258 



W. J. CROZIER 



The degree of correlation between the sizes of individuals 

 mating together is further shown graphically in a regression 

 plot (fig. 8) . The two components of a copulating pair are usu- 

 ally not of identical size. The value of the index of correlation 

 signifies, however, that as a rule large^ individuals mate with 



16 IT 16 15 14 13 12 11 10 9 8 T 



Fig. 8 Showing the relation between the length of specimens of C. zebra 

 (abscissas) and the average lengths of their mates (ordinates), for 148 pairs. 

 If there were no correlation between the lengths of components of mating pairs 

 the observed points, [2], would fall upon the line m-m' (= the general mean) ; 

 if assertive mating were perfect, the line [2] would coincide with [1]. The 

 smoothed line of regression is shown by the dot-and-dash line; in this and suc- 

 ceeding plots it is assumed that the regression is linear throughout. This is 

 possibly not quite correct, but does not affect the special conclusion drawn from 

 these data, namely that large individuals mate together. The class unit is 1 

 cm. The index of correlation is r = 0.608. 



large, small individuals with small. For each 1.0 cm. difference 

 in the total lengths of two specimens of C zebra there will, on 

 the average, be found a difference, corresponding in sign, of 

 about 0.6 cm. between the lengths of the partners during their 

 conjugation under natural circumstances. 



