A genetico-nutritional experiment is described in which 22 Ayrshire twin cattle, both one-egg and two-egg, were reared and maintained for up to seven years on a standard all-pelleted ration. Six equally spaced constant levels of food intake ranging from 35 1b. to 140 1b. per week were used. Each twin was reared on nearly ad libitum feeding until it had grown sufficiently to be able to consume its pre-assigned level of constant food intake. Thereafter its weekly food intake remained unchanged until the end of the experiment. The number of animals on each of the six feeding levels varied from two to six.

On each constant level of food intake, animals continued to grow fairly rapidly for two or three years, but thereafter their growth tended to slow down and usually an equilibrium weight was eventually attained. The resultant mean weight curves constituted an apparently simple, regular and systematic family with each curve eventually separated from the next by a relatively constant weight difference.

Coefficients of variation were about 3% for unrelated animals strictly conforming to the design, but increased at later ages to around 6% when animals with different pre-treatment were included. Coefficients of variation, however, did not change over the ascending series of feeding levels.

The advantages of the design for exploring the triple relationship of gain, weight and food intake are outlined. The potential use of such results as an empirical feeding scale and the possibility of obtaining paths of maximum efficiency from such a controlled experiment are described.