VARIABILITY OF SUCCESSIVELY FORMED WHORLS. 



101 



The equations to the fitted curves, when I = 2, are 

 Straight Hne, y = 0.7812 ] 1 - 0.2297 (|) [ 



Parabola, y = 0. 7812 -{ 1. 0338 - 0. 2297 ( j) - 0. 1015 (|) [ 



The result is the same as before. There is very clearly a decrease 

 in the variability with successive whorl formation. Since the total 

 number of successive whorls dealt with in this case (only 10) is so small, 

 it is perhaps worth while to examine the results obtained when each 

 whorl is considered by itself instead of being- paired with the one 

 following. Fig. 20 shows the result in this case. Here each circle 



represents the ratio — for the whorls in the particular position desig- 



nated by the abscissal number. 



\^' 



.'^ 



0.9 

 0.8 

 0.7 



x> 



0.6 



ID 



^ 0.5 

 0.4 

 0.3 



<k— 



\ 

 \ 

 \ 



^- 



\ 



o 



IZ345&789I0 



Whorls 

 Fig. 20.— Scedastic curve for whorls on all branches. Series V. 



In fitting the straight Hne to the observations in this case I have used 

 only the first seven points. The last three are very irregular, on account 

 of the small number of whorls on which they are based. The equation 

 to the Hne, where now I = 3, is, 



?/ = 0.8108 1 1 -0.1326(^) 



We reach the same result as before, namely, that the variability 

 decreases as we go out on the branch. The smoothness of the scedastic 

 curve for the first five observations in this case is remarkable. 



