NOETHER'S THEOREM

Three topics. Three conservation laws. Energy is conserved. Momentum is conserved. Angular momentum is conserved. Each law has been independently useful — we used energy conservation to pin down the speed of a falling brick, momentum conservation to solve collisions, angular-momentum conservation to explain spinning skaters.

Stop and notice the pattern. Why exactly three? Why these three? Why not also m²·v, or m·v³, or the quantity m·x²? There are infinitely many things you could imagine being conserved, and the universe picks three. Is there a rule that picks them out?

There is. It is the most beautiful theorem in classical physics, and it was proved in 1918 by a mathematician who had spent the previous decade being refused a paid position at Göttingen because she was a woman. Emmy Noether did not invent a new conservation law. She did something stranger. She proved that every conservation law that exists, or can exist, is the shadow of a symmetry of nature. Change the symmetries — and only those — and you change the ledger of what is conserved.

The rest of this topic is what that sentence means, and why physics has not stopped reckoning with it since.

The word is used loosely in everyday speech. In physics it has a precise meaning: a system has a symmetry under some transformation if applying the transformation changes nothing about the physics. The laws of motion of the transformed system are identical to the laws of motion of the original system. An experiment run before and an experiment run after will produce indistinguishable results.

Three candidate transformations of the physical world are unlike anything else:

These three transformations share a critical property. Each of them can be applied by any tiny amount — slide the clock forward by a nanosecond, translate by a millimetre, rotate by a milliradian. That makes them continuous symmetries: symmetries parameterised by a real number you can slide smoothly from zero. A mirror flip, by contrast, is a discrete symmetry — you either flip or you don't, no halfway. Noether's theorem, as we'll see, lives and dies on continuity.

Flip through the three panels. The two instances of each setup are identical systems, transformed by one of the three operations. They tick in perfect lockstep, because the laws of motion don't care about the transformation. Each symmetry is paired, in the label, with the conservation law it turns out to be hiding.

Noether's theorem is stated most cleanly in Lagrangian mechanics. (We'll treat Lagrangians properly in FIG.29. For now, take it that for every mechanical system you can write a single function L of positions and velocities whose behaviour over time encodes all the equations of motion. L is usually defined as kinetic energy minus potential energy.)

Here is the theorem, in the spirit of Noether's 1918 paper:

Three words do all the work. Invariant means the Lagrangian looks the same after the transformation as before. Continuous means the transformation is parameterised by a real number ε you can make arbitrarily small. Conserved means dQ/dt = 0 along any solution to the equations of motion.

The symbol for the whole bargain:

Read the arrow both ways. The forward direction is what Noether proved. The converse is equally consequential: the observation of a conserved quantity in nature is direct evidence that some symmetry is hiding in the laws. When we discover a new conservation law, we have discovered a new symmetry of the universe. When a conservation law is broken, a symmetry we thought existed doesn't — or is broken by some new physics we hadn't yet accounted for.

The next three sections unpack, in order, the three symmetries we already met — and in each case read off the Noether charge.

The simplest symmetry to state is: the laws of physics don't depend on when you apply them. A pendulum released at noon swings the same way as one released at midnight. A falling apple falls the same in 1687 and in 2026. Formally, the Lagrangian of a closed system doesn't explicitly depend on t. Shift the clock by any Δt and L is unchanged.

What quantity does Noether's machinery hand back?

The conserved Noether charge associated with time-translation is what we have been calling the total energy — the sum of kinetic and potential. Every time we summed ½mv² and mgh on a bouncing ball and got the same number at the top and bottom of the arc, we were observing time-translation invariance. Energy conservation is not a separate fact about the world, independent of its homogeneity in time. It is the same fact, re-expressed.

Run it in reverse: a system that does not conserve energy must have a Lagrangian that depends on t. A child on a swing who pumps their legs is transferring energy into the swing from their metabolism — the swing-plus-child system has no time-translation symmetry, because the Lagrangian is being actively modified by the child's muscles, and sure enough the swing's mechanical energy is not conserved. The broken symmetry is the leak in the bucket.

Next symmetry: the laws of physics don't depend on where you apply them. Shift the whole experiment by a metre in any direction and L is unchanged. Space is homogeneous.

Feed this into Noether and the conserved quantity is momentum — one component of p = mv for each direction in which the Lagrangian is invariant:

The Noether charge is literally the same vector p you used in FIG.06 to solve collisions. Its three components are tied to the three directions in which space is translation-invariant. If gravity broke translational symmetry along z — say, by being stronger in Dortmund than in Berlin by some mysterious amount — then p_z would stop being conserved. In practice, gravity is nearly uniform on collision scales, all three components of p are conserved, and billiard balls obey the law you learned.

The clean way to hold it in your head: x-invariance gives p_x, y-invariance gives p_y, z-invariance gives p_z. Each translation direction is a separate continuous symmetry and pays a separate conserved component. Three symmetries, three components, one conservation law for the combined vector.

Last symmetry: the laws of physics don't depend on which way you orient the experiment. Rotate the apparatus by any angle about any axis and L is unchanged. Space is isotropic.

The Noether charge is L = r × p, the angular momentum we met in FIG.07. Rotation about each axis is its own continuous symmetry, and each pays its own conserved component of L. A skater pulling in her arms is not doing magic. She is exploiting the fact that her isolated system has rotational symmetry about the vertical, so L_z is conserved; contract the moment of inertia and ω has to rise to keep the product unchanged.

A planet tracing an ellipse is the same story. The Sun's gravitational field has rotational symmetry about the Sun's centre, so the planet's angular momentum about that point is conserved, and the orbit — as Kepler discovered empirically — sweeps out equal areas in equal times.

That bookkeeping is why Einstein wrote of Noether, the year after her theorem appeared, that she had given physics its most profound insight yet into the structure of its own laws.

Noether's theorem will reappear. It is one of those results in physics that refuses to stay in its original room.

In gauge theories — the formal machinery behind electromagnetism, the weak force, and QCD — Noether's theorem is how we learn that conservation of electric charge comes from the symmetry of quantum wave functions under a phase rotation ψ → e^(iα)·ψ, and conservation of colour charge from a more elaborate gauge symmetry (SU(3)). Every conservation law discovered in modern physics has turned out to be a Noether consequence of some symmetry, often a hidden or internal one with no obvious spatial picture.

In general relativity the picture gets subtler: the theory's local gauge symmetry — diffeomorphism invariance — is governed by what's now called Noether's second theorem, a less famous companion result she proved in the same 1918 paper. Einstein consulted Noether directly about the conservation-of-energy puzzles in GR, and Hilbert and Klein — two of the era's foremost mathematicians — credited her with resolving them.

The Lagrangian we invoked in §3 is the other loose end. In FIG.28 we will build it from scratch, derive the equations of motion from it, and see Noether's theorem drop out as an almost geometric inevitability. By the end of the classical-mechanics branch you will have met the theorem three times: as a pattern to notice, as a statement to believe, and as a consequence of the deepest formulation of mechanics we know. That sequence — phenomenon, law, structure — is how physics keeps going.

But first, rotation. We have been treating spinning objects as point masses dressed up in rotational language. Module 3 stops pretending. Things have shape, and shape has consequences. It turns out that gyroscopes don't fall over, the Earth wobbles on a 433-day cycle, and the rotational analogue of F = ma has a few surprises that point-particle mechanics never warned us about.