PHYSICS: E. THOMSON 
5 
returning and traversed by waves of light of varying intensity. These 
appearances were not, of course, peculiar to the particular post of 
observation but were observed from points far east and west without 
substantial differences. A vivid description by Dr. C. C. Nutting is 
found in Science for October 6, 1916, pp. 496 and 497. See also Science 
of November 10 to November 17 and December 8, 1916, for other letters 
concerning this great aurora. 
The belt covered by this display was evidently of very great extent 
east and west and spread far to the north in latitude. One is compelled 
to recognize that observers far apart seeing the same appearances are 
looking up between nearly parallel and vertical streamers, seeing those 
on end as a zenith crown when directly above them; while laterally 
they are superposed by being back of one another at varying distances. 
The same auroral appearances are possible to be seen ahke at different 
places simultaneously, only when a system of vertical streamers exists. 
Let us for illustration assume an extended horizontal fiat surface 
and that there be erected above it a set of vertical and very long rods 
in a vertical plane extending east and west, the rods being spaced apart 
like the paling in a long straight picket fence. An observer on the 
plane faces north looking towards the rods or paling the lower ends of 
which are high above his position, appearing say at an altitude of 60° 
in the north direction. The lower ends will now appear to lie east and 
west in an arch of curve convex upward owing to diminished angle of 
vision with distance and the rods or palings will appear to converge 
upward if long enough almost to the observer's zenith, much foreshort- 
ened; the whole effect being that of perspective. Removing the point 
of observation further to the south, the middle point of the arch drops 
more and more, the arch becomes flatter, and the vertical rods appears 
less foreshortened and longer, while still converging towards the ob- 
server's zenith. These varied appearances, modified to a minor degree 
by the earth's curvature, are just what are seen in auroras. If the 
vertical rods are spaced apart irregularly, increased in number and 
spread into a band so that they do not lie exactly in a vertical plane 
east and west, but in an arrangement like a long strip of forest extending 
east and west and of considerable width north and south (an arrange- 
ment corresponding to a belt or zone of streamers instead of a single 
line), the analogy to the auroral arrangement at any instant is much 
closer. 
It can readily be seen that the recognition of the vertical relation of 
streamers to the earth's surface and the nearly constant level of their 
lower ends simplifies to a great extent the study of auroras, particularly 
