nder, and if the atoms are drawn apart, pull them again together,
causing them, before coming to rest, to quiver for a certain time at a
certain definite rate determined by the strength of the spring. Now
the volatilised metal which gives us one bright band is to be figured
as having its atoms united by springs all of the same tension, its
vibrations are all of one kind. The metal which gives us two bands
may be figured as having some of its atoms united by springs of one
tension, and others by springs of a different tension. Its vibrations
are of two distinct kinds; so also when we have three or more bands we
are to figure as many distinct sets of springs, each capable of
vibrating in its own particular time and at a different rate from the
others. If we seize this idea definitely, we shall have no difficulty
in dropping the metaphor of springs, and substituting for it mentally
the forces by which the atoms act upon each other. Having thus far
cleared our way, let us make another effort to advance.

A heavy ivory ball is here suspended from a string. I blow against
this ball; a single puff of my breath moves it a little way from its
position of rest; it swings back towards me, and when it reaches the
limit of its swing I puff again. It now swings further; and thus by
timing the puffs I can so accumulate their action as to produce
oscillations of large amplitude. The ivory ball here has absorbed the
motion which my breath communicated to the air. I now bring the ball
to rest. Suppose, instead of the breath, a wave of air to strike
against it, and that this wave is followed by a series of others which
succeed each other exactly in the same intervals as my puffs; it is
obvious that these waves would communicate their motion to the ball
and cause it to swing as the puffs did. And it is equally manifest
that this would not be the case if the impulses of the waves were not
properly timed; for then the motion imparted to the pendulum by one
wave would be neutralised by another, and there could not be the
accumulation of effect obtained when the periods of the waves
correspond with the periods of the pendulum. So much for the
particular impulses absorbed by the pendulum. But if such a pendulum
set oscillating in air could produce waves in the air, it is evident
that the waves it would produce would be of the same period as those
whose motions it would take up or absorb most completely, if they
struck against it. P