98 Journal of Agricultural Research voi. xvi, no. 3 



From the practical side of increasing the soHds content of cow's milk, 

 the importance of the correlation between the butter fat and the solids- 

 not-fat should be pointed out. This high correlation between these varia- 

 bles allows us to use the determination of the percentage content of one 

 as a means of predicting what the content of the other will be. Thus, 

 butter-fat content is easily determined by almost any one familiar with the 

 Babcock test; but solids-not-fat are not so easily determined nor so fre- 

 quently recorded. We may, in trying to improve the solid content of 

 milk select as breeders those cows which test well with the Babcock appa- 

 ratus, and at the same time improve the solids-not-fat content of the 



milk. 



SUMMARY 



This paper is the fourth of a series of studies on milk now being con- 

 ducted in the Biological Laboratory of the Maine Agricultural Experi- 

 ment Station. The data for this study are taken from the semioflficial 

 year record of the pure-bred Holstein-Friesian cows, compiled and super- 

 vised by the Holstein-Friesian Association. 



(i) The means, standard deviations, and coefficients of variation are 

 given for these year records. The mean annual production of these ani- 

 mals was 15,417 pounds of milk, 528 pounds of butter fat, 1,303 pounds 

 of solids-not-fat at a mean age of four years. The standard deviations are 

 3,742 pounds of milk, 134 pounds of butter fat, 260 pounds of solids-not- 

 fat, and two years. The coefficients of variations are, respectively, 24, 

 25, 20, and 50 per cent. 



(2) Comparison is made of Holstein-Freisian milk with the milk of 

 other breeds and other species. 



(3) Correlations are presented between the variables butter-fat percen- 

 tage and amount of milk produced, butter-fat percentage and age at 

 test, solids-not-fat percentage and amount of milk, and solids-not-fat and 

 age at test. These correlations lead to the following conclusions: (a) 

 As the amount of milk given by the cows in this test increases, the per- 

 centage composition of butter fat decreases. The amount of this de- 

 crease is highly significant, measured statistically. Considered practi- 

 cally, this fall in butter-fat content would not be easily detected in small 

 samples, (b) The correlation between the age at test and butter fat is 

 not significant, (c) The correlation between the amount of milk pro- 

 duced and the percentage of solids-not-fat is not significant; or, put in 

 another way, the quantity of milk produced for one year is independent 

 of the concentration of the solids-not-fat. This, from a genetic view- 

 point, means that the hereditary factors for high or low milk production 

 are separate and distinct from those causing a high percentage of solids- 

 not-fat. (d) The correlation of age at test and solids-not-fat is — 0.2 191 ± 

 0.0351 — that is, as the age of a cow increases, the solids-not-fat percen- 

 tage of the milk decreases, (e) The constants for the linearity of re- 

 gression are given. They show the regressions to all be linear, (f) 



