OF SHADOW.

Derived shadows are of three kinds of which one is spreading, the
second columnar, the third converging to the point where the two
sides meet and intersect, and beyond this intersection the sides are
infinitely prolonged or straight lines. And if you say, this shadow
must terminate at the angle where the sides meet and extend no
farther, I deny this, because above in the first on shadow I have
proved: that a thing is completely terminated when no portion of it
goes beyond its terminating lines. Now here, in this shadow, we see
the converse of this, in as much as where this derived shadow
originates we obviously have the figures of two pyramids of shadow
which meet at their angles. Hence, if, as [my] opponent says, the
first pyramid of shadow terminates the derivative shadow at the
angle whence it starts, then the second pyramid of shadow--so says
the adversary--must be caused by the angle and not from the body in
shadow; and this is disproved with the help of the 2nd of this which
says: Shadow is a condition produced by a body casting a shadow, and
interposed between this shadow and the luminous body. By this it is
made clear that the shadow is not produced by the angle of the
derived shadow but only by the body casting the shadow; &c. If a
spherical solid body is illuminated by a light of elongated form the
shadow produced by the longest portion of this light will have less
defined outlines than that which is produced by the breadth of the
same light. And this is proved by what was said before, which is:
That a shadow will have less defined outlines in proportion as the
light which causes it is larger, and conversely, the outlines are
clearer in proportion as it is smaller.

[Footnote: The two diagrams to this chapter are on Plate IV, No. 1.]

On the relation of derived and primary shadow (163-165).