On the topological nature of Volterra's theorem