Mr. 0. Heaviside on Duplex Telegraphy, 43 



ohms, which could be increased to 3860 or diminished to 1760 

 ohms without interrupting the working. At the other end the 

 balancing resistance could be varied from 3000 to 6000 ohms 

 without interfering with rapid signalling. The variation 

 allowable above balance is always much greater than below., 

 because the interfering current is inversely proportional to -the 

 resistance external to the battery, which is increased when the 

 balancing resistance is increased. In the above example the 

 line was fairly insulated. When the insulation falls, the effect 

 is to strengthen the sent and weaken the received currents ; 

 consequently the interfering currents bear a larger ratio to the 

 received currents for a given change of balance ; and the balance 

 therefore requires nicer adjustment. The extreme case is 

 reached with the very low insulation which prevails in this 

 country in continuous wet weather, when not much difference 

 can be detected between the resistance of the wire whether it 

 is insulated or to earth at one end. Under such circumstances 

 a very small change of balance is sufficient to upset the work- 

 ing. The ratio of the interfering to the received currents 

 may be diminished ad lib. by increasing the resistance of the 

 apparatus, or more simply by inserting a constant resistance 

 in the main circuit. As, however, it is only when the insula- 

 tion is very bad and the received current very weak that the 

 interfering current due to inexact balance attains such a pro- 

 portional strength as to mutilate the signals, the increase in 

 the resistance of the apparatus would be an evil rather than a 

 benefit, on account of the reduction in the strength of the re- 

 ceived signals, already very weak, that would ensue. 



The two other systems described in my former paper (Phil. 

 Mag. June 1873) are not likely, in accordance with the prin- 

 ciple of the survival of the fittest, to come into practical use ; 

 and it is therefore unnecessary to enter into details concerning 

 them. But this I may observe, that in both of them the ar- 

 rangement which produces the strongest signals at both sta- 

 tions is also the most sensitive balance. That this should be 

 the case in four different systems, renders it probable that it is 

 universally true for all duplex systems, in which some kind of 

 balance is concerned. 



