ON THE snriKIC IIKAT OP VVATKR. 227 



Commencement of this experiment, its effect would only be to increase each of the 

 temperature intervals by 0'0025 C., which would be insensible. The first 40 C. 

 point is read on the second range thermometer. The third range thermometer was 

 then substituted for the first and the 40 C. to 55 C. interval determined by this 

 third thermometer. At 55 C. the fourth range thermometer was substituted for the 

 second. In each case the error due to a possible zero depression, which had not 

 entirely disappeared at the time of the experiment, would not produce an error of 

 greater magnitude than 1 in 4000 on the interval. 



At the close of the experiment the constants of the resistances MiM, were obtained 

 in the manner previously described, the data being shown on the notes (Appendix D). 



The allowance for heating by obturator and stirring is obtained from the curve 

 described in Section 7. It amounts in the middle range of temperatures to 0'21 C. 

 rise per hour. The correction for this, in the point-to-point experiments, is made by 

 deducting this rise from the temperature interval. Thus in Experiment 125 (see 

 Summary, Appendix D) there is a rise in the second period of 14*235 in 13 minutes. 



13 x 0'21 

 Of this - = 0'045 C. is due to obturator heating, stirring, &c. This is 



deducted from the total temperature interval to give the interval A$, which is the 

 rise of temperature due to electrical heating. 



With regard to small defects or excess of temperature of bath or obturator during 

 a run, the effort of the observer was to make these balance as far as possible. With 

 practice it was found fairly easy to regulate the rise of temperature in the bath, so 

 that no correction for temperature difference between the inside and outside of the 

 calorimeter was required. But it was deemed advisable to ascertain the effect of any 

 such difference, and an experiment (No. 51) was made for this purpose. It was 

 found that, with about 2950 gr. of water in the calorimeter, if $ represents the 

 difference of temperature between the bath and the contents of the calorimeter, the 

 change in temperature of the contents of the calorimeter was 



0-00322^ 



degree per minute. This amounts to a heating (or cooling) effect of 0'66($ watt. 

 The electrical heating employed being of the order of 230 watts, an excess of 

 temperature of one-tenth of a degree maintained throughout the whole of any 

 interval would have involved a correction of the order of 1 in 3000 on the result. It 

 was not, however, difficult to regulate the temperature so as to make excess and 

 defect of temperature substantially balance, and experiments where the regulation 

 went wrong were not corrected, but rejected. The only exception to this was in two 

 or three cases where towards the end of a run there was for the last few minutes a 

 drop in bath temperature owing to the gas pressure being insufficient. In such cases 

 the temperature difference could be represented as $ = at, and the necessary correction 

 could be obtained from the formula A0 = 0'001Gla a . Two or three corrections of the 

 order of 0'01 C. were made in this way. 



2 o 2 



