Sieve Size Statistics from Grain Measurement

The size statistics of indurated sediments can be best estimated from the study of grain size in thin section. The present work attempts to relate thin section to sieve size statistics, and extends earlier empirical work for which the sieve size distribution was obtained by the disaggregation of sandstones. In the present work loose sands of known sieve size distribution were impregnated to give artificial sandstones that were sectioned for size analysis. A total of 72 sand samples were created, impregnated and measured in thin section. The mean size, sorting and skewness of the sand were varied in a controlled manner, grains of different shape were used, and two different types of grain orientation were studied. The work was extended to include 11 gravel samples measured in cut slabs and 8 grain mounts of the sands. The empirical results support theoretical argument that measures of mean size for similar samples are related by simple additive constants, if all measurements are made in phi units. However corrections are necessary to allow for different grain shapes and grain orientations, and these have been derived empirically. For randomly oriented grains of average shape (B/A axial ratio = 0.7) the sieve mean size may be estimated to within 0.3 phi at the 95% level by the addition of 0.34 phi to the mean grain long axis measured in thin section. The correlation coefficient relating sieve to thin section measures decreases from the mean (0.99) to standard deviation (0.96), skewness (0.54) and kurtosis (0.25), and indicates that skewness and kurtosis cannot be reliably estimated from thin section measurement. Sieve standard deviation may be estimated to within 0.24 phi at the 95% level by multiplying the thin section statistic by 0.81. Analytic argument indicates that sieve estimates can be calculated from thin section statistics by an appropriate equation, but that thin section size distributions cannot be converted in a simple way to equivalent sieve distributions.