134 SCIENCE PROGRESS. 



must either assume the value of a temperature coefficient 

 of the specific heat of water obtained by one observer 

 before we can express the results of other experimenters 

 in terms of the same unit, or we must give the different 

 results in terms of different units according to the mean 

 temperature of the range over which the observations were 

 conducted, in which case it becomes almost impossible to 

 estimate how close is the agreement. This difficulty is 

 a very serious one, and to me it appears to be useless to 

 further increase the accuracy of the methods employed for 

 determining the value of J until this fundamental inquiry is 

 completed. The observations of Rowland indicate a mini- 

 mum value for the capacity for heat of water about 33° C. 

 My own observations have not extended beyond 26 C, so 

 I have deduced from his tables (for the purposes of com- 

 parison) the expression for the capacity for heat of water 

 over the range 14° to 26 C. which can be approximately 

 represented in the form — 



1 —-000400 (0—15). 

 My own result over the same range gives — 



1 —-000266 (#—15). 

 I regret that I have been unable to procure the full 

 work of Professors Bartoli and Stracciati on this subject, 

 and I have only this week succeeded in obtaining any 

 information beyond that conveyed by a note in Nature, 

 July 27, 1893. A copy of the Journal de Physique for 

 December, 1893, has, however, just come to hand, and it 

 contains a full table of their experimental results together 

 with a short description of the methods employed. This 

 summary does not give the numbers resulting from the 

 repetition of experiments under similar conditions, and it 

 is therefore difficult to estimate the probable limit of 

 experimental errors. 



The table contains eleven columns, of which II. to VII. 

 give the specific heat of water for every degree from o° 

 to 31 C. deduced from the results obtained by the im- 

 mersion in water of balls of different metals at the tem- 

 perature of steam, the mean result of these experiments 

 being given in Col. VIII. The numbers in Col. IX. were 



