tap immediately in front of the measuring tube is
momentarily closed, this having been proved to be without ill effect on
the progress of the test. In all experiments done by this test the air
correction is subtracted from each reading, and the remainder brought to
milligrams of nitrogen with the usual corrections. As objection has
frequently been taken to the test on the ground of difficulty in
interpreting the results obtained, Dr Robertson made a series of
experiments for the purpose of standardising the test, and at the same
time of arriving at the condition under which it could be applied in the
most sensitive and efficient manner. A variety of nitro-celluloses having
been tested, there were chosen as typical, of stable and unstable
products, service gun-cotton on the one hand, and an experimental gun-
cotton, Z, on the other. The first point brought out by these experiments
was the striking uniformity of service gun-cotton, first in regard to the
rectilinear nature of the curve of evolution of nitrogen, and secondly in
regard to the small range within which a large number of results is
included, 15 samples lying between 6.6 and 8.7 mgms. of nitrogen evolved
in four hours. In the case of service gun-cotton, little difference in the
rate of evolution of nitrogen evolved is obtained on altering the rate of
passage of CO_{2} gas through the wide range of 500 c.c. per hour to 2,500
c.c. per hour. With Z gun-cotton (see Fig. 52), however, the case is very
different. Operating at a rate of 1,000 c.c. of CO_{2} per hour, a curve
of nitrogen evolution is obtained, which is bent and forms a good
representation of the inherent instability of the material as proved to
exist from other considerations. Operating at the rate of 1,500 c.c. per
hour, as recommended by Dr Will, the evolution of nitrogen is represented
by a straight line, steeper, however, than that of service gun-cotton. The
rate of passage of CO_{2} was therefore chosen at 1,000 c.c. per hour, or
two-thirds of the rate of Dr Will, and this rate, besides possessing the
advantage claimed of rendering diagnostic the manner of nitrogen evolution
in Z gun-cotton, has in other cases been useful in bringing out
relationships, which the higher rate would have entirely masked.

[Illustration: Fig. 52.--Dr. Robertson's results.]

[Illustration: Fig. 53.--Service Guncotton for Cordite made at a Private
Factory.]

Readings are taken thirty minutes from the time the nitro-cel