Sampling error in the chemical analysis of rocks

Small portions taken from a prepared sample of rock powder will not contain the same proportions of the individual minerals as the bulk sample. Statistical theory indicates that when the number of grains in the portion taken for analysis is less than 106 the deviations may become significant. It is recommended that, for samples analysed in duplicate, the rocks should be crushed to pass a 120‐mesh sieve. For portions smaller than 0·5 g the powder should be crushed more finely. Rock powders for use as reference standards should be crushed to pass a 230‐mesh sieve. G1 and W1 as originally prepared are too coarse to be used as reference standards.