< 267 ) 
may be thoroughly mixed, but the average composition and density 
of the mixture will vary along the radius. In this paper we shall 
only deal with the influence which the selectively absorbing mixture of 
gases , constituting the solar atmosphere , exerts on the ivhite light , emitted 
by the lower layers of the sun . Perhaps there exist, local accumu¬ 
lations of separate gases : we leave that possibility out of consider¬ 
ation now ; nor shall we notice the fact, that selective radiation by 
the solar atmosphere may in a perceptible degree contribute to the 
snn-light observed. 
Besides in the direction of the radius the density is sure to vary 
also in other directions, and that in an irregular way, on account of 
the movement of the gases. The point in question is: can we reason¬ 
ably admit the existence of density gradients sufficiently great to 
produce any considerable refraction effects ? 
If irregular density gradients comparable in magnitude with the 
vertical gradient of our terrestrial atmosphere near the surface 
^--=16X10— 10 ^, occur in the sun, the distribution of the light 
over the solar disk must already be quite perceptibly influenced 
thereby. Indeed, in gases we have the approximate relation *) 
the gradient — , the refraction constant or specific refracting 
power R = ^—^, and the radius of curvature 9 of rays perpendi¬ 
cular to the gradient. Let us suppose the average value of R in 
the mixture not to differ too much from 0,5, then our equation gives 
9 = 1.25 X 10® c.m.; which means, that the radius of curvature of 
rays, propagated at right -angles with the supposed gradient, is less 
than one fiftieth part of the radius of the photosphere (r=r7X10 l *e.M.). 
Even within a relatively thin chromospheric layer the light coming 
from lower, levels may therefore alter its direction very sensibly. 
This inference holds a fortiori for layers, in which the pressure 
is several atmospheres, because there the gradients too are likely to 
exceed the one assumed. 
And the importance of ray-curving increases greatly again with 
waves belonging to the nearest vicinity of absorption lines. Such 
waves suffer anomalous dispersion. With them the refraction constant 
°f the absorbing gas may be several hundred- times greater than 
v ) Proceedings Roy. Acad. Amst. IX, p.352; Astroph. Journ. 25, p. 107 (1907). 
