Refinement OF Feeding Experiments for Milk Production 223 



breed, because the records of a much larger number of individuals were 

 available. No study was made of groups comprising different breeds, 

 because it is considered undesirable to use animals of different breeds 

 in a feeding experiment under any conditions. 



It has been stated that data were to be obtained as to the variation 

 in an experimental period of individuals and of groups selected on the 

 basis of their production during a preliminary period. Nearly all the 

 studies included in this bulletin are concerned with milk production. 

 The word production will therefore mean milk production unless other- 

 wise designated. In order to get comparable and reliable results, a period 

 of definite length, during which the records should be studied, was required. 

 A twenty- weeks period was accordingly chosen, consisting of a two- 

 weeks preliminary period and an eighteen-weeks experimental period. 

 It must not be assumed that eighteen weeks is considered the best dura- 

 tion for an experimental period. It is believed that this period should 

 be as long as the uniformity of conditions prescribed at the beginning 

 of a given feeding trial can be maintained. Twenty weeks was found to 

 be the longest period, between the time of freshening and the time when 

 the animals went on pasture, for which any considerable number of records 

 could be obtained. To combine records of animals on winter feeding 

 with those of animals in pasture would introduce a rather serious 

 irregularity. It therefore seemed desirable to limit the study to a twenty- 

 weeks period for a given lactation, and the results obtained must be 

 considered with respect to the period used. 



In a part of this study the records of cows freshening between August 15 

 and December 15 were used, and the twenty-weeks period was begun 

 exactly one month after calving. This eliminated the period during 

 which many cows were on test. Under this plan all the records were from 

 cows in the same stage of lactation, thus eliminating any variation due 

 to difference in this respect. It was realized, however, that under experi- 

 mental conditions — even tho, with a sufficient number of animals to 

 choose from, a group freshening at approximately the same time could 

 be made up — the exactness prescribed above could not be obtained. 

 Accordingly, in another part of the study the twenty-weeks period was 

 started on January i of a given year, using cows that had freshened in 

 the preceding simimer or fall. It is believed that this scheme approximates 

 that of the ordinary feeding experiment. 



In grouping together cows of equal productive capacity on the basis 

 of the preliminary period, the question arose as to what constitutes " equal 

 production." Inasmuch as no information on this point was found in 

 the accounts of feeding experiments, arbitrary limits had to be chosen. 



