1. Oenothera rubrinervis is a half mutant, produced by the copulation of a mutated gamete with a normal velutina gamete of O. Lamarckiana. 2. In consequence, it produces about one-fourth empty grains, a mass mutation of about one-fourth pure or double mutants, and one-half specimens of O. rubrinervis, which will repeat the splitting. 3. The pure or double mutant is called O. mut. deserens. It is very similar to O. rubrinervis, but the leaves of its young rosettes and the bracts of its flower spike are broader and more even. 4. O. mut. deserens is constant from seed. It has no hereditary empty grains. 5. The formula for the self-fertilization of O. rubrinervis is therefore O. (deserens + velutina) = des. x des. + velu. x velu. + des. x velu. The first combination gives the mass mutation, the second the empty grains, the third the normal plants of O. rubrinervis. 6. In crossing with other species the two kinds of gametes will produce twin hybrids, as, for example, laeta and velutina. This assertion has been controlled by making the corresponding crosses of O. mut. deserens and O. mut. velutina. The first produce the laeta and the second the hybrid velutina. The result of a cross of O. rubrinervis is equal to the sum of these two crosses. 7. Outside of the mass mutability into O. deserens, O. rubrinervis is not known to mutate to any noticeable degree. This shows that the internal constitution, which causes the mass mutation, is not in itself a cause for further mutability. 8. The constitution of the gametes of O. rubrinervis can directly be proven by a cross with O. deserens, since O. rubrinervis = (deserens + velutina) x O. deserens produces O. deserens and (O. deserens x velutina) or rubrinervis. 9. Crosses of O. rubrinervis with O. Lamarckiana give three types of hybrids, besides about one-fourth empty seeds. One type exactly resembles O. Lamarckiana and is constant in its progeny. A second type called lucida has broader and more shiny leaves, and after self-fertilization splits off brittle specimens. The third type is either subrobusta or rubrinervis, and in the first case may produce the brittle form in the second generation. All these phenomena are easily explained by the proposed formula for the constitution of O. rubrinervis as a half mutant. They were confirmed by means of crosses with O. nanella and some other mutants. 10. O. oblonga is quite analogous to O. rubrinervis, since it must arise through a mutation of the typical sexual cells of O. Lamarckiana, leaving the velutina gametes unchanged. Contrary to O. rubrinervis, however, the two lethal factors remain in their condition, and moreover the mutated gametes must be assumed to become suppressed in the pollen of the mutant. 11. O. nanella seems to arise through mutations in the velutina gametes of O. Lamarckiana, since after crosses with other species or mutants it is not split off by the laeta hybrids, but only by those of the type velutina.