When we now include the data from the amphibian limb, we must surely conclude that RA induces an ectopic ZP A which then organizes the anteroposterior axis by so far undiscovered mechanisms. Since RA can proximalize, postcriorize, and ventralize the cells of the amphibian limb by a application rather than a local application, the establishment of gradients of RA is not necessary and it is inconceivably simplistic to propose that RA is the morphogen whkh generates positional information in aU three limb axes at once. It is much more realistic to propose that RA is a powerful switching molecule which establishes boundary conditions during the early phases of limb development or regeneration from an initially homogeneous group of cells. The application of exogenous RA could establish new boundary conditions, e.g., by binding to the inhibitor in a reaction-diffusion system This idea would explain why RA-binding proteins and RARs are present, and unites the data on chicks and amphibians, but leaves undiscovered the identity of the molecules which actually specify pattern in each of the three axes of the limb.