The Literature on Radiometric Concordance

Every source the main article rests on, with what it actually measured and which side it helps.  Sources are grouped by the role they play in the argument, not by author, because several authors appear on both sides of the same question — and two of the most useful young-earth sources are useful precisely because they report results against their own thesis.

Everything below was downloaded and read as bytes except where marked not obtained.  Those are collected at the end as well, because an argument resting on a source nobody on this side has read should be flagged rather than buried.  No quotation in the article or in this page was established through a page-summarising tool;  where a fact came from an abstract rather than a full text, the entry says so.

The exchange that framed the question

Isaac, Randy 2007, "Assessing the RATE Project," Perspectives on Science and Christian Faith 59(2):143-146.  https://www.asa3.org/ASA/PSCF/2007/PSCF6-07Isaac.pdf

The essay review that put concordance at the centre of the dispute.  Isaac, then Executive Director of the American Scientific Affiliation and a solid-state physicist with 28 years at IBM, argues that RATE "fail to invalidate the vast amount of concordance."  His weakness is that he documents none of it, and he mischaracterises RATE's discordances as "10-15 %" when they run to factors of two and three.

Which way it cuts:  against RATE in conclusion, but it hands RATE a fair procedural objection.

RATE Group 2008, "RATE Responds to the Isaac Essay Review," with "Isaac Replies," PSCF 60(1):34-38.  https://www.asa3.org/ASA/PSCF/2008/PSCF3-08Isaac.pdf

Signed collectively by Vardiman, Snelling, Chaffin, Austin, Humphreys, DeYoung and Boyd.  Contains the challenge that the article treats as its brief:  "Does he have examples of concordant isochrons between U-Pb, Sm-Sr, Rb-Sr and K-Ar in suites of earth rocks?  If Isaac could provide this documentation, he would have one of the strongest arguments in favor of the accuracy of radioisotope ages."  Also correctly corrects Isaac's 10-15 % figure.  Isaac has the last word in the venue.

Which way it cuts:  it defines the test.  Both sides agreed on what would settle it and neither ran it.

The database study, and the database

Beachy, Micah D., Benjamin R. Kinard & Paul A. Garner 2023, "How Often Do Radioisotope Ages Agree?  A Preliminary Study of 29,000 Radioisotope Ages in the USGS National Geochronological Database," Proceedings of the Ninth International Conference on Creationism, pp. 387-411.  doi:10.15385/jpicc.2023.9.1.23

The best quantitative young-earth argument in print on this subject, and the reason this article exists.  A public database, a stated algorithm, published source code in an appendix, and a record list.  Twenty-five pages.

Six things a citer needs to know.  Garner is third author, not lead;  the first two were sophomore undergraduates at Cedarville.  Garner was also an Area Editor of the volume in which it appears, undisclosed in the paper.  The volume's editor writes in his own preface that "the peer-review process was rushed in too many cases."  The Crossref and OpenAlex title contains a typo — "Radioistope" — which defeats naive searching.  The publications.cedarville.edu/…/389/ URL circulating in secondary sources is a page-viewer link, not a citable landing page;  the canonical PDF is at cedarville.tind.io/record/21195.  And the paper has zero citations in Crossref, OpenAlex and Semantic Scholar as of August 2026 — nobody, on any side, has replied to it.

The paper is more careful than its popularisers.  It calls itself "preliminary," says "there are clearly limits to what we can say with confidence about the prevalence of concordance," concedes that a database record is not a rock unit, lists five items of future work, and nowhere claims that deep time is falsified.

Which way it cuts:  for the young-earth argument at the level of the headline;  substantially against it at the level of the paper's own database-wide numbers, which report a 0.64 average concordance and 53.2 % of multiply-dated records perfectly concordant.  It also partly refutes the young-earth claim it set out to test — Kurt Wise's assertion that in this database "the methods rarely yield the same ages."

Zartman, R. E., C. A. Bush & C. Abston, rev. Sloan, Henry, Hopkins & Ludington 2003, National Geochronological Database, USGS Open-File Report 03-236.  https://pubs.usgs.gov/of/2003/0236/  Public domain.

18,576 records, about 30,000 ages, representing roughly half the ages published for the United States through 1991.  Six relational tables plus a merged export.  Now on disk in full at sources/data/usgs-2003_national-geochronological-database/ with a README documenting its traps.

Its own documentation carries the warning the article leans on:  "We did not recalculate radiometric ages using more recently accepted decay constants…  To avoid misinterpretation, check for consistency between records; some recalculation may be needed where direct comparison is intended."  It also states plainly that the report "is preliminary and has not been reviewed for conformity with USGS editorial standards."

Three traps, all found the hard way.  The .txt tables use bare carriage-return line endings, so wc -l reports zero and any line-by-line reader silently produces nothing.  647 of the 2,621 non-blank ²⁰⁷Pb/²⁰⁶Pb ages are exactly zero, which the documentation identifies as converted blanks.  And 5,509 K-Ar rows record the ⁴⁰K abundance as 1.1167E-04, which matches no published value;  recomputing 4,126 ages from the file's own potassium oxide and radiogenic argon proves the intended value is the IUGS 1.167E-04.

Which way it cuts:  neutral.  It is the data.  Its limitation is that it is a photograph of 1976 — median publication year of the underlying studies — and it has not been extended since 1994.

U.S. Geological Survey 2025, USGS Geochron, ver. 4.0.  doi:10.5066/P9RZNPIF

The successor, released June 2025, and the reason the 2023 analysis is now replicable on modern data:  76,426 usable ages, 97.1 % carrying an uncertainty against the old file's 71.0 %, structured interpretation fields, and coverage of ⁴⁰Ar/³⁹Ar, (U-Th)/He, Re-Os and Lu-Hf — none of which exist in the NGDB.  Joined on field sample name it yields 970 samples with two or more methods and 43 with three or more.

Which way it cuts:  unknown, which is exactly why the article recommends someone run it.

A warning about mrdata.usgs.gov/geochron/.  Despite a 2024 page date and a description promising regular updates, the CSV distribution served there is the same 2003 compilation.  Verified in house:  identical record counts in all six tables and zero rows differing in (record number, age), plus a metadata file named ofr-2003-236.met.  Anyone who cites it as an update is citing the 2003 file.

RATE's own primary literature

Volume II is available chapter by chapter as free PDFs from ICR, linked from icr.org/rate2;  each chapter link goes to an icr.org/article/<slug>/ page whose body contains the real PDF href.  Volume I was already in the project library.  ICC proceedings from 1986 onward are at cedarville.tind.io.

Two entries in that table deserve emphasis because they are young-earth sources reporting against the young-earth thesis.

Snelling 2015 concludes:  "No pattern was found in the isochron ages for these groups of meteorites similar to the systematic patterns of isochron ages found in Precambrian rock units during the RATE project, so there is no evidence of past accelerated radioisotope decay having occurred in these meteorites…  This is not as expected."  The rescue offered is that accelerated decay happened on Earth only.

Beachy, Kinard & Garner 2023 report the ordering Pb-Pb > Rb-Sr > ²³⁵U-²⁰⁷Pb > ²³⁸U-²⁰⁶Pb > ²³²Th-²⁰⁸Pb > K-Ar > fission track, in which a β-decaying system sits second, above three α schemes:  "in our study ⁴⁰K followed this pattern but ⁸⁷Rb did not."

Which way the RATE corpus cuts, overall:  the data are real and were honestly gathered and published, including the results that hurt.  The α-versus-β interpretation is not supported by the corpus's own contents.

The provenance of RATE's analyses

Worth recording because it answers the reflexive charge that the numbers were produced in-house.  Whole-rock geochemistry went to XRAL Laboratories (SGS Canada) and Amdel in Adelaide;  K-Ar to Geochron Laboratories in Massachusetts under Richard Reesman and Activation Laboratories in Ontario under Yakov Kapusta;  Rb-Sr, Sm-Nd and Pb-Pb to the University of Colorado at Boulder under G. Lang Farmer and to the PRISE laboratory at the Australian National University under Richard Armstrong.  Isochrons were fitted in Isoplot/Ex v2.49 with York two-error regression, and MSWD is reported for nearly every plot.  Snelling's own acknowledgment notes that the four named analysts "may not agree with some aspects of the interpretation of the data and the conclusions of this study."

The mainstream cross-method measurements

These are the article's positive case.  For each, the thing to check is not the age but which uncertainty is quoted — analytical only, or full external including tracer calibration and decay constants.

Mitchell, William S. III 2014, "High-resolution U-Pb Geochronology of Terrestrial Cretaceous-Paleogene and Permo-Triassic Boundary Sequences in North America," PhD dissertation, UC Berkeley.  https://escholarship.org/uc/item/6vv5d1nb

The U-Pb half of the strongest comparison in the article:  seven named Hell Creek coal-seam tephras, each with both a single-crystal ⁴⁰Ar/³⁹Ar sanidine age and a CA-ID-TIMS U-Pb zircon age, agreeing to 0.041-0.562 % with a mean of 0.217 %.  Mitchell states his own weaknesses:  "the majority of samples analyzed show clear and pervasive inheritance, even among the most pristine zircon grains," and "Choosing the right tephra can make a big difference, and often there is no way to tell which tephra is the right one until the mass spectrometry is underway."  The one outlier, X coal at 0.562 %, rests on two zircon grains and he flags it.

Which way it cuts:  strongly for concordance, with the author's own caveats intact.

Renne, Paul R., et al. 1997, "⁴⁰Ar/³⁹Ar Dating into the Historical Realm:  Calibration Against Pliny the Younger," Science 277(5330):1279-1280.

The closest thing to a controlled accuracy test the method has, and it cuts both ways inside two pages.  Isochron age of AD 79 Vesuvius sanidine:  1925 ± 94 yr against a true 1918 (0.36 % high, true value inside the bar).  Total-gas age of the same sanidine:  3300 ± 500 yr, 72 % too old.  Each side of the debate quotes one number.

Which way it cuts:  for the method with its internal check applied;  against the method without it.  This is the single most useful case in the article because it isolates the variable.

Min, Kyoungwon, Roland Mundil, Paul R. Renne & Kenneth R. Ludwig 2000, "A test for systematic errors in ⁴⁰Ar/³⁹Ar geochronology through comparison with U/Pb analysis of a 1.1-Ga rhyolite," GCA 64(1):73-98.

The cleanest single Ar-Ar↔U-Pb test, and the source of the "~1 % offset" figure everyone cites.  Palisade Rhyolite:  argon 1088.4 ± 4.0 Ma, uranium-lead ²⁰⁷Pb/²⁰⁶Pb 1097.6 ± 2.1 Ma, 0.84 % apart and statistically distinct on those bars — "However, including systematic errors in both U/Pb (1097.6 ± 5.3 Ma) and ⁴⁰Ar/³⁹Ar (1088.4 ± 23.3 Ma) ages shows that they are indistinguishable."

Which way it cuts:  both.  It documents a real systematic offset and shows that the field publishes it.

Kuiper, K. F., et al. 2008, "Synchronizing Rock Clocks of Earth History," Science 320(5875):500-504.

The astronomical calibration giving Fish Canyon sanidine 28.201 ± 0.046 Ma.  Contains the sentence the article uses to establish that the offset is known and signed:  "a systematic offset, in which the U/Pb ages are older by 0 to 1% than the ⁴⁰Ar/³⁹Ar ages for the same rocks."

Renne, Paul R., et al. 2010 (GCA 74:5349-5367) and 2011 (GCA 75:5097-5100).

The competing calibration, 28.294 ± 0.036 Ma, from a joint optimisation of the ⁴⁰K decay constants.  The 0.33 % gap between this and Kuiper's value is the live argument in the field, and choosing between them moves the K-Pg boundary by 0.31 % with no isotope measurement changed.

Which way these cut:  decisively for the article's central framing.  The dispute the discipline is actually having is three orders of magnitude smaller than a percent-level error and five orders smaller than what a young earth requires.

Renne, Paul R., et al. 2013, "Time Scales of Critical Events Around the Cretaceous-Paleogene Boundary," Science 339(6120):684-687.  Not obtained;  its numbers were recovered from Sprain et al. 2018 and Mitchell 2014, and its abstract verbatim from Europe PMC.

Sprain, C. J., et al. 2018, "Calibration of chron C29r," GSA Bulletin 130(9-10):1615-1644.

The argon half of the Hell Creek comparison, and the source of the only quantified selection statistic in the article:  Table 2 reports, per sample, how many single-crystal fusions were used out of how many were run — 1,131 of 1,315 retained, 14.0 % rejected, with one sample retaining 13 of 69.

Which way it cuts:  for concordance on the ages;  for the sceptic on the file-drawer question, and it is to the authors' credit that the number is computable at all.

Burgess, Seth D., Samuel Bowring & Shuzhong Shen 2014, "High-precision timeline for Earth's most severe extinction," PNAS 111(9):3316-3321.

The most precise date in the article — Permian-Triassic boundary at 251.902 ± 0.024 Ma — and the clearest statement anywhere of which error bar to use when:  "If compared with other chronometers such as Ar-Ar or astrochronology, then ±z should be used."  Analytical uncertainty 0.015 %, full external 0.11 %.

Mohr, Michael T., et al. 2024, "High-precision U-Pb geochronology links magmatism in the Southwestern Laurentia large igneous province and Midcontinent Rift," Geology 52(3):193-197.  CC-BY.

Published nineteen years after RATE's book, and it dates RATE's flagship rock:  CA-ID-TIMS zircon at 1098.16 ± 0.59 Ma on a granophyre pod in a sill intruding the Bass Formation at Colorado River mile 107.  RATE's β systems come in 3-4 % low;  its α systems 14-26 % high.

Which way it cuts:  against RATE, and specifically against the α-versus-β direction, since the α side is the side that is wrong.

Dalrymple, G. Brent 2000, "Radiometric Dating Does Work!," Reports of the NCSE 20(3).  https://ncse.ngo/radiometric-dating-does-work

The mainstream's standard short answer.  Its data tables exist on the page only as images and have been captured and transcribed for this project.  St Severin:  four methods, six determinations, four laboratories, 4.38 ± 0.04 to 4.55 ± 0.33 Ga.  K-T boundary:  187 determinations, three minerals plus glass, three decay schemes, six laboratories, five localities, 63.9-65.1 Ma.

It overclaims and the article says so.  "The results are the same within analytical error" fails on his own Hell Creek Z-coal row — ⁴⁰Ar/³⁹Ar 64.8 ± 0.1 Ma against a 26-point Rb-Sr isochron at 63.7 ± 0.6 Ma, 1.7 % apart with no overlap.  His Figure 1 biotite/sanidine pairs are a cross-mineral check inside one method, not a cross-method check, and the piece does not say so.  And the St Severin Sm-Nd isochron at ±0.33 Ga is a ±7 % constraint being described as agreement "to within a few percent."

Which way it cuts:  for concordance on the data;  against the mainstream on rhetorical discipline.  A hostile young-earth reader will notice these;  the article notices first.

Wiens, Roger C. 2002, "Radiometric Dating:  A Christian Perspective," American Scientific Affiliation.  https://www.asa3.org/ASA/resources/Wiens.html

The piece most often handed to a doubting Christian.  Its three tables are half-lives, extinct radionuclides and ice-core layer counts — there is no same-rock cross-method data table in it at all, and its central quantitative claim ("usually a few percent, not orders of magnitude!") carries no citation.

Which way it cuts:  a good explainer and a bad citation.  Do not lean on it as evidence.

Method mechanics and community standards

Condon, Daniel, et al. (37 authors) 2024, "Recommendations for the reporting and interpretation of isotope dilution U-Pb geochronological information," GSA Bulletin 136:4233-4251.

The most candid mainstream source in the whole article, and the one a sceptic should read first.  Three admissions that the article quotes:  lead loss "may be undetectable … in rocks less than ca. 300 Ma, where Pb-loss trajectories parallel the concordia curve";  chemical abrasion "is imperfect";  and — the circularity concession — "decay constants used in other radioisotopic dating methods (e.g., Rb-Sr, ⁴⁰Ar/³⁹Ar, and Re-Os) have been intercalibrated with U-Pb dates from the same geological materials."  It also records the unexplained ~0.3 % systematic U-Pb discordance across closed-system suites spanning three billion years.

Which way it cuts:  for the sceptic on three specific technical points, and for the discipline on the meta-point that it publishes them.

Vermeesch, Pieter 2025, "High MSWDs are not the problem, low ones are," National Science Review 12:nwaf036.  CC-BY.

A mainstream geochronologist making a publication-bias argument in a top-tier venue:  "High MSWD datasets are virtually absent from high-impact publications," reviewers deem them "unpublishable," and this "may tempt some scientists to 'cherry pick' data."  His conclusion is that high MSWDs "are the rule, not the exception."

Which way it cuts:  for the sceptic, and the young-earth literature has not yet found it.  It is the strongest available version of the file-drawer charge and it comes from inside.

Wendt, I. & C. Carl 1991, Chemical Geology 86:275-285, and Wendt, I. 1993, Chemical Geology 104:301-305.

The two papers behind the article's statistics section:  the MSWD acceptance window 1 + 2√(2/f), and the test that separates a mixing line from an isochron.  Wendt publishes his own test's failure mode:  "If the Rb and Sr concentrations of the samples are well correlated, however, the data points are arranged on a straight line in the x,y,z-space at any time and an appearantly 'perfect' isochron with a possibly wrong age is obtained."

Zheng, Yong-Fei 1989, "Influences of the nature of the initial Rb-Sr system on isochron validity," Chemical Geology:  Isotope Geoscience Section 80:1-16.

The paper RATE leans on, reported in Snelling's own words rather than quoted;  the abstract sentence the article uses is verified at the publisher record.  Two facts the young-earth citations omit:  the qualifier "in the light of Rb-Sr isotopic data alone" is inside the quoted sentence, which makes it an argument for cross-method checking;  and Zheng is a Chinese Academy of Sciences geochemist who co-authored the characterisation of the 91500 zircon U-Pb reference material used worldwide, and accepts deep time without qualification.

Field, D. & A. Råheim 1979, "A geologically meaningless Rb-Sr total rock isochron," Nature 282:497-499.

A clean line — 1259 ± 26 Myr, MSWD 1.58 — for an event that never happened.  The mixing objection is mainstream, and it was published in Nature by geologists.

Which way it cuts:  for the young-earth objection, and the article concedes it in those terms.

Dodson, M. H. 1973, Contributions to Mineralogy and Petrology 40:259-274;  Nebel, O. 2014, "Rb-Sr Dating," Encyclopedia of Scientific Dating Methods pp. 687-694.

Closure temperature, defined and worked.  Nebel supplies the concrete case:  the Great Dyke of Zimbabwe cooled slowly enough that Rb-Sr ages come out ~23 Myr younger than U-Pb on the same rocks.

Jaffey, A. H., et al. 1971, Physical Review C 4:1889-1906;  Hiess, J., et al. 2012, Science 335:1610-1614;  Mattinson, J. M. 2005, Chemical Geology 220:47-66;  Schaen, A. J., et al. 2021, GSA Bulletin 133:461-487;  Schoene, B. 2014, Treatise on Geochemistry 4:341-378.

The technical underpinnings:  the uranium half-lives and their 0.107 % / 0.137 % uncertainties;  the ²³⁸U/²³⁵U revision;  chemical abrasion;  argon reporting standards;  and modern ID-TIMS precision ("about 0.2 % of the reported date, but 0.05 % is routinely achieved in some labs").

New Mexico Geochronology Research Laboratory, "K/Ar and ⁴⁰Ar/³⁹Ar Methods."  https://geoinfo.nmt.edu/labs/argon/methods/home.html

A working laboratory's own teaching pages, and unusually candid.  Source of the standard-calibration admission the article quotes, and of a worked inverse-isochron example in which a ~28 % error is found and corrected.  Note:  the "Damon et al." passage attributed to this page in this project's older research notes is not present on the current version.

Failure modes and inter-laboratory work

Dalrymple, G. Brent 1969, "⁴⁰Ar/³⁶Ar analyses of historic lava flows," EPSL 6(1):47-55.  NOT OBTAINED — see the gap list.

The most-cited paper in the young-earth case and the one nobody in this project has read.  Its abstract establishes that of 26 historic flows, five carried excess ⁴⁰Ar (too old) and three carried excess ³⁶Ar (too young) — the second group being the one the young-earth literature never mentions.

Henke, Kevin R. 2003, "Blind Leading the Blind."  https://www.noanswersingenesis.org.au/blind_leading_blind_kh.htm

Alleges that Austin, Snelling and Swenson printed Dalrymple's excess-⁴⁰Ar concentrations, in units of 10⁻¹² mol/g, as though they were ages in millions of years.  Six of the seven disputed numbers fall inside the excess-argon range Dalrymple's own abstract states, which is strong circumstantial support — but the charge cannot be closed without Table 2.

Which way it cuts:  probably decisively against a widely repeated young-earth table, and the article deliberately declines to assert it as established.

Dalrymple, G. Brent 1986, USGS Open-File Report 86-110, 76 pp.  Public domain.  https://pubs.usgs.gov/of/1986/0110/report.pdf

The best single mainstream response to young-earth dating arguments, and public domain so it can be quoted at length.  Contains the Hualalai xenolith explanation verbatim, the admission that the 1801 flow "is so badly contaminated by the xenoliths that it is not possible to obtain a completely inclusion-free sample," and the statement that the submarine-pillow-basalt studies were designed to find out whether such rocks could be dated and concluded they could not.

Austin, Steven A. 1996, "Excess Argon within Mineral Concentrates from the New Dacite Lava Dome at Mount St Helens Volcano," CENTJ 10(3):335-343.

The young-earth movement's showcase experiment, and it contains the two facts that answer it:  Austin discloses that the laboratory was told nothing about the rock's age, and the laboratory was Geochron, whose published K-Ar footnote stated that they could not analyse samples expected to be younger than 2 Ma.

Zashu, Ozima & Nitoh 1986 (Nature 323:710-712);  Podosek et al. 1988 (Nature 334:607-609);  Ozima et al. 1989 (Nature 337:226-229).  All three:  abstracts verbatim, full texts not obtained.

The Zaire cubic diamond sequence, and the cleanest example in the literature of an isochron giving a physically impossible age:  6.0 Gyr, reproduced by a second argon method on four more diamonds.  The authors' own rescue — anomalous ⁴⁰K — was tested by two independent laboratories and killed.  Ozima then published the real cause:  excess ⁴⁰Ar in sub-micrometre mantle-fluid inclusions, identified by its correlation with chlorine, "no age significance."

Which way it cuts:  against naive concordance, hard.  Reproducibility failed;  a second argon method failed;  a chemical correlation succeeded.  It belongs in the article and does.

The KBS Tuff sequence:  Fitch & Miller 1970 (Nature 226:226-228);  Curtis et al. 1975 (Nature 258:395-398);  Gleadow 1980 (Nature 284:225-230);  McDougall 1985 (GSA Bulletin 96:159-175);  McDougall et al. 2012 (JGS 169:213-226).

The full arc, from a scatter of 223 to 0.91 Ma, through the pig-fossil challenge and Curtis's identification of detrital feldspar under a microscope, to a modern consensus of 1.869 ± 0.021 Ma.  Two items decide it.  Against the mainstream:  Gleadow published that the earlier fission-track "confirmation" was not independent, agreeing "mainly due to the close communication between these authors on track identification."  For the mainstream:  Gleadow's own 1980 redetermination of the same pumice published 1.87 ± 0.04 Myr.  The two later tests on Fitch and Miller's surviving original crystals — Newcastle 1981, blind, 2.3 Ma;  Hurford's 1987 potassium-argon in Berne, not blind, 1.87 ± 0.04 Ma — were never published and are known only from Lewin 1987, pp. 249 and 252.

Thorne, A., et al. 1999 (J. Hum. Evol. 36:591-612) and Bowler, J. M., et al. 2003 (Nature 421:837-840).  Abstracts verbatim;  full texts not obtained.

Lake Mungo 3.  Four independent methods concordant at 62 ± 6 kyr in 1999;  25 new optical ages giving 40 ± 2 kyr in 2003.  Four concordant methods, concordantly wrong by 55 %.

Which way it cuts:  against this article's own thesis, and it is the strongest such item found.  It is in the article for that reason.

Roberts, R. G., et al. 1998 (Nature 393:358-362).  Abstract verbatim;  full text not obtained.

Jinmium.  Thermoluminescence 116-176 kyr, contradicted by radiocarbon, resolved by single-grain optical dating at under 10 kyr in favour of the radiocarbon.  The mirror image of Lake Mungo:  discordance as a working alarm.

Scott, E. M., G. T. Cook & P. Naysmith 2010, "The Fifth International Radiocarbon Intercomparison (VIRI)," Radiocarbon 52:859-865.

The most decision-relevant numbers in the failure-mode dossier, and a genuine blind test — sample identities were withheld from the ~50 participating laboratories, which is precisely the design the "labs tune the answer" charge calls for.  Between-laboratory 1σ of 0.22-0.77 pMC for AMS laboratories.  But the full range is much worse:  4,810 BP reported on a bone with a 9,528 BP consensus;  21,684 to 42,500 BP on a mammoth bone with a 39,305 BP consensus;  about 10 % of returned results statistical outliers.

Jicha, B. R., et al. 2021, National Science Review;  Horstwood, M. S. A., et al. 2016, Geostandards and Geoanalytical Research 40:311-332;  Phillips, D., et al. 2022, Chemical Geology 597:120815 (obtained;  green open access via CORE 523303200).

The argon and laser-ablation inter-laboratory picture:  0.73 % maximum spread across three argon laboratories on one plagioclase separate (the best case);  "ca. 2 % 2s" as the stated current limit for LA-ICP-MS U-Pb;  and a ~0.4 % systematic floor under all argon ages, whose reconciliation paper is titled "closing the interlaboratory gap."

Damon, P. E. & J. L. Kulp 1958, "Excess helium and argon in beryl and other minerals," American Mineralogist 43:433-459.  http://www.minsocam.org/ammin/AM43/AM43_433.pdf

Sixty-eight years old, mainstream, and the answer to the "excess argon was discovered by creationists" framing:  "All beryl crystals appear to contain a quantity of helium and argon in great excess over that which can be accounted for from radioactive decay."

The clocks that do not use decay

Bronk Ramsey, C., et al. 2020, "Reanalysis of the Atmospheric Radiocarbon Calibration Record from Lake Suigetsu, Japan," Radiocarbon 62(4):989-999, with its supplementary data table.

The most valuable single dataset in this half of the article, because the supplement allows the comparison to be computed rather than quoted.  Doing so gives a result nobody advertises:  the varve count runs 5.4 % short of the uranium-thorium-anchored calendar scale at 50 ka, with a worst bin of 7.5 %, and only 16 % of levels inside the 2σ band.

It also gives the article's sharpest single argument.  Radiocarbon ages run younger than calendar ages throughout the glacial period — ratio 0.83 to 0.95 across 11-55 ka — and faster ¹⁴C decay in the past predicts the opposite sign.

The underlying Bronk Ramsey et al. 2012, Science 338:370-374, is not obtained (Cloudflare at every repository).

Reimer, P. J., et al. 2020, "IntCal20," Radiocarbon 62(4):725-757.

The calibration curve, and unusually forthcoming about its own weak points:  it downweights the Bahamas stalagmite record and removes all coral data older than 25 cal kBP.  Both concessions are ones a young-earth critic is entitled to, and both were made by the curve's authors.

Sigl, M., et al. 2015, "Timing and climate forcing of volcanic eruptions for the past 2,500 years," Nature 523:543-549.

The documented failure of ice-core layer counting — seven years wrong at 775 CE — caught by a ¹⁰Be spike matched to tree-ring radiocarbon.  Two named causes:  a sulfate layer wrongly attributed to AD 79 Vesuvius and used as a fixed tie point (retracted here;  it is 87/88 CE), and "use of manual-layer interpretation techniques that may favor interpretations consistent with a priori knowledge or existing chronologies."

Which way it cuts:  for the sceptic on confirmation bias, in Nature, in the sceptic's own words;  for the discipline on self-correction, since the error is 0.2 % and an independent clock found it.

Kent, D. V., et al. 2018, PNAS 115:6153-6158;  Kent, D. V., G. Muttoni & P. Brack 2004, EPSL 228:369-377.

The best and the worst of cyclostratigraphy.  The 405-kyr Jupiter-Venus eccentricity cycle counted across the Late Triassic gives 8.50 Myr where U-Pb gives 8.48 ± 0.3 Myr — 0.25 %, between two clocks sharing no physics.  The Latemar platform gives 9-12 Myr by cycle counting where U-Pb, biostratigraphy and magnetostratigraphy give ~1 Myr.

Which way these cut:  the first is the strongest decay-free cross-check in the article;  the second is the honest failure, and it matters that the radiometric clock is the one that caught it.

Williams, G. E. 2000, "Geological constraints on the Precambrian history of Earth's rotation and the Moon's orbit," Reviews of Geophysics 38:37-59;  Wells, J. W. 1963, Nature 197:948-950.

Wells counted about 400 growth lines per year in Middle Devonian corals.  Williams says the palaeontological clocks are "of dubious reliability" and supplies the strong version instead — Elatina/Reynella tidal rhythmites at ~620 Ma giving 400 ± 7 solar days/yr, 13.1 ± 0.1 synodic months/yr and a 21.9 ± 0.4 h day, three mutually constrained quantities.  The implied lunar recession rate, 2.17 ± 0.31 cm/yr, is "little more than half the present rate" — which defeats the "extrapolate today's recession backwards" argument.

Friedrich, M., et al. 2004, "The 12,460-Year Hohenheim Oak and Pine Tree-Ring Chronology," Radiocarbon 46:1111-1122;  Svensson, A., et al. 2008, "A 60 000 year Greenland stratigraphic ice core chronology," Climate of the Past 4:47-57, Table 2.

Ring counting against decay counting:  8 years' disagreement over 11,000 (0.07 %).  Ice-layer counting against argon and uranium-thorium:  within 800 years over 60,000, with a stated counting error of "1 % in the Holocene and 5 % in the Glacial."

Decay-rate constancy

Uzan, Jean-Philippe 2011, "Varying Constants, Gravitation and Cosmology," Living Reviews in Relativity 14:2.  arXiv:1009.5514.

The source of the α-decay sensitivities that close the article's argument:  ∂lnλ/∂lnα = +548 for ²³⁸U, +466 for ²³⁵U, +774 for ¹⁴⁷Sm.  Also the published Re-Os-against-U-Pb bound, Δα/α = (−8 ± 16) × 10⁻⁷.

Chaffin, Eugene F. 2005, RATE Vol. II ch. 7, pp. 525-585.

RATE's own physics chapter, and the most damaging single source in this section — for RATE.  Its Table 2 gives the β-decay sensitivities as negative (−30 for ⁴⁰K, −180 for ⁸⁷Rb, −18,000 for ¹⁸⁷Re) against the α-emitters' large positive values, which means no single change in the fine-structure constant can accelerate both families.  Chaffin then offers the Re-Os sensitivity as the explanation for a measured discordance of about 1.6 — a step that, taken at the 10⁹ the model requires, puts Re-Os at 10⁵⁴⁰⁰.

His concessions are extensive and quotable:  the α ceiling is 10⁸ "in some cases," the work is "exploratory in nature," "whether the weak force can vary in a way consistent with observation seems doubtful," and on his own Oklo samarium study, "half-life differences of more than one order of magnitude would seem to be ruled out."

Ohtsuki, T., et al. 2004 (PRL 93:112501) and 2007 (PRL 98:252501);  Stone, N. J., et al. 2007 (PRC 76:054315).

The measured ceilings.  The largest environmental effect on any decay short of full ionisation is 1.6 % (⁷Be in C₆₀ at 5 K versus in gold at 293 K).  The firm ceiling on any effect on alpha decay is 1 %, from forty years of low-temperature nuclear-orientation data.

Pommé, S., et al. 2016, "Evidence against solar influence on nuclear decay constants," Physics Letters B 761:281-286.

Forty-eight authors across fourteen national metrology institutes, refuting the claimed annual periodicity in decay rates.

Meshik, A. P., C. M. Hohenberg & O. V. Pravdivtseva 2004, PRL 93:182302.

Oklo's duty cycle reconstructed from fission xenon:  roughly 30 minutes critical, 2.5 hours dormant.  The α-variation bound from Oklo is Damour & Dyson 1996, −0.9 × 10⁻⁷ < Δα/α < 1.2 × 10⁻⁷.

The three exotic acceleration claims, graded rather than dismissed:

• Reifenschweiler 1994 (Physics Letters A 184:149-153) reports a 40 % slow-down of tritium decay in small titanium particles, from unrepeatable 1958-68 experiments. Wrong sign for the young-earth argument.
• Simakin & Shafeev 2011 (arXiv:1112.6276; Physics of Wave Phenomena 21:31-35, 2013) claim laser-induced accelerated α-decay with gold nanoparticles. Internally inconsistent by ~10⁶ against their own stated pulse-energy budget; the field involved is ~3.6 × 10⁻⁹ of the field at a uranium nuclear surface, worth Δλ/λ ≈ 5 × 10⁻⁶.
• Proton-21 (Adamenko et al. 2015, Results in Physics 5:62-68). No independent replication anywhere in the citation graph. Its own text concedes the process "is likely to demand an external energy input" of ~100 J, contradicting the laboratory's earlier over-unity framing and removing the heat sink that Walt Brown's hydroplate model invokes it for.

Our own computations

Five re-runnable scripts sit in this article's research folder.  All are Python 3 standard library only;  pandas is not installed on this machine and none is needed.

Two of these were validated against published results before being used to disagree with anything:  the RATE script reproduces RATE's own Isoplot fits to a fraction of a percent wherever the data are well conditioned, and the database script reproduces Beachy, Kinard & Garner's published numbers to the digit.  That order matters.  This project has been burned before by a verification tool whose bugs all failed in the same direction, manufacturing false accusations;  a checker that errs that way is worse than none.

Not obtained

Listed together, because the article's credibility depends on this list being visible rather than inferred.  Each is flagged again at the sentence that depends on it.

Sources that turned out to be unusable

scienceagainstevolution.org — the creationist page on lunar sample ages, cited in this project's research notes.  The domain has been hijacked and now serves casino spam on every path, including archived ones.  A capture is stored as evidence, and the article cites the page from it — all 116 rows, with the conflation of crystallisation ages, lead model ages, low-temperature release steps and cosmic-ray exposure ages shown row by row.

The Jane Balme quotation ("there is a competition to get the oldest date and there's kudos in it") — traced to a creationist magazine article citing an Australian newsweekly feature of 24 June 2003, not to anything Balme wrote.  Third-hand and unverified at the primary level.  Not used.

The Lars Borg quotation ("we really struggle to get these ages ourselves") — real, verified verbatim from a 2012 Nature news feature, but about a proposed miniature dating instrument for Mars and a three-person clean laboratory's throughput.  Used only with its context, in the article's quotation audit.

The "6 readings from 5 different dating methods" claim about the KBS Tuff — not what happened.  A garbling of a secondary source, and it contradicts the accurate account of the same episode in Johanson's Lucy, pp. 242-243, which appears elsewhere in the same set of research notes.

Access routes worth recording

• ICR hosts RATE Volume II chapter by chapter as free PDFs, linked from icr.org/rate2. Each link goes to an icr.org/article/<slug>/ page whose body contains the real PDF href; fetch the article page first, extract the href, then fetch the PDF.
• ICC proceedings from all nine conferences are at cedarville.tind.io, with stable record numbers. The publications.cedarville.edu and digitalcommons.cedarville.edu URLs in circulation are page-viewer or redirect links, not citable landing pages.
• USGS Open-File Reports come down whole as a zip from pubs.usgs.gov/of/<year>/<number>/, public domain, no login.
• Dalrymple's 1986 reply to creationism is a USGS Open-File Report and therefore public domain and quotable at length — an unusually useful property for a source in this dispute.
• Two-column PDFs must have their whitespace normalised before any phrase search, and cropped column by column if that fails. Two of this article's PSCF quotations required a sliding-window column crop to verify.

Verification pass 2 — 2026-08-22

Four independent verifiers re-established every footnote and number against source bytes (reports in research/verify2-yec.md, verify2-methods.md, verify2-failure-clocks.md, verify2-numbers.md; all eleven scripts re-ran and ngdb_output.txt reproduced byte for byte).  About sixty defects were found and applied:  wrong locators and overstated characterisations dominated again, plus one argument that reversed on re-computation — the flawless-method null model had treated 36 pairwise overlaps as independent (0.2%), and the dependent calculation gives 11% per nine-age record and about five or six perfect scores expected among the 34, against the one observed (research/null-model.py, null-34-exact.py).  Corrections that ran against the article's own case were made:  Wiens 2002 does carry a same-rock table (Amitsoq gneisses), the Science Against Evolution lunar page was recoverable and is now cited, Lake Mungo's "four independent methods" were one uranium-uptake model reported several ways, the Hell Creek pair is two decay schemes from one laboratory's orbit, and the Snelling 2010 and Chaffin quotations had lost their qualifiers.  The Austin/Snelling transcription error on Dalrymple's 1969 excess-argon table is now settled from Dalrymple & Lanphere 1969 Table 8-5.  Still open:  Zheng 1989's full text, Fitch & Miller 1970, Curtis 1975, Bowler & Magee 2000 (quarantined), and the Balme quotation at the primary level.