

OXYGEN IN THE SUN. g 



depth of about eight thousand miles, show only, under spec- 

 troscopic scrutiny, the bright lines indicating gaseity. But 

 though this is perfectly true, it is also true that we have not 

 here a particle of evidence to show that clouds of liquid 

 particles, and of tiny crystals, may not float over the sun's 

 surface, or even that the ruddy clouds shown by the spectro- 

 scope to shine with light indicative of gaseity may not also 

 contain liquid and crystalline particles. For in point of fact, 

 the very principle on which our recognition of the bright 

 lines depends involves the inference that matter whose light 

 would not be resolved into bright lines would not be recog- 

 nizable at all. The bright lines are seen, because by means 

 of a spectroscope we can throw them far apart, without 

 reducing their lustre, while the background of rainbow-tinted 

 spectrum has its various portions similarly thrown further 

 apart and correspondingly weakened. One may compare 

 the process (the comparison, I believe, has not hitherto been 

 employed) to the dilution of a dense liquid in which solid 

 masses have been floating : the more we increase the quan- 

 tity of the liquid in diluting it with water, the more trans- 

 parent it becomes, but the solid masses in it are not changed, 

 so that we only have to dilute the liquid sufficiently to see 

 these masses. But if there were in the interstices of the 

 solid masses particles of some substance which dissolved in 

 the water, we should not recognize the presence of this sub- 

 stance by any increase in its visibility ; for the very same 

 process which thinned the liquid would thin this soluble 

 substance in the same degree. In like manner, by dispers- 

 ing and correspondingly weakening the sun's light more and 

 more, we can recognize the light of the gaseous matter in the 

 prominences, for this is not weakened; but if the prominences 

 also contain matter in the solid or liquid form (that is, drops 

 or crystals), the spectroscopic method will not indicate the 

 presence of such matter, for the spectrum of matter of this 

 sort will be weakened by dispersion in precisely the same 

 degree that the' solar spectrum itself is weakened. 



It is easy to see how the evidence of the presence of anv 



