Whales and the Land-to-Sea Transition

"At present creation theory has no good explanation for the fossil record of whales." — Kurt Wise, 1995, a young-earth creationist with a Harvard paleontology PhD, writing in a creationist journal.¹

"Thus it is questionable whether Dorudon and Basilosaurus were tail-powered swimmers." — Philip Gingerich, 2019, who collected most of the specimens under discussion.²

"When a representative skeleton is found, it will indicate how pakicetids swam and how they should be exhibited." — Gingerich again, in a 2006 conference abstract subtitled "Caveat emptor!"³

Summary

Whales are mammals.  They breathe air, nurse their young, and carry finger bones inside their flippers.  The mainstream claim is that they descend from a four-legged land mammal, and that a run of fossils from Pakistan, India, Egypt and the American South records the change from a deer-like animal on a riverbank to a fully aquatic one.

This article concludes that the pattern is real, that the usual creationist rebuttals to it fail, and that the honest young-earth position is that this is an unsolved problem rather than a solved one.

That is not the conclusion this site would prefer.  It is the conclusion the evidence forces.  Two earlier articles in this series concluded that a young-earth argument was simply wrong and said so.⁴⁵  This one concludes something slightly different and worse:  that the young-earth argument here is unfinished.

The best statement of the difficulty is not an evolutionist's.  It is Kurt Wise's, in a creationist journal, in 1995.  Wise took his doctorate in paleontology at Harvard, where he studied under Stephen Jay Gould.⁶⁶⁷⁷  He wrote that "the largest challenge from the stratomorphic intermediate record appears to this author to come from the fossil record of the whales," and that creation theory "has no good explanation" for it.¹¹

What this article concludes

• The sequence is a real pattern, not an arrangement. It runs in stratigraphic order and in morphological order at the same time, and Wise concedes that in more detail than any evolutionist would need us to.
• Wise's factual premise checks out. He claimed in 1995 that whales occur only in Cenozoic rock. A query against the Paleobiology Database run for this article returned 5,986 catalogued cetacean occurrences and zero older than 66 million years.⁸ Confirming him makes the young-earth problem worse, not better, and it is printed here anyway.
• The transitional part of the record is genuinely thin, and mainstream sources say so best. Philip Gingerich's own 2003 table grades thirty archaeocete genera on what is actually in the ground. Two are "virtually complete skeleton," and both are at the fully aquatic end.⁹
• Reconstruction practice has repeatedly outrun the bones, and every case in Part 3 is documented by the describers themselves rather than dug out by critics.
• The current mainstream position on the whale tail is more favourable to the creationist reading than the version apologists are using. In 2019 Gingerich measured fluke proportions across thirty-six modern whales and put a question mark over whether Dorudon and Basilosaurus had working flukes at all.²² Apologetics has not registered this. Part 4 is about it.
• The sequence crosses a taxonomic line that young-earth biology draws for itself, at the family level, in its own journal.¹⁰ It passes through six families and two suborders.
• Two arguments our own side uses here are dead, and this article kills them rather than quietly dropping them. The rate argument fails on a number computed by the field's leading collector. The waiting-time argument is not refuted so much as unformulated, and the one measured input it has points the wrong way.

What this article rests on

Everything factual below comes from mainstream sources, and where a number could be computed rather than quoted, it was computed against the public databases.⁸

Radiometric and absolute dating are not used as evidence anywhere in this article.  That is a constraint across this series.  Every argument here works from relative order:  what sits above what, and the matching of fossil order between continents by the fossils themselves.  Young-earth geologists accept both and use both;  the Institute for Creation Research's own model rests on "the same relative order of fossils in the Tertiary sediments across all continents."¹¹  Where an age in years appears, it is the mainstream's attributed figure, not something a reader is being asked to believe.

Every quotation was checked against a source that was actually opened.  The research dossier behind this article audits thirty-one passages with verdicts, including the ones that failed.¹²  Of the ten passages in this project's own notes that turned out to be altered or misread, eight had moved in the direction that made the young-earth argument look stronger.  That is the reason the audit exists.

The rank question, stated up front

Young-earth biology does not deny descent with modification.  It denies descent across a boundary it locates at roughly the family level.  So the only question that matters about any fossil sequence is whether it crosses that line.

The line is drawn in the young-earth side's own journal.  Answers Research Journal, in the standard paper on identifying created kinds:  "researchers have suggested that the level of the baramin tends to fall at or near the taxonomic level of family."¹⁰

Wise applies the same rule to exactly this material.  Reviewing the Cenozoic mammal series, he writes that where "the within-group morphology differences are within taxonomic families, it is likely that the differences are intra-baraminic in every case."¹  That is a real concession, and it is conditional on staying inside a family.

The cetacean sequence does not stay inside a family.  It runs from a terrestrial artiodactyl family through five successive archaeocete families to two living suborders.  Cetacea as a whole holds eighty-five valid family-rank names.⁸  Under the traditional classification, whales and even-toed hoofed mammals were separate mammalian orders.  Part 6 gives the count, concedes the fair objection that Linnaean ranks above genus are conventions, and answers it with anatomy instead of labels.  It also reports the young-earth side's own statistical tests of the archaeocetes, which found them a separate group from both hoofed mammals and living whales, and the genome data that now cut against those tests.

What this article concedes before it argues

The genuine weaknesses on our own side, stated before the case is made.

• "Whales appear abruptly with no transitional forms" is false, and the person who says so most plainly is Wise. The defensible claim is about density, not existence.
• A gap named in print was filled. In 2003 Gingerich wrote that the step from Rodhocetus to Dorudon was the larger of the two and that the intermediate morphologies were missing.⁹ In 2019 he described Aegicetus gehennae from that gap, with the intermediate anatomy predicted for it.² That is the fossil record behaving the way its defenders say it should, and an audit that omits it is not an audit.
• A biomechanical model built from living mammals agrees with Ambulocetus, but it has not been shown to be a prediction made in advance. Frank Fish's 1996 model has whale swimming pass through an otter-like stage of hind-foot paddling and spinal undulation before a tail fluke appeared, and Ambulocetus's limb proportions match modern river otters. Thewissen later called this a prediction written before Ambulocetus was discovered. The 1996 paper he cites was not: it cites the 1994 description of Ambulocetus and uses the fossil as evidence. An earlier draft of this article repeated Thewissen's wording without checking the dates.¹³¹⁴
• Morphology-only phylogeny was wrong about whales for thirty-five years and the fossils corrected it. From 1966 until 2001 the fossil evidence pointed at an extinct group called mesonychids. Ankle bones found in 2000 moved whales inside the even-toed hoofed mammals, where the molecular data had been putting them all along.¹⁵ That cuts both ways and this article prints both ways.
• A low fossil count is not evidence that animals did not exist. The Paleobiology Database records where people have dug and published. Pakistan, India and Egypt are not the United States, which alone holds 22.4% of its catalogued collections.¹⁶ Part of the thinness of the early sequence is a map of field effort.

And two arguments this article retires from our own side

The rate argument — "there was not enough time" — is dead, and it dies on a mainstream number.

Gingerich computed the rate of change across the whale transition himself, in Paleobiology, in 2003, in the standard units of the field.  His result is a rate about ten times higher than the background expectation for that timescale.⁹  Elevated, and he says the result contradicted his own prior expectation.⁹  But "ten times background" is not "impossible."  It is roughly 10⁻⁵ standard deviations of change per generation, four orders of magnitude below what selection demonstrably produces in living populations.

And the first pass of this project's own research compared it against the wrong baseline, in a way that flattered the evolutionary side.  Correcting the mistake makes the transition look faster:  six to sixteen times background on Gingerich's fourteen univariate measures, and forty-six to a hundred and twenty-six times on his multivariate ones.⁹  This article prints the corrected number, which is the less convenient one for the position it started from, and still concludes the rate argument fails.  An article that leads with "there was not enough time" and then meets Gingerich's own arithmetic from a hostile reader has lost.

The waiting-time argument is retired, for the same reason the companion article on the mammal ear retired its own, plus two more.⁴

Nobody has counted the coordinated genetic changes the transition required.  Without that count, any waiting time is manufactured.  Beyond that:  this project's own notes understated the time available per change by a factor of about 6.6, by applying a modern great whale's twenty-seven-year generation time to a wolf-sized Eocene mammal.  And the one measured count of genomic change in the cetacean stem lineage — eighty-five genes inactivated, from a 2019 genome screen — falls squarely in the cheap, single-mutation category of the very population-genetics framework the argument borrows from.¹⁷  Part 8 shows the arithmetic in both directions.

Part 1 — The best statement of the problem is a creationist's

Kurt Wise is the young-earth author with the strongest technical credentials on this exact subject.  In 1995 he published an assessment of transitional fossils in a creationist journal, and the section on whales is the most candid paragraph either side has produced.¹

Verbatim, from p. 219:

At this point in time, the largest challenge from the stratomorphic intermediate record appears to this author to come from the fossil record of the whales.

And in the next paragraph:

At present creation theory has no good explanation for the fossil record of whales.

Wise is not conceding evolution.  The same paragraph offers four possible creationist escapes, and they are graded below.  But he states the problem without softening it, and he gives three separate reasons.  Each one is checkable.

His three reasons

First, the order is a stratigraphic series and a morphological series at the same time.  Wise lists it himself.  Ambulocetus with the largest hind legs; Rodhocetus with hind legs about a third smaller; Prozeuglodon with hind legs a few inches long; then whales with virtually none.  Then, among the later whales, toothed baleen-whale relatives before true baleen whales, and squalodontid toothed whales with telescoped skulls and triangular teeth before modern toothed whales with telescoped skulls and conical teeth.  He calls it "a very powerful stratomorphic series."¹

That is the concession that matters.  It is not a claim that some animals resemble others.  It is a claim that the morphological ordering and the stratigraphic ordering are the same ordering, made by a creationist who had every reason to find a way around it.

Second, the direction is backwards for Flood geology.  The standard young-earth explanation of fossil order is a transgressing Flood, burying communities in order from the deep sea inland.  The whale sequence runs land-to-sea.  Wise's own statement is that the relative order of land mammals, archaeocetes and modern whales is not explainable by transgressing Flood waters.¹

Third, whales are absent from Flood sediments entirely.  Verbatim:

Furthermore, whale fossils are only known in Cenozoic (and thus post-Flood) sediments.

He adds that this runs counter to the intuitive expectation that whales should be found in, or throughout, Flood sediments.¹

The premise checks out, and confirming it makes things worse

Wise made that third claim from the 1995 literature.  It can now be tested against a database that did not exist when he wrote it.

The query run for this article was occs/list.json?base_name=Cetacea&show=coords&limit=100000 against the live Paleobiology Database, re-run on 8 August 2026.  It returned 5,986 catalogued cetacean occurrences.  Zero have a maximum age older than 66 million years.⁸

Two records touch exactly 66 Ma, and both are coarse bin assignments — Basilosaurus entries labelled only "Tertiary" and "Paleogene" — not pre-Cenozoic occurrences.  The oldest resolved records are Himalayacetus subathuensis and Pakicetus attocki, both binned to the Ypresian.

So a claim made from the printed literature in 1995 is confirmed by a 2026 compilation of 5,986 records.  Nobody on either side appears to have run this check before.  It is printed here because it is true, and because the result deepens the young-earth problem rather than easing it.  Whales are entirely restricted to rock that Wise's school of young-earth geology reads as post-Flood, and they arrive there already sorted into a land-to-sea order.

A second young-earth reading puts the whales inside the Flood

Not every young-earth geologist puts the end of the Flood where Wise does.  Wise's school draws the line at the base of the Cenozoic, the Cretaceous-Paleogene boundary.  Timothy Clarey and Jeffrey Tomkins of the Institute for Creation Research draw it near the top of the Cenozoic, at the Neogene-Quaternary boundary.  On their reading every archaeocete lies in Flood rock, buried as the waters drained off the continents.

They state it in their own words.  Creationists who end the Flood at the Cretaceous "have claimed that these large marine mammals evolved rapidly from ancestors that walked out of the Ark," and that "would have required numerous and exceptionally rapid changes in anatomy and physiology – all in only the space of about 200 years or less."  Their alternative is burial in the "late Flood runoff":  "large, likely bloated, and buoyant carcasses of dead marine mammals like whales in Tejas bone beds globally.  These would have been some of the last marine creatures to have been buried in the Flood."¹¹

Their target is real, and it is Wise himself.  By 2009 Wise had offered an answer of his own.  As Clarey quotes him, Wise proposed that whales may have arisen after the Flood and that "vestigial legs and hips in modern whales confirm legged ancestors of the whales existed only a short time ago."¹⁸  That is the author of the 1995 problem statement reading the whales as descended from legged ancestors after the Flood.  The 2009 chapter itself was not available for this article, so its argument is given here only as far as Clarey quotes it.

What the high boundary buys.  It removes the need for whales to descend from a land animal in two centuries, which is the price Wise's reading pays.

What it does not buy.  Both camps agree that the Paleozoic and Mesozoic marine rocks are Flood rock.  Those rocks hold large air-breathing marine predators in quantity:  the same database query style returns 845 mosasaur, 668 ichthyosaur and 1,160 plesiosaur occurrences, 2,673 in all, and none that the database places in Cenozoic rock.⁸  Against that stand 5,986 whale occurrences and not one below the Cenozoic.  The two groups never share a bed in either direction.  A model that buries whales last because their carcasses were "bloated, and buoyant" has to explain why the marine reptiles, which were also large air-breathing animals, all went down earlier.  And it turns the land-to-sea order of the archaeocetes inside the Cenozoic into a burial order that sorts by anatomy, which nobody has modelled.

The two camps refute each other in print.  Marcus Ross and Chris Arment, for the low boundary, answer that the "whale distributions" argument for a high boundary is "far better explained in a post-Flood recolonization setting," and that a high boundary forces the created kind down "from the rank of Family to that of Genus."¹⁹  So the whale problem does not go away under either boundary.  It changes shape:  a speed problem under the low boundary, a sorting problem under the high one.

Wise's four escapes, graded

He offers them as clues that an alternative might be forthcoming, not as answers, and they should be read at that strength.¹

(a) The archaeocetes may be a taxonomic waste-basket of several unrelated created kinds.  Live, tested once, and weaker on the anatomy than in 1995.  Wise's stated ground was that the archaeocete group is not defined by good synapomorphies — that is, by shared derived features that mark a real group rather than a convenience.  That was arguable when he wrote it.  Since then Gingerich and colleagues have supplied a discrete set of ankle characters shared with even-toed hoofed mammals, and the involucrum unites archaeocetes with every later whale (Part 5).¹⁵  A creationist taking this escape now has to argue that those features are convergent, which is a much heavier lift.  It is also the only escape young-earth biologists have put to a quantitative test.  Statistical studies in 2005 and 2006 found the archaeocetes separated from both living whales and hoofed mammals, on small samples that predate most of the fossils in this article (Part 6).

(b) Modern whales retain latent genetic information for hind limbs, so Basilosaurus may be an early post-Flood member of a whale kind.  The premise is real; the conclusion does not follow.  Whale embryos do begin hind-limb buds that then regress, and the developmental cause has been traced to loss of Sonic hedgehog expression in the limb bud.²⁰  Nothing about that makes Basilosaurus post-Flood, and no test is attached to the inference.

(c) Ecological zonation may explain the order, and collection bias may explain the Cenozoic-only distribution.  The bias half is legitimate and partly survives; the zonation half does not work.  The collection-bias point is real and Part 2 supports it in general terms.  But bias does not produce a complete absence across 5,986 occurrences on every continent.  And zonation does not predict a morphological series correlated with the stratigraphic one — which is exactly what Wise concedes exists.

(d) Loch Ness and similar lake sightings may be living Basilosaurus.  Dead, and it is printed here on purpose.  Wise suggests that capturing an actual Basilosaurus would go a long way toward solving the mystery.¹  There is no specimen, no carcass, and no photograph that has survived scrutiny.  Creation Ministries International's own list of arguments creationists should not use retires living plesiosaurs and dinosaurs on the plain ground that enough decades have passed and everyone now carries a camera.²¹

Quoting Wise's honest statement of the problem while suppressing his weakest proposed solution would be exactly the defect this series audits other people for.  Both halves are on the same page of the same paper.  Both are printed.

Part 2 — How much material there actually is

The commonest creationist claim about this sequence is that it rests on scraps.  That claim is testable two ways:  against the collector's own inventory, and against the public occurrence databases.  Both were checked for this article, and the answer is more interesting than either side's version.

The collector's own summary

Philip Gingerich named or co-named more of these animals than anyone else.  His University of Michigan research page summarises what his teams recovered, and it is the single most useful inventory statement in print because it is written by the man who dug them up and is candid about what is missing.²²

His one-sentence summary of the haul:

We have found and collected virtually complete skeletons of middle-to-late Eocene Basilosauridae (Dorudon and Basilosaurus), exceptionally complete skeletons of middle Eocene Protocetidae (especially Rodhocetus and Artiocetus), and a partial skull of earliest middle Eocene Pakicetidae (Pakicetus).

Read the shape of that sentence.  The late, fully aquatic end of the sequence is whole skeletons.  The early, land-mammal end is a partial skull.

Gingerich's genus-by-genus table

In 2003 Gingerich published a table classifying every archaeocete genus then known, with a right-hand column summarising the known skeletal remains of each.⁹  It is the collector grading his own field's material.  Nobody in apologetics on either side appears to have counted the column, and the count is the argument.

Thirty archaeocete genera.  Two are graded "virtually complete skeleton," and both are basilosaurids — the fully aquatic end.  Seven are known from dental remains, a partial dentary, or vertebrae alone.  The remaining twenty-one are "skull and partial skeleton" or less.

Gingerich's abstract states the constraint in one line:⁹

Archaeocete skeletons complete enough to be informative about locomotion are rare (principally Rodhocetus and Dorudon), and these deserve to be studied in comparison to the full spectrum of semiaquatic mammals.

One honesty note about that table, printed rather than smoothed.  His text on the same page names Rodhocetus and Dorudon as the two complete genera, while his table grades Rodhocetus only "skull and partial skeleton" and Basilosaurus "virtually complete."⁹  The text and the table of the same paper do not agree about which two genera are the complete ones.  This article quotes the table, which is the more conservative of the two, and flags the discrepancy rather than picking whichever version suits.

The database counts

The Paleobiology Database records fossil occurrences by taxon and collection.  It is not run by anyone with a stake in this argument.  Every figure below was recomputed for this article on 8 August 2026 against a complete bulk dump of 2,001,053 occurrences.⁸

The whole of the land-to-whale transition is 478 catalogued occurrences.  That is 0.37% of the mammal record and 0.024% of the database.  Cetacea as a whole — every whale, dolphin and porpoise, living and fossil, across fifty million years and every ocean — is 5,986 occurrences, or 4.7% of all mammal occurrences.

Thirty-four of the seventy-two named species in the sequence, 47.2%, rest on exactly one catalogued occurrence from exactly one collection.  For each of those, the species and the specimen are the same thing.

Where the material sits, animal by animal

Occurrence counts for the animals a reader will have heard of:⁸

• Himalayacetus subathuensis, the oldest cetacean — 1
• Pakicetus inachus — 2; Pakicetus attocki — 4; Ichthyolestes pinfoldi — 4
• Ambulocetus natans — 2
• Rodhocetus kasranii — 3; Rodhocetus balochistanensis — 1
• Artiocetus clavis — 1; Georgiacetus vogtlensis — 4
• Peregocetus pacificus, Phiomicetus anubis, Aegicetus gehennae, Tutcetus rayanensis, Perucetus colossus — 1 each
• Dorudon atrox — 52; Basilosaurus cetoides — 42; Basilosaurus isis — 21

Note the shape of that list.  Dorudon atrox alone, at fifty-two occurrences, is 62% of the entire catalogued record of the raoellids, pakicetids, ambulocetids and remingtonocetids put together — eighty-four occurrences across twenty-two species.  Basilosauridae holds 324 of the 478 occurrences in the whole sequence, or 68%, while the two earliest cetacean families between them hold nineteen.  The animals doing the most argumentative work are the ones with the least material behind them.

Four specimens that carry more weight than they can hold

Pakicetus inachus.  The holotype, GSP-UM 084, is described by Gingerich and Russell in 1981 as the "posterior portion of a cranium with the right auditory bulla intact."²³  That is the entire holotype.  The referred material in the original description is one jaw fragment holding three premolars and four isolated molars, one of them a broken crown.  The describers estimated the original skull length at 30 to 35 cm from the partial cranium they had.

Ambulocetus natans.  The holotype is a genuinely associated individual, with skull, vertebral column, one forelimb and parts of both hind limbs.  But Thewissen's own review states the family total flatly:  "Fewer than ten fossils of ambulocetids have been discovered."¹⁴

Phiomicetus anubis, described in 2021 from Egypt and widely covered in the press.  Its describers write:  "The holotype is the only known specimen."²⁴  What was recovered is a cranium, both mandibles, isolated teeth, one cervical and one thoracic vertebra, and ribs.  There are no limbs at all.  Any life restoration of this animal below the neck is inference from other protocetids.

Tutcetus rayanensis, described in 2023 as the smallest known basilosaurid and used to argue about whale life history.  The holotype is a partial cranium, dentaries, hyoid apparatus and the atlas vertebra of a subadult.²⁵  One vertebra, no limbs, and the paper notes the palatal surface of the cranium is mostly destroyed.  A life-history argument resting on one subadult partial skull is thin, and this article says so.

Where the record is genuinely dense

One place.  Wadi Hitan in Egypt — Zeuglodon Valley, about fifty kilometres west of the Fayum oasis.  Gingerich's 1990 paper gives the hard figures:  in the 1987 and 1989 seasons his team "mapped 243 partial skeletons of B. isis and 77 partial skeletons of smaller archaeocetes," all within a ninety-metre-thick stratigraphic section.²⁶

His web page gives a rounder figure — about 400 archaeocete and sirenian skeletons mapped over six seasons.²²  The two are not obviously reconcilable, and the peer-reviewed 1990 pair is the one used here.  Note also that "mapped" is not "collected."

So the record is dense at exactly one locality, at exactly the end of the sequence where the animals were already fully aquatic.

What these numbers do and do not license

They license this.  The transitional part of the cetacean record is sparse in absolute terms and most of its species are singletons.  Anyone calling this sequence "one of the best-documented examples of macroevolution" — the phrase is Thewissen's own¹⁴ — is making a relative claim about how it compares with other origin-of-a-body-plan cases, not a claim about absolute abundance.  Both things are true at once and this article says both.

They do not license this.  A low occurrence count is not evidence that the animals did not exist or are not related.  Three caveats travel with every number above:

1. One occurrence can be a complete articulated skeleton or a single tooth. The database does not record completeness. Rodhocetus balochistanensis has one occurrence and virtually complete fore and hind limbs; Basiloterus hussaini has one occurrence and very little. Occurrence count is a lower bound on knowledge, not a measure of it, and that cuts against this argument as much as for it.
2. The database records where people have dug and published. Apparent diversity tracks collecting effort closely, and the United States alone holds 22.4% of catalogued collections.¹⁶ Balochistan is not Wyoming.
3. New taxa lag. New archaeocete genera were named in 2019, 2021 and 2023, and every one of them has a single catalogued occurrence. A record still producing new families of transitional animals every few years is not a closed book, and an argument from absence in it is weak in both directions.

Gingerich's own verdict on the record, written for a general encyclopedia audience in 2004, is the fair summary, and it has to be printed whole because the second half is the part apologists cut.  He writes that because whales sit at or near the top of the food chain and are therefore rare compared with other animals, "they have a good, but not yet great, fossil record."  His very next sentence says the record is nonetheless "sufficiently dense and continuous" that broad lineages can be traced up and down through successive strata.²⁷  Print one without the other and you have done to Gingerich what this series audits other people for doing.

Part 3 — Reconstructions that outran the bones

This is where the creationist complaint is strongest, and the reason it is strong is that every case in this section is documented by the describing authors themselves.  None of it requires a critic's testimony.  A reader who wants to check it can read the same papers.  The companion article on fossil reconstruction sorts composite mounts, undisclosed reconstructions and outright forgeries from one another across the whole fossil record.²⁸

The Pakicetus skull, and two magazine covers the artist's client disowns

Pakicetus was described in 1981 from the back of a skull and a handful of teeth.  In 1983 it appeared on the cover of Science as a swimming, fish-catching animal.  In 2001 Thewissen's team published skeletons and it appeared in the textbooks as a running land mammal.  Neither picture came from the holotype, because the holotype is a partial cranium.

Gingerich now says both his covers were wrong.  On his own page, of the 1983 Pakicetus and 2001 Rodhocetus cover art:

Based on what we know today, these animals were probably less different than shown here, and the hands and feet reconstructed for Pakicetus probably looked more like those now known for Rodhocetus.

That is the author of both covers saying, in print, on his own site, that both covers are wrong.²²

The pakicetid skeletons are composites, and Thewissen says so

The 2001 Nature paper that produced the running land-mammal Pakicetus worked from H-GSP Locality 62 in the Kala Chitta Hills of Pakistan.  Locality 62 is a bone bed.

Thewissen's own 2009 review, describing his own work:

The sheer volume of bones of unrelated animals at one locality makes it impossible to identify all the bones of one individual.

He goes on that pakicetid skeletons are therefore composites built from bones of a number of different individuals, matched by size, by similarity to other primitive whales, by the chemical composition of the bone, and by relative abundance at the locality.¹⁴

Compare what the 2001 Nature paper's own figure caption disclosed.  It says the reconstructions are based on fossils from Locality 62 and that unknown elements have not been reconstructed.²⁹  It does not say that the reconstructed elements come from different animals.  That disclosure appears eight years later, in an education journal.

That is a legitimate criticism of reconstruction practice, and it is documented entirely from the describers' own publications.

The Indohyus skeleton is a composite too

Indohyus is the small deer-like raoellid put forward as the closest known non-cetacean relative of whales.  Same review, three pages earlier:  many skeletons of Indohyus were washed together and the bones are jumbled, it is not possible to determine which skull went with which other bones, and "the skeleton of Indohyus shown in Fig. 5 consists of bones of a number of different individuals."¹⁴

The most famous whale skeleton in the world is a chimaera

The Basilosaurus cetoides skeleton restored by Gidley in 1913 and Kellogg in 1936 and exhibited at the Smithsonian is, in Gingerich's words on his own page, a composite of two partial skeletons with important parts missing or borrowed from other marine mammals — and its true length is unknown because it is a composite.²²

The quantitative version of that criticism is in Gingerich's 1990 paper.  Excavations in 1989 recovered 7 cervical, 18 thoracic and 42 lumbar and caudal vertebrae in Basilosaurus isis — nine more vertebrae than the reconstructions of B. cetoides had shown.²⁶  The classic mounts were nine vertebrae short.

The Ambulocetus undulation claim rested on two vertebrae

Ambulocetus is the best-known transitional whale, and its best-known functional claim is that it shows spinal undulation appearing in whales before the tail fluke.  Gingerich's 2003 paper corrects it, and this is one mainstream paleontologist correcting another with a page reference:

Ambulocetus is cited as showing that spinal undulation evolved in whales before development of a tail fluke (e.g., Thewissen et al. 1994: p. 212; Fish 2001: p. 637), but that claim was made when only one lumbar and one caudal vertebra were known (Thewissen et al. 1994: p. 210).

One lumbar vertebra and one caudal vertebra.⁹  That is the strongest single instance in this whole subject of a load-bearing inference resting on almost nothing, and it is documented inside the mainstream literature.

Gingerich adds on the same page that placing Ambulocetus on his body-proportion analysis at all required estimating six missing elements — scapula, humerus, a hand phalanx, tibia and two foot phalanges — by scaling them up from Rodhocetus balochistanensis by a factor of 1.45.  He labels the placement "very tentatively" and says it remains to be tested by finding the missing bones.⁹  Any published Ambulocetus body-proportion figure is reproducing a scaled estimate for six of its fourteen measurements.

The Rodhocetus fluke — what actually happened

This episode circulates in apologetics in a form that has the chronology backwards.  The true version is better documented and less dramatic.

What was claimed.  Gingerich and colleagues described Rodhocetus kasranii in Nature in 1994 and inferred tail-powered swimming.  Their abstract says the skeleton has "the flexible sacrum, robust tail and high neural spines on lumbars and caudals required for dorsoventral oscillation of a heavily muscled tail," and that "it shows that tail swimming evolved early in the history of cetaceans."³⁰  Carl Werner, who later interviewed Gingerich on film, writes of a Rodhocetus diagram of the period:  "There were no fossils of the arms and tail yet they had flippers and a whale's tail on the diagram."³¹  The arms part is right.  The tail part overstates:  the 1994 abstract describes tail vertebrae, and Gingerich's own page says "most of the tail" was missing, not all of it.²²  The 1994 paper's own reconstruction figure was not available for this article, so whether it drew a fluke is not settled here.

What was preserved.  From Gingerich's own page, on that type skeleton:

The femur is preserved on one side of the original Rodhocetus kasranii skeleton, but with this exception, the forelimbs are missing, the hind limbs are missing, and most of the tail is missing.

So in 1994 the animal had no known forelimbs, no hind limbs beyond one femur, and almost no tail.²²  The fluke and the flippers were inference from the assumption that Rodhocetus swam like a later whale.

What corrected it.  New fossils.  In 2000 the team found Rodhocetus balochistanensis in Balochistan with fore and hind limbs virtually complete.  The forelimbs turned out to be unable to spread as flippers, and a fluke requires flippers to control the recoil from the tail stroke.  Gingerich published the correction himself, in Science, in 2001:

The forelimbs and hands could not be extended as broad pectoral flippers, which would be required to control recoil from undulation or oscillation of a caudal fluke (38); hence, it is doubtful that Rodhocetus had such a fluke.

Parenthetically on the same page:  the length of the tail is not known.¹⁵

And the figure caption in that paper is a model of how a reconstruction ought to be labelled.  It states that the limbs are scaled up 16% from the other species, that the scapula is restored from Artiocetus and the upper arm from later protocetids, and that the terminal tail vertebrae are conjectural and the tail was almost certainly longer than drawn.¹⁵

Three things follow, and the first is the one apologetics gets wrong.

1. The chronology. The correction was published in Science in 2001. Carl Werner's own website dates his filmed interview with Gingerich to 2001 as well.³¹ Presenting the film as having later extracted an admission is false, and the second half of the quotation that circulates — Gingerich saying "since then, we've found the forelimbs" — is him describing the Balochistan finds he was about to publish. This article does not print that quotation. The video cannot be fetched, no transcript is published, and this project's own notes contain two mutually inconsistent transcriptions of it, which is itself evidence that our copies are ear-transcriptions passed between websites.³¹ Everything in it is in the Science paper, so cite the paper.
2. The scope. The correction concerns Rodhocetus's tail and forelimbs. It does not touch the ankle bones from the same 2001 paper, which are the evidence that moved whales into the even-toed hoofed mammals, and which have only been strengthened since.
3. The mechanism is the strongest pro-evolution point in the episode. New fossils falsified an earlier restoration, and the same author printed the falsification in the same journal. That is the process working, and saying otherwise would be dishonest.

The field had already moved before the film in any case.  Thewissen's 2009 review notes that while early reports proposed a fluke in protocetids, it is now generally accepted that protocetids lacked one — citing both Gingerich's 2001 paper and Buchholtz's 1998 chapter, which questioned it on vertebral morphology three years earlier.¹⁴

The one reconstruction that was disclosed properly

For fairness, because the criticism above is a criticism of practice, not of people.

Gingerich's 1990 hind-limb reconstruction of Basilosaurus isis is also a composite, built mainly on one specimen with fused tarsals and other foot bones taken from two others, and the caption says so — including that the mixture makes it hard to measure joint excursion angles accurately.²⁶  He discloses it in the caption of the figure it applies to.  That is the correct practice, and it is the opposite of what the 2001 pakicetid caption did.

Part 4 — The finding apologetics has not registered

This is the most consequential thing in this article, and it is the one place where the current mainstream position is more favourable to the creationist reading than the version apologists are using.

Apologetic writing on whales fixes on Rodhocetus, an animal everyone now agrees was semiaquatic and paddled with its feet.  Meanwhile Dorudon and Basilosaurus — the fully aquatic archaeocetes at the end of the sequence — are drawn in every textbook with a whale's tail fluke, and creationist writers concede that fluke without argument.

They should stop conceding it, because the man who dug those animals up measured it in 2019 and does not concede it himself.

What the fluke evidence actually was

No archaeocete tail fluke has ever been fossilised.  A fluke is soft tissue.  Its presence in these animals is inferred from the shape of the tail vertebrae — specifically, from a run of flattened terminal vertebrae like the ones under a modern whale's fluke.

Gingerich's own 1990 paper states the inference at exactly that strength.  The vertebral length profiles of Basilosaurus cetoides, B. isis and the dorudontines all show a sigmoid pattern toward the back of the tail, and that pattern "is present in the fluked tails of modern cetaceans, suggesting that Basilosaurus and other archaeocetes also had tails with flukes."²⁶  The word is his.

What he measured in 2019

In 2019 Gingerich, Antar and Zalmout described a new protocetid, Aegicetus gehennae, in an open-access paper about the transition to tail-powered swimming.  As part of it they measured the proportion of the vertebral column occupied by flattened terminal caudal vertebrae, across a comparative sample of living whales.

Flattened terminal caudal vertebrae range from ca. 5.5% to 10.5% of vertebral column length (mean 7.5%) in a large sample of modern whales (N = 36).

The archaeocetes come out below the whole of that range.  Dorudon atrox: 4.1%.  Basilosaurus isis: 3.1%.  Both fall under every one of the twenty-one modern species and thirty-six individuals sampled.²

His verdict, in the same figure caption:

Thus it is questionable whether Dorudon and Basilosaurus were tail-powered swimmers.  A small fluke on a relatively long body may have served, initially, as a caudal stabilizer.

And in the discussion he asks whether flukes that short, on bodies that long, could provide enough surface area to drive a large mass through water at all.²  Of the new animal itself, the paper's abstract says it is doubtful that Aegicetus had a tail fluke.²

Read what that does to the sequence

• 1994. A fluke is placed at Rodhocetus.
• 2001. Gingerich withdraws it from Rodhocetus, in Science, after finding the forelimbs.
• 2019. Gingerich puts a measured question mark over it in Basilosaurus and Dorudon as well.

The single most iconic cetacean feature is inferred in every Eocene archaeocete from vertebral proportions alone, and by the collector's own 2019 measurement those proportions fall below every living whale.  A creationist arguing against a Rodhocetus fluke while conceding a Dorudon fluke is arguing a weaker case than the mainstream literature now supports.

And here is what it does not do

This is easy to over-read, and over-reading it would be the same failure this article criticises elsewhere.

Gingerich is not saying these animals were not aquatic.  The same paper documents Aegicetus with unfused sacral vertebrae and no articular surface on the hip bone for the sacrum — no weight-bearing joint between spine and pelvis at all — plus reduced limbs and a body weight estimated near 890 kg.²  He describes it as more fully aquatic and less specialised as a foot-powered swimmer than earlier protocetids.

What he is arguing about is the mode of propulsion, not the fact of aquatic life.  His reading is that these animals swam by undulating the mid-body and tail, as a modern otter or a crocodile does, and that the fluke as a discrete organ may postdate the Eocene entirely.  That is a change to the story, not a collapse of it.

But it is a change apologetics has not noticed, and it belongs here at full strength for a reason beyond the argument.  This site's claim is that data outranks consensus and outranks the opinions of notable scholars.  A site making that claim has to be willing to report when the data has moved in its own favour without anyone on its side noticing — and to have gotten there by reading the primary literature rather than by reading what the other side's popularisers wrote a decade ago.

The corollary for Aegicetus, printed in the other direction

The same 2019 paper is one of the strongest pieces of evidence against the young-earth reading, and it belongs in the same section so nobody can accuse this article of quoting the convenient half.

In 2003 Gingerich named the gap.  He wrote that the step from Rodhocetus to Dorudon was the more profound of the two, because the distance travelled was greater and the full spectrum of semiaquatic locomotor forms was crossed inside it.⁹  That is a paleontologist saying in print, in a peer-reviewed journal, that a specific part of his sequence was missing.

Aegicetus gehennae was collected in 2007 and published in 2019, from the base of the Gehannam Formation at Wadi Al Hitan — the middle of that gap.  It preserves a vertebral column complete from the first neck vertebra through the ninth tail vertebra, a nearly complete rib cage and sternum, and both fore and hind limbs.  And it has the intermediate condition the gap called for:  more aquatic than a protocetid, less committed to foot-powered swimming, and without the weight-bearing hip joint.²²

A gap named in print in 2003 was filled by a specimen with the predicted anatomy in 2019.  That is the fossil record behaving as its defenders say it should, and this article prints it as plainly as it prints the fluke.

One more thing from that paper, because it is in scope for this series in a way most dating disputes are not.  Its two independent age criteria disagree, and the authors say so:  it is impossible at present to reconcile the microfossil assessment of age with the evidence from global sea-level change.²  By sea-level sequence stratigraphy the specimens are earliest late Eocene; by the first appearances of two microfossil species they are latest middle Eocene.  The authors chose the former and explained why.  That is a live, stated uncertainty in the relative placement of a specimen, printed by the describers, and it is a fair measure of how much slack sits inside a published stratigraphic assignment.

Part 5 — What actually ties the sequence together

Four independent character systems link these animals.  They are not equally strong, and a skeptical reader will know which is which, so they are ranked here honestly rather than listed flat.

1. The involucrum — the strongest single character

The involucrum is a thickened, unusually dense inward-facing lip on the ectotympanic bone of the middle ear.  Thewissen's description of the contrast is concrete:  the equivalent bone in an even-toed hoofed mammal roughly has the shape of half a walnut shell, whereas in a whale the inner wall is much thicker.¹⁴

Every cetacean has an involucrum.  Among non-cetaceans, exactly one group does:  the raoellid Indohyus.

Why it carries weight, stated as a testable property rather than an assertion.  It is a single, discrete, complex, dense bone feature.  It preserves well.  It appears in no other mammal.  And it is on the holotype of Pakicetus, which is precisely the partial cranium with the right auditory bulla intact that Part 2 described as thin material.²³  Whatever Pakicetus did for a living, the diagnostic whale ear structure is in the type specimen of the type species of the type genus, and that is not a reconstruction.

The honest caveat.  A single shared character is consistent with common descent and also with convergence, and this series' own position is that morphology is an unreliable guide to relatedness — a position the next character on this list demonstrates from the inside.³²  What makes the involucrum better than average is its complexity and its restricted distribution, not its mere presence.  State it that way or not at all.

2. The ankle — the character that overturned the previous consensus

From 1966 until 2001 the fossil evidence said whales came from mesonychids, an extinct group of hoofed carnivores.  Molecular data said whales sat inside the even-toed hoofed mammals with hippos as the closest living relatives.  The two answers were incompatible and the fossils were on the losing side.

The 2000 Balochistan finds produced ankle bones associated with protocetid skulls and skeletons for the first time.¹⁵  Four features settled it:  the astragalus has a pulley-shaped surface at both ends rather than one; the heel bone has a prominent convex facet for the lower fibula; the cuboid is notched to receive a process of the heel bone; and the entocuneiform is a large flat plate.¹⁵³³  The first three are general even-toed-hoofed-mammal characters.  The fourth resembles anthracotheres in particular.

This is the cleanest documented case in this whole series of morphology getting a relationship wrong and being corrected.  Gingerich's own words on his page:  the ankle bones showed convincingly that whales did not originate from mesonychid condylarths as Van Valen had hypothesised — and, in his own parenthesis, "and we had expected."²²

Print what this cuts both ways on, because it does.

Against morphology-only phylogeny:  the fossils alone supported the wrong answer for thirty-five years, and the molecules were right the whole time.  That is the worked example for this series' narrow point about convergence — morphology is a demonstrably unreliable guide to relatedness even among animals whose relationships we can check independently, so fossil sequences, which have nothing but morphology, are unreliable too.³²

For the fossil record:  it corrected itself when new material arrived, and the corrected answer agreed with an entirely independent line of evidence.  That is a genuine success and hiding it would be dishonest.

And there is a sharper version of the same lesson in the same author's own back catalogue.  In 1990 Gingerich described the Basilosaurus foot and read it the other way.  His abstract says flatly:  "The foot is paraxonic, consistent with derivation from mesonychid Condylarthra."²⁶  So the same paleontologist, reading the same class of bone — an archaeocete foot — reached opposite conclusions eleven years apart.  Print both.  It is the honest illustration of how much interpretive latitude foot morphology carries.

Gingerich's own assessment of the hippo result is more guarded than popular accounts:  from the point of view of the fossil record, the sister-group relationship of whales and hippos promoted by molecular phylogeneticists is "now plausible, though still tenuous and unproven."²²

3. The blowhole — real, but weaker than usually presented

The earliest cetaceans have nasal bones and a nasal opening near the tip of the snout.  In protocetids the opening sits further back on the snout.  In basilosaurids it has shifted back toward the eyes.¹⁴  The trend is ordered and it matches the stratigraphic order.

The weakness a careful writer should raise even though a skeptic will not.  Nostril position is a continuous character measured on skulls of different sizes and different degrees of crushing, and the sample of protocetid skulls is small.  It is a trend, not a discrete shared feature that is either present or absent.  It should be presented at that strength.

4. Hind-limb reduction — ordered, with one weak joint

The sequence is real and it is ordered:  a weight-bearing pelvis in Ambulocetus; a pelvis apparently not attached to the vertebral column in Georgiacetus; basilosaurid limbs that still have femur, tibia, fibula, ankle bones, foot bones and toes, with the excavated ankle, foot and toes measuring only about twelve centimetres on an animal up to seventeen metres; and finally modern whales with an internal pelvic remnant anchoring genital muscles.²²³⁴

And here is the caveat that has to travel with it.  The Georgiacetus step is where the sequence changes from weight-bearing to not, and it rests on one bone.  Thewissen:  "In Georgiacetus, the only limb element known is the pelvis."¹⁴  The original 1998 description was not available when this article was drafted.  It was read in October 2026 and it says the same thing in its own words:  the ilium has "no distinct auricular surface," and "The absence of this synarthrosis in Georgiacetus indicates a loss of any significant locomotor function of the hind limb in terrestrial locomotion."³⁵  The describers add that the hip socket and muscle scars show the limb kept some use in swimming.  It is still one bone from one animal.

The environmental evidence, checked at one of its two primaries

Oxygen isotopes in tooth enamel track the water an animal drank.  The reported pattern is pakicetids in fresh water, Ambulocetus in partly fresh water — possibly near a river mouth — and protocetids marine.¹⁴

The first draft of this article held that only through Thewissen's review, and said so.  The review itself cites no isotope data for protocetids, only the marine rocks they are found in.  The older of the two primary papers, Roe and colleagues in 1998, has since been read, and it carries all three steps in its own conclusions:  the pakicetids "were closely tied to terrestrial sources of freshwater and food";  Ambulocetus and Gandakasia "may have been euryhaline and able to ingest seawater at least occasionally";  and the protocetids Gaviacetus, Indocetus and Georgiacetus, with the remingtonocetid Remingtonocetus, "appear to have been fully marine, both living and feeding at sea."³⁶  The second primary, Clementz and colleagues in 2006, was still not obtained.  Two limits stay attached.  The pakicetid result rests on thirteen samples from a single locality.  And the Ambulocetus values are spread so widely that the authors themselves suggest testing whether the spread is real.

What was counted as a filled prediction, and what the dates allow

An earlier draft of this article called the following the strongest evidence for the sequence, on the ground that a prediction published in advance and matched by a fossil outranks any after-the-fact resemblance.  That ranking is right in principle.  The dates do not support applying it here, and the reason is printed rather than the claim being quietly dropped.

In 1996 Frank Fish, a biomechanist, published a model of how whale swimming evolved.  In it the modern up-and-down fluke stroke was preceded by up-and-down undulation of the spine with paddling of the hind feet, helped by a long tail, as in a modern otter.

Thewissen's 2009 summary of what happened:  "Writing before the discovery of Ambulocetus, Fish predicted" that otter-like stage, and "The discovery of Ambulocetus showed that Fish's prediction is probably correct: limbs of Ambulocetus are proportionally similar to modern river otters."¹⁴

The paper Thewissen cites was not written before the discovery.  Ambulocetus was described in Science on 14 January 1994, under the title "Fossil evidence for the origin of aquatic locomotion in archaeocete whales."  That paper already concluded that the animal "corroborates that cetaceans have gone through a stage that combined hindlimb paddling and spinal undulation, resembling the aquatic locomotion of fast swimming otters (25, 26)."³⁷  Fish's paper is dated 1996.  It cites the 1994 description on four of its pages and uses the fossil as evidence:  "Spinal undulation evolved before development of a tail fluke as indicated by the fossil Ambulocetus natans (Thewissen et al., 1994)."  Its acknowledgements thank Thewissen for discussions, and the 1994 paper's acknowledgements thank Fish.¹³  In the 1996 paper the fossil is an input to the model, not a test of it.

What is left is real but smaller.  Fish built his model mainly from the measured swimming mechanics of living otters, seals and other mammals, and one fossil's limb proportions agree with the otter stage.  That is agreement between a fossil and living analogues.  It has not been shown to be a prediction published in advance.  One thing was not checked and could restore the stronger claim:  the 1994 Ambulocetus paper cites two earlier papers by Fish, from 1991 and 1993, and if either states the otter-stage model, then some form of the prediction does predate the fossil.  Neither was read for this article.

There is a symmetry here that still holds.  The same biomechanical reasoning — that a tail-driven swimmer needs broad flippers to control recoil — is what Gingerich used in 2001 to argue against a Rodhocetus fluke.¹⁵  One principle confirmed one reconstruction and refuted another.  That is what a working method looks like, as opposed to a story that absorbs any result.

Part 6 — The rank the sequence spans

Every article in this series states this, because the rank is the whole disagreement.  Young-earth biology is not the claim that species do not change.  It is the claim that change is bounded, and the boundary is drawn at roughly the family level.  So the only question that matters about any fossil sequence is whether it crosses that boundary.

Where young-earth biology draws its own line

The standard statement is in Answers Research Journal, in the paper on identifying which animals had to be on the Ark.  Two sentences from it:

From previous work in baraminology, researchers have suggested that the level of the baramin tends to fall at or near the taxonomic level of family (Wood 2006).

There is often a strong cognitum at the family level.  This suggests that the family is a good initial approximation of the level of the baramin.

Both verified from the publisher's own site.¹⁰¹⁰

Wise applies the identical rule to the Cenozoic mammal series that includes the whales:  where the within-group morphological differences are within taxonomic families, "it is likely that the differences are intra-baraminic in every case."¹

That is a real concession and it should be honoured.  Wise is willing to accept a documented stratigraphic-plus-morphological series as post-Flood diversification when it stays inside a family.  A creationist who accepts that rule has already accepted a great deal, including the horse series.

What rank the whale sequence spans

Under the traditional classification the claim is order-to-order.  Artiodactyla — cattle, deer, pigs, camels, hippos — and Cetacea were separate mammalian orders for the entire history of this argument.  The claim is that one order originated from inside another.

Under the current classification the claim is worse for the young-earth side, not better.  Since the 2001 ankle papers, whales sit inside the even-toed hoofed mammals, with hippos as the closest living branch.  A reader might think that shrinks the claim, since the lineage now never leaves the order.  It does not.  It means the whale lineage travelled from a terrestrial artiodactyl family, through five successive archaeocete families, to two living suborders — all inside one order.  The distance travelled is the thing to be explained, and relabelling the box it happens in does not change it.

Count the families, because the count is the argument:

That last figure was computed for this article:  a query for valid family-rank names under Cetacea returned 85, some of them synonymy noise but the great majority real, alongside 440 genera and 704 species-level names.⁸

So the claim spans, at minimum, six families and two suborders, and under the traditional scheme it spans an order boundary.  That is four to five ranks above the line young-earth biology draws for itself.  This article states that without softening it.

The fair objection, and the answer to it

A reader can object that taxonomic ranks above genus are conventions.  Nobody measures a "family."  Calling two groups different orders does not make the biological distance between them larger; it records a decision by taxonomists.

That objection is correct as stated and this article concedes it.  Ranks are conventions.  The answer is to drop the labels and name the anatomy the labels are standing in for.

Between a terrestrial even-toed hoofed mammal and a modern whale, the differences include:  the nasal opening migrating from the tip of the snout to the top of the skull;  the acquisition of the involucrum and of the mandibular fat pad that conducts sound to the ear;  the loss of the weight-bearing joint between spine and pelvis;  the loss of external hind limbs;  reorganisation of the kidney to excrete a seawater load;  loss of the first toe;  decoupling of the windpipe from the gullet;  and the acquisition of a caudal propulsor of some kind, whatever its final form.  A 2019 genome screen found eighty-five protein-coding genes broken in the cetacean stem lineage alone, with functions in blood clotting during dives, lung surfactant composition, oxidative stress, saliva production and melatonin signalling.¹⁷

No amount of arguing about Linnaean conventions makes that a family-level difference.  So the honest young-earth position here is not that the rank is smaller than claimed.  It is either that the rank rule itself is wrong, or that the archaeocetes are several created kinds rather than one — which is Wise's escape (a), and which now has to contend with the involucrum and the ankle.

The one segment where the young-earth reading is coherent, conceded

Between Basilosaurus and Dorudon on one side and modern whales on the other, there is no comparable rank problem, and this article will not manufacture one.

Wise's own reading — that basilosaurids might be early post-Flood members of a whale kind whose descendants are the living whales — is internally coherent for that segment.  A creationist can accept that segment as diversification within a kind at no cost to the position.  The unexplained segment is the earlier one:  raoellid to basilosaurid.

Keep the two apart.  Conceding the later one costs nothing, and it makes the argument about the earlier one much harder to wave away as blanket denial.

But it is not where young-earth baraminology came out when it measured the animals.  The next section shows that its own tests put the basilosaurids in a group of their own and split the living whales into four created kinds.  On those results the later segment is not one kind either.

What young-earth baraminology has tested, and what the genomes say back

Wise's escape (a), that the archaeocetes are separate created kinds, is testable, and young-earth biologists have tested it.  An earlier draft of this article said no such study existed.  That was wrong.  Three analyses were presented at the Baraminology Study Group's meetings in 2005 and 2006, and Todd Wood summarised them in a peer-reviewed creationist journal.  The method, baraminic distance correlation with multidimensional scaling, scores how many characters each pair of animals shares and looks for clusters separated by gaps.

1. Mace and Wood, 2005. A published matrix of 207 skeletal and soft-tissue characters for 66 living whales, one composite basilosaurid (Basilosaurus, Zygorhiza and Dorudon scored as one taxon) and five hoofed-mammal outgroups. The living whales fell into at least four groups: baleen whales, beaked whales, sperm whales and the remaining toothed whales. Of the archaeocete composite they wrote that "the Archaeoceti, while occupying a morphologically intermediate position between the extant cetaceans and the artiodactyls, are very much closer to the artiodactyls than to any of the extant whale holobaramins."³⁸
2. Wood, 2006. The same scaling on three published matrices holding two to seven archaeocetes each, Pakicetus among them. His first conclusion: "Archaeocetes are adjacent to modern whales but are not morphological intermediates between extant cetaceans and artiodactyls."³⁹
3. Wood's summary, which also counts an independent analysis by David Cavanaugh and Richard Sternberg: the archaeocetes, "which are thought to be intermediate between modern cetaceans and land mammals, suggest instead that they are discontinuous with both groups."⁴⁰

So escape (a), graded weakest on the anatomy in Part 1, is the one escape the young-earth side has measured, and the measurement came out its way.  That is a real result and it belongs in any fair account.

It is also a weak result, and its authors said so.

• These are one-page conference abstracts, not full papers. Mace and Wood wrote "Based on the limited sample" and asked for the pakicetids and ambulocetids next.³⁸ Their single composite archaeocete could not show a gradation among archaeocetes even if one existed.
• Every matrix predates Indohyus as a whale relative (2007), Maiacetus (2009), and Aegicetus and Peregocetus (2019).
• The two abstracts disagree about where the archaeocetes sit: "very much closer to the artiodactyls" in one, "adjacent to modern whales" in the other.
• In 2018 Chad Thompson and Wood surveyed 82 analyses of Cenozoic mammals and "excluded datasets of cetaceans and hominins, in anticipation of forthcoming and more detailed analyses than this simple overview."⁴¹ No such cetacean analysis could be found. Twenty years on, the test has not been rerun on the modern fossils.
• The skeptic's reply has a published basis. Phil Senter ran the same scaling method on dinosaurs and early birds and found "morphological continuity between Archaeopteryx, other early birds, and a wide range of nonavian coelurosaurs." The method can find continuity when the in-between animals are sampled, so a gap between a few sampled endpoints may only mean the middle was not sampled. Senter also reports that a 2007 study by Wood found continuity "between finned archeocetes and early mysticetes."⁴² Wood titled that study "Evidence that some toothed mysticetes are archaeocetes."⁴³ Moving the toothed early baleen whales into the archaeocete kind keeps the gap, but a skeptic will read it as relabelling the intermediates.

And the genomes now test the living-whale kinds directly, against them.  A 2019 genome screen found 85 genes broken by inactivating mutations shared between toothed whales and baleen whales.  The toothed whales screened were the bottlenose dolphin, killer whale and sperm whale;  the baleen whale was the common minke;  the shared mutations were then confirmed in the beluga, the Yangtze river dolphin and the bowhead.¹⁷¹⁷  Those species belong to three of Mace and Wood's four living-whale kinds.  In one gene the same block of exons is missing, with the same breakpoints, in the toothed and the baleen whales:  "Shared breakpoints imply that the deletion of KLKB1 coding exons 6 to 12 occurred in the cetacean stem lineage".¹⁷  Separately created kinds would have to carry the same breaks either by repeated identical mutation or because they were made that way.  No creationist treatment of these shared breaks could be found.  This is the strongest evidence against the creationist whale kinds, and it bears on the living whales, where the baraminology result was strongest.

Part 7 — The rate argument, corrected against ourselves

Verdict up front:  "there was not enough time" fails as an anti-evolutionary argument, and it fails on a number computed by the field's own leading collector rather than on an assertion of ours.

This section prints a correction that runs against the position this site started from.  The first pass of the research behind this article compared the whale transition against a baseline that made evolution's job look far easier than it is.  Correcting the mistake makes the transition look faster, which is the direction a creationist would prefer — and the argument still fails.  Both halves belong in print.

What the intelligent-design side claims

The whale-specific version comes from Richard Sternberg, presented in Illustra Media's film Living Waters and in Discovery Institute material.  The clearest published statement of the numbers is on Discovery's own news site:

Standard evolutionary models would either require very large breeding population sizes (greater than that of any species of mammals) or a waiting period four or more times longer than the given 8-9 million years.

That is the whole of it.  No population size, no generation time, no mutation rate, no count of required mutations, and no equation appears in any Discovery Institute source reachable on the open web, and no peer-reviewed publication by Sternberg presenting the calculation could be found.⁴⁴

A claim of the form "four or more times longer" with none of its inputs stated cannot be audited.  It is not a number.  It is an assertion with a number attached, and this article says so plainly rather than repeating it.

The measurement that does exist

Gingerich did the calculation twenty-three years ago, with better inputs, and published it in Paleobiology.⁹

The unit is the haldane — standard deviations of the natural-logged trait per generation.  It is the standard unit for comparing evolutionary rates because it corrects both for how variable the trait is and for how many generations passed.

His setup, with every input stated:⁹

• Transition 1: an Elomeryx-like anthracothere at about 55 Ma to Rodhocetus balochistanensis at about 47.5 Ma. Duration 7.5 million years. Generation time about five years, taken for a mammal intermediate in size between the two from Eisenberg's 1981 mammal compendium. That is 1,500,000 generations.
• Transition 2: Rodhocetus to Dorudon atrox at 37 Ma. Duration 10.5 million years. Generation time about seven years. Also about 1,500,000 generations.
• Trait standard deviation taken as 0.05 natural-log units for linear measurements.
• Fourteen separate skeletal measurements per transition, plus a combined multivariate rate.

His abstract states the result in one line:

Rates of evolution in both phases of this change of adaptive zone are about an order of magnitude higher than background rates for the timescale involved.

The correction, and it goes against us

The research behind this article first ran its own version of this calculation from body size:  a roughly 10 kg raoellid to a roughly 6,000 kg Basilosaurus over about eleven million years at five years per generation, giving about 1.9 × 10⁻⁵ haldanes.  That method was sound.  It agrees with Gingerich's independently computed multivariate rate for the first transition to two decimal places — 10⁻⁴·⁷² against his 10⁻⁴·⁷⁰, from different measurements and different data.

The comparison it was set against was not sound.  That first pass compared a rate measured over about two million generations with the roughly 10⁻¹ haldanes that selection can sustain over one generation, and concluded that the whale transition demanded four orders of magnitude less than the measured capability.

That is precisely the comparison Gingerich's own rate framework exists to forbid.  His thesis is that measured rates fall systematically as the measurement interval lengthens, so a rate measured over a million generations may only be compared with other rates measured over a million generations.  Mixing the timescales makes any long-interval change look trivially easy.

Against the correct, scale-matched baseline the whale transition runs 6 to 16 times faster than background on the univariate measures, and 46 to 126 times faster on the multivariate ones.  Gingerich's own abstract puts it at about an order of magnitude.

That is the number this article prints, and it is the less flattering one.  The earlier draft's sentence — that the transition needed four orders of magnitude less than the measured capability — is deleted, because it was an artefact of mixing timescales.  An audit that only corrects errors running against its own side is not an audit.

Why the argument still fails

Because "ten times background" is not "impossible," and the gap between those two is very wide.

10⁻⁵ haldanes is one hundred-thousandth of a standard deviation of change per generation.  Artificial selection experiments and field studies of living populations routinely produce rates near 10⁻¹ haldanes — four orders of magnitude higher.  An article claiming the whale rate is impossible has to explain why 10⁻⁵ standard deviations per generation is impossible when 10⁻¹ is measured in living populations.  It cannot.

Gingerich's own explanation for the elevation is a shift of adaptive zone — Simpson's old idea that unusually high rates go with an animal moving into a genuinely new way of life.  It is a standard explanation with independent support.

And the way he reports it is worth more than any of it.  He states his prior, measures against it, and prints the result that contradicted him:

Before quantifying the rates of evolution documented here, I expected that rates associated with changes of an adaptive zone would be more or less the same as background rates calculated on comparable timescales in other settings (Gingerich 2001).

He was wrong and he said so.⁹  A mainstream paleontologist stating a prior, measuring against it, and printing the contradiction is exactly the genre this series is written in.  Quoting him is worth far more than quoting anybody who wants the answer to come out a particular way.

One caveat on the number

The background constant of 10⁻⁶·³⁶ comes from a 2001 paper of Gingerich's.  He published a book-length synthesis of evolutionary rates in 2019, and this article could not obtain it.  Whether that synthesis revises the constant has not been checked, so the "about ten times background" figure is presented here as Gingerich 2003's result, dated, rather than as the field's current one.⁴⁵

Why the rate result does not settle the question either

A rate in haldanes measures the magnitude of change in a measurement.  It says nothing about the number of independent, coordinated genetic changes each anatomical novelty required.  Rebuilding a kidney to excrete seawater is not two more centimetres of femur.  The objection worth keeping is about novelty count, not about magnitude — and that is Part 8, where it turns out nobody on either side has the number.

Part 8 — The waiting-time argument, retired

Verdict up front:  the waiting-time objection is not refuted.  It is unformulated.  Nobody on either side has produced the number it needs, and this article will not manufacture one.

The companion article on the mammal ear retired this argument for the same reason.⁴  The cetacean case adds two more, and one of them is this project's own arithmetic error.

The population genetics the argument leans on

The real, checkable literature behind waiting-time arguments is Durrett and Schmidt's 2008 paper in Genetics.⁴⁶  It is open access and every input is stated.

They ask how long it takes a population to acquire a specific pair of coordinated mutations where the first one breaks something.  Their worked case is a transcription-factor binding site in humans:  mutation rate 10⁻⁸ per site per generation, effective population size 10⁴, a ten-nucleotide target, and a generation time of 25 years.  The answer:

Multiplying by 25 years per generation gives 216 million years.

Their own simulation cut it to about 162 million years, which they still call a very long time.⁴⁶

And here is the number that creationist citations of this paper almost always omit.  From the same paper:  in humans, a new transcription-factor binding site can be created by a single mutation in an average of 60,000 years.⁴⁶

So 216 million years is not the cost of adaptation.  It is the cost of one prespecified coordinated pair where the first step breaks a working part.  Single steps cost sixty thousand years.  The ratio between those two figures is the entire argument, in both directions, and quoting one without the other is the failure mode this series audits.

The published exchange with Behe, scored both ways

Michael Behe replied in Genetics; Durrett and Schmidt replied to him.⁴⁷⁴⁸  The exchange does not go all one way and this article prints it as it fell.

Behe wins one point, and they concede it in print.  He argued that applying a binding-site mutation rate — where any of ten nucleotides will do — to a protein point mutation, where one specific nucleotide is needed, understates the wait thirtyfold.  Their reply:  "Behe is right on this point.  This divides our previously computed overestimate of 5 million by 30."⁴⁸

Behe loses the bigger point.  He argued they had ignored the chance of one of the nine already-matching nucleotides breaking before the tenth arrives.  Their answer is that this assumes only one individual in the population carries the first mutation, when in fact very large numbers of individuals carry it before the second occurs, so removing one of them has no effect on the waiting time.⁴⁸

And the structural objection is the one that matters for whales.  Durrett and Schmidt point out that there are at least 20,000 genes in the human genome, and for each of them tens or hundreds of pairs of mutations that could occur.⁴⁸  Their analogy is the New Jersey lottery.  Evelyn Adams won it twice, in October 1985 and February 1986.  The newspapers quoted odds of 17.1 trillion to one — which is the odds for that person on those two dates specified in advance.  The probability that somebody wins twice in a given year is about 1 in 200.

The prespecification problem is fatal to the whale version of the argument as it is currently stated.  To claim that whale evolution required a particular pair of coordinated mutations, one must first show that the anatomical change in question could only have been reached that way.  Nobody on either side has done that for a single cetacean trait.

Our own arithmetic, corrected

This project's own research notes list twenty-four systems that must be rebuilt to turn a land mammal into a whale — counter-current testicular cooling, lung surfactants, blowhole musculature and its neural control, the mandibular fat pad, kidney reorganisation for seawater, a decoupled windpipe and gullet, and so on — and then say:  on average, that is about ten thousand generations per change.

The arithmetic behind that figure is recoverable, and it is wrong.  Nine million years divided by a 27-year generation time gives 333,000 generations; divided by 24 changes gives 13,900 per change.  That reproduces the "about 10k."

The 27-year generation time is a modern great whale's.  It cannot be applied to these animals.  Thewissen's own team describes Pakicetus attocki as wolf-sized and Ichthyolestes pinfoldi as fox-sized; Ambulocetus is put at roughly the size of a large male sea lion; Indohyus was about the size of a cat.²⁹¹⁴¹⁴  Applying sperm-whale demography to a wolf-sized Eocene mammal is the same error as applying it to a deer.  The link our notes give for the 27-year figure is a dead IUCN legacy path.  An archived copy shows it was the Red List page for one species, the sperm whale, with a calculated generation time of 27.3 years.⁴⁹

Redo it with the generation time Gingerich actually sources.  Eleven million years divided by five years per generation is 2.2 million generations; divided by 24 changes gives 91,700 generations per change.  On Gingerich's own two-transition split — eighteen million years, about three million generations — it is 125,000.

Our own notes understated the time available per change by a factor of about 6.6, in the direction that made our argument look stronger.

And the corrected figure is not an argument either.  91,700 generations at five years each is about 458,000 years per "change."  Durrett and Schmidt's measured single-mutation figure is 60,000 years.  On their numbers, 458,000 years is enough for several single-step changes and nowhere near enough for one prespecified coordinated pair.  Which category each of the twenty-four items falls into is the entire question, and the twenty-four-item list does not say.

The one measured input the cetacean case has, and it points the wrong way

In 2019 a team led by Matthias Huelsmann screened cetacean genomes for protein-coding genes carrying shared inactivating mutations in both toothed and baleen whales — that is, genes broken in the cetacean stem lineage, before the two living suborders split.¹⁷

We found 85 gene losses.

Sixty-two of the 85, 73%, had not been reported before.  They attach specific functions to specific losses:  F12 and KLKB1 reducing the risk of blood clots during dives; POLM affecting the fidelity of oxidative DNA damage repair; MAP3K19 in oxidative-stress lung inflammation; SLC6A18 in the diving vasoconstriction response; SEC14L3 in pulmonary surfactant composition; SLC4A9 with reduced need for saliva; and complete loss of melatonin synthesis and receptor genes as a possible precondition for sleeping one brain hemisphere at a time.¹⁷

Now look at what that does to the argument.  Several items on this project's own twenty-four-item list — lung surfactants, seawater tolerance, diving physiology, sleep — turn out on the best available genomic evidence to be associated with gene loss, not gene gain.

A loss-of-function mutation is the cheapest event in the Durrett–Schmidt framework.  It is their inactivating class, not their coordinated-pair class.  Any gene can be broken by a great many different mutations, so the target is huge and the wait is short.

Report this against ourselves, plainly.  In the one place where the cetacean transition has an actual measured count of genomic changes, the changes counted are of the type population genetics says are easy and fast.

That does not prove the whole transition was easy.  Eighty-five gene losses is not a complete account of the difference between a raoellid and a whale, and the paper does not claim it is.  But it means the article cannot assert that the required changes are of the hard type, because nobody has shown that for a single cetacean trait.

So the argument is retired here, for four stated reasons

1. Nobody has counted the coordinated changes, so any waiting time produced for this case is manufactured.
2. The one count that exists points at the cheap category.
3. Our own version of the arithmetic was wrong by a factor of 6.6, in our own favour.
4. The morphological rate, computed by Gingerich, is about 10⁻⁵ haldanes — elevated for its timescale, and four orders of magnitude below what selection produces in living populations.

An article that runs this argument will be checked on all four points and will lose all four.

What to write instead is the true statement, which is uncomfortable for both sides.  The mainstream has published no accounting of how many coordinated genetic changes this transition required.  Neither has anyone else.  So nobody knows whether the wait was 10⁵ years or 10⁸.  That gap is real, it is nobody's property, and naming it is worth more than a calculation nobody can defend.

There is a piece of original work sitting in that gap, and it is the strongest opportunity on this subject for whoever takes it.  Take the twenty-four-item list, or a better one, and classify each item:  which are gene losses, which are regulatory shifts, and which genuinely require novel coding sequence.  The published data to do it exists.  Nobody has done it.

Part 9 — Legs, pelvises, and what "vestigial" means

This site has a documented failure mode in both directions on this subject:  "it is a useless leftover, therefore evolution," and "it has a function, therefore not a leftover."  Both overreach.  Here is what the sources support.

The 1919 humpback with hind limbs, and why our own use of it fails

In July 1919 a whaling ship out of Kyuquot on the west coast of Vancouver Island took a female humpback with two protrusions on the underside of its body, toward the rear.  The crew cut one off and lost it.  The other was photographed in place by the whaling station superintendent, and the bones were sent to Roy Chapman Andrews at the American Museum of Natural History, who published on them in 1921.⁵⁰

The limb projected about 4 feet 2 inches from the body.  The skeletal material Andrews received was "two bones and two heavy cartilages," totalling 31 inches when assembled.⁵⁰  He identified a cartilaginous femur, a bony tibia, a cartilaginous ankle and a bony metatarsal — a hind-limb series — and concluded that "the protrusions actually do represent vestigial hind limbs and show a remarkable reversion to the primitive quadripedal condition."⁵⁰

Two corrections to this project's own notes are needed.  The passage our notes quote as Andrews describing the whale — a female humpback of average length, with the protrusion ending in a knob like a clenched fist — is not Andrews speaking.  It is a letter from Sidney Ruck of the Consolidated Whaling Company, dated 4 March 1920, which Andrews reproduces.  And our notes render "of the average length [52 feet, or 16m]," when the paper gives no length for this whale at all.  The bracket is an editorial insertion by whoever compiled the note.⁵⁰

The argument our own side builds on this does not work either.  Our notes conclude that although Andrews was convinced the protrusions were an evolutionary throwback, the rear flippers resembled the front ones in both soft and bony parts — so they were flippers, not legs.

That reads Andrews backwards.  He observed that the metatarsal resembles a whale's metacarpal.  But fore and hind foot bones are serially homologous in every four-limbed animal, so a resemblance there is expected and carries no weight at all.  What Andrews described is a femur–tibia–ankle–metatarsal series, which is a hind limb.

Recommendation:  drop this argument.  It is a single anomalous individual, described from one photograph and four elements, two of which are cartilage.  It is a curiosity, not a data set, and a creationist should be as unwilling to build on it as on any other one-specimen claim in this article.

The pelvis has a function, and that is not an argument against evolution

Modern whales retain a pair of small bones where a pelvis used to be.  For decades the textbook line was that these are functionless leftovers.  Creationist writing now points to research showing they are not functionless, and treats that as a defeat for evolution.  Brian Thomas of the Institute for Creation Research wrote in 2014 that the new research had uncovered "details that undermine this key evolutionary argument and confirm divine design," that whale hips "are not designed for limb articulation at all," and that "Whale hips are not vestigial."⁵¹

That is a category error, and it defeats a claim the primary literature abandoned long ago.

Thomas does answer a real textbook argument.  He quotes Mark Ridley's textbook Evolution asking why whales should "use bones that are adapted for limb articulation in order to support their reproductive organs."  But his own reply, "not designed for limb articulation at all," fails in the bowhead.  John Struthers dissected the bowhead's pelvic bones in 1881 and found a femur jointed to them:  "An acetabular cartilage, synovial cavity, and head of femur, together represent the hip-joint."⁵²  The companion article on whale pelvic bones takes the creationist replies one by one, including the stronger one that denies the bones are a pelvis at all.⁵³

The functionless reading is Kellogg's, from 1936.  Gingerich's 1990 paper names it and dates it:  the Basilosaurus cetoides pelvic and femoral remains collected in the 1890s "were considered vestigial and functionless," on an interpretation consistent with the loss of functional hind limbs in modern whales — with the citation going to Kellogg 1936.²⁶

It died inside the mainstream literature in 1990, not in 2014.  Gingerich, Smith and Simons argued in Science that the Basilosaurus hind limb was not simply degenerate:  the bones are present, the remaining joints are well formed with little suggestion of degeneracy, the kneecap and heel process are large for powerful muscle insertion, and there is a complex knee locking mechanism.  Their conclusion was that the limbs are most plausibly accessories facilitating reproduction, used as guides during copulation.²⁶

And they printed the weakness of their own hypothesis, which is worth honouring.  Characteristics related to reproduction are often different between the sexes, so they proposed dimorphism in a larger sample of limb elements as a test — and then wrote that a lack of dimorphism "would not necessarily refute" their interpretation.²⁶  A hypothesis whose authors say a negative result would not refute it is weakly testable, and that has to be printed alongside the copulatory-guide claim rather than the claim being repeated as established.

It was then measured in 2014.  Dines and colleagues, in Evolution, analysed pelvic bones from 130 individuals across 29 species, tested against inferred mating system with relative testes size as the proxy for the intensity of sexual selection.³⁴  Two results:  males of species under intense sexual selection have relatively large penises and relatively large pelvic bones for their body size; and pelvic bone shape diverges faster in species pairs that have diverged in mating system.

The control is what makes the study convincing.  Neither pattern appeared in the front-most pair of vertebral ribs, which they used as a negative control.  So the signal is specific to the pelvis and not a general body-shape artefact.  Their own closing line:  "These important functions may explain why cetacean pelvic bones have not been lost through evolutionary time."³⁴

Now handle it carefully in both directions.

Against the textbook line, Dines et al. is a straightforward refutation with a control, published in Evolution.  Anyone still teaching that the cetacean pelvis does nothing is teaching something that has been tested and failed.

Against our own side's use of it, the same abstract says cetaceans "lost external hind limbs and evolved a highly reduced pelvis," and the whole paper is framed as an evolutionary study of a reduced structure.³⁴  Citing it as though the authors had found the pelvis to be an unreduced designed organ is context-stripping, and it is exactly the defect this article's quote audit catalogues in other people's citations.  The finding is "reduced but co-opted," not "not reduced."

The general lesson is the one worth writing, and it is precise.  "Vestigial" never meant "functionless."  Darwin's own usage allowed retained function.  The confident assertion that biologists have been caught pretending otherwise is a failure of the popularisers, not of the primary literature — and the correction runs Kellogg 1936 → Gingerich 1990 → Dines 2014, entirely inside the mainstream literature, with no creationist involved at any step.  An article that presents this as a creationist catch is wrong by thirty-six years.  An article that presents it as the field correcting itself twice is right, and it is a better story.

The 12-centimetre foot

The number to print is this one, because it makes the point without adjectives.

The excavated ankle, foot and toes of Basilosaurus isis from Wadi Hitan measure about 12 centimetres long.²²  Basilosaurus reached about 17 metres.  The limb retains femur, tibia, fibula, ankle bones, metatarsals and toe bones — a complete leg, in miniature, on an animal the length of a school bus.  The describers state that these limb and foot bones "were all found in direct association with articulated skeletons" of the species.²⁶

Gingerich's own assessment in 1990:

Hind limbs of Basilosaurus appear to have been too small relative to body size (Fig. 1A) to have assisted in swimming, and they could not possibly have supported the body on land.

Thewissen's judgement is that because the basilosaurid hind limb preserves even the bones of the foot, it is "reasonable to assume" that some of it projected from the body.¹⁴  Preserve the hedging if you quote it.  Nobody has seen a basilosaurid from the outside.

One further observation that cuts against simple form-follows-function reasoning in either direction.  A 2015 bone-microstructure study found that basilosaurid hind-limb femora show strong osteosclerosis — dense, thickened bone — despite having no involvement in locomotion at all.⁵⁴  A dense bone in a limb that does nothing is a fact that ought to make everybody in this argument slightly more careful.

Part 10 — Objections, one at a time

"Whales appear abruptly.  There are no transitional forms."

False, and the best refutation of it is Wise's own paragraph.  He lists the sequence in detail and calls it a very powerful stratomorphic series.¹  Gingerich's 2003 table names thirty archaeocete genera in five families.⁹  Aegicetus landed in a named gap in 2019.  Do not make this argument.  The defensible version is about density, and Part 2 has the numbers.  Sparse is true.  Absent is false.

"Pakicetus was reconstructed as a whale from a skull fragment."

True, and it is the strongest reconstruction claim available.  Make it, and make it entirely out of mainstream words:  the holotype is the back of a cranium with one ear bulla, and the only other material in the original description is one jaw fragment and four isolated teeth²³;  the 1983 Science cover was drawn from that;  Gingerich now says both his covers were wrong²²;  the 2001 skeletons are composites from a bone bed by Thewissen's own statement¹⁴;  and Gingerich wrote in 2006 that when a representative pakicetid skeleton is found, it will show how pakicetids swam and how they should be exhibited — a leading collector saying no such skeleton exists.³

The over-reach to avoid:  "therefore Pakicetus is not a whale."  The involucrum is on the holotype.  That is not a reconstruction.

"Pakicetus was a running land animal, so the sequence starts with a land mammal that is not a whale at all."

This one is more complicated than either side reports, and the complication runs against the textbooks.

Thewissen's 2001 Nature paper concluded that pakicetids "were terrestrial mammals, no more amphibious than a tapir."²⁹  That sentence is quoted accurately in creationist literature — this project's notes have it word for word.  But two things must travel with it or the quotation becomes context-stripped in our hands.

First, the same paper's headline result is that the fossil evidence had been read wrong for decades:  morphological analyses had placed whales with mesonychians while molecular analyses placed them with hippos, and this paper's own analysis put whales closer to even-toed hoofed mammals than to any mesonychian.²⁹  A creationist citing it for "pakicetids were land animals" is citing a paper whose main point is a large correction to the fossil reading.

Second, Thewissen himself moved off the position.  By 2009 he concluded that pakicetids "were waders in shallow freshwater," citing eye position, limb-bone density, sedimentology and stable isotopes.¹⁴  Quoting the 2001 sentence as his settled view is out of date by seventeen years.

And Gingerich has disputed the terrestrial reading four times in print.  In 2003 on the age of the beds.  In 2004, in a general-science encyclopedia written for non-specialists:  skeletal remains of Pakicetus have been interpreted as terrestrial, "but these are much too fragmentary to interpret reliably."²⁷  In 2005:  "Definitive interpretation of Pakicetus will require associations of skeletal elements that have not yet been documented."³³  And in 2006:  "Some later studies interpreted pakicetids as fully aquatic, or fully terrestrial, but neither extreme is justified."³

Note the direction of that disagreement carefully.  It is not creationists against paleontologists.  It is the two leading collectors of early whales disagreeing about whether bones from a bone bed belong to the same animals.  A textbook or museum label showing a running land mammal is showing a contested reading.

One claim in this area could not be used in the first draft of this article, and now can.  Sandra Madar's 2007 paper on the pakicetid postcranial skeleton is cited in creationist and mainstream writing alike for the conclusion that pakicetid limb bones are incompatible with running.  Four routes to that paper failed when this article was drafted.  A fifth, the Internet Archive's microfilm copy of the journal issue, reached it in October 2026.⁵⁵  Her conclusion is stronger than the usual citation of it.  She grants that pakicetids show the whole list of running adaptations found in Eocene even-toed hoofed mammals.  But she finds dense, thickened bone in every region of the skeleton, and writes that this "would have left these early cetaceans wholly unsuited to running, or even prolonged terrestriality."  Her summary is that pakicetids "were fully committed to an aquatic lifestyle."  That is a third specialist reading the same bones the opposite way from the 2001 Nature paper.  The limit on her evidence is the same as the limit on Thewissen's.  It is the same bone bed, and she says of it:  "No fossils from this locality are found in articulation."  A 2015 study is sometimes brought in to arbitrate this question and it cannot:  its sample was three archaeocete families, none of them pakicetids, and the disagreement it records with Madar is about Basilosaurus, not Pakicetus.⁵⁴  That error was in this project's own research file first, and it was found by opening a paper we had only cited.

"Gingerich admitted on camera that he made up Rodhocetus's fluke and flippers."

The substance is right and the chronology is backwards.  He published the correction himself, in Science, in 2001 — the same year Carl Werner's own site says he filmed the interview.¹⁵³¹  Cite the paper, not the video.  The video cannot be fetched, no transcript is published, this project's own notes hold two mutually inconsistent transcriptions of it, and everything factual in the circulating quotation is in the Science paper in print.

"Basilosaurus and Dorudon could not be ancestral to modern whales, as Barbara Stahl showed in 1974."

Half of it survives, and it is the half nobody quotes.

The quotation is real.  Stahl wrote that these archaeocetes "could not possibly have been ancestral to any of the modern whales."  But this project's own version of it had two errors, both found by opening the book:  a dropped opening phrase that tells the reader she is contrasting two body types;  and an altered key clause — "any group of modern whales" for "any of the modern whales" — which runs in the direction that strengthens the creationist use.⁵⁶

And the context matters more than the wording.  Stahl wrote in 1974, before the dorudontine relationships were worked out.  The field's current position is that Basilosaurus itself is a specialised side branch, which agrees with her, while the dorudontines are at or near the stock that produced modern whales, which does not.  Cite it as a dated 1974 judgement with the change of view stated, or drop it.

"The cetacean pelvis has a function, so it is not vestigial."

A category error, answered in Part 9.  "Vestigial" never meant "functionless," and the functionless reading was retired inside the mainstream literature in 1990 and measured against in 2014.  Both papers frame the pelvis as reduced.

"Whales have long generation times and small populations, so they are a worst-case scenario for rapid evolution."

Dead, and it is ours.  The 27-year figure is the modern sperm whale's, its source link is dead and survives only in an archive, and the animals in question were fox-, wolf- and sea-lion-sized.⁴⁹  Gingerich uses five and seven years, sourced.  Part 8 shows what that does to the arithmetic.

"The Antarctic basilosaurid ages were chosen to avoid an inconvenient result."

Dead, and out of scope in any case.  This project's notes accuse the authors of a 2016 Ameghiniana paper of ignoring dates they did not like in coded language.  Opening the paper kills the charge:  they state a physical mechanism for distrusting the strontium data — reworked shells in the unit — note that the global seawater curve has poor resolution in exactly that interval, cite seven prior publications going back to 1982 that had already discussed the younger age, and prefer an independent method entirely.⁵⁷  The accusation of bad faith is unfounded and it is recorded here only so nobody resurrects it.

"Whales are one created kind, so the archaeocetes are just post-Flood diversification."

Fails on the young-earth side's own rank rule.  See Part 6.  Six families and two suborders, against a line the movement's own journal places at or near the family level.  The escape that remains open is Wise's — that the archaeocetes are several kinds.  The young-earth side's own statistical tests in 2005 and 2006 support a version of it:  archaeocetes separate from living whales, and living whales split into four kinds.  Those tests are small and old, and shared broken genes across the proposed living-whale kinds count against them (Part 6).

Part 11 — Numbers that have moved since they entered the argument

The companion article on the fish-to-tetrapod transition found a date its own authors had corrected three times while apologists went on citing the first version.⁵  The same check was run here.  Four load-bearing figures have moved, and in three cases the person who moved them is Gingerich, correcting himself.

1.  The basilosaurid fluke.  Inferred in 1990 from a vertebral profile, with the word "suggesting."  Questioned by measurement in 2019.  Part 4.

2.  Pakicetus is middle Eocene, not early Eocene.  Gingerich argued in 2003 that the mammal-bearing Kuldana Formation is middle Eocene, and stated it plainly in his 2004 encyclopedia entry.²⁷  The 2003 paper itself is hedged:  the microfossils allow late early Eocene or early middle Eocene, and he favours the second.  The Paleobiology Database still bins Pakicetus attocki in the Ypresian, the early Eocene stage.⁸  Any source placing Pakicetus at 52–53 Ma in the early Eocene is using a pre-2003 figure.  Himalayacetus, at about 53 Ma, is the early Eocene one.

3.  Perucetus colossus's mass estimate fell by roughly a factor of four in seven months.  Described in Nature in 2023 from the Eocene of Peru with a body mass estimated at 85 to 340 tonnes, and reported worldwide as possibly the heaviest animal that ever lived.  Motani and Pyenson re-analysed it in 2024:

Depending on assumptions and methods, we estimate that Perucetus weighed 60–70 tons assuming a length 17 m.

Against about 270 tonnes for the largest measured blue whales.⁵⁸  The stated causes of the original overestimate were assuming a fixed proportion between skeleton mass and body mass and extrapolating it far outside the range it was fitted on, and not correcting for pachyostosis — abnormally dense, thickened bone, which inflates a skeleton-based estimate precisely because the bone is unusually heavy for the animal.

Two lessons, and the second is the useful one.  Do not print the 340-tonne figure.  And note what this is:  a fully documented case of a headline fossil number being wrong by a factor of four because of an unchecked scaling assumption, corrected inside a year by a hostile re-analysis in an open-access journal.  That is the method this whole series depends on, performed by the mainstream on itself.  Use it that way rather than as a gotcha.

4.  The rate-of-evolution framework.  The background constant used in Part 7 comes from a 2001 paper.  Gingerich published a book-length synthesis in 2019 that could not be obtained for this article, so the "ten times background" figure is presented as dated rather than current.⁴⁵

And the errors this audit found on our own side

Listed together, because the credibility of everything above depends on not hiding them.

• The Stahl wording is altered, in the direction that strengthens our use. The page number in our notes, 489, is right.⁵⁶
• The Andrews passage is misattributed to Andrews when it is Sidney Ruck's letter, and carries a bracketed whale length that is not in the paper.⁵⁰
• The 27-year generation time is the modern sperm whale's, applied to fox-sized and wolf-sized Eocene animals, with a dead source link.⁴⁹
• The "about 10k generations per change" figure is wrong by a factor of 6.6, in our own favour.
• The bad-faith accusation against Buono et al. 2016 is unfounded and the paper states its reasons openly.⁵⁷
• A quotation attributed to Larry Moran carried no URL, no date and no venue in our notes. The first pass called it unsourceable. It has since been traced to a comment Moran left on his own blog in January 2016. The wording in our notes is exact, but a blog comment giving no calculation is not evidence of the genetic scale of the transition, so it is still not used in the argument.⁵⁹
• A newspaper quotation from Jim Dines is unnecessary, because the paper it relays is in hand.³⁴
• This project's own first-pass research misread a 2015 paper, claiming it contradicted Madar on Pakicetus when the taxon in the sentence is Basilosaurus and the study sampled no pakicetids.⁵⁴

Of the ten passages in this project's notes that turned out to be altered or misread, eight moved in the direction that made the young-earth argument look stronger.  That is not a coincidence and it is not unique to this subject.  It is what happens when quotations travel between websites instead of being read off the page.

What would change this verdict

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

• A cetacean occurrence in pre-Cenozoic rock, established from superposition rather than from a date. Wise's third reason turns on there being none, and 5,986 catalogued occurrences currently say there are none.⁸
• A full baraminic study of the archaeocetes on a modern matrix, testing Wise's escape (a) with pakicetids, ambulocetids, raoellids, Maiacetus, Aegicetus, Peregocetus and the toothed early baleen whales sampled, against the involucrum and the ankle characters. The 2005-2006 abstracts found a gap on small, older samples; the fuller analysis promised in 2018 has not appeared. A persistent gap on the full sample would strengthen the young-earth case; continuity would end escape (a).
• A creationist account of the inactivating mutations shared across the proposed living-whale kinds, such as the KLKB1 deletion with identical breakpoints in toothed and baleen whales.
• A count of the coordinated genetic changes the transition required, with each change classified as a gene loss, a regulatory shift, or a genuine requirement for novel coding sequence. That would make a waiting-time argument possible here for the first time. Until then, this article holds that the argument is not available.
• A young-earth model of ecological or hydrodynamic sorting developed to the point of predicting a morphological series correlated with the stratigraphic one, with stated failure conditions. Wise proposed the direction in 1995 and nobody has built it.¹⁶
• Confirmation or refutation of the 2019 fluke measurement. Gingerich's own figures put Dorudon and Basilosaurus below every living whale sampled. A larger comparative sample, or a specimen with preserved caudal soft tissue, would settle it either way.²
• A published response to the Aegicetus stratigraphic conflict, in which two independent criteria disagree and the describers say they cannot be reconciled.²
• A second Georgiacetus pelvis, or any of its limb bones. The primary description has now been read and it does say the ilium of the one known specimen had no joint with the sacral vertebrae.³⁵ The hind-limb sequence in Part 5 still turns on that one bone from one animal.
• A test of the copulatory-guide hypothesis that could actually fail. Its own authors wrote that a negative result would not necessarily refute it.²⁶

Sources

The research dossier behind this article — the full specimen inventory, the database query log, thirty-one quotations with verdicts, the sources that could not be reached, and the open questions — is published alongside it:  research/whale-sequence.md.¹²

What was read in the original, and what was not.  Gingerich 2003, Gingerich et al. 2001, Gingerich et al. 1990, Gingerich 2019, Gingerich 2005, Gingerich 2006, Gingerich 2004, Gingerich & Russell 1981, Thewissen & Cooper 2009, Thewissen et al. 2001, Durrett & Schmidt 2008 and both sides of the 2009 exchange, Huelsmann et al. 2019, Dines et al. 2014, Motani & Pyenson 2024, Houssaye et al. 2015, Andrews 1921, Wise 1995, Lightner et al. 2011 and Buono et al. 2016 were obtained and read.  Stahl 1974 was read as page images through controlled digital lending.  A full footnote pass in October 2026 re-read every source above against its page, corrected the page numbers and wording it found wrong, and opened ten sources the first draft had not:  Madar 2007, Hulbert et al. 1998, Thewissen et al. 1994, Roe et al. 1998, Buchholtz 1998, Fish 1996, Gingerich 2001 in Genetica, Gingerich 2003 on the age of the Kuldana Formation, Thewissen et al. 2006, and Larry Moran's 2016 blog comment.  Five of those were read through Internet Archive scans that return text by search only, so their figures and tables were not seen, and each footnote says how its page numbers were fixed.  Still not obtained:  Clementz et al. 2006, Bianucci et al. 2023, Gingerich's 2019 book, Fish's 1991 and 1993 papers, and the Werner video.  Every footnote says so where it applies.

One quotation that circulates widely in apologetics is deliberately not printed here.  Gingerich's on-camera remarks about the Rodhocetus fluke exist only in a video that returns a bot challenge to every fetcher, with no published transcript, and this project's own notes carry two mutually inconsistent transcriptions of it.³¹  Everything in it is in Science in print, and the print version is used instead.  A quotation that cannot be checked against a source is not a citation, whichever side it helps.