14 
MR. J. W. CAPSTICK OR THE RATIO OF THE SPECIFIC HEATS 
mixture, and the capillary end, DE, dips below the level of the liquefied gas, which 
is contained in a tube surrounded by a freezing mixture. By opening the clamp for 
a moment the liquid is drawn into C, and the capillaries sealed off by a small blow¬ 
pipe, at B and D. On weighing C with the parts drawn off we get the amount 
of liquid enclosed. 
Methyl chloride requires a temperature below — 20° to liquefy it, and, for this, ether 
and solid carbonic acid is most convenient, but has the disadvantage of giving off 
an inflammable vapour, which might take fire from the blow-pipe flame. Hence the 
tube C was packed round with solid carbonic acid, moistened with chloroform, which 
forms quite as effective a freezing mixture and does not readily take fire. The 
condenser, F, was closed with a stopper, through which passed a tube to carry the 
ether vapour beyond the reach of danger. 
§ 4. Marsh Gas. 
As was to be expected this gas gave much more trouble than any of the others. 
It cannot be freed from air by liquefaction, as was done with most of the others, and 
as the density of the gas is a factor in calculating the ratio of the specific heats, it 
was necessaiy to make a set of analyses after each experiment to determine the 
percentage of air. The correction for this is by no means inaj^preciable, on account of 
the low density of methane ; roughly speaking, one per cent, of air makes an alteration 
of one per cent, in tlie result. 
In the case of such gases as the paraflSns, the quantity which can be taken for 
analysis is so small that any error in its measurement from want of accuracy in the 
calibration of the measuring tube, or other causes, has a large effect on the calculation 
of the percentage of air. Adding to this all the other sources of error incidental 
to gas analysis, such as incomplete combustion, oxidation of the nitrogen present, 
temperature errors, &c., the result is that the accuracy is less than that attainable in 
the velocity of sound determination. 
It is unfortunate too that this additional source of error should enter most 
prominently in the case of methane, which is notably a most diflficult gas to prepare 
pure. The consequence is that the range of values found for the ratio of the 
specific heats is greater than for any other gas, and the most that can be said is that 
the mean is probably within one or two per cent, of the truth. 
For the preparation of marsh gas two methods were used, Frankland’s method 
by the action of zinc methyl on water, and Gladstone and Tribe’s by the action of 
the copper-zinc couple on methyl iodide and alcohol. 
The latter method appears simple when the original memoir describing it is read, 
but in practice it requires considerable care. It would be tedious to recount the dis¬ 
couraging series of failures before gas was obtained sufficiently pure for the experiments, 
so the conditions on which success was found to depend will be stated simply. 
