may contain before entering the boiler. Perhaps the rapid circulation in the constricted space of the water-tubes carries such deposit as may tend to form away from the interior of the tubes. On this point, as on many others, we must still look for the results of long-continued service before an authoritative opinion can be expressed. The marine engineers of Bristol are, however, carefully collecting and absorbing such information as can be obtained upon the subject.

THE members of the Institute of Marine Engineers in their last meeting at the end of April were much interested in the discussion as to the merits of white metal for marine engines, as compared with the usual brass or gun-metal. The experiences contributed were very conflicting, some critics speaking feelingly of the difficulties and trouble attending the use of white metal in bearings in their own experience, and others giving instances of excellent records of prolonged wear without renewal or trouble, and some instances where the substitution of white metal in eccentric straps had entirely remedied excessive wear which had formerly taken place on the brass liners. These varied, and apparently contradictory results would, we think, have formed a most valuable record as to the behaviour of white metal as compared with brass or gun-metal, had the speakers been able to have given the various compositions of the metals in the different cases, and the probable pressure per square inch upon the bearings in the different instances. The term "white metal" is most vague and unsatisfactory, as anyone may make up a white metal that, when made, may be of the most indefinable character. A small admixture of one or other white metal, such as arsenic or antimony, with other white metals such as tin, spelter, or lead, will entirely alter the ultimate character of the alloy, and it is therefore evident that a ductile, greasy white metal on a bearing where the pressure per square inch is small, may give excellent results in that position but would be useless in a bearing, where there is heavy pressure per square inch, or where there is a constant reciprocation of the bearing producing blows. The pressure per square inch is usually heaviest upon crank pins in any engine so we are not surprised to hear that big ends so fitted give trouble. On the other hand, eccentric straps. have a large surface for the pressure and a steady rubbing contact, where a greasy metal would have every chance to wear well. One of the known dangers also with white metal is that the bearings of this metal will get soft, and gall, and strip, thus increasing the trouble, at a much lower temperature than is the case with brass or gun metal, hence white metal bearings must be kept fairly cool and must have

large surface, that is, have small pressure per square inch to enable them to run cool. There seems to be a concensus of opinion that white metal does not run well where in contact with water, and that a mineral oil seems to give the bearings in white metal the best chance of good and easy wear.

OUR shipbuilders must look to their laurels, and the shipwrights and engineers must learn that internal quarrels are a rival's opportunity, when we examine the enormous strides that foreign shipbuilders, and particularly those of Germany, have lately been making. During the recent deplorable strikes in Scotland, foreign owner after foreign owner gave contracts to the German shipbuilder, and it looks as if such business means to stay with them. The tonnage now produced in German shipyards is three times that produced 10 to 12 years ago, and it is in the last few years that the progress has been most rapid, the increase in the last three years having been about 100 per cent. The enormous extension of the "Vulcan" yard at Stettin, the splendid new Schichau yard at Dantzic, and the development of the Blohm and Voss yard at Hamburg show plainly, to those who have seen them, that the Germans mean to meet any chance of orders with full determination to cope with them, and to keep repeat orders for the future. In one yard alone the increase of turnover has been from 7,000 to 30,000 tons in 12 years. The largest sailing ship afloat is of German, not of British build, and was launched a few weeks ago from Geestemunde, with a tonnage of 6,150 tons, and any of the large German yards will contract for the largest and fleetest steam greyhounds required. Apparently these facts are not generally recognized in England, and some journals pander to our insular vanity by trying to depreciate the efforts, and what is worse, the actually accomplished results of those efforts, of our continental rivals. It is said on good authority that the depreciatory statistics of the ship production of our rivals can only be arrived at, and is so arrived at, either by ignorance of the existence of many flourishing shipbuilding yards abroad, or by a wilful ignoring of their products. We believe it is much the same as regards other countries than Germany. The production of the Austrian Lloyd's is often overlooked, and also the production of the Imperial dockyards. Italy also is doing her best to build for her own necessities, and Belgium and France are not to be despised as rivals.

Hopper Dredger Contract.-Messrs. Wm. Simons & Co., Limited, Renfrew, have received an order to construct a large and powerful hopper dredger for Devonport. The vessel is to be employed in removing blasted rock, and is the second vessel ordered for this purpose.

double-acting, the steam being employed not only for lifting, but also for accelerating the fall of the tup. The massive base-block, which weighs 12 tons, is a single casting, and has firmly fixed in its upper surface four massive poppetts, with screws for setting and holding the lower die. The surface upon which the lower die rests is raised above the main level of the top of the block, so as to prevent the dies making a hollow in the upper surface of the block, after long use. The stamp is provided with the firm's usual stamping gear, by which, as soon as the stop valve is opened, the tup rises to the top of its stroke against a substantial buffer, and there remains until the lever is pulled, to cause it to fall and strike its blow, when it is thrown down by the top steam with tremendous force. On the lever being released it immediately returns to the top, and can either be left there while the metal is removed, or caused to strike the dies together again almost instantaneously if a second blow is required. The stamps are fitted with hydraulic kicking-out gear for loosening from the lower die stampings which are of such a form as to require it, and the hydraulic cylinder of this gear is shown near the front of our illustration. The great advantages of these steam stamps over the crude kick stamps still so commonly used, are the rapidity with which blows can be made to succeed one another, and the greater precision with which the dies meet. A decided movement is already taking place among stamping firms in favour of these more accurate and mechanical tools, as against the old appliances which have hitherto served. The firm are also busy with a large number of steam hammers, with wrought-iron rivetted framing, chiefly for export.

by Messrs. John Penn & Sons with the following result:TABLE NO. 1.-TEST OF SHELL PLATE OF MODEL. Length of sample

Width

Thickness...

...

=

=

2 in. 2.01 in. ⚫0675 in. Breaking strength 3 tons 27 tons per sq. in. Elongation ⚫575 in. 29 per cent. Experiments.-The model was first tested with water, with the ends loose and free to move, the end rings and all the end bolts being taken off. Under these conditions the longitudinal joint commenced to open out almost as soon as the cylinder was full of water and opened out in. under a head of water of 12 in. That is to say the resistances due to the friction of the joint, to the stiffness of the plate, and to the india-rubber liner, all added together amounted to about lb. to the inch,

a quantity which may be neglected in an experiment of this kind. The strength of the shell as an "endless ring" having Pressure been thus ascertained, the end rings were bolted on. was again applied, and the measurements recorded in Table II. were taken at various pressures.

When the pressure reached 270 lbs. the india-rubber liner gave way, and the leakage of water prevented any more measurements being taken, or the pressure being increased. On examination it was found that the liner had been punctured by being forced into a small crevice at the end of the longitudinal joint. As these crevices were certain to increase if the pressure was increased, it was thought useless to replace or repair the liner, as it would have been sure to be damaged again in the same way whenever the joint began to open to the same extent. The experiments were therefore concluded. The shell plate generally showed no signs to the eye of having been seriously strained, except in the immediate neighbourhood of the joint on the unriveted side, where two big bulges, one at each end, can be plainly seen. The loose edge of the shell plate bad opened out in. just about the same, as it had been drawn out with a head of water of 12 in. when the ends were disconnected.

Measurements -For reasons already given, it was thought unnecessary to make any elaborate arrangement for measurements of the model under pressure, and such appliances were used as are to be found in every workshop. It was expected that the T-iron butt strap would have proved sufficiently strong to enable a pressure sufficiently great to be applied to produce a decided and unmistakeable bulging of the plate, but the event proved that the joint was somewhat weaker and the plate somewhat stronger than anticipated, and consequently it was not possible to get a sufficient pressure to produce the permanent and prominent strains in the plate which had been expected, and which would have more than compensated for the absence of exact measurement. Probably a very little more pressure would have strained the plate beyond its elastic limit, but no chance was seen of joint or liner being strengthened so as to get any more pressure, and therefore the measurements which were taken form the basis of the argument.

When testing the model with its ends loose, a line was drawn on the free part of the shell plate at the edge of the butt strap. As the plate drew out of the joint the distance between this line and the butt strap increased. When this distance had reached a † in., the pressure was stopped for fear of damage to the liner. During this test the model was standing on its end, and a pipe was screwed into the coupling fitted for water gauge. The water in this tube rose 12 in. above the water in the model; there was, therefore, a head of water 12 in. at the top of the model, and of 22 in. at the bottom. The exposed part of the shell plate was, therefore, subject to a mean head of 17 in. of water, which is equivalent to a pressure of about 6 lbs. per square inch. Although the pressure was nearly twice as much at the bottom as at the top, the line on the plate was very nearly parallel with the edge of the butt strap. After the ends had been bolted on, the points 1, 2, and 3, were marked off at right angles to the joint, and 4, 5, and 6, near to the joint, as shown in Fig. 6, While the model was subject to the pressures given in Table II. the diameters at the points 1, 2, 3, 4, 5, and 6 were measured with a pair of ordinary callipers, and read off on a taper scale whose width increased at the rate of in. to the foot, or 1-384th in. for each 1-16 in. of height above a datum line. By this scale it was easy to measure fairly accurately the opening between the callipers, but in setting the callipers much would depend on the sensitiveness of hand and eye on the part of the measurer.

ends were fastened. At a pressure of 270 lbs. the total load on the surface of the shell of the model would be about 38 tons, a load which must have had some visible means of support, and there was no other means of support but the end bolts. We also know that the pressure on all parts of the shell must have been transmitted to the ends by stresses in the shell plate itself, because, again, there is no other means by which the pressure could have been transmitted. Knowing that a certain load was trasmitted through the shell plate to its ends, we can determine one set of stresses that must have been produced by the load.

From the form of the model it is evident that the load would be about equally divided between the ends and uniformly distributed over the bolts in each end. The washer rings at the ends were sufficiently stiff to ensure that at the inner edges of these rings the resistance of the bolts would be evenly distributed over the circumference of the shell plate.

Under these conditions, if Fig. 7 is a section through the model, X OX being its axis, and YÖ Y being the centre line of its length, it is assumed that the total load on the shell was evenly distributed over the circumferences of the circles A A B B'; and that this load was balanced by the resistance of the bolts, uniformly distributed, by means of the washer rings, over the circumference of the same circles. The shell plate at A A1 and B B must, therefore, have been subject to a uniform shearing stress.

Assuming that the pressure on the half of the shell between A A1 and Y Y1 was transmitted to A A1, and that the pressure on the half of the shell between Y Y1 and B B1 was transmitted to B B1 then the shearing stress per inch of circumference at A A1 and at B B1 would equal the total pressure on a longitudinal strip 1 in. wide and half the length of the cylinder. Taking the pressure at 270 lbs. per in., and the length at 10 in., the shearing stress at A A1 and B B1 would be at the rate of 5 x 270=1,350 lbs. per in. of circumference.

Similarly, if the pressure on the ring a a1, Y Y1 was transmitted to A A1 there must have been a shearing stress throughout the section of the plate at a a1 at the rate of 270 lbs. per inch of circumference. And in the same way it can be shown that if the pressure on the shell was transmitted to the ends, and balanced by stresses in the shell plate, there must have been a shearing stress at as a1, of 540 lbs. per in. of circumference.

Whatever other stresses may have been produced by the pressure on this model, we can safely say that a pressure of 270 lbs. on it must have produced a series of shearing stresses increasing uniformly from O at the centre to 1,350 lbs. at each end.

In so far as a cylindrical shell is a vehicle for the transmission of pressure to its ends, to that extent it must be regarded as a beam supported at the ends and uniformly loaded. And the distribution of shearing stress given above is in accordance with the universally accepted theory of stresses in beams.

As a general rule the shearing stresses in beams are so small compared with the bending moments, that they may be safely neglected. Under ordinary conditions it is generally true that a beam which can resist the bending moment due to a load, can be safely trusted to resist the shearing stresses due to that load. We must therefore consider what is the bending moment due to a pressure of 270 lbs. on the shell of a model, and what is the resistance to bending of the shell.

There are two, and so far as we can see, only two, ways of treating this question. We may consider each longitudinal strip, into which the shell of a cylinder may be divided, as a separate and independent beam, or as a subordinate member of cylindrical beam of which it forms a part. Thus a longitudinal slip 1 in. wide in the shell of the model may either be considered as a separate and independent beam, or as forming part of a hollow cylindrical beam 10 in. in diameter. In the first case the neutral axis of the beam would be at the centre of the thickness of the strip, the metal outside this axis would be in tension, and the part inside the axis would be in compression. In the second case the neutral axis of the beam would be the axis of the cylinder, and all the plate on the same side of the axis as the strip would be in tension, all on the other side of the axis would be in compression. The bending moment due to a given pressure on such a strip would be the same in both cases, but the resistance to bending would be very different. The total load on a strip 1 in. wide in the shell of the model, when the

M1 = M,, f

12.5 p

1536

=

f = 12.5 × 1536 x p 19,200 p.

=

If p 270, then f = 270 × 19,200 5,144,000 lbs., that is to say, that if a narrow longitudinal strip of the shell of the model behaved as an independent beam, then at the pressure of 270 lbs. its outer layer must have been subject to a tension, and its inner layer to a compression, at the rate of 5,144,000 lbs. per square inch of section. This condition could never have been fulfilled, the ultimate resistance of the plate being only about 60,000 lbs. A pressure of 2 lbs. to the inch would produce a stress of 38,400 lbs., which is probably much above the elastic limit of the plate's resistance to tension or compression. Now, steel plate, which stretches 28 per cent. of its length before breaking, cannot be strained beyond its elastic limit without showing outward and visible signs of its condition. If the longitudinal strips in the shell of this model had behaved as independent beams, then at a pressure of about 2 lbs. we would have seen the centre of the model swelling out visibly before our eyes. As it did not do this, we must conclude that the strips did not act as separate beams, but as subordinate members of the cylindrical beam of which they formed a part. In this case the bending moment due to a pressure p on a strip 1 in. wide would be the same as in the previous case, namely, 12.5 p Ms; but the resistance to bending of the beam would be, according to Rankine's formula for thin cylindrical beams,

M, = n. f. S x d,

=

a factor depending on form of beam, in this case =1.

the coefficient of resistance of the material,

= the area of the section through the beam

31.4 16

=

or 5 x ƒ 12.5 × P, 2.5 × P.

=

When p = 270, f becomes 270 x 2.5 =675.

That is to say, if a longitudinal strip in the shell of the model acts as a subordinate member of the cylindrical beam of which it forms part, then the bending moment due to a pressure of 270lbs. on this strip would be balanced by a series of tensile stresses increasing from 0 at the axis of the cylinder to a maximum value of 675 lbs. in the strip itself; and a series of compressive stresses, increasing from 0 at the axis to a maximum value of 675 lbs. per square inch of section, in the other half of the cylinder.

But exactly opposite to the strip which has just been considered there is another identical strip subject to an equal pressure but acting in the opposite direction. The pressure on this strip