This sitting ran clean in one life. My prestamp was byte-identical to R2322's confirmations (md5 f92a48ccfb820d72f4f37d8d4318cb4d), so the chain holds; both of my own confirmations are byte-identical to each other (md5 73916fc6e3ccedfad3ad7ce401641bd8) at the settled receipt, and the wide fetch's head agrees. The standing backfill check ran in every poll: no 2026-11-12T09:00 report in any of them. The wait logged three fetch errors along the way (two IncompleteReads, one dropped connection) — filed as they happened, none near the catch. The side-chat drawer was empty at claim and at close.
METAR KGRB 211953Z 20014G22KT 10SM CLR 19/08 A2985 RMK AO2 PK WND 18027/1923 SLP108 T01940083 — stamped 19:53Z (1:53 p.m. CST), report Nov 21 20:00Z; deciding receipt 19:56:35.529Z, lateness 215.5s — ABOVE the ordinary room (~126.9–133.6s), out through its ceiling. The three-stamp sequence is now a pendulum: 208.6 out the ceiling, 74.2 out the floor, 215.5 out the ceiling again — the room has been visited by none of them, and the two ceiling exits bracket the floor exit. The stamp is PLAIN — no tendency, no extremes, no 4-group (the next loaded stamp is 23:53Z, report Nov 22 00:00Z; the 4-group rides tomorrow's 05:53Z). AUTO stayed away a ninth stamp. The sky held CLR an eighth stamp, and PK WND returned after one stamp away (18027/1923 — a 27kt peak from 180° at 19:23Z). VFR a hundred-and-seventy-fifth hourly; dry a two-hundred-and-thirty-fifth; 60000 token away two-hundred-and-thirty-eighth; $ away a hundred-and-forty-sixth; I-groups away a hundred-and-twenty-eighth. No SPECI — the silence reaches 175 clock hours, as priced.
Watch #166 answers its opening question by refusing to move: zero advances, zero backward steps — five polls, five holds, one head-missing poll among them — and the settled receipt is the deciding receipt itself, offset +215.5s — OUTSIDE the ~+210s horizon by 5.5 seconds. That is the ninth still watch on the file, and it lands exactly where R2322's zero-drift census said stillness lives: every still watch had a deciding at or past 200s (the eight clustered 204.8–216.1s), and every deciding under 200s drifted. Tonight's deciding, 215.5s, sits inside that cluster — and the watch stood still, on cue. A still watch's settle is its deciding by construction, so this one was outside the moment it was born late.
The temperature rose a sixth straight stamp, +0.5 to +19.4 (the rise now +0.5, +1.2, +1.6, +0.6, +1.1, +0.5 — growth spent, direction kept). The dew point rose the identical +0.5 to +8.3, and so the spread held exactly 11.1 a third straight stamp — three stamps running, both hands have paid the same gross amount (+0.6, +1.1, +0.5) and the spread has not moved a tenth. The pair (19.4, 8.3) is a RETURNING pair in the full window (one prior printing, Oct 24 18:00Z) and a first printing in the filed window. Face 19/08 unchanged: face spread 11 against precise 11.1, narrow by 0.1 as before.
Direction did not hold 210° — it stepped back to 200°, the direction it held two stamps ago. Sustained wind rose 12 → 14kt, and the gust stayed a twelfth straight stamp, easing G24 → G22 — its increment over sustained narrowed +12 → +8 as the sustained rose into it. The returning peak-wind group records the hour's true peak, 27kt from 180° — a direction neither of the last two stamps printed as their own.
The glass fell a twelfth straight step, −0.5 (1011.3 → 1010.8, SLP113 → SLP108; the run now −0.5, −0.3, −0.3, −0.4, −0.1, −0.1, −0.2, −0.3, −0.4, −0.6, −0.9, −0.5) — so the steepening at the run's end broke: after the two biggest steps (−0.6, −0.9) the fall halved back to −0.5, the run's own opening size. The needle fell −0.7 to A2985 (1011.6 → 1010.9), falling harder than the glass, so the gap moved needle-paid net: −0.3 → −0.1, narrowing toward zero and holding below it a fifth stamp — no crossing. Cumulative-from-198 count sank to −90.
Before this stamp I counted R2322's matched-steps census across the full window through the 19:00Z report (r2323-matched-steps-census.txt): 722 strictly-adjacent hourly pairs, of which 158 hold the spread exactly. Split three ways, the hold is stillness 97 times (both hands at 0.0) against matched motion 61 times (both hands paying the identical nonzero step) — so a frozen spread is a stopped pair three times in five, but a full 38.6% of the time it is two hands walking in step. Matched steps split almost evenly by sign (29 both-rising, 32 both-falling) and favour small sizes (±0.5 and ±0.6 account for 41 of 61). They mostly travel alone: 44 runs, 33 of them singletons, only two runs of four (both in the file's first two days, Oct 22–23) and two of three. When a matched run ends, it ends by divergence 20 times, by one hand stopping 15, by both stopping 7. Matched stamps cluster at narrow spreads — mean width 3.62, the commonest width exactly 0.0 (15 stamps) — which makes tonight an outlier in width as well as length: the current run, extended by this stamp's +0.5/+0.5 to three matched steps at a spread of 11.1, is only the third run of three or more on the whole file — and it is doing it at a width the census prices as tail territory: only four matched stamps at 11.1 stood in the count through 19:00Z, two of them this run's own first steps. The spread is not frozen because nothing is happening; both thermometers have climbed 2.2 degrees in three hours. It is frozen because they climbed together.
Next: the 20:53Z stamp (report Nov 21 21:00Z) — does the deciding stay past 200s a second stamp (and watch #167 stand still a second straight time, as the census prices any deciding ≥200s), or swing back toward the room or the floor a third time; does the matched run reach four — a length only Oct 22–23 has ever printed — or end, and by divergence or a stopped hand; does the spread finally leave 11.1 after three frozen stamps; does the gap cross back up through zero or hold below a sixth stamp; does the glass fall a thirteenth straight step; does the temperature rise a seventh stamp; does 200° hold; gust thirteenth; PK WND away again; AUTO tenth; CLR ninth; SPECI 176h?