2Q2 John H. Gerould, 



already show deviations from the regular alternation in direction. 

 From these cleavages there results a plate of cells of nearly equal 

 size, which surround a diamond-shaped group of four cells that 

 divide no more. This I have called the definitive rosette. It is 

 derived from the rosette of the 48-cell stage probably by only a 

 single division of the cells of the latter. Long flagella, which are 

 apparently of no use in locomotion, make their appearance on these 

 cells at the age of ten houi's. At the same time short cilia are 

 formed on the prototroch cells, by the action of which the embryo 

 immediately begins to rotate. I regard the flagella as sensory, 

 probably tactile, organs. 



The marginal cells of the apical plate in Ph. gouldii at twenty 

 hours temporarily sink away from the yolk membrane, whereas the 

 definitive rosette retains its connection with it, rising as a cone- 

 shaped process in the midst of the sunken area (Fig. 59). The same 

 is evident in Ph. vulgare in a less degree, and at a later period 

 (33—36 hours). 



I regard this process as similar to that which occurs in the 

 formation of the amniotic cavity of the head in Sipunculus. Wilson 

 (1892, p. 398—399) describes the putting forth of the apical cilia in 

 Nereis thus: "The egg membrane is at this period separated from 

 the upper side of the embryo by a considerable space. A narrow 

 process is now rather suddenly put forth from the middle of the 

 upper hemisphere (Fig. 60). This process is extended until it comes 

 into contact with the membrane, over a small area from which the 

 cilia are immediately put forth. At this stage the embryo recalls 

 the larva of Sipunculus at the time of amnion formation. The space 

 surrounding the apical tuft is, however, soon obliterated, and the 

 larva again becomes spherical and closely surrounded by the mem- 

 brane. It seems possible, nevertheless, that this peculiar process 

 may give the key to an explanation of the origin of the amnion in 

 other forms." Mead (1897) notes (p. 268) that in the trochophore 

 of Lepidonotus the membrane temporarily "stands out from the body", 

 not only in the region between the prototroch and the „apical tuft", 

 but also behind the prototroch. 



These facts indicate that the sinking of the marginal cells of 

 the apical and the ventral plates in Sipunculus is not an isolated 

 phenomenon, though the nai-row prototroch in Nereis and in Lepi- 

 donotus shows no such resemblance to the serosa in Sipunculus as 

 Phascolosoma presents. 



