A line of mice, at its limit to selection for high body weight did not decline in performance over 11 generations of random mating, neither did it respond when selection was renewed. The experiment tested a method of improving body weight by a scheme which had earlier increased litter size under similar circumstances. The scheme was to derive partially inbred lines from the plateaued line, to select during inbreeding and, finally, to cross the best inbreds. Body weight was not increased, but the study allowed further examination of the residual genetic variance in the line.

During inbreeding, the inbred lines became clearly differentiated in body weight, proving that loci controlling body weight had not become fixed. There was also a significant response to selection for a lower body weight during inbreeding. The pattern of results suggested the segregation of recessive genes, detrimental to high body weight but which selection had become inefficient at removing. A genetic model compatible with the results accommodated several such recessives, perhaps as many as 10, each with an effect of about two-thirds of a standard deviation (or some equivalent combination of gene number and effect), and at frequencies of around 0·2. Nevertheless, the total improvement in body weight to be gained by their elimination was only half a gram, or less than 2 %. Thus, substantial genetic effects can occur at individual loci despite trivially low heritabilities and negligible potential gains.