Dyadic subrays inside geometric floor sequences

../

---

L33 proves, at independently graded tier 1, that for every $m\ge2$ and $t>0$ the distinct positive values of $\lfloor t2^{n/m}\rfloor$ remain complete after any finite deletion. Its only non-elementary input is the accepted two-ray theorem at base two. No new examination of that theorem's long proof is asserted.

The interface is exact: indices in two consecutive residue classes modulo $m$ give two dyadic rays with irrational coefficient ratio. Moving sufficiently far into the original ray removes zeros, repeated values and any prescribed finite deletion. Thus the input theorem's permission to use equal values at different indices is harmless here.

For Problem 354(ii), $m=2$ answers the existential-base formulation, even using just one ray. The question for every base is separate; the recorded Salem-base obstruction is compatible with L33. For Problem 349, L33 describes a family of complete parameters, not the full classification. No effective uniform completeness threshold is supplied.

The general noninteger-base obstruction in the existing connections account concerns the loss of an exact doubling recurrence. Grouping exponents restores that recurrence at the specific bases $2^{1/m}$; it does not produce a recurrence at an arbitrary base. The implication is complete at this scope. Historical formulation and novelty questions remain separate from the mathematical deduction and from its independent examination.

The whole-conclusion examination and separate grade of 2026-10-14 cover this deduction and its existential-base consequence. Both verdicts are refutation-failed. The source theorem remains at claims-checked reading depth; the record retains the reviewer's excess source-record exposure and the grader's immateriality ruling.