should be the amount of the ore taken. The following table will give a
general idea as to this:--

Percentage of the substance Amount of ore, &c. to
to be determined. be weighed.
100-10 1 gram.
10-5 2 grams.
5-1 5 "
1-0.1 10 "
0.1-0.01 20 "

[Illustration: ASSAY NOTE]

The rougher the method of assay adopted, the larger should be the
quantity of ore taken. If the degree of accuracy attainable with the
methods and instruments at the assayer's service is known, it is easy to
calculate what quantity should be taken for any particular case. If the
results are good within 0.001 gram, then, taking 1 gram of ore we can
report within 0.1 per cent., or if they are good within 0.0002 gram,
taking 20 grams of ore, we can report within 1 part per 100,000, or very
closely within 6-1/2 dwt. to the ton. If it is wished to be yet more
particular in reporting, larger quantities must be taken. The difficulty
of manipulating very small or very large precipitates, &c., must be
borne in mind. So, too, must the fact that the greater the weight of the
final product of an assay, the less, as a rule, is the percentage error.
The distinction between absolute and percentage error, often overlooked,
is important. If 0.5 gram of silver be cupelled with 20 grams of lead,
there may be obtained a button of 0.495 gram; the absolute loss is 0.005
gram, and this equals 1 per cent. of the silver present. Similarly,
cupelling 0.1 gram, the resulting button may be 0.098; the absolute loss