1. In a large-scale experiment, using about 500 cows, half the cows were fed a production ration consisting of decorticated ground-nut cake, maize meal, wheat offals and molasses supplying about 0.6 lb. protein equivalent per 10 lb. milk produced, and the remaining half were fed a production ration containing the same constituent foods in different proportions supplying a similar starch equivalent, but only two-thirds the amount of protein. No measurable difference was observed in the fatness, handling properties of the skin, hair on the body and percentage of fat and solids-not-fat in the milk.

The milk yields showed no difference during the winter feeding period of about 20 weeks, but during the succeeding period when all the cows were allowed ample grass the milk yield response of the cows previously given the low-protein ration was significantly greater than that of the corresponding cows previously given the high-protein ration.

The experiment was repeated the following winter, using about 600 cows, and confirmatory results were obtained.

2. Another experiment similarly planned, using 300 cows, compared highwith low-protein production rations consisting of the constituents noted above together with soya-bean meal and rice bran. No significant differences were observed between the effect of the two rations on milk yield or composition, fatness of the cows, etc. There was, however, an indication (not statistically significant) that the low-protein ration which contained a relatively high proportion of rice bran depressed the milkfat percentage.

3. A small but more completely controlled experiment confirmed the findings of the large-scale experiments.