FIELDS AND GRADIENTS 281 



original polarity at all. This is seen in the reconstitution-masses 

 formed from pieces of sponges or hydroids after being strained 

 through bolting-silk.i Yet here, too, axiation, or the development 

 of polarity, later appears, presumably in response to external differ- 

 entials in such factors as oxygen supply. This appears to be com- 

 parable with the determination of the main axis of polarity in the 

 oocyte (pp. 36, 65 ; figs. 27, 132). In passing, it may be mentioned 

 that in the sponge reconstitution-masses, one important step in 

 differentiation, namely the attainment of the two-layered condition, 

 appears to be caused by the migration outwards of the dermal and 

 inwards of the collar-cells from their original scattered positions. 

 Here the fate of the cells is not determined by their position, as is the 

 case with undifferentiated cells (e.g. blastomeres of regulation-eggs, 

 early regeneration-buds), but the already acquired differentiation 

 determines the position taken up. When an excess of collar-cells is 

 present, these cannot be overgrown by dermal cells, and they form 

 spheres or vesicles with the collars directed outwards instead of 

 inwards (see p. 250). 



(iii) Our third general rule is that in regeneration the apical region 

 or head is the first to be formed ; and that its formation, once initiated, 

 is an autonomous process, independent of the level of the cut, and 

 also of the formation of other regions, whether in the regenerated 

 material or within the old tissues of the piece. 



The autonomy of a limited apical region is most clearly seen in 

 the regeneration of Annelid worms. In many of these, the tissue 

 actually regenerated at an anterior cut surface, whatever its level 

 in the body of the worm, never forms more than a restricted head 

 region, composed of a definite number of segments (the precise 

 number varying with the species: it may be as low as two).^ In 

 Planarians the autonomous apical region is the head ; its posterior 

 limits, however, seem not to be quite so sharply fixed as in Anne- 

 lids. The formation of the cephalic ganglion appears here to 



1 H. V, Wilson, 1907; Huxley, 1911, 1921 a. 



2 In some Annelids, anterior regeneration is complete; i.e. the regenerated 

 tissue produces just those segments needed to complete the front end of the 

 worm, and not a fixed number of segments only (see Berrill, 193 1 ; E. J. Allen, 

 1921). This appears to depend on the power of growth in the regenerated tissue. 

 However, the extreme anterior end would here too be the dominant region 

 (p. 285). 



