From the Table, it is seen that aurorae are most likely to be seen in the sector 

 N.E. to E., and least likely to be seen in the sectors S.W. to W. and W. to N.W. The 

 Table also gives the corresponding figures for the period May 13th to July 31st, 

 comprising 78 days in the darkest portion of the year. Not only are the figures for 

 this period in qualitative agreement with those for the whole period, but they are 

 also very closely in quantitative agreement. Division by eight of the figures for the 

 whole period, and by fiva of the figures for the dark period, gives the numbers shown 

 in the last two lines, which are seen to be in very close accordance. Fig. 3 gives a 

 graphical representation of the variation in azimuth for the longer period. 



Not only are the figures accordant for the two periods shown in Table 4, but even 

 for very short periods of one or two weeks the same general tendency is shown. Numbers 

 similar to those in Table 4 have, in fact, been compiled for the first and last halves 

 of each month, and the azimuth of maximum frequency of aurora is, in each period, 

 found to lie either in the sector N.E. to East, or in one of the two adjacent sectors. 

 The minimum is even more definitely defined, generally in the sector S.W. to W., but 

 sometimes in the sector N.W. to W. 



Though the method of observation at Cape Evans differed from that employed on 

 the previous Shackleton Expedition, the figures in Table 4 are in very close agreement 

 with those of Mawson (1908), indicating that the relation shown in fig. 1 is characteristic 

 of the situation of Winter Quarters. (Only a few miles separated the two Headquarters 

 in question.) 



It is possibly worth noting that the azimuths for maximum and minimum 

 frequency are almost, but not quite, separated by 180. 



It is also important to note that the distribution in azimuth is apparently unaffected 

 by the presence of the daylight arch in the spring and autumn, which suggests that 

 the characteristic distribution in azimuth is about the same during daylight hours 

 as during the darker hours of the day. 



Mention has already been made of the fact that auroral displays appeared 

 commonly to originate in the N.E. or E., sometimes moving to a point directly overhead, 

 or passing overhead, during the maximum display. This tendency is expressed in 

 the statement that aurora? were only seen in the " rare sectors " (N.W. to W. 

 and W. to S.W.) at times of maximum display, but were seen in the " common 

 sectors " even at times of minimum display. As will be seen later, this tendency is 

 of a most persistent character. 



6. DISTRIBUTION OF AURORA WITH RESPECT TO TIME. 



An analysis of all observations between May 13th and July 31st, with respect 

 to both azimuth and time, is given in Table 5, which shows the number of occasions 

 during this period on which aurora was seen in each sector at each hour of the clay. 

 Results are almost entirely lacking for the twilight hours, as, even during the darkest 



12 



