Mr. W. Sutherland on Molecular Refraction. 151 



tures of 10° and 20°, and in the gaseous state at 100° with 

 the utmost accuracy. If, then, in the formula 



(,t-l)_/Za(N-l) 



d 



\' 



-\-h)+cd 



we determine c and the constant bracketed term from the 

 data for Kquid at 10° and 20°, the value of the bracketed term 



n— 1 



ought to be found to be the same as the value of , found 



by Lorenz for the gas, seeing that d is so small that the term 

 cd may be considered to vanish in the case of a gas. When 

 it is remembered how small the difference in density is for 

 10° difference in temperature with liquids, it will be seen how 

 strict the test is. 



In the follo\\ang table are given the values oi (n — \)/d as 

 calculated by Landolt from Lorenz's data (^Ann. der Chem. 

 213) for the different liquids at 10°, 20°, and the vapours at 

 100°. Water and ethyl alcohol are not included on account 

 of the exceptional character of their molecules ; the fifth 

 column contains the values for the vapours calculated accord- 

 ing to the above equation : — 



Values of (n — l)/d. 



Substance. 



Liquid. 



Vapour. 



c. 



Obse 

 10°. 



rved. 

 20°. 



Observed. 

 100° 



Calculated. 

 100°. 



Ethyl oxide 



•4935 

 •4174 

 •2663 

 •3000 

 ■4977 



•4930 

 •4172 



•2658 

 •2996 

 •4970 



•4599 

 •4024 

 •2356 

 •2694 

 •4348 



•461 

 •407 

 •223 

 •265 

 •435 



•044 

 ■017 

 •022 

 •023 

 •050 



Ethyl acetate 



Ethyl iodide 



Chloroform 



Carbonic disulphide 



The agreement between observed and calculated values in 

 the above table in all cases except that of ethyl iodide is good, 

 especially if the severity of the method of testing is taken 

 into account. As regards the applicability of the formula to 

 gases under varying pressure we have the experiments of 

 Chappuis and Riviere {Comptes Rendus, ciii.). They find 



