The SUP4 tRNATYr locus in Saccharomyces cerevisiae has been studied by the isolation and characterization of mutations at the SUP4 gene which result in the loss of suppressor function. Most of the mutations act as single-site mutations, whereas about a third of the mutations are deletions of the entire gene. Two meiotic fine-structure maps of the gene were made. The first mapping technique placed 10 mutations plus the sup4+ anticodon on a map by a measurement of levels of recombination between pairs of mutations. The second map utilized a more qualitative estimate of recombination frequency, allowing 69 mutations and the sup4+ anticodon to be mapped. The maps were compared with the physical structure of the gene for the 34 mutations whose nucleotide alteration has been determined by DNA sequencing (Koski et al., Cell 22:415-425, 1980; Kurjan et al., Cell 20:701-709, 1980). Both maps show a good correlation with the physical structure of the gene, even though certain properties of genetic fine-structure maps, such as marker effects and "map expansion," were seen.The use of genetic fine-structure mapping to order mutations within a gene has been based on the assumption that the recombination frequency between two mutations is proportional to the nucleotide distance between them. The first system in which this was actually tested was the tryptophan synthetase A gene in Escherichia coli (20). The results of that study indicated that the genetic fine-structure map was indeed colinear with the protein structure.Since that time, it has become clear that finestructure mapping in lower eucaryotic systems is subject to several features that put the accuracy of fine-structure maps into doubt. In yeasts, it has been found that both meiotic and mitotic Xray fine-structure maps show "map expansion" (2, 3, 6, 9), a property initially defined by Holliday (7) in which, given a map order abc, the recombination frequency between ac is larger than that of the summed frequencies ab + bc. This property of map expansion indicates that the assumption that recombination frequency is proportional to DNA distance is not strictly true. Fine-structure mapping data also often show marker effects, in which a particular marker may give unusually high or low levels of recombination with other markers (3, 6). Such marker effects will also lead to inaccuracies in the fine-structure map. Finally, an extensive study by Moore and Sherman (13,14) indicated that the level of recombination per nucleotide could vary extensively between different pairs of mutations. These studies utilized a large number of mutations within the Saccharomyces cerevisiae cycl gene for which the physical location of the mutations were known. Even two different mutations at the same site could give quite different levels of recombination with another mutation. This effect was significant enough to lead to improper ordering of mutations with respect to one another in some cases. Although this effect was shown to be significantly less extreme for meiotic recombination frequen...