Theory for leaf patterning, phyllotaxis, is usually expressed in terms of interactions in the surface of the apical dome of the shoot. Mechanisms for leaf formation, however, usually relate to phenomena in the longitudinal section, e.g. periclinal divisions. Studying epidermal cell file patterns and the directionality of cellulose in the outer walls of the dome we have found distinct patterns of cells and reinforcing cellulose on the surface. Changes in the epidermal pattern correlate with the phyllotactic sequence to suggest that: Recently established leaves are associated with fields of tangential cellulose reinforcement which extend toward the apical dome. Where such aligned fields come into contact so as to generate relatively abrupt angular changes in reinforcement pattern, a new leaf will appear. As this region of discontinuity develops into a hoop-reinforced structure, the visible primordium, a new single field of tangentially aligned reinforcement is generated. The new field interacts with other fields to continue the cycle. In whorled phyllotaxis two angular discontinuities appear to be involved with each new leaf; the pertinent older leaves are just one plastochron older than the leaf being initiated. In spiral phyllotaxis (3:2) a single angular discontinuity appears to be involved initially; the pertinent older leaves are three and five plastochrons older than the leaf being formed. There are two major differences from previous theories of phyllotaxis. First, the cyclic changes in leaf initiation are thought to be based on the constructive involvement of new leaves in modifying the dome's reinforcement pattern. Many theories assume that leaf formation is spontaneous and the role of nearby leaves is inhibitory. Second, the pattern sequence can explain the concurrent appearance of reinforced stem tissue along with leaves.