a
signal reaches the receiving station the current in the telephone
receiver is sudden reduced; and in the time between signals the leak
across the grid condenser brings the tube back to a condition where it
can receive the next signal. While the sending key is depressed the
current in the receiver is decreasing and increasing once for every
signal which is being transmitted. For each decrease and increase in
current the diaphragm of the telephone receiver makes one vibration.
What the listener then hears is a musical note with a frequency
corresponding to that number of vibrations a second, that is, a note
with a frequency of one thousand cycles per second. He hears a note of
frequency about that of two octaves above middle _C_ on the piano.
There are usually other notes present at the same time and the sound is
not like that of any musical instrument.

[Illustration: Fig 91]

If the key is held down a long time for a dash the listener hears this
note for a corresponding time. If it is depressed only about a third of
that time so as to send a dot, the listener hears the note for a
shorter time and interprets it to mean a dot.

In Fig. 91 I have drawn a sketch to show the e. m. f. which the signals
from a spark set impress on the grid of a detector and to show how the
plate current varies if there is a condenser and leak in the grid
circuit. I have only shown three signals in succession. If the operator
sends at the rate of about twenty words a minute a dot is formed by
about sixty of these signals in succession.

The frequency of the alternations in one of the little signals will
depend upon the wave length which the sending operator is using. If he
uses the wave length of 600 meters, as ship stations do, he will send
with a radio frequency of 500,000 cycles a second. Since the signals are
at the rate of a thousand a second each one is made up of 500 complete
cycles of the current in the antenna. It would be impracticable
therefore to show you a complete picture of the signal from a spark set.
I have, however, lettered the figure quite completely to cover what I
have just told you.

If the grid-condenser and its leak are so chosen as to work well for
signals from a 500-cycle spark set they will also work well for the
notes in human speech which are about 1000 cycles a second in frequency.
The detecting circuit will not, however, work so well for the other
notes which are in the human voice and are necessary to sp