the Inflection , Reflection, and Colours of Light. 269 
images were also increased in size, being more distended, and 
highly coloured. These things immediately suggest the ex- 
planation. Each of the small fibres forms an image, which, 
from the different reflexibility of the rays, is divided into the 
seven primary colours. But why does not a plain mirror form 
one of these upon the same principles ? In fig. 12. let AE be the 
curve surface of a very convex mirror, that is of a small fibre; 
GC a ray reflected by the small surface DC ; it will be sepa*- 
rated into Cl red, and CK violet, by the unequal action of 
FC on its parts. But if DC is continued to L in a straight 
line, then LC's sphere of reflection extending a little way be- 
yond it, to KC, the part nearest to C, and not to IC, will drive 
KC and also the indigo and part of the blue nearer to the per- 
pendicular ; then IC being within LC J s sphere of inflection, 
will, together with the orange, yellow, and part of the green, 
be brought nearer to KC ; so that IC and KC will both be 
brought to an angle equal to that of incidence, and will be 
reflected in a parallel white beam. If LC is removed a little, 
or the surface becomes more convex, IC is attracted, and KC 
repelled, but not so much as to reduce them to parallelism and 
whiteness, an image being formed narrower and less coloured 
than when LC is moved so far round that KC is attracted, and 
IC deflected or repelled. If LC is moved round so that the 
mirror is concave, then KC is repelled, and IC attracted, as 
before, unless the curvature be considerable ; and then KC 
and IC are both repelled, and an image formed in the caustic 
by reflection. In Obs. 3. we found that certain irregularities 
in the surface of the reflector caused the images to be in the 
inverted order of colours. How does this happen ? In fig. 13. 
let gf, fe, er, ri , and ih , represent the sections of the con- 
