No Wavering in Speed of Light

In a world of uncertainty, Einstein said, you can always count on the speed of light. Now a rigorous test of gamma rays from space shows that Einstein was right--at least to 20 decimal places.

Einstein's theory of relativity assumes that the speed of light doesn't depend on how fast the source of the light is moving. In a classic example, photons from headlights of a car traveling near the speed of light should not move any faster than those from a car observing the speed limit. Gamma ray bursts provide a good test of this idea. These violent explosions at the edge of the observable universe emit radiation so energetic that astronomers reason they must be hurling matter through space at high speeds, at least a few percent of the 300,000-kilometer-a-second speed of light. And if a burst flings particles in all directions, some of the sources of radiation must start out moving at different velocities with respect to Earth.

Astronomer Kenneth Brecher of Boston University used recordings of gamma ray bursts to test Einstein's assumption. If the speed of light did depend on the motion of the light source, Brecher says, some gamma ray burst photons would have started with slightly faster speeds than others. During the billions of years it takes such photons to reach Earth, those tiny differences would add up, spreading out the arrival times so that the signal from a burst seems to stretch and blur. Brecher compared the photon pulses from many gamma ray bursts and found that if there are light speed differences among the photons, they must be smaller than 3 billionths of a millimeter per second. "The speed of light is really constant to a precision of one part in 1020," Brecher says. He will present the findings 29 April at a meeting of the American Physical Society in Long Beach, California.

Why do astronomers bother torture-testing a theory that almost nobody doubts is true? "We push as hard as we can, hoping that something breaks," says astronomer Bradley Schaefer of Yale University. Brecher agrees: "No one expects great deviations, but one should test the theories as well as one can."