boiler, there are so many practical reasons against it.

The next best theoretical form is a cylinder, which may be of any size, and only suffers from weakness at two ends.

With cylindrical boilers there is a difficulty about strengthening the ends.

Theory would tell us to make the ends hemispherical, when their strength would be as much as twice that of the cylindrical sides. There is, however, a practical objection to forming a hemispherical or egg-end, as it is called, from the boiler plate; and the result is that the ends are commonly made of flat plate, which is strengthened artifically in such a manner as to give security against danger. As to the strength of boiler-plate the general rule is that a rolled plate is less strong per square inch of section than a thick bar of the same iron.

Also the reduction of strength is more marked in the transverse than in the longitudinal direction. There is a further subtraction from the strength by riveting; and according to Sir W. Fairbairn the breaking strains of riveted joints of boiler plate are estimated somewhat as follows:

Since the plates after rolling are stronger in the direction in which they are rolled than transversely, it is apparent that they should be so disposed that the direction of greatest strength should encounter the greatest strain.

But the strain on the longitudinal section is greater than that on the transverse, and hence the plates, which may be 3 feet wide, are wrapped round the circumference, being laid in three lengths in order that the longitudinal seams may clear the brick-work seatings.

As a consequence of the theoretical disproportion between the two strains, it is further recommended that the longitudinal seams should be double-riveted with 3/4-inch rivets, pitched about 21⁄2 inches longitudinally and 2 inches diagonally.

The transverse seams are single-riveted, the pitch being 2 inches. To double-rivet them. would appear to add but little to the strength of the boiler, though it would increase its weight and cost.

The tubes diminish the area of the flat ends and relieve the pressure tending to rupture the material on a transverse section. They further act as stays for holding the ends together.

On both these accounts a boiler with internal tubes becomes stronger than it would be without the tubes, so far as transverse rupture is concerned. The flat ends remain, however, a source of anxiety, and in order to support them gusset

stays and tie rods are employed, as to which something will be said after the disturbing action of heat has been noticed.

EFFECT OF HEAT.

The wear and ultimate strength of a boiler are greatly influenced by adopting a construction which shall provide for the inevitable changes of form caused by unequal expansion and contraction due to changes in temperature.

Heat is motion, and as soon as the fire inside a furnace tube is lighted, the metal on the top becomes more heated than under the surface, and the tube arches itself in consequence of the greater expansion of the hotter portion.

And not only so, but the tube lengthens as a whole, and the flat ends bulge outwards. Finally the water becomes heated and the whole structure elongates, and unless sufficient allowance be made for the pulsating movements, straining will occur, which may possibly end in rupture.

The linear expansion of wrought iron (soft forged,) under the action of heat is stated to be .0012204 for a rise in temperature from o° C., to 100° C.

Thus a bar of iron, 30 feet long, expands about inch for a rise in temperature of 132.2° C. The expansion of the parts of a boiler as caused by heat is of course capable of accurate

measurement; and in particular the so-called hogging of a boiler tube has been observed by applying three gauge rods attached at equal distances along the crown of the tube; each rod is carried vertically upwards, and passes through a stuffing box in the shell of the boiler, whereby it has been possible to observe very accurately the distortion of the tube.

One boiler experimented on was 28 feet long, and it was found that the tube rose 3/8 inch when the flame passed around the boiler in the ordinary way along the side flues, and that it rose. 1⁄2 inch when the flame was carried directly into the chimney without heating the outer shell.

The gauge-rod at one-fourth the length of the boiler from the front end rose as much as that, and in one case in more.

Also the colder the water at starting the greater was the distortion, and generally the action was more severe just after the lighting of the fires. As soon as the whole of the water became permanently heated the gauge rods retired to their primary position, the distortion of the tubes seldom lasting for more than an hour. Mr. Fletcher recommends that the end plates of boilers, to be used at a pressure of 75 pounds per square inch, should be 1⁄2 inch in thickness, increasing to inch for increased pressure within moderate limits, excessive thickness being undesirable, as confining or restraining the

necessary movement of the furnace tubes.

The

object is to strengthen the end plate; and yet to preserve its elasticity, and in carrying out this intention it is a rule to attach the plate at the front of the boiler to the shell by external angle iron. This mode of construction is not, however, adopted at the opposite end.

The furnace flues are a vulnerable part of the boiler, inasmuch as they are liable to yield by collapsing unless sufficiently strengthened.

The subject of strengthing the internal tubes of the internal fire flue boiler was investigated by Sir W. Fairbairn, whose experiments led to the following conclusions:

(1) The strength of a tube to resist collapse by external pressure is inversely as its diameter. (2) The strength varies inversely as the length.

(3) The collapsing pressure in pounds per square inch

(Thickness of plate in inches.) =S06300XLength in feet diameter in inches.

2.19

In these experiments the ends of the tubes were firmly attached to rigid plates, and the vessel in which the compressing force was applied was a cast iron cylinder 8 feet long, 28 inches in diameter and 2 inches thick, which could be safely strained as far as 500 pounds per square inch.

Into this cylinder air was forced by a pump, and produced any required pressure on the sur