The pattern of variation observed in crosses between certain cluster and "normal" genotypes are not adequately explained by the currently published mutant loci cl1, cl2, and cl3. Segregation observed requires either a new major locus for cluster phenotype, gene modifiers, or multiple alleles. Based on the variation observed in 44 F1 and F2 populations generated from 8 cluster and 2 normal parents a system involvings two active sites at each locus is proposed as the simplest explanation not involving significant contradiction between observation and theory. One site probably controls node length, and the other sympodial node number.