ts of the plate to draw electrons away from the filament.

[Illustration: Pl. VIII.--To Illustrate the Mechanism for the Production
of the Human Voice.]

Then the plate current is reduced practically to zero.

That's the way to arrange an audion so that the incoming signal makes
the largest possible change in plate current. We can tell if there is an
incoming signal because it will "block" the tube, as we say. The
plate-circuit current will be changed from its ordinary value to almost
zero in the short time it takes for a few cycles of the incoming signal.

We can detect one signal that way, but only one because the first signal
makes the grid permanently negative and blocks the tube so that there
isn't any current in the plate circuit and can't be any. If we want to
put the tube in condition to receive another signal we must allow these
electrons, which originally came from the filament, to get out of their
trapped position and go back to the filament.

[Illustration: Fig 90]

To do so we connect a very fine wire between plates 1 and 2 of the grid
condenser. We call that wire a "grid-condenser leak" because it lets the
electrons slip around past the gap. By using a very high resistance, we
can make it so hard for the electrons to get around the gap that not
many will do so while the signal is coming in. In that case we can leave
the leak permanently across the condenser as shown in Fig. 90. Of
course, the leak must offer so easy a path for the electrons that all
the trapped electrons can get home between one incoming signal and the
next.

One way of making a high resistance like this is to draw a heavy pencil
line on a piece of paper, or better a line with India ink, that is ink
made of fine ground particles of carbon. The leak should have a very
high resistance, usually one or two million ohms if the condenser is
about 0.002 microfarad. If it has a million ohms we say it has a
"megohm" of resistance.

This method of detecting with a leaky grid-condenser and an audion is
very efficient so far as telling the listener whether or not a signal is
coming into his set. It is widely used in receiving radio-telephone
signals although it is best adapted to receiving the telegraph signals
from a spark set.

I don't propose to stop to tell you how a spark-set transmitter works.
It is sufficient to say that when the key is depressed the set sends out
radio signals at the rate usually of 1000 signals a second. Every time