Those plant species that rely on animals to mediate the transfer of pollen usually attract pollinators with visual and/or aromatic stimuli and then reward them with pollen, nectar, or some other food (Faegri and van der Pijl, 1971; Proctor and Yeo, 1973). Under some circumstances (i.e., competition for pollinators) there may be selective pressures to increase the size of the attractants and rewards in order to secure more visits by the pollinators. One possible response of a plant to these pressures is to cluster flowers in time and space. This would make the plant more conspicuous to pollinators and it would also increase the pollinator's reward by decreasing the search time and distance between food sources (see Burtt, 1961; Janzen, 197 la; Heinrich and Raven, 1972; Gentry, 1974; Willson and Rathcke, 1974; Heinrich, 1975). There are many plant species that regularly produce inflorescences consisting of many flowers but initiate only a few fruits. Species with this flowering and fruiting pattern include Albizia lebbeck (Maheshwari, 1931), Aesculus spp. (Percival, 1965; Benseler, 1975), Cassia grandis (Janzen, 197 la), Eucalyptus spp. (Carr et al., 1971), various orchard trees (see Kozlowski, 1973), Asclepias spp. (Willson and Rathcke, 1974; Wilbur, 1976; Willson and Price, 1977) and Catalpa spp. (Stephenson, 1978). Because each flower on these species has a low probability of setting fruit, there is the possibility that inflorescence size has been selected for factors other than a flower's ability to set fruit (see Willson and Price, 1977). In this study, I use Catalpa speciosa (Warder ex Barney) Engelm. (Bignoniaceae) to test which of three alternative hypotheses explains the proximate factors involved in a high ratio of flowers to fruits and then to test several ultimate factors that infer the adaptive significance of this pattern.