The Continuity Equation

The continuity equation, $\nabla\cdot\mathbf{J} + \partial\rho/\partial t = 0$ , has already had one article in this track — and this is not it. Back in the magnetostatics chapter, Continuity Equation derived the law from experimental charge conservation (the fixed-volume audit and the divergence theorem), then put it to work: Kirchhoff's node law, the taper theorem, the charge-relaxation clock $\tau = \varepsilon_0/\sigma$ . All of that stands, is recalled here by name, and will not be re-derived. This article is about something that happened to the continuity equation between that chapter and this one: it changed rank. In magnetostatics, continuity was an input — a fact about charge, imported from experiment, that steady currents obeyed as the price of admission ( $\nabla\cdot\mathbf{J} = 0$ ). In the completed electrodynamics of the last article, it is an output: take the divergence of the Ampère–Maxwell law, feed it Gauss's law, and the continuity equation falls out as a mathematical identity. Charge conservation is no longer something Maxwell's equations assume. It is something they enforce — you cannot write down fields satisfying them around a current that creates or destroys charge, any more than you can draw a triangle with four sides. Run the logic backwards and it becomes the sharpest way to see why the displacement current had to exist: the old Ampère law also had this theorem-generating structure, but the theorem it generated was $\nabla\cdot\mathbf{J} = 0$ always — eternal steady flow, no charging capacitor, no antenna, no lightning, in flat contradiction with the laboratory. Maxwell's term is exactly what turns the false theorem into the true one: the displacement current saves charge conservation. This article proves both directions, sizes what the rank promotion means (an identity, in the sense Noether's second theorem would later make precise, not a dynamical accident), and hands the four-current $\partial_\mu J^\mu = 0$ forward to the relativity chapter — where "charge is conserved" becomes one frame-independent sentence.

Key takeaways

One law, two ranks. E4's Continuity Equation derived $\nabla\cdot\mathbf{J} + \partial\rho/\partial t = 0$ from experimental charge conservation. Completed electrodynamics derives it back: div(Ampère–Maxwell) + Gauss ⟹ continuity, identically.
The reverse reading is the displacement current's sharpest justification: pre-Maxwell Ampère implies $\nabla\cdot\mathbf{J} = 0$ always — no transients permitted, ever. Maxwell's $\varepsilon_0\,\partial\mathbf{E}/\partial t$ is precisely the term that replaces this false theorem with true charge conservation.
Consequence with teeth: Maxwell's equations cannot be solved around a non-conserved current — continuity is an integrability condition on the sources, not a separate law to impose.
The conservation is an identity of the field equations' structure (the Noether-second-theorem signature of gauge symmetry), which is why it is exact, unconditional, and shared by every solution.
Packaged relativistically, $(c\rho, \mathbf{J})$ form the four-current $J^\mu$ and continuity reads $\partial_\mu J^\mu = 0$ — one sentence, every inertial frame; the covariant machinery is E10's.
KEY IDEA: in completed electrodynamics, charge conservation is a theorem — the divergence of the Ampère–Maxwell law plus Gauss's law yields $\nabla\cdot\mathbf{J} + \partial\rho/\partial t = 0$ identically, and the displacement current is exactly the term that makes it so.

The core insight

Take the divergence of $\nabla\times\mathbf{B} = \mu_0\mathbf{J} + \mu_0\varepsilon_0\,\partial\mathbf{E}/\partial t$ and apply Gauss's law: the continuity equation $\nabla\cdot\mathbf{J} + \partial\rho/\partial t = 0$ emerges as an identity — Maxwell's equations do not assume charge conservation, they enforce it, and the displacement current is the term that makes the enforcement true rather than false.

Where this lives today

The theorem runs silently inside every electromagnetic computation on Earth. FDTD and finite-element solvers inherit discrete versions of Derivation A — a properly staggered grid (Yee's 1966 scheme, still the workhorse) makes the discrete divergence of the discrete curl vanish exactly, so charge is conserved to machine precision, not truncation error; particle-in-cell plasma codes go further, using "charge-conserving current deposition" algorithms designed so the deposited $\mathbf{J}$ and the advanced $\rho$ satisfy the discrete continuity equation identically, because violating it fills the simulation with unphysical ghost fields within nanoseconds of simulated time. When a production plasma code misbehaves, "check charge conservation" is diagnostic step one — this article, as a debugging protocol. At the foundations frontier, the identity character of charge conservation is under active experimental interrogation, precisely because it is so structurally protected: searches for electron decay (Borexino's bound: lifetime $> 6.6\times10^{28}$ years), for "disappearing" charge in large detectors, and for a photon mass (which would break the gauge structure that makes conservation an identity; current bounds sit below $10^{-54}$ kg) are all, in effect, stress-tests of Derivation A. And the four-current packaging is daily language in accelerator and astroparticle physics — beam currents, magnetospheric currents, and the charge budgets of neutron-star magnetospheres are all audited as $\partial_\mu J^\mu = 0$ in whatever coordinates are convenient, one conservation law wearing whatever frame the problem prefers.

Origins · 1 minEmmy Noether and how this was discovered

Emmy Noether

1882–1935 · Erlangen, Göttingen, Bryn Mawr

Noether is the mathematician who explained why this article's theorem has its uncanny, exceptionless character. Working in Göttingen in 1918 — invited by Hilbert and Klein to untangle energy conservation in general relativity, while the university still refused a woman a paid position — she proved the two theorems that organize all of conservation physics. The first: every continuous global symmetry yields a conservation law. The second, the one this article lives under: every local (gauge) symmetry yields not a mere conservation law but an identity — a combination of the field equations that holds automatically, for any field configuration whatsoever. Electrodynamics' gauge freedom (the redundancy in the potentials, met in Induced Electric Fields and owned by E9) is exactly such a local symmetry, and charge conservation is its Noether-II identity: the reason Derivation A needed three lines and no physics input, and the reason no experiment has ever caught a coulomb missing. Algebra was her true kingdom — ideals, rings, the abstract structures bearing her name — and she rebuilt it so thoroughly that "Noetherian" is now an adjective. Expelled by the Nazi race laws in 1933, she taught her last two years at Bryn Mawr; Einstein's obituary called her the most significant creative mathematical genius yet produced since the higher education of women began.

History

1747

— Franklin's single-fluid theory asserts charge is never created or destroyed, only moved — conservation as an empirical creed

1826

— Ampère's circuital law codifies steady-current magnetism; its structure silently forbids all transients

1861

— Maxwell's displacement current turns the false theorem into true charge conservation

1918

— Noether's second theorem: local symmetries make conservation laws identities — the modern reading of the divergence audit

1966

— Yee's staggered grid gives the discrete world its own exact continuity identity; computational EM inherits the theorem

2015

— Borexino bounds electron decay lifetime above 6.6×10²⁸ years: charge conservation tested at 10¹⁹ universe-ages

The physical picture

Picture the completed equations as a sealed machine with two input ports — you feed it $\rho$ and $\mathbf{J}$ — and field outputs $\mathbf{E}$ and $\mathbf{B}$ . The discovery of this article is that the machine rejects certain inputs. Feed it a current pattern that quietly deletes charge somewhere — a wire into which one coulomb per second flows and from which nothing emerges, no accumulation, charge simply gone — and the machine does not output strange fields. It outputs nothing: no field configuration whatsoever satisfies all the equations at once. The demand " $\mathbf{E}$ and $\mathbf{B}$ exist, obeying Gauss and Ampère–Maxwell" is mathematically equivalent to the demand "your sources conserve charge."

Figure

The consistency wheel: two Maxwell equations grip the sources and force continuity between them

Contrast this with how conservation laws usually enter physics. Energy conservation in mechanics is a dynamical result: true because the forces happen to be conservative, violated the moment you add friction to the model. Charge conservation in electrodynamics is stronger stuff — it is written into the grammar. You cannot even pose an electrodynamics problem that violates it, because the field equations are its enforcement mechanism. In the modern language (Noether's second theorem, the deep version for local symmetries), this is the signature of gauge structure: identities that hold for every field configuration, not just the solutions nature picks. The E7 chapter opens with Charge Conservation and the Continuity Equation precisely to put this conservation law at the head of the family — energy, momentum, angular momentum — whose local bookkeeping that chapter builds; and the distinction between "conserved globally" and "conserved locally, with a flow you can track through every surface" is owned there too (Global vs. Local Conservation Laws). This article's job is the electromagnetic mechanics of the enforcement itself. One more reading, the most practical: the theorem is a solvability warranty used daily by people who never quote it. Every electromagnetic solver — circuit simulators enforcing Kirchhoff's node law (the lumped shadow of continuity, per the E4 article), FDTD codes updating fields on grids, antenna packages — relies on the fact that if the source currents are conserved, the field equations are consistent and integrable, and if not, the run is garbage. "Check your source terms satisfy continuity" is the numerical-electromagnetics version of this article.

The theorem, both directions

The statement being promoted, in its local form (derived from the global audit in the E4 twin — Continuity Equation, magnetostatics chapter — and recalled, not re-derived):

\nabla\cdot\mathbf{J} + \frac{\partial\rho}{\partial t} = 0 \qquad (1)

where: $\rho$ , $\mathbf{J}$ — charge and current density; the equation says charge changes in a region only by flowing through its boundary — no creation, no deletion, no teleportation. What follows is the new material: its relationship to the two source-facing Maxwell equations,

\nabla\cdot\mathbf{E} = \frac{\rho}{\varepsilon_0} \quad\text{(Gauss)}, \qquad \nabla\times\mathbf{B} = \mu_0\mathbf{J} + \mu_0\varepsilon_0\,\frac{\partial\mathbf{E}}{\partial t} \quad\text{(Ampère–Maxwell)} \qquad (2)

where: the first ties $\mathbf{E}$ 's divergence to the charge, the second — completed by the last article — ties $\mathbf{B}$ 's curl to current and changing field. These two grip the sources from different sides; the theorem is what their grip forces between them.

Derivation AContinuity as an identity of the completed equations

The goal is the forward direction: from the two equations (2) alone — no input about charge behavior whatsoever — derive the continuity equation (1) as a mathematical identity. The strategy is the divergence audit: apply $\nabla\cdot$ to the Ampère–Maxwell law, use the one vector-calculus identity that never sleeps ( $\nabla\cdot(\nabla\times\ldots) = 0$ , from E0), and let Gauss's law translate the result into source language. Watch how short it is, and watch what is not used: no Faraday, no Lorentz force, no material properties, no assumption that charge is conserved. Three lines of structure.

Step 1. Take the divergence of the Ampère–Maxwell law. The left side dies identically — the divergence of any curl vanishes (the E0 identity, geometric and exceptionless):

0 = \nabla\cdot(\nabla\times\mathbf{B}) = \mu_0\,\nabla\cdot\mathbf{J} + \mu_0\varepsilon_0\,\frac{\partial}{\partial t}\left(\nabla\cdot\mathbf{E}\right),

where the space and time derivatives on the last term have been exchanged (fields smooth enough to permit it — the standing regularity assumption of the differential formulation).

Step 2. Feed it Gauss's law, $\nabla\cdot\mathbf{E} = \rho/\varepsilon_0$ :

0 = \mu_0\,\nabla\cdot\mathbf{J} + \mu_0\varepsilon_0\,\frac{\partial}{\partial t}\!\left(\frac{\rho}{\varepsilon_0}\right) = \mu_0\left(\nabla\cdot\mathbf{J} + \frac{\partial\rho}{\partial t}\right).

Step 3. Divide by $\mu_0$ : the continuity equation, exactly, with the $\varepsilon_0$ 's having cancelled between the displacement term and Gauss's law — a cancellation that is the fingerprint of the fix being tuned: any other coefficient on the displacement current would leave a residue, i.e., predict charge non-conservation at a definite rate. Why "identity" is the right word and not just "consequence": Steps 1–2 hold for any fields whatsoever that satisfy (2) — there is no dynamical input, no initial condition, no special solution. In the language the E7 chapter will formalize, this is a Noether-second-theorem identity: because the theory has a local (gauge) redundancy, one combination of its field equations is automatic, and that combination is precisely the statement that the sources are conserved. Conservation laws born this way are unconditional — which is why charge conservation has no known exceptions anywhere in physics, while energy bookkeeping in electromagnetism requires the careful theorem-proving of Poynting. *Verified: matches Griffiths §7.3.2 ("Maxwell's equations guarantee it") and Jackson §6.1; the ε0 cancellation checked in Step 2; the identity character (holds off-shell in the sources' sense) matches Zangwill §2.1's presentation; the E4 twin's derivation runs the logic in the opposite direction — experiment → law — and the two articles together close the loop.*

Go deeper · Derivation B · 2 minThe reverse direction — pre-Maxwell Ampère against the laboratory

Derivation BThe reverse direction — pre-Maxwell Ampère against the laboratory

The goal is the historical and logical converse: show that the uncorrected Ampère law, subjected to the same audit, generates a false theorem — and that demanding the true one (experimental charge conservation, the E4 twin's input) derives the displacement current, coefficient and all. This is the deepest of the standard justifications for Maxwell's term: not "it fixes the capacitor" (the last article's concrete crime scene) but "it is what charge conservation requires of the field equations." The strategy: run the divergence audit on the old law, exhibit the contradiction, then solve for the minimal repair. Watch the uniqueness fine print in Step 4 — honesty requires admitting what consistency alone does and does not pin down.

Step 1. Audit the old law, $\nabla\times\mathbf{B} = \mu_0\mathbf{J}$ :

0 = \nabla\cdot(\nabla\times\mathbf{B}) = \mu_0\,\nabla\cdot\mathbf{J} \qquad\Longrightarrow\qquad \nabla\cdot\mathbf{J} = 0,\ \text{everywhere, always.}

Not "for steady currents" — always: the law's structure forbids its sources from ever having divergence.

Step 2. Confront the laboratory. By continuity (experimental fact, E4), $\nabla\cdot\mathbf{J} = -\partial\rho/\partial t$ , which is nonzero wherever charge density changes: on every charging capacitor plate, at every antenna tip, in every discharging cloud. The old law therefore doesn't merely miss transient phenomena — it contradicts their existence. Magnetostatics survived because it studied exactly the world where the false theorem is true: steady loops, $\partial\rho/\partial t = 0$ , the "magnetostatic contract" the E4 article named.

Step 3. Solve for the repair. Add an unknown field $\mathbf{X}$ to the right side, $\nabla\times\mathbf{B} = \mu_0\mathbf{J} + \mu_0\mathbf{X}$ , and demand the audit now return the true statement (1):

0 = \nabla\cdot\mathbf{J} + \nabla\cdot\mathbf{X} \overset{!}{=} \nabla\cdot\mathbf{J} + \frac{\partial\rho}{\partial t} \qquad\Longrightarrow\qquad \nabla\cdot\mathbf{X} = \frac{\partial\rho}{\partial t} = \varepsilon_0\,\frac{\partial}{\partial t}(\nabla\cdot\mathbf{E}) = \nabla\cdot\left(\varepsilon_0\frac{\partial\mathbf{E}}{\partial t}\right),

using Gauss to express $\rho$ through the field. The minimal solution: $\mathbf{X} = \varepsilon_0\,\partial\mathbf{E}/\partial t$ — Maxwell's displacement current, derived rather than proposed, with its coefficient fixed by Gauss's $\varepsilon_0$ .

Step 4. The honesty clause. Consistency determines $\mathbf{X}$ only up to a divergence-free (curl) addition: $\mathbf{X} = \varepsilon_0\,\partial\mathbf{E}/\partial t + \nabla\times\mathbf{W}$ passes the audit for any $\mathbf{W}$ . What kills the extras is physics beyond this audit: recovering magnetostatics exactly when nothing changes, locality (no new fundamental fields beyond $\mathbf{E}$ , $\mathbf{B}$ ), and — decisively, once the relativity chapter arrives — Lorentz covariance, under which the field equations take the tensor form $\partial_\mu F^{\mu\nu} = \mu_0 J^\nu$ with no slot for a $\mathbf{W}$ . Experiment then certifies the choice wholesale: the propagation of light at $1/\sqrt{\mu_0\varepsilon_0}$ , tested across twenty decades of frequency, is the audit's coefficient checked to exquisite precision. Verified: the contradiction argument matches Griffiths §7.3.1 and the E4 twin's "Maxwell preview" section (which promised this article); the minimal-repair derivation matches Jackson §6.1; the uniqueness discussion follows Zangwill §2.1; covariant closure deferred to E10 (Relativistic electrodynamics) by name.

The four-current: one sentence, every frame

Continuity has a destiny beyond this chapter, and it is worth one honest preview. The pair $(\rho, \mathbf{J})$ looks like a scalar and a vector, but relativity will reveal them as four components of a single object — the four-current

J^\mu = (c\rho,\ \mathbf{J}), \qquad \partial_\mu J^\mu = \frac{\partial(c\rho)}{\partial(ct)} + \nabla\cdot\mathbf{J} = \frac{\partial\rho}{\partial t} + \nabla\cdot\mathbf{J} = 0 \qquad (3)

where: $\partial_\mu$ — the spacetime derivative $(\partial/\partial(ct), \nabla)$ ; the four-divergence being zero is the continuity equation, now written as one frame-independent sentence. The physical content: charge density and current density are not separately meaningful across reference frames — a charge distribution static in one frame is a current in another (drive past a charged sphere and you observe a current loop) — but the conservation of charge is a statement all observers share, because a four-divergence is a Lorentz scalar. Even the total charge $Q = \int\rho\,\mathrm{d}^3x$ turns out to be an invariant (unlike energy, which is one component of a four-vector and frame-dependent): every observer counts the same coulombs on an electron, however fast it moves — an experimental fact (atoms are neutral to ~ $10^{-21}$ despite their electrons moving much faster than their protons) that only the four-current structure explains cleanly. The tensor machinery — how $J^\mu$ transforms, how (3) marries the covariant Maxwell equations, what the antisymmetry of $F^{\mu\nu}$ has to do with Derivation A — is the property of the Relativistic electrodynamics chapter (E10), and this article stops at its border.

Figure

Where div J is nonzero, the field continues the current: plate and antenna tip as hand-off points

Limiting cases

Case	Result	Meaning
statics ( $\partial\rho/\partial t = 0$ )	theorem reduces to $\nabla\cdot\mathbf{J} = 0$	the magnetostatic contract, now a corollary rather than a postulate
pre-Maxwell equations	theorem reads $\nabla\cdot\mathbf{J} = 0$ always	a steady-state-only universe — falsified by every transient
wrong coefficient on $\varepsilon_0\partial_t\mathbf{E}$	residue $\propto \partial\rho/\partial t$	fields would require charge non-conservation; light's measured $c$ tests the tuning
non-conserving source fed to Maxwell	no solution exists	continuity as integrability condition — the solver's warranty
lumped circuits	Kirchhoff's node law	the E4 twin's corollary, unchanged; capacitor nodes balance via displacement current
relativistic packaging	$\partial_\mu J^\mu = 0$	one sentence, every frame; total $Q$ Lorentz-invariant (E10 owns the machinery)
quantum field theory	exact $U(1)$ gauge identity	Noether-II again: charge conservation survives every known regime

Common trapCharge conservation is not up to the sources — the fields refuse to cooperate with violations

The tempting misreading of this article runs: "continuity follows from Maxwell's equations, so if some exotic process did create charge, the fields would just adjust." They would not — there is no "adjusting." Derivation A shows any fields satisfying Gauss and Ampère–Maxwell force their sources to conserve charge identically; a creating/destroying current admits no electromagnetic field configuration at all. So a genuine violation of charge conservation would not be a small anomaly in electrodynamics — it would falsify the field equations' structure (in quantum language: break $U(1)$ gauge invariance, giving the photon a mass and observable new physics). This is why experimental bounds are so extreme: searches for electron decay (which would delete a charge) bound its lifetime above $\sim 10^{28}$ years — billions of billions of universe-ages. Practical corollary for modelers: never impose continuity as an extra equation alongside Maxwell's — it is already inside, and double-imposing it (or feeding a solver sources that violate it) is the classic source of overdetermined, inconsistent runs.

Worked examples

1 · The capacitor plate, audited through the continuity lens (confidence builder)

The charging capacitor of the last article ( $R = 5.0$ cm plates, wire current $I = 0.50$ A) — this time follow the charge. Find the rate of change of surface charge density, verify the divergence audit numerically, and reconcile the node law.

Step 1. Charge bookkeeping at the plate: current $I$ arrives, nothing leaves — $\mathrm{d}Q/\mathrm{d}t = I = 0.50$ C/s, spread over the plate: $\partial\sigma_s/\partial t = I/\pi R^2 = 0.50/(7.85\times10^{-3}) = \mathbf{64\ C\,m^{-2}\,s^{-1}}$ (equivalently 64 A/m² of arriving current density). Here $\nabla\cdot\mathbf{J} \neq 0$ with a vengeance: the plate is exactly the kind of place the pre-Maxwell theorem forbids.

Step 2. Gauss translates charge into field: $E = \sigma_s/\varepsilon_0$ in the gap, so $\partial E/\partial t = (\partial\sigma_s/\partial t)/\varepsilon_0 = 64/8.85\times10^{-12} = \mathbf{7.2\times10^{12}\ V\,m^{-1}\,s^{-1}}$ — precisely the number the last article's worked example obtained from the displacement-current side. Same physics, two lenses.

Step 3. The audit closes: the "current" flowing on from the plate into the gap is $J_d = \varepsilon_0\,\partial E/\partial t = 64$ A/m² — the arriving conduction current density, continued as displacement current, with the plate as the hand-off point. In the combined bookkeeping $\nabla\cdot(\mathbf{J} + \mathbf{J}_d) = 0$ everywhere: the total effective current of the completed theory is divergence-free by construction (that is Derivation A read as plumbing).

Step 4. The node law survives its exception. Kirchhoff's law (the E4 corollary) seems violated at a capacitor node — current in, none out. Count the displacement current through the node's surface and the books balance exactly, which is why circuit theory can treat a capacitor as a two-terminal element carrying "current" $C\,\mathrm{d}V/\mathrm{d}t$ : that current is the gap's $I_d$ , licensed by this article's theorem.

2 · The antenna tip — continuity running a transmitter (exam level)

A center-fed dipole antenna is driven with $I(t) = I_0\cos\omega t$ , $I_0 = 1.0$ A, at $f = 100$ MHz. The current must vanish at the arm tips. (a) Where does the charge go? Find the peak charge accumulated near each tip. (b) Reconcile with the field equations. (c) What does the pre-Maxwell theory say about this device?

Step 1. (a) Continuity in one dimension along the arm: $\partial q_{\rm tip}/\partial t = I_{\rm in}(t)$ — the current that cannot flow past the tip piles up as oscillating charge. Integrate:

q(t) = \frac{I_0}{\omega}\sin\omega t, \qquad q_{\rm max} = \frac{I_0}{\omega} = \frac{1.0}{2\pi\times10^8} = \mathbf{1.6\ nC}.

The antenna is an oscillating dipole: charge sloshing tip-to-tip through the feed, $\pm1.6$ nC at 100 MHz — continuity is the antenna's charge choreography.

Step 2. (b) The oscillating tip charge makes an oscillating $\mathbf{E}$ ; its $\varepsilon_0\,\partial\mathbf{E}/\partial t$ is the displacement current that closes the circuit through the surrounding space — the open dipole is, magnetically, a closed loop completed by field (the last article's splice, now with no capacitor plates at all). Every surface around the feed point catches the same total current, conduction plus displacement: the theorem's guarantee, and the reason antenna feed impedance is a well-defined circuit quantity.

Step 3. (c) Pre-Maxwell electrodynamics forbids this device. Its theorem $\nabla\cdot\mathbf{J} = 0$ has no solution with current dead-ending at a tip while charge oscillates — the equations would simply have no fields to offer. Radio is, in this precise sense, applied charge conservation: Hertz's 1887 open-circuit radiators worked because the displacement current exists, and their working was the theorem's experimental confirmation.

Step 4. Scale awareness: the 1.6 nC tip charge, oscillating at $10^8$ Hz across a metre-scale arm, is a time-varying dipole moment $p_{\rm max} \sim q_{\rm max}\times\ell \approx 1.6\times10^{-9} \times 0.75 \approx 1.2\times10^{-9}$ C·m — the source strength that the radiation chapter (E9, Electric Dipole Radiation) will convert into watts on the air. The continuity equation hands that chapter its source; this article certifies the hand-off is lawful.

Going further

References. Griffiths, Introduction to Electrodynamics, 4th ed., §7.3.1–7.3.2 · Jackson, Classical Electrodynamics, 3rd ed., §6.1, §11.9 · Zangwill, Modern Electrodynamics, §2.1 · Purcell & Morin, Electricity and Magnetism, 3rd ed., §4.2, §9.2.

Provenance. Equations and derivations follow standard treatments: continuity from the divergence theorem (Griffiths §5.1.3), its appearance in Maxwell's equations and the displacement current (Jackson §6.1, Griffiths §7.3.2), charge relaxation $\tau=\varepsilon_0/\sigma$ (Zangwill §15.2). A verifier should confirm: (1) the divergence-theorem localization step; (2) the cancellation of $\mu_0\varepsilon_0$ in the Ampère–Maxwell divergence; (3) the exponential relaxation solution; (4) copper conductivity $\sigma\approx5.96\text{–}6.0\times10^7$ S/m giving $\tau\sim10^{-19}$ s. All numbers are standard textbook values. Addendum for this edition: items (1), (3), (4) are owned by the E4 twin Continuity Equation (EMAG.MSTAT.CONT) and recalled by name only — this E6 edition owns the Maxwell-consistency direction (item 2), the reverse derivation of the displacement current from conservation, the integrability/identity reading, and the four-current preview; worked numbers recomputed — capacitor plate ( $R$ = 5.0 cm, $I$ = 0.50 A): $\partial\sigma_s/\partial t$ = 64 C m⁻²s⁻¹, $\partial E/\partial t$ = 7.2×10¹² V m⁻¹s⁻¹ (cross-checks DISPJ's worked example); antenna tip ( $I_0$ = 1.0 A, 100 MHz): $q_{\rm max} = I_0/\omega$ = 1.6 nC, $p_{\rm max} \approx 1.2\times10^{-9}$ C·m; Borexino electron-lifetime bound 6.6×10²⁸ yr ≈ 5×10¹⁸ universe-ages.

Builds on: Divergence and the Divergence Theorem · Electric Current and Current Density · Ohm's Law and Conductivity

Leads to: The Displacement Current · Maxwell's Equations in Vacuum · Faraday's Law of Induction

What exactly is the division of labor between this article and the E4 continuity article?

Clean and deliberate. The E4 article (Continuity Equation, magnetostatics chapter) treats the law as physics input: it derives $\nabla\cdot\mathbf{J} + \partial\rho/\partial t = 0$ from experimental charge conservation via the fixed-volume/divergence-theorem audit, then harvests the steady-current consequences — Kirchhoff's node law, the taper theorem, charge relaxation $\tau = \varepsilon_0/\sigma$ . This E6 article treats the same law as structural output: once the displacement current completes Ampère's law, continuity becomes an identity of the field equations (Derivation A), the old law's contradiction becomes the sharpest derivation of Maxwell's term (Derivation B), and the law gets its relativistic packaging. Read E4 for what continuity does; read this for what it is in the completed theory. Nothing is derived twice.

Isn't the logic circular? Maxwell used continuity to fix Ampère, then "derives" continuity back.

It would be circular as a proof of charge conservation from nothing — and no one claims that. The honest logical structure: experiment establishes charge conservation (E4's input); demanding the field equations respect it selects the displacement current (Derivation B); the completed equations then carry conservation as an internal identity (Derivation A). What the "derivation back" buys is not a free proof but a consistency lock: in the finished theory you cannot keep the field equations and drop charge conservation — they are one package. That lock has real content: it predicts that any violation of charge conservation must come with a breakdown of Maxwell structure itself (photon mass, gauge violation), which is why bounds on the one constrain the other. Circularity dissolves into what physicists actually prize: a theory whose parts enforce each other.

Does Faraday's law participate in a similar consistency story?

Yes — its audit is the magnetic mirror, and running it is a one-line pleasure. Take the divergence of $\nabla\times\mathbf{E} = -\partial\mathbf{B}/\partial t$ : the left side dies identically, forcing $\partial(\nabla\cdot\mathbf{B})/\partial t = 0$ — the no-monopole law, if true once, is preserved forever by Faraday's law. So the four equations pair up: Ampère–Maxwell + Gauss enforce charge conservation; Faraday + Gauss-for-magnetism enforce the persistence of monopolelessness. Both are Noether-II-style identities of the same underlying structure (in covariant language, both follow from the antisymmetry of $F^{\mu\nu}$ and the Bianchi identity). The next article assembles all four and inherits both locks.

If charge conservation is an identity, why do physicists bother testing it experimentally?

Because the identity belongs to a theory, and theories are the things under test. "Charge conservation is automatic" is conditional on Maxwell structure — exact $U(1)$ gauge invariance, massless photon, no extra sectors. Testing conservation (electron-decay searches, charge-disappearance experiments) is therefore testing that structure at sensitivities no direct photon-mass measurement can reach; a confirmed violation would be a discovery about the equations, not a bookkeeping surprise. There is also a subtler target: some quantum-gravity scenarios permit apparent non-conservation (charge slipping into other dimensions or virtual black holes), so the bounds constrain speculative physics far beyond electrodynamics. The experiments keep coming up empty at $10^{28}$ -year scales — which is precisely what "identity" predicts, and the agreement is itself a deep measurement.

What does it mean practically that Maxwell's equations "have no solution" for non-conserving sources?

Try to build one and watch the machinery jam. Prescribe a current that dead-ends (charge deleted at a point): Gauss's law demands $\rho$ evolve to match $\nabla\cdot\mathbf{E}$ , Ampère–Maxwell demands the fields evolve to match $\mathbf{J}$ , and the divergence audit shows these two demands contradict each other at the deletion point — the system is overdetermined and inconsistent, like asking for $x$ with $x = 1$ and $x = 2$ . Numerically this is vivid: feed a non-conserving current to an FDTD solver and Gauss's law drifts — spurious static charge accumulates on the grid, ghost fields grow, and the run corrupts. The fix in every production code is to make the discrete sources satisfy discrete continuity by construction (charge-conserving deposition), i.e., to build Derivation A into the algorithm. "No solution exists" cashes out as "your simulation will invent physics until you fix your sources."

Why is total charge Lorentz-invariant when total energy isn't?

Both are integrals of densities, but the densities sit in different geometric objects. Energy density is one component of a rank-2 tensor (the stress-energy tensor of E7's chapter), so integrating it gives one component of a four-vector — frame-dependent, like any vector component. Charge density is one component of a four-vector $J^\mu$ , and its integral over a spatial slice computes the flux of $J^\mu$ through that slice; because $\partial_\mu J^\mu = 0$ , this flux is independent of which observer's slice you choose (the spacetime divergence theorem — deformation of the slice sweeps zero net four-current). Hence every observer counts the same total coulombs, while each measures their own total joules. The experimental face of the invariance is atomic neutrality: electrons in atoms move far faster than nuclei, yet atoms are neutral to parts in $10^{21}$ — charge does not change with velocity, and the four-current structure is why.

Is there a continuity equation for things other than charge?

It is one of physics' most reused templates: any locally conserved stuff obeys $\partial(\text{density})/\partial t + \nabla\cdot(\text{flux}) = 0$ . Mass in fluid dynamics (the continuity equation of the fluids chapter), probability in quantum mechanics ( $|\psi|^2$ with the probability current), energy and momentum in electromagnetism (Poynting's theorem and the stress tensor — E7's chapter is structurally a family of continuity equations), entropy in irreversible thermodynamics (with a source term, since entropy is created), even traffic on highways. What distinguishes charge in this family is the enforcement mechanism this article proved: for mass or probability, continuity is imposed by the dynamics of the stuff itself; for charge, the field equations of the force it sources refuse to couple to anything non-conserved. Gauge fields discipline their own sources — a structural luxury the other conserved quantities don't enjoy.