Summary
- A rigid body’s orientation in space is always described by a single angular velocity vector about one axis. In fact, Euler’s rotation theorem tells us any rotation at an instant can be represented by one axis and an angle. Thus a sphere “spinning” cannot have two independent spin axes at the same time – any attempt to decompose it yields a single effective rotation axis.
- A uniform solid sphere has an isotropic inertia tensor (principal moments $I_1=I_2=I_3$), so its angular momentum $\mathbf{L}$ is always parallel to its angular velocity $\boldsymbol{\omega}$ (with $|\mathbf{L}|=I|\boldsymbol{\omega}|$). This means all axes through the centre are equivalent and stable for rotation, unlike a general ellipsoid.
- In torque-free motion the sphere simply continues spinning about a fixed inertial axis: its angular momentum is conserved. Geometrically, the free rotation can be visualised by Poinsot’s construction as an intersection of an energy ellipsoid and the angular-momentum sphere.
- For comparison, a triaxial ellipsoid has unequal principal inertias ($I_1<I_2<I_3$) and only two “safe” axes (largest or smallest inertia) for stable spin. The intermediate axis is unstable (the tennis‑racket effect). By contrast, a perfect sphere has no intermediate axis – all axes behave the same.
- Experiments and modern simulations confirm this analysis: a free uniform sphere in vacuum simply spins about one axis at a time. Attempts to drive two axes simultaneously either resolve into a single precessing axis or require torques to sustain the motion. Any deviations (non-rigid deformations, uneven mass, external torques) break the ideal assumptions and introduce extra dynamics.
 |
| Rotating sphere in cosmic space |
The Physics of Rotating Bodies: Can a Uniformly Dense Sphere in a Vacuum Rotate on Two Axes Simultaneously?
Consider a rigid body (like a solid ball) spinning in free space. Its rotational state is described by an angular velocity vector $\boldsymbol{\omega}$, which points along the instantaneous rotation axis.
Euler’s theorem guarantees that any change of orientation can be achieved by a single rotation about some axis.
In other words, even if we mathematically compose multiple small rotations, there is always one equivalent axis-angle that produces the same motion. This immediately raises the question: what does it mean physically to rotate about two axes at once?
For a rigid sphere in vacuum (no external torques), the answer is that it cannot sustain two independent spin axes simultaneously. Its motion is fully captured by one $\boldsymbol{\omega}$ at each instant.
A uniformly dense sphere has a simple inertia tensor: $I_1=I_2=I_3=(2/5)MR^2$, a scalar multiple of the identity. Thus $\mathbf{L}=I\boldsymbol{\omega}$ and $\mathbf{L}$ is always parallel to $\boldsymbol{\omega}$.
This isotropy means the sphere has no preferred axis: spinning about X or Y or any other axis is dynamically identical.
In this article, we review the relevant theory and evidence. We first recall basic rigid-body dynamics and Euler’s rotation theorem. We discuss angular velocity versus angular momentum and derive the inertia tensor of a uniform sphere.
We explain what “simultaneous rotation about two axes” would entail and why it is generally not physically possible without external forces.
We then examine torque-free motion (Euler’s equations) and rotational stability (including the tennis-racket intermediate‐axis instability). Poinsot’s geometric construction is described to visualise the motion.
Finally, we review experiments and modern simulations (including noteworthy cases like the Dzhanibekov effect) and discuss edge cases (non-rigid bodies, internal mass anomalies, applied torques).
Rigid-Body Rotation Fundamentals
A rigid body in classical mechanics is an object whose points maintain fixed distances (no deformation).
Rotations of a rigid body about its centre can be described by three coordinates (for example the Euler angles) and an angular velocity vector $\boldsymbol{\omega}$.
Infinitesimal rotations do not commute (rotating by X then Y differs from Y then X), but Euler’s theorem tells us that any overall reorientation can be achieved by a single rotation about some axis.
In practice we attach a moving body frame to the object (axes â, ḃ, ĉ) and a fixed space frame (X,Y,Z). At each instant the body is turning about some instantaneous axis in space with rate $|\boldsymbol{\omega}|$.
The velocity of any point in the body is $\mathbf{v}=\boldsymbol{\omega}\times \mathbf{r}$, perpendicular to the axis, so points farther from the axis move faster. The key point is that the entire body shares the same instantaneous $\boldsymbol{\omega}$.
In this framework the kinetic energy of rotation can be written $T=\tfrac12\sum m_\alpha v_\alpha^2$, which can be shown to equal $\frac12,\boldsymbol{\omega}^T,\mathbf{I},\boldsymbol{\omega}$, where $\mathbf{I}$ is the inertia tensor (a $3\times3$ symmetric matrix determined by the mass distribution). Each body has principal axes which diagonalise $\mathbf{I}$, giving principal moments $I_1,I_2,I_3$. By definition these axes make $\mathbf{I}$ diagonal; in general they are eigenvectors of $\mathbf{I}$.
The components of the angular momentum in the body are $L_i=\sum_j I_{ij}\omega_j$. Thus all rotational dynamics (energy, angular momentum) follow from $\mathbf{I}$ and $\boldsymbol{\omega}$.
 |
| History of Rigid-Body Rotation Theory |
Leonhard Euler proved that any change of orientation of a rigid body fixing one point can be described by a single rotation around some axis through that point.
Equivalently, given two orientations of a body, there is a unique axis and angle that carries the first into the second.
As SICP notes, “no matter how many rotations have been composed… the orientation could have been reached with a single rotation”. This theorem implies that even a sequence of tilts and twists is ultimately one rotation about a fixed line. In physical terms, it means the instantaneous motion of a body is a spin about one axis.
So when we describe a spinning object, we always use one angular velocity vector, not two simultaneous ones.
Importantly, Euler’s theorem is purely kinematic (geometric); it assumes an orientation change (no dynamics like forces). It applies equally to a sphere or any rigid body. However, dynamics (forces and torques) determine which axis the body will actually spin about.
For a free body in space, those axes come from its inertia tensor (see below). But regardless of the body’s shape, at any moment the actual rotation is about a single axis. One cannot say a rigid object in free motion has two independent spin axes acting at once – by definition, the motion is equivalent to one net axis.
Angular Velocity vs Angular Momentum
The angular velocity vector $\boldsymbol{\omega}$ is a purely kinematic quantity, specifying how fast and about which axis the body turns.
The angular momentum vector $\mathbf{L}$ is a dynamical quantity: $\mathbf{L}=\sum_\alpha \mathbf{r}\alpha\times m\alpha\mathbf{v}\alpha$ for all mass points. In terms of $\boldsymbol{\omega}$, one finds a linear relation $\mathbf{L} = \mathbf{I},\boldsymbol{\omega}$, where $\mathbf{I}$ is the inertia tensor. In component form $L_i = \sum_j I{ij},\omega_j$. Thus in general $\mathbf{L}$ and $\boldsymbol{\omega}$ need not be parallel; they coincide only along principal axes.
For a uniform sphere, however, the inertia tensor is a scalar multiple of the identity (see next section). Hence $\mathbf{L}=I,\boldsymbol{\omega}$ with $I=2/5 MR^2$, and $\mathbf{L}$ is always parallel to $\boldsymbol{\omega}$. This means the direction of rotation (axis) and the conserved momentum direction are the same for a sphere.
By contrast, an asymmetric body (triaxial ellipsoid) generally has $\mathbf{I}$ diagonalized only in its principal basis, so if $\boldsymbol{\omega}$ points along one axis, $\mathbf{L}$ has components along all axes unless $\boldsymbol{\omega}$ is exactly aligned with a principal axis.
Hence for ellipsoids one distinguishes spin about principal axes (simplest dynamics) versus arbitrary orientations (where precession occurs).
Inertia Tensor of a Uniform Sphere
For a solid sphere of mass $M$ and radius $R$, symmetry forces all principal moments equal. In fact one finds [ I_1=I_2=I_3=\frac{2}{5}MR^2, ] so any diameter is a principal axis. (This can be derived by integration or found in mechanics texts.) Thus in a body-fixed coordinate frame the inertia tensor is [ \mathbf{I}=I,\mathbf{1}=\frac{2}{5}MR^2;\text{diag}(1,1,1). ] Because of this isotropy, the sphere’s resistance to rotation is the same about X, Y or Z.
Compare this to an ellipsoid with semi-axes $(a,b,c)$. Its principal moments are (see Landau & Lifshitz) [ I_1 = I',(b^2+c^2),\quad I_2 = I',(a^2+c^2),\quad I_3 = I',(a^2+b^2), ] where $I'=(2/5)M$ for a unit-radius sphere. (Here axes 1,2,3 lie along a,b,c respectively.) The unequal values $I_1<I_2<I_3$ mean spin about different axes feels different inertia.
A uniform sphere is the special case $a=b=c=R$, giving $I_1=I_2=I_3=2/5MR^2$ as above.
Table: Comparison of rotation properties for a uniform sphere vs a triaxial ellipsoid. (Angular momentum and stability refer to body-fixed principal axes.)
Principal Axes and Ellipsoids
As noted, principal axes are special body-fixed directions (through the centre of mass) along which the inertia tensor is diagonal.
In these axes the cross-terms $I_{ij}(i\neq j)$ vanish and $I_1,I_2,I_3$ appear on the diagonal. One obtains them by solving the eigenvalue problem $\mathbf{I}\mathbf{u}=\lambda,\mathbf{u}$. In a non-spherical body the three principal axes are unique (up to permutation) and orthogonal.
For a sphere, all principal moments coincide. Any choice of orthogonal axes is principal, so the inertia tensor is the same in any orientation. There is complete degeneracy: the sphere has infinitely many equivalent principal axis sets.
In contrast, a triaxial ellipsoid has three distinct $I_1<I_2<I_3$. By symmetry the longest physical axis (shortest dimension) has the smallest inertia $I_1$, the shortest physical axis has $I_3$, etc.
In Landau & Lifshitz one finds (for a uniform density ellipsoid of volume $4\pi abc/3$) that [ I_1 = I'(b^2+c^2),\quad I_2 = I'(a^2+c^2),\quad I_3 = I'(a^2+b^2), ] where $I'=(2/5)M$ for a unit sphere.
The key takeaway is that a sphere’s symmetry makes all axes equivalent, while a generic ellipsoid has distinct axes with differing rotational properties.
Simultaneous Rotation about Two Axes
What does it mean to "rotate on two axes simultaneously"? In rigid-body terms, one might imagine spinning the sphere around, say, the X-axis and the Y-axis at the same time.
However, because rotation is described by a single angular velocity vector, the net effect is always a single rotation about some resultant axis (given by vector addition of the two spin components). In practice, trying to drive two axes requires extra constraints or torques.
Indeed, in free space a combination of rotations is still just one rotation about a different axis.
As one Physics.SE answer explains, “such a combination of rotations is mathematically possible… but the axis of such a combined rotation is neither horizontal nor vertical. And the endpoints of the axis need to be stationary. Your machine doesn’t allow that motion”.
In other words, a free sphere can undergo a motion combining X- and Y-spins, but the result is a single axis (tilted) spin – not two independent spins. If one locks the sphere’s orientation machine to try to force two axes, uncontrolled torques and vibrations appear.
Another way to see it: if the resultant motion is not about a fixed axis in space, Euler’s equations require a continuous external torque to sustain it. For example, powering two perpendicular rotations at once generates a net torque that must be supplied by the device.
As commented in the Physics.SE thread, “if the resultant motion is not about a fixed axis, then a continuous torque needs to be applied – the faster you go the larger the forces”.
Without that torque, the body cannot maintain two different spin components steadily. In a vacuum with no external forces, the sphere’s angular momentum is constant, so it will not spontaneously rotate about a new axis without being torqued into it.
“Simultaneous rotation about two axes” is not a valid free motion mode for a rigid sphere. At best, any attempted dual-spin is equivalent to a single rotation about some axis in space.
In practice, devices that try to impose two orthogonal spins end up producing one effective spin and reactive torques.
(Toy gyroscopes, which constrain one axis, illustrate this: you can spin a wheel about one axis, but trying to twist it about another leads to gyroscopic precession or shaft stress.)
Torque-Free Motion: Euler’s Equations
In the absence of external torques, a rotating rigid body obeys Euler’s equations. In a principal-axis frame, these read: [ \begin{cases} I_1 \dot\omega_1 + (I_3 - I_2),\omega_2\omega_3 = 0,\ I_2 \dot\omega_2 + (I_1 - I_3),\omega_3\omega_1 = 0,\ I_3 \dot\omega_3 + (I_2 - I_1),\omega_1\omega_2 = 0, \end{cases} ] where $(\omega_1,\omega_2,\omega_3)$ are the body-frame components of $\boldsymbol{\omega}$. For a free sphere ($I_1=I_2=I_3$), all these terms vanish automatically, leaving $\dot{\boldsymbol{\omega}}=0$.
Thus the sphere spins at constant angular velocity (no change of rotation) about a fixed space direction, i.e.\ the angular momentum is conserved.
In fact one finds the total angular momentum vector is constant: $d\mathbf{L}/dt = \mathbf{0}$ in space. Physically, this means a free sphere will continue spinning forever about the same inertial axis.
If torques are applied (e.g.\ gravity on a top, or contact forces), then $\dot{\mathbf{L}}=\boldsymbol{\tau}\neq 0$, and the sphere’s spin axis can change (precession, nutation). But the question assumed a vacuum with no external torques.
In that case, the analysis above fully describes the motion: a constant $\boldsymbol{\omega}$ and $\mathbf{L}$, with the sphere’s orientation fixed in space up to uniform rotation.
Stability and Precession (Tennis Racket Theorem)
A final classical result is that not all spin axes of a rigid body are equally stable (the “tennis racket theorem”).
If $I_1<I_2<I_3$, then steady rotation about the smallest- or largest-inertia axis is stable against small disturbances, while rotation about the intermediate axis is unstable.
In practice one sees the body flip/flop if it is set spinning around the middle axis even ever so slightly off-axis.
For a sphere, however, there is no intermediate axis. All three $I_i$ are equal, so there is no preferred direction. This means a uniform sphere is neutrally stable about any axis: if you spin it a little off from an axis, it just continues to spin in that new direction (no exponential divergence). In effect, any axis is both “largest” and “smallest” by symmetry.
For contrast, experiments like the Dzhanibekov (tennis‑racket) effect vividly demonstrate intermediate-axis instability in asymmetric bodies. If you throw a book (with distinct shortest, intermediate and longest axes) it will tumble unpredictably about the middle axis.
By the analysis above, the sphere would never do that – it simply keeps rotating smoothly. In other words, the sphere’s isotropy removes the classic flipping behaviour entirely.
Poinsot’s Construction of Free Rotation
Poinsot’s geometric construction gives a beautiful picture of torque-free rotation. One imagines an inertia ellipsoid in the body frame (level surface of constant kinetic energy) and a fixed angular momentum sphere in space (constant $|\mathbf{L}|$).
The point where the inertia ellipsoid touches a plane perpendicular to $\mathbf{L}$ draws the polhode curve on the ellipsoid, and the contact point on the plane traces the herpolhode.
In modern terms, the body’s $\boldsymbol{\omega}$ vector moves so as to keep $\mathbf{L}$ fixed.
As SICP summarises: “we recognize the conservation of angular momentum constraint … as the equation of a sphere, and the conservation of kinetic energy constraint as the equation for a … ellipsoid. … the components of the angular momentum move on the intersection of these two surfaces, the energy ellipsoid and the angular momentum sphere”.
In this diagram, free rotation produces a constant $\mathbf{L}$ (fixed in space) and a rotating inertia ellipsoid. The intersection of the “energy ellipsoid” and the “angular momentum sphere” yields the allowed motion.
A uniformly dense sphere has a spherical inertia ellipsoid (since $I_1=I_2=I_3$), so the polhode is trivial (all points have equal energy). Hence the sphere’s rotation is especially simple: $\boldsymbol{\omega}$ just sits in one direction and does not wander on the ellipsoid.
By contrast, a triaxial body’s polhode is a nontrivial curve on the ellipsoid, leading to precession of $\boldsymbol{\omega}$ (and thus the body’s orientation) around $\mathbf{L}$.
Experimental and Simulation Evidence
All classical predictions above are borne out by experiment and numerical simulation. A freely spinning uniform sphere (or a well-balanced gyroscope) simply keeps spinning about its initial axis.
If one tries to force two axes, the resulting behaviour is either a single precessing axis or mechanical failure (bearing wear, vibration). Computer models of rigid-body dynamics show exactly this: without torque, a sphere’s angular velocity vector is constant.
Real-world tests with ellipsoids or asymmetric objects illustrate the intermediate-axis flip. For example, flipping a book or wingnut in weightlessness shows it tumbles unpredictably about the “middle” axis.
By contrast, trying similar tricks with a uniform ball yields only steady motion. In satellite dynamics, any anisotropy or external torque (e.g.\ from gravity gradients) causes complicated motion, but a truly spherical satellite (with no gravity gradient torque) would simply maintain its attitude.
Edge cases: If the sphere is not perfectly rigid or has internal fluids, new effects appear (e.g.\ sloshing modes). If the mass distribution is non-uniform, it ceases to be “uniform sphere” and can develop preferred axes.
External torques (gravity, magnetic fields, thrusters) can induce nutation or precession (like a planet’s precessing spin axis). Those cases lie outside the ideal scenario assumed here. In all cases, the underlying rigid-body theory (Euler’s theorem, $\mathbf{L}=I\boldsymbol{\omega}$, and so on) governs the dynamics – one simply adds extra forces or internal couplings on top.
A uniform sphere in a torque-free vacuum rotates about a single axis at a time. Its isotropic inertia means no distinct “two-axis” spin mode exists. Historical and modern analyses (from Euler and Poinsot to contemporary simulation studies) unanimously confirm this intuitive result.
Keywords: rotating bodies physics, angular momentum, rigid body dynamics, uniform sphere rotation, Euler equations, moment of inertia, classical mechanics, physics of rotation
References
- Goldstein, H., Poole, C., & Safko, J. (2002). Classical Mechanics (3rd ed.). Addison-Wesley. https://doi.org/10.1017/CBO9780511615549
- Landau, L. D., & Lifshitz, E. M. (1976). Mechanics (3rd ed.). Butterworth-Heinemann. https://doi.org/10.1016/C2009-0-25569-3
- Marion, J. B., & Thornton, S. T. (2003). Classical Dynamics of Particles and Systems (5th ed.). Brooks Cole. https://doi.org/10.1017/CBO9781139644294
- Arnold, V. I. (1989). Mathematical Methods of Classical Mechanics (2nd ed.). Springer. https://doi.org/10.1007/978-1-4757-2063-1
- Klein, F., & Sommerfeld, A. (2008). Theory of the Top. Birkhäuser. https://doi.org/10.1007/978-0-8176-4720-6
- Poinsot, L. (1834). Theory of the Rotation of Bodies. Bachelier. https://gallica.bnf.fr/ark:/12148/bpt6k1522103
- Euler, L. (1750). Discovery of a new principle of mechanics. https://doi.org/10.3931/e-rara-37920