The Palimpsest That Proves Nothing: Mingana, Puin and the "Other Quran"

Two Manuscripts, One Promise

In the preceding article we followed the codification of the seven canonical qurra’ under Ibn Mujahid ﵀, and we saw how the reading tradition thereby received its formal seal. The text now lay entirely fixed: the written skeleton of Uthman ﵁, furnished with the diacritical dots and the vowel marks that Abu al-Aswad ad-Du’ali ﵀ and Khalil ibn Ahmad ﵀ had developed, and beneath it the oral chain that carried the pronunciation from generation to generation back to the Prophet ﷺ. Every word fixed, every reading tested, every deviation audible. Yet it was precisely against that fully grown certainty that orientalism, in the twentieth century, tried to play its strongest counter-card. For if the text lay so firmly fixed, so the reasoning went, then somewhere, beneath a layer of ink or in a forgotten attic, an older and different Quran must be found that would expose the canon as a later construction. Two manuscripts came into consideration for this, and two scholars sought to read that “other Quran” into them with all their force: Alphonse Mingana and Gerd-Rüdiger Puin.

This article is about the two manuscripts with which orientalism in the twentieth century tried to play its strongest card: the idea of the “other Quran”. The idea that somewhere, beneath a layer of ink or under the woodwork of an old mosque, a Quran lies waiting that is different from the Quran the Muslim recites today. A Quran with shifted verses, with different words, with a text that would expose the canon as a later construction.

There are two manuscripts, and one must keep them well apart, for journalism and polemic have repeatedly confused them with each other. The first is a palimpsest that Alphonse Mingana published in Cambridge in 1914. The second is the mountain of parchment that came to light in 1972 above the prayer hall of the Great Mosque of Sana’a and that Gerd-Rüdiger Puin studied in the nineties. Two manuscripts, two scholars, two centuries of dating claims. And, as we shall see, twice the same outcome: the “other Quran” dissolves the moment one looks closely.

It is worth saying beforehand what this article does not do. It accuses no scholar of malice without proof, and it exaggerates the evidential force of no fragment. It simply follows the text. Where Mingana believes he reads a variant, we lay his reading beside the manuscript he himself published. Where Puin made a concession in his own letter, we cite that letter. The facts speak louder than the framing that has been built around them.

The Man Whose Name the Collection Bears

Alphonse Mingana was the Chaldean priest who, between 1924 and 1929, with the money of the chocolate manufacturer Edward Cadbury, brought more than three thousand manuscripts to Birmingham, the collection that would later bear his name. But Mingana was not merely a collector. He was also a polemicist, and his views on the Quran were sharp and public.

Already in 1917 he had argued, in the journal The Moslem World, that the Arabs in the time of the Prophet ﷺ scarcely possessed a script of their own suited to recording the revelation. That thesis, that a people without a mature alphabet could not possibly have recorded a text like the Quran, was the foundation beneath a whole series of his conclusions. We have seen in an earlier article how that foundation collapsed: dated Arabic inscriptions at Zabad, at Jabal Usays and at Harran, well over a century before the birth of the Prophet ﷺ, display a mature Arabic script that closely approaches the palaeography of the first century after the hijra. Mingana’s premise had already been refuted by stones centuries older than his essay.

But Mingana took a further step. If the Quran had indeed been written down, then that written text, so he reasoned, must bear traces of a period before the canonization under 'Uthman ﵁. A pre-Uthmanic Quran, a text from the phase when there was as yet no standard, would have to contain the variants that the later standardization had smoothed away. And Mingana believed he had found that text.

Before we examine his find, a judgment of the man is fitting that arises not from hostility but from his own work. Azami ﵀ notes that Mingana, who has been called by some “a great scholar of Arabic”, in reality had a shaky command of the subject. When Mingana published a manuscript of the traditions of al-Bukhari ﵀ and in doing so transcribed only a handful of lines, he made a series of errors that Azami cannot classify as anything but incompetence: he transcribed words wrongly, omitted words he could not read, added letters that were not there, and translated terms incorrectly. This is no small matter. It concerns a scholar who could not even transcribe without error the clear manuscripts of a hadith compilation, let alone the far less legible ink of a palimpsest.

Hold on to that judgment. It is the key to everything that follows.

One might object that an error in transcribing a hadith text says nothing about someone’s ability to read a Quran palimpsest. But the opposite is true, and that is precisely why it weighs. The traditions of al-Bukhari ﵀ are transmitted in clear, carefully executed manuscripts, with dots and vowel marks, published and reprinted down to the smallest details. Whoever cannot transcribe even such a text without error lacks not a detail but a foundation. And if that same reader then ventures upon the most difficult category palaeography knows, a half-erased underlayer beneath a second text, in a script without dots, then one must approach his readings with the greatest possible suspicion. The easy text exposed the reader; the difficult text gave him the room to see what he wished to see. Azami makes of this sequence not a rhetorical trick but a methodological point: assess the reliability of a reader by the clear cases, before you trust him on the obscure ones.

What a Palimpsest Is

To understand Mingana’s find, one must first understand what he held in his hands. The word palimpsest comes from the Greek and means literally “scraped again”. In a time when parchment was costly, it happened regularly that an old text was scraped or washed off a sheet to make room for a new one. The old ink never disappeared entirely; beneath the new text a ghost of the old remained, a faint shadow that was often illegible to the naked eye.

The manuscript Mingana published was such a palimpsest, written on vellum. Originally it contained verses of the Quran. That Quranic text had later been washed away, and over it a Christian Arab had written. And here comes a crucial physical detail that one must not lose sight of: the two texts, the washed-away Quran beneath and the Christian text above, stand at right angles to each other. The new writer had turned the sheet a quarter turn before he began. That is precisely what a palimpsest is: two layers that cross each other, of which the lower one is half erased.

Because the lower layer was so faint, Mingana used infrared light to increase the contrast. He photographed three leaves under that light and tried to reconstruct the underlying Quranic text. The result he published in 1914 under a title that lays bare his whole agenda: Leaves from Three Ancient Qurâns, Possibly Pre-'Othmânic, with a List of their Variants, published at Cambridge, together with Agnes Smith Lewis. “Possibly pre-Uthmanic”, “with a list of their variants”. The title promises precisely what orientalism was looking for: a Quran from before the standard, with a list of deviations.

One must realize the difficulty of the task. A trained palaeographer reading a clear, unbroken text can already err. Mingana read a half-erased underlayer, cut straight across by a second text, in a script without diacritical dots and without vowel marks. In that bare skeleton, the rasm, every letter lies open to multiple interpretations. A single arc can be a ب, a ت, a ث, a ن or a ي, depending on dots that are not there. And it is precisely in that openness that Mingana found his “variants”.

The Four Variants of Mingana

Azami analyzes four of the variants that Mingana believed he read in the manuscript, and shows in each of the four the same pattern. It is not difficult, he writes, to detect Mingana’s dishonesty on this point, a dishonesty aimed above all at readers with little knowledge of Arabic. Let us go through the four, for together they form the heart of this article.

The first variant concerns a word that Mingana read as اللكم, which he translated as “blow, fist, boxing”. It is a passage corresponding to the printed Quran in Surah Ghafir. Where the text speaks of those who can do the believer no harm, Mingana constructed a reading in which there is talk of a “blow” or a “punch”. Azami notes drily that the word Mingana believed he saw belongs in the boxing arena, not in the Quran. The closing words of Mingana’s reading make no sense whatsoever in this context. The most charitable translation one can give of his construction runs something like: “out of wickedness they will not protect you against the punch”, a sentence that means nothing. That the last words are due to a scribal error is entirely obvious. But Mingana refuses to give it up; he clings to the absurdity as though it were a discovery.

The second variant comes from Surah Al-Isra, the first verse. Here Mingana laid his own reading beside the printed Quran. The Quran printed in Medina has at this place a particular spelling of the word “We have blessed”. Mingana introduced into his manuscript an alif that did not need to be there, and thus created a “variant”. Indeed, the word in question can mean both “to bless” and “to kneel”, and Mingana made use of that ambiguity: through his self-added alif he translated the one line as “blessed” and the other as “kneeled”. The variant was not present in the manuscript; it had been laid into the manuscript by Mingana, with a letter that came from his own pen.

The third variant comes from Surah At-Tawba, verse thirty-seven. It is no secret, writes Azami, that early scribes occasionally omitted vowel letters in their copies, the alif, the waw and the ya. At this place the scribe had omitted the last vowel in a word, simply because that vowel was silent. Mingana again abused this. He detached an alif from the one word and placed it after a negation, whereby an entirely new, ungrammatical phrase arose that is stripped of all meaning. Azami compares the trick to taking the English words “tigers hunting”, and reshaping them into “tiger shunting”. The letters are the same; by grouping them differently a nonsense arises that only the manipulator himself has created.

The fourth variant comes from Surah Ghafir, verse eighty-five. Here Mingana proceeds with somewhat more refinement, but the procedure remains the same. He shifts a ya from the one word to the other, and, and this is the crux, he adds his own dots to the dotless text to make a new word of it. The underlying Quran had been written in the bare rasm, without diacritical dots. Mingana placed the dots he needed to construct his variant.

Does one see the pattern? Four variants, one method. In each case the “variant” does not exist in the manuscript itself. It arises at the moment Mingana inserts an alif that was not there, or shifts a letter, or adds dots to a dotless skeleton. The “other Quran” of Mingana is no other Quran. It is Mingana’s own rereading of an erased skeleton, in a text that can be read every which way once one drops the dots and then oneself puts them back at will.

Here lies the deepest point of this whole article, and it deserves a pause to sink in. A text without diacritical dots is not ambiguous for the one who recites it, for the oral transmission holds the pronunciation fixed. But it is infinitely pliable for the one who manipulates it. Whoever may add dots where he wishes can make of almost any skeleton almost any word. And that is precisely what Mingana did. His “variants” prove nothing about the early Quran; they prove only that a dotless rasm allows itself to be tortured into a confession the tradition never made. The incompetence that Azami had earlier established in the hadith transcription returns here as method: the same man who could not transcribe the clear Bukhari without error presents his misreadings of an erased palimpsest as text-critical discoveries.

It is useful to see how ordinary are the building blocks with which Mingana raised his sensation. That early scribes occasionally omitted the vowel letters alif, waw and ya is no secret and no defect; it is the sparing nature of the early rasm, which left the oral pronunciation to memory and fixed on the parchment only the skeleton. Precisely those omitted letters formed Mingana’s raw material. Where a scribe had skipped a silent vowel, Mingana placed his own alif back in another spot, or he shifted one, and he called the result a variant. Where a skeleton lay open without dots, he chose the pointing that served his reading. The tool was in both cases the same innocent phenomenon that marks every early mushaf; only the hand that wielded it was dishonest. One must hold this well in mind, for exactly the same innocent phenomena, the omitted alif and the placement of marks, will presently return as Puin’s first two “peculiarities”. The raw material of the sensation is each time ordinary; it is the framing that makes it seem extraordinary.

The Leap to Sana’a

We now leave Mingana and his palimpsest of 1914, and we leap well over half a century forward, to an entirely different manuscript. It is of the greatest importance to keep this distinction sharp, for the two have been endlessly confused in the popular press. Mingana’s palimpsest is not the Sana’a find. They are two different manuscripts, separated by more than fifty years, by thousands of kilometres, and by two utterly different scholars.

In 1972, during restoration work on the Great Mosque of Sana’a, in the attic space above the prayer hall, enormous quantities of old parchment were discovered. It was a book grave, a gathering place of worn and damaged mushafs that the community through the centuries had not wished to destroy but had reverently stored away. Among those fragments were some of the oldest Quran parchments the world knows, including a famous palimpsest with an underlayer and an overlayer.

One of the foremost figures who became involved in the restoration and study of the Sana’a mushafs was the German Arabist Gerd-Rüdiger Puin. His work became known to the broad public through an article by the journalist Toby Lester in The Atlantic Monthly of January 1999, titled “What is the Koran?”. That article leaned heavily on Puin’s findings, and it presented the Sana’a fragments to an audience of millions as a possible bomb beneath the established history of the Quran. Puin’s words aroused, writes Azami, both sensational joy and deep anger, depending on whether one spoke with orientalists or with pious Muslims.

What then had Puin said exactly? Azami summarizes it in three “peculiarities” that Puin believed he had encountered in the Yemeni fragments.

The first peculiarity was the defective writing of the alif. In the Sana’a fragments the alif was omitted more often than in other manuscripts. The second peculiarity concerned variations in the position of the aya dividers within certain verses. And the third, called by Puin his “greatest find”, was a fragment in which the end of Surah twenty-six was followed by Surah thirty-seven, instead of the usual order.

At first sight these three points sound like something. A Quran with different spelling, with shifted dividers, with a different order of the surahs. Is that not precisely the “other Quran”? Let us examine the three points, just as with Mingana, one by one.

The Three Peculiarities Unravelled

The first peculiarity, the defective writing of the alif, is no discovery but a known phenomenon that Muslim scholarship has known for fourteen centuries. Azami notes that the defective writing of the alif has already been extensively treated, and that it is nothing new. We have seen it in earlier articles: the early rasm was a thrifty spelling in which certain long vowels were not written out, because the oral transmission held the pronunciation fixed and the script was merely the cornerstone. The Birmingham folios show the same thrifty spelling. Every early manuscript shows it. That the Sana’a fragments omit alifs places them precisely where one expects them: in the earliest phase of the mushaf tradition, wholly in accordance with what the transmission has always described.

The second peculiarity, the position of the aya dividers, is likewise not new. Azami writes that the irregularities in the placement of some aya dividers have already been noted and catalogued by early Muslim scholars. The early community knew various, all valid, countings of the verses: the Kufan counting, the Basran, the Medinan, the Meccan. These differences concerned the question of where a verse ended, not what the verse was. Not a single word of the revelation stood or fell by it. That an early fragment places its dividers in a slightly different spot is a reflection of those age-old, documented counting differences, not a discovery of a hidden variant.

Then the third peculiarity, Puin’s “greatest find”: a fragment in which the end of Surah twenty-six is followed by Surah thirty-seven. This is the point where the framing worked hardest in the press, and it is also the point where it collapses fastest. Azami notes that this find is not in the least unique, as he has demonstrated on the basis of other partial mushafs on pages seventy-three to seventy-six of his book.

Here we must understand the heart of the matter, for it touches on a principle that carries the whole history of the Quran. Muslim scholarship has always distinguished between two things: the text of the Quran, and the order in which the surahs are placed in a mushaf. The text lies fixed and is inviolable. The order of the surahs, by contrast, was for whoever wished to write a partial, personal mushaf no binding prescription. Whoever transcribed the whole Quran followed the order of 'Uthman ﵁. But whoever gathered a handful of surahs together for personal use was permitted to arrange them as best suited him.

Why could that freedom exist without endangering the text itself? Because the text did not rest on the arrangement of the parchment, but on something far more solid. The community grew, from the very beginning, with the ranks of the huffaz, the people who knew the whole Quran by heart. The script was an aid to memorization and a barrier against corrupting influences, not a replacement of the memory. Whoever set a handful of surahs on parchment for himself changed nothing of the text that he and his whole community carried in the heart; he chose only the order in which those surahs stood on his own leaves, usually beginning with the shorter surahs that one had memorized first. The order on the leaf was a matter of personal convenience; the text of each surah was a matter of shared, manifold transmission. Therefore the arrangement could be free without the text ever drifting.

Azami documents this with an impressive table. From one single library, the Salar Jung Museum in Hyderabad in India, he lists dozens of partial mushafs, with their surah order and their dating. The manuscripts bear the numbers two hundred forty-four to three hundred ten, drawn from the catalogue of Muhammad Ashraf. And the orders diverge wildly. One manuscript begins with Surah thirty-six, another with Surah seventeen, a third with Surah one followed by thirty-six and forty-eight. Scarcely two are entirely alike.

And, this is the decisive point against every sensation: these Salar Jung manuscripts are not old in the sense Puin suggested. They date largely from the tenth to the fourteenth century after the hijra. Manuscript two hundred forty-eight bears the year one thousand seventy-six after the hijra, that is sixteen hundred sixty-six of the common era. Manuscript two hundred seventy-five bears six hundred twenty-six after the hijra, that is twelve hundred twenty-eight of the common era. These are copyists who lived centuries after 'Uthman ﵁, in a time when the canonical Quran had long been available everywhere in the world in the same text and the same complete order. And yet they arranged, when they compiled a partial mushaf for private use, the surahs according to their own judgment.

Azami draws the conclusion that imposes itself inescapably: whoever wished to transcribe a partial mushaf felt free to place the surahs in whatever order seemed good to him. Puin’s “greatest find” is thus nothing other than an ordinary partial mushaf, in which a copyist arranged the surahs according to his own choice, precisely as a copyist from the seventeenth century did in Hyderabad. It does not prove that there ever existed another Quran with a different surah order as canon. It proves that the text was fixed and the arrangement in private copies free, a distinction that the tradition itself has always made explicit.

Thus the “greatest find” disappears not because one reasons it away, but because one places it in its proper context. A phenomenon that Muslim scholarship has openly described for fourteen centuries cannot be an overthrow of that same scholarship.

The Letter

And now we come to the most remarkable part of the whole story. For Puin has, despite the vagueness and the suggestion with which his findings were presented in the press, himself left a document that refutes his own sensation and strengthens the position of the tradition.

Toby Lester’s article in The Atlantic Monthly had sown great unrest and mistrust among the Muslims of the street. To allay that anger and remove that mistrust, Puin wrote a long letter in Arabic to al-Qādī al-Akwaʿ of Yemen. That letter subsequently appeared in the newspaper ad-Thawra of Sana’a, and Azami has reproduced it elsewhere in his book. In praising the Sana’a mushafs and strengthening the Muslim position, writes Azami, Puin nonetheless wrote with enough subtlety and vagueness to cast a shadow over the whole history of the Quran. But the core of the letter, when one removes the subtlety and reads the words themselves, is a concession of the highest weight. Azami translates the decisive passage:

“The remnants [of these old mushafs] go back, scientifically assured, to the first century after the Hijra! Because of the existence of these manuscripts in Sana’a … [we possess] the only monumental proof of the completion of the Quran in the first century of the Hijra! Of course, Muslims may ask what use there is in such information from a non-Muslim scholar, when Muslims are certain that the complete mushaf has existed since the third caliph, 'Uthman bin 'Affan ﵁. For them it is simply a belief cherished in good faith.”

From Puin’s letter to al-Qādī al-Akwaʿ, published in ad-Thawra, Sana’a

Let one read this slowly, for it is the pivot around which this whole article turns. The man who was paraded by the world press as the one who would expose the Quran writes with his own hand that the Sana’a remnants, scientifically assured, go back to the first century after the hijra. He calls them the only monumental proof of the completion of the Quran in that first century. Not “possibly”. Not “perhaps”. Scientifically assured, monumental proof, of the completion of the Quran in the first century of the Hijra.

Azami summarizes Puin’s own main points. First: the Sana’a manuscripts are the only monumental proof of the completion of the Quran in the first century of the hijra, a solid refutation of the many non-Muslim scholars who claim that the Quran was completed only in the early third century. Second: Muslims possess no proof that the complete mushaf has existed since 'Uthman ﵁; good faith is their only support.

The first point is a gift to the tradition. The second point is the subtlety with which Puin nonetheless tries to cast a shadow, and it is precisely that second point on which Azami corrects him.

Precision Is the Criterion

Puin makes, as Azami shows, two intertwined assertions. On the one hand he pulls back the dating claim for the completion of the Quran from the third century to the first century, an enormous concession in favour of the Muslim position. On the other hand, by refraining from any further specification than “first century”, he subtly opens a broad window of time within which he can work. “First century” is a period of a hundred years. By not becoming more precise, Puin leaves the door open for the assigning of any date whatsoever within that century, and thus for the suggestion that the Quran perhaps received its definitive form only late in that century.

Here Azami is inexorable, and rightly so. Precision, he writes, is a core element of serious scholarship, and to that we must hold. Vagueness is no virtue when the facts are sharp. And the facts are sharp. With the passing of the Prophet ﷺ at the beginning of the year eleven after the hijra, the revelation came to its natural end. The verses were then brought together into their outward form during the reign of Abu Bakr ﵁, who passed away in the year thirteen after the hijra. And the spelling was standardized and the copies were sent out under 'Uthman ﵁, between the year twenty-five and thirty after the hijra. That is the Muslim view, and it is not vague; it is dated down to the decade.

Puin, writes Azami, never claimed that the complete Quran did not come into being before 'Uthman ﵁, and if he does claim that, then he certainly does not speak on behalf of any Muslim tongue. The canonization under 'Uthman ﵁ between twenty-five and thirty after the hijra concerned not the creation of the text, but the standardizing of the spelling and the replacing of divergent private copies by one standardized rasm. The text itself was completed with the passing of the Prophet ﷺ. Whoever leaves “first century” vague in order to create room for doubt ignores the precise, documented chronology that the tradition has always transmitted, and that the material witnesses of Sana’a and Birmingham now confirm to within a few decades.

Why the “Other Quran” Dissolves Each Time

We have now examined two manuscripts, of two scholars, separated by more than half a century, and we have seen in both cases the same outcome. It is worth naming why the “other Quran” dissolves each time, for the pattern is no coincidence.

With Mingana the “other Quran” arose from a dotless, half-erased skeleton to which he himself added the dots and the alifs he needed. His variants were not readings of the manuscript; they were insertions by Mingana into the manuscript. Take away the self-added dots, and the variant disappears. There remains the ordinary Quran.

With Puin the “other Quran” arose from three phenomena that Muslim scholarship had all three known and documented for fourteen centuries: the thrifty spelling of the alif, the documented differences in the verse division, and the freedom of the surah order in partial private copies. Place these three in their proper context, and the sensation evaporates. There remains the ordinary Quran, with precisely the features that its transmission had always described.

In both cases the “other Quran” rested on a reader who lifted a phenomenon out of its context and presented it as a discovery to an audience that did not know the context. And in both cases the scholar in question, when one read his own words and his own material closely, delivered the proof of the contrary. Mingana assembled the collection that in 2015 would prove to contain a seventh-century Hijazi mushaf, in textual agreement with the canon. Puin wrote with his own hand that the Sana’a remnants are the only monumental proof of the completion of the Quran in the first century.

There is an old principle in jurisprudence that applies here: the strongest witness against an assertion is sometimes the witness who brought it forward. Mingana’s collection testifies against Mingana. Puin’s letter testifies against the framing of Puin. The “other Quran” has no other existence than in the framing itself; the moment one takes the material in hand, it dissolves between the fingers.

The Monumental Proof

Let one note the word that Puin chose. He called the Sana’a fragments the only monumental proof of the completion of the Quran in the first century. The word “monumental” is no small choice. A monument is something that keeps standing, something that testifies across time, something one cannot reason away because it physically lies there.

And that is precisely the force of the material witnesses that we have met in this article and the preceding one. The hadith transmission and the historical works about the compilation under Abu Bakr ﵁ and the canonization under 'Uthman ﵁ were always already detailed. But orientalism had set against them that the textual transmission had no material support, that the silence of archaeology was an argument for a later origin. Sana’a and Birmingham filled that supposed silence with parchment that one can date, can photograph, can touch. And the parchment did not say what the revisionists had hoped. It said what the tradition had always said.

It is a rare situation in the history of scholarship that the two manuscripts deployed as the heaviest artillery against the canon prove, on closer examination, to belong to the strongest witnesses in its favour. The palimpsest that Mingana published to show variants shows, on honest reading, that the variants were his own insertions. The Sana’a fragments that Puin made famous bear his own written declaration that they are the monumental proof of the completion of the Quran in the first century. Twice the sword was forged into a shield.

One must remain honest, at the same time, about what precisely Puin’s concession does and does not amount to. It does not make Puin a Muslim or an advocate of the tradition; he continued, even in that same letter, to write with enough vagueness to let a shadow hang over the whole. But that is exactly why his core statement weighs so heavily. It is no testimony of a fellow believer who confirms his own conviction; it is a declaration of an outsider, a non-Muslim scholar who studied the Yemeni material with his own eyes, and who arrives at the conclusion that it goes back to the first century after the hijra. A concession against one’s own interest carries in every courtroom the greatest weight, and that holds here too. When the pressure of the Muslim street forced him to be clear, Puin chose not the language of sensation but the language of the monumental proof. The vagueness was for the world press; the precision was for the qadi who called him to account.

And even where Puin remained vague, Azami does not let the vagueness stand, for the tradition need not content itself with a century when it knows the decades. The revelation ended with the passing of the Prophet ﷺ at the beginning of the year eleven; the compilation took place under Abu Bakr ﵁, who passed away in the year thirteen; the standardization and dispatch took place under 'Uthman ﵁ between twenty-five and thirty after the hijra. Where Puin left open a window of a hundred years, the tradition points to a window of at most two decades, and the material witnesses of Sana’a and Birmingham press that window from their side just as sharply together. The concession of the critic and the precision of the tradition point in the same direction; only the tradition is more precise than its opponent dared to be.

That promise has been fulfilled in this story in a remarkable way. It was fulfilled not only by the care of the believers who transmitted the text, but even by the attempts of those who wished to undermine it. Whoever published the palimpsest to weaken the canon assembled the manuscript that would confirm the canon. Whoever used the Sana’a find to sow doubt wrote the sentence that would remove the doubt. The Guardian of the Reminder has, it turns out, fulfilled His promise even by way of His opponents.

From the Refuted Variant to the Free Order

We have refuted the two great attempts to make the oldest manuscripts into witnesses against the canon. We have seen that the supposed variants of Mingana were his own insertions, and that the three “peculiarities” of Puin proved to be phenomena that Muslim scholarship had known and documented for fourteen centuries. In both cases the “other Quran” dissolved the moment one read the material closely, and the ordinary Quran remained.

One of those three peculiarities deserves yet its own treatment, for it touches on a principle that carries the whole history of the Quran. Puin’s “greatest find” was a partial mushaf in which the surahs stood in a divergent order, and we have already seen in passing how Azami disarmed that find with a table from the Salar Jung Museum in Hyderabad. But that table deserves more than a passing reference, for it shows with a force that no reasoning equals what the distinction between the fixed text and the free order means in practice. Whoever transcribed a partial mushaf was permitted to arrange the surahs as suited him, and yet the text of each surah remained inviolable, century after century, copyist after copyist.

For that is how this history works. An erased palimpsest in Cambridge, a book grave in Sana’a, a letter in a Yemeni newspaper, and presently a dusty bookcase in Hyderabad. No single place preserved it in full, no single scholar could undermine it, no single rereading could change the text that remained preserved for fourteen centuries in the hearts of the believers. The Quran carried itself through the centuries, and even whoever erased or contested it unintentionally delivered the proof of its preservation. In the next article we therefore go to the partial mushafs of Salar Jung, where the free order and the fixed text stand side by side in black and white.