The investigated material consists of 11 original plants collected in
the neighbourhood of Lund; four 1, families, totalling 18 plants (rang
e 3-8 plants); four F1 families, totalling 52 plants (range 10-18 plan
ts); and four backcross families, totalling 311 plants (range 60-87 pl
ants). Self-incompatibility prevailed both in original and in inbred a
nd crossbred plants, as judged from seed-setting data in situ under gr
eenhouse conditions. Of the intercrosses (no complete diallels) among
I1 and F1 siblings, 13.8 % and 0 %, respectively, were judged to be in
compatible. Parental plants pollinated by backcross families (P x BC,
308 combinations tried) had 4.2 % cases of cross-incompatibility. Calc
ulations were applied to these data, aiming at minimum degrees of cros
s-incompatibility to be expected in different intercrosses at differen
t numbers of S-loci, considering ( 1) pure disomic inheritance, (2) an
autotetraploid with chromosome or (3) random chromatid assortment, (4
) an amphidiploid with two S-gene systems brought together, acting ind
ependently of one another, or (5) interdependently, with functional in
compatibility specificities being formed through interaction not only
within but also between the two systems. It was concluded that the dat
a indicate an S-gene system with disomic inheritance of one team of at
least 4 loci with complementary cooperation. To explain the presence
of such an S-gene system in a tetraploid, it is speculated that self-i
ncompatibility genes on a polysomic level are accompanied by too high
degrees of cross-incompatibility to be retained in the long run and ha
ve gradually to adopt disomic formation of incompatibility specificiti
es.