45 o Wilson: [Vol. vi. 



by Eiipomatiis, Discoccelis, Nereis, etc., in which the third 

 cleavage is more or less inclined, so that the upper cells are 

 displaced (apparently in all cases towards the right — i.e. follow- 

 ing the hands of a watch), and in extreme cases are thus caused 

 to alternate with the lower four. 



I wish to make the point that the foregoing considerations 

 render it probable that the spiral type is a secondary deriva- 

 tive of the tnie radial type, for the latter conforms more nearly 

 to Sachs's law, and agrees with the cleavage of spherical plant- 

 structures. Both are primarily owing to the spherical form 

 of the ovum, but the spiral modification must be caused by 

 some additional secondary factor. The nature of this factor 

 seems to be very obvious, although when closely examined it is 

 found to lead to some conclusions so important that they cannot 

 be accepted without very strong evidence. The factor in ques- 

 tion is the effect of mutual pressure among the cells. It is well 

 known that free cells tend to assume a spherical form, whereas 

 in masses they tend primarily to assume the same geometrical 

 forms as soap-bubbles or other regular elastic bodies, — i.e. reg- 

 ular dodecahedrons when in bulk, regular hexagonal prisms when 

 arranged in a single layer upon a plane surface. In the case of 

 lifeless bodies {e.g. soap-bubbles) it is certain that these forms 

 are the result of mutual pressure, and it is mathematically 

 demonstrable that they are the forms which afford the greatest 

 economy of space. The approximate conformity of cell-forms 

 to these ideal mathematical forms makes it impossible to doubt 

 that they are due to the same mechanical laws. 



All these are elementary commonplaces. But the remarkable 

 fact, and one which does not seem to be very clearly recognized, 

 is that the effect of these mechanical conditions on the dividing 

 cells has become hereditary. This I think is clearly proved by 

 a comparison of the true radial and the spiral types of cleavage ; 

 the comparison at the same time explains a number of interest- 

 ing features in the early stages, among others the cross-furrows 

 of the four-celled stage. In the segmenting ovum we have to 

 deal essentially with cells arranged in a single layer, either sur- 

 rounding a central cavity or (as is especially clear in the Nereis- 

 embryo) spread out on the spherical surface of a group of larger 

 cells. They tend, therefore, towards the hexagonal form, though 

 probably they never actually assume the ideal form, since they 



