VT, into so many other long Images _lrsm_, _msvn_, _nvt[Greek:
t]_; and all these long Images would compose the four square Images
_[Greek: pt]_. Thus it ought to be were every Ray dilated by Refraction,
and spread into a triangular Superficies of Rays diverging from the
Point of Refraction. For the second Refraction would spread the Rays one
way as much as the first doth another, and so dilate the Image in
breadth as much as the first doth in length. And the same thing ought to
happen, were some rays casually refracted more than others. But the
Event is otherwise. The Image PT was not made broader by the Refraction
of the second Prism, but only became oblique, as 'tis represented at
_pt_, its upper end P being by the Refraction translated to a greater
distance than its lower end T. So then the Light which went towards the
upper end P of the Image, was (at equal Incidences) more refracted in
the second Prism, than the Light which tended towards the lower end T,
that is the blue and violet, than the red and yellow; and therefore was
more refrangible. The same Light was by the Refraction of the first
Prism translated farther from the place Y to which it tended before
Refraction; and therefore suffered as well in the first Prism as in the
second a greater Refraction than the rest of the Light, and by
consequence was more refrangible than the rest, even before its
incidence on the first Prism.

Sometimes I placed a third Prism after the second, and sometimes also a
fourth after the third, by all which the Image might be often refracted
sideways: but the Rays which were more refracted than the rest in the
first Prism were also more refracted in all the rest, and that without
any Dilatation of the Image sideways: and therefore those Rays for their
constancy of a greater Refraction are deservedly reputed more
refrangible.

[Illustration: FIG. 15]

But that the meaning of this Experiment may more clearly appear, it is
to be considered that the Rays which are equally refrangible do fall
upon a Circle answering to the Sun's Disque. For this was proved in the
third Experiment. By a Circle I understand not here a perfect
geometrical Circle, but any orbicular Figure whose length is equal to
its breadth, and which, as to Sense, may seem circular. Let therefore AG
[in _Fig._ 15.] represent the Circle which all the most refrangible Rays
propagated from the whole Disque of the Sun, would illuminate and paint
upon the opposite Wall if