IX] OF THE SKELETON OF SPONGES 453 



what is known as " closest packing," but in linear series ; so that in 

 their arrangement, and by their mutual compression, we tend to 

 get a pattern, not of hexagons, but of squares : or, looking to 

 the solid, not of dodecahedra but of cubes or parallelopipeda. 

 This indeed appears to be the case, not with the individual cells 

 (in the histological sense), but with the larger units or vesicles 

 which make up the body of the hexactinellid. And this being 

 so, the spicules formed between the linear, or cubical series of 

 vesicles, will have the same tendency towards a "hexactinellid" 

 shape, corresponding to the angles and adjacent edges of a system 

 of cubes, as in our former case they had to a triradiate or a 

 tetractinellid form, when developed in connection with the angles 

 and edges of a system of hexagons, or a system of dodecahedra. 



Histologically, the case is illustrated by a well-known pheno- 

 menon in embryology. In the segmenting ovum, there is a 

 tendency for the cells to be budded off in linear series ; and so 

 they often remain, in rows side by side, at least for a considerable 

 time and during the course of several consecutive cell divisions. 

 Such an arrangement constitutes what the embryologists call the 

 "radial type" of segmentation*. But in what is described as the 

 "spiral type" of segmentation, it is stated that, as soon as the 

 first horizontal furrow has divided the cells into an upper and 

 a lower layer, those of "the upper layer are shifted in respect 

 to the lower layer, by means of a rotation about the vertical 

 axisf." It is, of course, evident that the whole process is 

 merely that which is familiar to physicists as "close packing." 

 It is a very simple case of what Lord Kelvin used to call 

 "a problem in tactics." It is a mere question of the rigidity 

 of the system, of the freedom of movement on the part of 

 its constituent cells, whether or at what stage this tendency 

 to slip into the closest propinquity, or position of minimum 

 potential, will be found to manifest itself. 



However the hexactinellid spicules be arranged (and this is 



* Se^, for instance, the figures of the segmenting egg of Synapta (after Selenka), 

 in Korschelt and Heider's Vergleichende Entwicklutigsgescliichte (AUgem. Th., 3*<^ 

 Lief.), p. 19, 1909. On the spiral type of segmentation as a secondary derivative, 

 due to mechanical causes, of the "radial" type of segmentation, see E. B. Wilson, 

 CeU-Uneage of Nereis, Journ. of Morphology, vi, p. 450, 1892. • 



•f Korschelt and Heider, p. 16. 



