Affimty maturation of the bumoral immune response by somatic bypermutation is marked by a rapid and dramatic increase in affinity for the eliciting antigen. We suggest that tbe optimal mutation schedule is one in whicb periods of rapid mutation alternate with periods of mutationfree growth. The multicompartmental structure of the gerramal center, together with re-entry of positively selected B cells back into the germinal center, will naturally implement such a schedule, thereby prowding an anatomical basis for the efficiency of the germinal center reaction.

Affinity maturation through somatic hypermutation of B-cell immunoglobulin (lg) V-region genes 1 is a remarkably efficient 0rocess of mutation and selection by which high-affinity antibodies are generated. Its fundamental role m the immune response to foreign antigens has become increasmgly clear in the past few years-'-7. While our knowledge of the structure of the germinal centers (GCs), the loci of affinity maturation, and the kmencs of the germinal center reaction (GCR) has increased substantially 8-13, an anatomical basis for the impresswe efficiency of the GCR remains unexplored. Several theories detailing events at the molecular level have been advanced to account for the efficiency of affinity maturation t4-~ 6. Our approach, in contrast, assumes only ordinary mo', ecular mechanisms: diversity generated through random point mutations and da.'winian selection based solely on the ability of variant Ig to brad antigen. To explain the high efficiency of somatm mutation, we suppose that the mutation rate can be regu-lated 3s'v'lTas and we seek the mutation schedule [//(t)] that maxamizes, at the end of the GCR, the quantity: