This term is used in two entirely different senses: (a) to denote the actual loss of information in the ordinary sense, e.g. by the destruction of records; (b) to denote failure to extract all the information which exists in the available data about a particular matter. In the second case the failure may be due to avoidable causes, such as the use of inefficient statistics; or it may, in the technical sense of the word “information” be due to the fact that no single estimator exists embodying all the “information” which exists in the sample under scrutiny.