34 SHINKISHI HATAI 



varies with the diameter of the fiber in the sense that the fibers 

 of greater diameter have the longer internodes" is correct when 

 the fibers are from the same segment of the same animal. 



This conclusion is still true even when all the measurements 

 in the five - groups are added together, since in this case again 

 the greater diameter is still associated with the longer internode, 

 though the degree of correlation is lessened. 



The average of the coefficients of correlation in the five groups 

 is found to be 0.2216. 



2. The Relation Between the Internodal Length and 



Diameter 



a. From a Single Locality in Frogs of Different Sizes 



We have shown already that the diameter varies with the inter- 

 nodal length in the same animal (this is true either for an individ- 

 ual or for averages of several individuals) though the degree of 

 correlation is not high. 



When however, several frogs of different sizes are compared 

 with each other, the internodal length varies, as has been found 

 by Boycott, proportionately more than the diameter, so that in 

 a large frog the length of the internodes on a fiber of given 

 diameter is greater than in a small frog. This finding of Boy- 

 cott can plainly be seen when one examines either the tables 

 or charts of his original paper. I have also prepared by a differ- 

 ent method the tables and charts to demonstrate this point. 



The following table shows the observed values of the internodal 

 lengths corresponding to the different values of diameters in all 

 of the five groups of Boycott. These values have been deter- 

 mined from the correlation tables (see tables 10-14, pp. 45-47) the 

 internodal length being the means of arrays corresponding to 

 the various values of the diameter. 



