Archaeopteryx and the Origin of Birds

"Distance correlation using Senter's set of taxa and characters supports his conclusion of morphological continuity." — Todd Wood, a young-earth creationist, writing in the Journal of Evolutionary Biology in 2011 about the dinosaur-to-bird sequence.¹

"Modern birds and dinosaurs fall into separate created kinds… However, some extinct feathered animals (such as the famous Archaeopteryx) may be kinds of dinosaurs." — the plain-language summary of a creationist baraminology study by Paul Garner, Todd Wood and Marcus Ross.²

"Archaeopteryx is half reptile and half bird any way you cut the deck, and so it is a Rosetta stone for evolution." — Alan Feduccia, the ornithologist most often quoted by creationists against the dinosaur origin of birds, answering a direct question about creationists in 2003.³

Summary

Archaeopteryx is the most famous fossil in the world.  It has teeth, a long bony tail and clawed fingers, like a small dinosaur.  It also has a wishbone and flight feathers, like a bird.  For a hundred and sixty years creationists have argued that it is "just a bird," and evolutionists have argued that it is the first bird.

This article says both sides should stop, because the people on both sides who actually did the arithmetic cannot say which it is.

Creationists have a quantitative method for finding where one created kind ends and another begins.  It uses baraminic distance correlation and multidimensional scaling on published character matrices, looking for clusters separated by gaps.  Between 2010 and 2018 that method was run on this group five times, twice by an evolutionist and three times by creationists, and three of the five runs were published in the Journal of Evolutionary Biology.

The result is not what either side expects.  On the creationists' own statistic, Archaeopteryx sits on the border.  Todd Wood, a young-earth creationist with a biochemistry PhD, replied to Phil Senter in the same journal and conceded that on Senter's data the continuity is real.¹  Paul Garner, Wood and Marcus Ross then ran six more matrices.  They found gaps between most birds and the dromaeosaurs and troodontids, the "raptors."  But Archaeopteryx fell on the dinosaur side in four of the six, and their own summary says it "may be a kind of dinosaur."²⁴  Matthew McLain and two colleagues ran five newer matrices in 2018 and wrote that they "still cannot say with certainty" which side it belongs on.⁵

So this article does something unusual for a young-earth site.  It concedes that Archaeopteryx cannot be shown to be "just a bird," and moves the argument to the gaps the creationist statistics actually found.

What this article concludes

• "Archaeopteryx is just a bird" should be retired. Creationist statistical baraminology does not support it. The creationists who ran the numbers put Archaeopteryx with the dinosaurs in some matrices and with the birds in others, and say they cannot tell which.⁴⁵
• The gaps the creationist method found are not where creationists usually argue. Garner, Wood and Ross found them between most birds and the raptors, and inside the Mesozoic birds: the line that leads to modern birds separates from the older groups.⁴ McLain's 2018 study found the first gap again on five newer matrices.⁵ McLain's team also says plainly that there is no "enormous morphological gulf" between the groups and that the creationist case from gaps is "not yet conclusive."⁵⁵ None of these gaps sits at the end of the Cretaceous. Part 10 argues that the bird turnover at that boundary is the strongest unworked line in the subject, on superposition alone.
• Four standard creationist objections are dead, and this article gives the dates and the names for each. The temporal paradox, as it was framed, died between 2005 and 2018. The avian-lung objection died when unidirectional airflow was measured in alligators, monitor lizards and iguanas. The "feathered dinosaurs are just collagen" claim lost on the weight of several lines of evidence, and the largest creationist study of the group now calls many dinosaurs "indisputably feathered."⁵ The digit-homology objection died when the frame shift was induced experimentally in a chick wing bud.
• The temporal paradox was a fair kind of argument, and it failed on its evidence. It matches fossil order between continents: the Solnhofen limestone of Bavaria against the Cretaceous beds that hold Deinonychus and Velociraptor. That matching is open to a young-earth writer, and the young-earth paleontologist Kurt Wise and his co-author Donna Richardson call the global biostratigraphic column real.⁶ But fossil order puts a paravian of the Anchiornis group at Archaeopteryx's own level in Bavaria, which is enough to refute the paradox as it was framed. The further claim, that nine feathered paravians from China lie below that level, leans partly on radiometric dates, and a young-earth reader may discount that part. The one writer who cannot use the paradox at all is the one who calls the column circular, because the paradox is the column's output. Part 4 sets this out.
• Three things do survive, and they are the article's real content. Archaeopteryx's position on the tree is separated from the alternative by one parsimony step.⁷ A 2025 review in a Royal Society journal says a fully avian lung should not be reconstructed in non-avian dinosaurs.⁸ And after twenty-seven years the field still cannot agree which fingers a chicken has — the 2019 paper that cut the frame-shift hypothesis down to one digit was co-authored by the man who proposed it.⁹
• Two men our side routinely misrepresents get corrected here, in our own voice. Storrs Olson wrote his own account of the Archaeoraptor forgery and credited Xu Xing — a leading proponent of the theropod hypothesis — with catching it.¹⁰ Alan Feduccia calls Archaeopteryx a Rosetta stone for evolution.³ Neither man says what he is used to say.
• Archaeoraptor was never peer-reviewed. It was rejected before publication, named by a magazine staff writer, and exposed within four months by a paleontologist. A creationist who runs it as evidence of institutional fraud has picked the one case where the institution worked.

What this article rests on

Every quotation here was checked against a source that was actually opened.  Four quotations that circulate widely on creationist sites could not be traced past those sites, and their wording is not printed anywhere in this article:  John Ruben's diaphragmatic-hernia sentence, the much-repeated line from Storrs Olson's open letter of 1 November 1999, a one-word jibe attributed to Alan Feduccia in a 1993 Science news piece, and the text of the 2013 Nature retraction of the Triassic bird tracks.¹¹  In each case a verified substitute from the same man or the same event is used instead.  A reader who has seen those four quotations on an apologetics site should notice that they are missing, and should ask why.

Radiometric dating is not used as evidence anywhere in this article.  That is a constraint across this whole series:  ages in years and durations belong to the dating articles.  Every argument here works from relative order, meaning what sits above what in one section and which fossils match between distant sections.  Young-earth geology accepts both.  Where an age in years appears, it is the mainstream figure under discussion, not something a reader is being asked to believe.  Part 4 restates the temporal paradox in that form before judging it.

The stratigraphic positions were pulled from the Paleobiology Database rather than copied from a summary.  Every age bracket in Part 4 came from a live query on 3 August 2026, and the query is printed with it.¹²

The rank question, stated up front, with its own caveat

A fossil series only counts against young-earth biology if it crosses a boundary that young-earth biology draws.  Creationists do not deny common ancestry.  They deny it above a certain level, and their own published rule puts that level at roughly the family.

That is not a caricature.  Jean Lightner's An Initial Estimate of Avian Ark Kinds, the study behind the Ark Encounter's bird list, states the working rule directly:  the cognitum at the family level "is preferred when hybrid data is lacking."¹³  On that rule she identifies 196 bird kinds covering all ten thousand-odd living species.

Dinosaur to bird is a claim far above the family level, and that is what makes it worth arguing about.  The horse series sits inside one family.  The named australopith-to-Homo candidates sit inside one family.  A creationist can accept every fossil in those sequences and lose nothing, because within-family descent is already his own position.  This one is different.

Now the caveat, because a well-read skeptic will raise it and he is right.  On the old Linnaean scheme this transition runs from class Reptilia to class Aves, which sounds decisive.  But professional systematics has been rank-free since Jacques Gauthier's work in 1986.  Aves is not a "class" in current usage.  It is a clade inside Theropoda, inside Dinosauria, inside Archosauria.  Under that convention, saying "birds crossed a class boundary" is empty — birds are dinosaurs by definition, and no boundary was crossed because none was drawn.

So this article does not let the word "class" carry the weight.  It uses two things that do not depend on a naming convention.

Morphological distance.  A live Paleobiology Database query on 3 August 2026 returns 971 valid non-avian theropod genera in 138 families, and 1,884 valid avian genera in 341 families in the fossil record.¹⁴  The IOC World Bird List v15.2 counts 11,227 living bird species in 251 families.¹⁵  The claimed transition runs from somewhere inside the first block to the base of the second.

Functional distance.  Powered flapping flight.  A fused pygostyle in place of a long bony tail.  A keeled sternum.  A carpometacarpus.  A toothless keratin beak.  Lungs ventilated by air sacs.  An avian brain and inner ear.  Loss of the forelimb claw.  No proposed within-family sequence anywhere asks a creationist to accept a package that size.

That is the honest form of the rank claim, and it is the form this article defends.

Part 1 — The agreement nobody mentions

This is the most important section in the article, and almost nobody on either side has noticed it.

The method is the creationists' own

Statistical baraminology is a creationist research programme.  Its aim is to find, in a matrix of anatomical characters, the boundaries of the created kinds — clusters of organisms that are continuous with each other and discontinuous with everything else.  Two tools do most of the work.  Baraminic distance correlation (BDC) asks whether two taxa are more similar to each other than either is to a third.  Classic multidimensional scaling (CMDS) projects the distance matrix into two or three dimensions so that clusters and gaps can be seen.

The method has a real virtue for an argument like this one.  It is not a rhetorical device.  It takes a published character matrix as input, produces a number, and can be run by anybody on anybody's data.  Both sides have now done exactly that on the dinosaur-to-bird matrices, five times.

Run 1: Senter 2010

Phil Senter, an evolutionary paleontologist, applied CMDS to a coelurosaurian dinosaur matrix and published the result in the Journal of Evolutionary Biology.  His title says what he was doing:  "Using creation science to demonstrate evolution."¹⁶

From his abstract, verbatim:  the analysis "reveals morphological continuity between Archaeopteryx, other early birds, and a wide range of nonavian coelurosaurs."¹⁶  His stated conclusion is that creationists who use CMDS have to accept that these animals are genetically related, because that is what their own criterion says.

Run 2: Wood 2011, and the concession

Todd Wood is a young-earth creationist and one of the two people who built the statistical-baraminology toolkit.  He replied in the same journal.¹

Wood's reply is more interesting than either side's summary of it.  He re-ran the question using baraminic distance correlation rather than raw CMDS, and reported this, verbatim:  "Distance correlation using Senter's set of taxa and characters supports his conclusion of morphological continuity, but other sets of taxa with more characters do not."¹

Read the first half of that sentence.  Wood conceded that on Senter's own data the continuity is there.  His defence is that richer matrices break it, which is a methodological objection, not a denial of the result.  But his larger matrix, 42 taxa scored on 187 characters, still put the birds and the raptors in one group:  "three well-marked groups are evident, the Oviraptorosauria, the Avialae + Deinonychosauria and the remaining coelurosaurs and outgroups."¹  Only when he cut it down, first to the paravians and then to the birds plus the oviraptorosaurs, did he find negative correlations that "would be interpreted by a baraminologist as evidence of a phylogenetic discontinuity between birds and nonavian dinosaurs."¹  In the paravian subset Archaeopteryx correlated with nothing, either way.  And he closed against his own side, writing that future fossil discoveries may provide stronger support for morphological continuity between dinosaurs and birds.¹

That is what an honest researcher looks like when the data go the wrong way, and it deserves to be reported at full strength by a site that shares his conclusions about Genesis.

Run 3: Senter 2011

Senter then ran the broader question across all of Dinosauria, using taxon correlation.  His abstract states the headline result verbatim:  "Morphological gaps in the known fossil record separate only seven groups from the rest of Dinosauria."¹⁷

That sentence cuts both ways, and both cuts belong here.

In our favour:  Senter found gaps.  Seven of them, by his own criterion, in a dataset chosen to look for continuity.  A creationist may legitimately say that even a hostile analysis using the creationist statistic returns discontinuities inside Dinosauria.

Against us:  the first group on his list is "Therizinosauroidea + Oviraptorosauria + Paraves."  That single group contains Archaeopteryx, Velociraptor, the troodontids and the oviraptorosaurs.  They are all on the same side of the gap.  By the creationist's own criterion, that makes them one created kind.

Run 4: Garner, Wood and Ross 2013

The most thorough creationist treatment of the early birds is Paul Garner, Todd Wood and Marcus Ross, "Baraminological analysis of Jurassic and Cretaceous Avialae," presented at the Seventh International Conference on Creationism.⁴  They applied BDC and three-dimensional MDS to six published character matrices:  Chiappe 2001, Clarke and Norell 2002, Zhou et al. 2008, O'Connor et al. 2009, Senter 2011 and Xu et al. 2011.

Their abstract states the result both ways.  For their side:  "The results reveal discontinuities between most Avialae and the Deinonychosauria (Troodontidae + Dromaeosauridae) and little evidence of continuity between modern birds and dinosaurs, suggesting that recent claims that statistical baraminology supports the evolution of birds from dinosaurs are misplaced."  Against it:  "Nevertheless, we did find positive BDC and MDS clustering of some Avialae and deinonychosaurs in four of our analyses."⁴

The matrix-by-matrix results do not all point one way.⁴

1. Three matrices split the line leading to modern birds from everything older. In Clarke and Norell, Zhou and O'Connor, the Ornithurae (or Ornithuromorpha) form one group, and the older birds sit with the dromaeosaur outgroup. O'Connor's matrix splits three ways: Ornithuromorpha, Enantiornithes, and the other birds plus Dromaeosauridae. Archaeopteryx sits next to the dromaeosaur in all three.
2. One matrix with real raptor genera puts Archaeopteryx with the raptors. In Xu's 2011 matrix, Archaeopteryx and Wellnhoferia cluster with the dromaeosaurs and troodontids, apart from six other birds.
3. Two matrices put it with the birds. In Chiappe's, which has no dromaeosaur in the analysed subset, Archaeopteryx groups with Rahonavis, Changchengornis and Confuciusornis. In Senter's 2011 matrix, "Avialan taxa are only negatively correlated with dromaeosaurid and troodontid taxa."

Then the authors raise a caution against their own dinosaur result for Archaeopteryx.  In the three matrices where it sits with the dromaeosaurs, Dromaeosauridae is entered as a single composite "taxon" while every other entry is a real genus, and "using a composite 'taxon' within the ingroup may have skewed our results, at least in the case of Archaeopteryx."⁴  Their overall reading is that the affinities are real in places but "a closer examination reveals robust evidence of discontinuity."⁴

Their project page puts it in one line for a general reader:  "Modern birds and dinosaurs fall into separate created kinds, consistent with the claims of the Bible.  However, some extinct feathered animals (such as the famous Archaeopteryx) may be kinds of dinosaurs."²

Two things in this paper are easy to misread.  First, its main gap is not between the Mesozoic birds and modern birds.  It is between most birds and the raptors, with further gaps inside the Mesozoic birds.  The ornithurine cluster that holds the line to modern birds is full of Cretaceous animals:  Yanornis, Yixianornis and Hongshanornis in Zhou's matrix.  In Chiappe's matrix the living duck Anas correlates positively with the Cretaceous Ichthyornis and Patagopteryx.  Second, the paper does not concede that Archaeopteryx is a dinosaur.  It reports that result in four of six matrices and then gives a reason to doubt three of the four.

Run 5: McLain, Petrone and Speights 2018

Five years later Matthew McLain of The Master's University, with Matt Petrone and Matthew Speights, ran BDC and MDS on five newer coelurosaur matrices and many subsets of them, for the Eighth International Conference on Creationism.⁵  It is the largest creationist analysis of this group so far.

On birds:  "we consistently found evidence of discontinuity separating avialans from other non-avian theropods."⁵  On Archaeopteryx:  "At this point, we still cannot say with certainty whether Archaeopteryx is continuous with some avialan taxa or dromaeosaurids."⁵

They also say three things a creationist reader may not expect.  Feathered dinosaurs are real:  "many species of dinosaurs were indisputably feathered."⁵  The gaps are small:  their abstract speaks of created kinds of dinosaurs, early birds and modern birds "without an enormous morphological gulf between these groups."⁵  And no single line divides the two:  "we could not find a way to separate theropod dinosaurs and birds overall into two groups based on their anatomy."⁵  Their answer is many created kinds on each side of the old bird-dinosaur line, not one bird kind against one dinosaur kind.

What follows

Line the five runs up and the picture is clear, though it is not the picture either side's popular literature paints.

No run puts Archaeopteryx securely among the birds.  The evolutionist runs put it in a continuum with the raptors.  The creationist runs put it with the raptors in some matrices and with the birds in others, and the most recent one cannot decide.  "Archaeopteryx is just a bird" is not the result of anyone's arithmetic.

The creationist runs do find gaps, and an honest evolutionist should cite them.  Wood in his subsets, Garner and McLain all report negative correlations between birds and the dromaeosaurs and troodontids, and Garner finds more gaps inside the Mesozoic birds.  That is a real result from a stated method.  A writer who cites Senter 2010 as proof that creationist statistics confirm bird evolution, and leaves these out, is telling half the story.

The skeptic's reply to those gaps is also fair.  A gap in a character matrix is a gap in a sample of fossils.  Wood made the point himself in his reply to Senter:  "Even Darwin admitted that morphological gaps exist, but he argued that they could be explained by extinction and the imperfection of the fossil record.  Indeed, as Senter (2010) has ably demonstrated, gaps in the fossil record connecting nonavian coelurosaurs and Mesozoic birds that were evident early in the twentieth century have been filled in by more recent fossil discoveries."¹  McLain's team agrees that "creationist arguments from discontinuity are not yet conclusive."⁵

So the argument a creationist should be making is not that Archaeopteryx is a bird.  It is that three creationist studies, on more than ten published matrices, keep finding gaps between most birds and the raptors and inside the Mesozoic birds, with Archaeopteryx on the border and the gaps small.  Part 10 takes up a separate question, the bird turnover at the end of the Cretaceous, which rests on superposition rather than on these statistics.

One caution before leaving this section, because it runs against the argument just made.  Baraminic distance correlation is a creationist technique with a small literature and no external validation.  Its output depends heavily on which characters are in the matrix, which is exactly Wood's complaint about Senter, and Garner's own caution shows that the coding of one outgroup can move Archaeopteryx across the line.  Nothing in this article requires the method to be correct.  The point is narrower and harder to escape:  whatever the method is worth, both sides ran it, and neither side's runs put Archaeopteryx securely among the birds.  A creationist who rejects the method has to stop citing it when it favours him, and an evolutionist who accepts it has to accept the gaps it finds.

Part 2 — The specimens, and what they will and will not bear

How many are there?

There is no agreed number, and the disagreement is real rather than sloppy.  Three counting conventions were in print in 2025 alone.

O'Connor and colleagues, describing the Chicago specimen in Nature, open by calling it "the nearly complete and uncrushed 14th specimen of Archaeopteryx."¹⁸  Foth and colleagues, describing the Mühlheim specimen in Fossil Record, also call theirs the 14th — but their "overall" counts Urvögel, which their paper defines as any paravian from the Late Jurassic Franconian Alb, so it sweeps in Ostromia and Alcmonavis, which are not Archaeopteryx.¹⁹  Rauhut, Foth and Tischlinger in 2018 called the Schamhaupten specimen the 12th, numbering the London skeleton as 1 — while the convention used by museum signage numbers the isolated 1861 feather as specimen 1, shifting everything after it by one.²⁰

The defensible statement is this.  Fourteen skeletal specimens have at some time been referred to Archaeopteryx, one of which has since been removed from the genus, plus one isolated feather whose identity is itself now disputed.

Foth and colleagues concede the disorder in print, which is worth citing because it is a mainstream team saying it about itself:  "a lack of consensus abounds within the specialist community, even amongst the authors of this paper."¹⁹  Two of their own co-authors think the German Urvögel are one growth series of a single genus, with the differences down to preservation.  The rest do not.  They close by saying a full taxonomic study of all the specimens is still needed.

The list

Numbering follows Rauhut et al. 2018, skeletal specimens only, London = 1, extended by the two described since.  Repository numbers are from the primary descriptions.²⁰¹⁹¹⁸

Two facts about that table matter more than any single row.

One specimen is physically lost.  The Maxberg specimen has not been seen since about 1991, when its owner died.  It survives only as Heller's 1959 description, photographs and casts, and Rauhut and colleagues can classify it no more precisely than "cf. Archaeopteryx sp."²⁰

Several are in private hands.  The 11th and 12th specimens are privately owned.  The Chicago specimen spent thirty years in a private Swiss collection before the Field Museum bought it in 2022.  The Daiting specimen passed through the commercial market.  Germany banned the export of Archaeopteryx specimens in 2015.

That custody problem is real and this article prints it.  But it should be stated for exactly what it is:  a weakness about access, not about authenticity.  No Archaeopteryx specimen has ever been shown to be a forgery.  A creationist who slides from "privately held" to "suspect" is making an accusation the evidence does not support, and Part 8 shows what happens when that accusation is made carelessly.

The Haarlem specimen: a pterosaur for 113 years, an Archaeopteryx for 47, and now neither

This is the most instructive item in the entire subject for anyone arguing about the limits of morphological classification.  The real history is stranger than the version usually told.²¹

1. 1855. The specimen is found at Jachenhausen — six years before the London Archaeopteryx.
2. 1857. Hermann von Meyer describes it as a pterosaur, Pterodactylus crassipes.
3. 1970. John Ostrom re-examines it, identifies it as a theropod, and refers it to Archaeopteryx.
4. 2017. Christian Foth and Oliver Rauhut re-examine it again, remove it from Archaeopteryx, erect the new genus Ostromia — named for Ostrom — and place it with the Chinese anchiornithids.

So the specimen was a flying reptile for 113 years, then the type animal of bird origins for 47 years, and is now neither.  The version that says "forty years" understates it and skips the pterosaur century entirely.

The reason for the 2017 move is specific.  Foth and Rauhut found no diagnostic features of Archaeopteryx on it at all, found that its proportions differ markedly from real Archaeopteryx specimens, and found two characters it shares with anchiornithids.²¹  Their phylogenetic analysis then recovered it as the first anchiornithid known outside the Tiaojishan Formation of China.

Why Ostrom got it wrong is the part that matters, and it is not what a creationist expects.  Ostrom was not careless and he was not hiding anything.  Foth and Rauhut record that Ostrom himself noticed the proportional differences — in the metatarsus and the hand claws — and assigned them "either to preservational artifacts (in the case of the metatarsus) or individual variation."²¹

He saw the differences and explained them away.  What changed between 1970 and 2017 was not the fossil and not the ideology.  It was the discovery of the Tiaojishan anchiornithids in China, which for the first time supplied an alternative the specimen could be compared against.  In 1970 no other Jurassic paravian existed.

Foth and Rauhut also state the obvious objection against themselves — that the distinguishing features might be crushing artefacts — and answer it with evidence rather than assertion.²¹  Whether that convinces a reader is a fair question.  The point is that the argument is printed in the paper, testable, and not concealed.

Verdict:  the reassignment is real, it is sound, and it is a legitimate creationist point about the limits of morphological classification — provided it is told accurately.  The accurate telling is that an incomplete specimen was misidentified twice by careful workers over 162 years, and that both corrections came from mainstream paleontology, not from outside it.  It says nothing about the well-preserved specimens.  The Berlin and Chicago skeletons are not incomplete, and nobody has proposed removing them from the genus.

The feather: a second reassignment, still unresolved

The 1861 feather was the original holotype of Archaeopteryx lithographica and the first fossil feather ever described.  In 2011 the ICZN transferred the name to the London skeleton, precisely because a single feather cannot be diagnosed.

In 2019 Thomas Kaye, Michael Pittman, Gerald Mayr, Daniela Schwarz and Xu Xing used laser-stimulated fluorescence to recover the geochemical halo of the feather's long-missing calamus, settling a 150-year-old question about whether the quill von Meyer drew in 1862 had ever existed.  Their abstract then states that the completed morphology "excludes it as a primary, secondary or tail feather of Archaeopteryx," leaving covert or contour feather as the options, and raises the possibility that it came from some other feathered animal of the Solnhofen archipelago.²²

In 2020 Ryan Carney, Helmut Tischlinger and Matthew Shawkey replied that it is an upper major primary covert of Archaeopteryx after all.  Their central technical charge is that Kaye's plotted centreline "is incorrect and cannot be reproduced from the paper's results," and that redrawn from Kaye's own fluorescence image the feather falls inside the range of modern primary coverts.²³

One detail belongs in an article that audits its own side.  Carney and colleagues complain that the Kaye press release went further than the Kaye paper, headlining that the first fossil feather "did not belong to iconic bird Archaeopteryx."²³  The paper said the possibility remains.  The press release said it did not belong.  Creationist sources citing the 2019 result as settled are citing the press release, not the paper.

Status:  unresolved.  This article asserts neither way.  The honest statement is that the original holotype of the most famous fossil in the world is a single feather that two specialist teams still cannot agree belongs to the same animal.  Carney's team concedes the limit themselves:  "Given the isolated nature of the fossil feather, we can never know the anatomical and taxonomic provenance with 100% certainty."²³

What the specimens actually preserve

Four points from the primary literature that a reader is unlikely to have met.

The 12th specimen has no feathers at all.  Rauhut and colleagues:  "No feathers or impressions thereof are preserved, but some of the pedal unguals show remains of their horny sheaths."²⁰  Feather preservation in the Solnhofen limestones is a taphonomic lottery, not a property of the animal.

The variation between specimens is large and unexplained.  Rauhut and colleagues measured limb proportions across nine specimens and found tibia, metatarsal II, metatarsal III and scapula differences significant at p < 0.01.  Their dentition table records tooth number, spacing, orientation and shape varying so much that no two specimens show the same pattern.  Their own range of explanations runs from ordinary intraspecific variation up to the possibility that the known specimens are a "species flock" produced by island speciation.²⁰  Whether "Archaeopteryx" is one animal or several is a live, published question.

The genus cannot be diagnosed by any single character.  Verbatim from the same paper:  the genus "cannot be diagnosed on the basis of any single apomorphic characters, but only by an apomorphic set of combined characters."²⁰  The emended diagnosis is a list of eight features, only three of which can be checked on the neotype.

The range and duration are tiny.  Foth and colleagues give the genus a geographic range of at least about 64 km and a duration of between 700,000 and one million years.¹⁹  Every specimen comes from one small area of Bavaria.

None of that makes Archaeopteryx less transitional.  All of it limits how much any single fossil's classification can be asked to prove — which is the point Part 3 makes with numbers.

Part 3 — Where Archaeopteryx sits on the tree, and how thin the margin is

The short version

Archaeopteryx has moved.  Between 2011 and 2026 it has been recovered as the most basal bird, as a deinonychosaur — a "raptor" — and as a member of a clade whose own membership shifts from paper to paper.  The movement is real, it is documented by the people who did the analyses, and the mainstream authors themselves say the support is weak.  That is not a creationist gloss.  It is what the papers say.

Two things have to be said in the same breath or this section will mislead.

First, the instability is inside a small region of the tree.  Every one of these results puts Archaeopteryx within Paraves — the clade holding dromaeosaurs, troodontids and birds.  Nobody has recovered it as a lizard, a pterosaur or a modern bird.  The dispute is over which twig, not which branch.

Second, the reason for the instability is not mysterious.  The animals in this region are so similar that the characters cannot separate them.  Stephen Brusatte, Graeme Lloyd, Steve Wang and Mark Norell measured it in 2014:  birds evolved faster than other theropods, but "are indistinguishable from their closest relatives in morphospace," and there was "no great jump between nonbirds and birds in morphospace."²⁴  Brusatte does not mean that birds cannot be told from non-birds.  But his sentence does license the narrower point about classification, and that is how it is used here.

The sequence of analyses, with what each actually claimed

Xu Xing and colleagues, 2011, Nature.  Described Xiaotingia zhengi.  Adding it to a comprehensive analysis "shifts Archaeopteryx to the Deinonychosauria" — a result the authors qualify in their own abstract as having "only tentative statistical support" and as needing confirmation before the assumed ancestral condition of birds is rewritten.²⁵  The hedging is Xu's, in the abstract, not a critic's.  The numbers Xu reported were Bremer support 2 and bootstrap below 50 per cent.²⁶

Michael Lee and Trevor Worthy, 2012, Biology Letters.  Reanalysed Xu's own matrix under maximum likelihood and Bayesian inference instead of parsimony.  Archaeopteryx returns to the base of birds with 73 per cent likelihood bootstrap and a Bayesian posterior probability of 1.0.²⁶

Their explanation of why the two methods disagree is the most useful thing in this literature.  They counted the characters that differentially support each tree:  64 favour the parsimony tree against 49 for the likelihood tree, but the parsimony-supporting characters are the more homoplastic set — consistency index 0.312 against 0.354, retention index 0.665 against 0.728.²⁶  So parsimony wins on character count and likelihood wins on character quality.

Their own conclusion is one no honest article may omit:  the analysis does not settle it, bootstrap support is only about 73 per cent, Bayesian posteriors can overstate support when the model is inadequate, and new fossils or new characters are needed.²⁶  Their abstract puts it in one sentence:  "Both positions for Archaeopteryx remain plausible, highlighting the hazy boundary between birds and advanced theropods."²⁶

Alan Turner, Peter Makovicky and Mark Norell, 2012, Bulletin of the AMNH.  Re-scored two of Xu's characters after examining the specimens in person and re-ran the matrix.  Their result:  the paravian node has eight resolutions across 231 most-parsimonious trees, only one of which moves Archaeopteryx away from Avialae, and a majority-rule consensus puts 89 per cent of resolutions behind the traditional placement.²⁷

But on the same page they record a result against their own preference:  correcting one character alone still left Archaeopteryx closer to deinonychosaurs than to Avialae, which "draws into doubt Xu et al.'s (2011) contention that their result was predicated entirely on the inclusion of Xiaotingia."²⁷  Their verdict:  "it may be premature to declare Archaeopteryx a nonavian theropod."²⁷

Pascal Godefroit and colleagues, 2013, Nature.  Described Aurornis xui and reported that adding it recovers Aurornis as the basal-most avialan, confirms the avialan status of Archaeopteryx, places Troodontidae as sister to Avialae, gives powered flight a single origin within Paraves, and puts the early diversification of Paraves in the Middle-to-Late Jurassic.²⁸

Foth and Rauhut, 2017, in the Ostromia paper, recovered Archaeopteryx as an avialan — and in passing catalogued exactly how unstable the neighbourhood is.  Epidexipteryx has been placed at basal Avialae, at basal Paraves and at basal Oviraptorosauria in different studies.  Anchiornis has been placed at basal Deinonychosauria, basal Troodontidae and basal Avialae.²¹  Three positions each, for two of the animals the whole argument turns on.

Scott Hartman and colleagues, 2019, PeerJ.  Described Hesperornithoides miessleri from the Morrison Formation of Wyoming.  This is the strongest recent result against the traditional placement, and it is a mainstream open-access paper.  Once their new data on Archaeopteryx and Anchiornis are folded in, Archaeopteryx nests inside the anchiornithine clade and the whole group comes out as deinonychosaurs.

Then the sentence that this whole section exists for:  "both the traditional Archaeopteryx position closer to Aves and the common Anchiornis position sister to troodontids require a single additional step each."⁷  Making archaeopterygids sister to dromaeosaurids costs four more steps.

One parsimony step.  That is the whole margin between "Archaeopteryx is the first bird" and "Archaeopteryx is a raptor" in an analysis with more than 700 characters and more than 500 taxa.

Michael Pittman and Xu Xing, editors, 2020, Bulletin of the AMNH 440.  A 353-page, 49-author volume.  Its chapter 3, by Diego Pol and Pablo Goloboff, quantifies the instability directly.  Analysing 164 taxa and 853 characters, the strict consensus has two large polytomies at the base of Avialae and inside Dromaeosauridae, caused by four unstable taxa — one of which is the Haarlem Archaeopteryx.²⁹  Their protocol identified 40 unstable taxa as most damaging to node support in that dataset.  Their list of clades recovered as only moderately supported includes Maniraptora, Pennaraptora, Paraves, Dromaeosauridae, Troodontidae, Anchiornithinae and the early-diverging clades of Avialae — that is, essentially every clade this article is about.²⁹

Foth and colleagues, 2025, Fossil Record.  The equally-weighted analysis of the 13th specimen produced over 200,000 most-parsimonious trees of length 3,666, with a consistency index of 0.211 and a retention index of 0.733, and a strict consensus with a large basal polytomy.¹⁹

That consistency index deserves a sentence of explanation, because it is the best single number in this article.  A consistency index of 0.211 means that roughly four fifths of the character changes in that matrix are homoplasy — convergence or reversal — rather than shared inheritance.  It is the quantitative form of "morphology is a noisy signal here," published by the people who built the matrix.

Jingmai O'Connor and colleagues, 2025, Nature.  The Chicago specimen.  Treats Archaeopteryx as an archaeopterygid avialan and describes its palate as intermediate between troodontids and crownward Cretaceous birds.¹⁸

The honest summary

• The current majority position is that Archaeopteryx is a basal bird. That is what Turner 2012, Godefroit 2013, Foth and Rauhut 2017, Pittman and Xu 2020, Foth 2025 and O'Connor 2025 all recover.
• The minority position is that it is a deinonychosaur. Published by Xu in 2011 and independently recovered by Hartman in 2019. Neither has been refuted. Both were published with explicit statements of weak support.
• The margin is one parsimony step, or a switch from parsimony to likelihood. Not "evolutionists cannot agree," but "here is the measured margin, taken from their own papers."

And here is what does not follow.  It does not follow that Archaeopteryx is therefore not transitional.  Under every one of these topologies the animal has teeth, a long bony tail, clawed fingers, a furcula and asymmetric flight feathers.  Its character combination does not change when its branch does.  A creationist argument that leans on the phylogenetic instability in order to deny the mosaic anatomy is confusing two different claims, and a competent skeptic will catch it inside a minute.

The legitimate use of this material is narrower and stronger.  It is evidence about how much weight a fossil's classification can bear — and the answer, measured in the field's own units, is one step.

Part 4 — The temporal paradox, judged on relative order

The objection as it was framed

The argument runs like this.  Archaeopteryx is Late Jurassic.  The dromaeosaurs and troodontids said to be its closest relatives are Cretaceous.  So the supposed ancestors are younger than the supposed descendant, and the whole story runs backwards.

Alan Feduccia pressed this hard in The Origin and Evolution of Birds in 1996 and in the Auk exchanges of 2002–2003, and it passed from there into essentially every creationist treatment of the subject.  Richard Prum, summarising his opponents in The Auk in 2002, recorded how large a gap they were claiming:  "Feduccia and Martin (1998), Feduccia (1999b), and Dodson (2000) all characterized the temporal disjunction between Archaeopteryx and dromaeosaurs as between 60–80 my."³⁰

Those 60 to 80 million years are the critics' figure on the mainstream scale, and this article does not use them.  The argument does not need them.  Stated in relative order, it reads:  Archaeopteryx comes from the Solnhofen limestone of Bavaria, at a Late Jurassic level, and the dromaeosaurs and troodontids its critics compared it with come from Cretaceous beds above that level, on other continents.  So the supposed relatives sit higher in the column than the supposed early bird.  That is a claim about order, and it can be tested on order.

That was a fair objection in 1996.  It is not a fair objection in 2026, and this article says so plainly.

What was found after the objection was framed

Between 2005 and 2018 a run of small feathered paravians came out of the Tiaojishan Formation and the Daohugou beds of Liaoning and Hebei, in rock that sits below the Solnhofen level in the conventional column.  Nine of them.

The people who found them said what it meant, in the abstract, without hedging.  Hu Dongyu, Hou Lianhai, Zhang Lijun and Xu Xing, in Nature in 2009:  "This new find refutes the 'temporal paradox'."³¹

There is also a check that does not depend on the Chinese material at all.  Hesperornithoides miessleri is a troodontid from the Morrison Formation of Wyoming, at essentially the same level as Archaeopteryx.⁷  A deinonychosaur is now known from the same slice of the column as the first bird, on a different continent.

The ages, pulled directly

These are first-appearance and last-appearance ages as the Paleobiology Database reports them, from a live query on 3 August 2026.  They are printed as the mainstream figures under discussion, not as this article's evidence.¹²

The bins for the Tiaojishan animals are coarse, because the database records the outer limits of the formation's assigned range rather than a point age.  Do not quote a single figure for them.  The defensible statement is the ordinal one:  they occur below the Solnhofen level in the conventional column.

Deinonychus's 143.1 Ma figure is the same kind of artefact, and it has to be flagged here rather than left to be caught.  Forty-six of the forty-seven Deinonychus occurrences in the database fall between the Aptian and the early Cenomanian.  A single record binned to the Berriasian drags the genus-level first-appearance figure back by more than twenty million years.  The animal itself is Early Cretaceous, tens of millions of years after Archaeopteryx, which is what the next section says.¹²

What survives of the objection

Two things, and only two.

First, the famous animals really are younger.  Deinonychus, Velociraptor, Microraptor and Sinosauropteryx all postdate Archaeopteryx by tens of millions of years on the conventional scale.  A museum diorama that lines up Velociraptor → Archaeopteryx → pigeon shows a sequence that runs backwards in time, and a visitor who notices is right to complain about the diorama.

Second — and this is the real answer — the complaint is about the diorama, not about the claim.  No paleontologist says Deinonychus is the ancestor of Archaeopteryx.  The claim is that they share an ancestor.  In a branching tree a sister group can appear in the record at any time after the split, and its lineage is as old as the split even where its fossils are not.

Christopher Brochu and Mark Norell tested exactly this, by asking how well each competing hypothesis of bird ancestry matches stratigraphic order.  Their result, as Prum summarises it, is that Feduccia's own preferred basal-archosaur ancestry "requires hypothesizing even longer 'ghost lineages' than does the theropod hypothesis."³²

That point is decisive and it is quantitative.  Every alternative on offer creates a worse temporal paradox than the one it was raised to defeat.  The objection cannot be used to move a reader toward "birds came from something else," because "something else" is worse off.

One honest note on the sourcing.  Brochu and Norell's own papers were not obtained for this article; the result is reported here through Prum's summary of them, which is a hostile-to-Feduccia source.³³  A reader who wants the numbers should get the originals, and this article does not print any number from them.

Tracks move first-appearance dates before bones do

This distinction is almost never made in the creationist literature, and it decides how the argument is scored.

The live case.  Miengah Abrahams and Emese Bordy re-examined Paul Ellenberger's Trisauropodiscus tracks from four sites in Lesotho, plus an 80-metre tracksite at Maphutseng, and published in PLOS ONE in November 2023.  Their abstract:  the tracks have "demonstrable bird-like affinities, predating basal bird body fossils by c. 60 million years."³⁴

Then they state, against their own headline, exactly what the tracks do and do not show.  Their Morphotype II "has a convincingly avian affinity but is not distinctly avian," lacking a well-developed digit III metatarsophalangeal pad and preserving no hallux impressions.³⁴  Their conclusion on the trackmaker:  "it is likely that Morphotype II tracks were made by a yet-to-be-found tridactyl archosaur."³⁴  The authors themselves read the tracks as convergence in an archosaur, not as a Triassic bird, and they say so in the paper with reasons.

The dead case, and it is the honest answer to the suppression charge.  In June 2002 Ricardo Melchor, Silvina de Valais and Jorge Genise published apparently avian tracks from the Late Triassic in Nature, some 55 million years before the first bird skeleton.  The same three authors retracted it themselves in 2013, after redating showed the unit is Late Eocene rather than Late Triassic.³⁵³⁶

Three things follow, and all three belong here.

1. A trace fossil can move a first-appearance datum without any bone. Any temporal-paradox argument must state whether it is counting tracks or skeletons, because the two records give different answers.
2. The retraction is a clean answer to the charge that anomalies are suppressed. A result contradicting the standard story ran in Nature, stood for eleven years, and was withdrawn by its own authors when the evidence turned. Nobody had to be exposed by an outsider. A creationist who cites the 2002 paper without the 2013 retraction is doing to Nature exactly what he accuses Nature of doing to him.
3. But the retraction rests on radiometric dating, which this series excludes from its own evidence. It is presented here as what mainstream geology concluded and why, with the caveat stated, not as a closed case by this article's own standards. The same caveat applies to Abrahams and Bordy, whose age figure comes from uranium-lead dating of detrital zircons.

One detail worth a sentence:  Melchor and de Valais were among the peer reviewers of Abrahams and Bordy 2023, named in its acknowledgements.³⁴  The people who got the Triassic-bird-track claim wrong in 2002 vetted the next one.

What relative order between continents shows, and who may use it

The temporal paradox is an argument from fossil order matched between continents.  It sets the Solnhofen limestone of Bavaria against Cretaceous beds in North America and Asia.  The answer to it sets the same limestone against the Tiaojishan and Daohugou beds of China, and the tracks in the last section bring in the Elliot Formation of Lesotho.

Superposition alone orders the beds in one outcrop.  Linking beds on different continents takes correlation, and the oldest kind of correlation uses fossils, not radiometric dates.  William Smith identified the strata of England by their fossils in 1816.³⁷  Albert Oppel divided the Jurassic of England, France and south-west Germany into zones named for ammonites in 1856 to 1858.³⁸  Both worked decades before radioactivity was discovered.  Matching of this kind gives a relative time scale:  which bed is lower, not how many years lie between them.

That kind of matching is open to a young-earth writer, and leading young-earth geologists use it.  Kurt Wise and Donna Richardson tested the column against the Paleobiology Database and concluded that "the global biostratigraphic column seems to be real."⁶  Timothy Clarey and Davis Werner built their Flood model on nearly 3,000 stratigraphic columns, and they matched rock packages between them "across each continent and again, globally."³⁹  So a creationist who argues from the order of the Solnhofen and the Cretaceous beds is making a legitimate kind of argument.

And on that same order, the paradox fails.  The matching that puts Deinonychus above Archaeopteryx also puts nine feathered paravians from the Tiaojishan and Daohugou beds below it, and a troodontid from the Morrison Formation of Wyoming at about its level.⁷  The Paleobiology Database bins are coarse, but even their outer limits keep the order:  the Tiaojishan animals fall between the Bathonian and the Kimmeridgian stages, below the Tithonian, where Archaeopteryx first appears.¹²  A writer who accepts the first placement should weigh the second by the same standard.  Hu and his co-authors drew that conclusion in the abstract of their Nature paper.³¹  The difference, set out below, is that the Chinese placement leans more on radiometric dates than the Bavarian one does.

The narrower point is true and worth saying.  A writer who calls the whole biostratigraphic column circular, on the ground that the fossils date the rocks and the rocks date the fossils, cannot use the temporal paradox at all.  "Younger" and "older" between Bavaria and Montana are exactly what the column outputs.  He cannot reject the method and keep its result.  That inconsistency is his alone;  it is not a rule for anyone else, and Wise, Richardson and Clarey do not commit it.

How the Chinese beds were placed, and why it matters here.  This is the weak joint in the reply.  The Tiaojishan and Daohugou beds are lake deposits.  They share no marine index fossil with the ammonite zones of the Bavarian limestone, so they cannot be matched to the Solnhofen level the way Smith and Oppel matched beds.  Their placement rests on regional biostratigraphy together with radiometric dates of volcanic rocks, and for a time it was disputed.  Gao Keqin and Ren Dong, defending the Jurassic age in 2006:  "The fossil beds were first misinterpreted as of Early Cretaceous age, based on alleged occurrences of key fossils of the Jehol Biota.  Compelling evidence revealed by more rigorous research involving regional biostratigraphy, radiometric dating, and paleontology supports the Middle Jurassic age of the fossil beds."⁴⁰  So the claim that the nine Chinese paravians lie below the Solnhofen level leans partly on radiometric dating, which this series does not count as its own evidence.  A young-earth reader may fairly discount that part, and should say that is what he is doing.

What fossil order alone still shows.  A paravian of the Anchiornis group sits at Archaeopteryx's own level in Bavaria.  The Haarlem specimen, now Ostromia (Part 2), comes from the "Early Tithonian laminated limestones of the Painten Formation" at Jachenhausen, in the same chain of lagoon basins as Archaeopteryx, with "a slight stratigraphic difference between some of the eastern and western plattenkalks."²¹  Those basins are matched to each other by fossils within one region, with no dates and no ocean to cross.  The anchiornithids are the group the 2009 Anchiornis paper called troodontids and other analyses call the earliest birds;  on either reading they are among the close relatives the paradox said appear only higher up.  So the paradox as framed, that the relatives of Archaeopteryx appear only above its level, fails on fossil order alone.  What fossil order alone does not show is that they lie below it.

The mainstream says much the same about the record outside China.  Christian Foth and Oliver Rauhut, the paleontologists who named Ostromia:  "Prior to the recent description of abundant maniraptoran theropods from the Oxfordian of China …, no certain pre-Kimmeridgian representatives of this clade were known."  The older finds elsewhere are mostly isolated teeth, and "teeth similar to those found in dromaeosaurids are also present in basal tyrannosauroids … and, probably, the juvenile dentition of more basal theropods."²¹  The ghost-lineage answer above still holds either way:  a sister group may appear in the record after the split.

Verdict

The temporal paradox, as framed, is dead on the evidence, and on fossil order alone.  It said the close relatives of Archaeopteryx appear only above its level.  An anchiornithid at its own level in Bavaria refutes that, and the Chinese finds of 2005 to 2018, made by the same research group whose Xiaotingia paper is the creationists' favourite citation for Archaeopteryx's instability, persuaded the field.  The stronger reply, that feathered paravians lie below Archaeopteryx, leans partly on radiometric dates for the Chinese beds, and on this series' own rule it is weaker than Hu's abstract makes it sound.  It was a fair kind of argument, because fossil order matched between continents is open to both sides.  The one writer barred from it is the one who calls the column circular.  In its old form, "the relatives are all younger," it belongs on the list of arguments creationists should stop making.  The narrower statement, that no close relative is securely older on fossil correlation alone, is true and may be made.

Part 5 — The avian lung

What the objection actually is

The bird lung is not a bigger version of a reptile lung.  It is a different machine.

A mammal or a lizard breathes tidally:  air goes in and out of the same blind-ended spaces.  A bird's gas-exchanging lung is rigid and does not expand.  It is ventilated by a set of compliant air sacs that work as bellows, and air moves through the gas-exchange tubes — the parabronchi — in one direction on both inhalation and exhalation.  Blood runs across that flow, giving a crosscurrent exchanger that is more efficient than anything a mammal has.

The objection, in its strongest form, is that there is no functional intermediate between the two designs.  A half-converted lung, on this argument, is a lung that does not work.

It was published by John Ruben and colleagues in Science in 1997, and creationists cite him because he is not a creationist.  He is a respiratory physiologist at Oregon State who accepts evolution and simply denies that birds came from theropods.⁴¹

The single most-quoted sentence in the creationist literature on this subject is not printed in this article.  It is the diaphragmatic-hernia line — the claim that evolving abdominal air sacs from a diaphragm-ventilating ancestor would have required selecting for a hole in the diaphragm.  It appears on creationist sites footnoted to Ruben's 1997 Science paper.  That paper was never opened for this article, so nothing is quoted from it.¹¹

The substance of the argument is confirmed instead by a hostile source that paraphrases it accurately and cites Ruben directly:  Mathew Wedel's Berkeley doctoral dissertation on postcranial pneumaticity and the origin of the avian lung.  Wedel, p. 133:  "They further claimed that birds could not be theropod descendants because abdominal air sacs could not have evolved without herniating the diaphragm and such hernias would have been selected against."⁴²

That is the objection, reported by someone who disagrees with it, from a source that was actually read.

Jonathan Sarfati's 2009 article for Creation Ministries International carries a second Ruben argument, from Quick and Ruben's Journal of Morphology paper of the same year:  a bird's femur is held nearly horizontal and immobile, and that femoral-thigh complex is what stops the thin-walled abdominal air sacs from collapsing inward on inhalation.  Theropods had mobile femurs.  Therefore, on this argument, they could not have had working abdominal air sacs.⁴³

What happened next: three measurements, not three arguments

The objection assumed that unidirectional airflow is a package deal — that it requires air sacs, requires a rigid parabronchial lung, and exists to support flight and endothermy.  All three assumptions have since been measured, and all three are false.  This came entirely from mainstream physiology labs working on living animals.

Alligators, 2010.  Colleen Farmer and Kent Sanders, in Science:  "air flows unidirectionally through parabronchi in the lungs of the American alligator, an amphibious ectotherm without air sacs, which suggests that this pattern dates back to the basal archosaurs of the Triassic."⁴⁴

Read the qualifiers in that sentence.  Without air sacs.  Ectotherm.  Two premises of the objection fail in one animal.

Monitor lizards, 2014.  Emma Schachner, Robert Cieri, James Butler and Colleen Farmer, in Nature, demonstrating "region-specific unidirectional airflow in the lungs of the savannah monitor lizard (Varanus exanthematicus)."⁴⁵

This one matters most, because Varanus is outside Archosauria — a lepidosaur, on the other side of the diapsid split.  The authors state the two options honestly:  either the pattern evolved twice, or it is homologous across archosaurs and Varanus, having evolved once in ancestral diapsids.⁴⁵

Iguanas, 2014.  Robert Cieri, Brent Craven, Emma Schachner and Colleen Farmer, in PNAS.  The green iguana has none of the structures the avian pattern was supposed to require.  From their abstract:  "Here we show unidirectional airflow in the green iguana, a lizard with a strikingly different natural history from that of birds and lacking these anatomical features.  This discovery indicates a paradigm shift is needed."⁴⁶  From the discussion:  "These results indicate that unidirectional flow does not require either a bronchus that has grown deep into the lung as a mesobronchium or intercameral perforations."⁴⁶

The consequence for the objection is direct.  The "no functional intermediate" claim rested on the premise that the airflow pattern only works when the whole avian apparatus is present.  That premise is now falsified in three living animals:  an ectothermic crocodilian with no air sacs, a lizard outside Archosauria, and an iguana with a two-chambered lung and no parabronchi.  The intermediate states are not hypothetical.  They are walking around.

What survives of the objection, stated fairly

Three honest limits, because the reversal should not be overstated either.

1.  The lizard and crocodile results do not show that the flow pattern is useful.  Cieri and colleagues say so against their own headline:  "we know of no clear evidence for increased efficiency of gas exchange in nonavian reptiles over that seen in mammals."⁴⁶  They also concede that convergent evolution has not been ruled out.⁴⁶  So the finding kills the "impossible intermediate" argument without establishing that selection built the avian system for the reason usually given.

2.  The claim that theropods had bird lungs is currently over-claimed, and the correction comes from inside the field.  Emma Schachner and Andrew Moore, in Philosophical Transactions of the Royal Society B in 2025:  "we argue that without evidence for the horizontal septum, a fully avian lung should not be reconstructed in non-avian ornithodirans, despite the presence of post-cranial skeletal pneumaticity."⁸

Note who wrote that.  Schachner is the same researcher who found unidirectional airflow in the monitor lizard — one of the people whose work destroyed the strong form of the creationist objection.  This is a 2025 mainstream caution that the fossil evidence for avian lungs in dinosaurs is being pushed further than it will go, and a creationist reader deserves to know that his best point here has been made better by a specialist than by any apologist.

3.  The Quick and Ruben femur argument has not, so far as could be determined, received a dedicated published rebuttal.  Wedel's dissertation predates it by two years and does not address it.  Searches for a formal reply came up empty.  It is recorded here as unresolved rather than pretended away.

There is one soft spot in it worth stating without a citation crutch.  The argument infers a hard functional requirement from one point on a continuum of living birds, and basal birds did not have the modern avian femoral posture.  That is an observation, not a refutation, and it is offered as such.

The positive fossil evidence, and exactly how far it goes

Patrick O'Connor and Leon Claessens, in Nature in 2005, reported "evidence for cervical and abdominal air-sac systems in non-avian theropods, along with thoracic skeletal prerequisites of an avian-style aspiration pump."⁴⁷  Their method is worth understanding, because it is what makes the inference possible at all:  specific air sacs pneumatise specific bones in an invariant pattern, so the pattern of holes in fossil bone identifies which sac was present.

Andrew Moore and colleagues confirmed in 2025 that the framework is still standard — cervical diverticula invade the cervical and cranialmost thoracic vertebrae, abdominal diverticula pneumatise the post-mid-thoracic, synsacral and caudal vertebrae and the pelvic girdle.⁴⁸  And for the fossils, they conclude that the unambiguous pneumaticity correlates in sauropodomorphs, theropods and pterosaurs suggest "a heterogeneous lung with a flexible, caudoventral sac-like component—is ancestral for Ornithodira."⁴⁸

Note the specialists' wording:  a flexible caudoventral sac-like component, not a modern bird lung.  That is exactly as far as the bones go, and it is where this article stops.

Verdict

• The strong form of the objection is dead. "There is no functional intermediate between a bellows lung and a flow-through lung" is refuted by living alligators, monitor lizards and iguanas, measured with endoscopes and flow probes between 2010 and 2014. A creationist article still running the 1997 argument unmodified is nearly thirty years out of date and will be caught.
• The weak form survives, and should be adopted instead. The complete avian system — decoupled rigid lung, horizontal septum, air sacs, crosscurrent exchanger — is not demonstrated in any non-avian dinosaur, and a 2025 review in a Royal Society journal says so in as many words. Argue from Schachner and Moore, not from Ruben.
• The Quick and Ruben femur argument is unresolved, and should be presented as unresolved.

Part 6 — Feathered dinosaurs and the collagen dispute

Why this matters more than Archaeopteryx does

If non-avian dinosaurs had feathers, then Archaeopteryx's feathers stop being a bird character, and the "it's just a bird" reading collapses without anyone having to argue about a cladogram.

So the creationist case rests far more on the collagen dispute than on anything about the Solnhofen specimens — and the creationist literature has not caught up with where that dispute now stands.

The collagen hypothesis, at full strength

It deserves to be stated at full strength, because it was a serious argument published in serious journals by an anatomist who works on reptile skin.

Alan Feduccia, Theagarten Lingham-Soliar and J. Richard Hinchliffe, 2005, in the Journal of Morphology, from their abstract:  "Our findings show no evidence for the existence of protofeathers… we conclude that 'protofeathers' are probably the remains of collagenous fiber 'meshworks' that reinforced the dinosaur integument."⁴⁹

Lingham-Soliar, Feduccia and Wang, 2007, in Proceedings of the Royal Society B, on a new Sinosauropteryx specimen.  Their claim is specific and testable, not hand-waving:  in the proximal tail the fibres run in bands parallel to the tail's long axis and originate from the skin, and dorsal to the neck and back they form a stiffening frill.  Their conclusion:  "The fibres show a striking similarity to the structure and levels of organization of dermal collagen.  The proposal that these fibres are protofeathers is dismissed."⁵⁰

Feduccia restated the position in 2016, and the restatement narrows it in a way that matters.  He does not deny real feathers on the small Chinese maniraptorans.  He counts Microraptor, Anchiornis and Caudipteryx as probable early birds, and keeps the collagen reading for the simple filaments of animals he accepts as dinosaurs:  "there are perfectly good theropods such as Sinosauropteryx," and "The so-called dinosaur protofeathers are generally collagen fibers or even a variety of structures, and the fibers of Sinosauropteryx are clearly beneath the skin."⁵¹  So the claim to be tested is about simple filaments, not about vaned feathers.  A fossil with real vaned feathers does not answer it, because in Feduccia's scheme such a fossil is more likely a bird.

What answered it

Melanosomes, 2010.  Zhang Fucheng and colleagues, in Nature, reported melanosomes preserved "not only… in the pennaceous feathers of early birds, but also in an identical manner in integumentary filaments of non-avian dinosaurs, thus refuting recent claims that the filaments are partially decayed dermal collagen fibres."⁵²

The logic is the point, not the authority.  Collagen fibres carry no pigment of their own.  Feathers do.  Finding pigment organelles inside the filaments, in life position, would be a positive test between the two hypotheses.  The method rests on Jakob Vinther and colleagues' 2008 work on a colour-banded Crato Formation feather, where dark bands preserve carbon-rich oblate bodies one to two micrometres long and light bands preserve none, matching eumelanin-bearing melanosomes in living black feathers.⁵³

Lingham-Soliar answered in 2011, and his answer should be stated in his own words.  He argued that the supposed melanosomes in Sinosauropteryx were never measured statistically, that no cross-section showed them inside the filament, and that "alternative arguments, e.g. that the particles might be bacteria or colour from the overlying skin, are peremptorily dismissed or not considered."  He also ran decay experiments:  "from experiments on the decomposition of native collagen in fish and reptilian dermis, SEMs of their ultrastructure show that distinctive spherical, elliptical or oblate particles, even more so than those figured in Sinosauropteryx, typically form during degradation."⁵⁴  That is a real reply.  If round bodies of the right size can form as collagen decays, the melanosome test is less clean than the 2010 paper made it look.

The answer to that came in 2015 from eleven paleontologists and ornithologists led by Adam Smith.  They call Lingham-Soliar's alternatives "special pleading" but concede that his arguments are "too complex to fully tackle here."  Then they make a comparison that does not depend on melanosomes at all.  If collagen fibres fossilise in these beds, they should show up on other animals too, and they do not:  "The structures in question are not observed in unequivocally feathered avian specimens (e.g., Confuciusornis) from the Jehol Biota," and "no fibers that might be misinterpreted as collagen have been identified in any of the hundreds of choristodere specimens found alongside Sinosauropteryx in coeval Chinese deposits."⁵⁵  That is the strongest single argument against the collagen reading, because it uses the same beds, the same kind of preservation and many animals.  Feduccia's 2016 reply to Smith did not take it up;  it said the filament question "will be covered in a separate paper by T. Lingham-Soliar."⁵¹

The direct rebuttal, 2017.  Phillip Smithwick, Gerald Mayr, Evan Saitta, Michael Benton and Jakob Vinther, in Palaeontology.  They report that the collagen reading "has been based on observations of low-quality images of Sinosauropteryx," and that "integumentary structures have been misinterpreted based on sedimentary features and preparation marks."⁵⁶  Some of what was read as collagen fibres was damage left by the person who prepared the slab.

Amber, 2016.  Xing Lida and thirteen colleagues, in Current Biology, describing a feathered dinosaur tail in mid-Cretaceous Burmese amber.  Amber removes the compression problem that the whole dispute turned on.  The specimen preserves the spatial arrangement of follicles and feathers on the body, with feathers showing "a short, slender rachis with alternating barbs and a uniform series of contiguous barbules."⁵⁷

Follicles, a rachis, barbs and barbules, in three dimensions.  That is feather anatomy.  Dermal collagen has none of those things.

But the amber answers a narrower question than it is often used for.  These are branched feathers with a rachis, not the simple filaments of Sinosauropteryx.  The specimen shows real feathers on a small theropod with a long, unfused tail.  It does not show what the filaments of Sinosauropteryx are, and Feduccia, who already accepts real feathers on several maniraptorans and counts them as birds, could place it the same way.  It belongs in the case that small theropods had feathers;  it is not a direct test of the collagen claim.

The honest counterweight, which this article prints

Caitlin Moyer and colleagues, from Mary Schweitzer's laboratory — the dinosaur-soft-tissue group — argued in Scientific Reports in 2014 that the microbodies could be bacteria rather than melanosomes, because the two overlap in size, shape and distribution.  Their discriminating criterion:  "Melanosomes are always internal, embedded in a morphologically distinct keratinous matrix.  Conversely, microbes grow across the surface of feathers in continuous layers, more consistent with published images from fossil feathers."⁵⁸

That is a real objection from a serious lab and it is not settled by assertion.  But note precisely what it does and does not do.  It challenges melanosome identification.  It does not challenge feather identification.  If the microbodies were bacterial, the amber specimen would still have follicles, a rachis and barbules.

It does have a consequence worth keeping:  every published colour reconstruction of a feathered dinosaur — including Smithwick's own countershaded, bandit-masked Sinosauropteryx — depends on melanosome identification being right.  Those reconstructions should be treated as provisional, and this article treats them that way.

Verdict

• The collagen hypothesis has lost for Sinosauropteryx, on the weight of the evidence rather than on one decisive test. The 2010 melanosome finding is contested by Lingham-Soliar's decay experiments. What carries the case is the combination: a 2017 re-examination showing that the original evidence included preparation marks and sedimentary features, and the 2015 observation that collagen-like fibres do not appear on the feathered birds or the hundreds of choristoderes from the same beds.⁵⁵⁵⁶ The amber tail of 2016 shows real feathers on a long-tailed theropod but does not test the simple filaments directly.
• Creationist paleontologists who have worked through the fossils accept the feathers. McLain, Petrone and Speights, in 2018: "many species of dinosaurs were indisputably feathered."⁵
• The melanosome-versus-microbe question is genuinely open, and this article cites Moyer rather than pretending the colour work is secure.
• A creationist article that still asserts "feathered dinosaurs are just collagen" in 2026 is citing a position that creationist paleontologists have themselves dropped, and that lost on new physical evidence rather than on argument. Say so, and say who did it.

Part 7 — The digit problem, and what a good ad hoc rescue looks like

This is the best case study in the whole subject, and it gets real space here, because it is what a genuine anomaly looks like when it is handled well.  A creationist reader who has only ever been told "evolutionists explain everything away" needs to see one worked through honestly — including the part where the explanation was invented specifically to save the theory, and then went on to make predictions that were tested.

The conflict, stated straight

A generalised tetrapod hand has five digits, numbered I to V from the thumb.  Birds have three.  Which three?

Paleontology says I, II, III.  The theropod fossil record shows the outer two fingers shrinking and disappearing:  five in Herrerasaurus, four functional in Coelophysis, three from Allosaurus onward.  The bones themselves say which.  Günter Wagner and Jacques Gauthier:  "Archaeopteryx… has a three-fingered hand with a phalangeal formula (2-3-4) diagnostic of the inner three digits (DI, DII, DIII)."⁵⁹  That formula goes back to Tulerpeton in the Devonian.

Embryology says II, III, IV.  In a chick embryo the cartilage condensations that become the wing's fingers sit in the second, third and fourth positions of the digital arch.  The first position never forms a digit.  This goes back to Morse in the nineteenth century, and runs through Hinchliffe's work in the 1980s and Pere Alberch's studies of the primary limb axis.  Wagner and Gauthier accept the embryology as correct:  "We agree that Hinchliffe correctly identified the posterior hand margin."⁵⁹

If the embryology is right, the bird hand is not the theropod hand.  Alan Feduccia used exactly this to argue that birds are not theropods.  That is a real, hard, empirical conflict between two reliable methods, and it stood unresolved for over a century.

The hypothesis, and the fact that it was ad hoc

Wagner and Gauthier published their answer in PNAS in 1999, under a title that states it as an equation:  "1,2,3 = 2,3,4."⁵⁹

Their proposal is that both datasets are right, because the position a digit grows in and the identity it develops are separable things.  In the theropod lineage, a functional demand to keep the inner three fingers collided with a developmental rule that deletes the last-formed condensation, which is the first one.  In their words:  "we postulate a frame shift—a homeotic transformation—in the developmental identity of the initial condensations: Condensation CII developed into digit DI, CIII developed into DII, and CIV developed into DIII."⁵⁹

State plainly that this was invented to rescue a threatened conclusion.  It was.  This article does not pretend otherwise, and a skeptical reader would smell it immediately if it did.  Wagner and Gauthier say in the paper that removing birds from Theropoda is the alternative solution, and reject it because the theropod result survives deleting hand characters from the matrix entirely.

Then state, just as plainly, what makes an ad hoc rescue scientifically legitimate rather than a dodge:  it must have independent support, and it must make risky predictions.  Theirs had both on the day it was published.

Independent support already in hand in 1999.  Inject a mitosis inhibitor into an alligator egg and condensation CI is deleted.  The finger that then sits in front takes on a mixture of thumb and index-finger features — three phalanges like digit II, but a metacarpal shorter than the next one and a proximal joint identical to the wild-type first finger.  Kiwis, which have two fingers, sometimes show the phalanx count and shape of digit I on the next digit back.  Wagner and Gauthier's conclusion from those two cases:  "Natural variation of this sort demonstrates unambiguously that there is no one-to-one relationship between digit identity (D) and condensation identity (C)."⁵⁹

The mechanism is not exotic either.  Homeotic transformation — a structure keeping its position but changing what it becomes — is the standard explanation for fruit-fly segment-identity mutants, and Ann Burke had already shown that birds and mammals express the same Hox combinations at the same body regions despite building them from different somites.

The risky prediction.  If digit identity is assigned by molecular signals rather than by position, then the anterior wing digit — which grows in position 2 — must express the molecular signature of digit 1.  That is a falsifiable claim about gene expression, made in 1999, before the data existed.

How the prediction fared — the twenty-year test

Read the last row carefully.  Günter Wagner is a co-author on the paper that cuts his own 1999 hypothesis down to a single digit.  The 2019 answer is not 1,2,3 and it is not 2,3,4.  It is 1,3,4 — which nobody proposed in 1999.

Xu Xing and Susan Mackem's review states the unresolved position that is still roughly correct:  homeotic transformation is the likely driver, but is "more likely to have occurred in a partial and piecemeal manner," and recent mouse genetics invite consideration of new alternatives — I-II-IV or I-III-IV — for the homologies of the avian wing digits.⁶⁷

Verdict

• The problem was real and the creationist complaint was legitimate. Do not write it off. For a century two good methods gave two different answers about the most basic anatomy of the most-studied transition in vertebrate paleontology.
• The frame-shift hypothesis was ad hoc when proposed, and it was also testable, and it was tested. It made a specific prediction about gene expression in 1999. That prediction was confirmed for the anterior digit in 2005, 2011 and 2014, by three independent labs using three different techniques — Hox expression, cell-lineage tracing, and whole-transcriptome comparison.
• It has also been partially falsified, by its own author's group. The full shift is not supported. Only the anterior digit shifted, and the posterior wing digits match no model proposed before 2019.
• What this kills: the creationist claim that the digit conflict shows birds cannot be theropods. The conflict now has a documented developmental mechanism that can be induced experimentally in a chick wing bud. Whatever the final numbering turns out to be, the position of the wing digits is no longer evidence against theropod ancestry.
• What survives for our side, and it is worth having: after twenty-seven years, four experimental approaches and the field's leading laboratory, the field still cannot say with confidence which fingers a chicken has. That is a fair, sourced, quantitative statement about how much confidence morphological homology deserves — which is the theme of this whole series. Make that argument. Do not make the stronger one.

Part 8 — Archaeoraptor, told accurately

This is the creationist movement's favourite item in the whole subject, and almost every retelling gets at least one thing wrong.  The accurate version is better for a careful apologist than the exaggerated one, because the accurate version cannot be dismantled by a hostile reader in thirty seconds.

The account here comes from Storrs Olson's own three-page article in Backbone, the newsletter of the Smithsonian's Department of Vertebrate Zoology, published in April 2000.¹⁰  That is Olson in his own words, in a Smithsonian publication, not relayed through an apologetics site.

The sequence

1.  The specimen was smuggled out of China and bought on the commercial market.  Olson:  "By Geographic's own admission, the specimen was illicit, having been illegally exported from China."¹⁰

2.  The name was published by a magazine staff writer.  Christopher Sloan's article "Feathers for T. rex?" ran in National Geographic in November 1999.  Olson records the nomenclatural consequence:  the binomial Archaeoraptor liaoningensis "had never been previously published and thus became available for purposes of nomenclature as of its appearance in National Geographic, with Sloan as its author."¹⁰  The animal was named, formally and permanently, by a journalist.

3.  It was never peer-reviewed.  This follows from point 2 and nobody disputes it.

4.  It was rejected by Nature and then by Science before the magazine ran it.  This is the standard account, and this article believes it — but says openly that it could not be verified from a primary source.  The usual citations are Lewis Simons's October 2000 National Geographic piece and Rex Dalton's news report in Nature in 2000, and neither could be obtained.⁶⁸  What is established, and sufficient, is that the specimen was never peer-reviewed and was named by a magazine staff writer.

5.  Paleontologists knew before publication.  Olson, on drafting his warning letter, was amazed to hear from colleagues "that it was rather common knowledge at the October meeting of Society of Vertebrate Paleontology that the specimen of Archaeoraptor was probably a fake, a composite."¹⁰  He had no direct evidence, so he kept that out of his open letter and put it in a confidential cover memo instead.

6.  Olson's open letter went out on 1 November 1999, addressed to Peter Raven and copied without restriction to colleagues in avian paleontology.¹⁰

7.  Xu Xing caught it.  Olson, writing in April 2000:  "in December, Chinese scientist Xing Xu informed U.S. colleagues that he had found the counterpart of the tail of Archaeoraptor still associated with the partial skeleton of a small dromaeosaurid dinosaur."¹⁰

Weigh that sentence carefully.  It is Olson — one of the theropod hypothesis's fiercest opponents — crediting Xu Xing, one of its leading proponents, with catching the forgery.  A hostile witness, in print, at the time.

8.  A panel confirmed it on 4 April 2000.  It met at the National Museum of Natural History and "unanimously concluded that the specimen of Archaeoraptor is indeed a composite, most of the skeleton being that of a bird."¹⁰  The specimen was signed back over to China that same morning, as a condition of the viewing.

9.  A CT scan showed how it was built.  Timothy Rowe and colleagues, in Nature in 2001, used high-resolution X-ray computed tomography to determine the nature and extent of the forgery "as well as how it was built, by imaging the fracture pattern and distribution of materials through the entire specimen."⁶⁹

10.  The pieces were then described properly, as separate real animals.  The tail belongs to Microraptor zhaoianus — which turned out to be the first mature non-avian dinosaur smaller than Archaeopteryx, and a genuinely important fossil.⁷⁰  The front half was identified as the bird Yanornis martini.⁷¹  Olson himself designated the five tail fragments as the lectotype, so that the name would attach to the dinosaur and be removed from the literature of avian paleontology.¹⁰

What the case does and does not prove

It proves that the commercial fossil trade in Liaoning produces forgeries, and that a popular magazine can be induced to run one.  Both are true and both matter.

It does not prove that the peer-reviewed literature was fooled, because the peer-reviewed literature never published it.  The claim "Archaeoraptor shows how evolutionary science works" fails at the first step.  The mechanism that caught the forgery was ordinary scientific scrutiny, applied within four months, by an author of the very hypothesis the fake was supposed to support.  A creationist who runs this case as evidence of institutional fraud has chosen the one example where the institution worked.

The honest way to use it is as evidence about the market, not the science.  A smuggled slab, an eighty-thousand-dollar purchase, a magazine deadline that outran peer review, and a museum with a financial stake in the specimen.  That is a real story about incentives and it is fully documented.

And Olson's own testimony against National Geographic is worth printing, because it is primary and devastating on its own terms.  Sloan's article was, in Olson's words, "a model of sensationalistic, tabloid journalism containing practically no information that was true, verifiable, or anything more than the wildest conjecture."¹⁰  Olson also records the magazine's response to his warning — its editor dismissed him as a known opponent of the bird-dinosaur theory — and Olson's dry reply that the only time he had ever taken a stance on the issue was in his letter to the magazine.¹⁰

Part 9 — Two men our side misrepresents

Storrs Olson, Alan Feduccia and John Ruben are the three names a creationist article on this subject almost always cites.  All three are evolutionary biologists.  None of the three is a creationist.  All three reject the theropod origin of birds, and none of them rejects the origin of birds.

That should be said once, plainly, and then they should be used properly.  Two of them can now be corrected from primary sources.

Storrs Olson

Olson was Curator of Birds at the Smithsonian's National Museum of Natural History.  He is quoted on creationist sites for one sentence from his open letter of 1 November 1999, which accuses a group of scientists and editors of promoting the dinosaur origin of birds beyond the evidence.  That sentence is not printed in this article.  Every copy of it that could be located was on an apologetics site, and the letter itself could not be found.¹¹  An unverifiable quotation is not evidence, whichever side it helps.

What can be verified is better, and it is the passage in Part 8:  Olson, writing in the Smithsonian's own newsletter in April 2000, credits Xu Xing with catching the forgery.¹⁰  The man cited as the movement's hostile witness against the theropod hypothesis is on record crediting that hypothesis's leading proponent with the correction.

One further caution, and it cuts our way.  Olson's Backbone piece contains the sweeping claim that as of April 2000 "there is absolutely no uncontested evidence of any dinosaur having feathers."¹⁰  Those are his words and he meant them.  They have been overtaken by twenty-six years of evidence, including the melanosome and amber material in Part 6.  A 2000 quotation is evidence about the year 2000.  This site applies that rule to its own witnesses, and the sentence should never be quoted without its date.

Alan Feduccia

Alan Feduccia is S. K. Heninger Distinguished Professor Emeritus of Biology at the University of North Carolina at Chapel Hill, an ornithologist and evolutionary biologist, and the author of The Age of Birds, The Origin and Evolution of Birds, Riddle of the Feathered Dragons and Romancing the Birds and Dinosaurs.  He reviews books for Science.  He is still publishing: his paper on the avian acetabulum appeared in Diversity in December 2023.⁷²

His position is not that birds did not evolve.  It is that birds did not evolve from derived maniraptoran theropods, and that they came instead from an earlier, non-dinosaurian archosaur — a climbing Triassic reptile.  He further proposes the "neoflightless" hypothesis:  that animals such as Microraptor and the troodontids are not dinosaurs on the way to becoming birds, but secondarily flightless birds, descendants of flying ancestors that lost flight.

His current argument is anatomical and specific.  From the 2024 abstract:  hip anatomy "differs substantially between dinosaurs and birds… Given new evidence presented here, the relationship of 'dinosaurs' and birds must be re-evaluated."⁷²  Note the framing — re-evaluated, not abandoned — and note that the evidence offered is a named structure, not a general complaint.

And here is the sentence every creationist article that cites him should have to print.  Asked directly by Discover in 2003 how he felt about creationists using the bird-dinosaur dispute, Feduccia answered:

"Creationists are going to distort whatever arguments come up, and they've put me in company with luminaries like Stephen Jay Gould, so it doesn't bother me a bit.  Archaeopteryx is half reptile and half bird any way you cut the deck, and so it is a Rosetta stone for evolution, whether it is related to dinosaurs or not."³

He continues, in the same answer, that creationists are confusing an argument about minor details of evolution with the indisputable fact of evolution.³

Feduccia is the creationist movement's most-cited authority on Archaeopteryx, and Feduccia says Archaeopteryx is half reptile and half bird and a Rosetta stone for evolution.  There is no honest way around that, and this article does not attempt one.

What use of Feduccia is legitimate

• Legitimate: citing him for the specific anatomical objections he actually raises — the acetabulum and antitrochanter, the digit conflict, the collagen reading of Sinosauropteryx, the neoflightless reading of Microraptor. These are real technical claims in the peer-reviewed literature.
• Legitimate: citing him to show that published dissent from the consensus exists, against the claim that no qualified scientist doubts it.
• Illegitimate: citing him as evidence against evolution, or against Archaeopteryx being transitional. He says the opposite, in print, in a mainstream magazine, in answer to a direct question about creationists.
• Illegitimate: citing him as evidence that Archaeopteryx is "just a bird." Feduccia's own alternative requires a reptile-to-bird transition. He only relocates it to a different reptile.

How his opponents characterise him, and how far that is fair

Richard Prum's rebuttal in The Auk describes Feduccia's alternative as "so vague as to be literally untestable," and accuses him of framing the field as a vast cladistic conspiracy to deceive the ornithological community.³²

The first charge has force, and a creationist should feel it before relying on Feduccia.  Feduccia has never named the Triassic archosaur he thinks birds came from, never produced a fossil for it, and never published a cladistic analysis recovering it.  A hypothesis with no candidate ancestor and no analysis is not in a strong position to displace one with both.

The second charge is a fair description of Feduccia's rhetoric — he titled a book Forays in Postmodern Paleontology — but it is not a scientific refutation, and this article does not use it as one.

Verdict:  Feduccia is not an ally.  He is a hostile witness whose testimony is useful on narrow points.  Present him as what he is:  a mainstream evolutionary biologist who thinks the field has picked the wrong reptile, who has been quote-mined for thirty years, and who has said in print that he does not mind, because the quotations do not mean what the quoters think they mean.  An article that represents him accurately gains more credibility than it loses evidence.

Part 10 — Where the gap actually is

Everything above concedes ground.  This section is where the ground is worth holding, and it is the section nobody on either side has worked properly.

What the creationist statistics do and do not say about this boundary

Part 1 showed where the creationist statistics put their gaps:  between most birds and the dromaeosaurs and troodontids, and inside the Mesozoic birds, where the ornithurine line that leads to modern birds separates from the enantiornithines and the older birds.⁴⁵  None of those gaps sits at the end of the Cretaceous.  The ornithurine cluster is full of Cretaceous animals, and in Chiappe's matrix the living duck Anas correlates positively with the Cretaceous Ichthyornis.⁴

So the creationist statistics do not point at the boundary this section examines.  The case below rests on superposition and on which birds lie above and below one layer, not on baraminology.  It still deserves to be worked, and hardly anyone has worked it.

What the fossil record says at that boundary

The best-constrained data on it come from Nicholas Longrich, Tim Tokaryk and Daniel Field, in PNAS in 2011.⁷³  They described the bird fauna of the very top of the Cretaceous in western North America — the Hell Creek Formation of Montana and the Dakotas, the Lance Formation of Wyoming, and the Frenchman Formation of Saskatchewan.

Their findings, in their own numbers:

• Seventeen bird species in the assemblage, the most diverse known from the Late Cretaceous.
• Seven of the seventeen (41 per cent) are archaic birds — members of Enantiornithes, Hesperornithes, Ichthyornithes and Palintropiformes, four lineages entirely outside the modern bird crown group.⁷³
• None of the seventeen could be definitively referred to Neornithes, the crown group that contains every living bird. In their words: "we could not definitively refer any of these fossils to the avian crown; thus claims of a major neornithine radiation in the Cretaceous are not at present supported by the fossil record."⁷³
• Exactly one of the seventeen is known above the boundary. Their unnamed "Ornithurine C" is the only Maastrichtian bird in the assemblage known to cross into the Paleocene.⁷³
• No archaic bird has ever been reported from the Paleogene, and their own examination of North American Paleocene material found none.⁷³

That is a sharp, checkable pattern.  Below the boundary:  a diverse fauna dominated by animals with no living descendants.  Above it:  none of them, and a radiation of modern bird orders that appears in the Paleocene and Eocene.

Why this argument needs only superposition

This argument rests on the simplest kind of order there is.  The Hell Creek, Lance and Frenchman formations sit directly below the boundary clay, and the Fort Union and equivalent beds sit directly above it, in the same outcrops, in the same basin.  The claim "these seven lineages occur below that layer and never above it" is superposition in a single stratigraphic column.  It needs no matching of beds between basins or continents, so a skeptic cannot answer it by disputing a correlation, and a young-earth reader can state it without relying on anyone's timescale.

That makes this the strongest line in the whole subject for a young-earth writer, and it has been almost entirely ignored in favour of arguing about a single German fossil.

What the argument is, and what it is not

What can be said:  the Mesozoic avialans and the modern birds are two faunas, not one graded series.  The animals below the boundary layer are not the animals above it.  And identifiable members of most modern bird orders first appear above the boundary, not below it.⁷³

What cannot be said, and this article will not say it.  Four honest limits, each of which a skeptic will raise.

1.  Longrich's own explanation is a mass extinction, not a creation event.  His conclusion:  the evidence is consistent with a major extinction of archaic birds at the boundary, "which may have provided an ecological release, permitting the radiation of modern birds in the Paleogene."⁷³  That is a complete naturalistic account of the same pattern, and a creationist has to answer it rather than ignore it.

2.  The record is thin and the authors say so.  Most of these fossils are isolated bones — coracoids, mostly.  Species have been erected on non-overlapping elements, so some may have been named more than once, and the material was never subjected to phylogenetic analysis before Longrich's paper.⁷³  A gap defined by a fauna of isolated shoulder bones is a gap that could close.

3.  The pattern outside North America is not tightly constrained.  Longrich says this against his own thesis:  outside North America only a handful of archaic birds can be pinned to the last half of the Maastrichtian, and the lack of temporal control makes it hard to be certain that the extinction was abrupt.⁷³  He predicts the global pattern will match.  A prediction is not a result.

4.  One bird does cross, and one modern bird is Cretaceous.  Ornithurine C crosses the boundary.  And Vegavis, from Antarctica, is on Longrich's own account the one definitive Cretaceous member of the modern crown group — a waterfowl relative.⁷³  So the boundary is not a clean wall even in the mainstream account, and any creationist argument built on a clean wall is building on something the data do not supply.

The honest closing position

The creationist case on bird origins is not where the creationist literature has put it.

It is not Archaeopteryx.  Both sides' quantitative methods put that animal on the border between the raptors and the birds, and the largest creationist study says it cannot tell which side it belongs on.⁵

It is in two places.  The first is the gap the creationist statistics keep finding between most birds and the raptors, with more gaps inside the Mesozoic birds.  That gap is real on the method's own terms, small on its authors' own account, and open to the reply that gaps in fossil samples tend to fill.⁵  The second is the bird turnover at the end of the Cretaceous:  a place where the fossil record shows a fauna replaced rather than transformed, and where the argument can be made from superposition in one basin, with no correlation between continents for anyone to dispute.

And both are unfinished.  Nobody has run baraminic distance correlation across the end-Cretaceous boundary with a real sample of Paleogene and living birds;  the nearest thing, one living duck in Chiappe's matrix, groups with the Cretaceous Ichthyornis.  Nobody has published the specimen-level completeness statistics for that interval.  Nobody on our side has engaged Longrich's ecological-release explanation on its merits.  Until that work is done, the honest young-earth position is that there is a gap worth investigating here, and the investigation has not been done.

Saying that is more useful than another article defending Archaeopteryx.

Part 11 — Objections, one at a time

"Archaeopteryx is just a bird."

Retire it.  Under the creationists' own statistical method it sits on the border:  with the raptors in four of Garner, Wood and Ross's six matrices and with the birds in two, and the authors of the largest creationist study "still cannot say with certainty" which side it belongs on.⁴⁵  Garner's team wrote that it "may be a kind of dinosaur."²  The defensible creationist claim is narrower:  the creationist statistics find gaps between most birds and the raptors, with Archaeopteryx on the border.  It also has teeth, a long bony tail, clawed fingers and a dinosaurian palate.  Calling it a bird is a classification decision, not an argument, and it concedes everything the other side needs.

"Even evolutionists cannot agree whether Archaeopteryx is a bird or a dinosaur."

True, and the true version is more useful than the exaggerated one.  The margin is one parsimony step in a 700-character analysis, and the papers say so themselves.⁷  But the argument only supports a claim about classification, not about anatomy.  The animal's character combination is identical under every topology.  Do not slide from one to the other; a competent reader will notice.

"The supposed ancestors are younger than the descendant."

Dead on the evidence as framed.  An anchiornithid, Ostromia, comes from Archaeopteryx's own level in the Bavarian lagoons, a Morrison troodontid sits at about the same level, and Xu Xing's group wrote in Nature that the paradox is refuted.³¹²¹  Nine Chinese paravians are placed below the Solnhofen level, but that placement leans partly on radiometric dates, so a young-earth writer may say no close relative is securely older on fossil correlation alone.  The argument matches fossil order between continents, which a young-earth writer may do, but the same matching that puts Deinonychus above Archaeopteryx puts the Chinese paravians below it.  A writer who calls that matching circular cannot use the argument at all.  See Part 4.

"There is no functional intermediate between a bellows lung and a flow-through lung."

Refuted by direct measurement in three living animals between 2010 and 2014 — an alligator with no air sacs, a monitor lizard outside Archosauria, and an iguana with no parabronchi.⁴⁴⁴⁵⁴⁶  The replacement argument, which is better, is Schachner and Moore's 2025 caution that a fully avian lung should not be reconstructed in non-avian dinosaurs.⁸

"Feathered dinosaurs are just degraded collagen."

Lost on the weight of the evidence.  Melanosomes were reported inside the filaments in 2010, though Lingham-Soliar disputes them.  Collagen-like fibres do not appear on the feathered birds or the hundreds of choristoderes from the same beds.  A 2017 re-examination found preparation marks and sedimentary features read as fibres.^52545556  Creationist paleontologists now call many dinosaurs "indisputably feathered."⁵  Do not lean on the amber tail of 2016 here:  it shows real vaned feathers, not what the simple filaments are.

"Bird wing digits are II, III, IV, so birds cannot be theropods."

The conflict was real for a century.  It now has a documented developmental mechanism, and that mechanism has been induced experimentally in a chick wing bud.⁶⁶  What survives is weaker and still worth saying:  after twenty-seven years the field cannot agree on the numbering, and the 2019 answer of 1,3,4 was proposed by nobody in 1999.⁹

"Archaeoraptor shows that peer review failed."

It shows the opposite.  It was never peer-reviewed, it was named by a magazine staff writer, and it was exposed within four months by Xu Xing — a leading proponent of the very hypothesis it was supposed to support, credited by Storrs Olson at the time.¹⁰¹⁰

"Feduccia and Olson show that scientists doubt evolution."

They show that scientists doubt the theropod origin of birds.  Both are evolutionary biologists.  Feduccia calls Archaeopteryx "a Rosetta stone for evolution" and says creationists are confusing minor details with the fact of evolution.³³

"The Chinese fossils come from a commercial market that produces fakes, so they cannot be trusted."

Half true, and the half that is true is worth keeping.  The Liaoning trade does produce forgeries, and Archaeoraptor is proof.  But the inference does not go through.  Forgeries in this material have been caught — by CT scanning, by finding counterpart slabs, by ordinary examination — and the specimens that matter here have been examined by multiple independent teams and, in the amber case, imaged in three dimensions inside a transparent medium that cannot be assembled from pieces.  Suspicion is not evidence.  Name a specific specimen and a specific reason, or drop the argument.

"Only fourteen specimens of Archaeopteryx exist, one is lost, and several are privately owned."

All true, all printed in Part 2, and it is a real limit on the evidential base.  But it is a problem of custody and sample size, not of authenticity, and no Archaeopteryx specimen has ever been shown to be a forgery.  The argument that can be made from it is about how much any one fossil should be asked to prove — and Part 3 makes that argument with a number instead of an insinuation.

What would change this verdict

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

• A baraminic distance correlation analysis spanning the latest-Cretaceous ornithurines and a real sample of Paleogene and living neornithines, using a published matrix, with the method stated in advance. Nobody has run one. The nearest thing, Chiappe's matrix in Garner's study, has one living bird, the duck Anas, and it groups with the Cretaceous Ichthyornis.⁴ A real test would show whether the creationist method finds any gap at this boundary at all.
• A published answer to Longrich's ecological-release explanation of the K–Pg bird turnover, on its merits, from the creationist side. There is none.
• A demonstration that the well-preserved Archaeopteryx specimens are not anatomical mosaics — that the teeth, the long bony tail or the asymmetric flight feathers have been misread. This would have to survive the uncrushed Chicago specimen, which is where the anatomy is now clearest.
• Evidence for the horizontal septum in a non-avian dinosaur, which would settle the lung question in the mainstream's favour and retire the one surviving version of that objection. Schachner and Moore say it is not there yet.⁸
• A settled answer on avian digit identity, from whichever direction. If the numbering resolves to 1,2,3 the frame shift is vindicated; if it resolves to II,III,IV the fossil and embryological data are back in conflict. Either result is more informative than the present one.
• A stratigraphic inversion established from superposition rather than from a date — an archaic bird above the boundary clay in the same column, or a neornithine well below the latest Maastrichtian in a section that can be walked.

Sources

The research dossier behind this article — the full specimen table, the four baraminology runs, every objection assessed, and a 51-row quotation ledger with a verdict on each quotation including the four that failed — is published alongside it:  research/bird-origins.md.

A note on what is deliberately absent.  Four quotations that appear routinely in creationist writing on this subject are not printed anywhere above, because no source carrying them could be opened:  John Ruben's diaphragmatic-hernia sentence, the much-repeated line from Storrs Olson's open letter of 1 November 1999, a one-word jibe attributed to Alan Feduccia in a 1993 Science news piece, and the wording of the 2013 Nature retraction notice.  Verified substitutes from the same men and the same events are used instead.  This is the same standard this series applies to mainstream sources, applied to our own.