No. 3.] AMPHIOXUS AND THE MOSAIC THEORY. 58 1 



At first the four blastomeres are grouped about the vertical 

 egg-axis with perfect radial symmetry, leaving a symmetrical 

 rhomboidal space between them, as in Hatschek's Fig. 6. This 

 space disappears completely as the embryo settles down into 

 the succeeding quiescent period, and it is then that the varia- 

 tions first become apparent. Slight as they are, they deserve 

 attentive consideration; for they give, I believe, a key to the 

 more considerable deviations of later stages. In some cases 

 the four blastomeres meet precisely along the egg-axis. More 

 commonly, their mutual pressure gives rise to slight displace- 

 ments, this way or that, through which "cross-furrows," or 

 "Brechungslinien," are produced. The cross-furrows are, how- 

 ever, usually very short, and are not constant in direction. 

 Occasionally they have the arrangement shown in the typical 

 spiral cleavage of annelids, mollusks, etc., — i. e., they are 

 equal and at right angles to each other on the opposite poles 

 of the embryo. More commonly, they are parallel to one 

 another; that is, two of the blastomeres are in contact alone: 

 the entire length of the egg-axis. In other cases the cross- 

 furrow exists only at one pole, and disappears at the other. 

 A careful comparison of all these cases leaves no doubt, in 

 my opinion, that the arrangement of the blastomeres in the 

 4-celled stage is typically radial, and that the departures from 

 this arrangement are purely accidental, probably depending 

 on slight variations in the activity of the individual blasto- 

 meres as they settle down into the quiescent period. 



2. The third cleavage (equatorial) is of considerable interest, 

 for it shows distinctly three different forms (connected by 

 every imaginable transition) each of which is a fixed type of 

 cleavage elsewhere in the animal kingdom. These forms may 

 be designated, respectively, as (i) radial (as in some echino- 

 derms), (2) bilateral (as in tunicates and cephalopods), and 

 (3) spiral (as in polyclades and annelids). (For more precise 

 definitions of these forms see p. 599 and the accompanying 

 diagrams.) 



a. In the radial fonn the spindles are regularly arranged 

 around the egg-axis and somewhat inclined inwards at their 

 upper ends, as shown in Figs. 4, 5. The divisions result in 



