ASEXUAL REPRODUCTION IN SAGARTIA 177 



cance, they point to a tendency on the part of symmetrical 

 individuals to divide into symmetrical parts by a vertical 

 plane perpendicular to the directive plane. 



Since a precisely symmetrical old piece has like mesenteries 

 adjacent to its edges, whereas an asymmetrical piece may have 

 like or unlike bounding mesenteries, any tendency toward the 

 formation in division of accurately symmetrical pieces would be 

 indicated by an excessive number of like bounding mesenteries. 

 This gives us an opportunity to discover the limits of the 

 tendency. The numbers of like and unlike bounding mesen- 

 teries of tables 3 to 6 are summarized in available form on the 

 left of table 9. We may best disregard the less common classes 

 and consider only the twenty-eight old pieces with two com- 

 plete non-directive bounding mesenteries, thirty-two cases of 

 one complete and one incomplete mesentery, and nine specimens 

 with two incomplete bounding mesenteries. If we arbitrarily 

 distribute into two equal groups the thirty-two cases of pieces 

 with unlike bounding mesenteries, we have data suitable for the 

 use of Yule's 'Coefficient of Association' ('00, p. 272) as a 

 measure of a possible tendency of the division plane to pass 

 through similar spaces on opposite sides of the directive plane. 

 The distribution of the divisions in these cases is, then, as given 

 in table 1. 



Complete positive association, the invariable association of two 

 incomplete or two complete mesenteries on the two boundaries 

 of a piece, would be represented by a value for Q of + 1. Com- 

 plete negative association, the constant association of a complete 

 mesentery on one boundary with an incomplete mesentery on 

 the other boundary of a single piece, would be indicated by a 

 value for Q of — 1. Merely chance association would be indi- 

 cated by a coefficient of 0. The calculated coefficient with its 

 probable error is — 0.008 ± 0.02. The coefficient, being less 

 than its probable error, is practically zero. Our figures, there- 

 fore, give no indication of a departure from the chance associa- 

 tion of two incomplete mesenteries, an incomplete and a 

 complete one, or two complete mesenteries as old bounding 

 mesenteries lying in one piece adjacent to a single division plane. 



