Probing the structure of the proton

Fire an electron at a proton and watch how it bounces, and you are doing one of the most productive experiments in the history of hadron physics. The electron is, as far as anyone has measured, a true point — no size, no substructure — and it interacts through one clean, fully-understood channel: it exchanges a virtual photon. So every deviation of the scattering pattern from what a structureless target would give tells you about the other particle. The electron is the ruler; the proton is the thing being measured. What makes the experiment so powerful is that the ruler has an adjustable scale. A scattering probe can only resolve structure on scales comparable to or smaller than its own wavelength, and the relevant wavelength here is not the beam's — it is the reduced wavelength of the exchanged virtual photon, set by the four-momentum transfer $Q$ . So a single apparatus, run across a range of $Q$ , sweeps the proton through a sequence of guises: at low $Q$ an effectively point-like charge obeying Mott scattering; at intermediate $Q$ an extended charge cloud whose finite size shows up as a form factor that suppresses the cross section; and at high $Q$ a swarm of nearly free, point-like quarks, each of which the photon strikes individually — the regime of deep inelastic scattering. This article builds the machinery connecting the data to the structure: the resolution criterion, the momentum transfer, the Mott baseline, the form factor and the charge radius it encodes, and the variable that separates "proton stays whole" from "proton shatters."

Key takeaways

Electron–proton scattering is the cleanest probe of proton structure because the electron is point-like and interacts through a single, well-understood virtual photon.
The resolving power is set by the momentum transfer $Q$ , not the beam energy: the photon's reduced wavelength is $\hbar c/Q$ , and it resolves a scale $r$ only when $Q\gtrsim\hbar c/r$ ( $\hbar c = 0.197\ \mathrm{GeV\,fm}$ ).
At low $Q$ the proton looks point-like and follows the Mott cross section; a finite charge cloud multiplies it by a form factor $|F(Q^2)|^2<1$ .
The form factor is the Fourier transform of the charge distribution, so its slope at $Q^2\to 0$ gives the charge radius: $\langle r^2\rangle = -6\,dF/dQ^2$ , with $r_p\approx 0.84\ \mathrm{fm}$ .
At high $Q$ the proton breaks up; the process becomes elastic scattering off individual quarks, separated from elastic scattering by the invariant mass $W$ and Bjorken $x=Q^2/2M\nu$ (elastic is $x=1$ ).
Key idea: a scattering probe only resolves structure on scales comparable to or smaller than its wavelength — so by tuning the virtual photon's wavelength (via the momentum transfer $Q$ ) you progressively reveal the proton as a point, then an extended charge, and finally a collection of quarks.

The core insight

A scattering probe only resolves structure on scales comparable to or smaller than its wavelength — so by tuning the virtual photon's wavelength (via the momentum transfer $Q$ ) you progressively reveal the proton as a point, then an extended charge cloud (measured by its form factor and charge radius), and finally a collection of point-like quarks.

Where this lives today

The program Hofstadter began is more active now than ever, because the proton's structure has proved surprisingly hard to pin down. The most visible modern chapter is the proton radius puzzle: for a decade the charge radius from electron scattering and ordinary hydrogen spectroscopy (about $0.877\ \mathrm{fm}$ ) disagreed with the far more precise muonic-hydrogen value ( $0.8409\ \mathrm{fm}$ ) by seven standard deviations, forcing a reexamination of both the atomic physics and the low- $Q^2$ extrapolation of the form-factor slope. Dedicated low- $Q^2$ experiments — notably PRad at Jefferson Lab, reading the slope almost directly — now favour the smaller radius, and the CODATA value sits near $0.84\ \mathrm{fm}$ . Jefferson Lab's CEBAF maps $G_E$ and $G_M$ to high $Q^2$ , where polarization-transfer measurements overturned the old assumption that both share the dipole shape. And the next leap is the Electron–Ion Collider under construction at Brookhaven: a high-luminosity polarized $ep$ machine built not merely to measure the proton's size but to produce a three-dimensional tomographic image of how quark and gluon momentum, spin, and position are distributed inside it — the direct descendant of Hofstadter's first form-factor measurement.

The physical picture

Picture the electron not as a little ball but as a wave, and the interaction as the exchange of one virtual photon carrying four-momentum $q$ . The photon's spacelike momentum transfer $Q=\sqrt{-q^2}$ fixes its reduced wavelength, $\hbar c/Q$ , and that wavelength is the smallest feature the collision can resolve. If it dwarfs the proton, the photon couples to the total charge and nothing else — the proton is a point. As $Q$ climbs and the wavelength drops through a femtometre, the photon feels the charge spread out and the scattering falls below the point-like prediction. Push $Q$ past a couple of GeV and the wavelength shrinks below the proton itself: now the photon no longer sees "a proton" but resolves the individual quarks and scatters off one at a time. The first figure is that ladder of resolution, with the numbers attached.

Figure

Three guises of the proton set by the momentum transfer Q through the reduced wavelength ƛ = ħc/Q = 0.197 GeV·fm / Q. At Q ≈ 0.05 GeV the wavelength 3.9 fm dwarfs the 0.84 fm proton, which looks point-like; at Q ≈ 0.5 GeV the wavelength 0.39 fm resolves the finite charge cloud; at Q ≈ 5 GeV the wavelength 0.04 fm resolves individual quarks.

Now make it quantitative. At intermediate $Q$ , where the proton stays whole but its finite size is felt, the whole effect of structure is packaged into one number: the electric form factor $G_E(Q^2)$ , the fraction by which the measured amplitude falls short of the point-like Mott value. Measured across many $Q^2$ , $G_E$ traces a smooth falling curve — well fit by the dipole $G_E=(1+Q^2/0.71\ \mathrm{GeV}^2)^{-2}$ — whose slope at the origin is the proton's mean-square charge radius, the quantity an experimenter actually plots.

Figure

Proton electric form factor G_E versus momentum transfer Q², fit by the dipole G_E = (1 + Q²/0.71 GeV²)⁻². G_E(0) = 1 fixes the total charge; the origin slope dG_E/dQ² = −2/0.71 GeV⁻² gives ⟨r²⟩ = 12/0.71 GeV² = 0.66 fm², a charge radius r_p = 0.81 fm. The dashed line is the origin tangent whose slope encodes the radius.

The core equations

The starting point is the resolution criterion, which turns "wavelength versus proton radius" into an inequality on the momentum transfer. The exchanged virtual photon carries four-momentum $q$ ; its spacelike magnitude $Q=\sqrt{-q^2}$ sets a reduced wavelength, and a probe resolves a scale only once its wavelength has shrunk to that scale.

\delta \sim \frac{\hbar c}{Q}, \qquad Q \gtrsim \frac{\hbar c}{r}\ \ \Rightarrow\ \ \text{structure of size } r \text{ resolved} \qquad (1)

where: $\delta$ — smallest resolvable scale; $Q=\sqrt{-q^2}$ — momentum transfer (GeV); $\hbar c = 0.1973\ \mathrm{GeV\,fm}$ ; $r$ — size being probed. Reading it: to resolve the $0.84\ \mathrm{fm}$ proton needs $Q\gtrsim 0.24\ \mathrm{GeV}$ ; to resolve a $\sim 0.1\ \mathrm{fm}$ quark needs $Q\gtrsim 2\ \mathrm{GeV}$ . The momentum transfer is not a free knob — it is fixed by the beam energy, the observation angle, and the elastic recoil of the proton. Squaring $q=k-k'$ gives $Q^2$ (electron mass neglected), and energy–momentum conservation for an intact proton of mass $M$ fixes the scattered energy $E'$ .

Q^2 \equiv -q^2 = 4EE'\sin^2\tfrac{\theta}{2}, \qquad E' = \frac{E}{\,1+\dfrac{2E}{M}\sin^2\frac{\theta}{2}\,} \qquad (2)

where: $E,E'$ — incoming and scattered electron energies; $\theta$ — scattering angle; $M=0.938\ \mathrm{GeV}$ . Reading it: forward scattering ( $\theta\to0$ ) resolves nothing however large $E$ is — large angle and large $E$ are both needed for large $Q^2$ . The baseline against which structure is measured is the cross section for a structureless target — a point charge struck by a relativistic spin- $\tfrac12$ electron. This is the Mott cross section, Rutherford corrected by the $\cos^2\tfrac{\theta}{2}$ factor that electron helicity conservation demands.

\left(\frac{d\sigma}{d\Omega}\right)_{\!\mathrm{Mott}} = \frac{\alpha^2\cos^2\frac{\theta}{2}}{4E^2\sin^4\frac{\theta}{2}} \qquad (3)

where: $\alpha=1/137$ ; the $1/\sin^4\frac{\theta}{2}$ is the Rutherford tail; $\cos^2\frac{\theta}{2}$ suppresses backscattering because a relativistic electron cannot flip its helicity. This is what a point-like proton would give. A real extended charge distribution scatters less than a point at each $Q^2$ , and the entire effect of its shape is one multiplicative factor: the form factor, the Fourier transform of the normalized charge density $\rho(\mathbf r)$ .

\frac{d\sigma}{d\Omega} = \left(\frac{d\sigma}{d\Omega}\right)_{\!\mathrm{Mott}}\big|F(Q^2)\big|^2, \qquad F(\mathbf q)=\int \rho(\mathbf r)\,e^{\,i\mathbf q\cdot\mathbf r}\,d^3r \qquad (4)

where: $F$ — form factor, normalized to $F(0)=\int\rho\,d^3r=1$ (total charge); $\mathbf q$ — three-momentum transfer, $|\mathbf q|^2\simeq Q^2$ . Reading it: divide the measured $d\sigma/d\Omega$ by Mott and take the square root to get $F$ — a direct image of the proton in momentum space. Because $F$ is a Fourier transform, its behaviour near the origin encodes the distribution's size: expanding the exponential in equation (4) for small $Q$ isolates the mean-square radius.

F(Q^2)\simeq 1-\tfrac16 Q^2\langle r^2\rangle+\cdots, \qquad \langle r^2\rangle = -6\left.\frac{dF}{dQ^2}\right|_{Q^2=0} \qquad (5)

where: $\langle r^2\rangle$ — mean-square charge radius; RMS radius $r_p=\sqrt{\langle r^2\rangle}\approx 0.84\ \mathrm{fm}$ . Reading it: you need not reach high $Q^2$ to measure the proton's size — the slope of $F$ as $Q^2\to0$ is enough. Finally, when $Q^2$ is large enough to shatter the proton, the elastic picture breaks down and a second invariant is needed: the mass $W$ of the recoiling hadronic system, and with it the Bjorken variable $x$ labelling how deeply inelastic the collision is.

W^2=(p+q)^2=M^2+2M\nu-Q^2, \qquad x=\frac{Q^2}{2M\nu}, \qquad \nu=E-E' \qquad (6)

where: $\nu$ — energy transferred to the proton (lab); $W$ — invariant mass of the final hadronic state; $x\in(0,1]$ . Reading it: elastic scattering keeps the proton intact, $W=M$ , forcing $2M\nu=Q^2$ and $x=1$ ; any $x<1$ means the proton fragmented, and in the parton model $x$ is the struck quark's momentum fraction.

Derivation AScattering off a charge cloud and the birth of the form factor

The goal is to show, from first principles in the Born approximation, that scattering an electron off a static extended charge distribution factorizes exactly into the point-like (Mott) cross section times the squared modulus of the Fourier transform of the charge density, and then to extract the leading measure of the proton's size — the mean-square charge radius. This is the derivation that converts a scattering measurement into a picture of where the charge sits.

Step 1. In the first Born approximation the amplitude off a static potential $V(\mathbf r)$ is proportional to its Fourier transform, $f(\mathbf q)\propto\int V(\mathbf r)\,e^{i\mathbf q\cdot\mathbf r}\,d^3r$ , at momentum transfer $\mathbf q=\mathbf k-\mathbf k'$ . For a point charge the Coulomb potential reproduces the Rutherford/Mott amplitude. Why: the Born approximation says the wave reads out the spatial Fourier components of whatever it scatters from, so momentum transfer and spatial structure are conjugate from the start.

Step 2. Replace the point charge by an extended density: $V(\mathbf r)$ is now the potential of a cloud $\rho$ , the convolution of the Coulomb kernel with $\rho$ . The Fourier transform of a convolution is a product, so the amplitude factorizes into the point-charge amplitude times $F(\mathbf q)\equiv\int\rho(\mathbf r)\,e^{i\mathbf q\cdot\mathbf r}\,d^3r$ . Why: convolution in position space is multiplication in momentum space — this is why structure appears as a clean multiplicative factor rather than tangling with the Coulomb dynamics.

Step 3. Square to get equation (4): $d\sigma/d\Omega=(d\sigma/d\Omega)_{\mathrm{Mott}}\,|F(Q^2)|^2$ . With $\rho$ normalized to unit charge, $F(0)=1$ , so as $Q\to0$ the extended target scatters exactly like a point of the same charge — you cannot see structure you cannot resolve. Why: a long-wavelength photon counts only the total charge, recovering the point-like limit automatically.

Step 4. Expand for small $Q$ . Taking $\rho$ spherically symmetric and averaging $e^{i\mathbf q\cdot\mathbf r}=1+i\mathbf q\cdot\mathbf r-\tfrac12(\mathbf q\cdot\mathbf r)^2+\cdots$ over angles kills the linear term and gives $\langle(\mathbf q\cdot\mathbf r)^2\rangle=\tfrac13 Q^2\langle r^2\rangle$ , so $F(Q^2)=1-\tfrac16 Q^2\langle r^2\rangle+\cdots$ and $\langle r^2\rangle=-6\,dF/dQ^2|_0$ — equation (5). Why: the first non-trivial term in the multipole expansion is the mean-square radius, so the initial slope of the measured form factor is a direct, model-independent readout of the proton's size. *As a consistency check: $F(0)=1$ ; $\langle r^2\rangle=-6\,dF/dQ^2|_0$ ; and for the dipole $G_E=(1+Q^2/0.71\,\mathrm{GeV}^2)^{-2}$ this gives $\langle r^2\rangle=12/0.71\,\mathrm{GeV}^2=0.66\,\mathrm{fm}^2$ , $r_p=0.81\,\mathrm{fm}$ .*

Go deeper · Derivation B · 1 minMomentum transfer, not beam energy, sets the resolution

Derivation BMomentum transfer, not beam energy, sets the resolution

The aim is to derive the exact kinematics of elastic electron–proton scattering — the momentum transfer $Q^2$ in terms of beam energy and scattering angle, and the recoil relation that fixes $E'$ — and thereby make precise the claim that resolving power is controlled by $Q$ , not by beam energy alone. This is what tells an experimenter which energy and angle to choose to see the proton at a given scale.

Step 1. Write $k=(E,\mathbf k)$ , $k'=(E',\mathbf k')$ , neglecting the electron mass so $|\mathbf k|=E$ , $|\mathbf k'|=E'$ . The momentum transfer $q=k-k'$ has invariant square $q^2=-2k\cdot k'=-2EE'(1-\cos\theta)$ . Why: $Q^2=-q^2$ is a Lorentz invariant, so it fixes the resolution identically in every frame, unlike the frame-dependent beam energy.

Step 2. Using $1-\cos\theta=2\sin^2\tfrac{\theta}{2}$ gives $Q^2=4EE'\sin^2\tfrac{\theta}{2}$ , equation (2). At fixed $E$ this vanishes as $\theta\to0$ however large $E$ is: a high-energy beam grazing past the proton resolves nothing. Why: energy buys resolution only when actually delivered as momentum transfer, i.e. at appreciable angle.

Step 3. Impose conservation for an intact proton, $k+p=k'+p'$ with $p=(M,\mathbf 0)$ . Squaring $p'=p+q$ with $p'^2=M^2$ gives $2p\cdot q+q^2=0$ , i.e. $2M(E-E')=Q^2$ ; substituting $Q^2=4EE'\sin^2\tfrac{\theta}{2}$ gives $E'=E/[1+(2E/M)\sin^2\tfrac{\theta}{2}]$ . Why: the proton's finite mass shares energy with the electron — the heavier the target relative to $E$ , the less it recoils and the closer $E'$ stays to $E$ .

Step 4. Read off the elastic constraint. For elastic scattering $E'$ is locked to $\theta$ by the recoil relation, so at each angle the elastic events sit at one sharp energy (the "elastic peak"), equivalently $2M\nu=Q^2$ , i.e. $x=1$ . Inelastic events, where the proton breaks, populate lower $E'$ (larger $\nu$ ) and hence $x<1$ . Why: $2M\nu=Q^2$ is the boundary between "proton stays whole" and "proton shatters," and it is what a spectrometer scan across $E'$ maps out. *As a consistency check: $Q^2=4EE'\sin^2\tfrac{\theta}{2}$ and recoil $E'=E/[1+(2E/M)\sin^2\tfrac{\theta}{2}]$ ; for $E=0.5\,\mathrm{GeV}$ , $\theta=60°$ these give $E'=0.395\,\mathrm{GeV}$ , $Q^2=0.197\,\mathrm{GeV}^2$ , $Q=0.444\,\mathrm{GeV}$ .*

Go deeper · Derivation C · 1 minFrom the elastic peak to deep inelastic scattering

Derivation CFrom the elastic peak to deep inelastic scattering

The purpose here is to introduce the invariant mass $W$ of the hadronic final state and the Bjorken variable $x$ , and to show how these two numbers organize the whole electron–proton landscape into a single elastic line, a band of nucleon resonances, and a smooth deep-inelastic continuum in which the proton dissolves into quarks. This is the bridge from "measuring the proton's size" to "discovering its constituents."

Step 1. The final hadronic system has total four-momentum $p+q$ and invariant mass squared $W^2=(p+q)^2=M^2+2p\cdot q+q^2$ . In the lab $p\cdot q=M\nu$ with $\nu=E-E'$ , so $W^2=M^2+2M\nu-Q^2$ — equation (6). Why: $W$ is the mass of whatever the proton turned into — the true label of inelasticity, fixed by the kinematics even though the electron never sees it.

Step 2. Set $W=M$ for elastic scattering (proton intact and unexcited). Then $2M\nu=Q^2$ , a curve in the $(\nu,Q^2)$ plane on which every elastic event lies — the same locking of $E'$ to $\theta$ found before. Why: elastic scattering has one fewer free variable, so it collapses to a line — which is why the elastic form factor depends on the single variable $Q^2$ .

Step 3. Form $x=Q^2/2M\nu=Q^2/(W^2-M^2+Q^2)$ . Elastic is $x=1$ ; raising $W$ above $M$ (exciting the $\Delta(1232)$ , then the continuum) drives $x$ below 1. The deep-inelastic region is large $Q^2$ and large $\nu$ with $x$ finite. Why: $x$ compresses the two invariants into the one number with a partonic meaning, along which the messy inelastic data organize themselves.

Step 4. Interpret $x$ in the parton model (recalled from the quark–parton model that follows). If the photon scatters elastically off a single quark carrying fraction $\xi$ of the proton's momentum, the same $2(\xi p)\cdot q=Q^2$ algebra gives $\xi=Q^2/2M\nu=x$ . So Bjorken $x$ is literally the struck quark's momentum fraction. Why: the variable that merely bookkept inelasticity turns out to measure which quark was hit — why deep inelastic scattering images the quark momentum distribution. *As a consistency check: $W^2=M^2+2M\nu-Q^2$ ; elastic $W=M\Rightarrow x=1$ ; and $x=Q^2/2M\nu$ equals the struck-quark momentum fraction in the parton model.*

Limiting cases

Case	Result	Meaning
Long wavelength, $Q\to 0$	$F(Q^2)\to 1$	photon resolves only total charge; proton is effectively point-like, pure Mott scattering
Non-relativistic, static	Rutherford $1/\sin^4\frac{\theta}{2}$	recover classical point-nucleus scattering; no spin, no recoil, no structure
Intermediate $Q$ , elastic	$d\sigma/d\Omega=(d\sigma/d\Omega)_{\mathrm{Mott}}\|G_E(Q^2)\|^2$	finite charge cloud; form factor falls as dipole, slope gives $r_p\approx0.84$ fm
Large $Q$ , elastic	$G_E\to 0$	proton cannot recoil coherently as a whole; elastic rate collapses, inelastic takes over
Deep inelastic, $x$ fixed	scaling: $F_2(x,Q^2)\to F_2(x)$	photon strikes point-like quarks; $Q^2$ -independence signals structureless constituents

Common trapA higher-energy beam automatically gives higher resolution

It does not. What resolves structure is the four-momentum transfer $Q$ , through the reduced wavelength $\hbar c/Q$ — and $Q^2=4EE'\sin^2\tfrac{\theta}{2}$ depends on the angle as much as the energy. A multi-GeV electron skimming past at a tiny angle transfers almost no momentum ( $Q\to0$ as $\theta\to0$ ) and sees nothing but the total charge, however energetic it is. Resolution is bought only when energy is actually delivered as momentum transfer, i.e. at appreciable angle. This is why experiments are quoted at a given $Q^2$ , never a given beam energy, and why the same beam probes many scales at once — one per detector angle. The corollary: the relevant wavelength is the exchanged virtual photon's ( $\hbar c/Q$ ), not the beam electron's de Broglie wavelength.

Worked examples

1 · What momentum transfer do you need? (confidence builder)

The proton has a charge radius $r_p\approx 0.84\ \mathrm{fm}$ , and its quarks live on a scale of roughly $0.1\ \mathrm{fm}$ . What momentum transfer $Q$ is needed to resolve each? (Use $\hbar c = 0.197\ \mathrm{GeV\,fm}$ .)

Step 1. Resolution criterion, equation (1): to resolve length $r$ needs $Q\gtrsim\hbar c/r$ . For the whole proton, $Q\gtrsim 0.197/0.84 = \mathbf{0.23\ GeV}$ . Below this the photon's wavelength exceeds the proton and it looks point-like.

Step 2. For the quark scale, $Q\gtrsim 0.197/0.1 = \mathbf{2\ GeV}$ — why "seeing quarks" needed the multi-GeV beams of SLAC in the late 1960s, not the few-hundred-MeV machines that first measured the proton's size.

Step 3. The required $Q$ scales inversely with feature size, so each factor of ten in spatial resolution costs a factor of ten in momentum transfer — and, through $Q^2=4EE'\sin^2\tfrac{\theta}{2}$ , far more in beam energy and luminosity. To resolve the 0.84 fm proton needs Q ≳ 0.23 GeV; to resolve its 0.1 fm quarks needs Q ≳ 2 GeV.

2 · Elastic kinematics and the form-factor suppression (fully worked)

A $0.5\ \mathrm{GeV}$ electron scatters elastically off a proton at $\theta=60°$ . Find $E'$ , the momentum transfer $Q^2$ and $Q$ , the wavelength probed, and — using the dipole $G_E=(1+Q^2/0.71\ \mathrm{GeV}^2)^{-2}$ — the factor by which structure suppresses the cross section below Mott. Take $M=0.938\ \mathrm{GeV}$ .

Step 1. Recoil, equation (2): $\sin^2\tfrac{\theta}{2}=\sin^2 30°=0.25$ , and $2E/M=1.0/0.938=1.066$ , so the denominator is $1+1.066\times0.25=1.267$ . Thus $E'=0.5/1.267=\mathbf{0.395\ GeV}$ — the proton has carried off about a fifth of the electron's energy.

Step 2. Momentum transfer: $Q^2=4EE'\sin^2\tfrac{\theta}{2}=4\times0.5\times0.395\times0.25=\mathbf{0.197\ GeV^2}$ , so $Q=\sqrt{0.197}=\mathbf{0.444\ GeV}$ .

Step 3. Reduced wavelength probed: $\hbar c/Q=0.197/0.444=\mathbf{0.44\ fm}$ , about half the proton radius — squarely in the regime where finite size matters, so expect a significant form-factor suppression.

Step 4. Form factor: $G_E=(1+0.197/0.71)^{-2}=(1.278)^{-2}=\mathbf{0.61}$ . The cross section carries $|G_E|^2$ , so the measured elastic rate is $0.61^2=\mathbf{0.37}$ of the point-like Mott prediction. The finite proton scatters at only 37% of the point-charge rate at this $Q^2$ — the missing 63% is the experimental signature of structure.

3 · The charge radius from the form-factor slope

Extract the proton RMS charge radius from the dipole $G_E(Q^2)=(1+Q^2/M_D^2)^{-2}$ , $M_D^2=0.71\ \mathrm{GeV}^2$ .

Step 1. Differentiate at the origin: $dG_E/dQ^2|_0=-2/M_D^2$ , so by equation (5) $\langle r^2\rangle=-6\,dG_E/dQ^2|_0=12/M_D^2=12/0.71=16.9\ \mathrm{GeV}^{-2}$ .

Step 2. Convert with $(\hbar c)^2=0.0389\ \mathrm{GeV^2\,fm^2}$ : $\langle r^2\rangle=16.9\times0.0389=\mathbf{0.66\ fm^2}$ , so $r_p=\sqrt{0.66}=\mathbf{0.81\ fm}$ — within a few percent of the modern $0.84\ \mathrm{fm}$ . The small gap is why the dipole is only an approximation and the precise radius is still refined today. The dipole slope alone reproduces the proton charge radius to better than 5%.

Physics of the portrait

Probing the proton with electrons is inseparable from the experimentalist who turned the idea into a Nobel-winning program: Robert Hofstadter, who first measured that the proton has a size at all.

Origins · 1 minRobert Hofstadter and how this was discovered

Robert Hofstadter

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

Robert Hofstadter built, at Stanford's Mark III linear accelerator in the early 1950s, the first electron beams energetic enough to resolve nuclei and then the proton, founding the field of electron-scattering structure physics. Where Rutherford had used slow, low-resolution alpha particles to find the nucleus, Hofstadter used fast electrons — clean, point-like electromagnetic probes — and pushed the momentum transfer high enough that the scattering fell measurably below the point-charge prediction. In 1955, with Robert McAllister, he showed the proton is not a point: the elastic cross section required a form factor whose slope gave a finite RMS charge radius of about $0.7$ – $0.8\ \mathrm{fm}$ . It was the first direct evidence that the proton is an extended object, and it introduced the form factor as the working language of nucleon structure. Hofstadter shared the 1961 Nobel Prize. His electrons, sharpened into the multi-GeV beams of SLAC, would later break the proton open and reveal the quarks — but it was Hofstadter who first proved there was something inside to find.

History

1911

— Rutherford infers the point-like nucleus from large-angle alpha scattering

1933

— Stern measures the proton's anomalous magnetic moment, a first hint it is not a Dirac point

1955

— Hofstadter and McAllister measure the proton form factor and a finite charge radius at Stanford

1961

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

1969

— SLAC–MIT deep inelastic scattering reveals point-like partons inside the proton

Going further

Texts. Thomson, Modern Particle Physics, Ch. 7 (electron–proton elastic scattering) and Ch. 8 (deep inelastic scattering) · Halzen & Martin, Quarks and Leptons, Ch. 8 · Griffiths, Introduction to Elementary Particles, 2nd ed., §8.3 · Perkins, Introduction to High Energy Physics, Ch. 6 · Peskin & Schroeder, An Introduction to Quantum Field Theory, §17 for the deep-inelastic continuation. The key results developed in this article are: (1) $Q^2=4EE'\sin^2\tfrac{\theta}{2}$ and recoil $E'=E/[1+(2E/M)\sin^2\tfrac{\theta}{2}]$ ; (2) $F(0)=1$ and $\langle r^2\rangle=-6\,dF/dQ^2|_0$ ; (3) the dipole $\langle r^2\rangle=12/0.71\,\mathrm{GeV}^2=0.66\,\mathrm{fm}^2$ , $r_p=0.81\,\mathrm{fm}$ ; (4) the worked-example values $E'=0.395$ GeV, $Q^2=0.197$ GeV², $G_E=0.61$ .

Prerequisites. Rutherford and Mott Scattering · QED Feynman Rules · Interaction Cross Sections. Leads to. Form Factors · Relativistic Electron–Proton Scattering · Deep Inelastic Scattering.

Ask the tutor: Want to see how the single form factor $F(Q^2)$ splits into the electric $G_E$ and magnetic $G_M$ pieces of the Rosenbluth formula, or how pushing $Q^2$ higher turns this elastic story into deep inelastic scattering and the discovery of quarks?

Is it the electron's wavelength or the photon's that sets the resolution?

It is the exchanged virtual photon's, and the distinction matters. Naively "a fast electron has a short de Broglie wavelength and resolves small things," but the interaction happens through the photon the electron exchanges, and it is that photon's reduced wavelength $\hbar c/Q$ — fixed by the momentum transfer $Q$ , not the beam momentum — that acts as the resolving scale. The same beam scattered at different angles delivers different $Q$ and probes different sizes at once. So results are always quoted at a stated $Q^2$ , and a high-energy beam grazing at small angle can resolve less than a modest beam scattered hard through a large angle.

What is the difference between elastic and inelastic scattering here?

Physically the two are the same process — the electron scatters electromagnetically by exchanging a single virtual photon — and what changes is only what happens to the proton. In elastic scattering the proton absorbs the photon coherently and stays whole, so its final state has the proton mass, $W=M$ , and the kinematics collapses to one variable ( $Q^2$ ) with the events pinned to the elastic peak, $x=1$ . In inelastic scattering the photon delivers enough energy and momentum to excite or shatter the proton, so the final hadronic mass $W$ exceeds $M$ and $x<1$ . It is a common error to think elastic and inelastic involve different forces; they do not. Only the coherence of the proton's response differs.

Why is the Mott cross section the right baseline instead of Rutherford?

Rutherford scattering describes a non-relativistic, spinless point charge and gives the famous $1/\sin^4\tfrac{\theta}{2}$ dependence. But the probe electrons are highly relativistic, and a relativistic spin- $\tfrac12$ particle cannot flip its helicity, which forbids exact backscattering and multiplies Rutherford by $\cos^2\tfrac{\theta}{2}$ — that is Mott. Using Mott as the point-like reference means $|F(Q^2)|^2$ isolates only the proton's spatial structure, with the electron's own spin dynamics already accounted for. Dividing by Rutherford instead would misattribute the spin suppression to proton structure.

How can measuring a slope at Q²→0 give a size without reaching small distances?

Because the form factor is a Fourier transform, and a Fourier transform ties large-scale (low- $Q$ ) behaviour to the overall extent of the distribution rather than to fine detail. Expanding $F(Q^2)=1-\tfrac16 Q^2\langle r^2\rangle+\cdots$ , the very first correction to the point-like value is proportional to the mean-square radius, so the initial slope of $F$ already carries the size. You do not need to resolve the internal texture (that lives in the high- $Q^2$ tail); you only need to measure how quickly the cross section starts to fall below Mott. This is why the proton radius can be extracted from relatively low- $Q^2$ data — and also why getting that slope exactly right, by extrapolating to $Q^2=0$ , is at the heart of the proton radius puzzle.

What is Bjorken x really telling me?

Kinematically, $x=Q^2/2M\nu$ is just a dimensionless combination of the momentum transfer and the energy transfer $\nu$ , running from $x=1$ for elastic scattering toward zero as the collision grows more inelastic. Its deeper meaning is in the parton model: if the photon scatters elastically off a single quark carrying fraction $\xi$ of the proton's momentum, the same algebra that gives $x=1$ for the whole proton gives $\xi=x$ for the quark. So $x$ is the struck constituent's momentum fraction, and the cross section versus $x$ maps how the proton's momentum is shared among its quarks — the parton distribution functions.

Why does the proton have a form factor but the electron does not?

Because the electron is, to the limit of every experiment, genuinely point-like: its charge is not spread over any measurable volume, so $F(Q^2)=1$ at all accessible $Q^2$ and scattering off it follows the point-like prediction exactly. The proton, by contrast, is a bound state of quarks and gluons; its charge is smeared over a region of order a femtometre, and the virtual photon's interference with that extended distribution produces the falling $F(Q^2)$ . A form factor is the experimental signature of compositeness — and the electron having none is precisely why it makes such a clean ruler for the proton that does.