A unified theory of the reciprocity calibration of electromechanical transducers is presented in a form which facilitates the derivation of particular calibration theories. The theories of two types of calibration procedure are treated as special cases. One of the theories is shown to reduce, when lumped standards are used, to that of the original technique of Trent; the other is descriptive of a new procedure.The application of calibration theories is discussed—particular attention being devoted to the use of distributed‐constant standards and the consequent difficulties involved.