The Falling-Cat Problem

Equivalent forms

\Delta\theta = \oint \vec{A}(\text{shape})\cdot d\vec{s} \neq 0 \;\text{with}\; L=0

I_f\omega_f = -I_b\omega_b

Reorientation with zero angular momentum: the cat is a swimmer in the space of shapes, paddling through geometry.

Symbol	Name	SI unit	Dimension	Range
$Δθ$	Net reorientationoutput Total body turn accumulated over one shape cycle	rad	1	0 – 3.1416
$β$	Bend angle Angle between the front and back body segments	rad	1	0 – 1.5708
$I_f$	Front inertia Moment of inertia of the front half about the twist axis	kg·m^2	ML^2	0.005 – 0.05
$I_b$	Back inertia Moment of inertia of the back half about the twist axis	kg·m^2	ML^2	0.005 – 0.05

Unit systems

Symbol	SI	CGS	Imperial
$Δθ$	rad	rad	deg
$β$	rad	rad	deg
$I_f$	kg·m^2	g·cm^2	slug·ft^2
$I_b$	kg·m^2	g·cm^2	slug·ft^2

Where it holds

Deformable bodies that conserve angular momentum (no external torque); the net rotation is a holonomy of the shape-space connection and is exact for slow ('adiabatic') shape changes.

Dimensional analysis

[ML^{2}][T^{-1}] = [ML^{2}T^{-1}]

[ML^{2}][T^{-1}] = [ML^{2}T^{-1}]

Both terms are angular momentum; their sum is zero (also

[ML^{2}T^{-1}])

.

\checkmark

Discovery

Étienne-Jules Marey (photographs); Kane & Scher (model) · 1894

Marey's 1894 chronophotographs of a dropped cat stunned the Académie des Sciences — the cat clearly turned over without pushing off anything. In 1969 Kane and Scher modelled it as a 'bend-and-twist' of two coupled cylinders, showing a net reorientation is possible with L = 0. It is a textbook example of a geometric (gauge) phase.

Try this

A cat dropped upside-down with zero spin still lands on its feet. No outside torque, yet it turns over — does it break physics?

By bending in the middle and changing each half's moment of inertia, the cat counter-rotates its front and back to net a half-turn — all while total angular momentum stays exactly zero.

Research status: stable

Real-world applications

Astronaut self-rotation and satellite attitude maneuvers without thrusters
Springboard and platform diving / gymnastics twists
Free-falling robot and drone reorientation
Understanding geometric (Berry) phases in mechanics

Common misconceptions

The cat pushes against the air to turn — air is irrelevant; it works in vacuum
The cat must start with some spin — it can start at rest and still flip
Angular momentum is created — it stays exactly zero throughout; only orientation changes

Experimental verification

High-speed video confirms cats execute the bend-and-twist

in \sim 0.3\,\mathrm{s}

; astronauts and divers use the identical trick to reorient with zero angular momentum; robotic 'space-cat' models reproduce the predicted

\Delta \theta per

cycle.

Derivation

Split the cat into front and back rigid halves bent at angle

\beta

.

Zero total angular momentum about the twist axis requires

I_f\omega _f + I_b\omega _b = 0

at every instant,

so \omega _b = -(I_f/I_b)\omega _f

.

If the cat tucks the front (small I_f) while twisting one way then extends it while twisting back, the per-cycle rotations do NOT cancel because I_f, I_b change with shape.

Integrating

\Delta \theta = \oint (I_f-I_b)/(I_f+I_b) d\psi

over the closed shape loop gives a nonzero net turn — a geometric phase.

Limiting cases

\beta = 0

(rigid body)⟶

\Delta \theta = 0

A rigid body with

L = 0

cannot rotate at all — the cat must deform to turn.

I_f = I_b

, symmetric bend⟶ Maximal counter-rotation efficiencyEqual halves trade rotation symmetrically; tucking one and extending the other breaks the tie to net a turn.

Maximal tuck/extend asymmetry⟶ Largest

\Delta \theta per

twistPulling the front legs in (small I) lets it spin far while the extended back (large I) barely moves.

What if…

What if the cat were perfectly rigid?

It could never turn over — a rigid body with zero angular momentum stays put forever.

What if an astronaut tries this in space?

Same physics: by cycling limb positions they reorient with no thrusters and no net angular momentum.

What if the cat keeps repeating the bend-twist cycle?

Each cycle adds another $\Delta \theta$ — it could keep slowly spinning purely by changing shape.

1

Counter-rotation rates

Given ·

I f:: 0.01
I b:: 0.03

Find · ratio

\omega _f/\omega _b

Steps

$L = I_f\omega _f + I_b\omega _b = 0$
$\omega _f = -(I_b/I_f)\omega _b$
Tucked front (I_f small) spins $3\times$ faster than the extended back

Result ·

\omega _f/\omega _b = -I_b/I_f = -3

2

Minimum drop to right

Given ·

height:: 0.5

Find · time available

Steps

Fall time $t = \sqrt{2h/g}$
$= \sqrt{2\times 0.5/9.8} \approx 0.32\,\mathrm{s}$
Cats need $\sim 30\,\mathrm{cm}$ to complete the righting reflex

Result ·

t = \sqrt{2h/g} = \sqrt{1/9.8} \approx 0.32\,\mathrm{s}

, enough for

the \sim 0.3\,\mathrm{s}

flip

Conservation of Angular Momentum→Moment of Inertia (Point Mass)→Rotational Kinetic Energy→Noether's Theorem→