Refinement of Feeding Experiments for Milk Production 231 



TABLE 7. Effect of Size of Group on Variability 



Group production, 



two-weeks period 



(pounds) 



Number 

 of groups 



Number 



of cows 



per group 



Coefficient of variability 



Observed 

 value 



Calculated 

 value 



4,400-4,499 

 3,200-3,299 

 1,800-1,899 

 2,100-2,199 

 2,300-2,399 



23 

 33 



49 



4 . 33 ± o . 43 

 5.09 ± 0.42 



5.39 ±0.37 



4.19 



4.84 



5-94 



A comparison of the observed and the calculated coefficients of varia- 

 bility for groups of eight, six, and four individuals, given in table 7, shows 

 close agreement between the two values in each case. In two cases the 

 calculated values are within the limits set by the probable errors above 

 or below the observed coefficients. In the third case the variation is not 

 significant. This indicates that the data are trustworthy and that in a 

 similar way the probable coefficients of variability for groups of different 

 sizes may be computed from the coefficients of variability of individuals. 



Selection of groups on the basis of production during the preceding lactation 

 period. — It has been stated that the selection of groups on the basis of 

 yield during the preceding lactation period must be considered a poor 

 method. Inasmuch as several investigators have selected groups on that 

 basis, it seemed desirable to make some study of the method. Accord- 

 ingly, groups of six cows each were formed on the basis of their productive 

 capacity as shown during the preceding lactation period. The results 

 of this study are shown in table 8 : 



TABLE 8. Variation of Groups Selected on the Basis of Preceding Lactation 



Records 



Group production during preceding lactation period 



(pounds) 



60,000-61,999. 



Number 

 of groups 



Coefficient of 

 variability 



9. II ± 1.05 



A comparison of the coefficient shown in table 8 with the value 5.09 ± 

 0.42, given in table 2, indicates that when groups of cows are selected 

 on the basis of their yield during the preceding lactation period the coeffi- 

 cient of variability is nearly double that for groups selected on the basis 

 of a two-weeks preliminary period in the same lactation period. It is 

 thus proved that the latter is by far the better method. 



341 



