3IO McMURRICH. [Vol. IV. 



of the early development of at least four different forms. In 

 Alcyo7iiiim digitatum, according to Kowalewsky ('73), the finely 

 granular protoplasm separates in the form of spherules from a 

 central coarsely granular mass, which thus becomes enclosed by 

 protoplasmic spherules, and later on divides up into a number 

 of cells. In Claviilaria (Kowalewsky and Marion, '83) the yolk 

 does not exert so great an opposing influence to the division 

 forces, and we see the planes extending much further towards 

 the centre of the egg than in Alcyonium, although there is at 

 first a small portion of yolk which does not share in the divis- 

 ion. It gradually disappears, however, with the later divisions, 

 and an embryo similar to that of Alcyojiiiim is formed. In 

 Reiiilla (E. B. Wilson, '83) the phenomenon is essentially the 

 same as in Clavularia, but in Sympodhtni (Kowalewsky and 

 Marion, '83) the segmentation seems to be total. The early 

 phases of segmentation have been thoroughly worked out only 

 in Renilla, and we know that in it considerable variation may 

 occur in the manner of division, the ovum sometimes dividing 

 at once into eight, sixteen, or even thirty-two cells, besides show- 

 ing other peculiarities. This did not depend on the sudden 

 division of the nucleus into such a number of parts, but the 

 nuclear division went on, no doubt, in the usual manner, the 

 cytoplasmic division being inhibited to a greater or less extent 

 by the amount of food-yolk present. In the segmentation of 

 Alcyonuim the same process is apparently carried still farther. 



In all these forms, and in Gorgonia Cavolini described by 

 Von Koch (i^j), the segmentation leads to the formation of a 

 solid structure consisting of an external layer of cells (ectoderm) 

 surrounding a solid central mass of cells. This is the organism 

 for which Metschnikoff {^6) has proposed the term Parenchy- 

 mella or Phagocytella, — rather cumbersome names, which may 

 perhaps be more conveniently replaced by Sterrula. In only one 

 of the forms studied have we definite evidence as to the manner 

 of formation of the central mass, and that is in Rejiilla, in which 

 Wilson has shown that a Sterroblastula (Goette) results from 

 the segmentation, and that the inner ends of its cells delaminate 

 off to form the central solid mass. The similarity of the Ster- 

 rulas of the other forms to that of Renilla seems to indicate 

 that the same process of delamination occurs in them likewise. 

 It certainly cannot be a process of invagination which gives rise 



