On the morning of November 6, 1919, a crowd gathered outside the Royal Astronomical Society in London to hear results that would reshape humanity’s understanding of the universe. Arthur Eddington and Frank Dyson had spent months analyzing photographs taken during a solar eclipse earlier that year — photographs designed to answer a single, audacious question: does the sun bend light? Their answer was yes, and by exactly the amount Albert Einstein had predicted. Einstein became a household name overnight.
But here’s what the newspaper headlines missed: that moment wasn’t a coronation. It was the opening round of a scientific examination that has never stopped. More than a century later, physicists are still designing ever more ingenious experiments to find the crack in Einstein’s edifice — and the theory, almost infuriatingly, keeps holding. Understanding how scientists conduct that examination tells us something profound not just about general relativity, but about the nature of scientific knowledge itself.
What General Relativity Actually Claims
Before you can test a theory, you have to understand what it asserts. General relativity, published by Einstein in 1915, is not simply a story about gravity being a force. It is a geometric theory: massive objects curve the fabric of space-time, and what we experience as gravitational attraction is really objects following the straightest possible paths — called geodesics — through that curved geometry.
This produces a cascade of specific, quantitative predictions. Light should bend when passing near a massive object. Clocks should tick more slowly in stronger gravitational fields — a phenomenon called gravitational time dilation. Orbiting bodies should precess in ways that Newton’s theory cannot fully account for. Accelerating masses should radiate energy in the form of gravitational waves, ripples in space-time itself. And sufficiently compressed matter should produce regions from which not even light can escape: black holes.
Each of these predictions is a potential falsification target. A good physical theory, as Karl Popper famously argued, must be falsifiable — it must make predictions specific enough that an experiment could, in principle, prove them wrong. General relativity clears that bar with room to spare. The theory’s mathematical precision means it doesn’t just say “light bends near the sun.” It says light should deflect by approximately 1.75 arcseconds — a specific number that observations either confirm or contradict.
The Classical Tests: Eclipse Chasing and Mercury’s Wobble
The story of how general relativity survived its first generation of tests is a masterclass in experimental ingenuity under constraint. Eddington’s 1919 eclipse expedition is famous, but it had a predecessor that Einstein himself valued more: the anomalous precession of Mercury’s orbit.
Mercury’s orbit doesn’t trace a clean ellipse that repeats perfectly. Instead, the point of closest approach to the sun — the perihelion — advances slightly with each orbit, rotating around the sun over long timescales. Newtonian mechanics predicted most of this precession based on gravitational tugs from other planets. But there was a stubborn residual of about 43 arcseconds per century that Newtonian physics simply couldn’t explain. General relativity, when applied to the curved space-time near the sun, accounts for that discrepancy exactly. This wasn’t a post-hoc fit: Einstein derived the number from first principles, and it matched.
The eclipse observations tested a different prediction: gravitational lensing of light. By photographing stars near the sun’s edge during a total eclipse — when the sun’s own glare is blocked — astronomers could compare star positions to where they appeared at night, when the sun was absent. A deflection matching Einstein’s prediction would confirm space-time was curved. Eddington’s results were noisy by modern standards, with significant measurement uncertainty, but they pointed in the right direction. Subsequent eclipse expeditions across the 20th century refined the measurements and closed in on Einstein’s predicted value with increasing precision.
A third classical test involved the gravitational redshift of light. If photons climbing out of a gravitational well lose energy, their frequency should shift toward the red end of the spectrum. This was confirmed with exquisite precision in 1959 by Robert Pound and Glen Rebka at Harvard, who measured the redshift of gamma rays over a vertical drop of just 22.5 meters — proving that even Earth’s modest gravity affects the frequency of light.
Atomic Clocks, GPS, and Precision at the Parts-Per-Billion Level
The abstract machinery of general relativity has a very concrete footprint in modern life. Every time you use a GPS navigation system, you are relying on corrections derived from Einstein’s theory. The satellites in the GPS constellation carry atomic clocks, and those clocks run fast relative to ground-level clocks for two distinct reasons: special relativity slows them down because of their orbital velocity, while general relativity speeds them up because they sit higher in Earth’s gravitational field. The net effect amounts to a gain of roughly 38 microseconds per day. That sounds trivial until you remember that GPS systems accumulate positioning errors at a rate of about 300 meters per microsecond of clock drift. Without the relativistic corrections, navigation would become useless within hours.
This makes GPS a daily, global test of general relativity at a precision of parts per billion. And the theory passes every day.
Laboratory tests have pushed precision even further. In 2010, physicists at the National Institute of Standards and Technology demonstrated gravitational time dilation over a height difference of just 33 centimeters — about the length of a human forearm — using aluminum-ion optical clocks. By 2022, researchers had used optical atomic clocks to map gravitational time dilation at the millimeter scale, a sensitivity that opens the door to measuring subsurface density variations on Earth by detecting how they alter the local passage of time. These experiments don’t just confirm general relativity; they are pushing the theory into regimes where potential deviations might finally appear.
Gravitational Waves: Hearing the Universe’s Loudest Events
If the atomic clock experiments represent precision physics in a quiet laboratory, the detection of gravitational waves represents the opposite: catching the most violent events in the universe from 1.3 billion light-years away.
On September 14, 2015, the Laser Interferometer Gravitational-Wave Observatory — LIGO — detected a signal lasting less than a second. The signal was the gravitational wave signature of two black holes, each roughly 30 solar masses, spiraling together and merging. The entire event released more energy than all the stars in the observable universe emit in light over the same fraction of a second, almost entirely as gravitational waves. The stretching and squeezing of space-time that reached Earth amounted to a displacement of about one-thousandth the diameter of a proton across LIGO’s 4-kilometer detector arms.
That detection, announced in February 2016, confirmed a prediction Einstein himself had made in 1916 but privately doubted would ever be verified. Since then, the LIGO-Virgo-KAGRA network has logged more than 390 confirmed gravitational wave events, including neutron star mergers and black hole collisions, creating an entirely new branch of astronomy. Each event is a test. The waveforms — the rising chirp of frequency as objects spiral inward, the ringdown as a newly formed black hole settles — are predicted in exquisite mathematical detail by general relativity. The observed waveforms match the predictions to within measurement uncertainty, time after time.
The neutron star merger detected in 2017, known as GW170817, was particularly powerful as a test. It arrived simultaneously with a gamma-ray burst detected by the Fermi and INTEGRAL telescopes, arriving within 1.7 seconds of each other after traveling roughly 130 million light-years. General relativity predicts gravitational waves should travel at exactly the speed of light. The observation constrained any difference between the two speeds to less than one part in 10 to the 15th power — an almost incomprehensibly tight bound that ruled out dozens of competing gravitational theories overnight.
The Event Horizon Telescope and the Image That Silenced Skeptics
Some predictions of general relativity are so extreme that physicists spent decades debating whether the mathematics corresponded to anything real. Black holes — regions where space-time curvature becomes infinite at a singularity, surrounded by an event horizon from which escape is impossible — were long regarded by some physicists as mathematical artifacts. Einstein himself, in a 1939 paper, argued that black holes could not actually form in nature.
He was wrong. In April 2019, the Event Horizon Telescope collaboration released the first image of a black hole’s shadow: the supermassive black hole at the center of the galaxy M87, a beast of approximately 6.5 billion solar masses. The image showed a bright ring of glowing gas surrounding a dark central void — exactly the structure general relativity predicts, produced by photons orbiting the black hole before being captured or escaping to our telescopes. The “photon ring” has a specific predicted diameter relative to the black hole’s mass, and the observed size matched the prediction to within the image’s resolution.
In 2022, the same collaboration released an image of Sagittarius A*, the black hole at the center of our own Milky Way galaxy. Imaging Sgr A* was technically harder because the black hole is smaller and its environment changes on timescales of minutes rather than days, requiring novel computational techniques to reconstruct a stable image from hours of variable data. The result again matched general relativity’s predictions — a bright ring structure around a shadow consistent with a 4-million-solar-mass object.
The Event Horizon Telescope is not a single instrument. It is a network of radio telescopes spread across Earth — from Hawaii to Spain to the South Pole — linked by atomic clocks and combined using a technique called very long baseline interferometry. The effective aperture is the diameter of Earth itself. That this baroque, planet-spanning machine was built specifically to test a prediction made in 1915 speaks to the lengths to which the scientific community will go to probe a theory’s limits.
The Frontier: Where Exotic Matter, Quantum Gravity, and Alternative Theories Await
If general relativity keeps passing every test, why do physicists keep searching for its failure? Because they know, with certainty, that it cannot be the final word.
General relativity is a classical theory — it does not incorporate quantum mechanics. Yet quantum mechanics governs everything at subatomic scales with equal precision. The two theories are famously incompatible: when you try to apply quantum field theory in strongly curved space-time, the mathematics produces infinities that cannot be tamed. This is not a minor bookkeeping problem. It means that at the center of black holes, where quantum effects and extreme curvature coexist, neither theory can be trusted. Something deeper must exist.
String theory and loop quantum gravity are the most developed candidates for a unified framework, but neither has produced predictions that experiments have yet confirmed or denied. The search for quantum gravity effects at accessible energies is one of the driving motivations for ultra-precise tests of general relativity: deviations from Einstein’s predictions at some level might signal the first fingerprints of a deeper theory.
Exotic matter complicates the picture further. General relativity’s field equations permit solutions — valid mathematics — describing wormholes and certain types of faster-than-light shortcuts through space-time. These solutions require matter with negative energy density, which physicists call exotic matter. The Casimir effect, a quantum phenomenon first predicted in 1948 and confirmed experimentally decades later, demonstrates that negative energy densities can exist in nature, at least locally and transiently. But whether exotic matter can exist in the quantities or configurations needed to stabilize a traversable wormhole remains deeply uncertain. Most physicists regard such solutions as mathematical curiosities rather than engineering blueprints — but the fact that general relativity permits them at all is itself a profound statement about the theory’s structure.
The Laser Interferometer Space Antenna — LISA — a space-based gravitational wave detector approved by the European Space Agency and scheduled for launch in the mid-2030s, will test general relativity in new regimes: extreme mass ratio inspirals, where a small black hole spirals into a supermassive one, will trace geodesics in curved space-time with a precision impossible from the ground. Pulsar timing arrays, which use networks of millisecond pulsars as natural gravitational wave detectors, produced the first evidence of a gravitational wave background — a low-frequency cosmic hum — announced by multiple collaborations in 2023.
What makes general relativity extraordinary is not that it hasn’t been broken. It’s that every attempt to break it has required building instruments of previously unimagined capability — detectors sensitive to motions smaller than an atomic nucleus, telescopes the size of a planet, clocks accurate to 18 significant figures — and that each new instrument has revealed a universe that looks exactly as Einstein predicted. The theory is a century old. The tests are still getting harder. And the theory keeps passing.
That, more than any single result, is what scientific confidence actually looks like: not a declaration of certainty, but a steadily deepening trust earned through a century of failed attempts at falsification. The day a well-designed experiment catches general relativity in a lie will be one of the greatest days in the history of science. Until then, physicists will keep trying — because that is precisely what the scientific method demands.