4 
one of which it obscured the light. The diaphram was 
placed as before in the focus of the eyepiece, and by rotating 
it every portion of the sun’s limb could be in turn examined, 
and that too in the centre of the field, so as to be equally 
subjected to the minimum of the peculiarities of the instru- 
ment. When the sun’s lower limb was allowed to descend 
into the field of view the first rays were intensely red. After 
a momentary duration they gave place in succession to 
orange, yellow, and green, which were then lost in the ordi- 
nary refulgence of the sun. The upper limb gave green, blue, 
and finally purple, which latter colour I have thus far never 
seen upon tlfe natural horizon. It should be remarked that 
the colours seen were vivid and unmistakeable, and each one 
of them easily detained at will or the whole phenomenon 
recalled by the adjusting screws of the instrument. I appre- 
hend that the results here given sufficiently prove that 
atmospheric refraction is the cause of the coloured rays 
seen at the moment of the sun’s departure below the horizon. 
I have however thought it worth while to examine the light 
proceeding from the moon’s limb by the aid of the artificial 
horizon and of course by direct observation. The results 
were decisive and satisfactory, the spectral colours being 
easily observed. The green effect I have also frequently 
seen on the departure of the moon beneath the edge of a 
dark and well defined bank of clouds. Telescopic aid has 
however in every instance been required. 
The rapid changes in colour observable in the case of 
almost any large fixed star at an elevation of twenty or 
thirty degrees above the horizon, and which changes vary 
between red, green, and blue, may I think be fairly attri- 
buted to the same cause as the colour in the sun’s final ray. 
Particles of dust floating in the air act, I apprehend, for the 
moment in the capacity of diaphram or horizon, and thus 
enable the eye to perceive even in the light of the stars the 
prismatic action of our atmosphere. 
