tube expander which is used to force a parallel tube into a parallel hole may also be a cause.

Of the various attempted remedies, none have yet given permanently satisfactory results, and for the merchant service, where practically no troubles are experienced, no deviation from the present course would be advisable. It consists in fitting a sufficient number of stay tubes.

Stay Tubes. These are fitted before the others, in order to be able to screw up the nuts. If the tubes are left parallel, the back end requires to be screwed sufficiently long so that the inside nut of the front end can be inserted, as shown in fig. 390. The back end is

FIG. 390

FIG. 391

FIG. 392

beaded over, as in fig. 393, or nutted, as in figs 391, 394. Nowadays nearly all stay tubes are swelled at their front end (fig. 392), and both plates are tapped together. Fig. 395 shows a section of a tool which is very convenient for screwing or unscrewing stay tubes. Two halfround grooves, deeper than shown, are cut into the sides of a spindle which is just large enough to enter the tube, and two short lengths of steel of a lenticular section are placed into these recesses and held there by any convenient means. Their outer edges, being roughened,

grip the inside of the tube whichever way they are turned. Both tube ends are expanded and also caulked, and either beaded over or fitted with nuts. The other tubes are then placed in position and expanded. Stay tubes are insisted on in the merchant service, while the locomotives and American steamers do without them, but all the ends are beaded. Exhaustive experiments on the holding power of tubes are mentioned by W. A. Shock, 1880, p. 217. As regards taperended tubes experiments were carried out by Martens (Mitt., Berlin,' 1887, vol. v. p. 65). None of these results were obtained at steaming temperatures. (See p. 35.).

The time required to tap the tube plate, fit a stay tube, and bead

Light Blow

or nut it, is about one hour and a half. The taps are hollow, and can be adjusted on a long spindle to suit any length of tubes; these are threaded in a lathe.

Caulking. In order to explain what takes place during the duration of a blow, when the hammer, the caulking tool, and the plateare in contact, it will be necessary to divide the plate into layers of, say, 1 in. in thickness, as shown in fig. 396. The velocity of the hammer and caulking tool is imparted to the first layer, and quickly transmitted to the next, and further on. The pressure which is

PLATE

FIG. 396

required to transmit this velocity from the caulking tool to the first layer, and then from one layer to another, is at first in excess of the elastic limit, and produces a permanent deformation, as shown. But this pressure reacts on the caulking tool, causing it to rebound after only a few layers have acquired the high velocity; but, their mass being small as compared with that of the hammer, their pressure on the further layers is not sufficiently great to flatten and spread them The permanent effect will, therefore, be that which is illustrated in fig. 396. A little reflection will show that, the lighter the hammer, the sooner it rebounds, and the fewer the layers acted upon, and the

out.

Heavy Hammer

Light Hammer

Heavy Blow

FIG. 397

lighter the blow, the slower the imparted velocity, and the smaller the deformations. Thus, if twenty-five layers equal to in. are deformed by the blow of a 7-lb. hammer on a 2-lb. caulking tool, only half that number, equal to in., would be spread out when using a hammer weighing 3 lbs. and a 1-lb. tool. By increasing the velocity,, the force between the caulking tool and plate would be increased; and the swelling would be greater, but the distance to which this swelling extends would not be altered at least, not materially. In an exaggerated form the effects would, therefore, be

as shown in fig. 397. and then light ones.

For this reason heavy hammers are first used,

The next thing to be considered is the shape of the tool and the shape of the edge of the plate. In fig. 398 the edge of the plate is square, and the effect of the caulking tool is seen under it (fig. 399). The metal has simply been swelled up.

In fig. 400 the edge of the plate is bevelled, and the caulking tool is placed firmly against it. The result of the blow is seen above (fig. 401). Not only is the edge of the plate swelled up, as in the previous case, but the lower plate is scraped up, forming a small ridge; and, thirdly, the blow being directed downward, both plates are depressed; but as

there is more spring in the outer plate (lap), it will not suffer as much permanent deflection as the lower one, and the result will be that the edge remains slightly, but permanently, open. This view is further illustrated in figs. 402, 403. An exaggerated seam of this sort is shown in fig. 404, and it is quite clear that no amount of hammering on the slanting surface would caulk the joint effectively. The maximum angle met with in practice is 1 : 3.

A better result would be obtained by placing the tool as shown in fig. 405, but, on account of the thinness of the edge, the caulking would not be very deep (fig. 406). Another method is shown in figs.

FIG. 408

FIG. 409

FIG. 410

FIG. 411

407 and 408, and strongly recommended by locomotive engineers. The effect of a blow in this case is almost the very reverse of the previous one; for while the edge of the metal is being swelled, the plate under it is being struck down, and, as the hammer rebounds, the lower plate springs back and presses firmly against the caulked edge (see figs. 409, 410). A tool of this shape will cause the swelling to extend

FIG. 412

farther into the plate, and thereby give a larger bearing surface. Care should be taken not to make this tool too small, otherwise it will act like a wedge, and press up the outer surface (fig. 411). Special tools have to be used for inside corners and for some rivets; they are generally shaped as shown in fig. 412. It is painful to handle

them, and their work is never very satisfactory, The jar of the blow on all caulking tools is greatly reduced, if their ends are grooved, by striking them on a coarse file in a red-hot condition.

The time required for caulking is about fifteen to eighteen minutes per foot, and five minutes for one rivet head.

There is apparently no need for caulking the inside edges of seams, for they could never be relied upon for water-tightness; but when the plates are not close, as shown in fig. 413, the inside edge should at least be fullered to prevent a rocking of the plates and an occasional opening of the outer edge or loosening of the rivets.

FIG. 413

Pickling. Riveted seams are said to be tighter if all black oxide scale is first removed from their surfaces, and the Admiralty practice with regard to tube ends is to grind them bright, so that they shall be in metallic contact with the tube plate. The pickling fluid which is used for removing the black scale consists of 1 per cent. of hydrochloric acid in water. Sulphuric acid, more than any other, has the effect of making steel brittle, particularly the hard qualities (see p. 150).

Welding Operations.-Reference has already been made to the fact that many of the seams of the internal parts can be welded, thereby saving the labour of flanging, drilling, riveting, and caulking. At one time efforts were made to weld iron shell plates, but the results were not encouraging, many seams having subsequently to be covered with straps. The introduction of steel, and the difficulty of welding it, stopped all progress. For a time it was only the furnaces which were welded; for, as these are not subjected to circumferential tension strains, the Board of Trade and Lloyd's Register' raised no objection. It is, however, well known that even now great difficulty is experienced in keeping the welded seams of some patent flues intact during the process of manufacture.

No doubt can be felt that better results are now obtained than formerly they are doubtless due to improvements in the production of the milder qualities of steel (20 to 25 tons), which can be made almost absolutely free from sulphur and phosphorus, two of the most injurious impurities. The influence of various chemicals does not seem to be accurately known, but what information could be collected on this subject will be found on p. 140.

The most reliable test for ascertaining whether steel is weldable is to cut off a strip about 18 ins. long, heat its centre, bend it and weld it, bend back the two ends (fig. 415), and then pull the sample asunder in the testing machine. The welded surfaces will probably be smooth and bright, except at the edges, where small patches of metal have left one side and stuck to the other. The larger these patches, and the greater their number, the more weldable is the steel. This test is not applicable to iron, as this metal tears through at the corners. To test it welded joints are re-heated and bent; if good, the weld should not open (see fig. 414). The ordinary tensile test is of little use when applied to welded samples, for, provided the joint is sufficiently taper, the strength will appear satisfactory. A far better test is to trepan small rings out of a welded plate, tap them, and tear them asunder by