118 BULLETIN OF THE 
surfaces of the wall, I infer that they result from a process of migration 
inward, either at the time of cell division or independently of that pro- 
cess. Indeed, there is obviously no other possible source whence these 
cells could come, but the exact process of transfer is not easily determined. 
I believe that this increase in number is at first for a considerable time 
due exclusively to the migration of cells which once shared in forming 
the external boundary of the sphere, but later the division of cells which 
have already migrated into the deeper portion of the ectoderm undoubt- 
edly contributes to this increase. 
We have now to turn our attention to a phenomenon of considerable 
importance, the study of which from preserved material is, however, 
attended with difficulties. I refer to the ingression of cells from the 
wall of the blastosphere into the cleavage cavity, which begins a con- 
siderable time before the invagination commences. The latter does not 
take place until the number of cells forming the wall of the blastosphere 
has exceeded 400, whereas the ingression, as far as can be inferred from 
the cases which I have studied, may occur at any time after the blasto- 
sphere contains about 100 cells up to the period of invagination. The 
phenomenon of ingression in A. flavidula is not of constant occurrence, 
but when it does take place is similar to that represented by Goette 
(87, Taf. I. Figs. 1-5) for the earlier stages of the blastula in A. 
aurita, It consists of a migration into the cleavage cavity of one or two, 
rarely more than three, of the cells of the blastospheric wall. With 
the exception that they assume a spherical form, because relieved from 
pressure, they are at first similar in size, as well as in nuclear and other 
characters, to the cells remaining in the wall. 
The study of ingression upon preserved material is attended with diffi- 
culty, since in any one specimen we have the condition at only one stage 
of development, and cannot say wh certainty what its condition has 
been in past stages, or what it might have been during some subsequent 
period. This can be determined only by studying the conditions exist- 
ing in other embryos killed at other stages, and arranging all in their 
probable natural sequence. In view of this fact, I have sectioned and 
examined several hundred embryos which were killed at different stages of 
development. As far as possible the results obtained from these sections 
have been verified by the study of embryos cleared and mounted whole. 
Although this ingression occurs before invagination, I have deferred the 
discussion of it until now, because invagination is constant in its occur- 
rence, whereas the ingression does not appear to be so; indeed, the 
majority of the specimens have shown no indications of it. 
