FERTILITY AND HATCHING OF EGGS. 147 



2. With reference to the percentage of fertile eggs hatched 

 the case appears to be somewhat different. Here, as is shown in 

 Fig. 15, there is a distinct tendency for the zigzag hne of dangh- 

 ter averages to run more or less parallel to the fathers' line. 

 In other words, there is some indication of a distinct tendency 

 for the daughters of a male whose average pen record for 

 hatching quality of eggs is high to show the same characteristic 

 themselves. Too much stress must not be laid upon the result, 

 however, because of the small number of males included in the 

 statistics. 



3. These results, so far as they go, accord with those previ- 

 ously obtained in so far as that there appears to be a difference 

 in the behavior of the character "fertility of eggs" as distin- 

 guished from "hatching quality." Fertility seems to be much 

 more a matter of external factors than hatching quality, which 

 appears to be very largely determined by innate constitutional 

 characters. This point will be more fully discussed farther on. 



We may now look at the c^uestion of the inheritance of these 

 characters under discussion in still another way, namely from 

 the standpoint of collateral inheritance. Let us turn to an 

 examination of the so-called "fraternal" correlations respecting 

 fertility and hatching quality of eggs. The question here is 

 this : if one sister in a family has a percentage of fertility above 

 the average will her other sisters (i. e., birds of the same 

 family) tend to have fertility records above the average and 

 vice versa? And similarly if one sister shows an unusually 

 high percentage of fertile eggs hatched will the other sisters 

 of the same family in general show hatching records above the 

 average, and vice versa? It is plain that if sisters are in gen- 

 eral alike in respect to either of these characters fertility or 

 hatching power of eggs, it will indicate that to that extent these 

 characters are inherited. 



In order to determine whether there is on the average a closer 

 resemblance between sisters in respect to these characters than 

 exists between individuals taken at random it is necessary once 

 more to appeal to the method of correlation. In forming the 

 correlation tables in this case, however, it is necessary to adopt 

 a slightly different method than that wdiich was used in the case 

 of mother and daughter records. When the correlation 

 between mother and daughter is determined we are dealing with 

 two entirely separate classes of individuals belonging to differ- 



