The distributions of erythrocyte values (red cell count, hemoglobin concentration, etc.) of healthy male workers working in a certain factory were studied in order to evaluate the usefulness of the variables as health indicators. In addition to the estimation of erythrocyte values, anamneses concerning chronic bleeding, smoking habit and alcohol consumption were questioned and serum Fe, Cu, TIBC, ferritin, plasma erythropoietin, and serum enzymes (GOT, GPT, γ-GTP) were measured. Two-dimensional frequency distributions with axes for the red cell count and hemoglobin concentration, and frequency distributions of the score calculated from principal component analysis, showed bimodal patterns. Using nonlinear curve fitting methods, the distributions of principal component scores were fitted to a mixture of two different Gaussian distributions. The workers were then divided into two groups corresponding to the Gaussian distribution he belongs. Then the frequencies of the items and mean values of the variates were compared between them. There were no differences in the incidences of diseases that caused iron deficiencies, but the mean serum ferritin level was significantly lower in one group than in the other, in other words the workers belonging to the group with low serum ferritin level had smaller iron stores than the other. The mean serum erythropoietin level and the mean serum copper level were higher in that group than in the other. As a result, this analysis gives a new evaluation of the health status of a man who belongs to the so-called healthy male group.
