The Problem of Dating the Early Qur'ans

In 1947 Prof. Dr. Giorgio Levi della Vidal published a valuable and interesting contribution to Arabic palaeography by editing fragments of parchment Qur'ans, acquired by the Vatican Library in 1946; they formerly belonged to the collection of Tammaro de Marinis, and were probably acquired in Egypt. This publication was made with the usual thoroughness and accuracy characteristic of all the works of Levi della Vida. In his customary cautious manner, he avoids taking a definite position on the difficult problem of dating, especially since it is just this dating of the oldest Qur'anmanuscripts that is still controversial and undecided. However, I think I am right in assuming that he is more inclined to prefer an earlier dating, proposed by Bernhard Moritz and Nabia Abbott, for he considers the early dating, namely, the second half of the first century of the Islamic era, as at least not impossible, since inscriptions and papyri of this period show a type of writing completely analogous to that of the earliest Qur'an copies.

So Levi della Vida has, in passing, indicated the only passable way in which the problem of dating the oldest extant copies of the Qur'an can be solved in a satisfactory manner; and when I here-in a limited way-enter into a discussion of the problem, I would, first, like to express my thanks to the learned scholar on the occasion of his jubilee2 for his above-mentioned contribution, and then, to explain what more new material, recently brought to light, has contributed toward the solution of this problem of dating.

While-with one exception3-we hitherto possessed Qur'anic texts on papyrus written only for subsidiary purposes, i.e. in pericopes and included in magic texts,4 real leaves from a Qur'an codex have made their appearance in the Collection of George Michaelides in Cairo, which put scientific research on quite a different basis. As a welcome completion thereto I discovered in the National Egyptian Library in Cairo quite a good number of valuable fragments of early parchment Qur'ans, which had apparently been collected and mounted between two glass plates by the former director of the Egyptian Library, my ever memorable friend B. Moritz, but which obviously had not been utilized by him for his studies. Finally, the last finds of Arabic papyri in Khirbet Mird (Palestine-Jordan) which were liberally made accessible to me by G. Lankester Harding, have enriched our knowledge of the early Arabic script used outside Egypt, and furnished new, hitherto unknown material for palaeographic investigation. I now give a short survey of the state of the problem of dating the oldest Qur'an manuscripts up to the present.

Since Jacob Christian Lindberg5 made the first attempt at dating the parchment Qur'an manuscripts in the Royal Collection in Copenhagen in 1830, and Johann Heinrich Moller,6 in the appendix to his collection of thirty-one facsimiles of Qur'an fragments from various centuries in the Ducal Library in Gotha, laid down the first principles and distinctive marks for dating these manuscripts in 1842, showing a fine comprehension of this hazy problem, definite dating of old parchment Qur'ans was practiced by Josef Balthasar Silvestre.7 He dated specimens from Kufic Qur'ans, formerly belonging to the collection of Michelangelo Lanci (Rome), in the seventh (Pl. iv), ninth (Pl. v, vi), and tenth centuries (P1. vii, , viii), while William Wright8 dated the parchment Qur'an Or. 2165 in the British Museum in the eighth century C.E. But the most important and extensive contribution to Qur'an palaeography is due to B. Moritz,9 in whose monumental work Qur'an manuscripts formed the overwhelming majority. He dated, for example, the big parchment Qur'an in the Egyptian National Library (formerly in the 'Amr-Mosque in Old-Cairo) and the copy of the mosque of Sayyidna Husein in Cairo into the first to second century of the Higra, and a Qur'an from the mosque in Fuwwah (Delta) in the second/third century.

His datations met with a severe criticizm from the part of J. v. Karabacek, who dated the parchment Qur'ans in Pl. i-xii in the third century A.H. (instead of the first suggested by B. Moritz), xvii-xxx into the third century A.H. (instead of the second, as Moritz did), and Pl. xxxix, xl, xliv in the second century A.H. (eighth century c.E.) instead of Moritz's third century.' 0 The monumental publication on the masterpieces of Muhammadan Art, exposed in Munich in 1910, contained Qur'an manuscripts also, among them Codex Gothanus no. 565, ascribed to the eighth century A.D.," which dating was contested by J. v. Karabacek,12 who suggested the ninth century C.E. In 1914 Eugen Tisserant13 published four plates containing Qur'ans of the eighth to the sixteenth century C.E., and in the following year Eugenio Griffini,14 one page of a Kufic parchment Qur'an without dating it. An important contribution to Arabic palaeography and especially also the palaeography of Koranic writing was then published in 1919 by Prof. Gotthelf Bergstrasser,15 who, nevertheless, gives relative datings only ("alteste, altertumliche Korane"), qualifying Arabic palaeography Pl. 44, Brit. Mus. Or. 2165 and Paris. Arabe no. 328 (1) and 328 (4) as the oldest or most archaic Qur'an-copies. Some years later16 he published a short survey of the script of the Qur'an manuscripts of the first four centuries of the Higra, and stated that dated copies occur more frequently after the third century A.H.17 He discerns three styles of writing: (1) Lapidary Kufi (e.g., Arabic Palaeography Pl. 1); (2) a script between this lapidary style and the cursive writing known from the papyri (hijazi), e.g., the Qur'an in the British Museum Or. 2165; and (3) Maghrabi writing. Eight plates with ten photographs complete his short essential survey.

Therewith the main bulk of material, available for the study of the question of dating this kind of Arabic bookscript, was concluded.

The main difficulty for dating this material is obviously formed by the fact that clearly dated copies of the Qur'an, which could serve as landmarks for dating, are still exceedingly rare: one dated copy exists from the first century and two exist from the second,18 seven only from the third century of the Hijra.19 Among the Qur'ans of the first century A.H. there should be-according to I. Y. Kratchkovsky20-also included the Qu'ran from Samargand.21 At any rate, the opinion about the age of early Qur'ans differs considerably; I shall quote only some of the most important examples: on one hand, Nabia Abbott,21 for example, considers the Qur'ans reproduced in B. Moritz's Arabic Palaeography, Pls. i-i2, 31-34 as "the very earliest extant Korans," while Arthur Jeffery is inclined to doubt "whether any fragments we have are from so early a period as the first Islamic century, and one would hesitate to date more than a very few in the second century."23 I shall come back to these statements later.

But the most spectacular announcement was made by M. Minovi in his "Outline History of Arabic Writing,"24 where he said that the extant early Qur'anic specimens are all either forgeries or suspect, and that there was a widespread use of the various cursive scripts in early Qur'anic writing. This spectacular and autoritative statement deserves closer investigation. Minovi starts from the illustration of the Meccan-Medinan script in the Chester-Beatty codex no. 3315 of Ibn an-Nadim's Kitdb al-Fihrist in the National Library of Ireland (Dublin), fol. 3 verso last line (see Fig. 1), which he classifies as cursive, and concludes thereform that early Qur'ans, said to be written in the Meccan-Medinan scripts, were written in cursive scripts.25 He then refers to the Galil, Tumar, Thuluthain, and Thuluth, all of which he considers as predominantly and equally cursive scripts, restricted to Qur'anic writing since the coming of the 'Abbasids, that is, since the second half of the eighth century C.E. He further says that even the new Qur'anic scripts, developing at this time, were also cursive, for example, the muhaqqaq, and that cursive scripts were widely used for early Qur'anic writing. Then he says that not many early Qur'ans have survived from the early centuries, and that these Qur'ans were wholesale forgeries in Kufic script, which was considered by the forgers as the script current in the first centuries of Islam. He corroborates this theory by pointing to instances of forged documents and some Qur'ans attributed to the Caliph cAli ibn Abi Ta lib.

Naturally enough, these suppositions of Minovi met with a severe and just criticizm by N. Abbott, who, in her article already quoted,26 gave her opinion extensively on Minovi's statements.

Fig. 1. The Basmala in Meccan-Medinan script.

a. facsimile drawn by Minovi (Survey II, p. 1710. fig. 580)

b. the drawing in the original

(1) She surmises that the illustrative Basmala in the Chester Beatty codex of the Kitab al-Fihrist is in a different script from that of the whole text (p. 70a).

(2) If this specimen represents the true cursive variety of the MeccanMedinan script, other specimens in this script should be easily found in documents dating from these centuries or even from the third. As a matter of fact, N. Abbott could not find any documents from these early centuries written in the script of the sample of the Meccan-Medinan script in the Chester-Beatty codex. The peculiar Alif of this sample, with its sharphooked bend to the left of the top is to be met with in the fourth- and fifthcentury script (p. 70 b).

(3) Although Minovi is convinced that this sample really corresponds to the description of the Meccan-Medinan script given by Ibn an-Nadim, N. Abbott is convinced that it does not. Minovi has obviously misunderstood and misinterpreted this description when he says, "the Mecca and Medina scripts were charcterized by an alif with a slightly oblique lower terminus, slanting to the right, and a final hooked in the opposite direction, together with a slightly swaying rhythm."27 The correct interpretation is that there is in its alifs a turning to the right and a raising high of the fingers (i.e., the upright strokes) and there is in its form a slight slant (p. 71).28

(4) In contradistinction of Minovi's supposed mass destruction of old Qur'anic copies, Miss Abbott (p. 72) surmizes that a good number of early Qur'anic manuscripts have survived, although the extant number of manuscripts is not suspiciously large.

(5) While Minovi states his conviction that the great majority of extant Qur'anic specimens are forgeries of later centuries, he gives no real proofs for this supposition, but promises to do so in a forthcoming monograph. Miss Abbott correctly states that even forged specimens are expected to give a fair reprensentation of Qur'anic scripts and practices of the earlier centuries (p. 72b). Although cases of forged deeds and letters are recorded in Arabic literature,29 she has not come across any references to Qur'anic forgeries in general, and even if a few such references would appear, they would not prove that our extant Qur'anic specimens are all, or almost all, forgeries, and although one should be reasonably suspicious of extant Qur'an copies attributed to 'Ali, 'Uthman, or Ibn Moqla and Ibn al-Bawwab, it would be oversuspicious to extend these suspicions to all extant Qur'ans of the angular types, as does Minovi.30

So far Miss Abbott's objections.

It was clear from the beginning that the capital point in the discussion is the sample of the Basmala occurring in the Chester-Beatty codex of Ibn an-Nadim's Kitdb al-Fihrist, published in facsimile by Minovi31 (see Fig. 1). Miss Abbott has not had the opportunity to see the original drawing, which obviously forms the cornerstone of Minovi's statements. Since every and even minute detail could be of great importance for the questions under discussion, I asked for photographs of the pages referring to Arabic script in the Chester-Beatty codex no. 3315, and especially also of the Basmala in Meccan-Medinan writing, and I am greatly obliged to Mr. J. V. S. Wilkinson, librarian of the Chester-Beatty Library in Dublin, for sending me the respective films. A close inspection of the photographs had the following results:

(1) The text of the locus classicus concerning the oldest Arabic scripts given by Ibn an-Nadim (fol. 3 verso) is obviously identical with that offered in the Edition of G. Flugel, p. 63 ff. and the edition of Cairo, Rahmaniyya Press, 1929/30, p. 8 ff.; it runs:

Qala Muhammadu'bnu Ishaga: fa-'awwalu'l-hutati'l-larabiyati'l-hattu 'l-Makkiyu wa-ba'dahu `l-Madaniyu summa '1-Basriyu tumma 'l-KuJIyu. Fa-'amma 11-Makkiyu wa-'l-Madaniyu fa -ft alifatihi ta`wigun 'ila yamani 'l-yadi wa-'i'la'u 'l-'asabi'i wa-ft aklihi indiga'un yasirun wa- hada mitluhu: Bismi 'llahi'r-Rah'nani'r-Rahimi.

"Muhammad ibn Isha q says: The first Arabic scripts are: the Makki-script, and after it the Medinan, then the Basran, then the Kufan. As far as the Meccan and Medinan are concerned, there is in its Alifs a bend to the right side of the hand, and a raising high of the vertical stokes (hastae, fingers) and in its shape a moderate inclination to the side."

So it is quite evident that Minovi's translation is incorrect and misleading (cf. above p. 218); for nothing is said of an oblique lower terminus and a final hooked in the opposite direction, and the translation of wa-'i`la' ul'asabi`i (a raising high of the hastae) is missing.32

(2) It is surprising that the Basmala sample in this codex is not, as should have been expected, in a quite different script from that of the whole text, but in a hand not essentially differing from the text in which it is included. This becomes clear from a comparison of the Basmala on fol. I verso line 1 and the Meccan Basmala on fol. 3 verso last line; the former differs only in the shape of the slightly curved Alif, and vertical Lam, both of which lack the hooked tops; but those barbed Alifs and Lams recur several times within the text, just as in fol. 3 verso. Besides, Minovi's copy of the Meccan-Medinan Basmala sample is not quite exact; it shows throughout barbed tops of the Alifs and Lams, which really occur in Lam of ir-Rahmani r-Rahami and in the Alif of ar-Rahmani only, while the Alif in allahi shows a thickened head as in the Mushaf, no. 77 in the Egyptian National Library. The initial Ha has, in Minovi's facsimile, a small loop coming down below the basic line of the letters, quite different from the form in the original, which shows a pointed triangle on the basic line. The tail of the Mim in Bismi shows a slight swing in the facsimile, while in the original it is straight, and finally, the flat, slightly curved tail of final-Mim in ar-Rahimi is too short in Minovi's facsimile. All these are inaccuracies that should be avoided in such an important sample of writing.

But is there really any chance that the sample of the Basmala in Meccan-Medinan writing in fact represents this script?

If we compare it with the description if the Meccan-Medinan script given by Ibn an-Nadim, we must state that two main characteristics are completely missing in the Alifs of the facsimile, that is, the bend to the right and the pronounced height. Furthermore, the thickened head of the Alif, which may ressemble either a fishing hook or a knob, does not occur before the third century of the Higra, for example, in PER Inv. Ar. Pap. 1920, APEL II no. 50 1 Pl. vi (third century A.H., ninth century C.E.), and is more frequently found only in the second half of the fourth century A.H., for example, in APEL I no. 37 (Pl. III); that is just about the time when Ibn an-Nadim composed his Kitab al-Fihrist and the first copies thereof came into existence. So it seems doubtful whether the facsimile in the Chester-Beatty codex could be considered as a real representative of the old and original Meccan-Medinan style of writing. Already Prof. Dr. A. Jeffery33 has referred to the fact that Ibn an-Nadim's illustration of the Himyaritic musnad-script does not encourage confidence that his information regarding early Arabic script was much more accurate. Although the illustration of this musnad-script is much better in the Chester-Beatty codex (fol. 3 verso) than in the Cairo edition (p. 9), even here some characters are obviously misunderstood and others completely ficticious.

The illustration of the Meccan-Medinan script offers therefore no criterium for the dating of early Qur'an copies, and gives no clue respecting the genuinity of these copies.

But have we really to give up any hope of dating these Qur'an copies or to content ourselves with general and somewhat vague characterisations as, for example, "oldest, most archaic," and so on?

In the first place we have to state that it is in fact very difficult to date the first group of Qur'ans, written in a "lapidary style."34 The hieratic character of this script, the intention to create something special and extraordinary, has led the creative and ingenious calligraphists to develop a copious set of different styles, of which Ibn an-Nadim35 gives a long list. I refer only to plates I, iii-vi, viii in G. Bergstrassers treaty in the Geschichte des Qor'ans to give a general idea of some of the rich possibilities prevailing for the purpose of rendering the Book of God conspicuous and to distinguish it from other, secular books by a special type of large writing, which is significant of the large-sized Qur'ans of the mosques.

But the second style, called hijazi by Berstrasser and others, allows dating, owing to its affinity with the script of papyri. Dated papyri can thereby serve as landmarks,36 and it is quite important to state that this style of writing is-in contradistinction to the hieratic lapidary script-a secular script, used even for economic purposes, as, for example, PER Inv. Ar. Pap. 11077 and 11154, two lists of payments, 11161 a list of persons, 11163 a receipt, and so on, all representing Egyptian writing and coming from the ruins of al-Fustat. If we compare the Qur'ans Brit. Mus. Or. 2165, Mss. Paris. Arabe 328 (1), 328 (4)37 Codex Vat. Ar. 1605,38 Arabic Palaeography Pl. 44, Istanbul, Saray, Medina 1 a,39 the parchment no. 1700 in the Papyrus-collection of the National Egyptian Library in Cairo (Plate III a) and Inv. Perg. Ar. 2 in the Archduke Rainer Collection in the National Austrian Library in Vienna (Plate Va) with these papyri, it is fairly possible to ascribe them to the first century of the Islamic era (seventh or beginning of eighth century c.E.). Also the papyrus fragment Arabic Palaeography P1. 43, ascibed by B. Moritz to the third century of the Higra, belongs still to the first century of the Islamic era. Anyhow, this fragment is severely mutilated. But the Qur'an fragment no. 32 in the Collection George Michaelides in Cairo is large and much better preserved, and may therefore serve as a good example of the early Koranic bookhand on papyrus.

Plate I

Qur'an LIV.11-38, 45-55, LV.1-32

Brown, fine papyrus. 14.8 x 5.9 cm. On the recto twenty lines, containing Sura LIV. 11-38, are written in black ink parallel to the horizontal fibres, the verso bears eighteen lines, running at right angles to the vertical fibres, containing verses 45-55 of Sura LIV, and verses 1-32 of Sura

On the recto:

On the verso :

Plate II

LV. In line 4, at the end of verse 50, a verse-division mark is visible. Sura 54 is divided from LV by two parallel horizontal lines running over the full width of the page, and filled in with an ondulating line with pearls in the compartments.

Place of discovery unknown.

The script has become obliterated in some places and the papyrus is torn upon all sides. The fragment shown here (Plate 1) comes from the middle of the page, which as the line originally contained about threetimes as much as at present, would have been about 17.7 cm. wide. This Qur'an was certainly destined for private use only, and belongs to the small, oblong sizes, particularly used for private owners already in very early times.

Before entering into the question of dating40 it will be recommendable to give a palaeographical analysis of the script of P. Michaelides no. 32. The Alif (line 8), the most significant letter of the Arabic alphabet, as J. v. Karabacek has said,41 shows a definite bend to the right side, as it is already the case in PER Inv. Ar. Pap. 94 (ca. 30 A.H., 650 C.E.).42 It differs, therefore, from the straight, vertical Alif in PERF no. 558,43 in the bilingual protocol BM 1473,44 in Arabic Palaeography P1. 43 and in the Qur'an Paris. Arabe 32845 as well as from the Alif, curved to the right at the basis, shown by the Qur'an Medina la in the Saray in Istanbul.46

The Dal corresponds to the form offered by PERF no. 558, P. Berol. 15002 (Plate IIb), P. Mich. 6714 (Plate Ila), P. Berol. 9177 (124 A.H., 742 C.E.) and the parchment Qur'an Paris. Arabe 328 (1).

The Rd (Zay) is similar to the same letter in PERF no. 558, P. Berol. 15002, PERF no. 573 (Plate IIc, 57 A.H., 677 c.E.).

In Sin, the tail of the final letter (recto line 6) goes down in an almost straight line, as in the Nun occurring in PERF no. 558, 573, PER Inv. Ar. Pap. 94. It therefore differs from the final Sin with a curved tail occurring, for example, in PERF no. 558.

The Ta has a rectangular body and a vertical stroke bending to the right side, as, for example, in Paris. Arabe 328 (4).47 The latter characteristic feature already occurs in PERF no. 558 and P. Loth II, further in the legend of copper coins from the second decade of the first century of the Higra48 and in the Qurra-papyri,49 where it interchanges with the straight form. This characteristic feature is preserved in the third century of the Higra (ninth century C.E.) in the papyrus script,50 and in early Christian Arabic manuscripts forming the transition to Maghrabi-writing,51 in which it is preserved until recent times. Possibly this form of the Ta was also significant for the Makki-script, while the old scripts of al-Kufa, alBasra and Damascus preferred the vertical stroke.

Plate III

The Nun has about the same form as in PER Inv. Ar. Pap. 94.

An additional characteristic feature is the reverted Yd (e.g., in recto line 2 and in L-- verso lines 9, 16).

All these characteristics show, together with the general impression of the writing, that the script of P. Michaelides no. 32 cannot be dated later than the first century of the Higra (end of the seventh or beginning of the eighth century c.E.).

Some orthographical peculiarities should be added here.

The Alif madda is frequently omitted:

A remarkable thing here is -,L-JI in line 18 for -,I and in P. Michaelides no. 235 line 2 (first century A.H., seventh century C.E.). Diacritical dots are used sparingly and are formed in the same way as in PERF no. 558 already described in CPR III, 1/i, pp. 70-71. Once a short dash is used in Dal (on the recto line 17 in --- ^dL instead of a dot. Such diacritical dashes are used very early. We find them-apart from the parchment Qur'an Paris. Arabe 328 (1)52-in PER Inv. Ar. Pap. 8181 (first cent. of the Higra, seventh century c.E.), 10136 (second century of the Higra, eighth century c.E.), Inv. Perg. Ar. 186, in P. Mich. 6714 (22-75 A.H., 643-694 C.E.), where the short dashes are similar to oval dots. There can be no doubt whatever that in all these examples the dots or dashes respectively are set by the original hand.

The occurence of dots in such an early Qur'an as P. Michaelides no. 32 is of some importance, since Prof. A. Jeffery,55 the greatest living Western authority in Qur'anic science, emphasizes the fact that the oldest Qur'anic codices were generally without diacritical points, and the lines replacing dots (occuring only in some old codices and fragments) may well be nothing more than scribal fantasy.

Plate IV

It is quite possible that the dashes, used as diacritical marks in some old Qur'an manuscripts, for example, Arabic Palaeography Pl. 1 ff., were added later. But the occurence of such dashes-and dots-would no longer militate in se ipso against a dating in the first century of the Higra, since dashes are not unusual in papyri of the first century of the Higra and the use of dots is proved by papyri as early as even the first half of the first century of the Higra.54

Another important statement, made by Prof. A. Jeffery55 is that the oldest Qur'ans had no rubrics or marks for divisions, which were reproved by traditions traced back to early authorities, and that the occurrence of marks indicating divisions of ten verses would argue against so early a dating as the first to second century of the Higra, attributed by Miss N. Abbott to some specimens in the collection of the Oriental Institute in Chicago University.

As a matter of fact, there is one mark indicating the ending of a group of ten verses after verse 50 of Sura LIV on the verso of P. Michaelides no. 32 line 4. It consists of a a surrounded by dots.56 It is said57 that the marking of the endings of groups of five or ten verses was introduced by al-Haggag ibn Yusuf, the governor of the Iraq (75-95 A.H., 674-714 c.E.) or by Nasr ibn cAsim al-Laithi (died 89/90 A.H., 708/709 c.E.).58 The occurence of a ten-verses division mark in P. Michaelides no. 32 argues in favor of these authorities, and shows that Qur'an copies containing these marks could well be placed in the time suggested by Miss N. Abbott. Another question is the origin of these marks.

I have suggested already in 1929 that these division marks were an imitation of the use of plain circles as punctuation marks in Pahlawi literary papyri, for example, PERF no. 446 and P. Berol. 4442.59 Such plain circles recur still in Arabic papyri of the second and third centuries of the Higra (eighth/ninth century c.E.),60 and when it is just an `Iraqi governor who is credited with its introduction, Persian influence is thereby very probable.

Still more important is the occurrence of an ornamental sura division after Sura LIV on the verso of P. Michaelides no. 32 between lines 6 and 7. It is formed by two parallel horizontal lines, framing an undulating line, the curves of which are filled in with pearls.

Plate V

The oldest extant parchment Qur'ans are apparently destitute of any rubric or indication of the ending of a sura or the beginning of a new one, and there are even several traditions disapproving such a practice, which, nevertheless, made its way into copies of the Qur'an of the first century of the Islamic era, and even those destined for private use, as the example of P. Michaelides no. 32 shows. A comparison with another fragment of a papyrus Qur'an in the Collection Michaelides (no. 190, Pl. IV), forming the leaf of a quire, comprising two columns on each side, folded in the middle, which contains Sura LIX.11 to LXV.4, furnishes interesting details for the early adornment of Qur'an manuscripts (see Plate IV). This fragment is apparently considerably later than P. Michaelides no. 32, and may, according to the script, appartain to the end of the second or beginning of the third century of the Higra (first half of the ninth century C.E.).

Here we see a simple intertwined band ornament at the end of Sura LXI (on the recto, left column, line 8) and perhaps also at the end of Sura LXIV (on the verso, right column, line 16). The end of Sura LX (on the recto, right column, line 14) is marked with a hexagram and an intertwined band ornament. The simple design in this papyrus codex was obviously more elaborate in large copies of the Qur'an, destined for the use in the mosque; a good example of an ornamented band concluding the sura is offered by the codex Saray no. 50395 in Instanbul.61

The development of the practice of emphasizing the division of the chapters in Qur'an manuscripts may have happened in the following way:

The first step was to an empty space between the end of a preceding sura and the beginning of the following sura;62 such empty spaces divide also the various parts of documents and occasionally of letters.63

Then a simple ornament, for example, a composition of intertwined or intersecting lines, as we see it in P. Michaelides no. 32, 190, marked the end of a sura. Possibly such bands were taken over from Greek or Syriac manuscripts, in which they marked the beginning of a new chapter or paragraph.

In a more elaborate execution, an intertwined band, beautifully ornamented, extended over the whole lenghth of a page in big liturgical copies of the Qur'an, as we see it in the Qur'an from the 'Amr-Mosque64 or that from Sarmaqand.65 These ornamental flourishes are often a clear imitation of the clavi in late Roman textiles.

In the meantime, it had probably become customary to mark the beginning of as sura by a special formula mentioning its origin (whether Meccan or Medinan), the name of the sura and the number of its verses.

In P. Michaelides no. 190 k ' .-1 ~,,.., a.{~.,62 accompanied by a simple intertwined band, opens sura LX (on the verso, left column, line 13), while the same plaited band is repeated after the Basmala (line 14).

The opening of sura LXI is unfortunately partially destroyed. It begins with the Basmala, followed by an intertwined band, then a lacuna ands r,' t, '° (on the recto, right column, LINES 14-15).

Still more damaged is the beginning of sura LXIV (on the verso, right column, line 3), where j i c,L -,*, mil, followed by an intertwined band and the Baslama, opens the chapter. It is not to be made out, whether also here i € - and respectively, are to be supplemented in the lacuna.

Since this fragment comes apparently from a Qur'an destined for private use, the scribe was not restricted to the observation of a severe tradition of adornment, but was able to apply a liberal selection of various ways of indicating the beginning of a sura.

Finally, the title of the sura, including the mentioning of its place of origin and the number of its verses, thus forming a real rubric, was set within the intertwined band, which now formed its ornamental frame. This last step of a long development was then standardized, and survived in the old lapidary style of writing within the embellished copies of Qur'an, now written in beautiful, big Thuluth or other, younger scripts.

We may summarize our article under the following headings:

(1) The supposition of Minovi, that the extant early Qur'ans are wholesale forgeries, is not only unjustified, but in itself highly improbable.

(2) The sample of Meccan-Medinan writing in the Chester-Beatty codex of the Kitab al-Fihrist by Ibn an-Nadim does not really represent the original Meccan-Medinan script, but we can form an opinion of this script by a close study of the writing in the papyri of the first century.

(3) These papyri can indeed form the basis for the dating of early Qur'ans, and it is fairly possible to date some of them in the first century of the Islamic era.

(4) The existence of diacritical dots or dashes, or even of punctuation and verse-decade marks and rubrics in those early Qur'an manuscripts would not prejudice a dating in this period, and Arabic sources, mentioning these peculiarities as existing in Qur'an codices of the first century of the Higra, are reliable.

So a small fragment of a simply adorned Qur'an leaf has contributed essentially to our knowledge of the early history of the Holy Book of Islam, and it is to be hoped that future finds may widen this knowledge and help to solve the problems whose solutions are still pending.



1. Frammenti Coranici in carattere Cufico nella Bibliotheca, Vaticana, (Codici Vaticani Arabi 1605, 1606), Studi e testi no. 132, Citta del Vaticano, 1947.

2. The present article was originally destined as a contribution to the Scritti offerti a G. Levi della Vida, but could not be completed in time for reasons of health.

3. Arabic Palaeography by B. Moritz, Pl. 43, containing Sura XXVIII. 48-57. J. V. Karabacek, WZKM 20 (1906): 137 expressed his doubts concerning the existence of papyrus Qur'ans, since this plate did not convince him owing to the bad state of the verso of the papyrus.

4. Loth II, cf. ZDMG 34 (1880): 688.

5. Lettre a M. le Chevalier P. O. Bronsted sur quelques Medailles cufiques dans le Cabinet du Roi de Danemark ... et sur quelques manuscrits cufiques, Copenhagen, 1830, paragraph 8, pp. 33 ff.

6. Palaographische Beitrage aus den herzoglichen Sammlungen in Gotha, fasc. 1, Erfurt, 1842.

7. Paleographie universelle, Paris, 1839, English edition by Frederic Madden (London, 1850), Pl. xxxii-xxxvi.

8. Facsimiles of Manuscripts and Inscriptions, Palaeographical Society, Oriental Series, London 1875-1883, Pl. xlix; the date is accepted by J. v. Karabacek, WZKM 5 (1891): 324, 20 (1906): 137 who places the piece in early eighth century c.E. or end of the first century and beginning of the second century A.H., respectively.

9. Arabic Palaeography, A Collection of Arabic Texts from the First Century of the Hidjra till the Year 1000 (Cairo, 1905), Publications of the Khedivial Library no. 16; cf. B. Moritz's remarks on Kufic Writing in Qur'ans in Encyclopaedia of Islam 1, p. 388. In his article "Ausfluge in der Arabia Petraea," Melanges de la Faculte Orientale (Universite Saint Joseph, Beyrouth) III (1908), p. 430, B. Moritz, dated the Qur'an from the `Amr Mosque ca. 100 A.H.

10. WZKM 20 (1906): 133 if., Sb. Ak. Wien 184/3 (1917): 12f., 33 n. 2. In Th. W. Arnold and A. Grohmann, The Islamic Book (Florence 1929), p. 22, I have tentatively dated the Qur'an manuscript reproduced in Pis. I-XII at about 107 A.H. (725 c.E.) and that reproduced in Pls. XXXI-XXXIV at 102 A.H. (720 c.E.).

11. Die Ausstellung von Meisterwerken muhammedanischer Kunst I (Munchen 1910), Pl. 1.

12. Sb. Ak. Wien 172 (1913): 35 n. 1. Karabacek (in a note in his posthumously examined manuscripts) bases his contestation on the form of the Alifs, the foot of which shows a curved turn to the right, which form can not be eighth century according to him. But the evidence of the papyri disproves this argument.

13. Specimina codicum Orientalium (Bonn 1914), Pl. 41, 42.

14. ZDMG 64 (1915): 80 and P1. xvi.

15. "Zur altesten Geschichte der kufischen Schrift," Zeitschrift des Deutschen Vereins fur Buchwesen and Schrifttum 5/6 (1919): 54-66, especially pp. 55a n. 3, 66.

16. Noldeke-Bergstrasser-Pretzl, Geschichte des Qordns III (Leipzig 1936), S. 251-257.

17. Ibid., p. 253.

18. E. Herzfeld has mentioned a Qur'an copy, dated 94 A.H. (712/13 C.E.), among the collections of valuable books in Persia (cf. Ephemerides Orientales O. Harrassovitz 28. 1. 1926). For two copies, dating 102 A.H. (720 C.E. B. Moritz, Arabic Palaeography, Pl. 31-34) and 107 A.H. (725 C.E., ibid., Pl. 1-12), respectively, according to information obtained in the Egyptian National Library cf. Arnold and Grohmann, The Islamic Book, p. 22. Cf. also I. Y. Kratchkovsky, Among Arabic Manuscripts (Leiden, 1953), p. 150: "Kufic Qorans of the firstsecond century A.H. are extremely rare. . . ." The Qur'an Masahif no. 387 in the National Egyptian Library (Arabic Palaeography, Pl. 18) dated by B. Moritz, basing his conclusion on the waqfiyya, in the second century A.H., is, according to J. v. Karabacek, WZKM 20 (1906): 135, 136, from the third century A.H. Moritz read the date 168 instead of 268 (882 C.E.).

19. Codex Paris. Arabe no. 336 229 A.H. (843/44 c.E.); Cairo National Egyptian Library General number 40160 256-264 A.H. (870-877 C.E.), 33910 270 A.H. (883/84 c.E.), 33910 277 A.H. (890/91 c.E.); two fragments of parchment Qur'ans in the National Museum in Damascus 265-271 A.H. (878-885 c.E.) and 298 A.H. (910/11 C.E.); one Qur'an manuscript in Persia, dated 260 A.H. (873/74 C.E.), is recorded by E. Herzfeld, op. cit.; cf. B. Moritz, Encyclopaedia of Islam 1, p. 388, G. Bergstrasser, Geschichte des Qorans, III, p. 270.

20. Kratchkovsky, Among Arabic Manuscripts, p. 150.

21. S. Pissaref, Reproduction exacte du celebre Coran Coufique ecrit, d'apres la tradition, de la propre main du troisieme Calife Osman (644-656) et se trouvant maintenant dans la Bibliotheque Imperiale publique de Saint-Petersbourg (St. Petersbourg, 1905).

22. "Arabic Palaeography," Ars Islamica. 7 (1941): 73, 74 n. 14.

23. Moslem World 30 (1940), p. 192.

24. A. V. Pope, A Survey of Persian Art 11 (1939), p. 1718 and note 4. Cf. N. Abbott, op. cit., p. 68.

25. Pope, A Survey of Persian Art, p. 1710.

26. Ars Islamica 8 (1941): 70 ff.

27. Pope, A Survey of Persian Art, p. 1710.

28. Apparently Minovi found his "final hooked (terminus) in the opposite direction" substantiated in the fishing-hooks or barbs of the Alifs in the Basmala of the Chester-Beatty codex of the Kitab al-Fihrist. We have to return to this point again later.

29. E.g., in as-Su li, Adab al-Kuttab (Baghdad 1922), pp. 43f. cf. also F. Krenkow, "The Grant of Land by Muhammed to Tamim ad-Dari," Islamica 1 (1924): 529-532.

30. Pope, A Survey of Persian Art, p. 1718.

31. Ibid., p. 1710, fig. 580.

32. The same has happened to Karabacek in WZKM 5 (1891): 323.

33. Moslem World 30 (1940): 193.

34. This expression is not very luckily chosen, but if we compare, e.g., the inscription B of al-Medina (Gebel Sala`) published by M. Hamidullah in Islamic Culture 13 (1939): plate opposite p. 435, the resemblance with the Kufi Qur'ans is very striking. The rich collection of photographs of old Qur'ans in the Bavarian Academy of Sciences in Munich would have been very useful for a more detailed classification of this style of writing.

35. Kitab al-Fihrist, pp. 9 ff., 10 ff.

36. Apart from the well-known Qurra-papyri (90-96 A.H., 709-714 c.E.) the following date from the first century of the Higra:

The parchment letter from Sogdiana about 100 A.H. (719 C.E.) V. A. Kratchkovskaia and I. Y. Krachkovski, Recueil Sogdien (Leningrad, 1934), pp. 52-90.

37. Cf. above, n. 8, E. Tisserant, op. cit. P1. 41 a, b.

38. Scritti offertia G. Levi della Vida, P1. 1, pp. 1 f.

39. Bergstrasser, Geschichte des Qorans, P1. viii.

40. The dating second/third century of the Hijra, given in From the World of Arabic Papyri, p. 229 n. 268, refers to the papyrus Qur'an leaf P. Michaelides no. 190 (not to no. 32).

41. WZKM 5 (1891): 323.

42. Cf. my "Apercu de papyrologic Arabe," Etudes de papyrologie 1 (1930): Pl. ix.

43. Cf. the Table of Writing in CPR III, 1/2, p. xxii.

44. Cf. ZA 22 (1908): Pl. 1 line 6.

45. Cf. Bergstrasser, Geschichte des Qorans, Pl. vii.

46. Ibid., P1. viii.

47. E. Tisserant, op. cit., Pl. 41b.

48. Cf. J. v. Karabacek, Beitrage zur Geschichte der Mazjaditen (Leipzig 1874), p. 35; WZKM 5 (1891): 324.

49. APEL III no. 147, 3, 5 P1. 11(91 A.H., 710 c.E.).

50. APEL III no. 167, 94, 06 Pl. xi (140 A.H., 757 C.E.); 180, 5, 7 P1. xv (113 A.H., 731/32 c.E.); 201, 26 Pl. xxi (ca. 116 A.H., 734 c.E.); APEL 11 no. 79, 8, 9 Pl. iv (third century A.H., ninth century c.E.), 82, 4 Pl. vi (253 A.H., 867 c.E.); 91, 3 Pl. vii (second/third century A.H., ninth century C.E.); 120, 4 Pl. xix (third century A.H., ninth century C.E.), APEL IV no. 2335,7 Pl. vii (third century A.H., ninth century C.E.); J. D. Weill, Le Djdmi'd'Ibn Wahb 1 (Cairo 1939), Pl. 5, 7 (second half of the third century A.H., ninth century C.E.).

51. E.g., in the translation of the New Testament, which H. L. Fleischer, "Beschreibung der von Prof. Dr. Tischendorf zurackgebrachten christlich-arabis- chen Handschriften," ZDMG 8 (1854): 585, dates in the eighth, at the latest the ninth century C.E.

52. E. Tisserant, op. cit., P1. 41a.

53. Moslem World 30 (1940): 195, 198.

54. PERF no. 558, P. Berol. 15002 (both 22 A.H., 643 C.E.); PERF no. 573 is dated 57 A.H. (677 c.E.), one year before the inscription of the dam near at- T'if, in which diacritical dots occur frequently (cf. G. C. Miles, "Early Islamic Inscriptions near Ta'if in the Hijaz," JNES 7 (1948): 237 and P1. xviiiA). Diacritical dashes also occur in Christian Arabic manuscripts, e.g., in the fragment of an Arabic translation of Job from the ninth century c.E.; cf. H. L. Fleischer, "Zur Geschichte der arabischen Schrift," ZDMG 18 (1864): 288-291 and plate in front of pag. 288.

55. Op. cit., p. 196.

56. Cf. CPR III, 1/1, p. 73. In the codex Vaticanus Arabicus 1605 the versedecade mark consists of a red circle surrounded by black dots with a numeral filling in the circle (cf. Scritti Offerti a G. Levi della Vida, p. 2).

57. Cf. A. Jeffery, Two Muqaddimas of the Qur'anic Sciences (Cairo 1954), p. 276 if. Some historians traced the use of such verse-decade marks back to the Caliph al-Ma'mun (198-218 A.H., 813-833 c.E.), but this is obviously too late.

58. Cf. Bergstrasser, Geschichte des Qorans, III, p. 258.

59. Arnold and Grohmann, The Islamic Book, p. 23.

60. E.g., PERF no. 712, 734, PER Inv. Ar. Pap. 814; J. D. Weill, Le Djami`d'1bn Wahb, Pl. xxx pag. 61 line 15. Such a very regular circle also occurs in the parchment Qur'an Paris. Arabe 334, cf. Bergstrasser, Geschichte des Qorans, III, Pl. 1, fig. 2.

61. Bergstrasser, Geschichte des Qorans, III, Pl. V, fig 6a.

62. Ibid., p. 259.

63. Cf. my From the World of Arabic Papyri, p. 89.

64. Cf. Moritz, Arabic Palaeography, P1. Iff.

65. See n. 20

66. For this formula cf. Or. 2166 in W. Wright, Palaeographical Society, Oriental Series, Pl. xLix line 17

N. Abbott, The Rise of the North Arabic Script, P1. xvii no. 11 verso and Berstrasser, Geschichte des Qorans, III, p. 259. He also mentions a Qur'an codex in the possession of the grandfather of Malik ibn Anas (died 179 A.H.), written in the time of the Caliph 'Uthman, containing sura subscriptions, written in black ink on an ornamental band, which ran over the whole line

If you find an error please notify us in the comments. Thank you!