Rihaczek and Hershkowitz have proposed that radar scattering from man-made targets is dominated by cavity-type reflectors that do not meet the assumptions of point scatterer theory. The effect of such dispersive scattering is to produce spurious bright responses in the radar image that can appear as a number of peaks lying along an arc. If such arc-type responses do exist in radar imagery then their removal would enhance the ability to focus moving target imagery and to perform automatic target recognition. Rihaczek and Hershkowitz have outlined a method for identifying such arc-type responses that use the image phase. Algorithms have been written to automate the outlined procedure and used to analyse a large number of synthetic and inverse synthetic aperture radar images. Arcs are present in most radar images of static and moving ground targets. To investigate whether the arcs are due to a physical mechanism or a chance occurrence, randomised target imagery was generated and also analysed. No significant differences are seen between the distributions of arc numbers obtained from randomised and real imagery. In conclusion, the arcs present in the real imagery are likely to occur by chance, rather than because of an underlying scattering mechanism.