﻿the Rays producing Aurora Borealis. 231 



atmosphere taking place nearly perpendicular rothe magnetic 

 parallel along which the bands are usually directed. And if: 

 oar interpretation of the structure is right, the sharp bottom 

 edge of tha drapery bands will require rays with the same 

 penetrating power. 



Now, on the other hand, we know that a-rays given out 

 from a radioactive compound during one single radioactive 

 transformation are in fact homogeneous, and consequently 

 ot-rays from some radioactive compound are just the proper 

 agency for the explanation of drapery bands. 



Another most characteristic and peculiar property of aurora? 

 is the existence of equidistant bands (see fig. 2), and in order 

 to explain their formation we shall merely have to suppose 

 that each band corresponds to its own homogeneous radiation. 

 In other words, the radiation must be composed of a mixture 

 of distinctly different homogeneous groups of rays, and at the 

 same time we must suppose that these groups are all of them 

 given out simultaneously from the same source on the sun and 

 subject to the same initial conditions. 



These two conditions being present the mutual relation 

 found between the bauds will be easily understood. For the 

 sake of simplicity we suppose only two groups to be present 

 in the source. 



Two rays, one of each group, leaving the source simul- 

 taneously along the same line, we shall call corresponding rays, 

 and the points where they strike the atmosphere corresponding- 

 points. 



Suppose now the conditions to be such that one of the 

 groups gives rise to a band. If the difference of deflectibility 

 is fairly small the second group will also strike the atmosphere. 

 Now the difference of path of corresponding rays is entirely 

 due to a different deflexion in the magnetic fields traversed, 

 or to the difference of stiffness of the two groups which is a 

 definite quantity, consequently the distance between corre- 

 sponding points will vary continuously along the first band; 

 and according to Stormer's calculations, when the band is 

 extended nearly along the magnetic parallel, the corresponding 

 points should keep their distance nearly unaltered all along 

 the band, and each pair of corresponding points should be 

 situated nearly on the same magnetic meridian. Thus the 

 entire collection of corresponding points will form a hand 

 running nearly parallel to the first one. 



Suppose the system to undergo a gradual change, which, 

 e.g. may be due to a motion of the magnetic equator relative 

 to the sun. Such a change mav produce a motion o( the 

 bands ; but if the two groups of rays are Mill sent out from 



