Page:Eddington A. Space Time and Gravitation. 1920.djvu/34

18 which shows how the result depends on the ratio of the speed of the current to the speed of the swimmer, viz. $30⁄50$. A very famous experiment on these lines was tried in America in the year 1887. The swimmer was a wave of light, which we know swims through the aether with a speed of 186,330 miles a second. The aether was flowing through the laboratory like a river past its banks. The light-wave was divided, by partial reflection at a thinly silvered surface, into two parts, one of which was set to perform the up-and-down stream journey and the other the across-stream journey. When the two waves reached their proper turning-points they were sent back to the starting-point by mirrors. To judge the result of the race, there was an optical device for studying interference fringes; because the recomposition of the two waves after the journey would reveal if one had been delayed more than the other, so that, for example, the crest of one instead of fitting on to the crest of the other coincided with its trough.

To the surprise of Michelson and Morley, who conducted the experiment, the result was a dead-heat. It is true that the direction of the current of aether was not known—they hoped to find it out by the experiment. That, however, was got over by trying a number of different orientations. Also it was possible that there might actually be no current at a particular moment. But the earth has a velocity of $18 1⁄2$ miles a second, continually changing direction as it goes round the sun; so that at some time during the year the motion of a terrestrial laboratory through the aether must be at least $18 1⁄2$ miles a second. The experiment should have detected the delay by a much smaller current; in a repetition of it by Morley and Miller in 1905, a current of 2 miles a second would have been sufficient.

If we have two competitors, one of whom is known to be slower than the other, and yet they both arrive at the winning-post at the same time, it is clear that they cannot have travelled equal courses. To test this, the whole apparatus was rotated through a right angle, so that what had been the up-and-down course became the transverse course, and vice versa. Our two competitors interchanged courses, but still the result was a dead-heat.