Biasing Unbiased Data: Mine Versus Mill

The examination of relatively well behaved distributions subject to truncation by a cutoff grade in the mine clearly demonstrates that a bias is inevitably introduced when the selection is based on the results of samples which are subject to even the most idealised error distribution. The idealised error used in this paper is assumed to be normally distributed with a mean of zero and is thus unbiased. The bias introduced can be minimised but it cannot be eradicated. A better understanding of the bias can lead to improved procedures for mine sampling, sample preparation, assaying and estimation of selection units which will not only reduce the discrepancy but more significantly will increase the efficiency of mining selection and hence profitability.