86 



PROCEEDINGS OF THE AMERICAN ACADEMY 



Regnault's results are embodied, for the most part, in tables given on 

 p. 239 of the first volume of his Relation des Experiences. The 

 figures given there are obtained from curves drawn to represent the 

 mean of his experiments, and do not contain any theoretical results. 

 The direct application of my formula to his experiments could hardly 

 be made without immense labor in finding the most probable value of 

 the constants. 



But the following seem to satisfy the experiments quite well : — 



Cristal de Choisy-le-Roi 5 = 0, a = .000 000 32. 



Verre Ordinaire b = 245°, a = .000 000 34. 



Verre Vert b = 270°, a = .000 000 095. 



Verre de Suede b = +10°, a = .000 000 14. 



From these values I have calculated the following : — 



TABLE II. — Regnault's Results compared with the Formula. 



The formula, as we see from the table, represents all Regnault's 

 curves with great accuracy, and if we turn to his experimental results 

 we shall find that the deviation is far within the limits of the experi- 

 mental errors. The greatest deviation happens at 340°, and may be 

 accounted for by an error in drawing the curve, as there are few ex- 

 perimental results so high as this, and the formula seems to agree 

 with them almost as well as Regnault's own curve. 



The object of comparing the formula with Regnault's results at 

 temperatures so much higher than I need, is simply to test the formula 

 through as great a range of temperatures, and for as many kinds of 



* Corrected from 280.52 in Regnault's table. 



