In this method it is assumed that the amount of yeast cake added will not derange the proportions of protein fat, etc., in the basal diet enough to affect optimum conditions in these respects. This is a curative type of experiment. If we wish to develop a preventive experiment the yeast cake may be incorporated in the diet from the first and the amount necessary to prevent deviation from the normal curve determined. Both methods are utilized, the one checking the other. If however the amount of the substance necessary to supply the vitamine required for normal development is large such addition would of course disturb the proportions of nutrients in the normal diet and in that case analysis must be made of the substance tested to determine its protein, fat, carbohydrate and salt content and the basal diet corrected from this viewpoint so as to retain the optimum proportions of these factors. McCollum's cereal testing combinations are illustrative of such methods applied to cereals. Still another method is to add a small per cent. of the unknown and then add just enough of the vitamine tested to make sure that normal growth results. Such a method gives the results in terms of a known vitamine carrier. For example, if we add to a basal diet, sufficient in all but the "A" vitamine (Steenbock's mixture for example), a small per cent of a substance whose content in "A" is unknown and note that growth fails to result we can then add butter fat until the amount just produces normal growth. If now we know just what amount of butter fat suffices for this purpose when used alone we can calculate the part of the butter which is replaced by the per cent of unknown used. To put this in terms of figures will perhaps make the idea clearer. Let us assume that 5 per cent of butter fat in a given diet is sufficient to supply the "A" necessary for normal growth. Assume that the addition of 5 grams of the unknown in 100 grams of the butter-free diet fails to produce normal growth but that by adding 2 per cent of butter fat normal growth is reached. It is obvious under these conditions that 5 grams of the unknown is equivalent in "A" vitamine content to 5 minus 2 grams of butter fat, i.e., is equivalent to 3 grams of butter fat or expressed in per cents the substance contains 0.6 or 60 per cent of the "A" found in pure butter fat.