The ratio of standard deviation to mean is the coefficient of variation, the statistic C. This statistic is an excellent guide to the number of observations (ore samples) that should be taken in order to sample an ore deposit, or other geological entity whose variability is unknown. Moreover, the value of C indicates whether or not analysis of the observations by other than statistical normal theory is likely to be advantageous, and, for repeated sampling of a deposit, whether quality control has been maintained.The necessary number of observations is related to the value of C for the substance of interest in an ore deposit. If C is high, as in many deposits of precious metals or trace elements, more observations are required to estimate the average grade with a specified precision than if the variability is low, as in deposits of base and ferrous metals. A plot of means and C values, calculated from 50,057 observations for amounts of various substances in 484 ore deposits, establishes an empirical relation between the mean and C and indicates probable values of C for various substances. For substances present in trace amounts, C is usually high, but C must approach 0 as the mean approaches 100 percent.The appropriate statistical analysis also is related to C because, for observations to be normally distributed, C must be less than about 0.5. If C is larger, the observations cannot be normally distributed, although, they may be lognormally distributed, making lognormal analysis appropriate. However, unless C exceeds 1.2, little can be gained by lognormal instead of normal analysis, because efficiency of estimation increases only slightly and bias may be severe.The statistical distribution of the coefficient of variation is also discussed, sample problems are worked, and tables to calculate confidence intervals for the coefficient are given.