Some difficult problems still existing in the preparation and certification of CRMs.

Differences between particle size measurements of CRMs by various methods are discussed and the importance of the reliability of such data for proper estimation of the homogeneity of the material is emphasized. On the basis of a very simple model, the dependence of the Ingamells' sampling constant on the average mass of a single particle of the material is derived, and theoretical predictions are compared with the experimental results. Various approaches to the certification of the candidate RMs are briefly reviewed. The merits of the approach being used in this laboratory to evaluate data obtained in the interlaboratory comparison, and to assign certified and information values, is discussed. The conclusions are supported by results obtained for selected trace elements by use of "definitive" (primary) and "very accurate" methods. Some observations on the unusual resistance of some biological materials to wet ashing and the resulting possibility of making analytical errors are mentioned.