70 Prof. R. A. Sampson on the 



ascertainable in every case except when three of the points 

 A , A 1? A 2 , A 3 lie upon one straight line. The argument 

 then has no bearing on the Principle of Relativity for linear 

 motion, even if we include accelerated linear motion. The 

 latter stands just where it did. 



It is worth remarking that the Principle of Relativity, 

 owing to its generality in asserting that no experiment can 

 be derived that will falsify it, can never be established except 

 upon trial. It suffers under the difficulty always found in 

 proving a negative. Thus the Michelson-Morley experiment 

 is, of course, consistent with it, but the experiment is explained 

 without its aid by adopting the FitzGerald contraction, which 

 there is otherwise strong reason to accept. 



At each successive test of the principle there is the same 

 possibility of escape. Thus, though the test imagined above 

 fails as a criticism of its application to rectilinear motion, it 

 is easy to imagine a test case which would exclude acceleration 

 from its domain. The annual aberration of the stars is, in 

 fact, just an observation of the variation of the motion of the 

 Earth in direction ; and if we could observe with sufficient 

 refinement, it would equally reveal a variation in its amount. 

 The only field that remains is that of uniform linear motion ; 

 and even here 1 cannot see that it is free from attack. The 

 gradual transmission of light was first demonstrated by 

 Roemer from the eclipses of Jupiter's first satellite. As the 

 Earth approaches Jupiter, immersions alone are observed, 

 and as it recedes only emersions, the changing distance is 

 marked by an apparent reduction of the period of revolution 

 in the former case, and a compensating increase in the latter. 

 The quantities have been observed, and agree with dimensions 

 of the orbits of the Earth and Jupiter found by Euclidean 

 triangulation and a velocity of light found from terrestrial 

 experiments. The comparisons take place in all directions 

 in the ecliptic. The quantity directly determined is the time 

 taken by light to travel a distance equal to that travelled by 

 the Earth between two eclipses. There is here no question 

 of contraction of the measuring rod, for the phenomenon is 

 one involving the relative velocity of light in the first order. 

 Hence we can conclude that if there is any undiscovered 

 drift of the solar system through the aether, it must fake 

 place in a direction perpendicular to the plane of the ecliptic; 

 this is reliable to such a degree of accuracy as these obser- 

 vations will bear, and that is by no means an inconsi- 

 derable one. 



If we are unable to draw a conclusion one w T ay or another 

 with regard to relative drift of the ?ether perpendicular to the 

 plane of the ecliptic, that is only because it happens that we 



