188 (;OMETS' TAILS, THE CORONA, AND THE AURORA. 



radii are so much less than their height above us that the eflE'ect to a 

 beholder on the earth is as if the3" moved along the lines of force. 

 Over the equator there is little luminescence, for lack of atmosphere. 

 But as the lines of force travel north and south they dip downward, 

 making for the magnetic poles, over which thej" stand vertical. Soon 

 the particles find themselves in lower la^^ers of the atmosphere, com- 

 parable in density with our highest artificial vacua, and begin to give 

 out the darting and shifting lights of the cathode ray. But this can 

 only be at the cost of absorption, and )yy the time the denser la3"ers of 

 air are reached their energy is exhausted. Hence the dark circles 

 roiuid the magnetic poles from which, as from behind a curtain, the 

 leaping pillars of the Aurora rise. From this point of view it is sig- 

 nificant that Dr. Adam Paulsen, who has made a special study of the 

 northern lights, found so many points of correspondence between 

 them and cathode ra^^s that in ISO-i he was led to regard the aurora 

 as a special case of the latter, though unable to give an\^ account of 

 their origin in the upper atmosphere, such as is supplied Vjy Arrhenius's 

 theory. 



The most ol)vious test to which we can .subject such a theory is to 

 ask from it some explanation of the very remarkable periodic varia- 

 tions in the frequency of aurorie. If they are caused b}" streams of 

 particles ejected from the sun, there should be some connection 

 between the changes in the sun's activity, as indicated by the number 

 of sun spots, and the number of aurorie observed. Again, since a 

 negative charge in motion is (pace M. Cremieux) equivalent to a nega- 

 tive current, the passage of electrified particles through the upper 

 atmosphere should affect magnetic instruments on the earth. Sun 

 spots, aurora?, magnetic storms should therefore var}" together. 



It has long been known empirically that they do agree in a general 

 way. Arrhenius''s discussion of the mass of statistics of ol)served 

 aurorte forms so striking an example of the "Method of concomitant 

 variations" that at the risk of wearying the reader we shall give it in 

 some detail. 



i. jS/"ir .sccnliir pcrtodx. — {<() Both sun spots and auror;v show marked 

 maxima at the middle of the eighteenth and the end of the nineteenth 

 centuries. 



{h) Sun spots, uuroric, and magnetic storms go through a simul- 

 taneous increase^ and decrease in the well-known period of 11.1 years. 



The source of these slow variiitions nuist be looked for in the little 

 understood variations of the sun's activity. 



2. ihniaxd j>eriod. — The number of aurorte is greatest in March and 

 September, and least in .lune and December; and the mean frequency 

 for Ijoth hemispheres is somewhat less in June than in December. 



Now the sun's activity, as indicated l)v the nunil)er of sun spots, is a 

 mininnim at his ecjuatoi", the s})ots occurring j)rincipally in l)elts about 



