experience has proved that the alkali mingled with the ore to separate impurities, after a very few operations will partially dissolve the silica bottom. The metal will then mingle with it, and much of it waste away as a silicate of iron. For this reason silica bottoms have been discarded for the ordinary puddling furnace cinder bottoms, and these are found in every way satisfactory. They keep the metal clean, and

are themselves uninjured by use.

In preparing the mixture several points are necessary to be observed, and, although they are subsidiary to the separation of impurities, are still very important, if not essential, to the complete success of the system.

The mixture must be so compounded as, when moulded, to withstand the shock of rough handling without breaking in transportation to the furnace.

It must also be of such consistency as to preserve its original moulded shape, while in the furnace, until thoroughly penetrated by heat. If it settles down into a mass of uniform thickness over the hearth, before it is penetrated by a reducing heat, the cost and yield will render the process unprofitable.

Both of these conditions are attained by a judicious mixture of lime and clay with pulverized ore and carbon, when a preponderance of silica is combined with the ore. Magnesian lime is preferred. With aluminous ores the clay may be largely dispensed with. These substances must be varied in proportion with the analysis of the ore, but they may always be so combined as to make a firm compact mass at low cost. Salt and manganese have been used sometimes in addition to lime and clay to aid in separating phosphorus and other impurities, but lime and clay are the main dependence.

There is one exceedingly important office of this cheap alkali mixture to which is due almost entirely the commercial economy of the process. It must form a "non-flowing slag." Not simply a flux to dissolve impurities, but a combination which fills a range of usefulness very much wider than that of any ordinary flux.

As the moulded masses are spread over the hearth with spaces between them, they present not only larger surface areas for deoxidation, but also present shapes that would be very favorable for re-oxidation by the furnace gases, unless means can be provided to prevent it. While it is essential to drive the heat high from the beginning, in order to bring about quick and economical reduction to metal, it is

equally as essential that the material shall at the same time be covered and protected from the waste of high heat, which is always very oxidizing. This is one of the requirements of this "non-flowing" slag. By entering the pores or spaces or cells of the ore, which have been vacated by the oxygen as it passes out and forms with carbon, carbonic oxide gas, each little atom or filament of metal is immediately sealed and varnished by this glassy coating, so that the delicate particles of new-made iron are in this manner effectually saved from re-oxidation and destruction.

An ordinary flux would flow from the metal on to the bottom of the furnace, carrying, it is true, a portion of the impurities with it, but leaving the new spongy iron to melt into a protoxide, like snow melts before a south wind. Hence the importance of having this slag nonflowing, as well as fluxing, which is so very easily accomplished.

There is still another result accomplished by the proper mixture of these cheap alkali material with ore and carbon. Being highly basic, it dephosphorizes and desulphurizes as well as desiliconizes. This same combination, which may be depended upon to form a firm mass both out of and in the furnace, and that protects the ore from oxidizing influences by furnace gases, may be also effectually relied upon to break the affinity of iron both for phosphorus and sulphur as well as for titanic acid. A large class of rich ores may be thus utilized which are now useless, because they contain so large a percentage of these impurities. These deleterious substances are separated either by volatilization or pass off with the slag as it is expelled by the squeezer and rolls.

Still I am not yet through with extolling the virtues of this "nonflowing slag." Important as its influence is for the purposes here before stated, it also serves to secure a great economy of fuel. Metallurgists and chemists agree that ore should be brought to wrought iron bars with the consumption of half a ton of coal to the ton of iron. Notwithstanding these nice theoretical estimates, the best part of three tons of coal is required to a ton of bars. This is one of those marked cases where theory and practice widely differ, in spite of every effort to correct it. Some very powerful, but perhaps not very well understood, cause must lie at the bottom of this extraordinary waste in converting ores to metal. Allow me to venture the query whether it may not be produced by nitrogen. The chemist will create intense

combustion in a jar of pure oxygen, but let him mingle with it a large volume of nitrogen, and combustion will be arrested.

Until lately nitrogen has been considered rather a neutral substance in iron making, being neither favorable nor injurious to the metal. Now with every 1000 pounds of air thrown into the furnace, only 230pounds is oxygen, while of the remainder more than 700 pounds is nitrogen. The oxygen forms not one-fourth of the total blast, and yet it is only upon the oxygen we place dependence to secure combustion.. A gas which forms nearly three-fourths of the entire volume contained in the furnace, if it is not combustible itself, most certainly must retard the ignition of the other combustible gases.

May it not be possible that these imprisoned gases in the moleculesof metal are the prime cause of deteriorating its quality, so that puddled pig iron will so rarely produce high grades of crucible steel? Dr. Müller, of Brandenburg, has proved by a simple and ingenious method that hydrogen and nitrogen are actually contained in very considerable quantities in iron. He practically determined that in some cases these gases formed a volume of about fifty per cent. of that of the drilled hole from which the test was made, and these experiments have since been verified by others.

In the process under consideration there are no gases forced by the pressure of the blast through the metal as in the blast furnace. As the carbonic oxide is generated by the heated mixture, by its own pressureit finds its way to the surface of the moulds. The door is open for the exit of gases, but is closed and sealed against the inroad of injurious volatile furnace impurities. The blast pressure is not sufficient to counterbalance the outward pressure of carbonic oxide, and hence gases are not introduced into the recesses of the metal to its injury, as pig iron is contaminated.

Whether the foregoing is or is not a correct solution of the problem, it is very certain that iron produced by this system is invariably of a quality that may be relied upon for the finest grades of fine crucible or open-hearth steel.

In all previous direct processes it has been considered essential to keep the furnace in a red smoky atmosphere of carbonic oxide, as the best condition for rapid reduction. This is a condition, it may be remarked, where about one-half of the fuel passes unconsumed out of the stack, and the other half is not allowed to generate more than half the heat it is capable of, if sufficiently supplied with air.

In common with others, I formerly deemed this most wasteful combustion of the fuel to carbonic oxide an essential means for the best reduction. Although I had discovered that commercial economy necessitated the application of a high furnace heat at once, in order to penetrate to the interior and rapidly reduce to metal, yet I was unable to apply this high heat because of its waste by a re-oxidation of the iron. After I found that a non-flowing slag could be made to remain and protect the metal, I was able to push the heat high on introducing the charge; and thus, by more economically burning the fuel to carbonic acid, bring the ore to metal very rapidly in less than two hours. I am aware that the treatment of an ordinary puddle ball to such prolonged exposure to furnace blast would oxidize and waste it largely to cinder, but it must be remembered that this combination is not that of an ordinary puddle ball. It is iron ore mingled with carbon to deoxidize it, and protected at every point by a glazing slag which prevents re-oxidation.

Saturated, as the mixture becomes, with a high heat, as fast as it can be conducted through it, the gases are observed to begin to work immediately. Every square inch of surface of these moulded masses is covered with a flame of carbonic oxide which is quickly transformed to carbonic acid upon meeting the heated furnace gases. Thus combustion is intensified by the heat evolved from the moulds themselves, throughout the entire furnace hearth. The heat given off from one mould impinges upon the mould adjoining, while the furnace walls and roof receive and reflect back the radiated heat. Every little particle of ore gives off pure oxygen-not oxygen diluted with three-quarters nitrogen to retard perfect combustion, as in the blast furnace-but as pure oxygen as it is possible for the chemist to generate in his laboratory. This oxygen immediately finds an atom of carbon close to it with which it assimilates and forms carbonic oxide. Thus existing gaseous impurities are removed, while other gases injurious to the iron are not insidiously introduced with a strong blast pressure as they are in pig iron. The practical result follows, as has been before stated, the iron is better for fine steel than puddled pig iron.

It is no longer necessary to waste fuel, ore and labor by maintaining in the furnace an atmosphere of carbonic oxide, which so needlessly prolongs the operation. The moulded mixture, upon the application of high heat, is itself a flame at all points, like dry wood

would be under similar circumstances. They produce, of themselves, a high heat, and thus economize fuel from the fire-grates.

Indeed, so effectual is the heat generated at the surface of the moulded mixture that I am accustomed to lessen the supply of fuel at the stoking-hole as the gases begin actively to work, and then the blast is reduced while the damper is somewhat lowered. Under this treatment, the ore mixture continues to do its work until the whole is reduced to metal. It asks for no puddling, no exhausting physical labor, but simply to be let alone to do its own work in its own way.

I may have been tedious in detailing the value I have discovered there is in a non-flowing slag, and also in so minutely describing the working of the process, but I have purposely elaborated this part of the subject, at the risk of wearying you, because I believe the true secret of successful commercial economy in working the direct method lies in this direction.

There is nothing new in fluxes, nor in moulding the mixture into shapes for reduction. The few alkali or acid materials cheap enough to be used for fluxes have all long ago been known, and one inventor after another has used them in varied proportions in moulded shapes, and still failed to demonstrate commercial economy in competition with other methods of working. Like my own earlier experiences, the successive steps to bring ore to metal direct have not been well understood. A non-flowing slag to protect from re-oxidation, and a quick, high heat, seems to be the pivot upon which commercial economy turns. By coupling this with that other quite as important discovery -namely, placing the several masses of metal mixture sufficiently apart so that the heat generated from one may impinge upon another— the general outline for practical working seems to be complete. A wide range of experimental tests with ore mixtures, in cases and without them, and in various furnaces, have gradually pointed out the successive steps by which to make the method simple and easy. Now heat after heat may be withdrawn from the furnace in less than two hours with the regularity of puddling, but with very much lessened

cost.

Having succeeded in satisfactorily working ore-mixtures, my attention has latterly been directed to the utilization of puddle, forge and heating furnace cinder as well as roll scale and hammer slag.

Puddle cinder is a rich silicated protoxide of iron, containing from 50 to 58 per cent. metallic iron, 16 to 18 per cent. of silica, and from