A careful look at traceability in chemical measurement

Abstract

A traceable procedure in chemical measurement is taken to comprise a series of ‘calibrations’ in each of which the output quantity is calculated from several input quantities with known uncertainties. The starting points are the units of the SI. The endpoint is the analytical result. There is evidence that few analytical procedures conform to that prescription. Problems arise because neither (a) the recovery of the analyte nor (b) the matrix-matching between the treated test solution and calibrators can be guaranteed to be perfect. This gives rise to a continuum of analytical procedures in which the proportion of the true uncertainty of the result found by a traceability chain is mostly less than 100%.