At an early stage of the examination of most prospects, samples are taken; and the sample values obtained play an important part in determining the subsequent history of the prospect. It is, therefore, important that. eveJJy effort should be made to interpret asfully as possible the meaning of the observed results. In this connection, it is suggested that examination of the assay values of mine samples by simple statistical methods may bring to light information of considerable value, which might otherwise be missed.Although expert statisticians may derive more information from the available data, those who are, like the writer, less proficient mathematically, may also benefit by applying quite simple statistical methods. At least they will develop a useful habit of thought bywhich they become aware at all times that sampling results only provide estimates of the truth, but that an assessment of the precision of such results may be possible. The object of this paper is to provide some evidence of. the truth of these statements, by theuse of a few very elementary statistical calculations.General.The idea that any mine sample is subject to error from a variety of causes is a very familiar one. Doubt always exists with regard to the representativeness of a perfectly cut sample, and other errors may arise through imperfections of sample cutting, crushing, splitting and assaying. Some causes of error may give rise to "constant" errors, others may cause variable but one-sided "systematic" errors, and others again may bring about variable results without any tendency toward bias on the average. In statistical language, mine sample values are subject to errors which may have either assignable causes or chance causes. The former are those which can be isolated, or should be capable of isolation, while the latter are those remaining when all assignable causes have been removed. A single constant assignable cause brings about bias in all results, and one or more than one variable assignable cause brings about variation between individual results and perhaps an overall bias. The system of chance causes that would remain if all assignable causes were removed would cause variation between individual results but no overall bias. There is no sharp practical line of distinction between the two types of cause) of error; nor is it ever possible by any method of statistical analysis to be certain that there are no assign...