The actual quantity of silica required is determined by the factor known as the formation temperature of the silicates. Every silicate has a definite formation temperature—_i.e._, a definite mixture of iron oxide and silica requires a definite temperature in order that complete combination may occur and a chemical compound silicate be formed. Conversely, at any definite temperature, only those silicates having a corresponding formation temperature to this degree of heat can be produced. In consequence, if the oxidation of the sulphides at the tuyere zone produces any particular temperature, that particular silicate whose formation temperature corresponds to this will tend to be formed, and the required quantity of free silica must be present to yield this definite silicate with the whole of the iron oxidised. Only a limited quantity of silica can thus be taken up for any definite rate of oxidation of iron sulphide, and the presence of either more or less silica does not greatly affect the _composition_ of the slag. Thus the concentration (sulphide oxidation) is primarily dependent on the oxygen supply, which determines how much iron shall be burnt, but the success of the operation depends upon the presence of the correct amount of silica to flux off this iron oxide. This proportion is fixed by the temperature attained at the tuyere zone, which restricts the silicate produced to such a composition that its formation temperature coincides with this degree of heat. Hence the general law has been deduced and has been confirmed in practice, that “a pyritic furnace produces a slag corresponding in composition to the silicates whose formation temperature equals that prevailing at the tuyere zone,” accounting for the well-known observation “that the pyritic furnace tends to make its own slag.” If the smelting operation is to proceed satisfactorily, slag approaching this composition will be produced, and assuming the air supply to be adequate for the purpose, the absence of the requisite silica on the charge affects the quantity rather than the character of the slag. The amount of iron sulphide oxidised depends largely upon the presence of silica to combine with the iron oxide produced; so much will be oxidised as the silica can deal with, and in consequence, if the free silica supply is deficient, a smaller quantity of slag is formed, whilst the matte will be larger in amount but of lower grade. An addition of silica to the furnace charge under such circumstances would thus raise the grade of the matte by encouraging the slagging of more iron, and would produce slag of approximately the same composition as before, though in larger quantity.