408 



PROCEEDINGS OF THE AMERICAN ACADEMY 



phide can bo represented for so great a range of temperature by a linear 

 function. The same remark of course applies to the other liquids ex- 

 amined, thougli in a less degree. With respect to the determination of 

 / by observation, I may here remark that, although the interference 

 bands between any two spectral lines may be directly counted, so that 

 the spectroscope alone is available, it will still be better, whenever 

 possible, to measure the two indices directly, and then compute Xwith 

 the assistance of an observed density. It will be seen that t is for each 

 substance a linear function of the temperature. 



For laboratory purposes, the interferential constant will, I hope, like 

 the density, boiling-point, specific volume, &c., serve as a means of 

 recognizing the purity of a given compound. I shall now endeavor to 

 show that it may also find application in quantitative analysis. 



The values of / given above are suiBcient to prove that the inter- 

 ferential constant of a mixture is the sum of the interferential con- 

 stants of the component parts. If P be the weight of any mixture, 

 jOj and jOj the relative weiglits of its components, / the constant for the 

 mixture, I^ and I.^ the constants for the components, we shall have 



Table III. contains the values of P I as obtained directly from the 

 observed valuer of P and /, and also as computed by adding tlie values 

 of Pj^ I^ and jOj I.2 in the cases of all the mixtures cited in Tables I. 

 and II. 



