t}_
so as to bring its windings more and more parallel to those of _L_.
If we want _L_{t}_ to have a large effect on _L_ its windings
should be parallel and also in the same direction just as they were in
Fig. 54 of Letter 13 to which we just referred. As we approach nearer to
that position the current in _L_{t}_ induces more and more e. m. f.
in coil _L_. For some position of the two coils, and the actual
position depends on the tube we are using, there will be enough effect
from the plate circuit upon the grid circuit so that there will be
continuous oscillations.

[Illustration: Fig 92]

We want to stop just short of this position. There will then be no
continuous oscillations; but if any changes do take place in the plate
current they will affect the grid. And these changes in the grid voltage
will result in still larger changes in the plate current.

Now suppose that there is coming into the detector circuit of Fig. 92 a
radio signal with, speech significance. The current in the plate circuit
varies accordingly. So does the current in coil _L_{t}_ which is in
the plate circuit. But this current induces an e. m. f. in coil _L_
and this adds to the e. m. f. of the incoming signal so as to make a
greater variation in the plate current. This goes on as long as there is
an incoming signal. Because the plate circuit is coupled to the grid
circuit the result is a larger e. m. f. in the grid circuit than the
incoming signal could set up all by itself.

You see now why I said the tube talked to itself. It repeats to itself
whatever it receives. It has a greater strength of signal to detect than
if it didn't repeat. Of course, it detects also just as I told you in
the preceding letter.

In adjusting the coupling of the two coils of Fig. 92 we stopped short
of allowing the tube circuit to oscillate and to generate a high
frequency. If we had gone on increasing the coupling we should have
reached a position where steady oscillations would begin. Usually this
is marked by a little click in the receiver. The reason is that when the
tube oscillates the average current in the plate circuit is not the same
as the steady current which ordinarily flows between filament and plate.
There is a sudden change, therefore, in the average current in the plate
circuit when the tube starts to oscillate. You remember that what
affects the receiver is the average current in the plate circuit. So the
receiver diaphragm suddenly changes posi