Growth Bands and Tidal Rhythmites

"It is not claimed that coral growth proves that either is right;  but it is suggested that palaeontology may well be able to supply a third stabilizing, and much cheaper, clue to the problem of geochronometry, and that further search for diurnal or circadian records in groups other than corals may result in strengthening this weak anthozoan prop." — John W. Wells, Cornell University, 1963¹

"The great hopes held for paleontological 'clocks' have not, however, been fully realized." — George Williams, University of Adelaide, in Reviews of Geophysics, 2000²

"the mean rate of lunar recession since that time is 2.17 ± 0.31 cm/yr, which is little more than half the present rate of lunar recession of 3.82 ± 0.07 cm/yr obtained by lunar laser ranging." — George Williams, on rhythmites deposited about 620 million years ago²

Summary

Tides slow the Earth's rotation.  The angular momentum has to go somewhere, so it goes into the Moon's orbit, and the Moon moves away.  Both effects are measured today:  the day is lengthening by about 1.7 milliseconds per century, and laser ranging off the Apollo retroreflectors gives a recession rate of 3.82 ± 0.07 cm per year.

Run that backwards naively and the Moon was touching the Earth about 1.5 billion years ago, which is a serious-looking problem for a 4.5-billion-year Earth.  This is one of the better young-earth astronomical arguments, and the standard reply — that tidal dissipation depended on the arrangement of the continents and was lower in the past — looks like special pleading until somebody measures it.

Somebody measured it.  Not with a model:  by counting laminae in a rock.

Eight conclusions, stated first.

1. A tidal rhythmite records the day, the month and the year at once. Each lamina is one tide. Laminae thicken and thin in a fortnightly neap–spring cycle, so counting laminae per cycle gives days per month; the cycles themselves are modulated by the lunar nodal period, so counting cycles gives months per year. Nothing about this involves radioactivity.
2. The Elatina and Reynella rhythmites of South Australia give 400 ± 7 solar days per year, 13.1 ± 0.1 synodic months per year, and a day of 21.9 ± 0.4 hours.² The record is about 60 years of continuous tidal deposition.
3. Three of those quantities are independent, and they agree. The lunar nodal period (19.5 ± 0.5 yr), the number of sidereal months per year (14.1 ± 0.1) and the number of sidereal days per year (401 ± 7) each give the Earth–Moon distance by a different equation of celestial mechanics. The three answers are a/a₀ = 0.969 ± 0.017, 0.965 ± 0.005 and 0.968 ± 0.007 — agreement to about 0.4%.² This test uses no age at all.
4. The implied recession rate is 2.17 ± 0.31 cm/yr, not today's 3.82. So the Moon has been receding more slowly on average than it is now, which is exactly what the tidal-dissipation argument predicts and what critics said was unfalsifiable. Extrapolating today's rate backwards puts a close approach at ~1.5 Ga; using the measured past rate pushes it past 3 Ga.²
5. The mainstream has already discarded the coral clocks, and creationists do not need to. John Wells's 1963 Devonian corals gave "usually about 400, ranging between extremes of 385 and 410" growth lines per year, which looks like a beautiful confirmation.¹ Williams's assessment thirty-seven years later: the Paleozoic values are "of dubious reliability; two values have large errors, two other values are approximations, and five of the highest values are for maximum rather than mean value."² Do not lean on Wells.
6. Wells himself said so. His paper calls its own result "this weak anthozoan prop" and states outright: "It is not claimed that coral growth proves that either is right."¹ Almost nobody who cites Wells quotes that sentence.
7. Scrutton's follow-up contains a circularity that has to be admitted. Chris Scrutton's 1964 measurement of thirteen months in the Devonian year is often cited alongside Wells. But Scrutton could not identify annual markings on his corals, so he took the 399-day year from the astronomical extrapolation and divided by his measured band length.³ What he independently measured is the month — 30.59 days against today's 29.53 — not the year.
8. The young-earth reply to the rhythmites is a circularity charge, and it is half right. Jonathan Henry's Journal of Creation treatment grants that rhythmites are the mainstream's answer and responds that "varves themselves are dated with respect to the presumed age of the earth. Thus lunar recession rates derived from such varve chronologies constitute circular reasoning."⁴ He is right that converting the rhythmite data into a recession rate in cm/yr requires the radiometric age of the formation. He is wrong that the whole result does: the day count, the month count, and the three-way internal consistency check are pure counting, and it is those that show the Earth's rotation has slowed by a measured amount.

The thing that is actually surprising

This is the one place in the whole non-radiometric-clocks debate where a counting clock is used to refute a young-earth argument rather than merely to resist one.  The lunar-recession argument says the Earth–Moon system cannot be older than about 1.3 billion years.  A rock that counts tides says the Moon has been receding at about half today's rate, which is precisely the condition the argument assumes cannot be established.  And the rock says it three ways at once.

Part 1: Why a growth band could be a clock

Earth's rotation is slowing.  The tidal bulge raised by the Moon is dragged ahead of the Earth–Moon line by the Earth's rotation;  its gravity tugs the Moon forward, boosting the Moon's orbit;  the reaction torque brakes the Earth.  Angular momentum is conserved, so as the day lengthens the Moon recedes.

Two consequences follow, and both are countable in the geological record:

• There were more days in the year in the past, because the year is set by the Earth's orbit and does not change, while the day gets longer.
• There were more days in the month, and more months in the year, because the Moon was closer and orbited faster.

Anything biological or sedimentary that records a daily rhythm and an annual or monthly one is therefore an astronomical clock — with no nuclear physics in it anywhere.

Wells put the point sharply in 1963, and it is worth quoting because it is the founding statement of the whole enterprise:

"ABSOLUTE age determinations of points on the geological time-scale, based on radioactivity, are generally accepted as the best approximations now known, even though they rest on a series of assumptions, any one of which may be upset at any time.  At present there is no means of confirming or denying the accuracy of these determinations by independent methods.  Can palaeontology provide anything by way of verifying the pronouncements now emanating, at very considerable expense, from the black boxes?"¹

Part 2: The coral clocks, and why they failed

What Wells did

Wells noticed that coral epitheca — the outer skeletal wall — carries fine ridges parallel to the growing edge.  He argued they were daily.  His control was a modern coral:

"the number of ridges on the epitheca of the living West Indian scleractinian Manicina areolata hovers around 360 in the space of a year's growth.  This strongly suggests, subject to experimental confirmation, that the growth-lines are diurnal or circadian in nature."¹

Then the fossils, from the Middle Devonian of New York and Ontario — Heliophyllum, Eridophyllum, Favosites:

"in every instance there are more than 365 growth-lines per annum, usually about 400, ranging between extremes of 385 and 410.  It is probably too much, considering the crudity of these data, to expect a narrower range of values for the number of days in a year in the Middle Devonian."¹

And two Pennsylvanian corals:  Lophophyllidium from western Pennsylvania at 390 lines per year, Caninia from Texas at 385.

Astronomical theory at the time (Munk and MacDonald 1960) predicted a slowdown of "about 2 sec per 100,000 years," giving roughly 400 days per year in the Devonian.  Wells got 400.

Why it is not as good as it looks

Wells's own caveats.  He is unusually careful.  He notes that the annual annulations must be assumed ("We must now make an act of faith"), that "no experiments have been undertaken on living corals to determine directly whether or not there is a causal relation between such annulations and annual variations," and that his fossil material was "the best of the limited fossil material I have examined so far."  His closing sentence disclaims the inference entirely.¹

Scrutton's circularity.  Chris Scrutton's 1964 study measured 112 growth bands on ten Devonian corals and found a mean of 30.59 ridges per band, from which he inferred thirteen lunar months in the Devonian year.  The arithmetic is 399 ÷ 30.59 = 13.04.  But where does 399 come from?  From the astronomical extrapolation — because Scrutton could not find annual markings:  "it has been concluded from the material examined here that, for these specimens at least, no regular grouping of ridges larger than the bands described above can be distinguished."³

So Scrutton's often-quoted "thirteen months in the Devonian year" is not an independent measurement of the year.  What is independent is the month:  30.59 days per synodic month against 29.53 today.  And Scrutton was candid about the failure modes:

"The possibility must also be borne in mind that if these fine growth-ridges are diurnal as it is claimed, unusual circumstances may have caused an increment to be missing for certain days, or conversely, two increments to be deposited in the same day.  Thus all ridge counts must be regarded as approximations."³

The mainstream verdict.  George Williams, reviewing the whole field in Reviews of Geophysics, is blunt:

"The great hopes held for paleontological 'clocks' have not, however, been fully realized.  The Paleozoic data, if taken at face value, imply little overall change in the rate of tidal energy dissipation over the past ~500 Myr.  Sustaining that high rate of dissipation implies a catastrophic close approach of the Moon near 2000 Ma, so raising doubts about the reliability of much paleontological data."²

and on the numbers themselves:

"These particular values derived from fossils are of dubious reliability;  two values have large errors, two other values are approximations, and five of the highest values are for maximum rather than mean value.  Scrutton and Hipkin [1973] and Scrutton [1978] argued that to accept only maximum counts for fossil data could lead to serious overestimates of the length of cycles if semidiurnal growth increments are counted inadvertently as diurnal increments."²

And Scrutton himself changed his mind.  Williams quotes him in 1978 saying that the assumption that growth increments record the solar day and the synodic month "may not be justified 'because of earlier misunderstanding of the interplay between lunar and solar day stimuli and the effects of tide type'."²

A retraction worth knowing about.  A stromatolite reported by Vanyo and Awramik in 1985 gave about 435 days per year and was attributed to the ~850 Ma Bitter Springs Formation.  Later fieldwork reassigned it to the Olympic Formation, "broadly coeval with or slightly younger than the ~620-Ma Reynella Siltstone and Elatina Formation" — at which age its 435 days per year contradicts the rhythmite value of 400 ± 7.²

The young-earth side reached the same verdict, independently

This article's judgement on Wells and Scrutton is not a concession wrung out of anyone.  It is also Creation Ministries International's position, argued by Robert Carter, who holds a PhD in coral reef ecology:

"Wells was one of the giants among coral reef scientists in his day.  Yet, Scrutton (1964) said he found it impossible to demarcate annual bands, casting a strong negative light on the claims of Wells.  But Scrutton did notice strong monthly patterns.  If the bands are monthly, there is no annual pattern to be seen, and so nobody knows what Wells was talking about."⁵

Carter also makes three points that are correct and that any use of Devonian corals has to clear:  a coral is not a reef, and reef growth rates have nothing to do with individual coral growth rates;  there are no Devonian coral reefs in the sense modern reefs are reefs;  and Devonian rugose and tabulate corals are extinct, had different symmetry, and secreted calcite rather than aragonite, so extrapolating modern banding behaviour to them requires an argument nobody has made.⁵

Where his account should be checked is the arithmetic.  Carter concludes:  "If those are monthly signals, the number of days in a month is about the same as today."⁵  Scrutton's measured figure is 30.59 solar days per synodic month.  Today's synodic month is 29.53 days.  That is not the same;  it is 3.6% longer, and it is longer in exactly the direction a slowing Earth predicts.  The month count is the one quantity Scrutton measured without importing an astronomical assumption, and it does not support the claim built on it.

That said, one measurement on ten specimens with a range of 27 to 35 ridges per band is not much to rest anything on, in either direction — which is why this article rests its case on the rhythmites instead.

The one modern successor

De Winter and colleagues in 2020 redid Wells's experiment with laser-ablation mass spectrometry, reading trace-element cycles across ~40-µm daily growth laminae in a Campanian rudist bivalve.  The publisher's abstract reports that "the rudist precipitated 372 daily laminae per year, demonstrating that length of day has increased since the Late Cretaceous, as predicted by astronomical models."⁶  At about 70 Ma, 372 days corresponds to a day of 23 h 31 min.

The full paper could not be obtained — Wiley's servers refuse every client available here, despite the article being open access — so nothing beyond the abstract is used.

Part 3: The rhythmite clock

What a tidal rhythmite is

In a wave-protected setting behind a tidal inlet, each tide lays down one lamina of fine sand or silt.  Tidal range varies over the fortnightly neap–spring cycle, so lamina thickness varies in a regular pattern:  thick laminae at spring tides, thin at neaps.  The bundles of laminae between successive neaps are countable.  And the whole pattern is modulated on longer periods — the lunar nodal cycle of 18.6 years today, and the cycle of lunar apsides at 8.85 years.

So a rhythmite gives, directly by counting:

• laminae per neap–spring cycle → days per half synodic month
• neap–spring cycles per nodal cycle → months per nodal period
• and from those, days per year and months per year

The Elatina Formation and Reynella Siltstone of South Australia preserve about 60 years of continuous tidal record, in the finest-resolved case.  Williams's Table 1 (column 6) and Table 2:

The internal consistency test, which uses no ages at all

This is the part that matters most, and it is worth being precise about why.

Three of the measured quantities are primary:  the nodal period (19.5 ± 0.5 yr), the number of sidereal months per year (14.1 ± 0.1), and the number of sidereal days per year (401 ± 7).  Each is read off the rhythmite by counting.  Each can be converted into an Earth–Moon distance by a different equation of celestial mechanics — the nodal period depends on the Moon's orbital precession, the month count on Kepler's third law, the day count on angular-momentum bookkeeping including the solar tide.

If the laminae were not tidal, or the counts were wrong, the three would disagree.  They do not:

Williams's own conclusion:

"This accordance of results, derived from independent, widely separated values by applying three different equations of celestial mechanics that make allowance for lunar and solar tidal effects, demonstrates the internal self-consistency of the observed Elatina-Reynella values."²

Note carefully what this test does and does not use.  It uses no radiometric date, no assumed age, no calibration curve.  It is a check that three counted quantities in one rock are mutually compatible with Newtonian celestial mechanics for a single Earth–Moon distance.  On a young-earth reading, in which the whole formation was deposited within the Flood year, the rhythmite would have to record — by coincidence — a self-consistent Earth–Moon configuration about 3.5% tighter than the present one, in three different ways.

Williams is also careful about scope, and it should be repeated:  "the value of 400 ± 7 solar days/yr and a l.o.d. of 21.9 ± 0.4 hours at ~620 Ma should be viewed as the only validated paleorotational data currently available for the Neoproterozoic and early to middle Paleozoic (1000–350 Ma)."²  One good datum, not a series.

The Paleoproterozoic datum, flagged as provisional

Cyclic banded iron formation of the Weeli Wolli Formation, Hamersley Basin, Western Australia, dated 2450 Ma, gives 14.5 ± 0.5 synodic months per year and a/a₀ = 0.906 ± 0.029.²  Williams marks the derived values in that column as unverified and does not claim internal consistency for them.  Combining Weeli Wolli with Elatina gives a mean lunar recession of 1.24 ± 0.71 cm/yr through most of the Proterozoic — "suggesting that a close approach of the Moon did not occur during earlier time."

Part 4: The lunar-recession argument, and what the rhythmites do to it

The argument as its best advocate states it

Jonathan Henry's 2006 Journal of Creation paper is the fullest recent young-earth statement, and it is careful.  Its structure:

• The Moon cannot have formed inside the Roche limit — debris there cannot coalesce. So naturalistic origin theories put the Moon at the Roche limit, at r₀ = 1.84 × 10⁷ m, less than 5% of its present orbital radius.⁴
• Tidal recession goes as the inverse sixth power of distance: dr/dt = k/r⁶. Integrating from the Roche limit to the present distance, with the present recession rate fixing k, gives 1.3 Ga as the maximum age of the Earth–Moon system.⁴
• Run it forwards instead: starting at the Roche limit 4.6 Ga ago, the Moon should now be at 4.7 × 10⁸ m — 20% further away than it actually is.
• The standard reply is that the tidal dissipation constant was smaller in the past, because the arrangement of the continents matters and a globally open ocean dissipates least. Henry calls this "tantamount to saying that the proportionality constant k… is actually variable… and must be adjusted to bring lunar chronology in line with that of the earth," and quotes Mignard: "even if we have sound reasons to accept a substantial reduction of the dissipation in the past, we are still lacking evidence of what the Moon's orbit looked like 3 or 4 billion years ago."⁴

That last quotation is the hinge.  The reply to the young-earth argument requires the recession rate to have been lower in the past;  the young-earth response is that nobody has evidence of what it was.

What the rhythmites supply

Evidence of what it was.

Elatina–Reynella at ~620 Ma:  mean recession 2.17 ± 0.31 cm/yr.  Today:  3.82 ± 0.07 cm/yr by laser ranging.  The past rate is 57% of the present one.  Williams:

"employing an average rate of tidal energy dissipation consistent with the Elatina datum of a/a₀ = 0.965 and the indicated mean rate of lunar recession of 2.17 cm/yr since 620 Ma pushes back a possible close approach of the Moon to 3 Ga."²

And adding the Weeli Wolli datum gives 1.24 ± 0.71 cm/yr for 2450–620 Ma, which pushes it back further.

Henry's answer, and how much of it survives

Henry does engage the rhythmites.  His answer is threefold:  the data sets are "often short, and periodicities must be interpreted from selected data sets";  the same varves were previously interpreted as a solar record before being reinterpreted as lunar;  and, decisively for him, "varves themselves are dated with respect to the presumed age of the earth.  Thus lunar recession rates derived from such varve chronologies constitute circular reasoning as 'evidence' that the moon is old."⁴

The circularity charge is partly correct, and it should be granted.  Converting the Elatina data into a rate in centimetres per year requires dividing a distance change by an elapsed time, and the elapsed time — 620 million years — is radiometric and stratigraphic.  Anyone who presents "2.17 cm/yr" as an assumption-free measurement is overstating it.

But it does not reach the part of the result that matters.  Four things in the Elatina data are independent of the assigned age:

1. 400 ± 7 solar days per year. This is a ratio of two counted quantities in one rock.
2. 13.1 ± 0.1 synodic months per year. Likewise.
3. A day of 21.9 ± 0.4 hours, which follows from (1) by arithmetic.
4. The three-way internal consistency, which asks only whether the counted quantities are mutually compatible with celestial mechanics for one Earth–Moon distance.

Those four say, without reference to any age:  this rock recorded a faster-spinning Earth and a closer Moon, and it recorded them coherently.  On a young-earth chronology every rock in the geological column was deposited within a few thousand years of the present, so the Earth's rotation cannot have differed measurably — and this rock says it did, by 9%.

Henry's "previously interpreted as solar" point is fair and worth stating.  The Elatina laminae were read in the 1980s as a record of the sunspot cycle before being reinterpreted as tidal — Bracewell, and Williams himself, published on that reading.  A record that has been reinterpreted once can be reinterpreted again.  The reason to prefer the tidal reading is not authority but the internal consistency test:  a solar interpretation has no reason to make three different equations of lunar celestial mechanics return the same Earth–Moon distance.

And his "short data sets" point is also fair.  Sixty years of tidal record is sixty years.  Williams's own restriction — one validated datum for the whole interval 1000–350 Ma — concedes the thinness of the evidence.

The newest form of the objection

The most recent young-earth engagement with this literature is not about rhythmites at all but about their successors:  cyclostratigraphic reconstructions of the Earth–Moon distance from orbital cycles in layered rock.  Michael Thompson's August 2026 critique of Huang et al. 2024 grants the general shape of the enterprise and attacks the inputs — the assumed radiometric ages of the eight sections, the quality of the Milankovitch matching in some of them, and the fact that the Bayesian inversion producing the Earth–Moon distance is bounded by a global tidal model.⁷  It is treated in Orbital Cycles in Rock.  It does not touch the Elatina internal-consistency test, which uses no model and no age.

The independent theoretical answer

There is now also a modern dynamical treatment of the problem from the theory side.  Farhat and colleagues (2022) model the Earth–Moon tidal history with ocean resonances rather than a constant dissipation parameter, and recover a history in which the Moon's recession rate has varied substantially, avoiding a close approach.⁸  Two other studies constrain the Earth–Moon system independently at 2.46 Ga and 1.4 Ga from Milankovitch cycles preserved in banded iron formation and in shale.⁹

These are model results, not counts, and they do not carry the same weight as the rhythmite data.  They are noted because Henry's argument specifically charges that the reduced-dissipation reply is unmotivated, and by 2022 it has a physical mechanism attached.

Objections and responses

"Wells's corals gave 400 days in the Devonian year.  That is a striking confirmation of the radiometric time scale." It looks like one, and it should not be used.  Wells disclaimed the inference himself, calling his own result "this weak anthozoan prop."  Williams's review finds the Paleozoic paleontological values "of dubious reliability" — several are maxima rather than means, which inflates them, and taken at face value they imply a tidal history that puts the Moon at the Earth's surface 2 billion years ago.  The mainstream discarded these clocks;  there is no reason for anyone else to defend them.

"Scrutton found thirteen months in the Devonian year." He found 30.59 ridges per band and divided 399 by it.  The 399 came from astronomy, not from his corals, because he could not identify annual markings.  His independent result is the month length, not the year.

"Growth bands can be missing or doubled." Scrutton says so explicitly and it is the standard objection.  It is also why the argument in this article rests on the rhythmites rather than the fossils:  a missing or doubled tidal lamina shifts one count, but it cannot make three separate equations of celestial mechanics agree on the same Earth–Moon distance.

"Tidal rhythmites are dated by the same radiometric methods under dispute, so this is circular." For the recession rate in cm/yr, correct — that number needs an elapsed time and the elapsed time is radiometric.  For the day count, the month count, and the internal consistency test, incorrect:  those are ratios of counted quantities within one rock.

"The Elatina laminae were once interpreted as a solar-cycle record." They were, and the reinterpretation is a real weakness in the record's provenance.  What decides between the readings is that the tidal interpretation makes three quantitative predictions that check against each other and a solar interpretation does not.

"The rhythmite record is only 60 years long." It is, and Williams says the Elatina–Reynella datum is the only validated one for a 650-million-year interval.  Sixty years is enough to resolve the nodal cycle three times over, which is what the internal consistency test needs, but nobody should present the rhythmite evidence as abundant.

"The Moon would have been touching the Earth 1.5 billion years ago." That calculation assumes the present recession rate held throughout.  A rock that counts tides says it did not — it says the average since 620 Ma was 2.17 cm/yr, 57% of today's.  The young-earth argument requires the past rate to be unmeasurable.  It has been measured.

Issues and unanswered questions

• De Winter et al. 2020 could not be obtained. The paper is nominally open access and Wiley's servers refuse every client available here. Only the publisher-deposited abstract is used, and nothing is built on it.
• Scrutton 1978 — the source of the field's own retreat from coral clocks, quoted by Williams — has not been obtained verbatim.
• Lambeck 1980, the standard treatment showing that face-value Paleozoic paleontological data force a close approach near 2000 Ma, has not been obtained. It is used here only as Williams reports it.
• The Weeli Wolli values are flagged unverified by Williams himself, and no independent confirmation of them has been found.
• The Elatina reinterpretation history (solar → luni-solar → lunar) is real and has not been traced through the primary literature for this article. Henry's account of it is the only one consulted.

Verdict

The paleontological growth-band clocks are a cautionary tale, not evidence.  They looked like a beautiful independent confirmation of the radiometric time scale in 1963, they were promoted as such for decades, and the field quietly discarded them because the numbers, taken seriously, imply a lunar history that cannot have happened.  Anyone still citing Wells's Devonian corals as confirmation of deep time is citing something the specialists gave up on.

The tidal rhythmites are different, and they are the best non-radiometric clock in this series that does not depend on an assumed chronology at all.  Three counted quantities in one Australian sandstone, run through three different equations of celestial mechanics, return the same Earth–Moon distance to within half a per cent — a distance about 3.5% smaller than today's.

That result answers, with a measurement, the one thing the young-earth lunar-recession argument insists cannot be measured.  It is the clearest case in this whole subject of a counting clock deciding a question against a young earth, and it does it without touching a decay constant.

Further reading

• Non-Radiometric Clocks — the overview of this series
• Orbital Cycles in Rock — the other astronomical counting clock
• Lake Varves — the sedimentology of laminated deposits, treated at length
• The Solar System's Age

Sources