His own list outran his books

Claim (verbatim)

The leaders of the Sokoto jihad were bibliographers of themselves: Usman dan Fodio, his brother Abdullahi, and his son Muhammad Bello each compiled lists of their own writings, and the reference bibliography of the region (Arabic Literature of Africa II, Hunwick and O'Fahey) registers hundreds of titles for the trio. This is the best-documented, most recent, most venerated authorial corpus in premodern West Africa — and even here the self-catalogue outruns the surviving books. The mechanism is a loss floor: if attrition removes a measurable fraction of titles even for authors who listed their own works, were copied for two centuries by a triumphant movement, and are objects of veneration, then the loss rate for ordinary, unlisted, unvenerated West African authors must be far higher — the Fodio corpus calibrates the best case, and the best case is not lossless. Prediction: matching the works self-listed by Usman dan Fodio, Abdullahi dan Fodio, and Muhammad Bello (and registered in ALA II) against manuscripts actually located in the union catalogues, at least fifteen percent of their self-attested titles will have no located manuscript witness — a nonzero loss floor under near-ideal conditions — and the located-witness rate will be markedly higher for the Shehu and Bello than for lesser Sokoto authors of the same generation (primary clause: at least fifteen percent of the trio's self-listed titles are currently unlocated; the verdict follows it). Kill: Arabic Literature of Africa, Volume II: The Writings of Central Sudanic Africa (ed. Hunwick, Brill 1995), cross-matched title by title against the West African Arabic Manuscript Database (AMMS) and the Northwestern Herskovits collections; the loss floor is a bibliographic join and a count of titles with zero located witnesses.

Prediction clause (verbatim)

Prediction: matching the works self-listed by Usman dan Fodio, Abdullahi dan Fodio, and Muhammad Bello (and registered in ALA II) against manuscripts actually located in the union catalogues, at least fifteen percent of their self-attested titles will have no located manuscript witness — a nonzero loss floor under near-ideal conditions — and the located-witness rate will be markedly higher for the Shehu and Bello than for lesser Sokoto authors of the same generation (primary clause: at least fifteen percent of the trio's self-listed titles are currently unlocated; the verdict follows it).

Kill-dataset (verbatim)

Kill: Arabic Literature of Africa, Volume II: The Writings of Central Sudanic Africa (ed. Hunwick, Brill 1995), cross-matched title by title against the West African Arabic Manuscript Database (AMMS) and the Northwestern Herskovits collections; the loss floor is a bibliographic join and a count of titles with zero located witnesses.

Novelty / leakage triage

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

ALA II registers the self-listed corpora of Usman dan Fodio, Abdullahi and Muhammad Bello and notes, per title, located manuscript copies — so it is the raw material for the test. But the primary clause is a measured loss floor: at least fifteen percent of the trio's self-attested titles having zero located manuscript witness, plus a higher located-rate for the Shehu and Bello than for lesser Sokoto authors. That title-by-title join against AMMS and the Herskovits holdings, and the resulting unlocated-share, has not been published as a figure; ALA II lists the titles but no one has counted the zero-witness fraction. The census is enabled, not run.

• John O. Hunwick, Arabic Literature of Africa, Volume II: The Writings of Central Sudanic Africa (Brill, 1995)
• West African Arabic Manuscript Database (AMMS)
• Herskovits Library of African Studies (Northwestern University), Arabic manuscript collections

Predictions

No prediction registered yet.