The amphibian nervous system has long been thought to arise from within a large (greater than 10(3) population of dorsal ectodermal cells, that would otherwise differentiate only as epidermis, as a result of inductive signals from underlying dorsal mesoderm at gastrulation. It has recently been claimed, however, that small cell groups are set aside much earlier in amphibian development, as the sole founders of particular portions or 'compartments' of the body plan. This would imply a dramatic re-interpretation of classical experiments where a second dorsal blastoporal lip, grafted to the ventral side of a gastrula containing 10,000 or more cells, causes there the development of a second central nervous system (CNS) in host tissue. We show that such surgically induced second nervous systems are made by host cells which have lineages separate from those that contribute to the original CNS from the 32-cell stage. Thus neural induction can occur as traditionally supposed, by assignment of ectodermal cell fate in relation to dorsal mesoderm during gastrulation. We discuss the implications of this for the recent proposals about early compartmentation in vertebrate development.