from a glowing lime cathode in mercury vapour. 503 
pass. The mercury spectrum was still seen faintly and also the 
three orange lines. The inability to get rid of the mercury 
spectrum with this arrangement was probably due to the vapour 
combining with the aluminium anode during the former experi- 
ments. The aluminium anode was therefore replaced by a platinum 
disc and the whole tube carefully boiled in nitric acid to get rid of 
all traces of mercury. Repeating the experiment with the glass 
spiral in liquid air the spectrum showed no trace of the mercury 
lines or of the new lines in the orange and red. 
This experiment was repeated in a slightly different way. 
The apparatus was pumped out to as low a pressure as possible 
(less than ’001 mm.). The spectrum was that of mercury together 
with the lines in the orange and red. The spiral was then placed 
in liquid air and the spectrum carefully watched. The mercury 
spectrum gradually faded away and so did the new lines, those in 
the red being the first to disappear. The current through the tube 
at the same time decreased considerably and the luminosity was 
feeble and showed the spectrum of the residual gas. On removing 
the liquid air from the glass spiral, a brilliant band spectrum of 
carbon monoxide appeared ; this lasted for a few seconds only, and 
as it faded away the mercury spectrum was seen increasing in 
brightness and accompanied by the three lines in the orange. 
It took a long time for this spectrum to be brightly established 
again as the mercury vapour slowly came over from the spiral 
tube. The two red lines were never given very brightly by this 
tube with a platinum anode, and were only obtained at all by 
heating the cathode to a high temperature and having a large 
difference of potential between the electrodes. 
In connection with the appearance of the CO spectrum it 
should be mentioned that the discharge tube, spiral and connecting 
tubes had been carefully washed out with hot chromic acid before 
fitting up, and that the CO spectrum was only obtained after the 
spiral had been placed in liquid air. No trace of the CO spectrum 
was seen when the experiment was repeated with the cathode hot 
but with no discharge passing. It was also found that the gas 
pressure in the apparatus was greater after removing the liquid air 
from the spiral tube than it was before the spiral had been cooled, 
although the passage of the discharge under ordinary conditions 
tended to decrease the gas pressure. 
These experiments were repeated, using first oxygen and then 
hydrogen as the residual gas in the apparatus. It was found with 
hydrogen that the mercury spectrum and the new lines were 
brighter than in air or oxygen. The author has previously shown* 
that the discharge takes place more readily in hydrogen gas than 
in other gases. 
* Phil. Trans. A, Vol. 207, p. 149, 1907. 
