INTRODtJCTION.. JINKs and JONES (1958) showed that the additive genetic, [d] and the additive x additive, [i] and additive x dominance, [j] interaction components of means estimated from the means of two pure-breeding lines (P1 and P2) and of their F1, F2, B1 and B2 families are functions of the degree of association (r) of alleles of like effect in the parents. Their real magnitudes and signs are therefore revealed only when the alleles of like effect are completely associated (r= 1) in the parents. In order to obtain a true picture of the gene action and interaction in Nicotiana rustica we have therefore investigated two crosses between pure breeding lines in which there is a high degree of association for the genes controlling most of the characters of interest.
MATEPIALThe experiment consists of the P1, P2, F1, F2, B1 and B2 families of the cross between varieties 2 and 12 and between two of the D inbred lines, D10 and D17. Varieties 2 and 12 are two of the most extreme in the Birmingham collection. They display a high level of non-allelic interaction and produce the most heterotic F1 (Jinks, 1954). D10 and D17 are the extreme selections for final height from a random sample of 60D inbreds derived by single seed descent from the F2 of the V2 x V12 cross (see Jinks et a!., 1977). The coefficients of gene association/dispersion in these two pairs of parents estimated as described by Jayasekara and Jinks (1976) as the ratio of the difference between the parental means and the difference between the extremes of the random sample of D inbred lines for each character are as follows;Character "r" for V2 x V12 "r" for D10 X D7 Height-i, H1