There is an important discrepancy between the statements of Kowalevsky and Agassiz as to the course of the growth of the small cells. According to Agassiz the small cells grow most rapidly at the formative pole and cover this before they meet at the opposite pole. The reverse statement is made by Kowalevsky. It would seem that the above discrepancy is due to an interchange on the part of the one or the other of these authors of the two poles of the embryo, in that according to Agassiz the formation of the mouth takes place _at the formative pole_, and according to Kowalevsky _at the pole opposite to this_. Without attempting to decide between the above views, we shall speak of the pole at which the mouth is formed as the oral pole. [FIG 83. FOUR STAGES IN THE DEVELOPMENT OF IDYIA ROSEOLA. (After Agassiz.) _s.c._ sense capsule; _st._ stomodæum.] The formation of the alimentary cavity commences shortly after the complete investiture of the embryo by the epiblast cells. At the oral pole an invagination of epiblast cells takes place (fig. 83 B), which makes its way towards the opposite pole. More especially from the figures given by Agassiz, and from the explanation of his plates, it would seem that a large chamber is formed in the hypoblast at the end of the invaginated tube, into which this tube soon opens (fig. 83 C). The invaginated tube would seem to give rise to the so-called stomach, while the chamber at its aboral extremity is no doubt the infundibulum, which as may be gathered from Kowalevsky's statements, is lined by a flattened epithelium. At a later period the gastrovascular canals grow out from the infundibulum as four pouches, which are surrounded by, and grow at the expense of, the large central cells, which have in the meantime arranged themselves in four masses, and appear to serve as a kind of yolk. The nuclei of these large cells according to Kowalevsky disappear, and the cells themselves break up into continually smaller masses. The main difficulty in the above description of Agassiz is the origin of the infundibulum. In the absence of definite statements on this head it seems reasonable to conclude that it arises as a space hollowed out in the central cells, and that its walls are formed of elements derived from the yolk cells[84]. On this interpretation the alimentary canal of the Ctenophora would consist, as in the Acraspedote Medusæ and Actinozoa, of two sections: (1) A true hypoblastic section consisting of the infundibulum and the gastrovascular canals derived from it; and (2) an epiblastic section--the stomodæum--forming the stomach. [84] Chun (No. 174) gives a short statement of his observations, which accords with the interpretation in the text.