When a beam of light passes through a refracting surface, especially
if imperfectly polished, a portion of it is irregularly scattered, and
makes the surface visible in all directions, but most conspicuously in
directions not far distant from that of the light itself: and, if a
reflecting surface be placed parallel to the refracting surface, this
scattered light, as well as the principal beam, will be reflected,
and there will also be a new dissipation of light, at the return
of the beam through the refracting surface. These two portions of
scattered light will coincide in direction; and, if the surfaces be of
such a form as to collect the similar effects, will exhibit rings of
colours. The interval of retardation is here, the difference between
the paths of the principal beam and of the scattered light between the
two surfaces; of course, wherever the inclination of the scattered
light is equal to that of the beam, although in different planes, the
interval will vanish, and all the undulations will conspire. At other
inclinations, the interval will be the difference of the secants from
the secant of the inclination or angle of refraction of the principal
beam. From these causes, all the colours of concave mirrors observed
by NEWTON and others are necessary consequences: and it appears that
their production, though somewhat similar, is by no means, as NEWTON
imagined, identical with the production of those of thin plates.