312 THE ROYAL SOCIETY OF CANADA 



(2) Saha in his theory takes no account of the relative abundance 

 of different elements. For a spectrum to show, a certain number of 

 atoms must be in a disturbed condition. Now, since at a given 

 temperature and a given ionization potential a certain fraction of the 

 atoms of an element are ionized, it is evident that of two elements of 

 like ionization potential the more abundant will be the first to appear 

 and the last to disappear. The neglect of this factor of relative 

 abundance of elements accounts for Saha's prediction that the Balmer 

 series should disappear in 0-type stars (effective temperature 22,000°K) 

 just slightly before ili"g+4481. The fact observed here that the 

 Balmer series persists long after this stage is doubtless due to the 

 fact that there are more H atoms than Mg in the stellar atmosphere. 

 The analyses of Clark show that in the first 10 miles of the earth's 

 crust there are about 12 times as many H as M g atoms. Not only, 

 however, must the relative abundance of elements be considered but 

 also the possibility that the relative amounts are subject to change in 

 the course of stellar history. We would naturally anticipate as the 

 star grew hotter that certain of the more complex atomic nuclei would 

 become unstable and break down into H and He nuclei. The fact 

 that in an 0-type star at 25,000°K 4686 is a broad (10 A.U.) emission 

 line shows from Merton's experiments that He is probably much more 

 abundant relatively than on the earth. The absence of unknown 

 lines in such stars, whereas we would have anticipated many 9 N lines 

 of different elements, also points to a possible change in relative abun- 

 dance. It also seems probable that it is this gradual change in the 

 relative abundance of elements that accounts for the spectral changes 

 in giant and dwarf stars of the same temperature which have led to 

 the determination of spectroscopic parallaxes. 



In conclusion it is evident that if Saha's hypothesis is to be used 

 to determine stellar temperatures, it will have to be modified to meet 

 such criticisms as these. 



