1903.] On the Theory of Refraction in Gases. 25 



spectral lines of the gas, and, therefore, must have a frequency 

 independent of temperature. 



The view adopted in the present paper is that, instead of free- 

 vibrations, we^ have constrained motion. Regarded from Professor 

 J. J. Thomson's point of view, the atom consists of a large positive 

 particle and a large number of small negative ones. Instead of 

 supposing that these negative particles can vibrate radially, I regard 

 them as rolling on the surface of the positive one in constrained 

 motion. The effective control on transmitted waves is thus the 

 rotational energy of motion of the particles, and it must be pro- 

 portional to the absolute temperature. 



When, by collisions or otherwise the rotational motion becomes 

 so great that the electric attraction is overcome by the centrifugal 

 force, ionisation occurs. The frequency or frequencies of rotation 

 at which this occurs are determined by the electrical attractions and 

 are independent of temperature, although, of course, the higher the- 

 temperature the greater will be the amount of ionisation. I regard 

 these frequencies as corresponding to the spectral lines, and it will be 

 seen that the view explains the ionisation produced by ultra-violet 

 light, and also agrees with the fact that luminosity is probably always 

 connected with ionisation, e.g., the characteristic lines come out in 

 the electrical discharge through the gas. 



Regarded simply as obstacles, the molecules must contribute a term, 

 to /i*-l, which is proportional to N and practically independent of 

 the frequency. 



The final formula obtained is 



where ^ and fe are constants and f(p,0) is a function of p and 0. 

 The function is fully discussed in the paper. 



The formula is shown to be capable of accounting for all the known 

 facts connected with the dielectric constant and the refractive index 

 while the absorption of ultra-violet light and apparent absorption, due 

 to selective reflexion in the infra-red, is also explained. 

 ^ Notwithstanding the very complex and varied facts in Air, Hydrogen,. 

 Carbon Dioxide, Ammonia and Sulphur Dioxide, complete numerical 

 agreement between the measurements of K- 1 and p* - I, as regards 

 both absolute magnitude and dependence on pressure, temperature and 

 frequency, has been established. 



