The geonim survive as footnotes to their heirs

Claim (verbatim)

For three centuries the geonim of the Babylonian academies were the supreme halakhic authority of the Jewish world, answering thousands of questions and composing the first post-talmudic books. Yet almost none of their writing survives as their own books: a handful of works (Seder Rav Amram, the She'iltot, Halakhot Gedolot, Halakhot Pesukot, Saadia's siddur and his Tafsir) came down more or less whole; the rest of the vast geonic output - responsa, monographs, commentaries - survives only as quotation, embedded in the rishonim who cited them and, later, gathered by modern editors from those citations and from the Geniza (Lewin's Otzar ha-Geonim being a compilation, not a recovered book). Mechanism: the geonic word travelled as answer and as citation, not as circulating volume, so when the academies declined the books were not there to copy - only the quotations were, held alive inside the Spanish and Franco-German codes that superseded them. The measurable form of this is a gross asymmetry: the named geonim are cited by the later halakhists constantly while their complete original works number in the single digits. Prediction restated: across Sefaria's open rishonim corpus, attributed citations of the major named geonim run into the thousands, against a complete-original-work survival in the low single digits - a citation-to-book ratio well past a hundred to one.

Prediction clause (verbatim)

Prediction: counting attributed citations of the major named geonim (Rav Hai Gaon, Rav Sherira Gaon, Rav Natronai Gaon, Rav Amram Gaon, Rav Saadia Gaon, and "the Geonim" as a body) across Sefaria's open rishonim corpus (Rif, Rosh, Ramban, Rashba, Ritva, Ran, Tur, Beit Yosef and kin), the total citation-instance count will exceed the number of complete, independently-transmitted geonic books by more than one hundred to one - primary clause: the > 100:1 citation-instances-to-complete-books ratio; the verdict follows it. A complete book is counted only on independent whole-text transmission, not a modern citation-compilation such as Otzar ha-Geonim; the test voids for coverage if fewer than 500 geonic citations are recovered.

Kill-dataset (verbatim)

Kill: Sefaria's open rishonim and halakhic corpora (with the Bar-Ilan Responsa Project as the deeper instrument) - string-match the named-geon attribution formulae, count citation instances, and set the total against the enumerated list of completely-transmitted geonic works, with Lewin's Otzar ha-Geonim marked as compilation rather than surviving book.

Novelty / leakage triage

already answered in the literature

Both sides of the ratio are already in the published record. Brody's survey states the survival profile - the geonic output reaches us overwhelmingly as quotation in the rishonim and as Geniza fragments, with the independently-transmitted books an enumerated short list - and the citation masses are themselves compiled in print: Lewin's Otzar ha-Geonim fills thirteen volumes with thousands of geonic rulings recovered largely from the rishonim's citations, with Assaf's survey and Ginzberg's Geonica documenting the same asymmetry. With a printed denominator on the order of ten and printed citation-compilations on the order of thousands, the > 100:1 clause is guaranteed arithmetic on published compilations; a Sefaria recount can only re-instantiate an answer the reference shelf already contains.

• R. Brody, The Geonim of Babylonia and the Shaping of Medieval Jewish Culture (New Haven: Yale UP, 1998), part II on the geonic literary corpus and its transmission
• B. M. Lewin, Otzar ha-Geonim, 13 vols. (Haifa/Jerusalem, 1928-1943)
• S. Assaf, Tekufat ha-Geonim ve-Sifrutah (Jerusalem, 1955)
• L. Ginzberg, Geonica, 2 vols. (New York: JTS, 1909)

Predictions

No prediction registered yet.