54 Maine Agricultural Experiment Station. 1920. 



known for other similar data. This large size in the coefficients 

 is brought out by table 2. 



Of the substance studied annual egg production is the most 

 variable. The coefficient of variation for milk production cor- 

 responds remarkably well considering the diversity of the 

 sources from which they are taken. They all go to show that 

 milk production varies around a mean of 25 or is about 9 per 

 cent less variable than egg production. This difference is sig- 

 nificant. The mechanism of the secretion of the sum of the egg 

 parts to form the egg is shown to be more variable than that 

 for the secretion of the milk parts to form the milk. Taking 

 this reasoning back to its ultimate source, such a significant 

 difference shows that the secretory cells of the mammary glands 

 work with greater precision than do the cells of the oviduct. 

 Such a difference in the variation of action of the two sets of 

 cells would seem to indicate a greater approach to perfection, 

 in a mechanical sense, in the cells of the udder than is true of 

 the cells of the oviduct. 



The skewness of the two sets of data furnish another in- 

 teresting contrast. The skewness of annual egg production for 

 Barred Plymouth Rock hens is — 0.205, whereas the skewness 

 for Jersey milk production is +0.315. Xot only is the sign 

 different but the actual amount is different. This cannot be 

 explained on the basis of any selection for high producers that 

 may have taken place as both sets of data are about equally sub- 

 ject to such selection. The data for the successive age groups 

 all goes to show that where the distributions for milk production 

 are skew they are all plus. This lends further strength to the 

 belief that the distribution for the yearly production of the two 

 sets of glands is skew in opposite directions. One thing is com- 

 mon in the two cases, the skewness in each is small in amount. 

 This is of especial importance as it shows the approach to the 

 value where the typical Gaussian curve of error may describe 

 these functions. 



This question is of great importance in considering the 

 milking records of advanced registry animals where it is neces- 

 sary to form an opinion of the capacities of a breed from the 

 milking abilities of a selected sample where the selection is not 

 at random but removes those cows producing under a certain 

 amount. 



