I;I:KI;.\( nvi. IM>HT.S OK TIIK HI. I:\IKNTS 



multiplied by the ratio of the atomic weights of the element and H (taken as 1'007). 

 Hence, the results are comparable to the well-known figures for the permanent gases, 

 whose molecules are diatomic. 



If N be the number of bands observed to pass the pointer as the whole of the 

 substance evaporates, X the wave length of light employed, X the length of the tube, 

 m the refractive index observed and n the refractive index at the standard density, 

 we have NX = (m 1) X and 



/t 1 _ standard density 

 ml oliserved density 



0000899 x atomic weight of element volume of tube 

 atomic weight of hydrogen weight volatilized' 



Hence 



_ NX '0000899 x atomic weight of element x volume of tube 

 X atomic weight of hydrogen x weight volatilized 



The legitimacy of this method of comparison rests on the assumptions that the 

 refractivity is proportional to the density, and is independent of the temperature 

 if the density remains constant. The former is the conclusion accepted by 

 M. MASCART, and, I believe, all other observers. With regard to the latter, opinion 

 is divided, but, as a first approximation, and to the degree of accuracy sought in 

 these observations, it is almost certainly true.* So far as my experience goes it 

 confirms this belief. When the substance had completely evaporated, no subsequent 

 rise of temperature, even by as much as 20 C., had any effect on the position of the 

 bands. 



Sources of Error. 



The principal source of error detected in the refractometer method is that which 

 may be called the starting effect. It frequently happened that, when the heating 

 began, the bands shifted by an amount which generally varied between ^ and 1 band. 

 In a few minutes the displacement seemed to reach a limit and then, after a pause, 

 short or long in proportion to the volatility of the substance, the true procession of 

 bands began, f Sometimes the shift took place in the direction in which the bauds 

 were about to move, sometimes in the opposite direction. Similarly, when the 

 temperature had reached its highest point and the heat was cut off, the bands some- 

 times shifted slightly before the main fall began. 



This effect is probably due to unequal heating of the tubes, or of the columns of air 

 outside the tubes through which the rays respectively passed. The error could be to 



* See G. W. WALKER, ' Phil. Trans.,' A, vol. 201, p. 435, 190.3, " On the Dependence of the Refractive 

 Index of Gases on Temperature." 



t By a series of blank experiments it was shown that the effect did not increase as the temperature rose. 



2 C 2 



