The Library Without Walls

A scholar in the nineteenth century who wanted to read the Rigveda in the original needed a patron, a sea voyage and years. Today the same text answers a keyword search before the kettle boils. The distance between those two facts is the subject of this last chapter.

It is also, quietly, the subject of the whole work.

The corpus goes searchable

The texts that this treatise has followed across three thousand years are now, most of them, files.

There is GRETIL, the Göttingen Register of Electronic Texts in Indian Languages, hosted at the Göttingen State and University Library: a machine-readable, searchable collection particularly strong in Sanskrit. There is the digital library of the Muktabodha Indological Research Institute, founded in 1997, strong in the Shaiva, Shakta and Shrividya traditions, holding thousands of texts with hundreds of them searchable. There is the Sanskrit Documents site, sanskritdocuments.org, a volunteer collection that offers its texts in Devanagari, other scripts and IAST transliteration.

Then there is a more recent kind of thing. The Digital Corpus of Sanskrit, built by Oliver Hellwig, is not a library of texts to be read so much as a corpus to be computed on: sandhi split, lemmatized, grammatically tagged, millions of words marked up so that a researcher can ask the whole of recorded Sanskrit a question and get a statistical answer.

Older translations have their own homes. sacred-texts.com hosts the public-domain editions, including Max Müller’s Sacred Books of the East. The Internet Archive, archive.org, hosts manuscript scans and the full out-of-copyright editions: Müller’s Rigveda, Sukthankar’s Prolegomena, the Baroda Ramayana.

And there is the state’s effort. India’s National Mission for Manuscripts, founded in February 2003 with the Indira Gandhi National Centre for the Arts as its nodal agency, surveys, conserves and digitizes a manuscript heritage estimated in the millions. Its digitization totals climb year by year, so any single figure belongs to the Mission and to a date. In 2024 the Mission was restructured under the banner of the Gyan Bharatam Mission.

A small confession belongs here. Müller’s printed Rigveda and his Sacred Books of the East, the very monuments this volume described as the printed recovery, are now free scans on those archives. Sukthankar’s Prolegomena was read for this work as an OCR file online. These chapters have been passing, all along, through the stages they describe.

Access without precedent

Weigh first what has been gained, because it is enormous.

The corpus is searchable. It is copyable. It is globally available. With the Digital Corpus of Sanskrit it is even computationally analyzable, open to questions no single human reader could ever have answered by hand. A text that once cost a scholar a lifetime and an ocean crossing to consult is now a line in a search box.

This is democratization, and it is real. The manuscript that lay in one monastery, readable by the few who could reach it and the fewer who could read the script, is now in front of anyone with a connection. The gatekeeping of geography is gone. The gatekeeping of rarity is gone.

There is a second face to this, and honesty requires naming it.

A verse, pulled up alone by search, arrives stripped. It comes without the commentary that argued over it, without the ritual that placed it, without the teacher who once stood between the student and the words and said what they did and did not mean. The corpus this work has described was never meant to be read in fragments by strangers. Access and decontextualization arrive in the same instant. The reader gains the verse and can, just as easily, mistake it.

So the gain is genuine and double-edged. Hold that, because the next gain has the same shape.

The new fragility

This treatise has been, from its first volume, a record of fragility.

Palm leaf rots in the damp. Birch bark cracks. Fire takes a library in a night, insects take it slowly, and the death of a single reciter could once take a branch of the Veda with him. The whole drama of preservation across these six volumes has been a war against material decay. The oldest surviving manuscript of the Mahabharata is only about five centuries old, which tells you how completely the older copies lost that war.

Digitization looks, at first, like victory. A file does not rot.

A file faces other deaths. Links break as servers are retired and URLs go dead, a quiet erosion the field calls link-rot (the steady failure of web addresses over time, so that yesterday’s citation points at nothing). Formats fall out of use and the software that could read them stops being made. Character encodings drift. Above all, a digital archive depends on things a buried manuscript never needed: a running server, continued funding, and electricity.

This is the trade made plain. The old fragility was material. The new fragility is infrastructural.

A manuscript can survive centuries of pure neglect in a box. A dead server survives nothing. Forget a palm-leaf bundle for two hundred years and a conservator may still recover it; forget a database for twenty and it may be gone past recovery, its medium more perishable than the bark it replaced. The danger has not been removed. It has been moved, from the object to the system that serves the object.

That is the honest ledger of the pixel. Now step back from it, all the way back, to where this began.

Four states of a word

Set the whole journey in a line and its shape becomes visible. The scripture passed through four states, in sequence, and each one is a volume of this work.

First, memory. For roughly three thousand years the Veda was deliberately not written. It was kept as sound, in trained minds, through the recitation modes, the pathas: samhita (the continuous text), pada (word by word), and the interlocking permutations krama, jata and ghana that wove each word forward and back so that no syllable could slip without the error showing. The Veda is held by its tradition to be shruti, “that which is heard,” and apaurusheya, “authorless”: received and preserved by sound, never composed by any person. The point of the pathas was a person who was himself the archive. We traced that machine made of memory in Volume I.

Second, leaf and bark. The corpus was at last committed to palm leaf and birch bark, fragile, copied and recopied, the originals always perishing under the copies.

Third, print. The nineteenth-century recovery: Müller’s Rigveda, funded by the East India Company, the Sacred Books of the East, and then the great Indian critical editions at Pune and Baroda, the subject of the chapter just before this one.

Fourth, pixel. GRETIL, the Digital Corpus of Sanskrit, Muktabodha, the National Mission for Manuscripts portal, archive.org.

At every step, something is gained: access, preservation, reach, analysis. Each medium reached more people and lost fewer copies than the last.

At every step something is also lost, or put at risk. The living oral chain thins. The trained memory that was once the medium, not merely the carrier of the content, becomes unnecessary. The teacher-to-student transmission, examined in Volume V’s chapter on keeping the sound, becomes optional. And the deepest idea quietly dies: that the text lives in a person, not an object. When the word becomes a file, the reciter becomes optional.

The chain that never stopped

There is an irony in the previous chapter that this one can now name.

Sukthankar, building the critical edition of the Mahabharata, reconstructed the epic by reasoning backward up a chain of bards who, in his account, passed it “from bard to bard … by word of mouth.” He was rebuilding, on paper, a line of human transmission that had once been alive.

The Vedic tradition is that same chain, still running forward.

Read the two together and the symmetry is hard to miss. The critical edition is the modern world straining to reconstruct, with apparatus and variants and stemmata, what one tradition simply never stopped doing in the body. One scholar reasons his way backward up a vanished chain of voices. A few hundred miles away, a chain of voices is still passing the sound hand to hand, the way it always has. The paper is trying to recover the very thing the breath never let go of.

Which raises the only question that can close a work like this one. The breath: is it still there?

The oldest way still breathes

It is. That is the fact on which this entire library can rest.

The oral tradition is not dead. Vedic shrauta recitation, the solemn public sacrificial rites, with the patha permutations, is still transmitted teacher to student in living shakha lineages, the surviving “branches” of the Veda, each one a distinct line of transmission carried in human memory. The “Tradition of Vedic Chanting” was recognized by UNESCO: proclaimed a Masterpiece of the Oral and Intangible Heritage of Humanity in 2003, and inscribed on the Representative List of the Intangible Cultural Heritage of Humanity in 2008.

UNESCO’s own description of the practice reads almost as a summary of this work’s first volume. Its practitioners, it notes, are “taught from childhood complex recitation techniques … based on tonal accents, a unique manner of pronouncing each letter and specific speech combinations.” The machine made of memory is still being assembled, child by child.

Here tradition and record meet without contradiction. The tradition holds that the Veda is shruti, eternal, apaurusheya, properly kept by sound in an unbroken chain, living in a person rather than an object. The record shows a corpus transmitted orally with extraordinary fidelity for millennia, then committed to manuscript, printed, digitized. The UNESCO inscription is, in effect, the record’s institutional acknowledgement of the tradition’s living practice. Both are true. Here they hold each other up.

One thing must be said plainly. The recognition exists because the practice is endangered. UNESCO inscribed the chant for the same reason a species is listed: fewer shakha branches survive with each generation, and what is not protected may not last. The honour is also an alarm.

So the close is not a triumph. The unwritten library has finally been written, then printed, then uploaded to servers that hum in rooms its first reciters could not have imagined. Every step gained the world more of the text and asked the world for a little less of the voice. And still, in temples and schools that the search box will never index, a teacher faces a student and says the syllable, and the student says it back, exactly, the way it has been done for three thousand years.

The library without walls is everywhere now. The oldest way of keeping it still breathes, for now.

This is the last chapter of The Unwritten Library. The whole of it has tried to follow one word from a mind to a file. The word is still being said aloud. That is where the work began, and it is the only place fit to leave it.