* On removing the tubes G, H, it was found, that owing of the metal, the cast-iron ends of both had been fractured, c perhaps, before the outer shell had attained its maximum re

† Tube M had an iron rod down its axis to prevent the e each other during collapse; a tin ring had also been left which accounts for the increased pressure required to produ

the steam-pressure gauges at c, which exhibited the pressure in the cylinder during the experiment in lbs. per square inch to ensure accuracy two gauges were employed, one of Schaeffer's and the other of Smith's construction, and the indications of these were checked by an accurate safety-valve, d. A small cock, e, served to let off the air contained in the cylinder when necessary.

Fig. 1 is a section of the large cylinder. The top and bottom covers, f and g, were made of strength proportionate to that of the cylinder, to which they were secured by 1-inch bolts placed 3 inches apart. In the bottom cover, g, a hole was drilled, to receive the rod and screwnut k, which supported the tube D to be experimented upon; and through the top passed a 24-inch pipe, m, inserted in the cast-iron end of the tube D. On the end of this pipe was a large nut, which screwed down upon an indian-rubber washer on the cover of the cylinder, so as to close the opening round the pipe and make it watertight. The object of this pipe was to allow the air from the interior of the tube D to escape during the collapse, and so to place it, as nearly as possible, under the same circumstances as the flue of a boiler.

The whole of the experiments were effected by means of the hydraulic pump, by which water was forced through the pipe a a into the cylinder C; thus driving the air in a highly compressed state to the upper part of the cylinder, whence, when a very high pressure was required, it was deemed advisable to let it escape by the cock e, and to effect rupture through the medium of water only. In both these cases a perfectly uniform pressure was ensured upon every part of the tube to be collapsed.

These preparations having been made, and the pressuregauges carefully adjusted, the experiments proceeded as shown in the following Tables.

The first experiment was upon a tube 4 inches in diameter, and 1 foot 7 inches long between the cast-iron ends, to which it was riveted securely and brazed. It was composed, as in the other experiments, of a single thin plate, bent to the required form upon a mandril, and riveted, and also brazed to prevent leakage into the interior. This tube having been fixed to the cylinder covers in the manner described above, the pump was applied and a

gradually-increasing force given to its exterior surface, until its powers of resistance were overcome. During the experiments the precaution of allowing the air to escape at high pressures was found absolutely necessary, as the tubes generally collapsed with an explosion of the suddenly-compressed air in the tube D, fig. 1, accompanied by a loud report as it made its escape by the pipe m. These explosions give pretty correct indications of what takes place when the internal flues of boilers collapse.

It has long been a desideratum to determine some law by which the engineer could calculate the proportionate strength of the internal flues. Hitherto we have acted upon the principle that the cylindrical flues, as ordinarily constructed, were considerably stronger than the outer shell; but this opinion has in reality no foundation in experiment, excepting only uncertain deductions from occasional explosions and the failure of vessels under high pressures in circumstances of a very variable and doubtful character. There have been no definite rules to guide us hitherto in proportioning the diameter, length, and thickness of plates of the flues, so as to correspond with the strength of the boiler; and even in cases where explosions have taken place from collapse, we have, it is to be feared, too frequently mistaken the actual cause, in consequence of the débris covering the site, and the force which has torn to pieces the outer shell. The anomalous position in which these constructions are placed has greatly retarded the application of science to their improvement, and there appears, in fact, to be no rule known by which to attain uniformity of strength between those parts of a boiler exposed to an internal tensile, and those exposed to an external compressive force.

To supply this want, and to remedy certain anomalous results arising from defective forms of construction, it appeared desirable that these vessels should be subjected