Following the pioneering work on pollination biology of the late 18th and 19th centuries, the phenomenon of dichogamy (the separation of presentation of pollen and stigmas in time) was, together with a plethora of other floral features, interpreted exclusively as an adaptive mechanism enhancing outcrossing. More recently, however, David Lloyd and co-authors (Lloyd and Yates 1982, Lloyd and Webb 1986) have highlighted the inconsistency of the outcrossing argument in species which possess both dichogamy and one or more additional mechanisms (strong self-incompatibility, unisexuality, herkogamy) which preclude or limit opportunities for self-fertilization. Lloyd et al. elegantly resolved this dilemma by viewing dichogamy primarily as a mechanism that reduces interference between the sexual functions of pollen export and receipt, while accepting that dichogamy will also often reduce self-fertilization. Dichogamy may take two forms, with stigma presentation preceding (protogyny) or following (protandry) the presentation of pollen (Lloyd and Webb 1986). Which form of dichogamy a plant practices has important implications for the plant's effectiveness in avoiding self-interference and self-fertilization. The efficiency with which the respective modes of dichogamy avoid self-interference may depend primarily on the relative ease of moving androecia and gynoecia (see below), although other more specific aspects of flower design and vector behavior may also play a part (Lloyd and Yates 1982, Lloyd and Webb 1986, Webb and Lloyd 1986). Similarly, the effectiveness of the two modes of dichogamy in preventing self-fertilization depends on the degree of dichogamy, and on which 'mode' of selfing is encouraged (Lloyd and Schoen 1992). Where dichogamy is complete, self-fertilization is totally precluded irrespective of the order of stigma/anther presentation (Lloyd and Webb 1986). However, many species exhibit some overlap in presentation (partial dichogamy), and it is here that differences in the efficiency of preventing selling and the selling mode become apparent. In incompletely protogynous flowers, there is a period of stigma receptivity (and opportunities for outcrossing) before selling is possible, with delayed selfing th result (Lloyd and Webb 1986, Lloyd 1992, Lloyd and Schoen 1992). In contrast, in flowers which are incompletely protandrous, selling occurs as soon as cross-fertilization is possible, unless all pollen has been previously removed (Lloyd and Webb 1986). The result is competing selling (simultaneous opportunities for crossand self-fertilization; Lloyd and Schoen 1992), although strictly speaking selling may be prior (where opportunities for selling precede opportunities for outcrossing) if visitation rates are very low. Because delayed selfing allows cross-fertilization to occur before selling, entails no pollen or seed discounting, and ensures all un-crossed ovules are selfed late in the flower's life, this mode of selling is always advantageous (Lloyd 1979, 1992). Competing and prior selling can provide reproductive assurance; however, both selfing modes are subject to pollen and seed discounting (Lloyd and Schoen 1992), and conditions favoring competing and prior modes are more stringent (Lloyd 1979). Therefore, where the aim is total avoidance of self-fertilization, complete dichogamy is preferred, although this precludes the possibility of reproductive assurance through selling. Where plants are partially dichogamous, self-fertilization is effectively reduced while still permitting some selling. However, of the two possible modes of dichogamy, partial protogyny should be more advantageous than partial protandry in both avoiding