Without addition 2.9 0.3 2.9 With antimony 3.2 0.2 3.2 With copper 4.9 1.7 4.9 Perhaps the antimony has so small an effect because it is eliminated in the earlier part of the process, while the silver is still alloyed with, and protected by, a large proportion of lead; whilst the copper on the other hand makes its fiercest attack towards the close, when the silver is least capable of resisting it. The ill effects of copper are most strongly felt when the quantity of lead present is not sufficient to remove it: the coppery button of silver got under these conditions is very considerably less than the weight of silver originally taken. Although the above is a fair statement of the loss attending average work, it will not do in very important and exact work to place too much reliance on the figures given, or, indeed, on any other set of figures, with the object of correcting the result of an assay. Each man must rely on his own work. It is easy to determine what allowance must be made for the loss in cupellation by cupelling side by side with the assay piece an alloy of similar and _known_ composition. For, if the two pieces are very nearly alike, we may justly conclude that the loss on each will be the same; and if, further, we take the average of three or four such determinations we shall get results accurate within 0.1 per cent. The method of getting such results may be best explained by one or two illustrations. This method of working is termed "assaying by checks." Suppose we have an alloy of silver and lead in unknown proportions and that by cupelling two lots of 10 grams each there is got from I. 0.1226 gram of silver, and from II. 0.1229 gram. We should know from general experience that the actual quantity of silver present was from 2 to 4 milligrams more than this. To determine more exactly what the loss is, the following plan is recommended:--The two silver buttons are wrapped up each in 10 grams of lead, and cupelled side by side with two other lots of 10 grams of the original alloy. If now the two buttons I. and II. weigh 0.1202 and 0.1203, they will have suffered in this second cupellation an average loss of 2.5 milligrams. Suppose the two fresh lots of alloy gave 0.1233 and 0.1235 of silver, the average loss on these would also be 2.5 milligrams. Add this loss to each result, and take the mean; which is in this case 0.1259.