Leaf and Bark

A leaf is cut from a palm tree in the wet heat of the Tamil country. It is boiled, dried, rubbed smooth, and a man draws a metal point across its surface until letters open in the skin like fine scars. Then he wipes it with oil and soot, and the dark fills the wounds, and a verse that has lived for a thousand years in the mouth at last has a body.

The body is mortal. That is the whole problem of this chapter.

For most of the preceding volumes the word in India refused the page. It lived in breath and memory, guarded by recitation so exact it could survive without ink. Volume VI follows what happened when the word finally consented to a surface. The first chapter asked why writing was held back. The second traced the scripts that carried it once it came. This chapter asks the plainest material question of all. When the word met a physical surface, what was that surface, how fragile was it, and how did anything survive at all in a climate that eats books?

The leaf

The dominant surface across the south and east of the subcontinent, from Bengal down to the Tamil country, was the palm-leaf. In Sanskrit it is tala-patra or tada-patra, literally palm-leaf. Two species supplied it: Corypha umbraculifera, the talipot palm, and Borassus flabellifer, the palmyra. The leaves were cut, boiled or otherwise cooked, dried, and smoothed.

What happened next is the signature of the southern and eastern workshop. The scribe did not write on the leaf. He cut into it. A sharp metal stylus incised each letter into the surface, scratch by scratch. The leaf at this stage was nearly blank to the eye, the text a barely visible network of grooves. Then the scribe wiped the whole leaf with a cloth soaked in oil and lampblack, the fine soot of a burning lamp. The black pigment lodged in the incised lines and was rubbed away from the smooth surface. The letters surfaced out of the leaf, dark on pale.

This is why early Indian scripts of the south curve and loop as they do. A straight horizontal stroke would split the grain of the leaf. The hand learned to round its letters to spare the surface, and the scripts grew their curves from the discipline of the material itself. We met that branching in the previous chapter; here is one of its physical causes.

The leaf decided more than the shape of the letters.

The bark

Move north, into the foothills and into Kashmir, and the surface changes. Here the common material was birch-bark, bhurja-patra, taken from Betula utilis, the Himalayan birch that grows high on the slopes. The fine inner bark peels away in pale sheets. Kashmir was a noted centre of its manufacture, and birch-bark remained in use there remarkably late, as far as the seventeenth century.

Birch-bark was written, not incised. The scribe used a pen or stylus and ink, laying the letters onto the surface as one lays them onto paper. It was also the most fragile of the common writing supports, thin and brittle, prone to flaking and to splitting along its layers.

It is tempting to draw a clean map: south equals incised palm-leaf, north equals inked birch-bark. The truth is a tendency, not a law. The scholar Dominik Wujastyk, in his survey of Indian manuscripts, stresses that the deep distinction is one of technique rather than strictly of material. Incising versus inking is the real divide. Some older palm-leaf manuscripts, the wide leaves of Bengal and Nepal, were inked rather than incised. So the picture is a strong regional habit with real exceptions, and it pays to hold it loosely.

Two materials, then. One you cut and darkened. One you wrote on and watched flake. Both shared a single shape.

The shape

The characteristic Indian book is not a thing of spine and turning pages. It is the pothi, an oblong, landscape-format stack of loose leaves. The word descends from Sanskrit pustaka, book. The folios are wide and short, pierced with one or two holes; a cord runs through the holes to keep them in order; the whole stack is sandwiched between two wooden boards and bound tight with the cord. To read, you untie it and turn the leaves one by one.

That oblong shape is the shadow of the palm leaf. A palm leaf is long and narrow, and the page took its proportions from the plant. The cord-hole solved a real problem: loose leaves in that climate scatter and are lost, and a string through a hole keeps the sequence and the count.

Here the format does something that historians find quietly revealing. When paper at last reached India and was widely adopted, the pothi shape did not die. Paper manuscripts kept the same oblong, landscape proportions, even though paper can be bound in any form. More striking still, as David Weston of Glasgow University Library has noted, paper scribes often kept the cord-hole when it no longer held a cord. The string was gone, but the scribes drew the hole anyway, two circles interrupting the lines of text where the holes had been, sometimes decorated, a painted ghost of a fastening the page no longer needed. A habit of the leaf outlived the leaf.

The form survived. The substance did not, and that is where the real history begins.

The copying imperative

Set a palm leaf or a sheet of birch-bark in a house in the Gangetic plain and wait. The monsoon arrives. Damp swells the fibres; mould blooms in the dark of the binding; insects find the leaves and bore through them; rats find the bundle and gnaw. An organic surface in a hot wet country is a meal.

So the central fact of Indian manuscript culture is not the writing. It is the rewriting.

A manuscript decaying could be saved only one way: by copying it out again onto fresh material before it failed, and then copying that copy when its turn came, and so on, generation after generation, century after century. The library was not a building of permanent objects. It was a relay. Wujastyk makes the consequence explicit: a text may be transmitted through copy after copy for over a thousand years before we have any physical evidence for the complete text. The text is old. The thing in your hand is young.

How young depends on the material and the luck. For surviving South Indian palm-leaf the honest statement carries two numbers, not one. The corpus that has actually come down to us from the humid south is rarely older than three to four centuries, because that is roughly what the climate permits to survive in ordinary keeping. Yet a well-prepared and well-kept leaf can last a millennium or more; that is the ceiling of the material, not the average of the wreckage. Both numbers are true. They describe different things, the survivors and the possible.

This is why the search for an ancient autograph, an author’s own manuscript, almost always fails. There is rarely an original. There is a chain of copies, each scribe an unpaid link in the survival of a sentence.

And yet the corpus is enormous. India’s National Mission for Manuscripts, founded in 2003, has documented on the order of five million surviving manuscripts and more, probably the largest such body in the world. Five million young copies of much older words.

Someone had to keep the relay running. That work had homes.

The treasure-houses

The places that preserved manuscripts were not neutral storerooms. They were institutions with a reason to copy.

Among the most important were the Jain jnana-bhandaras, the knowledge-treasuries, where a religious community held the act of preserving texts as a sacred duty. Those at Jaisalmer and at Patan and Cambay in Gujarat are famous for keeping very old palm-leaf manuscripts through the centuries, in dry desert air and under disciplined custody. The word itself, jnana-bhandara, knowledge-treasury, tells you how the books were regarded: as wealth to be guarded.

Hindu institutions did the same work in their own forms. Temples held collections; so did the matha, the monastic establishment, the seat of a teaching lineage. A matha was a place where a text could be both housed and recopied, custody and reproduction under one roof, the relay institutionalised.

There were also royal libraries. The Saraswati Mahal Library at Thanjavur grew from the collections of the Nayak and then the Maratha rulers of the city. It was greatly expanded under Serfoji II, who reigned from 1798 to 1832 and was a serious collector, and it was opened to the public in 1918. It survives as one of the richest manuscript libraries in the world.

These were the keepers. A few survivals, against the odds, let us touch the page directly.

The survivors

In 1931, near Gilgit in the far north, a cache of Buddhist birch-bark manuscripts came to light. The Gilgit manuscripts are written in Buddhist Hybrid Sanskrit, in a Gupta and post-Gupta Brahmi script, and are usually dated to around the fifth and sixth centuries CE. They are inscribed on the UNESCO Memory of the World register, and they are often described as among the oldest surviving manuscripts available in India.

That exact qualifier, oldest surviving in India, matters. Genuinely older South Asian written material survives, but abroad and out of the killing climate. The Gandharan birch-bark scrolls, dated roughly between the first century BCE and the third century CE, sit today in the British Library. The plains that produced and recopied most Hindu manuscripts kept almost nothing that old. What survived from deep antiquity survived where the air was dry or the manuscript was buried.

The second survivor comes with a controversy worth telling carefully. In 1881, near the village of Bakhshali, now in Pakistan, a birch-bark manuscript came out of the ground, a mathematical text written in the Sharada script. It is held now at the Bodleian Library in Oxford. In 2017 the Bodleian radiocarbon-dated three of its folios and got three startlingly different results: one folio to roughly 224 to 383 CE, one to roughly 680 to 779 CE, and one to roughly 885 to 993 CE. The library publicised the earliest date and announced the manuscript as containing the world’s oldest zero.

Scholars pushed back at once. Kim Plofker, Agathe Keller, Dominik Wujastyk and others objected on a basic principle of dating composite objects: a codex assembled from leaves of different ages should be dated by its latest datable folio, not its earliest, since the whole could not have existed before its youngest leaf was made. The three carbon dates do not point to one ancient book. They point to a bundle assembled from material of different centuries. In 2024 Oxford revised the dating of the manuscript to roughly 799 to 1102 CE.

So the Bakhshali manuscript remains a treasure and a genuinely early mathematical document. The claim that it holds the world’s oldest zero is not settled, and the careful reader keeps all three carbon dates, the objection, and the revision in view at once.

The two accounts

Tradition and record do not quite agree, and the disagreement is the truth of this chapter.

Tradition holds that the Veda is eternal, that its ideal preservation is the flawless recitation we followed in the volume on memory, and that the manuscript is secondary, a crutch for the living voice. By that account the written leaf is almost beside the point.

The record shows what physically survives, and it is humbler. A vast corpus of relatively young copies. Almost no ancient autograph anywhere. The oldest survivals not the central scriptures of the plains at all, but Buddhist birch-bark kept alive by dry climates and burial, far from the humid country where most Hindu manuscripts were copied and recopied into the present.

Both accounts are honest. The voice carried the word through the centuries; the leaf carried it only as long as a leaf lasts, which is why the leaf had to be made again and again. The next chapter takes up the machine that would finally break the relay, fix the copy, and change the word forever: the press.

The leaf darkens under the soot. Somewhere a scribe begins again.