Blood Clotting

Summary

Michael Behe argued in 1996 that the chain of enzymes that clots blood is irreducibly complex.  He limited the claim to the part of the chain past its "fork", where four proteins remain:  fibrinogen, prothrombin, factor X and factor V.  Thirty years of tests bear on that claim, and the result is mixed.

1. Behe's narrow claim holds. Mice lacking prothrombin, factor X, factor V or the tissue-factor trigger die before or soon after birth. Mice lacking fibrinogen cannot form a clot or carry a pregnancy. Genes for all of them have been found in every vertebrate examined, from lampreys and hagfish to people. The one doubt is factor V in jawless fish: a 2022 survey missed it, and the gene found since in newer genome assemblies has factor V's structure but has never been tested. The six parts every vertebrate keeps are the six whose loss kills mice or stops their clotting. The five parts some vertebrates lack are the five mice can live without.
2. Russell Doolittle's 1997 counterexample was a misreading. He wrote that mice lacking fibrinogen and plasminogen were "normal." The study he cited says they are "indistinguishable from Fib-deficient mice": they cannot form a clot, and their pregnancies end in fatal bleeding. Behe was right, and no correction by Doolittle was found.
3. The whale and puffer-fish evidence misses Behe's claim. Whales, dolphins and all birds have lost factor XII, puffer fish lack three contact factors, and lampreys lack factors VIII and IX. Every one of these lies before the fork, which Behe set aside on page 86 of the 1996 book. Kenneth Miller at the Dover trial and the judge's opinion misstated his scope. The evidence does refute the popular version, repeated by Of Pandas and People and Answers in Genesis, that every clotting protein is essential.
4. Behe's version is weaker than it sounds. His four parts cannot work without a trigger, as he conceded at Dover. Mice and people with no fibrinogen survive in sheltered conditions. Thrombin has jobs outside clotting.
5. Shared ancestry is clear, and Behe accepts it. The proteases share one design and the helpers another. Four genome studies from 2020 to 2024 agree that the genome doubled once before jawless fish split off and again only in jawed vertebrates. Jawless fish lack factors IX and VIII, the pair that doubling would have copied together. That answers where an enzyme and its partner could come from at once, which was Behe's objection to each step. It does not show that the copied pair became a working new step.
6. No step has been measured. Ancestors of factors VIII and IX from early mammals have been rebuilt, to make better drugs for haemophilia. No ancestral protein of the clotting core has been rebuilt and tested, and no step that added a factor has been tested. Doolittle himself conceded that no living animal has a rudimentary system.
7. Every change seen inside a clotting system is a loss. Gains of clotting factors are inferred from which species have them. None has been observed. The clearest gain of a new use is outside the system: some snakes carry a copy of factor X in their venom.
8. Horseshoe crabs, crayfish and insects clot by unrelated systems.

Verdict.  Behe's narrow claim stands, and his critics' best-known replies do not touch it.  On origin, the critics have a plausible route inferred from sequence with no measured step, and the design side has no probability calculation.  Neither has shown its case.

Part 1: The system, with numbers

Blood has to stay liquid for a lifetime and turn solid within minutes at a cut.  Two things stop bleeding.  Platelets stick to the torn vessel and to one another, and the vessel narrows.  Then threads of the protein fibrin bind the platelet plug into a firm clot.  The clotting cascade is the chain of enzymes that makes the fibrin.  Most of its parts are proteases:  enzymes that circulate switched off and are switched on by being cut.¹

The textbook diagram has three branches.

• The trigger, called the extrinsic pathway. Tissue factor sits on cells outside the blood vessel and meets blood only when the vessel breaks. It binds factor VII, which circulates at about 10 nanomolar with 1 to 2 percent already active.² The pair switches on factors X and IX.
• The contact system, called the intrinsic pathway. Factor XII, prekallikrein and high-molecular-weight kininogen become active on charged surfaces such as glass, and switch on factor XI, which switches on factor IX. They circulate at about 375, 490 and 670 nanomolar.³ Factor IXa, bound to its helper factor VIIIa, switches on factor X about 100,000 to 1,000,000 times faster than IXa alone.²
• The common pathway. Factor Xa, bound to its helper factor Va, cuts prothrombin into thrombin 300,000 times faster than Xa alone.² Thrombin cuts fibrinogen into fibrin, which polymerises, and factor XIII cross-links it.

The point of so many steps is gain.  The trigger complex of tissue factor and factor VIIa reaches less than 100 femtomolar, the prothrombin-cutting complex about 3 nanomolar,⁴ and fibrinogen, the raw material, runs at 180 to 433 milligrams per decilitre,⁴ about 5 to 13 micromolar.⁵  The clot becomes visible when 10 to 30 nanomolar thrombin has formed, about 3 percent of the total;  more than 96 percent comes afterwards.²

Brakes are built in at every level.  The amount of thrombin "is largely regulated by the stoichiometric inhibitors tissue factor pathway inhibitor (TFPI) and antithrombin III (AT-III) and by the dynamic anticoagulant process of protein C activation".²  Protein C, once thrombin switches it on, "destroys accelerin and activated antihemophilic factor", that is, factors Va and VIIIa.⁶  The brakes are as necessary as the accelerator:  mice lacking antithrombin, TFPI, thrombomodulin or protein C die before or around birth.⁷

In 2001 Maureane Hoffman and Dougald Monroe proposed that clotting in the body "occurs not as a "cascade", but in three overlapping stages":  initiation on a cell carrying tissue factor, amplification in which platelets and helper factors are switched on, and propagation, "in which large amounts of thrombin are generated on the platelet surface."⁸  In this model the contact system matters little for stopping a bleed:  "mice and humans lacking fXII have no bleeding tendencies".³

What happens without each part is the most direct test of any claim about necessity.

Haemophilia shortens life expectancy in rich countries by 30 percent for type A and 24 percent for type B.¹⁴  Factor XII deficiency is the odd one out.  In 1955 Oscar Ratnoff and Joan Colopy found three patients whose blood clotted very slowly in a glass tube, and wrote that "this fraction may be named Hageman factor, after Patient I."¹¹  Yet "None of the patients had significant hemorrhagic symptoms."¹¹

Part 2: Behe's claim in his own words

Michael Behe's chapter on clotting in Darwin's Black Box (1996) is titled "Rube Goldberg in the Blood."  It makes five claims.

1. Scope. "Leaving aside the system before the fork in the pathway, where some details are less well known, the blood-clotting system fits the definition of irreducible complexity." The parts he named are "fibrinogen, prothrombin, Stuart factor, and proaccelerin", which are fibrinogen, prothrombin, factor X and factor V.⁶ The sentence is in the first printing of 1996 on the same page.¹⁷ The fork is where the trigger and the contact system meet at factor X. Tissue factor and factors VII, XII, XI, IX and VIII all lie before it.
2. Removal. "Yet in the absence of any one of the components, blood does not clot, and the system fails."⁶
3. Single use. "None of the cascade proteins are used for anything except controlling the formation of a blood clot."⁶
4. Each step. A new step needs a switched-off enzyme and a separate switch at the same time, so "not only is the entire blood-clotting system irreducibly complex, but so is each step in the pathway."⁶
5. No explanation exists. Of Russell Doolittle's published scenario he wrote that "no blood clotting appears until at least the third step" and that "clusters of proteins have to be inserted all at once into the cascade." His conclusion: "no one on earth has the vaguest idea how the coagulation cascade came to be."⁶

Behe did not deny that the proteins share ancestry.  In 2000 he wrote that "structural and sequence evidence for gene duplication and domain swapping in the clotting cascade is very clear."  His objection was to the mechanism:  "EVIDENCE OF COMMON DESCENT IS NOT EVIDENCE OF NATURAL SELECTION."¹⁸  In the same essay he offered the definition in terms of unselected steps that the parent article treats as his most testable version.  In 2006 he dismissed the genome comparisons that followed as "just an exercise in sequence comparisons, with a few paragraphs of speculation."⁶

Behe's latest reply is an appendix to Darwin Devolves (2019).  He restated the claim in a wider form.  He had argued, he wrote, that "a major part of the blood-clotting cascade was irreducibly complex", meaning:  "remove one of its necessary parts and the cascade breaks, either failing to clot at the right time or clotting at the wrong time."¹⁹  "Clotting at the wrong time" brings in the brakes, which the 1996 fork sentence did not name.  He summarised the genome work fairly:  "Jawless fish have most of the proteins that mammals have. Jawed fish have all but a few."  And he noted that "virtually no experimental work was done".¹⁹  He then added a new argument.  The cascade is balanced between parts that promote clotting and parts that stop it.  Breaking one part shifts the balance faster than building a new one, so selection should favour breaking:  "the average time needed to evolve a mini–irreducibly complex feature such as a disulfide bond (a number of which occur in clotting proteins) is a million times that needed to degrade a gene."¹⁹  The "million times" comes from his chapter 9.  Michael Lynch had estimated that a new feature needing several amino acids could evolve in "a million years (or much less)".  Behe took a hundred thousand years and set it against the weeks a damaging mutation took to spread in Richard Lenski's bacterial flasks.¹⁹  It is a general ratio, not a calculation for any clotting route.  He closed:  "not only has the Darwinian evolution of the blood-clotting cascade not been solved; it hasn’t even been addressed."¹⁹  That overstates.  Davidson's group drew a route with two intermediate states in 2003 as a direct reply to him, and Doolittle proposed first steps in 2009, as Part 5 shows.  What is true is that no step of either route has been tested.

At the Dover trial in 2005 Behe narrowed claim 2 further.  Under cross-examination he said the four were necessary, not sufficient:  "those four are needed for the system to work. But -- and I confined my discussion to them. But they're not sufficient for a functioning system."  And:  "while you can do without one way to activate the pathway, you can't do without both ways to activate the pathway."²⁰

Two claims are weaker.  Claim 3 is not true of thrombin, which also signals to cells through its receptors;  the authors of the prothrombin knockout study wrote that thrombin "may play an important role in development."¹⁵  Claim 4 speaks of "the entire blood-clotting system", and the sentence can be read as reaching back before the fork.  In context it continues the argument about the four named parts, as Behe later insisted.²¹

Part 3: The Doolittle exchange and the knockout mice

In 1997 the Boston Review asked Russell Doolittle of the University of California, San Diego, the leading student of clotting proteins, to answer Behe.  He cited a new study of mice lacking plasminogen, the enzyme that dissolves clots, and fibrinogen, the raw material of clots:  "For all practical purposes, the mice lacking both genes were normal! Contrary to claims about irreducible complexity, the entire ensemble of proteins is not needed. Music and harmony can arise from a smaller orchestra."²²

The study is Thomas Bugge and colleagues in Cell, November 1996.  Its summary says the opposite of what Doolittle needed:  "Mice deficient in Plg and Fib are phenotypically indistinguishable from Fib-deficient mice."²³  Mice lacking only plasminogen wasted away;  every one of the 25 in the study had died or been put down by day 301.  Adding the loss of fibrinogen rescued them from that, because with no fibrin there were no clots left undissolved.  But the double-knockout mice kept every problem of fibrinogen loss.  Females could not "survive pregnancy beyond about 10 days", and the rescue left the mice "with the same health and reproductive liabilities associated with a total lack of clotting function."²³  The earlier study of fibrinogen-null mice had found that their "blood samples fail to either clot or support platelet aggregation in vitro" and that "pregnancy uniformly results in fatal uterine bleeding around the tenth day of gestation."¹⁶

Doolittle misread the paper.  The mice had no working clotting system at all, so they were not a smaller orchestra playing the same tune.  Behe said so in 2000, quoted the summary, and was right:  "the double knock-out mice have severe problems and have no functioning blood clotting system. They are not models of evolutionary intermediates."¹⁸  No correction or acknowledgement by Doolittle was found in his later papers on clotting evolution read for this article, which do not mention the mice.²⁴

Two details keep this from being a clean win for either side.  Doolittle's own next sentence hedged:  "No one doubts that mice deprived of these two genes would be compromised in the wild, but the mere fact that they appear normal in the laboratory setting is a striking example".²²  In the cage that is roughly true.  Only 2 of 23 double-knockout mice became terminally ill in 320 days, and 12 lived past 400 days "without any overt signs of disease."²³  The second detail cuts against Behe's wording.  An animal with no fibrin at all is not dead.  Most fibrinogen-null newborns that bleed "ultimately control the loss of blood, clear the affected tissues, and survive the neonatal period."¹⁶  Kenneth Mann's group at the University of Vermont wrote that congenital fibrinogen deficiency "both in mice and man is not lethal and frequently only mildly symptomatic. Therefore, fibrin formation does not seem to be essential for survival."²  Platelets and narrowing vessels carry an animal in a sheltered setting.  So "the system fails" is exactly true of the clot and of a mammal's pregnancy, and false if read as "the animal cannot live."

The other knockouts sort cleanly by Behe's fork.  Every one of his four parts is essential in mice:  fibrinogen for clotting and pregnancy, and prothrombin, factor X and factor V for life itself.¹⁵⁷⁷  So is the tissue-factor trigger.⁹  Mice without factor XII, by contrast, "are healthy, fertile, and phenotypically indistinguishable from their wild-type littermates", and bleed no longer from a cut tail:  369.5 seconds against 355.9 for normal mice.¹²  Behe cited this sorting at Dover as support:  knocked-out "fibrinogen, prothrombin, and tissue factor" fall in the part he wrote about, and there "the blood clotting cascade is broken."²⁵

Part 4: Fish, lampreys, whales and birds

In 1997 Doolittle made a prediction and offered a bet.  From the sequences of mammalian factors he had estimated when their genes were copied, and concluded that jawless fish "should not have both Hageman Factor and prekallekrin".  He was "willing to wager a goodly sum about the outcome."²²  Whole genomes have since settled it.

Doolittle won his bet.  The lamprey result is also a laboratory result, not only a genome search:  Dana Beeler, William Aird and Michael Grant "observed no clotting of recalcified lamprey plasma (up to 75 min), indicating the intrinsic pathway is not conserved in lamprey".³⁵  The whale result is a broken gene in place:  Semba's group found "two nonsense mutations making stop codons and a single nucleotide insertion causing a reading frame shift".³³  Huelsmann's group found the same breaks shared by toothed and baleen whales, so the loss predates their split, and reported that the genes for prothrombin, tissue factor, factor VII and factor X "are intact".³⁴

The same lamprey plasma clots fast when its own tissue factor is added:  "using lamprey thromboplastin resulted in a significantly more rapid PT with lamprey plasma compared with human."³⁵  So a jawless fish has a working trigger and common pathway with no factors VIII, IX, XI or XII.

Factor V is the one part of Behe's four whose presence in jawless fish is in doubt.  Doolittle's group reported in 2008 that in lampreys "factors V and VIII seem to be represented by a single gene".³⁶  In 2022 Abdulbaki Çoban, Erich Bornberg-Bauer and Carsten Kemena of the University of Münster searched the lamprey and hagfish genomes and wrote that "we could not identify one of the cofactors of the cascade, FV, in jawless vertebrates."  They suggested the earlier lamprey gene was a relative of hephaestin, an iron-handling protein of the same family, and that factor V might have arisen only in jawed vertebrates.  They added a caution:  "since the available jawless vertebrate genomes are in low quality, this finding could be an artifact of the fragmented genomes."³⁰  Newer, chromosome-scale genomes favour Doolittle.  The automated gene annotations that the US National Center for Biotechnology Information made for a 2025 sea lamprey genome and a 2026 Atlantic hagfish genome each contain one gene with factor V's full layout:  six copper-binding "A" domains and two "C" domains, about 2,000 to 2,240 amino acids long.  Human factor V has 2,224.  The same genomes carry separate hephaestin and ceruloplasmin genes about half as long, with the A domains and no C domains.³¹  The C domains are what separate factors V and VIII from hephaestin, so the candidate is very likely a factor V-type gene.  But these are machine predictions, no paper has analysed them, and nobody has tested whether the protein speeds up thrombin formation in lamprey or hagfish blood.  If it does not, the core of a jawless fish is five parts, and factor V joins the list of parts that were added later.

Counting the eleven parts of the human diagram gives a sharp pattern.

The eleven are tissue factor, factors VII, XII, XI, IX, VIII, X and V, prekallikrein, prothrombin and fibrinogen.  The six kept in every vertebrate are tissue factor, factors VII, X and V, prothrombin and fibrinogen.  They are also the six whose loss kills mice or stops their blood clotting.  The five missing somewhere, factors VIII, IX, XI and XII and prekallikrein, are the five whose loss mice survive.³⁷  Every one of the eleven belongs to a larger protein family.  Even the sea squirt, which has none of them, has relatives or building blocks of virtually all.²⁶  Both sides expect the parts an animal cannot live without to be kept, so the pattern shows which parts make up the core.  It does not show how the core arose.

Kenneth Miller of Brown University used these results at the Dover trial in 2005 against the school textbook Of Pandas and People, which said "all of the proteins had to be present simultaneously".  Of whales he testified:  "they don't have any problems with their blood clotting."  Of puffer fish:  "they are missing three parts of the system and their blood clots perfectly well."³⁸  In Only a Theory (2008) he turned the same evidence on Behe:  "As Michael Behe has written, the entire system has to be in place for clotting to work properly, and in the absence of any of the components, blood does not clot and the system fails."  Then:  "The genome of the fugu, or puffer fish, lacks three of the clotting factors—and its blood clots just fine."³⁹  Judge John Jones accepted the argument and wrote that Behe's "redefinition of the blood-clotting system was likely designed to avoid peer-reviewed scientific evidence".⁴⁰

Behe replied at the trial that he had set the early part aside:  "I was very careful to specify exactly what I was talking about".²⁵  Casey Luskin of the Discovery Institute pressed the point from 2008:  "all of the components he cites come before the fork".⁴¹

On this narrow point Behe and Luskin are right.  The fork sentence is on page 86 of the 1996 printing, nine years before the trial, so it was not a redefinition.  Every factor missing from any vertebrate in the table lies before the fork.  Genes for all four of Behe's named parts are present in every vertebrate examined, including the lamprey and the hagfish, with factor V in jawless fish known only from its gene's structure.^42362931  Miller's paraphrase dropped the words "any one of" and the sentence before it.  His evidence does refute Pandas, and it refutes the popular version that young-earth ministries repeated, in which blood clotting "would not function if even a single component were missing."⁴³

The cross-examination, though, exposed a real limit in Behe's own version.  His four parts cannot clot blood by themselves;  something must switch on factor X.  Behe agreed:  "they're not sufficient for a functioning system."²⁰  A defensible irreducible core must therefore include a trigger.  Every vertebrate examined keeps the same one:  tissue factor and factor VII.  The claim that survives the comparative evidence is narrower than the one in the book, and stronger for being narrow:  in every vertebrate lineage examined, no animal has been found that lacks the genes for tissue factor, factor VII, factor X, factor V, prothrombin and fibrinogen.  The weakest link in that sentence is factor V in lampreys and hagfish, where the gene has been predicted by machine and never tested.

Part 5: How the cascade is thought to have been built

The evidence of shared ancestry among the parts is strong, and Behe grants it.  Colin Davidson, Edward Tuddenham and John McVey of Imperial College wrote in 2003 that the factors "have resulted from the reduplication and functional diversification of only two different gene structures".  One is the protease design shared by factors VII, IX and X and protein C.  The other is the helper design shared by factors V and VIII, which are related to the copper-carrying blood protein ceruloplasmin.⁴²  Their tree of the helper proteins "clearly supports two successive gene duplications", which they read as support for two rounds of whole-genome duplication early in vertebrate history, a proposal they called "controversial."⁴²  Doolittle's lamprey work points the same way:  the IX and VIII pair could have appeared together, "the simultaneous appearance of a second set of two functionally paired proteins".³⁶

The genome doublings are no longer controversial, and their timing is now known.  Four groups using chromosome-scale genomes reached the same answer between 2020 and 2024.  Oleg Simakov and colleagues found that "an initial 1R doubling occurred before the divergence of the lamprey and jawed vertebrate lineages", followed by "a second duplication ... in the jawed vertebrate lineage (that is, after the lamprey lineage had diverged)".⁴⁴  Yoichiro Nakatani's group, working from new lamprey and shark genomes, agreed, and found that "the cyclostome lineage experienced an additional hexaploidization", a tripling of the genome in the jawless line.⁴⁵  Both teams that sequenced hagfish genomes in 2024 came to the same timing.⁴⁶⁴⁷  None of the four papers discusses clotting genes.  But the clotting genes fit their timing, which is this article's inference:  jawless fish lack factors IX and VIII, the pair the second doubling would have copied in the jawed line, and the lamprey's three copies each of factors VII and X match a tripling in the jawless line.

The fit answers where a matched enzyme and helper could come from at once.  It does not answer the harder question.  A copied pair of factor X and factor V starts as a spare copy of a step that already exists.  It becomes a new step only when the copied enzyme stops cutting prothrombin and starts cutting factor X, and when the copied helper binds the new enzyme instead of the old one.  How many mutations that takes, and whether each one helped, has not been measured.

That is the best published answer to Behe's claim that each new step needs an enzyme and its switch at once.  If a whole block of genome is copied, an enzyme and its partner are copied together, already fitted to each other.  Davidson's group drew the route explicitly, with "two intermediate coagulation states":  first an ancestral protease with tissue factor activating prothrombin, then separate factors VII and X with an ancestral V-VIII helper, then the IX-VIII pair.⁴²  They presented it as a reply to what "has been described as ‘an irreducibly complex system that could not have arisen by a gradual step-by-step Darwinian process’".⁴²

Doolittle's own account of the very first steps is openly speculative.  He proposed in 1993, and repeated in 2009, that thrombin first clumped the ancestors of platelets by cutting their surface receptors, and that fibrinogen came later.²⁷  He also wrote that "The total absence of clotting factors in basal chordates and the existence of a relatively complex system in lampreys indicate that it is not likely we will find any extant creatures with truly rudimentary clotting systems, leaving us to speculate on how the system might have gotten started."²⁷

Count what has been measured.  The lamprey and fish show that systems without the contact factors or without factors VIII and IX clot blood.  That is one measured fact about intermediate states, and it concerns the part before the fork.  Ancestral clotting proteins have been rebuilt, but for medicine, not to test the cascade's origin.  Christopher Doering's group at Emory University rebuilt ancestral factor VIII from "47 available mammalian FVIII sequences" in 2017.⁴⁸  In 2021 the same group rebuilt "8 AnFIX variants inferred from nodes corresponding to early placental mammals, supraprimates, primates, rodents, carnivores, and ungulates", and found two, "inferred as the last common ancestors of humans and lemurs", with more clotting activity than human factor IX.⁴⁹  These ancestors already sat in a complete mammalian cascade and share 83 to 98 percent of their amino acids with the human proteins, so they say nothing about how a step was added.  For the steps that built the core of prothrombin, factor X, factor V and fibrinogen, and for the first trigger, no ancestral protein has been rebuilt and tested, and no step has a measured advantage.  Every step is inferred from sequence.  Behe's 2006 complaint that these papers are "just an exercise in sequence comparisons" is accurate as a description.⁶

Loss is observed;  gain is inferred.  Every documented change inside a living clotting system is a loss:  factor XII in birds and whales, prekallikrein in whales, or one of the many human disease mutations.  A lost gene leaves a broken copy in place, as in the whale.  A gained gene leaves only a pattern of presence and absence across species, such as factor XI appearing in mammals and nowhere else.¹³  No gain of a new clotting factor has been observed.  Extra copies of existing factors are common, like the lamprey's three copies of factor X, but none has been shown to add a new step.  The clearest case of a copied clotting gene taking on a new job lies outside the cascade.  The venom of the rough-scaled snake of Australia contains trocarin D, a copy of factor X that acts as a toxin by switching on prothrombin, while the snake's own factor X works in its blood.  Md Abu Reza, Sanjay Swarup and Manjunatha Kini of the National University of Singapore found that the two genes "have eight exons with identical exon-intron boundaries" and that "All the introns are 92-99% identical with the exception of intron 1, indicating a recent gene duplication event."⁵⁰  They proposed that an insertion of 264 DNA letters in the copy's switch region moved it to the venom gland.  So a duplicated clotting gene can be put to a new use.  The gain is still inferred from gene structure, not watched, and it added a weapon, not a step to the snake's own clotting.

The design reading treats shared architecture as reuse of good parts, as argued in Convergent Evolution and Design Reuse, and weighs such patterns as in The Design Inference.

The young-earth ministries mostly repeat Behe.  Jonathan Sarfati of Creation Ministries International gave an accurate account of the Bugge mice, marred by calling the critic "the atheist Doolittle", and added that the next step back toward a full system "would be selected against because of the uncleared clots."⁵¹  That is fair for one proposed intermediate, not for all.  No article on the cascade was found at the Institute for Creation Research.

Part 6: Other animals clot differently

Other animals seal wounds with systems unrelated to ours.

• Horseshoe crabs. Their blood cells hold "three kinds of serine protease zymogens, factor C, factor B, and proclotting enzyme, and a gel-forming protein, coagulogen", and the chain is triggered by bacterial endotoxin, not by injury.⁵² Coagulogen's shape resembles the nerve growth factor of vertebrates, not fibrinogen.⁵² Mats Hall and colleagues noted that these proteins "are not homologous to the vertebrate plasma coagulation proteins."⁵³
• Crayfish. There is no protease cascade at all: "Coagulation in crayfish blood is based on the transglutaminase-mediated crosslinking of a specific plasma clotting protein."⁵³
• Fruit flies. "Transglutaminase is the only component of vertebrate clotting with an orthologue in the Drosophila genome."⁵⁴

Parasites and predators that attack clotting are treated elsewhere:  tick saliva in Ticks, Fleas, Lice, Mites and Botflies and snake venoms in Predation, Venom and Carnivory.

To the critic, protease chains assembled more than once cannot be as hard to build as Behe says.  To the design side, unrelated solutions to one problem are what an engineer with a shared toolkit produces.  Neither reading has a number behind it, and separate origins do not show how any one system was assembled.

Objections and responses

"Behe moved the goalposts at Dover."  He did not.  The sentence setting aside the system before the fork is on page 86 of the 1996 printing.¹⁷  The court's charge of a "redefinition" is contradicted by the book.

"Whales and puffer fish clot without parts of the cascade, so it is not irreducibly complex."  True of the parts before the fork, which Behe did not claim.  False of his four parts and the tissue-factor trigger, which no vertebrate examined has lost.

"Lampreys and hagfish have no factor V, so even Behe's four are not all needed."  A 2022 survey could not find factor V in either and suggested the earlier lamprey gene was hephaestin.³⁰  Newer genome assemblies each show one gene with factor V's full set of A and C domains, separate from their hephaestin genes.³¹  So the evidence leans toward factor V being present, but it rests on machine-predicted genes and no test of the protein in jawless-fish blood.  If the objection proves right, the jawless core is five parts, not six, and Behe's four would need a different list for the earliest vertebrates.

"Doolittle showed that mice clot without two of the proteins."  He misread Bugge 1996.  The double-knockout mice cannot form a clot at all, and their pregnancies fail.

"Mice and people without fibrinogen survive, so even Behe's core is not necessary."  For life in a sheltered setting, that is correct, and Behe's phrase "the system fails" should not be read as "the animal dies."  For the function Behe named, a fibrin clot, fibrinogen is necessary, and every mammalian pregnancy without it ends in fatal bleeding.  Doolittle agreed that such animals "would be compromised in the wild".²²

"Gene duplication explains the cascade."  It explains where the parts came from, which Behe accepts.  It has not shown that each step paid its way, because no step has been tested.

"Each new step needs two proteins, so the odds are squared."  Behe estimated the chance of assembling the four domains of one clotting enzyme by picking among 30,000 gene pieces at random, about one in ten to the 18th, and squared it for an enzyme and its activator.⁶  Doolittle answered that shuffling is confined to "specific regions of specific chromosomes".²²  A whole-genome duplication also copies an enzyme and its partner together, so the squaring does not apply to that route, and the doublings are now well established.  What remains is the change that turns a spare copy of a step into a new step, which nobody has priced.  Neither side has calculated a probability for a stated route.

"Shared domains are reuse by a designer, not ancestry."  Reuse accounts for shared modules.  It does not by itself account for the broken factor XII genes that toothed and baleen whales share at the same positions, and a young-earth account must place those whales within one created kind for the shared loss to be inherited.  That question belongs to Whale Evolution.

"The losses support a young earth, because genomes are degrading."  Loss of factor XII fits a degradation model, but it fits the standard model too, and both expect losses to be easier than gains.  Losses cannot decide between the two.

Issues and unanswered questions

Against the design side:

• No number. Behe's 2000 definition counts the unselected steps on a route. Nobody has counted them for any proposed route to the clotting core. The one number in the book assumes domains are picked at random from the whole genome, which no critic proposes. The number in Darwin Devolves, a million-fold gap between building and breaking, is a general ratio built on one estimate by Michael Lynch and one bacterial experiment, not a count for any clotting route.
• The four are not a system. Behe conceded at Dover that his four parts cannot work alone. The defensible core has at least six parts. His 2019 restatement speaks of "a major part" of the cascade and its "necessary parts" without listing them.
• Factor V in jawless fish is unproven. The best evidence that lampreys and hagfish have it is a machine-predicted gene with the right domains. If it is not a working factor V, the earliest vertebrate core has five parts and one of Behe's four was added later.
• The duplications are real. The two genome doublings that the critics' route needs are now well established, and the clotting genes of jawless and jawed vertebrates fit their timing. Behe's objection that each step needs two new proteins at once has a documented source for such pairs.
• Single use is false. Thrombin also signals to cells and may shape embryonic development.
• Popular overstatement. Answers in Genesis described the cascade as failing if "even a single component" were missing. The whale and puffer-fish evidence refutes that version, and the ministries have not been found to correct it.
• Removal is not assembly. Knockouts test whether a modern system breaks without a part. As the parent article sets out, parts that came to depend on one another over time, the route Hermann Muller described in 1918, would also break when one is removed.

Against the critics:

• No step is measured. After thirty years, no ancestral clotting protein from the core has been rebuilt and tested, and no step has a measured advantage. The only rebuilt clotting proteins are recent mammalian ancestors of factors VIII and IX, made for drug design.
• A copied pair is not yet a new step. Duplication gives a spare copy of an existing step. Turning it into a new step upstream needs the copied enzyme to change what it cuts and the copied helper to change what it binds. Nobody has measured how many mutations that takes.
• The rudimentary system is missing. Doolittle concedes that no living animal has one.
• Misreadings went uncorrected. Doolittle's mouse claim, Miller's paraphrase of Behe's scope, and the court's "redefinition" charge are each contradicted by the documents, and none was found corrected.

Not read for this article:  Doolittle 1993, the full texts of Doolittle 2008 and 2015, and Doolittle's 2012 book The Evolution of Vertebrate Blood Clotting.  Any of these could contain a measured step or a correction.

What would change the verdict.  Toward the critics:  a resurrected ancestral protease that, with tissue factor, clots plasma and improves survival in a model animal, with the later steps each tested the same way;  or a test showing that lamprey or hagfish blood makes thrombin with no factor V.  Toward design:  a test showing that the jawless-fish factor V-type protein does the job of factor V, which would complete the six-part core in the earliest vertebrates.  Toward design:  a population-genetic calculation showing that the unselected steps on the best published route need more time than vertebrate history allows, or evidence that the gene pairs did not arise together by block duplication.

Verdict

Behe's claim, read as he wrote it, has held up better than his critics said.  The four parts past the fork, with the tissue-factor trigger, are essential in every knockout, and every vertebrate genome examined has genes for them.  For factor V in lampreys and hagfish that rests on gene structure alone.  Doolittle misread the mouse study.  The whale and puffer-fish evidence refutes the textbook Pandas and the ministries' popular version, not Behe.

The claim is also narrower than it sounds.  Behe's four parts do not work alone.  Animals without fibrin live in cages.  Thrombin has other jobs.  And a knockout tests removal, not origin.

On origin, neither side has shown its case.  The factors share ancestry, and the two genome doublings, now well established, offer a way to add an enzyme and its partner together.  But no step of that route has been tested, no one has measured what it takes to turn a copied pair into a new step, and Doolittle concedes that no living animal shows a rudimentary system.  What survives is a precise absence:  no measured route to the vertebrate clotting core has been published.

Sources