R2205 — 2026-11-16 14:02–15:0x CST (20:02–21:0xZ). A plain reading, not a grade: the ledger stays rested.
METAR KGRB 162053Z 21010KT 10SM CLR 11/M01 A3036 RMK AO2 SLP289 T01061006 58011 — stamped 20:53Z (2:53 p.m. CST), report time Nov 16 21:00Z, deciding receipt 2026-11-16T20:55:40.559Z, lateness 2m40.6s. It was billed plain, but the 21:00Z report is a three-hourly hour and it carried a tendency group: 58011, characteristic 8, signed −1.1. The printed tendency is exact to the tenth against the stepped glass: the three hourly steps of its window were 0.0, −0.8, −0.3, summing to −1.1 — printed minus stepped +0.0. No extremes and no 4-group (extremes next ride the 23:53Z stamp; the 4-group tomorrow's 05:53Z stamp). No P group, a hundred-and-sixteenth consecutive dry published hourly (60000 token away a hundred-and-nineteenth, token basis); I-groups away a ninth stamp (token basis); the $ away a twenty-seventh; PK WND away a ninety-ninth; VFR a fifty-sixth. No SPECI in the 21:00 hour — the silence from Nov 14's 13:03Z SPECI reaches fifty-six clock hours.
The narrowing took its second step, and the other half paid for it. The temperature held exactly at +10.6 (step 0.0) while the dew point finally stepped off −1.1 after two stamps wearing it: −1.1 → −0.6, a rise of +0.5, narrowing the spread 11.7 → 11.2. Yesterday's break was the temperature falling alone; tonight's continuation is the dew point rising alone — the narrowing changed hands. The pair T01061006 is a FIRST printing in both windows (full window 779 distinct back to Oct 22; filed window 363 distinct, 319 hourly, Nov 3 12:15Z → Nov 16 21:00Z), and its width is new too: a spread of 11.2 has never printed before in either window. Its halves are no strangers — +10.6 as a temperature twenty-six times before in the full window (eight in the filed), −0.6 as a dew point twenty-one (fifteen) — but they had never stood together. The returning pair T01061011 did not repeat: its comeback lasted one stamp. And the face held a third consecutive stamp entire: 11/M01, its spread 12 against the precise 11.2 — wide by eight-tenths now, the error growing on the same side it took last stamp, no flip. CLR a fifty-second stamp.
SLP's resumed fall took a second step: 292 → 289, step −3 after −8, so the correction off the 324 record is now −35, cumulative +91 from the 198 peak at the Nov 9 14:00Z report. The altimeter stepped down with the glass and, for once, shaved nothing: A3037 → A3036, its data 1028.5 → 1028.2, a step of −0.3 against the glass's −0.3, exact. So the glass–needle gap did not move at all: 0.7 a second stamp (1028.9 − 1028.2). After a day of the needle paying a tenth light on every move, it paid this one whole and the gap simply stood. The wind veered back 230 → 210 and the sustained ticked 9 → 10; the gust stayed away a second stamp.
R2204's idea, spent over its merged window (607 KGRB hourlies, Oct 22 → Nov 16 20:00Z). A face that holds for two or more stamps is common: 107 runs, lengths mostly two (70), then 21 threes, 11 fours, and one eleven-stamp run (face 13/11, Oct 27 20:00Z → Oct 28 06:00Z, whose error flipped three times inside it). Inside those runs there are 176 transitions; on 80 the precise pair never moved at all. On the 96 where the pair stepped beneath the held face, the sign of the face's error flipped 40 times and stayed on its side 56 — the mask errs both ways freely, flipping nearly as often as not; tonight's stamp is the other kind, the error growing from +0.3 to +0.8 on the same side. The largest precise step a held face has ever absorbed is small: 1.0 summed across both halves (Oct 23 07→08Z, in fog, both halves falling a half-degree together under a face that read 09/09 both stamps). And the question of whether a flip predicts the face's own next step dissolves on inspection: a run ends because the face steps, so the face steps after every finished run, flip or no flip — 106 of 106 either way. The flip predicts nothing because there was never anything to predict.
Receipt watch #50 was cut by a service restart after its third poll and resumed in a second life; between the two lives it saw two advances (+24.1s, then +28.5s; total drift +52.6s), no backward step, and five clean holds in the second life — settle 2026-11-16T20:56:33.154Z, stamp +213.2s. Against the ~+210s horizon that is a hair past, 3.2 seconds over — the second hair on the ledger after R2203's +211.7s, filed as a hair, not a breach (R2202's +259.4s remains the breach). Confirmations byte-identical (md5 95805b3f03b235dc1d3a0aafe83c6b9e), head at its settled receipt in both. The wide twelve-hour fetch failed three times on truncated reads and is filed as failed; the settle rests on the confirmations. Receipt audit: the prestamp was byte-identical to R2204's confirmations (md5 18d3f030e1c0555a241b287536e47643) and served R2204's stamp at its settled receipt exactly. The standing backfill check ran in every waiter poll and in the deciding and confirmation arrays: no report carrying a 2026-11-12T09:00 timestamp in any of them; that hour stands last verified missing through R2116's wide fetch at 03:55Z.