604 WILSON. [Vol. VIII. 



displacements in the 8-celled, or even in earlier stages, as the 

 blastomeres settle down to the resting phase after division. 

 Amphioxiis differs from Clavelina in the fact that the bilater- 

 ality is less firmly established, and in this respect agrees very 

 nearly with the frog-cleavage, as described by Roux (No. 24) 

 and Rauber (No. 23). It would be interesting to examine a 

 larger series of Clavelina embryos, for it seems, on a priori 

 grounds, very probable that even in this case the bilaterality 

 may undergo occasional modification without affecting the end 

 result. 



In view of these considerations I conclude that the bilateral 

 cleavage of Amphioxus is an hereditary form, which is in 

 process of development, or perhaps of disappearance. The 

 primitive type, from which both the others arose, was probably 

 the radial, and a comparison of Figs. 19-21 with 25-27 will 

 show how easily the bilateral type may be derived from it. 



I must finally call attention to the remarkable fact described 

 at p. 582, that in the 8-celled stage the occasional rotation of 

 the four micromeres is invariably in the same direction as in 

 annelids, mollusks and polyclades — i.e., with the hands of a 

 watch. In annelids and mollusks, as I have pointed out 

 (35' P- 454) the direction of the rotation stands in a definite 

 relation to the unilateral origin of the mesoblast. In Amphioxus 

 the mesoblast has a bilateral origin, and the direction of the 

 rotation must be otherwise explained. The rotation itself 

 might well be due to purely mechanical causes, but such causes 

 leave its constancy of direction unexplained, and the possibility 

 of its being an ancestral reminiscence must be held open. 



Part IV. — Regeneration and the Mosaic Theory of 



Development. 



Two independent and nearly simultaneous attempts have 

 recently been made to harmonize the facts of regeneration (or 

 "postgeneration") in animal embryos, with the so-called mosaic 

 theory of development. The first is contained in Weismann's 

 remarkable work on the Germ-plasm (33) ; the second is by 

 Roux (25), who has reached conclusions essentially in agree- 

 ment with those of Weismann. Both these attempts are 



