Stereological studies on red corpuscle size produce values different from those obtained using haematocrit- and model-based methods.

The routine methods for estimating erythrocyte size (volume and surface area) are not generally unbiased. Using geometric models introduces a bias determined by the ways in which real red blood corpuscles depart in size and shape from the simplistic ideal. Employing haematocrit and red corpuscle count to estimate mean corpuscular volume relies on an implicit assumption that erythrocytes fill three-dimensional space. However, a biconcave circular disc cannot be space-filling. The bias is positive and its magnitude depends on erythrocyte geometry and packing density within the centrifuge pellet. We have estimated the volumes, surface areas and shape factors of red blood corpuscles in normal adults (three males, three females) using transmission electron microscopy and stereology. We confirm the lack of sexual dimorphism and find that the mean corpuscular volume is 44 fl (between-subject coefficient of variation 5%), mean membrane surface area is 87 microns2 (6%) and the shape factor is 18.4 fl/fl (11%). The surface is 44% greater than that of a sphere of equal volume. The possible reasons for discrepancies with standard haematological values are discussed. On the basis of observed volumes, we predict that red blood corpuscles account for only 51% of the volume of the red column in standard haematocrit pellets.