505 
from a glowing lime cathode in mercury vapour. 
being only just visible. At this stage the brightness of these lines 
was sometimes increased for a few seconds by reversing the induc- 
tion coil, but the orange lines soon disappeared, leaving the tube 
quite hard and the spectrum the same as at the commencement of 
the experiment. 
This was repeated several times and the results were always 
practically the same as those recorded above. 
Two blank experiments were next made to see if the gases 
condensed in the vacuum tube were really produced by the 
passage of the discharge or merely formed by the heating of 
the lime-cathode tube. The vacuum tube was surrounded by 
liquid air; the lime cathode was raised to a high temperature; 
and the two tubes left connected for half an hour. Then, as 
before, the vacuum tube was isolated and connected to the 
induction coil. It was found to be quite hard and no gas 
developed as it warmed up to the temperature of the room. 
This experiment was repeated with exactly the same result. 
On repeating the experiment with the discharge passing in the 
lime-cathode tube the spectra of the condensed substances appeared 
just as in the first experiment. It thus seemed that the carbon- 
monoxide and the substance giving the orange lines were formed 
either by the passage of the discharge or distilled over owing to the 
extra warming of the mercury anode while the discharge was 
passing ; this latter being the more probable explanation, seeing 
that some mercury must certainly have distilled over to cause the 
extra brightness of the mercury lines in the vacuum tube after 
the discharge had been passing in the adjoining tube. To test 
this point the vacuum tube was placed in liquid air and the 
mercury forming the anode in the discharge bulb was heated by 
means of a bunsen flame until it boiled. On then examining the 
discharge through the condensed substances it was found to give, 
for a few seconds, a faint carbon monoxide spectrum and then a 
bright mercury spectrum and the new orange lines. These lasted 
for a few minutes and then disappeared, a faint mercury spectrum 
remaining. 
The fact that very little gas was formed in the vacuum tube 
by the evaporation of the substances condensed by the liquid air 
during these experiments, and that the gas quickly disappeared 
when subjected to the induction coil discharge, led me to repeat 
the experiments with the residual gas in the apparatus at a 
higher pressure, so that the discharge would pass more easily 
and the spectrum be better observed. Hydrogen and oxygen 
were tried in turn but were not very satisfactory because of the 
red and orange lines in the spectra of these gases. On this 
account helium was tried, and was found to answer very well. 
With this gas in the apparatus at pressures from ’005 to milli- 
