The available data on heterogeneity in centromere distances for a number of loci in several linkage groups are analysed and interpreted. When the crosses are grouped according to wild-type ancestry, heterogeneity is eliminated in any one group except those which consist of backcrosses or intercrosses. Abbott 4 and 12 are shown to be the source of the heterogeneity while Lindegren and probably Chilton wild strains give consistent, homogeneous distances. In a cross between Abbott 12 and Lindegren wild-types, the centromere distances of mt and asco show heterogeneous values between the spore pairs in an ascus indicating that significantly different distances are genetically determined and that the factors concerned show segregation. The genetic determination differs in the various wild strains; the data suggest that at least three factors are involved. In random spore analyses heterogeneity is present in recombination frequencies between linked markers either proximal or distal to their centromere. The mechanism by which heterogeneity in the data might arise is discussed. To obtain homogeneous data it is suggested that all markers used should be repeatedly backcrossed to the Lindegren wild-type.