Polynucleated urban landscapes

City systems show a degree of resilience and persistence that has rarely been emphasised in urban theory. There is a fascination for recent and contemporary change which suggests that phenomena such as the rise of the 'edge city', for example, comprise the predominant forces determining how a polynucleated landscape of cities is emerging. We argue here that such explanations of polynucleation are largely false. Urban settlement structures from much earlier times are persistent to a degree that is extraordinary. We show this in two ways: first, from empirical evidence of stable rank-size relations in the urban settlement system for Great Britain over the past 100 years; and, secondly, from simulations based on weak laws of proportionate effect which produce aggregate patterns entirely consistent with these empirical relations. We then propose various spatially disaggregate models of urban development which generate an evolution of polynucleated settlement from initial, random distributions of urban activity. These models simulate the repeated action of agents locating and trading in space which illustrate how early settlement patterns are gradually reinforced by positive feedback. These produce lognormally distributed settlement structures that are characteristic of city systems in developed countries. In this way, we begin to explain how aggregate urban structures persist in spite of rapid and volatile micro change at more local levels of locational decision-making. Polynucleated urban landscapes are clear evidence of this phenomenon.