THEORIES of the development of mammalian hepatic haemopoiesis have polarised around two viewpoints. One hypothesis suggests that multipotential haemopoietic stem cells (HSC) migrate from the yolk sac and colonise the developing hepatic primordia¹. According to the other view, hepatic haemopoietic tissue arises from the transformation of liver mesenchyme cells and thus has no direct relationship with vitelline haemopoiesis². To solve this developmental problem, it is necessary to establish techniques, either in vivo or in vitro, to maintain hepatic tissue in a foetal state. In vitro studies have demonstrated that hepatic haemopoietic tissue can develop if tissues are explanted after the 28-somite stage³. Thus, these experiments did not verify the inductive model² of hepatic haemopoiesis, although the in vitro conditions may have curtailed haemopoietic induction without altering hepatic differentiation. With an exogenous source of yolk-sac-derived HSC attempts have been made to colonise pre-28-somite hepatic explants in vitro, using the appearance of granulopoiesis as an indication of successful recombination. These experiments were not conclusive since granulopoiesis was observed with yolk sac cells in vitro in the absence of hepatic tissue (G.R.J., unpublished).