Two experiments were carried out, Expt. 1 with twelve cows in a 4×4 Latin square and Expt. 2 with twenty-six cows in two groups which spent 11 weeks on the experimental rations. In both experiments hay and silage were fed ad lib. In Expt. 1 four levels of concentrate feeding were offered, 1¼, 2½, 3¾ and 5 lb/gal, and in Expt. 2 two levels, 1¼and 5 lb/gal. The levels of feeding as a percentage of Woodman's standards for starch equivalent (S.E.) ranged from 82 to 125% in Expt. 1 and from 83 to 116% in Expt. 2; for digestible crude protein (D.C.P.) the corresponding figures were 70–100% and 105–107%. When concentrate consumption increased by 1 lb dry matter the consumption of bulk feeds was reduced by 0·18–0·22 lb dry matter but the total fibre intake remained constant. Milk yield showed small but significant responses to level of feeding. Average responses per 1 lb S.E. were 0·77 lb milk in Expt. 1 and 0·61 lb milk in Expt. 2. Total lactation yields of the cows in Expt. 2 were unaffected by the experimental treatments. The production S.E. fed per gallon of milk ranged from 1·71 to 3·72 lb in Expt. 1 and from 1·92 to 3·27 lb in Expt 2. Production D.C.P. ranged from 0·37 to 0·63 lb and from 0·67 to 0·70 lb, respectively.

High levels of feeding increased the S.N.F. and protein contents of the milk but did not affect other constituents. The responses in S.N.F. were lower than in some other experiments and some cows showed no response. The results are discussed with reference to the experimental methods and also with reference to the yield responses compared with published data, and to the practical importance of the high production obtained with low usage of concentrate feeds.