232 Messrs. Rosa and Smith on a Calorimetric 



constant by a preliminary run of several hours. The tem- 

 perature of the calorimeter as indicated by the thermometer 

 A' rose gradually for an hour, and hence the heat absorbed 

 was less than the average. During the last hour the tempe- 

 rature was reduced by quickening the rate of flow and cooling 

 the entering water, so that the temperature was substantially 

 the same at the end as at the beginning. The per cent, of 

 loss is 4*84 at an average temperature of the condensers of 

 30° C. The voltage employed on this low-frequency test was 

 much higher than for any other experiment, and yet there 

 was no evidence of brush-discharge or appreciable leakage- 

 current. 



In the third, experiment the same condensers were subjected 

 to a high-frequency electromotive force at 30° C, and the loss 

 found to be 5*06 per cent., that is, slightly greater than before. 

 Hence for a given voltage the energy dissipated, per period 

 would be slightly greater, and the energy dissipated per second 

 more than five times as much as for the low frequency. 



In the fourth experiment only the " best pair " of con- 

 densers was used, and with a slightly higher voltage the 

 temperature of the condensers rose to 40° C. Here the los3 

 was found to be 8*37 per cent., nearly as much as the maximum 

 value found by the resonance method. 



The fifth experiment, with the " best pair," was made 

 some days later at 30° C, and the percentage loss came out 

 6*35 per cent., that is greater than the average of the six. This 

 was unexpected, as well as the last result, which showed a loss 

 for condensers 3 and 4, the " poorest pair," of 4*82 per cent., 

 which was less than the average. These results were then 

 confirmed by an independent method, showing conclusively 

 that the so-called " poorest pair " had the smallest loss ; not, 

 of course, because it emitted a distinct sound and hissed on 

 high voltages, but in spite of that. The chief loss is doubtless 

 due to some cause quite independent of the singing and 

 hissing, and happens to be smaller where it would naturally 

 be expected to be larger. 



Thus we have confirmed by these calorimetric measure- 

 ments the large values of the losses which we found by the 

 resonance method in beeswax and rosin condensers, and also 

 the existence of a well-marked maximum as the temperature 

 rises, beyond which the loss decreases considerably. It is an 

 interesting fact that the residual charges of these condensers 

 are very large, that they increase with the temperature up to 

 40° C, and then decrease as the temperature is carried 

 higher. That is, the maximum point for the residual charge is 

 the same as for the energy loss. 



