293 



adjoining kenozooecia by means of canal-like perforations. On the second figure, 

 showing the lateral wall which has most connections, we find corresponding to 

 these 7 small rosette-plates, the situation of which near to the basal surface 

 shows, that the inner cavity belongs to the oldest ])art of the kenozott'cium, whilst 

 the remaining part of the thickness of the wall has arisen from the later deposited 

 calcareous layers. As Waters' has already remarked, every second layer extends 

 further out than the foregoing and therefore the walls of the adjoining kenozoa-- 

 cia dovetail into each other. The avicularia of the kenozooecia, which are con- 

 nected with the cavities mentioned, send out canals, wliicli partly and often 

 through separating walls stand in connection with corresjionding canals from 

 other avicularia (PI. XXIII, fig. 1 a), and partly open out through |)ores on the 

 outer surface. At the same time as the originally formed avicularia are overgrown 

 by the gradually deposited calcareous layers, new ones are constantly formed, and 

 we therefore find on the frontal wall partly quite superficial, partly more immersed 

 and finally overgrown, though still distinct avicularia (PI. X, fig. a c). As the keno- 

 zooecia are dependent for their nourishment on the zooecia, they must have an 

 internal connection with these, and we therefore find on the basal wall of more 

 or fewer zooecia a small, round opening, which must correspond to just as many 

 rosette-plates on the zoa'cial surface of the kenozoa'cia. In Ret. tesselata, which 

 sometimes consists of two layers of ordinary zooecia (PI. X, fig. 4 c), the quite 

 young kenozooecia contain a large cavity (PI. X, fig. 4 b), but it seems to dis- 

 aj)pear fairlj' quickly or to become reduced (fig. 4 a) and in older zoa?cia to ex- 

 ist only in connection with an avicularium. If we grind down the frontal wall 

 of the zoa'cia in this species, we find on the basal wall a number of pores, which 

 undoubtedly serve for connection with the kenozooecia. Whilst PI. X, fig. 4 b 

 shows the growing edge of a colony ground down transversely, fig. 4 a shows a 

 much older part of the colony similarly treated, and the three rows of cavities 

 seem to indicate that we have here three layers of kenozooecia. That the keno- 

 zotpcia may occur in several laj'^rs, the one outside the other, apjiears even more 

 distinctly from PI. X, fig. 2 b, which represents the proximal part of a colony of 

 Ret. mediterranea ground down transversely. This part, which is seen from the 

 side in fig. 2 a, shows in transverse section three originally separated branches, 

 a narrower and two broader, which have become connected to one mass by 

 means of numerous kenozooecia, which at this place have not only overgrown 

 both surfaces of the colony by several layers, but have also filled up the space 

 between the single branches. On the transsected surface we see a transverse row 



' 115, p. 78, I'l. VI, figs. Ga-Gb. 



