SPINAL CORD REGENERATION. II 427 
thick at first, gradually thins out into a typical dorsal fin, but in 
reversed orientation (fig. 1, B—-F). 
In the first paper of this series, it was noted that the epider- 
mis probably plays no réle in the reunion of the spinal cord, but 
that further evidence was needed on this point. The close jux- 
taposition of the epidermis to the cut ends of the cord renders it 
very difficult to settle this question definitely. In both series of 
experiments, where the wound edges were not in apposition, a 
V-shaped invagination of the epidermis occurred between the 
cord ends in the early stages which, on further development, 
became a solid fold and was finally withdrawn. In the early 
stages of this process, the epithelial cells le in nearly direct 
contact with the spinal cord ends and are morphologically in- 
distinguishable from the primitive neural cells. The problem is, 
therefore, the same in both series of experiments. A careful 
study of many embryos brings out certain facts which, if they 
do not afford definite proof, at least give strong evidence in 
favor of the conclusion that the epidermis contributes no ele- 
ments to the regenerating cord. 
An examination of the wound region in an embryo twenty- 
four hours after operation, of which figure 2 may be considered 
typical, shows that, even though the epidermal ingrowth les 
almost in contact with the cells of the spinal cord, it has a very 
definite boundary and, except for an occasional cell, its connec- 
tion with the epidermis covering the body may be clearly seen. 
In the lower part of the invagination the condition is somewhat 
more confused owing to the presence of cells from the notochordal 
sheath, but even here one may differentiate with a fair amount of 
accuracy between the two types. The presence of mesenchyme 
cells here and there between the epidermal and neural cells also 
complicates the problem, but it is usually possible to identify 
them. The slight space between the cord ends and the inter- 
posed tissue seen on the left in figure 2 is too characteristic and 
constant to be a shrinkage space. 
This space between the cord ends and the interposed tissue 
increases with the growth in length of the embryo, due to the 
separation of the cord ends from one another and the withdrawal 
