the tivo characteristic Constants of the JEther of Space. 469 



a square not a triangle, (which had some advantages and some 

 disadvantages : neither obvious) . I tried filling the magnetic 

 solenoids with water, bisulphide of carbon, and salts of iron, 

 as well as with air: and sometimes I passed the light four 

 times both ways round the square. The maximum drop of 

 magnetic potential applied was nearly 2,000,000 c.g.s., and 

 this did not affect the light by so much as 1/100 of a wave- 

 length. 1 could have detected an aether-drift of 1 foot per 

 second, or a thousand millionth of the velocity of light ; but 

 I did not press the experiment to further extremes such as 

 would have decided against a eery slow setherial flow. 



We must now examine reasonable laboratory data to see if 

 by any possibility we could hope so to improve the arrange- 

 ments as to make perceptible a drift of sefcher in a transparent 

 magnetic field of practicable strength, even if ?etherial density 

 is very great. 



Take then the simplest form of magnetic circuit, a 

 closed solenoid or anchor-ring of sectional area A and mean 

 periphery /, and let it be wound with n turns of wire ; then 

 its coefficient of self-induction is ~L = ±7r/ju>i 2 Ajl and the mass 

 included in the ring is M = A/p. 



A current circulating in the wire has the energy ^LI 2 ; 

 while the energy of a hypothetical aetherial flow round and 

 round the ring, is -JMr 2 . 



Assuming that the two expressions are equal, or that one 

 is the equivalent of the other, we get 



V 2 _ L _ 4:7Tfin 2 



P ~ M ~ T^ - ' 



This is the drift to be observed. 



If the current is expressed in electromagnetic units, fx will 

 be 1; and if we assume the density p as 10 12 , J (±ir\p) will 

 be 3i x 10 ~ 6 ; so in that case the velocity v will be very slow, 

 and to make it perceptible we must contrive to make nj. very 

 large. 



Magnetically the size of the solenoid does not matter, but 

 we must arrange to crowd on it as many turns of wire to the 

 centimetre as we can, and to send as bio- a current as possible 

 through the wire without overheating it. 



Let us see whether an extreme case can be thought of which 

 shall come within the limits of experimental possibility. 

 Take a closed magnetic circuit of any convenient sectional 



