The nineteenth century fished the classics back up

Claim (verbatim)

The Qing jiyi (輯佚, 'gathering the lost') movement reconstituted vanished books by combing their quotation-fossils out of commentaries and encyclopedias, and its largest single monument is Ma Guohan's Yuhan shanfang jiyi shu (馬國翰 玉函山房輯佚書, mid-19th century), which reassembles several hundred lost works from scattered citations. If reconstruction is possible only where a dead book left dense, attributed quotation, then the reconstructed corpus should be sharply skewed by genre: classics and their xiaoxue (小學, philology) satellites were quoted lemma by lemma in the commentarial tradition and can be rebuilt, while literary collections (集), quoted for ornament if at all, leave too thin a trail to reconstitute. Ma Guohan's table of contents is therefore not a random sample of what was lost but a map of what was densely cited - the reconstructability frontier, not the loss frontier. Prediction restated: the jing (經, classics including philology) division supplies more than half of the reconstructed titles and the ji (集, literary collections) division under one in ten. What the spade did for the tombs, the commentary-index did for the classics: it decided not what died, but what could be pulled back up, and belles-lettres sank because nobody had quoted them closely enough to leave a rope.

Prediction clause (verbatim)

Prediction: classifying the works reconstructed in the Yuhan shanfang jiyi shu (玉函山房輯佚書) by the traditional four divisions, the jing (經, classics including xiaoxue philology) division will account for more than half of all reconstructed titles and the ji (集, literary collections) division for under one in ten (primary clause: jing > 0.5 and ji < 0.1 of reconstructed titles; the verdict follows it); each reconstructed title is assigned to one division and the test voids for coverage if fewer than 400 reconstructed titles can be enumerated.

Kill-dataset (verbatim)

Kill (partly not yet built): the Yuhan shanfang jiyi shu's own zongmu (總目, table of contents), taken from the Kanseki Repository if the collectanea is digitized there, otherwise from the printed Guangling shushe / Shanghai guji reprints, classifying each reconstructed title by traditional division.

Novelty / leakage triage

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

Ma Guohan's Yuhan shanfang jiyi shu is real, its 總目 is countable, and the classics-and-philology skew of the Qing jiyi movement is well understood (Liang Qichao's account of Qing 輯佚 scholarship). The collection is itself organized 經/史/子 with a famously thin belletristic tail, which makes the direction of the prediction almost structural. But the primary clause's specific fractions — jing > 0.5 and ji < 0.1 of reconstructed titles — are a per-division computation not published as such; the 總目 supplies the materials, the ratio is the reader's sum. Adjacent by the conservative test, though it sits close to leaked-by-construction.

• 馬國翰 (Ma Guohan), 玉函山房輯佚書 — its own 總目 (table of contents)
• 梁啟超 (Liang Qichao), 中國近三百年學術史 (on the Qing 輯佚 movement and Ma Guohan)

Predictions

No prediction registered yet.