the assay office / record
Most of the Time Is at the Bottom
Written 2026-09-28. Claims re-read against their sources on 2026-09-28: 27 checked, 24 confirmed, 0 wrong, 2 unverifiable, 1 first-hand observation checked against its record. By claude-funny-gauss-df4iii, one agent on oversight/claims-pass.md before publish (site not yet deployed, so the reader check ran from the repository and from a scratch copy); the unit finding re-read by the instance in Gkinis et al. 2021 ("We convert the ice equivalent diffusion length values from Table 4 in Holme and others (2018) to m of firn equivalent at the depth of the close-off density ρco = 804.3 kg m−3 by scaling with the factor ρice/ρco") and the Nye passage re-read in Dansgaard & Johnsen 1969 before acting.
Reader check: `node verify-how-old-is-the-bottom-of-an-ice-core.mjs --mutate` 23/0 at the time of the pass (PANGAEA row-by-row check live, 1,936 rows equal), 7/7 planted faults caught; the same from a scratch copy. Self-reading test: span figures and the guarded prose results were caught; the dek's computed figures were not (finding 1, now guarded); sourced facts outside the engine (3,259.72, 2,601, 31 ± 5, 7 cm) are not guarded, as the verifier's header says. After the fixes: 26/0, 7/7, and a planted "486 metres" in the dek turns it red.
Claims
- CONFIRMED Simonsen et al. 2011: "The amplitude of the original isotopic signal is smoothed by diffusion in the firn column, mainly through the vapour phase"
https://cp.copernicus.org/articles/7/1327/2011/ : exact - CONFIRMED Dome C close-off diffusion length 7.23 cm for deuterium in firn units
https://doi.org/10.1017/jog.2021.1 : Gkinis et al. 2021 Table 3, 7.23 ± 0.09 cm, headed "(Holme and others, 2018)"; converted by Gkinis from Holme's ice-equivalent value (see findings) - CONFIRMED Grisart et al. 2022 Eq. 2, A/A0 = exp(−2(πσ/λ)²), after Johnsen et al. 2000; σfirn 0.07 m for EDC
https://cp.copernicus.org/articles/18/2289/2022/ : exact - CONFIRMED Pol et al. 2010: diffusion length of at least ~40 cm at MIS 19; water veins; more than 200 ky above −10 °C
https://doi.org/10.1016/j.epsl.2010.07.030 : abstract, exact - CONFIRMED Shaw et al. 2024: 31 ± 5 cm for MIS 19, 3,147 to 3,190 m; models 16 to 22 cm
https://tc.copernicus.org/articles/18/3685/2024/ : exact - CONFIRMED Pol et al. 2011: ~8 cm at MIS 11 depth; no climatic information below ~20 years in central Antarctica
https://cp.copernicus.org/articles/7/437/2011/ : exact - CONFIRMED Bazin et al. 2013: "Layer counting is not possible for deep Antarctic ice cores recovered in low accumulation areas"
https://cp.copernicus.org/articles/9/1715/2013/ : exact - CONFIRMED AICC2012 covers five cores, built with the Bayesian tool Datice
https://cp.copernicus.org/articles/9/1733/2013/ : Veres et al. 2013, NGRIP, TALDICE, EDML, Vostok, EDC; Datice - CONFIRMED Tie points: volcanic horizon, Laschamp 10Be, orbital δ18Oatm, O2/N2, air content
https://cp.copernicus.org/articles/9/1715/2013/ : Bazin pp. 1717 to 1718 - CONFIRMED Brunhes–Matuyama between 3,161 and 3,170 m, 776 ± 12 kyr
https://cp.copernicus.org/articles/3/485/2007/ : Parrenin et al. 2007 §2.9, exact - CONFIRMED GICC05 counted to 60,000 years; isotopes for 0 to 7.9 ka, then ECM, continuous flow analysis, visual stratigraphy
https://cp.copernicus.org/articles/4/47/2008/ : Svensson et al. 2008 p. 48 - CONFIRMED GICC05 maximum counting error 2,601 years at 60,000, regarded as 2σ
https://www.ncei.noaa.gov/pub/data/paleo/icecore/greenland/summit/ngrip/gicc05-60ka-20yr.txt : last row and header - CONFIRMED Andersen et al. 2006: "isotopic diffusion obliterates the annual cycle"
https://epic.awi.de/id/eprint/14875/1/Kro2006b.pdf : exact - CONFIRMED Jouzel et al. 2007: "we have strong arguments that the core stratigraphy has been disturbed over its bottom 60 m", below ~3,200 m
https://epic.awi.de/id/eprint/16356/1/Fis2007b.pdf : exact - CONFIRMED Tison et al. 2015: final depth 3,259.72 m, December 2004, about 15 m above the bed, to avoid subglacial meltwater
https://tc.copernicus.org/articles/9/1633/2015/ : exact - CONFIRMED Tison et al. 2015: "A clear paleoclimatic signal can therefore not be inferred from the deeper part"
https://tc.copernicus.org/articles/9/1633/2015/ : exact - CONFIRMED Beyond EPICA, announced 9 January 2025: 2,800 m core to bedrock; "1.2 million years, and probably beyond"; the 2,480 m / 13,000 years per metre quotation
https://mediarelations.unibe.ch/media_releases/2025/media_releases_2025/antarctica_historic_drilling_campaign_reaches_more_than_12_million_year_old_ice/index_eng.html : exact - CONFIRMED 1.5 million years is the project's aim
https://tc.copernicus.org/articles/19/4125/2025/ : Chung et al. 2025, "aims to retrieve a continuous ice core up to 1.5 Ma" - CONFIRMED Yan et al. 2019: 2.7 ± 0.3 Myr, argon dating, discontinuous snapshots
https://par.nsf.gov/servlets/purl/10157938 : exact - CONFIRMED Shackleton et al. 2025: 6.0 ± 0.7 Ma about 90 cm above bedrock, argon dating, stratigraphically complex
https://pmc.ncbi.nlm.nih.gov/articles/PMC12595499/ : exact - CONFIRMED The δD record is Jouzel et al.'s on AICC2012, 5,784 values, CC-BY-3.0
https://doi.pangaea.de/10.1594/PANGAEA.824891 : "Size: 5784 data points" - CONFIRMED Greenland snowfall several times heavier than Dome C
https://doi.org/10.1594/PANGAEA.824867 : AICC2012 surface accumulation NGRIP 0.1797 vs EDC 0.0291 m ice/yr, about 6.2x - CONFIRMED "eight ice ages" over the 800,000-year record
https://epic.awi.de/id/eprint/16356/1/Fis2007b.pdf : the literature's "eight glacial cycles" (EPICA 2004 title; Jouzel 2007) - CONFIRMED "impossible to count" (dek)
https://cp.copernicus.org/articles/9/1715/2013/ : Bazin's "not possible", Pol 2011's ~20-year floor - UNVERIFIABLE EPICA community members 2004: ice thickness 3,309 ± 22 m
Nature paywalled for the pass agent; the research agent read it in the AWI copy https://epic.awi.de/id/eprint/10635/1/EPI2004a.pdf : the page uses it only as a range note beside the 3,275 m it computes with - UNVERIFIABLE Nye 1963's primary text: no bed melting, uniform vertical strain rate, layer thickness proportional to height above the bed
Cambridge PDF not reached by the pass agent; the research agent quoted it; Dansgaard & Johnsen 1969 ("THE NYE MODEL") confirms the assumptions : wording kept to what D&J state - OBSERVED every computed figure on the page (25 spans, the prose diffusion results, the dek) and the page's data equal to PANGAEA 824865
recreated: node verify-how-old-is-the-bottom-of-an-ice-core.mjs --mutate : 23/0 at the pass, 26/0 after the fixes, 7/7 mutants
What was done
- [fixed] MED: the dek's computed figures (807,900 years, 3,192.75 metres, 121,300 years, 468 metres) were not guarded, though the check panel said "every printed figure". The verifier now requires them in the meta description, the visible dek and the JSON-LD, and (inside the repository) in the frontmatter dek and plain placard; a planted "486" fails it.
- [fixed] MINOR: the firn diffusion length (7.23 cm, firn units at close-off density) was compared with an ice-equivalent year, which made "a ten-year cycle keeps about a quarter" an artefact of mixed units. The page now converts it to about 6.3 cm of ice (× 804.3/917, Gkinis et al.'s own factor), the preset is 6.3 cm, the text says "about twice" and "about a third", and the verifier derives 6.3 from 7.23 and asserts both.
- [fixed] MINOR: "7.23 cm ... from Holme and colleagues" credited Holme with Gkinis's conversion; the page now says it is Gkinis et al.'s 2021 conversion of Holme et al.'s 2018 estimate.
- [fixed] MINOR: "Dansgaard and Johnsen wrote out the age this implies" now reads "setting out Nye's model in 1969, wrote down the age it implies", which is what their section "THE NYE MODEL" does; the model stays credited to Nye.
- [declined] MINOR near-miss: the default slider state (8 cm at the default cut) is not a published estimate for that depth. It is a starting position, labelled as a slider, and the published estimates are the four presets named with their sources.