SUMMARYIn a discussion of the phylogeny of the subgenusOnagraof the genusOenothera,based on genetical, morphological, cytological, and systematic results, it is indicated thatOnagrawas derived from the large‐flowered subgenusRaimannia,probably in Central America, and that from this ancestry the small‐flowered North American species were produced, through a series of dominant mutations, as the genus spread out and moved northwards following the retreat of the ice. The evidence for these conclusions is derived from (1) the present distribution of the small‐flowered forms in America, (2) the conclusion that e.g.0. Tracyiis derived fromO. grandifloraandO. purpuratafromO. Hookeri,(3) the fact that occasional small‐flowered mutations appear, e.g.de Vriesiiandbienniformisfrom0. Lamarckiana,(4) the fact that0. novae‐scotiaeis composed of two complexes,grandiflorensandparviflorens,the former having petals much larger than the phenotype of the species.Interspecific crossing has played an important part in the development of the subgenus, as a result of which most of the species are permanent crypthybrids, composed of two complexes and breeding true because of catenation or linkage of their chromosomes during meiosis. Thus the species with smallest flowers occur generally in the higher latitudes and usually show catenation of all their fourteen chromosomes into a closed ring. It has been shown experimentally that catenation can arise by crossing two homozygous species ofOenotheraeach having seven free pairs of chromosomes. Probably the hybrid vigour resulting from the heterozygous (heterogamous) condition leads to the survival and spread of such species, while the relatively homozygous derivatives which will occasionally arise through chromatin rearrangement in meiosis combined with inbreeding will be less likely to survive in the struggle for existence. This accounts for the fact that nearly all the small‐flowered species show complete catenation.Following the twin‐hybrid results of de Vries, Renner and others have shown by extensive crossing experiments that mostOenotheraspecies, including all which exhibit a high degree of catenation, are composed of two complexes, the phenotypic equivalents of which are often very different from the phenotype of the species.The conception of parallel mutations in the subgenus is important, because of the evidence that dominant mutations giving rise to smaller flowers have appeared independently and successively in different parts of the continent from different lines of descent. The species and varieties with cruciate petals constitute another series of independent parallel mutations.The large amount of specialization and adaptation in some species ofOenotheracan only be adequately accounted for by the accumulation of small germinal changes (mutations), many of which must have taken place to account for the amount of specific differentiation which appears in the genus. That genemutations occur is known from the existence of such Mendelian mutants asbrevistylis, rvhricalyxand the various dwarf types.Interchange of segments between non‐homologous (heterogamous) species, called by Lotsy internal hybridization or intra‐syngamic evolution, will account for the appearance of a certain number of new types, especially those which are more nearly homozygous than the parent form; but so far as known these usually fail to survive in competition with the more heterozygous, and hence more vigorous, species from which they are derived. For these and other reasons the value of segmental interchange as an evolutionary factor is limited in comparison with the importance of interspecific crossing. Such crossing probably took place on a large scale in the early evolution of the group, producing a swarm of crypthybrids with a high degree of chromosome catenation, which were successful in spreading on account of their hybrid vigour. Gene mutations occurring regularly throughout this swarm are sufficient to account for the further differentiation of species which has taken place, segmental interchange of chromosomes playing a minor role in the same forms.Two new species are described, and the recognized species of subgenusOnagraare listed, with their type‐localities, petal length, and indications of their relationships and chief differences. The complexes and chromosome catenation are also given in those species in which it has been worked out, and references are made to various studies of naturalized hybrid populations in Europe.The expenses connected with the cultivation of many species and their hybrids have been defrayed in part by grants from the Royal Society. Other faculties have been provided in Re