48 
Proceedings of the Royal Society 
possible, and the focus adjusted until the spectrum appears bright 
and clear, with the black solar absorption lines crossing it vertically. 
If there is dust on the slit each grain appears drawn out into a thick 
black line running through the spectrum horizontally. When these 
horizontal lines are seen the slit should be cleaned by gentle brush- 
ing with a camel-hair pencil. 
The spectrum of clear sky or a bright cloud, in the instrument 
used, is a coloured strip apparently about an inch long and half an 
inch wide. On one side a band of red, apparently f^ths of an inch 
wide, emerges from blackness and shades into a strip of 
yellow, which merges into a quarter inch space of green, passing into 
an equal area of blue which dies away in hardly visible violet. 
Two or, in favourable circumstances, three black lines (a, B, C) are 
seen in the red, one (D) appears to separate the red from the yellow, a 
wide nebulous line divides the yellow from the green, several very 
thin lines appear in the green together with two thicker ones (E 
and &), and in the beginning of the blue there is a still darker line 
(F); besides which a glimpse may sometimes be had of other lines 
in the darker part of the spectrum beyond F. On looking directly 
at the sun many of the lines are split up into a number of very fine 
components, and the whole spectrum seems ruled with lines 
intensely black and geometrically narrow, most of which are invisible 
in diffused light. The nebulous line between the yellow and .green 
is usually mistaken for the rainband at first, for it varies in intensity 
from time to time. A more particular observation shows that its 
intensity at any time is proportional to the sun’s nearness to the 
horizon, but it appears to be sometimes affected by other causes. 
Professor Piazzi Smyth has shown that this is a dry air absorption 
band, and he defines it as “a function of dry air and low sun.” The 
real rainband cannot be seen by itself in the little spectroscope, but 
an acquaintance with the spectrum soon show-s that the width of 
the D-line is not constant, and that when widest it seems to shade 
off gradually towards the red, resembling a line ruled by a fine pen 
with a very small hair in its point. It can easily be shown that this 
widening is not due to a widening of the solar sodium absorption 
line, by looking at the sky spectrum througlj the flame of a Bunsen 
lamp faintly tinged yellow with a sodium salt. The D-line proper 
is thus replaced by a brilliant yellow streak, and the rainband is seen 
