The porous support, which is called _a cupel_(fig. 5), should absorb the slag (oxide of lead, etc.) just as a sponge absorbs water, but must be sufficiently fine-grained to be impervious to the molten metal. At first sight it appears difficult to filter, as it were, a fluid slag from a fluid metal; but an ordinary filter-paper damped with oil will allow oils to run through and yet retain the water; but damped with water it will allow water to run through and retain oils. Similarly, fused slags damp and filter through a cupel, but the molten metal not damping it withdraws itself into a button, which is retained. Although, of course, if the cupel is very coarse-grained the metal may sink into the hollows. Copper, antimony, tin, and most other metals, form powdery oxides, which are not of themselves easily fusible, and it is necessary when these are present to add some solvent or flux to render the oxide sufficiently fluid. Fortunately, oxide of lead is sufficient for the purpose; hence, mixed oxides of copper and lead, provided the lead is present in proper proportion, form a fluid slag. In separating copper from silver or gold, advantage is taken of this fact; for, although we cannot cupel an alloy of copper and silver, it is easy to cupel an alloy of copper, silver and lead. If, however, the lead is not present in sufficient quantity, the whole of the copper will not be removed, and the button of silver, still retaining copper, will be found embedded in a coating of black oxide of copper. Copper oxidises less easily than lead does; and, consequently, the alloy which is being cupelled becomes relatively richer in copper as the operation proceeds. It is on this account that the ill-effects of the copper make themselves felt at the close of the operation, and that the oxide of copper is found accumulated around the button of silver. Tin and antimony, on the other hand, are more easily oxidised; and the tendency of their oxides to thicken the slag makes itself felt at the commencement: if the button of alloy once frees itself from the ring or crust of unfused oxide first formed, the cupellation proceeds quietly, and leaves a clean button of silver in the centre. But in either case the cupellation is imperfect, and should be repeated with a larger proportion of lead. An unfused and, consequently, unabsorbed slag tends to retain small buttons of alloy or metal, and thus cause serious loss.