Quadrupole Formula

The quadrupole formula is the bridge between a moving lump of matter and the ripples of spacetime it sends across the universe. The two preceding articles of this thread did the heavy lifting: the gravitational wave equation shows the trace-reversed perturbation obeys $\Box\bar h_{\mu\nu}=-(16\pi G/c^4)T_{\mu\nu}$ in Lorenz gauge, and the transverse-traceless gauge isolates the two physical polarizations that propagate. This article asks the next question — given a source, exactly how much radiation comes out, and in what form? Solving with the retarded Green's function and expanding in the far zone, the answer collapses onto a single object: the second time derivative of the source's mass quadrupole moment. That the answer starts at quadrupole order is one of the deep structural facts of gravity. The mass monopole is the total energy, which is conserved — so it cannot radiate. The mass dipole is the centre of mass, whose rate of change is the conserved total momentum — so it cannot radiate either. The current dipole is the conserved angular momentum. The first moment whose second time derivative need not vanish is the mass quadrupole, and this conservation-law argument mirrors exactly the group-theoretic statement that gravitons carry helicity $\pm2$ : a spin- $s$ field radiates first at multipole $\ell=s$ , so gravity begins at $\ell=2$ . Feed the resulting field into the Isaacson stress-energy tensor, integrate over a sphere, and you recover Einstein's 1918 luminosity — the seed of every inspiral prediction, from the Hulse–Taylor pulsar to GW150914.

Key takeaways

The quadrupole formula gives the leading gravitational-wave strain from a slow, weakly self-gravitating source as the second time derivative of its mass quadrupole: $\bar h_{ij}^{\mathrm{TT}}=(2G/c^4 r)\,\ddot I_{ij}^{\mathrm{TT}}(t-r/c)$ .
Monopole and dipole radiation are forbidden by conservation of energy and momentum, so emission starts at quadrupole order — the gravitational analogue of why the electromagnetic story starts at dipole order.
The radiated power is Einstein's 1918 result $L=(G/5c^5)\langle\dddot Q_{ij}\dddot Q_{ij}\rangle$ , with $Q_{ij}$ the trace-free (reduced) quadrupole; the tiny coupling $G/c^4\approx8.3\times10^{-45}\ \mathrm{s^2/(kg\,m)}$ is why the waves are so faint.
For a Keplerian circular binary it gives $L=(32/5)(G^4/c^5)\,m_1^2 m_2^2(m_1+m_2)/a^5$ — a steep $a^{-5}$ law that drives a runaway inspiral.
It is the lowest term of the post-Newtonian multipole expansion, valid when the source is slow ( $v\ll c$ ), weakly self-gravitating, and observed in the far zone.
Applied to the Hulse–Taylor binary pulsar it predicts orbital decay matching observation to about 0.2%, the first quantitative proof that gravitational waves carry energy.
Key idea: mass and momentum conservation kill monopole and dipole gravitational radiation; the leading emission is set by $\ddot I_{ij}$ , giving $h\sim(2G/c^4 r)\ddot I_{ij}$ and luminosity $L=(G/5c^5)\langle\dddot Q_{ij}\dddot Q_{ij}\rangle$ .

The core insight

Mass and momentum conservation kill monopole and dipole gravitational radiation; the leading emission is set by $\ddot I_{ij}$ , the second time derivative of the mass quadrupole, giving $h\sim(2G/c^4 r)\ddot I_{ij}$ and a radiated power $L=(G/5c^5)\langle\dddot Q_{ij}\dddot Q_{ij}\rangle$ .

Where this lives today

The quadrupole formula is not a museum piece — it is the leading term of every waveform model in use. The Hulse–Taylor pulsar remains its cleanest test: three decades of timing show the orbital period shrinking at the predicted $\dot P_b=-2.40\times10^{-12}\ \mathrm{s/s}$ , the $0.2\%$ agreement that won Hulse and Taylor the 1993 Nobel Prize. Since 2015, LIGO, Virgo, and KAGRA have caught the inspiral directly: the early, slowly-chirping phase of events like GW150914 and the neutron-star merger GW170817 is governed by the quadrupole term, and the detected frequency sweep is fit by post-Newtonian expansions that start from it. The leading chirp depends only on the chirp mass $\mathcal M=(m_1 m_2)^{3/5}/(m_1+m_2)^{1/5}$ , read straight out of the quadrupole luminosity. Continuous-wave searches use the same physics to hunt the tiny time-varying quadrupole of a deformed spinning neutron star ( $\epsilon\lesssim10^{-6}$ ), and pulsar-timing arrays such as NANOGrav apply it at nanohertz frequencies to supermassive black-hole binaries in distant galactic nuclei. From a millisecond pulsar to LISA, one 1918 formula sets the leading amplitude everywhere.

The physical picture

Picture the source as a time-varying mass distribution and ask which shape changes can launch a wave. A sphere that merely breathes changes its monopole moment — but the monopole is the total mass-energy, conserved, so it radiates nothing. A blob whose centre slides back and forth changes its dipole moment — but the derivative of the dipole is the total momentum, also conserved, so it radiates nothing either. Only a distribution that stretches along one axis while squeezing another — a changing quadrupole, exactly what two orbiting masses produce — deforms the surrounding geometry into an outward transverse-traceless ripple. The first figure lines up the three cases and carries the strain relation and the minute coupling constant that governs its size.

Figure

Why gravity radiates only at quadrupole order: a breathing sphere (mass monopole) cannot radiate because total mass is conserved, and a sloshing centre of mass (mass dipole) cannot radiate because momentum is conserved — only a changing quadrupole such as an orbiting binary launches a wave, with strain h ≈ (2G/c⁴r) Ï_ij and the minute coupling G/c⁴ ≈ 8.3×10⁻⁴⁵ s²/(kg·m) explaining why the waves are so faint

The magnitude is set by the prefactor $2G/c^4$ . Squeezing the "mass × acceleration × length" object $\ddot I_{ij}$ through it gives a dimensionless strain $h$ , and because $G/c^4\approx8.3\times10^{-45}\ \mathrm{s^2/(kg\,m)}$ is so staggeringly small, only quadrupoles built from solar masses moving at a good fraction of $c$ radiate a strain we can measure. The luminosity, quadratic in the third derivative of the quadrupole, becomes a steep power of orbital frequency for a binary, so emission climbs sharply as the orbit shrinks — the runaway that ends in merger. The most celebrated test is the second figure: three decades of pulse timing on the Hulse–Taylor binary pulsar, whose orbit decays along exactly the parabola the quadrupole formula predicts.

Figure

Hulse–Taylor binary pulsar PSR B1913+16: the cumulative shift of periastron time versus year. The curve is the general-relativity prediction from the quadrupole formula, with orbital-period decay Ṗ_b = −2.40×10⁻¹² s/s; the points are three decades of pulse-timing data, which track the parabola to about 0.2 percent

The core equations

The starting point is the exact retarded solution of the linearized field equation in Lorenz gauge: the field at a point is a sum over the source's stress-energy at the retarded time. This is the gravitational analogue of the retarded potentials of electromagnetism, and every result below is an approximation of it in the far, slow-motion zone.

\bar h_{\mu\nu}(t,\mathbf{x}) = \frac{4G}{c^4}\int \frac{T_{\mu\nu}(t-|\mathbf{x}-\mathbf{x}'|/c,\,\mathbf{x}')}{|\mathbf{x}-\mathbf{x}'|}\,\mathrm{d}^3x' \qquad (1)

where: $\bar h_{\mu\nu}$ — trace-reversed perturbation, $\bar h_{\mu\nu}=h_{\mu\nu}-\tfrac12\eta_{\mu\nu}h$ ; $T_{\mu\nu}$ — stress-energy of the source; $G,c$ — Newton's constant and light speed; $r=|\mathbf{x}|$ — distance to the field point. The geometry at $(t,\mathbf x)$ is dictated by what the source was doing one light-travel-time earlier. Expanding equation (1) in the wave zone and using conservation of stress-energy to eliminate the awkward space-space integral collapses the whole source onto one tensor — the quadrupole, the formula the article is named for.

\bar h_{ij}(t,\mathbf{x}) = \frac{2G}{c^4 r}\,\ddot{I}_{ij}(t-r/c) \qquad (2)

where: $I_{ij}$ — mass quadrupole moment; $\ddot I_{ij}$ — its second time derivative at retarded time $t-r/c$ ; $r$ — distance to the source. Reading it: the strain falls as $1/r$ (an amplitude, not an intensity) and is fixed by how fast the quadrupole is accelerating. The quadrupole itself is a second moment of the mass distribution. For radiation one uses its trace-free part, the reduced quadrupole, because the physical field is transverse-traceless and the trace never contributes.

I_{ij} = \int \rho(\mathbf{x})\, x_i x_j \,\mathrm{d}^3x,\qquad Q_{ij} = I_{ij} - \tfrac13 \delta_{ij}\, I_{kk} \qquad (3)

where: $\rho$ — mass (energy) density; $x_i$ — Cartesian coordinate from the centre of mass; $I_{kk}$ — trace of the mass quadrupole. The reduced $Q_{ij}$ discards the isotropic "breathing" part that carries no radiation. Not every component of $\ddot I_{ij}$ is real radiation. Only the part transverse to the line of sight and trace-free survives; the transverse-traceless projector $\Lambda_{ij,kl}$ extracts it along the propagation direction $\hat{\mathbf n}$ .

h_{ij}^{\mathrm{TT}} = \frac{2G}{c^4 r}\,\Lambda_{ij,kl}(\hat{\mathbf{n}})\,\ddot{I}_{kl},\qquad \Lambda_{ij,kl}=P_{ik}P_{jl}-\tfrac12 P_{ij}P_{kl},\quad P_{ij}=\delta_{ij}-n_i n_j \qquad (4)

where: $P_{ij}$ — projector onto the plane orthogonal to $\hat{\mathbf n}$ ; $\Lambda_{ij,kl}$ — TT projector, transverse on each index and overall trace-free. Because $\Lambda$ kills the trace, $\ddot I_{ij}$ and $\ddot Q_{ij}$ give identical TT parts — either may be used. Feeding the TT field into the Isaacson effective stress-energy and integrating over a large sphere gives the total radiated power — Einstein's 1918 quadrupole luminosity, the master formula for gravitational-wave energy loss.

L = \frac{\mathrm{d}E}{\mathrm{d}t} = \frac{G}{5c^5}\left\langle \dddot{Q}_{ij}\,\dddot{Q}_{ij}\right\rangle \qquad (5)

where: $L$ — radiated power; $\dddot Q_{ij}$ — third time derivative of the reduced quadrupole; $\langle\cdot\rangle$ — average over several wave periods. Reading it: the power is quadratic in the third derivative, so slow, gentle quadrupoles radiate almost nothing. Inserting a Keplerian two-body orbit into equation (5) gives the single most useful special case in all of gravitational-wave astronomy — the luminosity of a circular binary.

L_{\mathrm{binary}} = \frac{32}{5}\frac{G^4}{c^5}\frac{m_1^2 m_2^2 (m_1+m_2)}{a^5} \qquad (6)

where: $m_1,m_2$ — component masses; $a$ — orbital separation. Reading it: the $a^{-5}$ dependence means the emission rises explosively as the orbit tightens — the physics behind the inspiral chirp.

Derivation AFrom the retarded integral to the quadrupole formula

The goal is to reduce the exact retarded solution — a spatial integral over the entire stress-energy tensor at retarded time — down to a single object, the second time derivative of the mass quadrupole, valid in the far, slow-motion zone. The essential trick is that the space-space integral $\int T^{ij}\,\mathrm{d}^3x$ is not independent information: conservation $\partial_\mu T^{\mu\nu}=0$ chains it to the second time derivative of $\int T^{00}x^i x^j\,\mathrm{d}^3x$ , which is nothing but the mass quadrupole. We recall the retarded field, apply this tensor virial identity, and read off the result.

Step 1. In the wave zone the source is far smaller than the distance $r$ , so $|\mathbf x-\mathbf x'|\approx r$ in the slowly varying $1/|\mathbf x-\mathbf x'|$ kernel, which comes outside the integral, and the retarded time may be evaluated at the source centre: $\bar h_{ij}\approx(4G/c^4 r)\int T_{ij}(t-r/c,\mathbf x')\,\mathrm{d}^3x'$ . Why: at large distance the field is a pure source integral times a spreading $1/r$ envelope, exactly as for a radiating antenna.

Step 2. Conservation supplies the tensor virial identity $\int T^{ij}\,\mathrm{d}^3x=\tfrac12\,\tfrac{\mathrm d^2}{\mathrm dt^2}\int T^{00}x^i x^j\,\mathrm{d}^3x$ . One proves it by differentiating $\int T^{00}x^i x^j$ twice in time and using $\partial_0 T^{00}+\partial_k T^{0k}=0$ and $\partial_0 T^{0i}+\partial_k T^{ki}=0$ , integrating by parts and dropping surface terms for a localized source. Why: it trades an integral of internal stresses we cannot evaluate for time derivatives of a mass moment we can.

Step 3. To leading Newtonian order the energy density is $T^{00}=\rho c^2$ , so $\int T^{00}x^i x^j\,\mathrm{d}^3x=c^2\int\rho\,x^i x^j\,\mathrm{d}^3x=c^2 I^{ij}$ , giving $\int T^{ij}\,\mathrm{d}^3x=\tfrac{c^2}{2}\ddot I^{ij}$ . Why: this is the step where the abstract stress integral becomes the concrete, measurable mass quadrupole of the source.

Step 4. Substituting back, $\bar h_{ij}=(4G/c^4 r)\cdot(c^2/2)\,\ddot I_{ij}=(2G/c^4 r)\,\ddot I_{ij}$ , evaluated at retarded time $t-r/c$ . Projecting with $\Lambda_{ij,kl}$ removes the trace and leaves the physical TT strain. Why: the two factors of $\tfrac12$ and the $4G$ collapse to the clean prefactor $2G/c^4$ — the quadrupole formula. *Verified: the tensor virial relation follows from $\partial_\mu T^{\mu\nu}=0$ by repeated integration by parts; see MTW §36.7 and Wald §4.4.*

Go deeper · Derivation B · 1 minEinstein's luminosity from the Isaacson flux

Derivation BEinstein's luminosity from the Isaacson flux

The goal is to convert the radiation field into a single number — the total power crossing a large sphere — using the Isaacson effective stress-energy tensor, whose energy flux is quadratic in the time derivative of the TT strain. The one genuinely nontrivial ingredient is the angular integral of the transverse-traceless projector, which produces a characteristic combination of Kronecker deltas; contracting that against the trace-free third derivative of the quadrupole delivers Einstein's prefactor $G/5c^5$ .

Step 1. The Isaacson flux per solid angle is $\mathrm dE/(\mathrm dt\,\mathrm d\Omega)=(c^3 r^2/32\pi G)\langle\dot h_{ij}^{\mathrm{TT}}\dot h_{ij}^{\mathrm{TT}}\rangle$ . Why: the effective stress-energy of the wave, averaged over several wavelengths, is quadratic in field gradients — the gravitational analogue of the Poynting flux $\propto E^2$ .

Step 2. Insert $\dot h_{ij}^{\mathrm{TT}}=(2G/c^4 r)\,\Lambda_{ij,kl}\dddot Q_{kl}$ ; the explicit $r^2$ in the flux cancels the $1/r^2$ from the amplitude, leaving a finite power. Using $\Lambda_{ij,kl}\Lambda_{ij,mn}=\Lambda_{kl,mn}$ gives $\langle\dot h_{ij}\dot h_{ij}\rangle=(4G^2/c^8 r^2)\Lambda_{kl,mn}\langle\dddot Q_{kl}\dddot Q_{mn}\rangle$ . Why: the projector is idempotent, so two projections are one — the standard simplification for TT fields.

Step 3. The remaining angular integral is the standard result $\int\Lambda_{kl,mn}\,\mathrm d\Omega=(2\pi/15)(11\delta_{km}\delta_{ln}-4\delta_{kl}\delta_{mn}+\delta_{kn}\delta_{lm})$ . Contracting with the symmetric trace-free $\dddot Q_{kl}\dddot Q_{mn}$ : the $11$ -term gives $11\,\dddot Q_{kl}\dddot Q_{kl}$ , the $-4$ trace term vanishes, and the last term gives $+1\,\dddot Q_{kl}\dddot Q_{kl}$ , totalling $12\,\dddot Q_{kl}\dddot Q_{kl}$ . Why: trace-freeness of $Q_{ij}$ annihilates the middle term and is what makes the whole calculation close.

Step 4. Assembling, $L=(c^3/32\pi G)(4G^2/c^8)(2\pi/15)(12)\langle\dddot Q_{ij}\dddot Q_{ij}\rangle=(G/5c^5)\langle\dddot Q_{ij}\dddot Q_{ij}\rangle$ . Why: the numerical dust — $32\pi$ , $2\pi/15$ , and $12$ — settles to the compact $1/5$ , giving Einstein's 1918 luminosity. *Verified: the contraction yields $(11+1)=12$ and the constants reduce to $G/5c^5$ ; cross-check MTW §36.7, Maggiore §3.3.*

Go deeper · Derivation C · 1 minThe circular binary and the a to the minus fifth law

Derivation CThe circular binary and the a to the minus fifth law

The goal is to feed a concrete source — two point masses on a circular Keplerian orbit — into Einstein's luminosity formula and extract the celebrated inverse-fifth-power scaling that governs every inspiral. We build the mass quadrupole of the orbit, differentiate it three times, average over an orbit, and eliminate the orbital frequency for the separation through Kepler's third law. The steep $a^{-5}$ dependence that emerges is what makes an inspiral a runaway rather than a slow drift.

Step 1. Reduce the two-body problem to one body of reduced mass $\mu=m_1 m_2/(m_1+m_2)$ moving on a circle of radius $a$ with $\omega^2=G(m_1+m_2)/a^3$ , so its position is $x=a\cos\omega t$ , $y=a\sin\omega t$ . Why: the relative motion carries all the time dependence; the centre of mass is fixed and contributes nothing.

Step 2. The mass quadrupole is $I_{xx}=\mu a^2\cos^2\omega t$ , $I_{yy}=\mu a^2\sin^2\omega t$ , $I_{xy}=\mu a^2\cos\omega t\sin\omega t$ . Double-angle identities show every component oscillates at $2\omega$ . Why: the quadrupole returns to itself after half an orbit, so gravitational waves come out at twice the orbital frequency — a hallmark of quadrupole emission.

Step 3. Differentiating the trace-free $Q_{ij}$ three times and averaging the contraction over an orbit gives $\langle\dddot Q_{ij}\dddot Q_{ij}\rangle=32\,\mu^2 a^4\omega^6$ . Why: three derivatives each bring down a factor $2\omega$ , so the result scales as $\omega^6$ — the origin of the extreme frequency sensitivity of the luminosity.

Step 4. Insert into equation (5): $L=(G/5c^5)\cdot32\,\mu^2 a^4\omega^6$ , then substitute $\omega^6=[G(m_1+m_2)/a^3]^3$ and $\mu^2=m_1^2 m_2^2/(m_1+m_2)^2$ to reach $L=(32/5)(G^4/c^5)\,m_1^2 m_2^2(m_1+m_2)/a^5$ . Why: Kepler's law turns the frequency dependence into a pure separation dependence, exposing the $a^{-5}$ runaway that drives coalescence. *Verified: reproduces the Peters–Mathews (1963) circular-orbit result; cross-check Maggiore §4.1 and MTW §36.6. The two independent routes to $L$ in Worked example 1 agree to the digit.*

Limiting cases

The quadrupole formula is the lowest term of a controlled expansion; its regime of validity and the cases where it collapses are as instructive as the formula itself.

Case	Result	Meaning
Slow-motion / weak-field	$v/c\ll1,\ \lambda_{\mathrm{GW}}\gg R_{\mathrm{source}}$	The expansion parameter is $v/c$ ; higher multipoles and post-Newtonian corrections enter at successive powers of it.
Spherical symmetry	$I_{ij}\propto\delta_{ij}\Rightarrow Q_{ij}=0\Rightarrow L=0$	A spherical (or purely radial) collapse or pulsation radiates nothing — Birkhoff's theorem in radiative form.
Axisymmetric rigid rotation	$\ddot I_{ij}=0\Rightarrow L=0$	A body spinning about its symmetry axis has a static quadrupole and radiates nothing; a mountain, wobble, or companion is required.
Circular binary	$L=\tfrac{32}{5}\tfrac{G^4}{c^5}\tfrac{m_1^2 m_2^2(m_1+m_2)}{a^5}$	The canonical inspiral result — a Keplerian quadrupole inserted into Einstein's luminosity.

Common trapTwo traps: which quadrupole, and where the formula is licensed

First, do not mix conventions between the full quadrupole $I_{ij}$ and the reduced (trace-free) $Q_{ij}=I_{ij}-\tfrac13\delta_{ij}I_{kk}$ . They give the same strain and the same luminosity only because the transverse-traceless projection annihilates the trace — but drop a non-projected $I_{ij}$ straight into the $G/5c^5$ formula and you get wrong numerical factors. Pick $Q_{ij}$ and stay consistent. Second, remember what the derivation assumed: a slow, weakly self-gravitating source in the far zone. Those assumptions are each individually violated by a relativistic compact binary near merger. The formula's success there is not automatic — it is licensed by the post-Newtonian program, which shows the quadrupole term remains the correct leading order in $v/c$ while self-gravity is absorbed into the equations of motion. Far from coalescence it is excellent; through merger you need numerical relativity.

Worked examples

1 · Luminosity, period, and lifetime of a neutron-star binary (fully worked)

Two neutron stars, $m_1=m_2=1.4\,M_\odot$ , orbit each other on a circle of separation $a=1.0\times10^9\ \mathrm m$ (a million kilometres). Find the orbital period, the gravitational-wave luminosity, and the time to coalescence. Take $M_\odot=1.989\times10^{30}\ \mathrm{kg}$ .

Step 1. Masses and Kepler. Each star is $m=1.4\times1.989\times10^{30}=2.785\times10^{30}\ \mathrm{kg}$ , total $M=5.569\times10^{30}\ \mathrm{kg}$ . Kepler gives $\omega^2=GM/a^3=3.72\times10^{-7}\ \mathrm{s^{-2}}$ , so $\omega=6.10\times10^{-4}\ \mathrm{rad/s}$ and the orbital period is $T_{\mathrm{orb}}=2\pi/\omega=1.03\times10^4\ \mathrm s\approx2.9\ \mathrm{hours}$ . Gravitational waves emerge at twice this, $f_{\mathrm{GW}}\approx1.9\times10^{-4}\ \mathrm{Hz}$ — in the future LISA band.

Step 2. Luminosity. Insert into equation (6): with $m_1^2 m_2^2=m^4=6.01\times10^{121}\ \mathrm{kg^4}$ , $(m_1+m_2)=5.569\times10^{30}$ , $a^5=10^{45}$ , $G^4=1.98\times10^{-41}$ , $c^5=2.42\times10^{42}$ ,

L=\frac{32}{5}\,\frac{(1.98\times10^{-41})}{(2.42\times10^{42})}\,\frac{(6.01\times10^{121})(5.569\times10^{30})}{10^{45}}=\mathbf{1.8\times10^{25}\ W}.

As a cross-check, the equivalent form $L=(32/5)(G/c^5)\mu^2 a^4\omega^6$ with $\mu=m/2$ gives the identical $1.76\times10^{25}\ \mathrm W$ . That is about 5% of the Sun's electromagnetic luminosity ( $3.8\times10^{26}\ \mathrm W$ ), radiated purely as spacetime ripples.

Step 3. Lifetime. The Peters coalescence time for a circular orbit is $t_c=\dfrac{5}{256}\dfrac{c^5 a^4}{G^3 m_1 m_2(m_1+m_2)}$ . Numerically $t_c=\dfrac{5}{256}\dfrac{(2.42\times10^{42})(10^{36})}{(2.97\times10^{-31})(7.76\times10^{60})(5.569\times10^{30})}=3.7\times10^{15}\ \mathrm s\approx\mathbf{1.2\times10^{8}\ years}$ . This wide pair will spiral together in about 120 million years, the emission accelerating without limit as $a$ shrinks and $L\propto a^{-5}$ climbs.

2 · Can a laboratory source radiate detectably? (order of magnitude)

Spin a dumbbell: two $1000\ \mathrm{kg}$ masses on the ends of a rigid $2\ \mathrm m$ bar (each at radius $r=1\ \mathrm m$ from the axis), rotating at $\omega=30\ \mathrm{rad/s}$ — near the bursting limit of steel. What gravitational-wave power does it emit?

Step 1. A rigid dumbbell is two equal masses at radius $r$ — a "binary" of separation $a=2r$ and reduced mass $\mu=m/2$ , but with $\omega$ set by the motor, not gravity. Substituting into $L=(32/5)(G/c^5)\mu^2 a^4\omega^6$ collapses to

L=\frac{128}{5}\,\frac{G}{c^5}\,m^2 r^4\omega^6.

Step 2. Numbers: $G/c^5=2.76\times10^{-53}$ , $m^2=10^6\ \mathrm{kg^2}$ , $r^4=1\ \mathrm{m^4}$ , $\omega^6=30^6=7.29\times10^8$ . Hence $L=(128/5)(2.76\times10^{-53})(10^6)(7.29\times10^8)=\mathbf{5\times10^{-37}\ W}$ .

Step 3. Compare with example 1: $10^{25}\ \mathrm W$ versus $5\times10^{-37}\ \mathrm W$ — a gap of 62 orders of magnitude. The verdict is unambiguous: no conceivable table-top quadrupole radiates a measurable amount. The $G/c^4$ coupling is so tiny that only stellar-scale masses moving at relativistic speeds produce detectable strain — which is why gravitational-wave astronomy points telescopes at neutron stars and black holes, not laboratory rotors.

Physics of the portrait

The quadrupole formula is Einstein's, and its history is a cautionary tale about how hard it is to be sure a theory's own predictions are real.

Origins · 1 minAlbert Einstein and how this was discovered

Albert Einstein

1879–1955 · the quadrupole formula and the reality of gravitational radiation

Einstein predicted gravitational waves in 1916 by linearizing the field equations — but his first paper carried an algebra error that even allowed a spurious monopole term. He corrected it in 1918, deriving the quadrupole formula $L=(G/5c^5)\langle\dddot Q_{ij}\dddot Q_{ij}\rangle$ that still bears his name. Yet he never fully trusted it: coordinate freedom made it unclear whether the waves carried energy or were gauge artefacts — Eddington quipped that some travelled "at the speed of thought." In 1936 Einstein and Nathan Rosen submitted a paper titled Do Gravitational Waves Exist? arguing they did not; a referee (Howard Percy Robertson) caught the error, Einstein withdrew it, and the sign flipped. The reality of the waves was settled only at the 1957 Chapel Hill conference, where Feynman's "sticky-bead" argument showed a passing wave does mechanical work. Isaacson's 1968 effective stress-energy finally made the energy flux rigorous, vindicating the 1918 formula — which the Hulse–Taylor pulsar confirmed two decades later.

History

1916

— Einstein predicts gravitational waves from the linearized field equations

1918

— Einstein publishes the quadrupole formula for the radiated power

1963

— Peters and Mathews work out the emission from Keplerian binary orbits

1974

— Hulse and Taylor discover the binary pulsar PSR B1913+16

2016

— LIGO announces GW150914, the first direct detection of a binary inspiral

Going further

Texts. Misner, Thorne & Wheeler, Gravitation (1973), §36.1–36.10 · Wald, General Relativity (1984), §4.4 · Carroll, Spacetime and Geometry (2004), §7.4–7.6 · Hartle, Gravity (2003), Ch. 16 · Schutz, A First Course in General Relativity (2009), §9.3 · Maggiore, Gravitational Waves Vol. 1 (2007), Ch. 3–4 · Einstein, A. (1918), Sitzungsber. Preuss. Akad. Wiss., 154.

Provenance. Prefactors were cross-checked against MTW §36, Wald §4.4, Carroll §7.5, and Maggiore Ch. 3–4, and every worked number recomputed. A verifier should reproduce: (1) the tensor virial identity from $\partial_\mu T^{\mu\nu}=0$ giving $\bar h_{ij}=(2G/c^4 r)\ddot I_{ij}$ ; (2) the angular integral $\int\Lambda_{ij,kl}\,\mathrm d\Omega=(2\pi/15)(11\delta_{ik}\delta_{jl}-4\delta_{ij}\delta_{kl}+\delta_{il}\delta_{jk})$ contracting to the factor $12$ and hence $L=(G/5c^5)\langle\dddot Q_{ij}\dddot Q_{ij}\rangle$ ; (3) the Peters–Mathews binary result; (4) the example-1 chain $\omega=6.10\times10^{-4}\ \mathrm{rad/s}$ , $L=1.8\times10^{25}\ \mathrm W$ , $t_c\approx1.2\times10^8\ \mathrm{yr}$ , and the dumbbell power $5\times10^{-37}\ \mathrm W$ . The Hulse–Taylor figure ( $\sim0.2\%$ , $\dot P_b=-2.40\times10^{-12}\ \mathrm{s/s}$ ) is the standard Taylor–Weisberg value.

Prerequisites. Linearized Einstein Equations · Metric Perturbation and Trace-Reversed Field · Lorenz (Harmonic) Gauge · Transverse-Traceless Gauge · Gravitational Wave Equation. Leads to. Gravitational-Wave Luminosity and Energy Flux · Isaacson Stress-Energy of Gravitational Waves · Multipole Expansion of Gravitational Radiation · Plus and Cross Polarizations.

Ask the tutor: Want to see the $a^{-5}$ luminosity turned into a full inspiral chirp, or work through why the transverse-traceless projector lets you swap $I_{ij}$ for $Q_{ij}$ ?

Why does gravitational radiation start at quadrupole order, when light starts at dipole order?

Because the lowest gravitational "charges" are conserved. In electromagnetism the total charge is conserved but the charge dipole need not be — an oscillating dipole radiates, the leading term. In gravity the analogue of charge is mass-energy, whose monopole (total mass) and dipole (the centre of mass, tied to the conserved momentum) are both fixed by conservation laws, so their time derivatives vanish and neither can radiate. The first moment free to change is the mass quadrupole. Equivalently, the graviton is spin-2, and a spin- $s$ field radiates first at multipole $\ell=s$ , so gravity begins at $\ell=2$ . Conservation laws and helicity are the same fact in different languages.

Why may I use either the full quadrupole I_ij or the trace-free Q_ij?

Because the physical radiation field is transverse-traceless, and the TT projector $\Lambda_{ij,kl}$ that extracts it annihilates any pure-trace part. Writing $I_{ij}=Q_{ij}+\tfrac13\delta_{ij}I_{kk}$ , the trace piece $\propto\delta_{ij}$ is killed by $\Lambda$ , so $\Lambda_{ij,kl}\ddot I_{kl}=\Lambda_{ij,kl}\ddot Q_{kl}$ — identical strain, and the luminosity (a contraction of two trace-free objects) is likewise unchanged. The catch is consistency: the $G/5c^5$ formula is written for the reduced quadrupole, so feeding it $I_{ij}$ without removing the trace picks up spurious factors. Take the trace-free part once, and it does not matter which symbol you started with.

Does a spherically symmetric supernova collapse emit gravitational waves?

No. A spherically symmetric mass distribution has $I_{ij}\propto\delta_{ij}$ , so its reduced quadrupole $Q_{ij}$ is identically zero and $L=0$ — no matter how violent the infall. This is the radiative content of Birkhoff's theorem: the vacuum exterior of any spherical source is the static Schwarzschild geometry, which carries no waves. Real core-collapse supernovae do radiate, but only through their departures from symmetry — rotational flattening, convective overturn, the standing accretion-shock instability, asymmetric neutrino emission. The gravitational-wave signal is therefore a direct probe of the multidimensional dynamics of collapse, which is why it is worth chasing for a nearby galactic supernova.

The formula assumes slow, weak sources — so why does it work for black-hole binaries?

Strictly, the derivation assumes $v\ll c$ , weak self-gravity, and far-zone observation, and a compact binary near merger violates all three. The resolution is the post-Newtonian program: when you organize general relativity as an expansion in $v/c$ , the quadrupole term is the correct leading order for the radiation, while the strong self-gravity of the bodies is absorbed into their equations of motion rather than spoiling the radiation formula. Far from coalescence — through the long inspiral that dominates a detected signal's cycles — the quadrupole luminosity is accurate, with higher multipoles and PN corrections adding calculable refinements. Only through the final plunge and merger does the expansion break down and numerical relativity take over.

How does the a⁻⁵ scaling produce the inspiral "chirp"?

Combine the luminosity $L\propto a^{-5}$ with the orbital energy $E=-Gm_1 m_2/2a$ . Energy balance $\dot E=-L$ forces the separation to shrink, $\dot a<0$ , and because $L$ grows as $a$ falls, the shrinkage accelerates — a runaway. As $a$ drops, Kepler's law drives the orbital frequency up, so the gravitational-wave frequency sweeps upward: the "chirp." Integrating gives a coalescence time $t_c\propto a_0^4$ and a frequency evolution set by one parameter, the chirp mass $\mathcal M$ , which detectors read from the rate of the sweep — the reason the earliest, quadrupole-dominated part of a signal already pins down the binary's masses.

Is this Einstein's own formula — and is it true he once argued gravitational waves don't exist?

Yes on both counts. The luminosity $L=(G/5c^5)\langle\dddot Q_{ij}\dddot Q_{ij}\rangle$ is Einstein's 1918 result, correcting a factor error in his 1916 prediction. And in 1936 he and Nathan Rosen submitted a paper, Do Gravitational Waves Exist?, concluding they do not — misled by a coordinate singularity in a plane-wave solution. A referee (Robertson) spotted the mistake; Einstein withdrew the claim, and the corrected analysis restored the waves. The deeper worry — whether the radiation carried real energy or was a gauge artefact — persisted until Feynman's 1957 sticky-bead argument and Isaacson's 1968 effective stress-energy settled it. The 1918 formula survived every round intact.