Cady and Vinal — Electric Arc. 95 



electrodes, it is clear that the glow phase can begin promptly 

 after each reversal of current. 



The frequency is too inconstant to permit of an analysis of 

 the wave form by any of the ordinary types of oscillograph. 

 On the other hand, an estimate of the frequency on theoretical 

 grounds would require a knowledge of the resistance, self- 

 inductance and capacity not only of the circuit, but of the 

 discharge itself. To compute these factors would be difficult, 

 involving as it must an exact knowledge of the extent to 

 which the oscillations penetrate into the circuit. 



II. Peoduction and Appeaeance of Oscillations. 



§ 8. The oscillations when strong were detected by means of 

 a bolometer insulated from the arc circuit and 10 to 100 CU1 

 distant from it. In this case the greater part of the energy 

 received by the bolometer circuit was in the form of electro- 

 magnetic radiations from the nearest portions of the discharge 

 circuit. With feeble oscillations like those first described 

 below, it was necessary to connect one point of the bolometer 

 circuit with one terminal of the arc. 



Evidences of feeble oscillations were obtained with a copper 

 arc in free air, and with iron and silver arcs in nitrogen. (First 

 paper, pp. 391, 410.) In some of these cases the cathode was 

 also of carbon. 



Further study of the copper arc in air has yielded the fol- 

 lowing results, obtained in general with an e.m.f. of over 400 

 volts. 



For oscillations the copper cathode must be clean and free 

 from oxide. This is in conformity with our earlier observa- 

 tion that no glow discharge is possible from an oxidized 

 electrode. The oscillations can sometimes be detected by the 

 bolometer when the current is as large as one ampere, and 

 they reach their maximum intensity at 0*2 or 0*3 amp. The 

 length of arc may be as large as 2 mm , but best results are 

 obtained with a very short gap between the electrodes. 



Water-cooled electrodes do not offer an appreciable advan- 

 tage over those at high temperature. Ample evidence shows 

 that the change from first to second stage at the copper anode 

 has nothing to do with the effect, which is purely a cathode 

 phenomenon. Like effects are produced whether the anode is 

 of copper or carbon. 



The beginning of strong oscillations is indicated by the 

 appearance of the discharge, which assumes an aspect inter- 

 mediate between that of an arc and of a glow. A very charac- 

 teristic tinkling sound is heard and the needle of the voltmeter 

 connected across the discharge trembles violently and irregu- 

 larly between two extreme positions. 



