The Order of the Fossil Record

"We know of no undisputed human skeletal remains in Flood rocks." — Brian Thomas, Institute for Creation Research, 2015.¹

"It remains a mystery why no undisputed pollen has been found in the earliest megasequences." — Jeffrey Tomkins and Timothy Clarey, ICR, 2023, on the single hardest fact for their own model.²

"The geologic column, which, as a matter of fact, was put together before evolution was known." — J. E. O'Rourke, 1976, in the paper creationists quote most often about circular reasoning.³

Summary

Dig anywhere on earth and the fossils come out in the same order.  Trilobites below dinosaurs.  Dinosaurs below horses.  No rabbits at the bottom, no trilobites at the top.  Oil companies bet real money on this order every day, and it pays.

A global Flood lasting one year has to explain that order.  This article asks whether any young-earth explanation does.

One thing to get straight first, because popular writing on both sides gets it wrong.  Working young-earth geologists do not dispute that the order exists.  They have not for decades.  Their position is that the order records the sequence in which ecological zones were destroyed and buried — deep marine first, then shelf, then lowland, then upland — and the serious literature argues about the mechanism, the megasequence boundaries, and where the Flood ended, not about whether trilobites sit below dinosaurs.  Anyone who tells you creationists deny the order is describing a straw man.  This article engages the real position.

The verdict is still that no young-earth explanation accounts for the order, and this article says so.

That is not a comfortable conclusion for a site that argues for a young earth.  Two earlier articles here examined young-earth arguments about erosion rates and natural arches and concluded the arguments were wrong.⁴⁵  This is a third.  The reason to publish it is simple:  a reader who finds these facts somewhere else, after reading a version that hid them, has been given every reason to distrust everything else on this site.

What this article does that others have not

Most writing on this subject is a fight over quotations.  This one computes.

The Paleobiology Database holds 2,001,053 fossil occurrence records, each carrying a published reference, a collection number, and a stratigraphic assignment.⁶  It is open, free, and maintained by working paleontologists.  Macrostrat holds 35,481 rock units across 1,534 stratigraphic columns.⁷  Both were queried directly for this article.  The scripts are published alongside it, and anyone can re-run them.

That makes three things possible that a quotation fight cannot do.  The order can be measured rather than asserted.  Every claimed exception can be looked up individually.  And the best young-earth model of fossil order can be tested against a database its author never used.

None of this depends on radiometric dating

Radiometric dating is a separate subject and gets separate articles on this site.  Nothing here rests on it.

The order of the fossil record is a relative sequence.  It rests on superposition — younger rock sits on older rock — plus the lateral tracing of beds.  Nicolas Steno stated the principle in 1669, a century before anyone knew radioactivity existed.  Every period name in the column — Cambrian, Devonian, Carboniferous, Cretaceous — was fixed by British and Continental field geologists between 1799 and 1879, before a single absolute date had been measured.

Young-earth geology accepts superposition.  So a reader may reject every decay constant in physics and the order in this article does not move by one bed.

Ages in millions of years do appear below, because that is the vocabulary the databases use and because the young-earth authors under discussion use it themselves when they name a system or a stage.  Every one of them is a label for a rock interval, not a measurement offered as evidence.  Where a date is doing argumentative work — as in the retraction of the Triassic bird tracks — this article says so and states what is left of the argument without it.

That point has a creationist-friendly source.  J. E. O'Rourke's 1976 paper in the American Journal of Science is the most quoted stratigraphy paper in all of apologetics, usually for a line about circular reasoning.  The same paper says the column "was put together before evolution was known," and that "radiometric dating would not have been feasible if the geologic column had not been erected first."³³

What this article concedes before it argues anything

• The order is real and it is not close. Fifty-six major groups were surveyed against 2.0 million records. They sort into a strict staircase. The staircase is not built on lucky finds: the first appearance of trilobites rests on 17,885 Cambrian occurrences, of dinosaurs on 998 Triassic ones, of rabbits on 746 Paleogene ones.⁸
• Every claimed out-of-order fossil audited here failed. Twenty-nine were examined, one at a time, against a test stated in advance. Zero survived. Several were killed by creationists.⁹
• Hydrodynamic sorting is dead. Brachiopods and bivalves are near-identical in size, shape, shell chemistry and settling behaviour. Their ratio in the record swings by a factor of about 990 from bottom to top. No settling-velocity argument can produce that. The movement's own researchers abandoned this decades ago; it survives in popular apologetics.¹⁰
• There are no pre-Flood human dead. In 1,345,826 catalogued occurrences below the Neogene there is no human bone, no tooth, no worked stone, no hearth, no sherd, no metal, no wall. This is the strongest single objection to global-Flood geology, and it is about stone tools more than about bones. A flint axe does not bloat and float.¹¹
• "Marine fossils on mountaintops prove a global Flood" is overstated to the point of failure. The observation is real. The inference ignores mountain uplift, which is measured directly by GPS today.¹²
• Haldane never said "fossil rabbits in the Precambrian." No publication or recorded lecture of his contains it. The people who actually said it in print are Mark Ridley and Richard Dawkins.¹³
• Grasses did live with dinosaurs. Grass phytoliths have been recovered from Late Cretaceous dinosaur coprolites in India.¹⁴ Any article claiming otherwise will be corrected within a day.

What survives

Three things, and they are worth having.

The order is prior to evolutionary theory and independent of it.  It was assembled by field geologists, most of them creationists, decades before Darwin published.  The charge of circular reasoning is answerable, and O'Rourke — the source usually quoted to make the charge — answers it himself.³³

One young-earth prediction is confirmed by independent data.  Ecological zonation predicts a monotone marine-to-terrestrial succession up the column.  Computed here from Macrostrat units carrying Paleobiology Database occurrences, the non-marine share runs 0.0, 0.1, 1.7, 5.6, 23.3, 45.0 percent across the six megasequences.¹⁵  That is Tim Clarey's result and it survives an independent check.  It is the best thing young-earth geology has on fossil order and this article says so.

The movement has policed itself, in public, more than once.  A creationist geologist killed the Precambrian pollen claim with better sampling technique.¹⁶  The Institute for Creation Research withdrew the Paluxy man-track claim in its own words in 1986.¹⁷  An ICR scientist has written that the famous coal artifacts "need more thorough investigations that test against fraud before building a case on them."¹  Those episodes are the movement at its best and they deserve to be better known than the claims they retired.

And the strongest concession is recent, quantitative, and from inside.  In 2023 Kurt Wise — a young-earth creationist with a Harvard paleontology PhD — and Donna Richardson ran an original analysis of 114 stratigraphic boundaries using the Paleobiology Database, and concluded that "the global biostratigraphic column seems to be real."¹⁸  They also found that burial order bears no relation to the energy of the flood that is supposed to have produced it.¹⁸  Their statistic has been recomputed for this article on the same database, and it is corrected here where it fails — which is the treatment this article asks readers to accept when a creationist claim is checked, so it is the treatment a creationist's best work gets too.¹⁹

That is a young-earth researcher doing original quantitative work on open data and publishing a result against his own model.  It is the standard this site is trying to meet, and it means the central factual claim of this article is not a mainstream imposition on creationism.  It is something creationism's most qualified paleontologist established for himself.

Part 1 — What the order actually is

The method

The Paleobiology Database is a compilation of the published paleontological literature.  Every occurrence in it carries a reference to the paper that reported it, a collection number, and a stratigraphic assignment that can be traced back to a described section.  It is not a neutral arbiter — it is assembled by mainstream paleontologists — but it is primary in the sense that every row can be checked against a publication.⁶

Three properties of the database govern everything below, and getting them wrong is how both sides produce nonsense from it.

Every occurrence carries an age bracket, not a date.  A collection recorded only as "Paleozoic" spans 287 million years of the conventional scale.  A range-through diversity count credits such a record to every period it touches.  So a query that asks "does group X appear in period Y" will return false positives unless it demands that the record's entire bracket falls inside the interval being tested.  Every absence claim below uses that stricter rule.⁶

Some entries are shapes rather than animals.  Ichnotaxa are names for footprints, trackways and burrows.  Form taxa are names for isolated seeds, dispersed pollen grains and other material whose parent organism is unknown.  Both are indispensable to working stratigraphers and both are poison to a naive genus count, because a footprint name is a description of a shape and the same shape can be made by unrelated animals a hundred million years apart.  The database flags them, and its own switch for excluding them is pres=regular.²⁰

The database records the literature, not the rocks.  It is geographically lopsided in the way the published literature is lopsided:  the United States alone holds 22.4% of catalogued collections and the top ten countries hold 56.9%.²¹  That bias is real and it is stated here before any number depends on it.  It matters much less for the arguments in this article than it does for diversity curves, and the reason is stated in the objections section:  sampling bias can suppress records, but it cannot invert a sequence.

Queried on 8 August 2026, the whole-database occurrence count by period:

Three totals matter for everything that follows.  Precambrian rock holds 3,259 occurrences.  Paleozoic rock holds 548,543.  Everything below the Neogene holds 1,345,826.²²  Those are the denominators.  When this article says a group is absent from Paleozoic rock, it means absent from 548,543 chances to appear.

The staircase, printed in full

Groups appear in a fixed order, and they do not appear before it.  The table below is the whole survey — fifty-six groups, sorted by first appearance, with the number of occurrences standing behind each first-appearance datum.  That last column is the point of printing the table.  A first appearance resting on one record is a different kind of claim from one resting on twelve thousand, and an honest presentation has to let a reader see which is which.

The dagger marks the thirteen groups whose raw first appearance was pulled apart record by record.  Eleven of the thirteen did not survive.  Every one of those audits is set out in Part 3, and every correction moves a first appearance later, never earlier.  The two that survive the audit are flagged rather than corrected:  the Ordovician cartilaginous-fish record is genuine but taxonomically contested, and the Silurian cycad records are unresolved in either direction.  The corrected picture:  land plants first appear in the Silurian, not the Ediacaran; insects in the Devonian, not the Ordovician; amniotes and reptiles in the Carboniferous, not the Devonian; ginkgos in the Permian, not the Carboniferous; flowering plants in the Cretaceous, not the Carboniferous.²³

Printing the raw column rather than the cleaned one is deliberate.  A reader who runs the query will get the raw numbers, and an article that quietly printed only the corrected values would look like it had something to hide.  The corrections are arguments, and arguments belong in the open.

What the shape of the table means

Marine invertebrates with mineralised skeletons appear first.  Fish appear next.  Land plants appear after marine invertebrates and before land vertebrates.  Amphibians precede amniotes.  Amniotes precede mammals and birds.  Flowering plants appear after conifers and cycads.  Grasses appear after flowering plants in general.  The placental mammal orders — rodents, hoofed animals, primates, carnivores, whales, rabbits, bats, elephants — all appear inside one narrow band, and every one of them is absent from the 548,543 Paleozoic occurrences and the 3,259 Precambrian ones.

The important feature is not the sequence itself.  It is that the sequence has no exceptions of the kind a one-year Flood would produce.

The absent categories

Fourteen groups were selected because their absence from lower rock is what the argument turns on.  Every Paleozoic-or-older occurrence in the database was then enumerated for each — a complete count, not a sample.  Twelve of the fourteen return literally zero records.²⁴

A separate and stricter test was run across all 2,001,053 occurrences, counting an occurrence only if its entire age bracket falls past the boundary — so a loosely dated collection cannot generate a false hit.  On that test:

The two hits are both database errors, and both were traced to the paper the record came from.

The "Devonian amniote" is occurrence 2050687, from a 2006 paper on Devonian fish.  The database's own identified-name field reads Dipnoa indet. — a lungfish.  Dipnoa is a rank-changed synonym of the lungfish order Dipnoi, but the database contains a second taxon of the same spelling filed under class Reptilia, and the occurrence was matched to that one.  The single "Devonian reptile" in the world's largest fossil database is a Devonian lungfish with a duplicated name.²⁵

The "Ordovician insect" is occurrence 1661280, from the Volkhov Formation of Russia, identified in its source paper as Aulocopsis simplex — a Cambro-Ordovician sponge.  The database's name matching bound it to Humiryssus, a Cretaceous wasp genus with a colliding name.²⁶

That is the whole anomaly list.  In a database of 2.0 million records, the number of fossils that sit unambiguously below their group's first appearance is two, and both are typing mistakes.  An independent audit run for the companion article on fossil-record trends, using a separately written script and different filters, found the same two.²⁷

An important limit on that table: it counts bodies, not tracks

The table above excludes ichnotaxa — footprints, trackways and burrows — and form taxa.  That filter is necessary for counting named genera, because a footprint name is a shape, not an animal, and the same shape can be made by different creatures a hundred million years apart.  But it means the table answers a narrower question than it appears to:  "is there a body fossil of this group in this rock," not "is there any trace of this group in this rock."

That distinction matters, because trace fossils can and do move first-appearance datums without a single bone.  The Zachełmie trackways pushed the first appearance of land vertebrates back by roughly fourteen million years on nothing but footprints.²⁸

The best test case is a claim about birds in the Triassic, and it is worth telling in full.

In 2002 Ricardo Melchor, Silvina de Valais and Jorge Genise published bird-like footprints from what was then called the Santo Domingo Formation of northwest Argentina, in Nature.²⁹  The tracks show genuinely avian characters — including a well-developed backward-pointing first toe — and the formation was assigned a Late Triassic age.  That would place bird-like trackmakers at least fifty-five million years before the first known bird skeleton.

In 2013 the authors retracted the paper themselves.³⁰  Redating of the sequence, which has since been renamed the Laguna Brava Formation, returned a Late Eocene age.³¹  The tracks are real and the tracks are avian; the rock is simply not Triassic.

Two things follow, and the article states both.

First, this is what the system working looks like.  A finding that appeared to break the order was published in the most prominent journal in science, stood for eleven years, and was withdrawn by its own authors when better evidence arrived.  Nobody suppressed it.  That is the answer to the claim that anomalies get buried.

Second — and this cuts the other way — the retraction rests on radiometric dating, which this article has deliberately excluded from its evidence.  A young-earth reader who rejects the redating is entitled to say the Triassic assignment has not been overturned on stratigraphic grounds alone.  This article will not pretend otherwise.  What can be said without radiometrics is narrower:  the original Triassic assignment was itself a correlation, not a direct observation, and the two candidate ages differ by a factor that no stratigraphic argument in the original paper was designed to resolve.

The honest summary is that the strongest apparent exception to the order in the trace-fossil record was removed by mainstream science using a method this article does not rely on.  A reader who wants to press the point still has it available, and should know that its own discoverers do not.

A trap worth naming

Run the same query without excluding form taxa and the database returns 85 "flowering plants in Paleozoic rock."

Every one is a morphogenus.  There are exactly three of them.  Carpolithus, named by Linnaeus in 1768, is a wastebasket for any seed- or fruit-like object of undetermined affinity and carries 531 occurrences across the database.  Monosulcites, Couper 1953, is a form genus for dispersed monosulcate pollen — a grain type shared by cycads, ginkgos and bennettitaleans as well as by some angiosperms.  Tuberculatosporites, Imgrund 1952, is a form genus for dispersed spores.  The database flags all three with its own machine-readable marker for form taxa, and applying its own pres=regular filter removes every one of the 85.³²

These are filing categories used by palynologists for material whose affinity is undetermined.  Nobody is claiming a Carboniferous flower.

This matters in both directions.  A creationist quoting the raw number would be quoting a filing convention.  So would a critic using it to argue the database is a mess.  The article states it here so that neither happens.

Part 2 — The commercial test

Biostratigraphy is not a theory.  It is a business.

Index fossils are used to predict what rock the next borehole will enter, and companies spend capital on the prediction.  If the order were arbitrary, or an artifact of how paleontologists classify things, the prediction would fail and the money would be lost.

What the industry actually does

Wellsite biostratigraphy puts a micropaleontologist on the rig.  Rock chips come up the annulus from the drill bit.  The micropaleontologist washes, sieves and examines them in near-real time, identifies the foraminifera, pollen and spores, or calcareous nannofossils in them, and tells the drilling engineer where in the stratigraphic sequence the bit currently is.  That call determines when casing is set, when a core is cut, when to stop drilling, and — in a horizontal well — which way to steer.

Biosteering is the specialised form.  A pilot well is analysed bed by bed to build a biozonation specific to that field.  The horizontal well is then steered by watching which fossil assemblage is coming up the hole.  The reservoir layer and the barren rock above and below it are told apart by their fossils and by nothing else.

The technique was first applied in 1987, by Mærsk Olie og Gas on the Dan Field in the North Sea.³³  It has since been used in Qatar, the United Arab Emirates, Saudi Arabia, Oman and the Gulf of Guinea.  In Nigeria it has been used to pilot a hole along an oil–water contact.³⁴

The numbers

Saudi Aramco has biosteered over 1,000,000 feet of hole using microfossils, achieving over 90% reservoir contact.³⁵  That is the closest thing to a published hit rate that exists anywhere:  a million feet of hole steered by fossil content, with more than nine-tenths of it landing inside the target reservoir.

The detail that makes the number matter is the drilling method.  These are underbalanced coiled-tubing wells, and coiled tubing is too narrow to carry logging-while-drilling tools.  In those wells the fossils are not a cross-check on an electrical log.  They are the navigation.³⁵

BP's own account is that biosteering "frequently provides greater precision than seismic and greater discrimination than wireline logs."³⁶  Read that slowly.  Seismic reflection and wireline logging are hard-physics instruments — sound waves and electrical resistivity.  A BP team is saying that in their reservoirs, fossils beat both.

The Eldfisk field, Norwegian North Sea.  A nine-zone micropaleontological framework was built across the Tor and Ekofisk formations and five horizontal wells were drilled on it through January 1995.  The operators' own summary of the economics is that the technique was critical to drilling the horizontal wells while representing a minute portion of total expenditure.³⁷

Three Palaeocene turbidite reservoirs in the UK North Sea — Donan, Forties and Andrew — were subdivided into correlatable time slices using field-scale fossil events, and the same microfossils were used to model whether mudstones inside the reservoir act as barriers to fluid flow.³⁸

Resolution.  A quantitative correlation built for the Taranaki Basin of New Zealand used 1,376 species of foraminifera, dinoflagellates and pollen from 2,829 cuttings, sidewall-core and core samples across 20 offshore wells, and achieved an average composite age resolution of 130,000 years for the interval between about 90 and 65 million years.  The same study puts the summed depth error of the seismic method at typically 100 to 200 metres.³⁹

One honest gap: no published industry-wide error rate exists

This article looked for one and did not find it.  Rather than invent a figure, it reports that the metric has not been published.  What exists instead is three things, and all three are worth having.

A success rate for the operational task.  Aramco's >90% reservoir contact over more than a million feet is a measure of whether the bit stayed in the pay zone, not a measure of how often a fossil was misidentified.  Those are different quantities and conflating them would be dishonest.

Practitioners' own published list of failure modes.  The dominant one is downhole caving:  cuttings arriving at the surface are not tied directly to the depth the bit is at, and they are mixed with cuttings that fell in from shallower in the hole.³⁵  Note which direction that error runs.  Caving delivers younger material into a deeper sample.  Because of that asymmetry, the industry treats a fossil's last occurrence going down the hole as more reliable than its first, and builds the zonation from cores and sidewall cores, which cannot cave.  That is a working correction with a physical basis, not an excuse.

Published admissions of failure.  The Eldfisk operators state in their own abstract that local pinchouts and thinning sometimes prevented continuous positioning within the target zone.³⁷  The method is not perfect and its practitioners say so in print.

So the claim this article makes is narrower than "biostratigraphy has an error rate of X" and stronger than a slogan:  the technique is used at scale, by companies with no stake in evolutionary theory, in competition with hard-physics instruments, under conditions where failure costs money — and the operators publish success figures rather than quietly dropping it.

Why this argument is hard to deflect

Three properties do the work.

It is predictive, not retrodictive.  A biozonation built in one well is used to forecast what the next well will encounter, before it is drilled.  If the fossil order were an artifact of interpretation, the forecast would fail.

It is adversarially tested by cost.  Rig time in the North Sea runs to hundreds of thousands of dollars a day.  A method that failed would show up immediately in dry-hole and lost-reservoir statistics.

It is done by people with no stake in the debate.  A Saudi Aramco micropaleontologist steering coiled tubing through the Khuff carbonates is not defending Darwin.  He is being paid to hit a reservoir.

That last point disposes of a common reply.  It is sometimes said that an out-of-place fossil would not be recognised if it were found, because investigators are committed to the standard order.  The commercial case runs the other way.  A human tooth in a well sample is not a threat to a driller's worldview — it is a problem with the correlation the client is paying for.  The incentive is to notice it.  And a genuinely Devonian human would be the most valuable fossil ever recovered and the career-making paper of the century.  The bias argument requires a conspiracy that includes Saudi Aramco's micropaleontology group.

Part 3 — Every claimed exception, one at a time

The test, fixed in advance

A claim counts as a genuine order violation only if all four of these hold.  Stating the test before running the audit is what makes the count at the end defensible rather than rhetorical.

1. The identification is secure. The fossil is what it is claimed to be, confirmed by somebody other than the claimant, ideally by a specialist in that group.
2. The stratigraphic position is secure. The specimen came from rock in situ — not float, not spoil, not a cave fill, fissure fill or burrow cutting down into older rock — and the horizon can be relocated.
3. Reworking and contamination are excluded.
4. The order is genuinely reversed, not merely extended. Finding a group earlier than previously recorded moves a first-appearance datum. It is a violation only if the group appears below groups it must postdate on any reading of the sequence.

Point four is the one both sides get wrong.  A Cretaceous mammal found in the Jurassic is a range extension.  A rabbit in the Cambrian is a violation.  That is not a dodge.  It is the difference between a hypothesis being refined and being falsified, and every empirical field draws it.

Twenty-nine claimed or apparent violations were examined.  None survived.

Precambrian pollen in the Grand Canyon — dead, killed by a creationist

The claim.  Clifford L. Burdick published in 1966, and again in 1972 and 1974, that he had recovered pollen and spores — including grains of flowering-plant type, conifers, juniper and Ephedra — from the Hakatai Shale, a Precambrian unit roughly 10,000 feet of section below the Cambrian Tapeats Sandstone in the Grand Canyon.  If real, this would be the single most important fossil discovery ever made.  It would put modern seed plants below the trilobites.

The refutation came from a creationist, and it is a model of how this should be done.  Arthur V. Chadwick went to the Grand Canyon with Burdick in 1971 under a National Park Service permit, relocated Burdick's own sample localities, and re-sampled on three further trips.

His laboratory protocol is the whole argument, so it is worth stating in detail.  Samples were washed with filtered water and soap.  Every external surface was then removed with a trim saw running non-recirculating coolant, so that no part of the rock that had ever been exposed to air survived into the analysis.  The freshly exposed surface was examined and any sample showing a microfracture or other flaw was thrown away.  All solutions were filtered.  The laboratory was held under positive pressure with a filtered air supply, so that room air flowed outward rather than inward.  All glassware was cleaned with filtered water and soap.¹⁶

The result, in his words:  "No single example of an authentic pollen grain was obtained from any of these samples."¹⁶  Fifty samples, from Burdick's own strata, completely scanned.  The Hakatai slides were in most cases free of any material of biological origin at all, modern or fossil.

Why contamination is the better explanation, in Chadwick's own numbers.  He compared Burdick's reported assemblage against a published modern surface-pollen sample collected near Burdick's site.  The modern pollen rain ran 30% bisaccate conifer pollen, 12% juniper, 16% Ephedra, 42% angiosperms.  Burdick's eighteen identifiable grains ran 37% bisaccates, 11% possibly juniper, 11% Ephedra, 34% angiosperms and unassignable.¹⁶  Chadwick's conclusion is that the simplest hypothesis is modern contamination, picked up during collection and transport or infiltrated into the rock before collection.¹⁶

The decisive internal point.  Burdick reported modern pollen types — the plants growing on the canyon rim today.  A genuine Precambrian flora would not consist of junipers and Ephedra.  The very feature that made the claim spectacular is the feature that identifies it as contamination.  The grains were also preserved nearly fresh, unlike anything else in rock of that age.

The Creation Research Society's confirmation studies, and why they fail.  CRS members re-ran the work through the 1980s and reported that they had recovered pollen.⁴⁰  The sampling is the problem.  Their own published account states that samples were collected at depths of only three to four inches, and Burdick likewise sampled within a few inches of the surface.⁴¹  Shale is fissile — a dense network of partings — and a few inches below a weathered outcrop face in an arid canyon that takes runoff is exactly where infiltrated modern pollen would be.  The CRS workers also assessed "freshness" by macroscopic field inspection, and a healed microfracture is invisible that way.  Chadwick's trim-saw protocol is the correct answer to that objection, and Chadwick got nothing.

One correction to the record, which cuts against this site's own side.  It is sometimes said that creationists have retired this claim.  They have not.  Carl Froede's defence of it remains published on the Creation Research Society's own site, arguing that rejecting Precambrian plant material is required by neither framework and ignores documented evidence.⁴²

That defence fails, and it fails by equivocation.  Froede's opening pages establish that plant fossils occur in Archean and Proterozoic rock as old as 3.4 to 3.5 billion years, citing Schopf, Walter, Margulis and Strother.  That is true and completely uncontroversial.  What those authors are describing is stromatolites, cyanobacterial microfossils, and isotopically biogenic kerogen.  None of them is a vascular plant and none of them is pollen.  Froede then slides from "plant fossils exist in the Precambrian" to treating Burdick's modern conifer and angiosperm pollen as of a piece with them.  Cyanobacteria in the Archean is the mainstream position.  Ephedra pollen in the Hakatai would overturn botany.  These are not the same claim.

Verdict:  dead as an order violation.  What survives is the methodological lesson.  A creationist did the careful work, published the negative result, and said in print that the evidence is not credible.¹⁶

The Paluxy River "man tracks" — dead, withdrawn by ICR

The claim.  Human footprints alongside dinosaur tracks in the Cretaceous Glen Rose Formation, in the bed of the Paluxy River near Glen Rose, Texas.  It was the most widely promoted young-earth fossil claim of the twentieth century, carried by Stan Taylor's 1973 film Footprints in Stone and by John D. Morris's 1980 book.

The retraction is by the man who wrote the book.  Morris — later president of the Institute for Creation Research — published "The Paluxy River Mystery" in ICR's Impact series on 1 January 1986.  Its finding:  "none of the four trails at the Taylor site can today be regarded as unquestionably of human origin."¹⁷  His conclusion is one sentence and it should be quoted whole:  "Even though it would now be improper for creationists to continue to use the Paluxy data as evidence against evolution, in the light of these questions, there is still much that is not known about the tracks and continued research is in order."¹⁷

What changed his mind is specific and checkable.  The elongated "human" impressions in the Taylor Trail developed visible three-toed colorations in the years after excavation.  In almost every print in the trail, three large toes appeared, matching the nearby dinosaur tracks.  The toes are coloration only, with no impression.  Morris credits the discovery to Glen Kuban, who first noted the staining in 1982 and who claims neutrality on the creation-evolution question, and he records that he, Paul and Marian Taylor, and Marvin Herrmann — all associated with the 1973 film — went back to the Paluxy in October 1985 at Kuban's invitation to look.¹⁷

One caveat belongs with the retraction, and hiding it would be the same offence this article exists to catch.  Four lines before the withdrawal, in a list of open questions, Morris observes that a reddish stain can easily be applied to a rock surface with readily available chemical agents.  That is an insinuation of tampering, made without naming anyone and without evidence, in the same breath as the withdrawal.  He does not endorse it; he lists it.  Quote the retraction, and quote it knowing what sits next to it.¹⁷

The mainstream explanation is metatarsal dinosaur tracks.  A bipedal dinosaur that walks with the metatarsus — the long bones above the toes — in contact with soft mud leaves a long, roughly foot-shaped depression.  When mud slumps back into the digit impressions, the toe marks are subdued and the elongate heel-to-toe outline survives.  The remainder of the "tracks" are erosional features, indistinct marks, and a small number of deliberate carvings.  The reference field study is Milne and Schafersman's 1983 paper in the Journal of Geological Education, which is cited in Morris's own reference list as his footnote 1.⁴³

The film is gone too.  Footprints in Stone was withdrawn from distribution after its own production associates took part in the 1985 re-examination that killed the claim.

Creation Ministries International lists Paluxy among arguments creationists should not use, under the heading of arguments that are doubtful and inadvisable.  Its stated reason is that prominent creationist promoters have withdrawn support, that some of the allegedly human tracks may be erosion of dinosaur tracks obscuring the claw marks, and that research to date has not been promising.⁴⁴

A correction about that list, since it is often described loosely:  there is one list and it changed hands.  Answers in Genesis published it first, with captures running from December 2001; the URL has redirected since at least April 2023 and the page is gone from AiG's site.  Creation Ministries International — formerly Answers in Genesis–Australia — separated in 2005–2006 and kept the list.  So "Answers in Genesis says do not use X" is usually false as of today, whatever it was in 2015.⁴⁴

A second correction against this site's own side.  Of the five claims audited here, CMI's list names only Paluxy.  It covers the iron pot only as one of a class, under the entry "Gold chains have been found in coal," which says that coal artefacts are "strictly anecdotal" because "in each case the coal is no longer associated with the artefact."  It does not cover the Precambrian pollen, the Moab skeletons or the Meister print.  An article claiming the movement has retired all of these would be overstating what the movement has actually done.⁴⁴

The Moab and "Malachite Man" skeletons — dead

The find.  In 1971 the rockhound Lin Ottinger, leading a field trip in the Big Indian Copper Mine — later the Keystone Azurite Mine — near Moab, Utah, found major portions of two human skeletons stained blue-green by copper minerals.  A bulldozer had just stripped about fifteen feet of overburden.  The enclosing rock is the Cretaceous Dakota Sandstone.

The decisive observation was made within days, and it is stratigraphic rather than chronological.  John Marwitt, then Field Director of the Utah Statewide Archaeological Survey, led the rest of the excavation.  He reported that the bones lay in loose, poorly consolidated blowsand, in contrast to the hard consolidated sandstone away from them, and that the bones were unfossilised — not mineral-replaced, essentially modern in appearance apart from the copper staining.  His conclusion was intrusive burial, probably only centuries old.⁴⁵

That single observation settles it, and it needs no dating method at all.  A body buried in a hole dug into old rock is surrounded by the loose fill of the hole.  A body deposited with the sediment is enclosed by the same lithified rock as everything else at that level.  Marwitt found the first.

The radiocarbon confirmation came later and agrees.  Bones from the 1971 find returned 210 ± 70 years before present at a UCLA laboratory.  Further excavation at the mine in the early 1990s returned 1450 ± 90 BP, calibrating to roughly AD 540–670 — consistent with Native American use of the mine over several centuries.⁴⁵

The trail of the claim.  Clifford Burdick — the same Burdick as the Hakatai pollen — wrote the find up as a probable out-of-place fossil in the Creation Research Society Quarterly in 1973, apparently working from a sensationalised newspaper account rather than from Marwitt.  Carl Baugh bought one skeleton from Ottinger in the mid-1980s and displayed it at his museum in Glen Rose as an out-of-order fossil.  In 1997–98 Don Patton and David Willis relaunched the material as "Malachite Man."⁴⁵

Two errors in the circulating version, both checkable by anyone.  The depth is given as fifty feet; Marwitt and creationist authors alike report about fifteen for the 1971 finds.  And the argument that the bones are in the same Dakota Sandstone as the dinosaurs at Dinosaur National Monument is simply wrong:  the Dinosaur National Monument quarry is in the Morrison Formation, which is Jurassic, not the Cretaceous Dakota.  That is a factual error about a well-known American monument.⁴⁵

The Meister "sandal print" — dead

In June 1968 William J. Meister, a rockhound collecting trilobites near Antelope Springs, Utah, split a slab of the Middle Cambrian Wheeler Formation and found an oblong mark about ten and a half by three and a half inches containing a specimen of the trilobite Elrathia kingii.  He read it as a sandal print with a crushed trilobite under the heel.

The trilobite is real.  The print is a spall.  Spalling is the flaking-off of a curved fragment along a bedding-parallel or conchoidal fracture, and the Utah Museum of Natural History's assessment was that the object is nothing more than a slab of Wheeler shale with a fragment spalled off in the shape of a footprint, revealing a trilobite.⁴⁶

Three diagnostics settle it, and none of them requires a date.

It occurs on both sides of the slab in opposite relief.  That is what a fragment breaking out of a slab produces.  A foot pressed into mud leaves an impression on one surface and a cast on the counterpart — and critically, a real track deforms the laminations beneath it.  This one does not.⁴⁶

The sides are unnaturally angular and the whole feature unnaturally shallow.  The reported depth difference between "heel" and "sole" is about one-eighth of an inch.  That is not a footfall in soft sediment.

The Wheeler Shale produces these features abundantly, in many shapes, by solution penetration, spalling and weathering.  A shape-matching claim is only as good as the population of shapes the process generates, and this process generates a great many.

The Carboniferous "iron pot" — untestable

The claim rests on a single affidavit by Frank J. Kenwood, dated 1948, stating that in 1912 he found an iron pot embedded in a lump of coal at the municipal electric plant in Thomas, Oklahoma, sourced from the Wilburton mine.

There is no pot available for examination.  There is no photograph of it in situ in the coal.  There was no independent witness at the moment of breaking, and there is no chain of custody.  An affidavit about an object nobody can now examine fails criteria one, two and three of the test above, simultaneously.

The cleanest verdict here is ICR's own.  Brian Thomas has written that the coal artifacts — the iron cup, the brass bell, the aluminium gear — "need more thorough investigations that test against fraud before building a case on them."¹  An ICR scientist saying the coal artifacts need testing against fraud is a better warrant for retiring them than anything a critic could write.

Lazarus taxa and living fossils — not violations at all

A Lazarus taxon is a group that vanishes from the fossil record for a long stretch and then reappears, either later in the record or alive today.  The four standard cases:

Why none of this touches the order.  A Lazarus taxon extends a range upward.  It says the group persisted longer than the record showed.  It never puts a group below something it must postdate.  No coelacanth has been found in the Precambrian.  No monoplacophoran has been found below the Cambrian.  The order of first appearances — which is what Part 1 measured — is untouched by every one of these cases.⁴⁷

The young-earth side gains something real here and should be allowed to say it.  Lazarus taxa show that absence from the record over long intervals is not proof of absence in life.  That is a genuine limit on the fossil record, conceded by the mainstream, and it is why any argument from stratigraphic absence to extinction has to be made carefully.

But that concession has a price, and this article charges it.  If absence of evidence is weak evidence of absence, it is weak in both directions.  The same argument cannot be used to say "the gaps between fossil forms are real gaps rather than sampling artifacts."  A movement cannot have Lazarus taxa proving the record is patchy on Monday and gap-scaling proving the record is complete enough to show real discontinuities on Tuesday.  The companion article on fossil-record trends handles that tension directly and concludes against the completeness argument.²⁷

A creationist-friendly source already states the point correctly.  O'Rourke, discussing exactly these cases in 1976, notes that supposed exceptions such as Neocalamites, Neotrigonia and Neopilina either differ slightly from their ancestors or have a longer range than previously thought.³

The database's own anomalies, audited one by one

The five famous claims above are the ones a reader will have met.  They are not the hardest test.  The hardest test is to run the query yourself and look at what comes back, because a raw database query does return apparent order violations, and publishing the clean numbers without auditing the dirty ones would be dishonest in this article's own favour.

Every anomalously early first appearance across the fifty-six-group survey was pulled as individual occurrence records — identified name, accepted name, stratigraphic interval, formation and source reference — and judged.²³

Flowering plants in the Carboniferous — 15 raw occurrences.  Form-genus artifact.  All three genera are morphogenera, as set out in Part 1.  Under the database's own form-taxon filter the count is zero, and the cleaned series runs Triassic 23, Jurassic 12, Cretaceous 3,318, Paleogene 8,962.³²

Insects in the Ordovician — 1 occurrence.  Homonym artifact.  Aulocopsis simplex, a sponge, bound to a Cretaceous wasp genus.  The record survives even the form-taxon filter, so it is a straightforward error and the database's own hierarchy contradicts the source paper.  The earliest genuine insect records are Devonian.²⁶

Reptiles and amniotes in the Devonian — 1 occurrence.  Duplicate-name artifact.  The Dipnoa lungfish, described above.²⁵

Land plants in the Ediacaran — 3 occurrences.  Homonym artifact.  Striatella coriacea, from the Vendian of Russia.  The database's Striatella is a dispersed-spore form genus placed under the ferns.  The Vendian Striatella is a different organism entirely — a Precambrian macroalga-grade problematicum.  Same name, different taxon, wrong classification inherited.  The earliest defensible land-plant records are Silurian, which is where the published cryptospore and Cooksonia literature puts them.⁴⁸

Cycads in the Silurian — 2 occurrences.  Unresolved, and flagged as unresolved.  Harringtonia acutiplicata and a questionable Harringtonia sp., from two 1970s South American collections.  Harringtonia was named by Frenguelli in 1942 and has two occurrences in the entire database.  This is a poorly constrained genus resting on one mid-century author, not a Silurian cycad flora, and no modern re-study could be located.  It is listed here as open rather than settled, in either direction.⁴⁹

Ginkgos in the Carboniferous — 1 occurrence.  Broad-bin artifact, and the most instructive error in the whole audit.  The record is Baiera braun, whose stratigraphic assignment in the database is literally "Carboniferous to Permian" — a bin 107 million conventional years wide, covering two whole periods.  A range-through diversity query counts such a record in every period it spans.  All ten other early Baiera and Sphenobaiera records are Early Permian, which is where the published literature places the origin of the ginkgophytes.⁵⁰

That case is the single most important methodological caution in this article.  When a range-through count shows a group appearing early, check whether one record has an interval spanning the boundary before concluding anything.  It is how a database with no errors in it at all can still produce a headline that is false.

Cartilaginous fish in the Ordovician — 1 occurrence.  Genuine, contested taxonomy, not an order violation.  Areyongalepis oervigi from the Stokes Siltstone of Australia is a scale-based taxon, and whether Ordovician scale taxa belong in Chondrichthyes is an open question in the specialist literature.  The main Silurian block of 61 occurrences is standard and uncontroversial.

Foraminifera in the Cambrian — 38 occurrences.  Real, and worth knowing.  Platysolenites antiquissimus is a genuine agglutinated foraminiferan used as an index fossil for the lowest Cambrian, recorded from Norway, Poland, Newfoundland and England in the pulled records.  This is a real Cambrian first appearance of a group that is still alive.  It is not an order violation, and it is not the same thing as the planktic foraminifera used in petroleum biostratigraphy, which first appear in the Jurassic.

Vertebrates in the Cambrian — 9 occurrences.  Real and mainstream.  Myllokunmingia, Haikouichthys and Zhongjianichthys from the Maotianshan Shales of Chengjiang, China, plus two indeterminate records from the Chatsworth Limestone of Australia.  These are the accepted earliest chordates and craniates and nobody hides them.

Grasses in the Early Cretaceous — 3 occurrences.  Two dead, one open.  Eragrosites changii was described as a grass in 1998 and later reinterpreted as a gnetalean; the database still files it under the grasses.  Phragmites pratti, "Early Cretaceous," is a nineteenth-century-style leaf identification with no modern support.  An indeterminate grass record from the Early Albian Zhonggou Formation of China is the one that is not obviously dead, and its source paper could not be obtained for this article.  If it is genuine it is a range extension of grasses within the Cretaceous — which is a change to a datum, not a reversal of order.⁵¹

Birds in the Jurassic — 15 occurrences.  Mixed, and the mixture is the point.  The set includes Praeornis sharovi, interpreted over five decades as a feather, as a cycad frond, and as a real feather; four ichnotaxa, which are footprints rather than bodies; and an indeterminate archaeopterygiform from Portugal.  Archaeopteryx itself sits in the Tithonian Solnhofen limestone.  The earliest-bird datum rests on a handful of records of mixed quality — which is the normal condition of a first-appearance datum, and exactly why the absence claims in this article are stronger evidence than the presence ones.

Mammals in the Triassic — 15 occurrences.  Real, and half of them are footprints.  Indeterminate mammal material from the Fleming Fjord Formation of Greenland; Woutersia mirabilis and W. butleri from the Rhaetian of France; a questionable multituberculate from the same unit.  The rest are mammal-like trackways from the Elliot Formation of South Africa and from the Towaco and Redonda formations.  Mainstream reading:  mammaliaforms are Late Triassic.  Nothing here is Paleozoic.

Lizards and snakes in the Permian — 4 distinct records.  All footprints, all at the period boundary.  Three Rhynchosauroides ichnospecies from Italy, Brazil and Argentina, every one of them in the Lopingian — the last stage of the Permian, immediately below the Triassic.  These are trackways assigned to Squamata by inference about the trackmaker.  A footprint-based first-appearance datum at a period boundary is a normal and much-discussed kind of stratigraphic argument.  It is not a lizard in the Cambrian.⁵²

The exhaustive sweep

The fourteen groups whose absence the argument actually rests on were then swept completely.  For each, every occurrence in the database whose stratigraphic range does not extend above the Permian–Triassic boundary was retrieved and printed.  This is a full enumeration, not a sample.²⁴

The count, and what it does and does not license

Combining the machine sweep, the individually audited first-appearance anomalies, and the five famous claims:

• Claimed or apparent order violations examined: 29.
• Genuine order violations surviving all four criteria: zero.
• Database or classification artifacts: 7 — the Paleozoic angiosperm form genera, the Ordovician "insect," the Devonian "reptile," the Ediacaran "land plant," the Carboniferous "ginkgo," the Cretaceous "grass" Eragrosites, and the Permian "lizards."
• Contamination or intrusive emplacement: 2 — Burdick's Hakatai pollen, the Moab skeletons.
• Physical or biological misinterpretation: 2 — the Paluxy tracks, the Meister print.
• Untestable: 1 — the Carboniferous iron pot.
• Genuine early first appearances that are not violations: 5 — Cambrian foraminifera, Cambrian vertebrates, Ordovician chondrichthyan scale taxa, Triassic mammaliaforms, Jurassic birds.
• Unresolved and flagged as unresolved: 2 — the Silurian Harringtonia records, and the Early Albian grass record. Both, if confirmed, would be range extensions inside an already-established interval, not reversals of order. Neither can produce a rabbit in the Cambrian.

What this licenses.  After a complete enumeration for the fourteen groups the argument depends on, plus an individual audit of every anomaly the fifty-six-group survey surfaced, plus the five famous claims:  not one verified case exists of an organism deposited in rock below the groups it must postdate.

What it does not license, and this matters.  It is not a claim that all 2.0 million occurrences have been checked for every possible order violation.  That is a different and much larger computation — a full range-consistency check on every genus — and it has not been done here.  Anyone quoting this article should quote the narrower claim.⁹

How a real anomaly is told from an artifact

Every dead claim above died on one of a short list of known failure modes, and each mode has a diagnostic that any field geologist can apply.

Reworking.  An older fossil is eroded out of its rock, transported, and redeposited in younger sediment.  It then sits genuinely in situ in a bed far too young for it.  This is common in conglomerates, turbidites, transgressive lags and channel fills.  The diagnostics are non-overlapping ranges inside one assemblage, mismatched depositional environments in the same bed, and mismatched preservation — colour, abrasion, mineral infill, thermal alteration.  Palynologists routinely see reworked grains as darker and more thermally mature than the indigenous assemblage.⁵³

Downhole caving.  Cuttings from higher in a well fall in and mix with cuttings from the bit depth.  This is the drilling industry's own dominant error source, and it moves young material down.³⁵

Laboratory and field contamination.  Modern pollen is in the air and in unfiltered water.  The diagnostics are Chadwick's three:  does the assemblage match the modern pollen rain of that locality, does the preservation state match the other organic material in the sediment, and does it make sense in the stratigraphic setting.  Burdick's assemblage failed the first two.

Infiltration through fractures.  Shale is fissile.  Joints, bedding partings and healed microfractures conduct water and everything suspended in it.  This is why sampling three inches into a weathered outcrop is not a control.

Intrusive burial and fissure fill.  A grave, a burrow, a karst fissure, a mine working or a collapse void puts young material physically inside old rock.  The diagnostic is the matrix, and it is a pure field observation:  material deposited with the rock is enclosed by the same lithified rock; material emplaced later is enclosed by its own softer, differently sorted fill.  That is exactly what Marwitt recorded at Moab.

Structural repetition and inversion.  Thrust faults and recumbent folds can put old rock physically above young rock — the Lewis Overthrust in Montana, the Glarus thrust in Switzerland, the Heart Mountain detachment in Wyoming.  The diagnostic is the contact itself:  a thrust surface carries mylonite, breccia, gouge, slickensides and drag folds, and can be walked out for tens of kilometres.  The order is inverted by mechanical transport of a slab, which is a different claim from the fossils having been deposited in that order, and the deformation is visible.

Misidentification and database error.  Three were found in the Paleobiology Database for this article alone.  The diagnostic is to go to the source paper, and in all three cases the original publication says something different from the database.

Now notice the direction all of these run.  Six of the seven move material later — up the column — or sideways, into a different taxon.  Only structural repetition moves rock down, and it leaves a visible fault.

That asymmetry cuts both ways, and this article states both halves.  It means a single verified case of a modern organism deposited in ancient rock really would be devastating, because no ordinary sedimentary process can fake one.  And it means the failure of every claimed case to survive examination is correspondingly strong evidence, because the processes that generate false positives almost all point the other direction.

What a real surprise looks like

The audit would be dishonest if it only listed failures.  Genuine stratigraphic surprises do occur, and mainstream paleontology publishes them in its most prominent journals.

Tetrapod trackways from Zachełmie, Poland.  Digit-bearing trackways in Eifelian marine carbonates, published in Nature in 2010, roughly fourteen million years older than the oldest tetrapod body fossils and about seven to nine million years older than Tiktaalik, the fish that had been presented as the immediately pre-tetrapod form.²⁸  Those are the current figures, not the 2010 ones.  The discovery team revised the age of the tracks from about 397 to about 390 million years in 2018, which shortened every offset; creationist writing still recycles the superseded eighteen-million-year gap.²⁸  What happened next is the point.  The paper ran in Nature, the finding was accepted, the first-appearance datum for four-limbed vertebrates moved down, and the Tiktaalik-as-ancestor story was demoted to Tiktaalik-as-late-surviving-member-of-a-grade.  Nobody suppressed it.  Note also what it is not:  the tracks are still Devonian, still above every fish group and below every amniote.  It moved a datum inside the sequence.  It did not break the sequence.

Angiosperm-like pollen in the Middle Triassic.  Peter Hochuli and Susanne Feist-Burkhardt reported pollen grains from the Barents Sea and later from northern Switzerland whose morphology is comparable to Early Cretaceous forms generally attributed to flowering plants, and stated in print that these are the earliest such grains in the fossil record.⁵⁴  That is a claim that angiosperms, or their stem lineage, were present about a hundred million years before the conventional Early Cretaceous radiation.  The field's response was to argue about it in print, not to bury it.  It remains contested, and "angiosperm-like" is doing real work in the title, because the grains are known only as dispersed pollen with no attached parent plant.

Grasses in the Late Cretaceous.  Silicified plant tissues in latest-Cretaceous dinosaur coprolites from India represent "at least five taxa from extant grass (Poaceae) subclades," which pushed crown-group grasses out of the Cenozoic and back into the Cretaceous.¹⁴  This is why the claim in this article is stated as "no grass in Precambrian or Paleozoic rock" and never as "no grass with dinosaurs."  Grasses did live with dinosaurs.

Mesozoic mammal ecology.  The old picture of Mesozoic mammals as uniformly shrew-sized and marginal has been dismantled by finds including Repenomamus, a Cretaceous mammal preserved with a juvenile dinosaur in its gut, and by swimming, gliding and burrowing Jurassic and Cretaceous mammaliaforms.  The mainstream revised the ecology freely.  What did not move is the stratigraphic order:  no mammal has been found below the Triassic, in 127,820 mammal occurrences.

What this section establishes.  When a genuine surprise arrives it is published in Nature or Science, argued about by name in the literature, and absorbed by moving a datum.  None of the claims audited above has any of those three properties.  That asymmetry is not special pleading.  It is the difference between evidence and anecdote.

Part 4 — Where are the pre-Flood human dead?

This is the hardest fact in this article, and it is not primarily about bones.

The facts

Genus Homo has 238 occurrences in the Paleobiology Database:  seven Pliocene, 191 Pleistocene, forty Holocene.  Zero below the Pliocene.  The family Hominidae has 619, the oldest Miocene.  Against 1,345,826 occurrences below the Neogene, the human count is zero.¹¹

The contexts are as diagnostic as the counts.  Human remains occur in cave fills, rock shelters, river terraces, lake margins, alluvium, loess, dunes, peat and deliberate burials.  Those are late Quaternary surficial deposits.  They do not occur in marine limestones, deep-water shales, coal seams, turbidites or chalk — which is to say, they do not occur in the thick, laterally extensive marine units that Flood geology identifies as Flood rock.

The earliest stone tools sit in the same narrow band:  Lomekwi 3 in West Turkana,⁵⁵ and before that Gona in Ethiopia.⁵⁶  The earliest fossils attributed to Homo sapiens are from Jebel Irhoud, Morocco.⁵⁷

No human artifact of any kind has ever been recovered from Paleozoic, Mesozoic or Paleogene rock.  No worked stone.  No hearth.  No post-hole.  No pottery.  No metal.  No wall.  No midden.  No road.

Why the artifacts matter more than the bones

Genesis describes the Flood destroying a whole civilisation.  It has metallurgy in Tubal-Cain, musical instruments in Jubal, cities, agriculture and herding, and enough people that "the earth was filled with violence."  It sits on the earth for more than 1,600 years, with lifespans in the hundreds.

That produces a testable prediction, and bones are the weakest part of it.

Worked flint is close to indestructible.  Pottery survives burial for millennia.  Copper and bronze survive.  Fired hearths leave chemical and magnetic signatures that survive when the hearth itself does not.  Charcoal survives.  Grinding stones, hammerstones and anvils survive.  A single Neolithic settlement mound yields tens of thousands of lithics — and a lithic is a rock, in a rock, and it is identifiable as worked by the conchoidal fracture pattern that only deliberate percussion produces.

So every taphonomic argument about floating bodies and dissolving bone leaves the harder half of the problem untouched.  A stone axe does not bloat, float, or get eaten by fish.

The answers in print, assessed one at a time

"Bodies bloat, float, and are scavenged or decay."  Answers in Genesis sets this out at length.  Its arguments as given:  hydraulic forces were strong enough to be highly destructive, so few bodies would have remained intact; humans, mammals and birds flee rising water and seek shelter on higher ground, in trees or on structures, unlike slower amphibians and reptiles; bodies bloated and floated and were eaten by sea creatures, washed ashore, or decomposed; calcium-phosphate bone may have dissolved because of the acidity of the water's volcanic components; and remains buried late and near the surface would have been eroded away as the waters receded.⁵⁸

Every mechanism named is real.  None is selective.  Bloat-and-float applies to every air-breathing land vertebrate — and the database holds 37,815 dinosaur occurrences, 15,175 bird occurrences, 127,820 mammal occurrences and 1,707 bat occurrences.  Dinosaurs bloat and float too.  If flotation explains the absence of humans it must explain why it did not also remove the sauropods.  Water acidic enough to dissolve human bone dissolves everything else in the same bed.  And none of it touches stone tools.

"They fled to high ground."  This is a positive prediction, and it is testable.  If humans fled upward, they should be at the top of the Flood sequence, not absent from it.  On any young-earth placement of the Flood's upper boundary, the uppermost Flood units contain abundant mammals and birds and no humans.

"The pre-Flood population was small."¹  Available, but expensive.  Brian Thomas lists high violence rates in Genesis 6 as something that could have severely limited human numbers, and John Woodmorappe explained the absence of pre-Pleistocene human fossils primarily through a low antediluvian population.⁵⁹  It is hard to square with a text that describes a population filling the earth, and which motivates the Flood by the scale of human corruption.  Worse, it runs against the movement's own most-used population argument:  creationist writers routinely calculate large pre-Flood populations from long lifespans and high fertility, precisely in order to show that eight survivors can repopulate the world.  A movement cannot use exponential growth to get billions before the Flood and then plead a small population to explain their absence.  This article searched for a defended young-earth estimate of the pre-Flood population and found only assertions in both directions.  Until somebody publishes a number with a method, this answer cannot be priced.

"The remains were pulverised."¹  Same defect, sharpened.  Turbidity currents and debris flows are extremely destructive — and the record is full of vertebrate material preserved inside turbidites and debris flows.  A mechanism invoked to erase humans has to explain why it spared everything else in the same flow.

"They are diluted across the world's sediments."¹  This is the most quantitatively testable answer and it fails quantitatively:  the same dilution left 1.3 million recorded non-human occurrences below the Neogene.  If human bodies are needles in a haystack, so are titanosaurs, and the titanosaurs turn up.

"It would not be recognised if found."¹  Answered in Part 2.  The commercial incentive runs the opposite way from the one alleged, and a Devonian human would be the most valuable fossil ever recovered.

"The pre-Flood continent itself was destroyed — subducted, or driven beneath another plate — taking the whole civilisation with it."  This is the strongest version of the burial answer and it deserves a fuller hearing than the vaguer "buried continents" formulation, because unlike the others it comes with a mechanism already present in young-earth geology.  Catastrophic plate tectonics, as developed by John Baumgardner, has the entire pre-Flood ocean floor subducted during the Flood year at metres-per-second rates.  If oceanic lithosphere can be consumed on that scale in a year, the argument runs, a continental block carrying the antediluvian world could go the same way — and the reason there is no trace of pre-Flood humanity is that the rock it lived on is no longer at the surface.

It also has a theological logic behind it that the other answers lack.  The stated purpose of the Flood was to blot out a civilisation.  A model in which that civilisation's entire substrate was destroyed is not an evasion; it is what the text describes, taken seriously.

Four costs, stated plainly.

Continental crust does not subduct easily.  This is the mechanical objection and it is the serious one.  Continental crust is roughly 2.7 grams per cubic centimetre against about 3.3 for the mantle; it is buoyant, and that buoyancy is why continents persist while ocean floor is recycled.  Where continental crust does enter a subduction zone today the usual outcome is collision and mountain-building — the Himalaya — rather than disappearance.  Ultra-high-pressure terranes show that fragments of continental material can be dragged down and returned, but that is fragments, not a populated landmass.  A young-earth model that subducts a continent has to explain why buoyancy did not do what buoyancy does.

It predicts something, and the prediction is checkable.  Subduction is not clean disposal.  A subducting plate sheds its sedimentary cover into an accretionary wedge, and that material stays at the surface.  If a populated continent went down, the wedge should hold its debris — worked stone above all, which survives what bone does not.  So the model does not merely predict absence; it predicts a specific place to look.  That is a virtue, and it is testable.

It has to explain the animals.  Whatever destroyed the human record left 1.3 million non-human occurrences below the Neogene, including 127,820 mammal occurrences.  The subduction answer therefore requires that pre-Flood humans and pre-Flood animals lived on different and separately fated landmasses — that the human continent went down while the animal-bearing regions were merely buried.  That is a substantive additional claim about antediluvian geography, and it needs independent support rather than being assumed in order to make the first problem go away.

It sits awkwardly with the Ark.  The animals aboard came from somewhere, and Noah and his family lived somewhere.  If the human landmass is gone, the Ark's occupants boarded from a place that no longer exists and disembarked on crust that does.

Verdict:  this is the best of the eight answers, and it is not yet a scientific one.  It has a mechanism, it fits the purpose of the Flood, and it makes a real prediction about accretionary wedges.  What it does not have is anyone who has worked it out quantitatively — no published calculation of whether a continental block of the required size can be consumed in a year against its own buoyancy, and no search of the predicted wedge deposits.  Until someone does that work it is a promising conjecture rather than an answer, and this article records it as one.  It is also, by some distance, the most productive thing anyone could pursue on this question, because it is the only version that tells a field geologist where to go and what to look for.

"Move the Flood/post-Flood boundary."  This is the serious answer and it deserves the most care.  If Flood deposits end at the Cretaceous–Paleogene boundary, then all Cenozoic human fossils are post-Flood and their absence lower down is expected.

It relocates the problem rather than removing it, and in one version it makes the problem much worse.  Steven Austin and Kurt Wise have long placed the pre-Flood/Flood boundary at the Precambrian–Cambrian, and Wise and Richardson's 2023 database analysis argues from their own statistics that the pre-Flood/Flood boundary sits below the Ordovician.¹⁸  That second argument does not survive being recomputed, and Part 6 shows why — the low straddling counts it depends on are an artifact of how the database labels Cambrian stages.  The position itself long predates the statistic and does not depend on it, so the problem below is unaffected.  On that scheme the antediluvian world — Tubal-Cain's forge, the cities of Genesis 4, sixteen centuries of human history — is represented by Precambrian rock.  Precambrian rock contains no land plants, no land animals, no vertebrates of any kind, no soils with rooted floras, and 3,259 fossil occurrences in total, essentially all microbial mats, acritarchs and Ediacaran problematica.  A reader is being asked to believe that the world in which Tubal-Cain forged bronze is the world recorded in stromatolite-bearing dolomites.

And the Flood layer itself, on that scheme, is where the drowned should be:  Cambrian through Cretaceous.  That is the interval with 1.2 million occurrences and zero humans.

There is a second cost, internal to the movement.  The candidate boundaries are roughly sixty million conventional years apart, and the disagreement between them is not a detail.  A boundary placed to fit the data is a legitimate scientific move, but it cannot then be presented as a prediction of the data.

The concessions, in the movement's own words

A reader will trust these in a way they will not trust a critic's summary.

• Brian Thomas, ICR, 2015: "we know of no undisputed human skeletal remains in Flood rocks."¹
• Troy Lacey and Avery Foley, Answers in Genesis, 2018: "It is not altogether implausible that pre-flood human fossils could be found one day."⁵⁸

The second is the honest answer and it is a promissory note.  As a statement of research direction it is unobjectionable.  As an answer to the objection, it concedes the objection.

The verdict

A global Flood that destroyed a metal-working, city-building human population has left, in 1.2 million catalogued occurrences below the Neogene:  no bone, no tooth, no tool, no hearth, no sherd, no wall.

The taphonomic answers are real processes that are not selective, and none of them touches worked stone.  The boundary-relocation answer converts the problem into a larger one.

This is a genuine anomaly for the model, of the same kind and severity as the anomalies this site holds against mainstream claims elsewhere.  Printing it is the price of the rest of the series being believed.

Part 5 — The young-earth explanations, assessed

Six explanations of fossil order are in circulation.  Each makes a prediction.  Each prediction can be checked.

Hydrodynamic sorting — dead

The claim.  Denser, rounder, smaller and more compact organisms settle faster out of turbulent water, so they end up lower in the column.  It comes from The Genesis Flood (1961), and every later statement traces back to it.

A sourcing note this article owes the reader.  The Genesis Flood was not opened for this article.  The open scan available online is front matter only, and the complete text is available only through a lending arrangement that could not be completed.  So no words are attributed to Whitcomb and Morris anywhere on this page.  The claim is characterised here through the later creationists who describe it — Tomkins and Clarey, who list hydrodynamic selectivity and sorting by size, composition and settling velocity as Whitcomb and Morris's proposal, and Woodmorappe, who writes of hydrodynamic sorting and differential escape as proposed by them.⁶⁰  Both are creationists describing a creationist, which is the right way round.  But the primary has not been checked, and this article says so rather than letting a reader assume it has.

What the physics predicts.  In the Stokes regime, settling velocity rises with the square of particle diameter and linearly with the density contrast between particle and fluid.  Above a Reynolds number of about one the drag law changes, but the ordering by size and density survives in every regime.  So the model makes a hard prediction:  fossils should be arranged by skeletal density and body size, and by nothing else.

The densities are not in dispute.  Calcite is 2.71 grams per cubic centimetre, aragonite 2.93, apatite 3.2, compact bone about 2.0, chitin about 1.4, the sporopollenin of pollen and spore walls about 1.2 to 1.45, waterlogged wood 0.9 to 1.1.

Three independent failures.

Brachiopods and bivalves.  These two groups are as hydrodynamically alike as two animal groups get:  centimetre-scale, biconvex, two-valved, calcium-carbonate shells, benthic filter feeders in the same shelf environments.  North American genera per period, computed from the database for this article:¹⁰

The ratio swings by a factor of about 990 from the bottom of the column to the top, and it does so monotonically.  No settling-velocity difference between these two groups could produce that, because there is no meaningful settling-velocity difference between them.  This is the cleanest single refutation of hydrodynamic sorting available, and anybody can compute it in five minutes from a free database.

The same table kills the model twice more.  Trilobites — chitin-and-calcite cuticle, one to ten centimetres, dense — run 519, 404, 155, 85, 39, 14 genera and then stop dead in the Permian, while hydrodynamically comparable crustaceans continue upward without interruption.  Ammonoids and nautiloids have nearly identical chambered aragonite shells of similar size:  ammonoids appear in the Devonian and run to the Cretaceous, nautiloids appear in the Ordovician and are alive today.  Same shell, same density, same size, four hundred million conventional years of stratigraphic offset.

Pollen.  Pollen grains and spores are twenty to a hundred micrometres across and about 1.2 to 1.45 grams per cubic centimetre — the last things that should settle anywhere.  Sorting predicts all pollen in the topmost layer of every basin.  Instead pollen is the petroleum industry's primary biostratigraphic tool through the entire column, and it follows the same order as the plants that produced it:  spores in the Devonian and Carboniferous, gymnosperm pollen from the Carboniferous and Permian, angiosperm pollen only from the Early Cretaceous upward.

The people who hold the model concede this.  Jeffrey Tomkins and Timothy Clarey, in the 2023 International Conference on Creationism proceedings:  "It remains a mystery why no undisputed pollen has been found in the earliest megasequences."  Their own next sentences say why it is a mystery:  on their model flowering plants existed before the Flood and in its earliest moments, and so did their pollen.²  That concession is worth more than any critic's assertion.

Bones in mud.  Articulated skeletons of large vertebrates are routinely enclosed in chalk, diatomite and shale — sediments that settle from suspension orders of magnitude more slowly than a bone.  Under hydraulic sorting, a dense bone suspended in a slowly settling mud cloud is impossible.  Under ordinary deposition it is unremarkable.

Verdict:  dead.  Not weak, not incomplete.  The mechanism makes a specific prediction, the prediction is testable against open data, and the data contradict it by three orders of magnitude in a case where the mechanism has no room to manoeuvre.

And no serious young-earth researcher still defends it.  John Woodmorappe relegated it to a supporting role in 1983 and had to invent a new mechanism because the old ones failed.  Clarey and Tomkins list it in 2023 only as a historical proposal, and then argue for something else.  The argument survives almost entirely in popular apologetics, and it should be retired there too.

Differential escape — dead as a general account

The claim.  Mobile animals fled upward as the water rose, so they are buried higher.  Clarey and Tomkins retain it for one specific case, suggesting that the more mobile theropods escaped to higher ground and were buried in Jurassic layers.²

Five ways it fails, and the first three are fatal.

Plants cannot flee at all — yet plants are ordered as sharply as animals.  Lycopods, then seed ferns and conifers, then flowering plants.  Escape ability cannot touch this.

Sessile marine invertebrates cannot flee — yet corals, bryozoans, crinoids and brachiopods each have their own stratigraphic order.  Rugose and tabulate corals occupy the Paleozoic; scleractinian corals occupy the Triassic upward.  Neither swims.

Marine animals could not flee the water at all — yet they are the most finely ordered group in the entire record, and they are the basis of every biostratigraphic zonation ever built.

Juveniles should be sorted below adults of the same species.  They are slower, smaller and denser per unit of drag, on either mechanism.  No such bimodal sorting exists anywhere.  Bonebeds routinely contain full age series in a single layer — Maiasaura nesting horizons, the Coelophysis assemblage at Ghost Ranch, Centrosaurus bonebeds.  That is the opposite of what differential escape predicts.

Flying animals should be at the very top.  Pterosaurs first appear in the Triassic, below most dinosaurs.  Birds appear in the Jurassic.  Bats appear in the Eocene.  Insects — the most mobile and by far the most numerous flying animals — first appear in the Devonian, below almost every land vertebrate.

Verdict:  dead as a general account.  Between them, plants, sessile invertebrates and marine life are the overwhelming majority of the fossil record, and the mechanism cannot address any of them.  A mechanism that cannot address most of the data is not an explanation of the data.

It survives in a narrow local role.  Within a single flooding event it is entirely plausible that a mobile animal is buried further up-slope than an immobile one, and Clarey's use of it for the dinosaur succession from the Morrison to Hell Creek is not silly.  As a general account of the column it is dead.

Ecological zonation — sound but weak, and it contains the movement's best result

The claim.  Organisms were buried in the order their habitats were inundated:  deep marine first, then shelf, then coastal lowland, then upland, then mountain.  Clarey traces it to Harold Clark in 1968 and Harold Coffin in 1983, and it is the mechanism he and Tomkins actually adopt.

What it gets right, and this is a genuine success.  It predicts a monotone marine-to-terrestrial succession up the column.  Computed here from Macrostrat units carrying Paleobiology Database occurrences, the non-marine share across the six megasequences runs:

That is perfectly monotone across six intervals and spans a factor of several hundred.  Clarey's statement that the first three megasequences hold nearly 100% marine fossils is not rhetoric:  on this measure it is 100.0%, 99.9% and 98.3%.  A skeptical reader should be told that the pattern is real, that it was computed here from a mainstream database that owes creationism nothing, and that it is not in dispute between the two sides.  What is in dispute is the cause.  It is the single best thing young-earth geology has on fossil order and this article says so plainly.¹⁵

Where it fails, and the failures are hard.

No upland community appears anywhere below its predicted level, on any continent.  If burial order tracks elevation, then somewhere on earth an upland biota should have been caught early — a mountain valley breached by a surge, a plateau margin overrun on Day 20.  Across roughly 1.3 million occurrences there is no such case.  The elevation filter is asked to have operated with perfect fidelity, simultaneously, on five continents, through six successive surges.  A model that must be 100% efficient globally is not a physical model.

Plants follow animals, and pollen follows plants.  Zonation is about where an organism lived.  It says nothing about why flowering plants should be zoned above conifers and conifers above lycopods, in exactly the pattern a growth-of-diversity model predicts.  And pollen travels on wind across every elevation belt and settles last of all.

The three marine megasequences differ from each other and the model has no account of why.  Tomkins and Clarey concede it in print:  "Questions about why distinctly different fossil assemblages are found in the Sauk, Tippecanoe and Kaskaskia Megasequences remain, even though all are dominantly marine."²  If all three are shallow-marine habitat at the same elevation, zonation cannot separate trilobite-dominated Cambrian faunas from Ordovician graptolite-and-brachiopod faunas from the Devonian "age of fishes."  That is not a detail.  It is the first half of the column.

There are freshwater and terrestrial deposits inside the early megasequences, and they can be named.  The 1,211 non-marine occurrences in the first three megasequences are not a rounding error to be waved away; they are the exceptions the model has to absorb.  Nearly all are Devonian and Mississippian, in the Kaskaskia:  the Escuminac Formation at Miguasha in Quebec, the classic Eusthenopteron fish site, with 113 occurrences; the Battery Point Formation of the Gaspé, with early land plants, 60; the Diamond Peak, Pocono, Horton and Albert formations.⁶¹  On Clarey's own chronology these were deposited on Days 30 to 40, before the Ark floated, when he says the land was still dry and only rain was falling.  A fluvial and lacustrine fish-and-plant deposit is an awkward thing to lay down on Day 35 of a global Flood.

Trace fossils.  Trackways, burrows, nests and rooted horizons occur at hundreds of levels through the column.  Each requires an animal walking on, or a plant rooted in, a surface that was then buried — repeatedly, within one year, under kilometres of accumulating sediment.  Tomkins and Clarey concede this too, and offer a hope rather than an answer:  "Our interpretation still has a bit of difficulty explaining the trackways and footprints found in some layers of the Cenozoic."²

Verdict:  sound but weak.  Zonation is a filter, not a mechanism.  It explains which broad habitat is buried when.  It explains nothing about ordering within a habitat, which is most of the fossil record.  Its own proponents say so in print.

Woodmorappe's TAB provinces — needs work, and quietly abandoned

This is the most substantial single piece of young-earth work on fossil order ever published, and the movement almost never cites it today.  John Woodmorappe's "A Diluviological Treatise on the Stratigraphic Separation of Fossils" ran to fifty-three pages of the Creation Research Society Quarterly in December 1983.⁵⁹

What he did.  He plotted over 9,500 global occurrences of thirty-four major index fossils on thirty-four world maps, by hand, two decades before the Paleobiology Database existed, and derived 479 juxtapositional determinations — measurements of how often any two index fossils are found in the same place.  His headline results, from his own abstract:  one-third of fossil families span three or more geologic periods; periods with four intervening periods between them still show double-digit familial faunal similarity; and very rarely are more than a third of the thirty-four index fossils simultaneously present in any region 320 kilometres across, and never more than half.⁵⁹

What he concluded.  He evaluated the standard Flood explanations — pure chance, selective preservation, differential escape, hydrodynamic selectivity, ecological zonation — and rejected all of them.  That is worth pausing on:  a creationist did the demolition work on hydrodynamic sorting in 1983, which is why he had to propose something new.  His replacement is TAB, tectonically associated biological provinces:  the pre-Flood world had regionally distinct biotas, sedimentation is strongly controlled by tectonics, and if tectonics and biogeographic zonation are linked then biogeographic provinces must be superposed in a consistent manner, which produces biostratigraphic separation.⁵⁹

The clever move at the heart of it.  Because index fossils rarely co-occur, Woodmorappe argues, the sorting mechanisms do not have to be efficient.  They need to work consistently only a few times relative to any two organisms in order to bury them in consistent order.⁵⁹  That is the strongest defensive argument the movement has ever produced, because it removes the requirement of a perfect global sorting engine and replaces it with a modest number of local superposition relationships.

Three problems, and one of them is structural.

The statistics do not discriminate.  "One-third of families span three or more periods" is compatible with any model of the record including standard paleontology.  It does not distinguish between hypotheses, so it cannot support one.

The window is arbitrary.  The 320-kilometre co-occurrence radius is chosen without justification, and the headline result depends heavily on that choice.  A different radius gives a different number.

And the mechanism does not deliver the direction.  TAB supplies no reason why the tectonic succession should track the biological one the same way everywhere — which is precisely what needs explaining.

The structural problem is created by Clarey, and neither man mentions the other.  Woodmorappe's model works because the order is locally contingent, set by chance, biogeography and tectonic setting at each site.  A contingent process produces a different order in different places.  Clarey's entire programme rests on the claim that the same six-fold order appears identically on five continents.  The two flagship young-earth models of fossil order require opposite properties of the same data.  Woodmorappe does not appear in Clarey and Werner's 2023 reference list at all.

Verdict:  needs work.  Serious, quantitative, original, and never refuted in its own terms — but forty-three years without an update, never tested against a modern database, and structurally incompatible with the model the same movement now promotes.  It has been quietly abandoned rather than answered, and a reader deserves to know that.

Floating forests — needs work, and refuted from inside

The claim.  A pre-Flood floating ecosystem — Kurt Wise's "floating forest," developed from Joachim Scheven's work — supplied the biota that became the Carboniferous coal measures, explaining both the coals and the peculiar anatomy of the plants in them.  The modern statement is by Steven Austin and Roger Sanders.⁶²

The paleobotanical case is real and interesting.  Sanders and Austin argue that lycopsid and tree-fern anatomies are characterised by air-filled chambers for buoyancy, with rooting structures not suited for growth into and through terrestrial soil, and that lycopsid development involved buoyant, photosynthetic, branching rhizomorphs established before upright stem growth.⁶³  Lycopsid rhizomorph anatomy genuinely is odd for a soil-rooted tree.  Austin's independent line is coal petrology:  that Carboniferous coal is detrital, accumulated underwater, rather than a peat swamp compressed in place — a reading with a long non-creationist pedigree in the French allochthonist school.

Its strongest published refutation is Tim Clarey's, in three papers between 2015 and 2017.⁶⁴  Two arguments do the work.  He and Tomkins examined an in situ lycopod forest site whose root systems penetrate the substrate, which a floating mat cannot produce.  And the largest and most extensive coal beds in the world are Cenozoic, not Carboniferous, so a pre-Flood floating forest cannot be the source of most coal.  The second point is decisive on its own arithmetic, and it comes from the man whose own megasequence database holds the coal thickness figures.

Verdict:  needs work.  The model explains at most the Carboniferous coals, which are a minority of world coal.  It has an in situ counterexample published against it by a fellow creationist.  And it plays no role in the megasequence model the same institution now promotes.  This article could not obtain the full text of Clarey's three papers, and says so:  the summary above rests on their titles and on the back-references in his 2023 paper.

Part 6 — The best young-earth model, tested

Tim Clarey of the Institute for Creation Research has built the only young-earth account of fossil order resting on a continent-scale stratigraphic database:  nearly 3,000 columns from oil wells, measured sections and seismic lines across five continents, mapped onto Sloss's six megasequences.⁶⁵

He is a real geologist.  Two master's degrees and a doctorate in geology and hydrogeology, eight years as an exploration geologist at Chevron, seventeen years as a full professor and geosciences chair at Delta College in Michigan, and a pre-ICR publication record of ordinary secular structural geology in refereed venues.⁶⁶  This article treats his model as the serious proposal it is.

What the megasequences are, in conventional geology

This has to come first, because both the strength and the weakness of the model flow from it.

Laurence Sloss, in 1963, divided the Phanerozoic rocks of the North American craton into six unconformity-bounded packages and named them after Native American peoples:  Sauk, Tippecanoe, Kaskaskia, Absaroka, Zuni, Tejas.  Each is a body of strata bounded above and below by interregional unconformities that can be walked across the continent.⁶⁷

Three things about Sloss's units are load-bearing and are usually left out of creationist summaries.

They are rock units, not time units.  A sequence boundary is a surface, and the surface is diachronous — younger in the basin centre than at the craton margin.  Sloss said so.  This is why a sequence cannot be mapped onto "Days 40 to 90 of the Flood year" without doing violence to what it is, and Clarey's model needs the boundaries to be effectively synchronous worldwide.

They were defined for the North American craton and nowhere else.  The extension to other continents is real published work and Clarey cites it correctly, but those authors extended the concept, and the correlations they drew are interpretations rather than measurements.

The framework has moved on.  Since the late 1970s the field has worked at much finer scale, attributing sequences to relative sea-level change from a mixture of eustasy, subsidence and sediment supply.  Sloss's six are now treated as a coarse, useful, partly tectonic subdivision, and modern workers dispute how global they are.

One conclusion follows immediately, and it goes in Clarey's favour.  The charge sometimes made by his critics — that megasequences are a creationist invention — is false.  So is the charge from inside creationism that they depend on the evolutionary timescale, at least at the level of definition:  sequence boundaries are picked on unconformities and lithology and only dated afterwards with biostratigraphy.  Macrostrat's units partition cleanly into the six intervals on lithostratigraphic grounds alone.  What Clarey does with the framework is where the argument is.

What Clarey claims

The 2023 flagship states it compactly.  The earliest megasequences show the least extent and lowest average thickness of sediment and are almost exclusively filled with marine fossils.  Later ones show progressively more coverage, more sediment volume, and progressively more land animals and plants.  Five continents show a maximum peak in surface coverage and in volume during the fifth sequence, around the Cretaceous–Paleogene boundary, which he identifies with Day 150 of the Flood.⁶⁵

The mapping onto Flood days, with his own question marks preserved:

He is honest that Days 1 to 40 are subdivided by guess.  Where he puts the Flood/post-Flood boundary is the single most consequential thing about the model:  high, at or near the top of the Neogene.  Not the Cretaceous–Paleogene.  That is what puts him at war with half of creationist geology.

His empirical content is one table.  For North America alone, his areas run 12.16, 10.25, 11.04, 11.54, 16.01, 14.83 million square kilometres, and his volumes 3.35, 4.27, 5.48, 6.34, 16.45, 17.76 million cubic kilometres, across the six megasequences in order.⁶⁵  From the five-continent version he derives the numbers the model turns on:  Zuni 32.6% of global Phanerozoic sediment volume, Tejas 32.5%.

How he explains fossil succession is worth stating precisely, because it is not what popular apologetics says.  Not by hydrodynamic sorting.  By ecological zonation driven by progressive inundation of elevation belts:  the fossils reflect a steadily changing record of different ecological zones, and the earliest three megasequences inundated only shallow marine environments.⁶⁵  He reads the mass extinctions the same way — as the last occurrence of organisms in the Flood record, caused by the simultaneous inundation of a complete ecological environment at roughly the same elevation globally.⁶⁵

Venue, and why it proves less than it looks

Of twenty-eight creation-side items in his megasequence corpus, none appeared in a secular peer-reviewed journal.  Three secular items exist and all three are conference abstracts — two to the Geological Society of America, one to the American Association of Petroleum Geologists — which are screened for topicality, not refereed.⁶⁸

That fact is worth much less as evidence than it first appears, and honesty requires saying so.

Mainstream journals are not neutral about this subject.  The clearest demonstration is not a Flood paper at all.  In January 2016 PLOS ONE published a biomechanics paper on hand coordination by a group at Huazhong University of Science and Technology.  Its abstract described the tendon architecture of the hand as "the proper design by the Creator."⁶⁹  Within days the paper was retracted.⁷⁰  The lead author's explanation was that English was not the authors' first language and that they had meant nature and evolution rather than God.⁷¹  PLOS ONE's stated grounds were broader than the phrase — scientific rationale, presentation and language not adequately addressed in review — but the paper had stood without complaint until the theological reading spread, and the reaction is what prompted the review.

If three words in an abstract can do that to an ordinary biomechanics paper, a paper openly arguing for a global Flood is not going to be judged on its stratigraphy.  So the absence of Clarey's work from secular journals is close to uninformative about its quality.  A reader should discount the venue point accordingly, and this article does.

One thing does count in the venue's favour, and it is worth noting because it is the opposite of what a reader expects.  The proceedings volume in which Clarey's flagship appeared was edited by John Whitmore of Cedarville University — who is, as the next section shows, one of Clarey's principal opponents on where the Flood ended.  The venue is not captive.

Two criticisms survive the discount, and they are the ones the argument should rest on.

First, the stratigraphic compilation itself is publishable without the Flood.  A description of megasequence thicknesses and extents across five continents is ordinary sequence stratigraphy.  Sloss published the framework in 1963 and it is standard.  The Flood interpretation could be omitted, or confined to a discussion section, and the data would still be a contribution.  That route has not been taken.

Second, and decisively, none of this explains the data not being released.  Journal bias is outside an author's control.  Publishing a dataset is entirely within it.  That is where the weight of the criticism belongs, and it is not affected by anything in the paragraphs above.

The database is not public

No supplementary file, no repository, no data appendix, no DOI'd dataset accompanies any of the papers.  The published empirical content is six isopach maps and one table of twelve numbers, produced in commercial software from maps that were hand-trimmed by a judgement call applied to each one.⁷²

Two further specifics a reader should know.

The megasequence boundary picks are interpretations.  In an oil well, a sequence boundary is picked from a wireline log signature plus, usually, biostratigraphy.  Clarey argues the picks are independent of fossil content.  For the type sections that is broadly true.  For a well in the subsurface of Angola it is often not, and the papers do not distinguish the two cases well by well.

Volume is thickness times area, and area comes from a hand-trimmed map.  Clarey states that the machine-generated isopach maps were trimmed to match the extent of each megasequence shown by the basal lithology maps.  That trimming is a judgement call applied to every one of the twelve maps that produce his table, and it is undocumented.

This is not a partisan objection.  It is what any journal referee would say.  It means the model cannot be replicated — it can only be checked against other data.

Checking it against Macrostrat

Macrostrat is open, keyless and CC-licensed.  Its North America project holds 949 stratigraphic columns covering 24.8 million square kilometres — within 7% of the area Clarey reports for the same continent — carrying 24,141 rock units, each with a name, top and bottom ages, maximum and minimum thicknesses, a lithology, an interpreted depositional environment, and a count of Paleobiology Database fossil occurrences.  That is the same kind of measurement Clarey's isopach maps are built from, gathered by a different community for a different purpose.⁷

The test was run on his terms.  Megasequence boundaries were set from Clarey's own definitions in system terms — Sauk from the Lower Cambrian to the Lower Ordovician, Tippecanoe from the Middle Ordovician to the top of the Silurian, and so on — and converted to standard stage ages.  Two scripts do the work and are published with this article.⁷³

What survives the check

His fossil-succession claim survives, and survives more strongly than he states it.  The marine-to-terrestrial progression in Part 5 is his result, independently reproduced from data he did not use, and it is perfectly monotone across all six intervals.

His volume shape survives.  Macrostrat gives volume shares of 12.3, 11.7, 14.7, 18.1, 24.3 and 18.8 percent across the six megasequences — a dip at the Tippecanoe, then a climb through the Kaskaskia and Absaroka to a Zuni peak, which is the shape he reports.  The Zuni is the single largest package in Macrostrat's North America, exactly as he says, and the peak survives a sensitivity run using minimum rather than maximum unit thicknesses.⁷⁴

Whatever the cause, more North American sediment was laid down in Jurassic-to-Cretaceous time than in any equivalent slice of the Paleozoic, and the Paleozoic packages really are the thin ones.  That is a real, independently confirmed result and the article states it plainly.

And the framework itself survives.  Macrostrat's 24,141 units partition cleanly into the six intervals.  There is nothing artificial about the subdivision.

What does not survive the check

Problem 1 — the magnitudes disagree, and they disagree in exactly the direction that manufactures the trend.

The two datasets agree on the Tejas to within 18% and disagree on the Sauk by a factor of four.  The discrepancy is monotone in age — large and old, small and young.  That matters because the entire "progressive" signal in Clarey's table is carried by how small his early numbers are.  If Macrostrat is right about the Sauk, his Sauk-to-Tejas volume ratio of 5.3 collapses to about 1.5, and the smooth ramp that anchors Days 1 to 40 becomes a gentle rise.

This is not proof that Clarey is wrong.  It is proof that two independent compilations of the same continent's stratigraphy differ by up to four times on the quantity his model rests on, with no way to adjudicate, because one of them is unpublished.

Problem 2 — the areas do not rise.

Clarey's progressive-Flood claim is above all a claim about area:  the water covered more and more of the continent until Day 150.  His North American areas rise 32% between the Sauk and the Zuni.  Macrostrat's six areas span 12.1 to 16.9 million square kilometres and are essentially flat, with the largest being the Tejas and the second largest the Tippecanoe.  The Absaroka — the start of his rising-water phase — is the smallest of the six, and the Zuni, his high-water megasequence, is smaller than the Tippecanoe, the Kaskaskia and the Tejas.⁷⁵

Problem 3 — the direct flooding test goes the wrong way, and it reproduces the very curve Clarey rejects by name.

The area metric above is coarse, for a reason set out below.  So a sharper test was run:  in each 10-million-year bin, what area-weighted fraction of North American columns contains strata that Macrostrat's compilers class as marine?  That is a proxy for how much of the continent was under the sea, computed from rock facies rather than from anybody's sea-level model.⁷⁶

North America's marine coverage peaks in the Late Ordovician at 45.7% and never gets that high again.  The second peak is Early Cretaceous, at 35.6% — high, but lower than the Ordovician.  The Tejas has the lowest mean of all six.  Under Clarey's chronology the Ordovician maximum falls on roughly Day 25, when he says the water was still confined to shallow seas on the continental margins and the Ark had not yet floated.

Worse for the model, that is precisely the shape of the conventional two-peaked sea-level curve, which Clarey reproduces as his own Figure 1 and then rejects.  Discussing Andrew Snelling's attempt to tie Flood water levels to the Sauk and the Zuni, he writes that both show the highest sea levels on the uniformitarian curve "but are not reflective of the sedimentary rock record."⁶⁵  Macrostrat is a sedimentary rock record — 24,141 units of it — and it gives the two-peaked Ordovician-and-Cretaceous shape.  On its own terms, using the same kind of data, the independent database supports Snelling's reading of the flooding history over Clarey's.

Problem 4 — the mean-thickness ramp does not reproduce.  Clarey's North American mean thicknesses climb smoothly from 0.275 to 1.198 kilometres, a 4.4-fold rise with no reversals, and it is the cleanest-looking number in his paper.  Macrostrat gives 1.109, 0.917, 1.186, 1.666, 2.102, 1.234.  The Zuni maximum is there.  The smooth ramp is not, and the Tejas falls back to Kaskaskia levels.

Problem 5 — the Zuni peak that dates Day 150 is inside the noise.  Clarey anchors Day 150, the high-water mark of the whole event, on the Zuni being the peak.  In his own five-continent table the Zuni beats the Tejas by 32.6% against 32.5% of volume, and by 99,964,200 against 98,772,900 square kilometres of area — a difference of 1.2%.  These are figures produced by hand-trimming computer-generated isopach maps, and no uncertainty is quoted anywhere in the paper.  A margin of one-tenth of a percentage point cannot carry the weight of dating the peak of the Flood.

Problem 6 — the Tejas volume is metric-dependent.  Re-running the Macrostrat comparison with minimum rather than maximum unit thicknesses gives shares of 12.8, 14.0, 19.6, 20.3, 22.1 and 11.3 percent.  The Zuni peak holds.  The Tejas swings from second-largest to smallest.  Clarey's Tejas figure — 32.5% of all Phanerozoic sediment, the number that carries his high Flood/post-Flood boundary — is the least robust number in the whole comparison.

The limits of this test, stated plainly

An audit that does not audit itself is worthless.  Six things weaken the check above, and three of them cut against the conclusions just drawn.  This section is here because without it the check is just another partisan number.

1.  Macrostrat is onshore-dominated and Clarey includes the shelf and deep water.  His Tejas argument leans heavily on offshore deposits — the Gulf of Mexico "Whopper Sand," offshore Asian coals, the seventeen kilometres of Neogene fill in the South Caspian Basin.  Macrostrat's North America project barely samples any of that.  This biases the test against the Tejas, which is where the two datasets disagree most about interpretation.  It does not affect the Ordovician flooding peak.

2.  The area metric is coarse, and the bias runs in favour of this article's own conclusion.  A Macrostrat column averages about 26,000 square kilometres, and a column counts its whole footprint toward a megasequence if it holds even one unit of that age.  That inflates the apparent extent of thin, patchy early megasequences relative to Clarey's finer isopach trimming.  Problem 2 above is therefore suggestive rather than decisive, and must be discounted accordingly.  The flooding curve in Problem 3 is much less exposed, because it compares the same columns against themselves across time, so a constant granularity bias cancels out of the shape.  The flooding curve is the robust result.  The area table is not.

3.  Maximum thickness overestimates volume.  Macrostrat's maximum reported thickness for a unit is not a mean thickness over the column's footprint.  Both metrics were run.  The disagreements in Problem 1 shrink under the minimum-thickness metric but do not vanish.

4.  Age apportioning assumes uniform accumulation.  A unit spanning a megasequence boundary is split in proportion to its age range.  For units with loose age constraints that smears material across boundaries.  It affects only straddling units and cannot create a trend.

5.  "Marine" in Macrostrat is an interpretation, not a measurement.  It is a facies call made by a compiler from lithology and sedimentary structures.  A young-earth reader may reject the whole interpretive apparatus.  That is the honest weak point of the strongest finding here, and it must be said.  Note that the fossil result does not depend on it in the same way, because a brachiopod is a marine animal and a hadrosaur is not, whatever anybody thinks of the enclosing sandstone.

6.  Both compilations measure preserved rock, not deposited rock.  Clarey confronts this directly and his three counter-arguments are good ones and deserve to be stated.  The internal sandstone-shale-carbonate cycle of the Sauk survives intact across much of North America, which vast interstitial erosion should have destroyed.  Reworked early Paleozoic fossils are not abundant in Zuni and Tejas strata, as they should be if the Paleozoic had been stripped and redeposited.  And Africa, which has no Cenozoic mountain belt to shed sediment, shows the same thin-early, thick-late pattern as continents that do.  None of these is decisive, but "it was all eroded away" is not a free move and this article does not pretend otherwise.

The honest summary

The megasequence programme is real geology doing real work.  Its fossil-succession result is sound and worth citing.  Its central claim of progressive continental inundation is not supported by the only open database large enough to test it.

Three components, three verdicts.  The fossil succession is sound, and is the best-supported quantitative claim in young-earth geology of the fossil record.  The sediment-volume claim is sound but weak:  the Zuni peak reproduces, the magnitudes do not, and the margin that dates Day 150 is a tenth of a percentage point.  The progressive-inundation claim needs work, bordering on dead:  areal extent does not rise, and the direct marine-coverage test peaks in the Late Ordovician and declines.  Until the 3,000 columns are released there is no way for an outside reader to see why the two datasets disagree, and the burden is on the party holding the closed data.

The movement has not settled where the Flood ended

The young-earth community has not agreed where the Flood ended in the rocks, and the disagreement is substantive rather than cosmetic.  The candidate boundaries are roughly sixty million conventional years apart.  A reader deserves to know that the people who share Clarey's starting assumptions have been arguing with him for a decade, in print, with data.

Froede, Akridge and Reed — the boundaries cannot be time lines

Carl Froede, Jerry Akridge and John Reed argued in the Journal of Creation in 2015 that the whole approach is compromised by its dependence on the geological timescale, and that physical correlation independent of the timescale, spanning entire continents, "remains to be demonstrated."⁷⁷

Half of that is wrong and Clarey answers it well.  Sequence boundaries are picked on unconformities and lithologic change and only dated by biostratigraphy afterwards, and Macrostrat's units partition into the six intervals on lithostratigraphic grounds, which supports Clarey.

The other half is strong and he has never answered it.  If a global Flood deposited the entire Phanerozoic in one year, then a "sequence boundary" is a surface produced by a change in current velocity or sediment supply — and there is no reason such a surface should be synchronous across five continents.  Similar processes at different moments of the Flood produce similar-looking boundaries at different times.  Clarey's model needs the six boundaries to be time lines.  Sloss's own framework says they are diachronous even within North America.  This is the deepest unresolved problem in the model and it comes from inside the movement.  Froede and his co-authors wrote in 2015 that Clarey's detailed work remained unpublished and that they looked forward to seeing how he handled these problems.⁷⁷  The 2018 and 2023 papers do not handle them.

Whitmore and Garner — criteria instead of levels

John Whitmore and Paul Garner take the opposite approach.  Where Clarey maps a boundary onto a chronostratigraphic level and extends it globally, they refuse to use chronostratigraphy at all:  their method "is independent of chronostratigraphic indicators (that is, radioisotope dates and zone fossils)."⁷⁸  They instead ask what kinds of deposit a Flood should and should not produce.  They would not expect glacial, lake and windblown deposits during the Flood, and would expect marine deposits on the continents, global unconformities, massive tectonic activity, mass-kill deposits, and deposits of unparalleled extent.⁷⁸

Applied to Wyoming, the criteria put the boundary below the Green River Formation — that is, in the early Cenozoic, far below Clarey's line.  Their case is physical rather than chronological, and it is worth stating in detail because it is field evidence collected by creationists.  Continental plants and animals unconformably overlying marine Cretaceous strata.  Closed lake basins formed by surrounding uplifts, with palaeocurrents directed into the basin centres.  Bird tracks, bird nests, large stromatolites, bioturbated sediment and caddisfly mounds preserved in place at the basin margins.  Thick trona deposits, which cannot precipitate from seawater because seawater is rich in calcium and magnesium, and which form today in Lake Magadi in Kenya.  And Whitmore's own fish taphonomy:  most Green River fish are not perfect specimens, but show days of decay before burial, with margin fish bursting from decay gases in shallow water while deep-basin fish did not.⁷⁸

That last point deserves emphasis.  It is a young-earth creationist geologist, working from field data he collected himself, arguing that a formation Clarey assigns to the receding Flood records the passage of ordinary time.

Ross — the biostratigraphy does not support a late boundary

Marcus Ross argued in 2012 that the fossil record shows no significant paleontological discontinuity at the Pliocene–Pleistocene boundary, and that the only substantial one is at the Cretaceous–Paleogene.  Challenged on whether the underlying database is contaminated by evolutionary assumptions, he replied that his own experience with fossil collections and evolutionary paleontologists over twenty years gave him no sympathy for unrestrained skepticism of their work, and that the multitude of fossil occurrences incongruent with a proposed geological model should make us reassess the model, not the fossils.⁷⁹

That is a young-earth paleontologist telling his own side to fix the model rather than the data, and it is the same posture this article is trying to take.  One caveat:  Ross's own paper could not be obtained for this article, and his words reach it through Chad Arment's quotation of them.

Wise and Richardson — the statistic recomputed, and half of it fails

Wise and Richardson's 2023 paper extends Ross's argument with a statistic of their own, and it is the only quantitative young-earth attempt to place both Flood boundaries from a public database.¹⁸  Their measure is the number of local, stage boundary-straddling species — NLSSS.  At each of 114 stratigraphic boundaries, count the species recorded both in the unit immediately below and in the unit immediately above, keeping only those whose two occurrences lie within two degrees of latitude and longitude of each other.  A high count means the same species was buried at the same place on both sides of the boundary.  From the shape of that curve they read five zones, and from the zones they place the pre-Flood/Flood boundary below the Ordovician and the Flood/post-Flood boundary just above the Cretaceous–Paleogene.

The right response to a statistic like this is not to quote its abstract but to recompute it, and that has been done here.  The same database, the same definition, the same 114 boundaries, with every ambiguity in their method run both ways.¹⁹  Three results, and they do not all go the same way.

First, the statistic itself reproduces.  Boundary by boundary, the recomputed counts track theirs closely — Spearman correlation 0.926 at their stated tolerance and 0.940 at the tighter one.  This is a real measurement and it does what they say it does.  It is also larger: the recomputed counts sum to 25,997 against their 16,998, because the database has grown and because the stage-matching used here recovers records theirs missed.¹⁹

Second, their Cambrian numbers are wrong, and the error runs in the direction that helps their conclusion.  Their Table 1 reports zero species in Cambrian Stage 2, Stage 4, the Jiangshanian and Stage 10, and zero straddling species at nine of the ten Cambrian boundaries.  The database is not empty in those stages.  It holds 2,149, 6,848, 945 and 2,090 species-level occurrences in them.¹⁹  The reason their query missed them is checkable:  the Paleobiology Database records the Cambrian almost entirely under regional stage names — Tommotian, Atdabanian, Botomian, Delamaran, Dyeran, Aksaian — and only 31 occurrences anywhere carry the literal label "Stage 2", 118 "Stage 4" and 19 "Jiangshanian".  Their two-million-year rule for equating regional stages with global ones was written to catch exactly this and it caught some units and missed others.  Recomputed, their summary sentence — "only 59 (5.2%) of 1133 species reported from a Cambrian stage are reported also in the following stage"¹⁸ — becomes 418 of 5,038, or 8.3%, and their three named zeros become 129 of 839, 29 of 586 and 33 of 456.¹⁹  Every Cambrian boundary straddles.

Third, and most damaging, the zones do not survive a sampling correction.  NLSSS is a raw count, so it has to rise with collecting effort:  a species can only be seen straddling a boundary at a place where somebody collected on both sides of it.  It does rise with effort, at a Spearman correlation of 0.887 against the smaller of the two occurrence totals and 0.848 against the number of two-degree grid cells sampled on both sides.  Standardise the effort — draw the same number of fossil collections from each side of every boundary, and repeat — and the picture changes:¹⁹

The Pennsylvanian–Permian zone — their "very high" zone, the one they read as the Flood's first inundation of the land — runs 2.58 times the Ordovician–Mississippian zone in raw counts, which is a strong result on a rank-sum test.  At equal sampling it runs 0.98 times it, which is no result at all.  Their own percentage measure says the same thing and they did not press it:  as a share of the species available, the Pennsylvanian–Permian zone is 13.2% against the Ordovician–Mississippian zone's 15.0%.  The zone exists in the raw count and nowhere else.

The Precambrian–Cambrian zone fails for a blunter reason.  All ten boundaries that return zero return it because no place on earth has been sampled to species level on both sides.  Zero paired localities forces zero straddling species; the statistic is not measuring continuity there, it is measuring that nobody has looked.  There is a second reason as well, and it is one Wise and Richardson raise themselves in a different context:  their Precambrian units are systems and erathems with a median length of 85 million conventional years, against 2.19 million for their Cenozoic stages.  A species has to persist far longer to straddle a Proterozoic system boundary than a Pleistocene stage boundary, whatever the sampling.¹⁹

What survives is the Cenozoic zone, and it survives every test.  It runs 2.12 times the Mesozoic raw, 2.18 times after equal sampling, and 1.63 times when the comparison is restricted to boundaries between units of similar length.  Cenozoic biostratigraphic continuity really is exceptional.

So half their conclusion stands and half does not, and the half that stands is the half that cuts against Clarey.  Their argument that the Flood/post-Flood boundary cannot sit high in the Cenozoic rests on Cenozoic NLSSS being genuinely high, and it is — that argument is strengthened, not weakened, by this recomputation, and it lands on the same side as Ross and Arment.  Their argument that the boundary sits just above the Cretaceous–Paleogene rests on a "substantial worldwide dip" at their boundaries 94 and 95, and that dip is a sampling dip.  Both of those boundaries touch the Selandian, the sparsest stage in the Paleogene:  1,644 species-level occurrences against the Danian's 9,167 and the Thanetian's 7,434.  The number of localities sampled on both sides falls from 150 at the Cretaceous–Paleogene boundary itself to 40 and 33 at the two boundaries they rely on.  Measured per doubly-sampled locality the dip inverts — 3.15, then 3.73, then 5.36, with the second of their two "dip" boundaries the highest of the four.¹⁹  The dip is the shape of the Selandian's collecting history, and it appears twice because a stage has two edges.

One thing this does not touch.  Their headline finding — that "the global biostratigraphic column seems to be real," and that burial order bears no relationship to fluctuations in Flood energy — does not depend on the five zones at all.  Both stand.  So does the value of the exercise:  a young-earth paleontologist mining a public database and publishing a result that embarrasses his own movement's sorting models is doing science, and the appropriate response to it was to check the arithmetic rather than to cite the abstract.  The arithmetic was checked here and the correction is published against a creationist by a creationist article.  That is the standard this series is asking of everyone else.

Arment — the sharpest argument in the whole literature, and it is unanswered

Chad Arment's 2020 paper in Answers Research Journal is the most quantitative internal critique of Clarey that exists, it names his model directly, and it is entirely internal to young-earth premises.  That is what makes it hard to dismiss.  It does not depend on radiometric dating, on evolutionary phylogeny, or on rejecting a global Flood.  It depends only on taking Genesis 7 literally and counting genera.⁸⁰

The premise.  Genesis 7 puts one pair of every unclean terrestrial kind on the Ark.  Everything below the Flood/post-Flood boundary is pre-Flood.  Everything above it is descended from that one pair.

The observation.  If the boundary sits between the Neogene and the Quaternary, then a great many genera appear on both sides of it, on the same continent.  Arment counts them.  Seventy-three genera in only ten terrestrial carnivore families cross the Neogene–Quaternary boundary worldwide.  Forty-four associations of still-living genera occur together at thirty-eight North American Miocene and Pliocene fossil sites.

The problem.  For each of those, the model needs the same genus to have existed pre-Flood on that continent, to have been wiped out, and then to have been re-established on that same continent from a single pair off the Ark.  Once is a coincidence.  Arment computes the probability of the coincidence repeating dozens of times and finds it negligible.

His conclusion, verbatim:  "There do not appear to be any arguments that can explain worldwide Cenozoic fossil data and trends from a Neogene-Quaternary Flood/post-Flood boundary perspective."⁸⁰

He also closes the obvious escape hatch, and names the person who offered it.  Michael Oard suggested privately that Australian fossils recorded from Miocene and Pliocene strata might be better characterised as Pleistocene, on the grounds that their ages were pushed back mainly because of allegedly primitive characteristics.  Arment's reply is that this does not explain the South American marsupial fossils, and that the argument seems entirely ad hoc, intended to protect a pet model from contrary data.⁸⁰

Nobody has answered him.

Where the dispute stands

There is a further exchange this article could not read.  Michael Oard published "Are global megasequences real?" in the Journal of Creation in 2025, Clarey and Werner replied in the next issue, and Oard replied again.  The journal serves subscribers only and all three pieces were inaccessible.  What can be said without them is that Oard is not a Cretaceous–Paleogene man — he has argued for a late-Cenozoic boundary since at least 2008, the same side of that question as Clarey.  Their fight is therefore about the megasequence framework itself, not about the boundary.  The two leading advocates of a high boundary reached it by different routes, and one of them thinks the other's route is invalid.⁸¹  Oard and Reed's own 2020 review of the whole field concludes that "all models need further development."⁸²

The trade-off nobody escapes

There is a conservation law under this dispute and almost no popular presentation states it out loud.

Whatever is post-Flood must be produced by diversification from the Ark's founder pairs in roughly 4,500 years.  So the higher the boundary, the less diversification is required — and the more boundary-crossing genera have to be explained.  The lower the boundary, the fewer boundary-crossers — and the more diversification has to be crammed into a few centuries.

On a Cretaceous–Paleogene boundary, the entire Cenozoic mammal, bird and flowering-plant record becomes post-Flood.  From a few thousand Ark kinds that requires producing every genus from the Paleocene through the Pleistocene — whales, bats, horses, elephants, primates, rodents — plus their extinctions, in a few centuries.  The movement knows this:  Whitmore's own companion paper at the same conference is titled "Rapid and Early Post-Flood Mammalian Diversification Evidenced in the Green River Formation," which is an attempt to pay exactly that bill.

On a Neogene–Quaternary boundary the diversification burden nearly vanishes, and the price is Arment's.

The two burdens trade off exactly against each other, and no published young-earth model pays both.  That is not a rhetorical point.  It is a constraint every model must satisfy, and stating it is more useful than adjudicating between the two positions.

Objections and responses

"The geologic column is circular reasoning — rocks are dated by fossils and fossils by rocks."

The column was assembled before Darwin published and before radiometric dating existed, by field geologists comparing vertical sections, most of whom were creationists.  O'Rourke, the source usually quoted to press this objection, says exactly that in the same paper.  His account of what a stratigrapher actually does is induction, not circularity:  on a thousand previous occasions certain taxa have been found in a certain vertical order, and on the thousand-and-first the expert arranges them in the customary order.³  Once a sequence has been observed a thousand times, using it to predict the thousand-and-first is ordinary inductive inference.  And the prediction is checked commercially, by people who lose money when it fails.

The quotation itself needs correcting, and the correction is ours to make.  The famous sentence about geologists never bothering to think of a good reply is cut in the middle.  It continues:  they feel explanations are not worth the trouble as long as the work brings results — and O'Rourke then labels that attitude and spends the paper replacing it.³  His conclusion is that once the process is acknowledged as the pragmatic basis of stratigraphy, both local and global sections can be treated as chronologies without reproach.³  Quoting the first half is quoting the setup and dropping the punchline.

"The complete geologic column does not exist anywhere on earth."

It does, in at least thirty-one basins.  The Williston Basin's Larson #1 well in North Dakota is documented formation by formation from the Fort Union at 100 feet down to Precambrian basement at 14,945 feet, total depth 15,064.⁸³

More usefully for a young-earth reader:  the paper that made this argument famous concedes it.  Woodmorappe's 1981 study measured how many of the ten Phanerozoic periods are present at each place on earth — nearly 13% of the land surface has five, slightly under 1% has all ten — and titled the result "the essential nonexistence" of the column.  But in the same paper he writes that there apparently are regions where all ten periods are found superposed, and names one:  all ten are "undoubtedly represented" in the Holy Cross Mountains of south-central Poland.  He also names West Nepal and the Bolivian Altiplano.⁸⁴

And the measurement does not support the title.  Woodmorappe measured completeness, not order.  A missing period is a gap, and mainstream geology has never claimed otherwise — that is what an unconformity is.  His twenty-five pages do not report a single locality where the periods are present in inverted order in undeformed rock.  A column that is patchy everywhere but never out of order is exactly what conventional geology predicts.  Anyone citing his 15-to-20-percent figure needs to know the same author concedes the counterexample eight pages later.  Print both or print neither.

"Marine fossils on mountaintops prove a global flood."

Marine strata really do cover the continents to a remarkable degree, and that is worth saying.  The summit rocks of Everest are a well-bedded limestone with fragments of trilobites, brachiopods, ostracods and crinoids in them, collected within metres of the top.  That is a striking fact and it is not in dispute.¹²

But marine rock on a mountain has two candidate explanations, not one.  Either the sea rose over the mountain, or the sea floor rose to become the mountain.  The second is directly observed:  Himalayan uplift is measured today by GPS in millimetres per year, and the rock is folded, faulted, cleaved and metamorphosed in ways that require hundreds of kilometres of lateral shortening — deformation a rising water level cannot produce.  The summit fossils are also Ordovician shelf faunas in normal marine stratigraphic order within the section, not a jumbled drowning assemblage.

This argument should be retired, or restated as the observation rather than the inference.  And one more thing should be said with it:  the claim that mainstream geology denies the sea was ever there is a straw man that nobody holds.  Mainstream geology says the sea was there repeatedly, for long intervals, and names the transgressions.

"Out-of-place fossils are found all the time and get suppressed."

Twenty-nine claims were audited here and none survived.  Several were killed by creationists using better technique.  Meanwhile genuine surprises — the Zachełmie trackways, Middle Triassic angiosperm-like pollen, Cretaceous grass phytoliths — were published in Nature and Science, argued about by name, and absorbed by moving a datum.  Notice also the direction of the known failure modes:  six of the seven move material up the column or sideways.  Only a thrust fault moves rock down, and it leaves a visible fault surface.  There is no ordinary sedimentary process that fakes a rabbit in the Cambrian.

"You are relying on a mainstream database assembled by evolutionists."

Every record in it carries a literature citation and a collection number.  The analyses in this article are published as scripts and can be re-run by anyone, including anyone who wants to argue the classifications are wrong.  That is the opposite of asking a reader to take the mainstream on trust.

Note also that this article reports the database's own errors when it found them.  A lungfish filed under Reptilia.  A sponge bound to a wasp.  A ginkgo record whose stratigraphic bin is 107 million years wide.  Eighty-five "Paleozoic flowering plants" that are three filing categories.  An article using this database as an oracle would not have found any of that.

"Your Macrostrat check is as unreplicated as Clarey's model."

It is not, and the difference is the whole point.  Macrostrat is open, free and keyless.  The two scripts that produced every number in Part 6 are published with this article.  Anyone can run them and get the same answers, or change the boundary definitions and get different ones.  Clarey's 3,000 columns cannot be examined by anybody outside ICR.

And this article states the limits of its own check in its own text, including the two places where the bias runs in favour of the conclusion drawn.  The area result is inflated by coarse column footprints.  The marine facies calls are compiler interpretations.  The flooding curve is the robust result and the area table is not.  A check that hid those would deserve the objection.

"Even if the order holds, that only shows a sequence, not evolution."

Correct, and this article does not argue otherwise.  Order is not ancestry.  What the order does is constrain any model that has to produce it, and a one-year global Flood is such a model.  The marine-to-terrestrial succession in Part 5 is equally consistent with a progressive Flood and with a long history of ecological expansion onto land.  It discriminates between young-earth models; it does not discriminate between young-earth geology and mainstream geology.  Whether the sequence reflects common descent is a separate question, handled in companion articles.²⁷

"Then the boundary disputes prove creationists cannot agree, which proves nothing."

Also correct, and this article is not making that argument.  Disagreement inside a research programme is normal and healthy.  The reason the boundary dispute is in this article is narrower:  the young-earth answer to the missing pre-Flood dead depends on where the boundary is, and the two leading placements make the problem worse in opposite ways.  That is a fact about the argument, not a debating point about the arguers.

What would change this verdict

Stating this in advance is what makes the verdict worth anything.

• One unambiguous, in situ, correctly identified human artifact from Paleozoic or Mesozoic rock, recovered under controlled conditions with documented provenance. Not a bone — a worked stone would do, and would be harder to explain away. This is the sort of thing that turns up in a single field season if it exists at all.
• A young-earth sorting model that predicts the brachiopod-to-bivalve ratio, or any comparable pair, quantitatively. Not an argument that sorting could in principle contribute: a number, against a measured series.
• An upland biota found below its predicted level anywhere on earth. One mountain-valley community caught early, on any continent.
• Release of Clarey's 3,000-column database, so his continental-flooding result can be replicated rather than only checked against Macrostrat. The disagreement between the two compilations reaches a factor of four on the Sauk and nobody outside ICR can currently diagnose it.
• A published answer to Arment on the seventy-three carnivore genera crossing the Neogene–Quaternary boundary and the forty-four associations at thirty-eight sites.
• Wise and Richardson's per-boundary species lists. Their paper prints totals, not the species behind them, so the gap between their Cambrian counts and the recomputed ones can be diagnosed only by inference. Their query strings and output files would settle it in an afternoon, and would show whether the fault is the regional-stage mapping identified here or something else. The same demand is being made of Clarey two sections above, and it has to be made evenhandedly.
• A straddling statistic that is sampling-standardised from the start. The recomputation here corrects NLSSS after the fact. A version built on a subsampling protocol — and computed at genus level, which Wise and Richardson themselves recommend for the sparse intervals — would be a better instrument than either, and it would test the one zone that survived.
• A second independent check of the Ordovician flooding peak. It is the single most consequential number computed here and it rests on one database and one facies classification. Two more independent datasets giving the same maximum would settle the progressive-inundation claim.
• A defended young-earth estimate of the pre-Flood human population, with a method. Without one, the small-population answer in Part 4 cannot be priced in either direction.
• The three sources this article could not open: Michael Oard's 2025 megasequence exchange in the Journal of Creation, Marcus Ross's 2012 boundary paper, and The Genesis Flood itself. Each is named at the point where it is used, with the route that failed.

Sources

Research files for this article, including every script and its raw output:  research/order-and-anomalies.md, research/yec-order-models.md, research/first-appearances.py, research/audit-earliest.py, research/sweep-absent-categories.py, research/megasequence-macrostrat-check.py, research/flooding-and-fossils-check.py, research/nlsss.py, research/nlsss-cambrian-labels.py, research/README-nlsss.md.  The quotation audit for this series, with every verdict including the failures, is published separately.⁸⁵