The RNA world hypothesis proposes that RNA once functioned as the principal genetic material and biological catalyst. However, RNA is a complex molecule made up of phosphate, ribose, and nucleobase moieties, and its evolution is unclear. Yakhnin has proposed a period of prebiotic chemical evolution prior to the advent of replication and Darwinian evolution, in which macromolecules containing polyols joined by phosphodiester linkages underwent spontaneous transesterification reactions with selection for stability. Although he proposes that the nucleobases were obtained during this stage from less stable macromolecules, the ultimate source of the nucleobases is not addressed. We propose that the purine nucleobases arose in situ from simpler precursors attached to a ribose-phosphate backbone, and that the weaker and less specific intra- and interstrand interactions between these precursors were the forerunners to the base pairing and base stacking interactions of the modern RNA nucleobases. Further, in line with Granick's hypothesis of biosynthetic pathways recapitulating evolution, we propose that these simpler precursors were the same or similar to intermediates of the modern de novo purine biosynthetic pathway. We propose that successive nucleobase precursors formed progressively stronger interactions that stabilized the ribose-phosphate polymer, and that the increased stability of the parent polymer drove the selection and further chemical evolution of the purine nucleobases. Such interactions may have included hydrogen bonding between ribose hydroxyls, hydrogen bonding between carbonyl oxygens and protonated amine side groups, the intra- and interstrand coordination of metal cations, and the stacking of imidazole rings. Five of the eleven steps of the modern de novo purine biosynthetic pathway have previously been shown to have alternative nonenzymatic syntheses, while a sixth step has also been proposed to occur nonenzymatically, supporting a prebiotic origin for the pathway.