Abstract: Data generally representative of presumed noise in an environment are colted and compared with data from one or more statistically symmetric observations of a selected portion of that environment. The data are ranked in order of magnitude and the rank order is stored in a shift register by associating one of two values to each datum according to its source. A plurality of register outputs are combined in summers to produce a representation of the rank order in terms of a complete orthogonal set of functions corresponding to the two-valued function stored in the register. A plurality of adjustable weight function multipliers are associated with the summers, their values having been determined according to statistical assumptions as to the nature of the underlying suspected signal and presumed noise. The outputs of the multipliers are summed and then are compared to a pre-determined threshold value, indicating the presence or absence of signal in the selected portion of the environment.