THE GROWTH OF OUR KNOWLEDGE OF HELIUM. 273 



" I now proceed to show that the same conclusion holds 

 good for the gases obtained by Professors Ramsay and 

 Cleve from cleveite. 



" For this purpose, as the final measures of the lines of 

 the gas as obtained from cleveite by Professors Ramsay and 

 Cleve have not yet been published, I take those given by 

 Crookes and Cleve, as observed by Thalen. 



" The most definite and striking result so far obtained is 

 that in the spectra of the minerals giving the yellow line 

 I have so far examined, I have never once seen the lines 

 recorded by Crookes and Thalen in the blue. This demon- 

 strates that the gas obtained from certain specimens of 

 cleveite by chemical methods is vastly different from that 

 obtained by my method from certain specimens of brog- 

 gerite, and since, from the point of view of the blue lines, 

 the spectrum of the gas obtained from cleveite is more 

 complex than that of broggerite, the gas itself cannot be 

 more simple. 



" Even the blue lines themselves, instead of appearing 

 en bloc, vary enormously in the sun, the appearances being 



4922 (4921-3) = thirty times 

 4713 (47 1 2-5) = twice. 



" These are not the only facts which can be adduced to 

 suggest that the gas from cleveite is as complex as that 

 from broggerite." 



It is seen that quite early in the inquiry we had not only 

 spectroscopic evidence in the laboratory which was com- 

 plete in itself, but that the case was greatly strengthened 

 when the behaviour of the various lines in the sun and stars 

 was also brought into evidence. 



In the first case we had the laboratory separation of D, 

 from the lines 5048, 5016, and 4922. 



Later on in the same month I showed that the lines at 

 D 3 and 447 behaved in one way, and that at 667 behaved 

 in another. 



In order to test this view I made some observations 

 based on the following considerations : — 



(1) In a simple gas like hydrogen, when the tension of 

 the electric current given by an induction coil is increased 



