the Thermal Unit. ^3/ 



Tlie conclusion arrived at was as follows : — " That over the 

 temperature range 15° to 26° (nitrogen scale), the capacity 

 for heat of water, in terms of its capacity at 15°, is 

 (approximately) 



1- -000266 (0-15)." 



The function, however, is not absolutely linear, as the rate 

 of decrease diminishes slightly above 22°. During the reading 

 of a communication to the Royal Society in November* last, 

 Professor Schuster stated that although the temperature range 

 covered by the experiments by Mr. Gannon and himself was 

 small, the change over that range was similar to that resulting 

 from my experiments. 



Plate VI. shows the discrepancy, over the range of tem- 

 perature 10° to 30°, between the results of those observers 

 whose work appears to be most worthy of attention. I have 

 added Regnault's values, for although his conclusions with 

 regard to the changes at this range are undoubtedly erroneous, 

 they are so generally accepted as to render their inclusion 

 important. The results of each observer are expressed in 

 terms of his own value at 15° as unity. A study of the 

 diagram will show that if the results of thermal experiments 

 at 20° were expressed in terms of the "thermal unit at 10°" 

 by means (a) of Regnault's formula, (h) of Rowland's curve, 

 the resulting differences would be as great as 1 in 300 ! — an 

 error of serious dimensions when we consider the smallness 

 of the temperature range. 



The range 10° to 30° is of particular importance, as the 

 majority of thermal determinations are expressed in terms of 

 the capacity of water at some point within this range, and as 

 a consequence great attention has been devoted to it. The 

 diagram shows that, nevertheless, the results are by no means 

 satisfactory. 



The difficulty is one not so much of experiment as of the 

 measurement of temperature, and I would repeat that any 

 one of the results which shows a decrease, when expressed in 

 the nitrogen, would show an increase if expressed in the 

 mercury scale. Before, therefore, any satisfactory conclusion 

 as to the behaviour of water can be arrived at, it is necessary 

 to decide upon and establish some standard temperature scale. 

 Further refinements in experiment are useless until a 

 general agreement as to temperature measurements has been 

 established. 



As an illustration, I may mention that in July I wrote to 

 ask Dr. Guillaume if he would favour me with his opinion on 



* Proc. Roy. Soc, November 1894. 



