CHAP. IX.] THE STELLAR EVIDENCE. 79 



very widely. We have a complex mixture of the vapours of metallic 

 substances and gases with paramount calcium, hydrogen, and the 

 cleveite gases. Temperature cannot -therefore vary the relative inten- 

 sities of the lines. H and K, the chief lines of calcium, must always 

 remain predominant, iron must remain because it cannot be destroyed, 

 and since the quantity of hydrogen and the cleveite gases present 

 cannot be increased, their lines cannot therefore become more impor- 

 tant in the spectrum. 



It is also clear that any change of relative density on the usual 

 hypothesis cannot be brought about by an increase of temperature ; 

 this, then, cannot alter, it cannot change the relative proportions of 

 chemical substances present in any layer, and therefore the relative 

 intensities of the lines which indicate the existence of the various 

 substances in the different layers. 



If now we turn to the other hypothesis, that, namely of dissociation, 

 we see at once, in the light of laboratory experiments, that with every 

 considerable increase of temperature in all such masses of vapour and 

 gas as those which now constitute the solar chromosphere and revers- 

 ing layer, a fundamental change in the appearance of the spectrum 

 must be brought about ; complex molecules would be broken up into 

 simpler ones, and the result of this action would bring new lines into 

 the spectrum, indicating the vibration of the molecules produced. 

 Now let us come to facts. Were the temperature of the reversing 

 layer to be increased, if dissociation takes place at this temperature, the 

 dissociation products must become visible, and we must look for them 

 among those lines which expand at the expense of those which contract 

 and disappear. Is any such experiment as this going on even at this 

 moment ? The answer is beyond question. 



The lower, hotter chromosphere differs from the reversing layer 

 precisely because this change has taken place. As I have said before, 

 we pass on descending the sun's atmosphere from the arc lines in the 

 reversing layer to the enhanced lines in the chromosphere, from the arc 

 spectrum to the " test spectrum," from the metals to the proto- 

 metals. 



What could only be pointed out with regard to only a line or two 

 20 years ago can now be proved for a whole set of lines, and the dis- 

 sociation argument is seen to be vastly strengthened the more it is 

 tested. 



Next, let us see where the stellar evidence helps us ; here I shall 

 deal with the main outlines merely. If in the sun the chromosphere is 

 hotter than the reversing layer in a star slightly hotter than the sun, 

 the reversing layer which builds up the stars' absorption should 

 resemble the chromosphere. 



