Cn. VI] FORMATIOX OF THE GERM-LAYERS 73 



As before, its ventral wall is composed of larger and yolk-bear- 

 ing- cells. Above and laterally the walls are formed of smaller 

 cells. The latter have now arranged themselves in a definite 

 layer, and have become somewhat flattened (Fig. 26, B, C, D). 

 This layer is also sharply separated from the mesoderm. The 

 mesoderm, as compared with its previous condition, has under- 

 gone important changes. It has extended further ventrally, 

 and has met from the right and left sides in the mid-ventral line 

 along most of the ventral surface. Over the dorsal and dorso- 

 lateral walls of the archenteron it forms a thinner layer of cells 

 than in the earlier embrj'o (Fig. 25, B). 



There is still a ventral region of the embryo where the ecto- 

 derm and the yolk-cells are in contact, i.e. a region into which 

 the mesoderm has not extended (Fig. 26, C). The medullary 

 plate is seen in cross-section. It will be noticed that the plate 

 is much thinner in the mid-dorsal line than at the sides. On 

 each side the medullary plates show a differentiation into two 

 parts. The most lateral and ventral edge of the plate is formed 

 of cells less closely held together than those nearer the mid- 

 dorsal line. This mass of rounded cells is the beginning of the 

 neural crest. 



The mesoderm in the mid-dorsal line is thickened in the 

 posterior sections. According to some writers, this median 

 mesoderm has always up to this time remained closely fused with 

 the layer of endoderm beneath it. It marks the beginning of the 

 notochord. 



The formation of the notochord takes place from behind 

 forwards, so that in the same embryo different stages of its 

 development may be found (Fig. 26, D, E). 



The account given above of the formation of the notochord 

 is not generally accepted, particularly since the formation of 

 the notochord from the endoderm is the method followed by 

 many, perhaps by all other vertebrates. That a median mass 

 of tissue stretches at first across the dorsal median wall of the 

 archenteron in the frog cannot be denied, but many embryolo- 

 gists have preferred an interpretation different from that which 

 I have followed. It is affirmed that there is always a closer con- 

 nection hetiveen the endoderm and the tissue lying above it in the 

 dorsal median line than between the endoderm on each side of 



