that unless the edge of the plate is well bevelled, and the rivet heads countersunk on the water side, steam-bubbles will lodge there. It is difficult to caulk this seam on the inside because of the furnace, and on the outside because of the tubes. The curvature of the tube plate side flanges is a very sharp one, in order to get the tubes as near the edge as possible, and when the furnace saddle flange is inside of this tube plate this curvature is often so sharp as to be more like a corner, and therefore very liable to crack (see fig. 43, p. 26). One way out of this difficulty is to increase the radius of curvature of the tube plate side flanges where they meet the saddle corners. This part

FIG. 347

FIG. 348

FIG. 349

must therefore be made deeper, and it is necessary to cut away the combustion chamber side plates, as shown either in black or dotted lines in fig. 304, p. 274. Another arrangement is shown in fig. 347 (see fig. 60, p. 40). The saddle being flanged over the tube plate, it can be made with a gentler curve than in figs. 344 and 345. The saddle seam is sometimes double riveted, and believing that the large amount of metal at this point would lead to burning, some engineers plane or chip this seam (see fig. 348). When it is necessary to fit very thick plates to the combustion chamber bottoms, they are sometimes planed thin at the seam before bending, as shown in fig. 349, but no object seems to have been gained by adopting this plan.

It is objected that the saddle seams of the last three constructions expose a caulked edge to the impinging action of the flame, but in practice they answer well. It is certainly far easier to caulk both edges of this seam, and there is no projection under which steambubbles could find a lodgment. A similar seam is shown in fig. 350,

which is adopted with ribbed furnaces and with those fitted with several Adamson's rings. The tube plate hole is not flanged, and can be carried to the very bottom of the combustion chamber.

The arrangement shown in fig. 351 is adopted when there is little spare space between the furnace top and the tubes, but as the work of flanging is doubled, and the caulking difficult, it cannot be recommended.

A somewhat more objectionable arrangement is shown in fig. 352, the tube plate being flanged to meet the furnace, and the seam being

exposed both to the radiant heat of the incandescent fuel and to the convection of the hot gases and flame.

The very greatest care has to be taken to secure metallic contact of the plates, and the flanged tube plate is sometimes bored to fit the furnace which has been turned (see fig. 353). In order to minimise the chance of leakage, and to reduce the labour of caulking, it is a good plan to remove the scale of all plates by pickling them in a 1 per cent. solution of hydrochloric acid, or to sponge the seams with sal-ammoniac. Other remarks about riveting these seams will be found on p. 39.

The back end seams are sometimes welded (see figs. 424, 425, p. 303), and an excellent job can be made of them, but in case of collapse they are apt to crack.

Riveting Saddle Seams.-While being riveted, seams invariably stretch, more particularly if the holes are countersunk, and generally the two plates do not stretch equally, and the result in saddle seams is either that the plate with the outer flange grows longer than the inner one, whereby the previously well-fitted corner opens, or else the inner plate stretches more than the outer one, causing the flange of the latter to crack. Therefore, under no circumstances should riveting be commenced at one side and carried across; it is also not well to commence in the centre and work to both sides. The best plan is to commence at both sides and work towards the middle, alternately riveting one side and then the other. If the reheating of corners without subsequent annealing were not objectionable it would be well if the saddle corners could be closed up hot after riveting (see p. 270).

The Screwed Stays between the combustion chamber backs or sides and the back end or shell of the boiler are sometimes fitted before the circumferential seam of the front plate is riveted, some

times afterwards. This being done by hand, all the holes will have been previously drilled or punched, and the plates annealed; they are tapped when in position, and the stays screwed and beaded over, or caulked and nutted. If the tapping and screwing of stays are done by machinery, the drilling is also done by the same machine (see p. 237). The sawing off of the ends should not be necessary, for accurate measurements of the lengths could have been taken, and it is certainly cheaper to saw off the correct lengths in a small machine than in one of these very expensive ones. It takes about 10 minutes to tap a pair of holes and to screw in the stay, and another 20 minutes to caulk and to screw up the nuts. When done by hand, the whole operation takes about one hour.

When screwed in by hand, these stays are often made with a square head (fig. 354). The labour of cutting off the ends would be

saved by the use of a closed nut with a shank (figs. 355, 356). Its depth should not exceed the thickness of a stay nut. Stay holes are always tapped when the plates have been placed in their final position,

FIG. 356

and immediately before the stays are fitted; otherwise it would be impossible to enter them. It does not seem to matter much that the threads of the stays and of the tap are not exactly alike. This difference is due both to the tap and the stays altering their length respectively during hardening and during screwing. The stays should be a tight fit, particularly at their inner end. This is not always the case, even in some of the best boiler-shops, and often leads to leakage with the least overheating. As the nutting of loose stays is difficult, they are at once scragged with a caulking tool, and inspection does not reveal the slackness.

The Pitch of Threads of the stays varies in different shops, the natural tendency being to make the pitch as fine as possible, because the effective diameter is measured from the bottom of the thread; but it takes longer to tap and screw them. There is little harm in fine threads when nuts are screwed on the ends, but a coarse and deep thread is necessary when the ends are only beaded over.

Cases in which explosions or mishaps have been traced to the use of fine threads are to be found in Engineering,' 1887, vol. xliii. p. 396, and 1890, vol. 1. p. 85.

Should the plate bulge, as it often does when hot, its inner (water) side will leave the screw threads entirely (fig. 357), and only the outer edge will hang on to the stay by the small riveted head and by a very few threads (fig. 357); whereas with a coarse thread, as in fig. 358, it would require a serious amount of bulging before any of the threads are quite clear of each other.

The threads of the taps and stays are never quite the same, which, when fine ones are used, may cause them to strip.

Stay Nuts ought not to be thicker, or only a very little thicker, than the plates which they have to support: not because of the danger of

burning them, but for fear of stripping the thread in the plate. This is illustrated in fig. 359, where, if the full power which such a nut could stand were applied, the threads in the plate would certainly strip. The drawing office habit of showing the stays horizontal instead of normal to the combustion chamber plate, and the carelessness with which the holes are drilled, placing the stays at an angle, and making the nuts bear on one side only, add to their power of doing harm. The taper washers which are then fitted often make matters worse by turning round, the thick part being found where the thin part ought to be. To prevent this, the bearing side of the nut should be faced. The

FIG. 359

FIG. 360

FIG. 361

thick external plates may be recessed, as shown in fig. 360. This is easily done with a rose disc of the size of the nut it is slipped over the stay, the nut screwed down on it, and it is then turned with a spanner, its outer edge being formed square or hexagonal. In many works the stays are bent by striking long steel nuts temporarily screwed on with a hammer (fig. 361). Thick stays are not likely to bend, but some of the threads will get damaged.

One often finds red lead cement, and also flat washers, under the nuts in the combustion chambers. Neither are wanted, doing more harm than good. Nobody would contend that the cement could stop

U

any leakage, the circumference of the stay having been caulked; it can, therefore, only act as a non-conducting layer between the boiler plate and the nut, causing the latter to burn when exposed to the flame. The interposition of a washer doubles this danger. The cement is also sure to get into the threads between the nut and the stay, where it acts as an efficient non-conductor and prevents the nut being cooled by the stay. Fig. 362 shows stay bolts which are used by some builders.

FIG. 362

FIG. 363

FIG. 364

An ideally perfect stay should allow of the nut in the combustion chamber being brought into as absolute a metallic contact with both stay and plate as is possible: it will thereby be prevented from burning, and will give the most solid and efficient support to the perforated plate. Taper washers ought therefore to be dispensed with as much as possible on the combustion chamber ends of the stays, and in no case should the angle be so large as shown in fig. 363; in fact, the angle should not exceed where t is the thickness of plate and d the effective diameter of the stay. With a -in. plate and a 13-in. stay this would give an angle of, or 10° (see fig. 364). Near flanged seams it is difficult to fit nuts; then riveting should be allowed.

t

Caulking Screwed Stays.-Screwed stays are caulked with an ordinary tool before the nuts are fitted (fig. 365). Locomotive engineers,

FIG. 365

who do not seem to like nutted stays, have used hollow bars and made them tight by drifting. Mr. Yarrow has also adopted this plan, but on account of the impossibility of getting at the outside ends, he makes the hole larger at the inner end than at the other, and uses both drifts from the fire side ('N. A.,' 1891, vol. xxxii. p. 102, plate 22). The drift used should be threaded

so that by screwing on a nut it can be drawn out again.

Girders. These are fitted to combustion chamber tops and other flat parts in the inside of boilers, transmitting the pressure from these parts to others which are better adapted to support it. If the girders end near stays, these will have to be made of a larger sectional area, in order to support the extra load which is thrown upon them.