66 
ANNIE MATTHEWS. 
the yolk globules become fewer but larger in size, and 
continue this decrease in number and increase in bulk for 
some little time, apparently by fusion of the smaller globules 
(PL 3, fig. 18, and Pl. 3, fig. 14, Y. V.). Meanwhile the 
segmentation cavity, which has persisted until now (Text-fig. 
34), disappears. The columnar ectoderm cells increase greatly 
in length and number, becoming very slender, while smaller 
rounded ones appear among them (PL 3, fig. 15, Ect. and 
R. C.). The outer halves of the columnar cells have become 
finely granular, the inner portions still containing yolk 
globules (Pl. 3, fig. 15, Gr. E. and Y. E), while the round 
cells each contain about four large yolk globules. 1 The 
innermost endoderm cells, which are also the largest, now 
begin to degenerate (PL 3, fig. 16, c.), and by the continued 
absorption of these cells a series of cavities forms which 
presently fuse into one — the coelenteron — the wall of which 
consists of a layer of endoderm about three cells deep, with 
the columnar ectoderm outside. As development continues 
the eudodermic layer becomes progressively thinner by the 
degeneration and ultimate absorption of the inner cells, until 
it consists of only one row of columnar cells about half as 
long and twice as broad as the columnar ectoderm cells. 
Round cells are at this time found lying between the tall 
•ectoderm cells at both their inner and outer ends (PL 4, 
fig. 27, a drawing of the next subsequent stage, which would 
also serve as a figure of this stage). At times the columnar 
ectoderm cells proliferate very rapidly, and so the ectoderm 
becomes temporarily multilayered, soon, however, regaining its 
normal columnar character (Text-fig. 35). In the final stages 
(Text-fig. 3), the ectoderm cells of the pre-planula are very 
deep indeed, and become separated from tlie endoderm by a 
thin structureless membrane (Text-fig. 35, S.M.). There is 
no direct evidence to show which layer secretes this, but it 
seems identical with and indeed an early part of, the meso- 
1 Hickson (4a) describes the endoderm in the planula of Alcyonarians 
as a plasmodium, but from the present account it is clear that in 
Alcyonium digitatum tlie endoderm is always a definite cellular 
layer. 
