516 WILSON. [Vol. V. 



In whichever way the flagellated chamber is formed, it at 

 first has no connection with the canals. It, like any particular 

 canal, is, in its origin, a lacuna, its cavity being an intercellular 

 space. 



In deciding phylogenetic questions, perhaps not much weight 

 should be attached to a development like this ; but whatever 

 weight it has, is in favor of Metschnikoff's theory of the solid 

 ancestry of sponges. The solid swimming larva itself, and the 

 details of the metamorphosis (the origin of the flagellated 

 chambers, excurrent and incurrent canals, and subdermal spaces, 

 as independent lacunas in a matrix of amoeboid cells) are all 

 understood on this theory. Conversely, if we hold to the view 

 which regards the calcareous sponges (Ascons) as the primitive 

 type, the development of Esperella may, of course, be regarded 

 as an extreme case of coenogeny. 



The development of Esperella, it seems to me, has perhaps a 

 bearing on problems out of the range of pure morphology. 

 Without discussing the matter in detail, I may point out the 

 striking resemblance between this asexual development and 

 the egg development of many silicious sponges. As in the 

 egg embryo, there are formed in the gemmule embryo two 

 germ layers. In the two embryos the layers are alike in many 

 details of structure. The egg larvae, again, are characterized by 

 the absence over one pole of the columnar ectoderm (Isodyctia 

 and Desmacidon, Barrois (i) ; Reniera, Marshall (2) ; Chalinula, 

 Keller (3) ; Esperia, Schmidt (4), etc.). The account according 

 to which the endoderm protrudes at this pole, is probably not 

 correct, but it is likely that the ectoderm is only greatly flat- 

 tened over this region. It is this characteristic more than any 

 other, which I should pick out as a point of exact resemblance 

 between the sexual and asexual larvae. Barrois has described 

 the egg development of two forms, in one of which (Isodyctia) 

 the non-ciliated pole of the planula is never covered with 

 columnar cells. This is paralleled by the Esperella larva. In 

 another form (Desmacidon), the larva has at first a complete 

 covering of columnar ectoderm (and cilia), which then disap- 

 pears over one pole (and much later over the whole surface). 

 This case is more or less similar to the gemmule development 

 of Tedania ; in this form, the embryo has at one time a com- 

 plete covering of columnar cells (unciliated however), which 



