52 



CELL-DIVISION AND GROWTH 



II. i 



inversely with its length, so that when all the drops are equal 

 the cross furrow lies in the same plane with the first division, 

 and so disappears. 



By another division it is possible to make a ring of eight drops 

 whose surfaces of contact all meet in one line, or in a 'seg- 

 mentation' cavity (Fig. 28). To realize this condition, however, 

 it is necessary that the division should be equal, and its direction 

 accurately radial. If unequal, the larger drop invariably passes 

 towards or wholly into the inside. If oblique or tangential the 

 inner drop passes into the segmentation cavity (Fig. 32). 



f 



FIG. 32. Three stages in the passage of a large drop (a") into the 

 centre of the system. The first stage extremely unstable. (From 

 Korschelt and Heider, after Roux.) 



Unequal division of all four equal drops produces very in- 

 teresting patterns, some of which recall the appearance of bi- 

 laterally segmenting ova, when the divisions are correspondingly 

 unequal on each side of the first or second division (Figs. 29, 30), 

 while others resemble certain phases of 'spiral' division when small 

 and large cells regularly alternate (Fig. 31). It is a rule for the 

 smaller of the two drops to go to the periphery, while the larger 

 assumes an oblong or wedge shape, passing towards the centre if 

 it does not slip entirely inside. The latter occurs with clean oil, 

 when the large drop is flanked by small ones on both sides. 



It is also possible to divide four equal drops horizontally 

 into two tiers. The upper drops, however unless absolutely 

 undisturbed quickly come to alternate with the lower. 



In these systems of drops the final arrangement is due to, 

 first, the principle of least surfaces; secondly, the circum- 

 scribing boundary ; thirdly, the size of the drops ; and fourthly, 

 in some cases, the direction in which they are divided. 



It only remains for us to consider, with Roux, to what extent 



