welded up solid were inch thick, and the furnace tubes of 3 inch plate, the material throughout being of best Snedshill iron. All longitudinal joints were double-riveted, and circumferential ones singleriveted. The holes were punched.

101

ONGITUDINAL

FIG. 7.

CENTRE

In the first experiment that was made the boiler was complete in all its details. The manhole mouthpiece was of wrought iron, such as recommended by the association. The cast-iron pipe or seating for the blow-off elbow pipe to attach to was also as is usual, while the gusset stays and longitudinal stays were the full complement provided for such a boiler for 75 pounds working pressure. The front end manhole casting was placed on the inside of the front plate and furnished with internal door and usual cross bar fixings. The wroughtiron neck on the top of boiler and on the central plate was the special object of this first experiment, and it is clearly shown in Fig. 6, in which all the dimensions are fully given. Fig. 7 is a sectional plan of this neck, and Fig. 8 shows, as does also Fig. 7, the lines where fracture ultimately took place. Before pressing the boiler up to the bursting point, careful records were taken as to the behavior of the boiler at all points, as the pressure gradually rose. The flat ends were carefully gauged to ascertain at what pressure and to what extent they gave way, and if any permanent "set" remained on withdrawal of

pressure.

The furnace flues were carefully gauged; the length of the boiler shell measured by rods fixed at one end and free at the other end,

and having pointers at several places, and the circumference was also measured by two encircling steel bands passed round the boiler and weighted at each end, so that by horizontal lines drawn across them the least enlargement could be rendered clearly visible; some of these measuring preparations are shown in Fig. 9.

Up to the pressure at which the wrought-iron neck (Fig. 6) gave way, viz., at 250 pounds on the square inch, there was no movement of the ends beyond a slight permanent "set," which has remained the same throughout the whole of the test, even after the removel of the two longitudinal stays, by which fact it is clearly shown that the arrangement of gusset stays here used is sufficient to secure the ends up to and beyond the limit of the strength of the shell.

The first rupture took place at the base of the 14 inch neck, and is shown in Figs. 7 and 8. Of course these fractures are, in themselves, as here shown, only the commencement of the work of destruction, but it is self-evident to any one understanding the subject that had the volume of water which was supplied at the pressure of 250 pounds been greater, the rent would have extended instantly, and had steam pressure been the agent used instead of water then complete destruction must have followed, by the development of rents in various directions, and with the usual accompaniments of loud report, damage to property, and probably loss of human life.

The boiler having been repaired by the attachment of a strong circu

lar plate where the neck had been placed, as may be seen in Fig. 9 on the top of the boiler, was furnished with a cast-iron manhole mouthpiece of the usual form. This casting was circular and was 163 inches inside diameter, and is shown in Fig. 10, and may here be compared with the wrought-iron manhole mouthpiece, Fig. 11, which is 17 inches internal diameter, and upon a hole in boiler shell 183 inches diameter. It will be seen that the cast-iron manway is 163 inches diameter inside, but requires a larger hole in boiler shell, viz., 20 inches diameter, to bring the edge of the plate to its proper distance from the centers of the rivets. The casting, which was of strong and substantial pattern and quite sound throughout, gave way at a pressure of 200 pounds per square inch, a large part of it flying off with considerable force, and so much more violent was the rupture than had been the case in the instance of the wrought-iron neck at 250 pounds that the shell plate was torn completely through on each side of the manhole, the rents being longitudinal until diverted by the resistance offered by the two ring seams through which they forced their way, and thence they began to pass circumferentially through the neighboring plates. Here again are clear lines of commencing rents whose ultimate development it is self-evident must have been utter destruction. The fractured portions of this casting are shown in Fig. 9, as is also the first wrought-iron neck with its fractured flange.

At the next or third experiment a steam dome 3 feet in diameter was applied, only a small portion of the boiler shell having been cut away, but the strain at the base of the dome was such that violent leakage ensued, so that no more than 235 pounds pressure could be maintained with the pumping power at command. A second trial was made with this dome, when it had been re-riveted with stouter rivet heads, suited to resist more satisfactorily the strong upward strain that was produced by the internal pressure, and in this case rupture was produced by the tearing of the bottom flange of the dome on the centre line of the boiler at a pressure of 200 pounds. Up to this period the gauging of the flues and ends had been carefully noted, but no signs of distress were discovered, nor had any strain of the flat ends taken place.

The fifth test was made to ascertain the value of a single-riveted joint as compared with a double-riveted one, and it was found that at 250 pounds the leakage at the single-riveted joint was so great that

the pumps were overpowered, and no increase of pressure could be obtained; the double-riveted joint remained quite tight.

The sixth test was to ascertain the strength of an ordinary unguarded manhole with internal door and cross bars, such as have been in time past so very generally used, and are indeed but too often met with even now. The oval hole cut in the shell measured 17 inches by 13 inches, and in this case it must be remembered that the whole apparatus was new and of first-rate construction, so far as the imperfect system of such a manhole and cover could be good. The inner cover

lapped well to the boiler shell, and the joint having ample surface, was made with new sheet india-rubber. The cross-bars were of the proper length, so that their feet rested upon the shell plate, not beyond the limit of the cover below, but immediately over the lap, and thus the chances of the door were far above the average rate of strength. There was, of course, no loss of strength from repeated renewal of such a joint. or from corrosion, but the door threw out, as is shown in Fig. 12, at a pressure of 200 pounds per square inch, a longitudinal rent starting from one end of the manhole and running across the plate, and then extending some distance along the ring seam of rivets, and fracturing. the next plate through a rivet hole.

Test number seven was a repetition of number five, to try if a rent instead of a simple leakage could be produced with a greater amount of pumping power, the idea having arisen that perhaps a singleriveted joint might be taken as a sort of safety valve through which sufficient leakage and relief from pressure might ensue in case of overpressure, and thus lead to the prevention of actual explosion. Such a theory, if not disproved, might have a very dangerous tendency, and the necessity of complete assurance was felt to be absolute.

With augmented pumping power, this question was set at rest by the complete fracture of a single riveted seam (machine work) at 275 pound pressure, when the rent extended the whole width of the plate, and passed into each side ring plate, the rent being about 5 feet 6 inches long, and such as would most inevitably have caused the whole shell to open out with most destructive results.

Double-riveted work executed by hand was then compared with machine work, with a result similar to the last, in so far as this rent