There are many difficulties in the way of understanding this partial segmentation and the gastrula that arises from it. We have only recently succeeded, by means of comparative research, in overcoming these difficulties, and reducing this cenogenetic form of gastrulation to the original palingenetic type. This is comparatively easy in the small meroblastic ova which contain little nutritive yelk—for instance, in the marine ova of a bony fish, the development of which I observed in 1875 at Ajaccio in Corsica. I found them joined together in lumps of jelly, floating on the surface of the sea; and, as the little ovula were completely transparent, I could easily follow the development of the germ step by step. These ovula are glossy and colourless globules of little more than the 50th of an inch. Inside a structureless, thin, but firm membrane (_ovolemma,_ Fig. 52 _c_) we find a large, quite clear, and transparent globule of albumin (_d_). At both poles of its axis this globule has a pit-like depression. In the pit at the upper, animal pole (which is turned downwards in the floating ovum) there is a bi-convex lens composed of protoplasm, and this encloses the nucleus (_k_); this is the formative yelk of the stem-cell, or the germinal disk (_b_). The small fat-globule (_f_) and the large albumin-globule (_d_) together form the nutritive yelk. Only the formative yelk undergoes cleavage, the nutritive yelk not dividing at all at first. The segmentation of the lens-shaped formative yelk (_b_) proceeds quite independently of the nutritive yelk, and in perfect geometrical order. When the mulberry-like cluster of cells has been formed, the border-cells of the lens separate from the rest and travel into the yelk and the border-layer. From this the blastula is developed; the regular bi-convex lens being converted into a disk, like a watch-glass, with thick borders. This lies on the upper and less curved polar surface of the nutritive yelk like the watch glass on the yelk. Fluid gathers between the outer layer and the border, and the segmentation-cavity is formed. The gastrula is then formed by invagination, or a kind of turning-up of the edge of the blastoderm. In this process the segmentation-cavity disappears. The space underneath the entoderm corresponds to the primitive gut-cavity, and is filled with the decreasing food-yelk (_n_). Thus the formation of the gastrula of our fish is complete. In contrast to the two chief forms of gastrula we considered previously, we give the name of discoid gastrula (_discogastrula,_ Fig. 54) to this third principal type.