Letters to the Editor Cutting gold assays

Regarding the letter from Dennis Fairbairn: It’s senseless to cut gold assays, March 28.

It is with interest that I have read the shareholder’s point of view on the classical problem of “cutting” gold assays. Generally, I agree with Fairbairn that these should not be cut, but I would like to nuance this issue with some technical insights.

Being a geostatistician, my task, among other things, consists of deciding the numerical treatment to be applied to the assay data.

A major concern of analyzing assay data is determining the average grade of the material that has been sampled. This is often difficult with many problems arising. One of these is related to the fact that the data-points are usually not uniformly distributed in space, therefore some assays represent the sampling of larger or smaller areas of the deposit and should be weighted accordingly.

Another major problem not to be confused with this is that often very high grade values represent mineralization events with a small extension in space compared to medium or low grade values. This is almost always the case with high grade veins or fractures.

In addition, significant problems related to statistics of skewed distributions and grade interpolation, and to sampling and assaying, also contribute to the difficulty in calculating a meaningful average grade from the available assay values. Cutting off high assay values is, in fact, a wrong reaction to a very real problem.

In any case, a mere arithmetic average or a single projection of all the assay values as they stand, is unsuitable because it ignores all the problems mentioned above and thus almost always gives a non- conservative biased estimate of the average grade, especially with small data sets; on the other hand, cutting the high values off, or even scaling them down to a constant, is likely to bias the result the other way, by an equally unknown quantity, unless the “cutting” or “scaling” has been derived from production data from the same orebody.

The apparently obvious task of calculating the average grade is difficult to perform properly. The good news is that common sense and basic statistical and geostatistical expertise, allied to a reasonable understanding of the geology of the orebody are usually all it takes to come up with an appropriate calculation of the average grade of the ore from a set of assay data. The bad news is that many companies and professionals do not have that basic expertise and therefore either resort to arbitrary methods, like the cutting off of high assays, or choose to ignore the problems that plague a proper grade determination.

From the shareholder’s point of view, I think this clearly demonstrates that it makes sense to give weight to the technical expertise available to the companies in which they contemplate investing their money. This should be done keeping in mind that the most appropriate geostatistical methods have been in widespread use in the industry for not much more than a decade or so, and that a lack of state-of-the-art expertise is sometimes disguised into some “long practical (outdated) experience in the business.” D. Francoise-Bonogarcon, Ph.D Head of Geomathematical Services City Resources (Canada) Ltd.