SAMPLING AND STATISTICAL EVALUATION OF DIAMOND DEPOSITS

To evaluate a diamond deposit, a sampling programme needs to be design ed to obtain a reliable estimate of the average grade of the deposit, expressed as carats per tonne or cubic metre, and of the average value of the diamonds, traditionally expressed as US dollars per carat. An analysis of the statistical distribution of the stone sizes, occurrenc es and values is essential for calculating the confidence limits on th e estimated averages. Too wide confidence limits will indicate the nee d for more sampling. If variables are lognormally distributed (such as is often the case with stone sizes and stone values), a more efficien t estimator, such as the t-estimator, can be used, instead of the arit hmetic mean. Knowledge of the statistical distribution is also useful when microdiamond counts from early stage exploration drilling are use d to make preliminary estimates of potential grades of commercial-size d diamonds. Stone occurrences (density distribution) are more homogene ous in kimberlites and lamproites than in alluvial or beach deposits. Grade distributions of secondary deposits are very skew, with often a large proportion of barren samples.