Parametrized Post-Newtonian (PPN) Formalism

Einstein's general relativity is not the only self-consistent theory of gravity built on curved spacetime. Scalar–tensor, vector–tensor, and bimetric theories all treat gravity as spacetime geometry and all reduce to Newton in the appropriate limit — so how does one experiment sift among them without recomputing light bending, time delay, and precession theory by theory? The Parametrized Post-Newtonian (PPN) formalism is the answer. Its strategy is disarmingly simple: write the weak-field, slow-motion metric any such theory produces as one fixed catalogue of gravitational potentials — sourced by mass, kinetic energy, internal energy, pressure, and the motion of matter — each times an arbitrary dimensionless coefficient. Every metric theory then collapses, at first post-Newtonian order, to a particular choice of those coefficients. Ten coefficients survive the gauge-fixing. Two, $\gamma$ and $\beta$ , carry the everyday physics: $\gamma$ measures how much space curvature a unit rest mass produces, $\beta$ the nonlinearity in gravity's superposition law. The other eight — $\xi$ , $\alpha_1$ , $\alpha_2$ , $\alpha_3$ , and $\zeta_1$ through $\zeta_4$ — are symmetry detectors that stay exactly zero unless gravity secretly cares where you sit in the universe, how fast you move through it, or whether total momentum can leak away. General relativity is the single point $\gamma=\beta=1$ with all eight others vanishing. The payoff is that abstract theory comparison becomes number-fitting. Light deflection and the Shapiro delay measure $\tfrac12(1+\gamma)$ ; Mercury's perihelion advance measures $\tfrac13(2+2\gamma-\beta)$ ; lunar laser ranging measures the Nordtvedt combination $4\beta-\gamma-3$ . Half a century of ever-sharper measurement has driven every parameter to its Einstein value — Cassini alone fixes $\gamma-1$ to two parts in a hundred thousand — making PPN the scoreboard on which general relativity has won every weak-field match.

Key takeaways

The PPN formalism expands the weak-field, slow-motion metric of ANY metric theory into a fixed set of standard potentials, each times a theory-dependent dimensionless coefficient.
Ten parameters survive: $\gamma,\beta,\xi,\alpha_1,\alpha_2,\alpha_3,\zeta_1,\zeta_2,\zeta_3,\zeta_4$ — spatial curvature per unit mass, nonlinearity of superposition, preferred-location and preferred-frame effects, and violations of total-momentum conservation.
General relativity is the unique point $\gamma=\beta=1$ with all eight others zero; any deviation flags a different theory of gravity.
Classic tests map straight onto the parameters: deflection and Shapiro delay give $\tfrac12(1+\gamma)$ , perihelion advance gives $\tfrac13(2+2\gamma-\beta)$ , the Nordtvedt effect gives $\eta_N=4\beta-\gamma-3-\dots$
Cassini fixes $\gamma-1=(2.1\pm2.3)\times10^{-5}$ ; lunar laser ranging gives $|\beta-1|\lesssim10^{-4}$ — GR passes every weak-field test.
PPN presupposes weak fields and slow motion ( $v\ll c$ ); it says nothing about strong-field or radiative regimes, where theories sharing all ten parameters can still differ wildly.
Key idea: write the weak-field metric of any metric theory as a fixed catalogue of potentials times arbitrary coefficients; the ten surviving coefficients — the PPN parameters — carry all the theory-dependence, and experiments measure them directly.

The core insight

Write the weak-field, slow-motion metric of ANY metric theory as a fixed catalogue of gravitational potentials times arbitrary coefficients; the ten surviving coefficients (the PPN parameters) then carry all the theory-dependence, and experiments measure them directly.

Where this lives today

The PPN parameters are the working currency of experimental gravitation. The tightest bound on $\gamma$ still comes from the 2002 Cassini superior-conjunction experiment, whose radio link grazed the Sun and returned $\gamma-1=(2.1\pm2.3)\times10^{-5}$ from the time-varying Shapiro delay; ESA's Gaia mission now measures the same spatial curvature astrometrically for all two billion of its stars. On $\beta$ , decades of lunar laser ranging hold $|\eta_N|$ near $10^{-4}$ , and planetary ephemerides fitted to MESSENGER (and soon BepiColombo) at Mercury pin the perihelion combination. The preferred-frame parameters are constrained far more tightly by systems that amplify any cosmic-rest-frame dependence: solar-spin alignment and pulsar timing give $|\alpha_1|\lesssim10^{-4}$ , $|\alpha_2|\lesssim10^{-7}$ , and the extraordinary $|\alpha_3|\lesssim10^{-20}$ . Gravity Probe B's gyroscopes measured frame-dragging at $37.2\pm7.2$ mas/yr against the GR prediction $39.2$ mas/yr, testing $\tfrac14(4\gamma+4+\alpha_1)$ directly. The open frontier lies where PPN cannot reach: binary pulsars and compact-object mergers probe the strong-field, radiative regime through the parametrized post-Keplerian formalism, which begins where the post-Newtonian expansion ends.

The physical picture

Think of a master template for gravity's weak field: a printed form with blanks. The blanks are the standard potentials — the Newtonian $U$ , plus potentials built from kinetic energy $\rho v^2$ , internal energy $\rho\Pi$ , pressure $p$ , and mass currents $\rho v_i$ — and every metric theory fills in the SAME blanks with different numbers in front. PPN names those numbers and lets experiment read them off. The diagram below shows the template, the two "dials" $\gamma$ and $\beta$ , and the eight symmetry detectors GR pins to zero.

Figure

The PPN template: any metric theory reduces to the same weak-field metric with ten coefficients. The two dials γ and β sit in g₀₀ = −1 + 2U − 2βU² + … and gᵢⱼ = (1 + 2γU)δᵢⱼ; the other eight parameters vanish identically in general relativity, whose point in the space is γ = β = 1

Two of the dials have a vivid reading. $\gamma$ asks: how much does a kilogram curve SPACE compared with how much it warps TIME? Newton curves only time ( $\gamma=0$ ); GR curves space and time equally ( $\gamma=1$ ), and light racing at $c$ samples both, so bending a ray weighs $\gamma$ directly. $\beta$ asks a different question: when you stack two masses, does the field of the pair equal the sum, or is there an extra nonlinear piece — does gravity gravitate? GR is mildly nonlinear ( $\beta=1$ ), and Mercury's orbit feels it. Everything experimental happens near the GR point, and the second figure shows how tightly two experiments have boxed it in.

Figure

The experimental box around general relativity in the γ–β plane. Cassini pins the vertical strip γ − 1 = (2.1 ± 2.3)×10⁻⁵; lunar laser ranging through the Nordtvedt effect pins the horizontal strip β − 1 ≲ 1×10⁻⁴. Their intersection, a sliver around the GR bullseye at γ = β = 1, is the surviving allowed region; the Brans–Dicke line β = 1 enters from the left as ω → ∞

The remaining six parameters never enter the Solar System's leading dynamics unless a symmetry is broken. $\xi$ would make local gravity depend on the surrounding matter distribution (a preferred location); $\alpha_1,\alpha_2,\alpha_3$ would make it depend on motion relative to a cosmic rest frame (a preferred frame); and $\alpha_3$ with $\zeta_1,\dots,\zeta_4$ would let a body's total momentum leak away. General relativity honours all of these symmetries exactly, so the six vanish — which is why the experimentally interesting story is so nearly a two-parameter one in $\gamma$ and $\beta$ .

The core equations

The formalism is organized by a single small parameter counting orders of smallness. In a bound, self-gravitating system the Newtonian potential, the squared velocity, the pressure-to-density ratio, and the specific internal energy are all comparable, so one bookkeeping symbol controls them together.

\epsilon \equiv U \sim v^2 \sim \frac{p}{\rho} \sim \Pi \ll 1 \qquad (1)

where: $\epsilon$ — the post-Newtonian expansion parameter (units $c=1$ ); $U$ — Newtonian potential, $U=\int \rho/|\mathbf{x}-\mathbf{x}'|\,\mathrm{d}^3x'$ ; $v$ — characteristic matter velocity; $p$ — pressure; $\rho$ — rest-mass density; $\Pi$ — specific internal energy. Reading: a term scaling as $\epsilon^n$ is $n$ -th post-Newtonian ( $n$ PN); PPN keeps everything through 1PN, exactly what the motion of planets and light rays requires. The heart of the formalism is the time-time metric component in the standard PPN gauge. It collects every 1PN potential, each weighted by a combination of the ten parameters — this equation is the "template."

g_{00} = -1 + 2U - 2\beta U^2 - 2\xi\,\Phi_W + (2\gamma+2+\alpha_3+\zeta_1-2\xi)\Phi_1 + 2(3\gamma-2\beta+1+\zeta_2+\xi)\Phi_2 + 2(1+\zeta_3)\Phi_3 + 2(3\gamma+3\zeta_4-2\xi)\Phi_4 - (\zeta_1-2\xi)\mathcal{A} - (\alpha_1-\alpha_2-\alpha_3)w^2 U - \alpha_2 w^i w^j U_{ij} + (2\alpha_3-\alpha_1)w^i V_i \qquad (2)

where: $\beta,\gamma,\xi,\alpha_i,\zeta_i$ — the ten PPN parameters; $\Phi_1,\Phi_2,\Phi_3,\Phi_4$ — potentials sourced by $\rho v^2$ , $\rho U$ , $\rho\Pi$ , $p$ ; $\Phi_W$ — the Whitehead anisotropy potential (preferred-location effects); $\mathcal{A}$ — a velocity-line-of-sight potential; $w^i$ — velocity of the PPN frame relative to the universe rest frame; $V_i, U_{ij}$ — momentum and tensor potentials. Reading: the only vacuum survivors far from an isolated mass are the leading $2U$ and the nonlinear $-2\beta U^2$ ; everything else needs matter, motion, or a broken symmetry to switch on. The time-space, or gravitomagnetic, component is where the motion of matter enters. Moving masses source it as moving charges source a magnetic field, and its coefficients set frame-dragging and gyroscope precession.

g_{0i} = -\tfrac12(4\gamma+3+\alpha_1-\alpha_2+\zeta_1-2\xi)\,V_i - \tfrac12(1+\alpha_2-\zeta_1+2\xi)\,W_i - \tfrac12(\alpha_1-2\alpha_2)w_i U - \alpha_2 w^j U_{ij} \qquad (3)

where: $V_i=\int \rho v_i/|\mathbf{x}-\mathbf{x}'|\,\mathrm{d}^3x'$ — momentum potential; $W_i$ — a second momentum potential built from the line-of-sight velocity. Reading: in GR the leading coefficient is $-\tfrac12(4\gamma+3)=-\tfrac72$ , and the Lense-Thirring gyroscope precession scales as $\tfrac14(4\gamma+4+\alpha_1)$ , the Einstein value when $\gamma=1$ , $\alpha_1=0$ . The space-space component is the simplest and most consequential: it carries $\gamma$ alone, the spatial metric being conformally flat to 1PN order.

g_{ij} = (1 + 2\gamma U)\,\delta_{ij} \qquad (4)

where: $\gamma$ — the space-curvature parameter multiplying the Newtonian potential in the spatial metric; $\delta_{ij}$ — the flat-space Kronecker delta. Reading: $\gamma$ is literally "how much spatial curvature per unit gravitational potential" — GR sets $\gamma=1$ , Newtonian gravity (flat space) sets $\gamma=0$ . This lone term is what light bending, the Shapiro delay, and geodetic precession all reach for. The first observable follows immediately: a light ray grazing a mass is bent by an angle in which $\gamma$ appears at full strength, because a photon at $c$ feels the spatial curvature as strongly as the temporal.

\delta\theta = \frac{1+\gamma}{2}\,\frac{4GM_\odot}{c^2 b} \qquad (5)

where: $\delta\theta$ — deflection angle; $M_\odot$ — deflector mass; $b$ — impact parameter. Reading: half the deflection comes from the theory-independent $g_{00}$ term and half from the $\gamma$ -carrying $g_{ij}$ term, giving $\tfrac12(1+\gamma)$ ; at the solar limb ( $\gamma=1$ ) this is $1.75''$ , and the identical prefactor governs the Shapiro delay $\Delta t=\tfrac12(1+\gamma)(4GM_\odot/c^3)\ln(4r_\oplus r_p/b^2)$ . Mercury's orbit brings in $\beta$ . The relativistic perihelion advance per revolution is a specific mixture of both dials.

\Delta\varpi = \frac{1}{3}\left(2+2\gamma-\beta\right)\frac{6\pi GM}{c^2 a(1-e^2)}\quad\text{per orbit} \qquad (6)

where: $\Delta\varpi$ — perihelion advance per orbit; $a$ — semimajor axis; $e$ — eccentricity; $M$ — central mass. Reading: the factor $\tfrac13(2+2\gamma-\beta)$ reduces to $1$ when $\gamma=\beta=1$ , recovering Einstein's $43''$ /century for Mercury; measuring the advance once $\gamma$ is known isolates $\beta$ . The last equation is the sharpest probe of $\beta$ and the strong equivalence principle: if a body's own binding energy feeds its inertial and gravitational mass differently, massive bodies fall differently — the Nordtvedt effect.

\eta_N = 4\beta - \gamma - 3 - \frac{10}{3}\xi - \alpha_1 + \frac{2}{3}\alpha_2 - \frac{2}{3}\zeta_1 - \frac{1}{3}\zeta_2 \qquad (7)

where: $\eta_N$ — the Nordtvedt parameter governing strong-equivalence-principle violation; the mass ratio is $m_g/m_i = 1 - \eta_N\,\Omega/(mc^2)$ with $\Omega$ the gravitational self-energy. Reading: $\eta_N=0$ in GR ( $4\cdot1-1-3=0$ ); a nonzero $\eta_N$ makes Earth and Moon fall toward the Sun at different rates, polarizing the lunar orbit — the signal lunar laser ranging hunts for.

Derivation AReading gamma and beta off a static point mass

The goal here is to show that the two everyday PPN parameters are not abstract fitting knobs but read directly off the metric of an isolated static mass, and that general relativity's Schwarzschild solution lands exactly on $\gamma=\beta=1$ . We write the exterior metric of a static spherical mass $M$ in isotropic coordinates through second order in $GM/c^2r$ , match it term by term to the PPN template, and compare with the expanded Schwarzschild solution.

Step 1. In vacuum outside the mass, every PPN potential built from moving matter, pressure, or internal energy vanishes, and the only survivor is the Newtonian $U=GM/c^2r$ . Why: the integrals $\Phi_1,\dots,\Phi_4$ and the momentum potentials all carry factors of $v$ , $\Pi$ , or $p$ , zero for a static exterior — only $U$ and its powers can appear.

Step 2. Insert $U$ into $g_{00}$ of equation (2). All that remains is $g_{00}=-1+2U-2\beta U^2$ , since the nonlinear $\beta U^2$ term is the one piece built purely from $U$ . Why: this exposes $\beta$ as the coefficient of the field's self-interaction — the deviation from the linear $-1+2U$ is entirely $\beta$ 's doing.

Step 3. Insert $U$ into $g_{ij}$ of equation (4) to get $g_{ij}=(1+2\gamma U)\delta_{ij}$ . Why: the spatial metric carries $\gamma$ and nothing else, so $\gamma$ is fixed purely by how much space curves outside a static mass — the cleanest operational meaning for the parameter.

Step 4. Expand Schwarzschild in isotropic coordinates via $r=\bar r\,(1+GM/2c^2\bar r)^2$ and read off $g_{00}=-1+2U-2U^2+\dots$ , $g_{ij}=(1+2U+\dots)\delta_{ij}$ . Matching to Steps 2 and 3 gives $\beta=\gamma=1$ . Why: GR sits exactly at the special point, and any competitor announces itself — a different $U^2$ coefficient is a different $\beta$ , a different spatial factor a different $\gamma$ .

Go deeper · Derivation B · 1 minWhy light deflection and the Shapiro delay both measure one-half of one plus gamma

Derivation BWhy light deflection and the Shapiro delay both measure one-half of one plus gamma

The aim is to make transparent why two very different-looking observables — the bending of starlight and the extra round-trip time of a radio signal — carry the identical prefactor $\tfrac12(1+\gamma)$ , a fifty-fifty split between time and space curvature. We read an effective refractive index off the static PPN metric, note that both metric potentials contribute equally, and integrate along the nearly straight ray; the same index that bends the ray also slows it.

Step 1. Impose the null condition $g_{00}c^2\mathrm{d}t^2 + g_{ij}\mathrm{d}x^i\mathrm{d}x^j=0$ on the static PPN metric and solve for the coordinate speed of light to first order in $U$ : $v=c\,[1-(1+\gamma)U]$ . Why: the " $1$ " comes from $g_{00}$ and the " $\gamma$ " from $g_{ij}$ ; a photon at $c$ weights the two equally — the physical origin of the sum $1+\gamma$ .

Step 2. Define an effective refractive index $n=c/v=1+(1+\gamma)GM/c^2r$ . Why: gravity now acts like an optical medium with $n>1$ , and every ray-optics tool applies directly, with the whole theory-dependence in the single factor $(1+\gamma)$ .

Step 3. Integrate the transverse gradient of $n$ along the undeflected path, $\delta\theta=\int_{-\infty}^{\infty}\nabla_\perp n\,\mathrm{d}z$ ; the elementary integral $\int b\,\mathrm{d}z/(b^2+z^2)^{3/2}=2/b$ yields $\delta\theta=\tfrac12(1+\gamma)\cdot4GM/c^2b$ . Why: for $\gamma=1$ this is the full GR deflection $1.75''$ at the solar limb, exactly twice the pure-time-curvature ( $\gamma=0$ ) value.

Step 4. For the Shapiro delay, integrate the excess travel time $\mathrm{d}t=(\mathrm{d}\ell/c)[1+(1+\gamma)GM/c^2r]$ along the signal path, giving $\Delta t=\tfrac12(1+\gamma)(4GM_\odot/c^3)\ln(4r_\oplus r_p/b^2)$ . Why: the identical index gives the identical prefactor — so timing a radio echo and imaging a starlight bend measure the same parameter, and the more precise timing (Cassini) sets the tighter bound.

Limiting cases

The value of the parameter vector picks out a theory or a physical regime; the table collects the landmarks.

Case	Result	Meaning
General relativity	$\gamma=\beta=1,\ \xi=\alpha_1=\alpha_2=\alpha_3=\zeta_i=0$	The unique fully conservative, Lorentz-invariant point — no preferred frame or location.
Brans–Dicke (scalar–tensor)	$\gamma=\dfrac{1+\omega}{2+\omega},\ \beta=1$	Coupling $\omega$ ; $\gamma\to1$ as $\omega\to\infty$ . Cassini's $\|\gamma-1\|\lesssim2.3\times10^{-5}$ forces $\omega\gtrsim4\times10^{4}$ .
Fully conservative theories	$\alpha_3=\zeta_1=\zeta_2=\zeta_3=\zeta_4=0$	A global energy-momentum conservation law kills all five conservation-law parameters.
Semiconservative theories	$\alpha_3=\zeta_i=0,\ \alpha_1,\alpha_2\ \text{free}$	Conserve momentum and angular momentum but may still show preferred-frame effects.
Newtonian limit	$U$ only, all post-Newtonian potentials dropped	Recovers $g_{00}=-1+2U$ , $g_{ij}=\delta_{ij}$ — time warp alone, no light-bending surplus

Common trapThe PPN parameters are free knobs you tune to fit the data

They are not. Each metric theory predicts definite values for all ten; the parameters are properties of the theory, read off once and for all, not dials fitted per experiment. Fitting data to the PPN metric measures where nature sits, and then you ask which theories predict that point. Three cautions travel with this. First, $\gamma$ and $\beta$ are strictly 1PN, weak-field, slow-motion quantities — they say nothing about strong fields, where two theories agreeing on all ten parameters can differ drastically and need the parametrized post-Keplerian machinery instead. Second, the formalism is gauge-fixed: parameter values are only comparable across references in the standard PPN gauge. Third, the eight non- $\gamma,\beta$ parameters are not corrections to neglect casually — several are bounded far more tightly than $\gamma$ and $\beta$ themselves ( $|\alpha_3|\lesssim10^{-20}$ ), and those bounds are physics, not rounding.

Worked examples

1 · Mercury's perihelion advance from the PPN factor (fully worked)

Show that the PPN combination $\tfrac13(2+2\gamma-\beta)$ with $\gamma=\beta=1$ reproduces the famous $43''$ per century, using $a=5.79\times10^{10}\,\mathrm{m}$ , $e=0.2056$ , $M_\odot=1.989\times10^{30}\,\mathrm{kg}$ , $G=6.674\times10^{-11}$ , $c=2.998\times10^{8}\,\mathrm{m/s}$ .

Step 1. Package the Sun's mass as a length: $GM_\odot/c^2 = \mathbf{1477\,m}$ (half the solar Schwarzschild radius).

Step 2. Form the semi-latus rectum $a(1-e^2)=5.79\times10^{10}\,(1-0.0423)=\mathbf{5.545\times10^{10}\,m}$ .

Step 3. With the GR factor $\tfrac13(2+2-1)=1$ , the advance is $\Delta\varpi=6\pi\,GM_\odot/[c^2 a(1-e^2)]=6\pi\,(1477)/(5.545\times10^{10})=\mathbf{5.02\times10^{-7}\,rad}$ /orbit; with $1\,\mathrm{rad}=206265''$ that is $0.1036''$ /orbit.

Step 4. Mercury completes $36525/87.969=415.2$ orbits per century, so the total is $0.1036''\times415.2=\mathbf{43.0''/century}$ . The factor equals 1 only because $\gamma=\beta=1$ ; a theory with $\beta=1.001$ would shift the prediction by $\tfrac13\times0.001\times43''\approx0.014''$ /century — within reach of modern ephemerides.

2 · Turning the Cassini bound on gamma into a Brans–Dicke bound on ω (fully worked)

Brans–Dicke theory predicts $\gamma=(1+\omega)/(2+\omega)$ , $\beta=1$ . Given Cassini's $\gamma-1=(2.1\pm2.3)\times10^{-5}$ , what lower bound does it place on the coupling $\omega$ ?

Step 1. Rearrange: $\gamma-1=\dfrac{1+\omega}{2+\omega}-1=\dfrac{-1}{2+\omega}$ , so $|\gamma-1|=\dfrac{1}{2+\omega}$ . The deviation from GR shrinks as $1/\omega$ .

Step 2. The measurement is consistent with zero; take the $1\sigma$ bound $|\gamma-1|\lesssim2.3\times10^{-5}$ . Then $2+\omega\gtrsim1/(2.3\times10^{-5})=4.35\times10^{4}$ , so $\mathbf{\omega\gtrsim4\times10^{4}}$ .

Step 3. Interpret: a single spacecraft-tracking experiment forces any Brans–Dicke scalar to couple more than forty thousand times more weakly than the "natural" order-unity value, squeezing the theory to the brink of indistinguishability from GR — which is why scalar–tensor alternatives now focus on strong-field or cosmological regimes.

3 · From lunar laser ranging to a bound on beta (fully worked)

Lunar laser ranging finds no Nordtvedt polarization of the Moon's orbit, giving $|\eta_N|<4\times10^{-4}$ . Assuming the preferred-frame and conservation-law parameters are negligible and using the Cassini $\gamma$ , what does this imply for $\beta$ ?

Step 1. With $\xi,\alpha_i,\zeta_i\to0$ the combination collapses to $\eta_N=4\beta-\gamma-3=4(\beta-1)-(\gamma-1)$ .

Step 2. Since $\gamma-1\approx2\times10^{-5}$ is a hundred times smaller than the $\eta_N$ bound, it is negligible, leaving $\eta_N\approx4(\beta-1)$ .

Step 3. Therefore $|\beta-1|\approx|\eta_N|/4<10^{-4}$ . This bounds the nonlinearity to $|\beta-1|\lesssim10^{-4}$ — an independent, oblateness-free confirmation that complements Mercury's perihelion.

Physics of the portrait

The PPN formalism is the work of many hands — Eddington and Robertson gave gravity its first parameters, Schiff sharpened the tests — but its modern ten-parameter form and relentless confrontation with experiment belong above all to one physicist.

Origins · 1 minClifford Will and how this was discovered

Clifford Will

b. 1946 · completing and testing the ten-parameter PPN framework

Clifford Will, a Canadian-born physicist trained under Kip Thorne at Caltech, turned the scattered parametrizations of the 1920s through 1960s into the systematic ten-parameter formalism used today. Eddington had noticed in 1923 that one could write the light-bending metric with adjustable coefficients, and Robertson and Schiff refined the idea for Solar-System tests, but it was Will, working closely with Kenneth Nordtvedt around 1971–72, who fixed the standard PPN gauge, catalogued the full set of potentials, and gave each surviving parameter a clean meaning tied to a symmetry principle — Lorentz invariance, conservation laws, the presence or absence of a preferred frame or location. His monograph "Theory and Experiment in Gravitational Physics" became the field's standard reference, and his periodically updated Living Review "The Confrontation between General Relativity and Experiment" is where every new bound on $\gamma$ , $\beta$ , and the rest is recorded. Will's deeper contribution was as much philosophical as technical: he insisted that testing general relativity means measuring where nature sits in a space of alternatives, not merely checking Einstein's predictions — a reframing that turned a handful of classical tests into a quantitative, open-ended experimental program.

History

1923

— Eddington parametrizes the light-bending metric, seeding what become gamma and beta

1962

— Robertson and Schiff refine the parametrized post-Newtonian bookkeeping for Solar-System tests

1968

— Nordtvedt predicts a strong-equivalence-principle violation in massive-body free fall

1972

— Will and Nordtvedt cast the full ten-parameter formalism in its standard gauge

2003

— Cassini fixes gamma minus one to two parts in a hundred thousand

Going further

Texts. Misner, Thorne & Wheeler, Gravitation (1973), Chapter 39 (Box 39.2, Table 39.2) · Will, Theory and Experiment in Gravitational Physics, 2nd ed. (2018), Chapters 4–5 · Will, "The Confrontation between General Relativity and Experiment," Living Reviews in Relativity 17 (2014) 4 · Carroll, Spacetime and Geometry (2004), Chapter 7 · Nordtvedt, Phys. Rev. 169 (1968) 1017.

Provenance. The standard-gauge PPN metric and parameter definitions follow MTW Chapter 39 and Will's monograph Chapter 4; every equation and number here was re-derived or recomputed. The static-mass match ( $\gamma=\beta=1$ ) reproduces by expanding Schwarzschild in isotropic coordinates $r=\bar r(1+GM/2c^2\bar r)^2$ ; the Mercury $43.0''$ /century from $GM_\odot/c^2=1477\,\mathrm{m}$ , $a=5.79\times10^{10}\,\mathrm{m}$ , $e=0.2056$ , $415.2$ orbits/century; the deflection $1.75''$ from $R_s=2954\,\mathrm{m}$ , $R_\odot=6.96\times10^{8}\,\mathrm{m}$ , $1\,\mathrm{rad}=206265''$ . Measured values — Cassini $\gamma-1=(2.1\pm2.3)\times10^{-5}$ (Bertotti, Iess & Tortora, Nature 425, 2003), Brans–Dicke $\omega\gtrsim4\times10^{4}$ , $|\eta_N|<4\times10^{-4}$ and $|\beta-1|\lesssim10^{-4}$ (lunar laser ranging), and Gravity Probe B frame-dragging $37.2\pm7.2$ mas/yr versus GR $39.2$ mas/yr — are from Will's Living Review. A verifier should recompute the isotropic-Schwarzschild match, the Mercury and deflection arithmetic, the inversion $|\gamma-1|=1/(2+\omega)$ , and the collapse $\eta_N\approx4(\beta-1)$ .

Prerequisites. Newtonian Limit of General Relativity · Weak-Field Limit and Linearized Gravity · PPN Parameters Beta and Gamma. Leads to. Deflection of Light by the Sun · Shapiro Time Delay · Perihelion Precession of Mercury · Frame Dragging and the Lense-Thirring Effect.

Ask the tutor: Ask me to derive the $\tfrac12(1+\gamma)$ deflection prefactor step by step, or to work out how a given theory's field equations translate into its ten PPN parameter values.

How is the PPN formalism different from an ordinary post-Newtonian expansion?

A post-Newtonian expansion is a calculational tool for one theory: you take general relativity (or any single theory), expand its field equations in powers of $v/c$ , and compute observables order by order. The PPN formalism is a comparison tool across theories. Instead of committing to one set of field equations, it writes the most general 1PN metric any metric theory could produce, with arbitrary coefficients in front of a standard catalogue of potentials; each theory then corresponds to a particular set of those coefficients. So the post-Newtonian expansion answers "what does this theory predict?" while PPN answers "which theory does the data pick?" You use the former to compute a candidate's PPN values, then PPN to confront them with experiment.

Why are there exactly ten parameters and not more or fewer?

The count is what survives after two prunings. You start by writing every 1PN potential that could appear in the metric and attach an arbitrary coefficient to each. Then you discard the coefficients removable by a coordinate change — fixing the standard PPN gauge eliminates the gauge artefacts. What remains, restricting to theories that are at least approximately Lorentz-invariant, is exactly ten physically meaningful numbers: $\gamma$ and $\beta$ , the preferred-location $\xi$ , three preferred-frame parameters $\alpha_1,\alpha_2,\alpha_3$ , and four conservation-law parameters $\zeta_1,\dots,\zeta_4$ . The number reflects how many independent symmetry properties a metric theory can have at this order — it is a classification result, not an arbitrary choice.

What is the physical difference between gamma and beta?

They answer two independent questions. $\gamma$ is about geometry: how much space curvature does a mass produce relative to how much it warps time? Newtonian gravity curves only time ( $\gamma=0$ ); GR curves space and time equally ( $\gamma=1$ ). Because light at $c$ feels spatial curvature as strongly as temporal, $\gamma$ is measured cleanly by light bending and the Shapiro delay. $\beta$ is about nonlinearity: when fields superpose, is the result the simple sum, or does the field's own energy gravitate and add an extra piece? GR is mildly nonlinear ( $\beta=1$ ). Since the nonlinear term $-2\beta U^2$ is second order in the potential, it shows up in the slow, cumulative dynamics of bound orbits — Mercury's perihelion — and in the Nordtvedt effect, not in a single ray's deflection.

Why can't PPN test strong-field gravity like neutron stars or black holes?

Because it is built on an expansion that assumes weak fields and slow motion, $U\sim v^2\ll1$ (with $c=1$ ), and keeps only first order in that small quantity. Near a neutron-star surface or a black-hole horizon, $U$ approaches order unity and the expansion fails — the discarded higher-order terms become as large as the ones kept. Two theories can share all ten PPN parameters, agreeing on every Solar-System test, yet make wildly different predictions for a neutron star's maximum mass or a binary's gravitational-wave signal. Those regimes need other tools: the parametrized post-Keplerian formalism for binary pulsars, and numerical relativity or higher-order PN expansions for mergers.

What is the Nordtvedt effect, and why is it such a good probe of beta?

It is a predicted violation of the strong equivalence principle: if a body's gravitational binding energy contributes differently to its gravitational and inertial mass, massive bodies fall at slightly different rates. The size is set by $\eta_N=4\beta-\gamma-3-\dots$ , which vanishes in GR. It is an excellent handle on $\beta$ for two reasons. First, using the Earth and Moon amplifies it — their self-energies are large enough that a tiny $\eta_N$ produces a coherent, phase-locked oscillation of the lunar orbit toward and away from the Sun. Second, once $\gamma$ is fixed by light bending, $\eta_N$ depends essentially only on $\beta$ , and unlike Mercury's perihelion it needs no uncertain solar-oblateness subtraction — so lunar laser ranging delivers $|\beta-1|\lesssim10^{-4}$ cleanly.

What does the preferred-frame velocity wⁱ represent?

$w^i$ is the velocity of the local PPN frame — say, the Solar System's rest frame — relative to the "universe rest frame" that the preferred-frame parameters would single out, usually identified with the frame in which the cosmic microwave background is isotropic (the Solar System moves through it at about $370$ km/s). If any of $\alpha_1,\alpha_2,\alpha_3$ were nonzero, gravitational dynamics would depend on $w^i$ , and orbits and spins would feel a preferred direction set by our motion through the cosmos. General relativity has no such frame, so all three $\alpha$ 's vanish and $w^i$ drops out entirely; the tight bounds on the $\alpha$ 's are, in effect, statements that no such cosmic-frame dependence has ever been seen.