In gold ores, the range of value disclosed by ordinary mine sampling is very wide, and the number of 'samples required for a reliable average is correspondingly great. Even where a large number of samples is available the weighted mean value of the samplesalmost invariably exceeds the true gold content. In operating mines this disparity is overcome by arbitrary reduction of high assays until reconciliation is achieved, but it is frequently necessary, especIallyin exploration, to base an estimate or to reach a decision on a relatively small number of samples from underground openings or even on a few drill hole intersections. In these circumstances, a decisive influence may be exercised by the weight allowed to high assays and to low grade intersections, both of which events are, of course, dependent on the metal distribution in the ore. Where sufficient data are available, statistical analysis will measure the risk involved, but as this information is seldom to be had, the final choice becomes largely a matter of personal judgment.With this problem in mind, a study was made some years ago of the distribution of values in four Western Australian gold ores, with the primary purpose of learning from actual ore-bodies of differing types, something of the metal distribution in the smallquantities of ore involved in individual routine mine samples, and of the importance in each case of high assays. Adequate and suitable data were available in .four cases, in three of which high assays were of frequent occurrence due to the presence of free gold. The results give concrete expression to well known differences of metal distribution, and it is proposed, firstly to record the observed distribution; secondly to examine some of the consequences of the differences in the distributions limd thirdly, to comment, in the light of this data, on the vexed question of the treatment of high assays. Standard statistical procedure was employed in the compilation of the data. The assays were classified by value into groups, each group covering a range of value of 1 dwt. per ton. Assays of less than 1.0 dwt. per ton form the first group; assays equal to or greater than 1.0 dwt. and less than 2.0 dwt. comprise the second group; those equal to or greater than 2.0 dwt. and less than 3.0 dwt. appear in the third group; and so on. Where the number of samples is large, it can be assumed that the mid-value of the group is the mean value of all the samples in the group, thus the mean values of the first three groups are 0.5 dwt., 1.5 dwt. and 2.5 dwt. respectively.The number of samples in the group is the frequency of the group, but for convenience of plotting this frequency is converted into percentage of all the samples in the series. Plotting percentage frequency as ordinate against dwt./ton as abscissa results in a frequency curve-a curve showing the frequency...