Selection was carried on in the vestigial-winged, bar-eyed stock for three generations and in the long-winged, bar-eyed stock for seven generations. The facets in the eyes of 9,000 flies were counted. Selection in the vestigial-winged stock had to be discontinued on account of sterility and low production in the single lines. The sterility and low production were not due to inbreeding. Sterility was not due to inability on the part of the females to lay eggs, nor does it appear to have been the final stage in the reduction of the number of offspring. Selection in the vestigial-winged stock was effective for one generation but failed to produce further effects in the second and third generations. Return selections from the low lines were not effective. A mutation appeared in a male in line 16. It was a reducing factor and was sex-linked. The complete sterility of the offspring of this male prevented further study of the character. Selection in the long-winged stock was effective for six generations. In the downward selected lines most of the high flies were eliminated in the first three generations and no further effect of selection was noted. Return selections from the sixth generation, however, were still effective. The mean of the upward selected lines continued to increase at approximately uniform rate for six generations. The seventh generation must be disregarded because it was not an upward selection. The range in the facet number was not increased by selection. The results of these experiments indicate that the hereditary differences in this race of Drosophila are due to a large number of small factors. Crosses made in the sixth generation between low and high lines indicate that the hereditary factors are not sex-linked. In both the long-winged and the vestigial-winged lines the mean facet number was highly variable. This variability appeared to be due chiefly to changes in the temperature of the room, but may have been due also to other causes. The mean facet number of flies reared in vials as single lines was slightly higher than that of stock flies reared in bottles. The mean facet number of the vestigial-winged flies was somewhat lower than that of the long-winged flies. Six full-eyed males and, five heterozygous females appeared in the stock bottles and the selected lines during these experiments. They must be regarded as reverse mutations. The reverse mutation can be explained on the theory of a reversible chemical reaction between two compounds one of which is more stable than the other. The present facts, however, favor an explanation on the basis of partial non-disjunction.