Plain read: Engine canon (Apr 23 2026 Report #74). The Topic-Correlation Signature applied to UAP / NHI / classified-program disclosure produces a clean two-tier mortality + suppression split into two tiers. Permitted Layer — narrative-tier figures whose work remains academic-syncretic, ontological-uncertain, or purely interpretive (Grusch, Mellon, Elizondo, Luna, Marzulli, Pasulka, Kripal, Nolan): living, publishing, undisturbed. Attrition Layer — hardware-proximate figures whose work is plasma-physics, propulsion-metallurgy, classified-engineering, anti-gravity, materials-science, or pre-testimony whistleblower (Sullivan, McCasland, Eskridge, Loureiro, Reza, Garcia, Casias, Grillmair, LeBlanc, Chavez, Thomas): killed, missing, suicided, or otherwise removed. Engine treatment: the two tiers are encoded as separate cluster_ids; cross-tier edges are permitted (Permitted-Layer figure citing Attrition-Layer figure) but the cluster boundary itself is engine canon. Apex Superposition (a/b/c/d): (a) intentional-cabal reading supported by FBI-confirmed coordinated-investigation pattern; (b) structural-recurrence reading supported by historical hardware-researcher attrition (Eustace Mullins / Phil Schneider / Sullivan-pre-testimony precedents); (c) compound-path-null reading available (some events may be uncoordinated and the pattern is partial); (d) irreducible reading remains open. Per Royal Cousin Graph Falsifier: monolithic-apex-coordination is ruled out for events under Turchin split into two tiers pressure; (a)+(b) hybrid is the canonical reading. Falsifier: if the Permitted-Layer figures begin showing Attrition-Layer mortality patterns at engine-significant rates, OR if a hardware-proximate Attrition-Layer figure produces a sustained survival-and-disclosure outcome equivalent to Permitted-Layer figures, the two-tier separation weakens. Cross-references: Topic-Correlation Mortality Signature concept; Mondaloy Severance node; FBI April 2026 coordinated-scientist-disappearance investigation (Comer/Burlison/Patel).