It has long been suggested that the generation of biological patterns depends in part on gradients of diffusible substances. In an attempt to bridge the gap between this largely theoretical concept and experimental embryology, we have examined the physiology of diffusion gradients in an actual embryonic field. In particular, we have generated in the chick wing bud concentration gradients of the morphogenetically active retinoid TTNPB, (E)-4-[2-(5,6,7,8-tetrahydro-5,5,8,8-tetramethyl-2-naphthalenyl)-1-prope nyl] benzoic acid, a synthetic vitamin A compound. Upon local application of TTNPB the normal 234 digit pattern is duplicated in a way that correlates with the geometry of the underlying TTNPB gradient; low doses of TTNPB lead to a shallow gradient and an additional digit 2, whereas higher doses result in a steep, far-reaching gradient and patterns with additional digits 3 and 4. The experimentally measured TTNPB distribution along the anteroposterior axis, can be modeled by a local source and a dispersed sink. This model correctly predicts the site of specification of digit 2, and provides an empirical estimate of the diffusion coefficient (D) of retinoids in embryonic limb tissue. The numerical value of approximately 10(-7) cm2s-1 for D suggests that retinoids are not freely diffusible in the limb rudiment, but interact with the previously identified cellular retinoic acid binding protein. In addition, D affords an estimate of the time required to establish a diffusion gradient as 3 to 4 h. This time span is in a range compatible with the time scale of pattern specification in developing vertebrate limbs. Our studies support the view that diffusion of morphogenetic substances is a plausible mechanism of pattern formation in secondary embryonic fields.