236 



Bulletin 397 



viduals producing between 400 and 449 pounds, the other of individuals 

 producing from 450 to 499 pounds. Thus the measure of variation 

 assuming a 50-pound range as representing equal production was obtained. 

 Similarly, the cows were further divided into four classes, with a range 

 of 25 pounds; and finally into eight classes, to obtain an even smaller 

 range. The results are given in table 12. • 



A slight decrease in the size of the coefficients as the limits are narrowed 

 is shown in table 12. However, the decrease is not significant when the 

 probable errors are considered. The difference between the values for 

 the widest and the narrowest range is 1.97 ± 1.22. The odds are only 

 2 to I that this difference is significant. Thus it appears that there is 

 no object in using narrower limits than 100 pounds in yields for a two- 

 weeks period in selecting individuals of equal production. Accordingly 

 the 50-pound range used for the studies in this bulletin must be sufficiently 

 narrow to secure the object desired. 



Group variation 



The variation of groups of equal production during the trial period 

 is shown in table 13. No selection of individuals was made, the only 

 consideration being to make the production of the groups equal within 

 the 100-pound limits. 



TABLE 13. Variation in Groups 



Group production, two- weeks period 

 (pounds) 



2,700-2,799 

 3,100-3,199 



Total . . . 

 Average , 



Number 

 of groups 



18 

 16 



34 



Number 



of cows 



per group 



Coefficient of 

 variability 



3.86 ± 0.43 

 5 25 ±0.63 



4.51 ± 0.37 



It is shown in this table that the variation in groups of six cows in 

 different stages of lactation, under the conditions of selection obtaining 

 in this study, is no greater than the variation in a similar group of the 

 same stage of lactation. The probable reasons why the variation in 

 stage of lactation does not result in a larger coefficient under the con- 

 ditions in question, have been discussed in connection with table 10. 



Effect of including aborting cows.-^ In the statistical studies reported 

 in this bulletin, the importance of numbers was felt to be so great that 

 not all the records of animals that had aborted were excluded. Inasmuch 



346 



