The caesura at twelve-twenty

Claim (verbatim)

The Mongol conquest of Khurasan and Transoxiana - Bukhara, Samarqand, Balkh, Marv and Nishapur sacked in 1220-1221, their populations massacred and their libraries destroyed - fell on the very region that had produced the first three centuries of Persian poetry. The transmission consequence is a survival watershed keyed to a date. A poet who died before the storm depended, to reach us, on copies that had already dispersed westward or southward out of Khurasan before 1220; a poet who died after it wrote into the surviving book-cultures of Fars, Anatolia and the Delhi Sultanate, which escaped the same devastation. The mechanism predicts not a smooth decay with age but a step: survival should drop sharply across the pre/post-1220 boundary, over and above any gentle trend, because the loss was an event and not an erosion. Measured on Ganjoor, poets whose deaths fall in the half-century before 1220 should carry systematically smaller surviving corpora - and a higher rate of no-corpus-at-all - than poets of the half-century after, even though the earlier group had more time to be canonized and recopied. The counter-intuitive part is exactly that: more runway, less survival, because the runway ran through a fire. Prediction restated: the pre-1220 cohort's median surviving corpus is a fraction of the post-1220 cohort's, a step you can see at the conquest date.

Prediction clause (verbatim)

Prediction: binning tazkira-attested poets by death-date, the median surviving Ganjoor corpus in bayts of the cohort dying 1170-1220 will be under half that of the cohort dying 1220-1270, and the no-surviving-corpus rate will be higher in the earlier cohort despite its longer canonizing runway (primary clause: the pre-1220 median corpus below half the post-1220 median; the verdict follows it). Death-dates are taken from Safa's Tarikh-i adabiyat dar Iran and de Blois; both cohorts are restricted to tazkira-attested poets so each was once canonical; and the test voids if either cohort has fewer than 25 datable poets.

Kill-dataset (verbatim)

Kill: the Ganjoor (ganjoor.net) author roster with per-poet bayt totals, joined to death-dates from Safa's Tarikh-i adabiyat dar Iran and de Blois' Persian Literature: A Bio-Bibliographical Survey; compute the median corpus size and the zero-corpus rate on each side of the 1220 boundary.

Novelty / leakage triage

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

The Mongol watershed as a literary catastrophe is the organizing narrative of the standard histories, and qualitative statements that pre-conquest Khurasani poetry survived poorly are ubiquitous in Browne and Safa. But the cohort design - median Ganjoor corpus and zero-corpus rate for poets dying 1170-1220 versus 1220-1270 - has never been computed anywhere. The direction is moreover genuinely at risk (Nizami, Khaqani and Anvari die inside the earlier window with large transmitted corpora, and Attar's death sits on the boundary), which is precisely what keeps this a live conjecture rather than a leak.

• E.G. Browne, A Literary History of Persia, vol. II (London, 1906), on the Mongol invasion's literary consequences
• Z. Safa, Tarikh-i adabiyat dar Iran, vols II-III, on poetry before and after the catastrophe
• J. Rypka, History of Iranian Literature (Dordrecht, 1968)

Predictions

No prediction registered yet.