DAVENPORT AND BULLARD. — CORRELATED VARIATION. 93 



of individuals per niille. The mean number of glands is 3.50 and the 

 index of variability is 1.3G4G. With these data, we can draw a prob- 

 ability curve including about the same area as our observed curve. This 

 curve, the continuous line, is drawn from the equation 



h is the so called index of precision, and is equal to the reciprocal 

 of the index of variability divided by the square root of tt, thus, 



n 



h = 



'^ '\/t 



e is the base of the Napierian system of logarithms; namely, 2.1718. 



h 

 k is a constant determined by multiplying the quotient of —i^ by 



Y TT 



the interval {dx) between successive values of x, in this case, 1 ; thus, 



hdx . 



X indicates deviations from the mean value, and y the corresponding 

 ordinates. 



When X = 0, 7/ =: k, which is thus the length of the ordinate at the 

 mean value of x. Its value gives the percentage of cases which should 

 theoretically occur at the mean; it is in this case 23.3%. Like h, k 

 might be taken as a measure of precision, since it increases as varia- 

 bility diminishes. 



3. The Degree of Correlation betioeen the Number of Glands on the 

 Right and the Left Legs of Individuals. 



To get quantitative results in this matter we must employ a method 

 devised by Galton.* This method depends upon the following procedure 

 and considerations. Separate the right legs into as many lots as there are 

 degrees of deviation from the mean number of glands. These lots may 

 be called the subjects. Find for each of the subjects the mean deviation 

 in the number of glands on the left legs of the corresponding individuals 

 (the relative). The deviation of any subject and the deviation of the 

 corresponding relative are to be compared. In order to make this com- 

 parison instructive, we must take into account the fact that left legs (for 

 example) are more variable than right legs. In order to eliminate this 



* Galton 's method is explained in liis paper in the Proc. Roy. Soc, Vol. XLV. 

 p. 135, 1889. 



