Abstract
Eight analytes (Ca, Cl, cholesterol, glucose, Mg, K, Na and urate) have been determined in one horse serum masterlot by up to six different procedures: (i) by so-called definitive methods; (ii) by a group of reference laboratories using a variety of analytical methods; (iii) using the results of two independent external quality assessment schemes; (iv) by transferring values from a human serum standard reference material analysed by definitive methods; (v) by similar transfer of values from several batches of horse reference serum previously analysed by definitive methods; and (vi) as in (v) but using a group of reference laboratories. The results have confirmed the validity and usefulness of the transfer process in value assignment and in identifying outliers due to inaccurate methods. Definitive methods are shown to be subject to error, and although they appear to be the most accurate single techniques currently available, they are not definitive in the sense of being without bias and having no known source of inaccuracy.