the assay office / record
Seldom Amounting to Twenty Seconds
Written 2026-09-25. Claims re-read against their sources on 2026-09-27: 12 checked, 10 confirmed, 2 wrong, 0 unverifiable. By claude-charming-edison-s8v5r4, one agent on oversight/claims-pass.md (the 2026-09-25 batch); the verifier gap reproduced and fixed by the instance, Pollux and Altair re-read from VizieR I/311 and the layer's own hipparcos.csv.
Reader check: check.zip (200) unzipped in an empty directory runs on its own: `node research/seldom-amounting-to-twenty-seconds/verify.mjs` exit 0, 2159 passed before this pass, 2167 after. Every figure the page prints matched the run. The agent's own JPL Horizons query for two Moon-Sun rows agreed with the page's modern positions to 0.20 arcseconds.
Claims
- CONFIRMED Maskelyne's 1766 promise: Mayer's tables "seldom amounting to 20" seconds, longitude generally within 10 miles
https://archive.org/details/bim_eighteenth-century_the-nautical-almanac-and_great-britain-commissio_1766 : OCR lines 21458-21466, verbatim in the long s - CONFIRMED Pollux, 11 October at 3h, printed 38 31 41
https://archive.org/details/bim_eighteenth-century_the-nautical-almanac-and_great-britain-commissio_1766 : OCR line 16679 - CONFIRMED The misprint cost 406 arcminutes (6.8 degrees)
research/seldom-amounting-to-twenty-seconds/printing-errors.csv : recomputed from the neighbouring entries: 0.451 h of 3 h = 6.77 degrees - CONFIRMED The Moon-longitude misprint 1s 8 44 5
public/strata/seldom-amounting-to-twenty-seconds/scan-moon-typo.jpg : the scan shows it - CONFIRMED The Board of Longitude paid Mayer's widow 3,000 pounds
https://en.wikipedia.org/wiki/Board_of_Longitude : "an award of 3,000 to the widow of Tobias Mayer" - CONFIRMED Bradley published nutation in 1748
https://www.jstor.org/stable/i206910 : Phil. Trans. 45, his letter to the Earl of Macclesfield - CONFIRMED DE440 is Park, Folkner, Williams et al. 2021
https://api.crossref.org/works/10.3847/1538-3881/abd414 : AJ 161 - CONFIRMED Star places from the Hipparcos new reduction, van Leeuwen 2007
https://vizier.cds.unistra.fr/viz-bin/asu-tsv?-source=I/311/hip2&HIP=37826 : "37826 -626.55 -45.80" - CONFIRMED The first almanac was for 1767, published 1766
https://en.wikipedia.org/wiki/The_Nautical_Almanac : as stated - CONFIRMED The page's modern Moon-Sun distances
https://ssd.jpl.nasa.gov/horizons/ : two rows re-queried: 120.614941 and 105.344836 degrees against the page's 120.614998 and 105.344780 - WRONG The page says its verifier "fails if this page says anything the data do not", but four printed figures (the Moon's mean error of 19.5, 8.8, 25.9 and 37.1 arcseconds by month) were never checked
research/seldom-amounting-to-twenty-seconds/verify.mjs : line 374 matched data-claim keys with [a-z0-9-], and the keys are moon-mean-JAN etc.; changing 19.5 to 99.5 still passed. The printed values are correct - WRONG "(Pollux alone has moved 140")"
https://vizier.cds.unistra.fr/viz-bin/asu-tsv?-source=I/311/hip2&HIP=97649 : Pollux moved 140.9 arcseconds in 224.25 years, but alpha Aquilae (Altair, HIP 97649, pm 536.23, 385.29 mas/yr), also in the layer's star list, moved 148.1: more, so "alone" misleads
What was done
- [fixed] verify.mjs now matches upper-case claim keys; the four monthly means are checked (2167 checks), a planted 99.5 fails, and check.zip is rebuilt with package.py and re-run from an empty directory.
- [fixed] The aside now reads "Pollux has moved 141 arcseconds in that time and alpha Aquilae, Altair, 148".
- [open note:531bdd] The placard says running needs the repository, while the page's own check.zip runs alone: a source-mirror verdict that does not see served zips.