The work is divided into three sections. Section I deals with a cytochemical and cytological investigation into the nuclear changes which occur with the induction of heritable changes in flax. Section II is concerned with the genetical basis of variation between genotrophs and epitrophs (selection lines) of the flax variety Stormont Cirrus. Section III is a report of an investigation of the pattern of DNA synthesis at the chromosome level. It was originally intended to carry out this study on flax, but in view of very small size and the large amount of heterochromatin in chromosomes of this species, it was felt that very little information would be gained. Consequently, Scilla campanulata a species with large chromosomes and no large blocks of heterochromatin was chosen for this study. Feulgen photometry measurements showed significant DNA differences between large and small genotrophs of two flax varieties, Stormont Cirrus (16iß difference) and Lyral Prince (iC difference). The plastic genotroph had intermediate values in both cases. There were no significant nuclear DNA differences between eight commercial varieties of flax investigated although there was some indication that two linseed varieties had slightly more nuclear DNA than the flax varieties. Nuclear DNA measurements of plants of the plastic genotroph of Stormont Cirrus grown in inducing environments indicated that the changes occurred during the first five weeks of growth and were completed prior to flowering. Since the DNA difference is evident at metaphase of mitosis it is concluded that the DNA variation involves the chromosomes. Such DNA variation must be spread over many chromosomes within the complement because there is no detectable asymmetry between bivalents at M1 in LxS hybrids. Further evidence of chromosome divergence was obtained from a comparison of the pattern of chromosome configuration at MI of meiosis in p. m. c. 's of autotetraploids of large and small genotrophs and in the reciprocal crosses between them. It was found that the frequency of quadrivalents was lower in the crosses than in the two parental genotrophs, the converse being true of bivalent frequency. This variation in the frequency of both quadrivalents and bivalents was shown to be independent of chiasma frequency and is exactly what would be expected if a small divergence in chromosome structure had influenced the 'pairing' pattern at prophase of meiosis in the autotetraploid crosses.