DEVELOPMENT AND LIFE-HISTORIES OF TELEOSTEAN FISHES. 711 



Balfour at one time held the view that the floor of the cavity in Selachians was 

 not truly blastodermic, the floor-cells arising as concretions around yolk-nuclei at the 

 base of the disc (No. 11), and such a cavity would be a germinal, not a segmen- 

 tation-cavity like Van Bambeke's ; but later, Balfour relinquished this view, a com- 

 plete floor being established, he states (No. 11, p. 43), by the growth inward of lower 

 layer cells along with cells formed in the periblast. The cells which Oellacher 

 describes on the floor of the " Keimhohle," he says fall from the roof of the cavity, sink 

 into the yolk, and multiply (Zeitsch.f. w. Zool., xxiii. pp. 12, 13). The real nature of the 

 blastodermic vesicle of Lereboullet is by no means clear, for though Bambeke regards 

 Lereboullet's cavity as no other than his own, yet it must be remembered that 

 Lereboullet's mucous layer is not necessarily a blastodermic layer in the strict sense ; 

 and Van Bambeke himself admits this possibility when he points out the likeness 

 of this layer with his intermediary layer (No. 20a, p. 4), a point E. van Beneden 

 also insists upon. That Lereboullet himself regarded his " feuillet muqueux " or 

 " vegetatif " as extra embryonic, is clear from his denying that it is formed of blastomeres 

 — " in fishes and Crustacea (the crayfish) the mucous layer," he says, " is not of the same 

 origin as the serous layer" (No. 95, p. 14), the one being the true or animal blastoderm, 

 and the other the nutritive blastoderm.* It is not necessary here to decide the real 

 nature of the mucous layer, whether it be truly hypoblastic, or hypoblast and mesoblast, 

 or neither ; it is sufficient to note that the floor of the cavity, according to Lereboullet, 

 has a different origin from the roof, and is not composed of cleavage-products, so that his 

 cavity would not seem to be a segmentation cavity at all, and though he considered 

 himself justified in stating that the blastoderm is " creuse et forme uneveritable vesicule 

 .... dont les parois sont plus ou moins rapproches l'une de l'autre" (No. 93, p. 487), 

 yet it must not be regarded as the segmentation-chamber of a blastosphere, but the 

 germinal-cavity underlying a morula. If Oellacher be right, that only cells resulting 

 from cleavage form the blastoderm, then a cavity, if not floored by such cells, is not 

 a segmentation-chamber according to the accepted view regarding that cavity. The 

 nature of the floor of any cavity appearing in an early blastoderm is all important, while the 

 nature of the roof is not so, being, indeed, subject to variation in very closely allied forms 

 like Rana and Triton, one layer of cells forming the roof in the latter (No. 147, p. 453), 

 whereas in Rana the roof is two or more cells thick. The lamprey has a multicelled 

 roof, which thins out to a single layer, as Shipley has found, in agreement with 

 Calberla, and as opposed to M. Schultze ; whereas in Elasmobranchs, as also in 

 Ganoids (Acipenser), the ectodermic roof is thickened by endodermic cells which creep up 

 the walls of the cavity and pass along the roof. The roof of the germinal cavity in Teleos- 

 teans is formed by the whole of that portion of the blastoderm which is raised to form it 

 (PI. II. fig. 15&, odm). It therefore includes epiblast (or ectoderm) and lower layer or 



* That Lereboullet's upper layer cannot be the epiblast, and his second layer the entoderm or " lower layer cells," 

 is shown by the fact that he speaks of the lower as a single layer (No. 93, p. 492), and the upper as of many regular 

 layers of smaller cells, so that our interpretation holds best. 



