CORRELATION AND HOMOLOGY. 27 



carpopodite correlations which is strong enough to outweigh any poten- 

 tial influence of position. The question as to whether the rule of con- 

 tiguity holds within the metamere can not be definitely settled until a 

 larger number of segments than is here considered has been dealt with 

 from each leg. 



The results set forth in this section of the paper may be summarized 

 as follows: The rule that structures occupying positions within the 

 organism contiguous to one another are more highly correlated together 

 than non-contiguous structures is found to hold (with a single insig- 

 nificant exception) for the correlations of the homologous segments of 

 different legs; it also holds for about half of the correlations between the 

 non-homologous joints of different legs; it fails signally in those cor- 

 relations involving a carpopodite as one of the variables, both in the case 

 of the non-homologous joint correlations of the different legs and in the 

 case of the correlations of the joints of the same leg. These exceptions are 

 probably due to the influence of some special factor concerned in the 

 carpopodite correlations. 



CORRELATION AND HOMOLOGY.* 



The problems to be considered in this section are as to the influence of 

 morphological homology on the degree of correlation between parts. Are 

 homologous structures in general more highly correlated than function- 

 ally similar but non-homologous structures? Our data furnish abundant 

 material from which light may be gained on this and the subsidiary 

 questions which are suggested by it. We a b c 



may first attack the general problem accord- 

 ing to the following plans: It is evident that a b 



<J CJ J. D 



the same segments of the different legs are 



serially homologous. The correlations between 



these Will furnish data for the "homologous" FlG ' '-For explanation see tezt. 



side of the comparison. What shall be compared with these to make a 

 fair test ? It is clear that there are several possible ways to proceed. The 

 correlation between joint a of leg A (fig. 7) and its homologue, joint a of 

 leg B, may be compared with the correlations (i) between joint a' of leg A 

 and joint 6' of leg B; (n) between joint a of leg A and joint c' of leg B. 



*A preliminary paper dealing with the subject taken up in this section has been 

 published by Clawson (1905). -An entire recalculation of the constants after the publi- 

 cation of this preliminary paper led to the discovery of several minor arithmetical errors. 

 These have now been corrected with the effect of making the whole system of correla- 

 tions somewhat smoother, but without essentially affecting the conclusions. 



