Watterson's homozygosity test of selective neutrality, which is based on the Ewens sampling formula, has been applied to the polymorphic paracentric inversions of Drosophila subobscura in populations near ZiJrich, Switzerland, in order to deterrrrtne whether the test is sufficiently powerful to detect selection. In three cases, involving chromosomes O, J, and E, the test does lead to the inference of nonneutrality. For chromosome J, there is a deficiency of the most frequent structural type, indicating some sort of balancing selection. For chromosomes O and E, there is an excess of the most frequent structural type, indicating some sort of directional selection against rare structural types. In the remaining two cases, chromosomes A and U, the distribution of frequencies of the structural types is not imcompatible with selective neutrality. Thus, the homozygosity test has detected nonneutrality in three cases out of five.
formula may be so low as often to prevent rejection of the neutrality hypothesis even though selection may be occurring. In this paper, we stand the problem on its head and consider a case in which all 'alleles' can be identified in a system that is almost certainly subject to natural selection. The homozygosity test of neutrality (Watterson, 1978a), which is based on the Ewens sampling formula, fares rather well. In three cases out of five, the hypothesis of selective neutrality can be rejected. The result is also of some interest in verifying that the 'alleles' in question -which are actually polymorphic paracentric inversions in Drosophila subobscura -are subject to selection under natural conditions; while this result has been widely believed for decades, the inference has rested largely on laboratory studies (see Dobzhansky, 1970, for a summary).