ARE ANALYTICAL MEASUREMENTS REALLY IMPROVING

Using data from the first five rounds of the QUASIMEME project, the re lationship between the coefficient of variation (CV) and assigned conc entration has been examined. For a number of determinand/matrix combin ations it is demonstrated that CVs increase at low concentrations. It is concluded therefore, that since achievable targets for performance are in practice concentration dependent, setting single targets in the QUASIMEME project is unrealistic. Judgement as to whether performance has improved between rounds can only be made by comparing results for test samples having similar concentrations. Crown Copyright (C) 1997 Published by Elsevier Science Ltd. All rights reserved.