On the topography of potential-energy surfaces for heavy-ion reactions