Currently favoured models postulate that gene conversion is due to the correction of mis-matches in heteroduplex DNA. If heteroduplex is formed reciprocally on both chromatids participating in recombination, the mis-matches due to a heterozygous site will be different on the two chromatids, and there will be four correction probabilities to be taken into account. It is shown that, given the frequencies of the five different kinds of aberrant ascus ratios, it is possible to calculate four alternative sets of values for the four correction probabilities and the total number of asci in which heteroduplex is formed. These four solutions reduce in effect to two when there are no other markers distinguishing the two chromatids. With the aid of flanking markers and the assumption that heteroduplex formation is chemically polarized, it is possible, in principle, to choose one best solution.
The method has been applied to the five one-point crosses in Sordaria fimicola from which most data are available. The data from four different mutants crossed to wild type are compatible with a restricted model in which the correction frequencies, from mutant to wild and from wild to mutant, are the same on both chromatids. In the case of the fifth mutant the data are not consistent with this restricted model, and indicate different correction frequencies in the two chromatids.