The concept of uncertainty as applied to chemical measurements.

The calculation of uncertainty as recommended for physical measurements cannot be transferred readily to chemical measurements. Physical measurements and chemical measurements have entirely different error patterns that behave differently on replication. Correctable local bias predominates in physical systems and random error is minor; random error predominates in chemical systems and bias is difficult to identify and eradicate. Therefore bias must be monitored by randomizing in the interlaboratory environment, a concept not handled by the conventional ISO treatment of uncertainty.