair accuracy since the caliph to whom it is dedicated reigned 132-136 A.H. (750-754 A.D.). In another passage this poet alludes to the Arabic alphabet12:

'Tis as though they were lines of a pointed text, uttering the qaf or the lam.

It is possible or even probable that this script was chosen in order to maintain the esoteric character of the book; the Koran is a work of this sort, and indeed in its opening sentences declares itself guidance to the pious, i.e., to those who observe the ordinances Islam. Unbelievers are not to handle it, or indeed know anything about it. There may then have been an oral tradition of the way wherein it should be read. But it is clear from the Various Readings that this tradition was to a great extent lost. In order to facilitate reading diacritic points, distinguishing the letters (as in the case cited, R from Z and B from T) were invented, according to the author of the work on textual corruption at the instance Hajjaj b. Yusuf by one Nasr b. 'Asim. Somewhat later vowel-signs were introduced. The caliph Ma'mun (198-218 A.H.) is said to have forbidden the use of both.13 The use of both came in very gradually as students of Arabic papyri know. The caliph Walid b. Yazid (125 A.H.) notices that an epistle is "dotted," if the line be genuine.14 Abu Tammam a century later compliments a correspondent for so marking his script that it leaves no doubt to the reader; it not only has dots, but signs which indicate the cases, etc.15

It is surprising that the introduction of these signs into the text of the Koran should have taken place without a civil war or the like. We may define the business of the Readers whose work became canonical as the proper assignation of these points and vowel-signs. On the one hand they had to build up a system of grammar from the Koran; on the other to apply that system to its interpretation. In numerous cases the ambiguity of the script which led to Various Readings was of little consequence; when, e.g., Allah was the subject, the verb might be read He shall, or We shall, without affecting the sense. Yet there are places wherein this ambiguity is by no means unimportant; in the account of the miracle of Badr (iii. 11) the nature of the miracle varies seriously according as we read ye saw them or they saw them.

We should have expected the Various Readings to be based on Tradition; the commentators rather assume that they are based on consideration of the evidence. In Surah vi. 91 it depends on the location of a couple of dots whether we read ye make it or they make it. "Ibn Kathir and Abu 'Amr," (two of the canonical readers) says Baidawi, "only read the third person to suit the preceding they did not esteem" (where the form is unambiguous). They were not, then, reproducing what they had learned from teachers, but doing their best to decipher a text. It is surprising to find a various reading in the short "Opening" Surah, which enters largely into the ritual. Some read maliki yaumi 'l-din, others mâliki, meaning respectively "king of" and "possessor of" the Day of Judgment. Parallels are cited from the Koran in defense of the one and the other reading.

Where the readings are traced to contemporary authorities, there is at times suspicion that this evidence is fictitious. In iv. 117 the text before the Readers ended in an obscure word: They do not invoke in lieu of Allah other than * * *; the last word was ordinarily read INATHAn (females). It was not clear that this statement was accurate; certainly many of the deities worshiped in pagan Arabia were male. In xxix. 16, however, the Koran says Ye only worship in lieu of Allah AUTHANan (idols). The emendation idols for females was clearly plausible; only the form AUTHANan involved the insertion of a letter, whereas the form UTHUNan was doubtful Arabic. Tabari16 tells us that some one had found the former in A'isha's copy,17 while others averred that Ibn 'Abbas, the interpreter of the Koran par excellence, read the latter, which might be an alternative form of plural.

Ordinarily the readers did not venture to tamper with the consonants. Thus Mubarrad (ob. 285 A. H.)18 dealing with the difficult verse lxxii. 4, wherein Ja(D)Du Ra(B)BiNA is ordinarily read, ascribes a variant JaDA to Sa'id b. Jubair (ob. 95 A.H.), but says it cannot be accepted because it disagrees with the writing; and the same objection, he states, would apply to a reading JiDDAn. Any reading which did not involve such alteration would be permissible. On the whole this is the view maintained in the great grammatical work of Sibawaihi (ob. 180 A.H.). The vocalization was settled by critical and grammatical considerations. In xxxi. 26 he quotes the reading wal-bahra but, he adds, some people read wal-bahru, in accordance with certain usage19. The principle whereon the "people of Medina" prefer the reading in kullan to inna kullan in xi. 113 is elaborately explained. "I am informed" he writes "that one of them read in cxi. 4 hammâlata for hammâlatu, treating the word not as a predicate, but as though he had said I mean etc., by way of vituperation."20 Yet he occasionally records variants which imply a difference in the consonantal text. The ordinary reading of lxviii. 9 fayudhinûna seems certified by the rhyme, while the grammar requires the subjunctive; Sibawaihi says "Harun asserted that fayudhinu was to be found in certain copies."21 "They aver," he says, "that in the text of Ibn Mas'ud of xi. 75 there was shaykhan" (i.e., with a final A) in lieu of shaykhun. Sibawaihi's formulae imply that he accepts no responsibility for the statements which he records.

Did these Readers ever go outside the Koran and the grammatical rules which they had formulated in order to determine the correct reading and vocalization of the text? The tradition that one of them read Ibraham in lieu of Ibrahim suggests such research; it was a bold alteration, for the form Ibrahim seems certified by the rhyme in xxi. 61, 63. It would appear that some historical study was bestowed on the opening words of Surah xxx. 1, containing a famous oracle. The natural way to vocalize the words would appear to give the sense The Romans have been victorious in the nearest part of the earth (the Near East!) and they after their victory shall be victorious. This was rejected by the Readers, who read either The Romans have been conquered in the nearest part of the earth and they after their conquest (defeat) shall conquer; or The Romans have conquered in the nearest part of the earth and they after their conquest shall be conquered. The former, which is the ordinary, view made the first reference to the defeat of the Byzantines by the Persians, and the second to their later defeat of the Persians, which is foretold; the second view made the first reference to the defeat of the Persians by the Byzantines, and the second to the defeat of the Byzantines by the Muslims. Since the text proceeds to say and that day shall the Believers rejoice, probably the second view is really right.

The introduction of diacritic points and vowel-signs stabilized the text of the Koran so far as it was possible to stabilize it; gradually out of the large number of recensions made by the processes that have been sketched a certain number became authoritative, and such various readings as were presented outside this number were quoted chiefly in support of one or other of the interpretations which the canonical recensions were supposed to admit. The process whereby this came about bears a resemblance to what happened in the case of Schools of Law, among many rival schools ultimately four came to be regarded as orthodox, and the others, even when attached to illustrious names, fell into oblivion.

One who deciphered the Koran afresh in these days - i.e., based a new edition on an unpointed text - would be likely to adopt many uncanonical readings, and might even introduce some that were new. He might, e.g., substitute Yuhanan for Yahya as the name of the Baptist. He might be able to obelize certain texts as later than the Prophet's time. But in the endeavor to produce a Koran such as the Prophet might have approved he would be confronted with the difficulty which was too great for the original collectors - the theory of substitution, the limits of which he would be unable to fix. Where the same narrative is repeated, should one version only be retained? The exercise of this theory would reduce the Koran to a fraction of its present bulk; yet repetition of narratives has no place in a single book, equally with or without discrepancies. The case of the four Gospels offers no parallel, since here we have not one book, but four books ascribed to different authors.

In normal cases a collection of various readings furnishes the history of the corruption of a text through carelessness or interpolation. Occasionally, it may include a record of alterations made by the author himself. The case of the Koran differs from these in some important respects. The greater number of variants are different attempts at deciphering the same text; often there is agreement about the consonants intended, but disagreement about the vowels to be supplied; the oracle about the Romans shows that such disagreement may affect the meaning seriously. Owing to the ambiguity of the signs used for consonants there is often disagreement about the interpretation of these signs; the variation that has been quoted between females and idols shows that the difference of sense which results may be considerable. Less frequently, the difference of reading extends to the number of signs in the text; the addition of one letter in vii. 142 changes I shall show to you into I shall cause you to inherit. "The latter" says Zamakhshari "is a good reading, because it agrees with vii. 133." At times, however, it extends to the omission or addition of whole words or phrases. We have good reason for believing that parts of the Koran were obtained by the collectors from oral tradition and not from manuscript; and here we have the possibilities that the same passage was variously reported by those who remembered it, that one form was substituted for another by the Prophet himself, and that the alterations were due to copyists and made intentionally or unintentionally. And thus we ascend by easy stages from lower to higher criticism.

It is worthy of note that in spite of the reputation of the Umayyads for impiety the stabilization of the Koran, so far as it was accomplished, was achieved during their dynasty. In this context it should be remembered that the founder of that dynasty appealed to the arbitrament of the Koran when his rival Ali would have preferred that of the sword, and the Koran appeared to favor the former's claim. And we have seen that the founder of the 'Abbasid capital was by no means Bibelfest, if that phrase may be applied to the Koran.


1 Umm vii. 60.
2 Ed. Krehl ii. 209.
3 Bulletin of the John Rylands Library, p. 201.
4 Tabari iii. 2175, 7; 2176, 17.
5 Irshad al-Arib vi. 94. Cf. Aghani iv. 177 end (ed. 2).
6 Mubarrad, Kamil ii. 322 (Cairo ed.).
7 Chronicle iii, 211.
8 See Fath al-Bari vi. 342.
9 Kitab al-tashif of Al Hasan b. 'Abdallah 'Askari, Cairo, 1908.
10 Diwan, Beirut, 1905, p. 215.
11 Ru'bah ed. Ahlwardt, p. 149. Ahlwardt's translation, Der Juden Bibel, drin der Kritzler malte die Züge die sein Griffel schrieb mit Tinte, most seriously misrepresents the meaning.
12 P. 144, 13. Ahlwardt is again misleading.
13 Iqd Farid ii. 166 (ed. 1)
14 Aghani vi. 101, ed. 2.
15 Diwan, p. 418.
16 Comm. v. 165, ed. 1.
17 Her private copy is mentioned by Ibn Hanbal vi. 73.
18 Kamil ii. 91 (Cairo ed.).
19 i. 285, Cairo ed.
20 i. 252.
21 i. 422.

Source: The Muslim World, Volume 15 (1925) pp. 334-44.

The Text of the Qur'an
Answering Islam Home Page