546 TRANSMISSION 



Different individuals of the same ancestry inherit differently, 

 and in general the behavior of characters in transmission sug- 

 gests that they are in some way made up of combinations, so 

 that a high degree of variability is inevitable, even with the 

 same elements ; as, for instance, a great variety of color effects 

 can be produced with the same three primaries, red, blue, and 

 yellow. 



Some characters blend and others are mutually exclusive, 

 each tending to preserve its identity. On this account, as well 

 as* from other causes, such as relative fertility, races often 

 exhibit distinct polymorphism. Inheritance is not so much con- 

 nected with sex as is popularly supposed. Characters often 

 do not develop until late in life. This is not to be regarded as 

 belated inheritance but as belated development. 



The only proper way to study the principles of heredity is by 

 statistical methods, using groups instead of single individuals, 

 from which no general conclusions can be safely drawn. 



The regression table brings out clearly the fact that like 

 parents beget unlike offspring ; that, in general, the offspring 

 is more mediocre than the parent, but that for selected offspring 

 the ancestry is comparatively mediocre ; that the coefficient of 

 heredity between the nearest relatives is seldom above 0.50; 

 that the mean of the offspring is not necessarily the same as the 

 mean of the parent ; that the means of a race are its most fertile 

 portions ; that, in general, a few offspring exceed the previous 

 limits of the race, that is, progress away from the type if 

 favored by selection ; that exceptional individuals may arise 

 either from exceptional or from mediocre parentage ; and that 

 successive offspring from the same parents are not identical. 



Nothing is clearer than that the inevitable consequences of 

 bisexual reproduction and of the manner of growth by the halv- 

 ing of the cell contents is to insure that character combinations 

 effected in this manner are brought together in definite mathe- 

 matical proportions not far from those expressed jn the expan- 

 sion of a binomial. This is the real foundation of Mendel's law, 

 for characters that do not blend, and it also expresses the rela- 

 tive proportions of characters that do blend. 



The statistical methods of study enable us to develop the law 

 of ancestral heredity, which agrees closely with experimental 



