D doubles the strength
of the current traversing wire A. This is sufficient to not only
neutralize the magnetism which the current in wire B would tend to set
up, but also--by reason of the excess of current in wire A--to make the
bar a magnet whose polarity would be determined by the direction of the
flow of current around it.

In the arrangement shown in Fig. 4 the batteries are so connected that
current flow is in the same direction, thus doubling the amount of
current flowing through wire A. But suppose the batteries were
so connected that the current from each set flowed in an opposite
direction? The result would be that these currents would oppose and
neutralize each other, and, therefore, none would flow in wire A.
Inasmuch, however, as there is nothing to hinder, current would
flow from battery C through wire B, and the bar would therefore be
magnetized. Hence, assuming that the relay is to be actuated from
the distant end, D, it is in a sense immaterial whether the batteries
connected with wire A assist or oppose each other, as, in either case,
the bar would be magnetized only through the operation of the distant
key.

A slight elaboration of Fig. 4 will further illustrate the principle of
the differential duplex. In Fig. 5 are two stations, A the home end,
and B the distant station to which a message is to be sent. The relay at
each end has two coils, 1 and 2, No. 1 in each case being known as the
"main-line coil" and 2 as the "artificial-line coil." The latter, in
each case, has in its circuit a resistance, R, to compensate for the
resistance of the main line, so that there shall be no inequalities
in the circuits. The artificial line, as well as that to which the two
coils are joined, are connected to earth. There is a battery, C, and a
key, K. When the key is depressed, current flows through the relay
coils at A, but no magnetism is produced, as they oppose each other. The
current, however, flows out through the main-line coil over the line and
through the main-line coil 1 at B, completing its circuit to earth
and magnetizing the bar of the relay, thus causing its armature to be
attracted. On releasing the key the circuit is broken and magnetism
instantly ceases.

It will be evident, therefore, that the operator at A may cause the
relay at B to act without affecting his own relay. Similar effects would
be produced from B to A if the battery and key were placed at the B end.

If, therefore, like instru