Their plan did not admit of the inclusion of large amounts of fat in their food mixtures, because this procedure diluted so much the mineral content, protein content, etc., of any natural food under investigation, that it made it impossible to feed sufficient amounts of grains, legume seeds, etc., to give an accurate picture of their dietary properties. It was imperative, for their purposes that rats should be fed with dry powders. A device was finally designed which left little to be desired for the quantitative feeding of such powders. This consists of a tin pie plate suspended from the roof of the cage by means of three wires which are attached to the rim of the pan so that they occupy the position at the angles of an equilateral triangle. These wires are brought together like the tepee lodgepoles, being connected with a ring by means of a small link for each wire. A wire is passed through perforations in the rim of the pan so as to cross it at a line, the center of which lies half the diameter of a pint tin cup from the center of the pie pan. This wire is bent so as to form a circle which circumscribes the center of the pan. The circle is just large enough to admit of the insertion of a pint tin cup without a handle. The feeding cup is thus held securely in the center of the pan. Inserted in the mouth of the tin cup is a cover so made as to fit firmly into the cup like the lid of a dinner pail. The top of this cover is shaped much like the top of a cuspidor, being funnel-shaped and having a hole in the center, through which the rats can secure the powdered food. In order to minimize the chances of scattering food, the cover is made about double the diameter of the tin cup, so that it forms a large funnel which tends to return the traces of powdered food to the cup should any be removed and not eaten by the rat. If this device is not filled more than half full, it is impossible for a rat to scratch any of the food from the cup. It must eat through the hole in the center. The size of this hole may be gauged, if desired, to the size of the rats used in experiment.

With this device McCollum and Simmonds found it easy to secure records of food intake which certainly are accurate to within 1 per cent. Since they did not keep their animals in individual cages, but in groups of four to eight or ten in one large cage, most of their records showed the food consumption of the group and not of any one individual. This practice was followed because of the great increase which it made possible in the number of animals which could be kept under observation, thus making feasible the conduct of a much larger number of experiments.

Experiments with feeding various sources of proteins at different percentages of the food mixtures brought to light the fact that there are few sources of proteins among our ordinary foods sufficiently high in quality that a diet containing but 9 per cent of protein, and otherwise satisfactory will be adequate for the promotion of growth at the maximum rate and to the full adult size, and for the maintenance to an advanced age of a high health standard in the rats confined to it. Only such proteins will, when fed at this plane of intake, serve to maintain the vitality of rats at a point where their fertility will be high, and the nursing of their litters satisfactory.