Hydrologic phenomena are dynamic, so their understanding and prediction are difficult and challenging. Many are cyclic, ranging from diurnal, seasonal, and annual cycles to climatic fluctuations of many hundreds or thousands of years. Predicting the magnitude and recurrence intervals of long-term fluctuations of these phenomena is a primary concern of hydrologists. The best data upon which to base predictions are those that directly measure a hydrologic phenomenon, such as stream discharge, lake level, and precipitation. Programs for collection of these data over a large area on a scientific basis, however, are less than 100 years old-not a very sound statistical base for predicting events that have recurrence intervals of at least that long.