696 Prof. C. Kinsley on 



The average distance between the mirrors A and B was 

 25 mm. 



_ 4zci \a 2a 



COS a COS [a. — 2x) COS (a — 4.f) 



D ]0 = -la<f— + / , + -A— — + 7^—7^+ j^-jr-. V 



[ cos a cos (a — 2#) cos (a— 4a?) cos(a — b^) cos (a— ox) J 



The tenth order was used in every case given in this paper 

 except on October 28th, when the fifth order was used. 



a^22\°. For the tenth order x = 2± 9 . 



Substituting in the above equations 



If the green light of mercury is used, A, = 546jua*« When 

 sodium light is used X=589/^/i. 



.\ d = 28'6fi , for mercury light and the tenth order. 



d=61'4:f 5 , for sodium light and the fifth order. 



The spark was obtained by discharging a condenser across 

 the gap separating the surfaces. The condenser was charged 

 by a set of small storage-cells, and the charge was renewed 

 each four seconds to ensure the constancy of the potential. 

 The leakage, however, was small, as the insulation resistance 

 of the glass sleeve was 16,500 megohms, of the switches and 

 conductors 170,000 megohms, which gave a total insulation 

 resistance of 15,000 megohms. The diagram of connexions 

 is shown in fig. 3. 



The capacity, O, was 1 microfarad. The voltage used with 

 the high-resistance galvanometer for finding the zero was 

 about 5^00 °f a vo ^* Non-inductive resistances i\ and r 2 

 were bridged across the inductive parts of the circuit. The 

 difference of potential that produced the spark- discharge was 

 measured by the Kelvin electrostatic voltmeter, or by the 

 ballistic throw of the galvanometer upon discharging the 

 condenser. The use of the single galvanometer was desirable, 

 as it was necessary to note the position of the fringes with refer- 

 ence to the cross-hair of the microscope at both the instant that 

 the circuit was broken in obtaining the zero, and also at the 

 instant the spark-discharge occurred, as shown by the ballistic 

 throw. This was accomplished by mounting the galvano- 



