of Duplex Telegraphy* 531 



the first method, where the adjustment can only be carried on 

 in one branch by varying d in increments *. Which of the two 

 methods, however, is to be finally chosen, depends on other 

 considerations, which will become clear further on. We know 

 now that both these modes of adjustment are convenient and 

 practicable, and contain immediate balance without special con- 

 ditions. In fact in this respect the compensation method is 

 preferable to the differential method, where immediate balance 

 by varying resistances could only be obtained when varying 

 the four branches simultaneously according to a fixed relation. 



Rapid approximation of the two functions S and D towards 



zero. 

 On account o£f=w + /3 we have 



where 



A=EVa-K\Wk 



Now suppose A = ; then this equation may be disturbed by 

 K, R, X, v, a, or b varying, a and b are wire resistances which 



* It has been suggested to adjust balance by a continuous variation 

 of resistance, as, for instance, by moving a contact-point along a thin pla- 

 tinum wire in the same manner as Dr. W. Siemens has done it in his 

 bridge employed for comparing accurately comparatively small resistances. 

 It is, however, scarcely necessary to point out that such a method, if ap- 

 plied for duplex working, must result in a failure, at all events so long as 

 electromagnetic instruments are used for producing the signals. For in 

 such a case the resistance of any branch, no matter what special duplex 

 method may be employed, must bear a certain ratio to the given re- 

 sistance of the line in order to get the signals with sufficient force. This 

 ratio,, as my investigations have shown, is by no means a small one; and 

 hence the resistances of all branches, even for a short line, cannot be made 

 small. Therefore the platinum wire constituting part of one or two branches 

 of the duplex method employed, must also offer a considerable resistance, 

 i. e. must be of great length. Hence to alter such a large resistance conti- 

 nuously and perceptibly, as is indicated by the balance-disturbance, must 

 evidently involve a considerable movement of the contact-point, which, 

 even choosing the thinnest possible wire and the shortest telegraph-line, 

 becomes already for the daily variation so large as to make its application 

 impossible. Unless another material of much higher specific resistance 

 than platinum wire can be found, which at the same time allows of the 

 sliding contact being made securely, the adjustment of balance by a conti- 

 nuous variation of resistances must be dispensed with. Such a material 

 does not appear to exist. I thought of acting on Phillips's suggestion to 

 use pencil-marks for the adjustable resistance ; and although I found that 

 pencil resistances can be adjusted very accurately, and can be enclosed in 

 a very small space, and that they keep sufficiently constant, it is difficult, 

 if not impossible, to alter them by a sliding contact ; the " TJebergangs- 

 tmderstand " is too variable and too great. Besides, if the contact is made 

 with sufficient pressure, its sliding along alters the thickness of the pencil- 

 mark, and hence the resistances become inconstant and uncertain, 



2N2 



