Results characteristic of excellent nutrition can be secured with diets planned as just described, in which the proteins are derived solely from milk, kidney, or a mixture of certain seeds, notably of rye and flaxseed oil meal. So far as experience has progressed, however, such effects cannot come from restricting rats to any diet containing less than 9 per cent of protein from any source. McCollum, Simmonds and Parsons (7) felt justified in classing as excellent, any protein or protein mixture which would support nearly optimal nutrition over periods approximating two-thirds or more of the normal span of life of the rat, when fed in amount corresponding to 9 per cent of the food mixture. Instead of trying as Osborne and Mendel had done, to establish absolute values for the proteins, they undertook only to compare any protein or mixture of proteins with some protein which, according to the definition just given, would be classed as excellent, or in other words, about as good as any proteins known from the nutritive standpoint.

In order to do this, McCollum, Simmonds and Parsons simply-confined groups of four to six rats in a cage 24x24x20 inches in dimensions, screened on all sides, or on three sides with wire cloth of one-quarter inch mesh. The food records always showed the amount consumed in the group. Note was kept of the number of times, with dates, when the standard quantity of one kilogram of food was prepared. The animals were not able to dispose of their food otherwise than through eating. It was believed that a more detailed record of individual food consumption would add little if any value to the results or make their interpretation more significant. Since Osborne and Mendel have repeatedly emphasized their belief that feeding studies of this nature fail of their purpose unless accurate food consumption records are kept, it is incumbent upon McCollum and his co-workers to defend their system against this criticism.