NERVOUS SYSTEM. 73 



tiarniensis, hedleyi, ruficollis, and in Balanoglossus apertus. In other species 

 it is as a rule only present at the anterior and posterior ends, being repre- 

 sented in the middle by a number of cavities (sometimes arranged in a 

 double row [Fig. 59] ). In two cases, already mentioned, all traces of the 

 central canal are absent (unless the anterior and posterior pockets are 

 regarded as parts of it), the cord being solid. But even within the limits 

 of the same species it frequently happens that the central canal or its repre- 

 sentative exhibits considerable variation. 



In the Ptvchoderida. 1 the dorsal side of the medullary cord is 



/ *. 



connected with the dorsal ectoderm by a variable number of 

 cords which constitute the dorsal nerve roots (Figs. 60, 8). These 

 roots are frequently hollow (Fig. 58), particularly those which 

 arise from the anterior end of the cord, and their axial canal 

 communicates with the central canal of the cord or with one of 

 the isolated cavities which represent that structure. The axial 

 canals of the roots do not however open to the exterior, but in 

 some species they may be continued for a short distance in the 

 epidermis as intraepidermal canals (Balanoglossus canwsus). 



The dorsal nerve roots vary much in number and in the course which 

 they take in reaching the ectoderm. In Glossobalanus rnfiro//fs there 

 may be from 12 to 18 of them, in Ptychodera flava 1. to 6. As to 

 course, they may be straight or directed obliquely backwards or even 

 curved. In the other families dorsal roots are rarely present. There 

 seems to be a trace of one such structure (not, however, reaching the skin) 

 in SpengeMa porosa. In Dolichoglossus kowalevskii the posterior part of 

 the medullary cord possesses a dorsal keel-like projection, which is solid 

 and does not reach the epidermis. The roots whether solid or hollow 

 consist of cellular prolongations of the cord and are coated with fibrous 

 matter which is a prolongation of that of the cord and is continuous with 

 the nervous stratum of the epidermis. 



The ectoderm over the dorsal and ventral nerve cords of the trunk is 

 thickener) and sometimes projects as a ridge : in some cases it is partially 

 invaginated and so placed at the bottom of a groove. Moreover it differs 

 from the rest of the epidermis in containing fewer gland cells. As an indi- 

 cation of the fact that the nerve cord of the collar is derived from the com- 

 pletion of such an invagination by the union of the sides of the groove over 

 the gland-free nerve-cord ectoderm, may be mentioned the fact that the 

 ectoderm of the floor of the central canal of the collar cord is more or less 

 free from gland cells, while they abound in the roof. Further, as already 

 mentioned, fibrous nerve matter is usually found only in ^he ventral wall 

 of the ventral canal. The ectodermal lining of the central canal and iso- 

 lated medullary cavities possesses a cuticle-like border. It is impossible 

 to arrive at any satisfactory conclusions as to the structure of the nerve 

 tissue. The most that can be said is that the external ends of some of the 

 superjacent ectoderm cells extend into and branch in it. According to 

 Bateson, processes from the fibrous matter of the nerve cords may be 

 traced through the basement membrane into the adjacent muscles. 



There do not appear to be anv organs of special sense. 



