The greater the concentration effected by the smelting operation, the higher is the grade of the matte produced; at the same time, the actual weight of matte is smaller. On the other hand, since more iron is oxidised from the charge and slagged off, the quantity of slag produced increases proportionately. Contrasting then, the likely losses of copper which would result from the association of a small quantity of high-grade matte with much slag, compared with those resulting from the association of a considerable quantity of low-grade matte in the presence of but little slag, the former condition is obviously the more productive of heavy loss, for not only will many more shots of matte be held in suspension, but each shot of high-grade matte represents a larger quantity of copper. It is found in practice that it is most economical to make a fairly low-grade matte on the first or “green-ore” smelting, and to re-concentrate this matte pyritically up to converter grade by a second smelting operation. The extra cost of casting the low-grade matte, of breaking up, rehandling it, and resmelting, with all the extra charges on capital, etc., involved, is less than the losses which would be incurred if higher-grade converter matte were made at the first smelting, although there is no difficulty at all in producing such mattes so far as the actual furnace operations are concerned. It is entirely a question of the slag losses involved. Under ordinary smelting conditions (not truly pyritic), when using some coke for fuel, it would be readily possible to alter the density of the slag by adding suitable constituents, such as limestone or additional silica, but in pyritic smelting this is not practicable. The furnace chooses to make at the tuyere zone its own slag, and that a highly basic one. High concentration and a slag low in iron content cannot be obtained together in true pyritic smelting, since high concentration means rapid oxidation of iron sulphide, and this necessitates high temperature and produces a highly ferruginous slag in consequence. Additional silica added to the charge could not alter the slag composition markedly and still yield the same grade of matte. The silica content of the slag depends on the temperature at the tuyere zone, and this is governed by the rate of oxidation of the iron sulphide. If the slag is to be more siliceous it must be produced at a lower temperature, which would be obtained by oxidising the iron less rapidly. This would lead to the production of low-grade matte, and probably would so reduce the furnace activity that there would not be sufficient heat to keep the slag molten.