This is
best illustrated by a famous experiment of Sir David Brewster's,
modified to suit present requirements. Into a glass cylinder, with
its ends stopped by discs of plate-glass, a small quantity of nitrous
acid gas is introduced; the presence of the gas being indicated by its
rich brown colour. The beam from an electric lamp being sent through
two prisms of bisulphide of carbon, a spectrum seven feet long and
eighteen inches wide is cast upon the screen. Introducing the
cylinder containing the nitrous acid into the path of the beam as it
issues from the lamp, the splendid and continuous spectrum becomes
instantly furrowed by numerous dark bands, the rays answering to which
are intercepted by the nitric gas, while the light which falls upon
the intervening spaces is permitted to pass with comparative impunity.

Here also the principle of reciprocity, as regards radiation and
absorption, holds good; and could we, without otherwise altering its
physical character, render that nitrous gas luminous, we should find
that the very rays which it absorbs are precisely those which it would
emit. When atmospheric air and other gases are brought to a state of
intense incandescence by the passage of an electric spark, the spectra
which we obtain from them consist of a series of bright bands. But
such spectra are produced with the greatest brilliancy when, instead
of ordinary gases, we make use of metals heated so highly as to
volatilise them. This is easily done by the voltaic current. A
capsule of carbon filled with mercury, which formed the positive
electrode of the electric lamp, has a carbon point brought down upon
it. On separating the one from the other, a brilliant arc containing
the mercury in a volatilised condition passes between them. The
spectrum of this arc is not continuous like that of the solid carbon
points, but consists of a series of vivid bands, each corresponding in
colour to that particular portion of the spectrum to which its rays
belong. Copper gives its system of bands; zinc gives its system; and
brass, which is an alloy of copper and zinc, gives a spectrum made up
of the bands belonging to both metals.

Not only, however, when metals are united like zinc and copper to form
an alloy, is it possible to obtain the bands which belong to them. No
matter how we may disguise the metal--allowing it to unite with oxygen
to form an oxide, and this again with an acid to form a salt; if the
heat applied