BLASTOGENIC VARIATIONS. 131 



mid-parents, viz., 68 inches, was very nearly that of the 

 whole population, and so obviously the filial height 

 could undergo no regression, but would be practically 

 the same value. The values in this table thus illus- 

 trate the existence of regression, but they also indicate 

 that the offspring produced by these mid-parents are, as 

 a whole, no less variable than they themselves are. The 

 offspring are, in fact, more variable, as a mid-parental 

 stature, being the mean of two parental statures, is ob- 

 viously, on an average, less variable than either stature 

 individually. Thus the mid-parents vary roughly be- 

 tween about 74 and 64 inches, but the children between 

 75 and 62 inches. This table therefore teaches us that 

 though the children are, on an average, more mediocre 

 than their parents, yet the general variability of the 

 race is not diminished. The reason why the variability 

 remains undiminished may be seen by studying the com- 

 ponents of the vertical columns of the table. For ex- 

 ample, with reference to children 71 to 73 inches in 

 height, we see that mid-parents of 71 inches and up- 

 wards contribute proportionately more of these tall 

 children than do any other mid-parents, but still mid- 

 parents of 69 to 71 inches contribute (proportionately) 

 a good many, and parents of 67 to 69 inches no small 

 number. Even mid-parents of 65 to 67 inches con- 

 tribute a very minute number of these children, who 

 are thus no less than 6 inches taller than their parents. 

 By these several contributions, therefore, the number 

 of tall and similarly of other children is kept up to 

 the same level in each generation. One may accord- 

 ingly sum up the contents of this table as follows: Tall 

 parents have many tall children, a moderate number 



