1891.] Luminous Discharge of Electricity through a Gas. 99 



filled with sulphuric acid, which serves to indicate any difference 

 in the absorption of the chlorine by the two c.oils of wire. The 

 vessel was exhausted and then filled with chlorine, and A and B 

 were placed in parallel with the electrodes of an induction coil, 

 giving sparks about an inch and a half long. In this way one coil 

 was positively and the other negatively electrified, and any dif- 

 ference in the rate of combination of the chlorine with the metal 

 would show itself by the motion of the sulphuric acid in the gauge. 

 Only a very small motion of the sulphuric acid occurred, and this 

 seemed to be accidental, as it was not reversed on reversing the 

 coil. The difference between the rate of combination of chlorine 

 with a positively and negatively electrified metal must therefore be 

 small. 



Again, if the difference between the behaviour of the positive and 

 negative discharge were due to purely chemical action between the 

 gas and the electrode, we should expect this difference to be absent 

 in the case where the electrodes consisted of a volatile liquid or 

 solid, and the gas was the vapour of the electrode. I tried three 

 cases of this kind : one in which the electrodes were water and the 

 gas water vapour; a glass tube was completely filled with water, 

 then placed on the pump, and the water boiled away until only just 

 enough was left to serve as electrodes ; the tube was then sealed off 

 and cooled down until the vapour pressure was low enough to allow 

 the electric discharge to pass without difficulty ; this tube, however, 

 had all the usual characteristics of the discharge through vacuum 

 tubes, including the negative dark spaces and the striations. In the 

 next experiment a similar tube was taken, the water being replaced 

 by bromine ; this, too, showed the usual differences between the dis- 

 charge at the two electrodes, and similar appearances were presented 

 by a tube in which the electrodes were re-distilled arsenic and the gas 

 arsenic vapour. 



Another difficulty in the way of explaining the difference at the 

 two electrodes by chemical action is that no difference seems to be 

 made in the appearance when a strongly electronegative gas, such as 

 chlorine, is substituted for a strongly electropositive one, such as 

 hydrogen. 



I next endeavoured to get rid of the electrodes altogether by trying 

 to get a circular discharge in an exhausted re-entrant tube without 

 any electrodes. For this purpose the primary was generally a piece 

 of copper rod bent into a horse-shoe shape; the secondary circuit 

 was an endless circular glass tube from which the air had been ex-- 

 hausted. A Leyden jar, charged by a Wimshurst machine, was dis- 

 charged through the primary, and produced by induction an electro- 

 motive force round the exhausted tube. When the secondary was 

 not shielded from the electrostatic induction of the primary, it was 



