1891.] Luminous Discharge of Electricity through a Gas. 97 



visible : we see the action on the visible current, but not the 

 opposite one on the other. 



The most obvious explanation of the remarkable difference in the 

 behaviour of the discharges from the anode and cathode is that it 

 arises from some difference in the chemical action between the gas 

 and the two electrodes. I have made a series of experiments in 

 order to test this view, and have been led to the conclusion that 

 an explanation of this effect by purely chemical action is delusive. 

 At the same time I think that the necessity for the existence of 

 some action between the gas and the electrode is shown by the fol- 

 lowing experiment. In the ' Philosophical Magazine/ vol. 29, 1890, 

 p. 441 (On the Passage of Electricity through Hot Gases), I described 

 an experiment in which cold electrodes were plunged into a hot 

 gas, such as iodine, heated until it dissociated, when it was found 

 that no current passed through the gas until the electrode got hot, 

 when it passed freely. The effect produced by the cold electrodes 

 seemed too abrupt to be altogether due to the cooling of the adja- 

 cent gas by the electrodes. I therefore made the following experi- 

 ment in order to test this point. If the effect is due to the cooling 

 of the gas, the temperature of the electrodes when the system begins 

 to conduct ought not to vary much, whatever may be the material of 

 which they are made ; while if the effect were due to chemical 

 action between the gas and the electrodes, we should expect con- 

 siderable variation with different electrodes in the temperature at 

 which conduction begins. I therefore attempted to measure roughly 

 the temperature at which conduction began (a) when the electrodes 

 were iron, (b) when they were platinum. This was done by making 

 one of the electrodes into a thermopile placed in circuit with a 

 dead-beat galvanometer; in case (a) the thermopile consisted of an 

 iron plate with a German-silver wire, in case (6) of platinum, foil 

 and a German-silver wire. The electrode used as the thermopile 

 was dropped cold into the hot gases and connected up with the 

 main circuit. When the galvanometer in the main circuit first began 

 to show decided indications of the passage of a current, the ob- 

 server who was watching this galvanometer called to the observer 

 at the dead-beat galvanometer in the thermopile circuit, and this 

 observer read the deflection of this galvanometer. From this reading 

 the temperature of the hot junction could n pproximately be determined. 

 Ine experiment was repeated, using, instead of the iron- German 

 silver couple, a platinum- German one, the platinum foil being 

 wound round an iron plate to make it heat up at approximately 

 the same rate as the first couple. It was found that the conduction 

 began at a much lower temperature when the electrode was iron 

 than when it was platinum, indicating that some action between the 

 electricity and the gas was necessary for conduction. I could not, 



