Every renewal notice is a death notice

Resolution: Inconclusive

Caveats: (a) THE VOID, MECHANICALLY: the registration places the coverage guard first in the clause precedence - INCONCLUSIVE_BY_DESIGN if fewer than 30 dated occurrences match the pinned stems. The compute found n_matched=26 of N_all_dated=13000; 26 < 30, so clause (1) fires and clauses (2) and (3) are never consulted. The headline median lag of 39.75y (short of the 100y threshold) and the post-1300 ratio of 1.43x (short of 1.5x) are diagnostic only and carry no verdict weight. This is the conjecture's own void rule, written verbatim into the registered prediction, firing as designed: not a kill, not a survival. (b) THE COLLISION FINDING, a real methodological result about the pinned stem list: the compute's mechanical lexical QC showed 20 of the 26 hits are substring collisions with unrelated Greek lexemes - 17 are the lemma στάχυς/ἀστάχυς 'wheat-ear' (the Byzantine funerary harvest metaphor: thematically a death notice, lexically nothing to do with στάχωσις 'rebinding'); 1 is προσταχθείς (aorist passive of προστάσσω 'command', a coincidental substring across the prefix/root boundary); and both palai- hits are the dynastic surname Παλαιολόγος/Παλαιολογίνα, which passed the registered book-word proximity rule only because these epigrams' patrons bear the surname and the books they commissioned are new - the opposite of the intended 'old book' sense. Corpus calibration: zero of the 13000 dated rows contain the genuine binding-sense substring σταχω- anywhere. The lesson: σταχ- and palai- are too collision-prone in Byzantine Greek to pin as bare substrings without tighter sense disambiguation. (c) RE-REGISTRATION PATH: a disambiguated matcher - match binding-sense σταχω- forms (στάχωσις, ἐστάχωσα/ἐσταχώθη) while excluding the στάχυς/ἀστάχυ- wheat-ear family and the προστασσ- family; exclude the Παλαιολ- surname from palai-; and handle the Greek augment (ἀνεκαίνισα contracts ana+e- to ane-, so the literal anakain- substring undercounts inflected renewal verbs, a gap the compute itself disclosed). The honest arithmetic must lead: the 6 clean anakain-/neourg- hits are the real floor under any cleanup of this stem list, and 6 is still far below the registered floor of 30 (and even these 6 show register drift: two theological anakain- uses, one neourg- renewing a building rather than a book) - so a re-registration cannot merely disambiguate, it must widen the lexeme net (augmented and contracted forms, further repair vocabulary) or justify a lower pre-committed floor before registering. Disclosed for the future registrant with no verdict weight: the 6 clean hits' diagnostic median lag is -95.75y, opposite in sign to the predicted >= +100y, so at n=6 the clean-subset signal currently points against the conjecture's direction. (d) COMPUTE HONESTY: the firewalled compute reported the below-floor n_matched=26 plainly rather than padding past 30, surfaced the collision QC unprompted, kept its diagnostic recount explicitly separated from and subordinate to the registered result, and disclosed its own matcher's undercount; computed_at 2026-07-18T10:23:07Z postdates the registration batch written 2026-07-18T09:38:25Z.