and other Spectra with the Spectrum of the Aurora. 265 



that the spectrum appears to be totally changed when the mass of 

 vapour is altered" I suppose, too, we may now add magnetism 

 as capable of effecting a change in certain spectra — though this 

 subject is at present new and only partially explored. 



Thus far I have given the results of some observations made 



o 



with a view to test experimentally Professor Angstrom's con- 

 clusions on the aurora-spectrum. It will no doubt be remarked 

 by some who look at my drawings, that I have obtained a 

 number of negatives, but that there is no positive result any- 

 where. I quite admit this, and am disappointed in so doing; 

 but then it must be remembered that one way of finding out what 

 a thing is, is to find out what it is not. I venture to think, 

 however, we may not consider the time entirely thrown away, 

 and that, although no one examined spectrum complies with all 

 the auroral conditions, we may still usefully examine the set in 

 regard to their relative proximity thereto. 



Tested by coincidence or close proximity of lines to those of 

 the aurora we may arrange the spectra in the following order : — 

 1, iron; 2, air-spark; 3, hydrogen; 4, air-tube spectrum; 

 5, phosphoretted hydrogen ; 6, carbon and oxygen. 



The air-tube spectrum might perhaps stand higher in the 

 scale but for its broad bands, which make comparison doubtful. 

 Lines of oxygen possibly escape detection in the aurora from the 

 faint character of its spectrum. 



The phosphorus and iron spectra are specially interesting in 

 connexion with Professor Nordenskiold's "metallic and mag- 

 netic cosmic dust in the polar regions " (see Phil. Mag. vol. xlviii. 

 No. 321, page 546). 



If asked to give an opinion in the present state of our know- 

 ledge of the aurora question, I should say : — 



1st. That the yellow-green line, and possibly also the red, are 

 due to phosphorescence or fluorescence. 2nd. That the fainter 

 lines are partly due to the air-spectrum (but not specially the 

 violet-pole), in which H lines probably play a prominent part, and 

 the spark-spectrum appears nearer the mark than the tube- 

 spectrum ; and that the remaining bands or lines may be due to 

 phosphorus and iron (the close coincidences in this latter spec- 

 trum with the aurora-lines being certainly very striking). All 

 these spectra seem to me to claim special and further attention ; 

 and to them I would add that of OH 2 in relation to aurorse of a 

 dense and misty character. 



For measuring auroral and other lines, a cheap and very effec- 

 tive micrometer is constructed by making the whole slit-plate of 

 the spectroscope (and consequently the spectrum itself) traverse 

 the field with a fine micrometer- screw, a pointer or pointers being 

 fixed in the eyepiece. My early observations were made with 



Phil Mag. S. 4. Vol. 49. No. 325. April 1875. T 



