Sampling of non-gaussian mineralogical distributions

Since the various new techniques of automation and computing have enabled vast numbers of mineralogical observations of relatively small samples to be made cheaply and rapidly, the statistical assessment of these measurements has become of vital importance. Much of the theory and most of the procedures of sampling that are used in the mineral industry are based on Gaussian statistics, but modern measuring techniques show that many mineralogical parameters do not necessarily follow the Gaussian distribution function. Consequently, it has become necessary to examine the application of other, more suitable statistical methods in order to derive the maximum amount of useful information from the examination of mineralogical samples. Some of the errors that can be introduced by the use of incorrect distribution functions are discussed and the value of determining the function that is appropriate to the particular parameter that is being measured is shown. Examples are provided of non-Gaussian distribution functions in mineralogical materials that were encountered during prospecting and during mineral treatment operations: a Canadian uranium ore, a Thai tin ore, and gold ore tailings.