Summary1. Atavisms emerge as evidence of localized modifications in development of an organ or of one of its parts. Different developmental processes can be triggered within the same organ rudiment, presumably in response to the same stimulus. We saw that that stimulus can have a genetic basis in a mutational event, which can be selected for. We also saw that atavism can be produced by experimental manipulation within developing systems ‐increased growth of the chick fibula, enamel production from avian ectoderm, and balancer formation in amphibians. Such atavisms arenotbased on heritable genetic changes. They indicate the developmental plasticity that exists within embryos and the relative ease with which development can be switched from one programme to another.2. Examination of mutants (wingless chicks), limbless vertebrates and experimental manipulation of embryos, shows that cell death, inductive tissue interactions and altered patterns of growth are developmental mechanisms used in the formation of atavisms.3. Differential development mechanisms can be triggered within the same organ at the same time to produce atavisms. In the guinea pig, formation of atavistic digit V involvesprolongationof growth of metatarsal V whereas formation of atavistic digit I involves development of anewmetatarsal I.4. Secondary functional modifications ensure that the atavism is integrated with the other components of the functional unit, as illustrated by extra digits in horses or guinea pigs and fibulae in birds. Atavistic 2nd and 4th digits in horses arise by continued growth of their primordia. A consequent reduction in the growth rate of digit 3, the normal single functional digit, enables all three digits to attain approximately equal lengths and so potentially to function. The altered functional load transmitted to the limbs results in secondary but correlated alterations in muscles and skeletal elements in other portions of the limbs. The fact that embryonic digit 2 normally develops to a more advanced state than digit 4 explains why digit 2 more often develops atavistically, for if variation in growth rate is the basis for the atavistic digit, digit 2 has an advantage over digit 4.5. Atavisms should not be an embarrassment to the evolutionary biologist. They are the outward and visible sign of a hidden potential for morphology change possessed by all organisms. Neither basic capacity to form the organ nor patterning information is lost. Modification of components of inductive tissue interactions helps to explain how organs are lost during evolution (also see Regal, 1977); retention of the basic mechanism explains how structures can be revived as atavisms (also see Rachootin&Thomson, 1981). Frequency of atavisms thus provides an indication of the degree of modification or loss of the underlying developmental programm