here is an
inductance and a condenser. Near the coil is another coil which has a
battery and a key in circuit with it. The coils are our old friends of
Fig. 33 in Letter 10. Suppose we close the switch _S_. It starts a
current through the coil _ab_ which goes on steadily as soon as it
really gets going. While it is starting, however, it induces an electron
stream in coil _cd_. There is only a momentary or transient current
but it serves to charge the condenser and then events happen just as
they did in the case where we charged the condenser with a battery.

[Illustration: Fig 49]

Now take away this coil _ab_ with its battery and substitute the
oscillator of Fig. 36. What's going to happen? We have two circuits in
which oscillations can occur. See Fig. 49. One circuit is associated
with an audion and some batteries which keep supplying it with energy so
that its oscillations are continuous. The other circuit is near enough
to the first to be influenced by what happens in that circuit. We say it
is "coupled" to it, because whatever happens in the first circuit
induces an effect in the second circuit.

Suppose first that in each circuit the inductance and capacity have such
values as to produce oscillations of the same frequency. Then the moment
we start the oscillator we have the same effect in both circuits. Let me
draw the picture a little differently (Fig. 50) so that you can see this
more easily. I have merely made the coil _ab_ in two parts, one of
which can affect _cd_ in the oscillator and the other the coil
_L_ of the second circuit.

But suppose that the two circuits do not have the same natural
frequencies, that is the condenser and inductance in one circuit are so
large that it just naturally takes more time for an oscillation in that
circuit than in the other. It is like learning to dance. You know about
how well you and your partner would get along if you had one frequency
of oscillation and she had another. That's what happens in a case like
this.

[Illustration: Fig 50]

If circuit _L-C_ takes longer for each oscillation than does
circuit _ab_ its electron stream is always working at cross
purposes with the electron stream in _ab_ which is trying to lead
it. Its electrons start off from one condenser plate to the other and
before they have much more than got started the stream in _ab_
tries to call them back to go in the other direction. It is practically
impossible under these conditions to get a