In cross‐sensor beamforming the narrow‐band quadrature signal at one sensor is multiplied by the complex conjugate of the narrow‐band quadrature signal at a second sensor to form a stable element of the cross‐sensor field. After suitable time averaging, the beam response can be calculated as the weighted correlation between the cross‐sensor field that is measured and the field that is expected. IfNsensors form a sparse line array, thenNlines form a sparse square array. Examples are given showing the performance of a sparse square array and a random planar array in resolving a realistic multiwave signal plus noise field. A simple adaptive algorithm is used to resolve weak signals close to strong signals and to reduce sidelobe problems associated with a random array.