The results show enhanced differences in metastatic potential of tight and loose clones after selective cloning and that there may be important differences in motility and cell-substrate interactions. Foulds (1949) introduced the term tumour progression to describe the process by which some tumours become more malignant during their growth. Nowell (1976) attributed this to genetic instability, although genetic or epigenetic processes could result in the emergence of variant cells with growth advantages which differ from the main population in the expression of specific characters (Frost & Kerbel, 1983;Schimke et al., 1985).The existence within tumours of cells heterogeneous for the expression of a wide range of in vivo and in vitro characters is well documented (for reviews see Heppner, 1984;Nicolson, 1984).We have previously reported heterogeneity of clonal morphology in a metastatic melanomaxlymphocyte hybrid tumour cell line (Clark & Sidebottom, 1984). These variants grow either as tight or loose colonies and the isolation of cell lines, stable for clonal morphology, which have low or high metastatic phenotypes has been described (Clark & Sidebottom, 1984). The morphology and patterns of growth of these variants suggests a number of ways in which these cells may differ and which might account for their behavioural differences in vitro. These include intercellular and cell-substrate adhesion and motility, which in turn are dependent upon cell surface properties. Various stages of the metastatic cascade, for example local invasion or tumourcapillary endothelium adhesion are partially dependent upon intercellular and cell-substrate adhesion and motility (Weiss, 1980;Varani et al., 1980;Nicolson et al., 1986). The cell surface has an important role in these interactions and has been extensively studied for correlations with metastasis and organ colonization (Turner, 1982;Nicolson, 1984).Observation and measurement of adhesion and motility in vivo have inherent difficulties, but these characters can be investigated in vitro and some clues about their possible significance in vivo thereby gained (Briles & Kornfeld, 1978; Hart, 1979;Vollmers & Birchmeier, 1983;McCarthy et al., 1985).We have, therefore, looked at the adhesive and motile properties of tight and loose cells and also examined their cell surfaces by lectin binding to PAGE separated glycoproteins, radio-iodination of proteins and quantitation of sialic acid content. (Sidebottom & Clark, 1983). The cell lines 1G8(6F) and 1G3(4E) were selected from F87.CI.6T2 cells for tight or loose clonal morphology respectively and their in vitro and in vivo properties have been described elsewhere (Clark & Sidebottom, 1984). Cells were maintained in MEM +10% newborn calf serum (Flow) (cMEM) and subcultured when confluent; they were regularly tested for mycoplasma and found to be negative. Wiscm cells were derived from explanted hearts from one day old Wistar rats and were maintained in DMEM + 10% foetal calf serum (Gibco) +10% tryptose phosphate broth (Difco). To prod...