Complication may be introduced through the presence of a toxic substance in the food, in which case, according to Hart, Miller and McCollum43 an additional supply of the essential factors will be required to counteract the injurious effect. Little is known as to the toxicity of common foodstuffs when fed as part of a restricted ration. Davis,44 believes that many of them may be toxic to certain animals if fed in sufficient amounts for a sufficient time. The toxicity of cottonseed is well attested and toxic effects of the whole wheat grain have been reported several times45 although Osborne and Mendel dispute this.46 A substance like magnesium which is not only harmless but essential in proper concentration may be distinctly toxic in larger amounts. That slight toxicity or other injurious character in a ration may easily exist without recognition and thus lead to inaccurate conclusions is brought out by McCollum, Simmonds and Pitz47 in another connection.

"A single factor (protein) in a ration may appear to admit of the maximum performance of the animal with respect to growth, without itself representing the optimum amount or character. When this circumstance prevails, it may entirely escape notice, yet if in another ration exactly like it, except that a second factor tends to injure the animal, nutritive failure may result. In such a case as the latter, improvement of the protein factor by the addition of more protein or by the substitution of a better protein, the plane of protein intake remaining unchanged, the animal may make the maximum performance notwithstanding the unfavorable character of the injurious factor of the ration".

Osborne and Mendel48 point out that the problem of dosage for vitamins may be quite different in the case of rats which are supplied with a minimum of the essentials throughout the period of growth and those which have declined through complete deprivation of some requisite food factor 49

43 Hart, Miller and McCollum, J. Biol. Ghem. 25, 241, 1916.

44Davis, J. Home Ec. 12, 207, 1920.

45Withers and Ray, J. Biol. Chem. 14, 53, 1913; Withers and Brewster, Ibid. 15,161, 1913; Crawford, J. Pharm. Exp. Therap. 1910, i. 519; Richardson and Green, J. Biol. Chem. 30, 249; Hart and McCollum, J. Biol. Chem. 19, 373, 1914; McCollum, Simmonds and Pitt, Ibid. 25, 105, 1916; Hart, Miller, and McCollum, Ibid. 25, 239, 1916; Hart, McCollum, Steenbock and Humphrey, Proc. Nat. Acad. So. 3, 374, 1917; J. Agric. Res. 10, 175, 1917; Hart. Steenbock, and Humphrey, Univ. Wis. Agric. Exp. Sta. Bull, 287, 1918.

46 Osborne and Mendel, J. Biol. 37, 595, 1919.

47 McCollum, Simmonds, and Pitt, J. Biol. Chem. 25, 111, 1916.

48 Osborne and Mendel, J. Biol. Chem. 41, 452, 553, 1920.

49Osborne and Mendel, J. Biol. Chem. 41, 452, 1920.

There are two feeding methods of testing products for the presence of water-soluble vitamin. One consists in feeding the material to be investigated to animals that have declined on a diet deficient in this food factor. Such tests, however, will not demonstrate the comparative vitamin content of different foods. They are essentially restorative in character. If the outcome is positive in the sense of a renewal of nutritive well-being, the test for the vitamin may be qualitatively successful. However, animals which have suffered a vitamin deficiency may be so badly malnourished or underfed that relatively excessive amounts of the vitamin-bearing food may be required to restore them to normal condition. In that event a failure to promote nutrition by the extract to be tested may be ascribed to the condition of the animal rather than the product. Comparisons of vitamin-bearing foods may not be reliable when they involve attempts to restore nutrition in animals that have suffered to a variable degree from unlike deficiencies.

The other method consists in feeding the supposed source of vitamin to animals which are normal in health and development, and observing whether the product offered supplies what is needed to promote normal growth when the diet affords an adequate supply of all essentials except that to be investigated.

The first of these methods is the one which has been chiefly resorted to in testing for vitamin A in plant and animal products, and also very commonly in the study of antineuritic B. In much of the more recent work on both B and C, however, both methods have been used and the results given for comparison.

Sherman, LaMer and Campbell50 suggest the following procedure for the quantitative determination of the antiscorbutic vitamin. Guinea pigs are fed a basal diet consisting of 59 per cent oats, 30 per cent skim milk powder heated two hours at 110° C, 10 per cent butter-fat, 1 per cent NaCl. The minimum protective dose of antiscorbutic is determined, and in addition the degree of scurvy produced, as measured by the autopsy findings, retardation in growth, and symptoms in life, of a series of animals receiving graduated sub-protective doses of antiscorbutic food. When the dosage is calculated per unit of body weight it is possible to distinguish the degrees of scurvy produced for addenda of antiscorbutics differing by 15 per cent or less. The percentage destruction due to a deleterious process is obtained by comparison of the degree of scurvy produced in a series of standard animals fed a similarly graduated series of doses of the treated product. The probable error of the mean in a series of five or more animals is less than 4 per cent.

50 Sherman, LaMer and Campbell, Science, Aug. 26, 1021, 170.

The accurate evaluation of much of the earlier research is difficult on account of the lack of data as to the actual food intake. With reference to this, Osborne and Mendel51 say:

If no account of the food eaten is taken the outcome of the experiments may be wholly misleading. Some preliminary experiments appeared at first sight to indicate a marked superiority of barley proteins compared with those of wheat or rice; but when gains of weight per gram of protein eaten were calculated this superiority entirely disappeared. It is thus plain that experiments which have hitherto purported to show the relative nutritive value of different food products in respect to any one of their constituents have no comparative value unless the amount eaten is known.