204 MESSRS. W. R. BOUSFIELD AND W. ERIC BOUSFIELD 



In order to determine the variations in the value of J from point to point it was 

 necessary to know what, if any, variation of capacity there might be in the calo- 

 rimeter. For this purpose we made a separate series of experiments on the specific 

 heat of glass. They showed a considerable rise in the specific heat of glass in the 

 range from C. to 80 C., the expression for the mean specific heat of glass from 



to 6 being 



S* = 0'1949 + 0-000090. 



From this we were able to deduce the mean heat capacity of the a calorimeter at 

 various parts of the scale, which was requisite for the next experiments. 



The fifth set of experiments was designed to determine the variation of the specific 

 heat of water from point to point between C. and 80 C. 



It has been already observed that the apparatus was primarily designed so as to be 

 specially convenient for making, with reasonable rapidity, a large number of observa- 

 tions on salt solutions. This involved the use of the open-mouthed Dewar vessel 

 loosely closed by an obturator. From 80 C. to 100 C. the rise in the vapour 

 pressure brings in corrections of too great magnitude for work of the order of 

 accuracy required, and hence our experiments have been limited to the range from 

 C. to 80 C. 



In these experiments a known weight of water was placed in the calorimeter and 

 heated from C. to 80 C. with various breaks. It has been already observed that 

 any temperature interval, small or great, may have a possible error of 0'01 C. in 

 its estimation. For this reason we took a 40 C. interval for our fundamental 

 experiment, involving a possible error in any individual result of 1 part in 4000. 

 With shorter intervals of only 12 or 15 degrees, we now have possible errors due to 

 this cause of 1 part in 1200, but, by taking the average of repeated observations, the 

 probable error due to this cause is again reduced to errors of the order of 1 part in 

 These experiments were conducted in the same manner as the third series 

 of observations above described. The results are as follows : 



Temperature interval. Mean value 



C. of J. 



Otol3 4-1937. 



13 27 4-1752. 



27 40 .... 4-1756. 



40 55 4'1935. 



55 73 4-2024. 



55 .. 80 4-2056. 



No value is to be attached to the fourth place of decimals, except for the purpose of 

 the value in the third place. But it may be noted that whilst the result of the 

 third series of experiments was to give 



Ji3 M = 4-1821, 



