The clonal selection theory proposed by Burnet required a genetic process, for which there was then no precedent, which randomizes the region of the gene(s) responsible for the specification of gamma-globulin molecules. Work over the subsequent half-century substantiated Burnet's speculation, revealing two distinct novel genetic processes. During early development (when Burnet first thought the randomization took place) programmed gene segment rearrangement catalysed by the RAG1/RAG2 recombinase generates a substantial diversity of immunoglobulin molecules (the primary repertoire). Somatic hypermutation (triggered by the activation-induced deaminase (AID) DNA deaminase) then occurs following antigen encounter in man and mouse, yielding a secondary repertoire. This hypermutation allows both limitless diversification as well as maturation of the antibody response by a process of somatic evolution akin to that envisioned by Burnet in later formulations of the clonal selection theory. AID-triggered antigen receptor diversification probably arose earlier in evolution than RAG-mediated repertoire generation. Here I trace our insights into the molecular mechanism antibody somatic mutation from when it was first proposed through to our current understanding of how it is triggered by targeted deamination of deoxycytidine residues in immunoglobulin gene DNA.