Overview
W1 version 2.0.1 withdraws its former proof that the Q5a admissibility weights have a common positive limit [H-w]. The old argument used a relative profile estimate never established by U1 and identified Q5a's fingerprint-response averages with sums of shell capacities without a bridge. U1 version 2.0 also withdrew its former absolute estimate.
The finite five-block fitting window is bounded by depth 22 for sufficiently large primes. Thus the former infinite-sum target and tail argument do not follow from that window. This does not disprove [H-w]; the actual Q5a weights still require a separate limit argument.
What the retained results say
Theorem 4 bounds a separately defined partial-sum proxy, conditional on a supplied relative profile estimate and a generator-to-character map. Lemma 6 gives the depth-22 target for a supplied profile if the specified calibrated cutoff converges to 22. Corollary 7 states the additional proxy-to-response estimate needed to infer a common limit of Q5a's actual weights. None of these missing premises is established by O25, Q5a or U1.
O25 reports finite-prime exponent fits, including a historical pair value of 7.44 and a raw global value of 7.61 at q = 601; it does not determine an asymptotic exponent. Those fits do not supply a global summability bound for a limiting profile.
Relation to the Cosmochrony programme
The programme registry version 2.23 records the withdrawn proof and keeps both Q5a weight hypotheses open. Q10 version 2.0.1 derives no coefficient value, and Q11 version 2.0, which no longer uses [H-w], shows that spatial $\mathrm{SU}(2)$ invariance leaves the temporal coefficient free; W1 supplies neither a missing weight bridge nor a universal coefficient.
Reference
Jérôme Beau. What Weight Stabilisation Would Require: A Finite-Window Audit of Admissibility Weights, version 2.0.1, 2026. doi:10.5281/zenodo.19886319