A rigorous application of the traditional definition of sonar equation terms leads to the appearance of an unexpected factor, not routinely included, equal to the ratio of the characteristic impedance at the receiver to that at the source. An omission of this factor can lead to non-negligible errors for realistic conditions. It is further argued that a gradual change in the de facto definition of propagation loss occurred between 1965 and 1980. Two alternatives to the traditional sonar equation are suggested, each using one of the two propagation loss definitions and both eliminating the unwanted impedance ratio.