Estimations of random mating frequency were computed for a series of sequential autumn sowings of populations of winter barley . The estimations were by means of the maximum likelihood scoring method and three varieties, each carrying a recessive genetic marker, provided three independent estimations for each population . High levels of out-crossing were found in sowings made early in September, and a trend towards absence of out-crossing in populations sown in late November was evident . Although fluctuations in this trend could be accounted for by fluctuations in meteorological factors, a more profound underlying effect was evident in that the three marker varieties behaved similarly, though to different degrees . It is postulated that the principle effect of sowing date, as it affects out-crossing, is upon the development of the flowering apex, and that environmental conditions at flowering time merely modify flowering behaviour .