A display case in Hyderabad
In the previous article we saw how the orientalist school played its strongest card in the twentieth century, the idea of a “different Quran” beneath a layer of ink or in a forgotten attic, and how that “different Quran” dissolved every time the material was read with care. Both in Mingana’s palimpsest and in Puin’s Sana’a fragments, after all the framing, what remained was the ordinary Quran, with a text that stood fixed down to the letter. But one of Puin’s objections we deliberately left there not yet fully worked out: his “greatest find” of a partial Mushaf in which the surahs stood in a deviating order. That very find opens a question the previous articles still left lying. In this article we move ourselves twelve centuries forward and several thousand kilometres to the east, to a museum in Hyderabad, in order to answer that question there. The question runs: if the text was so strictly guarded, then what of the order of the surahs. Were the one hundred and fourteen surahs fixed in the same iron grip as the letters themselves, or was the believer free to arrange them otherwise. And if the latter was so, does that not undermine the whole thesis of an unchangeable Book.
The answer lies, remarkably enough, in a dusty bookcase of the Salar Jung Museum and its attendant library in Hyderabad, India. There resides a collection of Arabic manuscripts that was catalogued by the scholar Muhammad Ashraf in his work A Catalogue of Arabic Manuscripts in Salar Jung Museum and Library. Al-Azami ﵀ drew a few dozen examples from that catalogue, from pages 166 to 234 of it, and set them in a table in his own work. That table is, at first glance, a dry list of manuscript numbers, counts of surahs, and dates. On closer inspection it is one of the most illuminating documents in the entire textual history of the Quran. For that list shows, in black and white, something that in words must be explained over and again, but which here speaks for itself in columns: that the order was free and the text was fixed.
Before we enter that table, we must first understand what precisely lies on the table. And for that we must draw a distinction that in the popular imagination is almost always confused: the distinction between a complete Mushaf and a partial Mushaf.
The difference between a whole and a half Mushaf
A complete Mushaf is the entire Book, from Al-Fatiha to An-Nas, one hundred and fourteen surahs in the established Uthmanic order. Such a copy was written in the early centuries on parchment, which was considerably heavier than paper, so that a complete manuscript could weigh several kilograms. Al-Azami describes in his work a page from the Yemeni collection whose calligraphy was so large that the thickness of a complete Mushaf written on that scale would have exceeded well over a metre. When he takes the Mushaf that the King Fahd complex in Medina prints as his standard, with its roughly six hundred pages and nine thousand lines, then one such Yemeni line would amount to half a line in the Medinan print, so that a complete Mushaf in that hand would run to eighteen thousand pages. Such heavy calligraphy was by no means rare.
Now such a book was unwieldy. A traveller who wished to carry the entire Quran bore a burden with him. Al-Azami even draws a light parallel here with his own habit: when he travels by aeroplane and wishes to take his work with him without hauling a bulky volume in his suitcase, he simply copies the portions he needs during his journey. The Muslims of the first centuries did precisely the same. Whoever did not need to write out the entire Quran, but only those surahs he used most in prayer, in teaching, in daily recitation, made a partial Mushaf: a manuscript with a selection of surahs, thin enough to carry, practical enough to use.
And here comes the decisive point. For a complete Mushaf the rule held inexorably: whoever writes out the Quran in its entirety must follow the established surah order of Uthman’s ﵁ Mushaf. Muslim scholars are unanimously agreed that the present ordering of the surahs is identical to that of Uthman’s ﵁ copy. But for a partial Mushaf it lay otherwise. Whoever copied only a handful of surahs, for his own personal use, was not bound to the arrangement of the complete edition. He was permitted to set the surahs down in whichever order suited him best.
That sounds, to the modern ear, alarming. Does this not mean that countless different Qurans were in circulation, each with its own ordering. Does this not mean that the text was unfixed. The answer is an emphatic no, and to understand that we must draw a distinction that forms the core of this whole article, and at the same time the core of the whole thesis of this map.
The ordering is no part of the revelation
There is a difference between the text of the Quran and the order of its surahs. The text, the consonantal skeleton, the words, the letters, is divinely revealed and unchangeable. The order in which an individual copyist arranges his selected surahs in a personal copy is not so in the same binding way.
Al-Azami sets this out at the beginning of his treatment. Scholars are unanimously agreed, he writes, that following the surah order is not obligatory in prayer, in recitation, in learning, in teaching, or in memorizing. Each surah stands on its own. A later surah does not necessarily carry more juridical weight than an earlier one; sometimes an abrogated verse, a naskh, even appears in a surah that follows the surah with the verse that replaces it. And he points to a habit that anyone who has memorized the Quran will recognize: most Muslims memorize the Quran from the back, beginning with the shortest surahs, numbers one hundred and fourteen, one hundred and thirteen and so on, and work their way forward. No one supposes that he thereby mutilates the Book. The ordering is a service to memory and to use, not a dogma that makes every relocation into heresy.
There even exists a report, recorded by Muslim in his Sahih, that the Prophet ﷺ once recited in a single rak’a the surahs al-Baqara, an-Nisa, and Al-Imran, that is to say surahs number two, four, and three, in that order, which deviates from their order of appearance in the Mushaf. The recitation was not wrong. The order of surahs in the recitation is not the same fixed quantity as the text of the surahs themselves.
Who ordered the surahs at all in the complete Mushaf is an old discussion among the classical scholars, and Al-Azami renders it honestly. There are, so far as he knows, no ahadith in which the Prophet ﷺ expressly indicates the order of all the surahs. The opinions diverge and may be summarized in four positions. The first position, which Al-Azami himself holds, maintains that the ordering of all the surahs as it now lies before us goes back to the Prophet ﷺ himself. The second position holds that the entire Quran was ordered by the Prophet ﷺ, with the exception of surah number nine, which was placed by Uthman ﵁. The third position, defended by al-Baqillani, ascribes the ordering of all the surahs to Zayd ibn Thabit ﵁, to Caliph Uthman ﵁, and to the Companions ﵃. The fourth position, of Ibn Atiyya, holds that the Prophet ﷺ ordered some surahs and the rest were ordered by the Companions ﵃.
Whatever one holds, and herein lies the nuance that undergirds the whole article, in all four positions it is the order that is at issue and not the text. No classical scholar ever disputed that the words of every surah stood fixed. The question concerned solely the arrangement. And precisely because the arrangement of a personal partial copy was not binding, a copyist could take that liberty without touching a single letter of the revelation.
It is worth pausing for a moment on the character of this classical disagreement, for it is illuminating for the honesty of the tradition out of which it arose. The classical scholars had no objection whatever to acknowledging openly that they did not know with absolute certainty who exactly had set the surahs in their final order. They recorded the four positions side by side, with their respective proponents, without elevating any one of them by artifice into some certainty. Al-Azami remarks that, so far as he knows, there exists no hadith at all in which the Prophet ﷺ expressly fixes the order of all the surahs. Where such an explicit text was lacking, the scholars did not invent a text. They preserved the disagreement.
That is in itself a sign of reliability. A tradition inclined to falsification would have plastered over the gaps in its knowledge with invented certainties. A tradition that places truth above the appearance of certainty leaves the open question open. And the striking thing is that this openness concerned solely the order. Over the text of the surahs there existed no disagreement whatever worth mentioning. The scholars quarreled over the arrangement and were of one mind in their silence over the content, and that is precisely the relation we expect when the first is a human ordering and the second a divine revelation. That the second position ventures one single exception, namely surah nine, and ascribes it to Uthman ﵁, changes nothing in that picture: there too it is a matter of the placement of a surah, not of a word within that surah.
Here lies the key that opens the door to Hyderabad. If a copyist was free to order his surahs but the text within each surah was inviolable, then a large collection of personal partial Mushafs from different centuries ought to show precisely that pattern: a wild diversity in order, and an utter uniformity in text. And that is precisely what the Salar Jung table shows.
The table of Salar Jung
Let us enter the table as Al-Azami presents it, drawn from the catalogue of Muhammad Ashraf. It is a list of a few dozen partial Mushafs from that one library in Hyderabad, each with a manuscript number, the number of surahs it contains, the order in which those surahs stand, and, where known, a date. We need not walk through the whole list line by line, but a few examples suffice to make the pattern visible.
Manuscript number 244 contains twenty-nine surahs, in the order 36, 48, 55, 56, 62, 67, 75, 76, 78, 93, 94, 72, 97 and then 99 through 114. Manuscript 246 contains sixteen, beginning with surah 62, of which only the first eight verses, followed by 110, 1, 57, 113 and onward. Manuscript 247 contains ten: 1, 36, 48, 56, 67, 78, 109 and 112 to 114. Manuscript 250 contains nine: 20, 21, 22, 63 and 24 to 28. Already with these first four, two things stand out. First: no two of them carry the same order. Second: certain surahs, in particular numbers 36, 48, 56, 67 and 78, recur again and again, as though they belonged to the most copied and most beloved. Surah 36 is Ya-Sin, surah 48 is Al-Fath, surah 56 is Al-Waqi’a, surah 67 is Al-Mulk, surah 78 is An-Naba. They were the surahs the believer recited most often, and therefore had written out separately most often.
When we look further down the list, we see the same pattern repeat itself in endless variation. Manuscript 248 contains nine surahs in the order 73, 51, 67, 55, 62, 109 and 112 to 114. Manuscript 249 carries nine as well, but an entirely different nine: 17, 18, 37, 44, 50, 69, 51, 89 and 38. Manuscript 251 carries eight: 6, 36, 48, 56, 62, 67, 76 and 78. Manuscript 252 carries eight in yet another arrangement: 1, 6, 18, 34, 35, 56, 67 and 78. Manuscript 255 contains eight surahs in the order 1, 36, 48, 55, 67, 73, 56, 78. Manuscript 258 carries seven: 18, 32, 36, 48, 56, 67, 78. Manuscript 268 carries five: 36, 48, 56, 67 and 78. Manuscript 275 carries five: 36, 78, 48, 56 and 67. Manuscript 279 carries five as well, and the catalogue even records the name of the copyist, a certain Yaqut al-Musta’simi. Manuscript 280 carries five, but an entirely different five: 1, 6, 18, 34 and 35. Manuscript 296 carries five: 1, 36, 44, 67 and 78. Manuscript 308 carries only four: 6, 18, 34 and 35. Manuscript 310, the shortest in the series, contains surahs 6 to 9. The partial Mushafs range from four surahs at the small end to twenty-nine at the large, and within that span almost every conceivable combination and order is represented.
Two observations force themselves upon us the moment one surveys the table as a whole. The first is that the recurring core surahs, Ya-Sin, Al-Fath, Al-Waqi’a, Al-Mulk, An-Naba, present themselves each time in different mutual positions. In one copy Ya-Sin opens the series, in another it stands in the middle, in yet another it comes only after Al-Mulk. The same five, in dozens of different arrangements. If the identity of these surahs had known even a trace of textual movement, a collection that contains them so often and in so many orders could not possibly have concealed that trace. The second observation is that the surahs which did not belong to the core, surah 6 for instance, or the block 34 and 35, or the block 20 to 28, cluster together in smaller groups that likewise change from manuscript to manuscript. The copyist assembled his little book according to his own devotional habit, and that habit varied from person to person and from region to region. What did not vary, in none of these cases, was what stood inside those surahs.
Now bring the dates alongside. And here one must speak with the greatest care, for it is precisely on this point that the most common misconception about this table arises. These manuscripts are no Uthmanic folios. They do not stem from the first century AH. They are no remnants of the commission’s work in Medina. By far the most carry dates from the eleventh through the fourteenth century after the Hijra, that is to say from what on the Christian reckoning is the seventeenth through the nineteenth century.
A few exact dates that Al-Azami takes over from the catalogue make that concrete. Manuscript 248 carries the year 1076 AH, which corresponds to the year 1666 of the Christian reckoning. Manuscript 275 carries the year 626 AH, or the year 1228. Manuscript 262 is from 1115 AH, or 1704. Manuscript 263 is from 1278 AH, or 1862. Manuscript 266 is from 989 AH, or 1581. Manuscript 267 is from 1075 AH, or 1664. For the manuscripts that carried no inscribed date, Al-Azami notes that he took over the approximately estimated hijri date from the catalogue and prefixed it with the circa symbol, so that a reader can always see which date stood fixed and which was an estimate. That is a detail that marks the accuracy of the whole list: no number is presented with more certainty than it possesses.
We thus have here a collection that stretches over many centuries. A partial Mushaf from the year 1228, another from 1581, yet others from 1666, 1704, 1862. Copyists who lived centuries apart, in different generations, in different times, each with his own choice of surahs and his own arrangement of them. And that is precisely what makes the table so powerful. For despite all that diversity in order, across all those centuries, there is one thing that never varies.
What never varies
Muhammad Ashraf catalogued these manuscripts, and Al-Azami drew his table from them, with one column that speaks and one column that is silent. The column that speaks is the column of order: it cries out with difference. The column that is silent is the column that is not there, because it never needed to be filled in: the column of the text. For in none of these manuscripts, from whichever century, did the text of a given surah deviate from the text of that same surah in any other copy. Surah 36 was in manuscript 244 the same Ya-Sin as in manuscript 275, the same Ya-Sin as in the complete Uthmanic Mushaf, the same Ya-Sin as is printed today in Medina. Only the place where it stood in the personal copy differed.
This is the distinction that the previous articles built up theoretically and that this table confirms empirically. In the article about Uthman’s ﵁ commission we saw that the consonantal skeleton, the rasm, was guarded with the greatest strictness: no verse was included that had not been verified against living memory. In the article about the seven ahruf we saw that the controlled variation in pronunciation had been permitted by the Prophet ﷺ himself, without a letter of the text moving on account of it. And here, in this list from Hyderabad, we see the third level of that same principle: that even the order of the surahs, in a personal partial copy, could be free without the text changing so much as a hair’s breadth.
When a copyist in the year 1666 ordered the surahs in a partial Mushaf as 36, 78, 48, 56, 67, and another in 1704 ordered those same five as 36, 48, 56, 67, 78, then these were not two different Qurans. They were two personal selections from one and the same Quran, each arranged for the convenience of its owner, and each word for word identical to the text from which it was copied. The variation lay in the folder, not in the content. In the index, not in the text. In the bookmarks, not in the words.
Al-Azami draws from this whole table one sober inference, and he formulates it without any embellishment. Whoever wished to write out a partial Mushaf, he concludes, would have felt free to place the surahs in whichever order he saw fit. He claims no more, and no more is needed. The freedom of arrangement is no concession, no weakness, no crack in the wall. It is the logical consequence of a simple and centuries-old distinction: the ordering of the surahs in a personal copy was never part of the revealed text, and the revealed text was never part of the free ordering.
Why this disarms the falsification thesis
Here we must pause for a moment on the reason why this subject is more than a technical footnote for lovers of manuscripts. For the free order of partial Mushafs has been used in certain circles as a weapon against the reliability of the Quran, and precisely when one sees through the misunderstanding, that weapon turns back upon itself.
The reasoning of the skeptic runs roughly as follows. Look, he says, here we have manuscripts with different orders of surahs. So the text of the Quran was not fixed. So there were different versions in circulation. So the idea of one unchangeable Book is a later myth. This reasoning is old, and it is served up over and again in various guises.
The reasoning breaks on precisely the distinction that this table makes so clear. What varied was the order of complete, untouched surahs, not the text within those surahs. A manuscript that places Ya-Sin before Al-Mulk and another that places Al-Mulk before Ya-Sin do not differ over what stands in Ya-Sin or what stands in Al-Mulk. They differ only over the page on which the one comes after the other. That is no textual variant. It is an organizational choice, comparable to the question of whether one prints the one poem before or after the other in an anthology: the poems themselves remain word for word the same.
And the force of the argument grows still greater when one brings the centuries into it. If the text of the Quran had truly been fluid, if copyists had truly taken the liberty to change it, then a collection of partial Mushafs from the thirteenth, the sixteenth, the seventeenth, the nineteenth century would have shown a trace of textual deviations, a slow drift such as one does indeed encounter in the manuscript transmission of many another ancient writing. But that trace is entirely absent. Over more than six centuries of copying, by copyists who had never met one another and lived generations apart, the text of every surah remained identical while solely their mutual order varied. That is not the picture of an unstable text. That is the picture of a text that stood so fixed that even the full freedom the copyist possessed, the freedom to order, could bring forth no deviation whatever in the words.
Let one attend here to the nature of the freedom. The copyist was permitted to choose which surahs he included and in which order. He was not permitted to choose what stood in those surahs. The first was his personal domain; the second was beyond his reach, for that stood fixed in the rasm that the whole umma passed down from generation to generation, upheld by the oral chain of the huffaz who knew every word by heart. A copyist who had wished to change a word of Ya-Sin would have been immediately corrected by the very first person who knew the surah by heart, and in every generation those were innumerably many. But a copyist who wished to place Ya-Sin before Al-Mulk in his own pocketbook wronged no one and transgressed no rule whatever.
Two kinds of variation that may never be confused
To weigh the force of this proof properly, it is useful to set the partial Mushafs of Salar Jung alongside another phenomenon that came up in earlier articles: the seven ahruf and the qira’at. There too there was variation, and there too the text remained in essence one. But it is a matter of two very different kinds of variation, and it is illuminating to see how they relate to each other.
The variation of the ahruf concerned pronunciation: manners of reciting the same revelation permitted by the Prophet ﷺ himself, carried by an unbroken chain of teachers, without loss of meaning. That variation lived in the mouth of the reciter and was not excluded by the undotted Uthmanic skeleton but rather accommodated by it. It was, if one will, a variation in the text itself, but a legitimate, bounded and documented variation, not a free one.
The variation of the partial Mushafs concerned something entirely different: not the pronunciation of a word, but the place of a whole surah in a personal copy. It lived not in the mouth but on the page, and it did not touch the text at all. Where the qira’at were guarded by strict isnad chains, because they concerned the recitation of the revelation, the arrangement of a partial Mushaf was guarded by nothing, because it added nothing to the revelation and took nothing from it. It is a striking interplay: the variation that touched the text, the qira’at, was precisely strictly regulated, while the variation that was free, the arrangement, did not touch the text. The tradition laid its strictness precisely there where the revelation was at stake, and let go where that was not so. That is the signature of a community that knew the distinction between the divine and the human precisely and applied it consistently.
Whoever confuses these two kinds of variation, and mistakes the free arrangement of the partial Mushafs for textual variation, arrives at the skeptical conclusion that the manuscripts themselves belie. Whoever keeps them apart sees two layers of the same discipline: one that guarded the recitation within permitted bounds, and one that let the organization of a personal copy go free without ever surrendering the text.
A collection that testifies of itself
There is something special about the kind of proof that the Salar Jung table provides, and it is worth the trouble to name it. Muhammad Ashraf did not catalogue these manuscripts in order to make a theological point. He was a librarian who described a collection, manuscript after manuscript, with the meticulousness of his craft. The dates, the numbers of surahs, the orders, he noted them because that was his work, not because he wished to prove a thesis about the reliability of the Quran. And precisely therein lies the force of the testimony. It is proof that was not gathered in order to prove. It is a by-product of accurate cataloguing, and such unintended testimonies are in historical scholarship often the most convincing, because they are free of any suspicion that they were constructed for the occasion.
Al-Azami also did not choose this collection because it was exceptional. On the contrary, he emphasizes that the libraries of the world are filled with partial Mushafs, and that the Salar Jung collection is only one single library from which he drew a few dozen copies by way of example. The pattern that this list shows, free order and fixed text, is not a curiosity of one Indian collection. It is the normal, worldwide picture of how Muslims dealt with the Quran for centuries: the text preserved with reverence down to the letter, the arrangement of personal copies handled with practical freedom.
That the collection is found in Hyderabad, far from the Arab heartlands of early Islam, strengthens that picture still further. For it shows that the distinction between the fixed text and the free ordering was not a local Arab custom, but a universally understood principle that had spread with Islam as far as India, and that there was applied with the same accuracy by copyists who handled Arabic as a sacred language and not as a mother tongue. A copyist in the Hyderabad of the seventeenth century knew just as well as a copyist in the Medina of the seventh century that he must not touch the words of Ya-Sin, and that he might place Ya-Sin before or after Al-Mulk as it suited him.
Let one consider what that geographical and temporal spread means for the evidential force. The early copies from the heartlands have been treated in other articles of this series, where the manuscripts of Sana’a, of Istanbul, and of Tashkent had their say. Those copies testify to the fixity of the text at the beginning of the chain, close to the source. The partial Mushafs of Salar Jung testify to that same fixity at the other end: centuries later, thousands of kilometres further, in the hands of copyists who could never have met the first generation of Muslims. When the text stands fixed at the beginning of the chain, and at the far end of the chain still proves to be exactly the same text, then the unbroken fixity over the whole length in between is no longer an article of faith but an observable fact. The early manuscripts and the late partial Mushafs are two gauges in the same stream, and they indicate the same level.
Precisely because the copyists of Hyderabad allowed themselves every freedom of arrangement, their fidelity to the text becomes all the more telling. They were not copyists who out of fear of authority took over letter by letter what was prescribed to them; they made their own choices, assembled their own little books, ordered by their own judgment. And precisely these people, who dared to order as it pleased them, did not dare to change so much as a single word of the text. That was not because it was forbidden them on pain of anything, but because in their understanding of the Quran it was simply unthinkable: the text was not theirs. The order of their own selection was.
The text as anchor, the order as service
It is useful at the close of this article to set the two layers apart once more, because it is precisely their conjunction that carries the thesis of this whole map.
The first layer is the text itself: the consonantal skeleton, the words, the letters of every surah. This layer is divinely revealed and was preserved with the utmost strictness. We saw in earlier articles how Zayd ibn Thabit ﵁ under Abu Bakr ﵁ included no verse without written confirmation, how the commission of Uthman ﵁ tested every letter against living memory, how the whole umma bound itself for fourteen centuries to the rasm Uthmani down to the spelling of a single alif. This layer does not move. It is the anchor.
The second layer is the arrangement of the surahs in a personal copy. This layer was, for partial Mushafs, free. It was a service to use, to memory, to the convenience of the traveller and the reciter. Whoever gathered his beloved surahs together in a thin little book ordered them as suited him, and no one reproached him for anything. This layer moves along with the need of the user. It is the service.
The confusing of these two layers is the source of nearly all the misunderstanding about this subject. Whoever takes the freedom of the second layer for unfixity of the first draws a conclusion that the manuscripts themselves contradict. For precisely because the second layer was so free, and the first nevertheless remained over all those centuries so immovably fixed, these partial Mushafs prove with a force that no argument from words can equal, that the text of the Quran was inviolable. The freedom of the order is not in spite of the fixity of the text, but as it were a control experiment that measures that fixity. Give the copyist all the freedom he can reasonably have, let him choose and order at will, and behold: the text he never touches, for the text is not his to touch.
Thus the display case in Hyderabad, with its dusty partial Mushafs from the seventeenth and nineteenth centuries, joins seamlessly to the room in Medina in which the commission of Uthman ﵁ fixed its rasm twelve centuries earlier. What was sealed there in 25 AH is what returns here imperturbably in a thousand later copies. The order was permitted to breathe; the text was not, and never has done so.
In the whole of these articles this is one of the quietest pieces of evidence, and perhaps for that very reason one of the most convincing. No dramatic discovery, no sensational manuscript, no radiocarbon date that makes the newspapers. Only a list in a catalogue, taken over into a table, from which a scholar drew the simple and inescapable inference that the copyist was free in his ordering. And precisely in that simplicity lies its weight: that the fixity of the Quran was so self-evident that even the full freedom of the copyist could not touch it, and that the proof of it did not need to be sought, but simply lay waiting in the columns of a library catalogue for whoever took the trouble to read them.
The partial Mushafs of Hyderabad testified to that fixity at the late end of the chain, centuries after the revelation. But the strongest testimony is the one that lies closest to the source. Where this catalogue knew no sensational manuscript and no radiocarbon date, in the next article precisely those await: the physical, early parchments that in 1972 came to light from a collapsed ceiling of the Great Mosque of Sana’a, and that let the text be read off not in a nineteenth-century copy but in the oldest parchment known to us. From the late partial Mushafs we thereby return to the beginning of the chain, to the first great material witness.
والله أعلم