Relativistic electron-proton scattering

Rutherford and Mott built their scattering formulae on a comfortable fiction: a proton so heavy it never moves, an infinitely massive point of charge that deflects the electron and absorbs nothing in return. That fiction is excellent when the momentum transfer is small compared with the proton mass, but it fails once an experimenter turns up the beam energy. Once the four-momentum transferred to the target becomes comparable to $m_p$ , the proton visibly recoils, the scattered electron comes out with less energy than it went in with, and a new interaction switches on — the coupling of the electron's current to the proton's magnetic moment, invisible in the static limit. This article treats $e^-p \to e^-p$ at those energies with the proton still idealized as a structureless Dirac point particle, the last stop before its real internal structure has to be admitted. The payoff is a single differential cross section that is the honest relativistic descendant of Mott's. It carries a recoil factor $E_3/E_1$ that Rutherford never needed, and it splits cleanly into two pieces: an electric term $\propto \cos^2(\theta/2)$ that remembers the charge, and a magnetic term $\propto (Q^2/2m_p^2)\sin^2(\theta/2)$ that remembers the spin and grows as $Q^2$ rises. Everything is expressed in the two quantities an experiment actually records — the scattering angle $\theta$ and the outgoing energy $E_3$ — because the recoiling proton usually escapes undetected. Attach form factors to the two terms and this becomes the Rosenbluth formula; that is why the point-proton result is worth deriving with care.

Key takeaways

At high energy the proton recoils, so the scattered electron loses energy: $E_3 \ne E_1$ , and the magnetic spin coupling can no longer be dropped.
Elastic kinematics collapse onto two observables — the angle $\theta$ and energy $E_3$ — tied by $Q^2 = 4E_1E_3\sin^2(\theta/2)$ and the recoil relation $E_1 - E_3 = Q^2/2m_p$ .
For a point-like Dirac proton the cross section is $\dfrac{d\sigma}{d\Omega} = \dfrac{\alpha^2}{4E_1^2\sin^4(\theta/2)}\dfrac{E_3}{E_1}\left(\cos^2\tfrac{\theta}{2} + \dfrac{Q^2}{2m_p^2}\sin^2\tfrac{\theta}{2}\right)$ .
The $\cos^2(\theta/2)$ piece is the electric (Mott-like) contribution; the $\sin^2(\theta/2)$ piece is the magnetic contribution, which vanishes as $Q^2\to 0$ and dominates at backward angles.
Setting the two proton form factors to unity ( $G_E=G_M=1$ ) turns this into the elastic charge–plus–point-magnetic-moment limit of the Rosenbluth formula.
Key idea: once proton recoil is included, elastic $e^-p$ scattering is fixed by just $\theta$ and $E_3$ (equivalently $Q^2$ ), and the cross section separates into a charge term and a spin term whose relative size is set by $Q^2/m_p^2$ .

The core insight

When proton recoil is accounted for, elastic electron–proton scattering kinematics collapse onto just two observables ( $\theta$ and $E_3$ , equivalently $Q^2$ ), and the cross section splits cleanly into an electric (charge) piece $\propto\cos^2(\theta/2)$ and a magnetic (spin) piece $\propto (Q^2/2m_p^2)\sin^2(\theta/2)$ that grows with momentum transfer.

Where this lives today

The point-Dirac cross section is the null hypothesis of nucleon physics: every modern elastic electron–proton experiment is, at bottom, a measurement of how the real proton departs from equation (4). At Jefferson Lab, few-GeV electrons scattered off hydrogen are fit with the Rosenbluth formula to extract the electric and magnetic form factors $G_E(Q^2)$ and $G_M(Q^2)$ across a wide range of $Q^2$ . A live controversy sharpens the point: the classic Rosenbluth-separation method and the newer polarization-transfer method disagree about the ratio $G_E/G_M$ at high $Q^2$ , a discrepancy traced to two-photon-exchange corrections beyond the single-photon amplitude derived here. The proton-radius puzzle is another descendant — muonic-hydrogen spectroscopy gives a radius near $0.841\ \mathrm{fm}$ while some electron determinations sat higher, prompting the PRad experiment at Jefferson Lab (which measured $0.831\ \mathrm{fm}$ from very-low- $Q^2$ scattering) and MUSE at PSI, which compares electron and muon scattering off the proton in one apparatus. In every case the analysis begins from the two-term structure of this article and asks what form factors the data demand.

The physical picture

A relativistic electron is essentially a moving current. When it passes a proton it exchanges a single virtual photon, and two things happen at once. The photon carries away four-momentum, kicking the proton forward so that it recoils — which is why the outgoing energy $E_3$ is locked to the scattering angle and always less than the incoming $E_1$ . And the photon couples not only to the proton's charge but to the current of its spin, an interaction with no static analogue that strengthens as $Q^2$ grows. The first figure fixes the kinematics: it shows the lab-frame geometry and the two relations that turn "angle and energy" into "momentum transfer".

Figure

Lab-frame elastic geometry for e⁻p → e⁻p: the electron enters with energy E₁, scatters through angle θ with reduced energy E₃, and the proton recoils. The single virtual photon carries momentum transfer Q² = 4E₁E₃ sin²(θ/2), and energy conservation locks the electron energy loss to E₁ − E₃ = Q²/2m_p

The second figure is the physical heart of the result: the cross section is a sum of an electric term and a magnetic term, and which one wins depends entirely on the scattering angle. Forward, near $\theta=0$ , the charge term $\cos^2(\theta/2)$ dominates and the scattering looks essentially Coulombic. Swing to backward angles and the roles invert — the magnetic term $(Q^2/2m_p^2)\sin^2(\theta/2)$ takes over because $Q^2$ is largest there and the spin coupling scales with it. For a $0.5\ \mathrm{GeV}$ beam the two contributions cross near $128^\circ$ ; beyond that the "magnetic" scattering off a point Dirac proton is the larger effect.

Figure

The two contributions to the point-proton bracket versus scattering angle for a 0.5 GeV beam: the electric term cos²(θ/2) falls from 1 while the magnetic term (Q²/2m_p²)sin²(θ/2) rises to 0.28 at 180°. They cross near θ = 128° — charge dominates forward, spin dominates backward

The core equations

The whole calculation is bookkeeping around one exchanged photon, so the first quantity to pin down is what that photon carries. Neglecting the electron mass against its energy, the space-like invariant $Q^2 = -q^2$ with $q = p_1 - p_3$ evaluates in the lab frame to a compact product of the two measured quantities.

Q^2 \equiv -q^2 = 4E_1 E_3 \sin^2\!\frac{\theta}{2} \qquad (1)

where: $Q^2$ — positive four-momentum transfer squared (units of energy $^2$ ); $E_1,E_3$ — incoming and outgoing electron energies; $\theta$ — electron scattering angle. Reading it: $Q^2$ grows with energy and, at fixed energy, with angle — it is largest for backscattering, smallest in the forward direction. Elasticity — the proton stays a proton — forces a second relation. Requiring the recoiling proton to sit on its mass shell, $p_4^2 = m_p^2$ , ties the electron's energy loss rigidly to the momentum transfer, so the electron cannot scatter at a chosen angle and keep its energy.

E_1 - E_3 = \frac{Q^2}{2m_p}, \qquad E_3 = \frac{E_1}{1 + \dfrac{2E_1}{m_p}\sin^2(\theta/2)} \qquad (2)

where: $m_p$ — proton mass, $0.938\ \mathrm{GeV}$ . The second form is (1) and the first combined and solved for $E_3$ : fix the beam energy and the angle, and the outgoing energy is determined. The heavier the target, the smaller the recoil and the closer $E_3$ creeps back to $E_1$ . The dynamics live in the spin-averaged squared matrix element, obtained by contracting the electron's current tensor with the proton's and summing over final spins while averaging over initial ones. For two point-like Dirac particles, with $m_e$ neglected, the lab-frame result condenses to the same two-term bracket that will survive into the cross section.

\langle |\mathcal{M}|^2\rangle = \frac{16\,e^4\,E_1 E_3\,m_p^2}{Q^4}\left(\cos^2\!\frac{\theta}{2} + \frac{Q^2}{2m_p^2}\sin^2\!\frac{\theta}{2}\right) \qquad (3)

where: $e^4/Q^4$ — the two photon vertices and the photon propagator squared; the bracket holds the electric ( $\cos^2$ ) and magnetic ( $\sin^2$ ) pieces. The $1/Q^4$ is the long-range Coulomb signature; the magnetic term is the piece that Rutherford and Mott never see. Folding (3) through the lab-frame flux and two-body phase space delivers the central result of the article — the point-proton differential cross section, the relativistic generalisation of the Mott formula.

\frac{d\sigma}{d\Omega} = \frac{\alpha^2}{4E_1^2\sin^4(\theta/2)}\,\frac{E_3}{E_1}\left(\cos^2\!\frac{\theta}{2} + \frac{Q^2}{2m_p^2}\sin^2\!\frac{\theta}{2}\right) \qquad (4)

where: $\alpha = e^2/4\pi \approx 1/137$ ; the prefactor $\alpha^2/4E_1^2\sin^4(\theta/2)$ is the Mott/Rutherford envelope; $E_3/E_1$ is the recoil factor; the bracket is charge-plus-spin. As $m_p\to\infty$ the recoil factor $\to 1$ , the magnetic term $\to 0$ , and (4) collapses to the Mott cross section. It is often cleaner to factor the electric term out and write the bracket with the dimensionless variable $\tau = Q^2/4m_p^2$ , which exposes the angular structure that a Rosenbluth separation exploits.

\frac{d\sigma}{d\Omega} = \left(\frac{d\sigma}{d\Omega}\right)_{\!\mathrm{Mott}}\frac{E_3}{E_1}\left(1 + 2\tau\tan^2\!\frac{\theta}{2}\right), \qquad \tau \equiv \frac{Q^2}{4m_p^2} \qquad (5)

where: $(d\sigma/d\Omega)_{\mathrm{Mott}} = \alpha^2\cos^2(\theta/2)/4E_1^2\sin^4(\theta/2)$ . Setting the proton's electric and magnetic form factors to unity, $G_E=G_M=1$ , in the Rosenbluth formula reproduces (5) exactly — this is what "point Dirac proton" means quantitatively.

Derivation AKinematics of elastic scattering with recoil

The goal is to derive the two relations that make elastic electron–proton scattering a two-variable problem: the momentum transfer $Q^2 = 4E_1E_3\sin^2(\theta/2)$ and the recoil constraint $E_1 - E_3 = Q^2/2m_p$ , and from them the explicit outgoing energy $E_3(\theta)$ . Everything follows from four-momentum conservation plus the single physical statement that the proton survives intact, so its final four-momentum squares to $m_p^2$ . We neglect the electron mass throughout, valid because every experiment of interest has $E_1 \gg m_e = 0.511\ \mathrm{MeV}$ .

Step 1. Label the four-momenta $p_1$ (incoming electron), $p_2 = (m_p,\mathbf{0})$ (target proton at rest), $p_3$ (scattered electron), $p_4$ (recoil proton), and define $q = p_1 - p_3$ , the momentum handed to the proton. Form the invariant $q^2 = (p_1 - p_3)^2 = m_e^2 + m_e^2 - 2p_1\cdot p_3 \approx -2E_1E_3(1-\cos\theta)$ , using $|\mathbf{p}_1|\approx E_1$ , $|\mathbf{p}_3|\approx E_3$ . The half-angle identity $1-\cos\theta = 2\sin^2(\theta/2)$ gives $Q^2 = -q^2 = 4E_1E_3\sin^2(\theta/2)$ . Why: $Q^2$ is a Lorentz invariant, so we may evaluate it in whichever frame is convenient — here the lab, where it becomes a product of the two things the detector reads.

Step 2. Impose elasticity through the recoil proton. Since $p_4 = p_1 + p_2 - p_3 = p_2 + q$ , square it: $p_4^2 = p_2^2 + 2p_2\cdot q + q^2 = m_p^2 + 2m_p(E_1 - E_3) - Q^2$ , because $p_2\cdot q = m_p(E_1-E_3)$ in the rest frame and $q^2 = -Q^2$ . Why: the proton remaining a proton is the entire dynamical input of "elastic" — it is the constraint that removes one kinematic freedom.

Step 3. Set $p_4^2 = m_p^2$ . The $m_p^2$ cancels and $2m_p(E_1 - E_3) = Q^2$ , i.e. $E_1 - E_3 = Q^2/2m_p$ . The electron necessarily loses energy: recoil is not optional. Why: this is the relativistic replacement for "the heavy target absorbs no energy" — the loss is small only when $Q^2 \ll m_p^2$ , exactly the Mott regime.

Step 4. Eliminate $Q^2$ between Steps 1 and 3. Substitute $Q^2 = 4E_1E_3\sin^2(\theta/2)$ into $E_1 - E_3 = Q^2/2m_p$ to get $E_1 - E_3 = (2E_1E_3/m_p)\sin^2(\theta/2)$ , then solve for $E_3$ : $E_3 = E_1/[1 + (2E_1/m_p)\sin^2(\theta/2)]$ . For $E_1 = 1\ \mathrm{GeV}$ , $\theta = 30^\circ$ this returns $E_3 = 0.875\ \mathrm{GeV}$ and $Q^2 = 0.234\ \mathrm{GeV}^2$ . Why: this closes the system — angle in, energy out — and shows the recoil is a few-percent effect at a GeV but total for hard backscattering.

Go deeper · Derivation B · 1 minThe spin-averaged matrix element for two Dirac particles

Derivation BThe spin-averaged matrix element for two Dirac particles

The aim is to obtain equation (3), the lab-frame spin-averaged $\langle|\mathcal{M}|^2\rangle$ for $e^-p \to e^-p$ with a point Dirac proton, and to see the electric and magnetic terms emerge from a purely covariant calculation. The strategy is the standard QED factorisation: one photon exchange means the amplitude is a product of an electron current and a proton current tied by the photon propagator, so the squared, spin-summed amplitude factorises into a leptonic tensor times a hadronic tensor, each a trace of gamma matrices. We keep the proton point-like (bare $\gamma^\mu$ vertex) and neglect $m_e$ .

Step 1. Write the tree amplitude $\mathcal{M} = \dfrac{e^2}{q^2}\,[\bar u(p_3)\gamma^\mu u(p_1)]\,[\bar u(p_4)\gamma_\mu u(p_2)]$ : two currents joined by the photon propagator $-g_{\mu\nu}/q^2$ . Squaring and using the spin-sum completeness relations turns $\langle|\mathcal{M}|^2\rangle = \tfrac14\sum_{\mathrm{spins}}|\mathcal{M}|^2$ into $(e^4/q^4)\,L^{\mu\nu}_{(e)}\,L^{(p)}_{\mu\nu}$ . Why: with only one propagator the problem separates — the electron and proton never talk except through $q$ , so their traces can be computed independently.

Step 2. Evaluate the electron tensor with the trace theorems: $L^{\mu\nu}_{(e)} = \tfrac12\mathrm{Tr}[\slashed{p}_3\gamma^\mu\slashed{p}_1\gamma^\nu] = 2\left(p_1^\mu p_3^\nu + p_1^\nu p_3^\mu - g^{\mu\nu}\,p_1\cdot p_3\right)$ , dropping $m_e$ . The proton tensor $L^{(p)}_{\mu\nu}$ has the same structure with $p_2,p_4,m_p$ retained. Why: the gamma-matrix traces convert operator spinor algebra into ordinary four-vector dot products — the machinery that makes cross sections computable.

Step 3. Contract the two tensors. The result, valid for any frame, is $\langle|\mathcal{M}|^2\rangle = \dfrac{8e^4}{Q^4}\left[(p_1\cdot p_2)(p_3\cdot p_4) + (p_1\cdot p_4)(p_2\cdot p_3) - m_p^2\,(p_1\cdot p_3)\right]$ . Why: this compact invariant is the workhorse — it is the same object that gives $e^-\mu^-$ scattering, since a point proton is just a heavy muon as far as QED is concerned.

Step 4. Insert lab-frame values. With $p_2=(m_p,\mathbf 0)$ : $p_1\cdot p_2 = E_1 m_p$ , $p_2\cdot p_3 = E_3 m_p$ , $p_1\cdot p_3 = Q^2/2$ , and via $p_4 = p_1+p_2-p_3$ , $p_3\cdot p_4 = Q^2/2 + E_3 m_p$ and $p_1\cdot p_4 = E_1 m_p - Q^2/2$ . The bracket becomes $2E_1E_3 m_p^2 + Q^4/4 - m_p^2 Q^2/2$ , which — using $E_1E_3\sin^2(\theta/2) = Q^2/4$ — regroups exactly into $2E_1E_3 m_p^2\big[\cos^2(\theta/2) + (Q^2/2m_p^2)\sin^2(\theta/2)\big]$ . Hence $\langle|\mathcal{M}|^2\rangle = (16e^4 E_1E_3 m_p^2/Q^4)\big[\cos^2(\theta/2) + (Q^2/2m_p^2)\sin^2(\theta/2)\big]$ , equation (3). Why: the abstract dot products reorganise into the two physical terms — charge and spin — that the experiment separates by varying the angle.

Go deeper · Derivation C · 1 minFrom matrix element to the cross section

Derivation CFrom matrix element to the cross section

The purpose here is to convert the dynamical input $\langle|\mathcal{M}|^2\rangle$ into the measurable $d\sigma/d\Omega$ of equation (4), and to isolate where the recoil factor $E_3/E_1$ comes from. The route is the fixed-target two-body cross-section formula: divide the matrix element by the incident flux and multiply by the density of final states, then integrate over the unobserved recoil proton, whose momentum is fixed by the delta functions. The one subtlety is that the outgoing electron energy is not independent of its angle, and accounting for that constraint is precisely what produces the recoil factor.

Step 1. Start from the lab-frame master formula for $2\to 2$ scattering off a fixed target, which after doing the recoil-proton integral reads $\dfrac{d\sigma}{d\Omega} = \dfrac{1}{64\pi^2 m_p^2}\,\dfrac{E_3^2}{E_1^2}\,\langle|\mathcal{M}|^2\rangle$ , with the factor $E_3^2/E_1^2$ carrying the phase-space Jacobian for the constrained final energy. Why: because $E_3$ is pinned to $\theta$ by recoil, integrating the energy delta function leaves a Jacobian $\partial E_3/\partial(\dots)$ that is the origin of the recoil factor — not an afterthought but a phase-space effect.

Step 2. Insert equation (3): $\dfrac{d\sigma}{d\Omega} = \dfrac{1}{64\pi^2 m_p^2}\dfrac{E_3^2}{E_1^2}\cdot\dfrac{16 e^4 E_1 E_3 m_p^2}{Q^4}\big[\cos^2(\theta/2)+\tfrac{Q^2}{2m_p^2}\sin^2(\theta/2)\big]$ . The $m_p^2$ cancels — the point proton's mass survives only inside the magnetic term and the kinematics, not as an overall scale. Why: the cancellation is the check that a "heavy target" leaves no spurious mass dependence in the overall normalisation.

Step 3. Simplify with $e^4 = (4\pi\alpha)^2 = 16\pi^2\alpha^2$ and $Q^4 = 16E_1^2E_3^2\sin^4(\theta/2)$ . Collecting powers gives $\dfrac{d\sigma}{d\Omega} = \dfrac{\alpha^2}{4E_1^2\sin^4(\theta/2)}\,\dfrac{E_3}{E_1}\big[\cos^2(\theta/2)+\tfrac{Q^2}{2m_p^2}\sin^2(\theta/2)\big]$ , equation (4). Why: the $1/\sin^4(\theta/2)$ Rutherford envelope, the $E_3/E_1$ recoil factor, and the charge-plus-spin bracket now stand cleanly separated, each with a distinct physical origin.

Step 4. Check the non-relativistic limit. As $m_p\to\infty$ at fixed $E_1$ , equation (2) gives $E_3\to E_1$ so $E_3/E_1\to 1$ , and $Q^2/2m_p^2\to 0$ so the magnetic term vanishes, leaving $d\sigma/d\Omega \to (\alpha^2/4E_1^2\sin^4(\theta/2))\cos^2(\theta/2)$ — the Mott cross section. Why: recovering Mott in the heavy-target limit is the decisive consistency test, confirming that (4) is the honest relativistic parent of the formulae the reader already trusts.

Limiting cases

Case	Result	Meaning
$Q^2 \ll m_p^2$ (low energy)	$E_3\to E_1$ , magnetic term $\to 0$	recoil and spin negligible; (4) collapses to the Mott cross section
Forward, $\theta\to 0$	bracket $\to \cos^2(\theta/2)\to 1$ , but $1/\sin^4(\theta/2)\to\infty$	pure charge (Coulomb) scattering; the familiar forward divergence
Backward, $\theta\to\pi$	electric term $\to 0$ , magnetic term survives	spin (magnetic) scattering dominates; the $\cos^2$ piece switches off entirely
$m_p\to\infty$ (static target)	$d\sigma/d\Omega \to$ Mott	Rutherford/Mott is the infinitely-heavy-proton corner of this result
$G_E,G_M\to 1$	equation (5)	this whole article is the point-form-factor limit of Rosenbluth
Ultra-relativistic, both massless	$\langle\|\mathcal M\|^2\rangle \propto (s^2+u^2)/t^2$	reduces to massless $e^-\mu^-$ scattering

Common trapAssuming the electron keeps its energy and dropping the magnetic term

The seductive error is to carry the Rutherford/Mott mental model into the relativistic regime: treat the proton as a fixed scattering centre so that $E_3 = E_1$ , and keep only the electric $\cos^2(\theta/2)$ term. Both approximations fail together once $Q^2$ is not small compared with $m_p^2$ . The electron must lose energy — the loss $E_1 - E_3 = Q^2/2m_p$ is fixed by kinematics, not a correction you can switch off — and the magnetic term, which scales as $Q^2/2m_p^2$ , is exactly the piece that grows in the same regime where recoil matters. At backward angles it is not a small correction but the dominant contribution: for a $0.5\ \mathrm{GeV}$ beam at $\theta=150^\circ$ the magnetic term is nearly four times the electric one, so dropping it loses most of the cross section.

Worked examples

1 · Kinematics and cross section at a forward angle (fully worked)

A $1.00\ \mathrm{GeV}$ electron scatters elastically off a proton ( $m_p = 0.938\ \mathrm{GeV}$ ) at $\theta = 30^\circ$ . Find the outgoing energy $E_3$ , the momentum transfer $Q^2$ , the energy loss, the relative size of the magnetic term, and the differential cross section.

Step 1. Outgoing energy from equation (2). With $\sin^2(15^\circ) = 0.06699$ and $2E_1/m_p = 2.132$ : $E_3 = 1.00/[1 + 2.132\times 0.06699] = 1.00/1.1428 = \mathbf{0.875\ GeV}$ .

Step 2. Momentum transfer from equation (1): $Q^2 = 4(1.00)(0.875)(0.06699) = \mathbf{0.234\ GeV^2}$ . Cross-check the recoil relation: $E_1 - E_3 = Q^2/2m_p = 0.234/1.876 = 0.125\ \mathrm{GeV}$ , and indeed $1.00 - 0.875 = 0.125\ \mathrm{GeV}$ . The kinematics are consistent.

Step 3. Relative size of the two terms. $Q^2/2m_p^2 = 0.234/(2\times 0.880) = 0.133$ , and $\tan^2(15^\circ) = 0.0718$ , so the magnetic-to-electric ratio is $2\tau\tan^2(\theta/2) = 0.133\times 0.0718 = \mathbf{0.0096}$ — the magnetic term is about 1 % here. Forward scattering is nearly pure charge scattering.

Step 4. Cross section from equation (4). The Mott envelope is $\alpha^2/[4E_1^2\sin^4(\theta/2)] = (1/137.0)^2/[4(1.00)(0.06699)^2] = 5.33\times 10^{-5}/0.01795 = 2.97\times 10^{-3}\ \mathrm{GeV}^{-2}$ . Multiply by the recoil factor $E_3/E_1 = 0.875$ and the bracket $[\cos^2 15^\circ + 0.133\sin^2 15^\circ] = 0.933 + 0.0089 = 0.942$ : $d\sigma/d\Omega = 2.97\times 10^{-3}\times 0.875\times 0.942 = 2.45\times 10^{-3}\ \mathrm{GeV}^{-2}$ .

Step 5. Convert to physical units with $1\ \mathrm{GeV}^{-2} = 0.3894\ \mathrm{mb}$ : $d\sigma/d\Omega = 2.45\times 10^{-3}\times 0.3894 = 9.5\times 10^{-4}\ \mathrm{mb/sr} = \mathbf{0.95\ \mu b/sr}$ . A GeV electron at 30° recoils the proton by 125 MeV, transfers Q² = 0.234 GeV², and scatters at about 0.95 μb/sr — overwhelmingly electric.

2 · Why backward angles are "magnetic"

Take a $0.50\ \mathrm{GeV}$ beam and compare $\theta = 60^\circ$ (forward-ish) with $\theta = 150^\circ$ (backward). Show that the magnetic term goes from negligible to dominant.

Step 1. At $\theta = 60^\circ$ : $\sin^2(30^\circ) = 0.25$ , so $E_3 = 0.50/[1 + (1.066)(0.25)] = 0.395\ \mathrm{GeV}$ and $Q^2 = 4(0.50)(0.395)(0.25) = 0.197\ \mathrm{GeV}^2$ . Then $Q^2/2m_p^2 = 0.112$ , electric term $\cos^2 30^\circ = 0.750$ , magnetic term $0.112\times 0.25 = 0.0280$ . Ratio magnetic/electric $= \mathbf{0.037}$ — about 4 %.

Step 2. At $\theta = 150^\circ$ : $\sin^2(75^\circ) = 0.933$ , so $E_3 = 0.50/[1 + (1.066)(0.933)] = 0.251\ \mathrm{GeV}$ and $Q^2 = 4(0.50)(0.251)(0.933) = 0.468\ \mathrm{GeV}^2$ . Now $Q^2/2m_p^2 = 0.266$ , electric term $\cos^2 75^\circ = 0.0670$ , magnetic term $0.266\times 0.933 = 0.248$ . Ratio magnetic/electric $= \mathbf{3.70}$ .

Step 3. Interpret. Swinging from $60^\circ$ to $150^\circ$ the magnetic term climbs from 4 % of the electric to nearly four times it — a factor of $\sim 100$ swing in relative importance — driven by two compounding effects: $Q^2$ more than doubles (so the coupling $Q^2/2m_p^2$ grows) and $\sin^2(\theta/2)/\cos^2(\theta/2) = \tan^2(\theta/2)$ blows up. The same point proton looks like a charge forward and like a magnetic moment backward; this angular lever is exactly what a Rosenbluth separation later uses to isolate G_E from G_M.

Physics of the portrait

The point-proton cross section is a beautiful piece of theory whose real importance is that it is wrong in a productive way — measure the deviations from it and you read off the proton's size and shape. The man who turned that idea into a Nobel-winning programme was Robert Hofstadter, who bombarded protons and nuclei with the electron beams of Stanford's linear accelerators and watched the data peel away from the point prediction.

Origins · 1 minRobert Hofstadter and how this was discovered

Robert Hofstadter

1915–1990 · Stanford electron scattering and the structure of the nucleon

Robert Hofstadter built his reputation on a single, powerful idea: a high-energy electron is the cleanest probe of a charge distribution ever devised, because it interacts only electromagnetically and its scattering can be computed exactly for a point target. In the mid-1950s, using the electron beams of Stanford's Mark III linear accelerator and a precision magnetic spectrometer, he measured elastic electron scattering off protons and light nuclei and compared it with the point-Dirac cross section that this article derives. The data fell below the point prediction at large angles and momentum transfer — the unmistakable signature of a spread-out charge. Hofstadter and Robert McAllister reported in 1955 that the proton is not a point but an extended object roughly $0.7$ – $0.8\ \mathrm{fm}$ across, the first direct measurement of the proton's finite size, and he generalised the framework into the language of electric and magnetic form factors multiplying the two terms of the cross section. That programme — elastic electron scattering read as a Fourier transform of the nucleon's charge and magnetisation — earned him the 1961 Nobel Prize and set the template for every structure measurement that followed, from the deep-inelastic experiments that revealed quarks to today's proton-radius determinations.

History

1929

— Mott computes the relativistic Coulomb scattering of a spin-½ electron

1932

— Chadwick discovers the neutron, sharpening the question of nucleon structure

1950

— Rosenbluth derives the elastic electron–proton cross section with form factors

1955

— Hofstadter and McAllister measure the proton's finite size at Stanford

1956

— the proton rms charge radius is reported near 0.8 fm

1961

— Hofstadter shares the Nobel Prize for electron-scattering studies of the nucleon

Going further

Texts. Thomson, Modern Particle Physics, §7.4 (relativistic electron–proton elastic scattering) and §7.4.1 (scattering kinematics), eqs. (7.18)–(7.30) — the primary source for the notation and the lab-frame result used here · Halzen & Martin, Quarks and Leptons, §6.6 (electron–proton scattering and the Rosenbluth formula) · Griffiths, Introduction to Elementary Particles, 2nd ed., §7.7–7.8 (spin-averaged amplitudes and $e^-\mu^-$ scattering) · Peskin & Schroeder, An Introduction to Quantum Field Theory, §5.4 (trace technology, $e^-\mu^-$ cross section) · Schwartz, Quantum Field Theory and the Standard Model, §13 (elastic scattering and form factors).

Provenance. The kinematic relations $Q^2 = 4E_1E_3\sin^2(\theta/2)$ and $E_1-E_3 = Q^2/2m_p$ and the point-proton cross section (4) follow Thomson §7.4. The spin-averaged matrix element (3) is the standard leptonic-times-hadronic tensor contraction (Halzen–Martin §6.6; Peskin–Schroeder §5.4), specialised to a point Dirac proton; it coincides with the $e^-\mu^-$ result under $m_\mu \to m_p$ . A verifier should confirm: (1) $Q^2 = 4E_1E_3\sin^2(\theta/2)$ and $E_1-E_3=Q^2/2m_p$ from four-momentum conservation with $p_4^2=m_p^2$ ; (2) the lab-frame reduction of the covariant bracket $[(p_1\cdot p_2)(p_3\cdot p_4)+(p_1\cdot p_4)(p_2\cdot p_3)-m_p^2(p_1\cdot p_3)]$ to $2E_1E_3m_p^2[\cos^2(\theta/2)+(Q^2/2m_p^2)\sin^2(\theta/2)]$ ; (3) the assembly $d\sigma/d\Omega = (1/64\pi^2 m_p^2)(E_3^2/E_1^2)\langle|\mathcal M|^2\rangle$ into equation (4); (4) the worked numbers $E_1=1\ \mathrm{GeV}$ , $\theta=30^\circ \Rightarrow E_3=0.875\ \mathrm{GeV}$ , $Q^2=0.234\ \mathrm{GeV}^2$ , $d\sigma/d\Omega \approx 0.95\ \mu\mathrm{b/sr}$ .

Prerequisites. Rutherford and Mott Scattering · Mandelstam Variables · The Dirac Equation · Matrix Elements and Spin Averaging. Leads to. The Rosenbluth Formula · Electromagnetic Form Factors · Elastic versus Inelastic Scattering · Deep Inelastic Scattering.

Ask the tutor: Ask me to run the recoil kinematics and the electric-versus-magnetic split for any beam energy and angle you like, or to walk the point-proton cross section forward into the full Rosenbluth formula with real form factors.

Why does the electron have to lose energy when it scatters?

Because the proton is not infinitely heavy and must recoil. The instant the electron transfers momentum $q$ to the proton, the proton picks up kinetic energy, and that energy can only come from the electron. Elasticity — the proton staying a proton, so $p_4^2 = m_p^2$ — turns this into an exact bookkeeping relation, $E_1 - E_3 = Q^2/2m_p$ . The loss is tiny when $Q^2 \ll m_p^2$ (the Rutherford/Mott regime, where we pretend the target is fixed) and total for hard backscattering. There is no elastic geometry in which a relativistic electron scatters through a finite angle and keeps all its energy; the recoil is locked to the angle.

What does "point-like Dirac proton" actually mean here?

It means we treat the proton exactly as QED treats any fundamental spin-½ particle: a bare photon vertex $\gamma^\mu$ , no internal structure, a magnetic moment of exactly one nuclear magneton's worth of Dirac value ( $g=2$ ). Computationally the proton is then indistinguishable from a heavy muon, which is why the cross section is identical in form to $e^-\mu^-$ scattering. The real proton has a finite size and an anomalous magnetic moment, and those facts enter as the form factors $G_E(Q^2)$ and $G_M(Q^2)$ multiplying the two terms. The point-proton result is the baseline you measure deviations from; setting $G_E=G_M=1$ recovers it.

Where does the recoil factor E₃/E₁ come from?

From phase space, not from the dynamics. Integrating the two-body final state over the unobserved recoil proton, the energy-conserving delta function pins $E_3$ to the scattering angle and leaves a Jacobian. The fixed-target cross section carries $E_3^2/E_1^2$ from phase space, and combining with the $E_1E_3$ in the matrix element leaves one net power, $E_3/E_1$ , multiplying the Mott envelope. Physically the available final-state density is reduced because the electron's energy is no longer free — some is spent recoiling the proton.

Why is the backward direction called "magnetic" scattering?

Two things conspire at large angles. First, $Q^2$ is maximal for backscattering, and the magnetic term carries an explicit factor $Q^2/2m_p^2$ , so the spin coupling is strongest there. Second, the angular weights are $\cos^2(\theta/2)$ for the electric term and $\sin^2(\theta/2)$ for the magnetic term; their ratio is $\tan^2(\theta/2)$ , which diverges as $\theta\to\pi$ . So the charge term switches off while the spin term switches on. At exactly $180^\circ$ the electric contribution vanishes entirely and the scattering is purely magnetic — a helicity-flip process that only exists because the electron and proton both carry spin.

How is this different from the Rosenbluth formula?

It is the special case of the Rosenbluth formula in which the proton is structureless. Rosenbluth writes the cross section as the Mott envelope times $(E_3/E_1)$ times a bracket $[(G_E^2 + \tau G_M^2)/(1+\tau) + 2\tau G_M^2\tan^2(\theta/2)]$ , where $G_E$ and $G_M$ are the electric and magnetic form factors and $\tau = Q^2/4m_p^2$ . Put $G_E = G_M = 1$ : the first term becomes $(1+\tau)/(1+\tau) = 1$ and you recover equation (5), $[1 + 2\tau\tan^2(\theta/2)]$ , exactly the point result. So this article supplies the skeleton; Rosenbluth adds the flesh of real proton structure onto the same two terms.

Can I ignore the electron mass everywhere in this derivation?

For every practical purpose, yes. The electron mass is $0.511\ \mathrm{MeV}$ , while the beam energies of interest run from hundreds of MeV to many GeV, so $m_e/E_1 \lesssim 10^{-3}$ and $m_e^2$ terms are negligible against $E_1E_3$ and $Q^2$ . Setting $m_e=0$ is what lets $Q^2 = 4E_1E_3\sin^2(\theta/2)$ take its clean form and simplifies the leptonic trace. The proton mass, by contrast, must be kept exactly — it sets the recoil, the energy loss, and the size of the magnetic term. One particle is effectively massless, the other decidedly not.