Oady and Arnold — Electric Arc. 407 



a resistance of about 12 ohms, was connected in series with a 

 mica condenser of one-third m.f. capacity, the two forming 

 a shunt across the arc. Recent experiments of Austin* call 

 attention to the importance of having a low resistance in series 

 with the condenser. * Nevertheless, the resistance of our bolom- 

 eter did not prevent the production of measurable oscillations 

 as will be seen. The platinum wire was connected as one 

 arm of a Wheatstone bridge. 



At the beginning of the experiment the arc was struck and 

 the bridge balanced with battery current flowing, the condenser 

 circuit remaining open. When the condenser switch was 

 closed, the resulting deflection of the galvanometer in the 

 bridge was taken as a measure of the intensity of the oscilla- 

 tions. Current for the arc was taken from the 240 volt mains. 



As a preliminary test, carbon terminals were employed. No 

 deflection could be observed when the arc was quiet, but the 

 light went violently off the scale when the arc hissed. On the 

 quiet state of the carbon arc oscillations are not to be expected, 

 on account of the small self-inductance and large resistance of 

 the shunt circuit. 



§36. Iron terminals in free air were next used. The self- 

 inductance of our shunt circuit cannot have been appreciably 

 greater than Lecher's. Yet we found it possible to make the 

 iron arc "sing'" audibly, even on the first stage, with the cur- 

 rent a little below one ampere. The tone was a high-pitched 

 squeak or whistle. Inserting a self-inductance of 0*000 04 

 henry caused a lowering of pitch, though not as much as would 

 be expected, showing that under these conditions the simple 

 formula for period, 



T = 2tta/X(7 



does not hold. 



When the arc was singing, the heating of the bolometer 

 threw the light off the scale ; the arc current increased, while 

 the voltage dropped several volts, as has been observed by 

 Austin (1. c). 



Whether the arc sang audibly or not, the galvanometer was 

 always found to be deflected, but, in general much more on the 

 second than on the first stage. As the oscillations were gener- 

 ally of too high frequency to produce an audible sound, no 

 direct estimate of their period was made, nor could their 

 dependence on current and length of arc be observed. The 

 galvanometer deflections depend on the total quantity of elec- 

 tricity passing per second through the shunt circuit, and this is 

 a function not only of frequency, but also of amplitude and 

 wave-form. We found the deflections to increase as the cur- 

 *Bull. Bur. Stand., iii, 325, 1907. 



