Conductivity of Gases. 309 



The level of mercury in the communicating tubes b V con- 

 sequently sank, and the distance of the electrodes increased 

 considerably, often up to 2 centims., without extinguishing the 

 light. On the contrary, the light was remarkably steady. 

 The temperature of the carbons was then only a little different, 

 but the difference could be recognized with certainty, and 

 upon interruption of the current the kathode ceased to glow 

 somewhat sooner than the anode. When the arc had been 

 several times restored, the apparatus being each time exhausted 

 as much as possible, the evolution of gas gradually decreased ; 

 and when at last no further increase in pressure could be 

 perceived, the difference in temperature between the carbons 

 disappeared at the same moment. But since the slow disin- 

 tegration of the electrodes continually exposes new portions 

 to ignition, I have not succeeded in obtaining pressures 

 under 1 to 2 millim., so that it remains undetermined whether 

 at still lower pressures the kathode would be more strongly 

 ignited than the anode. 



In the usual forms of glow-discharge it is always observed 

 that the kathode becomes red or even white hot, while the 

 anode remains dark. That, however, forms of glow-dis- 

 charge exist in which the temperature of the anode is higher 

 than that of the kathode appears from the experiments of 

 Hittorf described in the first paragraph. 



§ 3. The Spectrum of the Arc and of the Carbons. — If we 

 observe the spectrum of the arc-light as it is formed in 

 atmospheric air of ordinary pressure, it appears continuous, 

 except that we occasionally see a few metallic lines appear. 

 But at the moment that the current is interrupted, a very 

 large number of bright lines blaze up, which have their 

 origin in the mineral impurities of the carbon. Spectroscopic 

 observations of the carbon -light are made more conveniently 

 by producing the arc in vacuo, and with comparatively feeble 

 currents ; we have then at once the advantage that the 

 distance between the carbons may be increased to 1 centim. 

 without the arc becoming broken, and that consequently the 

 spectrum of the kathode, anode, and arc can be easily 

 observed separately. But the main advantage is that the 

 extraordinarily intense continuous spectrum of the particles of 

 carbon which exist in the arc at a brilliant white heat, does 

 not then obliterate the bright lines of the metallic vapours 

 present in the arc. 



It has always been supposed that the gaseous atmosphere 

 in which the arc-light is produced is without influence upon 

 the spectrum of the arc ; Liveing and Dewar * first observed 

 * Proc. Eoy. Soc. vol. xxxv. p. 75 (1883). 



