It is emphasized that the use of statistical methods is a consequence of the incomplete isolation of physical systems. The description of a partially specified quantum state by a density operator is shown to be equivalent to a description in terms of an ensemble of systems. Several definitions of the statistical operator are shown to be identical.