288 Maine Agricultural Experiment Station. 1919. 



tation period, another cow producing more than 10,000 pounds 

 for her eight months' lactation period. The height of the modal 

 ordinal for the frequency polygons is greatest in the earlier and 

 later years of the cows' lives. Only one mode is evident in any 

 of the histograms. From this it appears likely that these curves 

 are unimodal curves. Too much stress should not be laid on 

 this however since certain of the distribuitons appear quite flat 

 topped. 



The analytical constants show that the eight months' milk 

 production curves are of the Pearson types I, II and IV groups. 

 The skewness of these curves is all in the positive direction and 

 rather small in amount. This small size of the skewness is of 

 considerable interest to those investigators who deal with ad- 

 vanced registry data since it would indicate that these data 

 can with but small error be fitted by normal curves to deter- 

 mine certain of their needed constants. 



The correlation coefficient between age at the commence- 

 ment of test and milk production in comparison with the cor- 

 relation ratio for the same variables shows that the regression 

 of milk production on age is far from linear. This regression 

 curve when analyzed turns out to be a logarithmic function. 

 Attention is called to the fact that this same kind of logarithmic 

 function also describes the relation of growth to age. This 

 paralleling of the two functions may indicate that a causal re- 

 lation exists between them such that increased milk production 

 chiefly depends on increase in size of the mammary gland 

 caused by growth. 



The standard deviations of milk production progressing 

 from the heifer to the aged cow are described by a cubic parab- 

 ola. This regular change of the standard deviation of milk 

 production groups shows that the curve formed by them falls 

 into the heteroscedastic, nomic class as defined by Pearson. 



Correlation tables for milk production of the different age 

 groups are presented. 



The means derived from these tables of a given age remain 

 within the limits of random sampling irrespective of the age at 

 which the same cows were subsequently tested. 



The standard deviations for the various age groups also 

 remain at the same value within the limits of sampling irre- 

 spective of the age at which a subsequent test may be made. 



