A one-half microfarad condenser is placed in the receiver circuit at each station so that the line will not be tied up should some subscriber inadvertently leave his receiver off its hook.  This permits the passage of voice currents, but not of the direct currents used in stepping the relays or in releasing them.

The circuit of Fig. 193 is somewhat simplified from that in actual practice, and it should be remembered that the hook switch, which is not shown in this figure, controls in the usual way the continuity of the receiver and the transmitter circuits as well as of the generator circuits, the generator being attached to the line as in an ordinary telephone.

Broken-Line System.  The broken-line method of accomplishing selective signaling and locking-out on telephone party lines is due to Homer Roberts and his associates.

[Illustration:  Fig. 194.  Roberts Latching Relay]

To understand just how the principles illustrated in Figs. 186 and 187 are put into effect, it will be necessary to understand the latching relay shown diagrammatically in its two possible positions in Fig. 194, and in perspective in Fig. 195.  Referring to Fig. 194, the left-hand cut of which shows the line relay in its normal position, it is seen that the framework of the device resembles that of an ordinary polarized ringer.  Under the influence of current in one direction flowing through the left-hand coil, the armature of this device depresses the hard rubber stud 4, and the springs 1, 2, and 3 are forced downwardly until the spring 2 has passed under the latch carried on the spring 5.  When the operating current through the coil 6 ceases, the pressure of the armature on the spring 1 is relieved, allowing this spring to resume its normal position and spring 3 to engage with spring 2.  The spring 2 cannot rise, since it is held by the latch 5, and the condition shown in the right-hand cut of Fig. 194 exists.  It will be seen that the spring 2 has in this operation carried out just the same function as the switch lever performed as described in connection with Figs. 186 and 187.  An analysis of this action will show that the normal contact between the springs 1 and 2, which contact controls the circuit through the relay coil and the bell, is not broken until the coil 6 is de-energized, which means that the magnet is effective until it has accomplished its work.  It is impossible, therefore, for this relay to cut itself out of circuit before it has caused the spring 2 to engage under the latch 5.  If current of the proper direction were sent through the coil 7 of the relay, the opposite end of the armature would be pulled down and the hard rubber stud at the left-hand end of the armature would bear against the bent portion of the spring 5 in such manner as to cause the latch of this spring to release the spring 2 and thus allow the relay to assume its normal, or unlatched, position.