The study of the genetic determination of longevity in Drosophila melanogaster has made use of the technique of late-age reproduction. At low larval density, that indirect selection showed no effect. At high larval density, however, increased mean life-span in lines reproduced at late age was observed. When these last data are examined as a function of the number of days after the beginning of the experiment, instead of as a function of generations, the difference in life-span between early and late lines at high larval density disappears. The erratic evolutions of mean longevities in all the experiments here described may be attributed to unexplained variations in life-span previously observed in a 4-year experiment. Considering the time lag between the measurement of a given generation in early and late lines, the experiments of reproduction at late age cannot demonstrate the genetic determinism of longevity.