at the Contact of Solids and Gases. 13 



into them. Between a and b a current was closed, so that 

 chlorine was evolved upon the side of b turned towards a. 

 On the other hand, b and c could be connected with the gal- 

 vanometer by momentary closings. To my astonishment, 

 after a time there was shown an electric difference in which, 

 not b, but c appeared negative. Of the liberated chlorine, 

 traces passed through the atmosphere to the surface of the 

 liquid in the other vessel, and through it arrived at the elec- 

 trode c. That slight traces of chlorine act at once electromo- 

 tively on platinum also had already been remarked by Maca- 

 luso ; and I am now of opinion that the oxygen gas which I 

 used for my first measurements on gas batteries, and which 

 had been prepared from chlorate of potass, always carried 

 with it traces of chlorine, although I thought I had sufficiently 

 purified it by washing ; for with oxygen obtained by electro- 

 lysis I could as little excite platinum electromotively as palla- 

 dium. I now altered my apparatus by giving it the form of a 

 U-shaped tube, the horizontal part of which, 80 centims. long, 

 was divided in the middle by a thin plate of palladium into 

 two halves. I first filled both sides with diluted sulphuric 

 acid, and evolved hydrogen at the side of b facing the plate a, 

 and that by closing the circuit for only a few seconds. The 

 action of hydrogen that had penetrated through the palladium 

 was very soon perceptible : the plate b also became positive on 

 its reverse side. The experiment cannot be long continued ; 

 for the plate bends so much that it soon breaks loose from 

 its attachment. A fresh tube was now filled with diluted 

 hydrochloric acid. The long layer of liquid permitted none 

 of the evolved chlorine to escape, while the electrode c re- 

 mained perfectly indifferent until the plate b was eaten quite 

 through. In order to fix more exactly the instant at which 

 this took place, I filled the vertical parts of the U-tube up to 

 as many different heights as possible with the liquid, and re- 

 peated the experiment. Again b and c remained indifferent 

 to one another : suddenly there was a violent deflection of 

 the galvanometer-mirror ; but at this moment the liquid on 

 both sides began to place itself in equilibrium. According 

 to these experiments, chlorine does not penetrate palladium as 

 hydrogen does. 



From this I think I am warranted in maintaining, generally, 

 that, strictly speaking, we never have to do with any electro- 

 motive force of gases, but either with tension-differences 

 called forth by conducting liquids of different kinds, or with 

 alterations of metals by gases which have lost their gaseous 

 state by occlusion in, or condensation on the surface of, 

 metals ; for an actually coherent layer of gas that covered a 



