Network ideas are confronted with current hypotheses for the origin of antibody diversity and self-nonself discrimination. The difficulties of reconciling the promethean evolution of the antibody system with "germ line" theories are discussed, as well as the problems of "somatic" hypotheses to explain the completeness of the antibody repertoire. The formal incompatibility of the network theory with ideas basing self-nonself discrimination on the elimination of self-reactive cells is demonstrated, as well as the difficulties of these and other environment-dependent hypotheses for lymphocyte activation, to encompass the internal activity in the immune system. It is argued, on the other hand, that the limitations of the network theory in providing a functional basis for the idiotypic network and in accounting for self-nonself discrimination, can be solved by finding in a complete repertoire of antibody-combining sites the complementary structures to growth receptors on B lymphocytes, and by using these as internal mitogens in the expansion of the precursor cell pools and in the maintenance of the mature steady states. Letting self-nonself discrimination be accounted for by such growth receptors, both the integrity of the antibody repertoire and the internal activity in the system can also be ensured. Moreover, by postulating a germ line origin for the antireceptor antibodies and by accepting idiotypic cross-reactivity between growth receptors and other germ line antibodies, the possibilities are set for a phylogenetically and ontogenically autonomous immune system embodied with the capabilities for self-expansion, diversification and selection of available repertoires. Its promethean characteristics are explained by its completeness, and this is achieved by idiotypic interactions between growth receptors and a limited number of complementary or cross-reactive germ line antibodies, naturally selected on the basis of their structural relationships with growth receptors.