32 



SUuhj of N. obsofcta and N. trivittata 



Likcwisc it cannot be a^sumed that the coefficicnt of variability has the saine 

 significance fm- any two cliaracteristics. 



If now \ve run ovcr the coefficients of variability wc sce that the niost variable 

 character is dcpth of suturo, wherc the variability is about oiie-third of the dopth. 

 It is interesting to observe that this inost variable character is to geiieral 

 Observation onc of the most distinctivc of all those raeasured. That is, it is 

 what Brewstcr ('9Ü) would call a " conspicuous " character, aiul to the degrce 

 to which it represents roughncss of the shell it would have great wcight in 

 detennining the species. Its great variability is, therefore, in accord with the 

 thesis of Brewster ('97) that "any measurable qnality is, in general, variable in 

 individuals in proportion as it is a distingiiishing character of the gi'oiip to which 

 the individuals belong," though the variability is in this case much niore extreme 

 than in the case of any evidence Brewster offers. 



(9) Gorrelations. 



Table III. shows the amount of corrclation between some of the characters. 

 The characters of N. trivittata show a much groatcr amount of correlatiou 

 than those of iV. obsoleta, and the grcatest amount of correlation is shown 



TABLE III. 

 The Amount of Gorrelation between Various Gkaracteristics. 



between the angle and the ratio of diameter to lcngth (which we have called 

 globosity). Since globosity is the most important factor in determining the 

 apical angle, it is natural that the correlatiou between these two characters should 

 be great. The relatively small correlation between length and numbcr of whörls 

 Icads US to lock for other factors in the dctermination of length tiiau number of 

 whorls. These factors we find in the .size of the whörls and the incliuation of the 

 axis of the whorls to the axis of the shclls. The values of the correlations between 

 the characters length, index, and apical angle arc largely detormined, not onl}' 

 in magniuulc but in sign, by the fact that, other things bcing constant, the index 



