Essay on the Interpretation of Milk Records. 165 



the whole length of lactation period, their correlation with 

 other estimates of yield based on shorter periods would naturally 

 tend to be in positive ratio to the duration of those periods. 



(iv.) That the means even of the maxima seem to increase, 

 in normal cows, with the lactation period. The increase is 

 negligible for practical purposes, but nevertheless it does seem 

 to exist. The point needs further investigation, but without 

 committing oneself, it seems probable that the better cows 

 would tend on the whole to remain longer in milk, and so form 

 an undue proportion of the longer lactation groups. They 

 would often receive better food and attention, and the distance 

 between their maximum yield and zero would be greater. 

 These, however, are merely general suggestions. It is of course 

 possible that the connection is a physiological one. 



How little the maximum is influenced by the number of 

 weeks in milk is shown in the following diagram, where total 

 and maximum frequencies are plotted for both short (30 — 34 

 weeks) and long (55 — 60 weeks) lactations. 



The mean maximum of the long lactation group is in this 

 case less than that of the 30 — 34 weeks group. This does not 

 fall into line with the results just stated that were found within 

 the range of normal lactations. It would seem that the suggestion 

 there put forward does not hold good for cows remaining in 

 milk for more than about a year, but the present data are 

 insufficient on this point to justify definite conclusions. 



In considering the foregoing results, it became a matter of 

 some difficulty to decide between the " average " and the 

 " maximum " for general use. On the one hand the maximum 

 showed the less variation, while on the other the average gave 

 better correlation with totals. But more important than totals 

 is the physiological capacity of the mammary gland, and the 

 point really at issue is, which of these two figures indicates 

 this with the less exposure to error. 



When the possible sources of error come to be considered, 

 they maj' be classed into two groups :-^ 



1. Firstly, those influences which are general in character 

 and tend to act in the same direction for considerable periods, 

 such as temperature, time of year, food, method of feeding, 

 rainfall, housing, &c. These, it is suggested, would have more 

 influence oti averages than maxima. If they are generally 

 unfavourable during the two months, the average will be 

 reduced. But there seems a very good chance that on one day 

 during all this time the cow will have everything in her favour 

 (rise in temperature, new pasture, little extra cake, &c.), i.e., 

 that in spite of generally adverse conditions, there would occur 

 one day on which the environment would approach the ideal, 

 and would enable the cow to show " what she was worth." If 



