The Measurement of Economic Relationships presents a critical review of the established approach to measuring relationships in business and economics and of the one that preceded it. The first approach is based on the notion of a process "generating" the observations in a certain "random" manner, the second on the concept of approximating the observations of the variable of interest as closely as possible. It is shown that both approaches offer measures of the contribution of the determining variables, interval estimates and tests concerning the effects of the variables, and interval forecasts; in general, however, their solutions are different. In reviewing the two approaches since their first appearance at the end of the 18th century, little justification is found for the manner in which the established approach perceives the economic world. Doubts are raised that substantial real progress has been made since the advent of the so-called "probabilistic revolution." It is suggested that the simplicity and transparency of the approximating approach should be preferred.