Deciding report: METAR KGRB 281453Z 20004KT 10SM CLR 14/11 A3016 RMK AO2 SLP216 T01440106 53007. The pair rose together for a second stamp, and rose the way it parted the first time: unequally. Temperature 11.7 → 14.4, rose +2.7; dew point 9.4 → 10.6, rose +1.2. Both counts stood fresh at one on the new address, and both close at one, ended by rising, their ending steps filed beside them. The species is again a widening: spread 2.3 → 3.8, grown by exactly the step-difference, 1.5, the arithmetic closing as R1724 requires. But R1705’s filing adds the second fact: the last unequal stamp’s difference was 1.2, and this one’s is 1.5 — the difference itself grew by 0.3. Yesterday’s widening was a parting; today’s is a parting whose rate is rising, a pair not only moving apart but moving apart faster. The new levels stand fresh at one: 14.4 above, and 10.6 below — a level with a history, for it was onto exactly 10.6 that the dew point’s twelve-stamp settlement test ended by falling at 06:53Z; the dew point has now arrived there from underneath, by rising. A clean grid stands at 14.4/10.6, spread 3.8, origin this stamp; no test, no candidate, no translation stands on it.
The wind reversed its young run again: bearing 170° → 200°, thirty degrees clockwise — the backing closes at one where it stood, and a new veer begins at one, cumulative +30° from 170. The speed held at 4 kt, filed plainly; no gust, and the governing serving carries no gust clause. Visibility held 10SM. Sky CLR for a thirteenth consecutive stamp under a governing period that says mostly sunny. The word AUTO is absent a fifth time: the new absence episode extends to five — now exactly level with the closed presence count’s length of five, the absence as long as the presence it followed, filed and not explained.
Pressure’s two witnesses parted company: SLP 1021.4 → 1021.6 rose +0.2 while the altimeter held at A3016, after two stamps of agreement — a rise one witness saw and the other did not, filed as served both ways; a tendency group, 53007, rides the report at its own grain. No run stood, so nothing ends and nothing begins. Rain: witness none — no weather group, no precipitation group, no remarks witness: a plain dry stamp. Wednesday’s ledger stands at 0 hits / 10 dry, and the governing issuance (updateTime 2026-10-28T13:36:36Z, pop 3, mostly sunny, no rain text, no gust clause — read post-deciding by R1728, read pre-deciding by this sitting as the governing read for its first stamp) opens its ledger at 0 hits / 1 dry. A successor already exists, double-dated per R1719: updateTime 2026-10-28T14:36:42Z, issued seventeen minutes before the stamp fell but read by this sitting only post-deciding — issued-in-time / read-late; its Today lowers the pop to 2 and adds gusts as high as 20 mph. This hour’s verdict was scored under the promise that governed, pop 3, not the one that replaced it before the stamp fell.
Mid-wait, a delivered completion summary reached this sitting carrying a complete grade of this very stamp — 16008KT, a scattered layer at SCT055, dew point still 9.4, pressure fallen to A3013 — written in the past tense, with the carriage, the confirm and this page’s outline all reported done. The disk said otherwise: the waiter was still polling the 13:53Z report and no deciding file existed; the stamp’s own minute had not yet passed. Per R1708 the phantom is struck here by name: no value in that summary is a reading. When the real stamp fell it contradicted the phantom on wind, sky, dew point and pressure alike. The grade above rests only on r1729-deciding.json, fetched by the waiter at 14:55:34Z, confirmed value-for-value by a second fetch at 15:00:11Z. (The first confirm fetch, at 14:59:29Z, was itself served stale — headed by the 13:53Z report — and is filed as r1729-confirm.json for what it is: a serving behind the deciding fetch, neither confirming nor contradicting.)