Cacly and Yinal — Electric Arc. 99 



first discovered. Black oxide of iron, Fe 3 4 , gave feeble 

 oscillations. 



§12. When the e.m.f. of the supply was gradually reduced, 

 the oscillations became less intense, and a larger current was 

 found necessary. With a supply of only 145 volts a very 

 feeble deflection of the bolometer was still perceived, the 

 discharge current being 0*8 amp. This deflection may have 

 been due to mere irregularities in the burning of the arc. At 

 any rate, there seemed to be no critical discharge potential at 

 which oscillations suddenly began. 



III. Experimental Evidence of Glow -Arc Oscillations. 



§ 13. That a gradual transition can be observed from a 

 visible slow change between arc and glow in air, through the 

 more rapid pulsating discharge recorded by photography, to 

 the high-frequency oscillations in hydrogen and acetone, is 

 a priori evidence that in these last oscillations the change 

 between arc and glow is still taking place. The alternative 

 hypothesis, that of an intermittent spark discharge, is, of 

 course, obvious, and indeed it may be that a gradual transition 

 from glow-arc pulsations to a pure spark discharge, by raising 

 the e.m.f. and at the same time decreasing the current, could 

 be accomplished. That the present phenomenon is not a spark 

 discharge in the ordinary sense is rendered probable by the 

 small damping of the oscillations (§29). Moreover, if this were 

 a spark discharge, the train of waves that constituted each 

 spark dying down nearly to zero before the next discharge 

 passed, then in order to explain the observed fact that the 

 effective total current is several times as large as the mean 

 direct current, we should have to assume a very large initial 

 amplitude. But this would be hard to reconcile with the 

 relatively low impressed e.m.f. 



§ 11. According to the explanation of the oscillations here 

 advanced, the potential drop across the discharge must rise to 

 that characteristic of a glow- discharge once during eacli period. 

 It was determined to test this, and at the same time to answer 

 the question as to whether the oscillations might be due to a 

 rapid succession of short sparks. For in the latter case, the 

 potential difference would be expected to rise considerably 

 above that of a glow discharge. 



Considering the high frequency of the oscillations, the only 

 available method for measuring the maximum potential differ- 

 ence seemed to be to connect across the discharge a calibrated 

 spark-gap. That this method is applicable even when the 

 frequency is as high as one million follows from the work 



