466 Mr. G. W. Walker on the 



The observers already mentioned measured the amount of 

 dispersion, and in every case the refractive index increased 

 witn the frequency. They found no variation in the dis- 

 persive power with temperature ; but as the whole amount 

 of dispersion in the visible spectrum is very small and not 

 much in excess of possible experimental errors, it is doubtful 

 whether this can be taken as more than an approximate state- 

 ment of the truth. 



It appears that the presence of a spectral line only pro- 

 duces a measurable effect on the refractive index at points 

 in the immediate vicinity of it, and the experiments show 

 that dispersion is controlled by something of a general nature 

 and common to all gases. 



It is well known that all gases have intense absorption of 

 ultra-violet lio-ht*, this fact beinor intimatelv connected with 

 ionization. Again, there is evidence that several, if not all, 

 gases have absorption in the infra-red f. 



We have next to note that while for some gases the value 

 of yu,'"^ — 1 agrees with K — 1, this does not hold for all. In 

 some of the denser gases, such as SOo and XHg, the value 

 of K — 1 is much greater than the value of yu-^ — 1. I have 

 already referred to the fact that in such cases the temperature 

 coefficients are very different. 



P 



The deviation from the lawya^— 1 x -5- , as regards pressure, 



is extremely small, and has been satisfactorily explained by 

 Mascart as due to the deviation from Boyle\s law. 



(2) Demands on a Correct Theory. 



It appears that a satisfactory theory of refraction must 

 account for the following general features : — 



(A) That the refractive index very nearly obeys the 

 Gladstone and Dale Law yu- — la p, where p is the density, 

 with small variations positive or negative as regards 

 temperature. 



(B) That the refractive index increases regularly with the 

 frequency, and that the dispersion is substantially independent 

 of temperature. 



(C) That there is absorption in the ultra-violet, and 

 possibly also in the infra-red. 



(D) That in some cases the value of K — 1 may be much 

 greater than the values of fji' — l in the visible spectrum, and 

 that in such cases the law K — 1 cc pjO should be nearly true. 



* Mascart, Traite cfoptique, t. iii. p. 373. 



t Langley, Phil. Mag. ii. p. 119 (1901) ; also Rubens and Aschkinass, 

 Annalen^der Phi/sik, Ixiv. p. 584 (1898;. 



