It is assumed that random sampling by means of blocks of constant weight yields lognormally distributed copper, lead, and zinc concentration values in the Canadian Appalachian Region. The cumulative frequencies of the highest values belonging to these distributions are determined by approximation from the average grade values and ore tonnages of all types of sulfide deposits in the region. When the average value of the lognormal distribution for a metal is set equal to this metal's clarke, it is possible to construct a separate lognormal distribution for any observed cumulative frequency value. These different distributions are shown to have approximately the same shape for each metal when the metal grade exceeds 1 percent for Cu, 2 percent for Pb, and 5 percent for Zn, respectively. Application of these lognormal distributions to lower grade material yields estimated frequencies which are significantly larger than the corresponding observed frequencies. It is demonstrated that the fitted lognormal distributions are not sensitive to rather large changes in the assumed weight of the segment of the earth's crust used for control and in the assumed clarke values. This approach may be useful in estimating subeconomic resources.