1. Two herds of cows were divided into two groups for winter feeding. One group was given the usual ration of hay, roots and “concentrates”, the other was fed on a similar diet in which part of the hay and concentrates was replaced by artificially dried grass.
2. The carotene content of the butterfat in the control group fell steadily from 0·8 mg. per 100 g. of fat to about 0·2 mg. between October and March. That of the group to which dried grass was given did not fall below 0·5 mg. per 100 g. of fat.
3. The milks from the two groups were employed in a nutritional investigation on the boys of two institutions. During a “fore-period” from July until the end of September, the boys received the ordinary milk of the herd as part of their diet. They were then divided into two groups which, during the experimental period from October until March received the “winter” or the dried grass “summer” milk respectively. Then followed an “after-period” during which all the boys received the mixed milk of the herds, then out at pasture. The composition of the basal diet of the boys at these institutions was known. Boys attending an elementary school in the vicinity of one institution were also investigated in a similar experiment but no details of their home diet were obtained.
4. Excluding children whose records were incomplete there were 159 in the group on “winter” milk, and 163 in that receiving “summer” milk.
5. Periodically records of height, weight, and chest measurement were made. All the boys were under medical supervision.
6. Statistical examination of the results fails to show that the artificially produced “summer” milk exerted a beneficial effect detectable by the improvement in growth or general condition of health.
7. The most probable explanation is that the basal diets of the two institutions were already adequate in respect to carotene and vitamin A.
8. The variations exhibited by the records of the groups provide another example of the very great difficulty of eliminating or adequately controlling sources of error in a big-scale nutrition experiment of this type.