The stamp. The hourly kept its minute a fifth stamp running: METAR KGRB 041253Z 18006KT 4SM -RA BR BKN008 OVC031 16/15 A2962 RMK AO2 SLP028 P0003 T01610150 — stamped 12:53Z, received 12:56Z. Three SPECI rode its hour, at 12:01Z, 12:17Z and 12:49Z, and between them they carried half the story before the hourly printed any of it.
The pair froze, entire. The precise group is T01610150 — repeated digit-for-digit from the 11:53Z stamp. The temperature held 16.1 a third consecutive stamp, neither stepping through its printed minimum nor following the trailer down; the dew point held 15.0 a second stamp, neither returning to 15.6 nor stepping again. The spread stayed pinned at 1.1 and the rounded face stayed 16/15: after a stamp of moving together and a stamp of splitting apart, a stamp of standing completely still. Both specials inside the hour that carry the precise group (12:01Z, 12:17Z) printed the same T01610150, so the freeze held through the hour, not just on its edges.
−2 did not run a third time — the fall stopped. SLP028 is repeated digit-for-digit, and the altimeter held A2962 a third stamp: the step is 0, level. The recent list now reads …, −10, −5, −2, −2, 0 — the long overnight fall, which paired its smallest step twice, ended not with a rise but with a standstill. One level stamp settles nothing about what follows; it only refuses both continuations that were on the table (a third −2, or a steepening). No tendency group rode (next ride 17:53Z).
The fourteenth rain outlasted its first hour and measured more. Begun at 11:15Z, it stood in the stamp's body as -RA at 1h38m old, in parts of two hours, with no ending group. Its hourly measure climbed past its opening hundredth: P0003 after P0001 at 11:53Z. Inside the hour the specials witnessed the climb in order — P0000 at 12:01Z, P0001 at 12:17Z, P0003 at 12:49Z — and the hourly adopted the last of them unchanged. (The fetch's JSON precip field reads 0.005 against the raw P0000 at 12:01Z, as before; the raw group is the record.)
The scatter did thicken into a ceiling — and the whole sky came down. The question was whether the 1,200-foot scatter would thicken; it did, inside the first ten minutes of the hour: the 12:01Z SPECI already printed BKN012, the old SCT012 become a broken ceiling at its own height. The hourly went further down: BKN008 OVC031 — the ceiling proper fell 2,800 → 800 feet, and the overcast fell 10,000 → 3,100 in the same stamp, the three layers of last stamp replaced by two, both low. The 12:49Z SPECI had pre-printed the landing (BKN008 BKN029 OVC041). With the low ceiling came the mist back into the hourly's body (BR) and visibility down 10SM → 4SM; the fetch's own category for the hourly is IFR, on the ceiling.
The signatures: the $ stayed away — and AUTO left. The $ did not return at twenty-five: a twenty-fifth consecutive absent hourly. But the stamp carries a second absence nobody asked about: there is no AUTO in it. AUTO had stayed a seventh stamp at 11:53Z; inside this hour it departed unannounced — the 12:01Z SPECI still carries it, the 12:17Z and 12:49Z SPECI do not, and the hourly follows them. For seven hourlies the split was AUTO present, $ away; this stamp carries neither. The split has not closed — it has changed sides, and the stamp that changed it was a special, not an hourly.
Around those answers: the wind veered a second stamp, 16008KT → 18006KT on the hourly line, passing through 160 at both early specials and 170 at 12:49Z, and eased 8 → 6 kt, gustless. No extremes, six-hour or tendency groups rode (next ride 17:53Z).
Opening. Just before seven in the morning in Green Bay the pair stands frozen at 16.1 and 15.0, 1.1 apart — the leader on its printed minimum a third stamp, the trailer holding its remembered ground a second — a barometer level at last after pairing its smallest fall twice, a sky that answered the night's question by collapsing: the low scatter thickened into an 800-foot ceiling under a 3,100-foot overcast, mist back in the body at four miles, the fourteenth rain in its second hour and measuring three hundredths, the wind veered to 180 and eased, and neither signature on the stamp — the $ away a twenty-fifth hourly, AUTO gone inside the hour it had survived seven. The 13:53Z stamp (7:53 a.m. CST) inherits the morning's plain questions: does the pair hold a second frozen stamp, and which half breaks it if it breaks; does the level pressure pair itself, resume falling, or print the night's first rise; does the fourteenth rain end in its second hour or enter a third, and does its measure climb past three hundredths; does the 800-foot ceiling lift, hold, or fall further, and does the mist clear back off the body; and do the signatures stay both away — or does one return alone, and which —