Kennedy-Thorndike experiment
The Kennedy–Thorndike experiment was an interferometer test carried out by Roy J. Kennedy and Edward M. Thorndike and published in 1932, designed to detect a dependence of the speed of light on the velocity of the apparatus rather than on its orientation. It is the companion to the Michelson–Morley experiment, and the two together are what actually pin down the Lorentz transformation — a point that is easy to miss, and that a good deal of the argument catalogued on this wiki turns on.
What it did, and why the difference matters
The 1887 Michelson–Morley apparatus had two arms of equal length and was rotated. A rotation test asks whether light takes different times along different directions. If the arms are equal, an isotropic slowing of clocks in the moving frame cancels out of the comparison, so the experiment is sensitive to length contraction but blind to time dilation.
Kennedy and Thorndike therefore built an interferometer with arms of deliberately unequal length — differing by about 16 cm, the largest difference their light source's coherence would tolerate — and did not rotate it. Instead they left it fixed and photographed the fringe pattern over months, letting the Earth's rotation and its orbital motion change the instrument's velocity. Under the Lorentz–FitzGerald contraction hypothesis alone, with no time dilation, an unequal-arm interferometer's fringe pattern should shift as the speed of the apparatus changes. The apparatus was mounted in vacuum in a temperature-controlled bath, since with unequal arms a fraction of a degree of drift would swamp the effect. They found no shift attributable to the Earth's motion.
The logical structure was made explicit by H. P. Robertson in 1949 and refined by Mansouri and Sexl in 1977. Three classes of experiment are needed:
- Michelson–Morley constrains the dependence of light speed on the direction of travel — length contraction.
- Kennedy–Thorndike constrains the dependence of light speed on the speed of the laboratory — time dilation.
- Ives–Stilwell (1938) measures the transverse Doppler shift of moving ions and so measures time dilation directly.
Take any two and the third is not implied. Take all three, and the Lorentz transformation follows: no alternative combination of contraction and dilation factors survives. Modern repetitions of the Kennedy–Thorndike test use cryogenic optical resonators compared against atomic clocks and have tightened the original bound by many orders of magnitude.
On this wiki
The Kennedy–Thorndike experiment matters here for a specific reason. A large fraction of the aether and relativity material on this wiki proposes to explain the 1887 null result by a physical contraction of the apparatus in the direction of motion — a real FitzGerald–Lorentz contraction in a real medium, without Einstein's kinematics. That proposal is coherent, and it was Lorentz's own. But contraction alone does not survive Kennedy–Thorndike: an unequal-arm interferometer under contraction-without-dilation should have shown a seasonal fringe drift, and did not. Any aether model offered here has to say what it does about both experiments, and readers should check whether a given paper does.
Contributors who take the pair together include:
- William L Hughes, "Michelson-Morley, Trouton-Noble, Kennedy-Thorndike, and the Fitzgerald-Lorentz Contraction" — treats the three classic null experiments jointly against the contraction hypothesis, which is the right way to pose the question.
- Randy Reukauf, "A Particle Explanation of the Michelson-Morley and Kennedy-Thorndike Experiments" — an emission-style account intended to give both null results at once.
- Helmut Hansen, "The Hidden Face of c, or The True Meaning of the Kennedy-Thorndike Experiment" — argues that what the experiment fixes is a property of c itself rather than of the apparatus.
The wider dispute over what the interferometric evidence shows is set out at Michelson–Morley experiment, Michelson interferometer and Albert A. Michelson, and the alternative-kinematics positions at Lorentz ether theory and Preferred frame. Reginald T Cahill's gas-mode argument, which holds that vacuum-based instruments must be null while gas-mode ones need not be, applies to Kennedy–Thorndike as it does to Michelson–Morley — their apparatus was evacuated.