and Nature of the Cosmic Electric Rays. 51 



We see that the height of the lower limit even for the 

 same form comes out very different in the different cases. 

 Values have heen found varying between 73 and 166 km. 



Now we must remember that even if the aurorse always 

 reached the same height above the ground, the heights deter- 

 mined would be spread out between fairly wide limits on 

 account of inevitable errors. The most serious errors are 

 due to the fact, that it is often difficult to mark off corre- 

 sponding points on the bottom edge on the two pictures. 

 The parallax must often be found from the displacement of 

 outlines on the pictures, and in such cases there may be an 

 error due to what we have called "volume effect," and 

 which will have the effect of making the height of the lower 

 limit to come out too large. 



Errors may also enter into the results from the fact that 

 the outlines are not sharp, and as the pictures do not always 

 come out equally -well on both pictures the lines drawn as 

 corresponding lines may not really be corresponding. 



I think that the extreme values given in the table are due 

 to errors, as these are always found in cases where the 

 definition of corresponding points has-been less certain. 



In the case of very distinct pictures we never found such 

 extreme values, but the lower limits were found at a height 

 of about 100 km. in all such cases. 



The height distribution for a number of determinations 

 under the most favourable conditions are given in T;ible IV. 





Table IV. 



- 



Interval. 



Number of points, 



95-97 









97-99 





6 



99-101 





18 



101-103 





5 



103-105 





8 



105-107 





27 



107-109 





o 



109-111 









Average 103"2. 





In all these 66 cases the lower limit has hem found 

 between 98 and 108. 



Although the height distribution curve is influenced from 

 inevitableVrors there can be no doubt that real differences 



exist. 



I] 2 



