84 CORRELATED VARIATIONS. 



and female flounders (Pleuronectes flesus), and the 

 values obtained showed that the correlation was affected 

 by sex in 17 out of the 40 instances. The coefficients 

 were greater in the male than in the female fish in 11 

 instances, and in the female than in the male in 6 in- 

 stances. Several of the pairs of bilateral homologous 

 measurements (such as the numbers of rays in the right 

 and left pectoral and ventral fins) showed distinctly 

 lower correlation constants than were shown by the 

 corresponding pairs of measurements in the sym- 

 metrical fish Acerina cernua and Coitus gdbis. This 

 was doubtless due to their possessing slight differences 

 of function in the asymmetrical fish. 



It should be mentioned that Pearson, "Warren, and 

 Duncker employed a somewhat modified and improved 

 formula for determining these correlation constants,* 

 as compared with that originally suggested by Galton. 

 G. O. Yule,f and also Pearson and Filon,$ have recently 

 shown that the correlation can be determined in the 

 case of skew variation, as well as of normal variation. 



All these results may be taken to show that every 

 part and organ of the body is correlated with every 

 other part in a greater or less degree, though such cor- 

 relation may sometimes be of the negative order. The 

 immense importance in evolutionary processes of such 

 correlation, whereby when one organ becomes modified 

 by the action of an agency such as Natural Selection, 

 others are modified also, is sufficiently, obvious to need 

 no discussion. 



*Phil. Trans., 1896, A. p. 264. 

 fProc. Roy. Soc., Ix. p. 477, 1897. 

 {Phil. Trans., 1898, A. p. 229. 



