He wrote longest over the graves

Claim (verbatim)

Photios' Bibliotheca (the Myriobiblon), composed for his brother Tarasios before the embassy to the 'Assyrians' (conventionally c. 845; Warren Treadgold, The Nature of the Bibliotheca of Photius, 1980), reviews 280 numbered codices covering, on Treadgold's count, some 386 works. The reviews are wildly unequal: a few lines for Theodore the Presbyter (cod. 1), page after page of epitome for Ctesias' Persica (cod. 72), Memnon of Herakleia (cod. 224), Agatharchides (cod. 250), the novels of Iamblichus (cod. 94) and Antonius Diogenes (cod. 166), Conon's Narratives (cod. 186), Ptolemy Hephaestion's New History (cod. 190). The famous arithmetic - roughly half of what he reviewed is lost - is a blunt average that hides the instrument's design. Photios says he is summarizing books his brother will be away from; he epitomizes at length precisely what is rare, hard to re-access, off the school circuit, and glances briefly at what any reader could find. Review length is therefore a ninth-century access forecast: the patriarch invested words where he sensed scarcity. If so, the Bibliotheca's longest reviews should sit over the works that subsequently died - he wrote longest over the graves, and each notice's length is a prediction of loss that we can now grade. The long-review dead are already suggestive: Ctesias, Memnon, Agatharchides, Olympiodorus of Thebes (cod. 80), Candidus (cod. 79), Philostorgius (cod. 40, alive only as Photios' own epitome); Theophylact Simocatta (cod. 65) is the long-review survivor that keeps the test honest.

Prediction clause (verbatim)

Prediction: ranking the works reviewed in the Bibliotheca by review length (TLG character count of Rene Henry's text per codex, divided evenly within multi-work codices), works in the top length quartile will show an independent-survival rate - a substantially complete text transmitted outside the Bibliotheca, per the TLG canon and Pinakes - at most half that of works in the bottom quartile (primary clause: the <=0.5 top-to-bottom-quartile survival ratio; the verdict follows it); secondary clause: the gap persists within secular prose alone, excluding the theological codices; coverage guard: void if per-work review spans cannot be delimited for at least 80% of the works. Confidence: direction argued from the instrument's stated purpose; the exact ratio MODERATE.

Kill-dataset (verbatim)

Kill: the TLG full text of Photios' Bibliotheca (Henry's Bude edition, 8 vols) for per-codex review lengths, joined to the TLG canon and Pinakes witness records for independent-survival flags - a quartile survival-rate comparison.

Novelty / leakage triage

already answered in the literature

Treadgold's monograph already runs this join: it tabulates the Photian codices by review length, genre and survival, states the design fact the claim argues - full-dress epitomes for the rare, off-circuit books, brief notices for the accessible - and its statistics are the source of the standard half-lost arithmetic, with the long-summary class dominated by the dead (Ctesias, Memnon, Agatharchides, the novelists) and Theophylact the acknowledged survivor. Re-deriving the gradient as a TLG-character quartile ratio changes the units, not the answer, and Hagg's study of Photios' excerpting technique independently documents the length-of-treatment gradient.

• W. Treadgold, The Nature of the Bibliotheca of Photius (Dumbarton Oaks Studies 18, Washington, DC, 1980), tables and catalogue of the works reviewed
• T. Hagg, Photios als Vermittler antiker Literatur: Untersuchungen zur Technik des Referierens und Exzerpierens in der Bibliotheke (Uppsala, 1975)
• N.G. Wilson, Scholars of Byzantium (London, 1983), chapter on Photios

Predictions

No prediction registered yet.