(JRAI'TOIJTES OF NEW YORK. TART 1 



523 



development was <also a siciila. In the Dendroidea it is possible that the 

 flanking tubes of the tliecae, considered 1)y Holm and Wiman to be 

 gonangia, contained the sexual products. In the Axonolipa nothing has 

 been found as yet Avhicli M'ould suggest receptacles for the development 

 of the generative elements ; for the entire rhabdosome is composed of 

 nothing but thecae or supposed nourishing individuals. 



Fig. 3 



,(:§|(|p^ 



Fig. 6 



Fig. 7 Fig. 8 



Fig. 3-8 Eiidendrium ramosum Linn. Gi-owth stages (Copy from Allman ) 



The initial part of the sicula is obviously to be considered as the embiyo 

 sheath, as also suggested by Holm. 



The hollow nema is, as the writer's shale material shows, extended 

 into a primary disk [pi. 2]. The complete embryonic shell consists hence of a 

 disk and a short cone, connected by a holloAv tube. The question arises, 

 then, whether one of these parts was formed before the other or whether all 

 originated together. The answer to this problem is probably furnished 

 by the development of certain hydroids, that pass through a growth stage 



