210 EVOLUTION, GENETICS, AND EUGENICS 



the mean annual rainfall, and the general character and direction of air 

 movements remain essentially the same for very long periods. Such 

 constancy of environment could hardly fail to exercise a standardizing 

 effect upon animal and plant communities, and thus aid in maintaining 

 constancy in the expression of racial characters. 



Persistence and diversity mechanisms contrasted. — It is custom- 

 ary in courses in genetics to introduce the study of heredity by present- 

 ing an outline of Mendelian heredity. This seems to us a mistake, for 

 Mendelian heredity is really not essentially a persistence factor, but 

 rather a most effective diversity factor. It breaks up constancy of 

 combinations of unit characters and promotes multiplicity of recom- 

 binations of such characters. For this reason we shall begin our dis- 

 cussion of the persistence factors in evolution with the study of heredity 

 in pure lines, in which the persistence mechanism is free to operate 

 without being complicated by the diversity mechanism. Since bio- 

 metric methods are necessary for the study of variation and heredity 

 in pure lines, it is necessary to introduce a brief statement about these 

 methods. 



A short lesson in biometry. — When character differences are either 

 qualitative or are sharply defined, they are easily handled by Mendeli- 

 an methods. If for example, all individuals are either black or white, 

 tall or short, heavy or light, and no gradations occur between the two 

 alternatives, it is easy to follow the distinct types through successive 

 generations. If, on the other hand, there occur all gradations between 

 black and white, all gradations between tall and short, and all grada- 

 tions between heavy and light, it is no longer possible to classify each 

 individual in some particular category. When this is the case the only 

 possible method of finding out how much is inherited, and how much 

 is not, is to study whole generations of progeny as units, and to deter- 

 mine what is the characteristic of one whole generation as compared 

 with the next whole generation. Such a comparison requires statistical 

 methods. 



Suppose, for example, we want to find out whether the size of bean 

 seeds is hereditary or merely environmental, we shall have to measure 

 the seeds of the parent and compare them with those of the offspring. 

 There would be nothing gained by comparing a selected seed of the 

 parent with one of the offspring. We must compare the total of one 

 with the total of the other. It is necessary, then, to find a method or 

 methods of comparing the parent condition as a whole with the off- 

 spring condition as a whole. Some simple method must be devised 



