reversibility.[5] In the case of refraction, for instance, when the
ray passes obliquely from air into water, it is bent _towards_ the
perpendicular; when it passes from water to air, it is bent _from_ the
perpendicular, and accurately reverses its course. Thus in fig. 5, if
_m_ E _n_ be the track of a ray in passing from air into water, _n_ E
_m_ will be its track in passing from water into air. Let us push this
principle to its consequences. Supposing the light, instead of being
incident along _m_ E or _m'_ E, were incident as close as possible
along C E (fig. 6); suppose, in other words, that it just grazes the
surface before entering the water. After refraction it will pursue
say the course E _n_''. Conversely, if the light start from _n_'', and
be incident at E, it will, on escaping into the air, just graze the
surface of the water. The question now arises, what will occur
supposing the ray from the water to follow the course _n_''' E, which
lies beyond _n_'' E? The answer is, it will not quit the water at all,
but will be _totally_ reflected (along E _x_). At the under surface of
the water, moreover, the law is just the same as at its upper surface,
the angle of incidence (D E _n_''') being equal to the angle of
reflection (D E _x_).

[Illustration: Fig. 6]

Total reflection may be thus simply illustrated:--Place a shilling in
a drinking-glass, and tilt the glass so that the light from the
shilling shall fall with the necessary obliquity upon the water
surface above it. Look upwards through the water towards that surface,
and you see the image of the shilling shining there as brightly as the
shilling itself. Thrust the closed end of an empty test-tube into
water, and incline the tube. When the inclination is sufficient,
horizontal light falling upon the tube cannot enter the air within it,