cutting away of a longitudinal quarter
section, Fig. 28.

[Illustration: Fig. 28. Armor-Piercing Shell. Showing position of
flaw.]

There are, therefore, two great forces with which to contend in the
design of projectiles, to one of which, compression, has been given the
greatest attention because of its recognized tendency to cause the base
of the shell to crowd upon the head and cause the shell to break up
about the ogive. The other force, torsion, seems not to have been
considered prior to the present instance, at any rate so far as the
author has been able to ascertain, not because thought to be
unimportant, but because of oversight or failure on the part of
investigators to take into consideration in this instance, an element
of reaction commonly considered in mechanical engineering practice, as
in shafting for vessels and for power transmission in shops, etc.

The writer maintains that immediately upon impact the metal in a shell
assumes a state of physical unrest, due to stresses similar to those in
a propeller shaft when in motion, except that in the former case the
intensity of the compression stresses greatly exceed those in the
latter. Because a shell is only 3-1/2 calibres in length is no
criterion that the same stresses do not exist there as would exist in
the theoretical projectile considered of twenty calibres, or one of
even more exaggerated proportions--there would be merely a difference
in the _intensity_ of these stresses.

In a projectile making one complete revolution about its major axis in
every twenty-five calibres flight, any one elementary unit area or mass
in that shell likewise makes one complete revolution in the same
distance of travel, and the path traversed by that unit area or mass is
that of a spiral of radius equal to the distance of that unit area or
mass from the major axis of the shell, the diameter of which spiral
would be the diameter of the shell in question--and the pitch
twenty-five calibres--if said unit area were on the surface of the body
of the shell.

Upon impact the tendency of this unit area would be to continue its
flight along the continuation of that spiral or along the line _ed_
of our theoretical shell of twenty calibres. The result would be for
each disc element theoretically considered to crowd upon the next
corresponding disc element and these two upon the third corresponding
disc element etc., such crowding taking place along the line _ed_.
Therefore the