We study the time required for speciation in a species that is divided
into small versus large populations. Following Dobzhansky and Muller,
we assume that hybrid sterility or inviability is caused by ''complem
entary genes,'' that is, by the accumulation of genes that cause steri
lity or inviability when brought together in hybrids but that have no
deleterious effect on their normal species genetic background. When di
vergence between populations is caused by genetic drift, we show that
the time to speciation is independent of population subdivision: speci
ation occurs just as quickly in a species split into a few large popul
ations as into many small populations. When divergence is driven by na
tural selection, however, the time to speciation is very sensitive to
population subdivision and speciation occurs most rapidly when a speci
es is split into two large populations. These results contradict sever
al popular intuitions about the effect of population size on speciatio
n.