August 22, 1907J 



NA TURE 



429 



shunt circuit to an arc. At the moment of connection a 

 current flows from the arc circuit into the condenser circuit, 

 which tends to reduce ihe current flowing through the arc. 

 This reduction of the current through the arc tends to 

 raise the potential difference between its terminals, and 

 causes still more current to flow into the condenser circuit, 

 and I now have a condenser charged above the normal 

 voltage of the arc. The condenser, therefore, begins to dis- 

 charge through the arc, which increases the arc current 

 and decreases the potential difference, so that the condensei 

 discharges too much ; the reverse process then sets in ; the 

 condenser becomes successively overcharged and under- 

 charged, due to the fact that, instead of the potential 

 difference between the terminals of the arc remaining con- 

 stant and allowing the condenser to settle down with its 

 proper corresponding charge, the potential difference 

 actually decreases when the condenser is discharged and 

 increases when it is charging, so as to help to keep up 

 Ihe flowing backwards and forwards of the current in- 

 definitely. 



The oscillcgraph wave forms show what is going on 

 very clearly, and ihey show that in general the swing of 

 Ihe current in the condenser circuit attains such a magni- 

 tude that when the condenser is charging it takes the 

 u-hole cf the current away from the arc, so as to make 

 the arc, although burning on a direct current circuit, a 

 pulsatory arc. The pulsation of the current through the 

 arc causes the vapour colunm to grow bigger and smaller, 

 and the light to vary. When the vapour column grows 

 bigger and" smaller it displaces the air around it and pro- 

 duces a note the pilch of which is determined by the 

 frequency of the current in the shunt circuit. 



The values of the capacities of a series of condensers 

 have been calculated by Kelvin's formula to give the fre- 

 quencies corre.sponding with a musical octave, and the 

 nearest values in an ordinary laboratory box of condensers 

 have been taken and connected to a keyboard. The result 

 shows how nearly Kelvin's law is obeyed. 



With this apparatus I can demonstrate the importance 

 of tuning in electrical t-ircuits, and perform electrically 

 some e.xperiments which I have already performed 

 mechanically earlier this evening. I use the large coil 

 which form's the self-induction in the circuit shunting the 

 arc as a transmitting circuit for wireless telegraphy u\ 

 the magnetic induction or Preece method ; and I have 

 a receiving circuit consisting of a coil of wire connected 

 to a small lamp, and not connected in any way to the 

 transmitting circuit. At a certain short distance between 

 the transmitting coil and the receiving coils, the indicating 

 lamp lights if I cause my arc to sound any one of the 

 r.otes of the octave, and so produce an alternating current 

 of corresponding frequency in the transmitting coil. If I 

 now tune the receiving circuit, by connecting a condenser 

 in it, the lamp on the receiving circuit will light at about 

 five times the distance ; but it will only light when one 

 definite note is sounded by the arc. These are the two 

 distinct advantages of tuning, namely, greater distance 

 and syntony, or responding to only one definite note. 



For wireless telegraphy by means of Hertzian waves, 

 based on my arc method, we require much higher fre- 

 quencies in the shunt circuit. If we attempt to obtain 

 (his higher frequency from the ordinary arc burning 

 between solid carbons in air, we find that above a certain 

 limit the oscillations will no longer take place. This is 

 due to the fact that we are varying the current through 

 the arc at this higher frequency too quickly for an increase 

 in current to be accompanied by a decrease in potential 

 difference. I have demonstrated that if I only vary the 

 current through the ordinary current arc sufficiently rapidly, 

 then an increase in current is accompanied by a propor- 

 tionate increase in the potential difference, and the arc 

 behaves just like an ordinary resistance. If we work with 

 very small current arcs, we can obtain high-frequency 

 musical arcs burning in air either between carbon or metal 

 electrodes. 



In a paper read before the International Electrical Con- 

 gress at .St. Louis in H)04, Mr. Poulsen showed that, by 

 placing the arc in a flame, it was possible to obtain highci 

 frequencies than when the arc was burning in air. Hol- 

 lowing this up, Mr. Poulsen came to the conclusion that 

 the best results were obtained when the src was burning 



xo T973. '^'O^-- 76] 



in hydrogen, or a gas containing hydrogen ; and he further 

 added a magnetic field around the arc somewhat similar 

 to that which has been previously used by Elihu Thomson. 



The arc burning in coal gas in a powerful transverse 

 magnetic field was used by Poulsen in his early experi- 

 ments to produce the high-fret^uency current necessary fcr 

 wireless telegraphy between Lyngby and Esbjerg, in Den- 

 mark. This apparatus has been further improved, and is 

 now employed by the Amalgamated Radio-Telegraph Com- 

 pany in their station at CuUcrcoats and the otner stations 

 that they are erecting. 



In both the arc and the spark methods of wireless 

 telegraphy we employ a high-frequency alternating current 

 in the aerial conductor. The essential difference between 

 the two methods lies in the fact that with the spark method 

 our alternating current in the aerial conductor first in- 

 creased to a maximum value and then dies away rapidly, 

 making only a limited number of oscillations, whereas in 

 the arc method the oscillations are maintained continuously 

 of unvarying amplitude. 



With the arc method we are further able to choose 

 the number of consecutive oscillations which make up 

 each signal sufficiently great to obtain the very best syn- 

 tony. On the other hand, improvement in the arrange- 

 ment and construction of the apparatus for the spark 

 method has so increased the number of oscillations 

 corresponding with each spark that it may be that we shall 

 be able to obtain a sufficient number in each train to give 

 as good syntony by this method as that obtained with the 

 arc method. 



The arc method seems eminently suitable for very high 

 speeds of working. As the oscillations are quite continu- 

 ous, we can cut them up into groups to form the dots 

 and dashes of the Morse alphabet, just as if we w-ere 

 working with a continuous current such as is used on 

 land lines, so that there seems no reason w^hy as high a 

 speed of working should not be obtained from the arc 

 method of wireless telegraphy as is obtainable by auto- 

 matic signalling on land lines ; for it is to be noted that 

 the dot or shortest signal of the Morse alphabet, even at 

 a speed of three or four hundred words per minute, will 

 last long enough to consist of many hundreds of oscil- 

 lations of the current in the aerial, so that there will be 

 plenty of oscillations in the group forming the dot to give 

 good syntony. 



Turning to the spark method for high working speeds, 

 we find a difficulty in that the dot of the Morse alphabet 

 must at least occupy the average time required to charge 

 the condenser or aerial and produce one spark, and pie- 

 ferably sufficiently long for several. We are therefore 

 obliged in the spark method to use a high rate of sparking 

 for "high-speed signalling. This difficulty has not become 

 very serious with the present low speeds of sending. 

 When we come to use considerable amounts of power to 

 transmit messages over long distances, and we also require 

 a high speed of working, the practical difficulty in con- 

 structing apparatus suitable for sufficiently rapid sparking 

 will become serious. 



Mr. Marconi in 1905 claimed to have already reached 

 a speed of 100 words per minute by the spark method, 

 and lately there has appeared in the technical Press 

 examples of high-speed signalling by the British Post Office 

 over a distance of fifteen miles in which readable 

 signals were received at a speed of seventy words per 

 minute. 



Turning to the receiving end, almost all the receivers 

 that have been used in the spark method can be equally 

 well used for the arc method ; for it must be remembered 

 that the transmission in either case is affected by Hertzian 

 waves traversing space, and that the only fundamental 

 difference consists in the number of oscillations in each 

 train of waves. It must be noted, however, that in those 

 methods in which a telephone receiver is used it is neces- 

 sarv to break up the continuous oscillations of the arc 

 method into groups succeeding one another sufficiently 

 rapidly to produce an audible sound in the receiver ; for 

 in the spark method the sounds we hear in the receiver 

 correspond with the succession of impulses of the diagram, 

 one for each spark at the transmitter. This chopping up 

 of the continuous wave-train so as to produce audible 

 signals in the receiving apparatus can be done either at 



