250 BUMPUS. [VOL. I. 



The measurement of 1000 shells from Tenby, Pembrokeshire, 

 Wales, reveals the following facts : The breadth of the most 

 elongated shell is 83^ of the height. The breadth of the 

 most ventricose shell is 98^0 of the height. There are four 

 shells having an index of stature of Sq.fi, four of 85/0, twenty- 

 five of 86 f /o, thirty-eight of Sf'/o, etc. 



Chart I. l --\i from these data we construct a "curve of 

 frequency," it is evident that the location of the curve upon the 

 base line indicates the general shape of the shells from Tenby. 

 The length of the base line, inclosed by the limbs of the curve, 

 is an index of the amplitude of variation in respect to stature of 

 the 1000 shells, and may be numerically expressed as 1 5 (98 83 

 = 15). The altitude of the curve is an expression of conservatism, 

 i.e., it represents the segregation of the shells around a mean ; 

 and the flowing trend of the curve is at once an indication that 

 a sufficient number of individuals has been collected to be of 

 statistical value, and that the tension which tends to draw 

 varieties away from the mean is constant. 



Chart IL - - If we now examine the curve of distribution of 

 1000 shells from South Kincardineshire, Scotland, we note that 

 it lies further to the left; i.e., the shells are more elongated, 

 though its altitude and flowing trend are not considerably dif- 

 ferent from those of Chart I. The amplitude of variation, 

 however, is slightly less, being represented by 14 (95 --81 = 



14). 



Chart III.- -The curve of distribution for 1000 shells from 



the H umber District, while having practically the same altitude 

 and the same flowing trend as curves I and II, and resting at a 

 mean position upon the base line between the ordinates of 85^0 

 and 977' presents, nevertheless, extreme fixity - - the amplitude 

 of variation being indicated by only 12 (97 --8 5 = 12). This 

 remarkable constriction of the base line gives a precipitous 

 appearance to the curve which Charts I and II do not have, 

 and indicates a striking paucity in examples which depart, even 

 a little, from the ideal mean. 



1 The subdivisions of the base line from left to right represent the grades of 

 increase in ventricosity expressed in per cents, i.e., by indices of stature. The 

 length of the ordinates represents the number of individuals of each grade. 



