The Seam, Closed

What AP46 opened. Under the standard equation at the corpus’s partition an age of 13.83 needs a rate of 67.45, and a rate of 74.3 gives an age of 12.55 — two lanes below the cycle. (AP46 §9 quotes 12.52 at the sky’s partition; same relation, one share apart.) The corpus held both a one-cycle age and a rate of 74.3, and AP46 pre-registered that AP48 would derive a non-standard relation reconciling them or one edge of the family would fire: KS-45.1 or KS-STRETCH.3.

How it closed. By adjudication, not reconciliation. The derivation is the standard relation; the age survives; the corpus’s previous rate does not. This paper does not save both of AP46’s numbers.

One rate. The author’s words close the one door a second rate could have come through: records get written at a consistent rate; it cannot speed up; it can write more records at once, which might look like speed and is not; the now’s tug does not pull — it is just now. RULED, and narrowly: there is no structural mechanism in the corpus for a rate read from here and now to sit above the rate written in the record. The uniqueness of the rate itself is the equation’s — one integral, one number — and the describing is not pressed into proving anything about the Cepheids.

The side the structure is on, with its source lines. The relic glow: 67.4 ± 0.5 (Planck 2018 results VI, Table 2, TT,TE,EE+lowE+lensing; with BAO, 67.66 ± 0.42). The Cepheid ladder remeasured with JWST: 73.49 ± 0.93 (Riess et al. 2025, ApJL 992, L34, doi 10.3847/2041-8213/ae0ad6; the 2026 decade review, Cai and Wang, RAA, doi 10.1088/1674-4527/ae842f, beside it). The red-giant ladders in the Chicago–Carnegie programme’s hands (Freedman et al. 2025, ApJ, doi 10.3847/1538-4357/adce78): the programme’s best TRGB-alone value 70.39 ± 1.22 (stat) ± 1.33 (sys) ± 0.70 (σ_SN); from JWST data only, 68.81 ± 1.79 ± 1.32 for the tip of the red giant branch and 67.80 ± 2.17 ± 1.64 for the JAGB method. The same methods in the SH0ES team’s hands do not sit lower — their complete HST and JWST red-giant calibration sample (Li et al. 2026, ApJ 997, 115, doi 10.3847/1538-4357/ae1f17) returns 72.1 to 73.3 ± 1.8, near its Cepheids, as the decade review reports it — so the lower values are one programme’s position, not the method’s. The corpus’s window, 64.4 to 70.8, contains the glow and the central values of the Chicago–Carnegie red-giant ladders and excludes the Cepheid ladder and the SH0ES red-giant values; the TRGB-alone value sits at the window’s upper edge, inside on its central value and straddling it on its total error. So this paper says: the Cepheid ladder’s 73 is not the expansion rate. Either that ladder’s systematics are found, or this paper dies on KS-ASM.1 — and it can die within a few years of JWST samples.

What the side is worth. The side this paper takes is the standard model’s. If the Cepheid ladder is confirmed as the expansion rate, KS-ASM.1 fires and flat ΛCDM at the sky’s partition fails with it; on that edge the corpus gains nothing over the standard model and claims nothing. What is the corpus’s here is not the number but the closure — an age and a partition derived by counting that, under the standard equation, sit on the glow’s side and not the ladder’s. The distinctive prediction the corpus used to carry on H₀, a value above both camps, is gone.

The floor, re-read, with the file open. The corpus’s 74.3 came through AP18. Its §4 derives a₀ = 1.0445 × cH₀/(2π). Its §5 tests it against an empirical acceleration scale of 1.20 ± 0.02 × 10⁻¹⁰ m s⁻², citing McGaugh et al. 2016 and Lelli et al. 2017, lands 0.4% low at H₀ = 74, and calls the match sub-percent; its §8 arms KS-45.1 as H₀ = 74.3 ± 1.2. The ±0.02 is the random error alone. The sources AP18 cites give the scale as 1.20 ± 0.02 (random) ± 0.24 (systematic) × 10⁻¹⁰ m s⁻², the systematic from the stellar mass-to-light ratios and the distances. Inverted, 0.24/1.20 is twenty per cent, so the floor’s rate was 74.3 ± 14.9; it cannot tell 67 from 74. At this paper’s 67.45 the floor gives 1.089 × 10⁻¹⁰, 9.2% below the scale and 0.46σ on the honest error; across the window it stays between 0.24σ and 0.67σ. The coefficient KS-45 tests, 0.1662, stands against the sky’s 0.183 ± 0.037 at this rate, 0.46σ, live. The floor’s relation survives; its registered width did not.

KS-45.1 fires. On the width it registered, KS-45.1’s claim sits 5.7σ from the corpus’s own derived rate, (74.3 − 67.45)/1.2; folding in the corpus’s window, ±3.2, the separation is about two sigma, and the switch fires on its own terms, which are its registered width. AP46 pre-registered that if AP48 derived the standard relation this switch or KS-STRETCH.3 would fire; the age stands; this is the edge. Two dispositions were open. FIRED with erratum: the switch fires as registered, the corpse is shown, and the erratum records that its width omitted the systematic, so that the floor’s relation survives and KS-45’s coefficient test stands live. SUPERSEDED by erratum: a mis-registered width never tested anything and a dated erratum retires it unfired. The desk ruled the first on the author’s delegation, for one reason — the house’s law is that nothing fired is repaired, and re-sizing the width of a live switch after the corpus’s own derivation has contradicted it, then not counting the switch as fired, is a repair — and the author confirmed it in his own word on 3 September: fired. The width correction is right and stands beside the corpse, not instead of it. KS-ASM.1 freezes beside the corpse as the corpus’s H₀ entry.

The edge that fires if this paper is wrong. If the Cepheid ladder is confirmed as the expansion rate, the balance stands and the age at 73 and the corpus’s partition is 12.78 billion years — outside AP46’s window. KS-STRETCH.3 fires, as AP46 pre-registered, and AP46’s chain reverts to a candidate. This paper’s form, counts, and corpse stand either way.