he spectrum. A few trials show that this
intensification of the D lines is due to the vapour of sodium arising
from the salt burning in the lamp through which the sunlight has passed.

It is quite impossible that this marvellous connection between sodium
and the D lines of the spectrum can be merely casual. Even if there were
only a single line concerned, it would be in the highest degree unlikely
that the coincidence should arise by accident; but when we find the
sodium affecting both of the two close lines which form D, our
conviction that there must be some profound connection between these
lines and sodium rises to absolute certainty. Suppose that the sunlight
be cut off, and that all other light is excluded save that emanating
from the glowing vapour of sodium in the spirit flame. We shall then
find, on looking through the spectroscope, that we no longer obtain all
the colours of the rainbow; the light from the sodium is concentrated
into two bright yellow lines, filling precisely the position which the
dark D lines occupied in the solar spectrum, and the darkness of which
the sodium flame seemed to intensify.

We must here endeavour to remove what may at first sight appear to be a
paradox. How is it, that though the sodium flame produces two _bright_
lines when viewed in the absence of other light, yet it actually appears
to intensify the two _dark_ lines in the sun's spectrum? The explanation
of this leads us at once to the cardinal doctrine of spectrum analysis.
The so-called dark lines in the solar spectrum are only dark _by
contrast_ with the brilliant illumination of the rest of the spectrum. A
good deal of solar light really lies in the dark lines, though not
enough to be seen when the eye is dazzled by the brilliancy around. When
the flame of the spirit-lamp charged with sodium intervenes, it sends
out a certain amount of light, which is entirely localised in these two
lines. So far it would seem that the influence of the sodium flame ought
to be manifested in diminishing the darkness of the lines and rendering
them less conspicuous. As a matter of fact, they are far more
conspicuous with the sodium flame than without it. This arises from the
fact that the sodium flame possesses the remarkable property of cutting
off the sunlight which was on its way to those particular lines; so
that, though the sodium contributes some light to the lines, yet it
intercepts a far greater quantity of the light that would oth