PART II. POLAR MAGNETIC PHENOMENA AND TERRELLA EXPERIMENTS. CHAP. IV. 583 



According to the above, however, it is also conceivable that some rays of the group called A, 

 whii more especially bend round the terrella above and below the plane of the magnetic equator may 

 alscbe made partially responsible for the precipitation found upon the east side of the branching 

 scnns; for we have seen that some of these rays will loop upon themselves, and it is then clear that 

 ow)ranch of the ray-trajectory will be turning back. This branch may then just occasion precipitation 

 on ie screen in more southerly latitudes of the terrella with a tangential component the reverse in 

 diretion of that to which the ray would originally have given rise. 



iN AN INTIMATE CONNECTION BETWEEN RAYS OF THE TWO GROUPS A AND B. 



114. In continuation of the experiments which have last been described, I have succeeded at 

 lenjh in obtaining complete clearness as to the relative connection between rays of group A and those 

 of joup B. 



Further experiments were first made with an eight-armed star-screen with arms 3 centimetres in 

 hi-i^it instead of 1.5 centimetres as they had previously been, the purpose being to see whether the 

 preipitation and shadows on the two star-screens corresponded. 



The first eight photographs of fig. 213 show the conditions. The first four are from an experiment in 

 whii Nos. i and 3 were taken from directly opposite the magnetic poles, north and south, from posi- 

 tion with hour-angles of 90 and 270 without elevation, while No. 2 was from a position with hour- 

 ang 235 and 24 declination, and No. 4 with an hour-angle 295 and 24 declination. The discharge- 

 curint employed was 22 milliamperes with a tension of about 3000 volts and a magnetising current of 

 20 nperes to the terrella. The pressure sank from 0.022 mm. before the experiment, to 0.043 after it. 



The next four photographs were taken from the same respective positions, with discharge-current 

 of :, milliamperes and tension about 3000 volts, while the magnetising current was 36 amperes. The 

 prt-Hire was 0.012 mm. before the experiment, and 0.066 mm. after it. 



A comparison with the phenomena represented in fig. 212, in which the star-screen was about 

 1.5 n. in height, shows, on the whole, a similarity. One difference that may be mentioned is that the 

 posive precipitation does not extend so far down towards the terrella itself, as when the height of the 

 scrtn was less. The negative precipitation, on the other hand, extends right in, and the polar ring on 

 the ;rrella itself is now quite as well formed as with the lower screens. One especially characteristic 

 feafe is that the dark shadows in the ring of light on the terrella just behind the screening branches, 

 an; 10 longer now than when the screens were only 1.5 cm. high, but are, if anything, narrower. This 

 shos that the rays do not strike so straight down towards the terrella as might be thought from the 

 is experiments, but that rays that come in contact with the higher parts of the screen first move 

 a Hie away from the screen, and then turn in towards it again. 



It will further be observed from the extremely interesting negative precipitation on the screens 

 (froi which an idea can actually be formed of the manner in which the rays approximately move from 

 the orthern polar light-ring to the southern, see Nos. 4 and 8), that the precipitation nearest the terrella 

 s (inter and thinner than farther out. This suggested the thought that possibly one of the eight 

 brarhes of the star-screen might cast a shadow upon the neighbouring branch, that again upon the 

 nex and so on. In order to determine this question, one of the branches was cut off, as shown in 

 Nosg and 10. The positions here are similar to those in Nos. 2 and 4, and the discharge-current employed 

 in t; experiment was of 23 milliamperes. It will at once be seen from the photographs that the already- 

 meroned narrow precipitation of light nearest the terrella upon the branching screens is not caused 

 by ie casting of the shadow by one branch upon its neighbour. 



Photographs n and 12 were taken in two experiments, both in the same position as in No. 2. The 

 exp-iments were made very much as before, but with 10 and 20 amperes to the terrella. The tension 



