182 Mr. H. E. Schmitz. On the Determination of [Feb. 2, 



fcy the line representing the results of Experiments 8 to 14. This 

 would be explained by supposing that in Experiments 1 to 7, or in 

 some of them, some of the liquid air had not been got rid of by 

 shaking and entered the calorimeter. From later work it appeared 

 that anomalous results, such as those of Experiments 3 and 4, were 

 probably due more to a chance variation in the amount of liquid air 

 carried over than to other errors of experiment. 



In Table II the numbers of Table I are reduced according to the 

 following plan: Experiments 8 to 14 are represented by the 

 equation y = c- mx, where y is the apparent specific heat for time 

 of transference x. The method of least squares gives c- 0-0801 35, 

 7H, = 0-0000671. Table II gives the values of y for Experiments 1 to 

 7 as well as for Experiments 8 to 14, and in the last column the 

 differences between these numbers and the numbers in Table I. 



The mean of Experiments 1 to 5 (Table I) is 0-0800, which differs 

 but little from the value of c above, and is adopted by the author 

 as representing the specific heat of copper, the principle being that 

 the small amount of liquid air, not got rid of by shaking during a 

 short transference, approximately compensates for the heating effect 

 of the atmosphere during the transference. Results for other metals 

 were obtained on the same principle. As a standard shape for the 

 specimens, a short cylinder with approximately hemispherical ends 

 was adopted where possible. The number in brackets after each 

 result for the specific heat of a given metal indicates the time of 

 transference in seconds. This time was generally 3 or 5 seconds ; 

 it was in general found that any difference between a result for 

 3 seconds and a result with the same piece of metal for 5 seconds was 

 covered by the experimental errors. Five seconds was found to be a 

 more convenient time than three, allowing for a more deliberate 

 treatment and greater uniformity of shaking. 



The mercury thermometer was read to 0-001, the temperature 

 observations being made in accordance with the following scheme : 



Reading of platinum thermometer. 



Period of 2J minutes, with readings of mercury thermometer at end 

 of every half -minute. 



