Discharge of Electricity through Gases. 321 



an asymmetry. I have noticed a very curions tendency of the 

 discharge at the negative pole, to start rather from the glass out of 

 which the electrode protrudes, than from the metal electrode. The 

 glass is heated up in that case and a bright yellow, sodium spot is 

 seen at the point from which the current in the vapour sets out. 

 When the discharge sets out from the metal an amalgam of mercury 

 and platinum, or mercury and aluminium, is formed. The subject 

 requires further investigation. July, 1884.] 



It is more difficult to prevent stratifications than to prevent the 

 dark space and glow. A mixture of air and mercury shows beautiful 

 stratifications. When the air is expelled as much as possible, I never 

 saw stratifications while the tubes were attached to the pump, but iu 

 order to study the phenomena a little better I sealed off some of the 

 tubes and heated them up either in a steam or paraffin bath, so as to 

 give them the same temperature throughout. It is then that the 

 stratifications are very apt to occur at a temperature of 100 C., when 

 the tension of mercury vapour would, according to Eegnault, be 

 three-quarters of a millimetre. The extreme difficulty, however, of 

 excluding all traces of air easily explains this, while the irregularity 

 of the phenomenon, and its absence in the experiments which were 

 carried on with the greatest care, seem to me to be conclusive in 

 proving that pure mercury vapour never shows any stratifications, for 

 the experiments were continued until the tension of mercury vapour 

 was 20 millims. 



My assistant, Mr. Arthur Stanton, has taken some photographs of 

 the spark in mercury vapour at different pressures. 



Figs, (la) and (16), Plate 4, represent the spark in mercury at a 

 uniform temperature of 50. The tension of mercury vapour at that 

 temperature according to Regnault is O'l millim. Figs. 2a and 26 

 represent the discharge at a temperature of 100 and a tension of 0'7, 

 while in figs. 3a and 36 the temperature was 150 and the tension 

 4'3 millim. In the figures marked a the pole on the left h-ind side 

 was negative, while in the figures marked 6 the negative pole was to 

 the right. The electrodes were platinum wires surrounded by glass 

 tubes open at the end ; and in figs. 2 and 3 the tendency of the 

 discharge at the negative pole to start as far back as possible from 

 the glass, rather than from the metal, is well seen. 



I have reason to think that the vapour of sodium resembles that of 

 mercury, and does not show any dark intervals. From the fact that 

 at the temperature of the spark sodium shows its line spectrum instead 

 of the bands, we may conclude that we have also to deal in that case 

 with a monatomic molecule. 



Evidence of Dissociation during the Discharge. 

 There is very strong spectroscopic evidence that gases through 



