The Missing Transitions

"We conclude that the gap in the fossil record has been an artifact of collection failure." — Smithson, Wood, Marshall and Clack, PNAS, 2012, on Romer's Gap, which had stood for about eighty years.¹

"The new species shows that the plastron evolved before the carapace." — the paper describing Odontochelys, in Nature, 2008.  The belly shell came first.  Nobody had predicted it from fossils.  Embryologists had already seen it in living turtle eggs.²³

Of the 16,804 genera in the Paleobiology Database with occurrences in three or more stratigraphic stages, 75.9% have a hole in their own range.  That is not an estimate.  It is counting.⁴

Summary

An argument from absence has a shelf life.

Somebody publishes a list of transitions the fossil record has not produced.  The list is accurate on the day it is printed.  Then people go digging, and the list gets shorter, and nobody goes back and updates the page.

This article was going to be that list.  Two of the entries closed while it was being researched, so it is built differently.

The two that closed are turtles and Romer's Gap.  Both are told here first, fully, and without grudging — because a page that reports the fossils it likes and skips the fossils it does not is worthless, and because the way these two closed teaches the reader exactly what to distrust about the rest.

Then the article does what the popular literature on both sides rarely does.  It replaces the list with a measurement:  how leaky the fossil record demonstrably is, computed from the mainstream's own database, by counting.

What this article concludes

• The turtle gap closed. In 2004 it was true that no fossil stood between turtles and any other reptile. Since 2008 there is a stepwise sequence — and the sequence delivered a result nobody expected, which turned out to match what embryologists already knew. Any page still saying "Proganochelys appears fully formed and there are no intermediates" is misinforming its own readers.
• Romer's Gap closed too, in about four years of targeted digging in one Scottish valley, after standing for eighty. Its describers called it "an artifact of collection failure."¹
• Both closures left a residue that runs the other way. Filling Romer's Gap pushed tetrapod diversification earlier, and the newest work says so in Nature: the Devonian tetrapod record "is much less complete than has been thought."⁵ And in June 2026 the base of the turtle sequence was pulled out from under it, reopening a gap at the bottom.⁶
• The real argument is a number, not a list. Three-quarters of reasonably well-sampled fossil genera have a demonstrated hole in their own range. That is superposition applied to the mainstream's own data, and it does not expire. But the same data say a mammal genus known from the stages on both sides of a stage turns up inside it 69% of the time, so a long blank before a whole group needs a specific reason, not a general appeal to gaps.⁷
• Bats, pterosaurs and snakes each need their claim restated more narrowly, and one of them — snakes — mostly does not survive at all.
• Three wing origins are still blank: bats, pterosaurs and insects. No fossil shows any of the three wings in the making, and the specialists in each group say so in print. All three sit in rocks that are thinly sampled for small, delicate animals, so each is a real absence, not a proof.⁸⁹
• Darwin's "abominable mystery" was about the sudden appearance of the flowering plants, and the two creationist organisations checked here read it fairly. In other letters Darwin called that sudden appearance "a great difficulty in the views which I hold" and guessed at a hidden birthplace. Popular versions mostly drop the letter's date and qualifiers, and that is the only fix needed.¹⁰
• Several standing creationist claims are dead. This article names them, and it names them from the creationist pages themselves, which were fetched and read for this article rather than summarised.

What this article rests on

Radiometric dating is not used as evidence anywhere in it.  That is a deliberate constraint across this series.  Every argument here works from relative order — what sits above what — which young-earth geology accepts.  Where an absolute age appears it is the mainstream figure under discussion, not something a reader is being asked to believe.

The central number was computed for this article, not quoted.  The Paleobiology Database gap statistics come from a script run against the full occurrence dump, and the script is published beside this page along with its output.⁴¹¹

The creationist sources were read, not summarised.  Nine pages from Creation Ministries International, Answers in Genesis and the Institute for Creation Research were fetched and read in full on 3 August 2026.  Doing that changed four of this article's verdicts, and in every case it changed them in the creationist source's favour.¹²

Every quotation was checked against a source actually opened, and the ones that could not be checked are named and left out.  Three are named in the Sources section below.¹³

The rank question, stated up front

Young-earth biology accepts common ancestry within a created kind, usually somewhere around the family level.  A fossil series only counts against that view if it crosses a higher boundary.

All five of these do, and none of them is close.  A shell-less diapsid reptile and a turtle are not one family.  A cursorial non-flying archosaur and a pterosaur are not one family.  A four-legged lizard and a snake are not one family.  Whatever else is disputed about these cases, the rank is not.

So there is no cheap exit here.  If the sequences are real, they are a problem for the young-earth position, and this article says so where they are real.

Part 1 — Turtles: the gap that closed

This is the most important section in the article, and it goes against the position this site holds.  It goes first for that reason.

The claim was true, and the date matters

A turtle is one of the strangest animals alive.  Its shoulder blades sit inside its rib cage.  No other four-limbed animal does that.  Its ribs are fused into a bony box.  Its shell has bones found in no other vertebrate.  Explaining how an ordinary reptile turned into that is a hard problem, and creationists have been right to say so.

Until 2008 the oldest turtle known was Proganochelys quenstedti, from the Late Triassic of Germany, and it had a complete shell.  Nothing stood between it and any other reptile.

That is not a creationist characterisation.  It is the opening sentence of the paper that broke the deadlock.  Li Chun and colleagues, writing in Nature in 2008:

"The oldest known turtle, Proganochelys from the Late Triassic period of Germany, has a fully formed shell and offers no clue as to its origin."²

So a creationist article written in 2004 that said turtles appear fully formed with no transitional animal was factually correct at the time.  A creationist article that says it in 2026 is not.  Between those two dates four fossils were described, and they do not add up to nothing.

What was found, in order

Ages below are the mainstream figures quoted from each describing paper.  The argument this section makes needs only the relative order, which is superposition.

The surprise: the belly shell came first

Ask anyone to guess how a turtle got its shell and they will describe armour plates spreading over the animal's back.  That is what a predator-driven story predicts, and it is what almost every popular account assumed.

Odontochelys has it the other way round.  Its plastron — the belly shell — is complete.  Its carapace is not there.  Dorsally there are only ossified neural plates and broadened ribs.  Li and colleagues state the result in one sentence:

"The new species shows that the plastron evolved before the carapace and that the first step of carapace formation is the ossification of the neural plates coupled with a broadening of the ribs."²

A creationist reader should sit with that for a moment, because it is the opposite of a story fitted to a fossil.  Nobody wanted this result.  It made the standard armour explanation harder, not easier, and it is the reason the Odontochelys paper set off years of argument.

And it matched what embryologists already knew

Here is the part that makes the fossil count for something rather than nothing.

A modern turtle embryo does not build its shell out of skin bones.  It ossifies the plastron early, and it builds the carapace by expanding the ribs and vertebrae out sideways into the skin.  That was worked out from living turtle eggs, in developmental biology, before Odontochelys was dug up.

Schoch and colleagues state the relationship plainly in a 2019 open-access paper:

"Although this was known from embryological studies in extant turtles, important steps in this evolutionary sequence have recently been highlighted by the Triassic taxa Pappochelys, Eorhynchochelys and Odontochelys, and the Permian Eunotosaurus."³

Two independent lines of evidence converged.  Embryology said the shell is expanded ribs, not fused armour plates, and said the plastron forms early.  Then a fossil turned up with expanded ribs, no armour plates, and a finished plastron.

That is what a real test looks like.  The prediction was in a different data domain from the confirmation, and the confirmation arrived afterwards.  This site applies exactly that standard to the mammal jaw-to-ear sequence and finds it decisive there.¹⁴  It has to apply the same standard here.

But it is a bush, not a ladder

The sequence in the table is not a line of ancestors, and the describing authors do not present it as one.

Look at the two middle rows.  Eorhynchochelys is older than Odontochelys and has no shell at all — while it does have the toothless beak that Odontochelys lacks.  The characters arrived out of order, in different animals.  Li Chun and colleagues say as much in their 2018 abstract, describing a Triassic turtle "that has a mixture of derived characters and plesiomorphic features."¹⁵

This cuts both ways and the article prints both.

For the mainstream: a mosaic is what a branching bush predicts.  A ladder would demand each character appear once, in order, and stay.  The mosaic is evidence for a branching pattern, not against it.

For us: the animals in that table are contemporaneous side branches, and no one of them is claimed to be the ancestor of the next.  A diagram that runs Eunotosaurus → Pappochelys → Eorhynchochelys → Odontochelys → Proganochelys as an arrow is overselling, and the people who described these fossils did not draw it that way.

And the mechanism is still disputed

The stepwise series is not disputed.  How the shell was actually built still is, and the dispute is between two describing papers.

Li and colleagues concluded from Odontochelys that the shell is not made from fused osteoderms — the skin bones that armour a crocodile — because Odontochelys has none.²  A year later Joyce and colleagues concluded from Chinlechelys that the shell "was originally a complex composite in which dermal armour fused with the endoskeletal ribs and vertebrae."¹⁶

Both statements are in published abstracts.  The order of the steps is settled better than the mechanism of the steps.

The mainstream says its own record here is bad

Worth quoting because it comes from a describing paper rather than from a critic.  Joyce and colleagues, on why the early turtle record is so thin:

"An extremely poor pre-Jurassic fossil record … only exacerbates this situation."¹⁶

Their explanation for that poverty is a taphonomic filter: a thin-shelled turtle living on land is much less likely to fossilise than a thick-shelled turtle in a river.  Their abstract says the thin shell and probable terrestrial habits of Chinlechelys "hint at taphonomic filters that basal turtles had to overcome before entering the fossil record."¹⁶

That is a legitimate explanation and it is also unfalsifiable as worded, and both halves of that sentence should be said.  A filter invoked to explain the absence of the fossils that would test it cannot itself be tested by the absence.  Part 3 is an attempt to put a number on the general phenomenon so that appeals like this one can be checked rather than accepted or dismissed.

June 2026: the base of the sequence was pulled out

This is ten weeks old at the time of writing and it changes what this article can claim.

For thirteen years Eunotosaurus africanus was the oldest arm of the turtle stem.  Four teams argued it — two papers in Nature and two in Current Biology between 2013 and 2016.  Bever and colleagues claimed in Nature in 2015 that its skull gave "strong cranial support for Eunotosaurus as a critical transitional form in turtle evolution, thus fortifying a 40-million-year extension to the turtle stem."¹⁷

In May 2026 Xavier Jenkins and fourteen co-authors put Eunotosaurus and the candidate stem turtles through X-ray computed microtomography and reported that it is not a turtle relative at all:

"Our comprehensive phylogenetic analysis recovers Eunotosaurus as a millerettid stem reptile not related to turtles."⁶

Their explanation for the resemblance is that both groups burrowed.  Broad flat ribs are useful for digging, and two unrelated diggers evolved them separately.  The journal's own companion piece is titled "Turtle evolution: Missing a missing link."¹⁸

Four things follow, and they do not all run the same way.

One.  A gap reopened at the base.  Jenkins and colleagues put turtle origins at about 255 million years, in the latest Permian, among the earliest archosauromorphs.  The oldest surviving stem turtle in the table above is Pappochelys at about 240.  That is roughly fifteen million years with no stem turtle in it, plus everything before.  A creationist may report that, and should.

Two.  Morphology misled the specialists for 130 years.  Eunotosaurus was described in 1892 and its turtle-like look was noticed immediately.  Mike Day of the Natural History Museum, who curates the original specimen, described the history in the museum's own release: opinions changed as more fossils were found, and the stem-turtle reading had been increasingly debated.¹⁹  It took µCT of the inside of the skull to settle it.  That is a genuine and instructive failure of morphological inference, and this series has a separate article on how far that failure extends.²⁰

Three.  It destroys a creationist argument that was in wide use.  Turtles were the standard example of morphology and molecules disagreeing.  Molecular data put turtles next to crocodiles and birds; morphological data put them with lizards, or with pareiasaurs, or somewhere else.  Creationists cited the conflict as evidence that phylogenetic methods are unreliable.

Jenkins and colleagues report the conflict resolved — in favour of the molecules.  Serjoscha Evers, one of the authors: "Our results show that turtles do indeed belong to the group of archosaurs and occupy a very basal position in the family tree."¹⁹  Morphology was the side that was wrong.  A creationist page arguing from that conflict now has to argue that the morphological methods it was defending are the ones that failed.

One scope note, because this series does not argue about molecular phylogeny.  What is reported here is that the conflict a creationist argument depended on has gone away.  Nothing in this article uses a molecular tree as evidence for anything, and nothing here should be read as endorsing the method.

Four.  The rest of the sequence is untouched.  Pappochelys, Eorhynchochelys, Odontochelys, Chinlechelys and Proganochelys are unaffected.  Jenkins and colleagues call Proganochelys one of the oldest uncontentious stem turtles.  The plastron-before-carapace result stands.

And a warning about relying on this result.  It is ten weeks old.  As of 3 August 2026 it has one citing work — the companion dispatch in the same journal — and no published rebuttal that could be found.⁶  This article reports it and flags it as new, on the same principle by which the companion fish-to-tetrapod article flags a one-year-old Nature trackway claim.²¹  A result that overturns thirteen years of consensus on new scans of an old fossil is exactly the kind of result that gets overturned again.

What the creationist organisations actually say about turtles

The pages were fetched and read for this article.  Doing that produced a correction, and it runs against the assumption this article started with.

Neither major creationist organisation is still printing the bare claim.

Answers in Genesis published on Odontochelys on 6 December 2008, within days of the Nature paper.  Its report of the fossil is accurate: "the turtle appears to have a bony underbody (plastron) but no upper shell (carapace)."²²  It did not deny the fossil.  Its argument was that specialists disagreed about what the fossil meant — and it named them correctly.  Olivier Rieppel called it the most primitive turtle.  Robert Reisz and Jason Head suggested it might be a land turtle that had lost its carapace, the way a modern soft-shelled turtle has.  Walter Joyce wondered whether it was an oddity.  Chun Li called it an ideal missing link.²²

All four of those readings were real in December 2008.  The page's defect is that it has never been updated, and Pappochelys arrived in 2015 and Eorhynchochelys in 2018.

Jerry Bergman published a full review for Creation Ministries International in March 2022.  That page names and discusses Odontochelys, Eunotosaurus and Pappochelys, cites Li et al. 2008 with the correct title, and quotes Schoch and Sues's own Nature abstract accurately on the difficulty of the problem: turtle origins have long been contentious "due to conflicting character evidence from molecules and morphology and a lack of transitional fossils from the critical time interval."²³

So the argument to answer is not "no intermediates exist."  It is "the proposed intermediates are not intermediates."  That is a different claim and a better one, and it deserves a direct answer rather than a dismissal.

Here is the direct answer.  Three of the reasons the page gives for rejecting the fossils fail on facts printed elsewhere in the same page.

• "Odontochelys … it is the only evidence of an animal with broadened ribs."²³ It is not. Eunotosaurus has them and Pappochelys has them, and the same page says so — it describes Eunotosaurus's ribs as "wide and flat, forming broad plates," and captions a Pappochelys figure with its ribs marked in orange. Three paragraphs apart.
• "The presence of teeth in Odontochelys and lack of a horny beak is evidence that it was a reptile unrelated to turtles."²³ This does not follow, and the page refutes it itself two sections later, where it calls Proganochelys "the most primitive known turtle." Proganochelys also has teeth. A character cannot exclude an animal from turtles if the uncontested oldest turtle has it.
• "Even the earliest known turtle, the Jurassic turtle called Kayentachelys…"²³ The same page later says Proganochelys — Late Triassic — "was, and still is, the most primitive known turtle." Both cannot be right.

And the strongest version of the page's own argument concedes the point it needs to win.  Its case against Pappochelys is that the animal is "clearly a lizard with no evidence of either a carapace or plastron."²³  But nobody claimed it had a shell.  Schoch and Sues claimed it has broad T-sectioned ribs and paired gastralia in the position where the plastron later appears, and they said in their own abstract that it has neither carapace nor plastron.  Answering "it has no shell" to a paper that says the animal has no shell is not an answer.

There is also a page that no longer exists.  Answers in Genesis's "Turtle in the Gap," on Eunotosaurus, is cited in creationist and critical literature alike.  Every path to it under answersingenesis.org now returns a 404 or a topic index.  It is not being served.  This article does not know whether that was a deliberate retraction or a content migration, and it will not guess — but it will not cite the page either.

What honestly survives about turtles

Stated as narrowly as the evidence allows.

• The turtle shell transition is documented in a stepwise series of fossils, in the right relative order, and it produced a result that independently matched turtle embryology. This is the strongest of the five cases for the mainstream and the article says so.
• The turtle origin is not documented. Which reptile lineage turtles came from, and what the animal looked like before any shell-related character appeared, is unknown — and as of June 2026 the best-supported answer to it changed.
• There is now no known stem turtle between roughly 255 and 240 million years on the mainstream scale.
• The mechanism of shell construction is disputed between two describing papers.
• The pre-Jurassic turtle record is described as "extremely poor" by the people working on it, and the explanation offered for that poverty cannot be tested by the thing it explains.

Verdict:  the claim as usually stated is obsolete.  "There are no transitional forms between other reptiles and turtles" was true in 2004 and is false in 2026.  What remains is narrower, and it is enough to be worth saying — but it is not the claim in circulation.

Part 2 — Romer's Gap: the same story, told faster

The turtle case took twenty years to turn over.  This one is cleaner, and it is worth telling because it is the pattern.

What the gap was

Alfred Romer noticed an interval at the start of the Carboniferous — the Tournaisian, and early Viséan — from which almost no tetrapods were known.  It ran about fifteen to twenty million years on the mainstream scale.

The animals on either side of it are very different.  Below the gap sit the Devonian tetrapods: aquatic, primitive, with many digits.  Above it sit animals that are terrestrial and modern in aspect, and the group already includes the ancestors of amniotes.  All the interesting change happened inside an interval with no fossils in it.

That is a real, large, well-known hole.  It is named after a real paleontologist who noticed it.  It was not a creationist invention, and creationists were right to point at it.

How it closed

By people going out and digging in one Scottish formation.

Timothy Smithson, Stanley Wood, John Marshall and Jennifer Clack described their own starting position in PNAS in 2012: before their work, only two localities anywhere had produced Tournaisian tetrapod specimens — one in Scotland with a single articulated skeleton, and one in Nova Scotia with isolated bones, many of uncertain identity.¹

Then they went and collected, deliberately, from the Ballagan Formation in the Scottish Borders.  Their conclusion is one sentence:

"We conclude that the gap in the fossil record has been an artifact of collection failure."¹

In 2016 Clack and colleagues named five new Tournaisian tetrapods in a single paper.²⁴  In 2018 Otoo and colleagues added more from Burnmouth on the Scottish coast.²⁵

The standard explanation for the gap died at the same time.  For years it had been argued that atmospheric oxygen through the Tournaisian was too low to support land animals.  The 2016 work reported oxygen stable across the boundary.  So the fossils removed the gap and the geochemistry removed the reason it was supposed to exist.²⁴

The residue, which runs the other way

Filling the gap did not simply vindicate the mainstream, and the article that skips this part is not telling the whole story.

Clack's 2016 analysis recovered three of the five new animals as stem tetrapods and two as stem amphibians, which means the deep split among modern tetrapod groups had already happened inside the Tournaisian.  Filling the gap did not produce a gradual ramp.  It pushed the diversification earlier and compressed the time available for it.

That pressure is now explicit.  John Long and colleagues reported in Nature in 2025 on early-Tournaisian clawed trackways from Victoria, Australia, which they read as made by a crown-group amniote.  If they are right, crown amniotes are 35 to 40 million years older than thought and every stem-tetrapod and stem-amniote lineage must have originated within the Devonian.  Their conclusion:

"It seems that tetrapod evolution proceeded much faster, and the Devonian tetrapod record is much less complete, than has been thought."⁵

Two caveats travel with that and they are real.  The paper is one year old and has not been through the field's response cycle.  And it is a trackway claim without body fossils — which is the same evidential structure as the Zachełmie trackways this series treats elsewhere, and as a set of Argentine "bird" tracks that had to be retracted.²¹

The companion article on the fish-to-tetrapod transition works through Romer's Gap and Long et al. 2025 in detail, and this article does not duplicate it.²¹

The rule this establishes

A gap argument has a shelf life set by the field's collecting rate.

Romer's Gap lasted about eighty years and then closed in four years of targeted fieldwork in one Scottish valley.  The turtle gap lasted a century and then produced four fossils in ten years.

Any page on this site that rests on a named empty interval will eventually be wrong, and nobody will go back and update it.  That is not a hypothetical failure mode.  Answers in Genesis's 2008 turtle page is still live, still accurate about December 2008, and eighteen years out of date.  This site would do exactly the same thing.

A distribution does not have that problem.  Which is the next part.

Part 3 — The measurement

Every argument in Parts 4 through 7 is an argument from what has not been found.  Such arguments are not illegitimate.  Absence of evidence really is evidence of absence — in proportion to how hard you looked.

Palaeontologists have measured how hard the record has looked for more than forty years:  gap counts in the ranges of fossil groups, confidence intervals on the ends of ranges, per-stage preservation rates, and comparisons between the order of first fossils and the order of branching in family trees.  The sister article How Complete Is the Fossil Record? sets them out.⁷  The popular debate rarely uses any of them.  Creationists say the record is complete enough that missing intermediates mean something.  Evolutionists say it is too incomplete for absence to mean anything.  Both usually just assert it.

This section adds one count of its own, applies it to the five groups this article is about, and then says what the published measures add.

Ghost lineages inside the record itself

The measurement needs no model, no phylogeny, no molecular clock, and no assumption about rates.  It rests on one idea a young-earth geologist accepts without argument.

If a genus occurs in stage A and again in stage C, it existed in stage B.

So if there is no record of it in stage B, that is not a hypothetical gap in the record.  It is a demonstrated one.  The animal was there.  The rocks of that age exist.  Nobody has found it.

Counting those gaps gives a base rate for how leaky the fossil record is, computed from the mainstream's own data by the mainstream's own reasoning.

The result, run against the full Paleobiology Database occurrence dump for this series:

Of the 16,804 genera with occurrences in three or more stratigraphic stages, 75.9% have at least one internal gap in their own range.  The median gap is 6.8 million years.⁴

Three-quarters of reasonably well-sampled fossil genera have a hole in the middle of their own record, and the typical hole is several million years wide.

The five clades, computed for this article

The 75.9% is a global figure.  A fair objection is that it might not apply to the specific groups this article argues about.  So the same computation was re-run, restricted to them.  Completeness has been measured group by group before, including for bats and pterosaurs, but no source found for this article gives this particular count for these five groups.¹¹

Every one of the five clades is above the global rate.  These are not unusually well-sampled groups whose gaps would be surprising.  They are groups whose records are worse than average, and the transitions in dispute sit inside them.

The caveats, printed in the same section as the number

An article that prints a statistic and buries its limits somewhere else is doing what this article criticises other people for.  So here they are, next to it.

The long-gap tail is contaminated.  The script cannot filter trace fossils and form taxa, because the bulk CSV carries no preservation flag.  A footprint genus and a body-fossil genus are counted the same way.  That inflates the extreme long gaps.  It does not move the median, which is why the median is the number quoted.

The database is geographically lopsided.  The United States alone holds 22.4% of catalogued collections.  The top ten countries hold 56.9%.  Across 255 country codes the Gini coefficient is 0.861, and apparent diversity tracks collecting effort at a Spearman rank correlation of 0.888.²⁶  So a substantial part of what is being measured is where paleontologists live, not where animals lived.

Some of the clade rows are too small to carry weight on their own.  Pterosauria's 88% is seven genera out of eight.  The flowering-plant row counts only the 55 genera the database files under the class Magnoliopsida.  The sister article, taking every genus in the database's flowering-plant phylum, counts 898 genera and finds a Lazarus gap in 94% of them, which agrees with the 93% here.⁷  Snakes cannot be separated from lizards at the taxonomic rank the dump carries, so that row is both.  Read the clade table as corroboration of the global figure, not as five independent results.

And the measurement is of the database, not of the rocks.  A genus missing from a stage may be missing because nobody has published on rocks of that age in the right place.  That is exactly why the number is a lower bound: a genus with a demonstrated gap certainly existed through it, so the record certainly missed it.  It cannot be turned into an upper bound on how complete the record is.

Corroboration from a hostile source

The best-searched group in vertebrate paleontology is the fish-to-tetrapod transition.  It has had decades of deliberate, funded, targeted prospecting by people who knew exactly what they were looking for.

Long and colleagues concluded in Nature in 2025 that its record "is much less complete than has been thought."⁵

A creationist can quote that without having to argue for it, and it says the same thing the 75.9% says.

The same data limit the excuse

A genus with one hole in its range is what good but imperfect sampling looks like.  It does not show that the record misses whole groups for long stretches.  On that question the arithmetic runs the other way.

The sister article measures, from the same database, how often a genus known from the stages on either side of a stage also turns up inside it.  The rate is 69% for mammals, 66% for reptiles and dinosaurs, 63% for flowering plants and 35% for insects.  At 69%, a mammal genus that really lived through three stages in a row is missed in all three about 2.9% of the time.⁷  Several genera missed together through the same stages is rarer still, unless they all lived in a setting that seldom preserves.²⁷  These rates are measured on genera already known, so they say nothing about a lineage that left no fossil at all.

So a long blank before a whole group's first fossils needs a specific reason, not the general leakiness of the record.  The reasons offered for bats and pterosaurs are specific:  habitats that seldom preserve small animals, and records that hang on a few exceptional deposits.  Parts 4 and 5 give them with the studies behind them.  Michael Foote, who built several of the standard completeness measures, wrote in 1996 that the question had not been settled with numbers.  An estimate of how many major transitions the record ought to show, he said, "would have important implications for whether the small number of documented major transitions provides strong evidence against evolution."²⁸  No source found for this series has made that estimate.

The tree-and-rock comparisons do not settle it either.  Michael Benton and two colleagues compared the order of first fossils with the order of branching in 1,000 published family trees and found "no evidence of diminution of quality backwards in time," while conceding that "Ancient rocks clearly preserve less information."²⁹  The same year Peter Wagner and Christian Sidor showed by simulation that these agreement scores also rise with things unrelated to sampling, among them low extinction rates and "cladogram inaccuracy," which they said compromises "their validity as measures of the quality of the fossil record."³⁰  Neither side should cite the agreement scores as a clean measure of completeness.  What they do show is that first fossils and family trees mostly come in the same order, a pattern any young-earth model has to explain rather than deny.

Punctuated equilibrium, in its authors' words

The best-known mainstream statement about gaps is punctuated equilibrium, and it is usually quoted in pieces.  Here it is whole.

What Niles Eldredge and Stephen Jay Gould proposed in 1972.  If new species arise "very rapidly in small, peripherally isolated local populations," they wrote, "then the great expectation of insensibly graded fossil sequences is a chimera."  And:  "Many breaks in the fossil record are real; they express the way in which evolution occurs, not the fragments of an imperfect record."³¹  Gould put it more sharply in an essay collected in 1980:  "The extreme rarity of transitional forms in the fossil record persists as the trade secret of paleontology.  The evolutionary trees that adorn our textbooks have data only at the tips and nodes of their branches; the rest is inference, however reasonable, not the evidence of fossils."³²

What they did not say.  The theory is about the step from one species to the next.  It says that step happens fast and in a small population away from where most fossils of the parent species are found, so a local rock column shows a break.  It does not say the intermediates never lived.  Gould drew the line himself in 1981:  "it is infuriating to be quoted again and again by creationists—whether through design or stupidity, I do not know—as admitting that the fossil record includes no transitional forms.  Transitional forms are generally lacking at the species level, but they are abundant between larger groups."³³

What a creationist may fairly take from it.  Two leading paleontologists said that the breaks between species in a local column are mostly real, not the work of a poor record.  That is accurate, and it cuts against anyone who answers every gap with "the record is incomplete."  Their explanation for those breaks is speed and place, which leaves the intermediates unrecorded by design of the theory.  A reader may judge whether that makes the theory hard to test.  Michael Foote noted the cost in 1996:  if speciation is that fast, "the chance of capturing transitions, especially major transitions among classes and orders, is substantially diminished."²⁸

Where Gould's own line is too strong.  "Abundant between larger groups" holds for the reptile jaw and mammal ear, which was Gould's own example, and for whales.¹⁴³⁴  It does not yet hold for the two flying groups in this article.  No fossil shows a bat wing or a pterosaur wing in the making (Parts 4 and 5).  Quoting Gould's sentence against those two cases would be the same selective use he objected to, run in the other direction.

What a creationist may not take from it.  "Gould admitted there are no transitional fossils" is false, and Gould said so in print.  Quote the "trade secret" line with Gould's 1981 sentence beside it, or do not quote it.

The creationist case, in its own words

The young-earth argument from gaps has two classic statements, and they disagree with each other.

Duane Gish, 1995.  Gish, of the Institute for Creation Research, wrote the best-known book on the subject.  He stated the creation model's prediction plainly:  "we would predict an explosive appearance in the fossil record of highly complex forms of life without evidence of ancestral forms.  We would predict that all of the major types of life, that is, the basic plant and animal forms, would appear abruptly in the fossil record without evidence of transitional forms linking one basic kind to another."  His examples included bats.  For pterosaurs he named the test:  if they evolved from a non-flying reptile, "we would predict that the fossil record would produce transitional forms showing a gradual increase in length of the fourth finger."  And he shut the door on the excuse Part 3 measures:  "Sampling of the fossil record has now been so thorough that appeals to the imperfections in the record are no longer valid."³⁵

How Gish's prediction has fared.  His two tests in this article's groups still stand.  No fossil shows a bat wing part-built, and none shows a pterosaur's fourth finger part-lengthened (Parts 4 and 5).  That is a prediction made in print that has not yet failed, and a creationist may say so.  But the general form, "without evidence of transitional forms linking one basic kind to another," has failed wherever the record is good:  turtles in Part 1, and the jaw-to-ear, fish-to-tetrapod and whale series in the sister articles.¹⁴²¹³⁴  And his sampling claim is too strong.  The bat record is the least complete of any land-vertebrate group so far measured, and a known mammal genus is re-found in only about two of every three stages in its range.⁹⁷

Kurt Wise, 1995.  Wise, a young-earth paleontologist with a Harvard doctorate, answered Gish's generation from inside the movement.  He called fossils that sit between two groups in both rock position and form "stratomorphic intermediates," listed the early bird series, the tetrapod series and the whale series among them, and wrote:  "Creationists therefore need to accept this fact.  It certainly CANNOT be said that traditional creation theory expected (predicted) any of these fossil finds."³⁶³⁶  His explanation is a Flood model:  life before the Flood lived in zones, the rising water buried the zones in turn, and animals of in-between form lived where two zones met.³⁶  For the remaining cases he suggested that "a few stratomorphic intermediates might be expected in even a random burial process."³⁶

Where Wise's answer is weak.  It explains an order after the fact.  It did not predict any particular fossil, and Wise said so.  Its key quantitative claim is untested:  he expected that "when this is finally quantified," the unexplained cases would fit a random process, and no one has done the count.³⁶  That is the same position as Foote's unmade estimate on the other side.  The decisive number for each side, how many major transitions an evolutionary history should leave visible and how many a burial process would produce by chance, has been proposed by each side and, in every source this series has found, computed by neither.²⁸

What this licenses, and what it does not

The measurement is symmetric, and that is the point of using it.  It is not a rhetorical device that happens to help one side.

Licensed: "the absence of a particular intermediate is weak evidence, because demonstrated non-preservation on this scale is normal."  That is a statement about the record.

Licensed: "gap arguments expire, and here are two worked cases."

Not licensed: "therefore the intermediates existed."  The measurement weakens the argument from absence in both directions.  It supplies no positive evidence for anything.  A creationist who reads it as support for special creation has misread it exactly as badly as an evolutionist who reads it as support for undiscovered ancestors.

Not licensed: using incompleteness to dismiss a fossil that was found.  This is the abuse to watch for, and it is the most common one.  If the record is too poor to prove absence, it is good enough to establish presence when something turns up.  Turtles turned up.  Limbed snakes turned up.  A clawed non-echolocating bat turned up.  A page that pleads incompleteness against the fossils and then treats the remaining gaps as decisive is arguing both sides of the same fact.

Where the burden actually sits

The strongest form of this section is a question, not a defence.  For each origin, ask: what would have to be found, and are the rocks that would hold it being searched?

A gap in a well-searched interval is worth far more than a gap in an unsearched one.  Here is the honest answer for each of the five, and it does not run the way a creationist would prefer.

Only two of those six gaps are in rocks that have been searched hard.  A creationist argument built on the other four is much weaker than it sounds, and no creationist source found for this article makes the distinction at all.

A sixth gap: the insect wing

The origin of insect wings is the largest gap of the kind this article examines, and the specialists concede it.  Insects are the most diverse animals alive.  David Grimaldi and Michael Engel's standard text, Evolution of the Insects, says that "there is, unfortunately, no readily identifiable structure that can easily account for the appearance of wings, and debates over the origin of insects wings have raged for over a century," and that "the fossil record presently lacks conclusive evidence about the evolution of flight."  Winged insects appear already varied in the earliest Late Carboniferous rocks, with several lineages and wing types, and no fossil shows a wing being formed.⁸  The authors called the discovery of Early Carboniferous insects "the most significant problem for insect evolution" and perhaps "the only way" to settle how wings arose.⁸

The one older fossil once read as a winged insect has weakened since.  Rhyniognatha, a fragment of a head from the Early Devonian Rhynie chert of Scotland, was read by Engel and Grimaldi as showing features of winged insects.⁸  Carolin and Joachim Haug imaged it again in 2017.  They concluded that the structures "do not support an interpretation as an insect" and make a centipede more likely, and they describe every Devonian insect find as "problematic and controversial."³⁸

The mainstream reply is that the gap is partly real, not an artifact.  Sandra Schachat and five colleagues argued in 2018 that the blank between the first Devonian insect and the varied insect faunas of the early Late Carboniferous "requires no special explanation if insects were neither diverse nor abundant prior to the evolution of wings."  In their reading, "the fossil record may accurately record the transformative evolutionary impact of insect flight."³⁹  That explains why there are few insects before wings.  It does not supply a wing in the making.

Verdict:  sound, and conceded.  "No fossil shows the origin of the insect wing" meets the same standard as the bat and pterosaur claims, and the field's standard reference says it.  The limit is the same too.  The rocks that would hold the answer are thin, and insects have the lowest re-find rate of any group the sister article measured, 35% per stage.⁷  A creationist may state it as a real absence.  He may not state it as a proof.

The Cambrian phyla, the other gap usually named alongside these, belong to The Cambrian Explosion.  It finds the explosion real, since burrows, which need no special preservation, jump in variety at the base of the Cambrian, but finds that Ediacaran animals and Cambrian stem groups refute the claim that all phyla appeared at once without intermediates.⁴⁰

Part 4 — Bats: a fossil that split two problems apart

The claim, and why it is strong

Bats appear in the Early Eocene already flying.  The oldest ones — Icaronycteris, Archaeonycteris, Palaeochiropteryx, Hassianycteris, Onychonycteris — have complete wings.

There is no known gliding bat.  There is no known half-wing.  There is no pre-Eocene bat body fossil at all.

All of that is still true in 2026 and this article does not pretend otherwise.  Bats are the strongest of the five cases for the creationist side, and the reason is simple: a bat wing is a hand with the fingers grown out to the length of the body and a membrane stretched between them, and nothing shows that hand halfway.

A 2019 survey of the whole bat fossil record says the same, and one of its authors described Onychonycteris.  Emily Brown, Daniel Cashmore, Nancy Simmons and Richard Butler scored 441 fossil bat species.  They wrote that "the lack of any taxonomically useful Paleocene fossils is puzzling," and that "without Paleocene fossils researchers can only theorise as to how and why ancestral bats transitioned to flying locomotion and the unique bat morphology, as the first bats known from complete specimens have all the unique anatomical features of extant forms."⁹

The same paper gives the best reply from the other side.  Most of the bat record is isolated teeth, and bat skeletal completeness is "the lowest of any previously assessed tetrapod group."  The authors suggest that early bats lived in forests, which seldom preserve, and they treat the Palaeocene blank as unexplained, not as explained away.⁹  So the bat gap is not a gap in well-searched rocks of the right kind.  It is a gap the specialists themselves call puzzling, and nothing yet found shows the wing being built.

What changed in 2008

Nancy Simmons, Kevin Seymour, Jörg Habersetzer and Gregg Gunnell described Onychonycteris finneyi from the Green River Formation of Wyoming.⁴¹

The result that matters is a decoupling.  Every previous account had to bundle flight and echolocation together, because every known bat had both.  Onychonycteris has one and not the other:

"Forelimb anatomy indicates that the new bat was capable of powered flight like other Eocene bats, but ear morphology suggests that it lacked their echolocation abilities, supporting a 'flight first' hypothesis for chiropteran evolution."⁴¹

The rest of the description is a list of primitive features in a flying animal.  It has claws on all five fingers, where later bats have two or three and most living bats have one.  Its limb proportions suggest an agile climber that walked on all fours and hung under branches.  Its short broad wingtips suggest an undulating gliding-and-fluttering flight style.  And it has a long calcar — the spur that stiffens the tail membrane — which shows the tail membrane was there early.⁴¹

That is a character-level intermediate.  It also made a testable prediction that still stands: the earliest bats should be climbers.

The echolocation half of that result is contested, and the creationists found the paper

This article's research dossier stated flatly that Onychonycteris flew without echolocating.  That was too confident, and a creationist page is the reason this article knows it.

In February 2010 Nina Veselka and eight co-authors published a different criterion for echolocation in Nature.  They µCT-scanned 26 species and found that in laryngeally echolocating bats the stylohyal bone — part of the throat skeleton — articulates directly with the tympanic bone of the middle ear, and that this distinguishes echolocators from non-echolocators without exception.  Applying it to the fossil:

"…our findings suggest that O. finneyi may have used laryngeal echolocation because its stylohyal bones may have articulated with its tympanic bones."⁴²

Note the double hedge — "may have" twice.  Veselka and colleagues were careful, and they said the finding reopens basic questions about the timing of flight and echolocation.

In August 2010 the four authors of the original description replied in the same journal:

"We disagree with their interpretation of key fossil data and instead argue that O. finneyi was probably not an echolocating bat."⁴³

So the question is genuinely open, and this article now says so where its dossier did not.

And here is the defect, stated precisely.  Creation Ministries International's page on Onychonycteris, by Matthew Cserhati, was published on 13 September 2019.  It cites the February 2010 challenge.  It does not cite the August 2010 reply — in the same journal, by the people who described the fossil, nine years earlier.⁴⁴

That is not fabrication and it is not quote-mining.  It is a citation that stops at the point where the literature became inconvenient.  It is the same failure this article documents on the other side in Part 1, where a 2008 page reporting a real disagreement was never updated after the disagreement resolved.

What the creationist page gets right, and where it stops

The page is a reader-feedback answer.  A reader had read Wikipedia and concluded that Onychonycteris was "more shrew-like."  Cserhati corrects that, and he is right to.  The fossil is a bat.  Simmons and colleagues describe a bat.  Anyone presenting Onychonycteris as a shrew turning into a bat is wrong, and a creationist correcting that error is doing useful work.

Three of the page's other points are also fair.  The undulating gliding-fluttering flight style is not unique to the fossil — living mouse-tailed bats do it, and Simmons and colleagues say so themselves.  The calcar is present in modern bats too.  And the observation that the oldest bat has the most claws, with later bats having fewer, is a real pattern and the page is entitled to call it loss.

Two of its arguments do not work.

"Echolocation is an irreducibly complex structure that is too complicated to evolve.  Therefore, the lack of echolocation cannot be said to be a primitive trait."⁴⁴  This does not follow.  How complex a structure is has nothing to do with whether its absence in the oldest known member of a group is the primitive condition.  The two claims are about different things.

"The calcar (otherwise known as calcaneum)."⁴⁴  These are not the same bone.  The calcaneum is the heel bone.  The calcar is a separate spur arising from it that stiffens the tail membrane, and it is the calcar that Simmons and colleagues report.

And then the page ends here:

"O. finneyi is not a transitional fossil between non-bats and bats.  Such species don't exist."⁴⁴

This is the move the companion article on Tiktaalik retired, and it gets the same ruling here.  That article concluded that "Tiktaalik is just a fish" should be dropped from creationist argument, because it is a description masquerading as a rebuttal: a transitional form between fish and tetrapods will necessarily be classified as one or the other, so saying which one is not an argument about anything.²¹

"Onychonycteris is just another bat" is the same sentence with a different animal.  It is true.  It is also empty.  Simmons and colleagues did not claim it was half a bat.  They claimed it was a bat with a specific combination of characters no other bat has, and that combination decouples two things that had always been bundled.  Answering a character-level result with a taxon-level label concedes the result without engaging it.

This article's ruling:  drop it.  There is a stronger claim available and it is in the next paragraph.

The claim that survives

Onychonycteris resolves the order of two bat characters.  It does not touch the origin of the bat wing.

It is an Early Eocene bat with fully formed, fully functional wings.  The question the creationist objection actually asks — how a running or climbing mammal's hand became a wing — is untouched by it.  No fossil shows an incipient bat wing.  Not one.

So the bat gap narrowed at one joint and did not narrow at the joint under dispute.

Verdict:  partly survives, with the wording fixed.  "There are no fossils showing the origin of the bat wing" is accurate.  Say exactly that, cite Simmons and colleagues directly, and describe what Onychonycteris shows before saying what it does not.  Anyone claiming the fossil answers the bat objection is overselling.  Anyone claiming nothing has changed since 2007 has not read the paper.

Part 5 — Pterosaurs: a gap that got smaller twice, and one that did not

The concession is in the describing papers' own abstracts

This is the case where the mainstream states the problem more bluntly than any creationist does.

Martín Ezcurra and sixteen co-authors, opening a 2020 Nature paper on pterosaur origins:

"…their origin has remained an unresolved enigma in palaeontology since the nineteenth century, and there is still a major morphological gap between those forms and the oldest, unambiguous pterosaurs from the Upper Triassic series."⁴⁵

Davide Foffa and ten co-authors, opening a 2022 Nature paper on the same problem: the origin and early evolution of pterosaurs are "poorly understood owing to a substantial stratigraphic and morphological gap between these reptiles and their closest relatives, Lagerpetidae."⁴⁶

A creationist arguing this case should quote those sentences and stop.  They are stronger than any paraphrase, they come from the people making the opposite case, and they are in the published abstracts where anyone can check them.

What was found

Two results, both from µCT scanning of previously ambiguous material.

2020.  Ezcurra and colleagues recovered lagerpetids — small, running, non-flying dinosaur precursors — as the sister group of pterosaurs, sharing derived features across the whole skeleton.  They report that this "substantially shortens the temporal and morphological gap," and that neuroanatomical features tied to pterosaur sensory abilities are already present in lagerpetids, so those evolved before flight.⁴⁵

2022.  Foffa and colleagues scanned Scleromochlus taylori, a small Scottish reptile described in 1907 whose position had been argued about for a century — some had doubted it was even an archosaur.  They place it inside Pterosauromorpha, the lagerpetid-plus-pterosaur group, and closer to lagerpetids than to pterosaurs.⁴⁶

"Pterosaurs have no known relatives" is dead, and it died in 2020.  Any page still saying it should stop.

How strong is the link, honestly

Strong as a phylogenetic result.  Weak as a transitional series.  Those are different things and the difference is the whole argument.

Nothing found is partly winged.  Lagerpetids run.  Scleromochlus runs.  The pterosaurs fly.  Every step in the transition to flight is still missing, and both describing papers say so about their own subject.

The stratigraphic gap did not close as much as the morphological one.  Lagerpetids are Middle to Late Triassic; the oldest unambiguous pterosaurs are Late Triassic.  The word both papers use about their own gap is "substantial."

The result rests on scans of fragmentary material whose interpretation was disputed for a century.  That is a good result and it is also exactly the kind of result that gets overturned again — as Eunotosaurus just was, on the same technique, in Part 1.

Ezcurra's own 2025 paper pulls against Ezcurra's 2020 paper

This is worth printing, and creationists have not noticed it.

In 2025 Ezcurra and colleagues compared brain shape across pterosaurs, lagerpetids and the dinosaur-to-bird transition.  The 2020 paper's picture was of pterosaur characters assembling gradually, with sensory features arriving before flight.  The 2025 paper finds that pterosaurs share only the ventrolateralisation of the optic lobe with lagerpetids, and concludes:

"…in contrast to the bird-line archosaurs, where exaptation may have played a central role in the stepwise assembly of the avian brain configuration, brain evolution in pterosaurs seems to have unfolded rapidly at the origin of flight."⁴⁷

The same lead author, five years apart, weakened his own gradual-assembly reading for the brain.  This does not overturn the lagerpetid relationship, which rests on the whole skeleton.  It does mean the "characters were already accumulating before flight" story is thinner than the 2020 paper made it look.

What the creationist page got right, and the one thing it could not be checked on

Creation Ministries International published on Scleromochlus in November 2024, and this page is a correct media-hype correction.

The BBC headline had been "Scottish fossil revealed to be pterodactyl ancestor."  The CMI page points out that the paper places Scleromochlus closer to lagerpetids than to pterosaurs, and it cites Foffa et al. 2022 directly rather than through a summary.⁴⁸  That is right, and the page deserves the credit.

Two things about it should be said anyway.

Its quotation is spliced and only half of it could be verified.  The clause about Scleromochlus being more closely related to lagerpetids than to pterosaurs checks out against the published abstract.  The continuation — that the lack of flight-related adaptations means the origin of the pterosaur body plan "remains to be found" — is not in the abstract, and the full text could not be reached to check it: three university repository copies returned HTTP 403 and Europe PMC has no deposit.  This article therefore does not print that clause as a quotation.⁴⁶

It omits the abstract's closing sentence, which is a positive claim: the results "support the hypothesis that the first flying reptiles evolved from tiny, probably facultatively bipedal, cursorial ancestors."⁴⁶  A page correcting somebody else's selective reporting should not select.

Verdict

Largely survives, with the wording fixed.

"There are no fossils showing the origin of the pterosaur wing" is accurate, and it is conceded in the describing papers' own abstracts.  Cite those, not a paraphrase.

"Pterosaurs appear abruptly with no known relatives" is false since 2020 and must be dropped.

What would settle it: an Early or Middle Triassic archosauromorph with an elongated fourth digit or any patagial support.  None exists — and per Part 3, those rocks are poorly searched, which is why this gap is worth less than it looks.

Part 6 — Snakes: the case that mostly does not survive

Najash kills the standing claim

Snakes are the strongest of the five cases for the mainstream, and it is not close.  Limbed snakes are not hypothetical.  Several are known.

Guillermo Garberoglio and eight co-authors µCT-scanned the first three-dimensionally preserved skulls of Najash, a Cretaceous snake from Argentina with hindlimbs and a pelvis, and used them to work out the order in which snake skull bones were lost.  Their conclusion is not that limbs were a brief phase on the way to losing them:

"Morphological and molecular analyses including the new cranial data provide robust support for an extensive basal radiation of early snakes with hindlimbs and pelves, demonstrating that this intermediate morphology was not merely a transient phase between limbed and limbless body plans."⁴⁹

So the claim "snakes appear in the fossil record fully formed, with no limbed intermediates" is false, and Najash alone is enough to make it false.  Pachyrhachis, Haasiophis and Eupodophis are further limbed snakes, though whether they are stem snakes or derived burrowers has been argued since the 1990s and this article did not audit them.

The same paper states the honest limit in its own abstract: understanding of early snake evolution "is hampered by the very limited fossil record of early snakes."⁴⁹  That is the real claim, and it is a claim about sampling.

Tetrapodophis: three separate problems, kept separate

In 2015 David Martill, Helmut Tischlinger and Nicholas Longrich described Tetrapodophis amplectus in Science as a four-legged snake.  Science's own editors' summary said the fossil "shows clearly the early transitional stages from a lizardlike body plan to the smooth legless snakes."⁵⁰

It went wrong in three different ways, and conflating them is how a fair criticism turns into an unfair one.

One:  the identification was overturned.  In 2021 Michael Caldwell, Robert Reisz, Randall Nydam, Alessandro Palci and Tiago Simões published a 60-page reassessment concluding it is a dolichosaurid lizard, not a snake.⁵¹  An independent 2018 analysis by Paparella and colleagues, describing the Italian dolichosaur Primitivus manduriensis, had already recovered Tetrapodophis outside snakes under parsimony.⁵²  Two separate analyses put it outside Ophidia.

Two:  the provenance cannot be checked.  The specimen has no recorded locality and no recorded collector.  Its attribution to the Crato Formation of Brazil rests on comparison of the rock matrix.  Brazil has banned fossil export since 1942.  The 2021 team argued for repatriation.

Three:  there is a claim about the holotype's physical condition, and this article will not print it.  Reports say the specimen was compromised during study and that access was subsequently restricted.  Caldwell et al. 2021 is paywalled and could not be opened for this article, and the claim traces through press coverage.  It is left out, and it is named here so a reader knows it was left out on purpose.⁵¹

How to use this, and how not to.

Not as "evolutionists faked a fossil."  Nobody alleges fabrication.  The 2015 authors described what was in front of them.

Use it as the cleanest available case of a spectacular claim published from a specimen whose provenance could not be verified, and reversed within six years by direct re-examination.  The reversal came from paleontologists, in a paleontology journal, using the counterpart slab the original team had not studied closely.  That is the point worth making, and it is a point about how the field corrects itself.

The creationist page on this is better than most secular coverage

This article expected to find a page alleging fraud.  It did not, and the correction is worth printing at length.

Philip Bell's creation.com article on Tetrapodophis was published on 29 July 2015, five days after the Science paper.  It states that he read the original paper.  It reports the identification dispute as it stood — including that Michael Caldwell, who would go on to lead the 2021 reassessment, already doubted it was a snake in 2015.  It quotes Martin Cohn's enthusiastic comparison to Archaeopteryx as well as the doubts.  And it was updated twice, in December 2016 and November 2021, the second update reporting the Caldwell reassessment and its conclusion accurately.⁵³

It also does something this article did not expect.  It tells creationist readers what not to claim:

"If you're a creationist tempted to think that Tetrapodophis represents something out of Genesis 3, don't go there."⁵³

That is quoted from Creation Evolution Headlines, which the page endorses.  A creationist page telling its own audience to stop before an attractive but unsupported inference is doing the thing this site exists to do.

Two small defects, recorded for accuracy.  The 2021 update says the fossil is "housed in a private collection in Brazil," while the 2015 article on the same page says a German museum held it for a private owner.  This article could not establish the current location and does not state one.  And the reference list misspells the lead author as "Marthill."

Neither is the kind of error that changes anything.  The page is a model of maintenance and should be cited as one.

Verdict

Largely does not survive.

"Snakes appear fully formed with no limbed intermediates" is false, and the 2019 Najash paper's conclusion is that limbed snakes were a radiation, not a transient stage.

What survives is narrower: which lizard group snakes came from, and in what environment — marine or burrowing — remains disputed, the early snake record is thin by its own workers' account, and the fossil most publicised as settling it turned out to be a lizard.

Part 7 — Flowering plants, and Darwin's "abominable mystery"

This part corrects how the quotation is usually printed.  It also withdraws a ruling this article first made against the creationist reading of it.

Get the quotation right

Charles Darwin wrote to his friend Joseph Hooker on 22 July 1879.  The sentence, verbatim, from the Darwin Correspondence Project's transcription of the letter:

"The rapid development, as far as we can judge, of all the higher plants within recent geological times is an abominable mystery."⁵⁴

Two things popular versions often get wrong.

It is a private letter, not a publication.  Darwin never printed the phrase.  It first reached print in 1903, in More Letters of Charles Darwin, after his death.⁵⁵

The qualifiers are usually dropped.  "As far as we can judge" and "within recent geological times" are in the original.

What the mystery was about.  This article first ruled that the sentence is about the rate of diversification and not about origin.  Darwin's other letters on the same subject do not allow that split.

1. 8 March 1875, to the Swiss palaeobotanist Oswald Heer: "The sudden appearance of so many Dicotyledons in the Upper Chalk appears to me a most perplexing phenomenon to all who believe in any form of Evolution, & especially to those who believe in extremely gradual Evolution." Then: "But I fully admit that this case is a great difficulty in the views which I hold."¹⁰
2. 22 July 1879, the "abominable mystery" letter itself. The letter answers John Ball's essay on alpine plants, which proposed that the higher plants could at first live only at high altitudes. Darwin calls that idea "far too rash." A few sentences later he offers his own way out: "I have fancied that perhaps there was during long ages a small isolated continent in the S. hemisphere, which served as the birth place of the higher plants; but this is a wretchedly poor conjecture."⁵⁴ A birthplace is an origin.
3. 12 August 1881, to Hooker again: "I have been so astonished at the apparently sudden coming in of the higher phanerogams, that I have sometimes fancied that development might have slowly gone on for an immense period in some isolated continent or large island, perhaps near the S. Pole."⁵⁶

So Darwin's problem was a large group that enters the fossil record suddenly and already diverse, with no visible ancestors.  His answer was that the ancestors had lived for a long time somewhere nobody had yet looked.  That is a missing-transition problem, and Darwin said it counted against his own view.

Two botanists have studied what Darwin meant, and neither supports a rate-only reading.  William Friedman concluded in 2009 that "Darwin was deeply bothered by what he perceived to be an abrupt origin and highly accelerated rate of diversification of flowering plants in the mid-Cretaceous."  For Friedman the heart of it was Darwin's "abhorrence that evolution could be both rapid and potentially even saltational," that is, by leaps.⁵⁷  Richard Buggs argued in 2021 that Darwin "was referring to the origin only of a subset of what are today called angiosperms," the dicotyledons, whose record "began abruptly and with great diversity in the Cretaceous."  Buggs adds that the mystery may have been "abominable" to Darwin because William Carruthers, Keeper of Botany at the British Museum, read it "as evidence for divine intervention in the history of life."⁵⁵

Buggs's narrower point is the one correction a careful writer should still make.  In 1879 "the higher plants" meant the dicotyledons.  The monocotyledons were then thought to have an older record, so "all flowering plants" stretches Darwin slightly.

Verdict on the quotation as it is usually deployed:  right in substance, often stripped of its date and qualifiers.  "Darwin called the origin of flowering plants an abominable mystery" puts the word "origin" in his mouth where he wrote "development."  It does not change his meaning.  Print the whole sentence with the letter's date, and the Heer letter beside it, which is the stronger admission.

What the creationist organisations actually did with it

Both pages were read for this article.  Both read Darwin fairly.

David Catchpoole, writing for Creation Ministries International in December 2022, prints the sentence correctly, with the commas dropped but the words right, and cites the Darwin Correspondence Project itself.⁵⁸  The sourcing is good.

On the same page he glosses the sentence with the BBC's framing:  "The famous naturalist was haunted by the question of how the first flowering plants evolved."⁵⁸  This article first held up that gloss as the cleanest example of a slide from rate to origin.  That ruling is withdrawn.  Darwin's 1881 letter, which wonders whether "development might have slowly gone on for an immense period in some isolated continent," is exactly a man asking how the first flowering plants evolved.⁵⁶

The Institute for Creation Research, May 2018, frames it the same way Friedman does.  Jeffrey Tomkins and Tim Clarey print the same sentence from More Letters of Charles Darwin and frame the problem as "why these angiosperms appeared so suddenly and so widely diversified so late in the fossil record."⁵⁹  Sudden appearance plus rapid diversification is Darwin's meaning.  They should get the credit.

Where the creationist use is weak is not the quotation but the inference.  Darwin's admission describes the record as it stood in 1875.  It is evidence of what he saw, not of what the rocks show now, which is taken up under "The current state of play" below.

A separate error in the ICR page, checkable against a public-domain source

The same ICR page opens with a different claim, and this one does not survive.

"Charles Darwin frankly acknowledged that the profound lack of transitional forms in the fossil record for one fundamental type of creature evolving into another was 'a valid argument' against his idea of progressive gradualistic evolution over deep time."⁵⁹

The reference given is On the Origin of Species, 1859, page 308.

The page number is right.  The characterisation is wrong.  The sentence on page 308 reads:

"The case at present must remain inexplicable; and may be truly urged as a valid argument against the views here entertained."⁶⁰

"The case" is not transitional forms.  The section heading two pages earlier is "On the sudden appearance of groups of Allied Species in the lowest known fossiliferous strata."  What Darwin is conceding is that no fossiliferous formation is known below the Silurian, despite the vast stretch of time his theory required to have already passed.  That is the missing Precambrian record, which is a different problem — and, as it happens, a much better one for a creationist to raise, since it is still a live issue.

This article verified the passage against the 1859 first edition and cross-checked its page position against Darwin Online's page-tagged transcription.  A creationist page misdescribing a Darwin passage that supports a real creationist point is a self-inflicted wound, and it is exactly the sort of thing this series exists to stop.

The current state of play, and why this article stops there

The gap is no longer "no flowering plants before the Cretaceous."  It is a disagreement between two methods about how far back they go.

Mario Coiro, James Doyle and Jason Hilton argued in 2019 that the pollen record documents angiosperms appearing in the Valanginian and spreading from there, that this agrees with molecular inferences about the course of pollen evolution but "conflicts strongly with Triassic and early Jurassic molecular ages," and — the sentence a creationist article must quote fairly — that on scrutiny "supposed pre-Cretaceous angiosperms either represent other plant groups or lack features that might confidently assign them to the angiosperms."⁶¹

Daniele Silvestro and six co-authors argued the opposite in 2021, from about 15,000 fossil occurrences in 198 families plus living diversity, reporting that several families originated in the Jurassic, "strongly rejecting a Cretaceous origin for the group."⁶²

Both are working from fossils.  Both are mainstream.  They disagree by roughly 50 to 100 million years.

This article goes no further, and the reason is a scope line rather than a dodge.  Everything live in that dispute is dating work — molecular clocks against fossil calibrations — and this series does not use absolute dating as evidence in either direction.  Coiro and Silvestro are reported here as the mainstream figures under discussion, not as anybody's evidence.

What the series can say from relative order is narrow: the first unambiguous angiosperm pollen occurs below the first unambiguous angiosperm macrofossils, and the Paleobiology Database's apparent Palaeozoic angiosperms drop from 85 records to zero once form taxa are filtered out.⁶³

The claimed Precambrian pollen from the Roraima Formation, which the CMI page raises, is a claim about fossils out of order rather than about missing transitions, and this article does not assess it.  Part 3 of Claimed Fossil Anomalies does, and finds it unresolved but leaning strongly to contamination.⁶⁴

The honest verdict

"Abominable mystery" is a 147-year-old private remark about the sudden appearance of the flowering plants, which Darwin admitted was "a great difficulty in the views which I hold."¹⁰

What survives:  the abruptness Darwin noticed is still in the pollen record, and Silvestro and colleagues, arguing the opposite case on origin timing, still report the fast Cretaceous radiation as a real signal.  Darwin's own way out, a long hidden history somewhere unsampled, is still the evolutionary answer, and it is still unobserved:  on the fossil side's own reading, every supposed pre-Cretaceous angiosperm so far has turned out to be another plant group or to lack the features that would assign it to the flowering plants.⁶¹

What does not survive:  using the quotation to suggest that flowering-plant origins are an unexplained blank on which nobody has data.  Two teams disagreeing with each other by 50 million years, each with data, is an active dispute.  And a remark from 1879 cannot settle what the rocks show in 2026.

Part 8 — The claims, one at a time

Blunt where blunt is warranted.  Several of these were true when written and are false now, and this article's credibility depends on saying so before a critic does.  Each attribution to a creationist web page below is to a page fetched and read on 3 August 2026, not to a summary.

What would change this verdict

Stated as conditions, so a reader can hold this article to them.

• Any fossil showing an incipient bat wing, pterosaur wing or insect wing. Each would close one of the three cases this article says still stand. None exists.
• The two unmade estimates. How many major transitions an evolutionary history should leave visible, given measured completeness (Foote's proposal), and how many stratomorphic intermediates a burial process would produce by chance (Wise's). Either number would turn this article's judgements about absence into tests.
• A latest-Permian or Early Triassic archosauromorph with a shell-related character, which would close the base of the turtle sequence that Jenkins et al. reopened. Those rocks are well searched, so continued absence there would mean more than absence elsewhere.
• A published rebuttal of Jenkins et al. 2026, in either direction. A ten-week-old result with one citing work is not settled, and this article's Part 1 would change shape if it fell.
• Access to Caldwell et al. 2021, which would settle what the paper actually says about the Tetrapodophis holotype's condition. Until then this article prints nothing about it.
• A re-run of the gap statistic with a preservation filter, so that trace and form taxa can be excluded and the long-gap tail cleaned up. The Paleobiology Database bulk CSV does not currently carry the flag.
• A demonstration that Pappochelys is misassigned, the way Eunotosaurus just was. It is known from one locality, and the specimen count could not be established for this article.
• A creationist source that applies the "is the right rock being searched?" test. None found for this article does. It is the single most useful thing our side could add to this argument, and it would make some of our own claims weaker and the rest much stronger.

Sources

The research dossier behind this article — the full evidence by sub-topic, thirty-three quotations with verdicts, the creationist-source audit, and the list of sources that could not be reached — is published alongside it: research/missing-transitions.md.  The clade gap script and its output are at research/clade-gaps.py and research/clade-gaps.txt.

Three quotations that appear in the literature on this subject are deliberately not printed here.

The sentence Creation Ministries International attributes to Foffa et al. 2022 about the pterosaur body plan "remaining to be found" is not in the published abstract, and the full text could not be reached to check it — three repository copies returned HTTP 403 and Europe PMC has no deposit.  The sentence the Institute for Creation Research attributes to Richard Buggs's 2017 Nature Ecology & Evolution comment could not be verified, because the comment is closed access and no repository copy exists.  And the claim about the physical condition of the Tetrapodophis holotype rests on Caldwell et al. 2021, which is paywalled.  All three may well be accurate.  None of them was opened, so none is printed as a quotation.

What was read in the original, and what was not.  The Darwin–Hooker letter, the 1859 Origin, the Natural History Museum release, Joyce et al. 2009, and nine creationist pages were read in full.  For the remaining journal papers only the publisher's abstract could be reached — through PubMed E-utilities or the Crossref abstract field — and every footnote below says so where it applies.  Jenkins et al. 2026 and Caldwell et al. 2021 could not be obtained at all.