1. The growth of strains of Drosophila melanogaster selected for large size under different nutritional conditions has been recorded on a variety of different media and compared with that of the unselected population. The experiments were designed to test the inference from earlier work that selection for the same ‘character’, body size, on different diets leads to more or less different changes in growth and metabolism. The inference has been amply confirmed.

2. When compared on a number of deficient synthetic diets, the strains which had been selected either on a low-protein diet or on one in which all the essential nutrients had been reduced, suffered a much smaller reduction in body size than either the unselected population or, especially, a large strain selected on the favourable live yeast medium. Some diets which drastically reduced the body size of the unselected population lead to no change in the size of strains selected on the synthetic media, although development time was prolonged. Hence selection had extended the capacity for maintaining a characteristic adult body size to diets which normally would lead to a decline. This is taken as evidence of improved adaptation to such conditions. There is also some evidence that selection on the synthetic diets had lowered the level of adaptation to the usual live yeast diet, since body size tended to be lower on this medium than on some of the normally sub-optimal diets.

3. To provide comparisons in adverse conditions which are probably more closely related to those commonly encountered by populations in nature or the laboratory, the performance of the strains has been compared in a graded series of competitive conditions on the live yeast medium. By using genetically marked files of the foundation population, which were shown to react in the same way as unmarked flies—in terms of survival, body size and development time—the competitive ability of the different strains has been tested against that of unselected individuals. The latter are generally superior to the selected strains, which differ among themselves, however, in a way which can be related to the conditions in which they were selected.

4. Under such competitive conditions, the strains selected on the synthetic diets suffer a much greater decline in body size than do the unselected individuals. For the strain selected on live yeast, the proportional reduction of body size is about the same for the unselected flies at lower levels of crowding, but is clearly greater under more severe conditions of competition.

5. The low-protein strain has been backcrossed to the unselected stock. When reared on a variety of synthetic diets, the performance of the F1 was generally intermediate between that of the parents.

6. Nutritional variation may be responsible for either a high environmental correlation between the two measures of growth, body size and duration of larval period, or no apparent correlation. Provided the diet is not too unfavourable, body size remains constant although development time may be lengthened to a variable degree. With more adverse conditions, body size is reduced and development time is lengthened more or less proportionately. Such differences in reaction probably depend on the particular stage of larval growth and development primarily affected by the treatment; this problem is being examined further. The inverse relations between body size and development time may represent the operation of a kind of safety mechanism which ensures that the adult reproductive state is attained sooner than would be so if the capacity for maintaining a characteristic body size were more effective in relation to deficient diets. Populations and species adapted to different conditions are likely to differ as to where the balance is struck between effective maintenance of a characteristic adult size, with maximum potential egg production, and the alternative response, according to their ecology. This possibility must be borne in mind when the response to selection for, say, body size is compared in different species.