715 



implies that among the eighteen vertebrae thus brought together there 

 are really two kinds: seventeen old ones and one new one. In the 

 six specimens of Necturus containing nineteen instead of eighteen pre- 

 sacral vertebrae, the most careful scrutiny failed to reveal anything 

 by which I could detect the new vertebra. The vertebrae ditfer in no 

 essential respect except that of ordinal position. Therefore, I believe 

 that in the case of Necturus the facts do not warrant the statement, 

 that in the change from the typical condition with seventeen intervening 

 vertebrae to the exceptional one with eighteen such, a new vertebra 

 has been intercalated. What can be maintained is, that in place of 

 seventeen vertebrae in one specimen, eighteen appear in another. 



I do not wish to imply in this connection that a kind of inter- 

 calation of vertebrae may not in some instances take place. Baur 

 ('91, p. 331) in a recent article has shown very conclusively that, in 

 place of one vertebra, two or parts of two may arise, the process 

 evidently being a partial division of the material from which a single 

 vertebra ordinarily arises ^). This process, which I should call the multi- 

 plication rather than the intercalation of vertebrae, may in some cases 

 account for the increase of presacral elements, but I am not inclined 

 to ascribe to it the wide-spread importance that Baur does. In Nec- 

 turus, at least, I have never observed half-vertebrae or other indi- 

 cations that a process of multiplication was going on, though variations 

 in the number of presacral vertebrae occur once in every five or six 

 specimens according to my observations. 



If, then, we assume that the sacrum is fixed in position, we are 

 obliged to admit that eighteen vertebrae in one specimen represent 

 seventeen in another, and it follows from this that exact homologies 

 between the members of two such groups are no longer possible, for 

 example, the tenth vertebra in one group cannot be said to correspond 

 exactly to any one vertebra in the other group. 



In attempting to avoid this difficulty one naturally turns to the 

 second general assumption, namely, the travelling of the pelvic girdle 

 backward or forward over a fixed vertebral column. This explanation 

 seems at first sight all sufficient, but, as Bateson ('94, p. Ill) has 

 pointed out, it implies a process which very probably does not occur. 

 Thus, in Necturus the sacral ribs develop typically on the nineteenth 

 vertebra and occasionally on the twentieth. In these occasional in- 

 stances, however, there is not the least evidence, so far as I am aware, 



1) An instance of this kind among the Urodela has lately been 

 described by Fieu) ('95, p. 353) in Amphiuma means. 



47* 



