A semiobjective method for condensing classifications

A simple, semiobjective method is described to reduce the number of groups in a classification to an arbitrary level without losing contact with the geologic information contained in the evolving groups. The method, operated in a stepwise or cyclic manner, employs some of the commonly used numerical techniques, but avoids strict adherence to them to obtain geologically more meaningful results. The method is illustrated in a facies study of the upper Paleozoic rocks of southeastern Utah.     

Operator error in petrographic point-count analysis: A theoretical approach

Operator error in petrographic point-count analysis introduces bias into the estimates of proportion in a thin section. A correction for this bias, leading to an unbiased estimator of the true proportion in that thin section, is here proposed. Operator error also affects the confidence interval, and in this situation, too, an adjustment is possible. The approach proposed requires that the probabilities associated with operator error, categorized into A-type and B-type errors, are known or assumed. The A-type operator error tends to underestimate the true proportion in a thin section, whereas the B-type operator error tends to overestimate it.     

Counting error in petrographic point-count analysis: A theoretical and experimental study

The binomial model, commonly used to estimate counting error in point-count analysis, misestimates this error when the observation points on a grid are positively or negatively correlated. A model, called the cell model, is proposed as an alternative to the binomial model for use in studies, especially with coarse-grained rocks, in which such correlation is known or thought to exist. In the new model the thin section is conceptually partitioned into a number of cells (six is recommended), and the assumption is made that the proportions in the individual cells are statistically independent and that their variance does not differ from cell to cell. Empirical relations obtained from a suite of 200 thin sections of limestones are in reasonable support of the prediction that large particle size adversely affects counting error estimates based on the binomial model.