Abstract
Variation in recombination frequencies may lead to a bias in the estimated interference value in a linkage experiment. Depending on the pattern of variation, the bias may be toward negative interference or toward positive interference, even when there is positive interference at the cytological level. In this paper we have mainly concentrated on the case of negative interference. We use models to quantify this effect when data are derived from a backcross experiment or from the selfing of F1 individuals. The effect is quantitatively similar in the two cases. There is an upper limit to the size the bias may reach for every given level of recombination. Two reported cases of negative interference in Drosophila and cultivated barley fall within this possible parameter range, i.e., the observed negative interference values could--at least in principle--be due solely to a variation in the recombination frequencies in the experiments.