Classical Electron Theory and Self-Energy

Between 1881 and 1906, some of the best physicists alive tried to build the electron out of pure electromagnetism — mass as nothing but the inertia of a charge's own field. The attempt failed, and the way it failed is permanent physics. This article works the three classic calculations honestly and to the end: the self-energy of a charged shell, which diverges as the radius shrinks and defines the classical electron radius $r_e = 2.818$ fm; the field momentum of the moving shell, which comes out $\tfrac{4}{3}(U/c^2)\mathbf{v}$ — the notorious 4/3 problem, an apparent violation of $E = mc^2$ sitting in elementary electrostatics; and the Poincaré stresses, the non-electromagnetic cohesion that a purely electromagnetic electron provably needs, which — computed from nothing but mechanical equilibrium — repairs the 4/3 exactly. The modern payoff is the diagnosis. The 4/3 is not an algebra slip: it is a theorem about open systems. The electromagnetic field of a charged shell is not closed — something non-Maxwellian must hold the charge together — and the energy-momentum of an open subsystem simply is not a four-vector; the covariant conservation article's fine print, cashed here as the root cause. Add the minimal closing stress and the books balance to the digit: the tension required for equilibrium contributes exactly $-U/3$ where the field's stress integral contributes $+U/3$ . The verdict band is equally honest: scattering finds the electron pointlike below $10^{-19}$ m, where the classical model demands a mass of 7 GeV; QED softens the linear divergence to a logarithm and buries it in mass renormalization — the classical theory's renormalization move, industrialized. The classical electron died; its bookkeeping runs modern physics.

Key takeaways

The electrostatic self-energy of a charged shell is $U_{\rm em} = q^2/8\pi\varepsilon_0 a$ — the assembly work — diverging linearly as $a \to 0$ : a point charge costs infinite energy.
Setting $e^2/4\pi\varepsilon_0 a \sim m_ec^2$ defines the classical electron radius $r_e = 2.818$ fm $= \alpha\,\lambdabar_C = \alpha^2 a_0$ — a bookkeeping length (the Thomson cross-section scale), not a measured size.
The field momentum of the slowly moving shell, computed honestly, is $\mathbf{p}_{\rm em} = \tfrac{4}{3}(U_{\rm em}/c^2)\mathbf{v}$ — the 4/3 problem: energy account says $m = U/c^2$ , momentum account says $\tfrac43 U/c^2$ .
Root cause (a theorem, not a bug): the electromagnetic field of a charged body is an open system — its energy-momentum is not a four-vector; the mismatch is the slice-dependence the covariant conservation article warned about.
Poincaré stresses: mechanical equilibrium of the shell requires an inward tension $p = -\sigma^2/2\varepsilon_0 = -q^2/32\pi^2\varepsilon_0 a^4$ ; its stress integral is exactly $-U/3$ , cancelling the field's $+U/3$ and restoring $\mathbf{p} = (U_{\rm tot}/c^2)\mathbf{v}$ — von Laue's theorem in action.
The classical model dies on data: the electron is pointlike below $10^{-19}$ m, where the model predicts $m_{\rm em} \approx 14{,}000\,m_e$ ; QED's self-energy diverges only logarithmically ( $\delta m/m \approx 0.18$ even with a Planck-scale cutoff) and is renormalized away.
Key idea: treating the electron's own field as the source of its inertia produces a divergent self-energy and a self-momentum that violates $E = mc^2$ by a factor 4/3; the consistent classical fix requires non-electromagnetic Poincaré stresses, and the true cure is quantum renormalization.

Where this lives today

The classical electron program is dead; its three moves run the Standard Model. Renormalization — isolate the divergent self-interaction, absorb it into the observed parameter, keep the finite remainder — is the 1904 shell logic industrialized: in QED the electron's self-energy diverges only logarithmically, $\delta m/m \sim \tfrac{3\alpha}{2\pi}\ln(\Lambda/m_ec^2)$ , so mild that even a Planck-scale cutoff gives $\delta m/m \approx 0.18$ (compare the classical linear divergence: the same cutoff demands $\delta m/m \sim 10^{20}$ ). The softening is the Compton-scale smearing of the quantum electron — the ladder (2) again, now with $\hbar$ doing the regulating. Open-system honesty — energy-momentum of a fragment is slice-dependent — recurs wherever anyone partitions a total: "electromagnetic momentum in matter" (the Abraham–Minkowski file), quark contributions to the proton's mass and spin, quasiparticle weights in condensed matter. And the Poincaré stress is the ancestor of every "what binds it" question since — bag-model pressure confining quarks, soliton stability analyses, and an early cameo of negative-pressure physics. Meanwhile $r_e$ itself, stripped of its size pretension, works honest jobs daily: the Thomson cross section $\sigma_T = \tfrac{8\pi}{3}r_e^2 = 0.665$ barn (spent next door in Thomson Scattering), the synchrotron energy-loss coefficients of every storage ring, and the self-force clock $\tau = 2r_e/3c$ of Abraham-Lorentz Radiation Reaction Force. The classical electron left no body, but the estate is enormous.

Origins · 1 minHenri Poincaré and how this was discovered

Henri Poincaré

1854–1912 · Paris — who paid the electron's binding bill and priced covariance exactly

Poincaré enters this track for the third time, wearing a different hat each visit: the 1900 radiation-recoil argument that made field momentum undeniable, and the 1905 mathematics of the Lorentz group, are told elsewhere. This article's Poincaré is the author of "Sur la dynamique de l'électron" (announced 1905, published in full 1906) — the paper that, alongside its celebrated group theory, confronted the mechanical scandal of Lorentz's electron head-on: a surface of charge cannot hold itself together, and whatever holds it must enter the energy-momentum books. His supplementary stresses — a negative internal pressure, tuned by nothing, fixed entirely by equilibrium — were introduced to make the deformable electron dynamically consistent; that the same stresses repair the 4/3 transformation defect exactly (Derivation C) is the kind of unforced coincidence Poincaré trusted as a sign of correct structure. Writing at relativity's birth, he supplied the first worked example of a lesson field theory spent the century relearning — you cannot weigh a piece of a system and expect four-vector answers — and the first respectable negative-pressure medium in physics. He died in 1912, before the quantum theory that retired his electron model while keeping its bookkeeping discipline intact.

History

1881

— J.J. Thomson computes the extra inertia of a moving charged sphere: electromagnetic mass is born

1897

— Thomson identifies the electron; the theory now has its object

1902

— Abraham builds the rigid-electron dynamics and its field momentum, flying the all-mass-is-electromagnetic flag

1904

— Lorentz's deformable electron: the shell expansion, the divergent inertia absorbed — renormalization in embryo

1906

— Poincaré's stresses close the system: statics and covariance repaired by one tension

1911

— von Laue's theorem: closed static systems need vanishing stress integrals — the 4/3 diagnosed in general

1922

— Fermi shows a covariant slice definition dissolves the 4/3 without new stresses: closure as bookkeeping

1948

— QED renormalization (Bethe, Schwinger, Feynman, Tomonaga): the classical subtraction becomes precision physics

The physical picture

Assemble an electron the way you assemble any charge configuration: carry infinitesimal bits of charge in from infinity against their mutual repulsion. The bill — Work and Energy of a Charge Configuration owns the machinery — grows as the target shrinks, because the last bits arrive against the full Coulomb wall: $U \propto 1/a$ . If the field energy is the mass, a smaller electron is a heavier electron, without bound. The classical program hoped to run this logic forward: pick the radius where the field energy equals $m_ec^2$ , and declare mass explained. That radius is $r_e \approx 2.8$ fm — and the hope was not crazy: the same length really does govern how electrons scatter light (Thomson Scattering, next door, spends it as $\sigma_T = \tfrac{8\pi}{3}r_e^2$ ). The program dies on a bookkeeping contradiction. Weigh the electron's field two ways. Energy account: the rest-frame field carries $U$ , so relativity demands inertia $U/c^2$ . Momentum account: boost the shell to velocity $\mathbf{v}$ and integrate the momentum density $\varepsilon_0\mathbf{E}\times\mathbf{B}$ over all space — the answer is $\tfrac43(U/c^2)\mathbf{v}$ . Two masses, one object, factor $4/3$ between them. The discrepancy is not in the integrals (they are three lines each); it is in the object. A bare charged shell is not a mechanical possibility: its own field tears it apart, so "the shell's field" is a fragment of a system whose other part — whatever binds it — has been silently omitted. And fragments of systems do not obey $E = mc^2$ as four-vector bookkeeping: the total energy-momentum in a piece of a system depends on the simultaneity slice used to add it up — the open-slice honesty that Conservation Laws and Energy-Momentum Flow flagged and deferred here by name. Poincaré's 1906 move is then forced, not clever: add the minimal non-electromagnetic stress that holds the shell in equilibrium. The remarkable part — the part worth doing to the digit, and Derivation C does — is that the same tension required by statics is exactly what covariance requires: no tuning, no second dial. Mechanical consistency and relativistic consistency are one condition.

Figure

The two ways to weigh a field — the energy account and the momentum account disagree by a factor four thirds until the binding agent is put back into the books

The objects: assembly energy, the radius ladder, and the constraint that will save the day

The rest-frame field of the shell (radius $a$ , charge $q$ , surface density $\sigma = q/4\pi a^2$ ) is pure electrostatics: $E = q/4\pi\varepsilon_0 r^2$ outside, zero inside. Its energy — by the energy-density machinery of Energy Density of the Electric Field — is the article's first object,

U_{\rm em} = \frac{\varepsilon_0}{2}\int E^2\,dV = \frac{q^2}{8\pi\varepsilon_0 a}, \qquad (1)

where: the divergence as $a \to 0$ is linear in $1/a$ — the classical theory's signature infinity, far harsher than QED's logarithm (the verdict band prices both). For a solid uniformly charged sphere the coefficient is $\tfrac35 \cdot \tfrac{1}{4\pi\varepsilon_0}$ rather than $\tfrac12$ : an $O(1)$ model dependence that already warns against reading any of these radii literally. The second object is the length this energy mints. Demanding $e^2/4\pi\varepsilon_0 a \sim m_ec^2$ (shape factor dropped) defines

r_e = \frac{e^2}{4\pi\varepsilon_0 m_ec^2} = 2.8179\times10^{-15}\ \text{m}, \qquad r_e = \alpha\,\lambdabar_C = \alpha^2\,a_0, \qquad (2)

where: the ladder on the right — verified to five digits — places $r_e$ at the bottom of the atomic hierarchy: a factor $\alpha \approx 1/137$ below the reduced Compton wavelength $\lambdabar_C = 386$ fm (where QED lives), which sits a factor $\alpha$ below the Bohr radius $a_0 = 52.9$ pm (where chemistry lives). Classical electron theory operates two rungs of $\alpha$ below the physics that actually governs the electron — the geometric reason its infinities were never physical. The third object is a constraint. For any closed static system, special relativity requires the space-space stress integrals to vanish — von Laue's theorem, $\int \theta^{ij}\,d^3x = 0$ — precisely so that the total energy-momentum boosts as a four-vector. The field of the shell alone will fail this test by $+U/3$ ; the failure is the 4/3, and the repair is Derivation C.

Derivation AElectromagnetic mass from the assembly energy

The goal is the energy account done cleanly: the shell self-energy (1), its reading as an electromagnetic mass $m_{\rm em} = U_{\rm em}/c^2$ , the radius ladder (2), and the honest statement of what diverges and what the divergence means. The force-side twin of this bookkeeping — the shell's self-force expansion, whose divergent $-\delta m\,\mathbf{a}$ term is absorbed by mass renormalization — is owned by the E9 article Radiation Reaction Force and recalled, not re-run; here we do the energy side and keep the accounts parallel.

Step 1. Integrate the energy density. Outside the shell $E = q/4\pi\varepsilon_0 r^2$ ; inside, zero (Gauss). So

U_{\rm em} = \frac{\varepsilon_0}{2}\int_a^\infty \left(\frac{q}{4\pi\varepsilon_0 r^2}\right)^2 4\pi r^2\,dr = \frac{q^2}{8\pi\varepsilon_0}\int_a^\infty\frac{dr}{r^2} = \frac{q^2}{8\pi\varepsilon_0 a}.

Why this equals the assembly work: charging the shell from $0$ to $q$ costs $\int_0^q (q'/4\pi\varepsilon_0 a)\,dq' = q^2/8\pi\varepsilon_0 a$ — the two routes (field integral, assembly ledger) agree exactly, as the E1 energy articles guarantee. For the uniform solid sphere the same integral plus the interior contribution gives $\tfrac{3}{5}\,q^2/4\pi\varepsilon_0 a$ : shape moves the coefficient, never the $1/a$ .

Step 2. Read it as mass. If field energy has inertia — and relativity says energy does — the rest-frame field contributes $m_{\rm em} = U_{\rm em}/c^2 = q^2/8\pi\varepsilon_0 ac^2$ . At $a = r_e$ : $U_{\rm em} = 0.2555$ MeV $= m_ec^2/2$ exactly (the shell's factor $\tfrac12$ ); full mass requires $a = r_e/2 = 1.41$ fm. The program's ambition: choose $a$ once, obtain all the mass, and the electron is field.

Step 3. The ladder and the divergence. Equation (2)'s identities, $r_e = \alpha\lambdabar_C = \alpha^2 a_0$ , are exact consequences of the definitions of $\alpha$ , $\lambdabar_C$ , $a_0$ — verified numerically to five digits. The divergence: as $a \to 0$ , $m_{\rm em} \to \infty$ linearly. The classical options are (i) keep $a$ finite (then explain rigidity and stability — Derivation C's problem), or (ii) absorb the infinity into the observed mass and keep only finite structure-independent physics — the renormalization move, executed on the force side by the E9 shell derivation in 1904 logic, and the direct conceptual ancestor of QED's subtraction. What cannot be done is ignore it: the divergence sits in the same field energy that every capacitor calculation uses. Verified: $U_{\rm em}(r_e) = 0.25553$ MeV vs $m_ec^2/2 = 0.25550$ MeV (CODATA); ladder $\alpha\lambdabar_C = 2.81794\times10^{-15}$ m $= r_e$ to 6 digits; assembly-route equality exact.

Go deeper · Derivation B · 2 minThe field momentum of the moving shell — the four-thirds problem computed

Derivation BThe field momentum of the moving shell — the four-thirds problem computed

The goal is the momentum account: $\mathbf{p}_{\rm em} = \tfrac43(U_{\rm em}/c^2)\mathbf{v}$ for the shell in uniform slow motion, every factor shown, followed by the root-cause theorem — the coefficient is a property of open systems, not an algebra artifact. Strategy: build $\mathbf{B}$ at $O(v/c)$ , integrate the momentum density with one angular average, then convert the "paradox" into a statement about stress integrals and simultaneity slices.

Step 1. The fields at first order. A charge in uniform slow motion carries its rest-frame Coulomb field $\mathbf{E}$ essentially undistorted ( $\gamma = 1 + O(v^2/c^2)$ ), plus the boost magnetic field $\mathbf{B} = (\mathbf{v}\times\mathbf{E})/c^2$ — the $O(v/c)$ term of the field-transformation rules owned upstream.

Step 2. Momentum density and the double cross. With $\mathbf{g} = \varepsilon_0\,\mathbf{E}\times\mathbf{B}$ :

\mathbf{g} = \frac{\varepsilon_0}{c^2}\,\mathbf{E}\times(\mathbf{v}\times\mathbf{E}) = \frac{\varepsilon_0}{c^2}\left[E^2\,\mathbf{v} - (\mathbf{E}\cdot\mathbf{v})\,\mathbf{E}\right],

by BAC–CAB. The first term is momentum along $\mathbf{v}$ everywhere; the second subtracts the part of $\mathbf{E}$ 's direction that correlates with $\mathbf{v}$ .

Step 3. The angular average. For the radial field, averaging the dyadic over directions gives $\langle(\mathbf{E}\cdot\mathbf{v})\mathbf{E}\rangle = \tfrac13 E^2\mathbf{v}$ (the $\tfrac13$ is $\langle\cos^2\rangle$ over the sphere — sympy-verified). Hence

\mathbf{p}_{\rm em} = \frac{\varepsilon_0}{c^2}\left(1 - \frac13\right)\mathbf{v}\int E^2\,dV = \frac{2}{3}\frac{\varepsilon_0}{c^2}\,\mathbf{v}\int E^2\,dV = \frac{4}{3}\,\frac{U_{\rm em}}{c^2}\,\mathbf{v},

using $\varepsilon_0\int E^2 dV = 2U_{\rm em}$ from (1). The 4/3 is now on the table: the momentum account weighs the field at $\tfrac43 U/c^2$ while the energy account said $U/c^2$ . The coefficient is model-independent (any spherical distribution gives it), and it is consistent with the force account: the E9 shell derivation found self-inertia $\delta m = \tfrac43 U_{\rm em}/c^2$ in the self-force — same defect, third symptom. One wrong object, three coherent wrongnesses.

Step 4. The root cause, as a theorem. Boost a static system with stress-energy $\theta^{\mu\nu}$ ; adding up energy and momentum on the moving frame's simultaneity slice gives, at $O(v)$ ,

p^x = \frac{v}{c^2}\left[U + \int\theta^{xx}\,d^3x\right].

For a closed static system von Laue's theorem forces $\int\theta^{ij}d^3x = 0$ and $p = (U/c^2)v$ : four-vector behavior. For the shell's field alone: with the sign conventions of The Electromagnetic Stress-Energy Tensor ( $\theta^{ij} = -\sigma^{ij}_{\rm Maxwell}$ ), the radial Coulomb field gives $\int\theta^{xx}_{\rm EM}\,d^3x = \tfrac{\varepsilon_0}{6}\int E^2 dV = +U_{\rm em}/3$ — and the boost formula returns exactly $\tfrac43(U/c^2)v$ , reproducing Step 3 from pure covariance bookkeeping. The 4/3 problem is therefore the statement $\int\theta^{xx}_{\rm EM} \neq 0$ : the field of a charged body is not a closed system. Its energy-momentum depends on the slice used to integrate it — the open-slice honesty Conservation Laws and Energy-Momentum Flow established covariantly and deferred, by name, to this article. Nothing is paradoxical; something is missing. Derivation C supplies it. Verified: angular average $\tfrac13$ symbolic (sympy); numeric audit at $a = r_e$ , $v = 0.01c$ : $p_{\rm em} = 1.821\times10^{-24}$ kg·m/s vs $(U/c^2)v = 1.365\times10^{-24}$ — ratio $1.3333$ ; stress integral $\tfrac{\varepsilon_0}{6}\int E^2 = U/3$ exact from (1); matches Jackson §16.4–16.5.

Go deeper · Derivation C · 2 minPoincare stresses — mechanical equilibrium repairs covariance exactly

Derivation CPoincare stresses — mechanical equilibrium repairs covariance exactly

The goal is the closure: compute the non-electromagnetic stress that a stable charged shell mechanically requires, show its stress integral is exactly $-U_{\rm em}/3$ , and watch the total system pass von Laue's theorem — restoring $\mathbf{p} = (U_{\rm tot}/c^2)\mathbf{v}$ with no tuning. This is the honest core of the article: the fix is not bolted on to save relativity; it is demanded by statics, and covariance follows.

Step 1. The shell bursts without help. The electrostatic stress on a surface charge is an outward pull per unit area $\sigma^2/2\varepsilon_0$ (each patch sits in the average of inside and outside fields — the standard conductor-surface result from the electrostatics chapter). Numerically, per unit area:

P_{\rm burst} = \frac{\sigma^2}{2\varepsilon_0} = \frac{q^2}{32\pi^2\varepsilon_0 a^4} \qquad\xrightarrow{\;a = r_e\;}\qquad 1.46\times10^{29}\ \text{Pa},

fourteen hundred billion trillion atmospheres. Pure electromagnetism offers nothing to balance it — the field pushes every patch outward. A stable shell needs non-EM cohesion; this is not philosophy but force balance.

Step 2. The minimal closing agent. Model the cohesion as a uniform isotropic tension filling the interior: stress $\theta^{ij}_P = p_{\rm in}\,\delta^{ij}$ with $p_{\rm in} < 0$ (negative pressure pulls the wall inward). Equilibrium of a surface patch fixes it uniquely:

p_{\rm in} = -\frac{\sigma^2}{2\varepsilon_0} = -\frac{q^2}{32\pi^2\varepsilon_0 a^4}.

Why "minimal": any spherically symmetric static closure works for the theorem below (only the volume integral of the stress enters); the uniform-tension model is the simplest and is Poincaré's own. Its energy density is model-dependent — the minimal choice carries none — and that freedom is flagged honestly in Step 5.

Step 3. The stress integral closes the books exactly. The tension's contribution:

\int\theta^{xx}_P\,d^3x = p_{\rm in}\cdot\frac{4\pi a^3}{3} = -\frac{q^2}{32\pi^2\varepsilon_0 a^4}\cdot\frac{4\pi a^3}{3} = -\frac{q^2}{24\pi\varepsilon_0 a} = -\frac{U_{\rm em}}{3},

using (1) in the last step — an exact cancellation of the field's $+U_{\rm em}/3$ from Derivation B, verified numerically to machine precision. Total: $\int\theta^{xx}_{\rm tot} = 0$ . Von Laue's theorem is satisfied; the boost formula now returns

\mathbf{p}_{\rm tot} = \frac{U_{\rm tot}}{c^2}\,\mathbf{v},

a genuine four-vector relation. The tension required by statics is exactly the tension required by covariance. One condition, two consistencies — the deepest single fact in classical electron theory.

Step 4. What was repaired, said precisely. Nothing about the field's momentum changed — it is still $\tfrac43(U_{\rm em}/c^2)\mathbf{v}$ on the lab slice; open-system numbers remain slice-dependent. What changed is that the total system (field + cohesion) is closed, $\partial_\mu\theta^{\mu\nu}_{\rm tot} = 0$ everywhere, and closed-system energy-momentum is a four-vector however you slice it. The alternative repair in the literature — redefining the field's four-momentum by always integrating on the rest-frame simultaneity slice (Fermi 1922, Rohrlich's covariant definition) — makes the 4/3 vanish by bookkeeping instead; it is legitimate, and it changes no physics, because the physical object is the closed total, not the fragment. The two "resolutions" are one lesson: only closure is frame-proof.

Step 5. The honesty ledger. The Poincaré agent is a placeholder: classically consistent, physically unidentified. Its energy density (hence its contribution to $U_{\rm tot}$ ) is model freedom; stability against deformations (not just uniform expansion) needs more structure than uniform tension; and the whole construction lives at $a \lesssim r_e$ — two factors of $\alpha$ below the Compton scale where, per the ladder (2), the electron stopped being classical anyway. The verdict band finishes that sentence. Verified: patch equilibrium $\sigma^2/2\varepsilon_0$ against the conductor-surface stress result; $p_{\rm in}V = -U/3$ exact symbolically and numerically (ratio $-1.0000000$ ); $P_{\rm burst}(r_e) = 1.456\times10^{29}$ Pa recomputed from CODATA; structure matches Jackson §16.5–16.6 and Poincaré's 1906 construction.

Figure

The Poincare closure audited — the outward electrostatic pull on every patch fixes the interior tension, and the two stress integrals cancel to the digit

Limiting cases

case	result	meaning
$a \to 0$	$U_{\rm em} = q^2/8\pi\varepsilon_0 a \to \infty$ (linear)	the classical divergence; point electron unaffordable
$a = r_e$ shell	$U_{\rm em} = m_ec^2/2$ ; full mass needs $a = r_e/2$	$O(1)$ shape factors: never read the radius literally
field alone, boosted	$\mathbf{p} = \tfrac43(U/c^2)\mathbf{v}$ ; $\int\theta^{xx} = +U/3$	open system: not a four-vector — the 4/3
field $+$ Poincaré tension	$\int\theta^{xx}_{\rm tot} = 0$ ; $\mathbf{p} = (U_{\rm tot}/c^2)\mathbf{v}$	von Laue passed; covariance restored exactly
electron at $a < 10^{-19}$ m (data)	$m_{\rm em} > 14{,}000\,m_e \approx 7.2$ GeV	the classical model is dead on arrival
QED, cutoff $\Lambda$	$\delta m/m \sim \tfrac{3\alpha}{2\pi}\ln(\Lambda/m_ec^2)$	logarithmic: $\approx 0.18$ even at the Planck scale

Common trapThe classical electron radius is not a size — and the 4/3 is not a contradiction in relativity

Trap 1. $r_e = 2.82$ fm is the length at which a hypothetical classical self-energy would equal $m_ec^2$ — nothing more. Scattering experiments bound the electron's structure below $10^{-19}$ m, four orders of magnitude smaller, where the classical formula would demand a 7 GeV electron. Yet $r_e$ appears in hard data constantly (Thomson's $\sigma_T = \tfrac{8\pi}{3}r_e^2$ , synchrotron formulas) — as the natural cross-section scale built from $e$ , $m_e$ , $c$ , i.e. as bookkeeping. Say "the length scale $r_e$ ," never "the size of the electron." Trap 2. The 4/3 does not mean relativity fails for electromagnetic energy. $E = mc^2$ as a four-vector statement applies to closed systems; the shell's field alone is open (Derivation B's theorem), and open fragments have slice-dependent energy-momentum with no obligation to transform properly. Citing the bare 4/3 as a "problem with Maxwell + Einstein" gets the logic backwards: the factor is relativity correctly reporting that your books omit the binding agent. Close the system — Poincaré stresses, or a covariant slice definition — and the factor is gone identically.

Worked examples

1 · Weighing the field — the shell at the classical radius (confidence builder)

Setup. Take the shell model at $a = r_e = 2.818$ fm. Find the self-energy, the electromagnetic mass fraction, and the radius that would deliver the whole electron mass — for both shell and solid-sphere models.

Step 1. Self-energy. $U_{\rm em} = e^2/8\pi\varepsilon_0 r_e = \mathbf{0.2555}$ MeV — exactly $m_ec^2/2 = 0.2555$ MeV, since $U_{\rm em} = \tfrac12(e^2/4\pi\varepsilon_0 r_e) = \tfrac12 m_ec^2$ by the definition of $r_e$ .

Step 2. Mass fraction. $m_{\rm em}/m_e = U_{\rm em}/m_ec^2 = \mathbf{0.5}$ : the $r_e$ -shell explains half the mass. For the full mass: shell needs $a = r_e/2 = \mathbf{1.41}$ fm; the solid sphere ( $U = \tfrac35 e^2/4\pi\varepsilon_0 a$ ) needs $a = \tfrac35 r_e = \mathbf{1.69}$ fm.

Step 3. Read the spread. Three "electron radii" within a factor two, from one physical idea — the $O(1)$ shape dependence that already tells you the model has no sharp prediction. Compare the ladder: all these lengths sit at $\alpha^2 a_0$ , far below the Compton wavelength $\lambdabar_C = 386$ fm where quantum field theory takes the pen.

The lesson: the energy account is easy and internally consistent — the crisis only arrives when you open the momentum ledger.

2 · The full audit — 4/3 exposed, then repaired to the digit (exam level)

Setup. Shell with $a = r_e$ , boosted to $v = 0.01c$ . Compute the field momentum, exhibit the 4/3, compute the Poincaré tension, and verify the repaired books.

Step 1. The two candidate momenta. $(U_{\rm em}/c^2)v = (0.2555\ \text{MeV}/c^2)(0.01c) = \mathbf{1.365\times10^{-24}}$ kg·m/s. The honest integral (Derivation B): $p_{\rm em} = \tfrac43 \times 1.365\times10^{-24} = \mathbf{1.821\times10^{-24}}$ kg·m/s. Ratio: $1.3333$ — the defect, in SI.

Step 2. The stress audit. Field: $\int\theta^{xx}_{\rm EM} = +U_{\rm em}/3 = +1.365\times10^{-14}$ J. Von Laue demands zero for a closed static system: the field alone fails.

Step 3. The tension. Equilibrium requires $p_{\rm in} = -\sigma^2/2\varepsilon_0 = -e^2/32\pi^2\varepsilon_0 r_e^4 = \mathbf{-1.456\times10^{29}}$ Pa ( $1.4\times10^{24}$ atm of pull). Its integral: $p_{\rm in}\cdot\tfrac43\pi r_e^3 = \mathbf{-1.365\times10^{-14}}$ J $= -U_{\rm em}/3$ , matching the field's excess to machine precision (verified ratio $-1.000000$ ).

Step 4. The repaired books. $\int\theta^{xx}_{\rm tot} = 0$ ⇒ $p_{\rm tot} = (U_{\rm tot}/c^2)v$ exactly: with the minimal (energyless) tension model, $U_{\rm tot} = U_{\rm em}$ and $p_{\rm tot} = \mathbf{1.365\times10^{-24}}$ kg·m/s — the four-vector value. The extra third was never "in" the electron; it was the signature of a truncated ledger.

The lesson: one line of statics ( $p_{\rm in} = -\sigma^2/2\varepsilon_0$ ) and one line of field theory ( $\int\theta^{xx}_{\rm EM} = U/3$ ) cancel to the digit — the 4/3 problem is solved by closing the system, not by new postulates.

Going further

References. Jackson, Classical Electrodynamics, 3rd ed., Ch. 16 (Radiation Damping, Classical Models of the Electron) · Griffiths, Introduction to Electrodynamics, 4th ed., §11.2.2–11.2.3 · Zangwill, Modern Electrodynamics, Ch. 23 · Purcell & Morin, Electricity and Magnetism, 3rd ed., §5.6 · Rohrlich, Classical Charged Particles (2nd ed.) · Poincaré, "Sur la dynamique de l'électron," Rend. Circ. Mat. Palermo 21 (1906)

Provenance. Verifier should recompute: (1) shell self-energy integral $\int_a^\infty (e/4\pi\varepsilon_0 r^2)^2\,\tfrac{\varepsilon_0}{2}4\pi r^2\,\mathrm dr=e^2/(8\pi\varepsilon_0 a)$ — Jackson §16.4, Purcell §5.6; (2) $r_e=e^2/(4\pi\varepsilon_0 m_e c^2)=2.8179\times10^{-15}$ m (CODATA), $\tau=r_e/c\approx9.4\times10^{-24}\,$ s; (3) the 4/3 field-momentum derivation via BAC-CAB and angular averaging — Jackson §16.4; (4) Abraham–Lorentz force $\mu_0 e^2\dot a/(6\pi c)$ — Griffiths Eq.11.80; (5) Thomson cross section $\sigma_T=\tfrac{8\pi}{3}r_e^2=6.652\times10^{-29}\,\mathrm{m^2}$ (CODATA). References: Jackson Ch.16; Griffiths §11.2; Zangwill Ch.23; Rohrlich, Classical Charged Particles. Addendum for this edition — two base corrections: the base's item (2) quotes " $\tau = r_e/c \approx 9.4\times10^{-24}$ s" while its limits card asserts $F_{\rm rad} = m_e\tau\dot a$ with that same $\tau$ , and a symbol card elsewhere states " $\tau = r_e/c \approx 6.3\times10^{-24}$ s"; these are mutually inconsistent. The correct pairing is $\tau = \mu_0e^2/6\pi m_ec = \tfrac{2}{3}r_e/c = 6.266\times10^{-24}$ s in $F_{\rm rad} = m_e\tau\dot a$ (while $r_e/c = 9.40\times10^{-24}$ s is a different number, not the force's clock); this edition uses the correct $\tau$ throughout and defers the self-force entirely to its owners. Per the chapter's division of labor: item (4) and the whole Abraham–Lorentz/runaway story are owned by the E9 pair Radiation Reaction Force and Abraham–Lorentz Equation and Its Pathologies and the covariant treatment by Abraham-Lorentz Radiation Reaction Force next door — all recalled by name; the E9 shell derivation's $\delta m = \tfrac43 U/c^2$ is cited as the force-side symptom of the same 4/3. This edition owns: the energy account with the radius ladder $r_e = \alpha\lambdabar_C = \alpha^2 a_0$ (verified 6 digits), the 4/3 momentum integral with its numeric audit ( $1.821\times10^{-24}$ vs $1.365\times10^{-24}$ kg·m/s at $v = 0.01c$ , $a = r_e$ ), the open-system root-cause theorem with the boost formula and $\int\theta^{xx}_{\rm EM} = +U/3$ , the Poincaré-stress closure computed exactly ( $p_{\rm in} = -\sigma^2/2\varepsilon_0 = -1.456\times10^{29}$ Pa at $r_e$ ; $p_{\rm in}V = -U/3$ to machine precision), von Laue's theorem, the Fermi/Rohrlich covariant-definition honesty, and the verdict numbers ( $m_{\rm em}(10^{-19}\,\text{m}) = 1.4\times10^4\,m_e$ ; QED $\delta m/m \approx 0.18$ at a Planck cutoff vs classical $\sim10^{20}$ ). All numbers recomputed from CODATA for this edition; angular average and stress cancellations verified symbolically.

Builds on: Larmor Formula from the Liénard–Wiechert Fields · Liénard–Wiechert Potentials: Covariant Formulation · Fields of a Relativistic Moving Charge Leads to: Relativistic Generalization of Larmor's Formula · Synchrotron Radiation · Angular Distribution of Radiation from Accelerated Charges

Which mass does a laboratory scale actually weigh — bare, electromagnetic, or total?

Total, always, and nothing else is measurable even in principle within the theory. Every dynamical experiment — a Penning trap, a mass spectrometer, Compton recoil — couples to the inertia in Newton's (or Dirac's) equation, which is the renormalized combination $m = m_{\rm bare} + m_{\rm em}$ . The split between the two summands is a statement about a cutoff-dependent model, not about the electron: change the assumed radius (or in QED the regularization) and the summands slide while the total stays pinned to $0.51099895$ MeV. Renormalization looks like sweeping infinity under a rug; the rug covers a quantity no experiment ever probed. The classical program's real error was not the subtraction; it was hoping the split was physical (all mass electromagnetic), a hope the 4/3 and then the data dismantled.

Is the 4/3 factor just a mistake someone made that textbooks keep repeating?

No — it is a correct calculation of a slice-dependent quantity, endlessly mistaken for a paradox. The integrals in Derivation B are flawless; Abraham's 1902 value was right; the "problem" is only the expectation that a fragment of a system should transform as a four-vector. Von Laue's theorem makes the failure quantitative: $p = \gamma(v/c^2)(U + \int\theta^{xx})$ , and the field alone has $\int\theta^{xx} = +U/3$ . Three legitimate cures exist and agree: close the system physically (Poincaré stresses — statics then gives $-U/3$ exactly); define the field's four-momentum covariantly on the rest-frame slice (Fermi 1922, Rohrlich — the offending term never appears); or stop attributing standalone reality to a subsystem's momentum. The factor survives in textbooks because it is the cleanest teachable instance of a mistake physicists still make in fancier settings — quoting "the momentum of the electromagnetic part" of anything without declaring the slice.

Why does the same 4/3 show up in the shell's self-force calculation?

Because it is the same incomplete object measured a third way. The E9 article Radiation Reaction Force expands the shell's retarded self-interaction and finds $\mathbf{F}_{\rm self} = -\delta m\,\mathbf{a} + m\tau\dot{\mathbf{a}} + O(\varepsilon)$ with $\delta m = \tfrac43 U_{\rm es}/c^2$ — the inertia the field adds against acceleration. Consistency check, not coincidence: the momentum account (Derivation B) says the field carries $\tfrac43(U/c^2)\mathbf{v}$ , so accelerating the charge must build field momentum at rate $\tfrac43(U/c^2)\mathbf{a}$ , which the charge experiences as exactly that inertial back-force. One open system, three symptoms — energy says $1$ , momentum says $\tfrac43$ , self-force says $\tfrac43$ — and one repair: with Poincaré stresses in the books, the total inertia is $U_{\rm tot}/c^2$ and all three accounts agree. (The renormalization of $\delta m$ into the observed mass is untouched by any of this: whatever the coefficient, it is absorbed.)

Does the Poincare stress mean the electron really contains a mysterious non-electromagnetic substance?

Classically: yes by necessity, identity unknown — that is an honest theorem, not hand-waving. Any finite charged equilibrium needs non-EM cohesion (the field alone pushes every patch outward; Earnshaw's theorem forbids electrostatic self-confinement), and relativity independently demands the closure. But the model died before the substance needed naming: QED's electron is pointlike as far as any experiment resolves, held together by nothing because there is nothing to hold together — the self-energy problem migrates from statics to the radiative corrections, where it is logarithmic and renormalized. The lasting content of Poincaré's move is structural: whenever a bound system's energy is partitioned ("how much of the proton's mass is gluon field energy?"), the binding stresses must appear in the same books, or 4/3-style nonsense returns. Lattice QCD's proton-mass decomposition does precisely this accounting, the trace anomaly playing Poincaré's part.

If QED's self-energy still diverges, in what sense is it better than the classical one?

Two decisive senses — rate and place. Rate: the classical divergence is linear ( $\delta m \propto 1/a$ ), so cutting off at any remotely credible scale produces absurdity ( $10^{20}\,m_e$ at a Planck-length cutoff); QED's is logarithmic, $\delta m/m \sim \tfrac{3\alpha}{2\pi}\ln(\Lambda/m_ec^2)$ , giving a placid $0.18$ even with the cutoff at the Planck energy — the electron's quantum delocalization over $\lambdabar_C$ smears the Coulomb singularity almost completely. Place: in QED the divergence is systematically confined to a redefinition of $m$ and $e$ , after which every observable — the Lamb shift, $g-2$ to twelve digits — comes out finite and correct; the classical theory had no such theorem, only the one subtraction and then pathology (runaways) at the next order. The classical electron theory was renormalization's proof of concept; QED is its proof of consistency.

Is there any connection between the negative Poincare pressure and dark energy?

A structural one, worth stating carefully. Both are stress-energy components with negative pressure entering the books to satisfy a consistency requirement — Poincaré's to close a static charged system ( $\int\theta^{ij}_{\rm tot} = 0$ ), dark energy's ( $p = -\rho c^2$ ) to source the observed accelerated expansion in Einstein's equations. The Poincaré stress was the first serious appearance of tension-as-substance in fundamental physics, and the formal habit it started — treat pressure as a first-class entry in $\theta^{\mu\nu}$ , let consistency fix its size — is exactly how a cosmological constant is handled. But the analogy ends at bookkeeping: the Poincaré agent is confined to $a \sim r_e$ inside a dead model, while dark energy is (as far as measured) uniform, cosmological, and very much alive. Use it as intuition for how tension enters the books, never as a claim of shared mechanism.

What would it take, experimentally, to revive an extended classical electron?

A structure signal above $10^{-19}$ m, and every relevant probe says no. High-energy $e^+e^-$ scattering at LEP tested QED's pointlike vertex to $\sim10^{-19}$ m (contact-interaction bounds); the electron's $g-2$ , computed pointlike through five loops, matches measurement at the $10^{-13}$ level — a compositeness scale low enough to matter would spoil it grossly (generic estimates put the bound at $10^{-21}$ – $10^{-22}$ m); no electron excited states or form-factor deviations have ever appeared. Meanwhile the classical model requires structure at $r_e \sim 10^{-15}$ m — four orders of magnitude into excluded territory, where the measured mass would have to be $\sim7$ GeV of field energy minus a finely tuned negative bare mass, precisely the unnaturalness renormalization exists to avoid. The extended classical electron is excluded as physics — and remains superb pedagogy.