railroads, but to everyone who travels on them.

A very great amount of work and time has been spent trying to determine their exact nature, and especially their cause; and a large amount of information has been secured in regard to them by competent observers who have every opportunity for so doing.

At the meeting of the American Institute of Mining Engineers in February last, George F. Comstock presented a paper on this subject, and the Rail Committee of the American Railway Engineering Association has recently reported on the matter, an abstract of its findings appearing in the Engineering News-Record of March 27, 1919, p. 610. Mr. Comstock's paper and the report of the Rail Committee contain probably the latest information in regard to transverse fissures.

The difficulty experienced with flaky gun steel has been considerable, at least in America, and a great amount of investigation has been carried on in order to determine the cause and eliminate the trouble. Of course, its prominence has been aggravated by the greatly increased demand for gun forgings during the war. Two papers on the subject were presented at the February meeting of the American Institute of Mining Engineers, which attacked the trouble from different standpoints, and which were thoroughly discussed. In the issue of CHEMICAL & METALLURGICAL ENGINEERING for March 15, 1919, there is a very important article by Dr. Giolitti, the Italian metallurgist. These articles contain probably all of the important information in regard to flaky steel.

It may seem a far cry from the above two subjects to that of fusion welds, but, as the materials in which the two former defects occur are originally steel castings, and as a fusion weld is the same thing except on a smaller scale, it would seem reasonable that the defects in all three cases would be of the same general nature, although likely to be aggravated in the case of welds because of the lesser degree of skill and knowledge possessed by the average operator, and from the fact that but little attention has been given to defects in welds and to the methods of their prevention. The writer presented a paper at the February meeting of the American Institute of Mining Engineers in this connection. Since that time he has made further examination of heat-treated welds, and finds no reason to change his conclusions in the paper.

Furthermore, as bearing on the general subject, attention is called to a paper by J. C. W. Humfrey, on "The Intercrystalline Fracture of Iron and Steel," in

*Proprietor, Rochester Welding Works,

frey's discussion of the paper by Prof. Zay Jeffries on "The Metallography of Tungsten" in the November Bulletin of the American Institute of Mining Engineers.

Short abstracts from all these references are given herewith in order to sustain the position which the writer has taken in regard to these defects, which briefly is that they all result from the same cause, which he believes to be films of oxide, or other foreign matter, which lie along the grain boundaries and which impair the physical qualities of the metal. These films may be so thin as to be ultra-microscopic, and yet may have a serious effect.

In Mr. Comstock's paper, the statement is made that there are two schools whose opposing beliefs are: first, that transverse fissures are the result of fatigue and independent of the quality of the metal; second, that the quality of the metal and the mill practice must have something to do with them. Mr. Comstock associates these fissures with high-phosphorus streaks, using 24 rails having them and 12 rails giving good service as a basis, and also concludes that reheating the rail blooms reduces the tendency toward forming transverse fissures. He does not take the position that the fatigue theory is disproved by his investigations, and attempts to reconcile the two opposing theories by showing that both are probably closely connected with the formation of fissures. He also states in closing that his "paper is presented with the idea of throwing a little light in a new direction on the internal conditions that help the beginning of transverse fissures in rails." In other words, Mr. Comstock tries to find the starting point of the transverse fissures found in service, and suggests strongly that segregation due to phosphorus is a probable contributing cause.

LATEST INFORMATION ON DEFECTIVE RAILS The statements of the Rail Committee are well summarized in the Engineering News-Record of March 27, 1919, in "Progress in Study of Rail Quality":

First Mr. Waring's statement that deep etching as carried out by him on rails found with fissures either in service or during test brings out irregular markings associated with the fissures. He also states that a rail that performed well under test, though from the same heat as some of the defective rails, showed none of the markings. His conclusion is: "Since these rails were all from one heat, of a fairly uniform chemical composition, and rolled at the same time, and since they were all subjected to approximately the same amount of service in the track, it would appear as if these defects found in the rail heads are associated in some manner with the process of manufacture." And further: "An examination of all these deep etchings brings out the fact that these small internal transverse ruptures correspond in their location and appearance with

the nuclei of transverse fissures, and it is possible that they may be the original cause of such fissures."

Second-Mr. Young's closely allied statement is that his findings "are conclusive evidence that flaws and cracks exist in new rails which have been subjected to no strains except those developed in the rolling."

Third-Mr. Cushing's report as to the very much better quality of rails from some mills than others, based on the comparison of elastic limits and deflections, and that annealing removes the brittleness and increases the deflection. Dr. Dudley also makes the same statement with regard to annealing.

Fourth-The percentage of transverse fissures in direct rolled rails is ten times that found in rails from reheated blooms. In the latter there was one fissure per 5466 tons, and in the former one fissure per 531 tons. Fifth-There is a strong belief that small cracks are produced in rail heads during the gagging.

With regard to flaky steel, the paper' by Mr. Clayton and his associates suggests seven possible causes, of which they seem to lean strongly toward No. 5, which is "overheating and non-uniformity of heat of ingot preparatory to forging."

The paper' by Mr. Rawdon concludes as follows:

"On the whole, the microscopic appearance of flaky steel leads to the following conclusions regarding the nature and origin of flakes. They originate in the ingot, in which state they have the appearance of inter

""Flaky and Woody Fractures in Nickel-Steel Gun Forgings," by Charles Y. Clayton, Met.E., Rolla, Mo., Francis B. Foley, Pittsburgh, Pa., and Francis B. Laney, Ph.D., Washington, D. C. "Microstructural Features of Flaky Steel," by Henry S. Rawdon, Washington, D. C.

X

crystalline shrinkage cracks. They occur in the cast metal along with the slag inclusions in the angles between the branches of the tree-like dendritic crystals. They persist throughout the history of the forging into the finished form as discontinuities in the metal, often associated with slag films, which have resulted from the working down of the slag inclusions with which they are associated from the beginning."

The present writer is strongly inclined to believe that the usual cause of flakes is as explained by Mr. Rawdon, but it would seem quite plausible that at times overheating of the ingot, as referred to in the paper by Mr. Clayton, would cause similar defects which would not be removed during subsequent forging. As evidence of this, the statement made by Mr. Humfrey in discussion of Prof. Jeffries' paper, and appearing in his original paper, shows that intercrystalline brittleness can be produced by overheating, and this brittleness again removed by heating in a reducing atmosphere, such as that of hydrogen. Mr. Humfrey's statement referred to is as follows:

We do, however, know of cases in which certain metals, which normally break through the crystals, may be so altered in character as to tend to break around them. Typical examples of this are:

(b) Pure iron which has been annealed at certain temperatures and in certain atmospheres, and slowly cooled through a certain range. In this case there is evidence to indicate that the weakness is due to the formation of an iron-oxide eutectoid.

Mr. Humfrey was unable to find any microscopical evidence of oxide films around the grains even at high power, yet since the heating in a reducing atmosphere caused the rupture to pass through the grains, as in normal steel, he felt that these films must exist.

DR. GIOLITTI'S OPINIONS

Dr. Giolitti's entire article is worthy of the most careful perusal, especially the following statements:

Flakes originate in tender alloy steels as intercrystalline cracks, probably intensified by inclusions and segregations. They can consequently be controlled by careful steel furnace practice, casting, soaking, reheating, forging and heat treating.

Of even greater importance, however, we feel that free iron oxide (scale and rust) be kept at the lowest possible point in the raw materials charged in the furnace. This refers to scale on the surface of the metal (rusty turnings and borings are never used) and free oxide in the puddled iron itself. . . .

Starting from this pure iron, extreme care is used to prevent superficial oxidation during melting by maintaining a reducing flame in the open-hearth furnace at all times. .

What I wish to emphasize is that chromium steel, properly made of pure materials in this manner under reducing conditions and containing no oxide, will never develop flakes. When so made you can pour it into an ingot of any cross-section and forge it as you likeeither fast or slow, with reheating or without, with great or slight reduction (always providing the work is done within the correct temperature limits and with usual skill) and flakes will not appear.

On the other hand, a bath of metal properly made from correct raw materials will develop woody structure excessively if only a few kilograms of scale are thrown into the furnace even a long time before tapping. Microscopic examination of this defective metal seldom reveals an increased amount of inclusions, still we feel that this "oxidized metal" is flaky by virtue of its content of emulsified iron oxide, or some substance produced thereby which has diffused or broken up beyond the detection of a microscope.

J. C. W. Humfrey: "The Intercrystalline Fracture of Iron and Steel," Carnegie Scholarship Memoirs, Iron and Steel Inst. (1912), vol. 4, pp. 80-105.

There are many other statements in Dr. Giolitti's article that are equally positive, and which all point to the fact that oxide is responsible for flaky steel.

The examination of welds made by the writer, and which are summarized in his paper already referred to, was made by bending sections of welds about in. wide

and in. thick, and examining them at different stages under the microscope. The same method applied to flaky steel showed two things: first, that the first evidence of rupture was at visible defects, such as dots of oxide or manganese sulphide; second, that where these defects did not exist, the first evidence of rupture was in the ferrite around the grain boundaries. As illustrating these points, photographs Nos. 1, 2 and 3 are shown. In none of the pieces from which these photographs were taken was any defect visible at 1200 magnifications that had the appearance of a crack or film of oxide.

RUPTURE IN WELDS DUE TO OXIDE FILMS

Coming now to the matter of welds, in his paper the writer concluded: first, that all metal-electrode arc welds rupture at the grain boundaries; second, that gas welds made with low carbon material, so that no pearlite exists in the weld, rupture through the grains by slipping as in normal steel; third, that where welds are made from welding rods having sufficient carbon, or other elements which preserve pearlite in the weld, the rupture is along the grain boundaries. Since this paper was presented, other conclusions have been reached, which apply to welds which received heat treatments. These conclusions are: first, that neither quenching from temperatures ranging between 1500 and 1800 deg. F. nor annealing within the same range will prevent the intergranular rupture of electric welds; second, that annealing pearlitic gas welds changes the path of rupture from intergranular to intragranular along the slip planes. The heat treatments referred to were done in an ordinary small electric furnace, no attempt being made to keep the atmosphere reducing. In many instances, a careful examination was made at 1200 magnifications of the polished and etched surfaces of the test pieces before bending in order to locate if possible any defects in the way of oxide films that might exist, but in all such cases, even where there were no visible defects, the paths of rupture were in accordance with the above conclusions. A few of the photographs showing these ruptures are reproduced in connection with this discussion.

It is well known that Armco iron is practically pure iron, the total impurities being less than 0.1 per cent. It has one peculiarity, as have all other materials of the same composition, which is its brittleness when heated to about 1700 deg. F. The writer made some attempts to determine the reason for this brittleness by heating pieces of it to different temperatures for different lengths of time, and finds the results shown in the photographs. Normal Armco iron when under strain develops slip bands, and the grain boundaries remain intact as in Fig. 14. When heated to 1700 deg. F. for ten minutes and quenched in ice water, the grain boundaries appear to be thickened, although it is difficult to judge this positively, the number of slip bands developed is much less than with normal material, and the giving at the grain boundaries is quite frequent. (See Fig. 15.) When heated to 1700 deg. F. and cooled

slowly to 1600 deg. F. and quenched in ice water, there appear to be films around the grains as shown in Fig. 16. Under strain the ruptures were clearly intergranular, as shown in Fig. 17. Another test that was made was to heat a piece, -in. x 2-in., with an oxy-acetylene torch to about 1700 deg. F. and then to strike the piece while red hot on the end of an anvil. The heated section broke off short when struck, and the piece was caught in a tank of ice water. In a section prepared for the microscope, for a short distance back of the fracture the piece was seamed with grain boundary cracks in which in the unetched section material re

sembling oxide could be easily seen at high power. Etching the section showed that these defects were at the grain boundaries as in Fig. 18.

These tests appear to show quite clearly that it is possible to produce in the normal iron a condition which is abnormal, and it appears that this condition is due to films around the grains. A similar condition is well known to exist in tool steel, and enough overheating does produce what is commonly called burnt steel. The cause of these films is not known, but the writer offers the theory that they are ultra-microscopic in many cases, and possibly of iron sulphide, due to the small

amount of manganese in the iron. The large rupture in Fig. 18 is probably due to the forcing open of the grain boundaries by the strain and the admission of air, which causes oxide to form.

CONCLUSIONS

Considering all the facts above presented, it seems to the writer quite clear that the following conclusions are justified:

First-That nothing in the evidence so far presented is inconsistent with the theory that transverse fissures, flakes and intergranular ruptures in welds are due to the same cause, that is, films, possibly ultra-microscopic, of oxide; also that it is the only theory that appears to fit all the facts.

Second-That these films of oxide may be present in the ingot and usually are.

Third-That these films of oxide may be caused by overheating of the metal at some stage of the process of manufacture.

Fourth-That if they are caused by the presence of oxide of iron in small amounts only, such as probably causes woody steel, they can be removed by proper heat treatments, such as are stated by Dr. Giolitti and Mr. Humfrey; but if they are caused by larger amounts of oxide of iron, which would form considerable quantities of oxide film, then such heat treatment will probably not remove them.

Fifth-That it appears probable that better mill practice will result in the elimination of flakes and also of transverse fissures.

Sixth-The writer knows that such defects in welds as are visible under the microscope can be entirely eliminated by proper welding methods, which involve nothing but thorough fusion and floating all the oxide and dirt, which is clearly visible during the welding operation, to the surface. In electric welds, this procedure is more difficult because the metal is being constantly added from the electrode, and, in addition to this, the oxidizing conditions are exceedingly severe, as the metal passes through the air in the form of vapor and must become seriously oxidized. It is quite possible that ultra-microscopic films in welds may be removed by heating in a reducing atmosphere.

Seventh-The writer would suggest the examination of rails known to contain transverse fissures that have been in service by polishing and etching small sections and bending them under the microscope. He feels that such an examination would throw some additional light on the matter. He would also suggest that the same tests be applied to sound rails, to rails that show little elongation under the usual tests, to gun forgings and to other samples of steel that do not meet specifications, especially those giving low elongation. It is the writer's experience that welds properly made with good materials will, in a test piece in. thick, give from 22 to 30 per cent elongation in the weld; while welds containing oxide films, though invisible, will give an elongation of from 2 to 10 per cent. The same low elongation appears to be characteristic of flaky steel and defective rail material.

Rochester, N. Y.

[EDITOR'S NOTE-In fairness to Mr. Miller, we wish to point out that his conclusions and those set forth in CHEMICAL & METALLURGICAL ENGINEERING, Aug. 1, 1919, p. 145. were arrived at by each writer in an entirely independent manner, and in fact Mr. Miller's manuscript antedates the one published Aug. 1.]

To the Collectors of Internal Revenue and Revenue Agents in Charge:

T. D. 2914, issued today and showing additional matter to be affixed to containers of completely denatured alcohol, is called to your especial attention.

Reports recently received in the Bureau establish that completely denatured alcohol is being used extensively for bathing and rubbing purposes. This is contrary to the law and regulations, and such uses cannot be tolerated, as the completely denatured alcohol is highly injurious to the skin and animal tissue. It is also established that completely denatured alcohol is being sold by irresponsible dealers under circumstances as to assure them that it is being used for beverage purposes. Where it is so used for any length of time blindness inevitably ensues, and the continued use can only result in death.

Collectors should use every means at their disposal to make known to the public the dangers of either external or internal uses of completely denatured alcohol. Wherever collectors or revenue agents in charge hear of a misuse of completely denatured alcohol, a most thorough and careful examination should be made immediately and all the facts fully reported to the Commissioner for the infliction upon the responsible parties of the ultimate penalties provided by law.

In view of the grave and extended abuses of the use of completely denatured alcohol reported, it is deemed necessary to print upon the labels affixed to wholesale and retail packages a further and more specific warning as to its use than is shown on the present required label. In addition to the present matter on the labels there will be required on all new labels hereafter the printing, in large letters in red ink, under the skull and bones symbol, the word poison, and at the bottom of the label there will be printed the following statement: "Completely denatured alcohol is a violent poison. It cannot be applied externally to human or animal tissue without seriously injurious results. It cannot be taken internally without inducing blindness and general physical decay, ultimately resulting in death."

Until the present stocks of labels are exhausted, this additional matter may be affixed to the containers on a separate label pasted above the present required label. We chemists know that we daily handle poisons of various kinds, many far more dangerous than wood alcohol, and there are comparatively few evil results from their handling and use. People do not drink them. The medical profession administers daily to patients many substances that are poisons except in the minute doses given. But in the case of alcohol, the "kicker," in drinks of quondam familiar kinds, means just that one desired fluid, hence attendant danger. We should do all in our power to protect those who use and some likely to abuse this important chemical and thus incidentally avoid lurid appeals with consequent hampering legislation.

College of the City of New York.