The lost book is a row of shared digits

Claim (verbatim)

Āryabhaṭa (born 476) wrote two astronomies: the Āryabhaṭīya, which survives, reckoning the day from sunrise, and a second system reckoning from midnight — the ārdharātrika system, the so-called Āryabhaṭasiddhānta — which is lost. Its afterlife is purely parametric: Brahmagupta built his working handbook, the Khaṇḍakhādyaka (665), expressly on Āryabhaṭa's midnight system, and Bhāskara I's Mahābhāskarīya devotes a chapter to reporting the midnight system's constants alongside the sunrise system's. A lost siddhānta reduced to numbers in other men's books is recoverable exactly insofar as the numbers were fixed: if one written text stood behind both witnesses, the parameter vectors they transmit must agree digit for digit; if "the midnight system" was a school habit rather than a book, the reports should drift the way remembered numbers drift. K. S. Shukla assembled the ārdharātrika testimonia half a century ago; the two-channel comparison is a table anyone can build from published editions, and it decides whether this ghost was a text or a rumour. Prediction: tabulating every parameter of the midnight system independently recoverable from the Khaṇḍakhādyaka and from the Mahābhāskarīya's account — revolution numbers in a yuga, civil-day counts, epicycle circumferences, apogee and node positions — at least 90% of the shared parameters will agree exactly, digit for digit, across the two channels, fixing a single written source behind both (primary clause: the 90% exact-agreement floor; systematic divergence in the core revolution numbers kills it). Kill: the Khaṇḍakhādyaka editions of P. C. Sengupta (Calcutta, 1934) and Bina Chatterjee (New Delhi, 1970), K. S. Shukla's edition of the Mahābhāskarīya (Lucknow, 1960) and his study "Āryabhaṭa I's astronomy with midnight day-reckoning" (Gaṇita 18, 1967), controlled against David Pingree's Census of the Exact Sciences in Sanskrit, s.v. Āryabhaṭa.

Prediction clause (verbatim)

Prediction: tabulating every parameter of the midnight system independently recoverable from the Khaṇḍakhādyaka and from the Mahābhāskarīya's account — revolution numbers in a yuga, civil-day counts, epicycle circumferences, apogee and node positions — at least 90% of the shared parameters will agree exactly, digit for digit, across the two channels, fixing a single written source behind both (primary clause: the 90% exact-agreement floor; systematic divergence in the core revolution numbers kills it).

Kill-dataset (verbatim)

Kill: the Khaṇḍakhādyaka editions of P. C. Sengupta (Calcutta, 1934) and Bina Chatterjee (New Delhi, 1970), K. S. Shukla's edition of the Mahābhāskarīya (Lucknow, 1960) and his study "Āryabhaṭa I's astronomy with midnight day-reckoning" (Gaṇita 18, 1967), controlled against David Pingree's Census of the Exact Sciences in Sanskrit, s.v. Āryabhaṭa.

Novelty / leakage triage

already answered in the literature

Shukla assembled the ardharatrika testimonia and reconstructed the midnight system precisely from the agreement of the two channels, and the standard accounts state that nearly all the system's constants are shared between the Khandakhadyaka and the Mahabhaskariya's report - the digit-level agreement fixing a single written source is the published basis of the reconstruction, with only the 90% bookkeeping left.

• K.S. Shukla, 'Aryabhata I's astronomy with midnight day-reckoning', Ganita 18 (1967)
• B. Chatterjee (ed.), The Khandakhadyaka of Brahmagupta (New Delhi, 1970), introduction and parameter tables

Predictions

No prediction registered yet.