Theory stayed home; the machines emigrated

Claim (verbatim)

Hero of Alexandria's Mechanics - the ancient world's systematic treatise on moving heavy loads, gears, screws, cranes and the five simple powers - has no Greek manuscript tradition at all. It survives complete in Arabic, translated by Qusta ibn Luqa in ninth-century Baghdad as Fi raf al-ashya al-thaqila (edited by Carra de Vaux, Journal Asiatique, 1893; re-edited with German translation by Nix and Schmidt in Heronis opera II, Teubner, 1900). Greek kept only what its two anthology-carriers chose to excerpt: Pappus' Collection Book VIII preserves the baroulkos and stretches of the theory of the powers, and the Byzantine technical corpus kept Hero's showpieces - the Pneumatica in over a hundred manuscripts, the Automata, the Belopoeica in the poliorcetic collections. Philo of Byzantium doubles the experiment with the signs reversed: his Belopoeica survives in Greek in those same war-technology corpora, his Pneumatics chiefly through the Arabic tradition (Carra de Vaux, 1902). The mechanism: Greek survival of mechanics ran through a mathematical anthology and a military-court compilation, filters that selected theory, spectacle and siegecraft; the complete working handbook of industrial load-moving had no Byzantine constituency, but exactly fitted the engineering patronage of Abbasid Baghdad - the milieu of the Banu Musa's own book on ingenious devices - which commissioned the whole text. What each civilization kept of the engineer is a portrait of what each asked engineers to do.

Prediction clause (verbatim)

Prediction: aligning Pappus' Collection Book VIII (Hultsch's edition) against the Arabic Mechanics in Nix-Schmidt's chapter divisions, the Greek excerpt channel will cover fewer than one third of the Arabic version's chapters (primary clause: the <1/3 excerpt-coverage fraction; the verdict follows it), and the chapters without Greek parallel will be concentrated in the applied machine-construction and machine-operation material at a rate at least twice that of the theoretical chapters; coverage means substantive content overlap as registered in the Nix-Schmidt concordance and Drachmann's analysis, not topical resemblance, and the test voids if fewer than 80% of the Arabic chapters can be classified as paralleled or unparalleled.

Kill-dataset (verbatim)

Kill: Heronis opera II (Nix-Schmidt 1900) with its Pappus concordance, Hultsch's Pappus Book VIII, Carra de Vaux 1893, and Drachmann's The Mechanical Technology of Greek and Roman Antiquity (1963) - an excerpt-coverage fraction over the Arabic chapter roster plus an applied-versus-theoretical classification.

Novelty / leakage triage

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

That the Mechanics survives only in Qusta ibn Luqa's Arabic while Greek kept the Pappus Book VIII excerpts and the showpiece corpora is stated in Nix-Schmidt and analyzed chapter by chapter in Drachmann, and the anthology-filter explanation is standard; but no located study computes the excerpt channel's coverage as a fraction of the Arabic chapter roster, still less the applied-versus-theoretical concentration ratio. The components are all in print (Nix-Schmidt's Greek parallels, Hultsch's Pappus, Drachmann's analysis); the <1/3 coverage fraction and the 2x genre concentration are an un-run cross-tabulation requiring the substantive-overlap audit the prediction specifies.

• L. Nix & W. Schmidt, Heronis Alexandrini opera quae supersunt omnia II (Leipzig, 1900)
• A.G. Drachmann, The Mechanical Technology of Greek and Roman Antiquity (Copenhagen, 1963)
• F. Hultsch, Pappi Alexandrini Collectionis quae supersunt (Berlin, 1876-1878), Book VIII
• B. Carra de Vaux, Les mecaniques ou l'elevateur de Heron d'Alexandrie, Journal Asiatique (1893)

Predictions

No prediction registered yet.