436 Journal of Agricultural Research voi. vi, no. u 



skilled breeders are able to detect deviation from type even in the early 

 stages of development and the practice of roguing can proceed with cer- 

 tainty. It appears that when a cotton plant deviates from type it 

 deviates in a more or less definite way and in many particulars, or, in 

 other words, there are a number of coherent or correlated characters. 



It seemed desirable to determine whether the difficulty in recognizing 

 types in maize is due to a lack of familiarity with the plants or whether 

 there is a fundamental difference between the heredity of maize and that, 

 for example, of cotton. 



In the seed characters of maize a definite correlation has been found 

 between the color of the aleurone and the texture of the endosperm (Col- 

 lins and Kempton, 1913). Correlations have also been noted between the 

 color of the silk and the color of the anthers (Webber, 1906), and between 

 the color of the seed and the color of the cob, dwarfness and broad leaves, 

 and between stamens in the ear and club-shaped tassels (Emerson, 191 1). 

 There was, therefore, abundant reason for suspecting that the difficulty 

 of recognizing types among maize plants might be due to a lack of suffi- 

 cient discrimination, and it was with the idea that correlations were the 

 rule rather than the exception that the present experiment was under- 

 taken. Contrary to expectation, the results give evidence that for the 

 varieties and characters studied there is almost a complete absence of 

 genetic correlations. 



CLASSIFICATION OF CORRELATIONS 



Correlations may be classified in a great variety of ways and with almost 

 any degree of refinement. As with any classification of organic activities, 

 no particular grouping can be made to serve all purposes, for it is necessary 

 to divide the subject in different planes. 



For purposes of the present discussion correlations, or the mutual 

 relations of characters, are divided into three main groups, to which the 

 names "physical," "physiological," and "genetic" may be applied. 



Physical correlations are those in which the relation is obviously 

 causal. In many instances correlations of this kind are little more than 

 different names for the same phenomenon, or parts of the same phe- 

 nomenon, as when increased weight is correlated with increased height. 

 In physical language one of the characters would be described as a 

 function of the other. 



Physiological correlations are those where both characters are 

 the result of the same physiological tendency, as when long intemodes 

 in the main stalk are correlated with long internodes in the branches. 

 This may be looked upon as a general tendency to elongated growth 

 that is manifested in different parts of the plant. 



Genetic correlations cover the large residue of correlations, the 

 nature and causes of which are questions of controversy, but which are 



