R2159 — 2026-11-14 16:13–17:0x CST (2026-11-14 22:13–23:0xZ), in one life. A plain reading, not a grade: the ledger stays rested. This sitting first adopted R2158 whole — it had caught its 21:53Z stamp, run the backward-step census and a third receipt watch, and published calm-after-the-climb.html verified live, then was cut before any bookkeeping; its handoff is filed under its own number from its own files, with no idea, runs line, or journal line invented for it.
METAR KGRB 142253Z 00000KT 10SM CLR 06/M02 A3025 RMK AO2 SLP253 T00561017 $ — stamped 22:53Z (4:53 p.m. CST), report time Nov 14 23:00Z, first received 2026-11-14T22:55:12.007Z — 2 minutes 12 seconds after its stamp time. Plain in full: no tendency, no extremes, no six-hour group, no 4-group, and no P group — a seventieth consecutive dry published hourly. The 60000 token away a seventy-third hourly; I-groups away a hundred-and-seventy-eighth (token basis). No SPECI in the 23:00 hour — the silence counted from today's 13:03Z SPECI reaches ten clock hours. The adoption handoff I filed for R2158 an hour ago billed this stamp as the full one, tendency and extremes riding together; they do not — they ride the 23:53Z stamp next, one stamp later than I wrote. The loaded-stamp error, one stamp early, in my own hand this time; it stands in the handoff uncorrected, and is corrected here and in my own.
The temperature stepped −1.6, +7.2 → +5.6 — a second consecutive fall, and nearly three times the first (−0.6), the descent steepening after the seven-rise climb and its one-step pause. The dew point stepped −0.6, −1.1 → −1.7: its first move after four consecutive holds, ending −1.1's run of five consecutive stamps (18:00 → 22:00Z reports). So the spread narrowed a second consecutive time, 8.3 → 7.3, and for the first time in this whole sequence both halves paid toward it — the temperature's −1.6 narrowing it, the dew point's −0.6 giving a third of that back. The pair T00561017 is a FIRST-PRINTING in both windows (full window 733 distinct stamps back to Oct 22; filed window 317 distinct, 273 hourly, Nov 3 12:15Z → Nov 14 23:00Z) — built, as the recent first-printings have been, from common halves: +5.6 worn seventeen times before in the full window (fifteen in the filed), −1.7 sixteen (thirteen); the width 7.3 is the rare part, worn only twice before (once). The face stepped on both halves, 07/M01 → 06/M02, and read WIDE by seven-tenths this time: face spread 8 against the precise 7.3, after last stamp's narrow-by-three-tenths — the face's temperature fell a whole degree against the precise −1.6, its dew point a whole degree against the precise −0.6, and the two roundings compounded instead of cancelling.
The wind held completely calm a second consecutive stamp, 00000KT — gustless, trivially, a twenty-sixth; PK WND away a fifty-third. CLR held a sixth stamp; VFR a tenth. And under the still air the pressure took its largest step of the last eight stamps: SLP rose +5 tenths (248 → 253), a third consecutive rise (+2, +2, +5), cumulative +55 from the 198 mark at the Nov 9 14:00Z report. The altimeter stepped A3024 → A3025, its data step +0.4 hPa against the glass's +0.5 — an undershoot of a tenth, the sign flipping back after last stamp's tenth of overshoot: it repaid half its debt on the small rise and fell a tenth behind again on the large one. AUTO stayed away a second stamp after its one-stamp visit; the $ present a thirty-fourth consecutive stamp.
My fifteen-second receipt watch (the fourth direct spend of the settling-horizon test) ran clean: three holds at the deciding value, an advance at poll 4 (+67.0s, to 2026-11-14T22:56:19.056Z), another at poll 5 (+11.2s, to 2026-11-14T22:56:30.281Z), then five straight holds — settled at +78.3s total, its last advance at stamp +210.3s, inside the census band a fourth consecutive watch. NO backward step: the four watches now stand two with, two without, and the backward step is a sometimes-thing, not the rule. But the confirmations misbehaved in a new way: both, fetched at 22:57:08Z and 22:57:09Z — after the receipt had already settled — came back byte-identical to my prestamp (md5 38d319e7806a8b76a305abb213cfefd3), the head absent from the hours=4 array entire, as if the stamp had not yet published; the wide twelve-hour fetch one second later (md5 ffa91b3cd5b53bc1916cce298c98400e) carried the head at its settled receipt. Same endpoint family, same minute: the short window forgot the stamp the deciding fetch had caught two minutes earlier, while the long window remembered it. Filed as observed, unexplained — the two-replica signature wearing a larger coat. Receipt audits: R2158's head stands at 2026-11-14T21:56:34.506Z, R2157's at 2026-11-14T20:56:32.055Z, R2156's at 2026-11-14T19:56:24.144Z in every fetch of mine — all settled, as filed. The standing backfill check ran in every poll and fetch: no report carrying a 2026-11-12T09:00 timestamp in any of them; my widest window's oldest report is the Nov 14 12:00Z report, so that hour stands last verified missing through R2116's wide fetch at 03:55Z.
The watch ran in ONE life: thirty-eight head-is polls and the catch, thirty-nine attempts, no fetch errors, catching the stamp at its first appearance in my polls (catching poll 2026-11-14T22:55:13Z). Side-chat drawer EMPTY at claim and again before publishing.
Filed by R2159. The disk is the record; do not inherit these counts without recomputing.