Polarization Density P

The last three articles built dipoles one molecule at a time: stretched atoms carrying $\alpha E$ , leaning water molecules carrying $p_0^2E/3k_BT$ , each twisted and tugged by the local field. But no capacitor measurement, no boundary-value problem, no formula in the rest of this chapter will ever mention molecule number $10^{24}$ individually. What macroscopic electrostatics needs is a field — a smooth function of position that remembers only what matters at the scale of instruments: how much dipole moment matter carries, per unit volume, where. That field is the polarization $\mathbf{P}$ , and this article is its biography: definition, units, how to read it as a vector field, and the precise sense in which a lumpy swarm of $10^{28}$ molecular dipoles becomes a differentiable function you can put under a divergence. Two readings of $\mathbf{P}$ compete for your intuition, and both are indispensable. The first is the census: $\mathbf{P} = n\langle\mathbf{p}\rangle$ , dipole moment per volume, assembled by averaging over a mesoscopic cell — large enough to hold millions of molecules, small enough that the capacitor's geometry cannot tell it from a point. The second is stranger and more powerful: $\mathbf{P}$ measures displaced charge. When a material polarizes, charge physically moves, and $\mathbf{P}\cdot\mathrm{d}\mathbf{a}$ is exactly the charge that crossed the surface element $\mathrm{d}\mathbf{a}$ on the way to the polarized state. The units agree with the hint: $\mathrm{C/m^2}$ — the same as surface charge density, a coincidence with consequences. Those consequences are the next two articles: wherever $\mathbf{P}$ 's charge-shipping is unbalanced — at surfaces, or wherever $\nabla\cdot\mathbf{P} \neq 0$ — the displaced charge piles up as the bound charge of Bound Charges: Surface and Volume, and the piled charge makes fields per Field of a Polarized Object. This article's job is the object itself: what $\mathbf{P}$ is, what it is not, and why a single vector field is a complete macroscopic memory of everything the molecular mayhem does.

Key takeaways

$\mathbf{P}$ is the electric dipole moment per unit volume: $\mathbf{P} = n\langle\mathbf{p}\rangle$ , a smooth macroscopic vector field summarizing $10^{25}$ – $10^{28}$ molecular dipoles per cubic meter.
The smoothness is earned by mesoscopic averaging: cells of $\sim(10\text{–}100\,\mathrm{nm})^3$ hold millions of molecules yet are invisible to laboratory geometry — the average is insensitive to cell size across that plateau, which is what makes $\mathbf{P}(\mathbf{r})$ well defined.
Units: $\mathrm{C/m^2}$ — dipole per volume is charge per area, and the second reading is literal: $\mathbf{P}\cdot\mathrm{d}\mathbf{a}$ equals the charge transported across $\mathrm{d}\mathbf{a}$ as the material polarized.
Time-varying polarization is therefore a real current, $\mathbf{J}_p = \partial\mathbf{P}/\partial t$ — measurable amperes flow "through" a capacitor's insulating gap while it charges.
Both microscopic mechanisms (stretch and lean) and all response types (linear, ferroelectric, electret) feed the same $\mathbf{P}$ ; the macroscopic formalism downstream never needs to know which.
Scale anchors: water in a $10^6\,\mathrm{V/m}$ field carries $P \approx 7\times10^{-4}\,\mathrm{C/m^2}$ — just $0.3\%$ of the fully-aligned ceiling $np_0 = 0.21\,\mathrm{C/m^2}$ .
KEY IDEA: $\mathbf{P}$ — dipole moment per volume, equivalently displaced charge per area — is the complete macroscopic descriptor of polarized matter; wherever its charge-shipping is unbalanced, bound charge appears, and everything downstream follows from $\mathbf{P}$ alone.

The core insight

A polarized dielectric is electrostatically equivalent to bound charges $\rho_b=-\nabla\cdot\mathbf{P}$ in its interior and $\sigma_b=\mathbf{P}\cdot\hat{\mathbf{n}}$ on its surface; everything about the macroscopic field follows from $\mathbf{P}$ — this article builds the field itself, the next one earns those charges.

Where this lives today

$\mathbf{P}$ as a state field with memory is a technology in itself. Ferroelectric memories (FeRAM, and the newer ferroelectric-HfO $_2$ transistors) store bits as the sign of a spontaneous $\mathbf{P}$ , read out through exactly the polarization currents of worked example 2 — flipping a cell's polarization launches a measurable $\int J_p\,\mathrm{d}t = 2P_r A$ of charge into the sense amplifier. Electret condenser microphones hold a frozen $\mathbf{P}$ for decades and convert sound-pressure geometry changes into voltage with no bias battery — billions ship annually. Piezoelectrics run the same object mechanically: strain changes $\mathbf{P}$ , and the resulting surface charges drive everything from lighter sparks to ultrasound imaging. On the theory side, the innocuous-looking definition (1) hides one of condensed-matter physics's celebrated subtleties: in an infinite periodic crystal, "dipole moment per unit cell" is ambiguous (which atoms belong to which cell?), and the honest fix — the modern theory of polarization (King-Smith, Vanderbilt, Resta, 1990s) — defines bulk $\mathbf{P}$ through the current that flows as the crystal state changes, i.e., through exactly this article's transport reading, elevated to a Berry-phase formula. That the shipping definition survives where the census definition falters is a satisfying verdict on which reading is deeper; every first-principles calculation of a ferroelectric's $P_r$ today computes it as accumulated polarization current.

Origins · 1 minHendrik Lorentz and how this was discovered

Hendrik Lorentz

1853–1928 · Leiden

Lorentz is the physicist who made "macroscopic field in matter" a defensible phrase. His electron theory (1892 onward, systematized in The Theory of Electrons, 1909) treated matter as swarms of point charges in vacuum governed by microscopic fields, and derived the smooth fields of Maxwell's macroscopic equations by explicit averaging — the mesoscopic contract of this article, executed with full nineteenth-century rigor, polarization emerging as the averaged dipole density of his bound electrons. The same program produced the local-field cavity construction (this chapter's later business), the Lorentz–Lorenz relation tying refractive index to polarizability, and the electron-oscillator model of dispersion still taught as "the Lorentz model." His force law $q(\mathbf{E} + \mathbf{v}\times\mathbf{B})$ named the whole enterprise's contact point between fields and matter. Awarded the 1902 Nobel Prize (with Zeeman), he chaired the first Solvay conferences, where Einstein — whose relativity grew directly out of Lorentz's transformations — called him "the greatest and noblest man of our times." The averaging that opens every matter chapter of electromagnetism is his quiet monument.

History

1837

— Faraday's dielectric experiments establish that insulating matter acquires a strained, polarized state

1850

— Mossotti and Clausius connect a medium's polarization to the responses of its constituent molecules

1892

— Lorentz's electron theory derives macroscopic fields, polarization included, by averaging over microscopic charges

1909

— The Theory of Electrons codifies the mesoscopic averaging contract behind every field-in-matter definition

1993

— King-Smith and Vanderbilt's Berry-phase theory redefines crystal polarization via the transport reading — P as shipped charge wins

The physical picture

Zoom into polarized water at the nanometer scale and the "polarization" is invisible: molecules tumble, dipole arrows point everywhere, and a cubic nanometer holds only $33$ molecules — too few for statistics, and the count flickers as molecules wander in and out. Zoom out to a $(100\,\mathrm{nm})^3$ cell and everything changes: thirty million molecules, fluctuations beaten down by $1/\sqrt{N}$ to parts in ten thousand, and a net dipole moment that points cleanly along the applied field. Divide by the cell volume: one vector. Slide the cell smoothly through the material: a vector field, $\mathbf{P}(\mathbf{r})$ . That two-step — average over a cell big enough for statistics, small enough for geometry — is the mesoscopic contract every macroscopic field in matter signs (Lorentz wrote the contract's fine print for electromagnetism), and it works because laboratory apparatus is a million times larger than the cell: no capacitor can tell a smooth $\mathbf{P}$ from the lumpy truth it summarizes.

Figure

Coarse-graining — molecular chaos becomes a smooth vector field

Now the second reading, which turns $\mathbf{P}$ from a bookkeeping density into a transport statement. Polarizing a material moves charge: every stretched atom shifted its electron cloud, every reoriented water molecule swung its charged ends. Imagine any small surface patch drawn inside the material, and run the movie of polarization switching on. Positive charge lurches one way across the patch, negative the other; the net charge shipped across is not zero, and the derivation below shows it equals exactly $\mathbf{P}\cdot\mathrm{d}\mathbf{a}$ . That is why $\mathbf{P}$ 's units are coulombs per square meter: polarization is displaced charge per unit area, frozen in place. The reading pays immediately in two currencies. Run the movie in time and you get a genuine current density $\mathbf{J}_p = \partial\mathbf{P}/\partial t$ — the amperes that flow through your capacitor's dielectric while it charges, no free carriers required. And audit the shipping ledger over a closed region: wherever more charge is shipped in than out, charge piles up — the pile-up is the bound charge, and cashing that observation into $\sigma_b = \mathbf{P}\cdot\hat{\mathbf{n}}$ and $\rho_b = -\nabla\cdot\mathbf{P}$ is precisely the next article's opening move.

Figure

Reading P as a vector field — balanced versus unbalanced charge shipping

One discipline completes the picture: $\mathbf{P}$ is a state field, agnostic about its cause. Linear dielectrics have $\mathbf{P}$ proportional to $\mathbf{E}$ (the susceptibility machinery of Linear Dielectrics, Susceptibility and Permittivity); electrets hold a frozen-in $\mathbf{P}$ with no field applied at all; ferroelectrics choose a spontaneous $\mathbf{P}$ below their Curie temperature and hold it against reversal (Ferroelectrics and Spontaneous Polarization). The formalism this article and its successors build — bound charges, fields, displacement — works verbatim for all three, because it consumes only $\mathbf{P}(\mathbf{r})$ itself. Learn the object once; the constitutive zoo comes later.

Definition and the two readings

The census definition, made honest by the mesoscopic average:

\mathbf{P}(\mathbf{r}) = \frac{1}{\Delta V}\sum_{i\,\in\,\Delta V}\mathbf{p}_i = n\,\langle\mathbf{p}\rangle \qquad (1)

where: $\Delta V$ — an averaging cell centered at $\mathbf{r}$ , sized on the plateau (large to molecules, small to apparatus); $\mathbf{p}_i$ — the individual molecular dipole moments, induced or permanent; $n$ — molecular number density; $\langle\mathbf{p}\rangle$ — the mean single-molecule moment. Note what the definition presupposes from Origin Dependence of Multipole Moments: each molecular $\mathbf{p}_i$ is origin-independent because molecules are neutral — for an ionic crystal one groups neutral units first, or the "dipole density" would be bookkeeping rather than physics. The transport identity, derived below:

\mathrm{d}q_{\rm shipped} = \mathbf{P}\cdot\mathrm{d}\mathbf{a}, \qquad \mathbf{J}_p = \frac{\partial\mathbf{P}}{\partial t} \qquad (2)

where: $\mathrm{d}q_{\rm shipped}$ — the net charge that crossed the internal surface element $\mathrm{d}\mathbf{a}$ (orientation chosen by its normal) during polarization; $\mathbf{J}_p$ — the polarization current density while $\mathbf{P}$ changes. Equation (2) is the article's workhorse: it explains the units, powers the capacitor's charging transient, and hands the next article its bound charges on a plate. (Its time derivative also foreshadows a Maxwell-scale idea: polarization current joins free current as a source of magnetic fields — the full accounting arrives with the displacement current in the electrodynamics chapter.)

Derivation AFrom molecular swarm to smooth field — the mesoscopic average and its plateau

The goal is to make definition (1) legitimate: to show there exists a range of averaging-cell sizes over which $\mathbf{P}$ is independent of the cell — so that "dipole moment per volume at point $\mathbf{r}$ " refers to something real rather than to an arbitrary choice. The strategy is a two-sided squeeze on the cell size $L$ , with the numbers run for water at room temperature. This is the honest version of a step most treatments wave through; running it once inoculates against both over-trusting $\mathbf{P}$ (it has no meaning below nanometers) and under-trusting it (it is exact for every purpose above micrometers).

Step 1. The lower squeeze: statistics. A cell of side $L$ holds $N = nL^3$ molecules; the fractional fluctuation of its net dipole scales as $1/\sqrt{N}$ (uncorrelated tumbling; correlations in a liquid tighten this only by factors of order one). For water, $n = 3.3\times10^{28}\,\mathrm{m^{-3}}$ : at $L = 1\,\mathrm{nm}$ , $N = 33$ — fluctuations of $\sim17\%$ , useless; at $L = 100\,\mathrm{nm}$ , $N = 3.3\times10^{7}$ — fluctuations of $0.02\%$ , a clean thermodynamic average. The cell must beat $\sim10\,\mathrm{nm}$ before $\mathbf{P}$ stops flickering.

Step 2. The upper squeeze: geometry. The average must not smear real macroscopic variation: $L$ must stay far below the scale on which fields and materials change — the capacitor gap, the sphere radius, the wavelength (in optics: hundreds of nm, which is why the same construction underlies the refractive index). For laboratory electrostatics that scale is $10\,\mathrm{\mu m}$ at the very least.

Step 3. The plateau. Between $\sim10\,\mathrm{nm}$ and $\sim1\,\mathrm{\mu m}$ the average is flat: enlarge the cell tenfold and $\mathbf{P}$ changes by less than the fluctuations you removed. A quantity insensitive to its defining convention across two decades of scale is, operationally, a field. This plateau is the entire license for treating matter as a continuum in this chapter — the same license fluid mechanics uses for density and velocity fields. Why this can fail, and where: in a nanometer-thin gate oxide or a protein pocket, no plateau exists — the geometry scale crashes into the statistics scale, "the polarization at $\mathbf{r}$ " stops being well defined, and one must retreat to explicit molecular models. Knowing when the continuum contract is void is part of owning it.

Step 4. What the average forgets, honestly. Coarse-graining discards the molecular near-field — the violent ångström-scale fields between molecules — and keeps only the smooth part. The claim that the smooth $\mathbf{P}$ suffices to reproduce the macroscopic potential everywhere is real and provable, but its proof is the integration-by-parts construction owned by Field of a Polarized Object; and the question "what field does one molecule actually sit in?" (not the macroscopic average!) is the local-field problem deferred to Local Field and the Lorentz Cavity. This article's average is the foundation both of them stand on. The cell counts confirm the argument: 33 molecules at (1 nm)³ give fluctuations of 17%, while 3.3×10⁷ molecules at (100 nm)³ reduce them to 0.02% — the plateau is real, and the failure cases at nanometer scales match the warnings familiar from the nanodielectrics literature.

Go deeper · Derivation B · 2 minP as displaced charge — the transport identity and the polarization current

Derivation BP as displaced charge — the transport identity and the polarization current

The goal is equation (2): the net charge shipped across an internal surface element during polarization equals $\mathbf{P}\cdot\mathrm{d}\mathbf{a}$ — the statement that gives $\mathbf{P}$ its units, its current, and (next article) its bound charges. The strategy is direct counting in the displacement picture: model polarization as positive charge $q$ per molecule shifting by $\boldsymbol{\delta}$ (with $\mathbf{p} = q\boldsymbol{\delta}$ ; any real mix of stretches and rotations is a superposition of such shifts). We count crossings of a fixed mathematical surface inside the material. The division of labor with the next article is explicit: here we ship charge across one open patch; there the ledger is closed around a region and the pile-up becomes $\sigma_b$ and $\rho_b$ .

Step 1. Who crosses? Fix a patch $\mathrm{d}\mathbf{a}$ (normal $\hat{\mathbf{n}}$ ) inside the material. As polarization switches on, the positive charge of each molecule moves by $\boldsymbol{\delta}$ . A molecule's positive charge crosses the patch precisely if it started within the oblique prism of base $\mathrm{d}a$ and edge $\boldsymbol{\delta}$ behind the patch — a volume $\boldsymbol{\delta}\cdot\hat{\mathbf{n}}\,\mathrm{d}a$ .

Step 2. Count and convert. With $n$ molecules per volume, the positive charge shipped across is

\mathrm{d}q = q\,n\,(\boldsymbol{\delta}\cdot\hat{\mathbf{n}})\,\mathrm{d}a = n\,(\mathbf{p}\cdot\hat{\mathbf{n}})\,\mathrm{d}a = \mathbf{P}\cdot\mathrm{d}\mathbf{a}.

(If instead the negative charge moves backward, or both move, the counting superposes to the same total — only the relative displacement, i.e. $\mathbf{p}$ , enters.) The identity is local and exact for the coarse-grained field: polarization is the surface density of charge displaced across any interior surface. The units $\mathrm{C/m^2}$ stop being a curiosity.

Step 3. Sanity check on the uniform slab. Slab of thickness $t$ , area $A$ , uniform $\mathbf{P}$ along the normal. Total dipole moment two ways: census, $P\times(At)$ ; displaced-charge picture, a sheet $+q_{\rm ship} = PA$ effectively advanced to one face and $-PA$ to the other, separated by $t$ — moment $(PA)\,t$ . Identical, and the face sheets of magnitude $P$ per unit area are exactly the $\sigma_b = \mathbf{P}\cdot\hat{\mathbf{n}}$ the next article will earn in general.

Step 4. Let $\mathbf{P}$ vary in time. Charge crossing per unit area per unit time is a current density:

\mathbf{J}_p = \frac{\partial\mathbf{P}}{\partial t}.

This is not a metaphor — it is measurable amperes. While a dielectric-filled capacitor charges, the bound charge rearranging inside the insulator constitutes a real current through the gap, continuous with the wire current; in our polystyrene capacitor (worked example 2) it carries $62\%$ of the charging transient. Its magnetic consequences (a current is a current) and its starring role inside the displacement current belong to the electrodynamics chapter; its existence belongs here.

Limiting cases

Case	Result	Meaning
uniform $\mathbf{P}$	charge shipping balanced in bulk	pile-up only at faces — next article's $\sigma_b$
non-uniform $\mathbf{P}$	unbalanced shipping, $\nabla\cdot\mathbf{P} \neq 0$	interior pile-up coming: $\rho_b = -\nabla\cdot\mathbf{P}$
$\mathbf{P} = 0$	vacuum or unpolarized matter	the chapter's reference state
linear response	$\mathbf{P} \propto \mathbf{E}$	susceptibility machinery, next-but-one article
electret / ferroelectric	$\mathbf{P} \neq 0$ at $\mathbf{E}_{\rm applied} = 0$	frozen or spontaneous polarization; same formalism
saturation ceiling	$P \to np_0$	all dipoles aligned: 0.21 C/m² for water — never reached statically
static $\mathbf{P}$	$\mathbf{J}_p = 0$	polarization current flows only while $\mathbf{P}$ changes
cell below ~10 nm	no plateau, $\mathbf{P}$ ill-defined	continuum contract void; molecular modeling required

Common trapP is not a charge density, and polarized is not charged

Two conflations to refuse. First, despite the units, $\mathbf{P}$ is not itself a charge density anywhere — it is a dipole density whose spatial structure will generate charge densities ( $\sigma_b$ , $\rho_b$ ) only where its charge-shipping is unbalanced; quoting "the charge $P$ " without the geometry that unbalances it is meaningless. Second, a polarized object remains rigorously neutral: polarization displaces charge, it does not create any, so whatever bound charge appears in one place is exactly compensated elsewhere (the next article proves total bound charge is always zero). A polarized slab attracts and is attracted — but a charge-meter integrating over the whole slab reads zero, and any calculation in which total bound charge comes out nonzero has made an arithmetic error, not discovered new physics.

Worked examples

1 · How polarized is polarized water? (confidence builder)

Water at room temperature sits in a field $E = 1.0\times10^6\,\mathrm{V/m}$ (take the measured $\kappa = 80$ as given). Find $P$ , compare it with the fully-aligned ceiling, and back out the mean alignment per molecule.

Step 1. From the measured response (using $P = \varepsilon_0(\kappa - 1)E$ , the linear-dielectric relation quoted here as an anticipation of its own article): $P = (8.854\times10^{-12})(79)(10^6) = 7.0\times10^{-4}\,\mathrm{C/m^2}$ .

Step 2. The ceiling: complete alignment of every molecule gives $P_{\rm sat} = np_0 = (3.3\times10^{28})(6.2\times10^{-30}) = 0.21\,\mathrm{C/m^2}$ .

Step 3. Fraction: $P/P_{\rm sat} = 7.0\times10^{-4}/0.21 = 0.33\%$ . Per molecule, $\langle p\rangle = P/n = 2.1\times10^{-32}\,\mathrm{C\cdot m}$ — each molecule contributes a third of a percent of its full moment, on average.

Step 4. Cross-examine with the microscopic articles. The bare Langevin estimate ( $u/3$ with $u = p_0E/k_BT$ ) gave $0.05\%$ — the measured liquid response is about seven times larger per molecule, the signature of correlated, hydrogen-bonded clusters leaning together (the Kirkwood enhancement flagged in the orientational article). The macroscopic $\mathbf{P}$ quietly encodes liquid-state physics the independent-dipole theory misses — and being a state function, it doesn't care: bound charges and fields downstream follow from $P = 7\times10^{-4}\,\mathrm{C/m^2}$ regardless of how the molecules conspired to produce it.

2 · The current through an insulator (exam level)

The polystyrene capacitor of the chapter opener ( $C = 57.6\,\mathrm{pF}$ , $\kappa = 2.6$ , plate area $A = 25\,\mathrm{cm^2}$ ) is charged to $12\,\mathrm{V}$ through a $50\,\Omega$ resistor. Show that a real current flows through the dielectric, find its share of the total, and its peak value.

Step 1. The transient. Charging follows $\tau = RC = (50)(57.6\times10^{-12}) = 2.9\,\mathrm{ns}$ , with wire current $I(t) = (V/R)e^{-t/\tau}$ , peak $I_0 = 12/50 = 0.24\,\mathrm{A}$ .

Step 2. Inside the gap, the growing field polarizes the polystyrene: $\mathbf{P}$ rises from 0 to its final $P = \sigma_b = 0.17\,\mathrm{\mu C/m^2}$ (the opener's audit). By the transport identity, the rearranging bound charge is a current density $J_p = \partial P/\partial t$ through every interior plane of the dielectric.

Step 3. Its share. At every instant, $P = \sigma_f(1 - 1/\kappa)$ tracks the free charge, so $J_p = (1 - 1/\kappa)\,\mathrm{d}\sigma_f/\mathrm{d}t$ : the polarization current is a fixed fraction $1 - 1/\kappa = 0.62$ of the conduction current arriving on the plates. Peak polarization current: $0.62\times0.24 = 0.15\,\mathrm{A}$ — a seventh of an ampere flowing through a perfect insulator, carried by electron clouds shifting femtometers, molecule by molecule, all at once.

Step 4. Where the rest flows. The remaining $1/\kappa$ share of the current's continuity is carried by the changing field itself ( $\varepsilon_0\,\partial E/\partial t$ — the vacuum displacement current, whose story belongs to Maxwell's equations). Together they make the current continuous through the gap: nothing accumulates anywhere except the plates and faces. Kirchhoff's current law survives inside capacitors because polarization current and field current are exactly as real as the wire's.

Going further

References. Griffiths, Introduction to Electrodynamics, 4th ed., §4.2–4.4 · Jackson, Classical Electrodynamics, 3rd ed., §4.3 · Purcell & Morin, Electricity and Magnetism, 3rd ed., Ch. 10 · Zangwill, Modern Electrodynamics, §6.2–6.3. Key numerical benchmarks: $\varepsilon_r \approx 80$ for water and $\varepsilon_0 = 8.854\times10^{-12}\,\mathrm{F/m}$ are standard CODATA values. The bound-charge results $\rho_b = -\nabla\cdot\mathbf{P}$ and $\sigma_b = \mathbf{P}\cdot\hat{\mathbf{n}}$ follow by integration by parts of the dipole potential, and $\varepsilon_r = 1 + \chi_e$ — both derived in the next two articles. Worked numbers: water at 1 MV/m — $P = 7.0\times10^{-4}$ C/m², ceiling $np_0 = 0.21$ C/m², fraction 0.33%, $\langle p\rangle = 2.1\times10^{-32}$ C·m; capacitor transient — $\tau = 2.9$ ns, $I_0 = 0.24$ A, polarization share $1 - 1/\kappa = 0.62$ , peak $J_p$ current 0.15 A; averaging cells — 33 molecules (1 nm)³, $3.3\times10^7$ (100 nm)³.

Builds on: Electric Dipole Moment · Origin Dependence of Multipole Moments · Induced Dipoles and Atomic Polarizability · Permanent Dipoles and Orientational Polarization · Dielectric Materials and Insulators

Leads to: Bound Charges: Surface and Volume · Field of a Polarized Object · The Electric Displacement D · Linear Dielectrics, Susceptibility and Permittivity · Gauss's Law in Dielectrics

Is P the sum over induced dipoles, permanent dipoles, or both?

Both, indiscriminately — and that indifference is the design. The census in equation (1) adds whatever molecular moments exist in the cell: stretched clouds, leaning permanent dipoles, displaced ions in a crystal lattice. Downstream formalism (bound charges, fields, $\mathbf{D}$ ) consumes only the total $\mathbf{P}(\mathbf{r})$ . The decomposition matters only when you ask constitutive questions — how $\mathbf{P}$ responds to $\mathbf{E}$ , to temperature, to frequency — and there the mechanisms separate cleanly (Debye's $1/T$ trick, the optical washout of orientation). State field first, response theory second; this chapter's article ordering is that sentence.

If the averaging cell is arbitrary within the plateau, is P really unique?

Yes, to the accuracy anything macroscopic can probe — that is exactly what the plateau means. Two legal cells ( $50\,\mathrm{nm}$ and $500\,\mathrm{nm}$ , say) give values of $\mathbf{P}$ differing by less than the thermal noise either one carries, and no laboratory-scale measurement can distinguish the resulting predicted fields. Uniqueness fails only where the plateau closes — nanostructures, interfaces a few molecules thick, biomolecular pockets — and there the honest statement is not "P is uncertain" but "P is not the right variable"; one computes with molecules directly.

How can current flow through a perfect insulator without any charge making it across?

Because current through a surface needs charges to cross it, not to travel far. During polarization every molecule's charges shift a femtometer; molecules straddling the surface carry their charge across it; the sum over a mole of molecules is macroscopic amperes. It is a bucket brigade in which every bucket moves one inch: no individual charge traverses the gap, yet charge crosses every interior plane. When the polarization stops changing, the current stops — an insulator conducts transients only, which is precisely a capacitor's job description in a circuit.

Does the polarization current produce a magnetic field like an ordinary current?

Yes — an ampere is an ampere. $\mathbf{J}_p = \partial\mathbf{P}/\partial t$ enters Ampère's law (in its Maxwell-corrected form) on the same footing as free current; inside a charging capacitor the polarization current plus the vacuum term $\varepsilon_0\partial\mathbf{E}/\partial t$ together sustain exactly the magnetic field that the wire current predicts, keeping the law consistent around the circuit. The full bookkeeping — where these terms live inside $\partial\mathbf{D}/\partial t$ — is the displacement-current story of the Maxwell chapter; the physical reality of $\mathbf{J}_p$ 's magnetism is settled here.

What exactly is an electret, and how does it hold P without a field?

An electret is matter with a frozen-in polarization: polar molecules aligned in a strong field while the material was molten or soft, then locked by cooling (or space charge trapped in a polymer — engineering electrets use both). Thermodynamically the aligned state is metastable — equilibrium at zero field wants $\mathbf{P} = 0$ — but the rotational relaxation time in the solid can be centuries, so the polarization simply doesn't decay on human timescales. The frozen $\mathbf{P}$ produces real bound charges and real external fields (next two articles), making the electret the electrostatic analogue of a permanent magnet — and the microphone in your laptop its most common incarnation.

Why does defining P get subtle for a crystal — surely one can just add up dipoles per unit cell?

Try it: in an infinite periodic lattice of alternating $+$ and $-$ ions, which negative ion "belongs to" which positive ion? Different (equally legal) choices of unit cell assign different partners and give dipole densities differing by finite amounts — the census definition is ambiguous modulo a "quantum of polarization." The modern resolution keeps this article's transport reading and discards the census: define changes of $\mathbf{P}$ by the current that flows while the crystal state evolves (a Berry-phase expression, in quantum-mechanical dress). Experiments only ever measure such changes — switching currents, pyroelectric currents — so the shipping definition matches the lab exactly where the bookkeeping one dissolves.

Is there a magnetization analogue of everything in this article?

Nearly line for line: $\mathbf{M}$ , the magnetic dipole moment per volume, is defined by the same mesoscopic average, has its own bound currents (rather than charges) where its circulation is unbalanced, and its own subtleties in crystals. The transport reading transposes too, with a twist: changing $\mathbf{M}$ ships no charge but drives electric fields (Faraday's law side), while changing $\mathbf{P}$ drives magnetic ones. The magnetic-materials chapter (E5) rebuilds the machinery; readers who internalize the shipping-versus-census duality here will find $\mathbf{M}$ 's bound currents almost anticlimactic.