On the Chemistry of the Hottest Stars. 



207 



13. The lines of the cleveite gases show a similar inversion on the 

 downward side of the teniperatnre cnrve. Strongly represented in 

 the hottest stars, they thin ont very rapidly in cooling stars, and dis- 

 appear before the arc lines have begun to show themselves. 



14. Utilising the iron lines as a method of bringing together stars 

 of approximately equal temperature, it is found that at each stage 

 the stars are divisible into two groups, which, in accordance with my 

 previous work, correspond to increasing and decreasing temperatures 

 respectively. 



15. As determined in this way, stars of increasing temperature 

 differ from those of decreasing temperature at the same stage of heat : 



(1) in the greater continuous absorption in the violet or ultra-violet, 



(2) in the generally greater intensity and breadth of the metallic 

 lines, (3) in the smaller thickness of the hydrogen lines, (4) in the 

 greater thickness of the helium lines at those stages in which they 

 are visible. 



16. These differences are all explained on the meteoritic hypothesis. 



17. There are stars, near and at the top of the curve, which cannot 

 be arranged in order of temperature by the criteria referred to in 15, 

 for the reason that the iron lines have disappeared, and the lines of 

 hydrogen and cleveite gases show little variation. 



18. The arrangement of stars about the top of the curve will 

 depend upon the conditioning of certain lines, at present of unknown 

 origin; the necessary criteria, therefore, require further investiga- 

 tion. 



19. The known facts with regard to changes in the line spectrum 

 of an element can be easily explained on the hypothesis of successive 

 dissociations analogous to those observed in the case of undoubted 

 compounds. 



20. Similarly, the differences in the lines representative of a metal 

 such as iron in the spectra of sun-spots, prominences, chromosphere, 

 or different stars, are explained by supposing that there are different 

 molecular groupings at each stage of temperature. 



21. The change from a continuous spectrum to one consisting of 

 nutings, and afterwards to one of lines, is now acknowledged to be 

 due to the existence of different molecular combinations. 



22. The recent investigations of Humphreys and Mohler on the 

 shifts produced in metallic lines, when the vapours are observed at 

 different pressures, confirm my view that the line spectrum of a metal 

 integrates for us the vibrations of several sets of molecules. 



23. It is argued that the existence of " series " of lines in the 

 spectra of some chemical elements is another indication of molecular 

 complexity, each series probably representing the vibrations of similar 

 molecules. 



24. The behaviour of the magnesium lines in stellar spectra is 



