E. Robertson 81 



The difficulties of breeding for milk arc best illustrated by the fact 

 that no one has made any real study of the question and by the nature 

 of the " points " that are supposed to indicate a good milker. 



Beef cattle have been wonderfully developed within the last century 

 but, if anything, the milk yields of cattle have declined ; they have 

 certainly not improved. 



It does not seem to have struck anyone that while the eye, and the 

 " touch " or " feel " of the skin, were quite good guides as to the flesh 

 and fat that an animal developed, there was a better standard for dairy 

 cattle in the milk yield. But in order to utilise the yield as a standard 

 to which to breed, it is necessary to weigh the whole yield of every 

 cow and her heifers for some years. This means no little trouble and 

 attention. Moreover, it means steady trouble twice a day regularly. 



The milk record kept for these experiments has been maintained 

 without a single break since the beginning of 1904. Not one single 

 milking has escaped record. 



The butterfat estimations have been as regular since 1905. At first 

 tests were made for every milking, but as this was found to be too 

 expensive both in money and time, a sample is taken at every milking 

 for each cow and the combined sample is tested regularly every Friday 

 afternoon ; the Gerbers method being used. 



This ensures as correct a record as is possible of each cow's perform- 

 ance in the year. 



Nevertheless, the best kept record suffers from certain imperfections 

 that must be understood and, if possible, corrected. The amount of 

 milk given by any cow follows two curves, the first a seasonal one, due 

 to the time of year at which she calves, and the second a physiological 

 one due to the date of service. The same cow calving in September 

 will have a different seasonal curve from what she would have Avere 

 she calved in April. And again the curve of yield will be different 

 Avhen she is put in calf in the second month after dropping a calf to 

 what it is when put in calf in the third or any subsequent month. 

 How far these two curves contradict each other has not been fully 

 worked out. I have tried to do so, but once a cow has started work in 

 any given month, one does all one can to keep her calving dates in that 

 month so that real control experiments are rare. So far, I find that 

 they vary a good deal while following a general law more or less 

 closely. 



A curve beginning in any of the late autumn months, say October, 

 never rises as high as one that starts in, say March, but it keeps at a 



Journ. of Gen. xi 6 



