When there is variation between individuals of the same plant species for a putative defensive chemical it is possible to prove beyond reasonable doubt that the chemical does indeed have a defence function for the plant. The polymorphism for cyanogenesis in Lotus corniculatus and Trifolium repens provides ample evidence that cyanogenesis has an important role in plant defence. This herbivore-plant interaction is, however, far from simple: the degree of selectivity varies between individuals of the same animal species and is influenced by differences in hunger and the type of food previously consumed. The primary effect of cyanogenesis is at the feeding and not at the digesting stage, although it seems that it is the carbonyl compounds released on cyanogenesis which have the deterrent effect and not hydrogen cyanide itself. Although cyanogenesis is not a particularly effective method of defence, it is a conservative system--the compounds involved are recycled and the plant has to be physically damaged before cyanogenesis occurs. Because there is no pre-damage volatile compound involved, cyanogenesis cannot be used by insects or other animals as a means of finding the plants.