Theoretical Evaluation of 7. 645 



the rotation about the axis of symmetry being excluded. We 

 have now 



7= if, 



/(E)=|E, 



if we assume the rotation to be uninfluenced by reaction with 

 the surrounding aether, and 



/(B) =§, 

 while equation (5) becomes 



§ = -x(E )(F-fE). 



There is considerable difficulty in forming a reliable esti- 

 mate of the value of x(^o)- The on b r case in which I have 

 succeeded in calculating such a value is the case in which 

 the molecules are supposed to consist of geometrically perfect 

 spheres (of radius R and mass m), in which the centre of 

 gravity is at a small distance r from the geometrical centre. 

 In this case the value is 



where k is the radius of gyration*. 



The factor which depends on the geometrical structure of 

 the molecule is r 2 R 2 /k 2 . For a molecule in which the diver- 

 gence from spherical symmetry is not small, we may pro- 

 bably, without error as regards order of magnitude, replace 

 rjk 2 by unity. This gives, as an approximate value, 



x (K)=8Wnc, 



where n is the number of molecules per cu. centim., and c 2 

 is the mean-square of their velocity. Taking R=10 -8 cm., 

 n=10 20 , c = 4x 10 4 , we obtain 



X (E) =3xW (20) 



In the same units as before, we may take ft = '00019, 

 ct = p V 2 /y = 8'6 , x 10 5 , and the value of expression (19) 

 becomes 



The numerical estimate of #(E) is so vague that very little 

 importance can be attached to the actual result obtained. 



* The calculation is too lengthy to be reproduced here. I have 

 indicated the method elsewhere (Phil. Trans, cxcvi. p. 399). 



Phil. Mag. IS. 6. Vol. 2. No. 12. Dec. 1901. 2 U 



