of corrosion, as time goes on and more and more of the metal is eaten away. The reason for this is that the cementite is in such extremely minute microscopic plates that the eating away of a very small quantity of the iron from above them ought to bring very nearly the full proportion of this cementite to the surface; whereas the much larger and more distantly scattered plates of cinder in wrought iron would not constitute their full share of the surface until a much thicker layer of initially overlying metal had been eaten away.

This, then, may be the true explanation. The reason why steel does not rust faster than wrought iron in our direct tests, though it does in actual use, is that our direct tests are too short to bring out the full protective action of the cinder of the wrought iron. Or the reverse may be true. As time goes on, the harmful effect of the difference of potential of the cinder may grow more than its protective action.

Let us therefore henceforth push our tests to destruction.

Two other points: Sheet steel roofing may rust faster than iron because it does not hold the paint so well, and yet steel in other forms, like tubing, may rust no faster than wrought iron.

Again let me emphasize the difference between different steels. Carelessly made steel containing blowholes may rust faster than wrought iron, yet carefully made steel free from blowholes may rust more slowly. Recognize that any difference between the two may be due, not to the inherent and intrinsic nature of the material, but to defects to which it is subject if carelessly made. Care in manufacture, and special steps to lessen the tendency to rust might well make steel less corrodible than wrought iron, even if steel carelessly made should really prove more corrodible than wrought iron.

Mr. Howe.

Mr. Cushman.

THE CORROSION OF IRON AND STEEL.

GENERAL DISCUSSION.

THE CHAIRMAN (MR. H. M. HOWE).-Gentlemen, steel has the misfortune to be called steel, whatever its composition; and we must not forget that there is less difference between wrought iron and the steel which is used for tubing, such as boiler tubes, etc., there is less difference between them than between that steel and the steel that is used for high carbon knife and spring steels, of which we have just heard. We must discriminate, therefore, between steels. Not all steels are alike.

The effect of zinc, of course, is a very protective thing. The danger, however, of relying on small scale experiments for our belief in respect to zinc, is that we have to fear that that protection would be very local, because experience alone could tell how far we could depend on it; and while it is distinctly effective for a short distance, I think it has not been shown that it would be effective for any long distance. Of course we might place pieces of zinc at very short distances along the beams and joists, but after the zinc had oxidized it would be of no effect. But it seems to me any impermeable coating of asphalt or whatever it might be, a mechanical coating, ought to last forever.

DR. ALLERTON S. CUSHMAN.-Mr. Chairman, I would like to go on record as stating that I believe that the cinder in wrought iron has a good deal to do with the protection of that metal from corrosion. I am going to begin by discussing the effect of electrolysis on the corrosion of iron and steel. When I was an undergraduate student I made what I thought was a discovery; I soon found, however, that it was not. I found that by taking two steel needles from the same package and attaching them to wires which led to a galvanometer and then inserting the needles in a solution of an electrolyte, that the galvanometer showed a deflection, proving that there was a difference of potential established between two steel needles taken from the same package. If the needles were physically and chemically exactly alike, there should be no difference in potential. The trouble is, no two needles are exactly alike. If that is so, how much more must the difference be in large

steel structures that are made up of a number of separate units. I Mr. Cushman. mean by that, a building of steel construction and the consideration of protection from oxidation of the structure that would be established by erecting zinc anodes in the neighborhood of the building. I am perfectly willing to believe that the zinc anodes would be very rapidly oxidized, but it seems to me that very little effect upon the building as a whole would be obtained, because it stands to reason that in a structure of that sort there must be local differences in potential. I believe that differences in potential in what might be called a physically homogeneous piece of steel or iron, like a strand of wire, are continually being established, when it is wetted with an electrolyte, if the wire is not chemically homogeneous in structure. I admit that the differences of potential are extremely minute, but if they are long continued there must be continually formed points of attack where the chemical action takes place and oxidation begins. Thus weak points are formed where oxidation goes on rapidly and leads to pitting.

This may sound speculative, but it can be shown that there are differences of potential established in wires, that are stretched or bent and wetted with electrolytes. Electrolytic corrosion is due simply to local disturbances of equilibrium; and I do not see how we are going to get any protective effects by considering a structure like a building as a cathode. It seems to me to be perfectly impossible.

Now to go on to the subject of the relative corrosion of iron and steel, I confess that I have very little to say. I am not a friend of steel as against wrought iron, nor a friend of wrought iron as against steel. The question that undoubtedly interests every metallurgist, is the underlying reason for the great differences that are observed in service. There can be very little question, I think, that no two kinds of steel will be equally resistant to corrosion whether the test is in the open or in the laboratory. I have in my laboratory two samples of wire made by the same concern. They are both Bessemer steel wires. Yet these two wires will show great differences, in short time tests, in the relative rate of corrosion. It is, however, quite true that it is not necessarily the rate of oxidation but the kind of oxidation which is of the utmost importance. One metal may soon become covered with a protecting coat of oxide, while another may become pitted to the point

Mr. Cushman. of destruction. In my opinion there is no subject that is more important to-day to the practical metallurgist, as well as the man who is interested in it purely as a scientific problem, than the corrosion of iron and steel. And I do not think really that any of us, either in this country or abroad, know a great deal about it yet. I think it is a subject that has got to be systematically investigated and without bias. I do not think it is a subject that ought to be investigated in order to show that wrought iron is superior to steel, nor do I think it ought to be gone into for the purpose of demonstrating that steel is superior to wrought iron. Those are questions that must be determined by the facts in the case as we go along. Why is it that two steels-leaving wrought iron out of the question for a minute-differ so diametrically in the way that they rust and in the rate of their rusting? I have been led to believe that the impurities that exist in all steel, or nearly all steel, have a good deal to do with the question. Manganese is well known to affect the electrical conductivity of steel to a considerable degree. If the manganese is not evenly distributed throughout a long piece of metal it is possible that differences of potential will be established, if the metal is subjected to the action of water. If there are differences of electric potential, naturally currents will flow at the expense of something, and it must be at the expense of the metal which is corroded thereby.

Mr. Christie.

MR. JAMES CHRISTIE.-I shall confine my remarks to the subject of corrosion of sheet metal, and while I have no recorded data of experiments, I can refer to the frequent complaint, so widely expressed, that the sheet metals so extensively used for roofing and sheathing do not now, when steel is used, give as satisfactory endurance as when sheets were rolled from puddled iron. It is quite true, as Professor Howe has stated, that we cannot rely entirely on popular belief which may be based on insufficient or unsatisfactory evidence, nevertheless, when this belief finds widespread expression, and is sustained by numerous observations, we must give credence to popular assertion, or else disprove it. Within my own field of observation I can point to roofs covered with sheet iron which have been in existence for periods of twenty to thirty years, with a promise of further longevity, while under apparently similar conditions, steel sheets have become so corroded as to be unfit for service, in half the period. This is not only

the case with the black or uncoated sheet, but appears to apply also Mr. Christie. to tinned and zinc-coated metal. In fact, it is now generally considered as useless to apply the uncoated metal for roofing purposes, whereas in past times the use of black sheet metal was quite common. Believing these to be facts so far as sheet metal is concerned, yet we are not sure that the same conditions exist when metals of greater thickness are used, and this might appear to be anomalous or inconsistent.

Some experiments have sustained the view that initial corrosion is more apparent with steel than with puddled iron, but after superficial corrosion has occurred, some degree of protection is afforded to the underlying metal, which tends to confer equality between iron and steel.

This, if correct, would account for the anomaly, if it exists, as a superficial corrosion would destroy very thin sheets, which would have comparatively little effect on thicker material.

It would appear probable that the cinder contained in puddled iron would protect the metal from corrosion to some extent, considering that this cinder is largely metallic oxide, and ordinarily occupies from 1 to 2 per cent. of the total bulk.

Another illustration is afforded in the susceptibility of the metal to oxidation at high temperatures. When iron was highly refined, that is, subjected to repeated heating and rolling, and much of the cinder contained in the original puddled bars expelled, the welding property of the metal was impaired, or, as the old-time blacksmith said, it became too dry, and burnt readily at the welding heat. The smith had then to use a protecting flux of sand or borax, and when much welding was required, highly refined metal was objectionable.

The constant loss involved in the corrosion of iron, and the expense incurred in the frequent application of protective coatings, form the principal objections to the use of the metal for many purposes to which it is otherwise well adapted.

It would appear to be a promising and not impossible quest, for the metallurgist, to seek for material which if combined with iron will confer some protection from corrosion without impairing the physical properties of the metal or unduly enhancing the cost of its production.

MR. MAXIMILIAN TOCH.-In reference to the criticism that Mr. Toch.