( 144 ) 

 Our theory can dispense with such additional hypotheses. 



Another characteristic peculiarity, observed in every series of 

 photographs taken at short intervals with the slit set at various 

 points on the broad H and K bands, is the following. When the 

 slit is set, e.g., at a remote point of K,, the structure of the solar 

 image appears relatively fine, sharp and detailed ; approaching the 

 central line, we see some of the brilliant spots ^'anish, others grow 

 more extensive, especially those lying in the vicinity of sun-spots; 

 at the same time their outlines become less sharp, so that finally 

 the whole image gives us the impression of a coarser and at the 

 same time a more woolly structure ^). 



Hale and Eli,erman hold that the successive photographs refer to 

 gradually higher levels and conclude that the masses of calcium 

 vapour must have a tree-like shape. W, S. Lockyer, in Natur 

 No. 1800, draws a scheme showing this conception. 



Against this interpretation we propose the following one. 



The amount by which the divergence of a beam of light is altered 

 in consequence of the presence of calcium vapour in the streaming 

 and whirling mass depends, of course, on the proportion of calcium 

 in the mixture, and besides on two other circumstances, viz. 1^"^ on 

 the position occupied in the spectrum by the selected kind of light 

 with regard to the absorption lines, and 2°'^ on the steepness of the 

 density gradients in the mixture along directions perpendicular to the 

 path of the beam. 



Let us suppose the selected light to correspond to the extreme 

 edge of Hi or Kj, then its index of refraction differs but little from 

 unity. Accordingly, very considerable inecpialities of densitj' are required 

 to cause a perceptible change in the divergence of such beams. 

 Similar great inequalities may indeed occur at many separate places, 

 but at each of them they cannot, of course, extend very widely. 

 This accounts for the fine and rather sharply defined reticulation 

 shown by the so called "low-level" photographs. 



If the second slit were set a little nearer to the centre of the line, 

 the distribution of the light in the solar image would at all events 

 differ considerably from that of the former case; for the indices of 

 refraction being very different for neighbouring waves within a dis- 

 persion band, the divergence of beams, starting from the same point 

 of the Sun, must vary largely with the wave-length. So it is clear 



1) Such series of photographs are reproduced in: Publications of the Yerkes 

 Obervatory, Vol. Ill, Part I, PI. V, VI, X, XI, XII, XIII. 



