TRANSACTIONS OF SECTION A. 633 



through space ? I can only see one loophole through which to escape, namely, that 

 Maxwell's medium is not homogeneous, but consists of two parts, and that if we 

 speak of the medium as moving we mean the motion of one of these parts relative 



to the other. . 



While we may hope to obtain important results from an investigation ottUe 

 relation between what we call electricity and the medium, we must not lose sight 

 of another avenue, namely, the relation between electricity and chemical efifects. 

 The passage of electricity through gases presents us with a complicated problem 

 to which a number of physicists havs given their attention of late years. There 

 seems no reasonable doubt that electricity in a gas is conveyed by the diffusion of 

 particles conveying high charges, probably identical with those carried by the 

 electrolytic ions. The fact that this convection is a process of diffusion with 

 comparatively small velocity is shown by the experimental result that the path of 

 the discharge is affected by anv bodily motion of the gas which conveys the current. 

 Even the convection currents due to the heat produced by the discharge itself are 

 sufficient to deflect the luminous column which marks the passage of the current. 



The most puzzling fact, however, connected with the discharge of electricity 

 through gases consists in the absence of symmetry at the positive and negative 

 poles."' There must be some diflereuce between a positively and a negatively charged 

 atom which seems of fundamental importance in the relation between matter and 

 what we call electricity. A discussion of the various phenomena attending the 

 discharo-e of electricity through gases seems to me to point to a conclusion which 

 may possibly prove a step in the right direction. 



A surface of separation between bodies having different conductivities becomes 

 electrified by the passage of a current, while at the surface between two chemically 

 distinct bodies we have, according to Helmholtz, a sheet covered at the two sides 

 with opposite electricities. These surface electrifications are not merely imaginary 

 layers invented to satisfy mathematical surface conditions. They can be proved 

 to be realities. Thus, when one electrolyte floats on another, the specific resist- 

 ances being different, we often observe secondary chemical eftects due to the action 

 of the ions which carry the surface electrification. 



If the passage of electricity from the solid to the gas involves some work done, 

 we must expect a double sheet of electricity at the boundary, the gas in contact 

 with the kathode becoming positively, and that in contact with the anode negatively, 

 electrified. A priori we can form no idea how a layer of gas, the atoms of which 

 carry charges, will behave. The ordinary proof that all electrification must be 

 confined to the surface implies that all forces act according to the law of the 

 inverse square ; but where we have also to consider molecular forces, I see no reason 

 why the electrification at a surface may not stretch across a layer having a thick- 

 ness comparable with the mean free path of the molecule. It is here that there 

 seems to be the fundamental difference between positive and negative electricity. 

 A negative electrification of the gas, like that of a solid or a liquid, seems always 

 confined to the surface, and no one has ever observed a volume electrification of 

 negative electricity. The case is different for the positively electrified part of the 

 gas. Wherever from other considerations we should expect a positively electrified 

 surface sheet, we always get a layer of finite thickness. The result implies a 

 different law of impact between positively and negatively electrified ions, but I see 

 no inherent improbability in this. That the kathode let into a gas is surrounded 

 by a positively electrified laver of finite thickness extending outwards must be 

 considered as an established "fact, and several of the characteristic features ot the 

 discharge are explained by it. The large fall of potential at the kathode can also 

 be explained on the view which I have put forward, for in order to keep up 

 the discharge there must be a sufficient normal force at the surface, and if this lorce 

 is not confined to the surface, but necessarily stretches across a finite hiyer, 

 the fall of potential must be multiplied a great number of times. Similarly 

 Goldstein has shown that some of the phenomena of the kathode are observed at 

 every place at which the positive current flows from a wide to a narrow part of 

 a column of gas. At such places we should expect a positive surface electrifica- 

 tion, and here, again, the whole appearance tends to show that we are dealmg 



