176 DONALD WALTON DAVIS 



before regeneration possessed no directive mesenteries, and, 

 thirdly, that the division plane cuts the directive plane. 



We may now consider more specifically how nearly per- 

 pendicular to the major axis of the mouth is the plane of division. 

 In other words, is there any tendency toward bilateral symmetry 

 of the structures of the old piece with respect to its directive 

 plane? The simplest cases in which this problem may be 

 studied are those of originally diglyphic specimens with sym- 

 metrically placed non-directive complete mesenteries. Eight 

 specimens of this nature are recorded in table 3. Three of them 

 divided in spaces other than complete endocoels. Of two 

 regular hexameric individuals in this class, one (no. 4) divided 

 in two incomplete endocoels forming two pieces, each sym- 

 metrical with respect to its directive plane; the other (no. 6) 

 divided in one complete and one incomplete endocoel forming 

 pieces as nearly symmetrical as could result from division in 

 such spaces. A third specimen (no. 1), with eight pairs of 

 complete non-directives, divided likewise through one complete 

 and one incomplete endocoel into two nearly symmetrical 

 parts. 



Five originally symmetrical polyps divided in two complete 

 endocoels — assuming that in two cases a bounding mesentery 

 recorded as doubtfully incomplete was really complete at the 

 time of division. Four of these (nos. 7, 11, 13, and 14) were 

 regular hexameric specimens and one (no. 9) was regularly 

 octameric before division. Each of these five divided into two 

 symmetrical pieces. Neglecting the possibility of division in 

 directive endocoels, the chances are even that a regular hex- 

 americ individual dividing in complete non-directive endocoels 

 will produce two sjrmmetrical or two asymmetrical pieces. The 

 number of symmetrical and asymmetrical pieces produced by 

 such divisions should be approximately equal. Actually four 

 divisions of this sort gave eight symmetrical pieces. The 

 chances in such a division of a regularly octameric individual 

 are two to one in favor of producing two asymmetrical pieces 

 as against two symmetrical pieces. One such division gave two 

 synometrical parts. As far as these few cases have any signifi- 



