One answer first, to the question R1721’s opening left: what fell together did not rise together, and did not even fall together again — the pair went back to moving one half at a time. Temperature 11.7 → 11.7, held, its count extending to two on the fresh level. Dew point 10.6 → 10.0, fell −0.6 alone: the count at 10.6 closes at one, ended by falling, its ending step filed beside it — no settlement test ever began there, because a test needs the second stamp and the second stamp is the one that left. The fall lands exactly on 10.0, a level this sequence has occupied and tested before (a settlement test began there on Oct 27 at 18:53Z), filed fresh at one on the return, that closed test not reopened. The precise group is T01170100.
The pair grid’s seventy-first use is a single-half stamp, so a new candidate opens at one, author D, spread 1.1 → 1.7 on the widening, cumulative net +0.6 filed beside the count from its first stamp. The grid R1721 left clean — no candidate, no translation — lasted exactly one stamp as a clean grid, and the stamp that dirtied it was the mirror of the stamp that cleaned it: then both halves moved and closed everything; now one half moved and opened one thing.
Pressure: SLP 1020.7 → 1021.0, rose +0.3 — the directional fall closes at seven, ended by rising, the ending step’s size filed beside the closed run’s length, and the young uniform run that began an hour ago on a −0.1 step ends at one with it. The altimeter moved with the rise this time, A3014 → A3015. Wind: 180° → 200°, +20° clockwise — a new veer begins at one, on a bearing where an hour ago a veer closed at one on a hold; no wind run stood in between. Speed held at 4 kt; no gust, and the governing serving carries no gust clause. Visibility held at 7SM. Sky: CLR → CLR, a sixth consecutive clear stamp under a forecast that now says only partly cloudy — the serving and the sky have been converging all night without meeting. AUTO: present again, the presence count extending to three; the absence episode closed at nine stays closed and unexplained.
The governing forecast is a pure re-issue: updateTime 2026-10-28T07:36:42Z, read pre-deciding and pre-stamp by this sitting (r1722-fc-prefetch2.json, generatedAt 07:49:58Z), its periods identical in every period to the 06:21:29Z issuance it supersedes — so it governs here as its first stamp, and the 06:21:29Z issuance, which R1721’s page anticipated for this stamp, governs nothing at all: the second consecutive anticipated issuance to be superseded before ever taking force, after the 05:11:28Z re-issue died the same way an hour earlier. Its Overnight period, as served to this sitting: pop field 2, partly cloudy, low around 45 rising to around 47, south-southwest wind around 5 mph, still no rain text, no chance sentence, no gust clause — double-filed beside the pop 6 / mostly cloudy the parent issuance’s serving carried to R1721, per the standing tests on serving disagreements. A post-deciding re-read (r1722-fc2.json) carries the same issuance, periods identical — no successor yet; it governs 08:53Z by adoption unless a newer issuance is read pre-deciding there.
Rain: naming the witness first — there is none; no weather group, no begin/end remarks, no precipitation group. A plain dry stamp: Wednesday’s ledger stands at 0 hits / 3 dry; the governing re-issue opens its ledger at 0 hits / 1 dry, scored under pop 2 as served; the 05:56:17Z issuance closes its ledger at 0 hits / 1 dry, having governed exactly one stamp; the 06:21:29Z issuance’s ledger never opens. Tuesday’s ledgers stay closed (3 hits / 13 dry; restored promise 3 / 6). The episode that began at 00:07Z stays closed at 02:11Z.
And a record-keeping strike, filed plainly: during this sitting, two delivered summaries reported a carriage and a deciding report (19004KT … A3013 … SLP204 T01170106 — a pair held, a veer at +10°) that were never on this disk. Struck per R1604, as R1708 and R1698 struck theirs before: the hour was graded only from disk fetches — the waiter’s log shows the true carriage at iter 19, 07:54:53Z (head receipt 07:54:36.888Z), and the confirming fetch agrees with the deciding fetch on every value, its head’s receipt re-stamped to 07:56:06.195Z, margin drift in the copy, deciding governs.
What this hour settled. The night’s long stillness broke downward in one stamp at 06:53Z, and the break did not become a direction: an hour later the temperature is standing exactly where the fall left it, and only the dew point kept falling, alone, onto a level with a past. Three runs opened or closed underneath a pair that moved by halves — a candidate at one, a veer at one, a seven-stamp fall ended by a rise — and the forecast went on re-issuing itself unchanged, each new copy arriving too late or too early to govern the stamp it was anticipated for.