Genetic Analysis 269and by Lederberg [Lederberg, 1947; Lederberg and Tatum, 1946] for a process of
sexual (or parasexual) mating and recombination of genetic factors in the K-12
strain ofEscherichia coli, convinced geneticists that bacteria might be amenable to
“classical” genetic analysis along similar lines as have been carried out in eukary-
otes. The feasibility of genetic analyses, such as the construction of linkage maps,
was taken as sufficient evidence for chromosome-like existence also in prokaryotes
[Lederberg, 1947; Lederberget al., 1951].
Two presumably unrelated observations eventually challenged the mechanistic
notion of recombination: a. When recombination between very close markers was
examined, it turned out that by far too many double recombination events oc-
curred; instead of the expected increase of positive interference with decreasing
distance, in very short intervals,negative interference was observed. b. When
products of individual meiotic events could be followed, as became possible with
“tetrad analysis” in fungi, infrequent but consistent deviation from the expected
1:1 recovery of the two alleles for which the parent was heterozygous turned up.
This phenomenon of “gene conversion” — for a long time suspected to be an exper-
imental “contamination” — was found to be often associated with recombination
of outside markers. Thus, instead of finding that a heterozygoteA/aproduced
4 Aspores and 4aspores in the ascus (spore-sac, containing the products of a
single meiotic division, followed in many species by a mitotic division) ofNeu-
rospora, some asci contained 6A:2aascospores, or even 5A:3aascospores, whereas
the nearby outer markers (which often were found to recombine) segregated nor-
mally, giving 4A:4aascospores [Lindegren, 1953; Mitchell, 1955].
The results with recombination in bacteria “illustrated the experimental concor-
dance of bacterial segregations to a generalized definition of mendelism,” although
“[i]n a purely formalistic way, these data could be represented in terms of a 4-armed
linkage group.” Formally a branched linkage map emerged, however, Lederberget
al.added: “without supposing for a moment that this must represent the physical
situation” [Lederberget al., 1951, 416-417]. Yet, consistent non-branched linear
linkage maps were constructed once linkage maps were derived by measuring the
time of transferof marker genes in mating different Hfr (high frequency recom-
bination) “male” bacteria to an F−“female” bacterial strains. The virtual maps
derived from different Hfr stocks proved to differ merely in the permutations of
the same linkage relationships. It was concluded from the genetic analysis of such
matings that the bacterial chromosome was a closed circle, which was “opened” at
different sites in different Hfr stocks [Hayes, 1964; Jacob and Wollman, 1961]. This
was eventually confirmed by cytological observations of the physical organization
of the DNA ofE. coli[Cairns, 1963].
The mechanical chiasmatype notion of recombination assumed that exchange
occurred precisely at the same site in the two chromatids involved, since as a rule
both products of recombination were viable. This, however, was not strictly neces-
sary at the DNA molecular level. It was Whitehouse who first suggested a model
of crossing-over, still based on breakage and rejoining of chromatids, in which the
breakage points in the two strands of each of the exchanging DNA molecules were