Discharge of Electricity through Gases. 319 



fact that the vapour-density of mercury vapour is anomalous. If, as 

 is generally supposed, the molecule of the simple gases contains 

 two atoms, that of mercury can only contain one. At any rate, we 

 are justified in asserting that, the molecule of mercury vapour has a 

 simpler composition than that of most gases. If, then, an essential 

 part of the glow discharge is due to the breaking up of the molecules, 

 we might expect mercury vapour to present other and much simpler 

 phenomena than the gases with which we are generally accustomed 

 to work. This, indeed, is the case ; for I find that if the mercury vapour 

 is sufficiently free from air, the discharge through it, shows no negative 

 glow, no dark spaces, and no stratifications. In view of the importance 

 of the fact, I have paid a good deal of attention to this point, and, in 

 spite of some experimental difficulties, I have considerable confidence 

 that the statement which I have made is true. The one fact which up 

 to recently has made me speak with caution consists in the altered 

 conditions tinder which the experiments with mercury have to be 

 carried on, but I think that I have now a complete answer to the 

 objection which might be raised on that ground. 



One difficulty consists in- the choice of electrodes. Aluminium is 

 attacked by mercury vapour, and experiments made with; aluminium 

 might therefore be objected to as not conclusive. Platinum is the 

 only metal which can be conveniently used, but platinum blackens 

 the tube very rapidly and prevents the examination of the parts sur- 

 rounding the negative electrode. I have found it convenient to use plati- 

 num electrodes surrounded by a small glass tube open at its end. The 

 discharge can then pass freely from the open end, and the glass tube 

 prevents the deposit of platinum in the outer tube. But the platinum 

 may even protrude to a small extent, as the deposit, unless it is thick, 

 seems to combine and alloy easily with the mercury in the tube. If 

 the glass tube surrounding the metal electrode does not fit too closely 

 the discharge will partly pass through the interstice and heat up the 

 glass, which then becomes a sufficiently good conductor to allow the 

 discharge to pass through it, which is an advantage, as the electrodes 

 then have a larger area of active surface. My tubes had two such 

 electrodes, and contained also a quantity of mercury, varying perhaps 

 from 1 to 4 cub. centims. They were attached to a Sprengel pump. 

 Both sulphuric acid and phosphoric anhydride were used for drying 

 purposes. After all the air was as much as possible exhausted, 

 the mercury was heated up to its boiling point. When this had 

 repeatedly been done and the tube kept at the pump for at least 

 a day, the discharge from my coil, which gives a good 4-inch spark 

 in air, would not pass, and there was not a trace of fluorescence. 

 Aluminium electrodes gave more trouble than platinum ones, for they 

 seemed to contain much more hydrogen. 



It is needless to describe all the tubes used and the experiments 



