300 



Journal of Agricultural Research 



Vol. XVII. No. 6 



2. From comparison of the results here given with those of Vigor {28) 

 it is seen that while, the general, the agreement is fairly close, there are 

 some rather striking differences. Taking weighted means from Tables 

 III and IV, we see that the mean weekly yield is slightly lower, and the 

 mean fat percentage slightly higher, in the whole group than in the 

 Fenwick district data alone. The differences in the means, however, 

 are small and probably of no significance. When we turn to variation 

 as measured by standard deviations, there is a striking difference in 

 weekly yield. The weighted mean standard deviation for the whole 

 group is 2.806 gallons, while ^^igor finds for the Fenwick district alone 

 4.0704 gallons. This is a large and statistically significant difference. 

 In fat percentage the standard deviations are practically alike for the 

 two sets of data, our weighted mean value being 0.330 and Vigor's 0.3229 



3. The explanation for the difference in variability in weekly yield 

 between our figures and Vigor's is not far to seek. It lies mainly in the 

 fact that Vigor has dealt with cows of all ages lumped together, while 

 in the present paper each year of age is dealt with separately. Natur- 

 ally when dealing with a character which changes with age so extensively 

 as does milk yield, as has recently been discussed by Pearl (12), the 

 variation exhibited will be markedly increased if animals of all ages are 

 lumped together. In order to determine how much of the difference in 

 variation was due to this cause and how much to other factors Table V 

 has been prepared. This table gives the distribution for weekly yield 

 obtained by adding together all of the "combined" distributions for the 

 several years, as set forth in Table I. 



Table V. — Distribution for weekly yield combi?ted for all ages and for the whole area to 

 compare with Vigor's data for the Fenwick district alone 



Mean = i5.99i±o.o27. Standard deviation=3.329±o.oi4. Coefficient of variation= 20.81 6 ±0.088. 



4. The difference between this value of the standard deviation and 

 Vigor's, while reduced, is still sensible. It amounts to about 0.742 ± 

 0.082. This remains to be accounted for. We find it difficult to suppose 

 that the selection of relatively long lactations in the present data can be 

 the cause, since Vigor himself has shown that there is no sensible cor- 

 relation between either mean weekly yield or fat content and duration 

 of lactation. We are much m.ore inclined to the view, especially in the 

 light of unpublished results on milk production in other breeds of cattle, 

 that the Fenwick district returns give somewhat abnormal values in the 



