Clocking a Sunbeam: The 300-Year Race to Measure the Speed of Light

In the summer of 1926, on a concrete pier near the top of Mount Wilson in the hills above Pasadena, a small eight-sided mirror spun so fast that it howled. Compressed air blasted its vanes until it turned more than 500 times a second. Each time one of its faces swung into line, it flung a slice of light from an arc lamp east across the San Gabriel Mountains to a second mirror on a ridge near Mount San Antonio, the peak locals call Old Baldy, 22 miles away. The light went out, bounced back and returned to the spinning mirror about two ten-thousandths of a second later. In that sliver of time the mirror had turned just far enough for the next face to catch the returning beam.

The man running the experiment was Albert A. Michelson, 73 years old, the first American to win a Nobel Prize in science. He had been chasing the speed of light for nearly half a century. His answer that summer, 299,796 kilometers per second, was off by about four kilometers per second, roughly one part in 85,000.

Michelson’s mountain was one stop in a chase that began with a moon that seemed to run late in 1676 and ended, in a way, in 1983, when the world’s metrologists decided that the speed of light would no longer be measured at all. It would be fixed by definition, and the meter would be measured by it. In between came lanterns on hilltops, a toothed wheel in a Paris suburb, a mirror driven by steam, a stone slab floating on mercury and a mile-long pipe in a bean field. And there was at least one stretch when nearly everyone agreed on the wrong answer.

Lanterns on the hills

For most of history, it was not obvious that light took any time to travel at all. Aristotle held that light was a state of things, not a movement. Heron of Alexandria argued that its speed must be infinite, because when you open your eyes at night the most distant stars appear at once. In the 11th century the Arab scholar Ibn al-Haytham, known in Europe as Alhazen, took the other side: light was something that moved from objects into the eye, he wrote, so it must take time, even if that time was hidden from our senses. Johannes Kepler disagreed, and René Descartes staked a surprising amount on the question. If light were not instantaneous, he reasoned, the Sun, Earth and Moon would be visibly out of line during a lunar eclipse. They weren’t, so the speed was infinite. Descartes said that if light turned out to have a finite speed, his whole system of philosophy might be demolished.

Galileo tried a direct test. In his 1638 Two New Sciences he described two people with covered lanterns on distant hills. The first uncovers his lantern; the second uncovers his the instant he sees it; the first times the delay. Galileo said he had tried it at “a short distance, less than a mile” and could not tell whether light was instantaneous, only that if it was not, it was “extraordinarily rapid.” In 1667 the Accademia del Cimento in Florence repeated the experiment with lanterns about a mile apart and saw no delay. They could not have. The true round trip took about 11 millionths of a second, far shorter than any human reaction.

A moon that ran late

The first real answer came from the sky, and it came out of a very practical problem. A navigator who knew the time back home and the local time could work out his longitude, but no clock of the 1600s could keep good time at sea. Galileo had proposed using the four moons of Jupiter he discovered in 1610 as a clock in the sky, since their eclipses could be predicted and seen from anywhere Jupiter was up. The method was hopeless on a rolling deck, as the longitude story of John Harrison and his sea clocks shows. On land it worked, and mapmakers used it for more than a century.

The master of Jupiter’s moons was the Italian astronomer Giovanni Domenico Cassini, whom Louis XIV brought to Paris to run the new Royal Observatory, which opened in 1671. One of Cassini’s projects was to fix the exact longitude of Tycho Brahe’s old observatory at Uraniborg, on an island near Copenhagen. The French astronomer Jean Picard went there to time eclipses while Cassini timed the same ones in Paris. Picard’s assistant was a young Dane named Ole Rømer, fresh from the University of Copenhagen. Picard was impressed enough to bring him back to Paris.

Jupiter’s innermost large moon, Io, circles the planet every 42 and a half hours and slips into Jupiter’s shadow on almost every orbit. From Earth you can see it vanish into the shadow for a few months before Jupiter is opposite the Sun in our sky, and see it pop back out for a few months after. Rømer, timing those eclipses year after year, noticed something odd. The time between eclipses was not quite constant. When Earth, on its own orbit, was moving away from Jupiter, the eclipses came a little later than predicted, and the delay added up over weeks. When Earth was moving toward Jupiter, they came early.

Rømer was confident that Io itself was not speeding up and slowing down. The simpler explanation was that the light carrying the news of each eclipse had farther to travel when Earth was farther away. On August 22, 1676, Cassini told the Royal Academy of Sciences that he was changing the way he calculated his eclipse tables, and may have mentioned that the new correction seemed to be due to light taking time to reach us. Rømer then made a bold forecast: the emergence of Io expected on November 16 would be seen about ten minutes late. No record survives of anyone watching on November 16, but an emergence seen on November 9 fit his new timing, and on November 21 he explained his reasoning to the academy. A short report appeared in the Journal des sçavans on December 7, and in English in the Royal Society’s Philosophical Transactions the next summer.

Rømer’s estimate was that light needed about 22 minutes to cross the full width of Earth’s orbit, or 11 minutes to get here from the Sun. The Dutch physicist Christiaan Huygens combined that with the best guess at the size of Earth’s orbit and got a speed of about 220,000 kilometers per second, a little over a thousand Earth diameters a minute, or, as he put it, more than 16 Earth diameters “in one beat of the pulse.” That was about 27 percent too low, mostly because Rømer’s 22 minutes was too long; the true crossing time is closer to 17 minutes. The modern figure for sunlight’s trip is 8 minutes and 19 seconds.

Cassini never fully accepted the idea. The other moons of Jupiter did not show the effect clearly, and he could not stomach a theory that left some irregularities unexplained. In England it found friends faster. Edmond Halley and the Astronomer Royal John Flamsteed used it, and Isaac Newton gave “seven or eight minutes” for light from the Sun in his 1704 Opticks. Most of Rømer’s own notes were lost when fire swept Copenhagen in 1728, though one manuscript with about 60 of his eclipse timings survived.

The wind in the telescope

The final proof came by accident, from someone looking for something else. In the 1720s the English astronomer James Bradley and his friend Samuel Molyneux set up a telescope aimed straight up at the star Gamma Draconis, which passes almost overhead in London. They hoped to catch stellar parallax, the tiny back-and-forth shift in a nearby star’s position as Earth swings around the Sun, which would prove at last that Earth moves.

The star did shift, but in the wrong rhythm. Parallax should have pushed it farthest south in December and farthest north in June. Bradley’s star reached its extremes in March and September instead. The story goes, probably too neatly, that the answer came to him on a boat on the Thames, when he noticed that the little flag on the mast changed direction every time the boat turned, though the wind had not changed. The direction of the wind you feel depends on how you are moving through it.

Starlight works the same way. Run through vertical rain and you have to tilt your umbrella forward. Earth moves around the Sun at about 30 kilometers per second, so a telescope has to be tilted very slightly in the direction Earth is moving to catch a star’s light, and the tilt swings back and forth over the year. The size of that tilt, now called the aberration of starlight, depends on the ratio between Earth’s speed and the speed of light. Bradley announced it in January 1729 and worked out that light travels about 10,210 times faster than Earth moves in its orbit, which meant 8 minutes and 12 seconds from the Sun. That was only about 1.3 percent off. It confirmed Rømer and, as a bonus, gave direct evidence that Earth really does move around the Sun.

A toothed wheel in Suresnes

For the next 120 years the speed of light was an astronomer’s number, and it depended on knowing the size of Earth’s orbit. Measuring it on the ground, over a distance you could survey, needed a way to time an interval far too short for any clock. Two young Frenchmen found it.

Hippolyte Fizeau and Léon Foucault were friends and partners in the 1840s, part of the first generation to work with photography. In April 1845, using the daguerreotype process, they made one of the first photographs ever taken of the Sun, showing sunspots. François Arago, the director of the Paris Observatory and the man who had championed Augustin Fresnel’s wave theory of light (the story of Fresnel and his lighthouse lens), pushed them toward a bigger prize. He had proposed in 1838 that the speed of light in water, compared with its speed in air, could settle the old argument over whether light was a stream of particles or a wave. Particles should go faster in water; waves, slower.

By 1849 the two had fallen out and gone their separate ways, and Fizeau got there first with the absolute measurement. He put a bright lamp and a telescope in the bell tower of his father’s house in Suresnes, west of Paris, and a mirror in a second telescope on the hill of Montmartre, 8,633 meters away. Between the lamp and the open air he placed a wheel with 720 teeth, spun by falling weights through a set of gears. When the wheel was still, light went out through a gap, hit the Montmartre mirror and came back through the same gap. When it spun slowly, the light still came back through the gap before the next tooth arrived. At about 12.6 turns a second, the returning light hit the next tooth instead and the image went dark. At twice that speed it came back through the following gap and reappeared.

From the speed of the wheel and the length of the path, a little arithmetic gives the speed of light. Fizeau’s figure, about 313,000 kilometers per second, was roughly 4.5 percent too high, largely because it is hard to judge by eye exactly when a faint light disappears. But it was the first measurement of light’s speed made entirely on Earth, over a distance a surveyor could check. Two decades later Alfred Cornu refined the toothed wheel over a much longer path and got within 0.2 percent.

Foucault’s spinning mirror

Foucault, meanwhile, went after Arago’s water test with a different device. The English physicist Charles Wheatstone had used a fast-spinning mirror in 1834 to study electric sparks, and Arago had suggested the same trick for light. Send a beam to a spinning mirror, out to a fixed mirror and back. By the time the light returns, the spinning mirror has turned a tiny bit, so the reflected beam comes back slightly off from where it started. The faster light travels, the smaller the shift.

Foucault had the instrument maker Paul-Gustave Froment build the apparatus and drove the mirror with a jet of steam at 600 to 800 turns a second, so fast that his first mirrors shed their silvering. He split the light into two paths, one through air and one through a three-meter tube of water. On April 27, 1850, he saw that the image from the water path was shifted more. Light was slower in water, as the wave theory predicted. Fizeau, working with the clockmaker Louis Breguet, got the same answer seven weeks later. The result was widely seen as the last nail in the coffin of Newton’s particle theory of light, at least until Einstein revived a version of it in 1905.

A year later Foucault became famous for a very different experiment, his pendulum that showed the Earth turning. But he came back to light. In 1862, working at the Paris Observatory under Urbain Le Verrier, he used the spinning-mirror method to make an absolute measurement over a folded path of only about 20 meters. The shift he had to measure was less than a millimeter. His result, 298,000 kilometers per second, was within 0.6 percent of the modern value. Le Verrier wanted it to pin down the distance to the Sun, and since Bradley’s aberration gave the ratio between Earth’s orbital speed and the speed of light, a good value for one gave the other. The old astronomical figures had put the speed of light a few percent too high, and Foucault’s measurement agreed with Le Verrier’s suspicion that they were wrong.

A midshipman with a mirror

Albert Michelson was born in 1852 in Strelno, then in Prussia, now Strzelno in Poland, and came to America as a toddler. He grew up in the mining camps of the gold and silver rush, in Murphy’s Camp, California, and Virginia City, Nevada, where his father kept a store. In 1869 President Ulysses S. Grant gave him a special appointment to the U.S. Naval Academy, where he excelled in optics. After two years at sea he came back to teach physics and chemistry. Asked later in life why he spent so much of it measuring light, he reportedly said, “Because it’s so much fun.”

In 1877, preparing a classroom demonstration of Foucault’s method, Michelson saw how to improve it. Foucault had kept his spinning mirror close to a lens, which meant the returning light faded badly if the path got long. Michelson rearranged the parts so that the path could be stretched to nearly 2,000 feet with a lens of 150-foot focal length. Instead of Foucault’s shift of less than a millimeter, he had one of about 115 millimeters, big enough to measure well. In June and July of 1879, in a small frame building along the north sea wall of the Academy in Annapolis, he timed his mirror with calibrated tuning forks and measured the speed of light at 299,910 kilometers per second, give or take 50. He was 26. Simon Newcomb, the head of the government’s Nautical Almanac Office, who was planning his own measurement, took Michelson on as a collaborator, beginning a long friendship.

The experiment that found nothing

By then the speed of light had become more than a curiosity. In the 1860s James Clerk Maxwell showed that his equations for electricity and magnetism predicted waves traveling at a speed that matched the measured speed of light, and he proposed that light itself was an electromagnetic wave. Waves seemed to need something to wave in. Physicists imagined a “luminiferous aether,” an invisible medium filling all space, and Earth plowing through it at 30 kilometers per second on its way around the Sun. If so, there should be an aether wind, and light should travel at slightly different speeds with and against it.

The effect would be tiny, about one part in a hundred million for light making a round trip. To catch it, Michelson invented the instrument that now bears his name. A half-silvered mirror splits a beam in two; the halves travel down two arms at right angles, bounce back and recombine. If one trip takes even slightly longer than the other, the light waves fall out of step and produce a pattern of bright and dark bands, called fringes. Turn the apparatus and, if there is an aether wind, the fringes should shift.

His first try, in Germany in 1881, was too crude to settle anything. In 1887, now a professor at the Case School of Applied Science in Cleveland, he tried again with the chemist Edward Morley of neighboring Western Reserve University. In the basement of a heavy stone dormitory, they mounted the optics on a block of sandstone about a foot thick and five feet square, and floated it on a pool of mercury so that one push would set it slowly rotating. Mirrors bounced the light back and forth to stretch each arm’s path to about 11 meters. Even so, passing horse carts and distant thunderstorms could shake the fringes out of sight.

They expected the fringes to shift by about four-tenths of a fringe as the stone turned. The largest shift they saw was about two-hundredths, and the average was much less, about what you would expect from experimental error. Michelson called the result “decidedly negative.” He and Morley stopped hunting for the aether wind and turned their instrument to other work, such as using wavelengths of light as a standard of length.

It turned out to be one of the most important null results in the history of science. The Dutch physicist Hendrik Lorentz suggested that objects moving through the aether might shrink very slightly in their direction of motion, just enough to hide the effect. In 1905 Albert Einstein went further. He took it as a starting point that light in a vacuum moves at the same speed for every observer, however they are moving, and built special relativity on it. The aether became unnecessary. Einstein later wrote that if the Michelson-Morley experiment had not brought physicists “into serious embarrassment,” no one would have regarded relativity as a redemption. In 1907 Michelson won the Nobel Prize in Physics, the first American to win a Nobel in the sciences.

Twenty-two miles across the San Gabriels

Michelson moved to the new University of Chicago in 1892 and spent the rest of his life refining his measurements. By the 1920s his summers were spent at the Mount Wilson Observatory, with a plan to stretch Foucault’s method across a mountain range. The hardest part was not the light. It was knowing the distance.

The U.S. Coast and Geodetic Survey took the job. From October 1922 to March 1923, its crews laid out a baseline about 40 kilometers long in the valley below the mountains, using 50-meter tapes made of invar, an alloy that barely expands with heat. Where the ground was steep they built wooden towers to keep the tape level. At one point a house stood in the way, and with the owner’s help they ran the line straight through it, window to window. From that baseline they triangulated the positions of Mount Wilson and the mirror station near Mount San Antonio. The survey team put its uncertainty at about one part in 500,000, a few inches over 22 miles, among the most accurate measurements of distance ever made at that time.

On the far ridge, at a spot then called Baldy Lookout, Michelson’s team set a two-foot concave mirror on concrete piers to catch the beam and send it home. At Mount Wilson, the spinning mirror had to turn exactly one-eighth of a revolution while the light made its 44-mile round trip, which meant 528 turns a second. One glass mirror shattered at 400. The observatory today says the air-driven mirror “screamed like a siren.” Smoke from forest fires blurred the image, and the magnitude 6.3 Santa Barbara earthquake of June 1925 may have shifted the surveyed points slightly. Still, the definitive runs of 1926 gave 299,796 kilometers per second, give or take four. The concrete piers are still on the mountain, one carrying a metal survey tablet dated 1922.

In January 1931 Einstein, visiting Caltech, addressed Michelson at a dinner. “It was you who led the physicists into new paths,” he said, “and through your marvelous experimental work paved the way for the development of the Theory of Relativity.”

A mile of pipe in a bean field

Michelson knew the air itself was his biggest remaining problem. It slows light slightly, and its shimmer smeared the image. So for his last experiment he took the air away. On the Irvine Ranch in what is now Orange County, near today’s John Wayne Airport, his team built a corrugated steel pipe a mile long and three feet wide, pieced together from 60-foot sections propped on wooden trestles. Sealing the joints took a homemade recipe of rope, canvas, glue, friction tape and old tire inner tubes. Pumps drew the air down to a small fraction of normal pressure, and mirrors bounced the beam back and forth inside to lengthen its path to several miles.

The public followed the work in the papers and came out to see it in such numbers that the team asked people to stay away. Einstein visited too. Michelson, nearly 79 and in failing health, ran the work from his sickbed in Pasadena. He died on May 9, 1931, with only 36 of the planned 233 series of measurements done. His colleagues Francis Pease and Fred Pearson carried on through tides, ground movement and the Long Beach earthquake of 1933, and published the result in 1935: 299,774 kilometers per second.

When everyone agreed on the wrong answer

That result was about 18 kilometers per second too low, and for more than a decade the world believed it. New electronic methods in the late 1930s, which chopped a beam of light with a fast electrical shutter called a Kerr cell, came in close to the same number. In 1941 the American physicist Raymond Birge, reviewing all the evidence, recommended a best value of 299,776 kilometers per second, with an uncertainty of just four. It was precise, consistent and wrong. Historians of measurement have pointed to it as a classic case of what the physicists E. Richard Cohen and Jesse DuMond called “intellectual phase locking,” the tendency of experiments in one era to agree with one another.

The spell was broken by radar. During World War II the British physicist Louis Essen, at the National Physical Laboratory, had worked on microwave equipment, including a device called a cavity resonator, a metal can in which microwaves bounce back and forth and resonate at frequencies set by its size. If you know the can’s dimensions exactly, you know the wavelength; measure the frequency, and the two multiplied give the speed of light. In 1946 Essen and A.C. Gordon-Smith got 299,792 kilometers per second, far higher than the accepted value. The response was skeptical. The laboratory’s director, Sir Charles Galton Darwin, observed that Essen would surely get the correct result once he had perfected his technique. Essen refined the apparatus, with some help on the calculations from Alan Turing, and in 1950 got 299,792.5 give or take a few kilometers per second. Within a few years other methods confirmed him, and his value was adopted internationally in 1957.

The speed that became a ruler

Lasers finished the job. In 1972 a group at the National Bureau of Standards in Boulder, Colorado, led by Ken Evenson, measured both the frequency and the wavelength of a laser locked to a line of methane gas, and multiplied them. They got 299,792,456.2 meters per second, give or take 1.1, about 100 times more precise than anything before. What limited it now was not the light. It was the meter itself, then defined by the orange light of krypton atoms, a line slightly too fuzzy for the job.

So the metrologists turned the problem around. In October 1983 the General Conference on Weights and Measures redefined the meter as the length of the path traveled by light in a vacuum in 1/299,792,458 of a second. Since then, the speed of light in a vacuum is exactly 299,792,458 meters per second, by definition. No experiment can change it. A better experiment today gives a better meter instead. Time had already been tied to atoms in 1967, when the second was defined by the vibrations of cesium. Length was now tied to time by light. It is the same arithmetic that lets a GPS receiver turn the travel times of satellite signals into distances, where a timing error of a billionth of a second means about a foot of error on the ground.

Mirrors on the Moon

The most dramatic use of the old trick, timing light out and back, is aimed at the Moon. In 1962 researchers at MIT bounced laser pulses off the bare lunar surface. Then a Princeton graduate student named James Faller proposed putting reflectors there. On July 21, 1969, Buzz Aldrin and Neil Armstrong set up a panel of 100 small glass corner prisms at Tranquility Base. Each one sends light straight back the way it came, whatever the angle. On August 1 the Lick Observatory in California reported the first returns. Apollo 14 and 15 left more reflectors, the Soviet Lunokhod rovers carried smaller ones, and India’s Chandrayaan-3 lander added another in 2023.

The task is absurdly hard. A laser pulse takes about two and a half seconds for the round trip. By the time it reaches the Moon, the beam has spread to several kilometers wide, and of roughly 300 million billion photons in each pulse, only a handful, typically one to five, find their way back into the telescope. Yet by timing those stray photons, stations in France, Germany and New Mexico track the Earth-Moon distance to within millimeters. The measurements have shown that the Moon is drifting away from us by about 3.8 centimeters a year and have confirmed that Einstein’s general theory of relativity predicts the Moon’s orbit as precisely as the lasers can measure it.

New reflectors are still arriving. In March 2025 Firefly Aerospace’s Blue Ghost lander delivered a Next Generation Lunar Retroreflector, a single large prism designed by a University of Maryland team, and the observatory at Grasse in France ranged it the very next day. The University of Maryland team reports that early measurements from Germany’s Wettzell station reached a precision of about a millimeter. China’s Chang’e-6 lander left the first reflector on the Moon’s far side in 2024, and a small rover due to fly on Astrobotic’s Griffin-1 lander is slated to carry a NASA reflector array to the south pole region.

Rømer would recognize the idea at once. He watched a moon of Jupiter blink out of sight and noticed that the news arrived late when Earth was farther away. Today’s astronomers fire a pulse at our own Moon and wait for a few photons to come home. The difference is that the number Rømer was groping for, the one that took Fizeau’s wheel, Foucault’s mirror and Michelson’s mountain to pin down, is no longer in question. It has become the yardstick. On a clear night when Jupiter is up, a small telescope will show Io sliding into Jupiter’s shadow, right on a timetable that now accounts for the minutes its light spends crossing the solar system.

Further reading