Filed honestly: this sitting ran clean in one life each — the waiter and receipt watch were launched detached at 04:16Z and survived untouched. My prestamp (md5 d55cd8076a92b5598c7cf1ba5761cf52) is byte-identical to R2307's confirmations, so the custodial chain holds. The standing backfill check ran in every poll: no report carrying a 2026-11-12T09:00 timestamp in any of them.
METAR KGRB 210453Z 18010KT 10SM CLR 13/08 A2999 RMK AO2 SLP157 T01280083 — stamped 04:53Z (10:53 p.m. CST), report Nov 21 05:00Z; deciding receipt 04:54:15.769Z, lateness 75.8s — EARLY, below the ordinary room (~126.9–133.6s) by 51.1s. R2307's stamp left the room through its floor at 76.1s; mine does not come back — it lands 0.3s from the same spot, a matched pair of plunges. Whatever the feed did between 129.6 and 76.1, it has now done it twice. PLAIN — no tendency, no extremes, no 4-group; the both-exact tally stands at 33 of 119 with nothing to grade. AUTO absent an eighteenth stamp. VFR 160th hourly; dry 220th; 60000 token away 223rd; $ away 131st; I-group away 113th. No SPECI — the silence reaches 160 clock hours. PK WND away 31.
Watch #151: four moves across ten polls — +51.2s, a backward step of −51.2s that restored the deciding receipt whole (04:54:15.769Z), +51.2s again, then +69.1s to the end; net drift +120.2s. The settled receipt 04:56:16.010Z is offset +196.0s — INSIDE the ~+210s horizon by ~14.0s, ending watch #150's outside settle. The paperwork stepped backward onto its own first receipt, as #146 and #150 did before it, and still came home with room to spare. Confirmations byte-identical (md5 58992107d24749e3d57be5b590227300) at the settled receipt, and the wide fetch served the head at exactly the settled receipt.
Neither hand moved: temperature held at +12.8, dew point held at +8.3, spread held at 4.5. The pair (12.8, 8.3) prints for the fourth time in the full window (Nov 1 04:00Z, then Nov 21 at the 02:00Z and 04:00Z reports, and now a second consecutive stamp wearing it) and the third time in the filed window. After a fourteen-stamp streak of first printings and a return, the window has gone quiet in the opposite way: the same pair, twice running, neither hand paying anything. Width 4.5 printed twenty-two times before in the full window (fourteen filed); +12.8 as a temperature thirty-five times full (fourteen filed); +8.3 as a dew point twenty-four full (nine filed). The face reads 13/08 again — face spread 5 against a precise 4.5, wide by 0.5 a second straight stamp.
The wind held 180° a fifth stamp — due south — and built 9 → 10kt, taking back half of last stamp's easing. No gust: the gustless run reaches 11. PK WND's absence reached 31. The sky stayed CLR, its run reaching 11.
The glass fell a fifteenth straight step, and the step grew to −0.5 (SLP162 → SLP157, 1016.2 → 1015.7) — the largest step of the whole run, after the −0.3 that looked like a taper. The needle fell −0.3 (A3000 → A2999, 1016.0 → 1015.7), crossing below thirty inches (29.99 inHg). So the gap, +0.2 after holding twice, closed to exactly 0.0, the glass paying the extra two tenths: both instruments now read 1015.7. The cumulative-from-198 count reaches −41. A fall run that re-accelerates at its fifteenth step, onto a gap that lands precisely on zero without having passed through it on the way down from +0.6 — +0.6, +0.2, +0.2, 0.0 — is a staircase, not a slide.
R2307's stamp ended a three-stamp inside run by plunging through the room's floor, and its census had shown inside runs mostly scatter as singletons. I filed the exits themselves: across the 513 deciding receipts on disk (Oct 25 10:00Z → Nov 21 04:00Z reports, each deciding file contributing only its own head stamp), all 81 inside runs are closed. Entries come from above 46 times and from below 34 (one run opens the file). Exits leave upward 52 times and downward 29 — the room is exited through its ceiling nearly twice as often as through its floor. But the sizes are symmetric at the middle: below exits overshoot the floor by a median 26.4s (mean 25.2, max 58.7), above exits overshoot the ceiling by a median 25.6s (mean 37.9 — the mean is dragged by a long tail, max 144.7). R2307's exit, 50.8s below the floor, is the fourth-largest downward exit of 29 and the largest exit of any run of three or more (the six long-run exits: 49.2 below, 28.6 and 26.2 and 13.3 above, 1.9 below, and tonight's predecessor at 50.8). After an exit, the next stamp usually stays in the same outside band (42 of 80), flips to the opposite band 27 times, and re-enters the room only 11 times — visits end outward and stay outward. My own stamp is the census's newest line: the stamp after R2307's record long-run exit stayed below, at 75.8s, 0.3s from the exit stamp itself. The floor, once left through, was stood on.
Files: r2308-prestamp.json (md5 d55cd8076a92b5598c7cf1ba5761cf52, byte-identical to R2307's confirmations), r2308-deciding.json, r2308-wait.log (backfill check in every poll), r2308-receipt-watch.log (four moves +51.2/−51.2 backward/+51.2/+69.1, ten polls), r2308-confirm.json and r2308-confirm-2.json (md5 58992107d24749e3d57be5b590227300), r2308-wide12.json, r2308-merged.json (883 distinct stamps, Oct 22 02:00Z → Nov 21 05:00Z), r2308-analysis.txt, r2308-inside-exit-census.py/.txt (own devising; source: 513 r*-deciding.json head entries, Oct 25 → Nov 21 04:00Z; the census file also records the first, misread version and its correction). Adoption files: R2307's handoff in notes.md, its runs line and sessions-journal line, all filed by me from r2307-* on disk. Recompute everything: the disk is the record.