72 MAINE AGRICULTURAI, i;XPERIMENT STATION. 1909. 



to the egg production during any portion of the year. The 

 coefficients are, with a single exception, not sensibly different 

 from zero. In all three of the cases where the correlation is 

 determined from the tables in which the mothers are weighted 

 with their fecundity the correlation coefficients are not sensibly 

 larger than their probable errors. When the mother's individual 

 performance is correlated with the daughter's average perform- 

 ance the coefficient also is sensibly equal to zero in the case of 

 the spring production. 



2. In three out of the five cases the sign of the correlation 

 coefficient is negative. In one of these cases, namely that in 

 which the mother's individual winter performance is correlated 

 with her daughters' average winter performance, the negative 

 coefficient is large enough to be sensible in comparison with its 

 probable error. The meaning of a negative coefficient of corre- 

 lation is, illustrating from this particular experiment, that the 

 mother with a record above the average for mothers in general, 

 produced daughters whose records were on the average below 

 those for daughters in general. In other words, the negative 

 coefficient indicates that the relatively high producing mother 

 produced the relatively low producing daughter. As has been 

 said, however, only one of these negative coefficients can, when 

 taken alone, be considered to be significant in comparison with 

 its probable error. 



3. Turning to the variation results it is seen that whether 

 the variation is measured absolutely or relatively, the daughters, 

 as would be expected, are much more variable than the mothers. 

 In the case of the coefficient of variation, the daughters are 

 from 3 to 4 times as variable as the mothers in respect to the 

 egg production of corresponding periods of the year. The 

 daughters show a greater variation than the mothers regardless 

 of whether the mothers are weighted with their fecundity. 

 The daughters' variability clearly must approach that of the gen- 

 eral population. Its actual relation to the variability of the 

 general population will be shown in the next section. 



4. The data obtained from this experiment give no evidence 

 whatever that there is any appreciable correlation between 



