Abstract The specific combining region of an antibody molecule has been assumed to be a fixed property of the molecule. However, this assumption is unlikely if the antigen plays an instructive role in antibody formation. A simpler model of the specific site must be considered. Model building experiments are reported which indicate that a small peptide, devoid of fixed three dimensional structure in the absence of the antigen, can distinguish between closely related haptens. These experiments suggest that antibody specificity may reside in such a peptide.