76 Prof. Snell on a singular case of Parhelion, 
which lie within the narrow limits between 21° 50’ and 23° 22’. 
Therefore, although the entire width of the halo is 21° 37%, 
(=43° 27’—21° 50’,) only one or two degrees of the inner bor- 
der will be noticeable. But the inner edge of what seems to be 
the entire halo will be tolerably well defined, and the outer edge 
will shade off gradually to the light of the sky. Every careful 
observer must have noticed, that the area lying within the halo 
is darker than that without. The reason is made obvious by 
what has just been stated; the space which is regarded as out- 
side, is in fact a part of the halo itself. 
Every crystal decomposes as well as refracts the light, and no 
one can send more than a single color to the eye. But other 
crystals, closely contiguous, by slight differences of position, 
may transmit other colors ; and the effect will, in general, be that 
of white undecomposed light. On the inner edge, however, the 
red may almost always be seen, uncombined with any other col- 
or, because no other can deviate so little from the original direc- 
tion,—red being the least refrangible, and the crystals being, by 
supposition, at the minimum limit. Outside of the red, the other 
colors are sometimes seen in the prismatic order, growing more 
and more faint, from the mixture already spoken of. For in- 
stance, suppose a crystal B, at such a distance from the axis, ES, 
that when lying in the position for minimum deviation, it throws 
the green ray to the eye; then other crystals in the same visual 
direction, on which the light falls at angles of incidence a little 
larger and a little smaller, will also refract the same color to the 
eye, on account of the slight changes in deviation which occur 
there ; but there are others, still farther revolved, which will bring 
the less refrangible colors, yellow, orange and red. And thus the 
green, though predominant, is rendered impure. The violet at 
C is partially neutralized by all the other colors, and can rarely 
be seen. Beyond the colored ring, whose width is represented 
by ABF, the light, though just as much decomposed, is reduced 
to whiteness again by the union of all the colors from crystals in 
different positions. In the rainbow, the several colors are equally 
pure and distinct. 'The reason of this difference is obvious. If the 
distance of a rain-drop from the axis of the bow be given, the color 
thrown to a certain point is given also, because a revolution of the 
drop in its place does not change the relations of its surface to the 
sun, or to the observer’s eye. Not so with a crystal; both dis- 
