the compound conical shell, must constitute one element of the calculation, the breadth of the sheets the second, and the thickness. of the metal the third. From these data may be found the radii of the curves in which are to be placed the rivets, and also the convergency of the two straight lines along the two ends of each sheet. The difference in the exterior diameters of the two ends of each frustrum must, at their respective rivet lines, be equal to twice the thickness of the metal. The larger arc for the rivet holes must manifestly be described by a longer radius than the smaller, and the difference of the two radii is the breadth of the sheet between the curves.

The larger radius will in every case be found by adding to the mean exterior diameter of the boiler in inches, the thickness of the plate in the same denomination, multiplying the sum by the breadth of the plate and dividing the product by twice the thickness of the plate.

Effects of use and long exposure on the strength of boiler iron.This topic may be regarded as one of the most important which came under the notice of the committee. To treat it in all its bearings would demand far more of time and means than were at our disposal.

Table C (Frank. Jour., vol. xx., p. 106) contains the results of trials on four bars or strips cut from pieces of plate which had been long in use, and which were more or less visibly affected by the strain, or by the action of water and steam to which they had been subjected. They were cut lengthwise and crosswise respectively from the sheets out of which they were taken, one bar in each direction out of each sheet.

The two strips, Nos. 245 and 246, taken from a specimen of plate which had been burst, are represented in position by the dotted lines in the accompanying sketch. The length of the specimen is the original breadth of the sheet, and lay, as usual, in the direction of the length of the boiler, thirty inches in diameter, from the bottom of which, near the fire end, this piece was taken. Hence the shorter strip (a) is a length sheet strip, and the longer (b) a cross sheet strip.

Agreeably to what has already been shown, the former possessed the greater strength; but the second experiment on it made at the point m, where the swelling commenced, was found, as shown in the table, considerably weaker than the sections near the two ends of the strip.

The amount of the swelling on this specimen of iron may be seen by the two sketches, page 158, one of which is longitudinal and the other transversal; and the lengths of the ordinates, from lines at right angles to each other and both passing over the centre of the swelled part, at b, will be found in the accompanying table. The ordinates on the inside or concave part of the plate, commence fig. 1 from the centre of the rivet holes nearest to the place of rupture b. The points on the two axes were 1.075 inches apart.

The original distance in the line a.c., figure 1, was 22 inches. The length of the curve a b c, by actual measurement, was found to be 24.6 inches. Hence, the extension of the metal in this di

1

rection had been 21.10 inches in 22 5.10 or 10. 7 of the whole length.

The lergh in the direction of the curve de f, figure 2, was ori

ginally 10.6 inches. Actul measurement over the curve d bf gave the length 12 inches, showing an elongation of fibres in this direc

1

tion equal to 1.4 inches in 10.6, or 7.5 of their whole length. This greater extension of the fibres in the longitudinal direction of the sheet accords with what has already been proved by direct trials on bars cut from new plates.

The two specimens, Nos. 97 and 98, table C, were deeply corroded in certain parts, indicating the existence of local chemical actions, arising, probably, from inequalities in either the purity or the mechanical structure of the sheet.

In both sets of the above trials on old boiler-plate it will be observed, that though the mean strength is low, being under 50,000 pounds per square inch, yet the principle is still preserved which assigns a greater tenacity in the longitudinal than in the transverse direction of the rolling. The difference in this respect between Nos. 245 and 246 is 50,017-48,515, or 3 per cent. of the strength. of the transverse strip..

Between Nos. 97 and 98 the difference is somewhat less.

Effect of annealing on the tenacity of iron.—In a variety of cases the committee have endeavored to extend the range of their experiments so as to embrace the condition of a steam-boiler which, without being exploded, has suffered from the exposure of its fire-surface when destitute of a due supply of water, to the action of heat above redness. In a few of the trials at elevated temperatures this point was attained or surpassed, and the subsequent trials on parts near the places of fracture, in such cases, gave evidence that the condition of the metal in regard to tenacity had been altered. In the case of bar No. 13, table XXXI., it will be observed that annealing, which in that instance was produced by the folding over of the ends of the bar to obtain a more certain hold by the wedges, determined at once the place of fracture.

On specimens 199 B and 199 C, table CII., (Frank. Jour., vol. XX., p. 110,) which were wires manufactured at Phillipsburg from Juniata iron, were made several experiments to ascertain its mean tenacity in the ordinary state of the article. On other pieces of the same wires the process of annealing at high temperatures was performed. The results show that the maximum effect of annealing on one of the sizes was a diminution of 27.5 per cent., and on the other of 46 per cent. of the original strength.

In these instances the annealing was performed in a common smith's fire, without using any other precaution to defend the wire from oxidation than merely covering it with cinder.

This prevented, for the most part, any action of the air on the wire during the short time of its remaining in the fire. But to obviate altogether the objection, that oxidation might have some share in producing the weakening effect, several specimens of iron were, after being carefully gauged, weighed, and having their specific gravities taken, rolled up in several folds of clean sheet iron; the ends of the folds turned over and hammered flat, to prevent

access of air, and the whole then exposed for 15 or 20 minutes to the blast of a smith's fire, gradually raising it to a very high welding heat. In this state the package was withdrawn from the fire, often exhibiting one or two folds of the sheet iron cut through by the blast, but in no case extending to the enclosed specimens. When taken from the fire it was immediately buried in dead cinders, where it was allowed to remain until quite cold. The wrapping of sheet iron was then removed, and the specimens generally found with no visible change except a slight discoloration by black oxide of the metal, insufficient, however, to affect sensibly the weight of the specimen.

The five trials on Nos. 224 C and 254 D were among those which had been reduced or constricted by straining before the specimens were annealed.

No essential change of specific gravity from annealing was detected, except in the case of a piece which had been beforehand excessively hammer-hardened. This appeared to have been slightly diminished in density by the annealing process. When the specimens treated in this manner had been previously strained, nearly to the limit of their tenacity, and their original areas of section consequently much reduced, the precaution was taken to regauge them before annealing, so that we could employ the then existing areas of section instead of the original ones, as the basis of calculation for the strength per square inch. This course was followed with Nos. 224 C and 224 D.

Calculating the strengths on the areas of these bars, taken just before annealing, they were found as follows:

Instead of 36,052 we obtained.....

45,090

41,980

42,700

43,200

41,980

From table CI. (Frank. Jour., vol. xx., p. 109) it appears, that computing the strength in the comparisons from 11 to 15 inclusive on the area before annealing, and excluding Nos. 1 and 3 of the series, as not made at a sufficiently high temperature to effect the purpose, we get the mean tenacity of iron, by seventeen comparisons after annealing, equal to 45,117 lbs. to the square inch.

In a considerable number of cases, where trials were made at high temperatures, especially those above 1,000, the iron was left in an annealed state. In several of these the effect of the process is nearly as striking as where the heating had extended to the point of welding.

In other cases the difference between the strength, previous to annealing and that exhibited afterwards, was so small that it was difficult to refer it to any other cause than the original inequalities of structure.

It will be seen that, from experiment 3 to experiment 11, (table CI.,) the loss of tenacity by annealing follows very nearly the order of the temperatures at which the process was performed.