Milk Yield of Jersey Cattle. 127 



heterogeneous including data from British Holstein and Short- 

 Horns. In all events, the correlations confirm our general con- 

 clusions that milk production of one lactation is quite closely 

 correlated with that of the other lactations. This reasoning 

 transferred to the individuals of the race of dairy cattle, appears 

 to prove beyond any shade of doubt that the individuals of the 

 race are innately differentiated as regards the capacity for milk 

 production. 



Of the quantitative data on other species of practical inter- 

 est perhaps the most complete is that of Harris and Blakeslee on 

 the White Leghorn. In this work they determine the correla- 

 tions between the monthly egg production and the other eleven 

 months production of the same bird. The correlation for these 

 monthly ovulations w r ith the other eleven months ovulation take 

 values ranging from +0.240;+: .033 to +0.573^1.023. The range 

 is there quite similar to those obtained in this study of milk se- 

 cretion although lower in value. The knowledge of these sets 

 of constants gives criteria for the fairly accurate prediction of 

 the records that may be expected as a subsequent date in the 

 life of these two extremely important economic species. 



From these correlation coefficients in table 2 it is possible 

 to form the straight line prediction of the milk yield of any age 

 from the yield of any other age. As emphasized repeatedly in 

 the various sections of this paper dealing with the separate sub- 

 jects, the predictions may be criticized on the following grounds ; 

 that the mean milk yield rises with age in a line described by a 

 logarithmic function ; that the standard deviation of this milk 

 yield rises in line described by a cubic parabola; and that the 

 values of the correlation coefficients differ from the values of 

 the correlation ratios by 2.42 times the probable error of r; 2 — r 2 . 

 Such criticisms are recognized and admitted at once. It is be- 

 lieved that even admitting these there are a number of impor- 

 tant points which may be elucidated by these equations. 



The general equation for these regressions is given by 



<r Y o- Y 



Y = (Mean Y — Mean y) + rYy y 



