476 SCIENCE PROGRESS 



seatation of the transmission of electromagnetic action by means of a medium, as 

 opposed to the theory of action at a distance, showed that any visible optical 

 effects due to the relative motion of ether and matter must, if existing at all, be 

 exceedingly small. For no effect proportional to vjc, the ratio of the velocity of 

 translation of the matter relatively to the ether, to the velocity of light in free 

 ether, about 300,000 kilometres a second, was to be expected, but only effects 

 proportional to the square and to higher powers of this ratio. 



It was suggested by Maxwell that it might be possible to detect a difference in 

 the time of propagation of a ray of light between two points at a fixed distance from 

 each other, when the straight line joining them was placed, first in the direction of 

 the earth's orbital motion, and then in a direction at right angles to it. The 

 highest value that could be attained by v would be that due to the motion of the 

 earth through an ether at rest, and the principal component of this, amounting to 

 about 30 kilometres a second, will be that due to the orbital motion. The com- 

 ponent due to the earth's rotation is not quite half a kilometre a second, and may 

 be neglected in comparison with the former. There will also probably be a 

 component of unknown amount due to the motion of the whole solar system 

 through space. But this will increase or diminish the component due to the 

 orbital motion according to the position of the earth in its orbit, so that for at least 

 half the year v will have a value of at least that due to the orbital motion. 



The experiment was first made by Michelson in 1881, no effect being observed. 

 It showed that, as predicted by theory, there was no first order effect, i.e. no effect 

 proportional to vjc, but the method was not considered sufficiently sensitive to 

 prove beyond doubt the absence of any second order effect, i.e. one proportional 

 to the square of v\c. The experiment was therefore repeated with much greater 

 refinements of detail by Michelson and Morley in 1887. The essentials of the 

 apparatus employed consisted of a metallic framework with two arms, OA and OB, 

 at right angles, and as nearly as possible of equal length, provided at their free 

 ends with metallic mirrors perpendicular to them, and with a mirror of unsilvered 

 glass at O, with its plane bisecting the angle between them. A convenient source 

 of light was placed on AO produced through O, and a telescope was fixed with its 

 axis in the line BO produced through O. The whole arrangement was bolted to 

 a stone block floating in mercury, so that the arms lay in a horizontal plane, and 

 could be rotated slowly and steadily about a vertical axis. A beam of light from 

 the source would then be split up at O, part being transmitted to A, reflected back 

 to O, and part of this reflected along the axis of the telescope to the observer's 

 eye. The other portion of the split beam would be reflected to B, reflected back 

 to O, and a part transmitted along the axis of the telescope. The smallest differ- 

 ence in the lengths OA and OB would give rise to coloured interference fringes. 

 The apparatus was adjusted so that the fringes were as sharply defined as 

 possible when OA represented the direction of the earth's orbital motion. The 

 ray would then travel from O to A with the velocity c + v, and from A to O 

 with the velocity c — v, while the other part would travel from O to B and back 

 again with the velocity v. Simple geometry will then show that the time taken by 

 the former will be greater than that taken by the latter by one part in a hundred 

 millions, the ratio v 2 c/ 3 . If the apparatus were then turned through a right angle, 

 making OA instead of OB perpendicular to the component of orbital motion, 

 calculation shows that for a velocity of 30 kilometres a second the fringes should 

 be displaced by an amount equal to about two-fifths of the distance between two 

 successive fringes, provided the arms OA and OB were 1 1 metres in length. The 

 actual arms were much shorter, but their effective length was increased to 1 1 metres 



