84 BALANOGLOSSIDA. 



left side are nearly always larger than those of the right. The 

 lining of the hind end of the dorsal canals, whether they end 

 blindly or not. appears to give off nests of cells into the chondroid 

 skeletal tissue (see below), which is found in the neck of the 

 proboscis. 



There is the greatest variation in the proboscis pores and in the dorsal 

 IM mils ;ui<l in the end vesicles leading to them. In Ptychodera ftava, Harri- 

 maniaku I'tftri . ;md in Stereobalanus canadensis there are almost always two 

 proboscis | Hires. In Glcssobalanus minutus, and Dolichoglossus koivalevskii 

 there ;ire t\vn in some individuals, but not in all. In Ptychodera flava 

 there i-, the most remarkable individual variation in the arrangement of 

 the dors<il canals leading to the pores. The following variations have been 

 nliserved : (1) the right dorsal canal ends blindly and is connected by a 

 solid cord of cells with its end vesicle which is smaller than the left and 

 opens outwards by a smaller pore. (2) The right dorsal canal communi- 

 cates with a wide terminal vesicle by a narrow canal, the lumen oL-which is 

 oecluded : the left canal ends in the chondroid tissue of the proboscis 

 skeleton (see below), in which it gives off numerous islets of cells ; the left 

 end- vesicle which is thus not even connected with its canal by a cord of 

 cells, is present and opens to the exterior by the left pore. (3) A similar 

 arrangement to (2) but the right vesicle is in open communication with the 

 ritiht dorsal canal, and there is a solid cord connecting the left dorsal canal 

 .UK! end-vesicle. (4) Both end-vesicles communicate with their corre- 

 sponding dorsal canals, but the left vesicle is the larger, (.">) The right 

 vesicle is larger than the left, and neither of them are in communication 

 with their dorsal canals. Finally, in Bal. carnosus, in which only the left 

 pore and vesicle are present, the end vesicle is continued behind the an- 

 terior neuropore as two csecal pockets placed ventral to the neural tube, 

 and in some specimens the pore opens into the front end of the medullary 

 t nl ie (for the behaviour of the first branchial pore in this species, see p. 77). 

 \Vhen there are two pores the right is said to arise later in development 

 than the left. The lining of the end vesicles is said to be due to an 

 ingrowth of ectoderm : if this is true the actual coelomic pore would be 

 th" aperture connecting; the vesiele \\ith the dorsal eoelomic canal. 



The central complex of the proboscis (Figs, 60, 64, 65, 67). 

 Before proceeding to our account of the coelom in the collar, 

 it will be convenient to describe this remarkable structure. It 

 occupies almost the whole of the neck of the proboscis and in the 

 dorsal middle line is in contact with the basement membrane of 

 the ectoderm (Fig. <>7). On the ventral side of it is the ventral 

 extension of the proboscis coelom. In some species (Ptychodera 

 flava and cn/!//rt, etc.) the ventral canal of the coelom extends 

 back as far as the union of the notochord with the buccal 

 epithelium, but in most it stops short of this and the posterior 

 pirt of the central complex is in contact in the ventral middle 



