but is totally reflected upward: when looked down upon, such a tube
looks quite as bright as burnished silver. Pour a little water into
the tube; as the liquid rises, total reflection is abolished, and with
it the lustre, leaving a gradually diminishing shining zone, which
disappears wholly when the level of the water within the tube reaches
that without it. Any glass tube, with its end stopped water-tight,
will produce this effect, which is both beautiful and instructive.

Total reflection never occurs except in the attempted passage of a ray
from a more refracting to a less refracting medium; but in this case,
when the obliquity is sufficient, it always occurs. The mirage of the
desert, and other phantasmal appearances in the atmosphere, are in
part due to it. When, for example, the sun heats an expanse of sand,
the layer of air in contact with the sand becomes lighter and less
refracting than the air above it: consequently, the rays from a
distant object, striking very obliquely on the surface of the heated
stratum, are sometimes totally reflected upwards, thus producing
images similar to those produced by water. I have seen the image of a
rock called Mont Tombeline distinctly reflected from the heated air of
the strand of Normandy near Avranches; and by such delusive
appearances the thirsty soldiers of the French army in Egypt were
greatly tantalised.

The angle which marks the limit beyond which total reflection takes
place is called the _limiting angle_ (it is marked in fig. 6 by the
strong line E _n_''). It must evidently diminish as the refractive
index increases. For water it is 48½°, for flint glass 38°41', and for
diamond 23°42'. Thus all the light incident from two complete
quadrants, or 180°, in the case of diamond, is condensed into an
angular space of 47°22' (twice 23°42') by refraction. Coupled with its