Most flowers are bisexual in function, but counting secondary allocation to attractive structures such as the corolla as equally male and female leads to the paradoxical conclusion that plants bearing perfect flowers invariably allocate much more to female than to male function. A method of calculating the gender of secondary floral allocation is described, and it is speculated that this allocation is predominantly male. Observations and experiments with natural populations of herbs, designed to test this hypothesis gave the following major results. (i) Insects visit larger flowers more frequently (Fragaria), and removal of floral biomass causes a reduction in the frequency of insect visits proportional to the fraction of biomass removed (Impatiens). (ii) Removal of attractive structures may cause a decline in the probability that a fruit will be formed but has no effect on the number of seeds set per fruit; thus, mutilation of essentially solitary flowers has no effect on seed-set per fruit (Impatiens), while removal of flowers from inflorescences in a species that forms several many-seeded fruits per inflorescence reduces fruit-set per inflorescence but has no effect on seed-set per fruit (Asclepias), and removal of sterile flowers from an inflorescence in which the fertile flowers yield one-seeded fruits is effective in reducing seed-set per inflorescence (Viburnum). (iii) Larger flowers may disperse a greater fraction of their pollen in unit time (Impatiens) and the removal of flowers from inflorescences causes a steep reduction in total pollen exported and a weak decline in the quantity of pollen exported per flower (Asclepias). These results are consistent with the hypothesis that a single insect visit (or a very few visits) suffices to fertilize almost all available ovules and is procured by a very small allocation to attractive structures, while much greater allocation is necessary to procure the numerous visits required to disperse a large fraction of the pollen. This inference is supported by a comparative survey of sexually dimorphic plants, in which male flowers are generally larger than female flowers, male inflorescences bear more flowers, and male plants bear more inflorescences. It is concluded that the flower is primarily a male organ, in the sense that the bulk of allocation to secondary floral structures is designed to procure the export of pollen rather than the fertilization of ovules. This conclusion may be sensitive to whether it is the flower or the inflorescence as a whole that represents the primary unit of attraction to insects. It was found that the performance of a given flower was substantially affected by other flowers in the same compact inflorescence (Asclepias), though not by other inflorescences on the same plant nor by those borne by nearby plants (Fragaria, Impatiens). A general quantitative theory of flower and inflorescence design is outlined, and used to organize the extensive experimental results for Asclepias.