His experiments were all concerned with young rats at a stage when rapid growth is normal. Their initial weight was mostly from 36 to 50 grams, a stage when weight is normally doubled on an efficient dietary in 20 days or less. The animals, unless otherwise mentioned, were always fed two in a cage, bucks being paired with bucks and does with does. When the amount of food eaten is to be determined, this grouping has the disadvantage that an average for the two animals must be accepted, but when food is given in excess of the quantity required, this is not an important point, and any disadvantages are more than counterbalanced by the fact that quite young rats progress more normally when they have a companion than when kept singly. Great attention was given to the maintenance of a uniform temperature, this being recorded by a registering thermometer. In particular a constant temperature for day and night was carefully secured. The rats were kept in round wire cages somewhat similar to those used by Henriques and Hansen and by Mendel and Osborne, though shallower and of wider mesh. They stood in trays filled with sawdust, over which was a layer of filter paper. Rats tend to eat both sawdust and paper, so the cages were raised upon legs four inches high.

Two basal diets were used, differing only in the protein they contained. In one the protein was Merck's pure casein prepared by Hammersten's method, in the other it was the commercial casein preparation known as "Protene." No roughage was provided, as with rats Hopkins considers it unnecessary.

The composition of the dry food mixtures was the following:

The salts added were obtained by incinerating the normal laboratory food on which the rats had been kept when not under experiment and consisted of equal parts of the ash of oats and dog biscuits. The commercial casein contained 3 per cent of ash, raising the amount of salts in the mixture from 2.6 per cent to 3.3 per cent. The calculated energy value of the pure casein mixture is just over 5 calories per gram, and that of the protein mixture almost exactly 5 calories per gram. A series of estimations in the bomb calorimeter gave mean value of 4.98 for both mixtures. In general, however, the energy values of the diet were directly determined in the mixtures as they were made up for feeding; the protein and starch not being previously dried.

15 McCollum and Davis, J. Biol. Chem. 20, 644, 1915. 16 Mitchell and Nelson, J. Biol. Chem. 23, 400, 1915. 17 Hopkins, J. Physiol. 1912, 44, 427.

"The protein, starch, and sugar and salts were mixed dry, and the fat rubbed in by hand, the mixture being worked up until of completely uniform composition. It was given to the animals in earthenware vessels. The day's ration for each pair of rats was weighed out dry, and given the desirable consistency in the following way. Half the ration was mixed with enough water to make it into a thin paste, and the remaining half of the dry material was then added little by little, and well stirred in with a glass rod. The final mixture thus obtained was of such a consistency and character that the rats ate it freely, and almost without any tendency to scatter the food. The amount of food given was] always in excess of the consumption. Water was supplied in a special vessel. Before the day's ration was administered to either set of rats, each individual of that set which was to receive milk was put into a separate cage, and the measured quantity of milk given as a separate ration. Only after this was consumed - and its consumption never occupied . more than a few minutes - were the rats returned in pairs to the original cages. Both sets were then fed simultaneously with the artificial mixture.

The composition of the milk was carefully determined from time to time, and the energy content of the solids was occasionally determined in the calorimeter. This scarcely varied from the value of 4.7 calories per gram".

Hopkins found it possible to determine the amount of food eaten with great accuracy. A very small quantity might fall through the bottom of the cage on to the filter paper beneath, but this could always be recovered quantitatively. Absolute accuracy was to be obtained by removing such spilled food at frequent intervals, so as to avoid any soakage with! the urine. The amount spilled was never more than a minute fraction of the food eaten, or of that weighed, back; and in practice no difficulty was found in dealing with it. At each day's feeding, the food left over from the previous day was carefully removed from the vessel which contained it, the small quantity of spilled food added to it, and the whole dried at low temperature till of constant weight. This weight was deducted from the dry weight of the food mixture as originally weighed out. During periods in each experiment the feces were collected in order to determine the energy content, so that data as to the absorption of the food might be obtained. The greater part of the feces fell through the bottom of the cage on to the paper beneath, and as this lay upon sawdust, any urine passed was rapidly soaked up, so that very little contamination of the feces occurred.

No attempt was made to demarcate the feces of the experimental period, but as they were collected for a week, and under precisely similar conditions in the two sets of rats under comparison, such demarcation seemed unnecessary.

Drummond and Coward18 have worked out a technique for carryin out tests for vitamin A. The basal ration which they employ is composed of:

18 Drummond and Coward, Bioch. J. 14, 601, 1020. See also Osborne and Mendel, J. Biol. Chem. 45, 277, 1920-21.

Purified casein ...................................... 18 parts.

Purified rice starch.................................. 52.

Refined hydrogenated vegetable oil .................. 15.

Yeast extract....................................... 5.

Orange juice ........................................ 5.

Salt mixture ........................................ 5.

They state that small rats of 50-70 grams (4-5 weeks old) should show very little growth at all and should remain stable for a week or two after the slight initial growth. Any considerable increase in body weight in rats of this age on the diet is to be taken as an indication that the basal ration is insufficiently purified. They regard rats of considerably over 100 grams weight as unsatisfactory for testing for this vitamin and prefer to use only rats of between 80 and 120 grams, of which the growth has been suspended for 10-14 days.