Embryonic Leftovers

"Fundamentally, therefore, the embryo of a higher form never resembles any other form, but only its embryo." — Karl Ernst von Baer, 1828, translated by T. H. Huxley, 1853¹

"The actual pattern — early differences followed by similarities, then differences again — is quite unexpected in the context of Darwinian evolution." — Jonathan Wells, Icons of Evolution, 2000²

Summary

The claim under test is that a human embryo passes through its ancestry as it is built, so leftovers of that ancestry are visible in it.  That sentence hides three claims with different fates.

• The old claim is dead, and evolutionists killed it. Johann Friedrich Meckel wrote in 1821 that a higher animal "durchläuft," runs through, the permanent adult stages of the animals below it.³ Ernst Haeckel made this evolutionary in 1866: "Die Ontogenesis ist die kurze und schnelle Recapitulation der Phylogenesis."⁴ Gavin de Beer concluded in 1940 that this "does not take place" and called it "a mental strait-jacket."⁵⁵ Stephen Jay Gould wrote in 1977 that genetics had made it "untenable."⁶
• The observation that replaced it is older than Darwin and is real. Karl Ernst von Baer, writing three decades before Darwin, reported in 1828 that he could not tell whether two unlabelled embryos were lizards, birds or mammals, and that an embryo "never resembles any other form, but only its embryo."¹¹ Darwin explained this by descent: "community in embryonic structure reveals community of descent."⁷
• The modern form is a conserved middle stage, and it is measured. Naoki Irie and Shigeru Kuratani compared gene expression through development in the mouse, chicken, frog and zebrafish in 2011 and found "the pharyngula stage is most conserved, whereas earlier and later stages are rather divergent."⁸ The same waist appears within six fruit-fly species.⁹
• It is also disputed, and the limits matter. Developmental timing varies most in the middle, not least.¹⁰ Between phyla the middle is the least alike.¹¹ Early embryos are not conserved at all.¹²
• The creationist literature mostly answers the dead claim. Answers in Genesis and the Institute for Creation Research argue that human embryos never have gills or an adult-fish stage, which mainstream embryology gave up by 1940.¹³¹⁴ Jonathan Wells and Russell Grigg do engage the live claim. The only worked design explanation found, Jerry Bergman's, predicts that the earliest stages should be the most alike, which the data reject.¹⁵

Verdict.  Haeckel's plates were inaccurate and recapitulation is false, but neither fact touches the evidence that is still offered.  Vertebrate embryos converge on a shared body plan in mid-development and then diverge, and common descent explains that pattern without strain, even though it did not predict its exact shape.  Common design can explain why embryos share parts.  It has not yet explained why a mammal builds, at mid-development, the same pouches with the same genes that a fish turns into gills, and then remodels them.  Each structure needs its own answer, so this page is a hub for three.

Part 1: Three claims under one name

The phrase "ontogeny recapitulates phylogeny" is a paraphrase.  It blurs three claims that their own authors kept apart.

Meckel and Serres.  Meckel wrote that "das höhere Thier in seiner Entwicklung dem Wesentlichen nach die unter ihm stehenden, bleibenden Stufen durchläuft":  the higher animal runs through the permanent stages below it.³  Gould, following E. S. Russell's name of 1916, calls this the "Meckel-Serres Law" and notes that both men grounded it in "the unity of nature's laws," not in descent.⁶

Von Baer.  Von Baer rejected parallelism in 1828 and put four laws in its place.  The first two say that the general features of a large group appear in the embryo before the special ones.  The third and fourth say:  "Every embryo of a given animal form, instead of passing through the other forms, rather becomes separated from them," and "Fundamentally, therefore, the embryo of a higher form never resembles any other form, but only its embryo."¹⁶¹  "In my possession are two little embryos in spirit, whose names I have omitted to attach, and at present I am quite unable to say to what class they belong.  They may be Lizards, or small Birds, or very young Mammalia."¹

Darwin.  Darwin took von Baer's pattern, not Meckel's, and gave it a cause.  "For the embryo is the animal in its less modified state; and in so far it reveals the structure of its progenitor."⁷  He looked at the embryo "as a picture, more or less obscured, of the common parent-form of each great class of animals."⁷  In the same chapter he retold von Baer's unlabelled-embryo story and credited it to Agassiz.⁷

Haeckel.  Haeckel's law is thesis 41 of the "Ontogenetische Thesen" in the second volume of his Generelle Morphologie:  "Die Ontogenesis ist die kurze und schnelle Recapitulation der Phylogenesis, bedingt durch die physiologischen Functionen der Vererbung (Fortpflanzung) und Anpassung (Ernährung)."  Ontogeny is the short, fast recapitulation of phylogeny.⁴  Even Haeckel hedged:  thesis 43 says the repetition is "verwischt und abgekürzt," blurred and shortened, by secondary condensation.⁴

The modern claim.  Klaus Sander named the "phylotypic stage" in 1983 for the point where the vertebrate classes "first exhibit the characteristics common to all vertebrates";  the same stage is called the tailbud stage or the pharyngula.²  Denis Duboule's "egg-timer" of 1994 and the "hourglass" of Duboule and Rudolf Raff, 1996, describe development as wide at the start, narrow in the middle and wide at the end.⁸  Duboule's proposed cause was the strict time order in which Hox genes switch on.¹⁷

Part 2: Biologists buried recapitulation, not embryology

De Beer's Embryos and Ancestors (1940) closed with a list of conclusions.  One reads:  "Recapitulation, i.e. the pressing back of adult ancestral stages into early stages of development of descendants, does not take place".⁵  The next reads:  "Similarity in ontogeny between any animals is a proof of their affinity, and no evidence as to the adult structure of the ancestor."⁵  He added that dropping Haeckel freed embryology from "a mental strait-jacket which has had lamentable effects on biological progress."⁵

Gould's Ontogeny and Phylogeny (1977) is the standard history.  He argued that nobody refuted the biogenetic law by counting exceptions:  it "fell only when it became unfashionable in approach (due to the rise of experimental embryology) and finally untenable in theory (when the establishment of Mendelian genetics converted previous exceptions into new expectations)."⁶  Once heredity was understood as Mendelian genes, Gould argues, the law's two needed rules, terminal addition and condensation, became untenable in theory.⁶

Gould also said plainly what the switch did not change.  Haeckel read human "gill slits" as "features of ancestral adult fishes";  von Baer read them as "a stage common to the early ontogeny of all vertebrates."  Then:  "Does it matter whether we are actually repeating the adult stage of a fish-like ancestor (as the recapitulationists claimed), or only developing a common embryonic feature that fish, as primitive vertebrates, retain throughout life (as von Baer claimed)?  The phyletic information is the same".⁶  He was "convinced that the vast majority of supposed recapitulations represent nothing but the conservative nature of heredity, as expressed in von Baer's laws."⁶

So an attack on recapitulation attacks a claim evolutionary biologists gave up by 1940.  The argument they kept is weaker than Haeckel's but harder to dismiss:  shared embryonic stages mark shared descent, without showing what any adult ancestor looked like.

Part 3: What the modern data measure

The modern evidence comes in two kinds:  what embryos look like, and which genes they switch on at each stage.

What the looks show.  Michael Richardson's team photographed tailbud embryos across the vertebrates in 1997.  They opened with the resemblance:  "Embryos of different species of vertebrate share a common organisation and often look similar."  Then they measured its limits:  "we find a greater than 10-fold variation in greatest length at the tailbud stage," and Haeckel's plates showed "not a conserved stage for vertebrates, but a stylised amniote embryo."¹⁸  In 2003 Olaf Bininda-Emonds and colleagues scored the order of developmental events:  "phenotypic variation between species was highest in the middle of the developmental sequence," and "The onus is now clearly on proponents of the phylotypic stage to present both a clear definition of it and quantitative data supporting its existence."¹⁰

What the genes show.  Alex Kalinka and colleagues measured gene expression through embryogenesis in six fruit-fly species and found that "expression is maximally conserved during the arthropod phylotypic period."⁹  Irie and Kuratani found the same waist across four vertebrate classes, at the pharyngula.⁸  Tomislav Domazet-Lošo and Diethard Tautz reported that the zebrafish phylotypic stage "does indeed express the oldest transcriptome set."¹⁹  That last result assigns gene ages by assuming the tree, so it cannot count as independent evidence for the tree, and its method is disputed.²⁰

What limits it.  Barbara Piasecka and colleagues found in 2013 that the hourglass holds for regulatory DNA but "not at the level of gene sequence, age, or expression," and concluded that "both models are valid, but with respect to different genomic features."²¹  Michal Levin and colleagues compared ten phyla in 2016.  Across phyla the middle of development is a "divergent 'mid-developmental transition' that uses species-specific suites of signalling pathways and transcription factors."¹¹  Hu and colleagues, with Irie as senior author, found in 2017 that the idea "would be better applied to vertebrates than chordates."²²  Kalinka and Pavel Tomancak concluded in 2012 that "early animal embryos are not highly conserved" and that early development "often reflects adaptation to diverse ecological niches."¹²

Sorted by strength:

• Measured and not in dispute: early embryos differ widely; late embryos diverge toward adult form; within vertebrates, embryos at the pharyngula share a body plan of notochord, neural tube, somites, pharyngeal arches and a postanal tail. Answers in Genesis lists the same features.¹³
• Measured, with dispute over what counts: gene expression is most alike at mid-development within a phylum. Timing of events is least alike there.
• Not supported: one stage shared by all animals; any stage at which a human embryo resembles an adult fish.
• Explanations still argued: why the middle is conserved. Proposals are Hox timing, dense signalling that makes any change "increase the risk of mortality,"⁸ and genes reused so widely that changing them is costly.²²

This is not what Darwin expected.  Kalinka's group notes that Darwin expected the earliest stages to be most alike, and that this principle "has not been supported by morphological studies."⁹  The hourglass was explained after it was found.  Descent still explains it more simply than any rival offered, because a body plan laid down by a shared ancestor is what descent would conserve.

Part 4: The creationist literature in its own words

Jonathan Wells, Discovery Institute, 2000.  Wells's chapter on embryos makes four claims.  Haeckel's drawings "are fakes."  They "omit the earliest stages entirely," which differ.  "Gill slits" in human embryos cannot "logically serve as evidence for evolution."  And the true pattern undercuts Darwin:  "if von Baer's laws apply only to the second half of ontogeny, descent with modification is deprived of what Darwin believed to be 'the strongest single class of facts' in favor of it."²²²  He calls the hourglass "quite unexpected in the context of Darwinian evolution."²  He also shows the textbook problem is current, quoting Raven and Johnson's Biology:  "early in their development, human embryos possess gill slits, like a fish."²

Wells answers the live claim, and he grants its central fact:  vertebrate embryos "superficially converge midway through development."²  His strongest point is fair.  Darwin did lean on early likeness, and the early likeness is not there.  His weakest point is the step from "unexpected" to "paradox."  Darwin had already allowed that embryos which must fend for themselves diverge:  "cases could be given of the larvæ of two species, or of two groups of species, differing quite as much, or even more, from each other than do their adult parents."⁷  A mid-development convergence still has to be explained by someone.  Wells offers no account of why it exists.

Elizabeth Mitchell, Answers in Genesis, 2013 and 2020.  Mitchell writes that "mammalian pharyngeal arches are no more related to gills — ancestrally or otherwise — than stars are to streetlights."¹³  She quotes Mader's Biology (2007), which asks why land vertebrates build pouches that have "lost their original function," and replies:  "No one has ever documented that pharyngeal pouches in the embryos of terrestrial vertebrates function as gills or that adult terrestrial vertebrates ever had gills."¹³  That answers a claim of gill function, which Mader did not make.  Mader's claim is homology.  "Ancestrally or otherwise" also denies a sameness of structure that other creationists explain as common design.

Randy Guliuzza, Institute for Creation Research, 2016.  Guliuzza writes of the pharyngeal folds:  "No evidence exists that they ever resembled an adult fish".¹⁴  That is true, and it is de Beer's point of 1940.  His positive claim is that "It was creationists who had first discussed common designs for common features."¹⁴

Russell Grigg, Creation Ministries International, 1998.  Grigg separates the claims correctly:  "When evolutionists say that the recapitulation theory is false, they usually do not mean to admit that comparing embryos gives no evidence of common ancestry."  He then answers the modern claim with the fraud, saying it is "based on faked drawings."²³  That overstates.  The mid-stage likeness is now measured in gene expression with no drawings involved, and Richardson's own paper opens by granting it.⁸¹⁸

Jerry Bergman, Creation Research Society Quarterly, 2000.  Bergman gives the only worked design explanation found.  "The stages of embryological development of many animals show some similarities but the major reason for this is design constraints," and "we would expect the earlier in development the fewer the design possibilities that exist."¹⁵  This predicts a funnel:  most alike at the start.  The transcriptome studies above, and Wells himself, report the opposite.  It explains simple-to-complex order, not a waist in the middle.

Part 5: How each structure is tested

The parent article's test asks what is alleged, what the structure does, and whether a feasible build beats it.²⁴  Embryonic structures sharpen it, because a transient structure may be needed only while it exists.  Each child answers four questions.

1. Name the structure in neutral terms: pharyngeal pouch, not gill slit; caudal eminence, not tail.
2. Say what it becomes or does, in the embryo and after birth.
3. Say whether its route is explained by function, by ancestry, or by both. A route is explained by function when a later step needs it, as scaffolding. It is explained by ancestry when the same program builds a working organ in a related animal and the embryo builds that program and then undoes it.
4. Name what would discriminate. Common design predicts shared stages wherever embryos share a building task. Common descent predicts shared stages that nest in the same tree as adult anatomy and DNA, and detours with no use to the embryo that builds them.

Haeckel's Drawings and Recapitulation.  It finds no record that any Jena court or university body ever ruled on the drawings.  It covers the 1868 woodcuts and 1874 plates, the charges against Haeckel in his lifetime, Richardson's photographic re-check, and the plates' life in textbooks.  The likely verdict is that the plates were inaccurate:  Richardson's team found more than tenfold size differences at the stage Haeckel drew as one "stylised amniote embryo."¹⁸  Whether Haeckel meant to deceive is argued among historians, and the child weighs that.  The verdict will also carry this page's point:  the fraud does not dispose of the embryology, because the embryology no longer depends on the plates.

Pharyngeal Arches and "Gill Slits".  This child covers the arches in a human embryo and what they and the pouches between them become:  parts of the jaw, the middle-ear bones, and, from the pouches rather than the arches, the thymus and the parathyroids.  The creationists are right that no human embryo has gills, and that "gill slit" is a poor name.  The argument they have to answer is different.  In zebrafish and dogfish the gene Gcm2 is switched on in the pharyngeal pouches and is "required for the elaboration of the gill buds";  in mammals and birds the same gene in the same pouches builds the parathyroid glands.²⁵  Anthony Graham and Jo Richardson conclude that "the gills were transformed into the parathyroid glands."²⁵  The likely verdict is that this is the strongest of the three cases for ancestry.

The Embryonic Tail, Lanugo, and the Pronephros.  It finds a real embryonic tail trimmed by cell death, an untested lanugo function, and no human pronephros at all.  Three structures get three answers.  The caudal end is partly a dispute over a word.  Fabiola Müller and Ronan O'Rahilly, who staged 52 human embryos, write that the caudal eminence builds the caudal notochord and "all caudal structures, but it does not produce even a temporary 'tail' in the human."²⁶  Lanugo, the fine fetal coat, is the item where the child must first establish what, if anything, it does.  The three-kidney sequence is the best test of the function-or-ancestry question, since a creationist article quotes a secular anatomy text that calls the three "regionally specialized parts" of one kidney field rather than a replay of ancestors;²⁷  the child must verify that text before relying on it.  The likely verdict is mixed, and the child should say plainly if any of the three does nothing.  The coccyx and born-with-a-tail cases are already covered in The Coccyx and Atavisms, and the general vestige argument in Vestigial Structures.

Objections and responses

"Haeckel faked his drawings and recapitulation is false, so embryology gives no evidence for evolution."  The plates were inaccurate, and their long life in textbooks is a real scandal.  But this conflates Haeckel with von Baer.  De Beer rejected Haeckel's law while holding that shared development is "a proof of their affinity, and no evidence as to the adult structure of the ancestor."⁵  Rejecting the second half does not reject the first, and the present evidence is gene-expression data that uses no drawings.

"Similar embryos just mean a common designer."  Common design explains why parts recur.  It does not by itself explain why a mammal builds the pouch series with the program that makes gills in fish, then closes it off.  A design reply needs a stated building reason for that route.  Bergman's attempt predicts the wrong pattern.¹⁵

"The phylotypic stage has been refuted."  Bininda-Emonds and colleagues showed that the timing of events varies most in the middle, and Richardson's team showed that tailbud embryos differ in size and segment number.¹⁰¹⁸  Those results refute a single identical stage.  They do not refute the gene-expression waist, which later studies measured directly.⁸

From the skeptic:  "The hourglass is a successful prediction of evolution."  It was not predicted.  Darwin expected early likeness, which "has not been supported."⁹  Wells is right about that much.

"Human embryos don't have gills, so nothing is left over."  Correct about gills.  The pharyngeal pouches are still the same structures, built by the same genes, that form gills in fish.²⁵  The leftover, if there is one, is a building program, not an organ.

"A shared middle stage might simply be the only way to build a vertebrate."  Possibly.  Hu and colleagues' pleiotropy explanation is a constraint of that kind.²²  But a constraint that holds only within vertebrates, and weakens toward other chordates and disappears between phyla, is shaped like a family tree.  The design reply has to explain that shape too.

Issues and unanswered questions

• No design account of the waist. No creationist source examined explains why embryos of fish and mammals converge in the middle of development and diverge on both sides. Bergman's constraint argument predicts convergence at the start.
• Most of the creationist effort has gone to Haeckel. The plates are an easy target and a real one, but they have not carried the evolutionary argument since 1940.
• Young-earth kinds make this harder, not easier. Genesis 1:20–25 has fish made on the fifth day and land animals on the sixth, and no young-earth model puts fish and mammals in one created kind. Their shared pharyngula must therefore be common design, not descent. The fruit-fly waist sits inside one genus and might fall within a single kind; the cross-class result of Irie and Kuratani is the one that matters here.
• Weak points on the other side. The hourglass was not predicted. Gene-age measurements assume the tree they are used to support. The morphological phylotypic stage has no agreed definition. Parts of the tail debate are about vocabulary.
• Not obtained. Richardson's 1998 letters in Science, which Guliuzza reports as saying that the data remain consistent with Darwinian evolution, could not be downloaded, so they are not quoted. The anatomy text behind the three-kidney quotation was not found.

Verdict

Recapitulation of adult ancestors is false, and Haeckel's plates misrepresented real embryos.  Evolutionary biologists conceded both decades ago.  Arguments against them win a fight that ended in 1940.

The live claim is von Baer's, restated with gene-expression data:  within the vertebrates, embryos converge on a shared body plan at mid-development and then diverge.  This is measured.  Its limits are measured too:  early embryos differ, timing varies, and the waist does not reach across phyla.  Common descent explains a within-group waist naturally, though it did not predict it.  The design reply is right that similarity alone does not prove ancestry, and right that no human embryo has gills.  What it has not yet done is give a building reason for the route itself:  why a mammal embryo lays down a pouch series with the gene that makes gills in fish, and then remodels it.  Until it does, this is evidence that favours common descent, and each of the three child articles should show how strongly for its own structure.

Sources