This paper describes an empirical approach to an otherwise intractable problem.Measurements were made at 1·48 Mc/s on short rubber cylinders, both with and without cylindrical air-filled holes, immersed in water. As a first approximation, simple calculations based on geometrical acoustics prove satisfactory.Conclusions are drawn regarding the minima in signal strength caused by interference between the two `highlights' on the front surfaces of the rubber cylinder and of the hole, and concerning the number of lobes in the backscattering polar diagram.