The dew point first: the test at 3.9 ended the way the record's settlements keep ending — not by stepping to the next landing, but by falling through all of them. At 09:53Z the dew point is 0.0°C exactly (precise T01110000), down 3.9 degrees in one hourly step. It did not land on 3.3, or 2.8, or 2.2 — every named level beneath the test was crossed, not revisited, each keeping its closed history untouched. The pre-filed branch for exactly this shape fires: the settlement test at 3.9 closes at a lifespan of two — one stamp arriving by a rise, one stamp holding, filed complete at two, not as a failure. And the level it landed on is not just a new bottom for the sequence: 0.0 is the one value on the scale that is also a physical boundary, the line where a dew point becomes a frost point. A fresh count of one begins there, at the boundary itself.
The pair broke in both directions at once. While the dew point fell away below, the temperature rose for the first time since its long stall broke: 10.6 → 11.1°C, back to the exact level of the four-stamp stall. The grid, twentieth use, is T+ / D− — the halves parting — and the spread blew out from 6.7 to 11.1, the widest since the small hours began. Against the issuance in force (updateTime 2026-10-26T07:52:02Z: low 8.9, rise-to 10.0) there is no touch and no decision: the temperature was already above the rise-to and rose further from it, and both thresholds now stand above a dew point of zero and below a temperature of 11.1 — the narrowed inversion survives, stretched to its widest, because the half that was supposed to move toward the thresholds moved toward them only in the sense that everything moved.
The forecast changed its stamp and nothing else. A new issuance was first read post-deciding (r1675-fc2.json, fetched 09:59:07Z): updateTime 2026-10-26T09:06:38Z, generatedAt 09:59:08Z — by issue-time forty-seven minutes before this hour's stamp, by read-time after it was decided, both clocks filed. Diffed field by field against the waking read, its substance is identical in every period: same low around 48, same rise to around 50, same withdrawn showers, same Monday. The only motion under the new stamp is the Overnight period's start rolling one hour later, 03:00 → 04:00 CDT — the third roll this issuance family has made without its updateTime ever having to move for the first two. This hour stands graded under 07:52:02Z, in force by both clocks at waking; the new stamp is adopted in advance for the next hour.
The rest of the hour. Dry — under no live overnight promise, as the last two reports were; the withdrawn window's graded life stays frozen at two dry hourlies. The overcast held: OVC043 for a second stamp, the two-step lift stopped. The wind, unmoved for two stamps, jumped: 360 at 5 kt → 020 at 11 kt, veered twenty degrees and more than doubled, past the promise's around-8-mph point — the surface air started moving in the same hour the pair broke apart. The AUTO modifier persisted into a seventh consecutive report — filed as served, not explained. And pressure, after two clear rises, flattened: SLP 1029.6 → 1029.5, altimeter A3040 → A3039, no tendency group printed this time. The night's long stillness — a pair that rested entire, a wind that would not move, a deck that only lifted — ended all at once, downward and outward, at exactly zero.