3 o8 



THE AM ERIC AX XA TC 'R A LIST. [Vol. XXXV 1 1 . 



1. From the latter part of Table 8 it will be seen that 

 length has little to do with the number of joints in the antennae. 

 If the — sign were significant it would mean an inverse correla- 

 tion, the longer the animal the fewer antennal joints. But the 

 probable error is ± .067 so that p may as likely as not lie any- 

 where from — .08 to + .054. There is thus essentially no corre- 

 lation. You can say, a priori, that the antennae of the larger, 

 presumably older specimens are more likely to have been broken 

 and to be found regenerating. There is no way of telling a 

 regenerated terminal segment from an original termination. 



2. Length of body and number of prosternal teeth have 

 little to do with each other, the coefficient of correlation varying 

 between .065 and .195. I picked out from the data the four 

 individuals with 14 prosternal teeth, the maximum number. 

 Their lengths were 24, 24, 22, and 21 mm. In the four indi- 

 viduals with 8 teeth, the minimum number measured, the lengths 

 were 20, 22, 23 and 23 mm. This tells roughly what the coeffi- 

 cient of correlation tells precisely. In the case of the curve for 

 the prosternal teeth, which is strongly bimodal, the bimodality is 

 due to the tendency towards bilateral symmetry. At the ends 

 of the series this tendency is overcome by the tendency of varia- 

 tions to revert toward or group around the mode. Hence the 

 larger number of individuals with 9 and 13 teeth compared to 

 those with 8 and 14 teeth. 



3. The length bears a more decided relation to the number 

 of coxal pores. The number of coxal pores on each of the right 

 legs was correlated in succession with the length. The coefTi- 



