( 135 ) 



remain parallel and, generally speaking, the greatest incurvations 

 will be noticed in those rays, for which the medium has refractive 

 indices dilfering most from unity, i. e. in those which, in the spectrum, 

 lie closest to the absorption lines on either side. These particular 

 kinds of light, while diverging into space, will spread in many more 

 different directions than the average waves, and, as a rule, a smaller 

 portion of them will fall into the spectroscope, than of waves with 

 refi-active indices nearer to unity. 



Accordingly, there must always be certain places in the absorp- 

 tion spectrum, from which light is absent owing to dispersion in 

 the absorbing vapour, for it may be taken for granted that the latter 

 is never absolutely homogeneous. These darker parts in the spectrum 

 we shall call dispersion bands. It stands to reason that these bands 

 will overlap the regions of real absorption; so they might easily be 

 mistaken for strengthened absorption lines, which no doubt has 

 often been done. 



We will now look somewhat closer into the characteristics 

 by which dispersion bands may be distinguished from absorption 

 bands. 



The curvature of a ray of light of a definite wave-length, at any 

 point of a non-homogeneous medium, not only depends on the gradient 

 of optical density at that particular spot, but also on the angle 

 which the beam makes with the levels of equal density. Its diver- 

 gence will be greatest when this angle is zero. 



Strong ray-curving through anomalous dispersion in vapours may, 

 therefore, be artificially produced in two ways: first, by using masses 

 of absorbing vapour, presenting in a small space considerable dif- 

 ferences in density, such as e.g. occur in the electric arc ^); secondly 

 in larger spaces where the density varies but moderately, by making the 

 light travel over a considerable distance under small angles with 

 the levels of equal density. 



I have chosen the latter method of investigation, especially on 

 account of the extensive use, which may be made of the phenomena 

 presenting themselves, by applying them to the interpretation of 

 numerous peculiarities of the spectra of celestial bodies ^). 



The absorbing medium was a Bunsen flame, of a peculiar shape, 

 containing sodium vapour and so arranged, that the introduction of 

 the salt could be easily regulated. 



1) H. Ebert, Wirkung der anomalen Dispersion von Metalldampfen, Boltzmann 

 Festschrift, S. 443. 



