whereas the final metal contained only 296 of silicon and carbon, showing a gain of metal of 5.5-296-5-204 per cent., or, including the 5 lbs. of increased weight, a total gain of 5.7 per cent. of metallic iron. Supported by these observations, I venture to assert that the removal of the Silicon and Carbon from the pig iron in the ordinary puddling or "boiling" process is due entirely to the action of the fluid oxide of iron present, and that an equivalent amount of metallic iron is reduced and added to the bath, which gain, however, is generally and unnecessarily lost again in the subsequent stages of the process. The relative quantity of metal thus produced from the fluid cinder admits of being accurately determined. The cinder may be taken to consist of Fe3 O1 (this being the fusible combination of peroxide and protoxide), together with more or less tribasic silicate (3 FeO, SiO3), which may be regarded as a neutral admixture, not affecting the argument, and silicic acid or silica is represented by Si O3, from which it follows that for every four atoms of silicon leaving the metal, nine atoms of metallic iron are set free; and taking the atomic weights of iron =28, and of silicon=22.5, it follows that for every 4 x 22.5=90·0. grains of silicon abstracted from the metal, 9 × 28=252 grains of metallic iron are liberated from the cinder. Carbonic oxide, again, being represented by CO, and the cinder by Fe3 O1, it follows that for every four atoms of carbon removed from the metal three atoms of iron are liberated; and taking into account the atomic weights of carbon=6 and of iron-28, it follows that for every grains of carbon oxidized, 6×4=24 28×3=84 grains of metallic iron are added to the bath. Assuming ordinary forge pig, after being remelted in the puddling-furnace, to contain about 3 per cent. of carbon and 2 per cent. of silicon, it follows from the foregoing that in removing this silicon 252 90 x2=5.6 per cent., and in removing the carbon 84 24×3=10·5 per cent. of metallic iron is added to the bath, making a total increase of 5.6+10.5—5=11·1 per cent., or a charge of 420 lbs. of forge pig metal ought to yield 466 lbs. of wrought metal, whereas from an ordinary puddling-furnace the actual yield would generally amount to only 370 lbs. (or 12 per cent. less than the charge), showing a difference of 96 lbs. between the theoretical and actual yield in each charge. This difference, amounting to fully 20 per cent., is due to the enormous waste by oxidation to which the iron is exposed after it has been "brought to nature" (by the removal of the carbon), when it is in the form of a granular or spongy metallic mass and during the process of forming it into balls. So great a waste of metal by oxidation seems at first sight almost incredible; but considering the extent of surface exposed in the finely divided puddled mass, it is not at all exceptional, and is in fact almost unavoidable in a furnace of the ordinary construction, maintained as a puddling-furnace is at a welding heat. Many attempts have been made (for example, by Chenot, Clay, Renton, and others) to produce iron directly from the purer ores, by reducing the ore in the first instance to a metallic sponge, and balling up this sponge, which is a loose porous mass somewhat similar to spongy puddled iron, on the bed of a furnace; but all these attempts have failed, simply on account of the great waste of iron, a waste amounting to from 25 to 50 per cent. in balling up the sponge. Indeed the loss in an ordinary puddling-furnace would probably be greater than 20 per cent. if the metal were not partly protected from the flame by the bath of cinder in which it lies; for in one instance in which the cinder accidentally ran out of a puddlingfurnace during the balling up of the charge, leaving the iron exposed to the flame, I found the yield reduced from the average of 413 lbs. down to 370 lbs., showing an increased waste of 43 lbs., or over 10 per cent., due to the more complete exposure of the metal to the oxidizing action of the flame. In order to realize the theoretical result, a sufficient amount of oxides must have been supplied to effect the oxidation of the silicon and carbon of the pig iron, and to form a tribasic silicate of iron (3FeO, SiO3) with the silicic acid produced. The amount of oxide required may be readily ascertained. In taking the expression Fe3 O', the atomic weight of which is of cinder or oxide of iron are requisite to produce the 46 lbs. of reduced iron which were added to the bath. There must, however, remain a sufficient quantity of fluid cinder in the bath to form with the silicon (extracted from the iron) a tribasic silicate of iron, or about 60 lbs., making in all 124 lbs. of fettling, which would have to be added for each charge, a quantity which is generally exceeded in practice, notwithstanding the inferior results universally obtained. There remain for our consideration the sulphur and phosphorus, which being generally contained in English forge pig in the proportion of from 2 to 6 per cent. each, can hardly affect the foregoing quantitative results, although they are of great importance as affecting the quality of the metal produced. It has been suggested by Percy that the separation of these ingredients may be due to liquation. This I understand to mean that the crystals of metallic iron which form throughout the boiling mass when the metal "comes to nature," exclude foreign substances in the same way that the ice formed upon sea-water excludes the salt, and yields sweet water when remelted. According to this view, pig metal of inferior quality will really yield iron almost chemically pure, to which foreign ingredients are again added by mechanical admixture with the surrounding cinder, or semireduced metal. It may be safely inferred that the freedom of the metal from impurities thus taken up will mainly depend upon the temperature, which should be high, in order to ensure the perfect fluidity and complete separation of the cinder. Led by these chemical considerations, and by practical attention to the subject, extending over several years, I am brought to the conclusion that the process of puddling, as practised at present, is extremely wasteful in iron and fuel, immensely laborious, and yielding a metal only imperfectly separated from its impurities. How nearly we shall be able to approach the results indicated by the che mical reasoning here adopted, I am not prepared to say; but that much can be accomplished by the means actually at our doors is proved by the result of the working of a puddling-furnace erected eighteen months since to my designs by the Bolton Steel and Iron Company in Lancashire. This furnace consists of a puddling-chamber of very nearly the ordinary form, which is heated, however, by means of a regenerative gas furnace, a system of which the principle is now sufficiently well established to render a very detailed description here unnecessary. The general arrangement of the furnace is shown in the accompanying illustrations. It consists of two essential parts: The Gas-producer, in which the coal or other fuel is converted into a combustible gas; and The Furnace, with its "regenerators" or chambers for storing the waste heat of the flame, and giving it up to the incoming air or gas. Fig. 1.-Section of Gas-producer. Scale inch to a foot. The Gas-producer is shown in fig. 1; it is a rectangular firebrick chamber, one side of which, в, is inclined at an angle of from 45° to 60°, and is provided with a grate, c, at its foot. The fuel, which may be of any description, such as coal, coke, lignite, peat, or even sawdust, is filled in through a hopper, A, at the top of the incline, and falls in a thick bed upon the grate. Air is admitted at the grate, and, in burning, its oxygen unites with the carbon of the fuel, forming carbonic-acid gas, which rises slowly through the ignited mass, taking up an additional equivalent of carbon, and thus forming carbonic oxide. The heat thus produced distils off carburetted hydrogen and other gases and vapours from the fuel as it descends gradually towards the grate, and the carbonic oxide already named, diluted by the inert nitrogen of the air, and by any small quantity of unreduced carbonie acid, and mixed with these gases and vapours distilled from the raw fuel, is finally led off by the gas-flue to the furnace. The ashes and clinkers that accumulate in the grate are removed at intervals of one or two days. E is a pipe for the purpose of supplying a little water to the ash-pit, to be decomposed as it evaporates and comes in contact with the incandescent fuel, thus forming some hydrogen and carbonic oxide, which serve to enrich the gas; & is a small plughole by which the state of the fire may be inspected, and the fuel moved by a bar if necessary; and D is a sliding damper by which the gas-producer may be shut off at any time from the flue. It is necessary to maintain a slight outward pressure through the whole length of the gas-fiue leading to the furnaces, in order to prevent the burning of the gas in the flue through the indraught of air at crevices in the brick work. Where the furnaces stand much higher than the gas-producers, the required pressure is at once obtained; but more frequently the furnaces and gas-producers are placed nearly on the same level, and some special arrangement is necessary to maintain the pressure in the flue. The most simple contrivance for this purpose is the "elevated cooling-tube." The hot gas is carried up by a brick stack, H, to a height of eight or ten feet above the top of the gas-producer, and is led through a horizontal sheet-iron cooling-tube, J (fig. 1), from which it passes down either directly to the furnace, or into an underground brick flue. The gas rising from the producer at a temperature of about 1000° Fahr., is cooled as it passes along the overhead tube, and the descending column is consequently denser and heavier than the ascending column of the same length, and continually overbalances it. The system forms, in fact, a siphon in which the two limbs are of equal length, but the one is filled with a heavier gaseous fluid than the other. In erecting a number of gas-producers and furnaces, I generally prefer to group the producers together, leading the gas from all into one main flue, from which the several furnaces draw their supplies. The Puddling-Furnace proper is shown in figures 2, 3, and 4. Fig. 2 is a front elevation of the furnace, showing the gas-reversing valve and flues in section. Fig. 3 is a longitudinal section at A, B, C, D (fig. 4). Fig. 4 is a sectional plan at L, M (fig. 3). The peculiarity of the regenerative gas furnace, as applied either to puddling or to any other process in which a high heat is required, consists in the utilization in the furnace of nearly the whole of the heat of combustion of the fuel, by heating the entering gas and air by means of the waste heat of the products of combustion after they have left the furnace, and are of no further use for the operation being carried on. The waste heat is, so to speak, intercepted on its passage to the chimney by means of masses of firebrick stacked in an open or loose manner in certain chambers, called regenerator chambers," C, E, E,, c, (fig. 3). 66 On first lighting the furnace the gas passes in through the gas-regulating valve, B (fig. 2), and the gas-reversing valve, B', and is led into the flue, M, and thence into the bottom of the regenerator chamber, c (fig. 3); while the |