64 THE ANTARCTIC MANUAL. 
various names have been given, according to the objects which their 
forms simulate. They appear to be collections of streamers arranged 
so as to give the appearance, sometimes of a feather, or of a fan, or of 
an irregular bundle of filaments. 
The various periodicities of the aurora, some well ascertained but 
others still very doubtful, form a very interesting field of enquiry. 
The difficulty of the investigation lies in the separation of true perio- 
dicity from a periodic variation of visibility. The aurora, for instance, 
cannot be observed in full daylight: hence, if records of the presence 
or absence of the aurora are examined, it is clear that at first sight a 
diurnal variation would present itself, and the aurora would seem to 
occur more frequently when the sun is below the horizon than when 
it is above. On second thoughts the observer might be inclined to 
reject this periodicity and put it down altogether to the difference in 
visibility. A closer examination, however, will again bring him back 
to the first view, viz. that he is dealing with a real phenomenon, 
because there is, even in the hours of the night, a clear difference 
between the evening and morning hours, the aurora appearing much 
more frequently before and near midnight than near sunrise. Also, 
in the long winter nicht of the arctic region, where differences of 
visibility are to a great extent excluded, the diurnal periodicity shows 
itself distinctly. 
The annual variation is more difficult to investigate, as observa- 
tions in the summer months are excluded, and even two months 
like December and March have such a great difference in the dura- 
tion of darkness that a comparison becomes very difficult. The 
only way to prove the annual variation would be to compare the 
number of aurore seen in different months during the same hour. 
A periodicity depending on the phases of the moon has been 
made probable, but cannot be said to be proved. Observers should 
pay special attention to this matter, and carefully note how far the 
difference in the visibility of the aurora during full and new moon 
may affect their results. 
To arrive at a satisfactory conclusion in these matters it is not 
only necessary to note down carefully the presence of an aurora, 
with, as nearly as possible, its time of appearance and disappearance, 
but also to give, for each time of observation at which no aurora 
appears, information as to whether the observer is satisfied that the 
aurora was really absent, or whether, in his judgment, it might have 
been present, but hidden by clouds or rendered invisible by one 
strong light of full moon. 
The spectrum of the aurora presents some interesting problems, 
