207 



of the latter at the time of observation. As an example, 

 it may be stated that in the neighbourhood of the hours of 

 6 a.m. and 6 p.m. aurorae are most likely to appear in the 

 eastern or western skies. Of course, aurorae may occur in 

 the east and west at other times of the day. The figures 

 show, however, that an exhibition is more likely to occur in 

 that direction than in any other at that particular time (see 

 pi. xx.). 



7. The Diurnal Variation in Trend. 



Auroral displays at Cape Royds are distinguished by the 

 large proportion of curtains traversing the heavens in a 

 linear or nearly linear direction. A remarkable daily 

 sequence was observed in their trends. Always (on the 

 average) they appeared directed approximately towards the 

 sun. Thus should a curtain persist for any length of time it 

 was noted always to exhibit a slow rotational movement 

 counter-clockwise. This swinging movement was more con- 

 spicuous sometimes than at others. On rare occasions a swing 

 of as much as 20° was observed in a few minutes; at other 

 times this would occupy several hours. Generally speaking, 

 the swing was fairly regular. 



Except in simple cases, where the curtain crosses the sky 

 in the vicinity of the zenith, considerable experience is 

 required in determining the trend (geographic orientation), 

 and it is quite certain that erroneous statements in this regard 

 are usually logged where observations are made by untrained 

 observers. Although most of the logging at Cape Royds was 

 made by a few capable observers, a small proportion of the 

 statements are likely to be in error in respect to trend. In 

 the case of bands and curtains near the horizon it is probable 

 that the directions stated are frequently in error ; in other 

 cases they may be correct, but refer to a portion of a curved 

 form whose character is obscured because of its low elevation. 



For this reason, in the following analysis, account is 

 taken only of those appearing above 60° in altitude. Further 

 bands, etc, showing any marked changes in direction will be 

 omitted, for obvious reasons. Finally, due weight has been 

 given to those observations made by the more competent 

 observers. An analysis of the log resulted in the accompany- 

 ing statistics. The direction is stated as N". and S., E. and 

 W., and in all other cases the direction from the zenith is 

 indicated : for instance, N.W. means a trend from N\W. 

 to S.E. 



Trend of Curtains. 

 April. 



A.M. P.M. 



7, 2.00, N. and S. 21, 10.15, N. by E. 



20, 11.00, N.andS. 



