The injection water was, on resuming the experiments, heated over a small furnace, to boiling, in a metallic vessel, from which it was drawn by the pump. When the bottom of the boiler was at a bright red heat, the lowest thermometer had attained a temperature of 570°, and was soon after removed. The quantity of water thrown in at each stroke of the pump, was now by no means so regular as when the action of the pump was not impeded by the formation of steam within it, from the injection water. The results obtained were:

A combustible gas had appeared for the first time. The injection water was now changed for cold water, and a gas obtained, which burned as before; 11 measures of air and one measure of the gas detonated slightly, also one to 8 measures of air; neither detonation was sufficiently violent to extinguish the candle held at the mouth of the jar in which the gases were fired.

The gas now came over, not in copious bubbles and during a short time, but slowly and continuously, as if resulting from a constant, but not violent, chemical action. After these results had been obtained, the violent and brief bubbling, when the water was injected, recommenced, and the combustible gas was no more obtained. The change of hot injecting water for cold, and the collection of the combustible gas after the change, showed that the gas was not derived from any effect produced by the increased temperature of the liquid introduced. The other circumstances which had been different from those of former experiments, were the superior cleanliness of the bottom of the boiler, and the repacking of the hand hole with cloth, oil and putty, and of the glass window with paper. Before proceeding to the detail of the experiments, in examination of the source of the combustible gas obtained, it may be well to mention that the glass tube, already spoken of, showed, after the water injected in some of the experiments had evaporated, a current of air, due to a force equivalent to a head of from to 1 inches of water, from the exterior into the boiler; in one experiment it is noted that the water, in the tube referred to, soon fell, which indicated a leak in some part of the boiler.

On the day following that upon which the experiments just given, were made, nothing conclusive was obtained; no combustible gas appeared, but the heat was hardly as high as on the preceding day. Small disks of wood, thrown into the boiler, gave a combustible gas, which came over just as was noticed in relation to the inflammable gas of the preceding day's experiments. That these inflammable gases, in mixture with the oxygen, remaining in the atmospheric air, within the boiler, produced no explosion, is in accordance with

the well known fact in relation to them; pure hydrogen, in such a mixture, combines with oxygen under the influence of a body heated to redness.

On the following day of experiment, circumstances proved entirely favourable; the bottom of the boiler was heated as intensely as on the former occasion. After much incombustible gas had been obtained, traces of an inflam mable one appeared. A strong smell of oil was noticed about the hand hole, at the back end of the boiler; the packings were now white on the exterior. The fire was urged, and the boiler became strongly heated throughout its whole length. The following results are from the journal of the experiments.

Each stroke of the pump threw in 5-8ths of an oz. of water.
The gas is that collected by one stroke of the pump.

No combustible gas was procured in these experiments, though the circumstances were more favourable to the production of hydrogen, by the boiler, than on other occasions. On examining the cloth packing of the hand holes at each end, it was found to have disappeared, except in spots; the putty was white. The boiler was not dirty enough to colour, with oxide of iron, clean water which was introduced.

These results point conclusively to the packing, as the source of the combustible gas obtained. The flame of that gas was that of carburetted hydrogen, and not of pure hydrogen. They further show that even in this intensely heated state of the bottom of the boiler no hydrogen was liberated by the decomposition of the water injected.

In conclusion, it appears from these experiments:

1. That the gas obtained by injecting water upon the bottom of a boiler which was at a bright red heat, was nitrogen gas, with a variable quantity of oxygen: it was, in fact, atmospheric air deprived, by the heated metal, of more or less of its oxygen.

2. That this air was derived, principally, from the current into the boiler when surcharged steam had ceased to be formed, and the boiler was left dry; there will, therefore, be no such quantity in a working boiler, where the air must be supplied from the cold water thrown in.

3. That water in contact with heated iron in a steam boiler, the surface being in its ordinary state, clean, but not bright, is not decomposed by the metal.

VIII. To observe accurately the sort of bursting produced by a gradual increase of pressure within Cylinders of Iron and Copper.

It has been contended by some, that ruptures produced by a gradual increase of pressure within steam boilers do not bear the character of explosions, but that a mere rending takes place, giving escape to the contents. This has been assumed to be especially the case with copper boilers. To make the observation required by the above question, cylinders of iron and copper were prepared, of sufficient size, to make a small thickness of material answer for rending by a pressure which was easily attainable. Two experiments made, one with an iron and another with a copper cylinder, afforded so direct an answer to the query that it was not deemed necessary to carry the experiments further, especially as they were tedious, and not without danger. A further experiment of the same tenor, resulted from a trial of Perkins's assertion in regard to the effect of making an opening in a vessel containing water, and heated to a high temperature.

The boilers used were cylindrical, eight and a half inches in diameter, and ten and twelve inches respectively in length, of iron .02 inch thick, and of copper .03 inch thick, having iron heads .05 inch thick, to which the convex surface was fixed by iron rivets, placed nearly touching each other. A single opening in the middle of one of the heads of each boiler was provided to introduce the water, and was furnished with a screw, into which to insert a tube and piston, connected with a small spring weighing machine, which is represented at a in the cut on page 224. Upon the cylinder of this machine a ring was placed, which was moveable along the cylinder by a slight pressure: this ring was forced towards the end of the cylinder nearest to the boiler head, as the spring was bent and remaining in its place when the spring relaxed, served to register the maximum pressure to which the piston had been exposed previous to observing it.

The iron boiler was placed in a heavy cylinder of wrought iron, which served as a furnace, the axis of the boiler being nearly horizontal, and that of the furnace cylinder vertical. The boiler, having been half filled with water, was placed upon a fire of charcoal, and when the water boiled, the register machine for the pressure was screwed in.

The place selected for the experiments was in a deserted quarry on the banks of the Pennypack, near Holmesburgh. The high bank served as a protection, by the aid of which the experiments were viewed with little danger. A wire and cord were attached to the head of the boiler, to draw it from the fire when the latter required to be replenished. A leak in the riveting of the iron boiler allowed so much steam to escape that the boiler did not give way on the first trial. As soon as the escape of steam was observed to cease, the boiler was removed from the fire and again half filled with water. The fire was urged, and the boiler settled lower into it, and by once replenishing the fuel, without removing the boiler, an explosion was produced. Part of the committee were engaged in observing the progress of the experiment at this moment. The fire was near the middle line of the boiler, burning not strongly near that line, but very rapidly below the boiler; the steam issued freely through the leak before alluded to, and the whistling sound which it produced, and which had increased gradually in strength as the experiment progressed, seemed constant. The length of time during which the steam had escaped showed the water to be low, and induced the supposition that a second time the object would fail; when an explosion occurred. The explosion tore off one of the heads, bc, of the cylinder, projecting the other parts of the boiler in an opposite direction, carrying with them for a portion of the distance, the iron cy

linder forming the furnace, and scattering the fuel in every direction. The report attending the explosion resembled that from a small mortar (eprouvette) fully charged, the steam mixed with the smoke was not considerable in quantity, and few marks of water were to be seen. The boiler head was thrown fifteen feet, the boiler and spring register about six feet, and the furnace, weighing about forty-five pounds, was overturned and carried four feet. The pressure indicated by the register was eleven and a quarter atmospheres.

In examining the boiler it appeared that the head, b, which was thrown off,

had first struck against the iron furnace, which had deflected it outwards; this is shown by the indentation, b c, in the figure. This head was forced off all around in the line of rivets which attached the head to the boiler, the metal remaining between the rivets being less than the space occupied by them. The convex surface and the other head were thrown likewise against the furnace, and the head indented at de, overturning the furnace and carrying it four feet, as already stated. The boiler finally struck against the side of the bank of earth. The piston of the weighing machine was somewhat bent in the experiment.

The circumstances of this experiment show that the steam rose quite gradually on account of leaks in the boiler, increasing, probably, more rapidly as the quantity of water diminished, the intensity of the fire meanwhile increasing. That at a certain period the tension within had attained about eleven atmospheres, when the boiler exploded violently.

The accompanying figure will serve to give an accurate idea of the appearance of this boiler after its rupture.

The cylinder of copper, before referred to, was next put in the place of the iron boiler, and the fire again kindled; the general arrangements being as before described. This boiler being longer than the former would not descend so far into the furnace, and an attempt to raise the steam sufficiently high to burst it failed: there was a considerable leak in the junction of the curved surface with one of the ends. When the water was nearly exhausted, the fire having passed its period of greatest heat, the cylinder was removed and water again introduced, filling about three-fourths of its capacity. A new furnace was constructed of stones, allowing the boiler to rest more closely upon the fuel and affording a screen from the wind which was blowing quite strongly. The part of the boiler in which the leak had been observed was turned downwards, but a similar escape was found for the steam in the part now uppermost. The tension of the steam appeared to increase very slowly, and the fire passed its best action without effect; it was renewed, and as the water became lower the tension of the steam increased considerably. As before, nothing remarkable occurred previous to the instant of explosion, and the members of the committee, employed in the experiments, were engaged in observing the boiler at the instant it exploded. A dense cloud of smoke and flame, capped by steam, rose from the pit; the stones and combustibles were

widely scattered, and the boiler was thrown, in a single mass, about fifteen feet from the furnace. The noise attending this explosion was like that from the firing of an eight inch mortar.

The boiler was rent as shown in the accompanying figure, giving way in

an irregular line just above the probable water line on one side of the boiler, but not conforming to it. d and b were the lowest points in the two heads before the explosion. The sheet of copper was torn from the heads, unrolled and irregularly bent, adhering to the heads for only a short distance near the top of each; and the heads were bent outwards. The thickness of the copper along the line of rupture varies from .025 to .035 of an inch, and the metal appears to have been highly heated at one end of the torn portion. The piston of the spring gauge was bent, the screw which attached it to the boiler broken, and the whole instrument otherwise injured; it appeared that the wire intended to draw the boiler off the furnace had slipped and impeded the action of the piston, so that no register of the amount of force producing this explosion was obtained.*

The circumstances, as before, show that the steam was allowed to rise gradually until the boiler gave way. It is possible that there may be a relation between the space occupied by the water and that in which the steam is formed most favourable to the production of steam, and that when this was attained a rapid rise in elasticity took place; but there were no circumstances observed which would confirm such a view, and if it were correct it would only affect the conclusion as far as the increase of tension might have been rapid from such a cause.

As in the former case the marks of the sediments remained in the boiler, and indicated that the water was about an inch deep when the cylinder exploded. Much more steam was formed, and more water left than in the first experiment.

These experiments, together with the one referred to in a subsequent part of this report, are direct and conclusive; they show that all the circumstances attending the most violent explosions may occur without a sudden increase of pressure within a boiler. There can be no doubt, however, that if particular

*Assuming the strength of copper at 36,000 lbs. to the square inch, and that it was uninjured by the heating, neglecting also the effect of temperature, the bursting pressure appears by calculation, to have been about sixteen atmospheres. It was, no doubt, less than this.