(356) 



bodies really be such, that the rajs are sufficiently curved to exert 

 such a distinct influence on the distribution of' light in the spectrum? 



In former communications ^) I showed that the sun e.g. may be 

 conceived as a gaseous body, the constituents of which are intima- 

 tely mixed, since all luminous phenomena giving the impression as 

 if the substances occurring in the sun were separated, may be 

 brought about in such a gaseous mixture by anomalous dispersion. 

 We will now try to prove that not only this may be the case, but 

 that it must be so on account of the most likely distribution of 

 density. 



Let us put the density of our atmosphere at the surface of the 



1 



earth at 0.001293. At a height of 1050 cms. it is smaller by — — of 



this amount, so that the vertical density gradient is 



0.001293 



= 16 X 10-10. 



1050 X 760 



The horizontal gradients occurring in the vicinity of depressions 



1 



are much smaller; even during storms they are only about—— 



of the said value ^). Over small distances the density gradient in the 

 atmosphere may of course occasionally be larger, through local heating 

 or other causes. 



Similar considerations applied to the sun, mutatis mutandis, cannot 

 lead however to a reliable estimate of the density gradients there 

 occurring. A principal reason why this is for the present impossible 

 is found in our inadequate knowledge of the magnitude of the 

 influence, exerted by radiation pressure on the distribution of matter 

 in the sun. If there were no radiation pressure, we might presuppose, 

 as is always done, that at the level of the photosphere gravitation is 

 28 times as great as on the earth ; but it is counteracted by radiation 

 pressure to a degree, dependent on the size of the particles ; for some 

 particles it may even be entirely abolished. The radial density gra- 

 dient must, therefore, in any case be much smaller than one might 

 be inclined to calculate on the basis of gravitational action only. 



Fortunately we possess another means for determining the radial 

 density gradient in the photoshere, at any rate as far as the order 

 of magnitude is concerned. According to Schmidt's theory the photo- 

 sphere is nothing but a critical sphere the radius of which is equal 



') Proc. Roy. Academy Amsterdam, II, p. 575; IV, p. 195; V, p. 162, 589 and 

 662; VI, p. 270; VIII, p. 134, 140 and 323. 



2) Arrhenius. Lehrbuch der kosmischen Physik, S. 676. 



