AbstractWhen making repeated measurements of a quantity it is often useful to be able to obtain a quick estimate of the standard error of the observations to ascertain whether they conform to a specified standard of accuracy. Calculation of the standard error by either Bessel's or Peters' method is a tedious process under field conditions and may consume valuable observing time. An alternative procedure, well known to statisticians, is to use the range of a sample,i.e.the difference between the greatest and least values of a set of observations, to provide an estimate of the standard error. On the assumption that the errors are normally distributed, it is possible to derive a theoretical distribution of the ranges of samples of various sizes in terms of their standard deviations. The mathematical derivation can be found in standard statistical texts,e.g.M. G. Kendall,“The Advanced Theory of Statistics”, Vol. I. The quantitydn=E(w/σ), that is, the expected rangewin terms of the standard deviationσ, for samples of sizenis tabulated forn= 2(1) 1000 in“Tables for Statisticians and Biometricians”, Part II, ed. K. Pearson. For our purposes we require, as multipliers of the range, the quantities 1/dnand 1/dn√nto give the standard error of a single observation and of the mean respectively, and these are tabulated below forn= 2(1)20(5)40, together with 0·6745/dnand 0·6745/dn√nto give the corresponding probable errors.