So far the experiments were of a negative character, and the next step was made with a view to discover the actual cause of the divergence referred to.  A single barrel was now taken, to which a template was fitted, in order to record its exact length.  The barrel was then subjected to a heavy internal hydrostatic pressure.  Under this treatment it expanded circumferentially and at the same time was reduced in length.  This, it was considered, gave a clew to the solution of the problem.  A pair of barrels was now taken and a template fitted accurately to the side of the right-hand one.  As the template fitted the barrel when the latter was not subject to internal pressure, upon such pressure being applied any alterations that might ensue in the length or contour of the barrel could be duly noted.  The right-hand barrel was then subjected to internal hydrostatic pressure.  The result is shown in an exaggerated form in Fig. 2.  It will be seen that both barrels are bent into an arched form.  This would be caused by the barrel under pressure becoming extended circumferentially, and thereby reduced in length, because the metal that is required to supply the increased circumference is taken to some extent from the length, although the substance of metal in the walls of the barrel by its expansion contributes also to the increased diameter.  A simple illustration of this effect is supplied by subjecting an India-rubber tube to internal pressure.  Supposing the material to be sufficiently elastic and the pressure strong enough, the tube would ultimately assume a spherical form.  It is a well known fact that heavy barrels with light charges give less divergence than light barrels with heavy charges.

After the above experiments it was hoped that, if a pair of barrels were put together parallel and soldered only for a space of 3 in. at the breech end, and were then coupled by two encircling rings joined together as in Fig. 4, the left-hand ring only being soldered to the barrel, very accurate shooting would be obtained.  For, it was argued, that by these means the barrel under fire would be able to contract without affecting or being affected by the other barrel; that on the right-hand, it will be seen by the illustration, was the one to slide in its ring.

A pair of able 0.500 bore express rifle barrels were accordingly fitted in this way.  Fig. 3 shows the arrangement with the rings in position.  Upon firing these barrels with ordinary express charges it was found that the lines of fire from each barrel respectively crossed each other, the bullet from the right-hand barrel striking the target 10 in. to the left of the bull’s eye, while the left barrel placed its projectile a similar distance in the opposite direction; or, as would be technically said, the barrels crossed 20 in. at 100 yards, the latter distance being the range at which the experiment was made.  These last results have been accounted for in the following manner:  The two barrels were rigidly joined for