A much more complicated set of reactions occur when the alkali of a soda-lime glass is introduced either partly or wholly in the form of sulphate of soda (salt-cake). We have already pointed out that the unaided action of heat and of silica is not sufficient to bring about the rapid decomposition of sulphate of soda which is required for successful glass manufacture, and that the intervention of reducing agents is required. For this purpose a certain amount of carbon in the form of coke, charcoal or anthracite coal, is introduced into all salt-cake mixtures, but the reducing gases of the furnace atmosphere also play an important part in the reactions that take place. Here again it is not possible to give anything but an incomplete account of what takes place. The rationale of the whole process lies, no doubt, in the fact that sulphite of soda (Na_{2}SO_{3}) is much more readily decomposed by the action of hot silica than the sulphate (Na_{2}SO_{4}) itself, so that the essential action of the reducing agents consists in robbing the sulphate of part of its oxygen, thus reducing it to the condition of sulphite and rendering it accessible to the attack of silicic acid. But if we attempt to express such a reaction in the usual manner by a chemical equation from which the quantity of carbon required to effect the reduction in question can be calculated, we find that the amount of carbon required in practice is very considerably less than that given by this theory; it follows therefore that either this very large amount of reducing action must be ascribed to the furnace gases, or that the actual reactions are not strictly of the kind we have described. Both explanations are probably partly correct, and in practice the amount of carbon to be used in a given mixture and furnace can only be found by actual trial, in which the manufacturer is, of course, guided by the results obtained with other furnaces of a similar type. The end-product of the reactions is again a mixture of silicates, but a certain amount of undecomposed sulphate is always found in such glasses, while gaseous oxides of sulphur escape from these furnaces in considerable quantity. Under exceptional circumstances the glass may even contain sulphides of soda or of lime, and sometimes even suspended carbon, but these are abnormal constituents and result in the serious discolouration of the glass. It is obvious that to a mixture containing carbon as a reducing agent such oxidising materials as nitrates cannot be added, but small quantities of arsenic and of manganese dioxide are added because their other properties are sufficiently valuable to outweigh their disadvantages as oxidising agents.