3 
phenomenon is observed with telescopic aid, I may mention 
that on the 28th of June the sea was calm and the sky 
quite cloudless at the setting of the sun. Of the final 
coloured rays fifteen diameters showed the first to be a full 
and splendid yellow, which was speedily followed by the 
usual green, and then for a second and a half by a full and 
perfect blue. Respecting the increased duration of the 
colour I have found that when the atmosphere is sufficiently 
favourable to allow a power of 60 diameters being employed 
with a three-inch object glass, the green effect is seen at 
that part of the sun’s limb in contact with the horizon even 
when one half the sun is still unset, and of course from then 
till final disappearance. 
The different colours seen, together with the order of 
their appearance, are suggestive of the prismatic action of the 
atmosphere as the cause of their production, and the inter- 
ception of the horizon or the cloud as the cause of their 
separation. 
Assuming the correctness of this view, it becomes evident 
that an artificial horizon would prove equally efficacious in 
separating the coloured bands, and also that if employed 
during an inspection of the sun’s lower limb, the least 
refrangible end of the spectrum would be disclosed. Accord- 
ingly I introduced into an eyepiece of my telescope a black- 
ened disc of metallic copper, having a slit cut in it of about 
the one hundred and fiftieth of an inch in width, and pro- 
ceeded to make an observation in July when the sun was 
about one half of its meridian height. The blinding glare 
however of that portion of the sun seen through the slit 
rendered the observation futile. By projecting a large 
image of the sun into a darkened room I was enabled to get 
the whole of the spectrum produced by the prismatic action 
of the atmosphere in a very satisfactory manner. In this 
case a semicircular diaphram was used, so placed that its 
straight edge divided the field of view into equal parts, from 
