122 EDWIN CARLETON MACDOWELL 



sons and daughters; nor are low points in the parents' curv'e 

 accompanied by low points in the curves of their children. 

 When the parallelism of the curves of the sons and daughters 

 is noted, the lack of such parallelism between the curves of the 

 parents and children is very striking. 



High points 



Parents' curve Generations 11, 16, 23, 31, 49 



Children's curve Generations 13, 30, 41, 45 



Low points 



Parents' curve Generations 13, 17, 27, 47 



Children's curve Generations 11, 26, 35, 42 



There is a sort of similarity between the curves of the parents 

 and children that results from the fact that the flies selected 

 as parents for one generation are averaged with the children in 

 the preceding generation. If the mean of a generation is very 

 low, there will be found few high grade flies to select as parents 

 for the next generation, and accordingly the means of the flies 

 selected will be low. In this way low points in the children's 

 curves will be followed by low points in the parents' curve in 

 the next generation; and in like manner high points in the 

 children's curves will be followed in the next generation by high 

 points in the parents' curve. In generation 1 1 the means of the 

 offspring are low, in generation 12 the parents are low; in gen- 

 eration 26 the offspring are low, in generation 27 the parents are 

 low; in generation 30 the offspring are high, the parents are 

 high in generation 31; Such relationships, of course have no 

 significance in the question of the genetic differences between 

 high and low grades of extra bristles. However, the following 

 facts do have significance for this question: there is a difference 

 of nearly 3| bristles between the means of the parents in gen- 

 erations 27 and 28, while their offspring differ by less than one- 

 fifth of a bristle; in generations 28, 29 and 30 the parents' means 

 are nearly alike, while the means of the offspring rise If bristles. 



A similar comparison is offered in figure 2A. In this figure the 

 means of the sons in individual families is compared with the 



