Non-Radiometric Clocks

"It is relatively easy to count annual layers downward from the surface through considerable depths in the Greenland ice sheet." — Larry Vardiman, Institute for Creation Research, 1992¹

"Can palaeontology provide anything by way of verifying the pronouncements now emanating, at very considerable expense, from the black boxes?" — John W. Wells, Cornell University, 1963²

"The extent of the presently available long dendrochronologies argues forcefully for a Flood date in excess of 10,000 years ago." — Gerald Aardsma, Institute for Creation Research, in the Creation Research Society Quarterly, 1993³

Summary

Almost every argument about the age of the earth is an argument about radioactive decay.  That is a strange place for the argument to sit, because some of the best-measured chronologies in science contain no radioactivity at all.  They are built by counting things:  layers of ice, laminae of lake mud, rings of wood, daily growth bands in a shell, tidal laminae in a sandstone, cycles of climate driven by the orbits of the planets.

A counting clock has no decay constant in it.  If every half-life on the periodic table were wrong by a factor of a million, an ice layer would still be an ice layer and a tree ring would still be a tree ring.  So the young-earth proposal that nuclear decay ran fast during the Flood — the central claim of the RATE project — does not touch these records at all.  They have to be answered separately, and on their own terms.

Nine conclusions, stated first.

1. The counting archives are shorter than people think, and that matters. Continuous, absolutely dated tree rings reach 12,410 years.⁴ Greenland ice is counted layer by layer to 60,000 years, and no further.⁵ Lake Suigetsu's laminae are counted to about 50,000, and half of that count is interpolated by algorithm rather than seen.⁶ Beyond those depths the "ages" come from flow models, sedimentation-rate assumptions, or radiometric dating — not from counting. Anyone who says "800,000 years of ice layers" is misdescribing the record.
2. Inside those limits the counts are checked against decay clocks, and they agree to within a few per cent. Greenland's layer count sits within 800 years of a uranium-dated Chinese stalagmite across the whole 60,000 years.⁵ Counted 41,250 ± 814 years at the Laschamp geomagnetic event against argon-dated 40,400 ± 1,000 — 2.1%.⁵ Counted lake laminae against uranium-thorium at Lake Suigetsu — agreement to a few decades down to 15,000 years, drifting to a 5–7% shortfall at 50,000.⁶ Twenty-one counted orbital cycles against uranium-lead zircon in the Late Triassic — 0.25%.⁷
3. The sharpest single test uses no calibration curve at all. Measure the raw carbon-14 left in a tree ring or a leaf, in percent modern carbon, and plot it against the counted layer number. Nothing is calibrated, nothing is converted to an "age." If one ring is one year and one lamina is one year and carbon-14 decays at today's rate, the points must fall inside a narrow band. They do, from the surface to about 50,000 counted layers, across two archives on opposite sides of the world.⁸
4. The archives are tied together by events that are dated to a single year. In 774 CE a burst of solar particles spiked atmospheric carbon-14 worldwide. Forty-four tree-ring chronologies on five continents, thirteen genera, 484 individually measured rings — all record the spike in the same ring.⁹ The same event shows as a beryllium-10 spike in Greenland and Antarctic ice.⁹ When the ice chronologies were tested against it, they were found to be seven years wrong, and were corrected.¹⁰
5. Counting clocks fail in specific, documented ways, and the failures are published by the counters. More than one lamina per year in a lake fed by turbidity currents.¹¹ Missing rings in a drought-stressed pine. Ambiguous ice layers at depth. Six hundred beautifully rhythmic beds in an Italian atoll read as twelve million years of orbital cycles and now thought to be about one million.¹² Every one of these was found by the mainstream, published in the mainstream, and forced a revision.
6. The young-earth case against these clocks is real but under-quantified. Its strongest form is Michael Oard's: in a post-Flood ice age with enormous snowfall and violent weather, sub-annual oscillations get miscounted as years.¹³ The best documented natural example of over-counting is a factor of about two, in a glacier-fed Swiss lake with year-round turbidity input.¹¹ A young earth needs a factor of ten to twelve, sustained, in archives with different physics, on four continents at once. That gap has never been closed in print.
7. The most decisive test of the young-earth reading was run by a young-earth creationist, and it failed. Gerald Aardsma, then at the Institute for Creation Research, reasoned that if post-Flood trees grew two or three rings a year, the century-scale wiggles in the radiocarbon calibration curve would be stretched by the same factor. He measured the widths of thirteen such wiggles. They do not broaden. His conclusion, printed in the Creation Research Society Quarterly: the rings are annual back at least 9,000 years, and "the Flood must, in fact, have occurred more than 10,000 years ago."³
8. Two counting clocks refute arguments the young-earth side uses. Tidal laminae in a 620-million-year-old Australian sandstone record 400 ± 7 days in the year, 13.1 months in the year, and a 21.9-hour day, three quantities that agree with one another through three different equations of celestial mechanics.¹⁴ They imply the Moon has receded at 2.17 cm/yr on average, not the 3.82 cm/yr measured today — which destroys the "the Moon would have been touching the Earth" argument, using a clock that counts.¹⁴
9. The honest verdict is that this evidence runs against a young earth, and it is not close. It does not depend on the constancy of decay rates, on the uniformity of geological process, or on any dating method being assumed correct. It depends on whether a tree puts on one ring a year and whether a lake puts down one couplet a year, and those two questions have been tested against each other and against three independent decay systems, and they agree.

The thing that is actually surprising

The most interesting fact in this whole subject is not a number.  It is that the strongest quantitative test of the "multiple layers per year" hypothesis was designed, executed and published by a young-earth creationist in a young-earth journal — and that when it came out against him, he printed it anyway and drew the uncomfortable conclusion.

The second most interesting fact is that his own side agrees with him more than the popular literature suggests.  A systematic sweep of the creationist periodical corpus for this series — the Creation Research Society Quarterly, Journal of Creation, Creation, Answers Research Journal, ICR's Acts & Facts and technical papers, all nine volumes of the International Conference on Creationism proceedings, and the Geoscience Research Institute's Origins — turns up four incompatible positions on whether tree rings are annual, and a 2018 review of that literature, by a creationist, calling for research "to build a consensus dendrochronology calibration model" because there is none.¹⁵  The most technically serious analysis of the bristlecone ring-width data, by a creationist using the Arizona Tree-Ring Lab's own software, opens by conceding that "No obvious grounds exist for questioning the overall validity of the bristlecone pine long chronologies."¹⁶

What this series covers

This article is the overview:  what a counting clock is, what the archives are, how they are tied to each other, how they fail, and what would have to be true for them all to be wrong together.  Five companion articles treat the individual archives in detail.

• Ice-Core Layer Counting — Greenland and Antarctica, the five independent seasonal signals, the maximum counting error, the seven-year correction, and where the counting actually stops.
• Lake Varves — Suigetsu, the Green River Formation, the Swedish time scale, and the cases where more than one couplet really does form in a year.
• Tree-Ring Dating — bristlecone and oak, crossdating and its statistical traps, false and missing rings, and the 774 CE marker that fixes 44 chronologies to a single year.
• Growth Bands and Tidal Rhythmites — days in the Devonian year, why the coral clocks were abandoned, and the one rotational clock that survives.
• Orbital Cycles in Rock — counting Milankovitch cycles, the best agreement with a decay clock ever measured, and the worst published failure.

Part 1: What a counting clock is, and why it is a different argument

A radiometric date is a calculation.  It measures a ratio of daughter atoms to parent atoms and converts it to a time using a decay constant λ.  Change λ and every date changes.  That is why the young-earth argument concentrates on λ:  the RATE project — Radioisotopes and the Age of the Earth, published by the Institute for Creation Research and the Creation Research Society — proposes that decay was accelerated by a factor of roughly a billion during the Flood year, with the further specification that alpha-emitters were accelerated more than beta-emitters.¹⁷

A counting clock contains no λ.  Nobody calculates anything.  A person looks down a core with a microscope and counts.  So the acceleration hypothesis, whatever its merits, is simply irrelevant here.  If the counts are wrong, they are wrong for a reason that has nothing to do with nuclear physics — because the thing being counted is not annual.

That is exactly where the young-earth literature puts the argument, and it is the right place to put it.  Jake Hebert, Andrew Snelling and Timothy Clarey state it plainly:

"there is actually no empirical evidence to back the claim that varves form as annual deposits over extended periods of time.  There are some varve pairs that form in a single year, but in many cases, the observational evidence shows that multiple supposed varve couplets can and have formed in a single year."¹⁸

That is a testable claim about sedimentology and tree biology, and this series treats it as one.

The four questions a counting clock has to answer

1. Is the unit annual? Does one couplet, ring or layer correspond to one trip around the sun?
2. Are any units missing? A tree in a bad year may put on no measurable ring at all; a lake may deposit nothing.
3. Can separate records be joined without error? No single tree is 12,000 years old; no single Swedish clay pit holds 4,000 varves. Long chronologies are chains of overlaps, and a bad overlap inserts or deletes centuries.
4. Is the count independent of the answer it is being used to support? If the layer count is tuned to a radiometric age, comparing the two proves nothing.

Question 4 is the one the young-earth literature presses hardest, and it deserves to be pressed.  Hebert, Snelling and Clarey's conclusion on Lake Suigetsu is:

"None of these dating methods are truly independent and thus objective.  They are inter-calibrated and adjusted to agree because of the assumption they are supposed to agree."¹⁸

Part 3 below is the answer to that charge, and it is an answer that uses no calibration curve of any kind.

Part 2: The archives, and how far each one really counts

The single commonest error in popular writing on both sides is to quote the length of a record as though it were the length of a count.  It usually is not.  Deep ice, deep ocean mud and deep lake sediment are dated by models, by radiometric methods, or by tuning to an orbital template — and only the top part is counted.

Read that table honestly and two things follow, one in each direction.

Against the popular old-earth argument:  the counting archives do not reach millions of years.  The longest genuinely counted sequence of annual layers on earth is about 60,000 years, and its own authors give it a 5% error bar at the bottom.  Antarctic ice that is advertised as "800,000 years old" is not counted at all;  it is modelled.  The counting clocks are a devastating argument against 6,000 years and a weak argument against 100,000.

Against the popular young-earth argument:  60,000 years is still ten times too long, and 12,410 counted tree rings is still three times too long.  A young-earth chronology puts the Flood at roughly 4,300 years ago and Creation at roughly 6,000.¹⁸  Every one of these archives has to be compressed by a factor of two to twelve, simultaneously, by mechanisms that differ from archive to archive.

Part 3: The test that uses no calibration curve

The circularity charge deserves a test that cannot be circular.  Here is one, due to the geologists Gregg Davidson and Ken Wolgemuth.⁸  It is worth walking through slowly, because it is the strongest single argument in this article and it is not complicated.

The setup

Carbon-14 is made in the upper atmosphere by cosmic rays and taken up by every living plant.  When the plant dies, the intake stops and the carbon-14 decays with a half-life of 5,730 years.  A laboratory does not report an age;  it reports how much carbon-14 is left, as a percentage of the modern atmospheric level — percent modern carbon, or pMC.  Converting pMC to a date requires a calibration curve.  Not converting it requires nothing at all.

So:  do not convert.  Plot the measured pMC of a sample against the number of the layer it came out of.  No dates.  No curve.  No assumption that the layer is a year.

What the conventional model predicts

If — and only if — one tree ring is one year, one lake couplet is one year, and carbon-14 decays at today's rate, then the pMC has to fall along a decay curve.  The one uncertainty is how much carbon-14 the atmosphere held in the past, and that is bounded independently by beryllium-10 in ice cores, which tracks the same cosmic-ray flux but does not decay appreciably.  Davidson and Wolgemuth set generous bounds:  atmospheric carbon-14 between 95% and 200% of modern across the whole 50,000-year range.  Those bounds convert into a narrow band of allowed pMC values at each layer number.⁸

What each young-earth claim predicts

Each competing claim moves the points in a different, identifiable direction:

• Multiple rings or laminae per year puts the points above the band — the material is younger than its layer number says, so it has more carbon-14 than expected.
• A bad crossdating match produces an abrupt step up or down at the join.
• Much lower past atmospheric carbon-14 puts the points below the band.
• Faster past decay also puts them below.
• Rapid deposition of many couplets in a burst produces a staircase: long flat runs where hundreds of couplets share nearly identical carbon-14, because they were laid down within days of one another.

What the data do

They fall inside the band.  Continuously.  From the surface to about 14,000 tree rings, then from about 11,000 to about 50,000 lake laminae, with the two archives overlapping and lying on top of each other.⁸  There is no staircase.  There is no step.  There is no systematic drift above or below.

Davidson and Wolgemuth's summary of what that leaves is fair, and it applies to the whole subject:

"Either God saw fit that 14,000 tree rings equals 14,000 years, or God manipulated unrelated and independent processes (tree rings per year, atmospheric carbon-14 production, and radioactive decay rates) in a precise manner over a much more abbreviated time frame such that they are indistinguishable from the expectations of conventional geology."⁸

The strongest objection to this test, and what survives it

Hebert, Snelling and Clarey's central complaint about the earlier 2010 version of this argument was that the "measured carbon-14" on the axis was not measured at all — that Davidson and Wolgemuth had taken radiocarbon ages off the IntCal04 calibration curve and back-converted them.¹⁸

That criticism was substantially correct, and it was answered by fixing the graph.  A "radiocarbon age" is a reporting convention, not a date:  it is the measured pMC pushed through a fixed formula using the obsolete Libby half-life and a fixed 100 pMC starting value.⁸  The conversion back to pMC is exact and reversible.  But Hebert was right that reading points off a fitted curve is not the same as plotting measurements, and the 2018 version plots the measurements.

Two residual objections are worth stating, because they are the honest limits of the test:

• The starting point of the varve chronology is not independent. Lake Suigetsu's laminae do not reach the surface — a 12-metre unvarved interval sits above them — so the varve count is a floating sequence whose top has to be pinned to something. It is pinned to the tree rings, by matching carbon-14. Davidson and Wolgemuth concede this outright: "the age of the uppermost varves is not determined independently."⁸ What the test then does is check the slope — whether varves and rings accumulate at the same rate through their 3,000-year overlap — and the slope is not affected by where the hinge sits.
• Below about 50,000 layers the carbon-14 runs out. The test simply stops there. It says nothing about the deeper part of any record, and it is not evidence for millions of years. It is evidence against thousands.

Part 4: How the archives are tied to each other

If each counting archive were an island, "they all agree" would be a weak claim, because agreement could be built in.  They are not islands.  Four kinds of marker link them, and none of the four was invented for the purpose.

Single-year cosmic-ray spikes

In 2012 Fusa Miyake and colleagues found a sudden 12‰ jump in the carbon-14 of Japanese cedar rings between 774 and 775 CE — "about 20 times larger than the change attributed to ordinary solar modulation."²²  A burst that large, that fast, is atmospheric:  it happened everywhere in the same year.

Six years later, an international consortium tested exactly that.  They measured the carbon-14 in 484 individual tree rings from 44 chronologies on five continents, from 42° S to 72° N, 40 to 4,000 metres elevation, 13 genera, 27 species, in living trees, historic timbers, archaeological wood and subfossil logs.  Every one of them records the excursion in the same ring.  The authors' conclusion:

"The global signature of the two cosmogenic excursions independently validates the precise dating of well-replicated tree-ring chronologies from five continents."⁹

This is the answer to the crossdating objection in its strongest form.  It does not matter how subjective a ring match may be in principle;  what matters is whether the matches are right, and here is an external, physical, single-year marker that says they are — in Europe, North America, Siberia, Japan, South America and New Zealand at once.

Four such events are now confirmed by more than one isotope:  774 CE, 993 CE, 660 BCE, and 7176 BCE.²³  A fifth, larger than any of them, sits at 14,300 cal BP and has been matched between subfossil pine rings and Greenland beryllium-10 to within "only a decade, at the most."²³

Volcanic ash and sulfate

A big eruption dusts the whole hemisphere.  Its ash carries a chemical fingerprint that identifies the volcano, and its sulfuric acid leaves a conductivity spike in ice.  So an eruption is a horizon that appears in ice cores, lake sediments and peat bogs at the same instant.

The Greenland chronology GICC21 puts historically recorded eruptions where history puts them:  Katmai 1912 at 1912 ± 2, Laki 1783 at 1783 ± 2, Öræfajökull 1362 at 1362 ± 2, Eldgjá 939 at 939 ± 3.8, the Okmok eruption of 43 BCE at 43 ± 5.6 BCE.²⁴  Its stated uncertainty is one year per century.

The same trick works in the other direction, and one case is instructive precisely because it went wrong.  A sulfate layer in the GRIP core had been attributed to the eruption of Vesuvius in 79 CE and used as a fixed tie point.  It was a misidentification;  the layer dates to 87/88 CE, and using it was one named cause of a seven-year error in the Greenland chronology.¹⁰  Vesuvius should not be cited as an ice-core tie point.  The identification has been withdrawn.

Bipolar synchronization

Greenland and Antarctica are counted independently by different teams using different chemistry.  Large eruptions appear in both.  Between 500 and 2000 CE, large sulfate signals in Greenland "frequently occurred within ±1 year" of independently dated signals in Antarctica;  back to 400 BCE, within ±3 years.¹⁰  Two counts, two hemispheres, converging.

The calibration curve as a meeting point

IntCal20 is the international radiocarbon calibration curve, and for present purposes it is best understood not as a dating tool but as a place where five archives with entirely different physics have to give the same answer:  tree rings (pure counting), lake varves (counting plus uranium-thorium), speleothems (uranium-thorium alpha decay), corals (uranium-thorium), and marine sediment.²⁵

Three results from its construction are worth having:

• Three separate Hulu Cave stalagmites, each independently uranium-dated, require the same dead-carbon correction to within eleven radiocarbon years across 43,000 years of growth: 472 ± 50, 470 ± 50, 481 ± 50.²⁵
• Irish oak latewood and Californian bristlecone whole rings, for 1700–1500 BC, give the same atmospheric carbon-14 to within 8.1 ± 1.9 radiocarbon years.²⁵ Two chronologies counted on two continents by two teams.
• An entire archive was thrown out because it disagreed. All coral data older than 25,000 cal BP were removed from IntCal20, because the corals scatter by ~500‰ in Δ¹⁴C and exceed 1100‰ around 27,000–29,000 years — attributed to freshwater alteration when those reefs stood above sea level during the glacial lowstand.²⁵

That last point cuts both ways and should be reported both ways.  It is a genuine concordance failure, in a whole class of material, admitted in print.  It is also the strongest available answer to the charge that the curve is unfalsifiable:  the builders identified a physical failure mode, checked that it correlated with tectonic setting, and deleted the affected data rather than fitting around it.

Part 5: How counting clocks fail

A clock is only worth citing if its failure modes are known.  These are, in rough order of how much damage they do.

More than one layer per year.  Real, documented, and the young-earth literature is right to lead with it.  In Lake Walensee, Switzerland, André Lambert and Kenneth Hsü found 300 to 360 laminae deposited over 165 years, generally two a year and locally up to five, produced by turbidity currents from snowmelt and rainstorms.¹¹  Their own conclusion is quotable and should be quoted:  their work "threw doubt on [De Geer's] second interpretation that varves or varve-like sediment are necessarily annual."¹¹  In a New Zealand glacier-fed lake, Pickrill and Irwin found an average of three couplets a year.¹¹  In Muir Inlet, Alaska, tides plus diurnal meltwater produce two couplets a day.¹¹

The limits of the argument are equally real.  Every one of those settings is a glacier-fed or tidewater system with year-round sediment delivery, and the sedimentological literature the young-earth authors cite says outright that such deposits "are not true varves."¹¹  The archives used for chronology are chosen to avoid exactly this:  deep, anoxic, non-glacial basins where the couplet is a seasonal biological signal rather than a flood pulse.  The gap between a documented factor of two and a required factor of ten has not been closed.

Missing layers.  The opposite error, and it is the one that actually dominates in practice.  Lake Suigetsu's counters state that "on average 50% of the varves are indistinguishable" over part of the record, and fill the gaps by interpolation.⁶  The measured consequence is that the varve count runs short of the uranium-thorium chronology by 5–7% below 20,000 years.⁶  Note the direction:  the counting clock is running fast, not slow.  Undercounting makes the record look younger, not older.

False and missing tree rings.  A drought-stressed conifer can lay down a second latewood band in one season, or skip a year entirely.  Both are documented, both are why crossdating exists, and neither is a secret — see Tree-Ring Dating.

Bad joins.  Correlating one short section to the next is where centuries get inserted or lost.  Jan Lundqvist's judgement on the Swedish varve chronology is as blunt as any creationist could wish:  "even much better connections than the ones from Forsa can be obtained between limited parts of varve series in a way that is indisputably wrong."²⁶  In dendrochronology, David Yamaguchi showed that a single crossdating statistic returned 113 "significant at the 0.001 level" positions for one sample, including dates in the future.²⁷

Cycle-counting run past its evidence.  The Latemar, a Middle Triassic atoll in the Dolomites, contains about 600 metre-scale shallowing-upward cycles in 470 metres of rock.  They were read as orbital precession cycles and yielded a duration of 9 to 12 million years.  Uranium-lead zircon dating, ammonoid biostratigraphy and magnetostratigraphy together put the same interval at about one million years — an order of magnitude less.¹²  The beds are now interpreted as millennial, roughly 1.7 thousand years each.¹²

The Latemar is the most useful single item in this article, because it answers two opposite bad arguments at once.  Against the charge that cycle-counting is unfalsifiable:  here it was falsified, by radiometric dating, and the field changed its mind in print.  Against complacency:  six hundred visually obvious, beautifully rhythmic beds were confidently misread by a factor of ten, so "it looks periodic" is not by itself evidence of anything.

Tie-point circularity.  The real version of the charge.  Once a chronology is anchored to an external marker, the anchor is no longer independent evidence for it.  The Suigetsu varve chronology is anchored to the tree rings;  its absolute scale below 15,000 years is tied to Hulu Cave uranium-thorium, and the authors say so:  "the Suigetsu record on its own would not be useful without the age constraints provided by the speleothem and coral data with their direct U-Th ages."⁶  Anyone presenting Suigetsu as a standalone 50,000-year count is overclaiming.

The answer is not to deny the anchoring but to point at the anchors that owe nothing to the thing being tested:  the 774 CE spike, the argon-dated Laschamp horizon, the beryllium-10 record, and the raw-pMC test of Part 3, none of which were used to build the counts they check.

Part 6: What a young earth requires, in numbers

Set out the requirement plainly.  On a chronology with the Flood about 4,300 years ago and Creation about 6,000, here is what each archive demands.

Three observations about that table.

The required rates are not absurd individually.  Two rings a year in a bristlecone is not physically impossible;  Walter Lammerts induced an extra ring in seedlings by imposing a late-season drought and then re-watering.³  Eleven diatom blooms a year in a lake fed by ice-age dust pulses is not physically impossible either;  Oard's proposal that "dozens of diatom/clay couplets could have occurred each year" is a real mechanism, not hand-waving.¹⁸

The problem is that they have to happen together, at matched rates, in unrelated systems.  The Suigetsu laminae and the German oaks are 9,000 kilometres apart and are controlled by nothing in common.  For their counts to stay in step — which they demonstrably do, through 3,000 years of overlap, checked by raw carbon-14 — the diatom blooms in Japan and the growth flushes in a German floodplain would have to have been multiplied by the same factor at the same time.  There is no proposed mechanism that does that.

And the required rate has to decay away smoothly.  Post-Flood conditions are supposed to have relaxed toward the modern world.  So the multiplier must start high and fall to one.  That is a specific, testable prediction — and it is exactly what Aardsma tested.

Part 7: Aardsma's test, and why it is the most important paper in this subject

The argument is elegant and takes one paragraph to state.

The radiocarbon calibration curve, plotted against tree-ring number, has small wiggles in it.  They are not noise:  they are the record of the Sun going quiet.  During a solar minimum the shielding weakens, cosmic-ray flux rises, more carbon-14 is made, and trees growing then take up too much of it — so their radiocarbon ages come out anomalously young.  The Maunder Minimum (roughly 1645–1715) is one of these wiggles;  the Spörer Minimum is the next one back.  Both are visible in the American and the European chronologies, measured by different groups using different counting techniques.³

Now:  these wiggles have a characteristic width in years, set by how long the Sun stays quiet.  Aardsma measured the widths of the five wiggles in the last 3,000 years — the period where, he notes, "all recent-creationist scientists of which I am aware grant the general validity of dendrochronology and radiocarbon."  They average 70.5 years, splitting into a wide (Spörer-like, ~100 yr) and a narrow (Maunder-like, ~50 yr) family.³

Here is the test.  The horizontal axis of the calibration curve is not years;  it is tree rings.  If trees were putting on two rings a year in the early post-Flood period, then a 50-year solar minimum would occupy 100 rings, and the wiggle would appear twice as wide on the graph.  Multiple ring growth must show up as systematic broadening of the wiggles with depth.

He measured all thirteen well-defined wiggles across the whole 9,000-year record and plotted width against ring number, together with theoretical curves for a range of assumed Flood dates.  The result:

"it is apparent that the measured peak width data does not support any chronologically significant multiple tree-ring growth per year.  Rather, this data seems only supportive of single ring growth per year back at least as far as 9,000 BP (c. 7000 B.C.)."³

He also disposed of the assembly-error alternative — that the chronology was mis-joined so that contemporaneous trees were placed end to end — on the grounds that the calibration curve changes monotonically with no repeats or discontinuities, and that the short-term wiggle structure correlates in detail between chronologies on separate continents.³

And then he printed the conclusion, in the Creation Research Society Quarterly:

"these long dendrochronologies appear to consist of true annual rings for at least the past 9000 years.  This is an extremely important result which, in my opinion, warrants critical scrutiny by all recent-creationist scientists."³

"The extent of the presently available long dendrochronologies argues forcefully for a Flood date in excess of 10,000 years ago.  Efforts to overturn these long tree-ring chronologies and thereby allow for a more traditional Flood date have not proven successful when examined quantitatively."³

Two things need saying about this, one on each side.

Aardsma's own resolution — moving the Flood to roughly 12,000 BC and positing large gaps in the Genesis 11 genealogy — is not a position most young-earth creationists accept, and he acknowledges the cost:  "one is left with two looming questions.  First, where exactly do these gaps fall in this genealogy?  Secondly, what is the significance of the numerical information given in Genesis 11 if it is not to be used for direct chronological reckoning?"³  A reader is entitled to reject his answer.

But rejecting his answer does not touch his measurement.  The peak-width test is a physical test of the multiple-ring hypothesis, it was designed by someone who wanted the hypothesis to be true, and it came out against it.  What has been said against it since is one sentence.  The movement's 2018 review of its own tree-ring literature describes Aardsma's test accurately, including its cross-continental control, and offers a single counter:  "Of course, this requires the assumption of uniformitarian conditions in sunspot activity and radiocarbon flux since the Flood."¹⁵  That is a fair objection in principle and nobody has developed it into a measurement.  Meanwhile Hebert, Snelling and Clarey's 2016 article — the most-cited recent young-earth treatment of tree rings, varves and radiocarbon — argues at length that multiple rings per year are "commonplace" and does not cite Aardsma at all, and neither does the 2026 Answers Research Journal simulation that is now the state of the art on that side.

And Aardsma had a second argument, from the opposite direction.  Answering Robert Brown's competing model, he computed what it required of the trees:  a bristlecone living about 6,000 years ago would have had to make "580 rings in 80 years, or 7 rings/year," with some cases needing "up to 20 rings/year."  Answering Russell Humphreys's proposal that a weak post-Flood magnetic field could compress the curve, he computed four rings per year — and added that a climate wet enough to produce that many growth flushes would not be bristlecone habitat.¹⁵¹⁸

Part 8: Three places where the clocks caught each other

The claim that these methods are merely tuned to agree is best answered by the cases where they did not agree, and something had to give.

A counting clock caught a decay clock.  Before 2008, the astronomical and radioisotopic time scales disagreed by "∼1.0% or more."²⁸  The resolution was to recalibrate the argon-argon standard against the astronomical scale, moving the Fish Canyon sanidine age to 28.201 ± 0.046 Ma and cutting the method's absolute uncertainty from about 2.5% to 0.25%.  That is a counting clock correcting a decay clock — and it is also the reason to use the Triassic uranium-lead case rather than the Neogene argon case as the headline test, because after 2008 the Neogene agreement is partly built in.

A decay clock caught a counting clock.  The Latemar, above.  Six hundred cycles, an order of magnitude wrong, corrected by uranium-lead.

An outside clock caught both.  Sigl and colleagues found that Greenland and Antarctic ice chronologies were seven years off at 775 CE, detected by matching ice beryllium-10 to tree-ring carbon-14.  Among the named causes, in the authors' own words:

"use of manual-layer interpretation techniques that may favor interpretations consistent with a priori knowledge or existing chronologies"¹⁰

That sentence is a genuine admission of the exact bias the young-earth critique alleges, published by the people responsible for the chronologies.  It should be quoted, and it should be quoted with what follows:  the revised scales are accurate "to within less than ±5 years during the past 2,500 years," and when the independently recorded volcanic dust-veil years were tested against the old chronology there was no statistically significant association at all, while against the new one 24 of 32 fall within ±3 years.¹⁰

A framework that is merely self-consistent does not do that to itself.

Part 9: Where the young-earth case actually stands

The literature this series answers is not static, and the popular version of it is a generation behind the technical version.  A sweep of the creationist periodicals turns up three distinct stages.

Stage one, roughly 1961–1990:  the units are not annual.  Multiple rings per year, multiple laminae per year, storm layers in ice.  This is the argument almost every popular article still runs, and it is the one with the best experimental support — Lammerts really did induce an extra ring in a bristlecone seedling, and Lambert and Hsü really did find five laminae in a year at Walensee.

Stage two, roughly 1976–2000:  the joins are not secure.  Complacent bristlecone rings, autocorrelated crossdating statistics, "accordioned" ice-core correlations, De Geer's varve matching.  The strongest statements of this are Sorensen's 1976 critique, Brown's 1995 Origins paper, and Woodmorappe's 2002 study of the GRIP–GISP2 disagreement.  It is a better argument than stage one because it does not require any layer to be non-annual.

Stage three, 2001–present:  the layers are annual, the statistics are sound, and the matches are misplaced.  Woodmorappe's disturbance model, in which contemporaneous trees crossmatch because they shared an earthquake or a landslide rather than a climate, and the bundles are chained by fortuitous bridge series;  and its 2026 computational implementation, which builds a statistically robust 8,000-year chronology out of trees that all lived within 4,500 years.²⁹

Stage three is the one worth arguing with, and this series argues with it in Tree-Ring Dating.  Its problem is that it makes a prediction its authors have not tested.  If a chronology is a chain of contemporaneous bundles, then wood at ring 8,000 and wood at ring 4,000 grew at nearly the same time and holds nearly the same carbon-14 — so the raw-pMC plot of Part 3 must show flat treads.  It shows a smooth curve.

That is the shape of the whole subject.  The young-earth case has moved steadily toward more sophisticated and more testable claims, and the tests keep coming back the same way.

Objections and responses

"These counts assume the layers are annual.  That is the whole question." It is the whole question, and it is not assumed;  it is tested, three ways.  Against an external single-year marker (the 774 CE spike, in 44 chronologies at once).  Against a decay clock on the same material (uranium-thorium at Suigetsu, argon-argon at the Laschamp horizon in Greenland ice).  And against raw carbon-14 with no calibration at all (Part 3).  The layers could in principle be non-annual;  they could not be non-annual and pass those three tests.

"Lambert and Hsü found five laminae in one year, so varve counting is worthless." Lambert and Hsü found up to five in one year at some locations in Lake Walensee, and averaged two.  That is a real and important result, and it is why chronology is not done in glacier-fed lakes with year-round turbidity input.  Two to five is also not eleven, which is what a young-earth Suigetsu requires.  The finding constrains which lakes can be used;  it does not abolish the method.

"Half the Suigetsu varves were interpolated, not counted." Correct, and it comes from the counters themselves.⁶  The measurable consequence is a 5–7% shortfall against the uranium-thorium chronology below 20,000 years — the count is too short, not too long.⁶  Interpolation is a real weakness of Suigetsu specifically, which is why the article's load-bearing argument is the raw-pMC slope test rather than the absolute varve number.

"The oldest living bristlecone has only about 4,800 rings, which is roughly the biblical age of the earth.  Why are there no trees with 20,000 rings?" Because trees die.  The limiting factor on a bristlecone's age is not the calendar but erosion of the substrate it stands on, and the chronology is built by overlapping thousands of dead trees, not by finding one very old one.  The Hohenheim oak chronology runs 12,410 years and the mean age of its oaks is 176 years, with 97% under 300;  it is long because more than 7,000 trees overlap, with a mean of 108 simultaneous witnesses in every year.⁴  The objection assumes a chronology needs one long-lived tree.  None is built that way.

"Tree-ring alignments from Northern Ireland, southwest England and southern Germany disagree strongly." This claim appears on creation.com without a source.³⁰  The primary literature says the opposite:  the Hohenheim links "were confirmed by comparisons to the independent Irish oak chronology (Pilcher et al. 1984) and the Göttingen oak chronology (Leuschner 1992) back to 9741 BP."⁴  And IntCal20 measures the Irish-oak-versus-Californian-bristlecone difference at 8.1 ± 1.9 radiocarbon years.²⁵  If the claim has a source it needs producing;  as stated, it is contradicted by the papers.

"Cross-dating is subjective — an art, not a science." Partly true, and dendrochronologists say so:  standard practice is to date visually first and use the statistics as a check, not the reverse.¹⁸  The reply is not that the method is objective but that it is externally verifiable, and it has been verified — by the 774 CE marker, and by the case of the Norfolk timber circle, where the dendrochronologists refused to publish a date on a correlation of t = 3.98 because it was not significant, went to high-precision radiocarbon instead, and got the same answer.²⁷

"The ice cores, the varves and the tree rings are all calibrated against each other, so of course they agree." Some of the links are real and must be conceded — Suigetsu's absolute scale is tied to Hulu Cave;  the Neogene argon scale is tied to astrochronology.  But the specific links that carry the argument are not of that kind.  The 774 CE spike was measured in tree rings and in ice beryllium-10 independently, and it disagreed with the ice chronology by seven years, which is how the error was found.  Argon-argon dates on the Laschamp and Z2 horizons were produced by different laboratories for different reasons and then compared to the layer count.  And the raw-pMC test uses no chronology at all on its vertical axis.

"Accelerated nuclear decay during the Flood explains the radiometric ages." Perhaps, but it does not explain these.  There is no decay constant in a tree ring.  Young-earth authors agree:  Hebert, Snelling and Clarey state that decay rates "have been essentially constant during most of the post-Flood period" and that carbon-14 in particular would have been barely affected even during acceleration.¹⁸  So on their own account the whole counting-clock problem has to be solved by multiplying layers, not by changing physics.

"Deep ice and deep ocean records are orbitally tuned, which is circular." That charge is correct for the deep records and this article does not use them as counting clocks.  Antarctic ice below the counted section is modelled;  deep marine chronologies are tuned.  The counting evidence stops at 60,000 years in ice and about 50,000 in lake sediment.  It is more than enough to decide the question at issue.

Issues and unanswered questions

Green River is not settled here.  Bradley's 1929 lamina count gave the Eocene "a little less than 23,000,000 years," which is within about 4% of the modern radiometric figure of 22.1 Myr and was obtained "entirely independent of" radiometric dating.²¹  Against that stands a 1988 conference abstract by Buchheim and Biaggi reporting that laminae counts between the same two tuff beds range from 1,160 to 1,568 — a lateral variation of about 35% in a supposedly isochronous interval.³¹  That abstract could not be obtained and has not been verified;  if it holds it is a serious problem for the Green River count specifically, and it should be printed.  See Lake Varves.

The Suigetsu 5–7% residual is real and unexplained in detail.  Its authors offer a mechanism — thin varves under-represented in low-accumulation intervals — and label it "theoretical" for want of independent proof.⁶

Nobody has published a quantitative young-earth model of layer multiplication.  The qualitative mechanisms exist (dust pulses, storms, turbidites, drought rings).  What does not exist is a calculation that produces eleven laminae per year in an anoxic Japanese lake and three rings per year in a German floodplain oak and fourteen layers per year in central Greenland, with the multipliers decaying to one on a common schedule.  Until that exists, the young-earth position on counting clocks is a set of possibilities rather than a model.

The 40,000-year kauri conflict is unresolved.  A floating New Zealand kauri sequence placed by beryllium-10 matching to the ice comes out about 1,000 years younger than its placement against Hulu Cave carbon-14, and the specialists conclude the ice placement is probably the wrong one.³²  Around 30,000 years the Greenland scale carries about 1,000 uncertain years.³²

Verdict

The counting clocks are the part of the age-of-the-earth debate where the young-earth position is weakest, and it is worth being direct about why.

Everywhere else, the argument turns on assumptions:  whether decay rates were constant, whether a rock was closed to migration, whether present rates are a guide to past rates.  Those are real arguments and the young-earth side has made real contributions to them.  Here the argument turns on something much more concrete — whether a tree puts on one ring a year — and the answer has been checked against a solar-particle event that fixes 44 chronologies on five continents to a single calendar year.

What survives from the young-earth critique, and survives well:  the counting records are much shorter than popular writing claims;  half the Suigetsu varves are interpolated;  ice-core chronologies have been tuned to markers that turned out to be misidentified;  crossdating statistics can return confident nonsense;  and a whole coral archive was deleted from the calibration curve for disagreeing.  Anyone who presents these chronologies as clean and unproblematic is not describing the literature.

What does not survive:  the conclusion.  Every documented over-counting mechanism gives a factor of two to five, in settings deliberately excluded from chronology;  a young earth needs ten to fourteen, sustained, in matched step, across archives with nothing physically in common.  The one quantitative test of that hypothesis ever published was run by a creationist and came out negative.  And the raw carbon-14 measurements — no calibration curve, no assumed ages, no conversion — sit inside the band the conventional model predicts, from the surface to 50,000 layers.

That is the honest state of the evidence, and it runs against a young earth.

Further reading

• Ice-Core Layer Counting
• Lake Varves
• Tree-Ring Dating
• Growth Bands and Tidal Rhythmites
• Orbital Cycles in Rock
• Radiocarbon Dating — the calibration curve, and the carbon-14-in-coal argument
• Do the Dating Methods Agree? — cross-method agreement among the decay clocks
• The Ice Age — the post-Flood ice-age model these arguments are embedded in

Sources