The Sana’a Manuscripts

The preceding article set the fixed text and the free ordering side by side. We saw in the partial mushafs of the Salar Jung Museum in Hyderabad how copyists across many centuries arranged the surahs as they saw fit, while the text of each surah remained inviolable, identical down to the letter from one copy to the next. That evidence came from a nineteenth-century library catalogue, at the late end of the chain. But the fixity of the text lets itself be read not only at the end of the chain but at its beginning as well, and it is precisely there that the witnesses lie which exist not in an oral recitation or a late copy, but in ink, on parchment that can be dated, photographed, and set against the present Mushaf. Islamic tradition tells of one Book, of an Uthmanic standardization, of Companion codices that continued to exist alongside that standard, and of a textual unity that would remain unblemished across fourteen centuries. Against that tradition stands, for more than a century now, an orientalist counter-narrative: that the Quran has a far murkier history than Muslims admit, that somewhere beneath the surface a lost or suppressed textual reality lies buried. The question was long theoretical. In 1972 it suddenly became tangible.

A roof caved in, and out of the collapse emerged a mountain of parchment that would rewrite the entire debate.

The history of the Quran knows many turning points. The first was the revelation itself, in the cave near Hira. The second was the oral and written recording in Medina, under the direction of Zaid ibn Thabit ﵁ and the sixty scribes the Prophet ﷺ had gathered before him. The third was the standardization under Uthman ﵁, in the year the young umma showed the first signs of regional differences in recitation. The fourth was the invention of the diacritical dotting system by Abu al-Aswad ad-Du’ali, and later its elaboration by al-Khalil al-Farahidi. And the fifth, for the scholarship of the twentieth century, was the Sana’a find, because it was the first time that a large, coherent collection of first-century-AH parchments became public for systematic study.

For those who approach the Quran as believers, no archaeological confirmation is needed. The Message is engraved in the breasts before it is poured into ink. But for the debate that has been waged around the textual integrity of the Quran since the nineteenth century, Sana’a delivers something unique: a physical measuring rod. A touchstone. A document whose age is not in dispute, whose provenance is documented, and whose contents, after forty years of careful palaeographic labour, can at last be spoken aloud.

A Roof That Caved In

The Great Mosque of Sana’a is one of the oldest buildings of Islam. According to tradition it was founded in the years when the Prophet ﷺ still lived, on his very instruction, so Yemeni sources say. It stands in the old inner city, surrounded by minarets and plasterwork that has been touched up for centuries, and its outer facades betray little of its age. Whoever steps inside sees above all the pattern of columns and arches that carries the prayer hall.

Above the prayer hall lies a ceiling. Wooden beams, plaster, roof. In the winter of 1972 an unusual amount of rain fell on Sana’a. The stucco of the loft deck became saturated, the load-bearing structure weakened, and one day a part of the ceiling tore open. Construction workers were called in to repair the damage.

What they found above the plasterwork was not emptiness. It was a forgotten space. Between the loft vault and the roofing lay bundles of dust, leather, and parchment, bound with ropes, some in old travelling trunks, others loosely stacked, all covered beneath a thick layer of pigeon droppings and centuries-old dust. At first the material was swept aside as rubbish. Only when someone picked up a fragment and recognized the Arabic letters did it dawn that something remarkable lay here.

What had slumbered above the prayer hall was a repository. A so-called qubur al-kutub, in Arabic literally a book grave, the place where worn-out Qurans and Quran fragments were reverently stored away when they could no longer be used. No throwaway pit. A deliberate place of safekeeping. Bundles that generations of mosque-goers had never seen, because no one knew that they were there.

Someone called the right man. The Yemeni state at that moment had its own antiquities authority, and at its head stood Qadi Isma’il al-Akwa’, a scholar with a deep love for his national heritage. He had everything that had been found above the ceiling brought down and gathered together. The scale of what he saw before him was without precedent in modern times.

For Muslims such a find is not only a scholarly matter but also an ethical one. A worn-out Quran or a torn page may not be thrown away. It is buried, or locked away in a sealed space, or reverently sunk in flowing water. The Great Mosque of Sana’a chose the second option. Generations of mosque custodians stored parchments that could no longer be in daily use in this upper chamber, with the same care with which a museum closes a collection. And then, at a certain point, the upper chamber fell into oblivion. No one knew exactly what lay in it any longer. Generations of imams came and went, and the book grave remained.

Only a collapsed ceiling brought it back to light.

The Loft of Forgotten Books

When the bundles were sorted, they counted roughly twelve thousand loose fragments of parchment. Some leaves were still recognizable as pages. Most were crumpled, torn, or glued together by damp into blocks that had to be carefully separated layer by layer. Beside this mass of parchment lay a much larger stack of paper fragments, younger and more numerous, but less crucial for the palaeographic question.

The parchment was the heart of the find. Two scripts dominated. Part of it was written in an upright, angular Kufic script, the standardized calligraphic form that prevailed in the second and third centuries AH. Another part was written in Hijazi, the slanting, rightward-leaning script of the earliest Islamic generations. The Hijazi style is precisely that written form of which al-Azami ﵀ notes that orientalists such as Jeffery sometimes wrongly took it for Kufi: a confusion that is not innocent, for the Hijazi script is older, and its presence signals a first-century-AH document.

Among the thousands of fragments were pages that possessed a still greater rarity. They were palimpsests. A palimpsest is a parchment whose original text has been scraped off, and on which a new text was then written. In an age when parchment was costly and hard to come by, reuse was a normal act. But for the historian a palimpsest is a gift. Under certain lighting, and with modern imaging techniques, the scraped-off lines can be brought out again.

Beneath the Uthmanic Quranic script that was visible on these palimpsests lay an older text that had once been erased. What was that older text? Was it Quran? Was it a divergent Quran? Was it something else?

That question would hang over the Sana’a find for the next forty years.

The loft yielded still more material than palimpsests alone. Some fragments bore multicoloured diacritical marks, red, green, yellow, and light blue, each a different vowel signal. Others contained only the bare rasm, the skeleton of the letters, without any dotting. Still others showed the transition from an earlier system to a later one: a Quran in Hijazi script where later generations had added diacritical marks with a thinner brush, as if a reader from the second century AH had carefully wanted to prolong the legibility of an older copy.

Here it became clear what al-Azami describes at length in his chapter on Arabic palaeography. The earliest Qurans possessed few or no diacritical marks. Not because the scribes were careless, but because the system of Abu al-Aswad ad-Du’ali was only applied to a Quran in the years around 50 AH under Mu’awiya ﵁, and only became truly widespread in the third century AH. A Quran without diacritical marks seems unreadable to the modern eye. For the first generations it was not, because they knew the text by heart. The written Quran was a support, not an independent source. The true Quran sat in the breast of the reciter.

That palaeographic logic is fundamental for understanding Sana’a. For without that understanding, the illusion arises that variants in early manuscripts automatically mean textual corruption. With that understanding, the opposite turns out to be true: variants in early manuscripts are precisely what one should encounter in a textual culture in which the oral transmission is primary and the written transmission secondary, a support for the memory.

The German Conservators

Qadi Isma’il al-Akwa’ understood that his own authority, however devoted, did not have the means to conserve this material. Parchment that had lain for centuries in a semi-dry loft, with a thin layer of droppings on top of it, had paradoxically been preserved reasonably well, but the exposure to damp in 1972 had made it vulnerable. Moreover the material was in pieces: thousands of fragments that belonged together but had been torn apart, and that could only be reassembled into pages by patient comparison.

In 1980 a request was directed to the German Federal Republic, and the German Ministry of Culture sent two specialists to Sana’a. The first was Gerd-Rüdiger Puin, an Arabist attached to the Universität des Saarlandes. The second was Hans-Caspar Graf von Bothmer, an art historian with expertise in early Islamic manuscripts. They arrived in a country where the working conditions were challenging, and were given access to a collection where no Western researcher had set foot before.

Their task was not to judge, but to preserve. They photographed. They unfolded bundles that had been folded up for generations. They separated leaves that had stuck together from damp. They made a first catalogue. By the end of their work in the nineteen-eighties they had made an estimated thirty-five thousand microfilm exposures of fragments that would otherwise never have been accessible.

Part of what they found fitted perfectly with the known story. A large part of the parchment fragments bore the ordinary, standardized Uthmanic text, written in Hijazi or early Kufi. Some fragments were exceptionally beautiful, with multiple ink colours for the diacritical marks, with graceful verse and surah markings. Al-Azami shows images from this collection in his own work. He places figure 10.6 as an example of a Quran in Kufic script with the dotting system of Abu al-Aswad ad-Du’ali, presumably second-century AH, and states emphatically: “Figures 10.6 and 10.7 (above) probably date from the second century A.H., while the next is an example of the Qur’anic script from the third century A.H.”

Another part of the find behaved less ordinarily. Within it was the fragment that, under the catalogue designation DAM 01-27.1, would go down in history.

Puin expressed himself cautiously in his first publications. He noted that the Sana’a collection offered a unique opportunity to follow the palaeographic evolution of the Arabic script from its earliest forms to the stable calligraphic conventions of the second and third centuries AH. He pointed to a few striking phenomena, such as surah orderings that on a few loose leaves did not match the standard order, or word forms that did not fit the Uthmanic rasm. For a palaeographer these are details that make the whole puzzle richer. For a journalist in search of a sensational story, they are the crowbar with which the whole construction can be prised open.

The microfilm catalogue that Puin and von Bothmer compiled was not intended as an interpretive framework. It was intended as documentation. But the interpretation came without his having asked for it, and it came askew.

The Palimpsest

DAM 01-27.1 is a group of leaves, a few dozen folios, all palimpsest. The upper text, that is, the text that is now visible and was written on the parchment after the first text had been erased, is a standard Uthmanic Quran in early Hijazi script. There is no secret to that upper layer. Whoever looks over it reads the same Quran that is recited today in every mosque in the world.

What makes the manuscript special is what lies beneath it.

Beneath the Uthmanic upper layer the shadow of an earlier text is visible. A scribe had scraped off the older parchment, but the ink had penetrated deep enough into the deer leather to leave traces. With ultraviolet lighting and multispectral imaging that lower text can be reconstructed. Not fully. Not always unambiguously. But enough to build up a legible structure.

And the lower text of DAM 01-27.1 is also Quran.

But it is not exactly the same Quran as the upper layer.

Here arose the moment when the Sana’a palimpsest became world news, and when unfortunate journalism pushed the story in a direction that did not fit the facts. For to whoever views the evidence without knowledge of the history of Islamic transmission, the tempting impression arises: a divergent Quran, beneath the official Quran, hidden in a corner of Sana’a and forgotten for centuries. It looks like a hidden truth that has now finally been recovered.

That was not how it was. But that was how it was told.

The Analysis of Sadeghi and Goudarzi

It would take until 2012 before a thorough, peer-reviewed analysis of the lower text became available. Behnam Sadeghi, professor of religious studies at Stanford, and Mohsen Goudarzi published in the journal Der Islam an extensive study of DAM 01-27.1. Their work stands in sharp contrast to the sensational reporting that had shrouded the find for twenty years, for they based themselves on the factual transcription of the lower layer, verse by verse, compared with the Uthmanic text and with the variants that the classical Islamic sources attribute to the Companion codices.

Their conclusions were crucial for the whole debate.

The lower text is not a different Quran. It is an early non-Uthmanic version of the same Quran. The verses follow the same content, the same meaning, the same theological charge. The surah order deviates in a few places, but no single surah is missing, no single surah contains verses that do not occur in the Uthmanic version, and no single passage contradicts the present text in substance.

The differences that do occur are small and specific. A synonym in place of another synonym. A personal pronoun that changes position. A conjunction that has been added or omitted. A verb form conjugated in a subtly different way. Precisely the kind of variation that Islamic tradition has always attributed to the seven ahruf, and that is reported in the classical sources concerning the codices of Ibn Mas’ud ﵁, Ubayy ibn Ka’b ﵁, and other Companions ﵃ who had assembled their own private copies before the Uthmanic standardization.

Sadeghi and Goudarzi compared the lower text line by line with the variants attributed to Ibn Mas’ud ﵁ and to Ubayy ﵁ as they are transmitted in the works of Ibn Abi Dawud, as-Suyuti, and Ibn al-Nadim. The correspondence was striking. The Sana’a palimpsest did not seem to be an exact copy of Ibn Mas’ud’s ﵁ codex, but its pattern of variants fell within the same family. In other words: the lower text belongs to the type of Companion codex about which the Islamic sources have been telling openly and honestly for fourteen centuries.

Sadeghi and Goudarzi went further than a comparison alone. They brought the deviations into two categories. The first was orthographic: variations in spelling that have no influence on the pronunciation, such as the presence or absence of an alif in certain words. That kind of variation is precisely what al-Azami documents at length for the Uthmanic Quran itself, with the well-known examples such as the word as-samawat that is written in one hundred and ninety places without alif, save for one exception in verse 41:12. It is a convention, not a corruption.

The second category was textual: words that factually deviate from the Uthmanic text. And it is here that the analysis became truly interesting. For the deviations followed, in the overwhelming majority, a pattern that the classical Islamic sources had already predicted. When Sadeghi and Goudarzi placed the Sana’a lower text alongside the reconstructions of Ibn Mas’ud’s ﵁ codex that Arthur Jeffery had assembled in 1937, it turned out that the Sana’a text sometimes agreed with Ibn Mas’ud ﵁ against Uthman ﵁, sometimes agreed with Uthman ﵁ against Ibn Mas’ud ﵁, and sometimes went its own way that followed neither. In other words: the Sana’a codex is not a copy of a known Companion mushaf. It is a separate Companion mushaf, of a name we no longer know, but with a pattern of variants that falls entirely within the expected spectrum.

On top of that came a second measurement. The parchment itself was subjected to radiocarbon dating, and that dating is not a literary matter but a laboratory result: the measured age concerns the animal hide on which the text is written, not the ink of the text itself, and it was carried out by the radiocarbon laboratories that Sadeghi and Goudarzi report in their study. The outcome leaves little room for doubt about the century. The tested standing leaf of the Sana’a codex yielded, according to Sadeghi and Goudarzi, a published interval that places the parchment with ninety-five percent probability between roughly 578 and 669 CE, with a narrower band of roughly 614 to 656 CE, and the other tested leaves gave overlapping seventh-century outcomes. That means that the hide on which this text was written stems from the time of the Prophet ﷺ himself, or from the years immediately after. It is materially a first-century-AH document. No other surviving Quran manuscript approaches the moment of revelation so closely.

And what stands on that parchment from the prophetic generation? The Quran as it is still recited now, with the predictable small deviations of a Companion codex.

It is worth pausing on this for a moment. Textual scholars who occupy themselves with the Bible have at their disposal a handful of fragments from the first and second centuries CE, small in scope, and some more integral manuscripts only from the fourth century CE. Between Jesus ﵇ and the Codex Sinaiticus lie three centuries. Between the revelation of the Quran and the Sana’a palimpsest lie, in the most pessimistic scenario, two decades. In the optimistic scenario the Sana’a dating even overlaps with the life of the Prophet ﷺ himself. No other world religion can lay anything comparable on the table.

Al-Azami places here in his own work a sober observation: a Book that has withstood fourteen centuries of material tampering, while other writings have undergone considerable adjustments, testifies to a unique preservation. “A Book that has resisted any universal alterations for fourteen centuries is living proof that the text within belongs to Allah, Who has appointed Himself as Guardian.” Sana’a is, in a material sense, a partial confirmation of that statement.

What It Proves

The Sana’a palimpsest is, contrary to what half a century of sensational journalism has suggested, no counter-evidence against the Islamic tradition. It is a material confirmation.

Three things are made visible, black on parchment.

The first is that Companion codices really existed, that they really contained small variations relative to the Uthmanic text, and that those variations had precisely the nature that the classical Islamic sources have described for centuries. No clandestine, theologically divergent Quran. No secret Quran of Ali ﵁ with unknown verses. No revelations that overturn the orthodoxy. But lexical and grammatical shadows, all within the seven-ahruf framework that the Prophet ﷺ himself had permitted, as al-Azami documents at length from Sahih Muslim and the Musnad of Ahmad.

The second is that the Uthmanic standardization took place exactly as the tradition tells. For what does the upper layer of DAM 01-27.1 do? It overwrites, physically and literally, an earlier Companion codex with the Uthmanic text. Someone in the first century AH had an older, non-Uthmanic manuscript in hand, carefully scraped it clean, and used the parchment again for the official text. That is no anomaly. That is the Uthmanic decree in action: older divergent copies were either burned, or, as here, rewritten with the uniform version. The loft of Sana’a is a time capsule of the standardization process.

The third is that the text the upper layer carries corresponds exactly to the text we hold in hand today. Between the Uthmanic transcript from the first century AH and the printed Quran from the King Fahd Complex in Medina lie fourteen centuries of manual transmission. It is precisely in that kind of transmission that textual drift could strike, that letters could shift, that whole words could evaporate. The oldest material document we possess establishes: there is no drift. The text was already fixed then. The huffaz, the thousands of Muslims who in every generation have learned the Quran by heart from beginning to end, have held on to it.

Al-Azami writes about that exceptional methodology. The Quran was preserved not only on parchment, but in the breasts of hundreds of thousands of believers, in a chain of qira’at that was taught by recognized teachers, with an isnad that goes back to the Prophet ﷺ. No reciter could casually propose a deviation, for his error would immediately be heard and corrected by his fellow pupils and his master. Zaid ibn Thabit ﵁ said it explicitly: “The qira’at is a sunna that must be strictly followed.” The Sana’a palimpsest is a physical footprint of that system, congealed in ink and scraped parchment.

To this belongs a fourth observation, which in the discussion is often lost. The Sana’a palimpsest is also a proof of honesty in the Islamic tradition. For if the Muslim community had wanted to hide something, it could have destroyed precisely this kind of material in the first centuries. Uthman ﵁ had commanded that divergent copies be burned, not reused. The fact that an early Companion codex nonetheless still, by the detour of a palimpsest, has come down to us, means that the uniformization was not absolute. Someone in the first century AH knew that an older codex lay there, scraped it off for reuse, and preserved the resulting manuscript instead of destroying it. The loft of the Great Mosque was for centuries a silent witness to a process that is described in the sources and that in the physical world has now also been made visible.

The Orientalist Misunderstandings

It would be unfair to write this chapter without honouring the other side of the story. For between the moment of the find in 1972 and the careful analysis of Sadeghi and Goudarzi in 2012 lay four decades in which the Sana’a palimpsest played a role that its actual contents do not justify.

Gerd-Rüdiger Puin, one of the German conservators, had worked above all in the field of early orthography and palaeography. He was no Islamic theologian, and he was no textual critic of the qira’at system. What he did see, and rightly saw, was that the Sana’a find made tons of unprecedentedly early material available. In conversations with colleagues and in loose remarks he expressed himself enthusiastically about the possibilities, and about a few matters that personally astonished him in the earliest fragments: small deviations in spelling, unusual surah orderings in some loose leaves, palaeographic variants that did not fit the neatly organized impression that later Qurans give.

In January 1999 the American journalist Toby Lester published in The Atlantic Monthly an article with the title “What is the Koran?” The piece quoted Puin at length, sometimes in fragments without the academic reservations that he probably would have made in specialist literature. Lester portrayed the Sana’a find as a potential upheaval: might the Quran perhaps, like the Bible, turn out to have a history of editing, revision, and artificial unification? Could the official Islamic version be dismantled the way scholarship had dismantled the Old and New Testament?

The article was written with a journalistic flair that deserves respect, but its premise was askew. For the orientalist methodology that Lester invoked was precisely that methodology which al-Azami, in his own work, especially in chapter 11, dismantles word by word. The assumption is that variants always mean corruption. But in the Islamic methodological framework a variant within the seven ahruf is no corruption. It is a consciously authorized polyphony, established by the Prophet ﷺ himself, and carefully passed down per isnad. The difference between malik and malik in Surah al-Fatiha, which al-Azami discusses at length, is no tear in the text. It is a duality of meaning taught by the Messenger ﷺ: Owner and King of the Day of Judgement.

Lester’s article warned readers that its implications for the Islamic world would be “explosive”. What in reality exploded was the distance between what the palimpsest actually says, and what the reporting suggested that it said.

Sadeghi and Goudarzi went, thirteen years later, patiently back to the source. They counted the variants. They classified them. They compared them with the codices of Ibn Mas’ud ﵁ and Ubayy ﵁ as transmitted in the classical Islamic sources. And they established what any unprejudiced researcher could have established: the Sana’a palimpsest confirms the Islamic story of transmission, it does not refute it.

Al-Azami himself, writing about manuscripts from Sana’a, uses the Sana’a fragments as an illustration of early orthographic and palaeographic conventions. He sees in them a continuum with the Uthmanic rasm, no rupture. His analysis of the variants of Jeffery and Goldziher in chapter 11 is a methodical demonstration that the orientalist theory of widespread corruption cannot hold up. “The variations they predict are nowhere to be found in countless instances where a skeleton can contextually admit more than one set of dots and markings; the rare cases of authoritative divergence in qira’at by their very nature harbour no impact on the meaning of the text.”

That is a sober sentence. It is no polemic. It is a measurement.

For the justice of the story something more belongs to it. Puin himself has never fully distanced himself from his early remarks, and he has become more cautious in later interviews without retracting them all. Some of his palaeographic observations, particularly about the variation in surah orderings on a few loose leaves, remain legitimate research questions. The academic world is no monolith. Not every orientalist shares Jeffery’s premises, and some modern researchers, such as Sadeghi himself, work from a secular methodology but arrive at conclusions that largely confirm the Islamic tradition. The line between scholarly debate and sensational journalism is not the line between Western and Islamic. It is the line between careful and careless.

Al-Azami’s merit in this whole dossier is that he has demanded the carefulness. He has tested the Jeffery thesis piece by piece, checked the examples verse by verse, and shown where the orientalist reasoning rests on assumptions that it does not present as assumptions. Sadeghi and Goudarzi have, decades later, from a different methodological framework, applied a comparable carefulness to the Sana’a find itself. Both approaches have confirmed the tradition more than refuted it.

The Silent Measurement of a Palimpsest

For the believer there is yet another dimension to the Sana’a find that the scholarly discussion cannot always capture. The palimpsest is, in a certain sense, a love letter between generations.

Picture the scene. Somewhere in Sana’a, somewhere in the first century AH, a scribe sits with a scraped-off sheet of parchment before him. He knows what stood on it, for he has just scraped it off himself. It had been a beloved codex, perhaps of a teacher, perhaps of a father, perhaps of an unknown Muslim from Kufa or Basra who had taken it along on his journey to the south. The codex contained the Quran, but in a Companion version that had been replaced by the Uthmanic standardization. The scribe decides not to throw the parchment away. He decides not to burn the old codex. He decides to reuse it, and to write the standard over it, so that the community has the official version, and so that the old codex no longer circulates beside that official version.

What he does not know is that the ink of his predecessor has penetrated too deep into the parchment to disappear completely. He does not know that fourteen centuries later two researchers with ultraviolet lighting will be able to reverse his work. He does not know that his scraping opens a window onto the time before his own writing.

He does what his community asks of him. He follows the Uthmanic rasm. He respects the older codex enough not to throw the parchment away, but respects the standard enough to no longer read on with it. He is the intermediate link between two generations. And his work survives.

When Sadeghi and Goudarzi publish their analysis in 2012, they in a certain sense complete a conversation that had lain open for fourteen centuries. We now know who their anonymous predecessor was, in a general sense: a conscientious Muslim of the first century AH, a scribe in Sana’a, who in his own action expressed the whole textual system of Islam. The Quran must be uniform, and therefore I reuse this parchment. History must be honest, and therefore I do not destroy it completely. The Book is Allah’s ﷻ, and therefore my work may serve, but never dominate.

That is the silent measurement that was found in the loft of Sana’a. No sensation. No revelation. No overturning. A reverent hand that treats a parchment the way a community treats its Book: with care, with adjustment, and with depth.

One Book, Many Streams of Ink

The Sana’a palimpsest is a first-century-AH document. It confirms three things that Islam has always told about itself. It confirms that Companion codices existed. It confirms that the Uthmanic standardization smoothed out precisely the deviations that the tradition has always mentioned. And it confirms that the text that was established in the years after the Prophet ﷺ is identical to the text that is recited today by millions of people, on every continent, in every language of origin.

What the palimpsest does not do is reveal alternative Qurans. No hidden revelations. No suppressed verses. No Book-behind-the-Book. The loft of Sana’a was no secret. It was a reverent book grave, where worn-out manuscripts had been stored by generations of Muslims who knew which words lay here, and who recited the same Quran that their ancestors had recited.

The depth of this evidence is hard to overstate. We now have, materially in hand, parchments that are older than the oldest integral manuscripts of the New Testament, written within one generation of the revelation itself, and carrying the text that we still recite today. What Sana’a shows is not a different Quran but the same Quran, early recorded, and that is precisely the point: the upper layer proves the early fixation of the standard text, and the lower layer proves that the variants that lie beneath it do not fall outside the tradition but stand right in the middle of it, as a Companion codex within the seven-ahruf framework. There is no history of copying that let the text drift, there is no editorial evolution that reshaped it, there is no lost original that we are missing. One Book, with many streams of ink, all flowing back into the same riverbed.

And yet Sana’a is not the only document that has been interrogated in the past half century. Other very early fragments have been preserved, some in libraries where they lay unnoticed for decades among older collections. Sana’a yielded a large, coherent find, but its dating rested on palaeography and on the comparison of the lower text, not on one sharp laboratory measurement of the parchment itself. It was precisely that sharp measurement that would come from another quarter. In a German university library lay a substantial Kufic Quran fragment that had always been suspected of being early, and that only under radiocarbon dating would surrender its age to within a few decades, this time with a number from a natural-science laboratory.

In the next article we travel to Tübingen, where the carbon-14 speaks.