NORMAL VARIATION IN FAT CONTENT OF MIXED MILK. 3OT 



environmental conditions, During each of the months of ^larch 

 and July there was no change in the character of the food or 

 the method of handling the animals. But in March the cows 

 were in the barn fthat is stall fed) all the time, whereas in July 

 they were on pasture continuously. M'ay and October repre- 

 sent months in which tliere were marked changes in the feed. 

 In the course of the month of Afay the cows were put on 

 pasture. In the month of October they were partly stall fed 

 and partly on pasture, and other changes were made in the ra- 

 tion as well. In this way a contrast was afforded between the 

 months of uniform feeding and months of varying feeding. 

 The details as to the actual feed used may be found in the 

 original paper. 



P'rom the data described the constants of variation given in 

 Table i and 2 of this paper have been calculated, under my 

 direction, by Air. John Rice j\Iiner, the stafl computer of this 

 laboratory. The constants were calculated directly from the 

 raw data without grouping. In one ca.se, namely the evening 

 milk during the month of May, it is evident that there are some 

 errors in the records as printed in the original paper. vSome 

 of these are clearly typographical. Such we have been able 

 to correct from internal evidence in the paper itself. Even 

 after this partial correction, however, there are evidently still 

 left some undetected errors, either in the original determina- 

 tions of the fat percentages, or in the recording and printing 

 of these. The variation shown in the milk of this milking in 

 per cent of fat is so very much greater than that of any of the 

 other data, that it can only mean some uncorrected error. In 

 the case of the absolute amount of fat, the errors in per cent 

 are apparently compensated for to a considerable degree (cf. 

 Table 2). In the case of variation in fat per cent the variation 

 constants for the evening milk of May are given first for the 

 data exactly as they stand, and then for the data after correc- 

 tion of the obvious errors. 



