July 7, 19 10] 



NATURE 



27 



Ever since the invention of tlie siphon recorder, efforts 

 have been made to turn it into a relay, but two difficulties 

 had to be faced. The extreme feebleness of the received 

 signalling currents was such that they were incapable of 

 opening and closing a battery circuit so as to do useful 

 work in that circuit. 



The reason for this is that a certain force is required 

 to press the relay contacts together to complete the circuit 

 and a certain force to break the circuit when formed ; 

 these forces of " make " and " break " are too great for 

 the cable relay to supply under normal working conditions. 

 ^ The second difficulty was the want of definition in the 

 signals received to operate a relay ; they were too ill- 

 defined, and the zero line wandered too greatly to ensure 

 that a relay with a fixed mechanical zero would work 

 satisfactorily. 



These two difficulties were overcome by the invention of 

 my '■ drum " cable relay and my magnetic shunt. The 

 drum cable relay (Fig. 8) is very similar to the siphon re- 

 cordpr. It Is the same, so far as the suspended coil and 

 connecting fibres are concerned, but in place of the siphon 

 tube a relay contact arm is provided. 



The end of this arm is arranged to press upon the 

 surface of a revolving drum. The outer drum surface of 

 gold or silver is divided into three parts : a central insu- 

 lated portion, upon which the end of the contact arm 

 normally rests when no signals are received, and portions 

 one on each side of the central one. These outer divisions 

 are included in the circuit of a local batterv and 

 two post-office pattern relays. 



When the relay arm is deflected to one side or 

 the other, upon the receipt of the signal, it slides 

 or skates into contact with one or other of the 

 outer portions of the drum, and thus closes circu't 

 of the battery through one or other of the post- 

 office relays; this second relay is thus operated, 

 and in turn works a " sounder " key to re-trans- 

 mit the signal into a second cable. 



To reduce the electrical resistance that is found 

 to exist in the contact botwcen the relay pointer 

 and the revolving drum, and to allow a largf 

 current to pass, condensers are placed across to ^ 



short-circuit the contact. 



These short-circuiting condensers are verv 

 important to the proper working of the relay, as 

 without their aid very little current indeed could 

 be obtained in the local circuit to do useful work. 

 The cable relay is a delicate instrument, and 

 mechanical effects had to be produced by means 

 of energy four-millionths of that required to 

 produce one candle-power of an ordinary carbon 

 lamp. The operation of the relay throughout 

 is quite automatic and trustworthy, and no clerk 

 is required to supervise. 



The drum relay has two properties that peculiarly 

 fit it for cable work : — (i) the relay contact is always made, 

 because the contact arm ne\'er leaves the surface of the 

 drum ; (2) by the rotation of the drum, the friction between 

 the arm, to side motion, and the surface of the drum is 

 reduced in a most wonderful way, so that the arm may be 

 moved by the extremely feeble forces received at the end 

 of the cables. 



The relay has a fixed mechanical zero, the centre of 

 the insulated portion, to which the end of the arm must 

 return after every signal or group of signals, and the zero 

 of the electrical signals has been made by electrical adjust- 

 ment to coincide with the mechanical zero. If there were 

 not this coincidence there would be mutilation of the re- 

 transmitted signals. 



The working of the relay is complicated by the require- 

 ments of the service, which demand that a condenser should 

 be included in the suspended coil circuit. The object of 

 this condenser is to exclude the possibility of interference 

 from " earth " currents, which sometimes flow along the 

 caole. 



The presence of the " earth " current is due to outside 

 electrical influences, atmospheric or celestial. 



Now these " earth " currents, if allowed to flow through 

 the suspended coil, would produce deflections that would 

 interfere with the proper working of the relay. 



The magnetic shunt which is always placed across the 

 coil does shunt the " earth " current to a verv great 



extent, but does not always get rid of it, and so to make 

 matters sure the " unshunted " series or Varley condenser 

 is included in the system. 



The condenser, unfortunately, polarises or charges up 

 under a series of signalling impulses of the same polarity 

 or sign, and for this reason itself causes a wandering of 

 the electrical zero of the signals. We are therefore trying 

 to stop one kind of variable zero effect by a device that 

 produces another one of its own. 



The effect of the wandering zero due to the series con- 

 denser can be cured, because the wandering, unlike that of 

 the " earth " currents, follows a regular law, viz. the law 

 of the signals themselves. The relay produces the signals 

 and combination of signals in its local circuit, precisely 

 the same as the signals or combination sent through the 

 cable that work it, and are at the same time causing the 

 variable zero. Current is therefore taken from the local 

 circuit and passed through an electrical retarding device, 

 which is called the " local correction circuit," consisting 

 of a series of inductances and shunting resistances. The 

 local circuit is so adjusted in its value that the current at 

 the far end rises exactly as there is a drop in the received 

 signalling current through the series condenser. 



The correction current is passed through a separate 

 winding on the suspended coil of the relay, and produces 

 an effect on the coil exactly opposite to that produced on 

 the main winding by the variable zero itself, that is to 

 say, two variable zeros of equal strength but of opposite 



directions are superimposed on the suspended coil, and 

 thus neutralise one another. The variable zero of the 

 signals themselves is thus eliminated. 



Local correction is a very important part of the relay 

 adjustment, and cannot very well be dispensed with. 



The Eastern Telegraph Company generously lent me 

 their lines for a trial of my " high-speed " system of 

 working. The cable over which the tests have taken place 

 stretches from Porthcurnow in Cornwall to Gibraltar, and 

 is normally worked at 170 letters per minute, each way, 

 with the siphon recorder as receiver. With the new 

 method, using a special relay (Fig. 9), traffic has been 

 carried continuously, duplex, at 230 letters per minute. 

 On special trial runs, not carrying traffic, and not sending 

 into the cable at the receiving station, although on duplex 

 conditions, a speed of 2.80 letters per minute has been 

 obtained. 



The principle of operation is as follows. When a sub- 

 marine cable is forced much beyond its normal speed of 

 working, the quick-changing signals, such as make up 

 the letter c, are the first to fail, or in other words, 

 do not arrive with sufficient strength to work the 

 receiver. 



It was found on trial that allowing more of the current 

 from the cable to flow through the receiver, say by in- 

 creasing the size of the receiving condenser, the first and 



NO. 2123, VOL. 84] 



