The whole chronology hangs by one letter

Claim (verbatim)

Almost everything the tradition claims to know about how its own foundational books were made - when the Mishnah was redacted, how the Talmud reached closure, the order and dates of the sages and the geonic succession - descends from a single document: the Iggeret Rav Sherira Gaon, a responsum written in 986/987 to the community of Kairouan in answer to exactly these questions. There is essentially no independent narrative source for the redaction-history of rabbinic literature; the field stands on one late-tenth-century letter. Worse for its authority, that letter itself survives in two divergent recensions - conventionally the "French" and the "Spanish" - which disagree at points that matter, including the crucial question of whether the Mishnah and Talmud were committed to writing or transmitted orally. Mechanism: when a whole historiography rests on one witness, every uncertainty in that witness propagates undamped into the entire edifice, and a text preserved in disagreeing recensions has uncertainty built in. The measurable form is the divergence between the recensions and the reach of their sole-source dependency. Prediction restated: the two recensions of the Iggeret differ substantially - a high variant density concentrated at the historically load-bearing passages - and the standard chronological "facts" of rabbinic literary history trace overwhelmingly back to this one text with no independent corroboration.

Prediction clause (verbatim)

Prediction: aligning the French and Spanish recensions of the Iggeret Rav Sherira Gaon passage by passage, the character-level variant rate will exceed that found among the manuscript witnesses of a single-recension geonic text of comparable length by at least a factor of three, with the divergence concentrated at the load-bearing chronological passages (the redaction of the Mishnah and the closing of the Talmud) - primary clause: the >= 3x cross-recension variant rate; the verdict follows it. The test voids for coverage if the two recensions cannot be aligned passage-for-passage.

Kill-dataset (verbatim)

Kill (not yet built): B. M. Lewin's critical edition of the Iggeret Rav Sherira Gaon (1921), which prints the French and Spanish recensions in parallel - align them passage by passage for a character-level variant count, flagging the chronologically load-bearing sections; runnable once Lewin's parallel text is digitized, with Sefaria supplying the downstream check on which standard chronological claims cite the Iggeret as sole source.

Novelty / leakage triage

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

Lewin's 1921 edition prints the French and Spanish recensions in parallel, and the divergence - including the load-bearing disagreement over whether Mishnah and Talmud were committed to writing - is a famous stated datum, discussed by Brody together with the sole-source dependency of geonic-era and redaction-history scholarship on the Iggeret, and at depth in Sussmann's oral-torah study. What no one has done is the quantification: a character-level variant-rate across the aligned recensions, a matched single-recension geonic control, and the >= 3x comparison are digital-philology operations with no printed counterpart, and the outcome depends materially on the control chosen. Adjacent, with Lewin's parallel columns as the ready-made instrument bed.

• B. M. Lewin (ed.), Iggeret Rav Sherira Gaon (Haifa, 1921), parallel French and Spanish recensions
• R. Brody, The Geonim of Babylonia and the Shaping of Medieval Jewish Culture (Yale UP, 1998), on the Iggeret, its recensions and its historiographic weight
• Y. Sussmann, 'Torah she-be-al peh - peshutah ke-mashma'ah', in Mehqerei Talmud III (Jerusalem: Magnes, 2005)

Predictions

No prediction registered yet.