Ars Inquirendi

← All conjectures · South Asian text cultures

The chronicle fills its hole with round numbers

Status: Supported

The verdict’s fine print — quoted from the resolution record: “SUPPORTED with ~6x room over the registered bar: the wager was a mere twofold ratio and the record gave 11.89x (div-by-10: 10.27x; >=10x under every registered variant).” Read the full caveats ↓

Status is derived only from the shepherd-authored triage/prediction data above -- community submissions and claims are a separate overlay and can never change it (see the participation panel below).

This is a conjecture imagined by a language model — drawn from its trained weights and held to falsifiability, novelty, and plausibility, not to any one method: it may join two or more fields, or none. It is not an article and not evidence: it sits below the evidence/publication boundary. A quantitative prediction and a named kill-dataset are attached (when registered) so the claim stays falsifiable rather than merely evocative.

Claim (verbatim)

Kalhaṇa opens the Rājataraṅgiṇī (1148-49) with a source audit unique in Sanskrit letters: he names predecessors now mostly lost — Suvrata's condensed poem, Kṣemendra's Nṛpāvali ("not a single part free from error"), Helārāja's twelve-thousand-verse Pārthivāvali, Padmamihira, Chavillākara — and confesses that the records of thirty-five early kings had already perished. Yet his frame demanded a continuous chronology anchored deep in the Kali era, so the confessed holes still had to be filled, and filled numbers have their own physics: rounded, astrologically convenient reign-lengths and impossible durations (Raṇāditya reigns three hundred years), where documentary stretches carry archival granularity — odd years, months, and days. The seam between invention and record should therefore be arithmetically sharp at the point his sources thicken, the Kārkoṭa accession that opens Book 4. Prediction: extracting every stated reign-length from Stein's chronological tables, the share of reigns divisible by five (or exceeding fifty years) in Books 1-3 will be at least double the share in Books 4-8, and the maximum stated reign will drop by an order of magnitude across the Book 3/Book 4 boundary (primary clause: the twofold divisibility ratio; the verdict follows it). Exact computation: tabulate reign figures with year-month-day precision flags from Stein's tables and compare the two distributions. Kill: M. A. Stein's edition (1892) and annotated translation (1900) of Kalhaṇa's Rājataraṅgiṇī, both digitized on archive.org, whose chronological tables give every stated reign length.

Prediction clause (verbatim)

Prediction: extracting every stated reign-length from Stein's chronological tables, the share of reigns divisible by five (or exceeding fifty years) in Books 1-3 will be at least double the share in Books 4-8, and the maximum stated reign will drop by an order of magnitude across the Book 3/Book 4 boundary (primary clause: the twofold divisibility ratio; the verdict follows it). Exact computation: tabulate reign figures with year-month-day precision flags from Stein's tables and compare the two distributions.

Kill-dataset (verbatim)

Kill: M. A. Stein's edition (1892) and annotated translation (1900) of Kalhaṇa's Rājataraṅgiṇī, both digitized on archive.org, whose chronological tables give every stated reign length.

Provenance

Run: Fresh agent generation · model: claude-fable-5

Fresh blind generation by claude-fable-5, 2026-07-16, breadth wave weighting India/South Asia by inferred textual production rather than survival; every item grounded in real works, authors, catalogues, and testimonia, including the real evidence of loss (citing authors, catalogue entries, translation corpora, rediscovery cases); no fabricated citations.

Novelty / leakage triage

anticipated in the literature — this exact test has never been run

Stein's edition itself flags the legendary chronology of Books 1-3 (Ranaditya's 300-year reign, the misplaced Mihirakula) against the documentary alignment that begins with the Karkota book, and this Book 3/4 reliability seam is canonical in the scholarship. A digit-preference/roundness analysis of the stated reign lengths across the seam has not been run.

Sources cited by the triage
  • M.A. Stein, Kalhana's Rajatarangini: A Chronicle of the Kings of Kasmir (London, 1900), Introduction and chronological tables
  • S. Kaul, The Making of Early Kashmir: Landscape and Identity in the Rajatarangini (Oxford University Press, 2018)

Predictions

Supported registered 2026-07-17 calibration prediction (parent triage: leaked/adjacent)

Resolution: Supported

Caveats: SUPPORTED with ~6x room over the registered bar: the wager was a mere twofold ratio and the record gave 11.89x (div-by-10: 10.27x; >=10x under every registered variant). CALIBRATION NOTE: this is the resumed loop's second supported (after breadth-india-003, which wagered 3x and got >=8.4x), and it refines the live finding -- not 'multiplicative always dies' but 'OVER-pinned magnitude dies; a low, conservatively-set bar survives with room to spare'. COVERAGE LIMITATION, stated honestly: the late group (n=20) is carried almost entirely by Book IV (16/20, the Karkota dynasty) plus 4 from Books V-VI; Books VII-VIII contribute ZERO year-stated reigns because their reigns are given only in months/days -- the granular archival form the conjecture itself predicts for the historical era -- so the measured contrast sits at the Book III/IV seam rather than spanning the full Books IV-VIII, which STRENGTHENS the registered seam-sharpness claim (the break at the Karkota accession) even as it narrows what the late-group share empirically samples. All analytic choices cut CONSERVATIVE (toward a lower R): including Books VII-VIII's short non-round reigns, or counting the 35 excluded days/months-only kings (concentrated late) as non-round, would only raise R; the sole reading that flips the verdict is the degenerate zero-fill treating a days-long reign as '0 years, divisible by 5 = round'. DISCLOSED-NO-WEIGHT: (a) OCR italic/roman type is lost, so some individual figures may be Stein-computed rather than Kalhana-stated (only Yudhisthira I's could be positively identified via footnote and is tagged as such); (b) two cells (Gokarna months, Yudhisthira I days) remain illegible without affecting any years digit; (c) single-edition extraction (Stein 1900, via the cleaner alternate scan of the same Appendix I) -- faithful to Stein's tables, not a manuscript re-collation.

Registered against M. A. Stein's chronological tables for Kalhana's Rajatarangini (1900 translation, archive.org). Claim under test (primary clause): the share of reign-lengths divisible by five (or exceeding fifty years) in the legendary Books 1-3 is AT LEAST DOUBLE the share in the historical Books 4-8.

Resolution criteria — the registered fine print

Resolution criteria: DATA: Stein's 1900 annotated-translation chronological tables (archive.org), which give every stated reign length with precision; freeze the source. POPULATION: every stated reign-length, grouped by Book (1-8); GROUP EARLY = Books 1-3, GROUP LATE = Books 4-8. ROUND-NUMBER SHARE(group) = (reigns divisible by 5 OR exceeding 50 years) / (total reigns) in that group. PRIMARY R = share(Books 1-3) / share(Books 4-8). CLAUSE PRECEDENCE: (1) INCONCLUSIVE_BY_DESIGN if Stein's tables cannot be obtained/parsed, OR fewer than 10 reigns fall in either group. (2) SUPPORTED if R >= 2. (3) KILLED if R < 2. Report both shares, per-book reign counts, R, and the SECONDARY (non-binding): whether the maximum stated reign drops by an order of magnitude across the Book 3/Book 4 boundary. If digit-preference is measured a different reasonable way (e.g. divisible-by-10, or a Benford/last-digit test), report whether the verdict is stable across it. computed_at postdates registered_at.

Known-priors disclosure — what the registrant already knew

Known priors disclosure: Held: the conjecture's reasoning (Kalhana confesses lost sources and fills the legendary early reigns with round numbers) and general knowledge Stein's tables are on archive.org. NOT computed: the divisibility shares or R. Twofold threshold is the conjecture's own.

Method and dataset — how it was measured

Firewalled blind compute (claude-sonnet-5, blind to the R>=2 threshold and to direction) fetched and parsed Stein's Appendix I with stdlib-only tooling and reported per-king reign rows, shares, and robustness variants. The Opus orchestrator independently recomputed the primary ratio from the reign list (R = 11.892, exact match) and verified key figures (incl. the 300-year Ranaditya outlier) against the source OCR. Verdict authored by claude-fable-5 per the registered clause precedence: guard (1) passed; clause (2) requires R >= 2 -- share_early = 22/37 = 0.5946, share_late = 1/20 = 0.05, R = 11.89, exceeding the wagered twofold bar by ~6x; SUPPORTED. Clause (3) never reached. Robustness: div-by-10 variant gives R = 10.27; last-digit distribution shows 40.5% of early reigns ending in 0 vs 5% of late -- the verdict is stable under every registered variant. SECONDARY (non-binding) confirmed: max stated reign drops 300 (Book III, Ranaditya) to 50 (Book IV) = 83.3% across the III/IV boundary, and even excluding the 300 outlier Book III's next-highest (Pravarasena II, 60) still exceeds Book IV's max; the one round late reign is Durlabhaka-Pratapaditya II (50), the transition king opening Book IV.

Dataset: Stein 1900 Rajatarangini Vol I, Appendix I 'Chronological and Dynastic Tables' covering all 8 Books, extracted from the cleaner-OCR alternate archive.org copy kalhanasrajatara01mast (sha256 0b33db1bac77883445e53a6348b8cf65d849c5f76e7099d0c9e16087208f73af) after the pinned RajataranginiVol1 copy's numeric columns proved garbled as pre-flagged. Extraction cross-validated three ways: (a) three spelled-out running-text reign phrases in Vol I matched the table exactly (Baka 63y13d, Gopaditya 60y6d, Nara I 40y9m); (b) accession-date chaining held across Books I-IV; (c) per-dynasty printed aggregate sums matched exactly for Books II, III, IV and to within 5/1002 years for Book I. Population: 57 reigns with an explicit stated-years figure (37 early Books I-III, 20 late Books IV-VIII); 35 kings with no years digit (days/months-only or no duration) excluded rather than zero-filled -- the conservative choice, since they concentrate in the late group and zero-filling would count them as round; Jayasimha excluded as an open reign per Stein's own footnote. Guard clause (1) NOT triggered: tables obtained and parsed, and both groups clear the >=10 floor (37 and 20).

computed 2026-07-20

Weigh in

No community feedback yet.

New here? Create an account first

Create an account or sign in and your feedback is tied to you — you can track it, get replies, and claim this conjecture so others know you’re working on it. Prefer not to? Just leave your take below as a guest — only the name you type is shown.

Add your take

Posted immediately (spam is removed). Community feedback is never an adjudicated verdict and never changes this conjecture's triage label or status above.