No. 3.] AMPHIOXUS AND THE MOSAIC THEORY. 601 



right (with the hands of a watch), the second group to the left, 

 the third to the right, and so on. 



These types are connected by various intermediate forms 

 and are obviously devoid of any great phyletic importance. In 

 annelids, for example, the cleavage is strictly spiral up to the 

 32-celled stage, after which it becomes bilateral, though not 

 strictly so. In the nemertean Linens (Barrois, 3) it is perfectly 

 radial up to the 8-celled stage, and then assumes the spiral 

 form by a bodily rotation of the micromeres. In gasteropods 

 the cleavage is at first strictly spiral; in cephalopods it is per- 

 fectly bilateral from the beginning. In Aniphioxus the first 

 two cleavages are perfectly radial, after which any of the three 

 forms may be shown. 



But although these forms of cleavage are thus seen to be 

 without value in the investigation of general phyletic questions 

 they are very important for an analysis of the factors that 

 determine the form of cell-division; for they show, I believe, 

 that cleavage-forms ave not determined by mechanical cojiditions 

 alone. In this regard it is necessary to distinguish clearly 

 between the three types. The spiral type, as I have elsewhere 

 pointed out (35) is a direct modification of the radial type, and 

 is an effect of mutual pressure among the blastomeres (however 

 caused) in accordance with Berthold's principle of minimal 

 contact surfaces.^ 



This is clearly shown in the case of Linens, cited above, 

 where a t^/^pically radial cleavage passes over at the 8-celled 

 stage, into the spiral form by an actual rotation of the blasto- 

 meres. The spiral type, therefore, owes its peculiarities entirely 

 to mechanical conditions, the blastomeres assuming the position 

 of greatest economy of space, precisely like soap-bubbles or 

 other elastic spheres. 



With the radial type the case is not so clear. This form 

 is obviously an expression of Sachs's law of the rectangular 

 intersection of division-planes operating on a spherical mass. 

 Hertwig has pointed out the causal basis of this law (11 and 



1 I am indebted to Dr. Driesch for calling my attention to Berthold's fine trea- 

 tise (No. 4), with which I was unacquainted at the time my paper on Nereis was 

 written. 



