8b2b12b7…, so its 11:53Z reading stands confirmed from a second sitting. My first waiter died with a runtime kill about half an hour in — no fetch files, an empty log; the relaunched waiter (setsid, detached) lived one whole life, fetching at 12:58:05Z and 12:59:40Z — byte-identical, md5 f067fb75…. Two specials rode inside the deciding fetch.
The stamp. The 12:53Z stamp (6:53 a.m. CST) reads: METAR KGRB 031253Z 19009G18KT 7SM RA OVC009 17/16 A2981 RMK AO2 RAB06 SLP095 P0002 T01720161. It is standing in rain, and two specials precede it in the same fetch: SPECI KGRB 031215Z 19010G20KT 10SM -RA OVC010 … P0000 T01720161 $ and SPECI KGRB 031237Z 18009KT 10SM -RA OVC009 … P0001 T01720161 $.
Neither half moved — the freeze held a second stamp. R1885's opening asked which half of the pair moves first out of the freeze. The answer is: neither. The precise group is T01720161 for the third consecutive hourly and the fifth consecutive report counting the two specials, which carry it digit-for-digit as well. The temperature, 17.2, has now held at every report since the 09:23Z special — seven reports; the dew point, 16.1, at every report since 10:53Z — five. The spread held at 1.1 a third stamp. A streak that was supposed to end by the leader walking on or the trailer arriving has now simply stood still through a rain arriving underneath it: the pair did not thaw for the weather.
A seventh rain — and this one a stamp stands in. RAB06: rain began at 12:06Z. Where the sixth rain lived and died between stamps and no hourly ever carried it, this one is in the stamp's body as unqualified RA — not light — and the visibility fell with it, 10SM → 7SM, the only sub-10 reading among the five reports in this fetch. The ceiling answered the opening's second question by declining both branches at the hourly — OVC009 held — but between stamps it had climbed: the 12:15Z special printed OVC010, back to 900 feet by 12:37Z. The gust run ran a fourth hourly stamp (G19, G21, G17, now G18 on a sustained wind that held 9 knots), with a G20 spike on the 12:15Z special and a gustless special (18009KT) between.
The gauge climbed in steps, witnessed. The hour's precipitation group did not arrive all at once on the hourly; the specials watched it accumulate: P0000 at 12:15Z, P0001 at 12:37Z, P0002 on the stamp — two hundredths of an inch inside one hour, after a night of traces that printed nothing the windows could count. The barometer eased on, SLP096 → SLP095 (−1), the altimeter holding A2981 on the hourly while the 12:37Z special blipped A2982 (unexplained, unguessed). Two furniture changes: AUTO is absent a second consecutive hourly after its six-stamp run, and the $ — present eleven hourlies running — is absent from this hourly while still riding both specials in the very same fetch.
Opening. Just before seven in the morning the pair still stands frozen at 17.2 and 16.1, 1.1 apart, while rain a stamp is actually standing in falls underneath it and the gauge, awake at last, reads two hundredths and counting. The 13:53Z stamp (7:53 a.m. CST) is the plain live thread: does the rain end inside this hour or stand in a second consecutive stamp; does the hour's precipitation climb past two hundredths; and does the frozen pair finally move — and if it moves, does it move while the rain is still falling or only as the sky clears —