Serpula is one of the few Chætopoda at present known in which the segmentation is quite regular[136]. In other forms it is more or less unequal. The formation of the germinal layers has been far more fully studied in the Oligochæta than in the Polychæta, and though unfortunately the development is much abbreviated in the former group, they nevertheless have to serve as our type; and unless the contrary is indicated the statements in the remainder of the section apply to the Oligochæta. The segmentation is nearly regular in Lumbricus agricola (Kowalevsky) and results in the formation of a flattened blastosphere, one of the sides of which is hypoblastic and the other epiblastic, the hypoblast cells being easily distinguished from the epiblast cells by their clearer aspect. An invagination takes place, in the course of which the hypoblast becomes enclosed by the epiblast, and a somewhat cylindrical two-layered gastrula is formed. The opening of this gastrula at first extends over the whole of what becomes the ventral surface of the future worm, but gradually narrows to a small pore--the permanent mouth--near the front end. The central cavity of the gastrula is lined by hypoblast cells, but the oral opening, which leads by a narrow passage into the gastric cavity, is lined by epiblast cells. [136] According to Willemoes-Suhm, Terebellides stroemii is also characterised by a regular segmentation. The segmentation of Lumbricus trapezoides (Kleinenberg, No. 341), and of Criodrilus (Hatschek, No. 339), is more unequal and more irregular than that of Lumbricus agricola, and there is an invagination which is intermediate between the embolic and epibolic types. The segmentation of Lumbricus trapezoides is especially remarkable. It is strangely irregular and at one period the segmentation cavity communicates by a pore with the exterior. Before the completion of the gastrula stage the ovum becomes partially divided into two halves, each of which gives rise to a complete embryo. The two embryos are at first united by an epiblast cord which connects their necks (fig. 141 A), but this cord is very early ruptured, and the two embryos then become quite independent. Some of the peculiarities of the segmentation may no doubt be explained by this remarkable embryonic fission.