OF ARTS AND SCIENCES. 147 



cylindrical part of each bottle under the neck was about six centimeters 

 high, and four centimeters in diameter, and each strip of metal was of 

 such a size as to expose a surface of seven square centimeters on each of 

 its sides to the liquid. The resistance of each cell as measured by the 

 use of alternating currents* was nearly 1200 ohms. Of a large number 

 of such cells at our disposal, we found 240 sufficient for our purpose, 

 and a switch-board enabled us to arrange these easily in various com- 

 binations of groups of twenty cells each. 



The condensers which we used in this part of our work were nine 

 in number, including one of a capacity of one microfarad subdi- 

 vided into fractional parts. This condenser, which was made by 

 Messrs. Elliott Brothers and marked by them No. 72, we took as a 

 standard. 



As we shall show when we come to discuss the characteristics of 

 different condensers, the ratios of the capacities of the other condensers 

 to the capacity of this one varied somewhat with the time of charging, 

 but, when this remained constant, were the same, whatever the kind of 

 battery used. 



The capacity of each of the condensers for each time of charging 

 was measured in terms of the capacity which No. 72 has when charged 

 for one second by a battery of small internal resistance. The capaci- 

 ties of five of the other condensers, as determined in the manner just 

 described, did not vary on the average by more than two per cent 

 when the time of charge was decreased from 2.0 seconds to 0.01 

 second, that is, in the ratio of 200 to 1. By combining the different 

 condensers, we were able to get within one twentieth of a microfarad 

 any capacity from to 11 mf., a range which was amply sufficient for 

 our purposes. 



Most of our work on water cells was done with the help of a four- 

 coil mirror galvanometer (A), of 3300 ohms' resistance. For some pur- 

 poses, however, only one of the coils was used. The period of the 

 complete swing of the galvanometer needles was 27 seconds, and the 

 ratio of two successive swings 1.058. Unlike some of our other gal- 

 vanometers, this one, when used ballistically, within the limits which 

 we set for our experiments, gave throws which seemed to be func- 

 tions only of the quantity of eleetricityf discharged through it. that is, 

 independent of the tension of this electricity at the instant of dis- 

 charge. A minute coil of wire, placed in the core of one of the 



* F. Kohlrausch, Wied. Ann., xi. p. 663, 1 



t See Lord Rayleigh, Phil. Trans., I't. 2, p. Gl!», 18S2. 



