Khalil ibn Ahmad and the Modern Diacritical System

In the previous article we saw how the governor Al-Hajjaj ibn Yusuf had copies of the Uthmanic text distributed across the great and lesser cities of the caliphate and upheld that text against every deviation, while the later accusation that he had altered eleven words foundered on the dotless skeleton that could bear no such intervention. His work did not touch the rasm; it multiplied and protected it. But in that same century the script itself was still in motion, for a generation earlier Abu al-Aswad ad-Du’ali ﵀ had placed the first vowel dots in red ink above, below and beside the letters in Basra. That was a breakthrough, not a completion. The Mushaf of Uthman ﵁ remained untouched, but atop the black of the rasm there now hovered small red dots that fixed the pronunciation for readers who had not taken in Quraishi Arabic with their mother’s milk. A second problem, however, remained unsolved, for several of the most frequent letters in the Arabic alphabet shared one and the same skeleton in the old Hijazi and the early Kufic writing style, so that whoever did not know the text by heart could not possibly tell them apart in an undotted line. Between roughly 100 and 170 AH that problem was solved in Basra, and in those same decades the Arabic script acquired the marks it carries to this day.

What Abu al-Aswad ﵀ Could Not Solve

The system of Abu al-Aswad ﵀ was ingenious in its modesty. One red dot above the final letter of a word meant fatha, one below the letter meant kasra, one beside the letter meant damma, and two dots signalled the tanween. The scholar had not touched the Quran; he had only made it passable for the non-Arab believers who were entering Islam in ever greater numbers, in Persia, in Egypt, in Khorasan. The red dots were a learning aid, not a change to the text.

But the learning aid covered only half of the difficulty. The Mushaf of Uthman ﵁ knew no consonantal dots. It knew no vowel marks. It knew not even aya separations nor surah titles. The rasm, that carefully chosen consonantal skeleton laid down in the year 25 AH, carried the Book by its form alone, borne by an unbroken chain of oral transmission. For whoever lacked that chain, the letter ب in the Arabic script was visually indistinguishable from ت, ث, ن or ي when it stood in the middle of a word. Five wholly different sounds attached to precisely the same skeleton.

It was no theoretical problem. The Islamic state now stretched from Azerbaijan and Armenia in the north to Yemen in the south, and from Egypt in the west to Iran in the east. In that expanding world Mushafs were copied by scribes for whom Arabic was no mother tongue, men who had learned the language in fits and starts. A copyist who was not certain which of the sounds belonged in a given position could do little but transcribe what he saw and hope that the pupil who followed him would consult the oral transmission. For the direct transmission of the Quran this was no acute danger, for recitation and instruction had always been primarily oral and literacy remained embedded in the isnad. For the broadening civilization that raised the Quran to an object of study and a calligraphic ideal, it was a problem indeed.

The evidence that the Arabs knew consonantal dots is, moreover, far older than Islam itself. The Raqush gravestone from Mada’in Salih, dated to 267 AD, already bears skeleton dots on the letters dhal, ra and shin. The presumably pre-Islamic inscription of Sakaka, in a curious script that combines Nabataean and Arabic letters, bears dots on the letters n, b and t. A bilingual papyrus document from Egypt, dated to the year 22 AH, thus during the reign of Umar ibn al-Khattab ﵁, bears dots on the letters n, kh, dh, sh and z. An inscription near Mecca from 46 AH contains a dot on the letter b. The dam of Mu’awiya near Madina bears dots on the t, and the second dam of Mu’awiya near Ta’if, dated 58 AH, contains consonant dots on ya, b, n, th, kh, f and t. By the end of the first century, ten letters in scattered, pre-Quranic inscriptions had already been furnished with distinguishing dots. The technique existed, sometimes sparingly, sometimes dispersed, but it was never institutionalized. What was lacking was a man who would carry the system through to the end, in such a way that it did not intermingle with the red vowels of Abu al-Aswad ﵀ nor with the sacred black rasm of Uthman ﵁ himself.

There exists a remarkable tradition that shows the breadth of the practice before the official standardization. Muhammad ibn Ubaid ibn Aus al-Gassani, the secretary of Mu’awiya ﵁, relates that his master once asked him to apply some tarqish to a document. When Gassani did not understand what that word meant, Mu’awiya ﵁ explained: give each letter its own dots. He added that he had once done the same for a document he had written on behalf of the Prophet ﷺ. Whether that last detail is historically watertight remains a question for the Hadith scholars, but it shows how deeply the idea of dots to distinguish letters was already slumbering in the Hijazi culture long before it was institutionalized in Basra. What would happen in the time of Khalil ﵀ was no invention out of nothing. It was the definitive shaping of something the community had already been practising unsystematically for a century.

The Intermediate Step: Yahya ibn Ya’mar ﵀ and Nasr ibn Asim ﵀

Under the governorship of al-Hajjaj ibn Yusuf, who ruled Iraq with a hard hand, the first systematic attempt at consonantal clarification was undertaken. The school of Abu al-Aswad ﵀ carried his legacy on to a chain of pupils, among them Yahya ibn Ya’mar ﵀ (died ca. 90 AH) and Nasr ibn Asim al-Laythi ﵀ (died ca. 100 AH). To them the classical tradition connects the task of applying the dots that would separate the identical skeletons from one another, a task that fanned out from the same Basran workshop from which the vowel system had also come.

These skeleton dots were set in black ink, deliberately distinguished from the coloured vowel dots of Abu al-Aswad ﵀. In the convention that tradition hands down, one dot above the basic form of the skeleton made it a ن, one dot below it made it a ب, two dots above it produced a ت, three dots above it gave a ث, and two dots below it formed a ي. The same principle held for the pairs ج, ح, خ and for ر, ز and for س, ش and for ص, ض and for ط, ظ and for ع, غ and for ف, ق. In total the Arabic alphabet counts fifteen letters that carry a distinguishing dot pattern.

The message was strict: the rasm remained untouched, this concerned supplementary markings and not modifications, and the Mushaf of Uthman ﵁ retained its consonantal skeleton word for word. The black dots were placed around it, as notation on the score, not as a change to the score.

But there was now a new problem, for on one and the same page two independent dot systems ran together: red vowel dots that hovered above, below and beside the final letters, and black consonant dots that marked the skeleton itself. An inattentive reader, or a weary scribe in the third hour of the evening, could easily confuse the two, and an inexperienced eye lost its way in a fragment where three red dots and two black dots almost touched. It could not be sustained in the longer term, not for a community that stretched from the Pyrenees to the Indus.

In the Mushafs from this transitional period, of which fragments have come down to us in the Sana’a collection, we see that tension rendered, with the vowel dots in colour and the letter dots in black on a single page, sometimes clearly separated, sometimes pressed almost against one another. It was, as scholars would later say, a workable but inelegant system, one that worked for whoever already knew it and made life harder for whoever still had to learn it.

In the same period ran another, more sensitive debate. Some Mushafs of early Companions ﵃ and Tabi’un knew local spelling variants in line with regional dialects. On one folio in a Kufic Mushaf the word حتى was sometimes spelled as عتى, and the word صراط took on the spelling سراط in certain manuscripts. These were no textual modifications, but phonetic renderings that fell within the variation of the seven ahruf. Yet they pointed to a deeper reality: without a standardized scribal apparatus, the beauty of the oral tradition would always have to bear the vulnerability of the written.

An incident has been preserved that vividly illustrates the tension of this period. Al-Hajjaj ibn Yusuf, the governor of Iraq, heard that a certain Ubaidullah ibn Ziyad had had roughly two thousand additional letters added to a Mushaf. Al-Hajjaj reacted furiously and had the man who had carried out the work, Yazid al-Farsi, summoned. Yazid was certain that he would be put to death. When al-Hajjaj had him brought before him, he asked what Ubaidullah had actually wanted. Yazid answered that it concerned reconstructions of omitted alif and waw marks, a form of standardization within the existing conventions and no modification of the text, and al-Hajjaj, after some reflection, let him go unharmed. Al-Azami ﵀ notes here that the isnad of this tradition is weak and defective, so that the scholars have sufficient reason to treat it with caution.

What Ubaidullah had done, and what Yazid carried out, was no assault on the Book. It was a clumsy attempt at orthographic uniformization within the existing Uthmanic conventions. Yet the story shows how sensitive the community had become to any touch of the rasm. A governor threatened death over what was in fact a matter of notation. With that the legal red line was drawn for all time: the addition of reading aids around the letters was permitted, the alteration of the letters themselves was under no circumstances allowed.

The Young Khalil

At that moment a new figure stepped forward from Basra, a man who would forever change the history of the Arabic language. Al-Khalil ibn Ahmad al-Farahidi ﵀ died around 170 AH. The classical biographies place his birth around 100 AH in Oman, from the tribe al-Azd, and they relate that he spent the greater part of his life in Basra, the city where Arabic linguistic science came to flower in those years.

Whoever reads the traditions about him recognizes a type that recurs more than once in Islamic scholarship, a man of sharp mind and utterly austere way of life. Khalil ﵀ lived in a simple hut on the edge of Basra, lived on almost nothing, and repeatedly refused gifts from the Abbasid caliphs and courtiers who wished to bind him to the court. He preferred bread and water in freedom over luxury in dependence. His students said of him that he had more knowledge than wealth and more wealth than he showed.

His intellectual achievements are, if one follows the classical traditions, hard to summarize. To him is ascribed the founding of ʿarud, Arabic prosody, the science of the metres. The tradition tells that he hit upon the idea for his metrical system while walking past the workshop of a smith and hearing the rhythmic striking of the hammer on the anvil, and that in that sound he recognized the patterns the pre-Islamic poets had followed intuitively. He codified what until then had only been felt, and his metrical circles are taught to this day.

More important still is the tradition that ascribes to him Kitab al-'Ayn, the first true Arabic dictionary, which he is said to have begun and which his pupils completed after his death. The ordering was remarkable, for it was not alphabetical but phonological, beginning with the letter ʿayn because, according to Khalil ﵀, it occupied the deepest place in the throat. The dictionary worked its way up through the speech organs, from throat to tongue to lips, and it would shape Arabic lexicography for centuries.

Among his students was Sibawayh ﵀, according to the classical tradition a Persian from Shiraz who travelled to Basra to study with Khalil ﵀ and who would eventually compose Al-Kitab, the greatest grammatical work of the Arabic tradition and the standard work on which all later Arabic grammar rests. Sibawayh ﵀ died around 180 AH. The relationship between teacher and pupil was exceptional, for Khalil ﵀ supplied the phonetic and metrical infrastructure and Sibawayh ﵀ built upon it the complete grammatical edifice. It is telling that Sibawayh ﵀ died before Hunain ibn Ishaq, the scholar whom orientalists sometimes suspect to be the silent source behind the Arabic diacritical system, was even born. The complete Arabic grammar already stood on its foundations long before any external influence was possible, and the question of who borrowed from whom answers itself the moment one lays the datings side by side.

Khalil ﵀ lived austerely, learned eagerly and passed on in abundance. A pupil who came to study with him needed to bring no money, no letters of recommendation, no social standing, but only patience and memory. Whoever follows the tradition hears the same portrait each time: a lean figure in a simple robe, with sharp eyes and a calm voice, who in his dwelling on the edge of Basra passed on from morning to evening what he had learned and worked out. The classical biographers tell that he declined a princely offer from the court because it would require moving to the seat of power, and that he returned a governor’s gift unopened with the message that his own bread was sufficient. He understood freedom of mind as the first condition for pure inquiry, and he let no one undermine that condition.

In that same disposition of life one recognizes the inner logic of his diacritical work. Whoever is himself imperturbable sees when something else is disturbed. Whoever seeks simplicity in his own life finds simplicity also in his designs. The system he would design for the Arabic text would carry the same traits as his personality: clear, austere, distinguishing, enduring.

It is in this threefold context, prosody, lexicography and grammar, that Khalil ﵀ also turned his attention to the diacritical problem. He was not only a theoretician of language; he was an engineer of legibility. He knew that the existing system of red and black dots would not hold up alongside the growing complexity of Arabic literature and the increasing need for accessible Mushafs.

Making the Problem Visible

What Khalil ﵀ saw was simple and, once stated, inescapable. The dot system for consonants and the dot system for vowels both used the same graphic tool, namely the dot. The only distinction was colour. Whoever had no red ink to hand, or who copied from a Mushaf in which the colour had faded, or who was simply in haste, could not reliably tell the two apart.

Moreover, there were centres where divergent dotting conventions were followed. Ibn Ushta reports that the Mushaf of Isma’il al-Qust, the imam of Mecca between roughly 100 and 170 AH, carried a dotting system that deviated from the Iraqi one, and ad-Dani ﵀ notes centuries later that the scholars of Sana’a followed yet another convention, that the Madinans differed from the Basrans, and that only by the end of the first century did the Basran conventions become so dominant that even the Madinans adopted them. Different regions, different hands, no visual unity.

The solution, Khalil ﵀ saw, was that the vowel system needed its own visual vocabulary, a vocabulary that could not in principle be confused with the consonant dots. Not only a different colour, but a different form. What the Arabic language offered to derive those forms from lay ready at hand: the letters themselves. Every vowel was determined by a long counterpart in the Arabic alphabet. The fatha sounded like the short ا. The damma sounded like the short و. The kasra sounded like the short ي. Whoever could print those letters in miniature above, on or below the consonant had a reading system that explained itself.

The Modern Marks Are Born

Khalil ﵀ set to work. He replaced the coloured dots with marks of their own form, marks that were borrowed shapes of the letters themselves, and that series has remained the same for the Arabic reader to this day.

The forms that the later qira’at tradition connects to him show that logic clearly. The fatha became a small slanted stroke above the letter, originally conceived as a reduced ا lying on its side. The kasra became the same stroke, but placed below the letter, likewise conceived as a reclining ا, this time mirrored. The damma became a small و above the letter, recognizable as the head of that letter. The sukoon, the mark for the absence of a vowel, became a small open circle, a zero above the consonant. The shadda, the mark for doubling, received a form derived from the letter ش, because shadda begins with that letter. The hamza, the glottal stop, received a form derived from the letter ع, because hamza and ʿayn spring from the same place in the throat.

Every marking was a shortened reference to the letter that represented the sound, not arbitrary and not decorative, but systematic and mnemonic. A pupil who knew the letters of the alphabet could recognize the diacritical marks as kin of those letters, and thus the system taught itself. What Khalil ﵀ added was therefore no new layer of meaning, but a making visible of what the fixed rasm had always contained: the sound now stood on the page that earlier had lived only in the mouth.

The tanween, the nunation at the end of words, had already been marked with doubled dots in the older system of Abu al-Aswad ﵀. Khalil ﵀ translated that principle into his own repertoire of forms, so that two fatha strokes above it gave fathatan, two kasra strokes below it kasratan, and two damma loops above it dammatan. On that same plane he brought structure to the wasla, the connecting mark that indicated that an alif need not be pronounced, and he refined the rendering of the madda, the lengthening of vowels. No detail of Arabic pronunciation escaped his classifying eye, and what Sibawayh ﵀ would then systematize grammatically in Al-Kitab, Khalil ﵀ supplied phonetically and visually as a complete tool.

Important to understand is that Khalil’s ﵀ system was not immediately and not exclusively intended for the Mushaf, for it first gained ground in the non-Quranic texts, where grammatical works, poetry anthologies, legal treatises and literary texts took it over quickly because they were free to experiment. For the Mushaf the older colour-dot method was deliberately maintained for generations, in order to distinguish the script of the Book visually from the script of ordinary texts. Only gradually did calligraphers slide Khalil’s ﵀ marks into the Mushaf as well, at first cautiously, then in a wider circle, until his system took over there too.

Equally crucial was that the marks were visually wholly distinct from the consonant dots of Yahya ibn Ya’mar ﵀ and Nasr ibn Asim ﵀, for a fatha stroke could not be confused with a ta dot, and a damma loop could not be read as a tha three-dot. The eye separated the two layers immediately, without colour as support, and precisely for that reason the multicolouring that later calligraphers introduced, in red, green, yellow and soft blue, had become an aesthetic choice and no longer a functional necessity.

The Adoption Spreads

The spread of Khalil’s ﵀ system proceeded gradually and over a long span of time. Al-Azami notes emphatically that centuries passed before the scheme of al-Farahidi finally displaced the older method. It first gained ground in the ordinary texts and only much later penetrated, calligrapher by calligrapher, into the Mushaf as well. The red dots of Abu al-Aswad ﵀ thus did not fade in a single generation, but wore slowly out of use, not from a lack of respect for their inventor, but because a more elegant alternative imposed itself step by step.

Manuscripts from the second century, among them the specimens preserved in the Sana’a collection, show the transition step by step, for the early specimens still combine red vowel dots and black consonant dots, while later specimens carry Khalil’s ﵀ forms, first in the same black as the text and thereafter in contrasting colours for the beauty of the page. By the third century the combination was complete: consonant dots in black on the skeleton, vowel marks in various colours above and below, aya separations between the verses, and special markings for every fifth and tenth aya.

On one essential point, however, nothing changed, for the rasm of Uthman ﵁ remained untouched, word for word and letter for letter, and what Khalil ﵀ added lay above, on or below the letter, never within the letter itself. Imam Malik ﵀ (died 179 AH) was asked in his time for a fatwa on the question of whether a Mushaf might be rewritten with the modernized spelling conventions of Arabic, and he answered emphatically in the negative: for pupils’ booklets it was permitted, but the Mushaf itself had to keep the Uthmanic rasm. Ad-Dani ﵀ (died 444 AH) noted centuries later that all scholars from Malik’s ﵀ time to his own unanimously shared that same conviction. The difference between reading aid and text was inexorably upheld.

There lay the legal backbone of the whole system. The consonant was sacred. The vowel was aid. The community could add and strengthen aids, it could adapt them in colour or form, it could lay new layers over them, but it could not alter the skeleton. When Ibn al-Bawwab produced his famous Mushaf copy in 391 AH, he still wrote the Uthmanic rasm. When the King Fahd Complex in Madina published its standard edition in 1407 AH, it wrote that same Uthmanic rasm. Between 25 AH and 1407 AH the aid changed; the text did not.

Two Layers, One Text

Here lies, when one steps back, the genius of Islamic textual transmission. What the community did was not only to preserve the Book, but to learn how one preserves a Book. It made a distinction that was never drawn so sharply in the Jewish and Christian tradition, namely the difference between the unchangeable and the aiding, between the contract and the manual.

The contract was concluded in 25 AH, when Uthman ﵁ in his council of Companions ﵃ fixed the definitive rasm and had all divergent specimens replaced by the uniformized Mushaf. That rasm was the text. The manual, by contrast, was allowed to grow. Abu al-Aswad ﵀ wrote the first chapter with his red vowel dots. Yahya ibn Ya’mar ﵀ and Nasr ibn Asim ﵀ added the second chapter with the black consonant dots. Khalil ﵀ wrote the third and definitive chapter with his modernized diacritical marks. Each generation improved the legibility; no generation touched the letter.

Moreover, the system was not only added to the Mushaf. Aya separations appeared, at first in free forms, three dots in two columns, four horizontal dots, a triangular arrangement, according to the taste of the scribe, and were thereafter standardized. Surah titles were added, written in a divergent colour, often gold ink, and always furnished with an ornament. At the end of the first century the Mushaf was, under al-Hajjaj’s urging, divided into seven manazil, for whoever wished to read the Book through in a week, and in the third century the thirty juz’ sections were added for whoever wished to recite it in a month. Markers for every fifth and tenth aya set the pace, and in Mushafs from the third century even the words ʿashr, ʿishrun, thalathun were inscribed within the tenths, so that the reader knew at a single glance where he stood in his daily portion. None of these additions touched the rasm; all of them were aids to the reading of a text that did not itself move along.

Here the inquiry touches a principle that occupies a central place in Islamic thinking about text. The community always distinguished between what is transmitted by tawatur, in mass and in every generation through an unbreakable chain of oral transmission, and what was added by learned initiatives. The rasm was tawatur. The dots and tashkil were ijtihad in the service of tawatur. The believer recited what he had learned from his teacher, who had learned it from his teacher, back to the Prophet ﷺ. The marks on the page were only a written reminder of an oral reality. When a present-day hafiz opens a Mushaf, he reads not so much the marks as that which he already knows. The marks reaffirm, they do not create.

That is also the reason the Quran has survived the centuries in a manner that no other ancient religious text equals. The Book has never been primarily a written artefact. It was first and always a recited Book, a living sound in thousands of hearts, and only in the second instance a text on parchment or paper. When in history a Mushaf was lost through fire or flood, it could be reconstructed without the loss of a single letter from the memory of the huffaz. The script preserved what memory carried, and memory preserved what the script showed, in a double chain of safekeeping that ran on undisturbed through the centuries.

Whoever now opens a present-day Mushaf sees the whole history at once. The black rasm on the paper dates in its form from 25 AH. The black dots on that rasm date from the time of Yahya ﵀ and Nasr ﵀, around 90 AH. The red or multicoloured diacritical marks above and below the letters date from the time of Khalil ﵀, around 170 AH. The aya separations, surah titles, juz’ markers and pacing symbols date from the second and third centuries. It is a palimpsest of centuries, built up by dozens of generations, with at its base one unchanged text.

A Century and a Half of Perfection

With the death of Khalil ibn Ahmad al-Farahidi ﵀ around 170 AH, the phase of textual equipping was in essence completed, for every instrument a reader needed to read the Quran accurately was now available. The rasm had been fixed by Uthman ﵁ in 25 AH, the vowels had been marked by Abu al-Aswad ﵀ in the generation thereafter, the consonants had been distinguished by Yahya ibn Ya’mar ﵀ and Nasr ibn Asim ﵀ under al-Hajjaj around 90 AH, and the diacritical marks had received their modern form through Khalil ﵀ around 170 AH. Three generations of scholars, a century and a half of careful addition, and no alteration of the skeleton. Each layer clarified the reading of a text that already stood fixed; no single layer shifted a letter of it.

In the same period the qira’at, the canonical modes of recitation, also crystallized. In the classical tradition the seven great readers were fixed in the collective memory by the end of the second century: Nafi’ al-Madani in Medina, Ibn Kathir in Mecca, Abu Amr in Basra, Ibn Amir in Damascus, and Asim, Hamza and al-Kisai in Kufa. Seven men, seven cities, one Book with the variation of seven ahruf recognized in the Hadith literature, stabilized in a fixed series of transmissions. Each of these seven readers had his own chain of teachers and pupils, each of those chains could be traced back to a Companion ﵁ of the Prophet ﷺ, and none of those chains came into conflict with the Uthmanic rasm, for the rasm left just enough room for the permitted variation and not one letter more. The believer in Khorasan could now open a Mushaf, read the rasm, decipher the dots and the tashkil, and know which of the canonical recitations he practised, and the believer in al-Andalus did the same, different in dialect, identical in text.

Alongside the Mushaf the Arabic script itself also grew in elegance. Alongside the Hijazi and the Kufic, new styles developed, and calligraphers such as Ibn Muqla (died 327 AH) and later Ibn al-Bawwab (died ca. 413 AH) raised the script to art. Ibn al-Bawwab wrote his Mushaf in 391 AH, and even he, a master who dared everything that had to do with beauty, shunned every deviation from the Uthmanic spelling. The splendour was free, the text was not.

We stand, at the end of this article, on a threshold. What the community had established in 25 AH was now fully equipped to be carried through the centuries. The rasm stood fixed, the vowels were marked, the consonants were distinguished, and the written Book henceforth carried its own pronunciation on the page. The question that still remained concerned no longer the written letter but the living voice. For within that one fixed skeleton ran, since the Prophet ﷺ, more than one permitted mode of recitation, and that multiplicity in turn called for ordering.

In this article we saw the seven great readers already appear on the horizon, Nafi’ in Medina, Ibn Kathir in Mecca, Abu Amr in Basra, Ibn Amir in Damascus, and Asim, Hamza and al-Kisai in Kufa. Each of them carried his own chain of teachers and pupils, each of those chains went back to a Companion ﵁ and before that to the Prophet ﷺ, and none of those chains collided with the Uthmanic rasm, for the skeleton left just enough room for the permitted variation and not one letter more. But so long as those seven ways were passed on only orally and in dispersed fashion, their canonical status was not yet anchored in a book.

That anchoring awaited a scholar from Baghdad, in the fourth century AH. His name was Abu Bakr ibn Mujahid ﵀, and his work would fix the seven canonical qurra’ with their isnad chains, so that the certainty of recitation received the same fixed form as the written certainty that Uthman ﵁, Al-Hajjaj, Abu al-Aswad ﵀ and Khalil ﵀ had built up together. What has been told technically about the script in this article finds in the next article its oral counterpart.

In the next article we thus follow that second line to its crowning. We meet Ibn Mujahid ﵀, the man who out of the many permitted recitation traditions selected and documented the seven, each in a continuous chain to the heart of the revelation. And we shall see how the written certainty of the rasm and the oral certainty of the qira’at together carry one indivisible Book, line for line, word for word, letter for letter, through fourteen centuries.