( 276 ) 



Here also the direction of the movement is constant. 



35 cases for the years 1891 and '92 gave on an average for the 

 magnitude of digression HAD=9yW; AH=:-}-45y; AZ=— 16y. 

 duration ,, ,, 3.5 min. ; 5 min. ; 12 min. 



Though the movement of H and D are not sudden in absolute 

 sense, that for Z is too gradual to justify the application of the word 

 sudden to it. 



The reproductions of disturbance-curves for Potsdam and Zi Ka Wei 

 also show some cases of the preceding impulse. At both stations the 

 direction of the movement is remarkably constant. In the publications 

 of the Cape Hoorn observations (1882/83) I found three cases exposing 

 also a constant direction. 



Z 



Consequently with one exception for D and one for Z we find that : 

 the commencing impulse of the S-storms is the i^everse of the vector 

 oj postturhation, being deflected however to the West of it. 



^ Suppositions concerning the origin of disturbance. 



The hypothesis on the existence of defined conical streams of 

 electric energy, which strike the earth, though not quite new, has 

 obtained increased plausibility by Maundbr's results. From the statistics 

 based on the record of disturbances at Batavia it might be concluded 

 that it is chiefly the S-storms that find their origin in the sudden 

 encounter of the earth with such a stream. 



And as the earth is first struck at its sunset-arc, it is not impos- 

 sible that the G-storms, which begin by preference shortly after sun- 

 rise and have no annual periodicity in their frequency as the S- 

 storms have, are only partly caused by these encounters. 



When in the case of the streams we admit that energy progresses 

 from the sun in the form of negative electrons, we might think the 

 G-storms find their origin by electrified particles being propagated 

 by the light-pressure according to the theory set forth by Sv. Arrhenius. 



Further we may suppose, that when the earth has received a 

 charge the following development of the storm is the same as it is 



