Perhaps the most difficult step to explain in the origin of life is that from the replication of molecules (RNA for example) in the absence of specific proteins, to the appearance of polymerases and other proteins involved in the replication of RNA and themselves coded for by that RNA. Suppose we start with a population of replicating RNA molecules. Without specific enzymes the accuracy of replication is low and hence the length of RNA which could be precisely replicated small. Before replication can be reasonably accurate, there must as a minimum be a specific polymerase, as well as synthetases and tRNAs, which in turn implies an RNA genome of considerable length. Thus, even if one supposes an initially very limited set of codons, one cannot have accurate replication without a length of RNA, say, 2000 or more base pairs, and one cannot have that much RNA without accurate replication. This is the central problem discussed in a series of papers by Manfred Eigen and Peter Schuster proposing the ‘hypercycle’ as a necessary intermediate stage.