140 Records of the Indian Museum. [Vol. X, 



some genera completely lost by degeneration. The skeleton 

 consists of a more or less well defined network in the structure of 

 which diactinial or occasionally tylote spicules take an important 

 part, if they do not compose it altogether. In cases in which the 

 spicules form definite fibres they either lie parallel or nearly 

 parallel to one another in the core of these fibres, the chitinoid 

 covering of which varies greatly in strength and thickness (if it 

 exists at all), or else are connected together in a chain-like forma- 

 tion by means of small patches of a similar substance. In some 

 species no spicule-fibres can be detected and the skeletal network 

 is constructed entirely of single spicules either joined together by 

 patches of chitinoid substance and each encased in a thin film of the 

 same substance so as to form a lattice-like reticulation, or else 

 merely massed in the parenchyma without any definite arrangement. 



In many genera of Haploscleridae the life-history is unknown, 

 but in those in which it has been investigated a free-swimming 

 larva is produced that is covered externally (except at the broader 

 end) with cilia and has a solid body. In many cases the larval 

 ciliated cells exhibit distinct signs of specialization in certain 

 legions, but a pigment-spot is not present. 



In the Spongillidae, which are closely related to the Haplos- 

 cleridae and by some authors given only subfamil}' rank, the 

 typical microsclere is a small amphioxous spicule covered with 

 minute spines, which are evenly disposed on its surface. That 

 this spicule ma}^ be convergent towards the sigma is proved by 

 a study of the gemmule-spicules of Spongilla, the most primitive 

 genus of the family, and Pedis pongilla, in both of which certain 

 microscleres have a distinctly C-like outline. The evolution of the 

 microscleres takes, however, a very characteristic course in the 

 family as a whole. In the first place a tendency for the differen- 

 tiation of the minute spicules that have no part in the formation 

 of the skeleton into two distinct, but not widely divergent 

 types makes its appearance in the most primitive forms. In 

 Spongilla it is already well established ; we find simple spiny 

 amphioxi, which are never strongh^ curved, lying free in the 

 dermal membrane and the parenchyma, and also other amphioxi 

 of stouter build and slightly more complicated structure associated 

 only with the gemmules. Although the latter spicules often 

 approach the sigma- type in outline more closely than the " flesh- 

 spicules " do, they are more highly specialized as regards the 

 spines that cover them, in that these spines are often distinctly 

 longer and more recurved at the extremities of the spicule than 

 on the middle part. Unimportant as this specialization usually is 

 in Spongilla (it is much more strongly marked in Peclispongilla^ 

 an offshoot from the direct line of evolution in the family), it has 

 a well-defined significance in other, more highly developed genera. 

 In Ephydatia,^ a genus closely resembling Spongilla in general 



^ In the genus lotrochota of the family Desmacidonidae, which as a family 

 is characterized by the presence of the chela or its derivatives, and again in certain 



