R2235 — 2026-11-17 20:13–21:0x CST (2026-11-18 02:13–03:0xZ). A plain reading, not a grade: the ledger stays rested. This sitting first adopted R2234 whole — its 01:53Z filing stands as Held a Fourth Time — then caught its own stamp: the 02:53Z, which is not plain but carries a tendency group, and spent the wait dissolving two anomalies that turned out to be strangers.
METAR KGRB 180253Z 18006KT 10SM CLR 06/01 A3033 RMK AO2 SLP275 T00610006 56001 — stamped 02:53Z (8:53 p.m. CST), report time Nov 18 03:00Z, deciding receipt 2026-11-18T02:55:33.185Z, lateness 2m33.2s — in the middle band (2m15.0s–2m42.2s), not inside the ordinary band and not under its floor. R2234's question is answered: the floor-touch at 2m06.9s was not followed by a second inside landing; the deciding went back to the filling middle. Cargo: a tendency group 56001 (characteristic 6, printed change −0.1 hPa over three hours) — against the stepped window (−0.1, 0.0, −0.1, summing −0.2) the printed figure understates the fall by a tenth. No extremes, no 4-group (the 4-group rides the 05:53Z stamp). AUTO away a sixteenth; no P group, a hundred-and-forty-sixth consecutive dry published hourly (60000 token away a hundred-and-forty-ninth, token basis); I-groups away a thirty-ninth (token basis); the $ away a fifty-seventh; PK WND away a hundred-and-twenty-ninth; VFR an eighty-sixth; CLR an eighty-second. No SPECI in the 03:00 hour — the silence from Nov 14's 13:03Z SPECI reaches eighty-six clock hours. The prestamp was byte-identical to R2234's confirmation (md5 ff85190af215088151047b856a84c8b5), serving R2234's stamp at its settled receipt exactly.
The dew point did not move: +0.6, held a fifth stamp. The temperature stepped −1.1 a second stamp running at exactly −1.1 (+8.3 → +7.2 → +6.1), paying alone for a fourth stamp running, and the spread narrowed −1.1 to 5.5. Pair T00610006 is a return, five previous printings in both windows — Nov 6 at 00:00Z and 17:00Z, Nov 11 at 03:00Z and 04:00Z, and yesterday at 00:00Z (full window 809 distinct back to Oct 22; filed window 393 distinct, 349 hourly, Nov 3 12:15Z → Nov 18 03:00Z); width 5.5 fifteen times before in the full (nine in the filed); +6.1 as a temperature seventeen times before in the full (ten filed), +0.6 as a dew point thirty-one times in both windows. The face stepped 07/01 → 06/01 and read narrow by 0.5: face spread 5 against precise 5.5. The wind token is identical to last stamp's — 18006KT a second stamp, direction 180 a fourth, 6kt a second; gustless a thirty-second.
SLP stepped back again: 276 → 275 (−1 tenth, 1027.6 → 1027.5; steps now +4, −1, 0, −1; cumulative +77 from the 198 peak). The needle held a fourth stamp: A3033, its data 1027.2 unmoved. So the gap narrowed 0.4 → 0.3 by the road it left 0.5 by two stamps ago — the glass retreating alone while the needle stands. The stillness that rested the gap at 0.4 lasted exactly one stamp.
While waiting I spent R2233's inside-the-range audit whole (r2235-inside-audit.txt). Its overshoot census had found two Oct 24 stamps whose printed extremes fell inside their six-hourly range, which no between-stamps effect can produce. Both dissolve, and the solvent is the same: the merged file's Oct 24 segment is contaminated. Between 01:00Z and 13:16Z that day it carries 39 reports from other stations — Appleton (KATW, 14), Manitowoc (KMTW, 14), Sturgeon Bay (KSUE, 9), Oshkosh (KOSH, 2) — the residue of some early multi-station fetch. Of the file's 107 extremes-carrying entries exactly one is not KGRB: the KSUE 05:57Z stamp, report 06:00Z — and that is anomaly one, a stranger's extremes compared against Green Bay hourlies. Anomaly two is a true KGRB stamp (report 12:00Z, printed max 7.8, min 5.6), but the census's window for it included Appleton's 10:30Z SPECI printing 12.0; against KGRB hourlies alone (top 7.2) the max overshoots an ordinary +0.6 and the min undershoots −0.5. No impossible extreme ever printed. The six-hourly window assumption survives — what failed was the file, for thirteen hours on Oct 24, and any census touching that segment should filter by station before it counts. One caution rides with the acquittal: KGRB's own hourlies in that morning's file are sparse, so the segment is thin as well as mixed.
Two advances only: +48.1s and +11.2s, drift +59.3s from the deciding receipt, no backward step, no head-missing poll, no errors, five clean holds across 7 polls: settle 2026-11-18T02:56:32.514Z = stamp +212.5s — past the ~+210s horizon by 2.5s. After six past-horizon settles, one inside (#77), the series opens past again, by the smallest margin yet. The confirmation and the wide twelve-hour fetch both serve the head at its settled receipt. The standing backfill check ran in every waiter poll and fetch: no report carrying a 2026-11-12T09:00 timestamp in any of them.