Newtonian Limit of General Relativity

General relativity would be worthless if it did not, in the ordinary world, become Newton. Apples fall and planets orbit exactly as the inverse-square law says they should, to a precision no relativistic correction disturbs at the level of a dropped stone. So any theory that replaces gravity with curved spacetime carries a debt: it must reproduce $\ddot{\mathbf x}=-\nabla\Phi$ and $\nabla^2\Phi=4\pi G\rho$ in the corner of parameter space where Newton reigns. The Newtonian limit is the accounting that pays that debt — and in paying it, extracts two things the field equations do not wear on their sleeve. First, it tells us what the metric means: the humble potential $\Phi$ is not a separate field bolted onto geometry but lives inside a single component, $g_{00}=-(1+2\Phi/c^2)$ , so that what we feel as weight is the curvature of time. Second, it fixes the one dimensionless number dimensional analysis cannot supply — the $8\pi$ in Einstein's constant $\kappa=8\pi G/c^4$ . This is why every course, having written $G_{\mu\nu}=\kappa T_{\mu\nu}$ , immediately takes the slow-motion weak-field limit: it is the only calibration available, the bridge tying abstract geometry to a laboratory value of $G$ . The limit rests on three logically independent smallness conditions — weak field, slow motion, and stationarity — and the whole richness of solar-system gravitation follows from relaxing them one at a time: drop stationarity and gravitomagnetism appears; keep the next order in $v/c$ and Mercury's perihelion creeps, starlight bends, radar echoes delay. Newtonian gravity is the ground floor of that tower, and this article builds it.

Key takeaways

The Newtonian limit is the regime of slow motion ( $v\ll c$ ), weak gravity ( $|\Phi|/c^2\ll1$ ), and quasi-static fields in which Einstein's equations and the geodesic equation collapse to Newton's law.
One identification does the decisive work: $g_{00}=-(1+2\Phi/c^2)$ — the time-time metric is the Newtonian potential, so gravity at leading order is the curvature of time, not space.
The geodesic equation reduces to $\ddot{\mathbf x}=-\nabla\Phi$ because the dominant Christoffel symbol $\Gamma^i_{00}=\partial_i\Phi/c^2$ acts as the gravitational acceleration.
The $00$ Einstein equation becomes Poisson's $\nabla^2\Phi=4\pi G\rho$ , which fixes the coupling $\kappa=8\pi G/c^4$ — a number no dimensional argument can give.
The three smallness conditions are distinct: dropping stationarity yields gravitomagnetism and frame dragging; dropping weak-field yields the full nonlinear theory of black holes and neutron stars.
The same double expansion in $\Phi/c^2$ and $(v/c)^2$ generates every classic test — redshift, light bending, perihelion precession, Shapiro delay — and the PPN parameters $\beta,\gamma$ .
Key idea: In the slow-motion, weak-static-field regime, $g_{00}=-(1+2\Phi/c^2)$ turns the geodesic equation into $\ddot{\mathbf x}=-\nabla\Phi$ and the $00$ Einstein equation into $\nabla^2\Phi=4\pi G\rho$ — fixing $\kappa=8\pi G/c^4$ and showing gravity at leading order is the curvature of time.

Where this lives today

The Newtonian limit is not a historical curiosity but a living constraint and a working part of technology. GPS and its cousins Galileo and BeiDou continuously correct the $+38.5\,\mu\mathrm{s}/\mathrm{day}$ clock offset computed above; without the $g_{00}=-(1+2\Phi/c^2)$ term, navigation would drift kilometres per day. At the frontier, optical-lattice atomic clocks now resolve the redshift over a single centimetre of height — a fractional shift near $10^{-18}$ — turning equation (9) into a tool for relativistic geodesy that maps the geoid through clock comparisons. The limit also anchors the parametrized post-Newtonian (PPN) framework for solar-system tests: general relativity predicts $\beta=\gamma=1$ , and data pin them there — Cassini gives $|\gamma-1|\lesssim2\times10^{-5}$ , lunar laser ranging bounds $|\beta-1|$ at $10^{-4}$ . Meanwhile the strong-field corner of the second figure is now directly observed: LIGO and Virgo record mergers at $|\Phi|/c^2\sim0.5$ , and the Event Horizon Telescope images the shadow at the horizon itself — precisely where the linearization that gives us Newton has nothing left to say.

The physical picture

Picture a test particle drifting freely through a nearly flat spacetime dented, very slightly, by a mass. In Einstein's theory it feels no force — it simply follows the straightest worldline, the geodesic. But a geodesic extremizes proper time, and here is the twist that makes gravity fall out: because the mass makes clocks tick slower where the potential is deeper, the elapsed proper time along a worldline is $\int(1+\Phi/c^2-v^2/2c^2)\,\mathrm{d}t$ — a competition between the faster ticking higher up and the speed needed to get there and back. Extremizing that integral is exactly Hamilton's principle for $L=v^2/2-\Phi$ , and it returns $\ddot{\mathbf x}=-\nabla\Phi$ : the worldline bends toward the mass. What Newton called a force is the tendency of free worldlines to bend toward regions of slow time. The spatial curvature is there too, but for a slow particle it barely matters — such a particle moves almost entirely in the time direction, so it is almost exclusively the time curvature, the variation of $g_{00}$ , that steers it. The first figure makes this concrete: a potential well, a clock that slows as it descends, and a test mass accelerating down the slope at exactly $-\nabla\Phi$ .

Figure

Gravity as the curvature of time. The blue well is the Newtonian potential Φ = −GM/r, which sits inside the metric as g₀₀ = −(1+2Φ/c²). A free test mass accelerates down-slope at a = −∇Φ, giving g = GM/R² = 9.81 m/s² at Earth's surface, while a clock deeper in the well ticks slower by dτ/dt = 1+Φ/c², a fractional −6.9×10⁻¹⁰ at the ground

Why is Newton so good, and where does he fail? The controlling number is the dimensionless field strength $|\Phi|/c^2=GM/(Rc^2)$ , the depth of the well in units of $c^2$ . Where it is tiny, spacetime is barely dented, $h_{\mu\nu}$ is small, and the linearized story of the first figure is essentially exact; where it climbs toward unity, linearization disintegrates and the full nonlinear equations take over. The second figure lays this number out on a log scale across the gravitating universe, explaining at a glance both the success of Newtonian celestial mechanics and the regimes — white dwarfs, neutron stars, black-hole horizons — where relativity becomes unavoidable.

Figure

The domain of the Newtonian limit. The field-strength parameter GM/Rc², the depth of the potential well in units of c², plotted on a log scale for real objects. Newton is superb where it is minuscule — Earth's surface 7×10⁻¹⁰, the Sun's surface 2×10⁻⁶ — post-Newtonian corrections become measurable for a white dwarf near 10⁻⁴, and linearization collapses at a neutron star near 0.2 or a black-hole horizon at 0.5, where the full nonlinear Einstein equations are mandatory

The core equations

The construction begins with one choice: the metric is Minkowski plus a small correction. Everything else follows from keeping only the leading term in that correction, in the particle's speed, and in time derivatives of the field.

g_{\mu\nu}=\eta_{\mu\nu}+h_{\mu\nu},\qquad |h_{\mu\nu}|\ll1 \qquad (1)

where: $\eta_{\mu\nu}$ — Minkowski metric, signature $(-,+,+,+)$ ; $h_{\mu\nu}$ — small metric perturbation, first order only. Reading: gravity is a faint ripple on flat spacetime, so every equation below is linearized in $h$ and $O(h^2)$ terms are discarded. The pivotal step identifies the one component that carries Newtonian physics. Demanding that free-fall reproduce Newton's second law forces the time-time component to encode the potential with a specific numerical factor.

g_{00}=-\left(1+\frac{2\Phi}{c^2}\right) \qquad (2)

where: $\Phi$ — Newtonian gravitational potential (for a point mass, $\Phi=-GM/r$ ); $c$ — speed of light. Reading: the factor of $2$ is not decorative — it is fixed downstream by requiring the $00$ Einstein equation to reproduce Poisson's law with the correct $4\pi G$ ; get it wrong and the surface gravity comes out wrong. Motion in curved spacetime obeys the geodesic equation — a freely falling particle parallel-transports its own velocity, the relativistic form of "no force means no acceleration".

\frac{\mathrm{d}^2 x^\mu}{\mathrm{d}\tau^2}+\Gamma^\mu_{\alpha\beta}\frac{\mathrm{d}x^\alpha}{\mathrm{d}\tau}\frac{\mathrm{d}x^\beta}{\mathrm{d}\tau}=0 \qquad (3)

where: $\tau$ — proper time along the worldline; $\Gamma^\mu_{\alpha\beta}$ — Christoffel symbols of the Levi-Civita connection. Reading: the Christoffel term is the gravitational force per unit mass, and in the Newtonian limit only one component survives. For a slow particle in a static weak field, the Christoffel sum is dominated by a single term — nothing but the gradient of the potential.

\Gamma^i_{00}=\frac{1}{c^2}\,\partial_i\Phi \qquad (4)

where: $\partial_i\Phi$ — spatial gradient of the potential, i.e. minus the gravitational field. Reading: for a slow particle only $\Gamma^i_{00}(u^0)^2$ contributes, and this depends solely on $\partial_i g_{00}$ — the spatial metric never enters. Assembling the surviving piece and cancelling the factors of $c$ delivers exactly the law schoolchildren learn, now derived not postulated.

\ddot{\mathbf x}=-\nabla\Phi \qquad (5)

where: $\ddot{\mathbf x}$ — coordinate acceleration of the test particle. Reading: Newton's equation of motion, recovered with no free parameters; the mass drops out on both sides — the equivalence principle. On the field side, the geometry that sources gravity is the Ricci tensor, whose time-time component in a static weak field collapses to the Laplacian of the potential — the geometric half of Poisson's equation.

R_{00}=\frac{1}{c^2}\nabla^2\Phi \qquad (6)

where: $R_{00}$ — time-time Ricci component; $\nabla^2=\partial_i\partial_i$ — flat-space Laplacian. Reading: the curvature of time is set by how the potential's gradient spreads, precisely $\nabla^2\Phi$ . Feeding the rest-mass density of slow dust through the trace-reversed field equations reproduces Poisson's equation, the cornerstone of Newtonian gravitation.

\nabla^2\Phi=4\pi G\rho \qquad (7)

where: $G$ — Newton's gravitational constant; $\rho$ — rest-mass density, the dominant part of $T_{00}=\rho c^2$ for slow matter. Reading: in this limit the source of curvature reduces to ordinary mass density, the enormous $c^2$ in $T_{00}$ compensated by the small couplings elsewhere. Requiring (6) and (7) to agree fixes the one undetermined constant, giving Einstein's equation its normalized form.

G_{\mu\nu}=\frac{8\pi G}{c^4}\,T_{\mu\nu} \qquad (8)

where: $G_{\mu\nu}=R_{\mu\nu}-\tfrac12 g_{\mu\nu}R$ — Einstein tensor; $T_{\mu\nu}$ — stress-energy tensor, with $T_{00}=\rho c^2$ for dust at rest. Reading: the coefficient $8\pi G/c^4$ cannot be guessed from dimensions (the $8\pi$ is pure number); only the Newtonian match supplies it. Finally, the same $g_{00}$ that produced the force also governs how clocks run, so the limit already contains gravitational time dilation as a bonus.

\mathrm{d}\tau=\sqrt{1+\frac{2\Phi}{c^2}}\;\mathrm{d}t\approx\left(1+\frac{\Phi}{c^2}\right)\mathrm{d}t \qquad (9)

where: $\mathrm{d}\tau$ — proper time on a static clock; $\mathrm{d}t$ — coordinate time. Reading: a clock deeper in the well (more negative $\Phi$ ) records less proper time — the effect Pound and Rebka measured and GPS corrects for every second.

Derivation AFrom the geodesic equation to Newton's second law

The goal is to watch the full four-dimensional geodesic equation, with its many-term Christoffel sum, collapse to the vector statement $\ddot{\mathbf x}=-\nabla\Phi$ . We impose the three smallness conditions in a deliberate order — slow motion to prune the connection sum, the static weak field to evaluate the one surviving Christoffel symbol, and the identification of proper with coordinate time to reduce the derivatives. The quiet payoff is watching the factors of $c$ cancel at the end, turning a purely geometric identity into a familiar law of physics with no adjustable constants left over.

Step 1. Impose slow motion, $\mathrm{d}x^i/\mathrm{d}\tau\ll c\,\mathrm{d}t/\mathrm{d}\tau$ , so that in $\Gamma^\mu_{\alpha\beta}u^\alpha u^\beta$ every term with a spatial velocity is negligible against the one built from $u^0=c\,\mathrm{d}t/\mathrm{d}\tau$ . Only $\alpha=\beta=0$ survives, and the spatial geodesic equation becomes $\mathrm{d}^2x^i/\mathrm{d}\tau^2 + \Gamma^i_{00}(c\,\mathrm{d}t/\mathrm{d}\tau)^2=0$ . Why: a non-relativistic particle moves almost entirely through time, so its four-velocity is overwhelmingly timelike and the time-time connection dominates.

Step 2. Evaluate $\Gamma^i_{00}$ . From $\Gamma^i_{00}=-\tfrac12 g^{ij}(2\partial_0 g_{j0}-\partial_j g_{00})$ , drop the time derivative (stationarity) and set $g^{ij}\to\delta^{ij}$ , leaving $\Gamma^i_{00}=-\tfrac12\,\partial_i g_{00}=\tfrac12\,\partial_i h_{00}$ . Inserting $h_{00}=-2\Phi/c^2$ from (2) gives $\Gamma^i_{00}=\partial_i\Phi/c^2$ . Why: in a static field the connection reduces to a pure spatial gradient of $g_{00}$ , exactly where the potential was hidden.

Step 3. Identify proper and coordinate time. At slow speed and weak field $\mathrm{d}\tau\approx\mathrm{d}t$ , so $\mathrm{d}t/\mathrm{d}\tau\approx1$ and $\tau$ -derivatives become $t$ -derivatives. The equation becomes $\mathrm{d}^2x^i/\mathrm{d}t^2 + c^2\cdot(\partial_i\Phi/c^2)=0$ . Why: corrections to $\mathrm{d}t/\mathrm{d}\tau$ are of order $\Phi/c^2$ and $(v/c)^2$ , exactly what we drop at this order.

Step 4. Cancel the factors of $c$ and read off the physics. The $c^2$ from $(c\,\mathrm{d}t/\mathrm{d}\tau)^2$ annihilates the $1/c^2$ inside $\Gamma^i_{00}$ , leaving $\ddot{x}^i=-\partial_i\Phi$ , that is $\ddot{\mathbf x}=-\nabla\Phi$ . Why: this cancellation is the quiet triumph of the limit — $c$ disappears entirely and a geometric equation becomes Newton's law with the right sign and magnitude.

Go deeper · Derivation B · 1 minThe Ricci tensor and the geometric side of Poisson's equation

Derivation BThe Ricci tensor and the geometric side of Poisson's equation

Having reduced the equation of motion, we now reduce the field equation's left-hand side. The goal is to show that the time-time Ricci component, in general a fearsome nonlinear expression in the metric and its derivatives, becomes in the static weak field the almost embarrassingly simple $R_{00}=\nabla^2\Phi/c^2$ — the geometric half of Poisson's equation, saying that the curvature of time is literally the Laplacian of the potential. It needs only the linearized Ricci formula, stationarity, and the Christoffel symbol already computed, and it sets up the calibration of Einstein's constant.

Step 1. Start from the linearized Ricci tensor, $R_{\mu\nu}=\partial_\alpha\Gamma^\alpha_{\mu\nu}-\partial_\nu\Gamma^\alpha_{\mu\alpha}$ , with the quadratic-in- $\Gamma$ terms discarded as second order in $h$ . Why: each $\Gamma$ is already first order in $h$ , so a product of two is $O(h^2)$ and beyond our accuracy.

Step 2. Specialize to the $00$ component in a static field. With $\mu=\nu=0$ and $\partial_0=0$ , the second term $\partial_0\Gamma^\alpha_{0\alpha}$ vanishes and the first keeps only its spatial derivative, so $R_{00}\approx\partial_i\Gamma^i_{00}$ . Why: stationarity kills every time derivative, collapsing the contraction to a single spatial divergence of the one Christoffel symbol that matters.

Step 3. Insert $\Gamma^i_{00}=\partial_i\Phi/c^2$ from the previous derivation to obtain $R_{00}=\partial_i(\partial_i\Phi/c^2)=\nabla^2\Phi/c^2$ . Why: the divergence of a gradient is the Laplacian, so the geometry delivers exactly the operator on the left of Poisson's equation — the two theories now speak the same language.

Go deeper · Derivation C · 1 minFixing the Einstein constant by trace reversal

Derivation CFixing the Einstein constant by trace reversal

The final derivation earns the number $8\pi$ . Einstein's equation as first written, $G_{\mu\nu}=\kappa T_{\mu\nu}$ , contains an unknown coupling $\kappa$ that dimensional analysis fixes only up to a pure number. The goal is to pin that number down by demanding the $00$ -component reduce to Poisson's $\nabla^2\Phi=4\pi G\rho$ . The cleanest route runs not through the Einstein tensor directly but through its trace-reversed cousin, whose right-hand side for slow dust simplifies dramatically; the algebra then shows how the factor of $2$ in $g_{00}$ and the trace term conspire to give $\kappa=8\pi G/c^4$ .

Step 1. Trace-reverse the field equations. The trace of $G_{\mu\nu}=\kappa T_{\mu\nu}$ gives $-R=\kappa T$ , and substituting back yields $R_{\mu\nu}=\kappa(T_{\mu\nu}-\tfrac12 g_{\mu\nu}T)$ . For pressureless dust the only nonzero component is $T_{00}=\rho c^2$ , and $T=g^{00}T_{00}\approx-\rho c^2$ . Why: the trace-reversed form isolates the Ricci tensor, exactly the object equation (6) already expressed through $\Phi$ .

Step 2. Evaluate the effective source for the $00$ component. Using $g_{00}\approx-1$ at leading order, $T_{00}-\tfrac12 g_{00}T\approx\rho c^2-\tfrac12\rho c^2=\tfrac12\rho c^2$ . Why: the trace subtraction halves the naive source — the one-half that combines with the factor of $2$ in $g_{00}$ to leave a clean $4\pi$ .

Step 3. Set left equal to right. With $R_{00}=\nabla^2\Phi/c^2$ from Derivation 2, the $00$ equation reads $\nabla^2\Phi/c^2=\kappa\cdot\tfrac12\rho c^2$ , i.e. $\nabla^2\Phi=\tfrac12\kappa\rho c^4$ . Why: both sides are now in the same physical quantities, matching geometry to matter at leading order.

Step 4. Compare with Poisson. Demanding $\tfrac12\kappa\rho c^4=4\pi G\rho$ forces $\kappa=8\pi G/c^4$ , giving $G_{\mu\nu}=(8\pi G/c^4)\,T_{\mu\nu}$ . Why: this is the only place in the theory where the dimensionless coupling is nailed down — the limit is not a consistency check but the source of the number.

Limiting cases

The Newtonian limit is the simultaneous intersection of three smallness conditions. Relaxing each one is not a failure but a doorway into a distinct regime — so the table below is really a map of gravitation with Newton at the origin.

Case	Result	Meaning
Strict Newtonian limit	$\vert\Phi\vert/c^2\ll1,\ v\ll c,\ \partial_0\ll\partial_i$	All three hold; recover $\ddot{\mathbf x}=-\nabla\Phi$ and $\nabla^2\Phi=4\pi G\rho$ — solar-system mechanics to leading order.
Drop stationarity	$\partial_0\sim\partial_i$ , moving sources	Mass currents $T^{0i}$ source off-diagonal $g_{0i}$ : a gravitomagnetic field giving frame dragging and Lense-Thirring. No Newtonian analog.
Keep next order in $v/c$	$\mathcal{O}(\Phi/c^2)^2,\ \mathcal{O}(v/c)^2$	Post-Newtonian regime: perihelion precession, light deflection, Shapiro delay; now $g_{ij}=(1-2\Phi/c^2)\delta_{ij}$ contributes.
Massless probe	$v=c$ exactly	Slow motion fails; spatial curvature $g_{ij}$ contributes equally, doubling the light-bending angle to $1.75''$ at the solar limb.
Strong field	$\vert\Phi\vert/c^2\sim1$	Linearization breaks down; full nonlinear Einstein equations required — black holes, neutron stars, mergers.

Common trapThe weak-field and slow-motion conditions are the same thing

They are not. Weak field ( $|h|\ll1$ ) and slow motion ( $v\ll c$ ) are logically independent, and conflating them is the most common error in setting up the limit. A field can be weak while its sources move fast (a gravitational wave sweeping past a relativistic particle), and a particle can crawl slowly through a field that is not weak at all (a slow probe skimming a black-hole horizon). For bound orbits the virial theorem happens to tie them together, $v^2\sim|\Phi|$ , which lulls people into treating them as one — but that is a coincidence of bound systems, not a theorem. The trap it breeds is keeping terms in $g_{ij}$ or $\Gamma^i_{jk}$ that are formally the same order as terms already thrown away. At strict Newtonian order only $\Gamma^i_{00}$ survives and only $g_{00}$ matters. Setting $g_{ij}=\delta_{ij}$ is fine for planetary orbits but wrong for light, which moves at $v=c$ and feels the spatial curvature every bit as much as the temporal — which is exactly why the correct deflection of starlight is twice the naive "Newtonian photon" value.

Worked examples

1 · Gravitational redshift over a tower — the Pound–Rebka number (confidence builder)

A gamma-ray photon is emitted at the base of a tower and absorbed at the top, height $h=22.5\,\mathrm{m}$ above. Using the Newtonian-limit clock relation, predict the fractional frequency shift near Earth's surface.

Step 1. Equation (9) gives a static clock at potential $\Phi$ the rate $\mathrm{d}\tau/\mathrm{d}t=1+\Phi/c^2$ , so two clocks differ in rate by $\Delta\Phi/c^2$ . A photon's frequency, set by the emitter and read by the absorber, shifts by the same fraction, $\Delta\nu/\nu=-\Delta\Phi/c^2$ . The photon climbing out of the well loses energy — redshift.

Step 2. Near the surface $\Phi=gz$ with $g=9.81\,\mathrm{m/s^2}$ , so $\Delta\Phi=gh$ and $|\Delta\nu/\nu|=gh/c^2$ .

Step 3. Insert the numbers: $|\Delta\nu/\nu|=(9.81)(22.5)/(2.998\times10^8)^2=\mathbf{2.46\times10^{-15}}$ .

Step 4. This is the shift Pound and Rebka measured in 1959 using the Mössbauer effect, confirming the prediction to about 10% (and to 1% with Pound–Snider in 1964). The Newtonian-limit metric $g_{00}=-(1+2\Phi/c^2)$ predicts a redshift of $2.5\times10^{-15}$ over 22.5 m, pulled directly from $g_{00}$ and confirmed in the laboratory.

2 · Why GPS would fail without relativity (fully worked)

A GPS satellite orbits at $r=2.656\times10^7\,\mathrm{m}$ with $v\approx3874\,\mathrm{m/s}$ ; receivers sit at $R=6.371\times10^6\,\mathrm{m}$ . Using $GM_\oplus=3.986\times10^{14}\,\mathrm{m^3/s^2}$ , find the net rate of the orbiting clock relative to the ground, and the daily positioning error if it were ignored.

Step 1. Gravitational rate (general relativity). Higher in the well, the satellite clock runs fast. From equation (9) the fractional rate is $\dfrac{GM_\oplus}{c^2}\!\left(\dfrac{1}{R}-\dfrac{1}{r}\right)$ ; with $GM_\oplus/c^2=4.435\times10^{-3}\,\mathrm{m}$ and $1/R-1/r=1.193\times10^{-7}\,\mathrm{m^{-1}}$ , this is $+5.29\times10^{-10}$ , i.e. $\mathbf{+45.7\ \mu s/day}$ .

Step 2. Velocity rate (special relativity). Its motion makes the clock run slow by $-\tfrac12 v^2/c^2=-\tfrac12(3874)^2/(2.998\times10^8)^2=-8.35\times10^{-11}$ , i.e. $\mathbf{-7.2\ \mu s/day}$ . Opposite sign: altitude speeds a clock up, motion slows it down.

Step 3. Net offset. The orbiting clock gains $+45.7-7.2=\mathbf{+38.5\ \mu s/day}$ — the dominant systematic in the system, cancelled by offsetting the satellite oscillators before launch.

Step 4. Positioning consequence. Ranging times signals at $c$ , so a clock error $\delta t$ maps to $c\,\delta t$ of range: uncorrected, $38.5\,\mu\mathrm{s/day}$ gives $\approx\mathbf{11.5\ km/day}$ of drift. Ignoring the Newtonian-limit clock shift would render GPS useless within minutes — general relativity here is an operational necessity, not an academic refinement.

Physics of the portrait

The Newtonian limit is above all Einstein's own creation. While building general relativity between 1907 and 1915 he did not yet know the field equations — only that whatever they turned out to be, they had to give back Newton. That constraint was his compass, and the limit was the instrument that read it.

Origins · 1 minAlbert Einstein and how this was discovered

Albert Einstein

1879–1955 · general relativity and the Newtonian correspondence

Albert Einstein did not reach general relativity by admiring a beautiful equation; he arrived under the discipline of the correspondence principle, the demand that any new theory reproduce the old one in the appropriate limit. As early as 1907, in what he called "the happiest thought of my life", the equivalence of gravitation and acceleration led him to predict gravitational time dilation and redshift — physics that lives entirely in the $g_{00}$ identified here. For the next eight years the Newtonian limit was his lodestar: he insisted the field equations reduce, for slow matter and weak static fields, to Poisson's $\nabla^2\Phi=4\pi G\rho$ , and it was this requirement, not aesthetics, that fixed the coupling to $8\pi G/c^4$ when the final equations appeared in November 1915. The $43''$ -per-century advance of Mercury's perihelion, computed that same month, was the first payoff of pushing the expansion one order beyond Newton; within months Schwarzschild's exact solution matched the limit term by term, and in 1919 Eddington measured the predicted light deflection. Einstein's genius here was conservative as much as revolutionary: he trusted Newton so completely that he made Newtonian correspondence the non-negotiable boundary condition on a new theory of space and time — turning a limit into a law.

History

1813

— Poisson casts Newtonian gravity in field form as ∇²Φ = 4πGρ

1907

— Einstein's equivalence principle predicts gravitational time dilation, the seed of the g₀₀ identification

1915

— Einstein fixes the coupling κ = 8πG/c⁴ by demanding the Newtonian limit, and computes Mercury's 43-arcsecond precession

1916

— Schwarzschild's exact solution reproduces the weak-field metric term by term

1959

— Pound and Rebka measure the gravitational redshift, confirming g₀₀ = −(1+2Φ/c²) in the laboratory

Going further

Texts. Misner, Thorne & Wheeler, Gravitation, §17.4 (Newtonian correspondence) and Chapter 18 · Wald, General Relativity, §4.4a (the Newtonian limit) · Carroll, Spacetime and Geometry, §4.1–4.2 · Hartle, Gravity, §6.6 · Schutz, A First Course in General Relativity, Chapter 8. Key numerical values used throughout this article: Earth-surface $|\Phi|/c^2=6.96\times10^{-10}$ , solar-surface $2.12\times10^{-6}$ , white dwarf $\sim1.3\times10^{-4}$ , neutron star $\sim0.17$ , horizon $0.5$ ; Pound–Rebka $gh/c^2=2.46\times10^{-15}$ for $h=22.5$ m; GPS gravitational $+45.7$ , velocity $-7.2$ , net $+38.5\,\mu\mathrm{s/day}$ , drift $11.5$ km/day. The PPN bounds $|\gamma-1|\lesssim2\times10^{-5}$ (Cassini) and $|\beta-1|\lesssim10^{-4}$ are standard solar-system test values.

Prerequisites. Weak-Field Limit and Linearized Gravity · Geodesic Equation · Einstein Field Equations. Leads to. Gravitational Redshift · Schwarzschild Geodesics and the Effective Potential · Parametrized Post-Newtonian (PPN) Formalism · Gravitomagnetism.

Ask the tutor: Ask me to push the expansion one rung higher and show you exactly where Mercury's $43''$ perihelion advance and the $1.75''$ bending of starlight come from.

Why does the potential live in g₀₀ rather than in the spatial metric?

Because the four-velocity that steers a slow particle points almost entirely in the time direction — its spatial parts are smaller by $v/c$ . The "force" term $\Gamma^\mu_{\alpha\beta}u^\alpha u^\beta$ is therefore dominated by $\Gamma^i_{00}(u^0)^2$ , and in a static field $\Gamma^i_{00}$ depends only on the spatial gradient of $g_{00}$ , so the potential must sit in $g_{00}$ . The spatial metric $g_{ij}$ deviates from flat too — carrying $-2\Phi/c^2$ — but that only bites near $v=c$ , which is why it matters for light and not planets.

Where exactly does the mysterious 8π come from?

From the factor of $2$ in $g_{00}=-(1+2\Phi/c^2)$ , forced by matching the geodesic equation to Newton, combined with the trace term: for slow dust the effective source of $R_{00}$ is not $T_{00}=\rho c^2$ but $T_{00}-\tfrac12 g_{00}T=\tfrac12\rho c^2$ . Threading both through $R_{00}=\nabla^2\Phi/c^2=4\pi G\rho/c^2$ leaves $\kappa=8\pi G/c^4$ . The $8\pi$ is a genuine prediction of the matching, not an input — which is why dimensional analysis, fixing only $\kappa\sim G/c^4$ , is not enough on its own.

Are the weak-field and slow-motion assumptions really independent?

Yes, entirely. Weak field constrains how far the metric departs from Minkowski; slow motion constrains how the four-velocity splits between time and space. Neither implies the other: a weak field can be probed by a fast particle (a passing gravitational wave seen by a relativistic electron), or a strong field by a slow one (a barely-moving object near a horizon). For bound orbits the virial theorem gives $v^2\sim|\Phi|$ , linking them numerically, but that is a property of bound systems, not a general truth. The strict limit demands both separately, plus stationarity — three independent knobs, all turned down.

If Newton already worked, what did the Newtonian limit actually buy Einstein?

Two things he could not get any other way. First, a physical reading of the metric: without the limit $g_{\mu\nu}$ is just ten functions, but the limit reveals $g_{00}$ is the Newtonian potential, so redshift and time dilation follow at once. Second, the coupling: the field equations are consistent for any $\kappa$ , and only the demand that they reduce to Poisson selects $8\pi G/c^4$ , tying the geometry to a laboratory value of $G$ . So the limit is not a sanity check that the theory "contains Newton" — it is what makes the theory quantitative and gives its symbols meaning.

Why is the deflection of light exactly twice the naive Newtonian value?

Because light is fast, it feels the spatial curvature slow matter ignores. A "Newtonian photon" calculation uses only the time curvature in $g_{00}$ and gives half the answer, about $0.87''$ at the solar limb. But at $v=c$ the spatial metric $g_{ij}=(1-2\Phi/c^2)\delta_{ij}$ contributes equally, doubling the bending to the observed $1.75''$ . This is the massless-probe row of the table, and the sharpest illustration that slow motion is a genuine, separate restriction: for light it fails, and the spatial curvature you dropped for planets comes roaring back.

Does the Newtonian limit include gravitational time dilation, or is that extra?

It is already included, for free. The same $g_{00}=-(1+2\Phi/c^2)$ that produces the force $-\nabla\Phi$ also sets a static clock's rate, $\mathrm{d}\tau=\sqrt{-g_{00}}\,\mathrm{d}t\approx(1+\Phi/c^2)\,\mathrm{d}t$ . So accepting the metric identification already commits you to clocks running slower deeper in the well, with fractional rate $\Phi/c^2$ . This is why Pound–Rebka and the GPS offset are both, at heart, tests of the very same $g_{00}$ that gives Newtonian dynamics — dilation is not a separate add-on but the temporal face of the geometry whose spatial gradient we call gravity.