Experience has shown that it is dangerous to draw conclusions from experiments of too short duration or to base them on too few animals. For complete data the experiments should be carried through the complete life cycle of the rat, including the reproductive period. Otherwise it may turn out that the amount in the unknown while apparently sufficient for normal growths is incapable of sustaining the drain made in reproduction. It is this consideration that makes the accumulation of authoritative data on vitamine contents of foodstuffs so slow and tedious and one of the reasons why we lack satisfactory tables in this particular at present. Osborne and Mendel raise another point of methodology and believe that more accurate results will be obtained if the source of the vitamine is fed separately than if mixed with the basal diet. It is easily possible that since one of the effects of lack of vitamine, especially of the "B" type, is poor appetite, the amount necessary to produce normal growth may be smaller than would appear from results obtained by mixing it in the basal diet. When so mixed the animals do not get enough to maintain appetite and really decline because they do not eat enough rather than because the amount of vitamine given is inadequate to growth. Details of this kind are matters however that particularly concern the experimentalist and as our purpose here is to merely describe the methodology we may perhaps turn now to other types of testing. Before doing so it is perhaps unnecessary to suggest that in all experiments it is important that the food intake consumed be measured. Also that in all such experimentation it is necessary to run controls on a complete diet rather than to rely too much on standard figures. For this latter purpose it is merely necessary to add to the basal diets the "A" as butter fat and the "B" as dried yeast or otherwise to make them complete. Various special mixtures have been tested out for this purpose and the data already presented supplies the information necessary to construct such control diets. Professor Sherman has given me the following as a control diet on which he has raised rats at normal growth rate to the fifth generation:

One-third by weight of whole milk powder.
Two-thirds by weight of ground whole wheat.
Add to the mixture an amount of NaCl equal to 2 per cent of the weight
of the wheat.

A control mixture based on Osborne and Mendel's data would have the following components:

Meat residue 19.6 per cent or casein 18 per cent.
Starch 52.4 per cent or 49 per cent.
Lard 15 per cent or 20 per cent.
Artificial protein-free milk 4 per cent.
Butter fat 9 per cent.
Dried yeast 0.2 to 0.6 gram, daily.

The preceding description has applied especially to testing for the presence of the "A" or the "B" vitamine. When we come to the methods of testing for the "C" type it is necessary to change our animal. Rats do not have scurvy but guinea pigs do. The philosophy of the tests for the antiscorbutic vitamines then will be identical with that of the polyneuritic methods with pigeons, viz., preventive and curative tests with guinea pigs. The "C" vitamine is especially sensitive to heat and this fact enables us to secure a "C" vitamine-free diet. La Mer, Campbell and Sherman describe their methods as follows:

First select guinea pigs of about 300 to 350 grams weight. Test these with the basal diet until you secure pigs that will eat the diet. Those that will not eat it at first are of no use for testing purposes, for a guinea pig will starve to death rather than eat food he doesn't like. Having secured pigs that will eat they should on a suitable basal diet die of acute scurvy in about twenty-eight days. Their basal diet is as follows:

per cent Skim milk powder heated for two hours at 110°C. in an air bath to destroy the "C" vitamine that might be present. . 30 Butter fat . . . . . . . . . . . . . . . . . . . . . . . . 10 Ground whole oats . . . . . . . . . . . . . . . . . . . . . 59 NaCl . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

They claim that when fruit juice addenda are given in minimal protective doses and calculated to unit weight bases, the results are comparable in precision to those of antitoxin experiments.

Old food should be removed every two days and replaced by new, cups being cleaned at the same time. Since this is a scurvy-producing diet its use is obvious. We can let the pig develop scurvy on it and then test the curative powers of the unknown by adding it to the diet or we can add it to the diet from the first and determine the dose necessary to prevent scurvy; or we can determine its effect in terms of a known antiscorbutic such as orange juice by combining it with measured quantities of the orange juice.

There are other diets that have been given for this purpose, e.g., Holst and Fröhlich induced scurvy by restricting animals to an exclusive diet of cereals (oats or rye or barley or corn). Hess and Unger have used hay, oats and water given ad libitum. All of these and others are subject to criticism on the basis that they are not necessarily adequate in other food factors and may therefore not be fair bases for testing the antiscorbutic powers of the unknown combined with them. Abels has recently shown that scurvy increases susceptibility to infections and believes that the scurvy hemorrhages are brought about by the toxic effects of infection. It is therefore desirable in testing for antiscorbutic power that the basal diet be itself as complete as possible in all factors except the absence of "C."