Hess and Unger (19) hold the view that in human nutrition there is not likely to be a deficiency of fat-soluble A sufficiently pronounced to be a factor of importance except under exceptional circumstances - as in times of war. They base their conclusion on an experiment conducted with five infants who were fed dried skimmed milk (Krystalak) 180 grams; sucrose 30 grams; cotton-seed oil 30 c.c; autolyzed yeast 15-30 c.c; orange juice and cereal. The autolyzed yeast was to insure a liberal supply of the anti-neuritic substance, water-soluble B. The orange juice was to supply the anti-scorbutic factor, water-soluble C, and the olive oil to furnish fat. The diet was assumed to be almost free from fat-soluble A, but entirely adequate so far as definite understanding of the necessary factors can be judged in the light of present knowledge. The babies were from four to nine months old. They were confined to this diet for periods varying from five to nine months. Rickets was not observed in any of the babies. In fact they grew in a fairly normal manner. From this experimental data Hess and Unger were led to doubt whether more than traces of fat-soluble A are necessary in the nutrition of the human infant when other factors in the diet are satisfactory.

There is a fundamental weakness in their experiment. The assumption that the diet employed was nearly devoid of fat-soluble A, the factor on which the results were supposed to turn, is erroneous. Soon after the discovery of the existence of this dietary factor, McCollum and Davis (20) studied the relative values of the fat and non-fat portion of milk as sources of fat-soluble A. They compared the growth of young rats fed on diets of similar composition with respect to the several components. They made 3 per cent of butter fat the sole source of fat-soluble A in one diet and 10 per cent of skimmed milk powder (Merrell-Soule) in another. It was observed that growth and well-being were about equally well maintained in the two groups of animals. In 100 c.c. of whole milk there is approximately 3 grams of fat and 10 grams of solids not fat, and in the skimmed milk powder the fat was not more than 0.6 per cent. The 10 per cent of skimmed milk powder used in the one experiment, therefore, represented the same amount of milk as did the 3 per cent of butter fat. From this observation McCollum and Davis drew the conclusion that skimmed milk contains much more fat-soluble A than could be accounted for on the basis of its fat content. They expressed their belief that about half the fat-soluble A contained in milk is removed with the butter fat in skimming with the centrifugal separator, and half remains behind in the skimmed milk. If this observation is trustworthy, the diet employed by Hess and Unger was not very deficient in fat-soluble A, and indeed did for their infants just what it would have done for the nutrition of young rats - animals which respond very sensitively to a deficiency of this substance. Hopkins (21) has recently stated in a symposium on diet held by the British Medical Association, that he has observed experimentally the effects of skimmed milk as a source of fat-soluble A, and is also convinced that it contains much more of this