Noether's Theorem: Symmetry Behind Conservation Laws

Emmy Noether proved in 1918 that every conservation law comes from a symmetry: energy from time, momentum from space, charge from gauge invariance.

Table of Contents

The Energy Problem Nobody Wanted

In 1916 David Hilbert noticed something wrong with the bookkeeping of general relativity. He had written down an energy equation for Einstein’s new theory of gravity, and on inspection it did nothing. It followed from the field equations by pure algebra, without ever asking what the matter in the universe was actually doing. In classical mechanics a conservation law is a hard constraint on behaviour. In Hilbert’s version of gravity it was a tautology. He called it an improper conservation law, improper precisely because it held whether or not anything obeyed the equations of motion.

Hilbert handed the question to a 34-year-old algebraist whom he and Felix Klein had brought to Göttingen the previous year, and who was being paid nothing to be there. Emmy Noether answered it in a way neither of them expected. Rather than patch the energy equation, she proved a general theorem about when conservation laws exist at all, and showed that gravity’s failure to produce a proper one was not a defect but a structural feature of any theory built on that kind of symmetry.

The paper, Invariante Variationsprobleme, was presented to the Göttingen scientific society on 26 July 1918 and runs from page 235 to page 257 of that year’s proceedings. It is arguably the single most reorganising result in theoretical physics. Our wider survey of why the universe keeps its books balanced catalogues the conservation laws; this is the article about where they come from.

One Sentence That Does Everything

Noether’s first theorem says: for every continuous symmetry of a system’s action, there is a quantity that does not change in time.

That is the whole thing. The word continuous is load-bearing. The symmetry has to come in arbitrarily small increments — rotating by a thousandth of a degree, shifting by a micrometre — because the conserved quantity is constructed directly out of the infinitesimal version of the transformation. Discrete symmetries give selection rules rather than conserved currents, which is why the discovery that the weak force violates mirror symmetry cost physics no conserved quantity whatsoever.

Run the dictionary and the familiar list falls out. If the law governing a system is the same today as tomorrow — time-translation symmetry — then energy is conserved. If it is the same here as three metres to the left — space-translation symmetry — then momentum is conserved. If it does not care which way you are facing — rotational symmetry — then angular momentum is conserved. There is even a fourth that gets left off the usual list: invariance under a uniform change of velocity gives you the constancy of the centre of mass’s motion.

The point is not that these pairings are elegant. The point is the direction of the arrow. Before 1918, energy conservation was an empirical law, checked and rechecked and fundamentally unexplained. After 1918 it was a consequence of something simpler, and a falsifiable one: build a system whose law changes with time, and energy will not be conserved in it.

Watch the Symmetry Do the Work

The honest way to see this is to perform the transformation and check whether the law survives it. Shift the clock on a mass hanging from a spring whose stiffness never changes, and the motion you get is the same motion you would have got anyway, started at a different moment. Shift a pair of colliding blocks three metres down the bench, with nothing in the room to tell them where they are, and the collision plays out identically. Rotate a planet’s whole orbit about a perfectly spherical star and it is still the same orbit. In each case something refuses to move: a number that stays fixed to five decimal places while everything visible changes.

Then take the symmetry away. Make the spring’s stiffness depend on time, k(t) = k₀(1 + 0.3 sin Ωt), and shifting the clock now gives genuinely different motion, because the spring is not the same spring it was. Noether’s theorem withdraws its guarantee, and energy starts to climb. That contrast is the entire content of the theorem, and it is worth a minute with the figure below.

Symmetry in, conserved quantity out Shift in time → energy. The spring is the same spring at every moment, so E does not move. Mass on a spring: m = 1 kg, k = 4 N/m, amplitude 1 m Energy E (joules) equilibrium 4.00 2.00 0.00 0 s 10 s conserved quantity at this moment 2.000 J change across the window: 0.0000 per cent symmetry applied none yet drag the second slider to shift the clock
t = 0.00 s
none
Energy E = 2.000 J, unchanged to within 0.0000 per cent across the whole 10-second window.
Pick a symmetry, drag the time slider, then drag the second slider to apply the transformation and watch the orange ghost land on a system that obeys the same law. Finish on the broken case, where the spring stiffness depends on time, and watch the energy climb instead.

Charge, and the Theorem People Forget

Electric charge conservation is the famous fourth entry, and the usual one-liner — charge is conserved because electromagnetism has a gauge symmetry — is close enough to be useful and loose enough to irritate specialists.

The precise version needs Noether’s second theorem, which she proved in the same paper. The first theorem handles symmetries that depend on a finite number of constant parameters: rotate everything by the same angle, shift everything by the same amount. The second handles symmetries whose parameters are arbitrary functions of position and time, which is what physicists mean by a local or gauge symmetry. Those do not hand you new conserved quantities. They hand you identities among the field equations, which means the equations are not independent. That is exactly the improper conservation law Hilbert had walked into.

Charge conservation comes from the global part: the freedom to multiply a charged field’s quantum phase by the same factor everywhere at once. That is a constant-parameter symmetry, so the first theorem applies, and the conserved Noether charge is built from the squared magnitude of the field. The local gauge symmetry does something different and arguably deeper — it forces the charged field to couple to the photon in the first place. In quantum field theory the same structure reappears as the Ward identities, and the pattern of symmetry group in, conserved current out is how the Standard Model was assembled: isospin, colour, hypercharge. Noether’s theorem stopped being a result and became a design tool.

The Universe Does Not Have the Symmetry

Here is where the theorem bites back.

Energy conservation requires time-translation symmetry. Our universe is expanding. The geometry it hands you at one moment is not the geometry it hands you at the next, and in the language of general relativity there is no timelike Killing vector — no direction in time along which the spacetime is unchanged. The symmetry is simply absent, and so, globally, is the conservation law.

The clearest case is light. A photon released at the surface of last scattering arrives redshifted by a factor of about 1,100, which means it arrives carrying roughly a thousandth of the energy it set out with. Where did the rest go? There is no accounting entry. Dark energy runs the trick in reverse: its density stays constant while the volume grows, so the total energy inside any expanding region climbs.

Physicists differ on how to say this out loud. One camp states flatly that energy is not conserved in general relativity and has not been since the 1920s. A second prefers the formulation that conservation survives exactly but only locally, as the vanishing divergence of the energy-momentum tensor — a differential law with no global counterpart, because adding up energy-momentum vectors defined at different points in a curved spacetime is not a well-defined operation. A third tries to balance the books by assigning energy to the gravitational field itself, which requires a mathematical object that can be made to vanish at any point you choose and is therefore not quite an energy density. The mathematics is agreed. The sentence is not. What everyone agrees on is why the question is hard, and the reason is Noether’s.

What She Was Paid

Noether arrived in Göttingen in April 1915 with no position at all, because a Prussian ministerial decree of 1908 barred women from habilitation, the qualification that let you lecture. The decree had been justified by a ministry survey reported as overwhelmingly negative, although Hilbert, Klein, Minkowski and Runge had all voted the other way. Her first application went in that July; a faculty vote in November failed to recommend her, saved from outright rejection only by the chair’s casting vote. A second application was refused by the ministry in November 1917.

So Hilbert used a workaround. The course announcement for the winter of 1916 reads: mathematical-physical seminar, invariant theory, Professor Hilbert with the support of Dr Nöther, Monday four to six, free of charge. She taught it. His name was on it.

Einstein wrote to Klein in December 1918 calling the refusal a great injustice and offering his signature on any petition to the ministry. A third application went in the following month and succeeded: the theorem paper was accepted as her habilitation thesis on 28 May 1919, and she gave her trial lecture a week later. Prussia lifted the ban in February 1920, nine months after her exemption. In 1922 she received the title of unofficial associate professor, which came with no chair and no salary. She was paid for her work for the first time in 1923, at the age of 41.

Her result is the reason we now know that if you want to find out what a theory conserves, you do not start with the forces. You start by asking what you can do to the theory without changing it.

Frequently Asked Questions

What is Noether's theorem?

Noether's theorem, proved by the German mathematician Emmy Noether and published in 1918, states that every continuous symmetry of a physical system's action corresponds to a quantity that does not change in time. A symmetry here means something you can do to the system without altering the law that governs it. If the law is the same at every moment, the system has time-translation symmetry and its energy is conserved. If the law is the same at every point in space, momentum is conserved. If it is the same in every direction, angular momentum is conserved. If a charged quantum field can have its phase multiplied by the same factor everywhere, electric charge is conserved. The symmetry has to be continuous, meaning it can be applied in arbitrarily small steps, because the conserved quantity is built out of the infinitesimal form of the transformation. Discrete symmetries such as mirror reflection give selection rules instead. The theorem also runs backwards: remove the symmetry and the conservation law goes with it.

Why is energy not conserved in an expanding universe?

Energy conservation follows from time-translation symmetry, the requirement that the law governing a system is the same from one moment to the next. An expanding universe does not have that symmetry. The geometry of space changes with time, and in the language of general relativity there is no timelike Killing vector, no direction in time along which the spacetime is unchanged. Noether's theorem then supplies no globally conserved energy. The clearest example is light from the cosmic microwave background, redshifted by a factor of about 1,100 since it was emitted, arriving with roughly a thousandth of the energy it set out with and no accounting entry for the rest. Dark energy works the other way: its density stays constant while the volume grows. What survives exactly is the local statement that the divergence of the energy-momentum tensor vanishes. Physicists differ over the wording, some saying energy is not conserved and others that conservation has become purely local, but they agree on the mathematics and on the reason, which is the missing symmetry.

Which symmetry gives conservation of angular momentum?

Rotational symmetry. If the law governing a system does not change when you turn the whole system about an axis, the component of angular momentum along that axis is conserved. For a single particle of mass m moving in a plane the quantity is L = m(x·v_y − y·v_x), and for an extended rotating body it is usually written L = Iω, with I the moment of inertia and ω the angular velocity. The symmetry has to be genuine. A planet orbiting a perfectly spherical star sees a potential that depends only on distance, so every orientation is equivalent and angular momentum is exactly conserved even on a highly elliptical orbit where the speed changes by a large factor. Add a bulge that singles out a direction and the orbital angular momentum starts to change. This is also why a collapsing stellar core can spin up to hundreds of rotations per second: nothing about the collapse picks out an axis, so the angular momentum it began with has nowhere to go.

Was Emmy Noether allowed to teach at Göttingen?

Not at first, and not under her own name. Emmy Noether moved to Göttingen in April 1915 at the invitation of David Hilbert and Felix Klein, with no post and no salary, because a Prussian ministerial decree of 1908 barred women from habilitation, the qualification required to lecture. Her first application was submitted in July 1915 and failed at a faculty vote that November. A second was rejected by the ministry in November 1917. Hilbert's workaround was to announce her courses under his own name, so the winter 1916 timetable reads mathematical-physical seminar, invariant theory, Professor Hilbert with the support of Dr Nöther. After Einstein wrote to Klein in December 1918 calling the refusal a great injustice, a third application succeeded, and her theorem paper was accepted as her habilitation thesis on 28 May 1919. Prussia lifted the ban in February 1920. She received the title of unofficial associate professor in 1922, which carried no salary, and was first paid for her work in 1923, at the age of 41.

Read Next