Although the experimental methods above described were developed for use with rats only they will serve to indicate the important points to be observed in the case of animals in general.19

Certain general considerations are further emphasized in the literature. It is' obvious, as Osborne and Mendel20 point out, that the experimental conditions, apart from the food, must be such that control animals on a normal diet can grow and maintain themselves over a satisfactory period of time in a perfectly normal manner. In this connection it should be noted that a "normal" growth curve is not complete proof of satisfactory nutrition. McCollum, Simmonds and Parsons21 observe that when the faults of a diet are of a certain degree of magnitude, animals with a fair amount of vigor may grow at the normal rate and reach full adult size and yet not be in a state of optimum physiological well being. If continued on such diets, they may be tardy in maturing sexually, or may be capable of but a fraction of the fertility of the vigorous, well-nourished individual. It has likewise become apparent that animals which have grown at a satisfactory rate, appear vigorous, and produce a few young at or near the usual age, may fail to maintain fertility to the age normal for the species. In all cases where the diet is below the optimum in character the signs of involution characteristic of old age appear earlier than in the well nourished.

These several variations from the normal in the usual functions of the adult animal serve as valuable criteria as to the state of nutrition and are more refined than the simple observation of the growth curve.

19 For methods of handling other species of experimental animals see: Wells and Swing, Georgia Agric. Exp. Sta. Bull. 119, 1916, for experiments with pigs; Osborne and Mendel, J. Biol. Chem. 26, 293, 1916; 33, 433, 1918. Drummond, J. C. Bioch. J. 10, 77, 1916, and Funk and Macallum, Ibid. 27, 3, 1916, for experiments with chickens. Mitchell, H. H. J. Biol. Chem. 26, 231, 1916, for experiments with mice; Cohen and Mendel, J. Biol. Chem. 35, 429,1911, for experiments with guinea pigs.

Hume (Biochem. J. 25, 30, 1921) advocates the employment of guinea pigs for trial, in testing the vitamin A value of non-fatty foods, when for any reason rats are unsuitable for the purpose. Hume fed guinea pigs with a mixture of oats and wheaten bran ad libitum, but weighed. To this was added the food to be tested in weighed or measured quantities; the residues were either fed by hand or were measured and the quantity actually consumed was calculated. Whenever necessary, extra antiscorbutic in the form of orange juice was added. Growth took place when green cabbage (raw or steamed), green cabbage juice, hay and milk (raw, heated or dried) were added to a diet of oats and bran and water. The growth varied according to the size of the ration given. Raw white cabbage, white cabbage juice, swede juice, orange juice, onion or germinated peas produced little or no growth. The growth-promoting accessory substance in these experiments corresponds in its distribution among foodstuffs with the known distribution of vitamin A.

20 Osborne and Mendel, J. Biol. Chem. 13, 234,1913.

McCollum and Davis22 are even more emphatic, urging that in their experience even normal growth to the normal adult size and continued maintenance does not necessarily indicate perfect nutrition. Only when animals reproduce and nourish their young normally, and repeat this at normal intervals, can it be said that the ration is physiologically sufficient. And again, 23 "Growth to the normal adult size at the usual rate and continued well-nourished appearance is not sufficient evidence that a ration is fully adequate. Only when normal reproduction and rearing of the young is repeated at normal intervals can a ration be said to be physiologically sufficient".

It follows therefore that only in those cases where the experiments have extended over a sufficient period of time to cover the life history of at least two generations can the evidence be regarded as entirely trustworthy.

The necessity for long continued experiments before any adequate conclusions can be drawn has been frequently emphasized by various writers and repeatedly demonstrated by experiments in which apparently satisfactory growth over a considerable period of time had ultimately resulted in decline and complete failure.

Hart, McCollum, Stccnbock, and Humphrey " in their study of the physiological effects on growth and acount of physical defects not rations derived from a single plant carried A second complication is period of three years, and conclude that unquestion of all artificial diets cal value of a ration is largely dependent upon its condition could be ents, but the usual determinations made on feeding mature. As an reveal the character or manner of combination of many of infection uents. Consequently, according to these investigators, the logical value can be determined, in the present state of our knowledge only by long continued observations of the reaction of the feed on the animal." Osborne and Mendel confirm this.25

21 McCollum, Simmonds and Parsons, J. Biol. Chem. 37, 159, 1919. 22 McCollum and Davis, J. Biol. Chem. 20, 416. 23 Ibid. 21, 620.

24 Hart, McCollum, Steenbock and Humphrey, Un. Wis. Agric. Exp. Sta. Res. Bull. No. 17, 1911.