the same opinion appears to be held by Prof. Faraday.* The opinion that an explosion is rather due to the weakness of the boiler than to the strength of the steam may, in fact, be said to be universal. There is, indeed, a very complex train of mechanical, chemical, and physicochemical forces, leading to the deterioration and consequent destruction of a steam boiler, and it is probable that no other metallic structure is subjected to such complicated conditions. The pressure of the steam and the heat of the fire produce mechanical effects, while both the burning fuel and the water react chemically on the plates and in accordance with their varying chemical properties. Each of these agents play, so to speak, into the other's hands, furthering and quickening the other's progress. It is difficult to distinguish with strictness between the effects of each; and it is mainly for the sake of convenient examination that they be classified into-1. The effects of the pressure of the steam; 2. the mechanical effects of the heat; 3. the chemical effects of the fuel; 4. the chemical effects of the feed-water.

The Direct Effects of the Pressure of the Steam.

In calculating the working strength of a cylindrical boiler, the plates are assumed to be under a static load, and to be submitted to a tensile strain. The former of these assumptions is seldom, and the second is never correct. There are two principal causes that tend to exert impulsive strains on the sides of a boiler: 1. The sudden checking of the current of steam on its way from the boiler to the cylinder; 2. quick firing, attended with too small a steam room; and both may sometimes be found to act in combination. To the first of these causes the explosion, for instance, of one of the boilers of the Parana steamer, at Southampton, a few years ago, has been ascribed by the Government engineer surveyor;† to the second, the explosion of the copper boiler of the Comte d'Eu yacht, in France. According to Dr. Joule, the mere dead pressure of an elastic fluid is due to the impact of its innumerable atoms on the sides of the confining vessel. When the motion of a current of steam is suddenly checked, as by the valve, in its passage from the boiler to the cylinder, its speed and weight cause a recoil on the sides of the boiler analogous to the effects of the, in this case, almost inelastic current of water in the hydraulic ram ;‡ this action is necessarily most felt with engines in which the steam is let on suddenly, as in the Cornish, and other single-acting engines, working with steam valves, suddenly affording a wide outlet, and as suddenly closing. It produces such phenomena as the springing or breathing of cylinder covers, and the sudden oscillations of gauges, noticed long ago by Mr. Josiah Parkes and others.§ Some years ago while standing on a boiler working a single-acting engine, and with a deficient amount of steam room, the writer noticed the boiler to slightly breathe

* Proceedings of the Institution of Civil Engineers, 1852, page 392.

Rudimentary treatise on Marine Engines and Steam Vessels, etc. By Robert Murray, C.E., Engineer Surveyor to the Honorable Board of Trade, pages 74-78. Instituto di Scienze. Milano, 1829.

? Transactions of the Institution of Civil Engineers. Vol. iii.

with every pulsation of the engine. The same action has been observed by others with boilers, the steam room of which is out of proportion to their heating surface. The intensity of the instantaneous impulses thus generated would be, as Mr. Parkes observes, difficult to measure, but their repeated action must readily affect the boiler at its mechanically weakest points. The more or less sudden closing of a safetyvalve while the steam is blowing off would evidently produce the same effect; and this view is strengthened by the fact that the great majority of locomotive boilers-in which while at work there is no such sudden call on the reservoir of steam as in the Cornish engineexplode while standing with steam up at the stations.* It is not denied that, in the case of a locomotive, the mere extra accumulation of steam from the safety-valves being screwed down above the working pressure will also come into play. But there can be little doubt that most boilers are subjected, sooner or later, and with more or less frequency, to an impulsive load. This being the case, this consideration alone would demand a factor of safety of six in the designing of steam boilers. The Commissioners on the application of iron to railway structures, in their third conclusion on a mass of evidence which has made their investigations the most valuable ever conducted on the strength of materials, say, "that, as it has been shown, that to resist the effects of reiterated flexure, iron should scarcely be allowed to suffer a deflection equal to one-third of its ultimate deflection, and since the deflection produced by a given load is increased by the effects of percussion, it is advisable that the greatest load in railway bridges shall in no case exceed onesixth of the weight which would break the beam when laid on at rest in the centre."t

Emerson showed, more than sixty years ago, that the stress tending to split in two an internally perfectly cylindrical pipe, submitted to the pressure of a fluid from the interior, is as the diameter of the pipe and the fluid pressure. He also showed "that the stress arising from any pressure, upon any part, to split it longitudinally, transversely, or in any direction, is equal to the pressure upon a plane drawn perpendicular to the line of direction.' As in a boiler the thickness of the metal is small compared with the radius, the circumferential tension has been assumed to be uniformly distributed; and the strain per unit of length upon the transverse circular joint being only half that upon the longitudinal joints, the strength of the latter has been taken as the basis of the calculations for the tensile strength of the joints. But in taking the internal diameter of the boiler as the point of departure, the internal section has been assumed to be a correct circle, which would only be practically true in the case of a cylinder bored out in a lathe, and never in that of a boiler. Two of Emerson's corollaries from

* Reports of the Inspecting Officers of the Board of Trade, 1850-64. (The four locomotive boilers which burst last year all did so while standing. Neither the primary rupture leading to the explosions, nor the secondary rupture caused by the explosion, took place through the rivet holes.)

Report of the Commissioners appointed to inquire into the Application of Iron to Railway structures. xviii.

his first proposition have, in fact, been neglected. He shows that if one of the diameters be greater than another, there will then be a greater pressure in a direction at right angles to the larger diameter, the greatest pressure tending to drive out the narrower sides till a mathematically true circle is formed. The second is that "if an elastic compressed fluid be enclosed in a vessel, flexible, and capable of being distended every way, it will form itself into a sphere."* A number of proofs can be adduced that both these influences are more or less at the bottom of the wear and tear caused by the direct action of the

steam.

From 1850 to 1864 forty locomotive explosions, causing a loss of human life, have occurred in the United Kingdom. The Board of Trade reports in the Blue-books presented to Parliament, and more especially those by Captain Tyler, R.E., probably form the most valuable and connected series of records extant on boiler explosions. This is more especially the case with regard to wear and tear caused by the direct action of steam unmasked by the effects of the fire, as the barrel and outside fire-box of a locomotive cannot be said to be under the direct action of the heat. Perhaps the vibration of the boiler through the motion on the line may intensify this action, but it is clear that vibration cannot be a primary cause. The majority of the reports are illustrated by careful drawings. Eighteen of the forty boilers gave way at the fire-box-eleven from the crown of the inside fire-box being blown down upon the tube plates; seven from the shells or sides giving way. Twenty burst at the barrel; and two explosions may be ascribed to miscellaneous causes, from an originally defective plate, and from running off the line. Leaving out all those which occurred at the firebox, as the majority of these might be ascribed to other influences than direct pressure, all the twenty explosions of the barrel could be traced either to internal furrows or to cracks, both running parallel with one of the longitudinal joints of one of the rings forming the barrel. All the joints which thus gave way were lap-joints; and the furrows or the cracks (and the former greatly preponderate in number) occur at the edge of the inside overlap, and, therefore, just at the point where the diminution of diameter caused by the lap-joint would be most affected by the pressure of the steam. (See Fig. 1.)

The plate at the channels shows distinct traces of lamination through the cross bending, and it is probable that plate of a good material will

The action of a fluid pressing with equal forces in all directions can be evidently represented as to force and direction by innumerable radii of equal length led from a single point in all directions. Upon this principle may be explained the spherical shapes of soap bubbles, of the bulbs of thermometers, (blown while the glass was in a plastic state,) of the thin india rubber balls, used as playthings, and which are formed by forcing air into india rubber tubes closed at one end. Gas and air bubbles in water are necessarily flattened by the hydrostatic pressure. It is upon that principle that a gun of soft ductile iron often bulges out at the breech.

FIG. 1.

gradually laminate, while inferior metal will crack through in much less time. Nor are these furrows found with only lap-joints. Butt-joints, with a strip inside the boiler, and thus destroying the equilibrium of internal pressure, have been found to be attended with similar furrows. Channels of exactly the same character have been observed in locomotive boilers with lap-joints, which have exploded in Germany.*

Similar furrows, again, have been noticed in marine boilers, and in old boilers generally, longitudinal furrows being of course about twice as dangerous as those appearing transversely. The smoke-box tube-plates of inside cylinder-locomotive engines have been found to be similarly influenced by the racking action of the engines, showing furrows around the cylinder flanges. A parallel case is often found in Lancashire with the end-plates of double-flued Fairbairn boilers, which may have been too stiffly stayed to the barrel. Circular furrows, caused by the confined motion of the end-plates, are sometimes found at the base of the angle iron rings jointing the internal flues to the end-plates. But furrowing seems with no kind of boiler to be more felt than with locomotive boilers. This is due to the high pressure, to the thicker plates causing a coarser lap, and more especially to the fact that the unstayed barrel cannot be thoroughly examined without drawing the tubes, thereby enabling the furrow to enlarge itself unnoticed.

(Half size cross section of the furrowed longitudinal joint in the fire box ring of a boiler which exploded at Overton station, on the 30th May, 1864. It does not differ from other furrows.)

The inside fibres of a plate bent up while cold are necessarily initially in a state of compression. When the pressure from the inside comes on, striving to form a perfect cylinder, the plate gets bent to and fro by its own elasticity on one side, and by the pressure on the other. If the iron be brittle, it may crack right through; if ductile, the outside fibres gradually lose their elasticity, and, necessarily aided by other causes, crack away. This action is progressive, and probably very rapid towards its later stages. Once a weak place formed itself it would have to do more and more of the work. Even when pulled by the direct tension of the testing machine, a lap-joint behaves in a somewhat similar way. For instance, a half-inch lap, solidly welded by Bertram's process, has only half the strength of the solid plate; while the 3-inch lap-weld has actually two-thirds of the strength of the entire plate.

Messieurs Jean Piedboeuf and Cie, of Aix-la-Chapelle, Düsseldorf, and Liege, who turn out annually upwards of one thousand steam Organ fuer die Fortschritte des Eisenbahnwesens. 1864, page 159. t "Recent Practice on the Locomotive Engine," page 5.

boilers, use a lap-joint which probably gives slightly better results as to furrowing, while it is much easier to caulk, and must be therefore less injured by that process. (See Fig. 2.)

FIG. 2.

There is, however, another important appearance to be noted with respect to these furrows. An iron cylindrical vessel under internal pressure would, of course, rupture long before it could assume a spherical shape, from its ranges of elasticity and of ductility being so short. But it may be said to be undergoing three distinct stresses in as many directions. There is a stress acting on the ends, and tending to rupture the boiler in two halves in a direction parallel to the axis; there is the stress which is hoop tension in a true circle, but which acts with a cross bending strain in an ordinary boiler; and there is the stress which tends to make it assume the shape of a barrel, or to bulge it out in the centre of its length. The precise action on a material of several strains like this is a portion of the strength of materials which is still completely unknown. Its probable effects might be illustrated by the ease with which a stretched india rubber ring is cut through with a knife, or that with which a column under compression is broken by a blow from a hammer, or by the similar ease with which a tube under tension is split by a sharp blow; in fact, the operation of caulking a defective boiler under steam seems thus to often give it the finishing stroke which causes an explosion. The new boiler which burst from a defective plate at the Atlas Works, Manchester, in 1858, and that which burst through a crack at a longitudinal joint last January, at Peterborough, both gave way whilst being caulked. This again accounts for the fact that adjacent boilers sometimes explode one after the other, pointing at the same time to the danger into which a sound boiler may be thrown by an explosion.

(The edges of the plates are cut to an

angle of 65° by means of inclined shears.)

Upon the same principle it is probable that the modern guns, built up from strained rings, will be easily put hors de combat by shot. The probability is that a number of simultaneous strains in different directions diminish the elasticity of the material that would allow it to yield in any given direction. However this may be, it will be seen that it is only the pressure on the ends of the boiler acting parallel to the axis and tending to tear the cylinder through transversely, which bears fairly on the rivetted joint, or rather on that metal between the rivets which is left after punching. Unless the cylinder be perfectly correct inside, the circumferential strain resolves itself into cross bending, shifting the dangerous strain from the iron left after punching to the metal at the overlap. With respect to the stress tending to bulge the cylinder in the centre, it is clear that if we suppose a strip cut out from the entire length of the boiler, each portion of the length of this strip could be regarded as a beam under a uniformly distributed load.