Referring back to the qualities of the material, it will be seen that the plates having the lowest ultimate strength, the highest elastic limit, the greatest elongation and contraction of area invariably made the most efficient joint, and the obvious explanation is that the more ductile steel more perfectly fitted around and bore upon the rivets after the joint was in tension, and the higher elastic limit allowed this to go on for a longer time than in the case of the lower.

The low efficiencies of some of the joints are due to their not being designed for strength, but rather for tightness in trying situations.

Good material, thick ingots, and proper manipulation seem to be the requisites of good steel for boiler plates.

It is much to be desired that a uniform size of specimens for testing should be adopted throughout the country, and that the elongation should be taken in the same length by all. The specimen should not be too short and narrow.

Mr. Oberlin Smith.-Some years ago I had a great many pipe dies to temper of the ordinary form, inch and a quarter to two inch, the smaller dies being for three-eighths, half inch, etc. The first named were four inches square and one inch thick. The smaller ones were usually a half inch thick and two inches square. Sometimes we would harden a large batch, a hundred, or two or three hundred, with very few breaks. We usually used English steel of the best kinds we could get. At another time the same brand of steel in different bars would show a loss of from ten to fifty per cent. of the dies when they were hardened. Sometimes one jaw in a die, sometimes all four, sometimes two or three of the jaws would crack a little way in, but occasionally they would crack clear through the die and tumble out. I reasoned that probably on account of these small members inside cooling first and trying to shrink, pulling themselves away from the hot part outside which had not yet shrunk, there was a tensile strain which pulled them off. The general tendency of a ring of steel in hardening would be for the inside to pull itself away from the outside. In practice this cannot happen, in the case of a ring, because of the arch principle; but, in the case of these dies, the same tendency occurred, and the jaws, having no keystone between them, so to speak, cracked off and went inwards. After trying various ways, I hit upon a system of hardening the outside first. I made a little notch in the edge of the tub, in which revolved

a rod carrying the die in clamps at the inner end while the crank at the outer end gave means of revolution. The hot die was slipped into the tub of water and revolved. The consequence was that the corners commenced to dip first, and then the crank end was raised slightly, so that the die went deeper and deeper into the water, but the outside was hardened before the inside. That remedied the trouble almost entirely. We did not have a jaw break off afterwards. We did occasionally have a corner crack off, but the percentage of loss was slight, so that the scheme was a practical success. As far as my experience goes with steel I am much inclined to believe that there is no mysterious dispensation of Providence about it. I think we can trace most of the trouble to simple mechanical action. There may be some chemical action also or some irregularity in the composition of the molecules of carbon or iron, and strains may be thus produced about which we do not know. I do know that a great deal of the cracking we see can be accounted for by the simple principle involved in trying to work a very brittle material under strains which are too great for it. If we take a large tap or large punch, especially if it is as large as five or six inches in diameter, and, heating it red-hot, dip it in water, we very often find that part of the outside will crack off. Now this is evidently due to the outside cooling first, before the inside has had time to cool, thus putting the outside under a tensile strain while the inside forms an abutment and prevents it from going inward. A thin hoop of steel dipped red-hot into water and made almost as brittle as glass, is not very apt to crack. It can freely go in. We all know that a hole is sometimes drilled in the end of a large cylindrical piece of steel for the purpose of preventing cracking, with very good success. It enables the hoop thus formed to shrink as it wants to. We are all aware that thin light pieces of steel are not so likely to crack in hardening as solid heavy pieces. I think the irregular strains in steel are due in the first place perhaps to some irregularities in the homogeneity of the steel when it is cast; afterwards to irregularities of structure caused by rolling or hammering; and to irregularities due to uneven heating. Here are three distinct reasons why the steel is not homogeneous all through, but probably irregular heating has a great deal the most to do with it. Take any plate of steel-say a thin large plate that is homogeneous to begin with, having no internal strains. Now if we heat it around the edges a little more than in

the center, it expands first around the edge; the edge goes out; the middle does not; this puts a tensile strain around the edge, and that strain remains there, or partially so. There are sometimes strains in there that are almost up to the limit of strength, so that the least jar, or some slight molecular change in the steel itself, changes its shape just a little, and away it goes. It tumbles down to the floor, broken. I think that our chief remedies for this trouble of steel cracking in hardening are, in the first place, to keep it homogeneous as nearly as we can in the hammering and the rolling, and in the heating, and then in hardening, but being very careful to treat it, when dipping it in water, according to its shape. Each particular shape must be studied by itself, with a view to not putting a strain in by leaving the hot part to hold back against some cooled part-some part that has already been cooled and made brittle-thus subjecting such brittle part to tensile strains. Nearly every shape wants particular arrangement of tempering. Sometimes you can get a very good result by squirting water through a tube-sometimes through an annular tube. I have had tempered a good many rings (of special section) in this latter way with good success, laying a plate over them and squirting the water down through.

Before leaving the subject of hardening steel, I want to say a word or two about the treatment of steel in the fire to prevent burning. I believe a good deal of the burning of steel is done by the rapid action of the blast upon it, even if it is not heated up beyond the proper cherry red. Not only is the mischief done by too quick heating, but by the action of the air impinging upon the metal. To remedy this we want very deep fires, and must heat slowly and let the flames pass slowly by, so that a great volume of air does not touch the steel. I am not chemist or metallurgist enough to know whether we can decarbonize steel by blowing a quick blast upon it at a moderate heat; but, so far as I have observed, I think there is considerable action of that kind, where you have heated it rapidly, by letting a flame of oxygen blow upon the surface in a shallow fire with the blast coming up from beneath and blowing directly upon the steel. Hence, one remedy is a deep fire, and another is to cover the metal up as much as possible.

Prof. John E. Sweet. I was asked to call the attention of the Society to one peculiarity in steel, that is hardened steel, used for standard gauges. A gentleman in Syracuse is making measuring

machines, and has occasion to make length pieces for use in the machine. He finds that they are constantly changing their length after hardening. The question is, how long he should keep them before sending them out as standards of length. I presume Mr. Bond can give us some information in regard to the same subject.

I would also say in regard to standard gauges, where they have been ground to absolutely true cylinders set up on their end and allowed to remain for several hours and then measured, they were found to be the largest in their north and south directions, and, taking the same piece and setting it on end, and putting the north east and the south west, their diameters would again change, so that they would be largest in the north and south directions. This I give on the authority of S. Ashton Hand, a gentleman who has been engaged in the standard gauge business.

That standard pieces an inch and two inches and three inches in length do change in length, is without doubt, and the thought has occurred to me about some way to prevent it. Should we subject these pieces to an end-pressure before they are finished? Would that help the difficulty? I believe they grow shorter, and by subjecting them to a certain amount of end-pressure, it might take out that tendency at once.

Mr. Geo. M. Bond.-I would like to answer the question as far as it is possible for me to do so. In regard to the practicability of preventing a change in length, I can only say that the best way of avoiding it would be not to finish the gauge for at least six months after it had been hardened, because I find that in cases of large-diameter cylindrical gauges, in hardening the gauge in water the end naturally becomes the hardest, being dipped usually end first, and this would tend to establish unequal internal strains in the steel, those at the end of the gauge being greater than in the body part of it. This strain would have to be resisted by the metal in its internal structure, and it is reasonable to presume that in time the "fatigue" of the metal would result in the end of the gauge becoming practically smaller than the diameter of the body.

We find in our experience that nearly all gauges, in the course of time, become slightly smaller at the end, sometimes as much as two ten-thousandths of an inch-a difference which is quite perceptible in the hands of a tool-maker. We find that in allowing gauges to "season" thoroughly before the final finishing is done

upon them, this tendency to become tapering at the end is practically obviated. The end-measure gauges which we have made and have had in stock for at least three years, and which I have measured during the past two years, show no perceptible change in length as compared with our standard reference line-measure bars, and the latter have been compared at various times by Professor Rogers, during the past five years, and show no change in relation to those in his possession.

I think, therefore, that the most rational method of treating standard gauges during the process of construction, is to allow them to pass through this "transition" stage merely partly finished, and have the changes due to settling of the differences of molecular strains occur before the final finishing is done. I also think much depends upon the degree of hardness of the gauge as to the amount of change which is likely to occur after hurried finishing I have taken pieces of steel which were hardened as thoroughly as fire and water will make them, and have made many experiments to determine if a rise of temperature and sudden cooling would change their length, and found in the case of an inch endmeasure piece thus treated, raising the temperature about 400 degrees F., and suddenly cooling, its length would be reduced about one-thousandth of an inch after the second cooling mentioned. Hence it seems desirable to keep standard gauges as much as possible from the influence of temperatures higher than that at which they were finished, for even the heat of the sun upon gauges left exposed to direct rays, raising the temperature of them to about 120, or even only to 110 degrees F., might have a tendency to shorten them when they return to the conditions of a normal temperature.

This seems to me to answer the question, so far as my own experience goes, and I can say that we find this knowledge to be of practical value in our work in this direction.

I would like to say, before closing, in relation to hardened steel, and referring to the suggestion of Mr. Smith,-that of drilling a hole in the end of a large mass of steel before attempting its hardening, that I think it might be supplemented by drilling the hole entirely through the steel if it be a tap or other article of extra size, and, if a tap, to have the hole-drilled through the shank, as well as through the body part. In hardening taps of diameters of from about two and one-half inches to eight or ten inches, we find it necessary to drill entirely through body and shank, even if