No. 3.] MORPHOLOGY OF THE ACT/NOZOA. 313 



ceeding stage a form in which the ectoderm and endoderm are 

 well differentiated histologically, the endoderm cavity is filled 

 up with food-yolk, and the st-omatodasum has already formed. 

 There is evidently a great hiatus between the two stages. 

 Kowalewsky's description ('75), likewise, leads one to infer that 

 he observed only isolated stages, and did not actually witness 

 the conversion of the early invagination stage (Mark, '84, PI. 

 XI, Fig. 27) into the fully formed gastrula (F"ig. 28), which, it is 

 to be noticed, is exactly similar to my figure of the gastrula of 

 Metridiwn (PL XIII, Fig. 10). 



These facts, taken into consideration with the difficulty of 

 observing the delamination by optical sections, render it not 

 improbable that the interpretation of the figures referred to is 

 erroneous, and that what really occurs is what I have described 

 for Metridiiim, and there is a further point which renders this 

 probability almost a certainty, Jourdan's Fig. 117 shows the 

 endodermic cavity to be filled with food-yolk, which is not pres- 

 ent in the stage represented in Fig. 116. Whence does this 

 food-yolk come } Kowalewsky's figures of the gastrula of Ceri- 

 antJius (Mark, '84, PI. XII, Figs. 2 and 3) show essentially the 

 same thing, though the amount of food-yolk present in the 

 endodermal cavity is much smaller. Kowalewsky believes ('73) 

 that the food-yolk is secreted by the endoderm cells after invagi- 

 nation, and is later on reabsorbed by them — a very remarkable 

 proceeding. In the gastrula of Metriduim food-yolk is to be 

 found in the endoderm cavity, since it occurs in the blastocoel, 

 and I think it exceedingly probable that future observations 

 will show that, instead of invagination occurring in these forms, 

 the process of endoderm formation is essentially what I have 

 described above for Metridiwn. The cases of supposed gastru- 

 lation by invagination which have been described for the Hex- 

 actiniae must be denoted, along with Haeckel's account of 

 Monoxenia, as not autJienticatcd. 



In all the Actinozoa, then, concerning which we have reli- 

 able information, the endoderm is formed by delamination. It 

 remains now to be considered whether what we find in Metri- 

 diuvi or in Rejiilla is the more primitive. This inquiry leads to 

 another ; namely. Was delamination the primitive method of 

 endoderm formation, or is it a secondary modification of a still 

 earlier phe7iomenon ? 



