lycerine left in it. The
nitro-glycerine is then obtained by difference, or the method suggested to
me privately by Mr W.J. Williams may be used. The residue obtained by
evaporation of the ether-alcohol solution, after weighing, is treated with
alcoholic potash to decompose the nitro-glycerine, water is added and the
alcohol evaporated off. Some ether is then added, and the mixture shaken,
and the ether separated and evaporated, and the residue weighed as
vaseline.

The moisture should, however, be determined by the method devised by Mr
Arthur Marshall, F.I.C., of the Royal Gunpowder Works, Waltham Abbey,
which is carried out as follows:--The cordite or other explosive is
prepared in the manner laid down for the Abel heat test, that is t say, it
is ground in a small mill, and that portion is selected which passes
through a sieve having holes of the size of No. 8 wire gauge, but not
through one with holes No. 14 wire gauge.

[Illustration: FIG. 40.--MARSHALL'S APPARATUS FOR MOISTURE IN CORDITE.]

The form of apparatus used is shown in Fig. 40. It consists of an
aluminium dish A, having the dimensions shown, and the glass cone B
weighing not more than 30 grms. Five grms. of the cordite are weighed into
the aluminium dish A. This is covered with the cone B, and the whole is
accurately weighed, and is then placed upon a metal plate heated by steam
from a water bath. It is left upon the bath until all the moisture has
been driven off, then it is allowed to cool for about half-an-hour in a
desiccator and is weighed. The loss in weight gives accurately the
moisture of the sample. For cordite of the original composition, one
hour's heating is sufficient to entirely drive off the moisture; for
modified cordite containing 65 per cent. of gun-cotton, two hours is
enough, provided that there be not more than 1.3 per cent. of moisture
present.

If the proportion of nitro-glycerine be higher, a longer heating is
necessary. The aluminium dish must not be shallower than shown in the
figure, for if the distance between the substance and the edge of the
glass cone be less than half an inch, some nitro-glycerine will be lost.
Again, the sample must not be ground finer than stated, else some of the
moisture will be lost in the grinding and sieving operations, and the
result will be too low. In order to be able to drive off all the moisture
in the times mentioned, it is essential that the glass cone shall not fit
too closely on the alum