28 REPORT OF THE AGRICULTURAL RESEARCH 



(Hi) In India it is usually difficult to persuade a cow 

 to yield milk to the hand without the calf sucking at the 

 same time. At Pusa this was found to be regularly so. But 

 obviously if the calf takes an unknown quantity of the milk, 

 the total yield cannot be correctly ascertained ; and without 

 the adoption of special precautions this factor affects the 

 percentage of fat also. The practice at Pusa with this herd 

 is to allow the calf to take all the milk of one side of the 

 udder whilst the other half is hand drawn, the idea being 

 to ensure that the calves have a liberal allowance. It 

 remained therefore to ascertain whether this practice is 

 accompanied by any systematic errors, e.g., is the one half 

 udder as thoroughly stripped by hand as the other half is 

 by the calf \ The system which we adopted for these tests 

 was to allow the calf to take the milk from one half udder, 

 say the right, whilst the left half was hand milked, for 

 two milkings; then for the following two milkings the calf 

 was given the left half, whilst the right half udder was 

 hand milked; and this sequence was followed day by day. 

 During the first half of the period, i.e., approximately for 

 one month, the change from right to left side was made at 

 the morning milking, whilst during the second half period, 

 it was made at the evening milking. By the adoption of 

 this system of milking it was evident that if the calf 

 obtained more milk than the man (i.e., if the hand milking 

 were the less perfect) or if more milk were secreted during 

 the night or day time, systematic differences of yield would 

 be perceptible in the records which could be traced to one of 

 these two causes. The data showed no such systematic 

 differences and it became evident that the calf and man 

 were equally efficient in stripping the udder. The source 

 of error due to not being able to milk these cows entirely 

 by hand, was thus avoided. 



An examination of the data, both of yield and of per- 

 centage of fat, showed that the probable error of individual 

 tests was unusually small. The probable error of yield for 

 individual milkings was only + -27, and for percentage of 

 fat it was only + -29, which are small when compared with 



