398 THE GRAMMAR OF SCIENCE 



but suppose we stretch the diagram vertically in the ratio 

 of (T to (J (p. 203), then the slope of the line of regression 

 will be r X cr ja- instead of r, and the standard deviation 

 of the associated character will be a' and no longer cr. 

 Further, provided the vertical scale represent equal parts 

 of <j , as the horizontal represents equal parts of cr, it is 

 quite indifferent what the numbering on the scale may 

 be; it might run, for example, from 15 to 26, instead of 

 5 to 16. All we must take care of is that a' is measured 

 about the mean of its own organ, which will no longer be 

 10.04, but whatever falls at this point on the scale. We 

 thus reach the conception of the correlation of two unlike 

 organs in the same individual. Nay, we may go a 

 stage further, and take two quantities not belonging to 

 the same individual and see if these be correlated. 

 Illustrations of this will be given later. Meanwhile, as 

 an example of different organs in the same individual, 

 I give a correlation table of the first joints of the 

 middle finger and of the ring finger of the left hand 

 obtained from measurements on 551 women. Here the 

 mean values are 2.37" and 2.19" ; the standard deviations 

 are .114" and .108" ; while the correlation is .914. 



Ample illustrations of the correlation of organs from 

 different individuals will be given in the following chapter. 



[Table 



