14 
THE AGRICULTURAL EXPERIMENT STATION. 
Having found, then, that the fluctuations in the per 
pent, of butter fat are reduced to a minimum in most 
cases after a change in a ration has been instituted 
seventy-two hours, we add twenty-four hours for safety, 
and include in our averages only analyses after ninety- 
six hours, or after twelve feeds have been consumed by 
the cows, except in starting the experiments. As some 
cows were moved, they were given from twenty-four to 
forty-eight hours still more, to become familiar with their 
new surroundings and feed before the analyses of their 
milk, were taken into the averages. Analyses, however, 
were made from which to observe changes. Occasionally 
•one has been thrown out, when by accident or other cause 
it is known to be wrong. 
METHOD OF ANALYSIS EMPLOYED. 
The most careful records of milk yield, feed, water, 
and animal weight, have been kept throughout the 
experiment, and the milk of each cow has been tested 
•daily for its fat per cent, by the Babcock method. This 
tester was adopted, as we consider it the most accurate 
and speedy, and less subject to errors. 
The analyses with the Babcock tester were n 
■duplicated by the Station Chemist with the gravimetric 
method. The difference in the results of these two 
methods is greater than was expected, but can be 
accounted for in the fact that the samples for the gravi¬ 
metric examination of each day’s milking (combined 
morning and evening) were drawn off the morning after, 
while those for the Babcock were pipetted from the com¬ 
bined milk as soon as the evening’s milking was over, 
and were placed in the test bottles ready for analysis. 
The variation being so great, not only in comparison 
with the Babcock, but frecpiently as compared with the 
same cow’s milk the day before, by the same method, 
