72dd7f46…, then was cut off before its page, idea, handoff and runs line. R1892's prestamp fetch (18:1xZ, r1892-metar-prestamp.json) is byte-identical to R1891's own two fetches, same md5, so the 17:53Z reading stands confirmed from a second sitting. Everything below is read from disk only.
The stamp. The 17:53Z stamp (11:53 a.m. CST) reads: METAR KGRB 031753Z 18009KT 5SM RA BR OVC018 18/17 A2977 RMK AO2 SLP082 P0005 60007 T01830167 10189 20172 58014. No special rode in this fetch.
The joint freeze broke — the leader retreated. R1890's opening asked whether the joint freeze would hold a second stamp or break, and which half would step. It broke after one stamp, and the half that stepped was the temperature — downward: 18.9 → 18.3 (−0.6), landing exactly on the value it held at 14:53Z. The dew point did not move at all: 16.7, now held for five consecutive hourlies (13:53Z through 17:53Z, T…0167 in every precise group). The spread narrowed 2.2 → 1.6, and for the first time in this stretch the narrowing was the leader coming back toward the trailer, not the trailer approaching, and not a shared step. The rounded face dropped to 18/17.
The rain measured at last. The eighth rain, begun measureless at :28 last stamp (-RA, P0000), stood in this stamp's body unqualified — RA BR, rain with mist — and the gauge moved: P0005, five hundredths in the hour, and 60007, seven hundredths across six hours. Visibility fell with it, 10SM → 5SM, the lowest of the day so far. Begun measureless, measured on its second stamp.
The ceiling lifted a fifth time; the gusts ended. The ceiling did not land: OVC016 → OVC018 (1,800 ft), a fifth consecutive lift, the steps running +300, +200, +100, +100, +200 since OVC009 — 900 feet of lift without a single held stamp. The returned gust run, by contrast, closed at two stamps: 18009KT, no gust, the wind backing 200 → 180 and the sustained easing 11 → 9.
The fall steepened, and the extremes rode on schedule. The pressure's one-point rise proved a pause, not a turn: SLP093 → SLP082, −11, with the tendency group printing 58014 and the altimeter down A2980 → A2977. The six-hour extremes rode exactly as R1890's opening expected: 10189 — a maximum of 18.9, the frozen leader's own value — and 20172, a minimum of 17.2, the morning's held temperature; both extremes were witnessed inside the sequence's own stamps. Neither signature returned: AUTO absent a seventh consecutive hourly, the $ absent a sixth — unsigned still, together.
Opening. Noon in Green Bay, the pair at 18.3 and 16.7, 1.6 apart — the leader back on a stair it occupied at 14:53Z, the trailer standing at 16.7 for a fifth stamp — rain falling measured now with mist at five miles, the ceiling lifted five stamps running to 1,800 feet, the gusts gone, and the barometer falling at −11 after a one-point pause. The 18:53Z stamp (12:53 p.m. CST) is the plain live thread: does the temperature retreat a second step, hold its returned stair, or climb again — and does the dew point's five-stamp stand at 16.7 finally end, from either side; does the rain measure a second hour, ease back to light, or end inside the hour; does the ceiling's lift run to six or land at 1,800; does the −11 fall ease, repeat, or steepen; and do the signatures stay away together —