16 Prof. S. U. Pickering. Determinations to test [Dec. 11, 



that at the initial and final temperatures ; it was generally very small, 

 since the temperature of the air was kept at a point such that there 

 was heating at the initial temperature, and cooling at the final tem- 

 perature, or vice versd. The time occupied in obtaining almost 

 identical temperatures in the bottle and calorimeter varied between 

 two and twenty minutes, the whole determinations, including the 

 interval allowed for determining the two rates of cooling, occupying 

 fifteen to forty-five minutes. Such a duration militates very much 

 against the accuracy of the results. 



The calorimetric thermometer read by estimation (0'05 mm.) to 

 0'0005 ; the stirring apparatus and other appliances were the same as 

 those described elsewhere (' Chem. Soc. Trans.,' 1887, p. 293). The 

 water equivalent of the platinum bottle and its thermometer was 

 ascertained by direct experiment to be 2*223 grams. The last men- 

 tioned thermometer possessed a range of 30, one estimation figure 

 being equivalent to 0'01, and, as the rise or fall measured sometimes 

 exceeded 30, it was in such cases set so as to register the initial 

 temperatures of the bottle, the final temperature of this being taken 

 to be the same as that of the calorimetric water, previous determina- 

 tions having shown that the two temperatures were identical within 

 the reading error of the instruments when this rate of cooling 

 became constant. The temperature of the calorimeter was generally 

 about 18. 



The determinations were all made in duplicate. The mean error 

 of a single observation was found to be about 0'8 per cent, of the total 

 rise or fall measured ; this corresponds to an error of 0'0075 in the 

 alteration of temperature registered in the calorimeter, or 0> 22 in 

 that registered in the bottle; considering the long duration of the 

 determinations and the magnitude of the total correction for cooling 

 which had to be applied, such an error must, I think, be regarded as 

 small. In many cases the error in the heat capacity found is the 

 same as that in the rise or fall measured, i.e., 0'8 per cent., of its 

 value, or, on the average, 0'0032 of the heat capacity per gram ; in 

 other cases it is much greater, for the heat evolved sometimes in- 

 cluded the heat of fusion, and, after subtracting this, the whole error 

 remained concentrated in the smaller quantity, which represented the 

 beat capacity. In some cases, again, the heat capacity for a given 

 interval had to be found by taking the difference between two different 

 determinations, and in such cases the error was greater. 



Results Obtained. 



The experimental results are collected in Tables I to X, pp. 23 32. 

 In these w is the weight of substance taken, r the rise or fall mea- 

 sured in the calorimeter, t and t' the initial and final temperatures of 



