Al-Hajjaj’s Regional Mushafs and the Accusation of Eleven Words

Four Decades After the Five Cities

In the previous article we followed Abu al-Aswad ad-Du’ali ﵀, the poet from Basra who placed the first vowel dots in red ink above, below, and beside the letters, so that the non-Arab reader could find the sounds that the dotless rasm still withheld. His dots did not touch the text of Uthman ﵁; they hovered like a serving second layer above the sacred black. But the script he thus made navigable went back to a still older event. In the Mushaf that traveled to the five cities, we saw how Uthman ibn Affan ﵁ in the year 25 AH had the standardized Mushaf copied into a handful of exemplars and sent by couriers to Mecca, Kufa, Basra, Damascus, and Yemen, with one master exemplar that remained behind in Medina. Five cities, five copies, and in each city a reader appointed by the Caliph who could henceforth measure the local recitation against the written reference; with that the written skeleton of the Quran, in the Quraishi dialect, had become publicly verifiable in five capitals at once.

Four decades later that umma had grown beyond recognition. The riders who had set out from Medina in 25 AH had carried in their saddlebags one manuscript per province, but by the end of the first century AH the mosques of Iraq, Sham, and Egypt counted many thousands of believers, not all of Arab descent, not all raised with the Quraishi tongue, and not all able to tell the dotless consonants of the earliest Hijazi script apart without a reading aid. What in the small Medina of Uthman ﵁ had sufficed as one written reference form for a still manageable umma called, in the expanded Caliphate of Abd al-Malik ibn Marwan and his sons, for a second, administrative operation: not to change the text, but to make it available in greater number, with better legibility, and in more cities at once.

That operation did not come out of Medina. It came out of the East of the Caliphate, from the governor who in these years exercised the actual power there, and whose name has clung down the centuries to an accusation that, had it been true, would have overturned an entire textual history. His name was Al-Hajjaj ibn Yusuf ath-Thaqafi, and his contribution to the Mushaf consisted, fully documented, of distributing copies of the existing Uthmani text to the cities. Onto that sober logistical act was later grafted the charge that in those copies he had changed eleven words of the revelation. It was not true. But its refutation, precisely because it was deployed so heavily by later skeptics, is indispensable for the theme of this article.

In this article, then, we stand before two movements at once. The one is administrative: what did Al-Hajjaj actually have produced, and how did he distribute those copies across the provinces. The other is polemical: how did the story arise that in those copies he had changed eleven words of the Uthmani text, on what tradition does that story rest, and why can we establish with great certainty that it is no more than a projection after the fact. Al-Azami ﵀ treats both movements in his sixth chapter on the history of the Quran text, and his refutation turns, as we shall see, not on pious assurances but on one hard palaeographic fact that renders the accusation impossible at a stroke.

Who Was Al-Hajjaj ibn Yusuf

The man whose name stands at the center of this article was no Companion ﵁, no scholar, and no reciter; he was, in the most literal sense of the word, an administrator in a young political order that was still forming itself around the Umayyad dynasty. He was born in Ta’if, presumably around the year 41 AH, into a family of the Banu Thaqif, and he made his ascent under Caliph Abd al-Malik ibn Marwan in the years after the end of the second fitna. In 75 AH he was appointed governor of Iraq, a post he held for two decades, until his death in 95 AH, and under the succeeding Caliph Walid ibn Abd al-Malik that position remained unchanged until the end of his life.

Al-Azami sums the man up in a few words: a governor of Iraq of considerable notoriety, whose unbending, iron-fisted rule earned him many an unflattering judgment in the chronicles of Iraq. What this article will not do is elaborate that portrait at length. His military campaigns against Abdullah ibn az-Zubayr ﵁ in Mecca in 73 AH are disputed among the umma to this very day, and his harsh style of governance has produced mixed judgments among later scholars, but that is not the matter to be treated here. The question that concerns us is purely textual: what did the governor of Iraq, in his official capacity, do with the Quran text, and what did he not do with it.

On one biographical point, however, we must remain sharp, because it will weigh in later in the refutation. Al-Hajjaj was an administrator with administrative decisiveness. In 84 AH he had a new garrison city built on the Tigris, halfway between Kufa and Basra, which received the name Wasit, the Arabic for the intermediate, because it lay precisely between the two older military capitals, and that city became his residence and his administrative nerve center. His involvement with the Quran was no private initiative; it bore the signature of a state affair, backed by Abd al-Malik ibn Marwan in Damascus and continued under his son Walid ibn Abd al-Malik.

What he lawfully held in hand was limited to Iraq and its dependent regions. He was no Caliph, he ruled no empire, he was the highest authority in his province, and when he had copies sent to the cities, those exemplars ended up in a world far larger than his own jurisdiction. It is precisely that geographical limitation that is, as we shall see further on, one of the reasons the later accusation of a textual change is structurally impossible: whatever he might have done to the Iraqi copy, outside Iraq lay the older exemplars that no governor could touch.

The Governor’s Copying Project

What Al-Hajjaj had produced was no new Quran, but a supplementary run of the existing Uthmani Quran. Al-Azami records his action here in the track of Uthman ﵁ himself: just as the Caliph in 25 AH had distributed exemplars across the provinces, so Al-Hajjaj in his turn distributed copies of the Quran to several cities. The source Al-Azami cites here, as-Samhudi following Ibn Zabala, leaves little room for doubt about the nature of the operation: Al-Hajjaj sent the Quran to the major cities, including a large one to Medina, and he was the first to have the Mushaf go out to the smaller towns as well.

Where the five Uthmani consignments of 25 AH had reached the capitals, the operation of Al-Hajjaj penetrated further, into the mosques of smaller places that in the first generation had not yet possessed their own exemplar. The why of that broadening belongs not to theological sighing but to cool demography. Al-Azami points, in a footnote to this report, to the reason the sources themselves give: the Muslim population had grown strongly in the more than half a century between the time of Uthman ﵁ and that of Al-Hajjaj, and that growth had invariably led to an increased demand for Mushafs.

That growth is not hard to imagine. The five original exemplars were, after four decades of daily use, worn out in the hands of their keepers; they were written by hand on parchment that lost its edges through constant consultation. At the same time the community in every capital had grown manifold, and where in 25 AH one mosque with one reference exemplar had sufficed, secondary mosques now arose in which an exemplar of their own became desirable. The demand for written Mushafs had, put simply, grown larger than the supply from the first run, and Al-Hajjaj had at his disposal precisely the administrative means to meet that demand on a large scale.

The pattern is a repetition, on a larger format, of what Uthman ﵁ had done in 25 AH: a central authority had copies produced and sent them under governmental authority to their destination, with this difference, that the reach was now wider and that the one commissioning it was this time no Caliph but a governor. And precisely because it was a governor and no commission of Companions ﵃, it is important to underline that textual control was not thereby left to officials. The skeleton text that was copied came from the Uthmani mother exemplars; the scribes had no freedom to deviate, for they copied letter by letter from a validated master.

We must underline one nuance here, because it is crucial for what follows. On the receiving side, in every city where a copy arrived, stood the local qurra’ who in their oral isnad chain already knew the text before the parcel reached their hands. The written copy did not land in a void; it landed in a community that carried the revelation by heart and that laid every written exemplar against that memory. A scribe who had changed one letter would have had to justify his deviation before people who carried the Book in their mouths and who would have heard the difference at once. That accountability mechanism, spread across many cities at once, is, as we shall see further on, the heart of the matter.

The Accusation of Eleven Words

The accusation has a name and a chain, and Al-Azami names both without preamble. It goes back to 'Auf bin Abi Jamila, a transmitter who lived from 60 to 146 AH, and it has come down to us via Ibn Abi Dawud, who recorded it in his Kitab al-Masahif. In that tradition Al-Hajjaj is accused of changing the Uthmani Mushaf in eleven places, each time one word that, according to the picture the accusation presents, was replaced by another word.

Before we look at the list itself, we must look at the transmitter, for Al-Azami dwells on him emphatically. 'Auf, he notes, may stand on record as reliable, but he had in addition Shiite tendencies, and that in this matter is no side issue but the core of the context. Wasit, the garrison city Al-Hajjaj had founded, was one of the strongest bulwarks of Umayyad power, and thus a natural target for anyone coming from the opposing party; every report originating from that hostile camp must, so Al-Azami holds, be approached with the greatest possible caution. To this comes a second consideration that weakens the accusation from within. Mu’awiya ﵁, the first Umayyad ruler, had fought Ali ﵁ precisely under the pretext of the blood of Uthman ﵁; an Umayyad governor who had tampered with the Uthmani text would thereby have damaged the very cause of his own dynasty. The accusation thus presupposes an act that runs against the interest of the accused himself, and that makes it, in Al-Azami’s words, particularly implausible.

We must nonetheless render the accusation honestly before we refute it. The alleged mechanism is as follows: Al-Hajjaj would, making use of his administrative control over the distribution of copies, have replaced one word with another in eleven specific verses, with the aim of a political-religious shift of emphasis that would benefit his dynasty. Al-Azami includes the full list of eleven places in a table, with the Uthmani reading in one column and the alleged Hajjajian replacement in the other, so that the reader has the matter before him in its concrete shape and does not remain stuck in vague suspicion.

So the accusation presents itself in its full form. It is, as Al-Azami demonstrates with one decisive argument, not true, and that argument is not rhetorical but graphic: it lies in the form of the earliest Arabic script itself, and it is precisely against that form that the charge unravels.

The Refutation

The First Refutation: The Dotless Skeleton

The heaviest argument Al-Azami brings against the accusation is at the same time the simplest, and it rests not on piety but on the physical form of the earliest Arabic script. Long before 'Auf bin Abi Jamila voiced his charge against Al-Hajjaj, scholars had already pored over the official exemplars of Uthman ﵁ and compared them with one another letter by letter; and the variants those early scholars established in that painstaking work do not correspond with the variants 'Auf names. That is the first crack. The second is decisive.

The Mushafs that Uthman ﵁ had produced contained no dots. In the earliest Arabic script the consonantal skeleton, the rasm, stood bare on the parchment, without the diacritical dots that were only later systematically applied to distinguish consonants sharing the same basic form from one another. And, as Al-Azami emphasizes, even in the time of Al-Hajjaj the use of those dots was by no means general. Whoever looks at the table of eleven alleged changes now with this fact in mind sees the problem at once: several of the words in that table become, once one thinks the dots away, utterly identical. The alleged original word and the alleged replacing word carry, in the dotless skeleton, precisely the same letters in precisely the same place.

On this Al-Azami poses the question that dislocates the whole accusation in a single sentence. How could Al-Hajjaj have changed these words, when the dots were absent and the skeletons were precisely the same. There is nothing in a bare skeleton to replace that would change the skeleton itself; the difference between the two readings lives solely in the dots, and those dots were not yet there. The alleged change is thus not so much implausible as graphically impossible: it presupposes an act upon a layer of the script that was not yet present in Al-Hajjaj’s exemplars. The consonantal skeleton, which was present, is in the Uthmani text and in the supposedly changed text one and the same.

To this comes a third observation that deprives the accusation of its last support. None of the eleven alleged changes touches the meaning of the verses concerned; even where the placing of dots could color a reading, the tenor of the verse remains the same. A political-religious shift of emphasis in favor of a dynasty, the motive the accusation ascribes to Al-Hajjaj, is nowhere to be found in these eleven places, for no meaning simply shifts. Al-Azami draws from this the conclusion that after all this speaks for itself: the accusation itself is baseless, without ground.

The Second Refutation: The Silence of the Community

Suppose, for the duration of one step of thought, that the charge had after all held some ground, and that Al-Hajjaj had truly tampered with the Uthmani text itself. What would then have happened. Al-Azami answers that counterfactual question with an argument that, precisely because it rests on the behavior of whole communities, is hard to evade. Had Al-Hajjaj made any change in the actual Uthmani Mushaf, then neither the Muslim community nor the established power would have kept silent about it. And the Abbasids, who displaced the Umayyad dynasty as successors and had every political interest in discrediting their predecessors, would have exploited such an act to the full.

That last weighs heavily, and it is worth unfolding. The Abbasids, once they were in power, blackened the memory of the Umayyads in countless ways; they dug up every misdeed that could cast the old regime in an evil light. An Umayyad governor who had dared to falsify the sacred text of Islam would have been a gift from heaven for their propaganda, a reproach they could have made resound from every minbar. That they never made this reproach, that the textual question is absent from their extensive anti-Umayyad literature, is not just a gap. It is the silence of precisely that party which would have shouted the loudest had there been anything to shout about.

Add to this the silence of the community itself. In every city where a copy of Al-Hajjaj arrived, there lay either an Uthmani mother exemplar from 25 AH, or a derived copy that in an unbroken chain went back to such a mother exemplar, and everywhere stood the qurra’ who knew the text by heart and recited it daily, not from the new copy but from their own memory and their already present predecessors. At the moment a new copy was laid beside an older exemplar and had shown a different reading in eleven places, the difference would have come to light within one session of reciters, and a protest, a correspondence, a corrective process would have been set in motion such as had unfolded under Uthman ﵁ as well between Hudayfah ﵁ and Medina. The historiography of the first two centuries AH would have recorded such a rupture with the same precision with which it has set down every other administrative-religious episode. None of that exists, and that silence, borne by both the hostile power-holders and the watchful community, is in historical methodology the heaviest proof that the supposed deviation simply never existed.

The Third Refutation: What the Oversight Really Meant

There is yet another report that Al-Azami brings forward, and it is illuminating because it shows the true nature of Al-Hajjaj’s involvement with the exemplars. On the authority of a report of Asim al-Jahdari ﵀ it is transmitted that Al-Hajjaj appointed him, together with Najiya ibn Rumh ﵀ and ‘Ali ibn Asma’ ﵀, to scrutinize the Mushafs, with the intent of tearing up every exemplar that deviated from the Mushaf of Uthman ﵁; the owner of such a deviating exemplar was compensated for it with sixty dirhams.

The meaning of this report stands at right angles to the accusation. Al-Hajjaj introduced no deviations into the text; he had deviations removed from circulation precisely in order to protect the Uthmani reading. His three appointees, Asim al-Jahdari ﵀, Najiya ibn Rumh ﵀ and ‘Ali ibn Asma’ ﵀, were no falsifiers but inspectors, appointed to track down and destroy exemplars that deviated from the standard, and not to disadvantage the owners but to compensate them. Whoever reads this role rightly sees an administrator who upheld the Uthmani text with administrative means, precisely the opposite of what 'Auf reproaches him with.

Precisely in that enforcement process lies, so Al-Azami suggests following Ibn Qutayba, a foothold for an alternative explanation of how the misunderstanding could arise. A few deviating exemplars escaped complete destruction because they were not torn up but corrected, by erasing the ink at the deviating spot and writing the correct reading over it with a fresh pen. Whoever later laid eyes on such a corrected leaf, without knowing the circumstances, could wrongly read into the erasure and the rewriting the hand of Al-Hajjaj as an attempt to change the Quran, whereas in reality it was a bringing back to the Uthmani standard. What the accusation presents as a change is at bottom the trace of a correction in the opposite direction.

What Al-Hajjaj Did Do

So much for the defense of what he did not do. It is illuminating to set against it what he did do, for his real contribution to the Mushaf is well documented in the sources and it lies entirely on the administrative plane. It was twofold. In the first place he had, as we saw, copies of the Uthmani text sent to the cities and he upheld that text by having deviating exemplars tracked down and destroyed; in the second place he ordered, out of a kind of administrative curiosity, a precise inventory of the text he was distributing.

That last report is too pleasant to skip, and Al-Azami records it with relish. Al-Hajjaj, so Abu Muhammad al-Himmani has it, once gathered the huffaz, those who knew the Quran by heart, and took his place in their midst, for he counted himself among that group. He charged them with counting the number of characters in the Quran, and when they were done, they arrived unanimously at the round figure of 340,750 characters. His curiosity was not yet stilled by that; he wanted to know at what point precisely the half of the Quran fell, and where each seventh part began, and the counters sought the answer down to the verse and the letter. Al-Himmani mentions that Al-Hajjaj followed the progress of the commission every evening, and that the whole undertaking took four months.

It is worth pausing for a moment at what this report shows. A governor who had wanted to falsify the text would have been the last thing one expects: he would not have openly had the text counted down to the letter by a full commission of memory-bearers, for every counting is a collation, and every collation would have brought a change to light. That Al-Hajjaj had precisely that counting carried out, evening after evening, four months long, in the company of people who knew the Book word for word by heart, is in itself already proof that he had nothing to hide. Al-Azami notes that, although we no longer know the precise counting method of Al-Hajjaj’s commission, the fact that their result lies so close to the modern count shows that those four months of meticulous counting truly took place.

What Al-Hajjaj did, then, can be summed up in a single line: he multiplied the Uthmani text, he distributed it, he upheld it against deviation, and he had it counted down to the letter. None of those four acts touches the text; all four confirm it. The historical chain of memory has, where it linked Al-Hajjaj to a textual change, inverted precisely the opposite of what the sources show.

On one point, moreover, a misunderstanding is conceivable that is worth naming, for it explains how out of an innocent development an accusation could grow. In the century in which Al-Hajjaj lived, the script was still in motion. With the first dots of Abu al-Aswad ﵀ we followed his work in Basra under Ali ibn Abi Talib ﵁, with which the first vowel dots were introduced, and further on in this series we run through to Khalil ibn Ahmad ﵀, who in the second century AH would give the diacritical system its mature form. Between those two men the consonant-distinguishing dots and the vowel marks gradually became common property, at first sparingly, later systematically. Such an added dot is no intervention in the revelation; it is a reading aid, a mark added to the existing consonant so that a reader not raised in the Hijazi script can distinguish the correct consonant from its phonetic neighbor. A Mushaf with dots and a Mushaf without dots carry the same Quran; whoever knows the text by heart can leave the dots away in thought and read the bare skeleton as if it lay before him.

Precisely in that lies, in all likelihood, the error out of which the later polemic grew. Whoever in a later century, without knowledge of the first-century palaeography, laid an exemplar from this transitional time beside an older, bare Hijazi exemplar, could mistake the added dots and marks for changes to the text itself, whereas they were nothing other than a reading aid on the outside of the skeleton. Where that misunderstanding coincided with the report about the erased and rewritten correction exemplars, and with a transmitter who came from a corner hostile to the Umayyads, the ground was laid on which the charge of eleven changed words could shoot up. It remained, even then, a misunderstanding; the skeleton was touched in none of the eleven places.

What This Means for the Textual History

What at the surface looks like a polemical episode is on closer view a test case for the robustness of the Quranic transmission as a whole. Three observations together carry that conclusion.

The first observation is that the umma controlled its own text in a way that no ruler, however powerful, could evade. Al-Hajjaj was in his province one of the most feared governors of the first century AH; he had at his disposal political power, military means, administrative infrastructure, and scribal workshops in his own residence. If any one man in that century was in a position to force through a textual change, it was he. That he did not do it, not because he could not but because the communal control would have reported the change at once, illustrates where the textual certainty really rests. It lies not with a political authority; it lies with a collective memory that carries the text across many cities at once, and that would have reacted to every deviation within one generation. The fact that the most feared governor of his time could not touch the text without being discovered says more about the strength of the transmission than any theological assurance could.

The second observation is that the distribution project itself, ironically enough, did not weaken the textual unity but strengthened it. By distributing the Uthmani skeleton text in many copies not only across the old capitals but also across the smaller towns, Al-Hajjaj widened the circle of places where the text was mutually verifiable. Where in 25 AH a handful of exemplars had watched over a handful of capitals, there now lay a denser net of parallel copies, and every new exemplar could be measured against more predecessors than had been possible in the time of Uthman ﵁. That did not lessen the chance of an unnoticed intervention but really made it impossible: the more parallel witnesses, the less room for a silent deviation.

The third observation concerns the palaeographic trace that the centuries have left us. Al-Azami discusses in this same chapter one of the oldest preserved exemplars, the Mushaf in the Topkapi Sarayi in Istanbul, number 44, of which some notations suggest that it was still written in the first century AH; he weighs the dating debate carefully and considers the doubt about that early dating unfounded. Further on in this series the Sana’a manuscripts also come up, of which the oldest layer is likewise placed in the first century AH, and the Topkapi codex, which we then enter into more extensively. What these early witnesses have in common with one another and with the present Mushaf is their consonantal skeleton: apart from a few orthographic particularities, they carry the same rasm that stands today in every printed Quran. And precisely that skeleton is the layer in which the eleven alleged changes would have to live; that it has remained unchanged across the centuries and across the continents is the tangible counterpart of the argument that in the dotless script those changes could never have existed.

One Quranic verse sums up the principle under which all this stands. It is not the verse from Surah al-Hijr 15:9, which in earlier articles has already pronounced the guarantee of preservation; here the anchor from Surah Fussilat fits better, in which Allah ﷻ confirms its untouchability in time itself.

From before it nor from behind it. The phrase has in exegesis always been given both a chronological and a spatial reading: no falsehood can enter it in its original revelation, and no falsehood can slip into it in its later transmission. The Al-Hajjaj episode, in the way it falls apart in this article, is a historical illustration of the second part of that pronouncement. A political authority at the height of its power, with administrative means at its disposal and with a scribal workshop within reach, was not able to change the text in eleven words, and even the charge that it had done so could not affect the factual continuity of the text. It remained, from before nor from behind, untouched.

Looking Ahead to Khalil ibn Ahmad

When Al-Hajjaj died in 95 AH, his contribution to the Mushaf had spread the Uthmani text in new copies across the major and the smaller cities of the caliphate and had it inventoried down to the letter. What he had upheld was the written skeleton; what still called for refinement in the centuries after him was the reading aid that had to grow within that skeleton. The red vowel dots of Abu al-Aswad ﵀ and the first black consonant dots that gradually became common property in Al-Hajjaj’s own century already served the reader, but they did not yet form a coherent and elegant whole. The written certainty stood; the diacritical apparatus still awaited its mature form.

That mature form awaited a scholar from Basra, in the second century AH. His name was al-Khalil ibn Ahmad al-Farahidi ﵀, and his work was the perfection of the whole system of reading marks, so that the vowels and the consonant-distinguishing dots could no longer cross one another and the rasm of Uthman ﵁ remained untouched beneath the new layer. What has been told in this article about the administrative distribution and enforcement of the written text receives, with Khalil ibn Ahmad ﵀, further on in this series, its technical completion: the skeleton certainty that the copies of Al-Hajjaj and the Uthmani mother exemplars together carried was provided by al-Khalil ﵀ with a reading system that to this day is visible in every Mushaf in the world.

What lies between the distribution under Al-Hajjaj and the perfection of the diacritical system is no rupture but a continuous line. The umma has, with the means of its own memory and its own isnad discipline, preserved one Book against every attempt at intervention, and the charge of eleven changed words is, viewed in that long line, no more than a footnote whose refutation lies enclosed in the sources themselves. The facts stand in the dotless skeleton that could carry no change, in the early manuscripts of Sana’a and Istanbul, in the qira’at chains of Basra and Kufa, in the silence of the Abbasids who had no misdeed to point to, and in the mouth of every Muslim who today, in whichever ahruf modality, recites the same revelation that in 25 AH was set down in five cities at once and that under Al-Hajjaj was distributed across many cities more.

No falsehood can approach it from before it nor from behind it. At the eleven supposed places, and at every other place in the written or oral transmission, the sources carry this same testimony. What in the first century AH under a feared governor proved graphically impossible and was noticed by no contemporary, has in every century since likewise remained impossible.