Comparison of two subsampling methods for electron microscopic morphometry.

The purpose of the present paper is twofold. Firstly, to compare the accuracy per unit cost achieved by two different methods of subsampling micrographs from sections of a material for stereology when the sections cannot be analysed as a whole at the required magnification. Secondly, to illustrate, by means of real data, the application of some of the methods and formulae proposed in the companion paper (Cruz-Orive & Weibel, 1981) for estimating ratios at the electron microscopic level. The final estimates of a same ratio obtained by either subsampling method (namely systematic (SQ) and systematic area-weighted quadrats (SAWQ)) agreed in the mean and they were about equally precise. The former fact indicates that the new SAWQ method is at least as reliable as the SQ method as far as bias is concerned. The latter result is a consequence of the well-known fact that subsampling is relatively unimportant in two-stage sampling. Yet, SAWQ subsampling enjoys definite practical advantages over other subsampling methods in certain situations.