One Thousand and One Conjectures

Palm-leaf manuscripts in most of South Asia decayed within a few centuries, so every old text we have is the survivor of repeated recopying — but the interval of that treadmill has only ever been guessed at, never measured. Scribes, however, sometimes…

The Kathmandu Valley combined a cool dry climate, political continuity, and unbroken scribal institutions, which made it a slow-decay reservoir inside a fast-decay subcontinent. If that is right, Nepal should hold the oldest surviving witnesses not just of Nepalese works but of…

The famous millions of South Asian manuscripts hide a simpler and more dangerous statistic: how many copies each distinct work survives in. A recopying economy driven by curriculum and ritual demand should concentrate copies on a small canon while leaving the long…

Histograms of Sanskrit intellectual activity — authors per century, works per century — show a striking swell between roughly 1400 and 1800, often narrated as an early-modern efflorescence. But authors are dated largely through surviving manuscripts, and manuscripts survive on a decay…

South Asianists agree the extant corpus is a fraction of what was written, but the fraction has rarely been given a number with a method behind it. Ecology has the method: mark-recapture, where the overlap between two independent samples of a population…

From about the twelfth century, Sanskrit legal culture produced nibandhas — massive topical digests that excerpted the older smṛti texts so thoroughly that a working jurist no longer needed the originals. On a recopying treadmill, not being needed is a death sentence:…

Educational canons look immovable until they are replaced, and then the replacement is nearly total, because a curriculum is a coordination game — teachers teach what other teachers teach. Sanskrit grammar offers a clean natural experiment: the seventeenth-century reorganization of Pāṇinian teaching…

Early śāstra assumed a living teacher: the terse mūla was written and the explanation was breathed. But a text on the recopying treadmill can outrun its own teaching lineage, arriving in places and centuries where nobody holds the oral key — and…

Everyone says commentaries kept Sanskrit texts alive; nobody says by how much, and an unquantified truism is doctrine, not knowledge. On the recopying treadmill a commentary multiplies its root text's survival through three concrete channels: it adds a second copying constituency, it…

For heavily taught śāstra texts the commentary was the working format: students met the sūtras already wrapped in explanation, and the bare root text became a specialist's object. Copying demand should therefore invert the intuitive hierarchy — the derivative outnumbers the original…

Sanskrit copyists closed manuscripts with stock verses of complaint and disclaimer — the broken back, the dimming eye, the plea to blame the exemplar for errors. These verses are texts too, but they travelled by a channel no catalogue models: scribe-to-scribe and…

Pre-print South Asia ran several calendars at once — Vikrama, Śaka, Nepal Samvat, Kali years — and a scribe's choice among them was not free: it followed community, liturgical habit, and local administrative practice. A colophon's era system is therefore a barcode,…

Manuscript populations have age pyramids, like human populations, and the pyramid's shape records the demography of the institution that did the copying. Hindu texts in Nepal were reproduced continuously by household paṇḍits and temple scribes into the nineteenth century, which should give…

Recopying frequency should be set by consultation frequency: a text checked constantly wears out and is replaced on a shorter clock than a text recited from memory or shelved for prestige. Jyotiṣa — the astral-science literature used for calendars, horoscopes, and the…

Every locality in pre-print South Asia needed a fresh calendrical almanac every year, produced by the thousands for immediate use — plausibly the highest-volume written genre of the subcontinent. Yet consumables do not enter archives: an expired pañcāṅga is waste palm-leaf, while…

Subhāṣita anthologies — the medieval collections of quotable verses — sampled the poetry of their day the way a sediment core samples a vanished lake: verse by verse, with attributions, from whatever was circulating. Because the anthologist sampled circulation while the recopying…

Two selection pressures squeezed Sanskrit works from opposite ends of the length scale: very long works cost too much to recopy whole, while very short works were absorbed into anthologies and compilations and lost independent circulation. The surviving population of independently transmitted…

Sanskrit works advertise their sizes — the śataka's hundred verses, the sahasra's thousand, the seven hundred of famous saptaśatīs — because a round or auspicious count was itself a literary form and a marketing claim. If the form mattered, the length distribution…

Nearly every Sanskrit work opens with a maṅgala verse saluting a deity, and the choice of deity was constrained by the author's sectarian formation even when the work's subject was neutral — a grammarian's Gaṇeśa or Śiva was inherited, not selected for…

A large share of South Asian manuscripts are multi-text bundles — several works copied together in one pothi — and what travels together was studied together: the bundle fossilizes a teacher's or owner's working set. Aggregated over thousands of manuscripts, co-occurrence should…

A pothi is an unbound stack: its first and last leaves absorb the abrasion, damp, and rodents for the whole book, and those are precisely the leaves carrying title, author, and colophon. Much of the anonymous and unidentified matter clogging South Asian…

A palm leaf's size is set by the palm, but how much writing a scribe packs onto it is set by economics and use: reference reading rewards density, recitation support rewards legibility, and copying-for-sale rewards material economy. As śāstra culture shifted toward…

When paper reached the Himalayan and North Indian copying world it did not replace palm leaf uniformly: substrate choice was a statement about a text's dignity, and ritual conservatism priced purity into the material. Adoption should therefore be genre-ordered — almanacs and…

Pre-print South Asia split its writing between media by time-horizon: perpetual claims — land grants, endowments — went onto copper and stone, while knowledge went onto perishable leaf, to be renewed by recopying. If that split was a functioning division of labour…

Thousands of copper-plate grants record their brahmin donees with gotra and Vedic śākhā — which recension each man was trained to recite — because the gift's ritual efficacy depended on it. Read together, the plates are something the manuscript record can never…

Hero-stones — memorials to men killed in cattle raids and local battles — carpet the pastoral and frontier zones of the Deccan and western India, dated and localized by the thousand. Sanskrit knowledge production, by contrast, ran on agrarian surplus channelled through…

The men who composed copper-plate praśastis were trained kāvya poets, and some signed their names on the plates themselves. If the same labour pool wrote plate eulogies and transmitted literature, a measurable fraction of named plate poets should reappear in the prosopography…

A copper-plate grant needed only the legal facts, yet over the centuries grants swelled from terse deeds into long genealogical praśastis running to hundreds of lines — and engraving cost scales with length, so the inflation was a paid-for choice. The cheapest…

Temple donative inscriptions across South Asia record substantial female donor shares: religious giving was a recognized women's economy, carved in stone by the thousand. Manuscript colophons also record commissioning patrons, and if the same merit economy governed book-making, female patronage should appear…

The Vedic recitational system, deliberately oral in its product, still generated a written maintenance literature — prātiśākhyas, śikṣā treatises, pada- and krama-text aids, accent manuals — because the error-correcting apparatus itself needed stable specification, and specification is exactly what writing is good…

The picture of South Asian copying as diffuse household piety predicts scribal output spread thin across many one-manuscript copyists. But copying was also a paid profession, and paid professions concentrate: if career scribes and workshop-like establishments did the heavy lifting, the distribution…

Early colophons are chatty and idiosyncratic; a professionalizing copying market should standardize them, the way notarial formulas and coin legends standardize — templates cut transaction costs and signal competence to patrons. Standardization is measurable as falling lexical diversity of colophon wording over…

A manuscript recited or read cover-to-cover is missed the moment leaves go missing, and gets replaced; a consultation tool — a lexicon, a table-book, a manual — remains useful with half its leaves gone, so damaged copies stay in circulation and enter…

Annotation is costly attention, and readers spend it where error hurts: a wrong drug dose or a wrongly timed rite has consequences that a wrong poetic reading does not. Marginalia and correction density should therefore rank genres by practical stakes — medicine…

In a thin intellectual market a new work waits generations for a commentator; in a crowded one, commenting on fresh work is a career move and the wait collapses. The lag between a root text's composition and its first commentary is therefore…

A text is most self-explanatory at home, where its idiom, curriculum, and oral support live; carried to another region, it needs scaffolding. Deep commentary stacks should therefore be disproportionately built abroad: the farther down the chain — ṭīkā on bhāṣya on mūla…

If commentary accumulated like sediment, older root texts would carry ever-deeper stacks: depth would grow with age without bound. But a commentary stack is a pedagogical object that students must traverse, so its depth should saturate at what a scholarly training can…

When a work composed in one region turns up in another region's collections, it could have arrived two ways: as a travelled object — an imported pothi — or as a re-inscription, a local copy made from dictation, memory, or a briefly…

Temple and monastic donative inscriptions occasionally record gifts of books — titles, sometimes counts, dated and localized — which makes them tiny dated library catalogues in a medium that survives without recopying. Because stone samples the act of donation while manuscript catalogues…

The standard awe-statistic — millions of South Asian manuscripts, most catalogued only by title — quietly treats titles as works, but title-level cataloguing errs in both directions: one work travels under several titles (splitting), and one title covers several works (lumping). The…