92 WFLLIAM BATBSON. 



Nowj returning to what is found in Balanoglossus, it is to be 

 noted that, first, the cord separates from the skin as a solid 

 rod connected at the two ends to the skin, and upon this con- 

 dition invagination supervenes at the two ends, forming a 

 neural tube in these regions. Let us follow the effect which 

 an extension of this system of invagination along the cord will 

 have upon the origin of the dorsal roots ; for it is nearly certain 

 that invagination in this case is secondary to delamination ; the 

 condition in Amphioxus, in which the medullary plate folds up 

 after being enclosed, offering a stage of transition between the 

 condition found in Balanogiossus and that of an Elasraobranch, 

 for example. Since the invagination of a plate of tissue differs 

 from the separation of a cord in the fact that it is not the 

 central line, but the two edges of the plate, which remain last 

 in connection with the skin, it follows that, as the process of 

 invagination phylogenetically arrives at the point of attachment 

 of any one of these median dorsal roots, it must take up its 

 new attachment at one of these two edges. It is thus not 

 possible, supposing these views correct, that the dorsal roots 

 could in the first instance have been paired, except on the 

 hypothesis that as the process of invagination phylogeneti- 

 cally reached its point of attachment each dorsal root split into 

 two ; which is almost impossible, and which the condition of 

 Amphioxus shows not to have occurred. The other alternatives 

 would be (1) that all the dorsal roots should remain attached 

 on one side to the cord j (2) that they should be attached 

 irregularly to one side or the other ; and lastly (3) that 

 they should have been attached alternately to either side. 

 From the nature of the case they could not be opposite. 

 Now, the fact of their alternate arrangement in Amphioxus 

 is almost a proof that the latter alternative was the one 

 which occurred. (It may be observed that, as a physio- 

 logical convenience, they probably supplied the two sides 

 of the body alternately while yet attached in the middle 

 line.) Thus the opposite origin of the dorsal roots is 

 almost certainly secondary to an alternate arrangement. 

 The fact that it is the foremost pairs which are opposite in 



