made to take up any of the positions shown in the end views (figs. 285, 286, 287). In the first case the lines radiate from the furnace centre, and it is clear that the metal will have been much stretched, and consequently much thinned, at the corner. This happens if the corner is made very hot during flanging, and is also flanged first. In the second case the lines which extend up to the corners are parallel, showing that thus far the metal has not been stretched or thinned. that is not the case from the corner to the crown, where the lines spread out very much. This can only happen if the centre has been heated and flanged first, and kept hot while flanging the corners. Fig. 287 But FIG. 288 FIG. 289 FIG. 290 shows how the lines should have distributed themselves. The available means for attaining this end are: firstly, not to heat the whole width of the flange, but only the curved part, leaving the edges of a dull red heat; secondly, to do the flanging gradually, and not to finish one corner outright, but to partly flange first one corner, then the other, and so on; thirdly, by leaving an extra amount of metal on those parts of the flanges which are most easily drawn thin, and then if their edges are not heated too much they might even thicken some parts near the curve by stumping them up. FIG. 291 The most difficult furnaces to flange are those with high saddles, as shown in figs. 288 and 289. As nothing is gained by making them of this shape, it is better to keep the flanges as low as possible, as shown in figs. 290 and 291. The deep flanges usually require from fifteen to twenty heats, while it is asserted that some smiths can produce the low flanged saddles in five or seven heats, but in that case the curvatures must be kept large. Each heat requires about one hour. The Flanging Operation is carried out as follows:-The anvil mould M is shaped as shown in section (fig. 292). It is part of a segment of a circle, and is bolted to a strong bed plate, B (fig. 293), to which another light frame, F, and a stop, S, have been secured. The furnace is heated at the point where it is to be flanged, and laid on the mould, and the flanging is then done with mallets. FIG. 292 Care has to be taken while doing this. If the edge of the plate were to be struck first, as shown in fig. 294, it would chiefly bend and stretch the edge (see fig. 295); it is therefore usual to strike at first nearer the anvil block (fig. 296). This blow produces little effect at the edge, which therefore remains thick. as shown in figs. 297, 298, 299. FIG. 297 Further blows produce curves, By starting from the inside more pressure is exerted there, and the tendency to contracting the furnace at the saddle is reduced. It has already been pointed out that the stretching of one part of a plate sets up compression stresses in other parts, which show themselves by a swelling up (see fig. 300); but this is easily removed with the help of a facing iron, as shown in fig. 301. The effect of a hard tool is to stretch the metal locally, and that naturally increases the diameter of the furnace at that part. This bulging was once a very noticeable feature in some patent furnaces. When allowed to re main, such faults have repeatedly been mistaken for indications of weakness or partial collapses, and much unnecessary anxiety or expense might have been saved if they had been removed at first. Very little flanging is required at the corner, for the corners help to draw this part out. When all this work has been done the furnace will have the appearance shown in fig. 302. To FIG. 302 finish it one of the corners is heated at a time, and placed on an anvil mould, as shown in fig. 303. It is then hammered to the correct T shape. If the preliminary flanging has not been carried out carefully, troubles may be expected, not only at the corners, but also with the round parts of the furnace, which easily lose their shape. FIG. 304 FIG. 303 FIG. 305 These anvil moulds (fig. 303) may be dispensed with if the curvatures of the corners are sufficiently large, but this demands a slight deviation from the usual design; either the furnaces have to be flanged round the tube plate, or the top corners of the saddle flanges have to be left long, and the radius of the tube-plate flange increased locally, while the combustion-chamber plate is cut away, as shown in fig. 304. This entails chipping and is expensive. A cheaper and equally efficient plan is to plane the combustionchamber sides parallel, and to make the tubeplate flange deeper at its lower end. The furnaces should now be thoroughly annealed, gauged for roundness, and corrected before cooling, and, above all, the vertical part of each saddle flange should be made perfectly flat, so that it fits the flat part of the tube plate without having to be reheated. Where the flues are not supplied ready flanged by the steel works it is best to fit them and the tube plates together before annealing, and to heat and work the corners when bolted together, as is done with the end plates. If the two are fitted together after this process, it is best, but difficult, only to heat the corners of the tube plate, because it is easier to anneal it than an entire furnace. If these various precautions are not adopted it will often be found that the furnace saddle and the tube plate fit as badly as shown in fig. 305. Such work can be detected by cutting out occasional rivets. In some works this fitting is done cold, but then the use of heaters cannot be prevented, and brittleness, due to working at a blue heat, may be the result. In some works the furnace saddle flange is left very deep, particularly at the corners, as shown in figs. 288, 289 (p. 271). In others it is kept as small as possible (figs. 290, 291). The latter plan, particularly if the curvatures are not too sharp, is by far the best; not only is it easier to do the flanging, reducing the number of heats from over twenty to less than ten, but, on account of the gentler treatment of the material, its liability to crack, after the boiler has been put in use, is so very much reduced that no fear need be entertained on this point. That these troubles are not alone due to unequal expansion of parts of the boiler, is proved by the fact that it is the corners of the central furnaces which generally crack, while those nearest the shell, where the strains are certainly most severe, do not suffer so often. But recent numerous failures cannot be said to have fixed the blame on the quality of the material, and the only alternative explanation is that the steel has been injured during flanging, either by burning it, by heating it too often or over fires giving off noxious vapours, by insufficient annealing, by manipulating the corners at a blue heat, or by overheating and straining when in use. Flanging Tube Plates.-The preceding remarks make it unnecessary to add much about the flanging of tube and combustion chamber plates. As previously mentioned (p. 257), it is usual to flange the straight edges first, and to leave all the corners to the last. This practice need not be adhered to when doing the work under a press. In order to obtain the correct shapes, machine flanging must be done with suitable moulds, and then it irregularly, as shown in fig. 306. By keeping the edges of the plates hotter than the curved part this thinning action is reduced to a minimum. If the edges have been flanged before the corners, the flanging mould M (fig. 308) should have a good taper, so as to press the finished flanges firmly against the mould block M, (fig. 307); for if M is left nearly square there will be a strong tendency to draw the plate away from M,. This is particularly the case when flanging the corners first, as shown in fig. 307; and then strong stops, SS, SS, have |