THE COMBINING VOLUMES OF HYDROGEN AND OXYGEN. 
411 
Any impurities accompanying the oxygen must have been sufficiently volatile to 
escape condensation in the fractionating tubes. To test whether oxygen could 
I’etain or acquire carbon dioxide, as such, after leaving the fractionating train, a 
considerable volume was pumped through the cooled spiral, >7. Nothing was retained 
by the spiral. Kammerling Onnes( u ) has shown that, at the boiling point of oxygen, 
the vapour pressure of carbon dioxide is about 6 x 10 -6 mm. of mercury, so that little 
could escape condensation when exposed to a large surface of glass cooled in liquid 
air. It seemed just possible that oxidation of dust in the permanganate tubes might 
give rise to carbon monoxide, a gas about as difficult to separate from liquid oxygen 
by fractional distillation as nitrogen would be. Some experiments carried out by 
Mr. J. N. Greenwood indicate that if 
any carbon monoxide is present in perman¬ 
ganate oxygen, the quantity is less than 
1 part in 40,000 by volume, even when no 
precautions are taken to avoid the presence 
of dust in the decomposition tube. a 
We next considered the possibility of 
impurities in the explosion vessel. There 
was no question of grease finding its way 
back from the tap, y, because mercury and 
gas travelled always in the opposite direc¬ 
tion. The pipette tap was separated from 
the explosion vessel by more than half a 
metre of capillary tubing in which there 
were several bends. Any grease carried by 
a mercury thread through a tap-bore is 
usually deposited in the first few centi¬ 
metres of connecting tube. In order to 
obviate any possible risk of contamination 
in this way, the device illustrated in fig. 5 
was adopted. Two vertical capillary tubes, A and B, were sealed to the transference 
capillary at C, as close together as possible : A was provided with a tap, and B was 
connected with a mercury reservoir, D, through an air-trap, E. The tube, B, had 
no tap, but the flow of mercury through it was controlled by a screw-clip on the 
reservoir tubing. 
All mercury that passed through the pipette tap was withdrawn by the tube, A. 
At the end of the process of transferring gas from the pipette to the explosion vessel, 
the mercury thread following the gas through the pipette tap and along the capillary 
was checked at C. The gas remaining in the capillary between C and the explosion 
vessel was then driven over by clean mercury from the reservoir, D. 
To remove any grease that might have previously found its way there, the 
3 k 2 
