480 Mr. a, W. Walker on the 



agreement with the experimental fact that ~ varies hv 



P . . '^ 



only a small amount when the temperature is increased 



from 0° to 100° 0. 



We must thus conclude that will not alter by any 



measurable amount as the temperature changes within 

 ordinary limits. 



As an approximation, wdien ^^ is large we get 



and this is independent o£ temperature dnce (d^ cc 6. 



If, however, we were very near the point B the approxi- 

 mation would not be valid, but I doubt whether the present 

 method of measuring dispersion in gases is sufficiently accurate 

 to reveal the variation with temperature. 



We have seen that values of fju greater than unity are re- 

 quired in the visible spectrum, but we may have values of yu, 

 less than unity in the infra-red. This must occur when 



k . 



J-—- is large, that is in cases where the value of K — l is 

 kit/ 



much greater than the values of fJi^ — 1 in the visible 

 spectrum. This has an important bearing on Langley's ex- 

 periments on the infra-red spectrum through the atmosphere. 

 For suppose the sun to be low^ on the horizon but still visible, 

 then if there is any constituent of the atmosphere for W'hich 

 K—l is much greater thnn ^Ji^ — l, infra-red rays in a certain 

 range of the spectrum will be bent upwards and not down- 

 wards. They will suffer total reflexion at some point above 

 the surface of the earth, and thus apparent absorption will 

 occur. Water-vapour is one of those gases for which K — l 

 is much greater than fi^ — 1, and it is always present in the 

 atmosphere. The locality of the absorption in the infra-red 

 must depend to a large extent on temperature. 



The theory which has been proposed is thus capable of 

 explaining qualitatively all the main features wdth which w^e 

 are acquainted. It now remains to test whether the know^n 

 facts can be explained quantitatively. 



