Refinement of Feeding Experiments for Milk Production 221 



lactation yield from another cannot be made with great accuracy, since 

 the correlation coefificient between successive lactations does not rise 

 above + 0.576 ± 0.025. As a result of an extensive statistical study, 

 Gavin (191 2) suggests the use of the " revised maximum " as a figure to 

 describe the milking capacity of a cow. This figure is the maximum 

 daily yield, three times reached or exceeded. The figure bears a close 

 relation to the total yield of a normal lactation and shows rather less 

 variation than the total. Gavin states that the figure is independent 

 of length of lactation and time of service, and is not greatly influenced 

 by general environment. 



Instead of selecting animals on the basis of production in a previous 

 lactation period, groups may be made up by considering the production 

 in the same period before placing the cows on experiment. Caldwell 

 (1913) fed a certain number of cows for thirty-one days on a basal ration, 

 and then made up groups for a feeding test on the basis of production 

 during that period. It is believed that this scheme is much better than 

 the one based on the yields of a previous lactation. However, some con- 

 sideration of the latter seems desirable in order that abnormal cows may 

 be excluded. Fraser and Hayden (19 12) report that they consider both 

 present records and previous production in forming groups. 



SUMMARY 



From this study of the methods of conducting feeding experiments, 

 it appears that the practical feeding trial still has its importance. Refined 

 methods are, however, desirable. A more thoro knowledge and appre- 

 ciation of the special variable factors involved in milk production is also 

 essential. The continuous system of carrying out feeding tests would 

 seem to be the more satisfactory method, provided the probable limits 

 of variation over the experimental period can be ascertained. In select- 

 ing animals for an experiment the previous records and history of treat- 

 ment and behavior should be consulted, in order that animals abnormal 

 in these respects may be excluded. This selected group of animals 

 should be fed the basal ration for a preliminary period, and should receive 

 for that period the same care and management as tho on experiment. 

 On the basis of production and behavior during the preliminary period, 

 the selection and grouping of animals for the experiment should be made. 

 The group selected for the preliminary period should contain a some- 

 what larger number of animals than will be required for the experiment, 

 in order to allow a certain range in selection. In choosing animals for 

 both the preliminary period and the experimental period, as many of the 

 variable factors should be eliminated as is possible in consideration of 

 the number of animals at the investigator's disposal. There will always 



331 



