It has been clear for over sixty years that the principal method whereby cells replicate and segregate their DNA is semiconservative. It is much less clear why it should be like this rather than, say, conservative. Recently, evidence has accumulated that supports the hypothesis that one of the functions of the cell cycle is to generate phenotypically different daughter cells, even in nondifferentiating bacteria such as Escherichia coli. Evidence has also accumulated that the bacterial phenotype is determined by the functioning of extended assemblies of macromolecules termed hyperstructures. One class of these hyperstructures is attached dynamically to a DNA strand by the coupling of transcription and translation. Previously, we proposed in the strand segregation model that one set of hyperstructures accompanies one parental strand into one daughter cell while another set of hyperstructures accompanies the other parental strand into the other daughter cell. This epigenetic mechanism results in daughter cells having different phenotypes. Here, I propose that one of the reasons why semiconservative replication has been selected is because it allows the generation of a population containing cells with very different growth rates even in steady-state conditions.