RI 6867 Investigation Of Correlation Between Assay Values And Unequal Sample Interval Lengths

Relationships that may exist between assays and sample interval lengths when using unequal sample intervals in mine sampling have been investigated as part of the larger research problem of adapting statistical methods to the analysis of weighted average grades. The theoretical relationship between the amount of difference in the weighted and unweighted average grade of ore and the size of the correlation coefficient between assay value and sample interval length was found to be linear. Numerous examples of actual field sample and assay data, which have been obtained by using unequal sample intervals, are documented, and the differences between weighted and unweighted average grades of ore and the correlation coefficients between assays and sample interval lengths have been computed and plotted on graphs. Differences between these actual field data and the expected theoretical linear relationship are examined. It is hypothesized that the use of unequal sample intervals in mine sampling is, in fact, stratified sampling in a statistical context. The sampler establishes stratification by selecting unequal sample intervals on the basis of differences in lithology, changes in rock type, characteristics of mineralization, or differences in metal content. Statistically} methods currently used in mine sampling, involving unequal sample intervals, fail to consider differences in the inherent variances in grade of ore within the various strata established by using unequal sample intervals. Consequently the strata are sampled with different precision. This difference in precision of sampling between strata may account for much of the deviation between the plotted points representing actual field sampling data and the theoretical linear relationship between the amount of the difference between weighted and unweighted average grade of ore and the size of correlation coefficient between assays and sample interval lengths.