EXPERIMENTS ON ETHER-DRIFT 25 



is the medium by which light travels from one point to another. 

 If it be similar to material media (such as air in the case of 

 sound), it is to be expected that the velocity with which light 

 travels with reference to a fixed observer will be affected by the 

 movement of the ether relative to him. With so small a drift 

 as even 31 miles per second, however, very little effect could be 

 produced on the velocity of light, 186,000 miles per second; 

 and only by very accurate means could it be detected. Perhaps 

 the most obvious method of procedure, yet, nevertheless, only 

 recently proposed, is that described by the Hon. R. J. Strutt 1 

 in November last. The proposal is to compare the velocity of 

 light differentially with and against the drift at a time when it 

 is horizontal. The method is an extension of Fizeau's well- 

 known method of determining the velocity of light, and virtually 

 amounts to measuring this velocity in the two directions. A 

 difference of velocity of twice that of the drift is to be expected ; 

 and there appears to be no reason why this experiment, if 

 practicable, should not at length settle the question of ether- 

 drift. Unfortunately, the experimental difficulties to be over- 

 come would be enormous. Apart from the very great one of 

 securing equality of phase in the two rotating wheels which 

 are essential to the determination, it is to be remembered that 

 Fizeau's method of determining the velocity of light does not 

 lend itself to great accuracy. In effect, this velocity would 

 have to be measured correct to one part in ten thousand, and 

 even the best of recent determinations can hardly claim such 

 freedom from error. 



But, although it has up to the present been impracticable to 

 carry out an experiment of this type to the necessary order of 



exactness, viz. y, where u is the velocity of the supposed drift 



and V is the velocity of light, observations of a different 



type have been conducted, in which an accuracy beyond (^ ) 



has been successfully attained. This means a measurement 

 correct to less than one part in a hundred million. The first of 

 these experiments — the famous one of Michelson and Morley 2 

 — is doubtless sufficiently familiar to make more than a brief 

 description unnecessary. It is, in principle, to compare the 



1 R. J. Strutt, New Quarterly, November 1907. 



2 Michelson and Morley, Phil, Mag. vol. xxiv. 1887. 



