Continuity Equation

Every account of electricity leans, sooner or later, on a fact so reliable it is easy to forget it is a law: electric charge is never created and never destroyed. But the strong form of the law says more. It is not merely that the universe's total charge is fixed — a bookkeeping that would permit a coulomb to vanish in your lab and reappear on the Moon. Charge is conserved locally: whenever the charge inside any region changes, an exactly compensating amount must physically flow across the region's boundary as a current. No teleportation, no offsetting entries at a distance — every coulomb that leaves does so through the wall, and the ledger balances room by room. Written in the language of the last article, that statement becomes a differential equation binding the two source fields of electromagnetism to each other: $\nabla\cdot\vec J = -\partial\rho/\partial t$ . The continuity equation is arguably the most consequential single line in this chapter, because its steady-state shadow, $\nabla\cdot\vec J = 0$ , is the definition of the magnetostatic regime: currents that neither begin nor end, flowing forever in closed loops. Every formula ahead — Biot–Savart, the divergence and curl of $\vec B$ , Ampère's law — silently assumes it, and the curl-of- $\vec B$ article will need it as a load-bearing ingredient in the middle of a proof. The same three lines of reasoning, pointed at a conductor, answer a deceptively deep question — how long can a blob of charge sit inside copper? — with one of physics' most violent timescales, $10^{-19}$ seconds. And pointed at Ampère's law, they expose the crack through which Maxwell drove the displacement current and discovered light. That climax belongs to the electrodynamics chapter; this article forges the tool it will be built with.

Key takeaways

The continuity equation $\nabla\cdot\vec J + \partial\rho/\partial t = 0$ is the local, differential statement that charge is neither created nor destroyed — outflow from any point depletes the density there, coulomb for coulomb.
It follows from the global balance "charge change = − current out through the boundary" plus the divergence theorem (E0), localized by shrinking the volume.
Steady state kills the time derivative: $\nabla\cdot\vec J = 0$ . Current lines close into loops, Kirchhoff's node law holds, and $J$ adjusts to the cross-section along a wire ( $J_1A_1 = J_2A_2$ ).
$\nabla\cdot\vec J = 0$ is the precondition of magnetostatics: the Biot–Savart law and $\nabla\times\vec B = \mu_0\vec J$ are only self-consistent for divergence-free currents.
Inside an ohmic conductor, continuity + Ohm + Gauss force free charge to decay as $\rho = \rho_0e^{-t/\tau}$ with $\tau = \varepsilon_0/\sigma$ — $10^{-19}$ s in copper, the number the conductor chapter quoted on credit.
In time-varying situations $\nabla\cdot\vec J \neq 0$ (charging capacitors, antennas) — the failure that forces Maxwell's displacement current, a story owned by the electrodynamics chapter's twin of this article.
KEY IDEA: charge conservation is local — the rate of decrease of charge in any volume equals the net current out through its surface, $\nabla\cdot\vec J = -\partial\rho/\partial t$ — and its steady-state form $\nabla\cdot\vec J = 0$ is the foundation stone of magnetostatics.

The core insight

Charge conservation is local: the rate of decrease of charge in any volume equals the net current flowing out through its surface, which in differential form is $\nabla\cdot\vec J = -\partial\rho/\partial t$ .

Where this lives today

Every circuit simulator on Earth runs Derivation B billions of times a day: Kirchhoff's node law is the entire mathematical skeleton of SPICE and its descendants, and its field-theoretic pedigree — steady continuity at a junction — is also its documented failure mode, which is why RF and high-speed digital design graduate from lumped circuits to transmission-line and full-wave solvers exactly when $\partial\rho/\partial t$ at the "nodes" stops being negligible. Semiconductor device physics leans on the relaxation clock: the dielectric relaxation time $\varepsilon/\sigma$ of doped silicon governs how fast carrier populations neutralize injected charge, a design number in everything from photodiodes to DRAM refresh. At the frontier, the continuity equation is the most tested conservation law in existence. Charge non-conservation searches — electron decay would violate it — bound the electron lifetime above $6.6\times10^{28}$ years (Borexino), and Noether's theorem explains the robustness: local charge conservation is welded to the U(1) gauge symmetry that defines electromagnetism itself, so breaking one means demolishing the other. And the template generalizes exactly as promised: the probability current of quantum mechanics, the mass continuity of every weather model, and the energy bookkeeping of the Poynting theorem (this track's conservation-laws chapter) are the same equation with a different noun — learning it once here is learning it everywhere.

Origins · 1 minGustav Kirchhoff and how this was discovered

Gustav Kirchhoff

1824–1887 · Königsberg, Heidelberg, Berlin

Kirchhoff wrote the two circuit laws that bear his name in 1845, while still a 21-year-old student in Königsberg — extending his teacher Neumann's work to arbitrary networks of conductors, with the current law at a node amounting to exactly the steady-state continuity audit of this article. He went on to co-found spectroscopy with Bunsen at Heidelberg (identifying elements by their spectral lines, discovering caesium and rubidium), to state the blackbody radiation problem whose solution by Planck ignited quantum theory, and to put the diffraction theory of light on rigorous footing. His circuit laws remain the most-executed physics in the world — evaluated numerically every time any circuit is simulated — and their known breakdown at high frequency, where charge accumulates and radiates, is precisely the transition from his laws back to the full continuity equation they came from.

History

1747

— Franklin's one-fluid theory asserts charge is conserved — moved, never made — the global law in colonial dress.

1813

— Gauss has the divergence theorem, the tool that will localize conservation laws.

1845

— Kirchhoff, aged 21, publishes the circuit laws; the current law is steady continuity at a node.

1861

— Maxwell, forcing Ampère's law to respect continuity, invents the displacement current — the repair that predicts light.

1918

— Noether ties charge conservation to gauge symmetry: the law is as deep as the theory's structure itself.

2015

— Borexino bounds the electron lifetime above 6.6×10²⁸ years — local charge conservation as the best-tested law in physics.

The physical picture

Picture charge as money in a bank with no printing press and no shredder. Draw any imaginary box — a cubic nanometre inside a wire, a sphere around a whole circuit. The only way the balance inside can change is by money physically crossing the walls. There is no creation, no destruction, and — the local part — no wiring funds between vaults without an armored car driving the road between them. Replace money with charge and the armored cars with current, and the continuity equation is just the audit. The divergence built in E0 is precisely the auditor's tool: $\nabla\cdot\vec J$ at a point measures net outflow per unit volume. Positive divergence means this point is shipping more charge than it receives; the stockpile must be dropping, $\partial\rho/\partial t < 0$ ; and the equation equates the two rates with the sign that makes the books balance.

Figure

The local audit: outflow through the walls is the only way the balance changes

Steady state is the regime where the stockpiles everywhere are frozen: whatever flows into any segment flows out the other side. Then $\nabla\cdot\vec J = 0$ , and the geometry follows immediately — a divergence-free field has field lines with no beginnings and no endings, so steady current must circulate in closed loops. That single sentence is why circuits are circuits, why a battery with one terminal connected does nothing, and why the magnetostatics of the coming articles can treat every source as a loop of one kind or another. One more thing deserves marquee billing before the derivations. Continuity is the electromagnetic instance of a universal template: any locally conserved stuff obeys $\partial_t(\text{density}) + \nabla\cdot(\text{flux}) = 0$ . Mass in a fluid, probability in quantum mechanics, energy in the Poynting theorem — same equation, different cast. Learn to recognize the shape and you can read conservation laws straight off the structure of a theory; deepest of all, Noether's theorem ties this one to the gauge symmetry of electromagnetism itself.

From conservation to the local law

Global charge conservation is an experimental bedrock (charge changes have never been observed, to better than one part in $10^{21}$ in dedicated searches for electron decay). The work of this section is to compress the global statement into a pointwise differential law — the standard maneuver, using the divergence theorem from the vector-calculus chapter, that will be replayed for energy and momentum later in the track.

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

where: $\vec J$ — current density ( $\mathrm{A/m^2}$ ); $\rho$ — charge density ( $\mathrm{C/m^3}$ ). Read it as the auditor's identity: outflow rate per unit volume plus rate of stockpile growth equals zero. Both terms have units $\mathrm{C\,m^{-3}\,s^{-1}}$ . The integral form it came from,

\frac{\mathrm{d}}{\mathrm{d}t}\int_V\rho\,\mathrm{d}\tau = -\oint_S\vec J\cdot\mathrm{d}\vec a \qquad (2)

where: $S$ — the closed boundary of a fixed volume $V$ , $\mathrm{d}\vec a$ outward — is often the more useful tool in practice (it is Kirchhoff's law waiting to happen), and the derivation below is nothing but the two-way traffic between (1) and (2).

Derivation AContinuity from conservation of charge

The goal is equation (1) from the physical statement alone: charge in a fixed region changes only by current crossing its boundary. The strategy is the standard localization ritual — write the balance for an arbitrary finite volume, convert the surface term with the divergence theorem (owned by the vector-calculus chapter and recalled here, not re-proved), and then argue that an integral identity holding for every volume forces the pointwise law. Watch the sign conventions: one honest minus sign, from the outward-pointing area element, carries the whole physics.

Step 1. Write the charge balance for a fixed volume $V$ with boundary $S$ :

\frac{\mathrm{d}Q_{\rm enc}}{\mathrm{d}t} = -\oint_S \vec J\cdot\mathrm{d}\vec a.

Why the minus: $\mathrm{d}\vec a$ points outward, so the integral counts net outflow — and outflow decreases the enclosed charge. This line is the entire physical input; everything after is calculus.

Step 2. Express the enclosed charge as an integral and move the derivative inside:

Q_{\rm enc} = \int_V \rho\,\mathrm{d}\tau \qquad\Longrightarrow\qquad \frac{\mathrm{d}Q_{\rm enc}}{\mathrm{d}t} = \int_V \frac{\partial\rho}{\partial t}\,\mathrm{d}\tau.

Why a partial derivative appears: the volume is fixed in space, so the only time dependence under the integral is $\rho$ 's own.

Step 3. Convert the boundary flux with the divergence theorem:

\oint_S \vec J\cdot\mathrm{d}\vec a = \int_V \nabla\cdot\vec J\,\mathrm{d}\tau.

Step 4. Combine into one integral:

\int_V\left(\frac{\partial\rho}{\partial t} + \nabla\cdot\vec J\right)\mathrm{d}\tau = 0.

Step 5. Localize. The volume $V$ was arbitrary — every box, at every location, of every size. An integrand whose integral vanishes over every volume must vanish pointwise (were it positive in some small ball, the integral over that ball would be positive). Hence

\nabla\cdot\vec J = -\frac{\partial\rho}{\partial t}.

Why localization is the physical point, not a technicality: the differential law is strictly stronger than global conservation. It outlaws the teleport scenario — total charge constant while a coulomb jumps between distant points — because at the departure point the books would read $\partial\rho/\partial t < 0$ with no outflow to blame. *Verified: matches Griffiths §5.1.2 (Eq. 5.29) and Jackson §6.1; dimensional check $\mathrm{A/m^3} = \mathrm{C\,m^{-3}s^{-1}}$ ✓; integrating back over all space with $\vec J \to 0$ at infinity recovers $\mathrm{d}Q_{\rm total}/\mathrm{d}t = 0$ , the global law.*

Steady currents: the magnetostatic contract

Magnetostatics studies fields of currents that have settled down: nothing depends on time, so $\partial\rho/\partial t = 0$ everywhere and continuity collapses to

\nabla\cdot\vec J = 0 \qquad (3)

where: the statement holds pointwise throughout the steady flow. This is the contract every article ahead signs. It is worth deriving its two most-used consequences now, because they are quoted constantly and their pedigree is exactly three lines each — and because the deepest consequence (that $\nabla\times\vec B = \mu_0\vec J$ requires (3) to be a consistent equation, since the divergence of any curl vanishes identically) is a check this chapter will cash, with interest, in Curl of B.

Figure

Two closed-surface audits: the circuit node and the tapering wire

Go deeper · Derivation B · 1 minWhat divergence-free current forces — Kirchhoff's law and the taper theorem

Derivation BWhat divergence-free current forces — Kirchhoff's law and the taper theorem

The goal is the two working consequences of $\nabla\cdot\vec J = 0$ : the node law of circuit theory, and the way current density redistributes when a wire changes thickness. The strategy is one tool used twice — the integral form $\oint\vec J\cdot\mathrm{d}\vec a = 0$ , applied to two differently chosen closed surfaces. The lesson to absorb is the method: in steady state you may audit any closed surface you like, and each clever choice yields a theorem.

Step 1. Establish the tool. Integrate (3) over any volume and apply the divergence theorem:

\oint_S \vec J\cdot\mathrm{d}\vec a = 0 \qquad\text{for every closed surface } S.

Why this is the whole game: net current through any closed surface vanishes — in equals out, no exceptions, no charge pile-up anywhere.

Step 2. Choose $S$ = a small ball enclosing a circuit node where $n$ wires meet. The flux integral picks up only the wire cross-sections, each contributing its current with a sign (out positive):

\sum_k I_k = 0 \qquad\text{— Kirchhoff's current law.}

Why this matters beyond circuits: the node law is not an axiom of circuit theory but a theorem of field theory — steady-state continuity applied to a junction. When circuits run fast enough that charge does pile up (antennas, transmission lines at GHz), the law fails exactly on schedule, and the fix is the full equation (1).

Step 3. Choose $S$ = a pillbox around a section of wire whose cross-section tapers from $A_1$ to $A_2$ , with caps perpendicular to the flow. In equals out:

J_1A_1 = J_2A_2 = I.

Why this is the taper theorem: the scalar current is the same through every cross-section of an unbranched wire — that is what "a wire carries 3 A" even means — while the density rearranges inversely with area. A wire narrowing from $2\ \mathrm{mm^2}$ to $0.5\ \mathrm{mm^2}$ at $I = 3\ \mathrm{A}$ runs $J$ from $1.5\times10^{6}$ up to $6.0\times10^{6}\ \mathrm{A/m^2}$ — with quadrupled local heating per volume, which is why undersized sections of a conductor burn first. *Verified: both results are the standard consequences in Griffiths §5.1.3/§7.1 register; the node law reproduces $\sum I_{\rm in} = \sum I_{\rm out}$ ; the taper numbers recomputed for this edition.*

Steady state is an idealization, and one should know how fast real conductors reach it. Drop a blob of free charge into the interior of a conductor: how long does it take to vanish to the surface? The conductor chapter (E1) quoted the astonishing answer on credit — copper clears its interior in $\sim10^{-19}$ seconds. Continuity is the missing ingredient that lets us finally derive it.

Go deeper · Derivation C · 2 minCharge relaxation — why conductor interiors are empty, on a clock

Derivation CCharge relaxation — why conductor interiors are empty, on a clock

The goal is the decay law for free charge density inside an ohmic conductor, $\rho(t) = \rho_0\,e^{-t/\tau}$ with $\tau = \varepsilon_0/\sigma$ — the timescale that justified the E1 doctrine "field inside a conductor is zero, all charge sits on the surface." The strategy is a three-law pincer: Ohm's law relates $\vec J$ to $\vec E$ , Gauss's law relates $\vec E$ back to $\rho$ , and continuity closes the loop into a differential equation for $\rho$ alone. The honest caveat at the end is part of the derivation: for metals, the answer comes out faster than the model's own validity, and knowing why is worth as much as the formula.

Step 1. In the conductor's bulk, take the constitutive law $\vec J = \sigma\vec E$ (uniform, linear, ohmic — the microscopic Ohm's law named in the previous article and owned in full by the electrodynamics chapter).

Step 2. Take the divergence and feed it Gauss's law $\nabla\cdot\vec E = \rho/\varepsilon_0$ :

\nabla\cdot\vec J = \sigma\,\nabla\cdot\vec E = \frac{\sigma}{\varepsilon_0}\,\rho.

Why divergence is the right move: continuity speaks in divergences, so we translate Ohm's law into that dialect — and Gauss's law happens to convert the result into $\rho$ itself, closing the system.

Step 3. Insert into continuity (1):

\frac{\partial\rho}{\partial t} = -\nabla\cdot\vec J = -\frac{\sigma}{\varepsilon_0}\,\rho.

Step 4. Solve the first-order decay:

\rho(t) = \rho_0\,e^{-t/\tau}, \qquad \tau = \frac{\varepsilon_0}{\sigma} \qquad (4)

Numbers across the conductivity ladder: copper ( $\sigma = 6.0\times10^{7}\ \mathrm{S/m}$ ) gives $\tau = 1.5\times10^{-19}\ \mathrm{s}$ ; seawater ( $\sigma \approx 5\ \mathrm{S/m}$ ) gives $1.8\times10^{-12}\ \mathrm{s}$ ; a decent glass ( $\sigma \sim 10^{-12}\ \mathrm{S/m}$ ) holds charge for $\sim 9$ seconds. The material's own properties set how fast "electrostatic equilibrium" and "steady state" become legitimate words. Why the copper number must not be taken literally: $10^{-19}$ s is shorter than the electron collision time ( $\sim 2.5\times10^{-14}$ s in copper), and $\vec J = \sigma\vec E$ is a collision-averaged law — it has no right to operate on timescales its own mechanism can't resolve. The honest reading: the ohmic model destroys interior charge faster than its own validity clock ticks, so the true relaxation is set by other physics (plasma oscillations at $\sim10^{-16}$ s in metals); either way, far faster than anything electrostatic, so the E1 doctrine stands. For poor conductors, where $\tau$ is long and slow, equation (4) is quantitatively trustworthy — it is the "dielectric relaxation time" of semiconductor device design. *Verified: derivation matches Jackson §7.7 register and Griffiths Prob. 7.42 territory; $\tau = \varepsilon_0/\sigma$ dimensionally $(\mathrm{F/m})/(\mathrm{S/m}) = \mathrm{s}$ ✓; copper value $8.85\times10^{-12}/6.0\times10^{7} = 1.5\times10^{-19}$ s recomputed, agreeing with the figure quoted by the E1 conductors article.*

What breaks when things change: the Maxwell preview

Everything above ran on $\partial\rho/\partial t = 0$ . Switch on time dependence and $\nabla\cdot\vec J = -\partial\rho/\partial t \neq 0$ wherever charge accumulates — at the plates of a charging capacitor, at the tips of an antenna where charge sloshes to a stop twice per cycle. This is not a nuisance case; it is where the continuity equation earns its keep in the full theory. The magnetostatic law this chapter is building toward, $\nabla\times\vec B = \mu_0\vec J$ , has a fatal allergy to such currents: the divergence of any curl is identically zero, so the law forces $\nabla\cdot\vec J = 0$ and flatly contradicts a charging capacitor. Maxwell's cure — adding $\mu_0\varepsilon_0\,\partial\vec E/\partial t$ so that the law's divergence reproduces exactly the continuity equation — is the birth of the displacement current and, a few articles later, of electromagnetic waves. The full derivation is owned by the electrodynamics chapter (Displacement Current, and that chapter's own continuity article, which revisits this equation as an identity of the completed Maxwell equations rather than an independent postulate). Here, the takeaway is directional: charge conservation is the senior law — when it collided with Ampère's law, it was Ampère's law that moved.

Limiting cases

Case	Result	Meaning
steady state	$\nabla\cdot\vec J = 0$	current in closed loops — the magnetostatic contract
circuit node, steady	$\sum_k I_k = 0$	Kirchhoff's law = continuity on a junction
unbranched wire	$J_1A_1 = J_2A_2$	same $I$ through every section; $J$ inversely with area
isolated system	$\mathrm{d}Q_{\rm tot}/\mathrm{d}t = 0$	$\vec J \to 0$ on the far boundary — global conservation recovered
one-dimensional flow	$\partial J_x/\partial x + \partial\rho/\partial t = 0$	thin-wire/transmission-line form
ohmic conductor	$\rho = \rho_0e^{-t/\tau}$ , $\tau = \varepsilon_0/\sigma$	interiors self-empty; copper $10^{-19}$ s, glass seconds
time-varying sources	$\nabla\cdot\vec J = -\partial\rho/\partial t \neq 0$	capacitor plates, antenna tips — Maxwell's displacement current required (E6)

Common trapdiv J = 0 is a regime, not a law of nature

The single most common misuse of this article is promoting the steady-state condition to an always-true identity. $\nabla\cdot\vec J = 0$ holds only when nothing depends on time. During any transient — a charging capacitor, a discharging cloud, an antenna, the first nanosecond after a switch closes — current dead-ends somewhere and charge piles up: $\nabla\cdot\vec J = -\partial\rho/\partial t \neq 0$ . Forgetting this is precisely the error frozen into pre-Maxwell Ampère's law, and it produces real wrong answers (apply Kirchhoff's node law to an antenna feed and you will conclude antennas cannot work). The full continuity equation is the law; the divergence-free version is a contract you must check you are entitled to sign. Related fine print: the fixed-volume assumption in the derivation is what makes $\mathrm{d}/\mathrm{d}t \to \partial/\partial t$ legal — for moving or deforming volumes, extra flux terms appear.

Worked examples

1 · Reading a divergence (confidence builder)

In some region a steady diagnostic reports $\nabla\cdot\vec J = +5.0\ \mathrm{C\,m^{-3}s^{-1}}$ at a point $P$ . (a) What is happening to the charge density at $P$ ? (b) A ball of radius $2.0\ \mathrm{mm}$ centred on $P$ has approximately this divergence throughout its volume; what net current leaves it?

Step 1. Continuity, solved for the density rate: $\partial\rho/\partial t = -\nabla\cdot\vec J = \boxed{-5.0\ \mathrm{C\,m^{-3}s^{-1}}}$ — charge at $P$ is draining: more current exits than enters. (And note the situation is by definition not steady in the strict sense: someone's "steady diagnostic" caught a transient.)

Step 2. Net current out of the ball is the volume integral of the divergence (divergence theorem run forward): $I_{\rm out} = (\nabla\cdot\vec J)\times\tfrac43\pi r^3 = 5.0\times\tfrac43\pi(2.0\times10^{-3})^3 = \boxed{1.7\times10^{-7}\ \mathrm{A}}$ .

Step 3. Sanity: the enclosed charge falls at the same $1.7\times10^{-7}\ \mathrm{C/s}$ — the two bookkeeping columns of equation (2), equal and opposite ✓.

2 · The tapering busbar and the vanishing blob (exam level)

A copper busbar carries a steady $I = 3.0\ \mathrm{A}$ through a section that tapers from $A_1 = 2.0\ \mathrm{mm^2}$ to $A_2 = 0.50\ \mathrm{mm^2}$ . (a) Find $J$ in each section and the ratio of local ohmic heating densities. (b) During assembly, a static-charged tool deposits a blob of free charge in the bar's interior. Estimate how long the blob persists, and contrast with the same accident in a glass insulator ( $\sigma \approx 10^{-12}\ \mathrm{S/m}$ ).

Step 1. Taper theorem (Derivation B): $J_1 = I/A_1 = 3.0/2.0\times10^{-6} = 1.5\times10^{6}\ \mathrm{A/m^2}$ ; $J_2 = I/A_2 = \boxed{6.0\times10^{6}\ \mathrm{A/m^2}}$ — same 3.0 A through both sections, density up fourfold in the neck.

Step 2. Heating density scales as $J^2/\sigma$ (ohmic dissipation per volume), so the neck runs $(J_2/J_1)^2 = \boxed{16\times}$ hotter per unit volume — the physics behind fuse design and the reason damaged, thinned conductors fail exactly at the damage.

Step 3. The blob: relaxation with $\tau = \varepsilon_0/\sigma$ . Copper, $\sigma = 6.0\times10^{7}\ \mathrm{S/m}$ : $\tau = 8.85\times10^{-12}/6.0\times10^{7} = \boxed{1.5\times10^{-19}\ \mathrm{s}}$ — gone before any instrument could notice, with the honest caveat of Derivation C (the ohmic model overruns its own validity; the true clearing is "merely" $\sim10^{-16}$ s).

Step 4. The glass tool-mark: $\tau = 8.85\times10^{-12}/10^{-12} = \boxed{8.9\ \mathrm{s}}$ — a static charge on glass genuinely sits there for seconds, which is why insulators spark, cling and shock while metals never store interior charge. Same law, twenty orders of magnitude apart.

Going further

References. Griffiths, Introduction to Electrodynamics, 4th ed., §5.1.2 and §7.3.4 · Jackson, Classical Electrodynamics, 3rd ed., §6.1 · Purcell & Morin, Electricity and Magnetism, 3rd ed., §4.2 · Zangwill, Modern Electrodynamics, §2.1.

Provenance. Equations are standard textbook results. A verifier should: (1) reproduce the divergence-theorem derivation of continuity (Griffiths §5.1.2); (2) verify the displacement-current consistency argument (Griffiths §7.3.4, Jackson §6.1); (3) confirm the charge relaxation result $\tau=\varepsilon_0/\sigma$ (Jackson §7.7). Constants: $\mu_0=1.25663706\times10^{-6}\,$ T·m/A and $\varepsilon_0=8.8541878\times10^{-12}\,$ F/m (CODATA). Cross-check covariant form $\partial_\mu J^\mu=0$ against Jackson §11.9. (In this edition the displacement-current derivation and the covariant four-current form are deferred by name to their owning articles — Displacement Current and the electrodynamics chapter's continuity twin, and Four-Current in the relativistic tier; this article previews both.) Worked numbers recomputed: ball outflow $1.7\times10^{-7}$ A; taper $1.5\times10^{6} \to 6.0\times10^{6}$ A/m², heating ratio 16; $\tau$ = $1.5\times10^{-19}$ s (Cu), $1.8\times10^{-12}$ s (seawater), $8.9$ s (glass at $10^{-12}$ S/m).

Builds on: Electric Current and Current Density · Divergence theorem

Leads to: Magnetic Field of Steady Current Configurations · Curl of B · Ampère's Law

Is this the same continuity equation I met in fluid mechanics?

The identical mathematical object with a different conserved noun: fluid mechanics writes $\partial_t\rho_m + \nabla\cdot(\rho_m\vec v) = 0$ for mass density, and the electromagnetic version substitutes charge density and $\vec J = \rho_{\rm mob}\vec v_d$ . Both are instances of the universal template "density rate + flux divergence = 0" that any locally conserved quantity must obey — probability in quantum mechanics and energy in the Poynting theorem complete the classic quartet. One physical difference worth noting: mass conservation in fluids is an excellent approximation (violated in relativity by $E = mc^2$ bookkeeping), while charge conservation, as far as every experiment can tell, is exact.

Charge is conserved globally — why insist on the stronger local version?

Because global conservation alone is compatible with nonsense: a coulomb vanishing in Delhi while another materializes in Geneva keeps the universe's total fixed. Locality outlaws it — and relativity demands the outlawing, since distant simultaneous events are not even unambiguously simultaneous for all observers; a "conservation law" that needed them would be frame-dependent gibberish. The local law is also what has been tested: every charge ever tracked moved continuously through space. Formally, the statement packages into the four-divergence $\partial_\mu J^\mu = 0$ , manifestly the same in every inertial frame — machinery this track builds properly in the relativistic tier's Four-Current article.

Does the continuity equation follow from Maxwell's equations, or is it an independent law?

Both, in a historically layered way. Take the completed Maxwell equations, and continuity falls out as an identity (divergence of the Ampère–Maxwell law plus Gauss's law) — the theory has conservation built into its bones. But the arrow of logic ran the other way: Maxwell imposed charge conservation on Ampère's law, which without the displacement current contradicts it, and the repair term is what made the equations complete. So at this chapter's level, treat continuity as the senior, independently-tested law and magnetostatics as valid exactly where its steady-state shadow holds; when you reach the full equations, you can re-derive your axiom as a theorem and admire the closed loop.

If div J = 0 in steady state, how does a capacitor in a DC circuit ever charge up?

It charges during the transient, which is precisely not steady state. While current flows onto a plate and stops, $\nabla\cdot\vec J < 0$ there and $\rho$ grows — the full continuity equation in action. As the plate charges, its growing field opposes the supply until current ceases; the final DC "steady state" of a capacitor circuit is $I = 0$ through the capacitor branch, perfectly consistent with $\nabla\cdot\vec J = 0$ . The interesting intermediate regime — AC, where the capacitor passes current forever without ever passing charge through the gap — is exactly where the displacement current becomes indispensable, and it is deferred to the electrodynamics chapter with a clear conscience.

How can charge relaxation in copper take 10⁻¹⁹ s when electrons drift at fractions of a millimetre per second?

Because relaxation is not a delivery problem — no electron has to travel to the surface. The interior blob's field pushes on the entire sea of conduction electrons at once, and each needs to shift only an imperceptible distance for the net density to redistribute; it is the same "already-full hose" logic that reconciles slow drift with instant signals. The legitimate criticism of the number is different (Derivation C): $10^{-19}$ s is faster than the collision time underlying Ohm's law, so the model has outrun its own assumptions, and the honest metal answer is set by plasma-oscillation physics at $\sim10^{-16}$ s. Either way: interior charge in a metal is gone on timescales no electrostatic experiment can see.

What exactly breaks in Kirchhoff's current law at high frequency?

The assumption that no charge accumulates at the node. At 50 Hz, the charge sloshing on a junction's stray capacitance is utterly negligible and the node law is exact for all purposes. At gigahertz, a millimetre of conductor is an appreciable fraction of a wavelength: charge piles up and drains twice per cycle ( $\partial\rho/\partial t \neq 0$ ), conductors radiate, and "the current into the node" differs measurably from "the current out." The engineering response is layered: add explicit capacitances to the lumped model, then switch to transmission-line theory, then to full-field solvers. Each layer is the continuity equation, honored at progressively finer resolution.

Noether's theorem connects charge conservation to a symmetry — which one, and what does that buy?

The global U(1) phase symmetry of the charged fields — rotating the quantum phase of every charged particle's field by the same constant angle changes nothing observable, and Noether's machinery converts exactly that invariance into a conserved current satisfying $\partial_\mu J^\mu = 0$ . Promoting the symmetry from global to local (angle varying point to point) is what forces the electromagnetic field itself into existence — gauge theory's founding move. What it buys: an explanation of why charge conservation is exact rather than approximate. Symmetries of this structural kind cannot be slightly broken without dismantling the theory, which is why experimentalists hunting charge non-conservation are really stress-testing the foundations of gauge theory — and finding them, so far, flawless.