[en] A simple procedure is presented for the statistical analysis of measurement data where the primary concern is the determination of the value corresponding to a specified average exceedance probability. The analysis employs the normal and log-normal frequency distributions together with a χ²-test and an error analysis. The error analysis introduces the concept of a counting error criterion, or ζ-test, to test whether the data are sufficient to make the Z²-test reliable. The procedure is applied to the analysis of annual rainfall data recorded at stations in the tropical Top End of Australia where the Ranger uranium deposit is situated. 9 refs., 12 tabs., 9 figs