560 



Scientific Proceedings, Royal Dublin Society. 



month, both included, neither curve diverges more than six per cent, from 

 the average. The actual divergencies for the Ayrshire cows are : — 



At the beginning of the 3rd month, 



4th „ 



5th „ 

 6th „ 

 7th „ 

 8th „ 



Per cent. 



5 



3-5 



1-3 



2-9 



5-4 



cent., and never 

 iod it represents. 



Now, the four-day test is usually not more than 5 per 

 more than 10 per cent., outside the average for the short per 

 There is, however, the possibility of the divergence being in the same direc- 

 tion as that recorded above ; and in that case the error is accumulated. 

 That is to say, when the four-day test is applied during a period extending 

 from the beginning of the third to the beginning of the eighth month, there 

 is the possibility of a 10 per cent, error in the majority of eases, and of a 

 maximum of 15 per cent, error in some cases. The possible error can be 

 reduced by limiting the time of sampling to a shorter period, and by confining 

 it to the fifth and sixth months of the lactation we ensure a result generally 

 not more than 5 per cent., and never more than 10 per cent., outside the 

 average for the entire period of milking. 



The returns of weekly tests from twenty cows during normal lactation 

 periods have been tabulated from data kindly supplied by Sir Gilbert Greenall 

 from his dairy herd at Kilmallock, where the quality of each cow's milk is 

 tested one day a week. Single-day tests are always liable to occasional severe 

 fluctuations such as were found in the milk of the cows at Glasnevin, which 

 also appear in the Kilmallock returns. When these are allowed for, the 

 Kilmallock data show that, when the lactation period is of normal length. — 

 that is, when the next calf is expected within eleven to thirteen months after its 

 predecessor — tests may be relied on from samples taken at any time from the 

 third to the eighth month (inclusive) after calving, the error limit appearing 

 less than in the returns from Hogstrom and Hansson. In a prolonged 

 . lactation the period of sampling may be extended ; but in any general 

 scheme it is well to adhere to the general rule. Postponing the date of 

 sampling might in cases be overdone. 



It is, therefore, suggested that, to ascertain the percentage of fat in a 

 cow's milk over a normal lactation at the smallest cost and with approximate 



