428 



NA rURE 



[August 22, 1907 



It i lake two electrodes sutlicicntly far apart in air, and 

 gradually raise ilie electrical pressure between them, the 

 rirst indication that anything is going to happen is the 

 formation of fine violet aigrettes on the more pointed or 

 rougher parts of the electrodes. This is known as the 

 brush discharge. Hy gradually raising the pressure, this 

 brush discharge extends further out into the air, until 

 finally the air between the two electrodes becomes so 

 -trained that it breaks down and the real spark passes. 



The long thin spark that occurs in this case is not very 

 suitable for wireless telegraphy, as its resistance is too 

 high. Ordinary lightning flashes are good examples of long 

 sparks on a very large scale. If instead of working with 

 the electrodes far apart they are placed nearer together, 

 and if the electrical pressure is supplied from a very power- 

 ful source, then dfrectly the spark passes it forms a thick 

 discharge having the appearance of a flame in which the 

 nitrogen of the air is actually being burnt ; a process which, 

 it is hoped, in the future may have inmiensc importance in 

 the supply of artificial nitrates for agriculture. This flame- 

 like discharge has a low electrical resistance, but has the 

 elTcct that it so heats or modifies the air that it is difiiculi 

 to get the air to insulate again, after one discharge, ready 

 for the next. 



If a large quantity of electricity is discharged through 

 the spark gap, and if the spark lasts a very short time 

 compared with the interval between successive sparks, then 

 a highly-conducting spark can be obtained, as well as a 

 good insulation between the sparking terminals when no 

 discharge is passing. 



In order to help to bring the gap back to its insulating 

 condition after each discharge, many devices are employed, 

 such as subdividing the spark into several shorter .sparks, 

 coaling the electrodes, blowing air across the spark gap. 

 &c. When the condenser or antenna discharges through tiie 

 spark gap, oscillations are set up which radiate Hertzian 

 waves. 



In practice in wireless telegraphy it is difficult to obtain 

 a large number of oscillations during each discharge as 

 corresponding with each oscillation ; the antenna radiates 

 energy. A large number of oscillations means, if we keep 

 amplitude of each the same, that we are radiating a large 

 quantity of energy. Besides this radiated energy, which 

 is useful for transmitting messages, there is also energy 

 wasted in heat in the spark gap, in the conductors, in 

 the glass or other insulation of the condensers. It is this 

 useless part which we require to make as small as possible. 



I have lately had an opportunity to determine how many 

 oscillations actually take place in a certain wireless trans- 

 . mission. The experiment was made by photographing the 

 spark as seen in a mirror rotated at a very high speed, 

 and it was found that each spark consisted of nine or 

 ten complete oscillations. 



If all the oscillations had been of equal strength or ampli- 

 tude, and if the receiving circuit had been similar to my 

 pendulum In my mechanical model, then there would he 

 very little to be gained by Increasing the number of oscil- 

 lations. As the oscillations die away in the spark method, 

 two or three times this number would probably be required 

 for the best effect. .\s a matter of experiment, verv good 

 tuning was obtained w^lth the wireless transmission referred 

 to above. 



.As an example of the sharpness of tuning obtainable by 

 the spark method, the following test carried out on the 

 Lodge-Muirhead installation at Hythe may be of interest. 



The station at Hythe had to receive messages from 

 Elmer's End at a distance of fifty-eight miles over land, 

 in spite of the fact that the Admiralty station at Dover, 

 only 9J miles distant, was transmitting as powerfully as 

 It could, in order to produce interference, and that the 

 regular communications were going on in the Channel 

 between the shipping. It was found possible with a 

 difference of wave-length of 6 per cent, to cut out the inter- 

 ference from the Dover station. 



In the arc method of producing continuous oscillations 

 we employ, as before, a condenser and self-induction ; but, 

 Instead of charging the condenser to a high voltage and 

 allowing It to discharge by means of oscillations which die 

 away, and then repe.-iting the process over and over again, 

 we actually maintain the condenser charging and discharg- 

 ing continuously without any intermission, so that we 



NO. T973, VOL. 76J 



practically obtained a high-frequency alternating current In 

 the aerial. 



To impress the difference on your minds, I have an 

 incandescent lainp, which 1 switch on and off rapidly about 

 ten times, and then after a short time I repeat the same 

 flickering of the light, and so on. The flickering of the 

 light corresponds with the oscillations in the ordinary spark 

 method, and the time spaces between the flick.-rs represent 

 the tiiTies during which the condenser or antenna Is being 

 charged ready to produce a fresh series of oscillations. In 

 practice we may have as many as, say, a couple of hundred 

 discharges of the condenser a second, and during each dis- 

 charge we may get, say, ten complete oscillations, each 

 oscillation lasting one-millionth of a second, if the wave- 

 length is 300 metres ; thus the total liine that the condenser 

 is discharging is only one one-hundred-lhousandth of a 

 second, or the five-hundredth part of the Interval of time 

 between tw-o successive discharges. My lamp here flickers 

 about five times per second, and makes ten flickers before it 

 goes out ; the total time that it is flickering Is two seconds, 

 and the time before it should start to flicker again to 

 correspond with the practical wireless case is therefore 

 1000 seconds, or rather more than a quarter i-t an hour. If 

 now I represent continuous oscillations, such as are 

 obtained by the arc method with this lamp, I shall simply 

 keep the lainp flickering continuously, and there will be 

 no intervals whatever. 



The arc method of producing continuous oscillations is 

 founded on my musical arc. In order to explain this 1 

 must demonstrate some of the properties of the direct- 

 current arc. If I vary the current flowing through the arc 

 very slowly and note the potential difference corresponding 

 with each value of the current, keeping everything else con- 

 stant, I obtain a curve generally spoken of as the charac- 

 teristic of the arc. These curves under different conditions 

 have been very thoroughly investigated by Mrs. .Ayrton. 



With the carbon arc between electrodes in air the voltage 

 decreases very rapidly when the current is gradually in- 

 creased, starling from very low values. As the current 

 becomes larger the rate- of decrease of the voltage becomes 

 less and less until it is, comparatively speaking, quite 

 stnall, with a currrint of ten or twelve amperes. With the 

 arc between metal electrodes similar results are obtained, 

 except that the discontinuity in the curves, called the 

 hissing point by Mrs. .Ayrton, takes place at very small 

 currents, generally well below one ampere. 



With arcs burning in hydrogen, Mr. Upson has found 

 that the curves are generally much steeper for the larger 

 values of the current than for the corresponding arcs 

 burning in air. This point is of great Importance as 

 explaining the value of the hydrogenic atinosphere used by 

 Poulsen and referred to later. 



In general, I may therefore say for the above arcs that 

 Increase in current through the arc is accompanied by 

 decrease of the potential difference between Its electrodes, 

 and vice vctsCi decrease of the current causes Increase in 

 the potential difference; on the other hand, certain arcs, 

 such as the mi, between cored carbons, behave in an oppo- 

 site manner, that Is to say, current and potential difference 

 increase and decrease together. 



I demonstrated In 1900 that if I connect between the 

 electrodes of a direct-current arc (or other conductor of 

 electricity for which an increase In current is accompanied 

 by a decrease in potential difference between the terminals) 

 a condenser and a self-induction connected in series, I 

 obtain in this shunt circuit an alternating current. I called 

 this phenomenon the musical arc. The frequency of the 

 alternating current obtained in this shunt circuit depends 

 on the value of the self-induction and the capacity of the 

 condenser, and may practically be calculated by Kelvin's 

 well-known formula. 



Besides the condition that an Increase of current must 

 be accompanied bv a decrease in potential difference. It is 

 necessary that the relative decrease in potential difference 

 produced by a given increase in current, that Is to say, 

 the steepness of the characteristic, shall exceed a certain 

 ininimum value which depends on the losses In the 

 shunt circuit. It is also necessary that an increase in 

 current shall be accompanied by a decrease in potential 

 difference, even when the current is varied very rapidly. 



Let us consider what tnk's place when I connect this 



