180 DONALD WALTON DAVIS 



plete endocoels below the first grade are added to the exocoels 

 as shown in my table 10, these form only 12 per cent of all the 

 divisions, whereas Torrey and Mery found 28 per cent of all 

 the positions of division planes to be in exocoels. Apparently, 

 S. davisi and S. luciae differ in the proportion of divisions in 

 exocoels. 



As between incomplete and complete endocoels the chances 

 cannot be accurately stated, since the cycles of incomplete mesen- 

 teries are variable in number. Probably division occurs very 

 rarely if at all in a region where the first cycle of incomplete 

 mesenteries is not yet established. The members of this cycle 

 are equal in number to the complete mesenteries. Usually in- 

 complete mesenteries of the second grade are also present. 

 These are twice as numerous as those of the first cycle. Later 

 cycles are sometimes present, at least in the older parts of a 

 specimen. We may say, therefore, that probably not less than 

 50 per cent nor greatly more than 75 per cent of endbcoels are 

 incomplete; and that while, according to chance, divisions should 

 occur in incomplete endocoels in at least as many cases as in 

 complete endocoels, the actual numbers found are 57 incomplete 

 to 103 complete endocoels, or 36 per cent of divisions in incom- 

 plete endocoels. There is, then, a marked tendency to divide 

 in complete endocoels. 



Since in regeneration the full number of complete mesenteries 

 to be formed appears very promptly and the directive or non- 

 directive characteristics are obvious at an early stage, the 

 chances of a later division plane passing in directive or non- 

 directive endocoels may be quite accurately determined. The 

 old parts of animals represented in tables 3 to 6 show 572 com- 

 plete mesenteries. Of these, 154 are directives. Therefore 

 approximately 27 per cent of the complete endocoels of these 

 animals are directive endocoels. Of 103 divisions in complete 

 endocoels (as calculated for table 10), six, or approximately 6 

 per cent, are in directive endocoels. That is, of the divisions in 

 complete endocoels there were less than one-fourth as many 

 divisions in directive endocoels as chance would demand. This 

 is only another expression of the fact, already demonstrated 



