the sun's rays was systematically crossed by
dark lines in various places. These lines always,
however, occupy the same relative position
when the coloured image ( which is usually called
the prismatic spectrum) . is obtained from sunlight;
while the light, either from fixed stars or
from artificial sources, gives a spectrum either
without lines or crossed by lines occupying
different positions. In every case the position and
number of the lines are found to be invariably
the same when the light has been obtained from
the same source. The light from the moon or
planets being reflected sun light, exhibits a
spectrum crossed by the same lines as those
of the usual solar spectrum.

For a long time there seemed no means of
accounting for these singular interruptions in the
beautiful image obtained by decomposing sun
light into its elements. Why should there be, in
this image or spectrum, dark portions having no
apparent reference to the various shades of
colour, but indicating a partial and imperfect
state in the pencil of rays itself as it emerged
from the luminous body? If the light from
the body of the sun were given off pure, and
were received in the same state, one can
hardly imagine that interruptions in the
spectrum could occur. If, however, owing to the
passage of these rays through some coloured
medium or atmosphere before they reached the
earth, they were deprived of a portion of their
effect, the place of the rays thus removed might
be expected to be marked by some such blanks
as we see. But if the absorption, or partial
absorption of light or colour was limited to our
atmosphere, there should be no difference in the
spectra obtained from the sun and star light,
whereas experiment shows that such difference
does exist. Whatever lines are due to absorption
while passing through the air must clearly
be common to all kinds of light received from
the heavenly bodies, and we are thus apparently
forced to the conclusion that dark spaces, not
common to sun and star spectra, are produced by
a loss of certain parts of a pencil of rays before it
finally emerges into space. In other words, we
are led to the assumption that the light of the sun
proceeds from the nucleus or central body of the
sun in groups of glorious and perfect and intensely
bright rays of pure white light which are shorn
of at least some portion of their brilliancy even
before they enter the measureless space around.
Of the light thus diminished only a few pencils
—few at least in proportion to the whole number
given off—are interrupted by and penetrate our
atmosphere, lighting up the surface of the earth,
while the rest either serve to illuminate other
planets or are lost in the starry waste. The rays
received on the earth are, of course, deficient
both in the part absorbed by the sun's
atmosphere and also in that part absorbed by the
atmosphere of the earth. The amount of the
latter we can pretty well estimate, and what
remains must belong to the former.

Making use of the strong and very pure light
obtained by burning lime in a jet of mixed gases
(known us the Drummond light), and passing
a pencil of rays of such light through a prism,
a complete spectrum may be thrown on a screen ;
but if instead of this, a light is used affected by
the combustion of some substance that colours
flame, the spectrum thrown on the screen is
no longer complete, but consists of one or more
coloured bands, each in its place, all the rest
of the image being dark. Whatever the state
of the substance may be that is burnt, the
colours are found to depend only on its
ultimate particles or elements. Thus common salt
(which is a combination of the gas chlorine and
the metal sodium) when burnt, gives a yellow
colour to flame, in which the human face looks
ghastly in the extreme, ln the spectrum it
shows a single vivid yellow band due to the
combustion of the metal, and in like manner
every other combination in which the same
metal exists, will always produce the same
result ; so that, by the presence or absence of
the particular band of colour that belongs to
sodium, the presence of that metal can be
detected with the most perfect certainty. What
is true of sodium, is true, in like manner, of
calcium, strontium, and other such substances ;
also of copper and other metals, each of which
communicates a colour to flame, and exhibits
coloured bands in the spectrum.

If an alloy of metals be burnt, the colours and
bands of the different component metals are
detected, and if colours and bands are found in
the spectrum, derived from coloured flame not
belonging to any known substance, it becomes
certain that there exists and is present in the
flame some substance hitherto unknown.

The prismatic spectrum, and a knowledge of
the various colours and dark bands of which
it is made up, and the various changes which
it exhibits under the action of differently
composed lights, thus becomes, in the hands of the
accurate experimenter of the present day, a
means of effecting minute and accurate chemical
analysis, but it has been found capable of doing
much more even than this.

The presence of a particular band of colour
in a spectrum obtained from coloured flame,
proves the existence of a certain metal or other
elementary substance in intense combustion; but
what does the absence of a particular band mean,
or, in other words, to what are owing the dark
lines originally discovered in the solar spectrum
by Wollaston and Fraunhofer? To answer this
query requires a little further inquiry. It might
be thought, as we have already hinted, that
the material known to give a band of colour
in the part of the solar spectrum, crossed by
a dark band, was absent from the source
of light; but this is not necessarily the case,
for if a lime light is burnt, and a continuous
spectrum thus formed, and we burn a little salt
between the light and the prism, the place where
the sodium band would be if there were no
lime light, instead of becoming brighter, is
immediately marked by a dark line. To be
satisfied of this, we have only to take away the lime
light when the bright yellow sodium band is at
once beautifully shown. A dark band, then, is