Darwin's orchid book (1862) has been cited as his first detailed example of how to study evolution (Ghiselin, 1969; Gould, 1986). The book starts as a presentation of observations showing that the morphology of orchids is, in most cases, wonderfully well suited to having insects remove and deposit pollinia. It ends by tracing how the enormous diversity of orchids can be seen as arising through modifications from ancestral forms. What Darwin did not do was to explain how orchid flowers come to be different. He probably thought that by showing how to study the origin of adaptation he had shown how to study the origin of diversity. Mayr has often pointed out that Darwin failed to see genetic isolation as a precondition for speciation, and thus for diversification (e.g., Mayr, 1959). Likewise, we contend that evolutionists have seldom clearly dissected the alternatives for how divergence occurs, given isolation. We shall concern ourselves here with how different environments-in our case, different pollinator regimes- do or do not provide heterogeneity in selection that might adaptively drive the divergence of flowers.