During the last summer, while amusing myself with verifying a statement
of Sir D. Brewster respecting the light of the rainbow, viz. that it is
polarised in particular planes, I observed a phenomenon which startled
me exceedingly, insamuch as it was quite new to me at the time; and not
withstanding subsequent enquiries, I cannot find that it has been
observed by any other person. I found that _the light of the blue sky is
partially polarised_. When analysed with a Nichols prism, the contrast
with the surrounding clouds is very remarkable; so much so, indeed, that
clouds of extreme tenuity, which make no impression on the unassisted
eye, are rendered plainly visible.
The most complete polarisation seems to take place near the horizon;
and, when the sun is near the meridian, towards the west and east. The
depth of colour appears to be immaterial, as far as I have been able to
ascertain with an instrument but rudely constructed for the purpose. The
light is polarised in planes passing through the {410} eye of the
observer, and arcs of great circles intersecting the sun's disc.
From the absence (so far as I am aware) of all mention of this
remarkable fact in works on the subject, I am led to conclude that it is
something new; should this, however, turn out otherwise, I shall be
obliged by a reference to any author who explains the phenomenon. The
greater intensity towards the horizon would point to successive
refractions as the most probable theory.