314 M. F. Stenger on the Electric 



feeblest with zinc, cadmium, or magnesium poles. The 

 increase in conductivity was very remarkable with carbon 

 poles. On the other hand, iron and platinum electrodes gave 

 no perceptible difference ; whence it results that the increase 

 in conductivity in the other cases cannot be brought about 

 by the high temperature of the nitrogen. The experiment 

 is, however, so far unfavourable, and allows no conclusion to 

 be drawn as to the true conductivity of the different metallic 

 vapours, since, as Hittorf's work sufficiently proves, the chief 

 resistance of the glow-discharge occurs in the neighbourhood 

 of the kathode ; and with the great distance between the 

 electrodes in De La Rue's experiment, the formation of 

 metallic vapours takes place exclusively in the neighbourhood 

 of the positive brush-discharge. 



Moreover, it is to be remembered that in an experimental 

 arrangement of this kind we have to consider, not only what 

 metallic vapours are present, but without doubt also their 

 quantity, that a less volatile metal may therefore afford better 

 conduction than a more volatile one, although the true con- 

 ductivity, in the first case, may be far worse than in the 

 second. 



As a second proof of the relatively good conducting -power 

 of two metallic vapours, I may quote certain observations of 

 Hittorf's * : — " At the temperatures of our flames, at which 

 the gases of which they consist possess so considerable an 

 electric resistance, other vapours have a much greater con- 

 ductivity. Of all gases, the vapour of potassium conducts 

 best. Next comes sodium. The other metals, so far as they 

 are volatile under these conditions, produce, in the state of 

 gas, little change in the deflection." 



Hittorf shows in another place that mercury vapour, in the 

 non-luminous Bunsen flame, conducts much worse than 

 potassium vapour. 



That most metals in the Bunsen burner produce no percep- 

 tible change in the resistance, is no doubt explained by the 

 fact that the temperature is too low for a sufficient evolution 

 of vapour. 



After these preliminary remarks, I will now pass on to con- 

 sider, in the particular cases where a stratum of gas possesses 

 a relatively small resistance, whether incandescent gaseous 

 vapours are present, and to show their influence. 



§ 7. As far as the arc-light between metallic electrodes is 

 concerned, it is universally known that the colour of the light 

 varies essentially with the nature of the metals employed, 



* Pogg. Ann. cxxxvi. p. 229 (1869). 



