Irreducible Complexity

Behe's 1996 definition rules out only direct routes, as he said on his page 40. Whether indirect routes are too improbable has never been calculated for any named machine. Muller's 1918 route was shown in a laboratory in 2012, for a duplicated part that lost function. No route to any of Behe's five systems has been reconstructed with measured fitness.

By AI (Claude Fable 5.1, reasoning effort 30) Draft:  September 2026

"Even if a system is irreducibly complex (and thus cannot have been produced directly), however, one can not definitively rule out the possibility of an indirect, circuitous route." — Michael Behe, Darwin's Black Box, 19961

"We are aware of no other mechanistic analyses of a molecular machine's evolutionary trajectory, so the generality of our observations is unknown." — Gregory Finnigan, Victor Hanson-Smith, Tom Stevens and Joseph Thornton, Nature, 20122

Summary

Michael Behe, a biochemist at Lehigh University, argued in 1996 that some cell machinery stops working when any one part is removed, and that such machinery cannot be built by small improvements to one function.  The argument is often stated as "no gradual route exists."  Behe did not say that.

  1. The definition rules out direct routes only. A direct route improves one function by one mechanism. Behe said on the next page that an indirect route cannot be ruled out, and argued that it becomes less likely as the parts multiply. The claim is a probability, and nobody on either side has calculated it for any named machine.
  2. The definition was revised repeatedly. Behe offered a version counting "unselected steps" in 2000, added the word "necessarily" in 2001, and admitted in the same paper that the definition tests removal when evolution must do assembly. By 2007 he called irreducible complexity a rough yardstick and moved the argument to counts of mutations.
  3. Hermann Muller published a route in 1918. A part added as an improvement becomes necessary when other parts come to depend on it. The date and the substance are right. The phrase "interlocking complexity" is not Muller's, and he offered no worked case.
  4. Muller's route has been shown once in a laboratory. In 2012 Joseph Thornton's group rebuilt a two-part protein ring that became a three-part ring in fungi. The new part is a copy of an old one, each copy lost an ability, and the pump works no better.
  5. The other routes are mostly lists of related proteins. Of twelve biological studies read for this article, three tested proteins in a laboratory and nine compared sequences or structures. None measured the selective value of each step, and only one of the three concerns a system Behe named.
  6. Several critics answered a claim Behe did not make. Kenneth Miller, Robert Pennock and the Kitzmiller opinion each say he ruled out parts or precursors with other functions. His book allows them and argues they are improbable. The philosopher Paul Draper, who rejects Behe's conclusion, says the same.
  7. Young-earth ministries state a stronger claim than Behe does, and apply it to eyes and knees, which he excluded.

Verdict.  The argument as popularly stated fails, because a universal negative cannot be shown.  The narrower claim survives:  for these systems no step-by-step route with measured advantage has been published in thirty years.  That is an absence of evidence, and the adaptive immune system shows how such an absence can close.  The live part of the argument is now numerical and is treated in Haldane's Dilemma and Plasmodium falciparum.

Part 1: The argument in Behe's words, and how it changed

Darwin's test

Charles Darwin set the test in 1859:  "If it could be demonstrated that any complex organ existed, which could not possibly have been formed by numerous, successive, slight modifications, my theory would absolutely break down.  But I can find out no such case."3  One page later he named a way around it.  The swimbladder, he wrote, shows "that an organ originally constructed for one purpose, namely flotation, may be converted into one for a wholly different purpose, namely respiration."3  Both halves of the later debate are there.

The 1996 definition

Behe's definition reads:  "By irreducibly complex I mean a single system composed of several well-matched, interacting parts that contribute to the basic function, wherein the removal of any one of the parts causes the system to effectively cease functioning."1  His example is a mousetrap with a platform, hammer, spring, catch and holding bar.1

The next sentence carries the argument, and its parenthesis carries the limit:  "An irreducibly complex system cannot be produced directly (that is, by continuously improving the initial function, which continues to work by the same mechanism) by slight, successive modifications of a precursor system, because any precursor to an irreducibly complex system that is missing a part is by definition nonfunctional."1  Then the concession quoted at the head of this article, followed by:  "As the complexity of an interacting system increases, though, the likelihood of such an indirect route drops precipitously."1

The first claim is true by definition and says little.  The second is about odds, and the book gives no number.

Behe named five systems:  the cilium, the bacterial flagellum, the blood clotting cascade, gated transport of proteins within the cell, and the immune system.1  He also accepted common descent and an old universe,1 which is why Henry Morris of the Institute for Creation Research wrote that "Michael Behe himself admits to being an evolutionist."4

The revisions

Year Where What changed
2000 Discovery Institute essay A tentative "evolutionary" definition that counts unselected steps in a pathway5
2001 Biology and Philosophy The word "necessarily" added. An "asymmetry" admitted6
2002 Dembski, No Free Lunch An "irreducible core", and the system's "original" function7
2007 The Edge of Evolution Irreducible complexity called a rough yardstick. Steps and coherence preferred8
2019 Darwin Devolves "Mini-irreducible complexity" inside one protein9

The 2000 version is the one that can be tested:  "An irreducibly complex evolutionary pathway is one that contains one or more unselected steps (that is, one or more necessary-but-unselected mutations).  The degree of irreducible complexity is the number of unselected steps in the pathway."5  It makes irreducible complexity a property of a route and a matter of degree.  It is often credited to the 2001 journal paper, where the word "unselected" does not occur.10

The 2001 paper makes two admissions.  Allen Orr's objection, given in Part 2, showed the definition was "equivocal", so Behe inserted a word:  the system must be "necessarily composed" of its parts.66  Robert Pennock's objection showed something worse.  "The current definition puts the focus on removing a part from an already-functioning system," Behe wrote, while evolution must assemble one.  "Thus there is an asymmetry between my current definition of irreducible complexity and the task facing natural selection.  I hope to repair this defect in future work."6

William Dembski's version requires that each part be "indispensable to maintaining the system's basic, and therefore original, function."7  The words "and therefore original" set aside by definition any route in which the function changed.  Dembski also conceded that a core of "one or only a few parts" may be no obstacle.7  His one calculation, for the flagellum, is examined in The Design Inference.  Dembski later set it aside.

In 2006 Behe restated the 1996 position and denied ever claiming more:  "I did not assert that indirect routes were logically impossible".1  He also shifted the burden:  "in the presence of manifest design, the onus of proof is on the one who denies the plain evidence of his eyes."1

In 2007 he wrote that the new measures "gauge difficulties for random mutation at a much finer level than does irreducible complexity", which he likened to "a yardstick with no markings".8  The reason he gave is that the concept "passes over the fact that a part might itself be a special piece that needs explaining in terms of many steps."8  In 2019 he applied the idea to features within one protein, such as a disulfide bond, which "require multiple amino-acid residues", and he named the problem as "the need for multiple coordinated mutations".99  He rejected "could not possibly" as the standard and proposed another:  "if there are good physical reasons to think Darwinian routes wouldn't work and if after a diligent search no evidence is found that they do, then the theory has failed."9

The repair promised in 2001 was never published.  The argument moved from missing parts to counted mutations.

Part 2: Muller's route, published in 1918

Hermann Muller, who won the Nobel Prize in 1946, was explaining in 1918 why most mutations in fruit flies are lethal.  His answer was that an organism is a machine whose parts have come to need each other:

"thus a complicated machine was gradually built up whose effective working was dependent upon the interlocking action of very numerous different elementary parts or factors, and many of the characters and factors which, when new, were originally merely an asset finally became necessary because other necessary characters and factors had subsequently become changed so as to be dependent on the former."11

The date is right.  The words are "interlocking action";  the label "interlocking complexity" came later.  And Muller did not claim to have shown it.  He wrote that the idea "had suggested itself to the writer in 1912" and that it "would manifestly have been very difficult to obtain experimental evidence for it".11

He returned to it in 1939.  A new character, once established "with or without the aid of selection", is built upon by later mutations until "the once novel character can no longer be withdrawn with impunity, and may have become vitally necessary."12  His examples were the gill slits of vertebrate embryos, now needed to form the parathyroid gland, and the hip rudiments of whales.  He reported no experiment.  He also wrote that such transfers of function are "largely of a non-adaptive, fortuitous nature", which is the idea later named constructive neutral evolution.12

Allen Orr, an evolutionary geneticist at the University of Rochester, brought Muller into the debate in 1996:  "Some part (A) initially does some job (and not very well, perhaps).  Another part (B) later gets added because it helps A.  This new part isn't essential, it merely improves things.  But later on, A (or something else) may change in such a way that B now becomes indispensable."13  He credited it:  "This scenario was first hinted at by the geneticist H. J. Muller in 1918 and worked out in some detail in 1939."13

Behe replied twice.  In 2001 he granted that the route works for one class of system, where a single protein does a job weakly and a second one helps.6  He denied that it reaches a system that needs several kinds of part to do its job at all.  In 2006 he asked what "part A" could act alone as a flagellum, and noted that Orr "had nothing at all to say" about any of the biochemical cases.1

Two critics of Behe agreed with him on this point.  Richard Thornhill and David Ussery wrote in 2000 that Orr's route "could only produce a complex, rather than an irreducibly complex, structure."14

Muller's route is sound logic that waited until 2012 for a demonstration.

Part 3: The proposed routes and what has been measured

Route Proposed by Best measured case Limit
A part is added, then made necessary Muller 1918, Orr 1996 The fungal proton-pump ring, 2012 The new part is a copy of an old one
Co-option of a part from another job Darwin 1859, Gould and Vrba 1982 The RAG transposon, 2016 One component of a larger system
Scaffolding that is later removed Cairns-Smith 1985 None The scaffold is inferred, not found
Gene duplication and divergence Orr 1996 The same ring Rates are disputed by model
Constructive neutral evolution Stoltzfus 1999 The same ring Explains dependence, not the parts
Digital organisms Lenski and others 2003 Avida Not biology

The ring that gained a part

The V-ATPase is a proton pump found in all cells with nuclei.  Part of it is a ring of six proteins.  In animals the ring has two kinds of protein.  In fungi it has three, and a yeast cell missing any one of them cannot grow in high calcium.  Finnigan and colleagues inferred the sequence of the protein that existed before the third kind arose, built it, and put it into yeast.  They place the change before the last common ancestor of fungi.  It replaced both of its modern descendants.  After the gene was copied, each copy lost the ability to sit on one side of its neighbour.  "These losses were complementary, so both copies became obligate components".2  One historical mutation in each copy was enough to recreate most of the loss.

Their abstract opens:  "Despite much speculation, strong evidence of the mechanisms by which these assemblies evolved is lacking."  They conclude that "simple degenerative mutations are sufficient to explain the historical increase in complexity".22

Behe accepted the history.  "The work of Finnegan et al. (2012) strikes me as quite thorough and elegant.  I have no reason to doubt that events could have unfolded that way."  His objection:  "All of the functional parts of the system were already in place before random mutation began to degrade them."1515

Both are right.  A part became required, by known mutations, with no foresight, and that is what Muller proposed.  The ring, the pump and every binding surface existed beforehand, and the pump gained nothing.  The experiment measured growth in a test medium, not whether any step was favoured.

Co-option

Stephen Jay Gould and Elisabeth Vrba coined "exaptation" in 1982 for "features that now enhance fitness but were not built by natural selection for their current role."16  Their cases were feathers and skull sutures.  Orr doubted the idea as a general answer for molecular machines:  "You may as well hope that half your car's transmission will suddenly help out in the airbag department."13

In 2006 Jamie Bridgham, Sean Carroll and Thornton rebuilt an ancestral hormone receptor and found it already responded to aldosterone, a hormone that arose later.  They concluded "that tight interactions can evolve by molecular exploitation".17  Behe replied that the system "consists of just a single protein and its ligand", and that the two mutations tested weakened a response the protein already had.18  The paper's own figure agrees on the second point.  The best case of co-option is in the immune system, given in Part 4.

Scaffolding

Graham Cairns-Smith, a chemist at Glasgow, asked in 1985:  "How can you build any kind of arch gradually?"  His answer was a wall of stones from which some are later removed.19  He granted the dependence that Behe would later name.  Of central biochemistry he wrote:  "Nothing can be touched or the whole edifice will collapse."19  No biological scaffold has been identified for any of Behe's systems.

Duplication, and evolution without selection

Behe and the physicist David Snoke modelled a duplicated gene that needs several changes before it gains a new feature.  They found that a feature needing two or more residues "would entail population sizes of no less than" a billion to be fixed in a hundred million generations.20  They added:  "results bearing on the efficiency of this one pathway as a conduit for Darwinian evolution say little or nothing about the efficiency of other possible pathways."20  Michael Lynch replied that the model "is non-Darwinian" because its intermediate steps are useless.21  That exchange is scored in Haldane's Dilemma.

Arlin Stoltzfus proposed in 1999 that excess capacity plus mutation bias can build dependence with no advantage at any step.  He claimed only that it "is possible, in theory."22  Julius Lukeš and four colleagues wrote in 2011 that such scenarios are "less forced than the selectionist or adaptationist narratives that are generally told."23

Digital organisms in Avida evolved a complex logic function when simpler functions were rewarded.  When they were not, "none of these populations evolved" it.24  See Evolutionary Computation.

Part 4: The critics, the court, and the cases

The critics

Jerry Coyne reviewed the book in Nature in 1996.  He granted that the pathways "are dauntingly complex, and their evolution will be hard to unravel", and that "we may forever be unable to envisage the first proto-pathways."  His objection was to the inference:  "It is not valid, however, to assume that, because one man cannot imagine such pathways, they could not have existed."25

Russell Doolittle, the leading student of clotting proteins, wrote in 1997 that mice lacking two clotting genes were normal:  "Contrary to claims about irreducible complexity, the entire ensemble of proteins is not needed."26  He had misread the study.  Behe quoted it:  in such mice "pregnancy uniformly results in fatal uterine bleeding".5

Kenneth Miller showed that working cilia exist with fewer microtubules than the usual pattern.27  All of them keep the three kinds of part Behe named.  Miller also wore a mousetrap missing two parts as a tie clip, and wrote that Behe "argues that no portion of an 'irreducibly complex' system (such as a mousetrap) could have any function."28  Behe had written in 1996 that the parts of a cilium have other jobs in the cell.1  His reply:  "The system is irreducible, not the parts."1

John McDonald drew a series of mousetraps from one part to five.  He said they were "not intended as an analogy of how evolution works."29  Behe answered that each trap needs several separate adjustments to become the next.30

Robert Pennock charged a bait and switch.  Losing a part stops the named function "by definition", but Behe "can't get the empirical conclusion from his 'by definition' conceptual argument."31  Pennock also noted that Behe had not listed all the parts of any system.31  Behe conceded the asymmetry Pennock exposed, and denied making the claim Pennock ascribed to him.61

Thornhill and Ussery sorted routes into four kinds.  They granted that direct routes cannot build such a system from indivisible parts, "and this is the valid conclusion to draw from Behe's thesis."14  The other two kinds can.

Paul Draper, an agnostic philosopher, defended Behe against Miller:  "Behe does not mean to rule out by definition functioning precursors".  He then named the gap.  A small system could arise indirectly and grow directly, so the argument "is at best incomplete."3232

Maarten Boudry, Stefaan Blancke and Johan Braeckman said in 2010 that the claim "becomes either trivial (weak version) or plainly wrong (strong version)", and that Behe moves between them.33  The weak version is what Behe showed.  His page 40 states it openly.

Kitzmiller v. Dover

Behe testified for the school board in October 2005.  Judge John Jones found that irreducible complexity "is a negative argument against evolution, not proof of design", that Behe had not repaired the defect he admitted, and that the claim "has been refuted in peer-reviewed research papers and has been rejected by the scientific community at large."3434

On the immune system the opinion says Behe was handed fifty-eight articles, nine books and textbook chapters, and "simply insisted that this was still not sufficient evidence of evolution, and that it was not 'good enough.'"34  The transcript differs.  The words were the lawyer's:

Q.  So these are not good enough?

A.  They're wonderful articles.  They're very interesting.  They simply just don't address the question that I pose.35

Behe's question was whether any of them gave a step-by-step account by mutation and selection.  He also admitted he had not read the books:  "I am quite skeptical, although I haven't read them".35  The opinion adds that Behe excluded co-option "by definitional fiat".34  His book discusses it and calls it improbable.  The court's finding that the argument is negative is sound.  Its finding of refutation cites testimony about studies of the kind listed below, which establish relatedness.

The cases

The bacterial flagellum.  The question is whether a rotary motor of some forty proteins has simpler working relatives.  The secretion system often offered as its precursor is now placed as its descendant.  A study of over 1,000 genomes concluded that it "arose from an exaptation of the flagellum".36  That shows a sub-assembly can be reused, and leaves the motor's own origin open.  The "forty parts" are genes;  about twenty proteins form the structure, about ten of 42 proteins have relatives outside the flagellum, and no proposed route has a measured step.  See The Bacterial Flagellum.

Blood clotting.  Behe set aside the early part of the cascade and called the rest irreducible.1  Lampreys clot blood with fewer factors than other vertebrates, and the missing ones belong to the early part.37  The four parts Behe named past the fork are essential in every knockout and present in every vertebrate examined, and every known loss falls before the fork;  no step in building the core has a measured advantage.  See Blood Clotting.

The cilium.  Behe argued in 2007 that intraflagellar transport, which builds the cilium, deepens the problem.8  Its proteins are related to the coat proteins of cell vesicles, and one study finds that two of its three sub-complexes "likely arose" from the third by duplication.38  That is a family tree of parts.  No step has been rebuilt.

The immune system.  Behe wrote that its complexity "dooms all Darwinian explanations to frustration."1  Antibody genes are cut and joined by two proteins, RAG1 and RAG2.  Before those proteins were discovered it had been suggested that their cutting signals came from a jumping gene.  In 2005 the core of RAG1 was matched to the Transib family of jumping genes.39  In 2016 a living jumping gene carrying both was found in the lancelet, and its proteins cut and rejoined DNA in the test tube.40  This is co-option with the precursor in hand.  It covers one component.  The 2016 authors also wrote that before their work "definitive evidence for this is lacking", eleven years after the court called the matter confirmed.40

Vesicle transport.  The coats of vesicles share a fold with the nuclear pore, which suggests "a common evolutionary origin".41  Behe's argument concerned the tag, receptor and gate.  The coat evidence does not address them.

The bombardier beetle.  This is a creationist example.  Morris wrote that Duane Gish "has always argued" from it.4  Behe declined it:  "All we can conclude at this point is that Darwinian evolution might have occured."1  The first protein survey of the gland, in 2025, began by saying its molecular basis "is not well characterized."42

The ministries

Answers in Genesis states the argument without Behe's limits:  "Irreducible complexity cannot be produced by evolution because evolution is restricted to step-by-step change where every change must give a survival advantage."  Its examples include "the eye, human knee joint, and the upright stature of humans."4343  Behe ruled out whole organs in 2001.6  Tas Walker of Creation Ministries International conceded the logical point, "Philosophically it is impossible to prove that a particular structure cannot be explained by small, incremental steps", and also claimed that "every biological pathway is irreducibly complex".4444  No ministry page found warns against overusing the argument.  The limits Answers in Genesis names for intelligent design are theological.

Part 5: The logic of the claim

"No gradual route exists" is a universal negative.  It would require checking every possible route, and nobody can.

"Indirect routes are improbable" is a different kind of claim.  It can be tested, but only after someone states a route and the numbers that price it:  population size, generations, mutation rate, and the count of steps with no advantage.  Behe's 2000 definition asks for this.  It has been done for single proteins and for malaria.  It has not been done for a cilium or a flagellum.

Each side has a test it could pass.

Side What would count Status
Design A system whose parts have no relatives elsewhere and which has no simpler working relative in any organism Not shown. Relatives are known for many parts of each system.45
Evolution A rebuilt route in which each step's advantage is measured Not shown for any of the five systems.46

Behe's 2019 criterion is reasonable in form.  It has two conditions, and the first is unmet, because "good physical reasons" means a calculation.  The critics' reply that routes are conceivable is also unfinished.  Relatedness shows where parts came from.  It does not show that each step paid.  Behe made this point in 2006:  "even in principle sequence comparisons can't show whether a structure arose by random mutation and natural selection."1  Lukeš and colleagues state the same duty from the other side:  it is "the job of evolutionary biologists to show how selective advantage can account for each step".23

Objections and responses

"Irreducible complexity was refuted at Dover."  The court found the argument negative and its definition defective, and Behe had conceded the defect in print.  The refutation it cites consists of homology papers.  The opinion also misattributes a phrase and misstates the book on co-option.

"Behe keeps moving the goalposts."  The definition did change.  The central claim did not.  Page 40 of the 1996 book already says indirect routes are possible and improbable.  The fair complaint is that the improbability was never priced.

"The flagellum's parts are found elsewhere, so it could evolve."  Behe listed other uses of ciliary parts in 1996.  Orr, a critic, called wholesale recruitment "unlikely".13  A list of relatives is a list of candidates.

"The fungal ring proves that machines gain parts."  It proves that a copy can become required by loss.  The authors call the generality "unknown".

"Evolutionists cannot explain it, so it was designed."  This is the weak form of the design argument.  Behe's stronger form is that purposeful arrangement is positive evidence.  That inference is examined in The Design Inference, which finds that no probability has been computed for any machine.

"Thirty years without a route is decisive."  The immune system shows why it is not.  A proposal older than the discovery of the two proteins waited until 2016 for its precursor.

Issues and unanswered questions

Weaknesses on the design side:

  1. No system has had its full parts list tested by removal, one part at a time and in pairs. Pennock's 1999 criticism stands.
  2. The defect admitted in 2001 was never repaired in print. The argument moved to numbers instead.
  3. The probability of an indirect route has not been calculated for any named machine. Behe and Snoke's model covers one route to one feature in one protein, by their own statement.
  4. The 1996 claim about the immune system was followed by the discovery of the precursor of its cutting enzyme.
  5. Ministry versions drop the limits Behe kept. A skeptic who reads page 40 will see that.
  6. The argument is silent on the age of the earth. Behe accepts common descent.

Weaknesses on the critics' side:

  1. No route to any of the five systems has been rebuilt with each step's advantage measured.
  2. The critics contradict each other. Orr called co-option unlikely. Miller made it his main answer. Thornhill and Ussery said Orr's own route cannot produce the result.
  3. Doolittle, Miller, Pennock and the court each misstated a fact or Behe's position.
  4. The best laboratory cases work by loss.
  5. Joanna Masel's group called multi-step adaptations whose intermediates are all worse "one of the longest standing problems in evolutionary biology."47 Their answer is a model.

Open questions:  can the count of unselected steps be estimated for a real machine, as the 2000 definition requires, and does any flagellar or ciliary sub-assembly work alone when tested?  Observed cases of several mutations arising together are scored in Measuring Biological Information, The Long-Term Evolution Experiment, Phage Lambda and the OmpF Switch and HIV Evolution.  How rare working proteins are is in How Rare Is a Working Protein?.  A parallel dispute over an organ is in The Inverted Retina.

Verdict

The original definition rules out one thing:  building these systems by steady improvement of a single function.  Two critics grant that.  It does not rule out indirect routes, and Behe said so in 1996.  Muller's route is sound and has been seen once, in a case that lost abilities.  Co-option has been seen once with its precursor in hand, for one component of one system.

The design side is weakest where it claims a probability and gives no number.  The critics are weakest where they offer relatives in place of routes.  What survives is a modest and accurate statement:  for the cilium, the flagellum, the clotting cascade and gated transport, no measured step-by-step route exists in print.  That is a gap, and gaps can close.  The argument that can be tested is the numerical one, and it belongs to the articles on waiting times and observed mutations.

Sources

  1. 1.

    Behe, Michael J.  Darwin's Black Box: The Biochemical Challenge to Evolution.  Free Press.  10th anniversary edition.  2006.  First published 1996.  The body text is unchanged and the afterword, pages 255 to 272, is new.  Read from the library PDF, whose running heads print the page number.  The 1996 wording of the definition was checked on archive.org item darwinsblackboxb0000behe by paragraph search.  Copyright.

    1. 1a.

      Page 5:  “For the record, I have no reason to doubt that the universe is the billions of years old that physicists say it is. Further, I find the idea of common descent (that all organisms share a common ancestor) fairly convincing, and have no particular reason to doubt it.” ↩

    2. 1b.

      Page 39:  “By irreducibly complex I mean a single system composed of several well-matched, interacting parts that contribute to the basic function, wherein the removal of any one of the parts causes the system to effectively cease functioning.” ↩

    3. 1c.

      Page 39:  “An irreducibly complex system cannot be produced directly (that is, by continuously improving the initial function, which continues to work by the same mechanism) by slight, successive modifications of a precursor system, because any precursor to an irreducibly complex system that is missing a part is by definition nonfunctional.” ↩

    4. 1d.

      Page 40:  “Even if a system is irreducibly complex (and thus cannot have been produced directly), however, one can not definitively rule out the possibility of an indirect, circuitous route. As the complexity of an interacting system increases, though, the likelihood of such an indirect route drops precipitously.” ↩ ↩

    5. 1e.

      Page 42:  “If the hammer were gone, the mouse could dance all night on the platform without becoming pinned to the wooden base.” ↩

    6. 1f.

      Page 65:  “Therefore we can conclude that the cilium is irreducibly complex”.  Page 72:  “it is irreducibly complex. Gradual evolution of the flagellum, like the cilium, therefore faces mammoth hurdles.”  Page 86:  “the blood-clotting system fits the definition of irreducible complexity.”  Page 109:  “Because gated transport requires a minimum of three separate components to function, it is irreducibly complex.”  Page 138:  “demonstrated that each individually poses massive challenges to a putative step-by-step evolution.” ↩

    7. 1g.

      Page 258:  “I did go on to argue that indirect routes seem very unlikely, and that the more complex the system the more unlikely indirect routes become. But I did not assert that indirect routes were logically impossible, as he implies.” ↩ ↩

    8. 1h.

      Page 265:  “in the presence of manifest design, the onus of proof is on the one who denies the plain evidence of his eyes.” ↩

    9. 1i.

      Page 263:  “Astoundingly, Orr had nothing at all to say either about the mousetrap or about any of the biochemical examples from Darwin's Black Box.” ↩

    10. 1j.

      Page 66:  “To begin, microtubules occur in many cells and are usually used as mere structural supports, like girders, to prop up cell shape.” ↩

    11. 1k.

      Page 261:  “The system is irreducible, not the parts.” ↩

    12. 1l.

      Page 86:  “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.” ↩

    13. 1m.

      Page 139:  “but the complexity of the system dooms all Darwinian explanations to frustration.” ↩

    14. 1n.

      Page 36:  “All we can conclude at this point is that Darwinian evolution might have occured.”  Spelling as in the library copy. ↩

    15. 1o.

      Page 269:  “even in principle sequence comparisons can't show whether a structure arose by random mutation and natural selection.” ↩

  2. 2.

    Finnigan, Gregory C. et al.  “Evolution of Increased Complexity in a Molecular Machine.”  Nature 481:360–364.  Jan 19, 2012.  Read from the authors' manuscript at PubMed Central, which has no journal page breaks.  Copyright.  Mirror:  PMC

    1. 2a.

      Section closing discussion:  “We are aware of no other mechanistic analyses of a molecular machine’s evolutionary trajectory, so the generality of our observations is unknown.” ↩

    2. 2b.

      Section abstract:  “These losses were complementary, so both copies became obligate components with restricted spatial roles in the complex. Reintroducing a single historical mutation from each paralogue lineage into the resurrected ancestral proteins is sufficient to recapitulate their asymmetric degeneration and trigger the requirement for the more elaborate three-component ring.” ↩

    3. 2c.

      Section abstract:  “Despite much speculation, strong evidence of the mechanisms by which these assemblies evolved is lacking.” ↩

    4. 2d.

      Section closing discussion:  “our data indicate that simple degenerative mutations are sufficient to explain the historical increase in complexity of a crucial molecular machine. There is no need to invoke the acquisition of ‘novel’ functions caused by low-probability mutational combinations.” ↩

  3. 3.

    Darwin, Charles.  On the Origin of Species by Means of Natural Selection.  John Murray.  1st edition.  1859.  First edition, chapter VI.  Read from the Darwin Online transcription, which marks each printed page.  The key sentence is unchanged in the sixth edition of 1872.  Public domain.

    1. 3a.

      Page 189:  “If it could be demonstrated that any complex organ existed, which could not possibly have been formed by numerous, successive, slight modifications, my theory would absolutely break down. But I can find out no such case.” ↩

    2. 3b.

      Page 190:  “an organ originally constructed for one purpose, namely flotation, may be converted into one for a wholly different purpose, namely respiration.” ↩

  4. 4.

    Morris, Henry M.  “Intelligent Design and/or Scientific Creationism.”  Acts & Facts 35(4).  Apr 2006.  Retrieved Sep 29, 2026.  Morris founded the Institute for Creation Research.  Copyright.

    1. 4a.

      Paragraph 11:  “Thus, theistic evolution is quite compatible with Intelligent Design (Michael Behe himself admits to being an evolutionist).” ↩

    2. 4b.

      Paragraph 10:  “Duane Gish has always argued that such creatures as the butterfly and the bombardier beetle could not possibly have arisen by chance variation and natural selection.” ↩

  5. 5.

    Behe, Michael J.  “In Defense of the Irreducibility of the Blood Clotting Cascade: Response to Russell Doolittle, Ken Miller and Keith Robison.”  Discovery Institute.  Jul 31, 2000.  Retrieved Sep 29, 2026.  A web essay, not a journal paper.  Copyright.

    1. 5a.

      Section final section, before the references:  “So with those thoughts in mind, I offer the following tentative “evolutionary” definition of irreducible complexity:”.  The page stores the quotation marks around the word evolutionary as control characters, so a plain search for the phrase fails. ↩

    2. 5b.

      Section final section, before the references:  “An irreducibly complex evolutionary pathway is one that contains one or more unselected steps (that is, one or more necessary-but-unselected mutations). The degree of irreducible complexity is the number of unselected steps in the pathway.” ↩ ↩

    3. 5c.

      Section II. Russell Doolittle’s Criticism:  “pregnancy uniformly results in fatal uterine bleeding around the tenth day of gestation.”  Behe is quoting Bugge and others, Cell 87:709–719, 1996, which was not read for this article. ↩

  6. 6.

    Behe, Michael J.  “Reply to My Critics: A Response to Reviews of Darwin's Black Box: The Biochemical Challenge to Evolution.”  Biology and Philosophy 16(5):685–709.  2001.  The pages are those printed on the article.  The publisher's metadata gives 683 to 707.  Copyright.

    1. 6a.

      Page 693:  “As I constructed it the definition is equivocal; it doesn’t distinguish between systems that necessarily must have several parts because of the”.  Page 694:  “mechanism by which they perform the task at hand, and systems which use a number of parts to do a task which could in theory be done with one.” ↩

    2. 6b.

      Page 694:  “a single system which is necessarily composed of several well-matched, interacting parts that contribute to the basic function, and where the removal of any one of the parts causes the system to effectively cease functioning.” ↩ ↩

    3. 6c.

      Page 695:  “The current definition puts the focus on removing a part from an already-functioning system.”  “Thus there is an asymmetry between my current definition of irreducible complexity and the task facing natural selection. I hope to repair this defect in future work.” ↩ ↩

    4. 6d.

      Page 694:  “If that were to happen further mutations might change the system such that, although it functioned when both components were present, in the absence of part B, part A wouldn’t have even a weak activity.” ↩

    5. 6e.

      Page 692:  “Whole organs, such as lungs or swim bladders, are not “single systems.”” ↩

  7. 7.

    Dembski, William A.  No Free Lunch: Why Specified Complexity Cannot Be Purchased without Intelligence.  Rowman and Littlefield.  2002.  Read on archive.org item nofreelunchwhysp0000demb, which cannot be borrowed, by paragraph search.  Not stored.  Printed page is leaf minus 32, from the public leaf map and from the book's own index entry "irreducible core, 285, 287", which matches the leaves found.  Copyright.

    1. 7a.

      Page 285:  “A system performing a given basic function is irreducibly complex if it includes a set of well-matched, mutually interacting, nonarbitrarily individuated parts such that each part in the set is indispensable to maintaining the system’s basic, and therefore original, function.” ↩ ↩

    2. 7b.

      Page 287:  “If the irreducible core of an irreducibly complex system consists of one or only a few parts, there may be no insuperable obstacle to the Darwinian mechanism explaining how that system arose in one fell swoop.” ↩

  8. 8.

    Behe, Michael J.  The Edge of Evolution: The Search for the Limits of Darwinism.  Free Press.  2007.  Wording from the publisher's e-book in the library.  Printed pages from archive.org item isbn_9780743296205, which cannot be borrowed, by paragraph search:  page is leaf minus 14, anchored on the running heads of leaves 150 to 160 and on the section heading found on leaf 134.  Copyright.

    1. 8a.

      Page 120:  “Although closely related to it, the new concepts gauge difficulties for random mutation at a much finer level than does irreducible complexity, and are more appropriate to defining the edge of evolution. If irreducible complexity is likened to a rough measuring tool—say, a yardstick with no markings—then the new criteria can be thought of as a ruler subdivided into millimeters.” ↩ ↩

    2. 8b.

      Page 121:  “The concept of irreducible complexity, with its broad focus on the “parts” of a system, passes over the fact that a part might itself be a special piece that needs explaining in terms of many steps.” ↩

    3. 8c.

      Chapter 5, section Irreducible Complexity Squared:  “IFT exponentially increases the difficulty of explaining the irreducibly complex cilium.”  Wording from the e-book.  The printed page was not found by search. ↩

  9. 9.

    Behe, Michael J.  Darwin Devolves: The New Science About DNA That Challenges Evolution.  HarperOne.  2019.  Read from the library e-book, which carries the print edition's page anchors.  Copyright.

    1. 9a.

      Page 237:  “On a much finer scale than the examples I discussed in Darwin’s Black Box, structural and functional protein features that require multiple amino-acid residues call to mind irreducible complexity.” ↩ ↩

    2. 9b.

      Page 246:  “Irreducible complexity is an additional independent hurdle for evolution. The need for multiple coordinated mutations is a different problem from the degradation of genetic information or the constricted fit of a system to its current role”. ↩

    3. 9c.

      Page 232:  “The appropriate straightforward criterion is this: if there are good physical reasons to think Darwinian routes wouldn’t work and if after a diligent search no evidence is found that they do, then the theory has failed.” ↩

  10. 10.

    The published PDF of Behe 2001 was searched for "unselected" and returns nothing.  Section 1.4 of that paper, "Defects in the definition of irreducible complexity", runs from page 692 to page 695 and contains the two admissions quoted here.  The definition by unselected steps is in the Discovery Institute essay dated 31 July 2000.5  Behe also quotes his own 1996 definition with small changes of wording in 2000, 2001 and 2006 ("which is composed of" and "and where the removal").  The meaning is the same. ↩

  11. 11.

    Muller, Hermann J.  “Genetic Variability, Twin Hybrids and Constant Hybrids, in a Case of Balanced Lethal Factors.”  Genetics 3(5):422–499.  Sep 1918.  A page scan.  Its text layer garbles the italic clause, so the wording was read from tesseract OCR of pages 463 and 464 at 300 dots per inch.  Public domain.

    1. 11a.

      Page 464:  “thus a complicated machine was gradually built up whose effective working was dependent upon the interlocking action of very numerous different elementary parts or factors, and many of the characters and factors which, when new, were originally merely an asset finally became necessary because other necessary characters and factors had subsequently become changed so as to be dependent on the former.”  The sentence begins at the foot of page 463. ↩

    2. 11b.

      Page 464:  “Although this conclusion had suggested itself to the writer in 1912 it would manifestly have been very difficult to obtain experimental evidence for it”. ↩

  12. 12.

    Muller, H. J.  “Reversibility in Evolution Considered from the Standpoint of Genetics.”  Biological Reviews 14(3):261–280.  1939.  Read from the public OCR text of the whole volume on archive.org.  Pages from the running heads.  The OCR has slips in other sentences.  Copyright.

    1. 12a.

      Page 271:  “after becoming once established (with or without the aid of selection)”.  Page 272:  “Being thus finally woven, as it were, into the most intimate fabric of the organism, the once novel character can no longer be withdrawn with impunity, and may have become vitally necessary.” ↩

    2. 12b.

      Page 271:  “For there is undoubtedly far more transfer of function in evolution, largely of a non-adaptive, fortuitous nature, than has commonly been realized.” ↩

  13. 13.

    Orr, H. Allen.  “Darwin v. Intelligent Design (Again).”  Boston Review 21(6).  Dec 1996.  The December 1996 and January 1997 issue.  Read from an archived copy of the magazine's page.  Copyright.

    1. 13a.

      Section Reducible Complexity:  “Some part (A) initially does some job (and not very well, perhaps). Another part (B) later gets added because it helps A. This new part isn't essential, it merely improves things. But later on, A (or something else) may change in such a way that B now becomes indispensable.” ↩

    2. 13b.

      Section Reducible Complexity:  “This scenario was first hinted at by the geneticist H. J. Muller in 1918 and worked out in some detail in 1939.” ↩

    3. 13c.

      Section Reducible Complexity:  “But this is also unlikely. You may as well hope that half your car's transmission will suddenly help out in the airbag department. Such things might happen very, very rarely, but they surely do not offer a general solution to irreducible complexity.” ↩ ↩

  14. 14.

    Thornhill, Richard H., and David W. Ussery.  “A Classification of Possible Routes of Darwinian Evolution.”  Journal of Theoretical Biology 203(2):111–116.  2000.  Read from the second author's own web copy, which has no page breaks.  Copyright.

    1. 14a.

      Section Introduction:  “However, this could only produce a complex, rather than an irreducibly complex, structure.” ↩

    2. 14b.

      Section (b) Parallel direct Darwinian evolution:  “Parallel direct Darwinian evolution can generate irreducibly complex structures, but not irreducibly complex structures of functionally indivisible components (Fig. 1), and this is the valid conclusion to draw from Behe's thesis.” ↩

  15. 15.

    Behe, Michael J.  “A Blind Man Carrying a Legless Man Can Safely Cross the Street: Experimentally Confirming the Limits to Darwinian Evolution.”  Evolution News.  Jan 11, 2012.  Retrieved Sep 29, 2026.  Published by the Discovery Institute.  Behe spells the first author's name Finnegan.  Copyright.

    1. 15a.

      Paragraph 5:  “The work of Finnegan et al. (2012) strikes me as quite thorough and elegant. I have no reason to doubt that events could have unfolded that way.” ↩

    2. 15b.

      Paragraph 5:  “All of the functional parts of the system were already in place before random mutation began to degrade them.” ↩

  16. 16.

    Gould, Stephen Jay, and Elisabeth S. Vrba.  “Exaptation: A Missing Term in the Science of Form.”  Paleobiology 8(1):4–15.  1982.  Read on the archive.org microfilm issue sim_paleobiology_winter-1982_8_1, which cannot be borrowed, by paragraph search.  Not stored.  Page is leaf minus 3, from the public leaf map at leaves 12 and 18 and from the title page on leaf 7, which is page 4.  Copyright.

    1. 16a.

      Page 4:  “features that now enhance fitness but were not built by natural selection for their current role. We propose that such features be called exaptations”. ↩

  17. 17.

    Bridgham, Jamie T. et al.  “Evolution of Hormone-Receptor Complexity by Molecular Exploitation.”  Science 312(5770):97–101.  Apr 7, 2006.  Read on the archive.org microfilm issue sim_science_2006-04-07_312_5770, which cannot be borrowed, by paragraph search.  Not stored.  Page is leaf minus 7, from the public leaf map at leaves 60, 104 and 107.  Copyright.

    1. 17a.

      Page 97:  “Our results indicate that tight interactions can evolve by molecular exploitation—recruitment of an older molecule, previously constrained for a different role, into a new functional complex.” ↩

  18. 18.

    Behe, Michael J.  “Michael Behe on Molecular Exploitation and the Theory of Irreducible Complexity.”  Discovery Institute.  Apr 2006.  Retrieved Sep 29, 2026.  Also circulated as "The Lamest Attempt Yet to Answer the Challenge Irreducible Complexity Poses for Darwinian Evolution".  The page carries no reliable date;  the text refers to the 7 April issue of Science.  Copyright.

    1. 18a.

      Paragraph 4:  “Their “system”, on the other hand, consists of just a single protein and its ligand.”  Paragraph 5:  “Nothing new was produced in the experiment; rather, the pre-existing ability of the protein to bind several molecules was simply weakened.” ↩

  19. 19.

    Cairns-Smith, A. G.  Seven Clues to the Origin of Life: A Scientific Detective Story.  Cambridge University Press.  1985.  Borrowed on archive.org and read by search.  Not stored.  Pages are from the public leaf map, which gives a constant leaf minus 16 from page 59 to page 115.  The pages were not checked against a running head.  Copyright.

    1. 19a.

      Page 59:  “How can you build any kind of arch gradually?”  Page 60:  “Then you would remove stones to leave the ' paradoxical ' structure.” ↩

    2. 19b.

      Page 60:  “Nothing can be touched or the whole edifice will collapse.” ↩

  20. 20.

    Behe, Michael J., and David W. Snoke.  “Simulating Evolution by Gene Duplication of Protein Features That Require Multiple Amino Acid Residues.”  Protein Science 13(10):2651–2664.  2004.  Publisher PDF in the library.  The exponents are printed as superscripts and are lost in the text layer.  Copyright.  Mirror:  PMC

    1. 20a.

      Page 2651:  “the production of novel protein features that require the participation of two or more amino acid residues simply by multiple point mutations in duplicated genes would entail population sizes of no less than”.  The figures that follow are ten to the ninth power organisms and ten to the eighth power generations. ↩

    2. 20b.

      Page 2658:  “We strongly emphasize that results bearing on the efficiency of this one pathway as a conduit for Darwinian evolution say little or nothing about the efficiency of other possible pathways.” ↩

  21. 21.

    Lynch, Michael.  “Simple Evolutionary Pathways to Complex Proteins.”  Protein Science 14(9):2217–2225.  2005.  Publisher PDF in the library.  Copyright.  Mirror:  PMC

    1. 21a.

      Page 2217:  “Behe and Snoke consider a situation in which the intermediate steps to a new protein are neutral and involve nonfunctional products.”  The sentence before it says the model they present is non-Darwinian, with the hyphen at a line break. ↩

  22. 22.

    Stoltzfus, Arlin.  “On the Possibility of Constructive Neutral Evolution.”  Journal of Molecular Evolution 49(2):169–181.  Aug 1999.  Read on the archive.org microfilm issue sim_journal-of-molecular-evolution_1999-08_49_2, which cannot be borrowed, by paragraph search.  Not stored.  Page is leaf plus 159, from the public leaf map at leaves 10 and 40 and from the title page on leaf 10.  Copyright.

    1. 22a.

      Page 170:  “it may be said that constructive neutral evolution is possible, in theory.” ↩

  23. 23.

    Lukeš, Julius et al.  “How a Neutral Evolutionary Ratchet Can Build Cellular Complexity.”  IUBMB Life 63(7):528–537.  2011.  Abstract only.  The body was not obtained.  Copyright.

    1. 23a.

      Section abstract:  “We suggest that CNE-based evolutionary scenarios are in these and other cases less forced than the selectionist or adaptationist narratives that are generally told.” ↩

    2. 23b.

      Section abstract:  “it is typically understood to be the job of evolutionary biologists to show how selective advantage can account for each step in their origin and subsequent growth in complexity.” ↩

  24. 24.

    Lenski, Richard E. et al.  “The Evolutionary Origin of Complex Features.”  Nature 423:139–144.  May 8, 2003.  Publisher PDF in the library.  Copyright.

    1. 24a.

      Page 143:  “At the other extreme, 50 populations evolved in an environment where only EQU was rewarded, and no simpler function yielded energy.”  “Indeed, none of these populations evolved EQU”. ↩

  25. 25.

    Coyne, Jerry A.  “God in the Details.”  Nature 383:227–228.  Sep 19, 1996.  A review of Darwin's Black Box.  Read on the archive.org microfilm issue sim_nature-uk_1996-09-19_383_6597, which cannot be borrowed, by paragraph search.  Not stored.  Page is leaf plus 175, from the public leaf map at leaves 60 and 100, which puts the review on the pages of its published citation.  Copyright.

    1. 25a.

      Page 227:  “There is no doubt that the pathways described by Behe are dauntingly complex, and their evolution will be hard to unravel.”  “we may forever be unable to envisage the first proto-pathways. It is not valid, however, to assume that, because one man cannot imagine such pathways, they could not have existed.” ↩

  26. 26.

    Doolittle, Russell F.  “A Delicate Balance.”  Boston Review 22(1).  Feb 1997.  The February and March 1997 issue.  Read from an archived copy of the magazine's page.  Copyright.

    1. 26a.

      Paragraph last before the notes:  “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.” ↩

  27. 27.

    Miller, Kenneth R.  Finding Darwin's God: A Scientist's Search for Common Ground between God and Evolution.  Cliff Street Books.  1999.  Read on archive.org item findingdarwinsgo00mill, which cannot be borrowed, by paragraph search.  Not stored.  Page is leaf minus 16, from the public leaf map at leaves 148, 159 and 280.  Copyright.

    1. 27a.

      Page 143:  “In nature, we can find scores of cilia lacking one or more of the components supposedly essential to the function of the apparatus.” ↩

  28. 28.

    Miller, Kenneth R.  “The Mousetrap Analogy, or Trapped by Design.”  millerandlevine.com.  c. 2000.  Undated.  It answers Behe's essay of July 2000.  Copyright.

    1. 28a.

      Paragraph second from last:  “I removed two parts from a mousetrap (leaving just the base, spring, and hammer), and used that 3-part device as a functional tie-clip.”  Paragraph last:  “and then argues that no portion of an "irreducibly complex" system (such as a mousetrap) could have any function.” ↩

  29. 29.

    McDonald, John H.  “A Reducibly Complex Mousetrap.”  University of Delaware.  2002.  The second version of the page.  McDonald is a geneticist.  Copyright.

    1. 29a.

      Paragraph 2:  “they're not intended as an analogy of how evolution works.” ↩

  30. 30.

    Behe, Michael J.  “A Mousetrap Defended: Response to Critics.”  Discovery Institute.  Jul 31, 2000.  Retrieved Sep 29, 2026.  Copyright.

    1. 30a.

      Section reply to McDonald:  “If several separate mutations have to occur before we go from one functional trap to the next, then a Darwinian process is effectively ruled out, because the probability of getting multiple unselected mutations that eventually lead to a specific complex structure is prohibitive.” ↩

  31. 31.

    Pennock, Robert T.  Tower of Babel: The Evidence against the New Creationism.  MIT Press.  1999.  Read on archive.org item towerofbabelevid00penn by paragraph search.  Not stored.  Pages from the public leaf map, leaves 294 to 298.  Second anchor:  Behe 2001, note 5, cites the same passage as pages 267 to 268.  Copyright.

    1. 31a.

      Page 267:  “Behe claims that there could never be any functional intermediates that natural selection could have selected for on the”.  Page 268:  “way to any irreducibly complex system, but he can't get the empirical conclusion from his "by definition" conceptual argument. The strong empirical premise he needs is false.” ↩

    2. 31b.

      Page 266:  “However, Behe has not done this for any of his examples. In part, this is simply because not enough is known about them.” ↩

  32. 32.

    Draper, Paul.  “Irreducible Complexity and Darwinian Gradualism: A Reply to Michael J. Behe.”  Faith and Philosophy 19(1):3–21.  2002.  A scan with a text layer, read from an archived copy of the journal's own file.  Pages from the running heads.  Copyright.

    1. 32a.

      Page 6:  “For Behe does not mean to rule out by definition functioning precursors to irreducibly complex systems that are missing parts but serve either a different function or the same function by a different mechanism.” ↩

    2. 32b.

      Page 3:  “I conclude that, while Behe successfully rules out some Darwinian paths to the biochemical systems he discusses, others remain open. Thus, his argument against Darwinian gradualism (and ipso facto his argument for intelligent design) is at best incomplete.”  Page 11:  “this leaves open the very real possibility of an indirect Darwinian”. ↩

  33. 33.

    Boudry, Maarten et al.  “Irreducible Incoherence and Intelligent Design: A Look into the Conceptual Toolbox of a Pseudoscience.”  The Quarterly Review of Biology 85(4):473–482.  Dec 2010.  Read from the authors' accepted manuscript at Ghent University.  Page numbers are the manuscript's, not the journal's.  Copyright.

    1. 33a.

      Section before "Dembski's Conceptual Remedy":  “non-functionality claim becomes either trivial (weak version) or plainly wrong (strong”.  Page 5 of the manuscript.  The journal page was not checked. ↩

  34. 34.

    United States District Court for the Middle District of Pennsylvania.  Kitzmiller v. Dover Area School District, Memorandum Opinion.  Dec 20, 2005.  Written by Judge John E. Jones III.  Reported at 400 F. Supp. 2d 707.  Pages are those of the 139-page slip opinion.  A court record.

    1. 34a.

      Page 72:  “Irreducible complexity is a negative argument against evolution, not proof of design, a point conceded by defense expert Professor Minnich.”  Page 74:  “however, he has failed to do so even four years after elucidating his defect.” ↩

    2. 34b.

      Page 79:  “been refuted in peer-reviewed research papers and has been rejected by the scientific community at large.” ↩

    3. 34c.

      Page 78:  “chapters about the evolution of the immune system; however, he simply insisted that this was still not sufficient evidence of evolution, and that it was not “good enough.”” ↩

    4. 34d.

      Page 75:  “Professor Behe attempts to exclude the phenomenon of exaptation by definitional fiat”. ↩

  35. 35.

    United States District Court for the Middle District of Pennsylvania.  Kitzmiller v. Dover Area School District, Trial Transcript, Day 12, Afternoon Session.  Oct 19, 2005.  Cross-examination of Michael Behe by Eric Rothschild.  Page numbers are printed at the head of each transcript page.  A court record.

    1. 35a.

      Page 16, line 16:  “So these are not good enough?”  Page 16, line 17:  “They're wonderful articles.” ↩

    2. 35b.

      Page 23, line 3:  “but again I am quite skeptical, although I”. ↩

  36. 36.

    Abby, Sophie S., and Eduardo P. C. Rocha.  “The Non-Flagellar Type III Secretion System Evolved from the Bacterial Flagellum and Diversified into Host-Cell Adapted Systems.”  PLoS Genetics 8(9):e1002983.  2012.  CC BY.  Mirror:  PMC

    1. 36a.

      Section abstract:  “Phylogenomic and comparative analyses of these systems argue that the NF-T3SS arose from an exaptation of the flagellum”. ↩

  37. 37.

    Doolittle, Russell F. et al.  “Genomic Evidence for a Simpler Clotting Scheme in Jawless Vertebrates.”  Journal of Molecular Evolution 66(2):185–196.  2008.  Abstract only, from Europe PMC.  The body was not obtained.  Copyright.

    1. 37a.

      Section abstract:  “a lamprey system with a smaller number of components than is found in other vertebrates in that factors V and VIII seem to be represented by a single gene, and factor IX, which is ordinarily a cofactor of factor VIII, is not present.” ↩ ↩

  38. 38.

    van Dam, Teunis J. P. et al.  “Evolution of Modular Intraflagellar Transport from a Coatomer-like Progenitor.”  Proceedings of the National Academy of Sciences 110(17):6943–6948.  2013.  Read at PubMed Central.  Copyright.  Mirror:  PMC

    1. 38a.

      Section abstract:  “We find that IFT-A and the BBSome likely arose from an IFT-B–like complex by intracomplex subunit duplication.” ↩ ↩

  39. 39.

    Kapitonov, Vladimir V., and Jerzy Jurka.  “RAG1 Core and V(D)J Recombination Signal Sequences Were Derived from Transib Transposons.”  PLoS Biology 3(6):e181.  Jun 2005.  CC BY.

    1. 39a.

      Section abstract:  “Our results provide the first direct evidence linking RAG1 and RSSs to a specific”.  Section Introduction:  “Even prior to the discovery of RAG1 and RAG2, it had”. ↩

  40. 40.

    Huang, Shengfeng et al.  “Discovery of an Active RAG Transposon Illuminates the Origins of V(D)J Recombination.”  Cell 166(1):102–114.  2016.  Read at PubMed Central.  Copyright.  Mirror:  PMC

    1. 40a.

      Section Summary:  “We demonstrate that ProtoRAG was recently active in the lancelet germline and that the lancelet RAG1/2-like proteins can mediate TIR-dependent transposon excision, host DNA recombination, transposition, and low efficiency TIR rejoining using reaction mechanisms similar to those used by vertebrate RAGs.” ↩ ↩

    2. 40b.

      Section Summary:  “RAG1/2 are proposed to have arisen from a transposable element, but definitive evidence for this is lacking.” ↩

  41. 41.

    Devos, Damien et al.  “Components of Coated Vesicles and Nuclear Pore Complexes Share a Common Molecular Architecture.”  PLoS Biology 2(12):e380.  2004.  CC BY.  Mirror:  PMC

    1. 41a.

      Section abstract:  “These similarities suggest a common evolutionary origin for nuclear pore complexes and coated vesicles in an early membrane-curving module”. ↩ ↩

  42. 42.

    Vogel, Heiko et al.  “Molecular Basis of the Explosive Defence Response in the Bombardier Beetle Brachinus crepitans.”  Royal Society Open Science 12(5):241823.  May 21, 2025.  CC BY.  Mirror:  PMC

    1. 42a.

      Section abstract:  “The molecular basis of this response is not well characterized.” ↩

  43. 43.

    Burgess, Stuart.  “What about Design Arguments like "Irreducible Complexity"?”  The New Answers Book 4.  Master Books.  Edited by Ken Ham.  2013.  Retrieved Sep 29, 2026.  Chapter 29, republished by Answers in Genesis.  Burgess is a professor of engineering design.  Copyright.

    1. 43a.

      Section Irreducible Complexity:  “evolution is restricted to”.  The full sentence, which runs across line breaks in the saved page:  Irreducible complexity cannot be produced by evolution because evolution is restricted to step-by-step change where every change must give a survival advantage. ↩

    2. 43b.

      Section Irreducible Complexity:  “Other examples of irreducible complexity are the eye”.  The sentence goes on to name the human knee joint and the upright stature of humans, each with a reference number. ↩

  44. 44.

    Walker, Tas.  “Irreducible Complexity and Cul-de-sacs.”  Creation Ministries International.  Sep 14, 2016.  Retrieved Sep 29, 2026.  A reply to a reader's letter.  Copyright.

    1. 44a.

      Paragraph second reply:  “Philosophically it is impossible to prove that a particular structure cannot be explained by small, incremental steps because Darwinists can always respond”. ↩

    2. 44b.

      Paragraph third reply:  “Furthermore, every biological pathway is irreducibly complex, requiring multiple steps to produce the required biochemical output.” ↩

  45. 45.

    Relatives are reported for the cutting enzyme of the immune system,40 for the transport proteins of the cilium,38 for vesicle coats41 and for the clotting factors.37  For the flagellum the count of proteins with no known relative has been put at figures from seventeen to forty by writers on both sides, and it is left to the child article. ↩

  46. 46.

    The twelve are the studies with a biological subject that are cited in this article or worked up in its literature review.  Three tested proteins in a laboratory for the function in question:  Bridgham 2006, Finnigan 2012 and Huang 2016.  Nine compared sequences, structures or gene lists:  Devos 2004, Kapitonov 2005, Jékely 2006, Liu and Ochman 2007, Doolittle 2008, Clements 2009, Abby 2012, van Dam 2013 and Vogel 2025.  Of the three, only Huang 2016 concerns a system Behe named.  None reports the fitness of an organism at each step.  Lenski 2003 measured every step, in software. ↩

  47. 47.

    Trotter, Meredith V. et al.  “Cryptic Genetic Variation Can Make "Irreducible Complexity" a Common Mode of Adaptation in Sexual Populations.”  Evolution 68(12):3357–3367.  2014.  Read from the arXiv preprint, whose wording may differ from the journal's.

    1. 47a.

      Section abstract:  “partial combinations are less fit than the original genotype) is one of the longest standing”.  Preprint abstract.  The sentence ends "problems in evolutionary biology." ↩