Phase 22.D Results: The Range Criterion Is Untestable on Available Unseen Mycenaean Data — A Pre-registered TEST-VOID Outcome.

Pre-registered results for Phase 22.D. The range-non-overlap criterion was to be tested on a genuinely unseen hold-out: Linear B sub-corpora from Thebes, Mycenae and Tiryns, harvested from DAMOS only after the pre-registration was published. The harvest yielded 617 tablets and 766 word tokens under the same L2_strict filter used for Knossos and Pylos, but fragmented across 27 series, of which only 3 reached the pre-declared minimum of 40 types. The pre-declared power gate required 5. No word-length class reached the 20-type floor, so the stratified primary test never ran. Verdict: TEST-VOID. With 3 observations the 99% prediction interval lower bound is -43.1 percent, an impossible value for a rate, which is precisely what the power gate exists to catch. The suspension of the Linear A / Linear B morphological gap imposed by Phase 22.B therefore becomes indefinite: the gap is real and is not a word-length artefact, but whether it demonstrates non-Mycenaean-Greekness remains untested. Descriptive figures are reported without verdict force. The only remaining route is the full DAMOS Knossos corpus rather than the def_strict subset. TERMINOLOGY: this project does not claim that Linear A "is not Greek"; the tested and rejected hypothesis is Mycenaean Greek. Pre-registration: doi 10.5281/zenodo.22749112. Version v2 (2026-09-15): OpenTimestamps proofs upgraded to confirmed Bitcoin block attestations. The pre-registration is anchored in blocks 966966, 966969 and 966971; the hold-out corpus and runner in blocks 967020, 967026 and 967042 — 54 blocks (approximately nine hours) later. The blindness of the test is therefore established by the Bitcoin blockchain rather than by declaration. No change to the data, the analysis, or the TEST-VOID verdict.