588 B1RKELAND. THE NORWEGIAN AURORA POLARIS EXPEDITION, igO2 1903. 



No. 16 shows the conditions with a magnetising current of 4 amperes, 3000 volts and 

 milliamperes. 



No. 17 shows the conditions with a magnetising current of 6 amperes, 3000 volts and 23 milli- 

 amperes. The pressure fell from 0.014 mm. to 0.022 mm. 



Lastly No. 18 shows the conditions with a magnetising current of 10 amperes and a discharge 

 current of 23 milliamperes with 3000 volts. The pressure fell from 0.017 mm - before the experiment, 

 to 0.021 mm. after it. 



It is a noticeable fact in all these experiments, that the remarkable occurrences of precipitation that 

 we have previously designated A and B, are also found here when the magnetisation is sufficiently strung 

 (see No. 18). Their shape is so exactly the same as that with the larger terrcllas, that we may conclude 

 that for these experiments at any rate, the vacuum-tube was large enough in our earlier experiments. 



In addition to these distinct, characteristic instances of precipitation on the afternoon side of terrella 

 and screen, we find upon the morning side that the pencil of rays is sharply defined, although the rays 

 evidently only graze the vertical screen. In reality it is, as we shall see, the greater part of the rays 

 from the cathode that are bent downwards in front of the terrella. This is immediately seen if the 

 screen is turned a little, so that the rays strike at an angle. This will be illustrated in the next plate. 



These experiments will be of service to us, as a subsequent paragraph will show, in explaining the 

 zodiacal light. 



In order to find out what became of the luminous patches upon the screen, when the plane o 

 latter no longer passed through the centre of the cathode, the screen was turned 23 in a positive 

 direction, and photographs were then taken. 



Nos. i 4 of fig. 215 were taken from places with hour-angles of 90 and 270. Nos. 2 and 4 

 show how the rays that turned off in front of the terrella, and only grazed the screen in its former 

 position, form a strong, sharply-defined precipitation in the new position. This shows that while the 

 rays near the magnetic equator almost follow that plane, those outside the equator curve more and more 

 away from it. We have seen this before, having found a bright precipitation of rays respectively above 

 and below the two magnetic poles, upon the floor and ceiling of the vacuum-box (fig. 200). 



Nos. i & 2 are of experiments with a magnetising current of only 1.5 amperes to the terrella. 

 A discharge-current of 23 milliamperes at 2800 volts. The pressure about 0.015 mnl - 



Nos. 3 & 4 are of a similar experiment, the only difference being that the magnetising current 

 was 5 amperes. 



Nos. 5 10 are of important experiments in which a small screen was introduced in front of the 

 south pole of the terrella, at about right angles to the magnetic axis. The introduction of this si 

 was for the purpose of studying more closely the above-described precipitation of light. The large screen 

 was turned back 23 to its original position. 



The hour-angles of the several positions corresponding to these photographs were 90, about 210, 

 and 270. The first three are of experiments in which the magnetising current was 2 amperes, the 

 discharge-current 22 milliamperes with 2900 volts. The pressure was 0.017 mm - 



Nos. 8 10 are of experiments like the above, with the difference that the magnetising current was 

 10 amperes and the tension 3000 volts. 



It will be noticed that the precipitation on the small screen moves outwards with increased mag- 

 netisation. 



When we compare Nos. 7 and 10 here with Nos. 4 and 8 in fig. 214, full light will be at once 

 thrown on a hitherto somewhat obscure point. We perceive how it is thai rays of group B and rays of 

 group A, before they have reached the terrella, fortn a single coherent group, but that the rays which come 

 nearest to the poles of the terrella when this is sufficiently magnetised, are thrown round and acquire 



