Closely sieved (a.k.a. calibrated) material, i.e. material obtained as retained between two screens, is commonly sampled. One example is some iron ores, when drill-hole samples are screened at multiple sizes and the size fractions are sampled in an attempt at predicting iron ore products that could be generated. Another, growing example is the heterogeneity testing method in which instead of crushing parts of a lot at different sizes to generate experimental series of samples, a single coarse crushing is used, then size fractions are obtained by screening and experimental series are generated from each one. Questions automatically then arise: How do we control the sampling variance when sampling calibrated material? And how can we use calibrated material in heterogeneity testing? Under deeper examination, the issue amounts to finding a model for the liberation factor of the calibrated material. A method was used to find a reasonable form for that model, to be customised, as in the general case of unsieved, crushed material, using heterogeneity testing.