( 358 ) 



percentage of sodium were larger, the "sodium chromosphere" would 

 appear higher. 



It is customary to draw conclusions from the size of the chromo- 

 spheric and flash crescents, observed during a total eclipse with the 

 prismatic camera, as to the height to which various vapours occur 

 in the solar atmosphere. According to us this is an unjustified con- 

 clusion. On the other hand it will be possible to derive from these 

 observations data concerning the ratio in ivhich these substances are 

 presetit in the gaseous mixture, provided that the dispersion curves 

 of the metallic vapours, at known densities, will first have been 

 investigated in the laboratory. 



Until now we only dealt with the normal radial density gradient. 

 By convection and vortex motion however irregularities in the density 

 distribution arise, with gradients of various direction and magnitude. 

 And since on the sun the resultant of gravitation and radiation 

 pressure is relatively small, there the irregular density gradients may 

 sooner than on the earth reach values that approach the radial 

 gradient or are occasionally larger. 



The incurvation of the rays in these irregularities must produce 

 capriciously shaped sodium prominences, the size of which depends, 

 among other causes, on the percentage of sodium vapour in the 

 gaseous mixture. 



So the large hydrogen and calcium prominences prove that rela- 

 tively much hydrogen and calcium vapour is present in the outer 

 parts of the sun ; but perhaps even an amount of a few percents 

 would already suffice to account for the phenomena ^). 



If we justly supposed that non-radially directed density gradients 

 are of frequent occurrence in the sun, and there disturb the general 

 radial gradient much more than on the earth, then not only rays 

 from the marginal region but also rays from the other parts of the 

 solar disc must sensibly deviate from the straight line. Chiefly con- 

 cerned are of course the rays that undergo anomalous dispersion. 

 Every absorption line of the solar spectrum must consequently be 

 enveloped in a dispersion band. 



To be sure, absorption lines of elements which in the gaseous 

 mixture only occur in a highly rarefied condition, present themselves 

 as almost sharp lines, since for these substances all density gradients 

 are much smaller than for the chief constituents, and so the curvature 

 of the rays from the vicinity of these lines becomes imperceptible. 



1) This result would be in accordance with a hypothesis of Schmidt (Phys. 

 Zeilschr. 4, S. 232 and 341) according to which the chief constituent of the solar 

 atmosphere would be a very light, until now unknown gas. 



