Conductivity of Gases. 313 



centre is surrounded by a helmet-shaped strongly luminous 

 envelope, about which again is a less luminous layer. Then 

 suddenly the lower carbon rose, and came into contact with 

 the upper one, and the process began anew. This occurred 

 some 50 times in a minute. Evidently an increase of 

 pressure occurred which increased the distance between the 

 carbons, so that the arc was extinguished ; but immediately 

 the pressure returned to its former magnitude, and the 

 carbons came into contact again. 



This phenomenon would be very well explained by the 

 hypothesis of A. Schuster *, according to which the process 

 of electric discharge in gases is accomplished by the dis- 

 sociation of the molecules, but that as soon as the current is 

 interrupted the old condition of the gas is restored. 



Part II. 



The final result of the first part of our investigation was 

 that the arc-discharge and glow-discharge cannot be sharply 

 distinguished, and that in particular the usually enormous 

 difference in the resistance of the layer of gas does not always 

 exist. In this second part I shall endeavour to show upon 

 what conditions the different magnitude of resistance depends, 

 and that the same cause is at work in all cases of gaseous 

 discharge in which the resistance of the gas is small. It will 

 be the more convenient at once to state my view, and then to 

 show that it is correct in particular cases (§ 6-11). 



When in gaseous discharge the resistance of the gas is small, 

 hot metallic vapours are present which conduct the current. 



§ 6. I would first of all recall certain experiments which 

 show that incandescent metallic vapours conduct infinitely 

 better than nitrogen, hydrogen, or air. 



De La Rue f employed for this purpose a globe-shaped 

 vessel with four tubulatures ; two opposite each other served 

 as electrodes of an induction-coil, and the two others for the 

 production of the arc ; the pressure of the nitrogen amounted 

 to from 2 to 3 millim. 



A galvanometer included in the induction-circuit served to 

 measure the current-strength, before the voltaic arc was 

 formed. Then the arc-light was produced, and the new 

 current-strength read off. 



The result obtained was a great increase in the conducting 

 power, when the arc passed between silver and copper elec- 

 trodes ; the change was less with aluminium electrodes, and 



* Proc. Roy. Soc. xxx^vii. p. 317 (1884). 



t Phil. Mag. i v . pp. 29, 553 (1865) ; Comptes Rendus, lx. p. 1002(1865). 



