400 



PROCEEDINGS OP THE AMERICAN ACADEMY. 



When we remember that the percentage error of the time is doubled 

 in the calculation of capacity, it is seen that these two computed values 

 of C agree with an error not greater than the error of observation, or 

 the errors possibly made in the computations of L^ and Lj. 



In this way the inductance of tlie leads and the capacities of the con- 

 densers were determined in a number of cases (Table III) to be used in 

 the discussion of the results obtained in section VI for the resistance of 

 the interrupter. 



TABLE III. 



n = the number of plates in condenser, 



T = period in millionths of a second for the discharge through the leads alone. 



T' = period through leads and .000106 henries in series. 



In putting in additional condensers up to nineteen plates the induct- 

 ance of the leads had to be increased, hence the progression in the first 

 three values of column four. To get tweuty-four plates, plates were 

 introduced back nearer to the interrupter by leads whose direction was 

 such as to diminish the inductance, giving the smaller value of induct- 

 ance in the last line. The capacities are correct within about 1%. The 

 inductances having been determined as a difference, may contain an 

 error as great as 2%. 



V. Photographs Showing Rapidity of Recovery of the 



Mercury Interrupter. 

 The revolving-mirror photographs of the mercury interrupter show 

 that when the condenser in series with the interrupter is charged to a 

 sufficiently high potential the matter in the globe in some way becomes 

 conducting, and that this conductivity continues during a series of oscilla- 

 tions. In this respect no difference is apparent between the action of 

 the mercury interrupter and the ordinary discharge of the condenser 

 between metallic terminals in air. It thus occurs to one that the advan- 



