and Nature of the Cosmic Electric Rays. 61 



The distribution of the values of Z x , Z 2 , Z 3 , on the various 

 height intervals is represented in the curves in fig. 4 



(pi. il). 



An invariable luminosity distribution would mean th;=t 

 at any rate the two quantities ^ and Z 2 ought to keep very 

 nearly constant ; the quantity Z 3 , however, might vary with 

 the strength of photographic impression, even if the lumi- 

 nosity distribution kept constant. 



The curves and the table show, however, great variations 

 of the quantities Z x and Z 2 . I n the case of arcs, l x varies 

 between 4 and 8 km. and Z 2 varies between 10 and 18 km., 

 and in this case l 2 represents practically the distance from 

 the bottom edge to the point where the photographic 

 impression ceases. 



The drapery-shaped arcs show variations of Zj from 2*5 

 to about 14 km. with a maximum at about 6 km., l 2 varies 

 between 6 and 30 km. with a maximum at 12 km., and Z 3 

 between 12 and 120 km. with a maximum at about 35 km. 



For draperies Z x varies still more (4-23 km.) with a 

 maximum between .6 and 7 km., Z 2 varies between 10 and 

 30 km. and Z 3 between 35 and 125 km. 



In the case of the ray-form, Z] and Z 2 are usually not well 

 defined. The luminosity often increases quite gradually, 

 keeps fairly constant for long distances, and then the lumi- 

 nosity gradually diminishes towards the upper limit. It is 

 thus often only Z 3 or the length of the ray which is a fairly 

 well-defined quantity. From the table giving the height of 

 the upper limit of rays, we see that the length of a typical 

 ray varies between about 100-200 km., and in some cases the 

 length of a ray is even greater. 



The typical luminosity distribution of the principal forms 

 is illustrated in fig. 5 (PL II.). How now are these results 

 with regard to the luminosity distribution to be explained 

 from a physical point of view ? 



• Let us suppose that a bundle of electric rays were entering 

 the atmosphere in orbits parallel to the direction found for 

 the auroral ray-streamers. Let us further suppose that there 

 was no effect of any electric or magnetic field to be con- 

 sidered. If the ray bundle were heterogeneous and composed 

 of a mixture of rays with different physical properties, a 

 -change of luminosity distribution might be due to a varia- 

 tion of the composition of the nrv bundle. Such an 

 explanation, which might naturally suggest itself, is ] think 

 excluded, because any definite form of aurora, a drapery or 

 an arc say, must consist of a homogeneous bundle ol rays, 

 for on account of the magnetic deflexion produced by the 



