Contribution to the quantitative analysis of neuronal parameters: The effects of biased sampling procedures on estimates of neuronal volume, surface area and packing density

Photomicrographs of motoneuron perikarya from the anterior horn of the rat cervical enlargement were analysed by various histometric and stereological procedures to compare the volumetric, volume‐to‐surface ratio, and packing density estimates obtained by several distinct sampling regimes. Information derived from unbiased motoneuron samples was compared with quantitative data derived from samples biased towards the nucleolus and from others containing cells above a minimal size (dimension‐biased samples). Nucleolarbiasing and dimension‐biasing techniques were found to produce only gross estimates of neuronal parameters. They produced substantial overestimates of cellular volume and perikaryal volume‐to‐surface ratio, together with significant underestimates of neuronal population density, perikaryal volume proportion and cytoplasmic volume/nuclear volume ratio. Unbiased sampling increased the amount of available material which could be analysed. It is concluded that meaningful estimates of these neuronal parameters are best obtained by the unbiased stereological approach, thereby giving a more accurate picture of the structural composition of nervous tissue.