The lexicon quotes the morgue, not the library

Claim (verbatim)

The Suda's most celebrated service - preserving hundreds of quotations from the lost historians - is second-hand. Carl de Boor demonstrated ('Suidas und die konstantinische Exzerptsammlung', Byzantinische Zeitschrift 21, 1912, and 23, 1914-19) that where the Suda quotes the classicizing historians, its excerpts coincide in extent, incipit and wording with sections of the Excerpta Constantiniana: the lexicographers, working in Constantinople within a generation of Constantine VII's death, were already mining the palace digest rather than the originals. That is a devastating terminus for the full texts: by c. 980 Priscus, Malchus, Menander Protector and their kind were effectively out of reach even for the best-connected compilation project of the age - the originals had dropped from circulation within decades of being excerpted, and the Suda's reservoir of lost historiography is a copy of a copy, its water drawn from the morgue and not the library. De Boor's demonstration was philological and selective; the claim here is that it generalizes into a stemmatic census with a hard rate, decidable passage by passage now that both sides are printed: every Suda quotation that overlaps a surviving Excerpta section either sides with the digest or witnesses the original, and the tally is the measure of how fast an empire lost books its own grandfathers had excerpted.

Prediction clause (verbatim)

Prediction: among Suda quotations attributed to the fragmentary historians carried by the surviving Excerpta collections (author set pinned: Priscus, Malchus, Menander Protector, Eunapius, Peter the Patrician, Nicolaus of Damascus, Dexippus, John of Antioch), of the passages where a surviving Excerpta section transmits the same content, at least 80% will agree with the Excerpta's text and excerpt boundaries against any independently transmitted alternative, placing the Suda stemmatically downstream of the Excerpta (primary clause: the >=0.8 downstream-agreement share; the verdict follows it); disambiguation: passage identity and attribution follow the critical apparatus of the Berlin editions and of Adler; coverage guard: void if fewer than 40 overlapping passages can be aligned.

Kill-dataset (verbatim)

Kill: Adler's Suda text (via Suda On Line) collated against the Berlin corpus of the Excerpta (de Boor's Excerpta de legationibus, 1903, and De insidiis, 1905; Buettner-Wobst and Roos, De virtutibus et vitiis, 1906-1910; Boissevain, De sententiis, 1906) - passage-level collation counting stemmatic agreement.

Novelty / leakage triage

already answered in the literature

The generalization claimed as new is what de Boor printed: his two articles demonstrate passage by passage, across the classicizing historians, that the Suda's quotations coincide with the Excerpta's sections in extent, incipit and wording, and Adler's prolegomena and apparatus register the parallels entry by entry; the >=0.8 downstream-agreement share re-states that finding as a rate, and Nemeth's monograph confirms the dependence as the standing consensus. A fresh collation would re-derive, not discover.

• C. de Boor, 'Suidas und die konstantinische Exzerptsammlung', Byzantinische Zeitschrift 21 (1912) and 23 (1914-19)
• A. Adler (ed.), Suidae Lexicon, 5 vols (Leipzig, 1928-1938), prolegomena and apparatus on the Excerpta parallels
• A. Nemeth, The Excerpta Constantiniana and the Byzantine Appropriation of the Past (Cambridge, 2018)

Predictions

No prediction registered yet.