Maximum likelihood (ML) estimations of gene frequency,deduced from dat
a on small pedigree samples, are considered. It is demonstrated that a
range (''gap'') of possible values of gene frequency exists, within w
hich no ML frequency estimation falls. The dependence of the size of t
his range on sample size is studied in samples consisting of nuclear p
edigrees and pedigrees of a more complex structure. The practical impo
rtance of the effect revealed is discussed.