84 Journal of Agricultural Research voi.xvi. No. 3 



butter fat is 7. 17 per cent, contained in the milk of the ewe, closely followed 

 by that of the Indian buffalo and the sow. The milk of the dairy cow 

 is about intermediate between these two extremes. The colostrum is 

 lower in its fat content than is either of the normal milks of the same 

 species. The solids-not-fat content varies from 8.61 per cent for the cow 

 to 1 1. 91 per cent for the normal milk of the species included in the table. 

 It reaches its highest value in the colostrum, where the cow's milk in- 

 cludes as much as 20.58 per cent. The same general association is also 

 seen in the milk of the different species that are present in the milk of 

 the different breeds — that is, the milk of species containing a low per- 

 centage of fat contains proportionately more solids-not-fat than does the 

 species which contains the higher percentage of fat. The species con- 

 taining a low percentage of fat contains less actual solids-not-fat per 

 hundred pounds of milk than does the species containing a higher per- 

 centage of fat. 



A survey of Table I shows that the most variable character in the 

 Holstein-Friesians, is the age included in the tests. This has a very high 

 coefficient of variation, the highest shown by any of the measured char- 

 acters, emphasizing the need of a proper age correction when the amount 

 of milk is to be studied for its hereditary behavior. 



The next highest variation coefBcient is that for milk (24.3). This is 

 closely followed by the variation coefficients for butter fat, 21.0; solids- 

 not-fat, 19.9; and total solids, 19.4 These last four coefficients may be 

 considered as dependent variables and owe part of their variation to 

 variations in other characters. Thus, the large constant of variation 

 for the amount of milk is due in some part to the age differences in the 

 animals included, for, as has already been shown, milk production rises 

 in a logarithmic curve with increasing age. Again, as will be shown 

 later, the relation of the milk constituents to the amount of milk is so 

 close that a large part of their variation may be explained by variations 

 in the amount of milk. As a matter of fact, these data are not needed 

 here, for when the coefficients of variation for the percentages of butter 

 fat, solids-not-fat, and total solids are considered, it is seen that the 

 coefiticients are reduced to 9.2, 3.9, and 4.5, respectively, or coefficients 

 which compare rather favorably with physical variables. The high 

 coef^cients of the milk solids are thus shown to be due to the variations 

 in amount of milk and not to variations in percentage contents of these 

 constituents. 



FACTORS AFFECTING THE COMPOSITION OF MILK 



Already some work has been done on factors affecting the composition 

 of milk by Wilson {37) and by Pearson {21-23). In his studies Wilson 

 attempts to show that the percentage of butter fat is not dependent on 

 the amount of milk. The methods used to draw this conclusion are, 

 according to Pearson, open to criticism on the following grounds : The tail 

 frequencies are clubbed together so that the real correlation can not be 



