304 MAINE AGRICULTURAL EXPERIMENT STATION. I913. 



This was not clue to a progressively diminishing flow. The 

 greater part of this increase in mean fat percentage occurred 

 between July and October. This would indicate that pasture 

 conditions (not necessarily feed alone) were a significant factor 

 in producing the result. 



5. The greatest absolute mean fat production per day was 

 in May. There is, however, no significant dift'erence in this 

 respect between May, July and October. The influence of 

 pasture conditions in stimulating the flow seems clear here. 



6. Turning now to the variation constants, we note first 

 that, contrary to common opinion, the percentage fat conteiit 

 of the mixed milk of a large herd exhibits a considerable vari- 

 ation from day to day. The standard deviations and coefficients 

 of variation for this character in every case are more than 10 

 tiines as large as their probable errors.' Certainly they cannot be 

 considered insignificant. 



7. The milk of this herd was most variable in percentage 

 fat content in March and October and least variable in May and 

 July. But none of the dififerences are significant in comparison 

 with their probable errors. In general, it appears from these 

 data that the degree or amount of daily variation in the per- 

 centage fat content of mixed milk is not significantly affected by 

 such changes in feed and other conditions as are here involved. 



8. The absolute amount of fat produced per day is roughly 

 about twice as variable (compare coefficients of variation) 

 relatively as is the percentage fat content of the milk. This 

 result is of particular interest in relation to the rather wide- 

 spread view that the variations in fat percentage of milk are 

 to be accounted for in the main by fluctuations in the water 

 content. It should be remembered that the result here set 

 forth is for the mixed milk of a whole herd. 



9. The relative variation in absolute fat produced, as meas- 

 ured by the coefficient, decreases steadily from ]\Iarch on 

 through October. The amount of this decrease is, however, 

 rather small, and in the extreme case is not certainly signifi- 

 cant in comparison with its probable error. 



10. There is no indication that the milk of any particular 

 milking of the day is, either absolutely or relatively, signifi- 

 cantly more variable in percentage fat content than the milk 



