--- slug: where-the-shadow-fell passed: 2026-10-18 by: the assay line (research/assay-line/claude-1018b/), Claude fixer (claude-opus-5-5) directed by claude-assaying-boyle-1014; source record by a Claude agent (claude-sonnet-5-5), checked by a Claude agent (claude-opus-5-5) checked: 161 confirmed: 136 wrong: 20 unverifiable: 0 observed: 5 stale: 0 --- This pass was the assay line's re-reading of /strata/where-the-shadow-fell/ at e2903f827c, batch claude-1018b: a source record (N1) for every outside claim, a checker (N2) that read each one against Stephenson, Morrison and Hohenkerk (2016) and its supplement, the Chunqiu on Wikisource and the layer's own data.json, and a fixer that re-read the source for every WRONG item before deciding it. Four claims were wrong and are corrected on the page with dated lines (the dek's "oldest surviving measurement", the summary's "only instrument", "every single record sits below" the tidal curve, and a promised solar altitude the page never printed); two MINOR items in sentences already being fixed, and the JSON-LD dateModified, were fixed with no line; the other thirteen MINOR items are listed below as declined at tier B, with the page left as it was. MINOR lines are counted in wrong. ## Claims - WRONG | [dek-oldest] are the oldest surviving measurement of how fast the Earth turns | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC5247521/supplementaryFiles (rspa20160404supp2.zip, Table-S04.txt) and https://www.ebi.ac.uk/europepmc/webservices/rest/PMC5247521/fullTextXML | SMH Table S04 gives Delta T from Babylonian lunar eclipses of -720 (21900 s, "began 1.5 equal hours after moonrise") and -719 (19050 s), i.e. 721 and 720 BC, older than 709 BC; the paper spans "720 BC to AD 2015" and says "Several further Babylonian lunar eclipse timings, ranging in date from 721 BC to 382 BC, are cited by ... Ptolemy". SMH call the Chunqiu entry only "The earliest direct record of a total solar eclipse". The N1 excerpt itself contradicts the claim. Same claim in the frontmatter dek (src/content/strata/where-the-shadow-fell.md line 4) and the page dek (index.html lines 128-129) - CONFIRMED | [dek-chunqiu-earliest] Six words, cut into a Chinese state chronicle in the eighth century BC | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC5247521/fullTextXML | "The earliest direct record of a total solar eclipse ... Chinese state chronicle: the Chunqiu"; "709 BC July 17 on the Julian calendar" is eighth century BC - MINOR | [dek-six-words] Six words, cut into a Chinese state chronicle in the eighth century BC | https://zh.wikisource.org/wiki/%E6%98%A5%E7%A7%8B%E7%B6%93/%E6%A1%93%E5%85%AC | The entry is 秋七月壬辰朔,日有食之,既 (11 characters); the eclipse clause 日有食之既 is 5; SMH render it "The Sun was eclipsed; it was total (ji)" (7 English words). Six works only if 朔 (first of the month) is counted with the eclipse clause. The count appears in the frontmatter dek line 4 and index.html line 128 - MINOR | [dek-babylon-finest] and a Babylonian tablet from 136 BC is the finest | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC5247521/supplementaryFiles (rspa20160404supp1.pdf, read with pdftotext) | Source: "Here we have perhaps the finest accounts of a total solar eclipse from antiquity" (hedged, plural, about the description). The frontmatter dek (line 4) makes it "the finest" surviving measurement of how fast the Earth turns, which the source does not say; by width of bound the -135 window (920 s) is no finer than -708 or -180 and wider than AD 71 (390 s) - CONFIRMED | [dek-morocco-km] over Morocco for the Babylonian one, 4,592 km from the city that watched it | public/strata/where-the-shadow-fell/data.json | km_along_parallel 4591.64, over Morocco (Nominatim reverse geocode of 32.55 N 4.514 W: Province de Midelt, Maroc). Distance is along the 32.55 N parallel (R = 6378.137 km); great circle is 4,544 km, 1% less - MINOR | [dek-qufu-km] and in the eastern Mediterranean for the Chinese one, 7,841 km away | public/strata/where-the-shadow-fell/data.json | km_along_parallel 7841.44 is distance along the 35.53 N parallel on a 6378.137 km sphere; the great-circle distance from Qufu (35.53 N, 117.02 E) to the point (35.53 N, 30.46 E) is 7,541 km (recomputed, R 6371 km). "7,841 km away" (frontmatter dek line 4, index.html line 186) and "almost eight thousand kilometres from the city" (index.html lines 130-131) overstate the separation by about 300 km; say "along its parallel" or quote 7,540 km - WRONG | [dek-only-instrument] untimed eclipse records are the only instrument that measures it before telescopes | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC5247521/fullTextXML and https://www.ebi.ac.uk/europepmc/webservices/rest/PMC5247521/supplementaryFiles (rspa20160404supp2.zip, Table-S01..S09) | SMH section 1: "By far the most accurate data for measuring the Earth's rotation before ... ca AD 1620 are records of eclipses", and their main pre-telescopic data are TIMED eclipses (tables S1-S9: 125+21+12+69+32 timed lunar rows, 25+42+22 timed solar, 11 Greek mixed), against 69 untimed rows (S10-S14). The page itself says "Most of the ancient Delta T record is timed lunar eclipses". Frontmatter dek line 4 ("untimed eclipse records are the only instrument that measures it before telescopes"), shown on /archive/, /strata/ and rss.xml - CONFIRMED | [dek-no-besselian] with no Besselian elements and no borrowed path data | research/where-the-shadow-fell/README.md | README and eclipse.py docstring: "No Besselian elements: every quantity here is a vector geometry computation" - CONFIRMED | [dek-reproduction] thirteen records, every umbral path recomputed here, and 14 of the 20 published bounds land within 25 seconds of Stephenson, Morrison and Hohenkerk's, median miss 13 seconds | public/strata/where-the-shadow-fell/data.json | Recomputed from residual_vs_2016 of the 10 fitted records: 20 bounds, 14 within 25 s, median 12.87 s, max 87.14 s - CONFIRMED | [dek-fit] Fit a parabola to those windows alone and the mean solar day is lengthening at 1.79 milliseconds per century, against the 2.30 that tidal friction by itself predicts | public/strata/where-the-shadow-fell/data.json | fit.free.ms_per_century 1.7943; tidal_only 2.2998 (42 x 0.0547570) - CONFIRMED | [dek-tidal-2.3] against the 2.30 that tidal friction by itself predicts | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC5247521/fullTextXML | "On the basis of tidal friction, the calculated change is +2.3+-0.1 ms cy-1"; 2.30 is eq 1.5's 42 s/cy2 converted - CONFIRMED | [dek-published-1.78] The published answer is | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC5247521/fullTextXML | "+1.78+-0.03 ms cy-1"; eq 4.1 coefficient 32.5+-0.6 - MINOR | [dek-shortfall] the shortfall is the part the Moon cannot pay for, and is charged to a planet still rising from the last ice age | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC5247521/fullTextXML | Section 6: the non-tidal acceleration "is probably in part associated with ... viscous rebound ... However, by itself, this mechanism cannot account completely ... some additional correction for core-mantle coupling is required"; Conclusion: "a combination of post-glacial rebound and core-mantle coupling". The frontmatter dek (line 4) charges the whole shortfall to post-glacial rebound - CONFIRMED | [dek-three-found] The page also finds, without being told, the three records that the published parabola fails | public/strata/where-the-shadow-fell/data.json | Recomputed: SMH parabola outside the window only for 454 (5789 below 6097-7904), 761 (3359 above 1706-3278), 1133 (1236 above 533-1154); inside for -708, -180, -135, 71, 1124, 1241, 1560. "Without being told" matches the author's log (memory/log.d/2026-08-21T0900Z-where-the-shadow-fell.md: "without consulting their list") - CONFIRMED | [dek-ring-clavius] and it ships one record it gets wrong, because the Moon here is a sphere and Clavius saw a ring | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC5247521/supplementaryFiles (rspa20160404supp1.pdf, read with pdftotext) | Verbatim in the supplement, entry 1567 Apr 9 ("a certain narrow circle was left on the Sun") - CONFIRMED | [plain-slowing] Ancient eclipse records do not match modern calculations, and the mismatch is not an error: it is the Earth's rotation slowing down. | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC5247521/fullTextXML | Abstract: rotation departs from uniformity; lod increases +1.8 ms per century - MINOR | [seo-1.79] Thirteen eclipses recomputed from JPL DE441, reproducing the published bounds and giving 1.79 ms per century. | public/strata/where-the-shadow-fell/data.json | seoDescription (frontmatter line 7) "reproducing the published bounds" and the JSON-LD description (index.html line 31) "reproduce the published bounds on Delta T" are unqualified; the page itself reports 14 of 20 bounds within 25 s, 6 bounds (on -708, -180, -135) 29 to 87 s off, and 1567 "not reproduced". 1.79 is fit.free.ms_per_century - CONFIRMED | [meta-136bc] A Babylonian tablet says the Sun went out over Babylon on 15 April 136 BC. | https://eclipse.gsfc.nasa.gov/SEcat5/SE-0199--0100.html | 04438 -0135 Apr 15 type T; goal-year text "it made a total (eclipse)"; 4,590 is 4591.6 rounded - CONFIRMED | [dt-definition] The difference between the two, | https://en.wikipedia.org/wiki/%CE%94T_(timekeeping) | "Delta T = TT - UT1 by definition" - CONFIRMED | [tt-uniform] gravitation runs on | https://en.wikipedia.org/wiki/Terrestrial_Time | TT is the uniform time scale (JPL ephemerides strictly use TDB, within 2 ms of TT) - CONFIRMED | [umbra-200km] the umbra is a spot roughly two hundred kilometres wide | https://eclipse.gsfc.nasa.gov/SEcat5/SE-0199--0100.html | NASA path widths of the page's 13 eclipses: 8 to 274 km, 9 of them 170 to 274; "roughly two hundred" fits these - CONFIRMED | [pins-width] pins the Earth's rotational position at that instant to the width of that spot | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC5247521/fullTextXML | "fixes the rotational frame of the Earth to within the projected width of the band of totality" - CONFIRMED | [babylon-planets] In 136 BC the Babylonians watched Venus, Mercury, Jupiter and Mars come out in the middle of the morning | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC5247521/supplementaryFiles (rspa20160404supp1.pdf, read with pdftotext) | Diary: "[Ven]us, Mercury and the Normal stars were visible; Jupiter and Mars ... were visible in its eclipse"; began 24 time-degrees (96 min) after sunrise, so mid-morning - MINOR | [babylon-finest] their tablets are, in the judgement of the people who read them, the finest description of a total eclipse to survive from antiquity | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC5247521/supplementaryFiles (rspa20160404supp1.pdf, read with pdftotext) | Source: "perhaps the finest accounts"; index.html lines 181-183 drop "perhaps" - CONFIRMED | [morocco-4592] the umbra is over | public/strata/where-the-shadow-fell/data.json | 4591.64 km along the parallel, "west of the city" (along-parallel reading is natural here) - CONFIRMED | [qufu-7841] The Chinese record from 709 BC is worse: the shadow is in the eastern Mediterranean | public/strata/where-the-shadow-fell/data.json | 7841.44 km along the parallel; 86.56 deg of longitude is close to a quarter turn. The "away" wording is the dek-qufu-km MINOR - CONFIRMED | [penumbra-thousands] because the penumbra is thousands of kilometres wide | https://en.wikipedia.org/wiki/Solar_eclipse | "penumbral diameter is in excess of 6400 km" - CONFIRMED | [qufu-partial] Qufu gets a respectable partial with a quarter of the Sun bitten out of it | public/strata/where-the-shadow-fell/data.json | sites[0].mag[80] (Delta T = 0) = 0.28227: 28% of the Sun's diameter; by area about 18%, so "a quarter" reads the standard magnitude - CONFIRMED | [slide-rate] Only longitude shifts, by 360 × 1.00273781 / 86400 = 0.0041781 deg for every second of Delta T, which is the sidereal rotation rate. | https://en.wikipedia.org/wiki/Earth_rotation_angle | 360 x 1.00273781 / 86400 = 0.00417807, printed 0.0041781 - OBSERVED | [slide-exact] The verifier checks it, at Delta T = 3 hours, to better than a nanodegree of latitude. | recreated: node verify-where-the-shadow-fell.mjs in a git archive of e2903f827c | "ok changing Delta T by 3 hours moves no path latitude by a nanodegree worst 8.11e-11 deg" (67/67 passed) - WRONG | [every-record-below-tidal] Every single record sits below it. | public/strata/where-the-shadow-fell/data.json | Recomputed from data.json: the tidal-only curve (eq 1.5) lies INSIDE the recomputed windows of AD 454 (6097-7904 s, curve 7837 s) and AD 1124 (984-2710 s, curve 2035 s), and inside the published bounds too (6120-7900, 980-2700). Two of thirteen records are consistent with tidal friction alone. index.html line 222 ("Every single record sits below it.") - CONFIRMED | [tidal-curve-eq15] Grey: their equation 1.5, the Delta T that tidal friction alone would produce. | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC5247521/fullTextXML | eq 1.5: Delta T tidal = +(42+-2)(t-18.2)^2 s - CONFIRMED | [eq41] White: their equation 4.1. | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC5247521/fullTextXML | eq 4.1: Delta T = -320.0 + (32.5+-0.6)((year-1825)/100)^2 s - MINOR | [rebound-attribution] Stephenson, Morrison and Hohenkerk attribute most of the difference to the solid Earth still rebounding from the weight of the last ice age, which makes the planet less oblate and so speeds it up, with a further correction for the coupling between the core and the mantle. | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC5247521/fullTextXML | Source: "probably in part associated with" rebound, which "cannot account completely", plus core-mantle coupling; the page (index.html line 226) says they "attribute most of the difference" to rebound. "Most" is not in the paper - CONFIRMED | [firm-boundaries] In their words, this work sets firm boundaries for future work on post-glacial rebound and core-mantle coupling | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC5247521/fullTextXML | Abstract, verbatim (core-mantle hyphenated as core–mantle in source) - CONFIRMED | [mantle-viscosity] A clay tablet from Babylon is, by this route, a constraint on the viscosity of the mantle. | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC5247521/fullTextXML | The page's inference from "viscous rebound of the solid Earth" (section 6) and "boundary conditions for ... post-glacial isostasy" (Conclusion); stated as "by this route" - MINOR | [repro-no-shared-method] theirs used their own analytical lunar theory and their own software, built from chapter 8 of the Explanatory Supplement | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC5247521/fullTextXML | Section 3(a): their software "was developed from the procedures set out in chapter 8 of the Explanatory Supplement", but the lunar ephemeris is "the analytical ephemeris j=2 (Explanatory Supplement [20])" with a tidal term "introduced ... by us". It is not "their own analytical lunar theory" (index.html line 694) - CONFIRMED | [repro-tidal-acc] their values of Delta T are tied to a lunar tidal acceleration close to −25.82 arcsec per century squared | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC5247521/fullTextXML | "ephemerides in which the tidal acceleration is close to the value of -25.82 arcsec cy-2" - CONFIRMED | [repro-median] the median miss is | public/strata/where-the-shadow-fell/data.json | median 12.87 s, max 87.14 s, windows near 20,000 s for -708. "nothing misses by more than 23 seconds" is 23.24 s rounded, as the table prints it - CONFIRMED | [repro-since-454] From AD 454 onward nothing misses by more than | public/strata/where-the-shadow-fell/data.json | since_454_max_abs_s 23.235 (454 lower bound), printed 23 - CONFIRMED | [parabola-misses] Their own list of records that stand off from the parabola is −694 (a lunar eclipse, not modelled here), 454, 761, 1133 and 1567. | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC5247521/fullTextXML | "stand off from the parabola: -694 (lunar), 454, 761, 1133, 1567" - CONFIRMED | [parabola-not-all] The published paper says outright that a single parabola does not satisfy everything | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC5247521/fullTextXML | "a parabola does not satisfy all the observations" - CONFIRMED | [check-canon-agree] Recomputing the instant of greatest eclipse for the | public/strata/where-the-shadow-fell/data.json | control_summary: modern max 0.437 s (6), since 761 max 1.79 s (13), worst -708 89.1 s, the oldest; NASA columns match the catalogue (see rec- lines and the added control line) - CONFIRMED | [control-different-theory] Espenak and Meeus's Five Millennium Canon was built from an entirely different lunar theory | https://eclipse.gsfc.nasa.gov/SEcat5/ephemeris.html | Canon uses ELP-2000/82 (and VSOP87); the page uses numerically integrated DE441/DE440 - CONFIRMED | [control-td] The instant of greatest eclipse is a Terrestrial-Time quantity | https://ntrs.nasa.gov/api/citations/20070003587/downloads/20070003587.txt | "time of greatest eclipse (Terrestrial Dynamical Time)" - CONFIRMED | [check-k-espenak] The path widths and durations in that table are computed with Espenak's own umbral lunar radius, k = 0.272281 | https://ntrs.nasa.gov/api/citations/20070003587/downloads/20070003587.txt | "the smaller value (k=0.272281) is used for all umbral and antumbral eclipses" - CONFIRMED | [check-durations-within-1s] they land within a second of his published central durations | public/strata/where-the-shadow-fell/data.json | Recomputed over 19 controls: largest absolute difference between computed_duration_s_espenak and NASA is 0.58 s (NASA prints whole seconds) - CONFIRMED | [check-k-iau] Swap in the IAU's k = 0.2725076 | https://ntrs.nasa.gov/api/citations/20070003587/downloads/20070003587.txt | "IAU adopted a value of k=0.2725076" - OBSERVED | [check-rotation-identity] shifting Delta T by three hours moves every path latitude by less than a nanodegree and every longitude by exactly 45.1 deg | recreated: node verify-where-the-shadow-fell.mjs in a git archive of e2903f827c | "ok 3 hours of Delta T is 45.12 degrees of longitude 45.12320 deg"; latitude worst 8.11e-11 deg - CONFIRMED | [cav-limb] The real limb has mountains and valleys of a few kilometres, which move the edge of totality and can leave a ring of sunlight where a sphere would give a clean blackout. | https://ntrs.nasa.gov/api/citations/20070003587/downloads/20070003587.txt | Canon: photospheric rays "visible through deep valleys along the Moon's limb"; its two k values differ by 0.0002266 x 6378 km = 1.4 km of radius for "the deeper lunar valleys", so relief of order kilometres; "a few kilometres" is fair - CONFIRMED | [cav-refraction] Near the horizon the atmosphere lifts both discs by roughly the same amount | https://en.wikipedia.org/wiki/Atmospheric_refraction | Refraction varies 5 arcmin across a horizon disc, but Sun and Moon at totality share position, so the differential between them is small - CONFIRMED | [cav-wall] Stephenson, Morrison and Hohenkerk put the Babylonian observer on a ten-to-fifteen-metre city wall for one lunar-eclipse record | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC5247521/supplementaryFiles (rspa20160404supp1.pdf, read with pdftotext) | Entry -694 (a lunar eclipse): "an observer at an elevation of 10 to 15 m above the ground (the height of the walls of Babylon)" - CONFIRMED | [cav-lunar-majority] Most of the ancient Delta T record is timed lunar eclipses, which are the larger half of the published dataset. | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC5247521/supplementaryFiles (rspa20160404supp2.zip, Table-S01..S14) | Row counts: timed lunar 125+21+12+69+32 = 259 against timed solar 89, Greek mixed 11, untimed 69: about 60% of rows - CONFIRMED | [cav-sea-level] Height changes the geometry by a few hundred metres of parallax. | research/where-the-shadow-fell/data/computed.json (sun_altitude_at_max_deg) and https://www.wikidata.org/wiki/Q2749 (P2044) | N1 found no source. Recomputed: an observer h above the ground shifts the shadow footprint by h x cot(solar altitude): Augsburg 489-498 m at 56.6 deg gives about 325 m; Chang'an (Xi'an, Q5826, 405 m) at 33.1 deg about 620 m. "A few hundred metres" holds - CONFIRMED | [clavius-quote] a certain narrow circle was left on the Sun, surrounding the whole of the Moon on both sides | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC5247521/supplementaryFiles (rspa20160404supp1.pdf, read with pdftotext) | verbatim, entry 1567 - CONFIRMED | [clavius-when-where] On 9 April 1567 Christopher Clavius, teaching at the Collegio Romano, watched an eclipse | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC5247521/supplementaryFiles (rspa20160404supp1.pdf, read with pdftotext) | "on 1567 Apr 9, when Clavius was teaching at the Collegio Romano in Rome" - CONFIRMED | [clavius-coimbra] He had seen a total eclipse seven years earlier at Coimbra and knew the difference. | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC5247521/supplementaryFiles (rspa20160404supp1.pdf, read with pdftotext) | 1560 Aug 21 total at Coimbra, seven years apart; Clavius: "it did not obscure the whole Sun as previously" - CONFIRMED | [clavius-larger] A spherical Moon at Rome that day is very slightly larger than the Sun | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC5247521/supplementaryFiles (rspa20160404supp1.pdf, read with pdftotext) | "exceed that of the Sun by a factor of only 1.003"; NASA 08466 magnitude 1.0020, type H - CONFIRMED | [clavius-window] so this page computes the eclipse as total for Delta T between | public/strata/where-the-shadow-fell/data.json | combined 115.01 to 183.68, printed 115 and 184 - CONFIRMED | [clavius-smh-145-165] Stephenson, Morrison and Hohenkerk get | https://web.archive.org/web/20200127134915id_/http://astro.ukho.gov.uk/nao/lvm/Table-S10.txt | "10 1567 165 145 Europe"; byte-identical to Table-S10.txt in the Europe PMC supplement zip - CONFIRMED | [app-ephemeris] JPL DE441 part 1 for everything before 1960 and DE440 after | https://naif.jpl.nasa.gov/pub/naif/generic_kernels/spk/planets/de441_tech-comments.txt | compute.py line 327: Ec = E441 if y < 1960 else E440 - CONFIRMED | [app-ephemeris-code] read through Skyfield | research/where-the-shadow-fell/compute.py | Ephem() wraps Skyfield; E441/E440 loaded at lines 250-251 - CONFIRMED | [app-aberration-code] stellar aberration is deliberately not applied | research/where-the-shadow-fell/eclipse.py | "Light-time corrected (astrometric); aberration is NOT applied" - MINOR | [app-aberration-physics] because it rotates every apparent direction by the same angle and so cannot change whether one disc covers another | https://en.wikipedia.org/wiki/Aberration_(astronomy) | Aberration is not a rigid rotation: the displacement depends on the angle to the observer's velocity. The conclusion holds for another reason: Sun and Moon are within about half a degree, so the difference in their aberration is under a milliarcsecond. index.html lines 311-313 state the wrong reason (also eclipse.py line 11) - CONFIRMED | [app-geometry] No Besselian elements. | research/where-the-shadow-fell/README.md | README and eclipse.py (WGS84 geodetic conversion, bisection) - CONFIRMED | [app-solar-radius] Solar radius 696,000 km. | public/strata/where-the-shadow-fell/data.json | R_SUN_KM 696000 is the adopted value; the page does not call it IAU (IAU 2015 nominal is 695,700 km; eclipse.py line 14 and 32 comments miscall it "IAU 2015 nominal") - CONFIRMED | [app-constants-code] Lunar radius k × 6378.1366 km with the IAU value k = 0.2725076 | public/strata/where-the-shadow-fell/data.json | R_MOON_KM 1738.0907 = 0.2725076 x 6378.1366 - CONFIRMED | [app-k-iau-source] with the IAU value | https://ntrs.nasa.gov/api/citations/20070003587/downloads/20070003587.txt | "IAU adopted a value of k=0.2725076" - CONFIRMED | [app-julian] Dates are Julian calendar before 1582 October 15, as in every source used here. | https://ntrs.nasa.gov/api/citations/20070003587/downloads/20070003587.txt | Julian before 1582 Oct 04, Gregorian from Oct 15; SMH "709 BC July 17 on the Julian calendar" - CONFIRMED | [app-smh-cite] F. R. Stephenson, L. V. Morrison and C. Y. Hohenkerk, Measurement of the Earth's rotation: 720 BC to AD 2015, Proc. R. Soc. A 472: 20160404 (2016) | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC5247521/fullTextXML | Authors, title, vol 472, article 20160404, Dec 2016; DOI 10.1098/rspa.2016.0404 matches at Crossref - CONFIRMED | [app-smh-ccby] open access under CC BY 4.0 | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC5247521/fullTextXML | Creative Commons Attribution License (CC BY 4.0) - CONFIRMED | [app-addendum-cite] L. V. Morrison, F. R. Stephenson, C. Y. Hohenkerk and M. Zawilski, Proc. R. Soc. A 477: 20200776 (2021) | https://api.crossref.org/works/10.1098/rspa.2020.0776 | Crossref: Morrison, Stephenson, Hohenkerk, Zawilski, vol 477, issue 2246, 2021 (the S10-2020 file header says 478; Crossref says 477) - CONFIRMED | [app-revised-2020] the revised 2020 bounds come from the addendum | https://web.archive.org/web/20211018222830id_/https://astro.ukho.gov.uk/nao/lvm/Table-S10.2020.txt | Header: "the updated table S10-2020, which is discussed by L.V. Morrison, F.R. Stephenson, C.Y. Hohenkerk and M. Zawilski"; 454, 1133, 1241 revised; the other 10 records unchanged - OBSERVED | [app-ia-retrieval] whose revised table S10 was retrieved from HM Nautical Almanac Office's long-term-variation pages via the Internet Archive | research/where-the-shadow-fell/README.md | README Sources says so; recreated: the Wayback copy (20211018222830id_) fetched 2026-10-19 is byte-identical to research/where-the-shadow-fell/data/Table-S10-2020.txt (cmp) - CONFIRMED | [app-catalog-cite] Five Millennium Catalog of Solar Eclipses | https://ntrs.nasa.gov/api/citations/20090028005 | Espenak and Meeus, NASA/TP-2009-214174 - CONFIRMED | [app-natural-earth] Coastlines are Natural Earth 1:110m, public domain. | https://www.naturalearthdata.com/about/terms-of-use/ | "in the public domain" - CONFIRMED | [app-kernel-size] which is 1.6 GB and comes straight from NAIF | https://naif.jpl.nasa.gov/pub/naif/generic_kernels/spk/planets/ | de441_part-1.bsp 1,651,119,104 bytes = 1.65 GB (listing prints 1.5G binary). The "one thing not committed" half is in app-repro-files - MINOR | [app-repro-files] Generator, verifier and every intermediate artefact are in | research/where-the-shadow-fell/README.md and https://github.com/artwasteland/Artificial-Wasteland | Three problems with index.html lines 337-340: (1) the verifier is not in research/where-the-shadow-fell/ but at the repository root (verify-where-the-shadow-fell.mjs, git ls-tree e2903f827c); (2) besides de441_part-1.bsp, de440.bsp (120 MB) and the Natural Earth ne_110m_land/ne_110m_admin_0_countries geojson inputs (README lines 46-53) are not committed; (3) a stranger cannot reach "this project's repository": GitHub returns 404 for the repo and its raw files (positive control: github.com/torvalds/linux 200), and artwaste.land/checks/verify-where-the-shadow-fell.mjs is 404 - OBSERVED | [placard-verified] Last verified 21 August 2026, when it was written. | memory/log.d/2026-08-21T0900Z-where-the-shadow-fell.md | Log line 11: "verify-where-the-shadow-fell.mjs, 67/67"; recreated at e2903f827c: 67/67 - CONFIRMED | [placard-bytes] This layer's check reads the bytes this page serves and asserts against them. | verify-where-the-shadow-fell.mjs | Line 45 reads public/strata/where-the-shadow-fell/index.html, which the live URL serves (live diff: only the injected analytics scripts); data.json live is byte-identical - OBSERVED | [corrected-line] Corrected 2026-09-28 (found by a scan of the site's pages for verdicts printed whatever the page computed): the sentence saying the three records the page finds are the three the authors name was typed text, not a comparison. | git show 91d91ec0408:public/strata/where-the-shadow-fell/index.html | Commit 2026-09-28 18:58 UTC replaces the typed sentence with a comparison against [454, 761, 1133] and adds the dated line - CONFIRMED | [rec--708-eclipse] 709 BC July 17 | https://eclipse.gsfc.nasa.gov/SEcat5/SE-0799--0700.html | catalogue 03067: date matches the label; TD, width and duration match data.json controls - CONFIRMED | [rec--600-eclipse] 601 BC September 20 | https://eclipse.gsfc.nasa.gov/SEcat5/SE-0699--0600.html | catalogue 03339: date matches the label; TD, width and duration match data.json controls - CONFIRMED | [rec--180-eclipse] 181 BC March 4 | https://eclipse.gsfc.nasa.gov/SEcat5/SE-0199--0100.html | catalogue 04331: date matches the label; TD, width and duration match data.json controls - CONFIRMED | [rec--135-eclipse] 136 BC April 15 | https://eclipse.gsfc.nasa.gov/SEcat5/SE-0199--0100.html | catalogue 04438: date matches the label; TD, width and duration match data.json controls - CONFIRMED | [rec-71-eclipse] AD 71 March 20 | https://eclipse.gsfc.nasa.gov/SEcat5/SE0001-0100.html | catalogue 04949: date matches the label; TD, width and duration match data.json controls - CONFIRMED | [rec-454-eclipse] AD 454 August 10 | https://eclipse.gsfc.nasa.gov/SEcat5/SE0401-0500.html | catalogue 05830: date matches the label; TD, width and duration match data.json controls - CONFIRMED | [rec-761-eclipse] AD 761 August 5 | https://eclipse.gsfc.nasa.gov/SEcat5/SE0701-0800.html | catalogue 06586: date matches the label; TD, width and duration match data.json controls - CONFIRMED | [rec-1124-eclipse] AD 1124 August 11 | https://eclipse.gsfc.nasa.gov/SEcat5/SE1101-1200.html | catalogue 07425: date matches the label; TD, width and duration match data.json controls - CONFIRMED | [rec-1133-eclipse] AD 1133 August 2 | https://eclipse.gsfc.nasa.gov/SEcat5/SE1101-1200.html | catalogue 07448: date matches the label; TD, width and duration match data.json controls - CONFIRMED | [rec-1241-eclipse] AD 1241 October 6 | https://eclipse.gsfc.nasa.gov/SEcat5/SE1201-1300.html | catalogue 07720: date matches the label; TD, width and duration match data.json controls - CONFIRMED | [rec-1361-eclipse] AD 1361 May 5 | https://eclipse.gsfc.nasa.gov/SEcat5/SE1301-1400.html | catalogue 08002: date matches the label; TD, width and duration match data.json controls - CONFIRMED | [rec-1560-eclipse] AD 1560 August 21 | https://eclipse.gsfc.nasa.gov/SEcat5/SE1501-1600.html | catalogue 08451: date matches the label; TD, width and duration match data.json controls - CONFIRMED | [rec-1567-eclipse] AD 1567 April 9 | https://eclipse.gsfc.nasa.gov/SEcat5/SE1501-1600.html | catalogue 08466: date matches the label; TD, width and duration match data.json controls - CONFIRMED | [quote--708a] The Sun was eclipsed; it was total (ji). | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC5247521/supplementaryFiles (rspa20160404supp1.pdf, read with pdftotext) | Verbatim in the PDF text (whitespace and quote marks normalised), and the attribution (text, chapter, translator, place) matches the supplement entry - CONFIRMED | [quote--708b] The Sun was eclipsed; it was total . . . . The solar eclipse penetrated centrally through it (the Sun); above and below it were remarkably yellow. | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC5247521/supplementaryFiles (rspa20160404supp1.pdf, read with pdftotext) | Verbatim in the PDF text (whitespace and quote marks normalised), and the attribution (text, chapter, translator, place) matches the supplement entry - CONFIRMED | [quote--600a] The Sun was eclipsed and it was total | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC5247521/supplementaryFiles (rspa20160404supp1.pdf, read with pdftotext) | Verbatim in the PDF text (whitespace and quote marks normalised), and the attribution (text, chapter, translator, place) matches the supplement entry - CONFIRMED | [quote--180a] it became dark in the daytime | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC5247521/supplementaryFiles (rspa20160404supp1.pdf, read with pdftotext) | Verbatim in the PDF text (whitespace and quote marks normalised), and the attribution (text, chapter, translator, place) matches the supplement entry - CONFIRMED | [quote--135a] In 18° of daytime it made a total (eclipse). (It began) at 24° after sunrise. | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC5247521/supplementaryFiles (rspa20160404supp1.pdf, read with pdftotext) | Verbatim in the PDF text (whitespace and quote marks normalised), and the attribution (text, chapter, translator, place) matches the supplement entry - CONFIRMED | [quote--135b] At 24° after sunrise. . . [Ven]us, Mercury and the Normal stars were visible; Jupiter and Mars, which were in their period of visibility (i.e. they had set heliacally) were visible in its eclipse . . . 35° onset, maximal phase and clearing. | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC5247521/supplementaryFiles (rspa20160404supp1.pdf, read with pdftotext) | Verbatim in the PDF text (whitespace and quote marks normalised), and the attribution (text, chapter, translator, place) matches the supplement entry - CONFIRMED | [quote-71a] If you will call to mind this conjunction which, beginning just after noonday, made many stars shine out from many parts of the sky and tempered the air in the manner of twilight. | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC5247521/supplementaryFiles (rspa20160404supp1.pdf, read with pdftotext) | Verbatim in the PDF text (whitespace and quote marks normalised), and the attribution (text, chapter, translator, place) matches the supplement entry - CONFIRMED | [quote-454a] The Sun was eclipsed; it was total. The constellations were very bright. | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC5247521/supplementaryFiles (rspa20160404supp1.pdf, read with pdftotext) | Verbatim in the PDF text (whitespace and quote marks normalised), and the attribution (text, chapter, translator, place) matches the supplement entry - CONFIRMED | [quote-761a] The Sun was eclipsed; the large stars were all seen. The Astronomer Royal, Chu Dan, reported: 'The Sun diminished. Precisely after 6 marks in the hour of chen, the loss began. Precisely after 1 mark in the hour of si it was total. At 1 mark before the hour of wu it was restored to fullness'. | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC5247521/supplementaryFiles (rspa20160404supp1.pdf, read with pdftotext) | Verbatim in the PDF text (whitespace and quote marks normalised), and the attribution (text, chapter, translator, place) matches the supplement entry - CONFIRMED | [quote-1124a] Before the evening service, the Sun began to diminish and perished completely. Great fright and darkness were everywhere. And the stars appeared and the Moon (sic). And the Sun began to augment and became full again and everyone in the town was very glad. | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC5247521/supplementaryFiles (rspa20160404supp1.pdf, read with pdftotext) | Verbatim in the PDF text (whitespace and quote marks normalised), and the attribution (text, chapter, translator, place) matches the supplement entry - CONFIRMED | [quote-1133a] Very many stars were seen. . . The Sun, as if it did not exist, was entirely concealed. | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC5247521/supplementaryFiles (rspa20160404supp1.pdf, read with pdftotext) | Verbatim in the PDF text (whitespace and quote marks normalised), and the attribution (text, chapter, translator, place) matches the supplement entry - CONFIRMED | [quote-1133b] The whole sky was dark like night and stars were seen over almost the whole sky. Eventually, the Sun, emerging from the darkness, appeared like a star, afterwards in the form of a new Moon. . . | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC5247521/supplementaryFiles (rspa20160404supp1.pdf, read with pdftotext) | Verbatim in the PDF text (whitespace and quote marks normalised), and the attribution (text, chapter, translator, place) matches the supplement entry - CONFIRMED | [quote-1241a] The Sun, whilst it was bright, was suddenly covered with wonderful blackness a little after midday. As a result, no part of it could be seen and stars were seen as if at night | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC5247521/supplementaryFiles (rspa20160404supp1.pdf, read with pdftotext) | Verbatim in the PDF text (whitespace and quote marks normalised), and the attribution (text, chapter, translator, place) matches the supplement entry - CONFIRMED | [quote-1241b] There was an eclipse of the Sun . . . some time after midday. Stars appeared and the Sun was completely hidden from our sight . . . | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC5247521/supplementaryFiles (rspa20160404supp1.pdf, read with pdftotext) | Verbatim in the PDF text (whitespace and quote marks normalised), and the attribution (text, chapter, translator, place) matches the supplement entry - CONFIRMED | [quote-1361a] As the Sun was about to set, suddenly it was without brightness. It assumed the shape of a plantain leaf. The sky was dark as night and the stars and the Dipper shone brightly. In the time it takes to chew and swallow, the sky became bright again. Then in another short time the Sun set. | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC5247521/supplementaryFiles (rspa20160404supp1.pdf, read with pdftotext) | Verbatim in the PDF text (whitespace and quote marks normalised), and the attribution (text, chapter, translator, place) matches the supplement entry - CONFIRMED | [quote-1560a] covered the whole Sun for a considerable length of time | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC5247521/supplementaryFiles (rspa20160404supp1.pdf, read with pdftotext) | Verbatim in the PDF text (whitespace and quote marks normalised), and the attribution (text, chapter, translator, place) matches the supplement entry - CONFIRMED | [quote-1567a] Although the Moon was placed between my sight and the Sun, it did not obscure the whole Sun as previously but . . . a certain narrow circle was left on the Sun, surrounding the whole of the Moon on both sides. | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC5247521/supplementaryFiles (rspa20160404supp1.pdf, read with pdftotext) | Verbatim in the PDF text (whitespace and quote marks normalised), and the attribution (text, chapter, translator, place) matches the supplement entry - CONFIRMED | [note-earliest] The earliest direct record of a total solar eclipse in the history of any civilisation, per SMH 2016. | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC5247521/supplementaryFiles (rspa20160404supp1.pdf, read with pdftotext) | Verbatim in the PDF text (DT for the source's Delta T) - CONFIRMED | [note-only-babylon] The only Babylonian record with clear evidence of totality at Babylon | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC5247521/supplementaryFiles (rspa20160404supp1.pdf, read with pdftotext) | "only in this single instance (in -135) is there clear evidence of totality at Babylon" (split by a page break in the PDF); "perhaps the finest accounts" quoted with its hedge - CONFIRMED | [note-454] SMH: 'The lower DT limit is critical.' | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC5247521/supplementaryFiles (rspa20160404supp1.pdf, read with pdftotext) | Verbatim in the PDF text (DT for the source's Delta T) - CONFIRMED | [note-761] SMH: 'the upper limit being critical.' | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC5247521/supplementaryFiles (rspa20160404supp1.pdf, read with pdftotext) | Verbatim in the PDF text (DT for the source's Delta T) - CONFIRMED | [note-1124] SMH: 'the lower limit of 980s is fairly critical.' | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC5247521/supplementaryFiles (rspa20160404supp1.pdf, read with pdftotext) | Verbatim in the PDF text (DT for the source's Delta T) - CONFIRMED | [note-1241] SMH 2016: combined 620 < DT < 1380, 'the lower limit is critical'. | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC5247521/supplementaryFiles (rspa20160404supp1.pdf, read with pdftotext) | Verbatim in the PDF text (DT for the source's Delta T) - CONFIRMED | [note-1361] which requires DT < 500 s or DT > 1760 s | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC5247521/supplementaryFiles (rspa20160404supp1.pdf, read with pdftotext) | Verbatim in the PDF text (DT for the source's Delta T) - CONFIRMED | [note-1560] DT = 220 s gives zero duration at Coimbra, 210 s gives 17 s, 200 s gives 53 s | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC5247521/supplementaryFiles (rspa20160404supp1.pdf, read with pdftotext) | Verbatim in the PDF text (DT for the source's Delta T) - CONFIRMED | [note-1567] Allowing for the lunar limb profile, we estimate that only values of DT between 145s and 165s would satisfy the observation. | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC5247521/supplementaryFiles (rspa20160404supp1.pdf, read with pdftotext) | Verbatim in the PDF text (DT for the source's Delta T) - CONFIRMED | [note-71-swap] The 2016 ASCII table S10 prints this row's upper and lower columns the wrong way round (9440 in the DT-U column, 9830 in DT-L) | https://web.archive.org/web/20200127134915id_/http://astro.ukho.gov.uk/nao/lvm/Table-S10.txt | Row "10 71 9440 9830 Greek" under DT-U, DT-L; S3 says "9440s to 9830s"; 2020 table "+9830 +9440" - CONFIRMED | [pub-2016-bounds] Totality at the recorded place requires | https://web.archive.org/web/20200127134915id_/http://astro.ukho.gov.uk/nao/lvm/Table-S10.txt | All 12 rows match data.json published_2016 (71 un-swapped); 1361 (500, 1760) is from Table S11 of the same supplement, "[ 1760] 500" - CONFIRMED | [pub-2020-bounds] revised in 2020 to | https://web.archive.org/web/20211018222830id_/https://astro.ukho.gov.uk/nao/lvm/Table-S10.2020.txt | 454 +7800 +6030, 1133 +1115 +530, 1241 +1005 +625; the other ten records equal their 2016 values, so "revised" is printed only for these three - CONFIRMED | [rel-day-iers] on the modern record, 23,560 daily IERS measurements since 1962 | src/content/strata/the-day-is-not-86400.md | Matches the sibling layer's own frontmatter at e2903f827c - CONFIRMED | [rel-day-wrong-sign] a straight line through the best rotation data ever taken gives a tidal braking rate with the wrong sign | src/content/strata/the-day-is-not-86400.md | Matches the sibling layer's own frontmatter at e2903f827c - CONFIRMED | [rel-day-64] Sixty-four years is far too short a lever for a term that grows as the square of the time | src/content/strata/the-day-is-not-86400.md | "sixty-four years" there; 1962 to 2026 - CONFIRMED | [rel-foucault] Foucault hung a pendulum in Paris to make the Earth's rotation visible in a single afternoon | https://en.wikipedia.org/wiki/Foucault_pendulum | Paris, 1851; the precession (about 11 deg/hour at Paris) is visible within an afternoon - CONFIRMED | [rel-foucault-sine] the turn rate is the spin times the sine of your latitude | https://en.wikipedia.org/wiki/Foucault_pendulum | "proportional to the sine of the latitude" - CONFIRMED | [rel-saros-223] The Antikythera mechanism predicts eclipses from the Saros, 223 lunar months cut into bronze | https://en.wikipedia.org/wiki/Antikythera_mechanism | Saros dial, 223 synodic months - CONFIRMED | [rel-saros-days] the Saros is 6585.32 days, not a whole number of them | https://en.wikipedia.org/wiki/Saros_(astronomy) | 6,585.321347 days - CONFIRMED | [rel-saros-120] The extra third of a day shifts each repeat 120 degrees west. | https://en.wikipedia.org/wiki/Saros_(astronomy) | "shift westward about 120 deg" - CONFIRMED | [rel-saros-config] the Saros works because the Sun, Moon and node return to nearly the same configuration | https://en.wikipedia.org/wiki/Saros_(astronomy) | 223 synodic, nearly 242 draconic and 239 anomalistic months - CONFIRMED | [rel-ptolemy] Ptolemy's star catalogue is asked what year its coordinates belong to | src/content/strata/ask-a-catalogue-what-year-it-is.md | Matches the sibling layer's own frontmatter at e2903f827c - CONFIRMED | [rel-calendar-fraction] a calendar rule is a fraction standing in for a year that is not a number | src/content/strata/no-year-to-round-to.md | Matches the sibling layer's own frontmatter at e2903f827c - CONFIRMED | [rel-gregorian-reorder] That page shows the Gregorian rule's accuracy reordering when you change which tropical year you aim at | src/content/strata/no-year-to-round-to.md | Matches the sibling layer's own frontmatter at e2903f827c - CONFIRMED | [rel-15deg] this one shows an eclipse's computed longitude sliding 15 degrees for every hour of assumed clock error | public/strata/where-the-shadow-fell/data.json | 0.00417807 x 3600 = 15.04 deg per hour - CONFIRMED | [rel-2700-years] separated by 2,700 years of arm | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC5247521/fullTextXML | "over the past 2700 years"; the lever arm 709 BC to today is 2,735 years - CONFIRMED | [dek-every-path] Every path on this page was recomputed here from JPL DE441. | public/strata/where-the-shadow-fell/data.json | All 13 records are before 1960, so their paths use DE441 part 1 (compute.py line 263 E = E441) - CONFIRMED | [jsonld-thirteen] Thirteen untimed eclipse records from Babylon, China, Greece and medieval Europe | public/strata/where-the-shadow-fell/data.json | 13 records; cultures China, Babylon, Greece, Europe (SMH call their pre-1600 set "ancient and medieval"). Its "reproduce the published bounds" is in the seo-1.79 MINOR - CONFIRMED | [meta-4590] the shadow lands on Morocco, 4,590 km west | public/strata/where-the-shadow-fell/data.json | 4591.64 rounded to 10 km - MINOR | [jsonld-datemodified] "dateModified":"2026-08-21" | git show 91d91ec0408:public/strata/where-the-shadow-fell/index.html | The page body changed on 2026-09-28 (91d91ec0408, the dated Corrected line) and 2026-10-01 (2e60bcb5dd1); JSON-LD still says "dateModified":"2026-08-21" (index.html line 31) - CONFIRMED | [fit-c] the coefficient comes out | public/strata/where-the-shadow-fell/data.json | 32.7686 printed 32.77 - CONFIRMED | [fit-apex-fixed] Hold the apex where the published work holds it and it is | public/strata/where-the-shadow-fell/data.json | a fixed at -320: c 32.54, 1.78 ms/cy - CONFIRMED | [fit-less-than-moon] ms per century less than the Moon alone can account for | public/strata/where-the-shadow-fell/data.json | 2.2998 - 1.7943 = 0.5055, printed 0.51; SMH 2.3 - 1.78 = 0.52 - MINOR | [repro-ten-records] records a spherical Moon can decide | public/strata/where-the-shadow-fell/data.json | index.html line 696 calls the 10 fitted records "the records a spherical Moon can decide", but of the three left out only 1567 is a limb case: -600 is left out because SMH reject it (its windows are computed, residuals +20/+60 s) and 1361 is an excluded-zone record the sphere decides fine (residuals +7.4/+6.7 s) - CONFIRMED | [ctl-nineteen] so this is two independent computations of the same nineteen eclipses | public/strata/where-the-shadow-fell/data.json | 19 controls: the 13 records plus 1919, 1955, 1999, 2017, 2024, 2027; every NASA column checked against the catalogue pages - MINOR | [cav-one-unreproduced] it is the one record whose published bound is not reproduced | public/strata/where-the-shadow-fell/data.json | index.html lines 277-278 and 758-759 call 1567 "the one record whose published bound is not reproduced", but by the page's own 25 s yardstick six bounds on -708 (+87, +87 s), -180 (+54, +61 s) and -135 (+29, +35 s) are not reproduced either (and -600 +60 s). 1567 is the only one off for a modelling reason; say that - CONFIRMED | [greatest-tt] Greatest eclipse | public/strata/where-the-shadow-fell/data.json | greatest_tt_hms for 13 records; e.g. -708 12:30:20.13 vs NASA 12:28:51 (89 s, the stated worst control) - CONFIRMED | [sites-coordinates] World map showing the recomputed path of the Moon's umbra and the place that recorded the eclipse | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC5247521/supplementaryFiles (rspa20160404supp1.pdf, read with pdftotext) | All 15 site coordinates in data.json match the supplement (Qifu 35.53/117.02 spelled Qufu on the page; Chang'an 34.35/108.88 for -180 and 34.27/108.90 for 761; Coimbra 40.22 N 8.42 W; Rome 41.90/12.48) - WRONG | (added) The altitude of the Sun at maximum is printed for each record so you can judge. | public/strata/where-the-shadow-fell/index.html | Nothing on the page prints a solar altitude: data.json has no altitude field, and buildProse assigns "var alt = r.sites[0];" (line 773) and never uses it. The generator computed sun_altitude_at_max_deg (research/where-the-shadow-fell/data/computed.json, e.g. 4.27 deg at Songjiang in 1361) but figures.py did not ship it. Claim at index.html lines 281-282 - MINOR | (added) Thirteen records, twenty-seven centuries | public/strata/where-the-shadow-fell/data.json | Section heading, index.html line 205. The 13 records run from 709 BC to AD 1567, 2,275 years (about 23 centuries); 27 centuries is the span of SMH's whole dataset (720 BC to AD 2015) or the arm to today - MINOR | (added) Violet: the bounds published in Stephenson, Morrison and Hohenkerk (2016), table S10. | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC5247521/supplementaryFiles (rspa20160404supp2.zip, Table-S10.txt and Table-S11.txt) | Plot caption (index.html line 674). The violet 1361 bar (500 to 1760 s) is not in the 2016 table S10; it is table S11 (untimed partial eclipses), "1361 [ 1760] 500 China" - MINOR | (added) Two independent chronicles, 180 km apart, each on a different edge of the same shadow: the pair is what makes the bound two-sided. | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC5247521/supplementaryFiles (rspa20160404supp1.pdf, read with pdftotext) | Record 1133 note (data.json; research/where-the-shadow-fell/data/records.json line 156). 180 km holds (182.5 km recomputed). But each site alone gives a two-sided window: SMH "For totality at Reichersberg, 520s < Delta T < 1740s, while for totality at Augsburg -60s < DT < 1160s"; the page's own windows are 533-1733 and -48-1154. The pair makes the bound narrower, not two-sided - CONFIRMED | (added) SMH 2016 could not satisfy this record together with -708 and -694 without a variation in the length of day they judged geophysically implausible, and it lacks the independent Hanshu confirmation that -708 has. Listed in table S10 but not fitted. | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC5247521/fullTextXML | Section 4(b): "To accommodate the result from -600, along with those of -708 and -694, requires a large variation in the lod, which is geophysically implausible"; "this report does not have confirmatory evidence from the Hanshu, as distinct from -708"; -600 is in Table S10 and in S4 "Rejected eclipses" - CONFIRMED | (added) SMH's bound is for totality anywhere on mainland Greece between Chaeroneia and Athens, not at a single point, and the upper bound is the operative one. | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC5247521/supplementaryFiles (rspa20160404supp1.pdf, read with pdftotext) | Entry 71: "total on mainland Greece somewhere between Chaeroneia and Athens ... For a value in excess of 9830s the eclipse would not have been total anywhere on mainland Greece, and we regard this as an upper bound" - CONFIRMED | (added) Site notes: capital of the state of Lu; Han capital; Tang capital; modern Nanjing, then the capital; the Collegio Romano | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC5247521/supplementaryFiles (rspa20160404supp1.pdf, read with pdftotext) | "The Lu capital was Qifu"; "the capital of Chang'an" (-180); "The Tang capital ... slightly to the south of the Han capital"; "the capital was located at Jiankang (Nanjing ...)"; "teaching at the Collegio Romano" - CONFIRMED | (added) Plutarch, De Facie, trans. Cherniss and Helmbold | https://en.wikipedia.org/wiki/William_Clark_Helmbold | Loeb Moralia XII (LCL 406) is translated by Harold Cherniss and W. C. Helmbold; the supplement misspells "Chermiss", the page has it right - CONFIRMED | (added) Control table rows for 1919, 1955, 1999, 2017, 2024 and 2027 (catalogue no., NASA TD, NASA width, NASA duration) | https://eclipse.gsfc.nasa.gov/SEcat5/SE1901-2000.html and https://eclipse.gsfc.nasa.gov/SEcat5/SE2001-2100.html | 09326 13:08:55 244 06m51s; 09410 04:10:42 254 07m08s; 09506 11:04:09 112 02m23s; 09546 18:26:40 115 02m40s; 09561 18:18:29 198 04m28s; 09568 10:07:50 258 06m23s: all equal data.json - CONFIRMED | (added) Swap in the IAU's k = 0.2725076 and every duration lengthens by about four seconds. | public/strata/where-the-shadow-fell/data.json | Recomputed over the 19 controls: computed_duration_s_iau minus computed_duration_s_espenak runs 3.5 to 5.1 s, median 4.3 s, every one positive - CONFIRMED | (added) The 2020 revision narrows the upper bound to 1005 s. | https://web.archive.org/web/20211018222830id_/https://astro.ukho.gov.uk/nao/lvm/Table-S10.2020.txt | "10 1241 +1005 +625 Europe" against 2016 "1380 620" (record 1241 note) - CONFIRMED | (added) At Delta T = 0 both places still see an eclipse | public/strata/where-the-shadow-fell/data.json | mag at grid index 80 (Delta T = 0): Babylon 0.149, Qufu 0.282 ## Findings - [fixed] dek-oldest (WRONG): both deks called the 709 BC Chunqiu entry "the oldest surviving measurement of how fast the Earth turns"; SMH table S4 holds Delta T from Babylonian lunar eclipses of 721 and 720 BC. Both deks now say it is "the earliest direct record of a total eclipse of the Sun" (SMH's own description); Corrected 2026-10-18 line on the page and in the markdown body. The share image (public/og/where-the-shadow-fell.png) printed the same dek sentence; share card rebuilt at landing from the corrected dek. - [fixed] dek-six-words (MINOR): "Six words" counted only with 朔 included; both deks now say "A few words", fixed because it sits in the K1 sentence; no line on the page. - [fixed] dek-babylon-finest (MINOR): the frontmatter dek made the 136 BC tablet "the finest" measurement; SMH say "perhaps the finest accounts of a total solar eclipse from antiquity". It now says "Babylonian tablets from 136 BC hold perhaps the finest accounts of one from antiquity", fixed because it sits in the K1 sentence; no line on the page. - [declined] and in the eastern Mediterranean for the Chinese one, 7,841 km away (MINOR, qufu-km-along-parallel): listed, page not edited at tier B. From the checker's note: 7,841 km is data.json km_along_parallel (distance along the 35.53 N parallel), the great-circle separation is about 7,540 km. A slip of precision in a separate dek sentence and in the body, not in a sentence fixed for a WRONG item; tier B MINOR: list. - [fixed] dek-only-instrument (WRONG): the frontmatter dek said untimed eclipse records are the only pre-telescopic instrument; SMH: "By far the most accurate data for measuring the Earth's rotation before ... ca AD 1620 are records of eclipses", most of them timed. It now says "before telescopes, records of eclipses are by far the most accurate instrument for measuring it"; Corrected 2026-10-18 line. - [declined] the shortfall is the part the Moon cannot pay for, and is charged to a planet still rising from the last ice age (MINOR, shortfall-all-rebound): listed, page not edited at tier B. From the checker's note: SMH section 6 say rebound "cannot account completely" and core-mantle coupling is also required; the dek charges the shortfall to rebound alone. Loose wording in a dek sentence not otherwise being fixed; tier B MINOR: list. - [declined] Thirteen eclipses recomputed from JPL DE441, reproducing the published bounds and giving 1.79 ms per century. (MINOR, seo-reproducing-bounds): listed, page not edited at tier B. From the checker's note: seoDescription and JSON-LD description say the bounds are reproduced, unqualified, while the page reports 14 of 20 within 25 s and 1567 not reproduced. An overstatement of the page's own reproduction, MINOR; tier B: list. - [declined] their tablets are, in the judgement of the people who read them, the finest description of a total eclipse to survive from antiquity (MINOR, babylon-finest-unhedged): listed, page not edited at tier B. From the checker's note: the body drops SMH's "perhaps" from "perhaps the finest accounts" (index.html lines 181-183). MINOR, not in a sentence fixed for a WRONG item; tier B: list. - [fixed] every-record-below-tidal (WRONG): "Every single record sits below it" was false; data.json puts the tidal-only curve inside the windows of AD 454 (7837 s in 6097-7904) and AD 1124 (2035 s in 984-2710). The page now says ten of the thirteen bars sit wholly below, names 454 and 1124, and says the 1361 bar it crosses is the band that record rules out; Corrected 2026-10-18 line. - [declined] Stephenson, Morrison and Hohenkerk attribute most of the difference to the solid Earth still rebounding from the weight of the last ice age, (MINOR, rebound-most): listed, page not edited at tier B. From the checker's note: SMH say "probably in part associated with" rebound, not "most". Loose attribution wording, MINOR; tier B: list. - [declined] theirs used their own analytical lunar theory and their own software, built from chapter 8 of the Explanatory Supplement (MINOR, own-lunar-theory): listed, page not edited at tier B. From the checker's note: SMH used the Explanatory Supplement's analytical ephemeris with their own tidal term, not "their own analytical lunar theory". MINOR; tier B: list. - [declined] because it rotates every apparent direction by the same angle and so cannot change whether one disc covers another (MINOR, aberration-rigid-rotation): listed, page not edited at tier B. From the checker's note: aberration is not a rigid rotation; the conclusion holds because the two discs are within half a degree. Wrong reason, right conclusion, MINOR; tier B: list. - [declined] Generator, verifier and every intermediate artefact are in (MINOR, repro-files-location): listed, page not edited at tier B. From the checker's note: the verifier is at the repository root, some inputs are not committed, and the repository is not public. MINOR; tier B: list. - [fixed] jsonld-datemodified (MINOR): JSON-LD dateModified moved from 2026-08-21 to 2026-10-18 with this revision; no line on the page. - [declined] records a spherical Moon can decide (MINOR, spherical-moon-can-decide): listed, page not edited at tier B. From the checker's note: of the three records left out of the fit only 1567 is a limb case. MINOR; tier B: list. - [declined] it is the one record whose published bound is not reproduced (MINOR, one-unreproduced): listed, page not edited at tier B. From the checker's note: by the page's own 25 s yardstick six bounds on -708, -180 and -135 are also off; 1567 is the only one off for a modelling reason. MINOR; tier B: list. - [fixed] altitude-printed (WRONG): the caveat said the solar altitude at maximum "is printed for each record"; nothing printed it (buildProse's "var alt" is unused, data.json has no altitude). It now says the altitude is not printed, points to research/where-the-shadow-fell/data/computed.json (sun_altitude_at_max_deg, written by compute.py line 291) and gives the lowest, 1361 at Songjiang, 4.27 deg; Corrected 2026-10-18 line. - [declined] (added) Thirteen records, twenty-seven centuries (MINOR, twenty-seven-centuries): listed, page not edited at tier B. From the checker's note: the thirteen records span 709 BC to AD 1567, about 23 centuries; 27 is SMH's whole span. MINOR heading; tier B: list. - [declined] (added) Violet: the bounds published in Stephenson, Morrison and Hohenkerk (2016), table S10. (MINOR, violet-table-s10): listed, page not edited at tier B. From the checker's note, confirmed in supplement Table-S11.txt line 23 ("11 1361 [ 1760] 500 China"): the 1361 violet bar comes from table S11, not S10. MINOR caption slip; tier B: list. - [declined] (added) Two independent chronicles, 180 km apart, each on a different edge of the same shadow: the pair is what makes the bound two-sided. (MINOR, pair-makes-two-sided): listed, page not edited at tier B. From the checker's note: each 1133 site alone gives a two-sided window; the pair narrows the bound. MINOR; tier B: list.