eech. For
example, if notes of about 2000 cycles a second are involved in the
speech which is being transmitted, the leak across the condenser will
not work fast enough. On the other hand, for the very lowest notes in
the voice the leak will work too fast and such variations in the signal
current will not be detected as efficiently as are those of 1000 cycles
a second.

You can see that there is always a little favoritism on the part of the
grid-condenser detector. It doesn't exactly reproduce the variations in
intensity of the radio signal which were made at the sending station. It
distorts a little. As amateurs we usually forgive it that distortion
because it is so efficient. It makes so large a change in the current
through the telephone when it receives a signal that we can use it to
receive much weaker signals, that is, signals from smaller or more
distant sending stations, than we can receive with the arrangement
described in Letter 14.

LETTER 18

AMPLIFIERS AND THE REGENERATIVE CIRCUIT

MY DEAR RECEIVER:

There is one way of making an audion even more efficient as a detector
than the method described in the last letter. And that is to make it
talk to itself.

Suppose we arrange a receiving circuit as in Fig. 92. It is exactly like
that of Fig. 90 of the previous letter except for the fact that the
current in the plate circuit passes through a little coil, _L_{t}_,
which is placed near the coil _L_ and so can induce in it an e. m.
f. which will correspond in intensity and wave form to the current in
the plate circuit.

If we should take out the grid condenser and its leak this circuit would
be like that of Fig. 54 in Letter 13 which we used for a generator of
high-frequency alternating currents. You remember how that circuit
operates. A small effect in the grid circuit produces a large effect in
the plate circuit. Because the plate circuit is coupled to the grid
circuit the grid is again affected and so there is a still larger effect
in the plate circuit. And so on, until the current in the plate circuit
is swinging from zero to its maximum possible value.

What happens depends upon how closely the coils _L_ and
_L_{t}_ are coupled, that is, upon how much the changing current in
one can affect the other. If they are turned at right angles to each
other, so that there is no possible mutual effect we say there is "zero
coupling."

Start with the coils at right angles to each other and turn _L_{