Letters to the Editor 853

et al., 1979), base their findings on ana-lytical coefficients of variation of 2 % for calcium and of 3 % for albumin, which they found at normal serum concentra-tions. These give a coefficient of variation for their 'adjusted ' total calcium con-centrations of 2.4 % (Sanderson, personal communication). Now most of their data refer to patients whose calcium and albumin levels are far removed from normality, and yet no figures are given for the analytical coefficients of variation at these abnormal levels. One can, therefore, only assume that they have extrapolated their findings obtained in the normal to the abnormal situation without modification-a somewhat dan-gerous practice I would have thought! However, I feel that this uncertainty at least entitles me to say that the coefficient of variation for their 'adjusted ' total calcium determinations in the patients they discuss is 2.4 % +. Now Mitchell (1978) states that 'for diagnostic purposes, calcium requires to be measured in serum with a high degree of precision and accuracy- ( 1 % is generally considered to be desirable) In my opinion, there is a marked difference between 1 % and 2.4 % +, which leads me to ask just who is right