complete reduction of all the iron is impracticable, which is the cause of so great a proportion of the metal being carried off as slag in low structures such as those last mentioned. It will hereafter be seen how admirably a well-appointed blast furnace discharges the threefold duty of intercepting heat, expelling the oxygen from the ore, and raising the product to the required temperature; while in the Catalan furnace these operations, from the nature of the apparatus, are all of them most inefficiently performed.

So much for the imperfections of the direct process—a word now for its advantages. In most iron ores phosphorus is found existing, either as a phosphate of iron or phosphate of lime. In the blast furnace the deoxidizing power is so intense, that both these compounds lose their oxygen, and are converted into phosphide of iron or phosphide of calcium, as the case may be. If the latter, phosphide of iron is still the final product, since the calcium is displaced by the molten iron in the lower zone of the furnace. The phosphide of iron so generated is dissolved by the metal, and, if in excessive quantity, gives rise to great inconvenience in the forge: and when exceeding 1 part in 1,000, it utterly unfits the pig for steel-making.

On the other hand we have the ancient, and to a great extent obsolete, method of direct procedure, which during the last 30 years has engaged more attention than its actual merits, as an economical process of making iron, perhaps deserve. To its apparent simplicity has to be added the desideratum of giving a product comparatively free from phosphorus; for, owing to the partially oxidizing tendency of the operation, the phosphates are chiefly left in the slag. Now that steel is the object which occupies the special attention of those engaged in the manufacture of iron, this direct process is deserving of some further examination, than that which would be its due apart, from this consideration.

There are two plans for accomplishing the reduction of the ores of iron, without the operation of smelting as it is commonly understood. In the first the mineral is heated in contact with charcoal in closed vessels; in the other a mixture of ore and carbonaceous matter is heated on the floor of a furnace of the reverberatory principle, either stationary or revolving.

The actual quantity of carbon required for the reduction of the particular compound of iron and oxygen usually found in our ores as they reach the smelter-viz.: the peroxide-is in reality inconsiderable. Heat, applied to a mixture of the fuel and ore, converts the carbon into carbonic oxide, or into a mixture of carbonic oxide and carbonic acid; the oxygen of course being derived from the ore. Supposing charcoal to be pure carbon, which however is not the case, and such carbon to escape entirely as carbonic oxide, then 6:428 cwts. of this fuel would suffice to reduce 20 cwts. of iron to its metallic state, from a condition of peroxide. It will, however, hereafter be shewn that carbon can carry away from iron more oxygen than that required to form carbonic oxide; and we may therefore assume that 6 cwts. of common charcoal, or thereabouts, would suffice for the mere reduction of the ore.

The moderate temperature required for the operation, and the small weight of carbon necessary for the chemical part of the process, doubtless offer great inducements to pursue the enquiry, both from the point of view of cost and of simplicity. How far expectations in either direction have been verified by actual experience, we will now proceed to consider.

I will pass over the attempts of Mr. Clay, who laboured long and assiduously in the cause, and who, after very many extensive trials made at the Walker Works in the year 1846, failed entirely to convince me, or himself, that it was not cheaper for the forge owner to buy pig iron at 60s. per ton than to treat Lancashire ore in a reverberatory furnace by the direct process, although that ore cost less than onethird of this money.

In 1855 M. Chenot received the highest honour the Jurors of the Paris Exhibition of that year could bestow, for his direct process, declared by a high French authority to be the "greatest metallurgical discovery of the age." In spite of this recommendation there is, I believe, only one work in the world where it is in operation; or at least was so in the year 1872, when I visited the locality in Spain. Even in this case, its continuance seemed dependent upon the possession of a quantity of charcoal screenings, which would otherwise have been wasted. By the courtesy of the proprietors, Messrs. Y. Barra & Co., I was permitted to examine the process, as it was being carried on at that time.

C

The ore, in small pieces previously calcined, is placed along with charcoal in an upright retort of brick, about 33 feet high, its horizontal dimensions being about 4 ft. 9 in. by 1 ft. 4 in. Heat from a coal fire is applied externally, causing a wasteful expenditure of fuel. In three or four days reduction of the ore is complete, and the charge is then withdrawn and cooled in a vessel closed so as to exclude the air, which would otherwise reoxidize the porous metal, or "sponge" as it is termed. This immunity from reoxidation cannot, of course, be secured when the product has to be reheated as a preliminary to being drawn into bars. The waste of metal, by the fire acting on so large a surface of iron, is very great: 30 to 40 per cent. being the usual loss from this cause.

Fifty years ago, when five tons and more of coal were frequently used to make a ton of pig iron, the Chenot system, when dealing with rich and cheap ores, might have had some pretensions to hold a place, in competition with the blast furnace and the puddling process. Against the present more perfect mode of smelting iron, where the ton of pig metal is obtained with 40 cwts. of coal, the struggle, in my opinion, would be hopeless; and still less has this direct plan a chance of holding its own against the Bessemer mode of treatment, in which the waste gases from the blast furnace, with proper machinery, suffice to expel, in the converter, the impurities absorbed during the smelting of the ore.

In support of this opinion, I have taken the cost of making rolled Bessemer steel as unity, and against this have estimated approximately, from data in my possession, the cost of bar iron produced by the

Notwithstanding the overwhelming disadvantages with which the direct process is thus burthened, even after some years of experience, the idea of avoiding the circuitous treatment by the blast furnace, and the subsequent operations, has not been suffered to be forgotten.

Mr. T. S. Blair, of the Glenwood works, near Pittsburg, U.S.A., considered that the large waste of combustible, in heating the exterior of Chenot's retorts, was due to the difficulty the heat had in penetrating their contents. Accordingly, he sought to remedy the evil by applying

fire to an annular space, which contained the mixture of ore and charcoal. By means of a central tube in the retort, the materials, having now a thickness of only 4 inches, were heated much more readily than when they occupied a space of 15 inches. In some cases the apparatus had a diameter of 4 feet, and a height of about 45 feet. Heat was applied only to the upper 8 or 10 feet; the remainder of the retort being used for allowing the product to cool before coming in contact with the air.

The weight of fuel consumed, per ton of sponge made, was stated to be about 8 cwts. of charcoal and 27 cwts. of coal-in all, 35 cwts. This is not so much as is consumed for making Bessemer pig metal; but, as nearly one-fourth of the quantity is charcoal, the cost will, generally speaking, be something in excess of that in the blast furnace, where only mineral coal is employed.

As much as two tons of sponge were obtained per day in such an apparatus, and at a very low cost for labour, according to the figures I received; but nevertheless, I am mistaken if it was not higher than that paid in England at the smelting furnace and Bessemer converter put together.

The inducement to revive this modification of the Chenot process was to use the product for making steel in the method known as the Siemens-Martin, which, as is well understood, consists in melting malleable and pig iron together. The sponge was thrown into the melted pig metal, when, according to Mr. Blair, the supernatant slag protected the porous iron from the oxidizing action of the fire. It was my misfortune not to agree with Mr. Blair, either as to the extent to which this protection was accomplished, or in his estimate of the advantages possessed by his direct process as a partial substitute for the blast furnace.

My reasons have been given above for regarding the sponge, in the matter of fuel and labour, as decidedly more expensive than an equal weight of pig iron. Now, in the so-called "Ore process" for obtaining steel to be hereafter spoken of-to 100 parts of melted pig iron, as much ore as contains 10 to 15 parts of iron is added; and the result is that the carbon and silicon in the pig reduce (by a much cheaper "direct process" than any hitherto practised) as much iron, as gives the maker about the same quantity of steel that he uses of pig metal.

The actual loss of metallic iron in the operation just described is very small-not exceeding 6 per cent. of the entire quantity delivered to the melter in the form of ore; and on this no expense for smelting charges, &c., has been incurred.

The ore in use for the Blair process, on the occasion of my visit at Glenwood, was from the Iron Mountain of Missouri, and was said to contain 66 per cent. of iron. If this was correct, the sponge should have yielded 90 per cent. of iron-supposing all the metal it contained to be reduced, as was supposed to be the case. But the average composition, as given to me, consisted of the following ingredients, in the proportions annexed:

1 Shewing a loss of 15.83 per cent. of Iron, or 143 parts.

When considerations of a commercial nature are applied to these figures, the result holds out little hope of the process occupying any important position as a means of making either iron or steel. It has already been shewn that the labour is higher, and the fuel quite as high, in producing a ton of sponge as in smelting a ton of pig iron. As there is practically no loss of metal in the blast furnace, it may safely be admitted that the cost of sponge will be at least as great as that of pig metal. We have thus, in the ore process, say 1,000 tons of pig iron added to 250 tons of ore, and producing 1,000 tons of steel; and in the Blair process, 1,000 tons of material costing fully more per ton than the pig, but only giving 760 tons of steel. Few manufacturers would, under such circumstances, hesitate in assigning a difference of 15/- per ton of steel in favour of the use of pig and ore

It is only right to mention that Mr. Morrison Foster, the partner of Mr. Blair, does not agree with me in the opinion I formed of the direct process, as carried on at Glenwood. In a letter addressed to the American manufacturer, owing to some differences in the figures we made use of, he makes the loss in actual metal 14.62 per cent. His numbers are no doubt entitled to be preferred to mine, but the discrepancy is not such as to alter the views expressed, which I regret are not those of Mr. Foster.