The results of this experiment seem to confirm Roux’s conjecture that the material of each of the first two blastomeres is of such a sort that it gives rise to half the embryo, and, if so, there would be some probability that there is a causal connection between the first cleavage and the separating out of the parts of the embryo. In fact, Roux drew this conclusion, and even attempted to show how such a qualitative division is brought about. It should not be overlooked, however, that this conclusion goes beyond the legitimate bounds of deduction from the results, since the half-development takes place while the injured half retains its connection with the developing half, the former still remaining alive. On the other hand, the presence of the injured half makes the experiment more suitable to demonstrate that each of the first blastomeres gives rise, under normal circumstances, to half of the embryo. If one half had been removed, we can foresee that its absence might lead to other complications that would affect the result. The most important outcome of this experiment is, I think, to show that a half-structure may develop by itself, _i.e._ that there is a certain amount of independent power of development in the parts of the egg. Roux also tried to show that if, after the second cleavage has been completed, the two blastomeres that lie on opposite sides of the first cleavage plane are killed by a hot needle, the remaining two produce either an anterior or a posterior half of an embryo. An embryo derived from the two “anterior” blastomeres is represented in Fig. 61, _D_. The anterior half of the body is present. Posteriorly the half-embryo abuts against the injured half. It is possible, I think, that this embryo may represent the anterior half of a whole embryo of half size that has been prevented from closing in posteriorly by the mass of injured material of the undeveloped blastomere. Roux did not determine positively whether the two “posterior” blastomeres could give rise to posterior half-embryos; one embryo in his opinion appeared to bear out this interpretation. This part of Roux’s work is, it seems to me, not so satisfactory as the part dealing with the first two blastomeres, and we may leave it, for the present, out of the discussion, and consider only the result of the first experiment, in which one of the first two blastomeres was injured. Since the problems involved in the two cases are essentially the same, nothing will be lost by dealing with the first case alone.