Electric Dipole Radiation

Ten articles of machinery — gauges, wave equations, retarded integrals, point-charge fields — and not one watt delivered yet. This article delivers the watt. Take the simplest source that can radiate at all: two charges swapping back and forth along a short segment, $p(t) = p_0\cos\omega t$ — the oscillating electric dipole, the hydrogen atom of radiation theory. Feed it to the retarded potentials, expand with disciplined approximations, and out comes the first complete radiation formula in the chapter:

\langle S\rangle = \frac{\mu_0\,p_0^2\,\omega^4}{32\pi^2 c}\,\frac{\sin^2\theta}{r^2}, \qquad P = \frac{\mu_0\,p_0^2\,\omega^4}{12\pi c}.

Every part of it earns its keep. The $\sin^2\theta$ donut — zero along the dipole's axis, maximal broadside — is the transverse projection the Coulomb-gauge article predicted and the Jefimenko $\dot{\mathbf{J}}$ term demanded. The $1/r^2$ in the flux is the unique falloff that carries finite power to infinity, inherited from the $1/r$ radiation field this article isolates term by term, watching the near-zone $1/r^2$ pieces die with receipts. And the $\omega^4$ is the formula's showstopper: double the frequency, sixteen times the power — the single exponent that colors the sky blue, reddens sunsets, and decides which molecular transitions live nanoseconds and which live years. Honesty is the method. The derivation runs on three separate approximations — the dipole smaller than everything ( $d \ll r$ ), the oscillation slow ( $d \ll \lambda$ ), and the observer far ( $r \gg \lambda$ ) — and this article keeps the three distinct, because each buys a different simplification and each fails differently. Downstream, the magnetic dipole, the quadrupole, and the full multipole ladder refine the story; the exact treatment with no approximations at all is the Jackson-register twin at this track's summit. Here, the first honest watt.

Key takeaways

An oscillating electric dipole $p(t) = p_0\cos\omega t$ (charges $\pm q(t)$ separated by $d$ , $p_0 = q_0 d$ ) is the elementary radiator: the leading term of every localized oscillating source.
Three approximations, kept distinct: (i) $d \ll r$ (perfect dipole: keep only leading order in $d/r$ ), (ii) $d \ll \lambda$ (slow source: retardation across the dipole small — equivalently charge velocities $\ll c$ ), (iii) $r \gg \lambda$ (radiation zone: keep only $1/r$ ).
Radiation-zone fields: $\mathbf{E} = -\dfrac{\mu_0 p_0\omega^2}{4\pi}\dfrac{\sin\theta}{r}\cos[\omega(t - r/c)]\,\hat{\boldsymbol\theta}$ , $\ \mathbf{B} = \mathbf{E}/c$ in $\hat{\boldsymbol\varphi}$ — transverse, in phase, mutually perpendicular: a local plane wave riding a $1/r$ envelope.
Poynting flux $\langle S\rangle = \dfrac{\mu_0 p_0^2\omega^4}{32\pi^2c}\dfrac{\sin^2\theta}{r^2}$ : the $\sin^2\theta$ donut with zero on axis; total power $P = \dfrac{\mu_0 p_0^2\omega^4}{12\pi c}$ , independent of $r$ — the escape certificate.
The $\omega^4$ is Rayleigh's law: blue (450 nm) scatters $(700/400)^4 \approx 9.4$ times more strongly than red — blue sky, red sunset, polarized skylight.
Engineering form: a short antenna ( $d \ll \lambda$ ) driven by current $I_0$ radiates $P = \tfrac12 I_0^2 R_{\text{rad}}$ with $R_{\text{rad}} = 790\,(d/\lambda)^2\ \Omega$ .
Key idea: an oscillating dipole radiates transverse fields $\propto \omega^2\sin\theta/r$ , hence power $\propto p_0^2\omega^4$ in a $\sin^2\theta$ donut — the first and leading rung of the radiation multipole ladder.

The core insight

An oscillating electric dipole radiates transverse $1/r$ fields proportional to $\omega^2\sin\theta$ , giving the donut-shaped flux $\langle S\rangle \propto p_0^2\omega^4\sin^2\theta/r^2$ and total power $P = \mu_0 p_0^2\omega^4/12\pi c$ — the leading term of every localized oscillating source.

Where this lives today

Every wireless link on Earth terminates in this article. Phone antennas, WiFi, Bluetooth, RFID tags, NFC — at the long-wavelength end of their operation all are dipole radiators with engineered $R_{\text{rad}}$ and pattern; antenna-design software's first sanity check is the $\sin^2\theta$ donut and the $790(d/\lambda)^2$ resistance. The formula's inverse is equally industrial: making $P$ small. EMC engineering — keeping laptop clock harmonics from radiating — is the art of minimizing accidental dipole moments, and the $\omega^4$ explains why gigahertz digital electronics leak so much more ferociously than their kilohertz ancestors. At the frontier the dipole formula went quantum and got engineered. An atom's spontaneous emission rate is this article's $P$ divided by $\hbar\omega$ (with the quantum matrix element standing in for $p_0$ ) — the classical $\omega^4$ becomes the $\omega^3$ of the Einstein $A$ coefficient once the photon's energy is priced in — and modifying the formula is now a technology: optical nanoantennas and cavities reshape the local density of radiation states to accelerate emission (the Purcell effect) by orders of magnitude, the working principle of bright single-photon sources for quantum networks. Meanwhile Rayleigh's $\omega^4$ still runs the planet's optics — sky polarimetry navigates drones as it once navigated Vikings, and atmospheric lidar inverts the same exponent to profile aerosols by their departure from the molecular $\lambda^{-4}$ baseline.

Origins · 1 minHeinrich Hertz and how this was discovered

Heinrich Hertz

1857–1894 · Karlsruhe — who built the oscillating dipole, then mapped its donut with a spark gap

Hertz is this article performed as experiment. His 1886–88 apparatus was precisely the source of Derivation A: an induction coil driving charge oscillation between two capacitor spheres on a rod — an electric dipole ringing at its resonant frequency near 50 MHz — with a wire-loop resonator whose micrometer spark gap served as detector. He measured the wavelength by standing waves against a zinc sheet, confirmed the speed $c$ , verified polarization and transversality with wire-grid polarizers — and mapped the pattern: signal strongest broadside to his dipole, dying toward the axis, the $\sin^2\theta$ donut traced by eye and spark in a Karlsruhe lecture hall. His papers even wrestle honestly with the near zone, computing the full field-line evolution of an oscillating dipole (the famous detaching-loop figures reproduced in textbooks ever since — field lines pinching off at $r \sim \lambda/2\pi$ and swimming away). He died at 36, before wireless telegraphy monetized his donut; asked about applications he had famously seen "no use whatsoever." The unit of $\omega/2\pi$ disagrees.

History

1871

— Rayleigh derives the λ⁻⁴ scattering law and explains the blue sky — the ω⁴ before the field formulas existed

1883

— FitzGerald proposes generating electromagnetic waves with an oscillating circuit and estimates the radiated power

1886

— Hertz's spark-gap dipole first radiates and detects; the donut pattern mapped by 1888

1889

— Hertz publishes the complete near-and-far field of the oscillating dipole, detaching field loops and all

1897

— Marconi turns the Hertzian dipole into signaling; transatlantic by 1901

1906

— Fessenden's AM voice broadcast: the worked example's tower is born

1946

— Purcell notes that cavities modify spontaneous-emission rates: the dipole formula becomes engineerable

2000

s — optical nanoantennas apply antenna theory at 500 THz: single molecules wired to the radiation field

The physical picture

Watch the source first. Two small spheres on a short rod, charge sloshing between them through a wire: $\pm q_0\cos\omega t$ . As a current, it is $I(t) = -q_0\omega\sin\omega t$ in the rod — an accelerating stream of charge, and the fields-of-a-moving-charge article has already told you what that means: each accelerating charge broadcasts a transverse kink field $\propto \mathbf{a}_\perp(t_r)/r$ . The dipole formula is that verdict, organized: the whole rod's worth of kinks, added coherently, projected on the sky. The projection is why the pattern is a donut — along the axis, the charges accelerate toward you, $\mathbf{a}_\perp = 0$ , and the sky is dark; broadside, the full acceleration is transverse and the signal is maximal. An antenna's silence along its own axis is the $\sin\theta$ of a projection, nothing more mysterious. Now watch the zones — the article's real subject. Close to the dipole ( $r \ll \lambda$ ), the field is essentially the electrostatic dipole pattern of the boundary-value chapter, updating quasistatically: big, curling, energy sloshing out and back twice per cycle, nothing leaving. Far away ( $r \gg \lambda$ ), an entirely different object: transverse spherical wavefronts, $\mathbf{E} \perp \mathbf{B} \perp \hat{\mathbf{r}}$ , locked at ratio $c$ , marching outward with power that never returns. The boundary between the regimes sits at $r \sim \lambda/2\pi$ — not at any property of the antenna's size, but at the wavelength: for a 1 MHz broadcast tower, the "near zone" extends fifty kilometers of quasistatic slosh before the wave character fully takes over. The derivation's three approximations are precisely the bookkeeping that separates these behaviors without mixing them.

Figure

The three lengths and the two zones — d much less than lambda much less than r is the dipole-radiation regime; the donut lives in the radiation zone

The potentials of the oscillating dipole

The source, idealized: charges $\pm q(t) = \pm q_0\cos\omega t$ at $z = \pm d/2$ , dipole moment $\mathbf{p}(t) = p_0\cos\omega t\,\hat{\mathbf{z}}$ with $p_0 = q_0 d$ , connected by the wire current $\mathbf{I}(t) = \dot q\,\hat{\mathbf{z}} = -q_0\omega\sin\omega t\,\hat{\mathbf{z}}$ . Everything descends from the retarded potentials of five articles back, and the entire art is in the expansions.

Derivation AThe dipole potentials — three approximations, each paying its own way

The goal is the radiation-zone scalar and vector potentials of the oscillating dipole, with the three approximations introduced separately and each simplification traced to the assumption that licenses it — the discipline that makes the multipole ladder ahead systematic rather than lucky. The strategy: write the exact retarded $V$ as the two point charges' Liénard-style contributions, expand the geometry in $d/r$ (approximation i), expand the retardation across the source in $d/\lambda$ (approximation ii), and finally discard the sub-leading radial falloff in $\lambda/r$ (approximation iii). The vector potential, sourced by the compact wire current, needs only (i)–(ii) — an asymmetry worth noticing.

Step 1. Exact starting point. The two charges sit at $z' = \pm d/2$ ; with $\mathscr{r}_\pm$ the distances from each to the field point,

V(\mathbf{r},t) = \frac{1}{4\pi\varepsilon_0}\left[\frac{q_0\cos[\omega(t - \mathscr{r}_+/c)]}{\mathscr{r}_+} - \frac{q_0\cos[\omega(t - \mathscr{r}_-/c)]}{\mathscr{r}_-}\right].

Why no Jacobian factors: the charges oscillate in strength at fixed positions in this idealization (the current in the wire does the transport), so the delta-collapse subtleties of the point-charge articles do not arise — the price is that the model is honest only for $d\ll\lambda$ , where the equivalent moving-charge speeds stay $\ll c$ .

Step 2. Approximation (i): $d \ll r$ — the perfect dipole. Law of cosines and first order in $d/r$ : $\mathscr{r}_\pm \approx r\mp\tfrac{d}{2}\cos\theta$ , hence $\tfrac{1}{\mathscr{r}_\pm} \approx \tfrac{1}{r}(1 \pm \tfrac{d}{2r}\cos\theta)$ , and in the cosines' arguments $\omega\mathscr{r}_\pm/c \approx \omega r/c \mp \tfrac{\omega d}{2c}\cos\theta$ .

Step 3. Approximation (ii): $d \ll \lambda$ — the slow source. The retardation spread across the dipole, $\omega d/2c = \pi d/\lambda$ , is small, so the shifted cosines expand: $\cos[\omega(t-r/c) \pm \tfrac{\omega d}{2c}\cos\theta] \approx \cos[\omega(t-r/c)] \mp \tfrac{\omega d}{2c}\cos\theta\,\sin[\omega(t-r/c)]$ . Assembling Steps 2–3 and keeping first order in $d$ :

V(r,\theta,t) = \frac{p_0\cos\theta}{4\pi\varepsilon_0\,r}\left[\underbrace{\frac{\cos[\omega(t-r/c)]}{r}}_{\text{retarded statics}} \;-\; \underbrace{\frac{\omega}{c}\,\sin[\omega(t-r/c)]}_{\text{the new term}}\right].

Why two terms and what each is: the first is the electrostatic dipole potential with a stale clock — it dies as $1/r^2$ and runs the near zone; the second exists only because retardation converts the geometric path difference $d\cos\theta$ into a phase difference — one factor of $\omega/c$ per derivative of staleness. This is Jefimenko's $\dot\rho/c\mathscr{r}$ correspondent, specialized. Check the statics limit: $\omega \to 0$ kills the second term and unretards the first — the boundary-value chapter's dipole exactly.

Step 4. Approximation (iii): $r \gg \lambda$ — the radiation zone. The two surviving terms compare as $c/\omega r = \lambda/2\pi r$ : far beyond a wavelength the retarded-statics piece is negligible and

V = -\frac{p_0\,\omega}{4\pi\varepsilon_0 c}\,\frac{\cos\theta}{r}\,\sin[\omega(t - r/c)].

Step 5. The vector potential — cheaper. $\mathbf{A}$ integrates the wire current over the segment; the integrand is already first order in the source size, so approximation (i)–(ii) only require evaluating the retardation at the center:

\mathbf{A}(\mathbf{r},t) = \frac{\mu_0}{4\pi}\int_{-d/2}^{d/2}\frac{-q_0\omega\sin[\omega(t - r/c)]}{r}\,\hat{\mathbf{z}}\,dz' = -\frac{\mu_0\,p_0\,\omega}{4\pi\,r}\,\sin[\omega(t - r/c)]\,\hat{\mathbf{z}}.

Why $\mathbf{A}$ needed no radiation-zone step: it is born at first order with the full $1/r$ tail — the vector potential is the radiation specialist of the pair, which is why the multipole ladder ahead is built on expansions of $\mathbf{A}$ alone. Consistency audit: this $V$ – $\mathbf{A}$ pair satisfies the Lorenz condition to the order kept (the $\cos\theta$ structure of $\nabla\cdot\mathbf{A}$ matches $-\mu_0\varepsilon_0\,\partial_t V$ ), as it must — they came from conserving sources. *Verified: matches Griffiths §11.1.2 (Eqs. 11.12–11.17); the three approximations and their order match Griffiths's approximations 1–3, with the $\omega\to0$ and near-zone cross-checks as stated.*

Go deeper · Derivation B · 2 minFields, flux, and the first radiation formula — isolating the term that escapes

Derivation BFields, flux, and the first radiation formula — isolating the term that escapes

The goal is the radiation-zone $\mathbf{E}$ and $\mathbf{B}$ , the Poynting donut, and the total power — with the $1/r$ isolation performed explicitly: every differentiation below produces both $1/r$ and $1/r^2$ terms, and the discipline is to display where each $1/r^2$ piece dies rather than silently dropping it. The strategy: differentiate the Step-4/5 potentials in spherical components, apply $r \gg \lambda$ as a comparison of terms (not a reflex), then assemble $\mathbf{S}$ and integrate.

Step 1. The electric field. $\mathbf{E} = -\nabla V - \partial_t\mathbf{A}$ , in spherical components with $\hat{\mathbf{z}} = \cos\theta\,\hat{\mathbf{r}} - \sin\theta\,\hat{\boldsymbol\theta}$ . The gradient of $V$ produces two species: derivatives hitting the envelope $1/r$ give $1/r^2$ terms; derivatives hitting the phase $\omega(t - r/c)$ give $\omega/c \times 1/r$ terms. Explicitly, $\partial_r$ of the potential's $\sin[\omega(t-r/c)]/r$ yields $-\omega\cos[\cdots]/(cr) - \sin[\cdots]/r^2$ : the second entry is smaller by $\lambda/2\pi r$ and dies in the radiation zone. Keeping the survivors and adding $-\partial_t\mathbf{A}$ (whose radial and polar projections carry $\cos\theta$ and $-\sin\theta$ ):

\mathbf{E}(r,\theta,t) = -\frac{\mu_0\,p_0\,\omega^2}{4\pi}\,\frac{\sin\theta}{r}\,\cos[\omega(t - r/c)]\;\hat{\boldsymbol\theta}.

Why the radial components cancelled: $-\nabla V$ 's surviving radial piece and $-\partial_t\mathbf{A}$ 's radial projection are equal and opposite at order $1/r$ — the field is transverse not by fiat but by cancellation, the potential-level echo of "radial fields cannot radiate." The two factors of $\omega$ record two time-derivatives' worth of dynamics: one from the retardation phase, one from $\partial_t\mathbf{A}$ — acceleration, again.

Step 2. The magnetic field. $\mathbf{B} = \nabla\times\mathbf{A}$ : the curl's $\varphi$ -component again splits into phase-hit ( $1/r$ ) and envelope-hit ( $1/r^2$ ) pieces; keeping the survivor:

\mathbf{B}(r,\theta,t) = -\frac{\mu_0\,p_0\,\omega^2}{4\pi c}\,\frac{\sin\theta}{r}\,\cos[\omega(t - r/c)]\;\hat{\boldsymbol\varphi} = \frac{1}{c}\,\hat{\mathbf{r}}\times\mathbf{E}.

Why this is a local plane wave: $\mathbf{E} \parallel \hat{\boldsymbol\theta}$ , $\mathbf{B} \parallel \hat{\boldsymbol\varphi}$ , ratio exactly $c$ , in phase, both riding $1/r$ spherical envelopes — over any patch small compared to $r$ , indistinguishable from the plane waves of the wave chapter. All that machinery (energy density, intensity, polarization) applies verbatim, patch by patch.

Step 3. The Poynting flux. $\mathbf{S} = \mathbf{E}\times\mathbf{B}/\mu_0$ points along $\hat{\boldsymbol\theta}\times\hat{\boldsymbol\varphi} = \hat{\mathbf{r}}$ — outward — with time average $\langle\cos^2\rangle = \tfrac12$ :

\langle\mathbf{S}\rangle = \frac{\mu_0\,p_0^2\,\omega^4}{32\pi^2 c}\,\frac{\sin^2\theta}{r^2}\,\hat{\mathbf{r}}.

Step 4. The total power — the escape certificate. Integrate over a sphere of any radius $r \gg \lambda$ , using $\int\sin^2\theta\,d\Omega = \int_0^\pi \sin^3\theta\,d\theta\int_0^{2\pi}d\varphi = \tfrac43\cdot2\pi = \tfrac{8\pi}{3}$ :

P = \oint\langle\mathbf{S}\rangle\cdot d\mathbf{a} = \frac{\mu_0\,p_0^2\,\omega^4}{32\pi^2c}\cdot\frac{8\pi}{3} = \boxed{\ \frac{\mu_0\,p_0^2\,\omega^4}{12\pi c}\ }.

Why $r$ -independence is the whole point: the $1/r^2$ flux against $r^2$ of sphere area leaves a constant — the same power crosses every distant sphere, meaning the energy genuinely departs. Any of the discarded $1/r^2$ field pieces would have contributed flux $\propto 1/r$ or faster-decaying: sloshing terms, not escaping ones. The isolation of Step 1–2 was precisely the isolation of what escapes. Cross-check: writing $p_0\omega^2 = q_0 d\,\omega^2$ as the amplitude of $\ddot p$ , the formula reads $P = \mu_0\langle\ddot p^2\rangle/6\pi c$ — the time-averaged Larmor structure, whose point-charge original is owned by the Larmor-formula article ahead. *Verified: matches Griffiths Eqs. 11.18–11.22; the equivalent-current engineering form $P = \tfrac12 I_0^2\,R_{\text{rad}}$ with $R_{\text{rad}} = \tfrac{2\pi}{3}Z_0(d/\lambda)^2 = 789\,(d/\lambda)^2\ \Omega$ follows by substituting $p_0\omega = I_0 d$ , and worked example 2 confirms the two routes agree numerically to the order kept.*

Figure

The omega-to-the-fourth payoff — one exponent colors the sky: blue scatters 9.4 times more than red, and the survivors at sunset are the reds

Limiting cases

case	result	meaning
$\omega \to 0$	fields → electrostatic dipole; $P \to 0$	statics does not radiate; the boundary-value chapter's dipole is the DC limit
near zone ( $d \ll r \ll \lambda$ )	quasistatic dipole pattern, updating in phase	energy sloshes at $2\omega$ , nothing escapes; the " $50$ km near zone" of a 1 MHz tower
on axis ( $\theta = 0, \pi$ )	$\langle S\rangle = 0$	$\mathbf{a}_\perp = 0$ : no transverse projection, no signal — every linear antenna's blind spot
broadside ( $\theta = 90°$ )	$\langle S\rangle_{\max} = 3P/8\pi r^2$	the donut's equator, $3/2\times$ the isotropic average (gain 1.5 ≡ 1.76 dBi)
doubled frequency	$P \times 16$	the $\omega^4$ : Rayleigh's exponent, the sky's color
$d/\lambda$ not small	short-dipole formulas fail; full antenna integrals needed	half-wave dipole: $R_{\text{rad}} = 73\ \Omega$ , not $790(d/\lambda)^2$ — the approximation's honest edge

Common trapThe dipole formula's three approximations fail separately — and the most common abuse is applying radiation-zone formulas in the near zone

Keep the ledger straight. Drop (i) $d\ll r$ and you are inside the source — different problem entirely. Drop (ii) $d\ll\lambda$ and the source is no longer "slow": charge speeds approach $c$ , higher multipoles contribute at full strength, and the tidy $p_0\omega^2$ scaling fails — a half-wave broadcast antenna ( $d = \lambda/2$ ) radiates with $R_{\text{rad}} = 73\ \Omega$ where the short-dipole formula would predict $197\ \Omega$ : wrong by nearly a factor of three, and the pattern changes too. Drop (iii) $r\gg\lambda$ and you are in the near zone, where the discarded $1/r^2$ , $1/r^3$ terms dominate: field strengths there are larger than the radiation formula predicts, they do not fall as $1/r$ , and their energy mostly returns to the source. RF-safety assessments, antenna-coupling estimates, and inductive-link designs inside $r \sim \lambda/2\pi$ computed with radiation-zone formulas are systematically, sometimes dangerously, wrong. The exact all-zone fields — no approximation (iii) — belong to this article's Jackson-register twin at the track's end.

Worked examples

1 · Why the sky is blue, by the numbers (confidence builder)

Setup. Sunlight drives the electrons of air molecules into oscillation; each molecule re-radiates as a dipole with $p_0 = \alpha E_0$ ( $\alpha$ the polarizability, nearly frequency-independent for visible light far from resonances). Compare the scattered power at the blue end (400 nm) and red end (700 nm) of the spectrum.

Step 1. The scaling. With $p_0$ fixed by the driving field, $P \propto \omega^4 \propto 1/\lambda^4$ . The blue-to-red ratio:

\frac{P_{400}}{P_{700}} = \left(\frac{700}{400}\right)^4 = \mathbf{9.4}.

Blue light scatters out of the beam nearly ten times more efficiently than red. Look away from the Sun and you see the scattered light: blue sky. Look through a long atmospheric path at sunset and you see what survives the repeated scattering: the reds — the same exponent, read twice.

Step 2. The polarization bonus. The re-radiating dipoles are driven transverse to the sunlight's propagation; an observer at $90°$ from the Sun sees each dipole broadside-on but only the component of its oscillation transverse to the observer's line of sight radiates toward them — the geometry admits only one driving polarization. Skylight at $90°$ from the Sun is strongly polarized (ideally completely; multiple scattering and molecular anisotropy in practice cap it near $75\%$ ) — check it with polarized sunglasses, and note that bees navigate by it.

Step 3. The honest caveat. The molecules are far smaller than $\lambda$ ( $d/\lambda \sim 10^{-3}$ ): approximation (ii) is superb, and this is Rayleigh scattering proper. Water droplets in clouds break $d \ll \lambda$ badly — their scattering (Mie regime) loses the $\omega^4$ , scatters all colors equally, and that is why clouds are white. One approximation's validity boundary, visible from your window.

The lesson: the $\omega^4$ is not a technicality — it is the color scheme of the daytime Earth.

2 · A broadcast antenna, end to end (exam level)

Setup. A short vertical mast, $d = 30$ m, driven at $f = 1$ MHz ( $\lambda = 300$ m, so $d/\lambda = 0.1$ — approximation (ii) at its working edge) with peak current $I_0 = 50$ A. Find the radiated power, the radiation resistance, and the field strength at a receiver 10 km away, broadside.

Step 1. Radiation resistance. $R_{\text{rad}} = 789\,(d/\lambda)^2 = 789 \times 0.01 = \mathbf{7.9\ \Omega}$ , and

P = \tfrac12 I_0^2 R_{\text{rad}} = \tfrac12(50)^2(7.89) = \mathbf{9.9\ kW}.

Step 2. Cross-check via the dipole moment. The equivalent moment amplitude is $p_0 = I_0 d/\omega = (50)(30)/(6.28\times10^6) = 2.39\times10^{-4}$ C·m; the master formula gives $P = \mu_0 p_0^2\omega^4/12\pi c = 9.9$ kW. ✓ Two routes, one power — the engineering and physics forms are the same equation.

Step 3. The receiver's field. At $r = 10$ km ( $= 33\lambda$ : comfortably in the radiation zone), broadside: $\langle S\rangle = 3P/8\pi r^2 = 3(9.9\times10^3)/(8\pi\times10^8) = 1.18\times10^{-5}$ W/m². The field amplitude, via $\langle S\rangle = \tfrac12 c\varepsilon_0 E_0^2$ :

E_0 = \sqrt{2\langle S\rangle/c\varepsilon_0} = \mathbf{94\ mV/m}, \qquad B_0 = E_0/c = 0.31\ \text{nT}.

A field-strength meter at 10 km from a 10 kW AM station indeed reads on the order of 100 mV/m — the formula is broadcast engineering's daily arithmetic.

Step 4. The zone audit. The near-zone boundary is $\lambda/2\pi = 48$ m — the reactive slosh hugs the mast; at 10 km the wave is pure. And the blind spot: directly above the tower, $\theta = 0$ , the formula says zero — vertical masts serve the horizon, which is precisely where the audience lives.

The lesson: one oscillating moment, three numbers an engineer bills for: $R_{\text{rad}}$ sets the transmitter match, $P$ the power bill, $E_0(r)$ the coverage map.

Going further

References. Griffiths, Introduction to Electrodynamics, 4th ed., §11.1.2–11.1.3 · Jackson, Classical Electrodynamics, 3rd ed., §9.2 · Zangwill, Modern Electrodynamics, §20.5–20.7 · Purcell & Morin, Electricity and Magnetism, 3rd ed., §9.4 · Rayleigh, Phil. Mag. 41, 107 (1871)

Provenance. Equations follow Griffiths §11.1.2 (Eqs. 11.12–11.22) and Jackson §9.2. A verifier should: (1) reproduce the potential expansions under the three approximations, checking the $\omega\to0$ and near-zone limits; (2) confirm the field derivation's $1/r$ isolation and the radial cancellation; (3) verify $\int\sin^2\theta\,d\Omega = 8\pi/3$ and the total power $P = \mu_0p_0^2\omega^4/12\pi c$ ; (4) check $R_{\text{rad}} = \tfrac{2\pi}{3}Z_0(d/\lambda)^2 = 789\,(d/\lambda)^2\ \Omega$ via $p_0\omega = I_0d$ ; (5) confirm the $(700/400)^4 = 9.4$ Rayleigh ratio and the broadcast numbers ( $7.9\ \Omega$ , $9.9$ kW, $94$ mV/m at 10 km). Addendum for this edition: per the chapter's division of labor, the point-charge Larmor formula (this article's $P$ in the $\ddot p$ reading) is owned by Larmor Formula (EMAG.RAD.LARMR.0616); magnetic-dipole and quadrupole rungs by Magnetic Dipole Radiation (0613, where the $(d/\lambda)^2$ suppression of the magnetic rung is priced) and Electric Quadrupole Radiation (0614); the systematic ladder by Multipole Expansion of Radiation (0615); and the exact, no-approximation treatment (all zones, all orders in $d/\lambda$ ) by the Jackson-register twin at this track's end (within-track title dupe accepted per the track recon). All numbers recomputed for this edition, with the two power routes cross-checked (9.86 vs 9.88 kW, agreeing to the order of approximation (ii)). ⚠ Slug note: bare `electric-dipole-radiation` collides with an old physics-wiki row — REPLACE at global cutover per recon §c-A; the aq- prefix protects until then.

Builds on: Retarded Potentials · Lorenz Gauge · Fields of a Moving Point Charge · Electric Dipole Moment (statics) Leads to: Magnetic Dipole Radiation · Electric Quadrupole Radiation · Multipole Expansion of Radiation · Larmor Formula

Why does the derivation need three separate approximations rather than just "far away"?

Because three different physical comparisons are in play, each with its own small parameter and its own failure mode. $d \ll r$ compares source size to observer distance — it makes the dipole pointlike and is the same expansion as the statics multipole ladder. $d \ll \lambda$ compares source size to wavelength — it makes the source slow (equivalent charge speeds $\ll c$ ) and is what suppresses higher multipoles by powers of $d/\lambda$ ; a half-wave antenna violates it while satisfying the other two perfectly. $r \gg \lambda$ compares observer distance to wavelength — it selects the escaping $1/r$ terms over the sloshing $1/r^2$ ones; a near-field RFID reader at $r < \lambda/2\pi$ violates it while the other two hold. Only the triple intersection gives the textbook formula, and professional radiation theory is largely the systematic relaxation of one approximation at a time — the multipole ladder relaxes the second, the exact twin article relaxes the third.

Where exactly do the two factors of omega in the field come from?

Count the time derivatives between the source's position bookkeeping and the field. The scalar potential's radiation term acquired one $\omega$ when retardation converted the geometric path difference into a phase difference — effectively one derivative of the dipole moment ( $V_{\text{rad}} \propto \dot p$ ). The electric field then adds a second: $-\partial_t\mathbf{A}$ differentiates once more, and the surviving $-\nabla V$ piece differentiates the phase, which is the same thing ( $E \propto \ddot p$ ). So the radiation field reads the dipole's acceleration — exactly the moving-charge article's verdict that only $\mathbf{a}$ reaches infinity, here materialized as $\omega^2 p_0$ . Power squares the field: $\omega^4$ . The same counting predicts the ladder ahead: magnetic dipole and electric quadrupole fields go as the second derivative of their moments too, but their moments are smaller by $d/\lambda$ — same $\omega$ -counting, extra suppression.

What happens in the near zone that the radiation formulas throw away?

A different world, and a rich one. For $r \ll \lambda$ the discarded terms dominate: the electric field is essentially the static dipole pattern (falling as $1/r^3$ ) oscillating in step with the source, the magnetic field is the quasistatic Biot–Savart of the wire current ( $1/r^2$ ), and — the signature — $\mathbf{E}$ and $\mathbf{B}$ are a quarter cycle out of phase, so the Poynting vector reverses sign twice per cycle: energy surges out and flows back, netting almost nothing. This reactive slosh is stored energy, not radiation, and it is why antenna engineers speak of reactive near fields and why an antenna's input impedance has a reactance to tune out. Hertz's own 1889 figures show the transition: near-zone field loops that stretch, pinch off at $r \sim \lambda/2\pi$ , and swim away as radiation — the most instructive figure in classical electrodynamics, and the exact content of the terms Derivation B dropped, kept.

Is the atom-emitting-light problem really this classical dipole?

Structurally yes, numerically almost. An excited atom's electron cloud oscillates between states at the transition frequency, forming a dipole moment whose magnitude quantum mechanics supplies (the matrix element $\langle f|q\mathbf{r}|i\rangle$ replaces $p_0/\sqrt2$ -style amplitudes); insert that into this article's $P$ and divide by $\hbar\omega$ per photon, and you have the spontaneous-emission rate to within the bookkeeping factors QED makes precise — the classical $\omega^4$ power becoming the $\omega^3$ Einstein $A$ coefficient because each photon costs $\hbar\omega$ . The classical estimate of optical lifetimes (~16 ns, computed in the advanced-potentials article by the equivalent oscillator route) lands on measured strong transitions. The genuinely quantum content is which transitions have nonzero moments: selection rules are statements about matrix elements, i.e. about which "antennas" an atom is allowed to build — electric-dipole-forbidden transitions then radiate by the ladder's higher rungs, at rates this chapter's next articles price.

Why is the radiation resistance called a resistance when nothing is dissipating?

Because from the transmitter's terminals it is operationally indistinguishable from one. The generator drives current $I_0\cos\omega t$ into the antenna and delivers average power $\tfrac12I_0^2R_{\text{rad}}$ that never comes back — Ohmic bookkeeping, satisfied exactly, with the "heat" deposited in the electromagnetic field at infinity rather than in a resistor's lattice. The concept earns its keep in design: the antenna's total input resistance is $R_{\text{rad}} + R_{\text{loss}}$ (real wire heating), and the ratio $R_{\text{rad}}/(R_{\text{rad}}+R_{\text{loss}})$ is the radiation efficiency. The short dipole's plunge as $(d/\lambda)^2$ is the practical tragedy of small antennas: at $d/\lambda = 0.01$ , $R_{\text{rad}} = 0.08\ \Omega$ , easily swamped by loss — why AM towers are enormous, why your fitness tracker's Bluetooth antenna is a compromise, and why electrically small antenna design is a dark art with its own theorems (Chu's limit) bounding what miniaturization can achieve.

Does the sin-squared donut mean a single dipole can never make a directional beam?

A single short dipole, correct — its pattern is fixed by the transverse-projection geometry, gain a modest 1.5. Directionality comes from arrays: many dipoles driven with chosen phases, whose fields add coherently in favored directions and cancel elsewhere. The retardation-across-the-source skew that approximation (ii) suppressed within one small dipole is deliberately engineered between dipoles — element spacing and phase gradients steer the interference, which is how a phased-array radar sweeps its beam electronically and how a 5G massive-MIMO panel aims narrow lobes at individual phones. The donut remains the array's atom: the total pattern is (element pattern) × (array factor), with this article supplying the first factor. Yagi antennas, dish feeds, and radio-telescope arrays are all the same theorem — superposition applied to the formula derived here — scaled from rooftops to continents.

How does the magnetic dipole rung compare — when does the next article's radiation matter?

Take the same physical footprint and price both. A current loop of radius $b$ carrying $I_0\cos\omega t$ has magnetic moment $m_0 = \pi b^2I_0$ and radiates (next article) $P_m = \mu_0m_0^2\omega^4/12\pi c^3$ — same structure, two extra powers of $c$ . Its ratio to an electric dipole of comparable construction works out to $(m_0/cp_0)^2 \sim (\omega b/c)^2 = (2\pi b/\lambda)^2$ : the magnetic rung is suppressed by the same small parameter — source size over wavelength, squared — that organizes the whole ladder. Consequences: whenever a system has an oscillating electric dipole moment, that channel dominates; magnetic dipole radiation rules only when symmetry kills the electric moment — closed current loops, certain atomic "forbidden" transitions (the 21 cm hydrogen line is a magnetic-dipole transition, hence its 10-million-year lifetime and its usefulness as an astronomical clock). The quadrupole rung sits at the same suppression order, and disentangling the two is the business of the next two articles.