The amount of numerical data in geology has increased markedly during the past several decades. The numbers arise from field and laboratory measurements on rocks and fossils, as well as from controlled laboratory experimentation on geological processes. Much of the numerical data represents sample observations on larger classes of objects or events about which inferences and predictions are made from the observational data. These larger classes constitute geological populations. In this framework of population and sample, modern probability and statistical methods provide the essential means for expressing generalizations and predictions at specified levels of reliability.