In Darwin's Black Box, Michael Behe defines irreducible complexity as a system composed of several well-matched, interacting parts that contribute to a basic function, such that removing any one part causes the system to stop functioning. He argues that this creates a problem for a direct Darwinian pathway in which parts are added one at a time: a proposed intermediate must contribute to survival or reproduction before the later parts exist, rather than becoming useful only after the complete system appears. The bacterial flagellum is his best-known example. This is a challenge to a particular account of the evolutionary route, not by itself a demonstration that no indirect route involving co-option, altered functions, or loss of parts could produce the system.
Behe's argument has faced substantial criticism from evolutionary biologists. A key line of evidence is that some proteins in the flagellar basal body and export apparatus are homologous to components of the type III secretion system, a different and simpler protein-export system that is functional in its own right. This is consistent with hypotheses involving common ancestry and exaptation: some components may have performed other functions before becoming incorporated into a flagellum. The comparison is not, however, a complete reconstruction of flagellar evolution. Homology alone does not establish the direction of evolutionary change or identify every historical intermediate, and the two systems are not identical. It therefore weakens the inference that all flagellar components had to arise simultaneously without by itself settling every detail of the pathway.
Moreover, co-option and changes of function are well-documented evolutionary processes: an existing structure can be modified or recruited for a new role, sometimes while retaining aspects of its earlier function. In discussions of the flagellum, Mark Pallen and Nicholas Matzke have examined homology, modularity, and possible routes by which systems with earlier functions might have contributed to a more elaborate motility apparatus. Such analyses make an indirect evolutionary history scientifically plausible and challenge the claim that every part had to originate for the flagellum's present function at once. They should not be presented as a uniquely established reconstruction of every step; historical pathways remain claims to be assessed against comparative, genetic, and structural evidence.
Another possible rebuttal involves scaffolding: an intermediate system may have included additional components that stabilized or enabled a function and were later lost after the remaining components became sufficiently integrated. Such intermediates could have had functions different from the final flagellum, allowing evolutionary change without requiring every present-day part to be useful in its present role from the beginning. This is a legitimate mechanism in evolutionary reasoning, but for the flagellum it is a proposed type of pathway rather than proof that a particular temporary scaffold existed. Comparative genomics, structural biology, and experimental work are needed to distinguish historically supported routes from merely possible ones.
The debate over irreducible complexity and the bacterial flagellum also received extensive public attention in Kitzmiller v. Dover Area School District (2005). The federal district court heard testimony about intelligent design, including testimony from Behe, and ruled that intelligent design could not be presented as a scientific alternative in the public-school science curriculum. The opinion criticized the evidential basis and scientific status claimed for irreducible complexity, while also addressing the religious context of the school policy. Its conclusions are important legal and educational history, but a judicial opinion is not an experiment or a scientific consensus statement; the biological assessment must rest on the evidence and arguments of evolutionary research.
In light of these considerations, the bacterial flagellum remains a legitimate case study in what a satisfactory evolutionary explanation must account for, and Behe's criticism successfully exposes the weakness of a simplistic story in which a complex machine is formed merely by adding one present-day part at a time. But homology, co-option, modularity, and possible scaffolding provide scientifically grounded reasons not to infer impossibility from complexity alone. They do not constitute a complete, uniquely demonstrated reconstruction of every flagellar component either. The measured conclusion is therefore that the flagellum is not a successful refutation of evolutionary biology, while unresolved details in its history are not proof that design or theism is true. A design conclusion would require independent, testable evidence rather than an argument from a gap in current evolutionary reconstruction.