1889.] of the various Species of Heavenly Bodies. 231 



When the spark first passed only a few of the strongest nitrogen 

 flutings in the violet were visible, but as the pressure was reduced, 

 the spectrum gradually extended towards the red. A line of oxygen 

 near 5316 was visible in the early stages, but it afterwards disappeared. 

 At the most luminous stage, nothing but nitrogen flutings were visible. 

 After a time the nitrogen flutings dimmed, and low-temperature 

 flutings of carbon appeared. Then the F line of hydrogen appeared, 

 and a little later the C line. Later still, the hydrogen line at G also 

 appeared. With the further dimming of the nitrogen flutings, an 

 oxygen line at 471 brightened, being sometimes as bright as the F line, 

 and brighter than the carbon flutings. The whole spectrum then 

 became very faint, but as the line at 471 dimmed, another oxygen 

 line at 465 appeared. Ultimately, the glow was so faint that only a 

 few of the nitrogen flutings were visible. 



The sequence of the various flutings and lines is shown in fig. 18. 

 Below the various air spectra the principal lines of the aurora 

 spectrum are given for comparison. The spectra of nitrogen, carbon, 

 and hydrogen are given as a key to the spectra observed. It should 

 also be stated that the line near 5316 is an oxygen line. I am now 

 working at this line. It will be seen at a glance that there is only 

 one coincidence with one of the most persistent flutings, which are all 

 that need be considered. Since equally persistent flutings are not 

 present in the spectrum of the aurora, this coincidence is obviously 

 of no importance. 



Y. Comparison with Uncondensed Meteor- swarms. 



In my preliminary communication I indicated the remarkable 

 coincidences between the lines in the spectrum of the aurora and 

 the bright lines in the spectrum of ry-Cassiopeiae, and also with the 

 absorption-bands in bodies of Group II. These bodies are uncon- 

 densed swarms of meteorites at a comparatively low temperature, 

 and hence the comparison suggests the probable meteoritic origin 

 of the spectrum of the aurora. 



I have since extended the tables which I then gave, and excluding 

 for the present Gyllenskiold's observations, they now stand as 

 telow : 



