The work of French and Bishop is compared critically with previous attempts to explain the images of grain boundaries in field‐ion micrographs. It is argued that the assumption that the regions adjacent to a boundary behave as rigid bodies is unrealistic for low‐angle boundaries and dubious for high‐angle boundaries in the light of existing theories and experimental observations. In the absence of relaxation it is difficult to relate stepped spiral contrast in the field‐ion image to dislocations unless they have very wide cores. Combined transmission electron microscopy and field‐ion microscopy can be used to clarify the nature of grain boundary dislocations.