Three wether lambs of the Mutton Merino type were kept confined in metabolism cages confined to either a calorimeter or a structure resembling the plant, from about birth to about 36 weeks. Measurements over regular intervals with aid of the tritium dilution technique were in agreement with protein and fat retention measured during alternate weeks in the calorimeter. Linearly independent measurements of feed intake, heat production, protein and fat during active growth indicate that cumulative feed intake, cumulative heat production, body protein and body fat are pairwise log-linearly related for each lamb, in such a fashion that the two-dimensional relationships extend to a straight line in 4 dimensions. The parameters of this straight line determine the growth rate constant for all these traits up to a multiplicative constant with time dimension. This suggests improved quantitative control of body composition by control of feed intake. Principal components analyses support the extrapolation to a straight line in 4 dimensions by giving, on the average, a first root of 99%. Cumulative feed intake against time is described by fitting first order auto- regression of ln (cumulative DE intake). In the feed domain, cumulative ME intake, cumulative heat production, body protein, body fat and body mass are log-linearly related by ordinary regression methods to cumulative DE intake. From these relationships efficiencies for protein and fat deposition are calculated and it is shown that the gross efficiency of energy deposition is, approximately, constant over much of a lamb's growth period, while no such constancy can be assumed for the biological efficiencies of protein and fat deposition. Both the dimension and curvilinearity of the system of relationships between the relevant variables invalidate the use of conventional multiple regression methods to calculate partial efficiencies of protein and fat deposition. Consequently, doubt is expressed about the existence of constant efficiencies of protein and fat synthesis in vivo.