( 183 ) 



To estimate the percentages of hydrogen the spectra of the just- 

 mentioned quantities were compared with the spectrum of a helium 

 which could not contain much more than 0.005 •/• hydrogen accord- 

 ing to an estimation founded on the quantities of hydrogen which 

 had been absorbed from the gas the last few times of successive 

 purification when it was led compressed over charcoal at the tempe- 

 rature of liquid hydrogen, and with the spectrum of this helium 

 after 0,1 % hydrogen had been mixed with it. 



The gas used for the experiment did not differ much from that which 

 served for comparison, and of which the red hydrogen and the 

 helium lines vanished simultaneously for the highest vacua, but it 

 seemed to be somewhat less pure, for the red hydrogen line prepon- 

 derated over the helium line for the highest vacua. In the different 

 spectra the hydrogen line was not to be seen at a pressure of 32 mm., 

 the inline with an intensity of 0,01 of He 5016; at 12-16 mm. 

 C was faint compared with He 6677, and F faint compared with 

 He 5016. An amount varying between 0,01 and 0,3 was assumed 

 for the ratio of the intensity. 



On the other hand at 32 mm. the G in the mixture with 0,1 pCt. 

 hydrogen had already the same intensity as He 6677, F 0,3 of 

 lie 5016, which remained the case at 16 mm. (somewhat less for 

 0, somewhat more for F). 



In spite of the precautions taken it was observed a single time 

 that the hydrogen lines increased in intensity during the deter- 

 mination, so when we proceeded to lower pressures the determinations 

 became unreliable. These comparisons are, therefore, very imperfect ; 

 but then, the examination how traces of hydrogen in helium may be 

 quantitatively determined by a spectroscopic method would constitute 

 a separate investigation. In connection with the above difference in 

 content of B and C with the original gas, the observations mentioned 

 may perhaps serve to show that these percentages have not been 

 much more than 0,004 and 0,008. 



The purity of the helium had already been beyond doubt before, 

 for the cock worked without the least disturbance, and no turbidity 

 was observed even in the last remaining 2 cm 8 , of liquid. 



The reliability of the helium thermometer was tested by the 

 determination of the boiling point of oxygen, for which 89° K. 

 was found instead of 90° K. We must however, bear in mind that 

 the thermometer has not been arranged for this temperature and 

 the accuracy in percents of the total value is considerably higher 

 for the much lower temperature of liquid helium. 



