Appendix E 



415 



many ears by chemical methods. Or if, after a relation had 

 been established, we could safely choose the longest or tallest 

 bean plants right in the field and know that they will bear the 

 greatest number of pods, it would be of great advantage to the 

 breeder. 



Now, an exact determination of the degree of correlation can 

 be obtained by the biometrical method. Let us follow the pro- 

 cess step by step, using our bean data. 



First of all, we take our data for the two characters for which 

 we wish to find the correlation, length of plant, and number of 

 pods. 



Our original observations will be somewhat as follows : 



In finding the constants mean, standard deviation, etc., for 

 each of these characters, the observations for length and those 

 for number of pods were distributed in separate tables. Now, 

 however, we distribute both sets of observations on one table, 

 in what are known as arrays of a correlation table. (See Table 

 1.) For example, the first observation tabulated above would 

 fall in the vertical array 25-34, as regards length, and in the 

 25-34 horizontal array, as regards number of pods. The second 

 observation would fall in the 5th column (vertical array 45-54) 

 and in the third row (horizontal array 25-34). 



Thus each vertical array would be a frequency distribution 

 of length of plant with respect to number of pods, and each 



