L. VEGARD. 



M.-N. Kl. 



Already in tlie first paper I came to the coiiclusiun that also the 

 green line belcng-ed to Nitrogen: Init at the same time it was mentioned 

 that the correctness of it might he tested hy observing auroral spectra corre- 

 sponding to various altitudes. J^or if the green line is produced by some 

 g-as different from Nitrogen this gas must have a different molecular 

 weight ^, and in that case we should observe, that the ratio between the 

 intensity of the green line and that of the negative N-bands varied con- 

 siderabl}' with the altitude. 



For the analysis of the light emitted at various heights I used a small 

 spectrogTai)h with a great ligh.t power. 



Before the slit of the instrument a lens 

 was mounted in such way that the slit was 

 situated in the focus plane of the lens. The 

 lens forms a picture of an aurora that falls on 

 the .slit, and if the optical a.xis of the instrument 

 hits the sky at a certain point only auroral 

 light from the nearest surroundings of this point 

 will enter the spectrograph and hit the plate. 

 With this instrument it is possible to 

 analyse separately light emitted near the battom 

 edge and from parts near the upper limit. 



For this anah'sis we select auroral forms 

 having fairly long ray streamers, because in 

 that case the two spectrograms will correspond 

 to the greatest difference of height. 



During the evenings and nights of March 

 II and 12. I succeeded in taking two pairs 

 of such photograms — one pair each day 

 (Fig. I and 2). 

 The spectrograms corresponding to the bottom edge (a on the plate) 

 was obtained with about half an hour's exposure. Near the upper limit 

 the light was fainter and the spectrum was here exposed for al:)out four 

 hours (/' on the plate). Still the intensities on the plate of the two spectra 

 were nearly the same. 



The aurorae used for these exposures were mostly draperv-shaped 

 arcs, some draperies and some isolated rays. 



The bottom edge is situated at a height of about 100 km. and the 

 two spectra to be compared correspond to a difference of height of about 

 50 — 80 km. 



The intensities of the spectral lines will l)e accurately measured later 

 on by means of a microphotom(?ter, but some results can be directly seen 

 on the plate. 



ie. r and 2. 



' The gas CO with a molecular weight equal to that of N cannot be supposed to be 

 present with a sufficient density. 



