Quasistatic Approximation

Here is an uncomfortable fact about the chapter you have just finished: almost nobody solves the full Maxwell equations. The power engineer designing a 50 Hz transformer, the audio designer laying out an amplifier, the physicist winding a magnet coil — all of them use Coulomb-style and Biot–Savart-style field laws, updated instant by instant, as if the last two hundred pages of electrodynamics had never happened. And they get the right answers to parts per billion. Are they cheating? They are exploiting the quasistatic approximation, and this article is its owner's manual. The full equations couple $\mathbf{E}$ and $\mathbf{B}$ through two time-derivative terms — Faraday's $\partial\mathbf{B}/\partial t$ and Maxwell's $\partial\mathbf{D}/\partial t$ — and that closed feedback loop is what lets fields detach and propagate. But the loop takes a light-transit-time $L/c$ to close across an apparatus of size $L$ . Drive the system slowly compared with that, and the loop never closes: the fields track their sources "instantaneously," corrections entering only at order $(L/\lambda)^2$ . Keep one coupling term and drop the other, and you get two complementary half-worlds: electroquasistatics, the world of capacitors and charge, and magnetoquasistatics, the world of coils, transformers, and eddy currents. Knowing which term to keep, why the choice is dictated by two material clocks whose geometric mean is always the light-transit time, and when the whole licence expires — that is one of the most transferable skills in electromagnetism. It is also, as the final sections show, the secret constitution of circuit theory itself.

Key takeaways

Quasistatics is a controlled expansion in $L/\lambda = L/cT$ : when the apparatus is much smaller than a wavelength, retardation is negligible and static-form field laws apply with time as a parameter; corrections enter at $(L/\lambda)^2$ .
Drop one coupling term, never both. Electroquasistatics (EQS) keeps $\partial\mathbf{D}/\partial t$ , drops induction: capacitors, electrometers, insulators. Magnetoquasistatics (MQS) keeps $\partial\mathbf{B}/\partial t$ , drops displacement current: coils, transformers, conductors.
Two material clocks decide the branch: charge relaxation $\tau_e=\varepsilon/\sigma$ and magnetic diffusion $\tau_m=\mu\sigma L^2$ , tied by the identity $\sqrt{\tau_e\tau_m}=L/c$ . Fast-relaxing media (conductors) are MQS; charge-holding media (insulators) are EQS.
Circuit theory — Kirchhoff's laws, lumped L and C — is quasistatic Maxwell: KCL is continuity with negligible stray charging, KVL is Faraday with flux confined to the inductors.
The licence expires when $L/\lambda\gtrsim1$ : retardation, phase slip, and radiation appear, and you must graduate to retarded potentials (E9).
KEY IDEA: quasistatics keeps the static field laws and restores exactly one time-derivative coupling — the choice dictated by whether charge relaxes faster or slower than the drive.

The core insight

Compare the time for light to cross the system, $L/c$ , with the timescale of variation $\tau$ . If $L/c \ll \tau$ , retardation is negligible and you may use static-form field equations with slow time dependence — keeping either the displacement current (EQS) or induction (MQS), but never needing both.

Where this lives today

Every electromagnetic simulation suite ships quasistatics as first-class products: field solvers offer explicit electrostatic, EQS, MQS, and full-wave modules, and choosing correctly is routinely a thousandfold saving in compute — a wireless-charging pad (MQS coupling at 100–300 kHz, $L/\lambda\sim10^{-4}$ ) meshed as a full-wave problem is an amateur's mistake that turns minutes into days. MRI machines are a quasistatic layer cake: the static magnet, kHz gradient coils designed in MQS (their eddy-current interactions with the cryostat are the hard part — next article's physics), and only the MHz RF coil requiring wave treatment. The sharpest modern edge is the discipline of electrically small design. The signal-integrity engineer's $\lambda/10$ rule, the EMC engineer's antenna-inadvertent-loop audits, the power-electronics designer racing switching edges (a 1 ns edge has spectral content to ~GHz — its $\lambda$ is 30 cm, and suddenly a PCB is not small) — all are applications of this article's single parameter. The approximation even guards the quantum frontier: superconducting-qubit couplers and ion-trap electrode stacks are designed EQS/MQS first, with full-wave corrections only where the error budget demands, because at millikelvin the designer's intuition, like Faraday's, is quasistatic.

The physical picture

Turn the dial on a signal generator feeding a coil a metre across. Any change in current announces itself outward at the speed of light — and has finished announcing in about three nanoseconds. Against the milliseconds of your dial-turning, the field's messenger service is instantaneous for every practical purpose: the magnetic field everywhere tracks the present current, Biot–Savart style, as though magnetostatics simply agreed to be re-evaluated at each tick of the clock. That instant tracking, plus the back-EMF that Faraday's law extracts from the changing flux, is magnetoquasistatics — the operating system of every inductor, motor, and transformer on Earth. The deeper picture is a feedback loop. A changing $\mathbf{E}$ sources $\mathbf{B}$ (the displacement current); a changing $\mathbf{B}$ sources $\mathbf{E}$ (Faraday). Run the loop once and you get a small correction to the static field. Run it endlessly, each field regenerating the other, and you get a self-sustaining wave — that was the crown argument of Maxwell's Equations in Vacuum, recalled here by name. Quasistatics is what happens when the loop is too slow to close: across an apparatus of size $L$ , one lap of the loop costs a light-transit time $L/c$ , and if the drive changes over a much longer time $T$ , the second half-step of the loop arrives as a correction of order $(L/cT)^2 = (L/\lambda)^2$ — usually absurdly small. You keep the first half-step (one coupling term) and truncate.

Figure

The Maxwell feedback loop — quasistatics keeps one half-step; radiation is the loop running forever

Which half-step do you keep? Usually the terminals tell you. An open, high-impedance device that stores charge — capacitor plates, an electrometer, a high-voltage bushing — is electroquasistatic: $\mathbf{E}$ stays essentially curl-free and derives from a scalar potential, while the displacement current closes its circuit. A closed, low-impedance device that carries current and stores flux — a coil, a transformer, a conducting slab bathed in changing field — is magnetoquasistatic: induction is the whole story, and the displacement current is the negligible garnish (the $\varepsilon\omega/\sigma$ regime map of The Displacement Current, recalled by name: for copper at 1 Hz the ratio is $\sim10^{-18}$ ). What is genuinely satisfying is that this terminal-reading rule is not folklore — it falls out of a two-clock race derived below, and the clocks' geometric mean is always, exactly, the light-transit time. The stakes of this article are also historical honesty: quasistatics is where all of electromagnetism lived before Maxwell. Faraday's induction, Henry's self-inductance, the telegraph, the dynamo — all MQS discoveries, made and mastered with no displacement current and no waves. The approximation is not a degraded modern convenience; it is the ancestral home of the subject, now equipped with an error bar.

The mathematics of slow

The claims — corrections at $(L/\lambda)^2$ , exactly two consistent truncations, a two-clock selector — deserve proofs at the register of this track.

Derivation AThe quasistatic ordering of Maxwell's equations

We want to establish that "slow" systems admit exactly two self-consistent simplifications of Maxwell's equations, each accurate to relative order $(L/\lambda)^2$ , and to see precisely which term each one drops. The strategy is nondimensionalization: measure lengths in units of the system size $L$ , times in units of the drive timescale $T$ , and let the equations themselves display the small parameter. Watch for one subtlety — the ratio of field scales $E_0/B_0$ is not ours to choose; each branch chooses it, and that choice is what distinguishes EQS from MQS.

Step 1. Write the two curl equations in vacuum-form (media add $\varepsilon,\mu$ decorations without changing the logic), with $\mathbf{r}=L\tilde{\mathbf{r}}$ , $t=T\tilde t$ , $\mathbf{E}=E_0\tilde{\mathbf{E}}$ , $\mathbf{B}=B_0\tilde{\mathbf{B}}$ :

\tilde\nabla\times\tilde{\mathbf{E}} = -\frac{LB_0}{TE_0}\,\frac{\partial\tilde{\mathbf{B}}}{\partial\tilde t}, \qquad \tilde\nabla\times\tilde{\mathbf{B}} = \mu_0 \tilde{\mathbf J}\,(\text{source term}) + \frac{L E_0}{c^2 T B_0}\,\frac{\partial\tilde{\mathbf{E}}}{\partial\tilde t}.

Every symbol with a tilde is of order one; the physics sits in the two dimensionless prefactors.

Step 2. Define $\epsilon \equiv L/cT = L/\lambda$ (up to the factor $2\pi$ that never changes a verdict) and the field-ratio parameter $\beta\equiv E_0/cB_0$ . The prefactors become $\epsilon/\beta$ on the Faraday side and $\epsilon\beta$ on the Ampère side, and their product is $\epsilon^2$ — fixed by geometry and drive, no matter how the field scales arrange themselves. This is the loop-closure cost from the physical picture, now an identity: whatever you do, running both couplings costs $(L/\lambda)^2$ .

Step 3. The MQS branch. Suppose the system is source-dominated by currents: $B_0$ is set by Biot–Savart, $B_0\sim\mu_0 J L$ , and the only electric field is the one induction manufactures. Then $E_0$ must balance the Faraday equation: $E_0/L \sim B_0/T$ , i.e. $\beta = E_0/cB_0 \sim L/cT = \epsilon$ . The Faraday prefactor $\epsilon/\beta\sim1$ — induction is a first-class citizen — while the displacement prefactor $\epsilon\beta\sim\epsilon^2$ : droppable. The consistent truncation is $\nabla\times\mathbf{E}=-\partial\mathbf{B}/\partial t$ with $\nabla\times\mathbf{B}=\mu_0\mathbf{J}$ , magnetic field computed instantaneously from the present currents.

Step 4. The EQS branch. Suppose instead charge dominates: $E_0$ is set by Coulomb, and the only magnetic field is what the displacement and conduction currents manufacture, $B_0 \sim E_0 L/c^2T$ , i.e. $\beta\sim1/\epsilon$ . Now the displacement prefactor $\epsilon\beta\sim1$ — the charging current matters — while the Faraday prefactor $\epsilon/\beta\sim\epsilon^2$ : droppable. The consistent truncation is $\nabla\times\mathbf{E}=0$ (scalar potential survives!) with the Ampère–Maxwell law intact. Why "never both": dropping both couplings forces $\beta$ to satisfy $\epsilon/\beta\ll1$ and $\epsilon\beta\ll1$ with neither field generated by the other — that is true statics, where nothing induces and nothing charges. A system with any inductive or capacitive behavior at all lives on one branch or the other; pretending otherwise (the classic exam disaster) predicts transformers that do not transform.

Step 5. Read off the error bar. In either branch the abandoned term re-enters as a source for a first correction field, smaller by $\epsilon^2$ ; iterating generates a power series in $\epsilon^2=(L/\lambda)^2$ whose infinite sum is full electrodynamics. Quasistatics is the first Padé of light. Verified: the two branches and the $\epsilon^2$ error estimate match Zangwill Ch. 14 and the canonical Haus–Melcher treatment; setting $\partial_t\to0$ in either branch recovers electro/magnetostatics exactly, and retaining both prefactors at order one ( $\beta\sim1$ , $\epsilon\sim1$ ) reproduces the wave regime of Maxwell's Equations in Vacuum.

Go deeper · Derivation B · 1 minThe induced-field correction, computed for a real capacitor

Derivation BThe induced-field correction, computed for a real capacitor

The ordering argument promised that the term each branch drops feeds back at relative order $(L/\lambda)^2$ . Here we compute that correction concretely — for the EQS workhorse, a circular parallel-plate capacitor of radius $R$ driven at angular frequency $\omega$ — because seeing the number for a real device converts "negligible" from faith to arithmetic. The route: start from the uniform oscillating field, let the displacement current build the ring magnetic field, then let that field's own time dependence induce a correction to $\mathbf{E}$ , and compare.

Step 1. Zeroth order: $E_z^{(0)}(t)=E_0\cos\omega t$ , uniform in the gap — the instantaneously-updated electrostatic solution.

Step 2. First order in the loop: the displacement current sources a ring field. By the Ampère–Maxwell law on a circle of radius $s$ (the gap-field calculation of The Displacement Current, recalled): $2\pi s\,B_\varphi = \mu_0\varepsilon_0\,\pi s^2\,\partial_t E_z^{(0)}$ , so $B_\varphi = -(\omega s/2c^2)E_0\sin\omega t$ .

Step 3. Second half-step: this oscillating $B_\varphi$ threads Faraday loops and induces a correction $E^{(1)}_z$ . Faraday's law on a rectangle with one leg on the axis and one at radius $s$ : $E^{(1)}_z(s)-E^{(1)}_z(0) = -\int_0^s \partial_t B_\varphi\,\mathrm{d}s' = \frac{\omega^2 s^2}{4c^2}\,E_0\cos\omega t$ .

Step 4. The relative size at the rim:

\left|\frac{E^{(1)}}{E^{(0)}}\right| = \frac{\omega^2R^2}{4c^2} = \frac{1}{4}\left(\frac{\omega R}{c}\right)^{2},

where $\omega R/c$ is exactly the $L/\lambda$ parameter of the ordering derivation (with $2\pi$ 's absorbed). For a bench capacitor, $R=5$ cm driven at 1 MHz: $\omega R/c = 1.05\times10^{-3}$ , correction $2.7\times10^{-7}$ — a quarter part per million of non-uniformity. Why you should still care about the form: iterate the loop and you generate the series $1 - (\omega s/2c)^2/1!^2 + (\omega s/2c)^4/2!^2-\dots = J_0(\omega s/c)$ — push the frequency until $\omega R/c\sim2.4$ (the first Bessel zero) and the "capacitor" has become a resonant cavity. The quasistatic error term is the embryo of microwave engineering. Verified: the $J_0$ resummation matches the standard cavity-mode result (and Feynman's classic iteration, Lectures II-23); the first correction's sign — field less reduced on axis, more at the rim — agrees with the exact $J_0$ profile's curvature, and the numbers above recompute to $2.7\times10^{-7}$ at $f=1$ MHz, $R=0.05$ m.

Go deeper · Derivation C · 2 minTwo material clocks and one identity - which quasistatic branch a system takes

Derivation CTwo material clocks and one identity - which quasistatic branch a system takes

The branch choice was framed by field sources ("charge-dominated vs current-dominated"). For real materials there is a sharper, quantitative selector built from two relaxation clocks, and an identity between them that is one of the quiet gems of the subject (Haus–Melcher). We derive the clocks, the identity, and the decision rule.

Step 1. The electric clock. In a medium of conductivity $\sigma$ and permittivity $\varepsilon$ , unbalanced charge decays as $e^{-t/\tau_e}$ with $\tau_e=\varepsilon/\sigma$ — the charge-relaxation derivation owned by the E4 Continuity Equation article, recalled by name, not re-derived. $\tau_e$ answers: can the medium hold charge over one drive period?

Step 2. The magnetic clock. In the MQS limit inside a conductor, combining Faraday, Ampère ( $\nabla\times\mathbf{B}=\mu\sigma\mathbf{E}$ ), and the curl-of-curl identity gives the magnetic diffusion equation $\partial_t\mathbf{B}=\nabla^2\mathbf{B}/\mu\sigma$ — derived properly in Eddy Currents, next article, deferred by name. Dimensional reading: field structure of scale $L$ diffuses in $\tau_m = \mu\sigma L^2$ . $\tau_m$ answers: can the magnetic field rearrange through the medium in one period?

Step 3. The identity. Multiply:

\tau_e\,\tau_m \;=\; \frac{\varepsilon}{\sigma}\cdot\mu\sigma L^2 \;=\; \varepsilon\mu\,L^2 \;=\; \left(\frac{L}{v}\right)^{2},

where $v=1/\sqrt{\varepsilon\mu}$ is the medium's light speed. The conductivity cancels: the geometric mean of the two clocks is always the light-transit time $\tau_{em}=L/v$ , for every material. The two clocks sit symmetrically astride the one timescale quasistatics cares about.

Step 4. The decision rule. Quasistatic validity demands the drive period $T\gg\tau_{em}$ . Because $\tau_{em}$ is the geometric mean, the clocks must straddle it: either $\tau_e\ll\tau_{em}\ll\tau_m$ (good conductor — copper at 1 cm: $\tau_e=1.5\times10^{-19}$ s, $\tau_m=7.5$ ms, $\tau_{em}=33$ ps, mean checks: $\sqrt{1.5\times10^{-19}\times7.5\times10^{-3}}=3.3\times10^{-11}$ s ✓) or $\tau_m\ll\tau_{em}\ll\tau_e$ (insulator). Now place $T$ : any valid quasistatic drive has $T\gg\tau_{em}$ , hence in a conductor $T\gg\tau_e$ automatically — charges relax to the surfaces within every cycle, the bulk carries conduction current only, and the system is MQS. In an insulator $T$ can exceed $\tau_{em}$ while remaining far below $\tau_e$ — charge stays where you put it, currents are capacitive, and the system is EQS. The material decides; geometry only sets the stakes.

Step 5. Connect to the regime map of The Displacement Current: its conductor criterion $\varepsilon\omega/\sigma\ll1$ is precisely $\tau_e\ll T/2\pi$ — the same clock, read from the current ratio instead of the charge budget. One selector, two dialects. Verified: the identity $\tau_e\tau_m=(L/v)^2$ is algebraic and checked numerically above; the classification reproduces Haus & Melcher's canonical EQS/MQS taxonomy, and the copper numbers use $\sigma=5.96\times10^7$ S/m throughout, consistent with every earlier article in the track.

Figure

The two-clock selector — the light-transit time is always the geometric mean, and the drive period lands in a regime

The constitution of circuit theory

The most consequential quasistatic system is the humble circuit, and it is worth seeing exactly which approximation licenses each of Kirchhoff's laws — because this is the bridge between the field theory you have learned and the circuit theory the world runs on (the dedicated DC/AC-circuits chapter of this wiki, built from the old high-school corpus, is the gap this track will fill by name).

KCL (currents into a node sum to zero) is the continuity equation plus the MQS decree that no appreciable charge accumulates at junctions: $\nabla\cdot\mathbf{J}=-\partial\rho/\partial t\approx0$ everywhere except designated charge warehouses — capacitor plates — whose charging current the displacement current smuggles across the gap. KVL (voltages around a loop sum to zero) is Faraday's law plus the decree that $\partial\Phi_B/\partial t$ is negligible around wiring loops — all appreciable flux being confined inside designated flux warehouses, the inductors, whose $-L\,\mathrm{d}I/\mathrm{d}t$ appears in the loop sum as a lumped "voltage." A circuit diagram is thus a treaty: fields shall live only inside components; the wires between them shall be field-free ideal connectors. The treaty holds precisely while the whole apparatus is electrically small, $L\ll\lambda$ — and every "weird" high-frequency phenomenon (ground bounce, crosstalk, trace resonance) is a treaty violation, the fields reasserting their right to exist in the space between components.

Limiting cases

case	condition	what survives
true statics	$\partial_t=0$	both couplings gone; Coulomb + Biot–Savart, no EMF
EQS	$L\ll\lambda$ , $T\ll\tau_e$	scalar potential, capacitive currents, charge relaxation
MQS	$L\ll\lambda$ , $T\gg\tau_e$	instantaneous Biot–Savart + induction, eddy currents, inductance
deep-MQS conductor	also $T\ll\tau_m$	field excluded from bulk: skin effect, diffusion (next article)
resistive limit	$T\gg\tau_m$	field fully penetrates; DC resistance physics
breakdown	$L/\lambda\gtrsim1$	retardation, phase slip, radiation — retarded potentials (E9)

Worked examples

1 · Auditing a lab electromagnet's quasistatic licence (confidence builder)

A solenoid of length $L=20$ cm is driven at $f=10$ kHz. Check every clause of the licence: retardation, branch selection, and the size of the neglected term.

Step 1. Retardation clause: $\lambda = c/f = 3.0\times10^4$ m — thirty kilometres. $L/\lambda = 6.7\times10^{-6}$ : the phase of the field is uniform across the coil to seven digits. Pass.

Step 2. Branch clause: copper windings, so compare clocks. $\tau_e = \varepsilon_0/\sigma = 1.5\times10^{-19}$ s against $T=10^{-4}$ s: charges relax fifteen orders of magnitude within a cycle — MQS. Equivalently $\varepsilon_0\omega/\sigma = 9.3\times10^{-15}$ : displacement current is fourteen orders below conduction. Pass.

Step 3. Size of the abandoned term: relative correction $\sim(2\pi L/\lambda)^2 = 1.8\times10^{-9}$ . The flux, the inductance, and the back-EMF you compute from instantaneous Biot–Savart are good to parts per billion. The licence is valid; magnetostatics-with-a-clock is the exact right tool.

Step 4. What you must keep: $\partial\mathbf{B}/\partial t$ at full strength. The back-EMF $-L\,\mathrm{d}I/\mathrm{d}t$ is first-order physics here (at 10 kHz it dominates the coil's impedance), and dropping it — the warning's first crime — would predict a coil indistinguishable from a wire.

2 · One frequency, two verdicts: the transformer and the transmission line (exam level)

Both run at 50 Hz ( $\lambda = c/f = 6.0\times10^3$ km). A distribution transformer has size $L_1=0.3$ m; an interstate power line has length $L_2=1000$ km. Deliver a verdict on each.

Step 1. Transformer: $L_1/\lambda = 5\times10^{-8}$ , correction $(L_1/\lambda)^2 = 2.5\times10^{-15}$ . Deeply MQS — the entire design (flux linkage, turns ratio, core eddy losses) is legitimately computed with instantaneous magnetostatics plus Faraday. This is why Mutual Inductance's treatment of transformer coupling, recalled by name, never needed retardation.

Step 2. Power line: $L_2/\lambda = 0.17$ . The phase slip along the line is $2\pi L_2/\lambda = 1.05$ rad $= \mathbf{60^\circ}$ — the voltage at the far end of the line belongs to a different part of the cycle than the near end. Quasistatics has expired.

Step 3. Consequence of the expiry: the line must be modelled as a distributed system — telegrapher's equations, surge impedance, travelling waves — and real grid phenomena live in the gap: open-ended long lines show voltage rise at the far end (the Ferranti effect), and switching surges propagate as genuine waves that protection engineers time in milliseconds-per-thousand-kilometres. Same frequency, same copper, same Maxwell; the only thing that changed between Step 1 and Step 2 is $L/\lambda$ .

Step 4. The modern rhyme: a 1 cm processor die at 3 GHz has $L/\lambda = 0.1$ — the same marginal ratio as the power line. A chip is a power grid problem at ten orders of magnitude higher frequency, and signal-integrity engineers re-derive the telegraphers' wisdom on every new package. Rule of thumb both communities share: lumped thinking is safe below $L/\lambda \sim 1/10$ , and this example is why.

Origins · 1 minGustav Kirchhoff and how this was discovered

Gustav Kirchhoff

1824–1887 · Königsberg and Heidelberg — the man who wrote the quasistatic constitution

Kirchhoff wrote the circuit laws in 1845 as a 21-year-old student — before Maxwell's equations existed to be approximated. The laws worked so well that their fine print went unread for a century. In 1857 he analyzed signal propagation along a telegraph wire and found the speed came out as the ratio of electrostatic to electromagnetic units: numerically $3.1\times10^8$ m/s, the speed of light — four years before Maxwell. Working within the quasistatic worldview, Kirchhoff noted the coincidence and set it aside; the framework that could have told him he was staring at light's electromagnetic nature was exactly the approximation his own laws enshrined. It stands as the great instructive near-miss of physics: an approximation good to parts per billion in its domain can still amputate the discovery waiting just outside. Maxwell, keeping both coupling terms, closed the loop; Kirchhoff's laws survive as the loop's slow-motion limit — and his spectroscopy (with Bunsen) went on to seed the other revolution, quantum theory.

History

1845

— Kirchhoff, aged 21, states the circuit laws — quasistatic electromagnetism before field theory

1857

— Kirchhoff computes telegraph-signal speed ≈ 3.1×10⁸ m/s and files the coincidence away

1873

— Maxwell's Treatise: both coupling terms, the full loop, light as the payoff

1885

— Heaviside's telegrapher's equations map exactly where lumped circuit thinking expires

1926

— long-distance power grids force wave treatment of 50 Hz lines: quasistatics fails at continental scale

1989

— Haus & Melcher's text canonizes the EQS/MQS taxonomy and the two-clock selector

Going further

References. Griffiths, Introduction to Electrodynamics, 4th ed., §7.2, §11.1.2 (quasistatic and the failure of the quasistatic approximation) · Jackson, Classical Electrodynamics, 3rd ed., §5.18, §6.3 (quasi-static fields, diffusion, skin effect) · Zangwill, Modern Electrodynamics, Ch. 14 (Quasistatics) · Purcell & Morin, Electricity and Magnetism, 3rd ed., Ch. 7 (electromagnetic induction) · Haus & Melcher, Electromagnetic Fields and Energy (the canonical EQS/MQS treatment).

Provenance. Validity condition $L\ll\lambda$ and the EQS/MQS split follow Zangwill Ch. 14 and Haus & Melcher. Verifier should recompute: (1) $\lambda=c/f$ examples; (2) the ratio $\varepsilon_0\omega/\sigma$ for copper ( $\sigma=5.96\times10^7$ S/m); (3) charge-relaxation time $\tau_e=\varepsilon/\sigma$ ; (4) skin depth $\delta=\sqrt{2/(\mu_0\sigma\omega)}$ ; (5) the $(L/\lambda)^2$ scaling of the induced field from the order-of-magnitude derivation. Equations cross-checked against Griffiths §7.2 and Jackson §5.18, §6.3. Addendum for this edition: skin-depth machinery is RESCOPED forward to Eddy Currents (0565) by name; this edition owns the nondimensional ordering, the capacitor-iteration correction (recomputed: $(\omega R/c)^2/4 = 2.7\times10^{-7}$ at 1 MHz, $R=5$ cm), the two-clock selector with the identity $\tau_e\tau_m=(L/v)^2$ (copper 1 cm: $\tau_e=1.5\times10^{-19}$ s, $\tau_m=7.5$ ms, geometric mean $3.3\times10^{-11}$ s $= L/c$ ✓), and the circuit-theory constitution. Worked numbers recomputed: 50 Hz $\lambda=6.0\times10^3$ km, line phase slip $60°$ , transformer $(L/\lambda)^2=2.5\times10^{-15}$ , solenoid audit $1.8\times10^{-9}$ , 3 GHz die $L/\lambda=0.1$ .

Builds on: Faraday's Law of Induction · The Displacement Current · The Continuity Equation · Ohm's Law and Conductivity.

Leads to: Self-Inductance · Mutual Inductance · Energy Stored in Magnetic Fields · Induced Electric Fields.

If MQS drops the displacement current, doesn't that violate charge conservation?

It restricts you to current distributions that don't need it. Taking the divergence of the MQS Ampère law gives ∇·J = 0 exactly — the steady-current contract of the E4 Continuity Equation article. MQS systems are loop-current systems: solenoids, transformer windings, eddy loops, all divergence-free to excellent accuracy. Where charge genuinely accumulates (a capacitor plate inside an otherwise-MQS circuit), you locally re-admit the displacement current — that is exactly what circuit theory does when it hands the capacitor its own lumped element. The E6 twin's consistency theorem is not broken; its jurisdiction is just partitioned.

Is the quasistatic zone the same thing as an antenna's near field?

Essentially yes, and the connection is worth owning. Within r ≪ λ of any oscillating source, retardation hasn't accumulated and the fields are quasistatic in structure — the "near zone" of the radiation articles (E9, by name) is Coulomb/Biot–Savart dressed with time dependence, which is why near-field probes read source structure rather than radiation. The radiation zone begins where the phase slip becomes order one, r ~ λ/2π. An MQS wireless-charging pad and the reactive near field of its coil are the same physics in different vocabularies.

How can I tell EQS from MQS at a glance, without computing clocks?

Read the terminals and the storage. Open circuit, high impedance, energy parked as charge/voltage → EQS (capacitors, electrometers, insulation stacks, MEMS actuators). Closed loops, low impedance, energy parked as current/flux → MQS (coils, transformers, busbars, eddy-current systems). If both storages matter — an LC resonator — you need both lumps, and circuit theory glues an EQS element to an MQS element with ideal wires; the fields still never need full Maxwell as long as each element and the whole layout stay electrically small. The clocks are the arbiter when intuition stalls: conductor-dominated → MQS, insulator-dominated → EQS.

Do 50 Hz power lines radiate at all?

Almost immeasurably, and the scaling is why. Radiated power from a current loop scales as (L/λ)⁴ (dipole radiation, E9 by name) — for a 100-m pylon span at 50 Hz, L/λ ~ 1.7×10⁻⁵, so the radiated fraction carries a factor ~10⁻¹⁹ relative to a matched antenna. A continental grid moves gigawatts while radiating microwatts; effectively all "loss" is ohmic and inductive, not radiative. The 60° phase slip of the worked example matters for wave propagation along the line, guided by the conductors — a different phenomenon from radiation into space, and vastly stronger.

Why does the same device change regime when I change frequency?

Because two of the three clocks in the selector are fixed by material and size, but the drive period T is yours. A copper block at 50 Hz has T ≫ τ_m ~ ms: field soaks through, resistive regime. At 50 kHz, T ~ τ_m: partial penetration, serious eddy screening — the induction-cooktop operating point. At 50 MHz, T ≪ τ_m but still ≫ τ_em: deep-MQS skin regime, field expelled to microns. Push to optical frequencies and T ~ τ_em: quasistatics itself expires and the block becomes a mirror problem. One object, four physics regimes, one dial.

Is EQS's electric field exactly curl-free?

No — approximately, to (L/λ)². The capacitor derivation shows the honest structure: the zeroth-order field is curl-free and carries a well-defined potential, the displacement current builds a small B, and that B's time dependence induces a rotational correction ~10⁻⁷ of the main field for a bench device. "The field derives from a scalar potential" is an approximation statement with an error bar, not an identity. When precision work needs the correction (resonant-cavity design, high-Q sensor calibration), you iterate the loop — and the iteration's resummation is exactly the transition to full electrodynamics.

Who decides the boundary between lumped circuit theory and field theory in practice?

The λ/10 rule, enforced by pain. Practitioners treat structures below ~λ/10 as lumped and above as distributed; between λ/10 and λ/4 experienced designers already add transmission-line models for anything carrying fast edges. The rule is this article's (L/λ)² error estimate wearing a hard hat: at L = λ/10 the phase slip is 36°, and reflections/resonances on unterminated traces become measurable. Digital design converted wholesale in the 1990s–2000s as clock edges outran board sizes; the power grid made the same conversion in the 1920s. Every technology crosses this boundary exactly once, on its way up in frequency.