1902.] on Problems of the Atmosphere. 225 



3^0 00 ^^ ^^ *^® volume of the air directly liquefied, and contained 38 

 per cent, of nitrogen, 4 per cent, of hydrogen, and 58 per cent, of mixed 

 helium and neon. On the average some 25 cc. of gas was collected 

 for every litre of liquid air produced, and as the apparatus works 

 almost automatically, the method becomes a practical one for the 

 separation of the most incondensable constituents of the atmosphere. 

 After the removal of the hydrogen and nitrogen by sparking in the 

 usual way with excess of oxygen over alkali, the helium and neon 

 can be separated. This is easily done by freezing out the neon by 

 passing the gaseous mixture through a tube cooled in liquid hydro- 

 gen ; or the alternative method of spectroscopic fractionation described 

 in a former lecture may be adopted to effect the general separation 

 of all the constituents. The execution of the process on the larger 

 scale will enable gases other than helium and neon to be detected and 

 ultimately separated. 



A similar sample of gas collected from air that bad been passed 

 through lead pipes and dried with strong sulphuric acid contained 

 about 33 per cent, of hydrogen, instead of the 4 per cent, found as 

 above, or say at the most loo^oo Q-th of free hydrogen in the original 

 sample of air. The increased proportion of hydrogen must have 

 originated from the contact of metal with water and acid vapours. 



The total volume of the very volatile gases, helium and neon, 

 collected in this way amounted to 5 «^^o o*^ ^^ ^^® original volume of 

 the air. This mixed gas had a density 8 • 7 times that of hydrogen, 

 so that the mixture was composed of 16 per cent, of helium and 

 84 per cent, of neon ; if the latter be taken as having a density of 10. 

 Thus air contains five times more neon than helium. These 

 experiments prove that air contains as a minimum TjTnrVoir^l^ ^^ 

 its volume of helium, about 7-0 Joo*^ ^^ neon, and not more than 

 ro-WT7o*^ of free hydrogen. In order to make the method strictly 

 quantitative, the correction factor for solubility of these gases under 

 the conditions of the experiment in the liquid air would require to be 

 known. With the use of air freed from the volatile gases by lique- 

 faction and standing for a day or two, the method was proved to be 

 capable of detecting the presence of s-owo-th of free hydrogen in a 

 specially prepared mixture. In a mixture containing goVo*^ ^^ ^^®® 

 hydrogen the loss from solubility, etc., amounted to about 50 percent. 



The spectroscopic examination of these gases throws new light 

 upon the question of the aurora and the nature of the upper air. On 

 passing electric discharges through the tubes containing the most 

 volatile of the atmospheric gases, they glow with a bright orange 

 light, which is especially marked at the negative pole. The spectro- 

 scope shows that this light consists, in the visible part of the spectrum, 

 chiefly of a succession of strong rays in the red, orange and yellow, 

 attributed to hydrogen, helium and neon. Besides these, a vast 

 number of rays, generally less brilliant, are distributed through the 

 whole length of the visible spectrum. The greater part of these rays 

 -are of, as yet, unknown origin. The violet and ultra-violet part of 



Vol. XVII. (No. 96.) Q 



