Three major theories of cancer--somatic mutation, virus causation, and faulty differentiation--are proposed to involve alterations in DNA structure. Each results finally in terms of failures in the normal intercellular communication that involves feedback between differentiated cells acting on less differentiated cells still capable of proliferation. The historical background of the latter idea is traced to Osgood, Weiss and Kavanau, and to Iversen. The historical background of concepts of initiation and promotion are traced to Berenblum and Mottram and the Boutwell concept of promotion as gene activation is cited. It is proposed that gene activation by promoters is a valid concept and that it results from the blocking of the normal intercellular communication postulated by Osgood and others. The problem of explaining the low probability of cancer following initiators or promoters acting alone is cited as a problem in basic science. A hypothesis to solve the problem is proposed: Cancer results from two of more relevant mutations: promoters enhance proliferation of cells with one relevant mutation, thereby increasing the probability of obtaining a cell with two relevant mutations. A new scheme of five stages of hepatocarcinogenesis is proposed in terms of the hypothesis and available data.