The magnetic moment & g=2

Every electron is a tiny magnet. Put one in a magnetic field and its spin feels a torque, its two spin states split in energy, and the whole apparatus of magnetic resonance follows. The strength of that magnet is set by one dimensionless number, the g-factor, through $\boldsymbol\mu = g\,(q/2m)\,\mathbf S$ . A classical spinning ball of charge gives $g=1$ , the same value that governs orbital motion. Yet the electron's spin magnet is twice as strong — $g=2$ — and nobody could say why until 1928. The answer is one of the great early triumphs of relativistic quantum mechanics: couple Dirac's equation to an electromagnetic field, take the slow-particle limit, and the Pauli equation drops out with the spin–magnetic term already carrying its factor of two. The value $g=2$ is not fitted; it is forced by the four-component structure of the Dirac spinor. That would be a tidy story if it ended there. It does not. In 1947 Kusch and Foley found $g$ slightly larger than 2, and within months Schwinger showed the excess to be $\alpha/2\pi \approx 0.00116$ — the electron radiating and reabsorbing a virtual photon nudges its own magnetism. This anomalous magnetic moment $a = (g-2)/2$ is now the most precisely tested prediction in physics, agreeing with QED to twelve significant figures, while its muon cousin is our sharpest tool for hunting physics beyond the Standard Model. This article derives $g=2$ from the Dirac equation, reads it off again from the QED vertex, and follows the anomaly from Schwinger's one loop to the Fermilab storage ring.

Key takeaways

The magnetic moment of a spin-½ particle is $\boldsymbol\mu = g\,(q/2m)\,\mathbf S$ ; the g-factor is what everything hinges on.
A classical spinning charge, and orbital motion, gives $g=1$ . The Dirac equation predicts $g=2$ for a structureless spin-½ particle.
The proof: minimally couple Dirac's equation to $A_\mu$ , take the non-relativistic limit, and the Pauli equation emerges with spin term $-(q/2m)\,\boldsymbol\sigma\cdot\mathbf B$ — exactly $g=2$ .
The QED vertex splits into $F_1\gamma^\mu$ (charge) and $F_2\,i\sigma^{\mu\nu}q_\nu/2m$ (magnetism); $g = 2[F_1(0)+F_2(0)] = 2(1+a)$ , so tree level ( $F_2=0$ ) gives $g=2$ .
Loops make $g$ exceed 2. Schwinger's one-loop result is $a=(g-2)/2=\alpha/2\pi\approx 1.16\times10^{-3}$ , the anomalous moment.
Measured $g_e/2 = 1.001\,159\,652\,181$ — twelve-digit agreement with QED; the muon's $a_\mu$ probes new physics via its $(m_\mu/m_e)^2\approx 4\times10^4$ enhancement.
Key idea: $g=2$ is not an empirical constant but a prediction squeezed out of relativity and quantum mechanics together, and the tiny excess $g-2$ is where virtual particles leave their fingerprint on the electron.

The core insight

The gyromagnetic ratio $g=2$ falls out of the Dirac equation the moment it is coupled to electromagnetism — it is relativity's gift, not a fitted number — while the anomaly $a=(g-2)/2=\alpha/2\pi+\cdots$ is the imprint of quantum loops, making $g-2$ the most stringently tested quantity in physics.

Where this lives today

The electron's $g-2$ is the crown jewel of precision physics. Single electrons stored one at a time in a cryogenic Penning trap yield $g_e/2=1.001\,159\,652\,181\,28$ , good to about 0.1 parts per trillion — and comparing it with the five-loop QED calculation gives both the most stringent test of any physical theory and one of the best determinations of $\alpha$ itself. The muon tells a more suspenseful story. Because its anomaly is enhanced by $(m_\mu/m_e)^2\approx4\times10^4$ against heavy virtual particles, Fermilab's Muon g-2 experiment — a 15-metre superconducting ring reusing the very magnet barged from Brookhaven — measures $a_\mu$ to parts per billion in search of Standard Model cracks. For years the measured $a_\mu$ sat about four standard deviations above the data-driven prediction, a tantalizing hint of new physics; the interpretation is now in flux as lattice-QCD computations of the hadronic contribution shift the theoretical target. Meanwhile the proton's anomaly $g_p\approx5.586$ is a daily reminder of what $g=2$ means: any particle whose $g$ departs from 2 has internal structure — which is how the proton's compositeness first announced itself.

The physical picture

A charged particle with spin behaves like a current loop: it carries a magnetic dipole moment $\boldsymbol\mu$ locked to its spin $\mathbf S$ . In a uniform field $\mathbf B$ the energy is $-\boldsymbol\mu\cdot\mathbf B$ , so the two orientations of a spin-½ particle split by $\Delta E = g\mu_B B$ , and the moment precesses about the field. All of "how magnetic is the electron" is packed into the constant $g$ between $\boldsymbol\mu$ and $\mathbf S$ — and the surprise is that it is 2, not the classical 1. The first figure fixes the geometry and the relation.

Figure

An electron in a field B: its magnetic moment μ is antiparallel to its spin S because the charge is negative, and the two spin states split by ΔE = g μ_B B with g = 2, so μ = (g/2) μ_B ≈ one Bohr magneton

The number $g$ is where the physics lives. Treat the electron as a classical top of charge $q$ and angular momentum $S$ : the circulating current gives $\mu=(q/2m)\,S$ , i.e. $g=1$ , the same ratio orbital motion produces. Experiment said otherwise — atomic spectra demanded twice that for spin — and Dirac's equation explained it. But the story did not stop at the integer. Once the electron emits and reabsorbs virtual photons, $g$ creeps above 2 by a calculable amount, and each layer of quantum corrections has been measured. The second figure is the punchline of the subject: the observed $g/2$ , magnified until the anomaly shows, sitting exactly where quantum electrodynamics puts it.

Figure

The measured g/2 for the electron magnified near 1: Dirac's tree level gives exactly 1, Schwinger's one loop adds α/2π = 1.1614×10⁻³, and higher-order QED plus experiment agree at g/2 = 1.001159652181, the anomaly a = (g−2)/2

The core equations

The moment-to-spin relation is the definition of $g$ , the object every later equation tries to pin down. Nothing about it is relativistic yet; it just names the proportionality between a particle's intrinsic magnetism and its intrinsic angular momentum.

\boldsymbol\mu = g\,\frac{q}{2m}\,\mathbf S \qquad (1)

where: $\boldsymbol\mu$ — magnetic dipole moment; $g$ — gyromagnetic ratio (the g-factor), dimensionless; $q$ — charge; $m$ — mass; $\mathbf S$ — spin. Orbital motion takes the same form with $g_L=1$ ; the whole subject is the claim that spin has $g=2$ , not $1$ . To get $g$ from first principles, couple the Dirac equation to the potential by the minimal substitution $p_\mu \to p_\mu - qA_\mu$ — the only gauge-invariant way to introduce the field. This is the equation whose slow-particle limit we take.

\big[\gamma^\mu(i\partial_\mu - qA_\mu) - m\big]\psi = 0 \qquad (2)

where: $\gamma^\mu$ — Dirac matrices; $A_\mu=(\phi,\mathbf A)$ — electromagnetic four-potential; $\psi$ — four-component spinor. In Hamiltonian form $H_D=\boldsymbol\alpha\cdot(\mathbf p-q\mathbf A)+\beta m+q\phi$ , ready for the non-relativistic reduction. The reduction lands on the Pauli equation, and the spin term arrives with its factor of two already attached — the coefficient of $\boldsymbol\sigma\cdot\mathbf B$ is twice the orbital coefficient.

i\frac{\partial\varphi}{\partial t}=\left[\frac{(\mathbf p-q\mathbf A)^2}{2m}-\frac{q}{2m}\,\boldsymbol\sigma\cdot\mathbf B+q\phi\right]\varphi \qquad (3)

where: $\varphi$ — the upper (large) two-component spinor; $\boldsymbol\sigma$ — Pauli matrices; $\mathbf B=\nabla\times\mathbf A$ . Reading $-\boldsymbol\mu\cdot\mathbf B$ off the middle term with $\mathbf S=\boldsymbol\sigma/2$ gives $\boldsymbol\mu=(q/m)\mathbf S=2(q/2m)\mathbf S$ , i.e. $g=2$ . The engine of that result is one algebraic identity for the square of $\boldsymbol\sigma\cdot\boldsymbol\pi$ , with $\boldsymbol\pi=\mathbf p-q\mathbf A$ the kinetic momentum — it manufactures the magnetic term from nothing but the non-commuting components of $\boldsymbol\pi$ .

(\boldsymbol\sigma\cdot\boldsymbol\pi)^2=\boldsymbol\pi^2 - q\,\boldsymbol\sigma\cdot\mathbf B \qquad (4)

where: the first term is the ordinary kinetic energy; the second, from $\boldsymbol\pi\times\boldsymbol\pi=iq\mathbf B$ , is the spin–field coupling. Dividing by $2m$ gives the spin coefficient $q/2m$ , and $\boldsymbol\sigma=2\mathbf S$ supplies the extra factor of two. Beyond tree level the electron–photon interaction is not a bare $\gamma^\mu$ but a vertex function with two form factors, and $g$ is read off their values at zero momentum transfer — the exact, all-orders statement of what $g$ means.

\bar u(p')\,\Gamma^\mu\,u(p)=\bar u(p')\!\left[\gamma^\mu F_1(q^2)+\frac{i\sigma^{\mu\nu}q_\nu}{2m}F_2(q^2)\right]\!u(p),\qquad g=2\big[F_1(0)+F_2(0)\big] \qquad (5)

where: $q=p'-p$ — momentum transfer; $\sigma^{\mu\nu}=\tfrac{i}{2}[\gamma^\mu,\gamma^\nu]$ ; $F_1$ — the electric (Dirac) form factor, fixed by charge to $F_1(0)=1$ ; $F_2$ — the magnetic (Pauli) form factor. At tree level $F_1=1,\ F_2=0$ , so $g=2$ ; loops give $F_2(0)\neq0$ . The anomalous moment is precisely that loop-generated excess; its leading value is Schwinger's — the most famous number in quantum electrodynamics.

a\equiv\frac{g-2}{2}=F_2(0)=\frac{\alpha}{2\pi}+O(\alpha^2)\approx 1.16\times10^{-3} \qquad (6)

where: $a$ — anomalous magnetic moment; $\alpha=e^2/4\pi\approx1/137$ — fine-structure constant. The full QED series continues $a=\tfrac{\alpha}{2\pi}-0.3285\,(\tfrac{\alpha}{\pi})^2+1.181\,(\tfrac{\alpha}{\pi})^3-\cdots$ , computed now through five loops.

Derivation AFrom the Dirac equation to g equal to two

The goal is to show that $g=2$ is not inserted by hand but is an unavoidable consequence of coupling Dirac's relativistic equation to an electromagnetic field and asking what a slow-moving electron looks like. We start from the minimally coupled Hamiltonian, split the four-spinor into large and small pieces, eliminate the small piece in the non-relativistic regime, and watch the spin–magnetic coupling appear at exactly twice the orbital ratio — the factor of two classical pictures cannot produce.

Step 1. Write the minimally coupled Hamiltonian $H_D=\boldsymbol\alpha\cdot\boldsymbol\pi+\beta m+q\phi$ , $\boldsymbol\pi=\mathbf p-q\mathbf A$ , in the Dirac representation with $\boldsymbol\alpha=\begin{pmatrix}0&\boldsymbol\sigma\\\boldsymbol\sigma&0\end{pmatrix}$ and $\beta=\begin{pmatrix}\mathbb 1&0\\0&-\mathbb 1\end{pmatrix}$ . Split the spinor into halves, $\psi=e^{-imt}\binom{\varphi}{\chi}$ , pulling out the rest-mass phase so the remaining time dependence is slow. Why: the rest energy $m$ dominates all other scales for a non-relativistic particle, so removing it exposes the small kinetic and potential energies that govern atomic physics.

Step 2. The coupled equations become $i\partial_t\varphi=\boldsymbol\sigma\cdot\boldsymbol\pi\,\chi+q\phi\,\varphi$ and $i\partial_t\chi=\boldsymbol\sigma\cdot\boldsymbol\pi\,\varphi+q\phi\,\chi-2m\chi$ . In the non-relativistic limit the lower spinor is small and slow, so $i\partial_t\chi$ and $q\phi\,\chi$ are negligible beside $2m\chi$ , giving $\chi\approx\dfrac{\boldsymbol\sigma\cdot\boldsymbol\pi}{2m}\,\varphi$ . Why: this is the standard elimination of the "small component," suppressed by order $v/c$ — keeping its leading piece is exactly the non-relativistic approximation.

Step 3. Substitute $\chi$ back into the upper equation for a closed equation in $\varphi$ alone: $i\partial_t\varphi=\dfrac{(\boldsymbol\sigma\cdot\boldsymbol\pi)^2}{2m}\,\varphi+q\phi\,\varphi$ . Everything now hinges on evaluating $(\boldsymbol\sigma\cdot\boldsymbol\pi)^2$ . Why: the entire magnetic physics hides inside this square, because $\boldsymbol\pi$ has non-commuting components once a vector potential is present.

Step 4. Apply $(\boldsymbol\sigma\cdot\mathbf a)(\boldsymbol\sigma\cdot\mathbf b)=\mathbf a\cdot\mathbf b+i\boldsymbol\sigma\cdot(\mathbf a\times\mathbf b)$ with $\mathbf a=\mathbf b=\boldsymbol\pi$ . The dot term gives $\boldsymbol\pi^2$ ; the cross term needs $\boldsymbol\pi\times\boldsymbol\pi$ , nonzero because $[\pi_j,\pi_k]=iq\,\epsilon_{jkl}B_l$ , so $\boldsymbol\pi\times\boldsymbol\pi=iq\mathbf B$ and $(\boldsymbol\sigma\cdot\boldsymbol\pi)^2=\boldsymbol\pi^2-q\,\boldsymbol\sigma\cdot\mathbf B$ . Why: a vector crossed with itself vanishes only for ordinary vectors; the kinetic momentum components fail to commute in a field, and that failure is the whole source of the spin–magnetic coupling.

Step 5. Insert this into the equation for $\varphi$ : $i\partial_t\varphi=\Big[\dfrac{\boldsymbol\pi^2}{2m}-\dfrac{q}{2m}\boldsymbol\sigma\cdot\mathbf B+q\phi\Big]\varphi$ , the Pauli equation (3). Compare the middle term to $-\boldsymbol\mu\cdot\mathbf B$ with $\mathbf S=\boldsymbol\sigma/2$ : $\boldsymbol\mu=\dfrac{q}{2m}\boldsymbol\sigma=2\Big(\dfrac{q}{2m}\Big)\mathbf S$ , so $g=2$ . Why: rewriting $\boldsymbol\sigma$ as $2\mathbf S$ pulls out precisely the factor of two — the orbital piece buried in $\boldsymbol\pi^2/2m$ carries only $g_L=1$ , so spin is twice as magnetic.

Go deeper · Derivation B · 1 minThe Gordon decomposition and the magnetic form factor

Derivation BThe Gordon decomposition and the magnetic form factor

The aim here is to recover $g=2$ a second, fully relativistic way — directly from the electron's electromagnetic current — and to isolate the object radiative corrections will later modify. We split the current sandwiched between Dirac spinors into a piece behaving like a moving point charge and a piece behaving like a magnetic dipole. The dipole piece is the coupling $i\sigma^{\mu\nu}q_\nu$ , whose coefficient gives $g=2$ at tree level and defines the form factor $F_2$ that carries the anomaly.

Step 1. Start from the tree-level current $j^\mu=\bar u(p')\gamma^\mu u(p)$ between on-shell spinors satisfying $\slashed p\,u(p)=m\,u(p)$ and $\bar u(p')\slashed p'=m\,\bar u(p')$ — the only extra input needed. Why: the on-shell conditions let a single $\gamma^\mu$ be rewritten, relating it to combinations of the external momenta.

Step 2. Use $\gamma^\mu\slashed p=p^\mu-i\sigma^{\mu\nu}p_\nu$ (and its mirror) with the on-shell relations to derive the Gordon decomposition $\bar u(p')\gamma^\mu u(p)=\bar u(p')\Big[\dfrac{(p+p')^\mu}{2m}+\dfrac{i\sigma^{\mu\nu}q_\nu}{2m}\Big]u(p)$ , $q=p'-p$ . Why: this splits one term into two physically distinct currents without approximation — exact for any on-shell electron.

Step 3. Interpret the pieces. The $(p+p')^\mu/2m$ term is the convection current of a structureless charge in motion — it knows nothing about spin. The $i\sigma^{\mu\nu}q_\nu/2m$ term is antisymmetric and couples to the field gradient, i.e. to $\mathbf B$ ; it is the spin current of a magnetic dipole. Why: $\sigma^{ij}$ generates spin rotations, and its contraction with $q_\nu$ is exactly how a dipole couples to a spatially varying field.

Step 4. Read the moment: the coefficient of $i\sigma^{\mu\nu}q_\nu/2m$ in the tree current is exactly $1$ , which corresponds to $F_2(0)=0$ and hence, through $g=2[F_1(0)+F_2(0)]$ with $F_1(0)=1$ , to $g=2$ . Radiative corrections dress the vertex, adding a genuine $F_2(0)$ so that $g=2(1+F_2(0))=2(1+a)$ . Why: the vertex (5) is the most complete parametrization allowed by Lorentz invariance and current conservation, so once loops shift $F_2(0)$ from zero, the anomaly $a$ is defined with no ambiguity — it is literally the magnetic form factor at zero momentum transfer.

Go deeper · Derivation C · 1 minSchwinger's one loop and the origin of the anomaly

Derivation CSchwinger's one loop and the origin of the anomaly

This derivation explains, without grinding through the loop integral, why the electron's $g$ -factor exceeds 2 and how the leading correction comes to be exactly $\alpha/2\pi$ . The mechanism is that a real electron is never bare: it emits and reabsorbs virtual photons, and while one is in flight its coupling to an external field is slightly altered. We trace how the one-loop vertex produces a finite magnetic form factor even though the same diagram's electric part is ultraviolet-divergent.

Step 1. Consider the order- $\alpha$ correction to the vertex: the incoming electron emits a virtual photon, interacts with the external field, then reabsorbs the photon before leaving. This one-loop triangle is the leading dressing of the bare $\gamma^\mu$ . Why: it is the first diagram where the electron's electromagnetic structure differs from a point particle's, so any deviation from $g=2$ must first appear here.

Step 2. Decompose the loop onto the two structures of (5). The $\gamma^\mu$ part renormalizes the charge — ultraviolet-divergent, absorbed into the definition of $e$ , leaving $F_1(0)=1$ . The $i\sigma^{\mu\nu}q_\nu$ part is ultraviolet-finite and needs no renormalization. Why: gauge invariance guarantees the magnetic form factor is finite — a divergent $F_2(0)$ would be an infinite, uncalculable moment, so the theory protects the very quantity we want to predict.

Step 3. Evaluating that finite piece at zero momentum transfer gives $F_2(0)=\dfrac{\alpha}{2\pi}$ , so $a=\dfrac{g-2}{2}=\dfrac{\alpha}{2\pi}\approx\dfrac{1}{861}\approx1.16\times10^{-3}$ , lifting $g$ from $2$ to about $2.00232$ . Why: this is Schwinger's 1948 result — a pure number times $\alpha$ , no free parameters, that matched Kusch and Foley and launched precision quantum field theory.

Step 4. This is the first term of an asymptotic series in $\alpha/\pi$ , each order adding more virtual particles — more photons, and eventually virtual leptons, hadrons, and whatever heavier states exist. Why: every particle coupling to the electron can appear in a loop, so $g-2$ sums over all of physics at the relevant scale — which is what makes it so sensitive a probe.

Limiting cases

Case	Result	Meaning
Orbital motion	$g_L=1$	the classical current-loop ratio; the baseline spin is measured against
Classical spinning charge	$g=1$	any rigid rotating charge gives $g=1$ — no classical model yields $2$
Dirac particle, tree level	$g=2$ exactly	structureless relativistic spin-½; forced by the four-spinor, independent of charge or mass
QED one loop	$g=2(1+\alpha/2\pi)$	virtual photons dress the vertex; $a=\alpha/2\pi\approx1.16\times10^{-3}$ (Schwinger)
Muon	$a_\mu\approx1.166\times10^{-3}$	heavier lepton, same tree $g=2$ ; loops enhanced by $(m_\mu/m_e)^2\approx4\times10^4$ for heavy states
Proton	$g_p\approx5.586$	huge anomaly ( $a_p\approx1.79$ ) — the proton is not a Dirac point particle; its quark substructure shows

Common trapThe electron is a spinning ball of charge, and g=2 just means it spins twice as fast

Neither half is right. No classical rigid charge distribution — ball, ring, shell — gives $g=2$ ; they all give $g=1$ , the same ratio as orbital motion, however the charge and mass are arranged. You cannot rescue $g=2$ by spinning faster, because $g$ is the ratio of moment to angular momentum, so "spinning faster" raises both and leaves $g$ unchanged — and a literal electron of classical radius would need its surface to move far faster than light to carry spin $\hbar/2$ . The value $g=2$ is not a mechanical fact about a rotating object; it is a consequence of the Dirac spinor's structure under relativistic quantum mechanics. Its doubled magnetism is the signature that spin comes from relativity plus quantum mechanics, not geometry. Likewise the later excess $g-2$ is not the electron finally acquiring a size — it is the vacuum's virtual particles contributing, calculable to a dozen digits with a point electron.

Worked examples

1 · The electron's magnetic moment is one Bohr magneton (confidence builder)

Show that a spin-½ Dirac electron ( $g=2$ , charge $q=-e$ ) has a magnetic moment of magnitude one Bohr magneton, and evaluate it.

Step 1. From equation (1), the moment along the spin axis for a spin-½ state ( $S_z=\hbar/2$ ) is $\mu=g\,\dfrac{e}{2m_e}\,\dfrac{\hbar}{2}=\dfrac{g}{2}\,\dfrac{e\hbar}{2m_e}=\dfrac{g}{2}\,\mu_B$ , with $\mu_B\equiv e\hbar/2m_e$ the Bohr magneton. The factor $g/2$ does all the work.

Step 2. Put in $g=2$ : $\mu=\mu_B$ . The Dirac electron carries exactly one Bohr magneton — which is why $\mu_B$ is the natural unit of atomic magnetism in the first place.

Step 3. Evaluate with $e=1.602\times10^{-19}\,\mathrm{C}$ , $\hbar=1.055\times10^{-34}\,\mathrm{J\,s}$ , $m_e=9.109\times10^{-31}\,\mathrm{kg}$ : $\mu_B=\dfrac{e\hbar}{2m_e}=\mathbf{9.274\times10^{-24}\ J/T}$ . Including the anomaly the true moment is $1.00116\,\mu_B$ — the extra $0.116\%$ is all of $g-2$ . |μ| = (g/2) μ_B = μ_B = 9.274×10⁻²⁴ J/T at tree level.

2 · Schwinger's number, and why one loop is not enough (fully worked)

Compute the one-loop anomaly $a=\alpha/2\pi$ using $\alpha^{-1}=137.036$ , compare to the measured $a_e=1.15965218\times10^{-3}$ , and show the two-loop term closes the gap.

Step 1. One loop: $a^{(1)}=\dfrac{\alpha}{2\pi}=\dfrac{1}{2\pi\times137.036}=\dfrac{1}{861.02}=\mathbf{1.161410\times10^{-3}}$ , matching the measured value to three significant figures — Schwinger's 1948 triumph — and giving $g/2=1.0011614$ .

Step 2. Compare to experiment $a_e=\mathbf{1.15965218\times10^{-3}}$ . The one-loop value is too large by about $1.8\times10^{-6}$ — a discrepancy at the fifth significant figure, far outside the measurement's precision. One loop is not enough.

Step 3. Add the two-loop term $a^{(2)}=-0.328479\,\big(\dfrac{\alpha}{\pi}\big)^2$ . With $\alpha/\pi=2.32282\times10^{-3}$ , $(\alpha/\pi)^2=5.3955\times10^{-6}$ , so $a^{(2)}=\mathbf{-1.772\times10^{-6}}$ . The negative sign pulls the prediction down, exactly where it needs to go.

Step 4. Sum: $a\approx1.161410\times10^{-3}-1.772\times10^{-6}=\mathbf{1.159638\times10^{-3}}$ , agreeing with $1.159652\times10^{-3}$ to six figures; the tiny remaining gap is filled by three- through five-loop QED plus hadronic and weak pieces. One loop gives 1.1614×10⁻³ (3 digits); two loops give 1.15964×10⁻³ (6 digits) — QED converges onto the data.

3 · Why the muon, not the electron, hunts new physics

The muon and electron are both spin-½ leptons with tree-level $g=2$ . Explain quantitatively why $a_\mu$ is far more sensitive to a hypothetical heavy particle of mass $M$ .

Step 1. A virtual heavy particle of mass $M$ contributes to a lepton's anomaly a piece scaling as $\delta a_\ell\sim(m_\ell/M)^2$ : the amplitude to "reach" a heavy state grows with the lepton's own mass, so the heavier lepton wins.

Step 2. The ratio of sensitivities is $\dfrac{\delta a_\mu}{\delta a_e}\sim\Big(\dfrac{m_\mu}{m_e}\Big)^2=\Big(\dfrac{105.66}{0.5110}\Big)^2=(206.8)^2=\mathbf{4.28\times10^{4}}$ — the muon is about forty-three thousand times more responsive to heavy new physics.

Step 3. That is why the Fermilab Muon g-2 experiment, measuring $a_\mu\approx1.16592\times10^{-3}$ , is a front-line new-physics search even though $a_e$ is measured far more precisely: a part-per-million discrepancy in $a_\mu$ could signal particles too heavy to make directly. The (m_μ/m_e)² ≈ 4×10⁴ enhancement makes the muon the discovery channel, the electron the precision-QED and α-determination channel.

Physics of the portrait

The magnetic moment carries two names above all others: Dirac, who predicted the integer, and Schwinger, who computed the first decimal — the anomaly $\alpha/2\pi$ , emblem of renormalized quantum field theory, carved on Schwinger's tombstone.

Origins · 1 minJulian Schwinger and how this was discovered

Julian Schwinger

1918–1994 · quantum electrodynamics and the anomalous magnetic moment

Julian Schwinger was a prodigy who published his first physics paper at sixteen and held a Harvard professorship by his late twenties. In 1947, at the Shelter Island conference, Kusch and Foley reported that the electron's $g$ -factor was not exactly 2 but about $2.00229$ — a small effect the naive theory could not explain and that, taken literally, involved infinite quantities. Within months Schwinger produced the calculation that turned quantum electrodynamics from an embarrassment into the most precise theory ever built: by separating the divergent charge-renormalization piece from the finite magnetic piece of the one-loop vertex, he obtained the anomalous moment $a=(g-2)/2=\alpha/2\pi\approx0.00116$ , in immediate agreement with experiment. It was the first hard number to come out of renormalization, and it convinced the community that the infinities of field theory were an artifact of faulty bookkeeping, not a fatal flaw. Schwinger shared the 1965 Nobel Prize with Feynman and Tomonaga for the consistent formulation of QED, which the three had reached by strikingly different routes — Schwinger's the most formal and operator-based, Feynman's the most pictorial. The formula $\alpha/2\pi$ is engraved on his tombstone in Mount Auburn Cemetery, a fitting monument, since that number remains, three-quarters of a century later, the gateway to the most stringently verified prediction in all of science.

History

1925

— Uhlenbeck and Goudsmit propose electron spin, and spectra demand g = 2

1928

— Dirac's relativistic equation predicts g = 2 with no free parameter

1947

— Kusch and Foley measure g slightly greater than 2, revealing the anomaly

1948

— Schwinger computes a = (g−2)/2 = α/2π, the birth of precision QED

2006

— Gabrielse's group measures the electron g/2 to better than a part per trillion

2021

— Fermilab's Muon g-2 confirms the earlier Brookhaven anomaly in a_μ

Going further

Texts. Peskin & Schroeder, An Introduction to Quantum Field Theory, §6.2–6.3 (the electron vertex function and the one-loop $g-2$ ) · Griffiths, Introduction to Elementary Particles, 2nd ed., §7.6 · Halzen & Martin, Quarks and Leptons, §6 · Thomson, Modern Particle Physics, §4.4–4.7 (spin and the Dirac equation) · Schwartz, Quantum Field Theory and the Standard Model, §17 (the anomalous magnetic moment). Key numerical checks: $(\boldsymbol\sigma\cdot\boldsymbol\pi)^2=\boldsymbol\pi^2-q\,\boldsymbol\sigma\cdot\mathbf B$ yields $g=2$ from the $\boldsymbol\sigma\cdot\mathbf B$ coefficient in the Pauli equation; $\alpha/2\pi=1.16141\times10^{-3}$ and the two-loop term $-0.328479(\alpha/\pi)^2=-1.772\times10^{-6}$ sum to $a_e\approx1.15964\times10^{-3}$ ; $\mu_B=e\hbar/2m_e=9.274\times10^{-24}$ J/T and $(m_\mu/m_e)^2=4.28\times10^4$ . The measured values $g_e/2=1.00115965218$ and $a_\mu\approx1.16592\times10^{-3}$ are standard CODATA/experimental figures.

Prerequisites. The Dirac Equation · Gamma Matrices and the Clifford Algebra · Non-relativistic Quantum Mechanics Review. Leads to. Spin and Helicity States · Plane-Wave Dirac Solutions · The QED Vertex and Loop Corrections.

Ask the tutor: Ask me to walk through the σ·π squaring that conjures $g=2$ , or to trace how Schwinger's $\alpha/2\pi$ grows into the twelve-digit test of the electron.

Why exactly 2, and not some other number close to 2?

Because the Dirac equation is a first-order relativistic wave equation whose four-component spinor is built so that its square reproduces the Klein–Gordon equation. That construction fixes how spin couples to the field with no freedom left over: in the non-relativistic limit the coefficient of $\boldsymbol\sigma\cdot\mathbf B$ comes out at exactly twice the orbital value. There is no adjustable parameter — no place to put a $1.9$ or a $2.1$ — so $g=2$ is a genuine prediction. The measured departures are not corrections to this number but a separate effect: quantum loops, absent from the bare Dirac equation.

Is g=2 special to the electron, or does every spin-½ particle have it?

Every elementary, structureless, charged spin-½ particle described by the Dirac equation has $g=2$ at tree level — electron, muon, tau, and the quarks all start there. Composite particles do not: the proton has $g_p\approx5.586$ and the neutron, though neutral, a nonzero moment, both being bound states of quarks and gluons rather than point Dirac particles. So $g=2$ is a test of elementarity: a $g$ at 2 (up to small calculable loops) means a point particle; a $g$ far from 2 means structure.

What is the difference between g, g−2, and a?

The g-factor $g$ is the full ratio of moment to spin in units of $q/2m$ . The quantity $g-2$ is the excess over the Dirac prediction — everything beyond tree level. The anomaly $a=(g-2)/2$ is the same halved, with a clean meaning: it equals the magnetic form factor at zero momentum transfer, $F_2(0)$ , and is what perturbation theory computes directly as a series in $\alpha/\pi$ . Experimentalists quote $g/2=1+a$ ; theorists quote $a$ . But $a$ is the natural object because tree level contributes zero to it, so every digit of $a$ is quantum physics.

How can a point particle with no size have a calculable magnetic moment correction?

This is the beautiful part. The correction $g-2$ does not come from the electron having a size; it comes from the vacuum. A physical electron is perpetually surrounded by a cloud of virtual photons and pairs it emits and reabsorbs, and while such a fluctuation is present its coupling to a field is slightly modified. Summing these gives $a=\alpha/2\pi+\cdots$ . The electron stays a point; the corrections live in the fields around it. That is why the calculation works to twelve digits with a strictly pointlike electron — the "structure" probed is that of the vacuum, not the particle.

Why is the muon g−2 considered a search for new physics when QED already predicts a?

Because $a$ gets contributions from every particle that can appear in a loop, not just photons and electrons. The QED and electroweak parts are calculable, but any undiscovered particle would also contribute — its effect scaling as $(m_\mu/M)^2$ , so heavy states matter more for the muon by a factor of about forty thousand. If measured $a_\mu$ disagrees with the complete Standard Model prediction, the gap signals something not in the Standard Model. The situation is subtle: the measurement is superb, but the hadronic (strong-interaction) piece of the theory is hard, and data-driven versus lattice-QCD estimates disagree, so the size of any anomaly is under active revision.

Does the anomalous moment mean Dirac's g=2 was wrong?

Not at all — it means Dirac's equation was the correct leading description and quantum field theory is the correction on top. Dirac's $g=2$ is exact for the idealized single-particle theory with no radiation. Real electrons interact with the quantized field, and that interaction, absent from the one-particle Dirac equation, produces the anomaly. Far from overturning Dirac, the anomaly was the first place QED was shown to make quantitative sense: Schwinger's $\alpha/2\pi$ builds on Dirac's $2$ , the two being consecutive terms of one expansion $g=2(1+\alpha/2\pi-\cdots)$ .