Experiments with Rubber Balls 71 



But beside the objection raised above that the conical shape of 

 the cells would seem to be more the effect than the cause, and the 

 subsequent change of shape during invagination likewise an effect 

 rather than a cause, it is very doubtful whether, if the changes in 

 assimilation and surface tensions suggested do occur, they could 

 produce a swelling of the inner part of the cell and corresponding 

 reduction of the outer, because we know that the cell when isolated 

 tends to assume a spherical form ; that is to say cells do not retain 

 a conical shape after the mutual pressure has been removed, thus 

 showing that they have no intrinsic tendency to be conical (v. 

 Herbst's figures), and so can hardly be so prone to become conical 

 in the obverse direction as to be able to exert the pressure necessary 

 to convert a blastula into a gastrula. 



In my original experiments on the process of gastrulation in 

 Amphioxus, I experimented with indiarubber balls, which were 

 supposed to represent isolated cells, and which like cells, when 

 pressed together become flattened. In most respects an india- 

 rubber ball behaves as an isolated blastomere does when at rest. 

 I tried by various means representing increase in number illus- 

 trating, that is to say, more rapid growth at various places, or 

 by variation in the size or shape of ball, to reproduce gastrulation 

 such as occurs in a typical way in Amphioxus, but by no means 

 could I do this until I used indiarubber cord applied so as to 

 represent a mutual attraction between cell and cell. 



By means of a circle of indiarubber balls strung together by 

 indiarubber cord I could reproduce exactly the process of con- 

 version of a blastula into a gastrula, or to be more accurate, the 

 conversion of a circle into a double crescent, one part of the circle 

 becoming " invaginated " into the other. 



The necessary conditions are: (i) that the balls shall be 

 sufficiently numerous to cause tension in the elastic cord the 

 balls themselves then becoming flattened against each other, the 

 free inner and outer surfaces remaining convex. 



(ii) The cord must pass nearer to the outward surfaces of 

 a certain number of the balls along one part of the circle than 

 elsewhere. That is to say, if the cord pass through the centre 

 of the balls for two-thirds of the circle, and pass, say half way 

 between the centre and the outer surfaces of the balls forming 



