Sign up ×
Physics Stack Exchange is a question and answer site for active researchers, academics and students of physics. It's 100% free.

Suppose a rigid body is invariant under a rotation around an axis $\mathsf{A}$ by a given angle $0 \leq \alpha_0 < 2\pi$ (and also every multiple of $\alpha_0$).

Is it true that in this case the axis $\mathsf{A}$ is a principal axis of the rigid body?

If so, how to prove it? Do you have any references for a proof?

share|cite|improve this question
I think that using Poinsot's ellipsoid (inertia ellipsoid) would serve as intuitive proof which with the aid of some formalism would legitimize it. – qoqosz Jun 6 '12 at 15:03
@qoqosz Could you provide some details for your idea? – martin Jun 7 '12 at 14:33
the moments of inertia around axes perpendicular to A must be all the same (they form a circle if $\alpha_0 < \pi$) - so it's a symmetrical top but I'm not sure if this leads to something. – qoqosz Jun 8 '12 at 17:20

3 Answers 3

up vote 3 down vote accepted

Your statement is true.


Let $\rho$ be the mass density of the rigid body.

Remember that the tensor of inertia $I$ is given by:

$$ \vec{v}^t I \vec{w} = \int d^3b\, \rho(\vec{b}) (\vec{v} \cdot \vec{w} - (\vec{v}\cdot \vec{b})(\vec{w}\cdot \vec{b})) $$

for all $\vec{v},\vec{w} \in \mathbb{R}^3$.

Now take an orthogonal matrix $O$ which represents a rotation around an axis $\mathsf{A}$ with direction vector $\vec{n}$, i.e. $O\vec{n} = \vec{n}$.

Your invariance means that $\rho$ is invariant under $O$, i.e. $\rho(O \vec{x}) = \rho(\vec{x})$ for all $\vec{x}$.

Next show that the inertia tensor commutes with $O$: $IO = OI$: $$ \begin{align*} \vec{v}^t I O \vec{w} &= \int d^3b\, \rho(\vec{b}) (\vec{v} \cdot O \vec{w} - (\vec{v}\cdot \vec{b})(O \vec{w}\cdot \vec{b})) \\ &= \int d^3b\, \rho(\vec{b}) (O^t \vec{v} \cdot \vec{w} - (\vec{v}\cdot \vec{b})(\vec{w}\cdot O^t \vec{b})) \\ &= \int d^3b\, \rho(O^t \vec{b}) (O^t \vec{v}\cdot \vec{w} - (O^t \vec{v}\cdot O^t \vec{b})(\vec{w}\cdot O^t \vec{b})) \\ &= \int d^3b\, \rho(\vec{b}) (O^t \vec{v}\cdot \vec{w} - (O^t \vec{v}\cdot \vec{b})(\vec{w}\cdot \vec{b})) \\ &= \int d^3b\, \rho(\vec{b}) ((\vec{v}^t O)^t\cdot \vec{w} - ((\vec{v}^t O)^t \cdot \vec{b})(\vec{w}\cdot \vec{b})) \\ &= \vec{v}^t O I \vec{w} \end{align*} $$

for all $\vec{v},\vec{w} \in \mathbb{R}^3$

Then one sees that $I\vec{n}$ is again an eigenvector of $O$ because $O(I\vec{n}) = IO\vec{n} = I\vec{n}$.

Now $I$ has only one real eigenvalue $1$ with eigenspace $\mathbb{R}\vec{n}$. This implies that there is a unique $\lambda$ with $I\vec{n} = \lambda \vec{n}$. Thus $\vec{n}$ is an eigenvector of $I$ i.e. $\vec{n}$ points along a principal axis of $I$.

share|cite|improve this answer

As a starting point, Wikipedia says on this issue:

The principal axes are often aligned with the object's symmetry axes. If a rigid body has an axis of symmetry of order $m$, meaning it is symmetrical under rotations of 360°/$m$ about the given axis, that axis is a principal axis. When , the rigid body is a symmetrical top. If a rigid body has at least two symmetry axes that are not parallel or perpendicular to each other, it is a spherical top, for example, a cube or any other Platonic solid.

But it gives no canonical reference.

share|cite|improve this answer
Thanks. Do you have a proof? – martin Jun 6 '12 at 13:31

At least as phrased ("by a given angle $\alpha_0$"), axis $\mathsf{A}$ can be trivially shown not to necessarily be a principal axis. If $\alpha_0$ is a multiple of $2\pi$, every rigid body is invariant under a rotation about any axis.

share|cite|improve this answer
Thanks, I added this constraint to the question. – martin Jun 6 '12 at 14:37

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.