166 



Scientific Proceedings, Royal Dublin Society. 



To measure times of the order of 100 micro-seconds (100 x 10"' sec), if ^o is 

 to be ten times q, we see, from (3), that the product BC should be about -j-oo-o- 

 Having a very well insulated mica condenser of one microfarad capacity, we 

 made B 1000 ohms in most of our observations; but, with a view to testing 

 the method somewhat, we varied ^ in a few cases. The actual values of B and 

 C required are to some extent determined by the sensitivity of the galvano- 

 meter, but the battery e.m.f. can also be modified so as to give q and j'q 

 suitable values. To measure q^ and q a ballistic galvanometer with a 

 universal shunt was employed, so that large, and nearly equal, deflections 

 were obtained for both q^ and q. An eight-volt battery was employed 

 throughout, and the galvanometer had a ballistic sensitivity of about 450 

 scale divisions per micro-coulomb. 



■ i 



B 



R 



-WA/\AAAAA\^ 



•-^VWWWV^ 



-J^ 



D 



Kf 



Kf 



E 







Fig. 1. 



The main difficulty to be overcome in practice is the means to be adopted 

 by which the condenser is caused to commence and cease charging at the 

 beginning and termination, respectively, of the time interval to be measured. 

 In order that accidental and constant errors should be reduced to a minimum, 

 it is necessary to have the commencement and termination of the charging 

 brought about by two operations as similar as possible. We attempted to do 

 this by arranging that these operations should each consist in the breaking of 

 fine copper wires, in which practically the same currents would be passing. 

 These are shown in the diagram (fig. 1) by the dotted lines. The wires were 

 soldered to the brass plates {A, B, D), and care was taken to have them under 



