Ten years ago, Walter Gilbert contemplated “an RNA world containing only RNA molecules that serve to catalyze the synthesis of themselves” [ 11: in other words, selfreplicating systems that function in the absence of protein and DNA. A recent comprehensive monograph describes the current understanding of the above concept and surveys recent progress concerning the biochemistry and molecular biology of RNA [Z]. Here I highlight some facts about the role of the four RNA bases in contemporary intermediary metabolism, and then offer some thoughts on what implications these facts have for the evolution of our present DNA/protein-based world from the postulated RNA world, a perspective rarely mentioned in the discussion of the RNA world’s central themes such as replication. We must start by looking at research at the beginning of the second half of this century, when biochemists began to focus on the monomeric precursors of the larger cell constituents: fats, carbohydrates, proteins and nucleic acids. Before the advent of isotopic tracers, these syntheses were thought to be simply a reversal of the corresponding lytic processes. How naive, in retrospect! Eventually, in the 1950’s, these views became untenable, not because they were rejected as unreasonable but because of chance discoveries pointing to obligatory monomer activation by nucleotides and their derivatives prior to polymerization. The critical event in several of these chance discoveries was reagent contamination. Accidental discoveries Although it was recognized early on as the universal source of chemical energy, ATP, commercial or home-made, proved to be of dubious purity. No one initially suspected that preparations of ATP might contain other nucleoside triphosphates. This unwarranted confidence came to an end in the 1950’s with dramatic consequences. At the time Kornberg and Pricer [3] and Kennedy’s laboratory [4] independently investigated the biosynthesis of phospholipids, but with divergent results. Both groups examined the energy-dependent conversion of precursors to phosphatidylcholine. In the Kornberg laboratory, amorphous ATP (Pabst) was used as a reagent. Somewhat later, Kennedy and Weiss repeated and confirmed the results obtained in Kornberg’s laboratory, but in addition they tested samples of the newly available crystalline ATP. These ATP specimens gave entirely negative results, pointing to a contaminant in the amorphous ATP as the source of the activity; this was later shown to be CTP. The cytosine-containing nucleotide proved to be the energy source that nature had selected not only for the synthesis of phosphatidylcholine, but also for phosphatidylethanolamine and other membrane phospholipids. The discovery of GTP as an energy source closely resembled the accidental encounter of CTP In studies in the 1970’s on a glycogensensitive adenylic cyclase from rat liver membranes [S], Martin Rodbell found that ATP was required for activation, but in unusally high concentrations. Moreover, the responses were variable depending on the source of ATI? Aware of the experiences of Kornberg and Kennedy, Rodbell traced the variable activities of ATP samples to contamination with GTI? Equally inadvertent, the discovery of the importance of uridine triphosphate came about when Luis Leloir found a thermostable cofactor necessary for the enzymatic breakdown of lactose [6]. By chance, his coworker Caputto found a published ultraviolet spectrum of uridine that proved to be identical with that of the cofactor needed for lactose breakdown, now known as UDP-galactose.