Efforts have been made to construct systems of transformism along Mendelian lines, but none of them has met with notable success. Lotsy, for example, sought to explain all variation on the basis of the rearrangement of preëxistent genetic factors brought about by crossing. But such a solution of the problem is very unsatisfactory. In the first place, the generality of hybrid (heterozygous) forms are ruled out on the score of instability. The phenotype of hybrids is directly dependent, not on the genes themselves, but on the diploid combination of genes contained in the zygote. This combination, however, is always dissolved in the process of gamete-formation, by the segregative reduction division which occurs in the reproductive organs of the hybrid. Hybrids, therefore, do not _breed true_, if propagated by sexual reproduction. To maintain constancy of type in hybrids, one must resort to somatogenic reproduction (_i.e._ vegetative growth from stems, etc.). Certain violets, in fact, as well as blackberries, are maintained in a state of constant hybridism by means of this sort of reproduction, even in nature. In the case of _balanced lethals_ (_i.e._ factors causing death in the pure or homozygous state), the hybrid phenotype may be maintained even by sexual reproduction, inasmuch as all the pure (homozygous) offspring are non-viable. Two lethals are said to be balanced, when they occur, the first in one and the second in the other homologous chromosome of the same synaptic pair. “Such a factorial situation would maintain a state of constant heterozygosis, the fixed hybridism of an impure species ... the hybrid will breed true until the relative position of the lethals are changed by a crossover, or the genetical constitution in these respects is altered by a mutation.” (Davis, _Science_, Feb. 3, 1922, p. 111.) As is evident, however, the condition of balanced lethals involves a considerable reduction in fertility.