Experimental Status

General relativity is among the most stringently tested theories in physics, confirmed across an extraordinary range of scales — from millimetre laboratory experiments to the merger of black holes a billion light-years away, from the $4 3^{''}$ -per-century wobble of Mercury to the expansion of the whole universe. This page collects the evidence, organized by regime, and notes where the theory is pushed hardest. Every confirmation is a constraint any quantum completion must reproduce.

We use $c = 1$ and keep $G$ explicit.

The classical tests

The original tests, all computed as geodesics in the Schwarzschild metric, remain benchmark confirmations:

Test	Effect	Current precision
Perihelion precession	Mercury's orbit advances $4 3^{''}$ /century beyond Newton	matches to $\sim 1 0^{- 4}$
Light deflection	starlight bends $1.7 5^{''}$ at the solar limb ( $2 \times$ Newtonian)	VLBI radio to $\sim 1 0^{- 4}$ (parameter $γ$ )
Gravitational redshift	clocks run slow in a potential well, $Δ ν / ν = ΔΦ$	Pound–Rebka (1959); optical clocks over cm heights
Shapiro time delay	signals passing the Sun are delayed	Cassini (2003) to $\sim 1 0^{- 5}$

The light-deflection and Shapiro results are usually quoted as bounds on the PPN parameter $γ$ (space curvature per unit mass), with Cassini giving $γ - 1 = (2.1 \pm 2.3) \times 1 0^{- 5}$ — general relativity predicts $γ = 1$ exactly.

The parametrized post-Newtonian framework

To test GR against alternatives systematically, the weak-field, slow-motion metric of any metric theory is expanded in a set of ten PPN parameters (the best-known being $γ$ above and $β$ , measuring nonlinearity in superposition). General relativity fixes $γ = β = 1$ and the remaining eight to zero (no preferred frames, no violations of conservation laws). Solar-system data pin every PPN parameter to its GR value, typically at the $1 0^{- 4}$ – $1 0^{- 5}$ level — tightly constraining scalar–tensor and other competitors.

Equivalence-principle tests

The foundational premise — the universality of free fall — is verified to extreme precision:

Weak EP (composition-independence of free fall): torsion balances (Eöt-Wash) and the MICROSCOPE satellite (2017–22) bound the Eötvös ratio $η ≲ 1 0^{- 15}$ .
Local position invariance: gravitational-redshift tests, including optical-clock comparisons over height differences of centimetres and clocks flown on rockets.
Strong EP (self-gravitating bodies fall alike): lunar laser ranging bounds the Nordtvedt effect (would the Earth and Moon fall differently toward the Sun?) to $\sim 1 0^{- 4}$ , and pulsar systems extend this to strongly self-gravitating neutron stars.

Strong-field and dynamical tests

The 21st century opened qualitatively new, strong-field regimes where the nonlinear theory is essential:

Binary pulsars. PSR B1913+16 (Hulse–Taylor) and the double pulsar PSR J0737−3039 exhibit orbital decay from gravitational-wave emission agreeing with the quadrupole formula to $\sim 0.01%$ (double pulsar), plus periastron advance, Shapiro delay, and time dilation — multiple simultaneous strong-field tests in one system.
Gravitational waves. LIGO/Virgo's detections (from GW150914, 2015, onward) confirm GR in the dynamical, strong-field regime: inspiral–merger–ringdown waveforms match numerical-relativity predictions, the ringdown tests the black-hole "no-hair" spectrum, and the graviton is bounded as massless ( $λ_{g}$ large). GW170817 (2017) fixed the speed of gravity to equal $c$ within $\sim 1 0^{- 15}$ , killing many modified-gravity models.
Black-hole imaging. The Event Horizon Telescope resolved the shadows of the supermassive black holes in M87* (2019) and Sgr A* (2022), matching the size and shape predicted by the Kerr metric.
Galactic-center orbits. Tracking stars (S2) around Sgr A* revealed the GR gravitational redshift and Schwarzschild precession of the orbit (GRAVITY collaboration).

Cosmological-scale evidence

On the largest scales, general relativity plus the FLRW framework underlies the $Λ$ CDM standard model, which fits an interlocking web of data:

the cosmic microwave background (Planck) — the acoustic-peak structure of a hot, expanding, nearly-flat universe;
Type Ia supernovae — the 1998 discovery of accelerating expansion (a positive cosmological constant);
baryon acoustic oscillations and gravitational lensing — the growth and distribution of structure;
primordial nucleosynthesis — the light-element abundances from the first minutes.

These successes come with the theory's two great open puzzles at this scale: dark matter and dark energy together dominate the cosmic budget, are inferred only gravitationally, and may signal either new matter or a breakdown of GR on cosmological scales (motivating modified-gravity tests).

Everyday and practical confirmation

General relativity is not only astrophysical: it is engineering. The Global Positioning System must correct for both special-relativistic time dilation (satellite motion, $- 7 μ$ s/day) and gravitational blueshift (weaker potential at altitude, $+ 45 μ$ s/day); the net $+ 38 μ$ s/day, uncorrected, would accumulate navigation errors of $\sim 10$ km/day. Working satellite navigation is a continuous, operational verification of general relativity.

The standing of the theory

Across every accessible regime — weak and strong field, static and dynamical, laboratory to cosmological — general relativity has passed each test, often at the $1 0^{- 4}$ – $1 0^{- 15}$ level, with no confirmed deviation. Its limits are those of principle, not observation: the singularities it predicts and its incompatibility with quantum theory mark where it must eventually yield, but no experiment yet contradicts it. Any successor must reduce to general relativity everywhere it has been tested.

Summary

The classical tests (perihelion precession, light deflection, redshift, Shapiro delay) confirm GR to $1 0^{- 4}$ – $1 0^{- 5}$ ; the PPN framework pins $γ = β = 1$ and rules out competitors.
The equivalence principle is verified to $η ≲ 1 0^{- 15}$ (MICROSCOPE) and, for self-gravitating bodies, by lunar laser ranging and pulsars.
Strong-field / dynamical confirmations: binary-pulsar orbital decay ( $\sim 0.01%$ ), gravitational waves (GW150914; $v_{GW} = c$ from GW170817), EHT black-hole shadows, and galactic-center stellar orbits.
Cosmological evidence underlies $Λ$ CDM (CMB, supernovae, BAO, lensing, nucleosynthesis), with dark matter and dark energy as the open puzzles.
GR is even operational engineering — GPS depends on it — and stands with no confirmed deviation, constraining any quantum-gravity successor.

This completes the general-relativity section. For the flat-spacetime foundation on which it is built, see Special Relativity; for the unfinished union with the quantum world, QFT in Curved Spacetime and Quantum Gravity; and for the conceptual questions the theory raises, Philosophy of Space and Time.

youyuanwu