THE LIFE-HISTOEY OK NFCULA DELPHTNODONTA. 331 



unequal growth causes it to be divided into two lobes. The 

 dorsal wall of the stomach re-forms, and the liver-cells begin 

 to be rearranged. The commissures between the ganglia are 

 distinctly visible. The otocysts (ot.) are quite large, and 

 contain granules. Although adults have canals leading from 

 the otocysts to the exterior, I have not been able to demon- 

 strate their existence in this or somewhat older stages. The 

 presence of the otocystic canal had been explained (13) as 

 the persistent opening of the otocyst, which was formed as 

 an invagination from the surface of the body. This seems 

 to be the natural explanation, but if canals are present at this 

 stage they are certainly very small. I am inclined to regard 

 the exceedingly small size or absence of these canals as 

 evidence against the view that the otoliths are foreign 

 particles. 



Thus far most of the embryos have been carried in the 

 brood-sacs, but many of them now become free. They are 

 not set free by any act of the mother, but they individually 

 find their way into the mantle chamber of the mother and so 

 to the exterior. 



Frequently younger embryos become free, but they 

 generally do not live long. Many embryos remain in the 

 brood-sacs until a much later period, but they do not seem to 

 be in need of its protection after the stage that has just been 

 described. The brood-sacs frequently remain intact after all 

 of the embryos have left them. 



The more dorsal of the gill lobes elongates into a finger- 

 like process, and the ventral lobe broadens and becomes 

 divided into two lobes (fig. 41, g.). New lobes are thus 

 formed as the result of unequal growth of the most ventral 

 lobe. 



About the time that the third lobe of the gill begins to 

 form a few papillae appear along the margins of the side flaps 

 of the foot (fig. 41,/.). The liver lobes also become hollowed 

 out and lose most of the rounded cells. Part of these cells 

 seem to go to pieces much as if digested (fig. 43), and it 

 seems quite possible that this is the case. 



