Speaking roughly the retained silver will vary with the weight of gold present; if one alloy contains twice as much gold as another the retained silver will be about twice as much also. On the other hand, as already explained, the cupellation loss on the poorer alloy is as much as, or even more than, with the richer one, because of the copper, &c. present. With rich gold alloys the silver more than compensates for the loss and the surcharge is positive; but with poorer alloys the loss is greater and the surcharge is negative. In Mints and places where bullion assays must be made with the highest attainable accuracy, the surcharge is determined by experiment, and the proper correction is made in the reports on the bullion. This is done by making assays of gold of the highest degree of purity alongside of those of the bullion whose quality has to be determined. These "checks" are so made that they do not differ from the actual assays in any material point. Thus, being of the same quality and weight and undergoing exactly the same treatment, they may reasonably be expected to have the same surcharge as the assays they imitate. Suppose the bullion being assayed varies only a little, up or down, from 900 gold and 100 copper in the thousand, and that .5 gram of it is used in each assay. A quantity of gold differing only a little from .450 gram would be very exactly weighed and placed with .050 gram of copper in the same weight of lead as is being used in the other assays. It would be cupelled, parted, &c., as nearly as possible under the same conditions as the actual assays. Suppose the pure gold weighed .45016 gram and the parted gold weighed .45025 gram, the gain in weight, .00009 gram, would be deducted from the actual assays. A surcharge correction is never applied except to bullion of the same quality as that represented by the "check assay" it was calculated from. It is evident that unless the gold is of the highest degree of purity these check assays will introduce an error almost equal to that which it is designed to remedy. Moreover, to work the checks to the greatest advantage, a very systematic and uniform method of working must be adopted.