Cosmogenic Nuclide Dating

"When transforming the Be-10 values into Biblical earth history, creation scientists run into four complications which make it impossible to be accurate.  In spite of this, it appears that the cosmic rays were very high during the Flood and tailed off rapidly during the Ice Age." — Michael J. Oard, Creation Research Society Quarterly, 2021¹

"In general, the results show that the calibrated production rates fail statistical tests of goodness-of-fit. … The results indicate that there are substantial unresolved difficulties in modeling cosmogenic nuclide production and the calibration of production rates." — Borchers and nine co-authors, reporting on their own calibration for the CRONUS-Earth project, 2016²

"The important thing about this relation is that we know the production ratio and the decay constants, so the slope of the line depends only on the burial age t_b.  Thus, we do not need any information about the sample depths, the exposure time, or the inherited nuclide concentrations to determine the burial age of the soil." — Greg Balco and Charles Rovey, American Journal of Science, 2008³

Summary

Cosmic rays hit oxygen and silicon atoms in quartz at the ground surface and smash them into rarer atoms — beryllium-10, aluminium-26 and a few others.  The longer a rock sits in the open, the more of these atoms it holds.  Count them and you have a clock.

The clock comes in two forms, and they break in opposite ways.  That single fact organises everything below.

1. A surface-exposure age is concentration divided by production rate, t = N/P. The production rate sits in the denominator, so the age is inversely proportional to the cosmic-ray flux assumed for the past. If the flux was forty times higher, every exposure age is forty times too old. Radioactive decay barely enters: over 12,000 years the beryllium-10 decay correction is 0.3 per cent. So the young-earth attack surface here is flux, not decay rate — the exact reverse of uranium-lead or potassium-argon dating.
2. A buried-sediment age is t = −ln(R/R₀)/(λ₂₆ − λ₁₀). The production rate does not appear in that equation at all — only the production ratio of the two nuclides and their two decay constants. Multiply the past cosmic-ray flux by any number you like and a burial age does not move by one digit.
3. The young-earth engagement is thin, and every part of it lands on the first clock. Michael Oard named the calibration problem in 1997, attacked the published field results at the Wind River terraces in 2014, and built the full flux reconstruction in 2021. The 2021 Creation Research Society Quarterly paper is careful, numerate, and right about several things this article concedes in full. But the flux history he derives is circular in a way one line of algebra exposes, and he cites the decisive burial-dating paper without engaging its result. Creation Ministries International has reported a ²⁶Al/¹⁰Be burial age twice, in 2022 and 2023, and both times the response was that dates keep changing — not that the burial clock is wrong.
4. The mainstream's own accuracy figures are worse than most defenders of it admit, and they are nowhere near what a young earth needs. The CRONUS-Earth team's hold-out test missed independently dated ages by 7.1 to 27.1 per cent depending on the nuclide. Five published scaling schemes disagree by about 11 per cent, and the field says there is "no compelling reason" to prefer any of them. Those are the honest error bars. They are percentages, not factors of a thousand.
5. Verdict: the young-earth criticism of exposure dating is real and partly correct; the young-earth position still has no answer to ²⁶Al/¹⁰Be burial dating. Four tills in Missouri, dated through the buried soils beneath them, give burial ages in correct stratigraphic order, with the one at 0.82 ± 0.07 Ma normally magnetised (so younger than 0.78 Ma) and the one at 1.25 ± 0.31 Ma reversely magnetised (so older than 0.78 Ma). The nuclide ratio, the layer order and the rock magnetism share no physical parameter, and all three agree.

Every source this rests on is catalogued, with what each one measured and which way it cuts, in a companion page: the literature.

Two clocks, not one

Cosmic-ray secondaries — mostly fast neutrons — reach the ground and knock chunks out of the nuclei they hit.  Oxygen and silicon in quartz become beryllium-10 and aluminium-26.  Calcium and potassium become chlorine-36.  Olivine holds helium-3.  Production falls off exponentially with depth, with an e-folding length near 60 cm in rock, and is effectively dead below about three metres.⁴

Clock one: how long has this surface been in the open?  Ignoring decay and erosion, the answer is the concentration divided by the production rate.  With decay included, N = (P/λ)(1 − e^−λt^).  Either way P is in the denominator.

That is the whole vulnerability.  Nobody measured the cosmic-ray flux four thousand years ago.  P is inferred, and if the inference is wrong by a factor, the age is wrong by that factor.

Clock two: how long has this sediment been buried?  Aluminium-26 and beryllium-10 are made in the same quartz grain at a fixed ratio of 6.75 to 1.³  Bury the grain deep enough to stop the cosmic rays and the two stop being made.  Aluminium-26 has a half-life of about 0.705 Myr and beryllium-10 about 1.387 Myr, so the ratio falls, and the fall is the clock:

t_b = -ln(R_measured / R_initial) / (lambda_26 - lambda_10)

Read that equation for what is missing.  No production rate.  No exposure time.  No burial depth.  No inherited concentration.  Balco and Rovey say so in as many words, in the sentence quoted at the top of this page.

The two clocks therefore fail under opposite attacks.  Change the past cosmic-ray flux and clock one moves in exact proportion while clock two does not move at all.  Change the decay constants and clock two moves while clock one barely notices.  A young-earth argument has to make both moves at once, and the second one is expensive.

A small Deno script, research/burial-dating.ts, works all the arithmetic in this article.  Its output is reproduced below in the places it is used, so a hostile reader can rerun it.

Where the young-earth argument comes from

There is very little of it, nearly all of it is by one man, and it has improved.  Four pieces, over twenty-four years.

1997: the calibration problem, named early and named well.  Michael Oard's first piece on the subject is three pages in the CEN Technical Journal, prompted by the very low granite erosion rates that ¹⁰Be and ²⁶Al give for the inselbergs of the Eyre Peninsula in South Australia.⁵  Most of it is an erosion-rate argument that belongs to a different subject.  One paragraph of it is the best criticism the young-earth movement has ever made of cosmogenic exposure dating, and it is set out and answered in the next section.

2013: dismissal without engagement.  Writing in the Journal of Creation on continental erosion, Oard set cosmogenic methods aside in one sentence.⁶  In his words:  "There are several methods for estimating the amount of erosion in a region.  Some of them are indirect, such as chemical and cosmogenic isotopic methods, which rely on radiometric dating.  Since these methods are based on uniformitarian assumptions, other more scientific methods will be used."⁶

That sentence is wrong on the physics, and this site holds its own side to the same standard it holds everyone else.  A cosmogenic exposure age does not "rely on radiometric dating" in the sense meant.  It relies on a production rate.  Run the decay correction on Oard's own showcase sites and it is 0.3 per cent at the 12.6 kyr Scottish moraine and 0.5 per cent at the 18.3 kyr Bonneville shoreline.⁷  For a late-glacial surface the method is very nearly a pure counting exercise with no radiometric content at all.  Calling it uniformitarian and moving on is not an engagement.

2014: the field results.  A year later comes eighteen pages in the Creation Research Society Quarterly on the river terraces of the Wind River in Wyoming.⁸  Its target is not the production rate but the published ages themselves, and on one landform it has a point.  It is dealt with below, under the honest limits of the method, because that is what it is about.

2021: the real paper.  Oard went and did the work.  "Much Greater Cosmic Rays During the Ice Age and Before," nineteen pages in the Creation Research Society Quarterly, is the fullest young-earth treatment of cosmogenic nuclide dating in print.¹  He reads the primary literature, quotes it accurately, tabulates measured concentrations from seven sites, does arithmetic, and publishes numbers.  It deserves to be answered at that level.

His conclusion, in his own words: "I have examined the amount of in situ Be-10 on quartz from Unaweep Canyon formed during the Flood, an inselberg, and several glacial surfaces.  It appears that the amounts of cosmic rays that form Be-10 were much higher during the Flood and tailed off rapidly during the Ice Age."¹

The part of his case that is right

Oard's central criticism of production-rate calibration is accurate, he made it first in 1997, and it is fairly sourced.

The 1997 version describes the procedure exactly.  A concentration is measured by accelerator mass spectrometry "in a material of 'known' age.  By dividing this concentration by the age, the production rate is calculated for a given locality, and then extrapolated for other latitudes and altitudes."  He names the anchor:  glacially polished quartz in the Sierra Nevada at 2–3.5 km altitude, whose surface "was estimated to be 11,000 years old based on the C-14 age of organic remains associated with the surfaces."⁵  Then he draws the conclusion:

"Because of this calibration, the in situ cosmogenic radioisotope dating method is not independent.  Its accuracy depends upon the accuracy of other dating methods, in this case carbon-14 dating."⁵

That was right in 1997 and it is still right.  It is a better argument than the flux reconstruction he built twenty-four years later, and this article says so plainly.

Here is the 2021 version at full strength, in his words:

"In the past, observed production rates have been estimated from physics (Masarik and Reedy, 1995), but the physics is so complex that researchers estimate the production rate based on surfaces of 'known' age (Balco et al., 2009; Balco, 2017b; Blard et al., 2019).  Such a procedure is in fact circular reasoning and adds an old-age bias to the measurements."¹

He then quotes the CRONUS-Earth calibration paper itself for the mechanism: "The production rate parameter is calibrated by finding the value that best fits measured concentrations from a collection of sites for which independent age measurements are available."²  That quotation checks verbatim against the source.  So does his quotation of Devendra Lal on why the physics-first route does not work: "Accurate estimations of production rates are generally not possible because of lack of knowledge of the probabilities of formation of nuclides in the different reactions."¹

He is right.  The production rate is a fitted parameter, not a measured constant.  Any article on this subject that hides that is not worth reading.

And the field concedes it just as plainly.  Gosse and Phillips, in the standard 86-page review of the method,⁴ write: "Unfortunately, there are very few locations with highly reliable independent estimates of exposure duration.  Because of the continuous development of the TCN method, most sites with reliable independent age control and known simple exposure history are actually used to calibrate the TCN timescale."⁴

And Oard's 1997 sentence still describes the calibration set of 2016.  Borchers and colleagues list their primary ¹⁰Be sites one by one:  Promontory Point on the Lake Bonneville shoreline in Utah, the Isle of Skye and the Scottish Highlands, Quelccaya in Peru, and New Zealand.  Every one of those four is anchored by radiocarbon, recalculated with CALIB 6.0.  Varve counting appears only at a secondary site, the Littleton–Bethlehem moraine in New England.²  So the concession has to be made in full:  the beryllium-10 production rate is, to a first approximation, tied to radiocarbon.

Three things bound that, and all three are set out at length below rather than asserted here.  The fitted rate has to predict concentrations at sites held out of the fit, and the published miss is 7 to 27 per cent, not a factor.  The radiocarbon it is tied to covers roughly 3,000 to 18,000 years at these sites, a span in which radiocarbon is itself tied to tree rings and varves, so the error it can pass on is percentage-sized.  And a ²⁶Al/¹⁰Be burial age contains no production rate at all, so the circularity Oard identified cannot reach it.  An argument that stops at calibration circularity has answered half the subject.

What the mainstream says about its own accuracy

Three published numbers, all from the people who built the calibration, all against their own interest.

The hold-out test failed.  The CRONUS-Earth project² fitted production rates to a set of primary calibration sites, then computed ages for a second set of sites held out of the fit and compared them with the independent ages.  From the abstract: "In general, the results show that the calibrated production rates fail statistical tests of goodness-of-fit. … The root-mean-square percent differences between the median computed ages for these secondary sites and independent ages range from 7.1% to 27.1%, differences that are much larger than the typical uncertainties in the site ages.  The results indicate that there are substantial unresolved difficulties in modeling cosmogenic nuclide production and the calibration of production rates."²

Per nuclide, at the held-out sites:

²

The scaling schemes disagree.  Balco and colleagues⁹ fitted the same calibration data set five different ways and printed all five reference beryllium-10 production rates side by side: 4.96, 4.88, 4.90, 5.39 and 4.84 atoms per gram per year at sea level and high latitude.⁹  The spread between the extremes is 5.39/4.84, or 11 per cent, and the age divergence reaches about ±10 per cent at high elevation and at ages far outside the calibration range.  Their own verdict: "In practice, there is no compelling reason for a user to prefer the results from any particular scaling scheme … Furthermore, there is no reason to believe that the results from the different scaling schemes are randomly distributed, so averaging them will not yield a more accurate age estimate."⁹

One of the half-lives was wrong by 14 per cent until 2010.  Chmeleff and colleagues, opening the paper that fixed it:¹⁰ "Yet no consensus exists on a reliable value for the half-life of ¹⁰Be.  Several determinations of this half-life have been done in the past … leaving a discrepancy of 14% between different estimates and splitting the community of users."¹⁰  They then measured 1.386 ± 0.016 Myr by one method while a separate laboratory measured 1.388 ± 0.018 Myr by another, giving a recommended 1.387 ± 0.012 Myr.

That last one cuts both ways and both should be said.  A field carried a 14 per cent disagreement in a fundamental constant for decades.  That is a real defect.  It was then closed by two independent techniques agreeing to 0.15 per cent — and closed in the direction that made every burial age about ten per cent older, against the interest of anyone who wanted quick answers.  Chmeleff's own note on who was affected matters too: the old disagreement "did not compromise exposure ages and erosion rate determinations using in-situ produced ¹⁰Be, as long as the half-life used for production rate determinations and unknown samples are the same," but it did compromise burial ages.¹⁰

So the honest accuracy of exposure dating, stated by the people who do it, is somewhere between 7 and 27 per cent.  Not 4,000 per cent.  That is the number the rest of this article has to be measured against.

Oard's method, and the line of algebra that undoes it

His procedure has three steps.

Step one: one global correction, derived from one site.  A moraine in the Scottish Highlands has radiocarbon on associated organic matter at 12,480 BP.  Oard's flood and ice-age chronology assigns that moraine about 4,000 years.  He divides: "This is about three times the Biblical age, suggesting the production rate in the Biblical timescale would be roughly three times that of the secular timescale.  We can probably apply this correction factor worldwide for all the measurements."¹

Step two: four complications, three of them dismissed.  His Table I lists the things that could change beryllium-10 production in a biblical timescale.  Beside three of the four he writes, in effect, small: accelerated radiometric decay is "Unknown but will assume small"; a post-Flood quadrupole field has an "Effect likely small"; a fast-oscillating field during rapid reversals has a "net effect likely was small."¹  So the ×3 and the site-by-site residual carry the whole argument.

Step three: apply the ×3 to seven sites and read the leftover as a flux history.  Yarwondutta Rock in South Australia comes out "74 times the expected concentration."¹  Hickory Run boulder field in Pennsylvania is "38 times the expected Biblical value."¹  The GISP2 bedrock under the Greenland ice is 6.7 times, or 60 times once he corrects for eroded rock.¹  Promontory Point on the Bonneville shoreline is "only about 1.5 times the Biblical estimate."¹  The Scottish moraine comes out at 1.07 and the Sierra Nevada pavements at 0.79 — that is, his model slightly over-predicts the Sierra measurement.¹  The pattern he reads out of this is very high cosmic-ray flux during the Flood, decaying rapidly through the Ice Age, and back to modern rates by the end of it.

Tabulating what he never tabulates

Oard never puts his seven sites in one table.  Doing so is the whole test.  For each site the total flux enhancement his reconstruction demands is the measured concentration divided by the present-day production rate and by the number of years his chronology assigns:

=== Oard 2021: flux enhancement demanded by each site ===
site                    N (at/g)   P_now  Oard yr  N/P (kyr)  factor needed
Yarwondutta inselberg   3.57e+6    3.57   4500     1,000.0    222.2x
Hickory Run boulders    3.26e+6    6.3    4500       517.5    115.0x
GISP2 bedrock surface   2.48e+4    8.2     150         3.0     20.2x
Bonneville Promontory   2.75e+5   15.1    4050        18.2      4.5x
Scottish moraine        6.93e+4    5.5    3950        12.6      3.2x
Sierra Nevada pavement  5.50e+5   60      3900         9.2      2.4x

These reproduce Oard's own published factors exactly.  74 × 3 = 222, 38 × 3 = 114, 6.7 × 3 = 20.1, 1.5 × 3 = 4.5, 1.07 × 3 = 3.2, 0.79 × 3 = 2.4.  The arithmetic is his arithmetic, checked and found correct.⁷

Now look at what the last column is

Write the flux factor out:

factor = N / (P × t_Oard)

and split the fraction:

factor = (N / P) / t_Oard

N / P is the mainstream simple exposure age.  So the last column is the published age of the site divided by the number of years Oard's chronology assigns to it.

That is not a measurement of cosmic-ray flux.  It is a restatement of the disagreement.  Six sites, six freely chosen durations, one residual per site — and the residual is then named as a discovery about cosmic rays.  Any set of published ages whatever can be reproduced this way.  No independent observation of past cosmic-ray flux enters the derivation at any point.

The apparent shape of the curve comes from the same place.  The flux "declines through the Ice Age" because the sites Oard chose have declining published ages, and because he assigned five of the six durations within 600 years of each other (4,500, 4,500, 4,050, 3,950, 3,900).  With the denominators nearly constant, the residual is just the numerator, and the numerator is the mainstream age.

Where this criticism does not reach.  A curve fitted to data is not automatically false.  If the cosmic-ray flux really was two hundred times higher during the Flood and decayed with a time constant of a century or so, Oard's numbers are exactly what would be seen.  The argument against him is not that his fit is a fit.  It is that the proposal makes other predictions, and those are checkable — which is what the rest of this article does.

An inconsistency inside his own paper

Oard's three well-behaved sites — Sierra Nevada at 0.79, Scotland at 1.07, Bonneville at 1.5 — are the calibration sites.  They are where the production rate was fitted in the first place, and they agree with his model to within tens of per cent.

His two badly discrepant sites, Hickory Run at 38× and Yarwondutta at 74×, are the two whose published ages are largest, up to 628 ka.  Those are also the sites where beryllium-10 is closest to saturation, where the decay correction stops being negligible — 16 per cent at Hickory Run and 39 per cent at Yarwondutta⁷ — and where the "minimum age" caveat bites hardest.

Oard notes both facts himself.  On the Australian inselbergs he writes that "the figures are minimums because a little erosion would have already taken place between the time of formation and today."¹  He does not connect them.  His measured factors track the published ages because that is what they are made of.

The clock with no production rate in it

Everything above concerns clock one.  Clock two is where the argument stops working, and it stops working for a structural reason rather than an evidential one.

Bury a quartz grain deep enough and production stops for both nuclides.  What remains is decay at two different rates, so the ²⁶Al/¹⁰Be ratio falls from its surface value of 6.75 towards zero.  The ratio itself has an effective half-life of ln 2 divided by the difference of the two decay constants, which is 1.434 Myr.  Balco and Rovey's isochron refinement uses several samples from different depths in one buried soil: they share a burial age but differ in how much they inherited from earlier exposure, so their ²⁶Al and ¹⁰Be concentrations fall on a straight line whose slope carries the age.

Here is the whole clock, with research/burial-dating.ts doing the arithmetic:

=== constants ===
lambda(10Be) = 4.9975e-7 /yr  (T1/2 = 1.387e+6 yr)
lambda(26Al) = 9.8319e-7 /yr  (T1/2 = 7.050e+5 yr)
lambda26 - lambda10 = 4.8344e-7 /yr
R_init (26Al/10Be at surface) = 6.75
half-life of the RATIO = ln2/dlambda = 1.434 Myr

=== burial age vs measured 26Al/10Be ratio ===
R_m     burial age (Myr)
6.75    -0.000
6.00     0.244
5.00     0.621
4.00     1.082
3.50     1.359
3.00     1.677
2.00     2.516
1.50     3.111
1.00     3.950

Now run the young-earth move through both clocks at once:

=== sensitivity to a factor-of-N change in past production rate ===
factor  exposure age (kyr) from N=3.26e6 at/g, P=6.3 at/g/yr    burial age (Myr) at R_m=3.5
1x      517.5                                                   1.359
3x      172.5                                                   1.359
10x      51.7                                                   1.359
38x      13.6                                                   1.359
74x       7.0                                                   1.359
100x      5.2                                                   1.359

Read the two columns.  The mechanism Oard's whole paper is built on moves the exposure age by a factor of a hundred and moves the burial age by nothing whatever.⁷

Oard supplies the input himself

He does address the ratio, once, in a single paragraph on page 39:

"A second complicating factor is that the Al-26/Be-10 ratio turns out less than its present-day formation ratio of about seven, because there should not have been enough time in 4,500 years for this ratio to measurably change.  But, secular scientists find values around three or four.  Such a ratio implies hundreds of thousands to a few million years of age.  This means that much accelerated radiometric decay has taken place after the Flood."¹

Take his own numbers.  A ratio of 4.0 gives a burial age of 1.082 Myr; a ratio of 3.0 gives 1.677 Myr.  To compress those into 4,500 years, the difference of the two decay constants must be larger by a factor of 241 to 373 — and because the ratio clock depends only on λ₂₆ − λ₁₀, that means both constants scaled up by roughly that factor, in the same window, for two light beta-decaying nuclides.⁷

Two sentences later, on the same page, discussing the same two nuclides, Oard writes:

"However, it is possible that accelerated radiometric decay is less for low atomic number nuclides (Vardiman et al., 2005).  On the other hand, accelerated radiometric decay could cause an increase in these isotopes, because of an increase in neutrons from the rapid fission of ²³⁸U and the alpha decay of ²³⁵U, ²³⁸U, and ²³²Th (Heisinger et al., 2002).  It is difficult to know the outcome of these two competing processes for cosmogenic nuclide dating.  It could be they cancel each other and the net effect would be small."¹

A factor of 250 to 370 is not a small net effect.  And his own Table I, on the facing page, assumes accelerated decay is negligible.  The paper needs accelerated decay to be enormous in one paragraph and negligible in the table two paragraphs later.

This is stated without heat, because it is not a gotcha.  It is a gap.  Nobody has filled it.

The Missouri till stack

This is the hardest single result for a young earth in the whole subject, and the paper that contains it is cited in Oard 2021 — once, in passing, on page 33 — and never returned to.¹

Central Missouri has a stack of glacial tills with buried soils between them.  Balco and Rovey dated four of the tills by running the isochron on the soil underneath each one.³  The isochron slopes they measured, and the ages those give:

³

The abstract states the result plainly: "The method yields ages for four tills in Missouri that are stratigraphically consistent, agree with paleomagnetic age constraints, and show that ice advanced into Missouri near 1.25 Ma, near 0.8 Ma, and twice between ca. 0.4 and 0.2 Ma."³

Why this is hard.  Three completely different things agree.

1. The nuclide ratio in the buried soil. It depends on two nuclear decay constants and nothing else — not on the cosmic-ray flux, not on burial depth, not on how long the grains sat at the surface first.
2. Superposition. Which till lies on which. It depends on nothing but the order they were laid down in.
3. Remanent magnetisation. Whether the till froze in a normal or reversed magnetic field. It depends only on the reversal history of the earth's field. "The Moberly till is magnetically reversed, indicating deposition prior to 0.78 Ma, and the other three overlying tills have normal polarity."³

These three share not one physical parameter.  Cosmic-ray flux appears in none of them.  And the sharpest case is the pair that straddles the boundary: Fulton is normally magnetised so must be younger than 0.78 Ma, and the isochron puts it at 0.82 ± 0.07 Ma — the authors note their "age estimate is consistent with this constraint and, in fact, appears to closely limit the age of the till to near 0.78 Ma."³  Moberly is reversely magnetised so must be older than 0.78 Ma, and the isochron puts it at 1.25 ± 0.31 Ma.³

A young-earth account has to explain, at once, why a decay-ratio clock lands on either side of a magnetic reversal boundary in exactly the two units whose magnetism brackets it, and why four ages come out in stratigraphic order — all as an artefact of something.  Nobody has attempted this.  The nearest thing to an attempt is two short Creation news items about a different burial-dating site, Sterkfontein, and they argue discordance rather than mechanism;  see the objections below.

The sequence in the young-earth literature is worth stating exactly, and then leaving alone.  Oard cites Balco and Rovey 2008 on page 33 for the proposition that cosmogenic nuclides have been used to date burial of sediments and soils, "although several uniformitarian assumptions go into the calculations."  He never mentions the Missouri result.  His one paragraph on ²⁶Al/¹⁰Be ratios, six pages later, concedes ratios of "three or four," attributes them to accelerated decay after the Flood, and two sentences later suggests that the competing effects on light nuclides may cancel to something small.

Testing the clock against events dated some other way

The cleanest possible test of exposure dating is a lava flow.  A flow starts with no inherited nuclides — the rock was molten — and its age can often be got independently, from radiocarbon on charcoal it buried, or from argon-argon, or from a written record.  Three results of that kind matter here, and they do not all point the same way.

Kurz and colleagues, Hawaii, 1990.  They measured cosmogenic helium-3 in olivine from Hualalai and Mauna Loa flows whose ages were fixed by radiocarbon, "ranging in age from 600 to 13,000 years," picked specifically because they show no erosion or soil cover.  The result is double-edged and should be reported that way.  The inferred sea-level production rate "was found to vary significantly, ranging from 47 to 150 atoms g⁻¹ yr⁻¹."  Averaged over the past 2,000 years it is 125 ± 30; between 2,000 and 7,000 years it drops to 55 ± 15; between 7,000 and 10,000 years it is back up at 127 ± 19.  The authors put the total swing at "a factor of 2.3 ± 0.8" and add at once that it is "poorly determined due to the large uncertainties in the youngest samples and questions of surface preservation for the older samples."  Theory predicts less: models of the same dipole variation give "a factor of 1.15–1.65."  Their bottom line for the method: "the data demonstrate that it is feasible to date samples as young as 600 years of age providing that there has been no erosion or soil cover."¹¹

Take that seriously in both directions.  It is a measured factor-of-2.3 wobble in Holocene production, larger than the 3 per cent that reversal modelling predicts for long exposures, and a young-earth writer is entitled to point at it.  It is also not monotonic — the low point is in the middle of the record, not the old end, which is not the shape a flux that "tailed off rapidly through the Ice Age" would leave.  And it is a factor of 2.3, not a factor of 74.

Licciardi and colleagues, Oregon, 1999.  They recalibrated the independent lava ages and renormalised all the published helium-3 rates, their own included.  As Gosse and Phillips report it: "The average normalized production rate reported by Licciardi et al. is 119 ± 1 atoms g⁻¹ yr⁻¹ based on 44 measurements that span from about 600–17,500 yr.  The precision and consistency of these measurements make the ³He system the best calibrated, and also suggests than any dipole or non-dipole field effects were not significant."⁴  Forty-four independently dated flows spanning six hundred to seventeen and a half thousand years, and the production rate needed to fit all of them is one number to about one per cent.  Gosse and Phillips add the honest caveat in the next sentence — because Kurz's low-latitude Hawaiian rates "have a large range (55 ± 9–231 ± 101 atoms g⁻¹ yr⁻¹) over the Holocene and are higher, it is possible that there was a magnetic (dipole or non-dipole) or surface effect on these data."

Ashik Kol volcano, Tibet.  This one is a different kind of test, because it checks a production-rate clock against a decay clock.  "The uncertainty-weighted mean of 21 ³⁶Cl age determinations from the Ashik Kol volcano in the Kun Lun Shan (northern Tibetan Plateau) was 66,000 ± 1000 years.  The result of a ⁴⁰Ar/³⁹Ar age determination on the same basaltic trachyandesite was 62,000 ± 1300 years."⁴  Two clocks with nothing in common — one counting atoms made by cosmic rays at the surface, one counting atoms made by potassium decaying inside the rock — land six per cent apart at sixty-six thousand years.  For a young-earth reconstruction both would have to be wrong by a factor of roughly fifteen, in the same direction, for unrelated reasons.

One classic test was not obtained for this article.  Zreda and colleagues dated the young Lathrop Wells basaltic eruption complex in Nevada with chlorine-36 in 1993, and that paper is the other standard zero-inheritance check.  It is closed at Geology, has no open-access copy anywhere that Unpaywall or OpenAlex knows of, and was not read here.  What can be said from a source that was read: Gosse and Phillips report that at Lathrop Wells, weathered olivines replaced with iddingsite gave chlorine-36 exposure ages "30% older than unweathered samples from the same surface," which is a caution about sample selection rather than a validation.⁴  The known-age eruption validation exists in the literature and is cited by the reviews; only two of its three legs have been verified here.

The ice-core cross-check, and what it does not show

Oard's paper closes with an argument from ice cores, and it does not fail in the way a hostile reader might expect.  This section says so plainly because overstating it would be easy.

His argument: measured beryllium-10 in the GRIP core rises from about 1–2 × 10⁴ atoms per gram in the Holocene to about 3–5 × 10⁴ during the Ice Age, and because his model puts metres of ice per year where the mainstream puts centimetres, the same concentration implies vastly more beryllium-10 falling out of the sky per year.¹

Work that through.  Deposition flux is concentration times accumulation rate.  A concentration ratio of 2.5 to 3, against Ice Age accumulation rates tens of times the Holocene value, gives a flux enhancement in the range of tens to low hundreds — which overlaps the 20 to 222 his in-situ sites demand.  His scheme does not self-destruct on the ice-core concentrations, and an article claiming it does would be overstating.

Three things do damage it, and only the third is decisive.

First, and weakest: the ice cores do not independently support his flux history; they are merely consistent with it once his accumulation rates are assumed.  The reason the mainstream thinks beryllium-10 fallout was roughly steady is that glacial concentration is up and glacial accumulation is down, so the product changes little.  Oard needs accumulation to have been an order of magnitude higher, and that requirement comes out of his ice-age model, not out of the cores.  This is not a refutation.  It is a note that the cores are not evidence for him either.

Second: he needs the flux to run in two directions at once.  He offers a high post-Flood cosmic-ray flux as the answer to the radiocarbon inventory problem — "47,000 C-14 years needs to be telescoped into 1,500 years after the Flood" — and notes that Gerald Aardsma, who identified that problem, "is no longer a young-Earth creation scientist as a result."¹  But his own Sierra Nevada and Scottish results require the flux to be back at modern values by roughly 3,900 years ago, at the end of his Ice Age.  That leaves the entire period since then at ordinary flux, and it is that period in which the radiocarbon calibration curve has to do most of its telescoping.  The two halves of the paper pull against each other and the paper does not reconcile them.

Third, and this is where the real damage is: beryllium-10 and carbon-14 are produced by the same cosmic rays, and their records are locked to each other and to tree rings, to the year.  A flux that swung by two orders of magnitude inside the last 4,500 years would have to have done so while leaving those joint tie-points synchronous.  Three of them, all documented with sources in the radiocarbon article and in this repository's ice-core notes:

• AD 774/775. Miyake and colleagues found "a rapid increase of about 12‰ in the ¹⁴C content from ad 774 to 775, which is about 20 times larger than the change attributed to ordinary solar modulation" in Japanese cedar rings.¹² The matching beryllium-10 spike appears in three separate ice cores.¹³
• 7176 BCE. A solar particle event found in the beryllium-10 of four cores — NGRIP, EGRIP, GRIP and EDML — matching the tree-ring radiocarbon date to about one year after the standard layer-count adjustment.¹⁴
• 14.3 ka. The oldest such event yet found: beryllium-10 in GRIP at 14,301–14,292 cal BP against tree-ring radiocarbon at 14,300–14,299, a difference Bard and colleagues describe as "only a decade, at the most."¹⁵

Those are two independent cosmogenic archives — ice and wood — agreeing on the timing of the same cosmic-ray events to within a year at 9,000 years ago and within a decade at 14,300.  The synchronism is the argument, not the concentration profiles.  That is where this case should be made, and it is where Oard's paper does not go.

The honest limits of the method

Stated in full, because a reader who finds them elsewhere first will rightly stop trusting this page.

Inheritance makes an age too old.  Nuclides accumulated during an earlier exposure, before the surface was made, are indistinguishable from nuclides made afterwards.  Oard's citation of Balco on this is fair: inheritance in a striated bedrock surface usually is not detected until the answer disagrees with an age known some other way.¹

Erosion makes an age too young.  Atoms leave with the rock they were made in.  So when the true age is unknown, an uncorrected exposure age is a minimum.  This asymmetry cuts against the young-earth argument as well as for it: it means Oard's showcase concentrations are floors, and the flux factors he derives from them are floors too.

On a real landform the methods can disagree by a factor of two to four.  This is Oard's 2014 argument and it is his second-best.  Terrace WR7 on the Wind River in Wyoming carries the Lava Creek B ash from the last Yellowstone super-eruption, dated by tephrochronology at 660 kyr.  Cosmogenic-nuclide depth profiles on the same terrace give about 300 kyr, and one study puts it as young as 250 kyr;  ²³⁰Th/U on pedogenic carbonate gives 140 to 315 kyr.⁸  Because every higher terrace is dated by extrapolating the incision rate down from WR7, the highest terrace WR15 comes out at 1.7 Ma on the ash date and at 431 to 511 kyr on the cosmogenic dates.⁸  Oard quotes the practitioners against themselves and quotes them accurately.  Hancock and colleagues concede "the continuing difficulty of precisely dating depositional surfaces older than a few 100 ka [thousand years] with CRN, because of the uncertainties in the geologic history of the surfaces and the large errors arising from calibrations of the rate of production which are inherent in dating surfaces with CRNs."⁸

Concede that, then size it.  The discordance is real, it is worse than the 7-to-27-per-cent hold-out figure, and it falls on the clock this article already calls the weak one.  But the methods disagree with each other by a factor of two to four, and Oard's chronology needs WR7 to be about 4,500 years old — a factor of 55 to 150 below the lowest published number on the terrace.  A scatter of two to four does not buy a factor of a hundred.  Nor does the 2014 paper claim it does:  it argues the terraces were cut by late-Flood channelised runoff and offers the dating discordance as a reason to distrust the glacial sequence.  And it does not touch burial dating.  The word "burial" does not occur in it, and Balco and Rovey are not cited.

Cover shields the surface, and the corrections are estimates.  Snow is the commonest cause.  Sediment matters as much: Gosse and Phillips work an example in which a surface deglaciated 25,000 years ago but covered by 5, 15 or 25 cm of sediment for its first 10,000 years yields ages "2.5, 7, and 11% too low if continuous exposure was assumed."⁴  Even very dense forest reduces the cosmic-ray flux "by up to 4%."⁴

The scaling schemes disagree by about 11 per cent and there is no principled way to choose.  Stated above; it is Balco's own conclusion, not a critic's.

The geomagnetic field has varied, and the correction runs in the young-earth direction.  Today's field "appears to be almost 1.5 times stronger than the average intensity over the past 1 Myr," so past production genuinely was higher on average.⁴  But the size is small when integrated.  Gosse and Phillips simulate a worst case — field intensity dropped to 20 per cent of modern for twenty thousand years across the Brunhes–Matuyama reversal, at 20° latitude and 3 km elevation, the most sensitive place on earth — and the time-averaged production rate moves by "approximately 3%."  Their conclusion: "even in the worst-case scenario, TCN production rates for long (>10 kyr) exposures are essentially indistinguishable from a constant long-term integrated production rate for the same period."⁴  Oard reaches the same verdict about the geomagnetic route in his own Table I, and it is to his credit that he does.

And the production rate is fitted to surfaces dated by other methods.  This is the big one and it is not hidden here.  The primary beryllium-10 calibration sites are the Bonneville shoreline in Utah, dated by radiocarbon at about 18.3 ka; a Scottish Highlands moraine with radiocarbon at 12,480 BP; and sites in New Zealand and Peru.¹  Nishiizumi's original calibration and Lal's used Sierra Nevada glacial pavements at about 11 ka.  A clock calibrated against radiocarbon and glacial chronology cannot then be used to prove that radiocarbon and glacial chronology are right.  That is a real limit on what exposure dating may be used for, and it is the strongest thing in Oard's paper.

What bounds it is the hold-out test.  A production rate fitted to a handful of sites then has to predict concentrations at sites left out of the fit, and Borchers and colleagues ran exactly that test and published the miss: 8.3 per cent for beryllium-10, 7.1 for aluminium-26, 17.7 for chlorine-36, 27.1 for helium-3.  A method that were purely circular could not fail such a test at all, and certainly could not fail it by different amounts for different nuclides.

And burial dating escapes the circularity entirely, which is the reason this article is organised the way it is.  There is no production rate to calibrate.  The ratio at the surface is 6.75 because that is what the nuclear physics of spallation on silicon and oxygen gives, checked directly in quartz; the two decay constants come from nuclear laboratories that never heard of a moraine.  Nothing in a burial age is fitted to a geological timescale.

Objections

"The production rates are calibrated on surfaces dated by other methods.  The whole thing is circular, so it proves nothing about deep time."

Correct as far as exposure dating goes, this article concedes it everywhere it comes up, and it is Oard's own argument from 1997 rather than a critic's.  Three answers, and none is a dodge.

The first is that circularity of this kind is bounded and testable.  Fit a rate to four sites, then predict fourteen more and see how far off you are.  That was done, and the answer is 7 to 27 per cent.  A parameter tuned to force agreement would not miss by 27 per cent on helium-3 and 7 per cent on aluminium-26; it would miss by nothing, everywhere, by construction.

The second is that the radiocarbon the production rate is anchored to covers roughly 3,000 to 18,000 years at these sites, and over that span radiocarbon is itself calibrated against tree rings and varve counts.  A clock tied to a clock tied to counted annual layers can inherit percentage errors.  It cannot inherit a factor of a hundred.

The third is that burial dating has nothing to be circular about.  No production rate enters the equation.  If every exposure age published since 1986 were thrown away, the Missouri till stack would still read 0.17, 0.33, 0.82 and 1.25 Ma, in that order, and would still bracket a magnetic reversal from the correct sides.

"A 27 per cent hold-out misfit on helium-3 means the method does not work."

It means the method is accurate to tens of per cent, which is what its practitioners say.  The relevant question is what the misfit would have to be to rescue a young earth.  A 27 per cent error is a factor of 1.27.  Oard's sites need factors of 2.4 to 222.  His mildest site needs the age to be wrong by 140 per cent — about five times the worst misfit the field has ever measured — and his most extreme needs it wrong by twenty-two thousand per cent.  That is not a difference of degree.

Note also which nuclide it is.  Helium-3 is the one whose calibration data set spans the widest range of ages, and it is the one Kurz's Hawaiian scatter sits inside.  The 27 per cent is a real measurement of real trouble in helium-3 systematics.  It is not evidence that beryllium-10 ages are wrong by orders of magnitude, and beryllium-10 is where the argument lives.

"Burial dating still assumes the two half-lives.  If decay rates changed, it collapses like everything else."

True, and it is the only route left.  So price it.  The ratio clock depends on λ₂₆ − λ₁₀, and Oard's own conceded ratios of three to four require that difference to have been 241 to 373 times larger.  Since both are beta decays of light nuclei, that means both constants up by roughly that factor.

Three problems follow.  First, it is the opposite of what the RATE volume Oard cites for the low-atomic-number caveat proposes: accelerated decay in that literature is largest for heavy alpha emitters, not for beryllium and aluminium.  Second, Oard's own Table I assumes accelerated decay is small, so the paper cannot have it both ways within two pages.  Third, a factor of 250 to 370 in decay rate over a few thousand years is a heat and radiation problem of exactly the kind the RATE project itself documented and did not solve.

The honest position is that this objection identifies the only surviving escape and shows how expensive it is.  It is not answered here by asserting that decay rates are constant.  It is answered by pricing the assumption and observing that nobody has paid.

"The Missouri magnetostratigraphy is itself dated by the timescale you are trying to prove.  The 0.78 Ma boundary comes from argon-argon and orbital tuning.  That is circular too."

The sharpest form of the objection, and it needs a careful answer rather than a brush-off.

The date of 0.78 Ma for the Brunhes–Matuyama boundary does come from radiometric and astronomical work.  If that number is wrong, the numerical agreement is worth less.

But the numerical date is not what the Missouri result turns on.  What it turns on is ordering and sign.  The polarity of a till is a yes-or-no fact recorded in the rock at the moment it was deposited: was the field normal or reversed.  Reading it requires no timescale at all.  The Moberly till is reversed; the three tills above it are normal.  That is an observation.  Independently, the ²⁶Al/¹⁰Be ratio in the soil under each till gives a number, and those numbers come out in the same order as the layers, with the change from oldest-to-youngest crossing exactly where the polarity changes.

For that to be an accident, a nuclide-ratio clock that knows nothing about magnetism would have to have produced a monotone sequence whose one crossing point coincides with the one polarity flip in the section.  And the ordering constraint is itself independent: superposition needs no dating method whatever.

The objection would land if the argument were "0.82 Ma equals 0.78 Ma, therefore both are right."  It is not.  It is "three unrelated measurements put the same boundary in the same place in the sequence."

"Burial dating failed at Sterkfontein.  Granger's ²⁶Al/¹⁰Be ages are more than a million years older than the uranium-lead, palaeomagnetic and electron-spin-resonance ages on the same cave fill.  Creation Ministries International has said so twice."

Every fact in that objection is true, and it is the only place a young-earth publication has engaged burial dating at all.  Creation magazine ran the case in October 2022 and again in April 2023.  The second item put it sharply:  "Earlier dates on flowstone using both uranium-lead and paleomagnetic methods ranged from 2.1 to 2.6 My.  The new dates, based on measuring aluminum and beryllium isotopes, ranged from 3.41 to 3.67 My.  Earlier electron spin resonance (ESR) dating of the actual fossils ranged from ∼1 to 4 My."¹⁶

The dispute is stratigraphic, not nuclear, and that is the whole of the answer.  Nobody in it argues that the ²⁶Al or ¹⁰Be decay constants are wrong.  The question is what the flowstones date.  Granger and his co-authors¹⁷ argue that the dated flowstones fill cavities dissolved into the breccia after it accumulated, so a flowstone age is a minimum for the sediment around it:  "We show that these previously dated flowstones are stratigraphically intrusive within Member 4 and that they therefore underestimate the true age of the fossils."¹⁷  Their verdict on the other methods is not that they failed — "We believe that previous dating of flowstones by U-Pb and paleomagnetism is correct but that the dates indicate the age of flowstone deposition rather than accumulation of the breccia."¹⁷

Is that ad hoc?  It would be, if it rested on nothing but the need to save a date.  Two things stop it being that.  The isochron carries an internal check that nothing was fitted to:  its intercept implies a post-burial ¹⁰Be production rate of 0.028 ± 0.003 atoms per gram per year against an expected 0.030 at ten metres depth, and the fit has a mean square weighted deviation of 1.09, so the samples are consistent with one burial age.¹⁷  And the method has been checked elsewhere against clocks that share nothing with it — ⁴⁰Ar/³⁹Ar on lavas interbedded with dated gravels in the West Kunlun Mountains, and palaeomagnetism at Lantian — with agreement, and with disagreement only at Sterkfontein, where the intrusive flowstones can be seen in the field.¹⁷

And notice what the objection cannot deliver.  Every number in it is in millions of years.  The widest spread Creation itself quotes, from ESR at about 1 My to burial dating at 3.67 My, is a factor of four.  The chronology the same paragraph offers — "The caves formed late in Noah's Flood some 4,500 years ago" — is a factor of two hundred to eight hundred below the bottom of that range.  A disagreement among methods that all read millions of years is not evidence for thousands.

"Cosmogenic dating is radiometric dating, and radiometric dating assumes deep time."

This is the Oard 2013 position and it is simply wrong on the physics.  A beryllium-10 exposure age on a late-glacial surface has almost no radiometric content: the decay correction is 0.3 per cent at 12.6 kyr.  What such an age depends on is a production rate, which is a different assumption with different failure modes.  Collapsing the two is how the 2013 paper missed the real weakness — and the 2021 paper found it.

"Boulder ages on a single moraine scatter wildly.  Hickory Run ranges over a factor of seven.  How can anything so noisy be a clock?"

The scatter is real and the field publishes it.  At Hickory Run the beryllium-10 runs from 0.44 to 3.26 × 10⁶ atoms per gram across one boulder field, and the original authors attribute it to boulders having rotated or flipped.  Oard reports this correctly.

But scatter of that kind is signed and diagnosable.  Boulders that turned over read young, not old, because the exposed face is fresh.  That is why the maximum is taken as most representative — Oard does exactly this — and why moraine studies sample many boulders.  Noise that has a known sign is a solvable problem.  It is not the same thing as a systematically wrong production rate, which is the thing that would actually matter.

Where the young-earth case is genuinely strongest

Stated at full strength, with no hedging, because all six of these are real and none is manufactured.

1. The production rate is a fitted parameter, not a measured constant. The physics-first route does not work well enough to use, and the field says so, from Lal in 1991 to Borchers in 2016.
2. The calibrated rates fail their own goodness-of-fit tests. That is the CRONUS-Earth team's verdict on the CRONUS-Earth team's work.
3. Five scaling schemes disagree by about 11 per cent in reference production rate and by ±10 per cent in age at high elevation, and there is no principled way to choose between them.
4. The past cosmic-ray flux is not directly measured. It is inferred, partly from beryllium-10 and carbon-14 themselves. And the geomagnetic correction runs in the young-earth direction: today's field really is about 1.5 times the million-year average, so past production really was higher.
5. The beryllium-10 half-life was disputed by 14 per cent until 2010, "splitting the community of users." Every burial age published before that revision was wrong by about ten per cent.
6. Exposure ages are minima whenever a surface has eroded, and inheritance is often only caught when the answer disagrees with something already known. Very few surfaces have genuinely independent age control, and Gosse and Phillips say outright that most of the ones that do have been used up as calibration sites.

An honest article has to concede all six.  This one does.

Where it fails

The positive reconstruction is circular, and one line of algebra shows it.  The flux factor derived for each site, N/(P × t_Oard), is identically the mainstream exposure age divided by the years the biblical chronology assigns.  Six sites, six freely chosen durations, six residuals, and the residuals are then named as a discovery about cosmic rays.  No independent measurement of past flux appears anywhere in the derivation.  The declining curve is the decline in the published ages of the sites that were chosen.

Burial dating is untouched by the entire argument.  The equation contains no production rate.  Every mechanism the paper proposes — supernova, high flux, rapid decline — changes burial ages by exactly zero.

And the escape route is priced at 241 to 373 times faster decay for two light nuclides, in a paper that assumes accelerated decay is negligible in its own summary table and that suggests, two sentences after invoking it, that the competing effects "cancel each other and the net effect would be small."

The decisive paper is cited and passed over.  Balco and Rovey 2008 appears once in Oard 2021, on page 33, for a general proposition about burial dating, and the Missouri result — four isochron ages in stratigraphic order straddling a polarity reversal — is never mentioned.

And the standard being applied here is a young-earth scientist's own.  At the first International Conference on Creationism in 1986, Kurt Wise told his own movement that its attack on radiometric dating was not good enough.¹⁸  Of the four things a radiometric age requires, only the present parent-to-daughter ratio is measured in a laboratory.  On the other three:  "Creationists typically call the remainder of the points invalid assumptions.  In most cases, though, they are not assumptions.  Though they lack complete proof, they do not completely lack proof.  Critics need to consider what proof there is before categorically rejecting the dating method."¹⁸  And on what a complete critique owes:  "a simple list of errors is insufficient to reject the method.  If creationism is true, radiometric dates are wrong, but the fact that they are wrong in the same direction (all too old) must be explained for a critique to be complete."¹⁸

That is the test this article has applied, and it was set inside the movement forty years ago.  Oard 2014 supplies the list of errors.  Oard 2021 attempts the explanation of why they all run one way, and that explanation is the flux reconstruction, and the flux reconstruction is the published ages divided by the durations Oard assigns.  Nobody has met Wise's standard here yet.

What would change the verdict

Three things, in order of how tractable they are.

1. A measured ²⁶Al/¹⁰Be ratio well below 6.75 in a sediment of independently known young age. This is the cheap experiment nobody has run. If buried Holocene sediment somewhere reads a ratio of four, the burial clock is broken and the news is enormous. It would cost one field season and some accelerator time.
2. A mechanism producing a sub-production-ratio ²⁶Al/¹⁰Be at the surface, so that the burial signature could form without burial. Nobody appears to have looked for one.
3. A demonstration that both decay constants can be raised by a factor of 250 to 370 without an accompanying heat or radiation catastrophe. This is the hardest and it is the one the literature keeps reaching for.

Until one of those lands, the position stated here is that exposure dating has the weakness Oard identified and burial dating does not.

Has anyone else in the young-earth literature looked?

A correction first.  An earlier version of this page said that cosmogenic nuclide dating had been engaged once, by one author, in 2021, and that ²⁶Al/¹⁰Be burial dating had never been engaged at all.  Both halves of that were wrong.  The claim was scoped to the three archives that could be searched at the time and said so, which was honest, but the two that could not be reached — Journal of Creation behind a Cloudflare interstitial, and the International Conference on Creationism archive apparently returning 403 — turned out to be the two holding the material.  A fuller sweep on 30 August 2026 reached all five venues.  Here is what it found, and how, so that a reader can check it.

Creation Ministries International has engaged burial dating twice.  Creation 44(4) for October 2022 reported Granger's Sterkfontein paper under the heading "New dates for South African fossils cause contention," and named the method:  "They used a method known as Cosmogenic Nuclide Burial Dating."¹⁹  Creation 45(2) for April 2023 ran the same case again as "What date would you like?"¹⁶  Both are short unsigned news items rather than technical papers, and neither disputes the physics.  The argument in both is discordance with other methods on the same cave fill, and it is answered in the objections above.

Michael Oard engaged the method in 1997, not 2021.  The CEN Technical Journal piece on the Australian inselbergs carries the calibration-circularity argument set out earlier on this page, and it is a better argument than the flux reconstruction he built twenty-four years later.  It is also the seed of that reconstruction.  In 1997 he already proposed multiplying the production rate "by two or three times," found that even so "we still have too much cosmogenic radioisotope accumulation in surficial rocks since the Flood," and concluded that the only remaining route was "postulating a significantly higher cosmic ray intensity over the early centuries following the Flood."⁵  The 2021 paper is that programme carried out.  He engaged it a third time in 2014, in the Wind River terraces paper, on the field results rather than on the production rate.

Nobody has replied to the 2021 paper in CRSQ, and he has not revised it.  All 1,403 Creation Research Society Quarterly records from 1964 through volume 62 (2026) were enumerated through the journal's OAI-PMH endpoint.  Its full-text search index was validated against controls first — zircon returns 41 hits and radiocarbon 150, both far more than the harvested metadata alone contains — and then searched.  cosmogenic returns 12 records and beryllium 11, and in both lists exactly two are on topic:  Oard's 2014 terraces paper and his 2021 flux paper.  There is no reply, comment, correction or later Oard paper that revisits it.²⁰  One caveat, and it matters.  The CRSQ repository types every item as a peer-reviewed article and does not appear to ingest the Letters and Comments columns, so a short letter answering Oard would not appear in it.  The claim is safe for articles, not for letters.

No paper on terrestrial cosmogenic-nuclide dating has ever been presented at the International Conference on Creationism.  This is now established from full text rather than from titles, and the earlier 403 does not reproduce:  the Cedarville repository serves every ICC record to a plain curl with a browser user-agent.  All 510 record IDs in the nine conference blocks from 1986 to 2023 were fetched, 458 paper PDFs were mapped, 457 were downloaded and text-extracted, and the whole corpus was swept for cosmogenic|beryllium-10|10Be|26Al|36Cl.  Seven papers match and none is about terrestrial exposure or burial dating:  two concern cosmogenic nuclides in meteorites, and five are single passing mentions.  Controls on the same corpus:  radiocarbon hits 37 papers, zircon 27.²¹

The one exception is a single forward-looking sentence.  Nathan Mogk, writing on accelerated nuclear decay at the ninth conference in 2023, notes:  "There may still be evidence of post-Flood AND in the ratios of cosmogenic radionuclides which are commonly used for Quaternary dating (Oard, 2021)."²²  That is the only place another young-earth author has looked at Oard's cosmogenic work, and it is a proposal for future work rather than an argument.  It also points at the ratio, which is the burial clock, which is the right place to look.

The Journal of Creation has never run a Letter, Perspective, Viewpoint or Forum piece on it.  A harvest of 6,741 creation.com items, covering Journal of Creation volumes 1 to 40 complete, contains no title in any of those columns with cosmogenic or beryllium in it.  Oard's 1997 piece is a Perspective;  his 2021 paper ran in CRSQ;  nobody answered either in print.²³

Answers in Genesis and the Answers Research Journal have nothing on topic.  An offline filter over all 436 ARJ titles and abstracts for cosmogenic, beryllium, 10Be, 26Al, 36Cl and exposure returns zero papers.  The AiG site search returns one near-miss, Snelling's "How Reliable Are Dates for Human Fossils?", which explains ¹⁰Be and ²⁶Al accurately and does not critique them.  There is a trap worth recording:  AiG's own site search does not index answersresearchjournal.org, so searching AiG is not searching ARJ, and the volume table of contents has to be enumerated instead.²⁴

The Institute for Creation Research has used ¹⁰Be data once, to argue for a young earth, and got the physics wrong in the same sentence.  Brian Thomas, reporting Portenga and Bierman's global compilation of ¹⁰Be erosion rates in 2011:  "Geologists have been measuring quantities of ¹⁰Be, an isotope of the element beryllium that becomes radioactive with exposure to the sun.  The more ¹⁰Be that is present at a given site, the longer it has been exposed to the sun without being carried off by erosion."²⁵  Beryllium-10 is not an isotope that becomes radioactive on exposure to anything.  It is made by cosmic-ray spallation of oxygen and silicon and is radioactive from the moment it exists, and the agent is the galactic cosmic-ray flux, which solar activity reduces rather than supplies.  The error is not the interesting part.  The interesting part is that ICR's one use of the data treats ¹⁰Be accumulation as a valid clock in order to date the continents — the opposite of the position taken in CRSQ.

One negative that can never be established from ICR's search box.  Its search endpoint has broken recall.  Queried on 30 August 2026, racemization and thermoluminescence each return no result list at all, while https://www.icr.org/article/amino-acid-racemization-dating-method/ and https://www.icr.org/article/examining-thermoluminescence-dating/ are both live ICR articles with those words in their titles.  Controls behave:  zircon and cosmogenic both return hit lists.  So the failure is silent and selective.²⁶  That is a fact about the tool rather than about the literature — no negative finding about ICR may be drawn from that search box — and the site's sitemap.xml, with 29,086 URLs and descriptive slugs, is the route that can carry one.

The corrected finding, with its limits.  Cosmogenic exposure dating has been engaged by one young-earth author, in three pieces, across twenty-four years, and his best argument is the oldest one.  ²⁶Al/¹⁰Be burial dating has been engaged twice, in two short unsigned magazine items, about one field discordance, with no treatment of the equation at all.  Nobody in the young-earth literature has yet addressed the fact that a burial age contains no production rate in it.  That gap is smaller than this article once claimed, and it is still the gap the article was written about.

Every source above, with what each one measured, the access route that worked, and which way it cuts, is catalogued in the literature — including the ones that were considered and not used, and the two that could not be obtained at all.

Sources