148 SILICISPONGLE. 



archenteron, and a mesoblastic layer between this and the now ciliated 

 epiblast. At the narrow hinder end of the embryo the mesoblast becomes 

 thickened, and largely obliterates the archenteron. In this part of the 

 mesoblast silicious spicula are formed. The larva becomes attached by 

 its hinder extremity, and in the course of this process flattens itself out 

 to a disc-like form. From the nearly obliterated archenteric cavity out- 

 growths take place which give rise to the ciliated chambers. These 

 are not placed directly in communication with the exterior, but open, if I 

 understand Ganin rightly, into a space in the mesoblast, which subsequently 

 acquires an exterior communication the primitive osculum. The subse- 

 quent pores and oscula are also formed as openings leading into the meso- 

 blastic cavity, which communicates in its turn with the ciliated chambers. 



It appears that in the present unsatisfactory state of our 

 knowledge the larvae of the Porifera may be divided into two 

 groups : viz. (i) those which have the form of a bla'stosphere or 

 else of a solid morula ; (2) those which have the amphiblastula 

 form. 



In the former type the mesoblast and hypoblast are formed 

 either from cells budded off from the outer cells of the blasto- 

 sphere or from the solid inner mass of cells ; while the outer 

 ciliated cells become the epiblast. This type of larva, which is 

 found in the majority of sponges, is very similar in its general 

 characters and development to many Ccelenterate planulae. 



The second type of larva is very peculiar, and though in its 

 fully developed form it is confined to the Calcispongiae, where it 

 is the usual form, a larval type with the same characters is 

 perhaps to be found in other sponges, e.g. amongst the Gum- 

 minese, and amongst the Silicispongiae where one-half of the 

 embryo is without cilia, though in the case of the Silicispongiae 

 the cells of the ciliated part of the embryo correspond to the 

 granular cells of the larva of Sycandra. 



The later stages in the development of the larvae of the Pori- 

 fera are not similar to anything we know of in other groups. 



It might perhaps be possible to regard sponges as degraded descendants 

 of some Actinozoon type such as Alcyonium, with branched prolongations of 

 the gastric cavity, but there does not appear to me to be sufficient evidence 

 for doing so at present. I should rather prefer to regard them as an 

 independent stock of the Metazoa. 



In this connection the amphiblastula larva presents some points of 

 interest. Does this larva retain the characters of an ancestral type of the 

 Spongida, and if so, what does its form mean ? It is, of course, possible that 



