Differences in the kinetic and/or endurance properties of enzymes encoded by alternative alleles have been suggested as the basic level at which selection could act on allozyme polymorphisms1,2. The correlation between spatial differences in gene frequencies and an environmental variable (hereafter designated cline, in its widest sense), when not spurious, has been proposed to show the working of selection3. Let us suppose that a given cline has originated through an environmentally imposed selection on enzyme activity. Because differences in enzymatic activities among individuals can be split into structural-gene and 'genetic-background' components4-6, selection should act on both components in the same direction. That is, a correlation should exist between the environmental variable and the background component and it should be of the same sign as that between the environmental variable and the frequency of the most active allele. It should be less likely for clines conforming to these expectations to be spurious. Here we have investigated the relationship between the alcohol dehydrogenase locus (Adh) polymorphism and temperature in Spanish populations of Drosophila melanogaster and have found that it meets the above criterion. © 1980 Nature Publishing Group.