Hertwig repeated Roux’s experiment and obtained results entirely different from those of Roux. He injured one of the first two blastomeres of the frog’s egg with a hot needle, or by means of a galvanic current. Hertwig states that after the operation the egg turns so that the uninjured part lies uppermost. This is owing, he thinks, to the appearance of a blastula or of a gastrula cavity in the developing part. The segmentation cavity is found in many cases surrounded by the cells of the segmenting half (Fig. 62, _A_), but at other times at the border between the new and the old parts. In still other cases the cavity may lie eccentrically, and in some cases the floor of the cavity may be bounded by the yolk substance of the injured half. An embryo appears on the upper, uninjured part, though it is not, according to Hertwig, a half-embryo, but a whole embryo, or at least one approaching that condition (Fig. 62, _D_, _E_, _F_, _G_, _H_). It is shorter than the normal embryo, and its posterior end is incomplete. When these embryos are cut into sections, it is found that the part that has developed corresponds to the dorsal part of a normal embryo, but the ventral part is continuous with the yolk substance of the injured half (Fig. 62, _B_, _C_, _J_, _K_). Hertwig interprets these embryos as forms in which the yolk portion of the developing half, together with the whole of the injured blastomere, represents a yolk mass that has not yet been enclosed by the margin of the developing part. In nearly all the embryos that Hertwig has described, the medullary folds appear eccentrically on the developing half (Fig. 62, _D_, _F_, _K_), and in some cases they may lie so far to one side that they are situated almost at the edge; and the less development of one of the folds makes the embryo appear almost like the hemi-embryos obtained by Roux. In fact, one embryo seems to have been a true hemi-embryo. Hertwig attributes the eccentric position of the embryo to the eccentric position of the blastopore of an earlier stage, but he does not attempt to account for the eccentricity of the latter.