NOTE 19P: In principle, this could be caused by a changing selection coefficient on the F allele itself: strongly negative, then weakly positive, and then approaching zero. However, conditions are supposedly constant, and frequency-dependent selection cannot account for the changing direction of selection: Allele frequency falls rapidly when in the range 0.4–0.1 in the first few generations, but then rises again through the same range of allele frequencies between generations 10 and 40.