pointing toward the shank. I am convinced that this break is from solid stock, not being caused by an imperfection in the steel.

Mr. W. E. Crane.-The peculiarity of steel shrinking when hardened is valuable in many industries, such as dies for drawing tubes, rivets, etc. When a die becomes worn, it is a simple matter to take it to the blacksmith and have it re-hardened and shrunk. If steel would do this indefinitely, these dies could all be worn out on one size, but there is a limit to the number of times that the same piece of steel will shrink, this number being from five to seven, after which it does not shrink. It is possible that steel might be re-heated and cooled seven or eight times-if it would not be injured—and then the tool ground to size and hardened and retain its size.

Mr. Ezra Fawcett.-We had occasion some time since to make some large taps and dies for bridge bolts, and being in a hurry, the forger in annealing left them in a bed of charred (bituminous) coal on the forge over night, to give them a good "soaking," as he called it. On working the steel, we found it to have a very coarse, crystalline structure and brittle. Needing them immediately, we finished them up, tempered, and put them to work. One of them broke after threading some hundreds of nuts, but did not show as large a crystalline structure as before tempering; the others have been in use ever since. The steel was ordered for the special purpose from a well-known manufacturer in Pittsburgh, and had every appearance of being first class.

Mr. Thomas S. Crane.-I am surprised that no one has alluded to the peculiar formation of the ingots from which high carbon steel is produced, and I will call attention to the fact that all such ingots are defective at one end, and that such defect is embodied in the bar when the ingot is worked up, and is only eliminated by a tedious process of inspection in the mills.

I believe that, with the exception of those breakages which arise in hardening from the peculiar shape of steel articles, most of the flaws and cracks are produced in hardening by the hidden imperfection ordinarily existing in the ingot and afterward preserved in the finished bar.

High carbon steel, used for making tools and for other purposes when hardness is required, shrinks a great deal in cooling, and the ingots, as shown in the illustration (Fig. 89), always have a pipe, P, in the upper end, extending from one-quarter to one-half of its length downward.

Р

The illustration (Fig. 89) is taken from a photograph of an actual ingot, split open to exhibit the extent of the pipe, and it will be apparent that the inner sides of the pipe, if at all exposed to the atmosphere before the ingot is worked down, become more or less oxidized, so that no amount of hammering or rolling will perfectly weld them together.

FIG. 89.

I

When the ingot is worked down into a bar of any size whatever, the lack of union between the opposite sides of the pipe forms a flaw or seam, which is quite discernible to the eye when the bar is broken upon its end; and it is common for the inspector to break foot after foot from the end of the bar to remove the injured portion, so that the remainder may be sold with confidence as a sound article.

It is very evident that a point in the bar would be reached where the defect would not be perceptible to the eye, but exist in sufficient degree to cause a crack when the metal was exposed to any internal strain in hardening.

It is not merely a theoretical conclusion "that a crack would arise when hardening where the defective union between the sides of the pipe remain, as it would weaken the cohesion of the steel at that point;" but it is a matter of common practice in testing samples of steel for such defects to break a piece from the end of the bar and harden it to see if it will crack.

No system of inspection is perfect enough to prevent infallibly the existence of such cracks in the steel, and it appears to me that it is the defect or pipe in the ingot to which we must trace many of the extraordinary cracks which arise at peculiar and unexpected points in steel articles when hardened.

I hope to present a paper at the next meeting upon the means used to prevent piping in ingots, and have some interesting examples of the defects. caused by the pipe in the finished bar.

Mr. F. W. Dean.-It is a matter for congratulation that failures

of steel are diminishing, or at all events such astounding failures as the splitting of boilers when under water pressure have not occurred for some time. If this were not so, it would indeed be a thankless task for those engaged in the steel industry to see no response to their efforts to perfect.

While we in this country can feel pride in the results of our efforts to perfect certain kinds of steel, we are lamentably behind in securing good quality to heavy pieces, whether cast or forged. Not having had demands from the Government for gun steel, which doubtless has been a potent factor in Europe in influencing for the better the qualities of forged steel, all processes have been of the most inefficient kind.

As for steel for general purposes, testimony is contradictory. Among railway master mechanics opinion is divided as to the relative merits of steel and iron for axles and crank-pins, while there are hardly any men in opposition to steel for boilers. I suspect the reason for this is partly due to our making better steel for boilers than for other purposes. We, in fact, probably make the best steel boiler plate in the world.

Steel appears to labor under the disadvantage of inertness of accommodation to conditions. It must be humored, and he who succeeds best in ascertaining its peculiar nature, and caters to its weakness of character, becomes the best designer of steel structures. It can be laid down as an axiom that, when it is to be in much stress, steel should be free from sudden changes in size and form. Large fillets should be used, and key-ways should be well rounded at the corners.

It is very satisfactory to know that there are at present several specifications out for locomotive boilers having butt joints and covering plates, the inside one wider than the outside. This is one step toward making American locomotive boilers equal to English. A higher ideal of a boiler is still desirable. The shell must not only be good, but all forgings for braces must be at least respectable. Crown bars for supporting crown sheets should be things of the past, and their places should be taken by stay bolts between parallel plates. It should no longer be possible to find in use a riveting machine with a plate-closing ram, which never fails to make a circular indentation around the rivet. The plate within this ring is nearly useless, because it is bounded by steel which has been mostly strained beyond its elastic limit.

Several years ago a peculiar possible cause of failure of a steel

boiler joint came under my notice, when doing some experimental work for Mr. Leavitt, to which I have never seen reference. I refer to the effect of a non-axial pull on a joint, of which the following is an account:

A competitive test was being made of steel furnished by two makers for some 90-inch boilers of the locomotive type, having 2,800 square feet of heating surface each. The steel itself was tested and all of its important physical qualities noted, and the remainder of the plates made up into joints. The test pieces were each 34 inches long, 8 inches wide at the ends, and reduced to a finished width of 5 inches for a mid-length of 24 inches. The qualities of the steel as shown by the Emery testing machine at Watertown, Mass., were as follows:

Among 12 joints made from these plates was a butt joint of inch. A plate with a inch covering plate on one side and a inch covering plate on the other, the latter extending sufficiently beyond the former to permit three rows of rivets, with rapidly increasing pitch, to pass through it and the main plate, while it— the main plate-and the thicker covering plate were double riveted on each side of the center of the joint. There were thus ten rows of rivets in the joint, and its width was 15 inches. The length of the specimen was 5 feet, and it broke with a pull of 450,000 lbs. The fracture was in the main plate, through the

outer row of rivets (of greatest pitch), was granular, scarcely reduced in area, and it broke with a loud report. The preceding table shows that steel A, of which this joint was made, had excellent qualities, and it was therefore surprising that the fracture of the joint was short. It was observed, however, that, although the rivet holes were drilled with the plates in place, one of the remote side rivets sheared some time before the joint failed, and this suggested to Mr. Howard, in charge of the testing machine-who had seen similar phenomena that the character of the break was due to a non-axial pull. A test of a piece of the steel adjacent to the joint showed that it still retained its qualities, and thus the theory seemed to be confirmed. Since having had this piece of experience, I have often wondered if mysterious failures of boilers might not be caused by a one-sided pull.

In connection with this, I wish to mention the value of a high elastic limit in steel, or other metals, provided it is accompanied with other good qualities. When this combination occurs, the high elastic limit is due to excellence in material and intelligence in manipulation. Hundreds of tests have convinced me that the elastic limit and ultimate strength are in no way dependent upon each other, and, as steel is useless after the elastic limit is passed, it seems absurd to specify any ultimate strength in particular.

In partial support of these statements and views, I have given the particulars of specimens rather fully, and below give some general results of tests of the twelve joints previously referred to, of various designs, six being made of plate A, and six of plate B, in pairs, each member of a pair being an exact duplicate of the other. All edges of plates were planed and nicely finished, all holes were accurately spaced, drilled in place, countersunk slightly, and the rivets were closed by a steam machine.

Table showing the efficiency of certain riveted joints in comparison with the strength of the solid plates.