Bearing of Rotation on Relativity. 69 



as proving that the Earth swept with it the aether in its 

 vicinity, spun two large heavy metal disks, parallel and close 

 to one another, with high angular velocity, in order to test 

 whether they exercised any viscous drag upon the aether 

 between them. Light was transmitted to and fro across the 

 intervening space, but the effect upon it was null, Now 

 return to the two spheres A and B. Let their surfaces be 

 perfect reflectors. Let light signals be emitted from A. 

 Then according to Lodge's experiment, whatever rotations 

 the spheres may have, the passage of this light will be un- 

 affected by it. Let B be fixed and diametrically opposite 

 to A at the moment of emission. The waves or rays will 

 spread in a sheet over the sphere, converge upon B at the 

 same moment and in the same phase, and issuing from B will 

 return again to the point from which they were emitted. 

 But if A has moved in the interval, this will permit the 

 observer A to ascertain the point in space which he occupied 

 at a given past moment, A say. It is the point upon which 

 the rays converge and are received all together and at once in 

 the same phase — that is, the point diametrically opposite to B. 



It may be said that when the globe A is in rotation, it 

 must be supposed to undergo the Fitz Gerald contraction 

 parallel to its equator, deforming it from a sphere; and 

 therefore the geodesies which are the paths of rays will in 

 general no longer converge upon a single point as they do 

 for a sphere, but pass alongside it. This, however, is a 

 second-order effect, and would not interfere with the deter- 

 mination of AA , which is of the first order ; moreover, it 

 would of itself demonstrate the rotation by the non-concurrence 

 of the rays. We should therefore conclude that by the use 

 of signals transmitted through the aether it is possible to 

 determine the motion of A, imagined as a rotation. There is 

 no sign of conflict here with results of experiments with 

 matter alone, referred to earlier ; but the conclusion bears 

 awkwardly upon the Principle of Relativity for linear 

 motions, and it is desirable to examine whether the division 

 is apparent only or real. 



Consider whether rotation is essentially connected with the 

 discussion. Let A . A ]5 A 2 , A 3 be four points successively 

 occupied by the observer. Describe a sphere through 

 A , A x , A 2 , A 3 , and let light pass round it as above, being 

 emitted from A . Adjust the rotation of this sphere so that 

 the time taken to carry the observer from the position A to 

 the position A] is equal to the time taken by light to girdle 

 the sphere. Then the argument runs just as before, and the 

 conclusion can be drawn that the previous position of A is 



