A review of the data available for settlement pattern development in the mountainous Mani peninsula of southern Greece between 400ooo-300ooo0 B.C. and A.D. 1830 reveals the existence of two distinct spatio-temporal unfoldings separated by a major discontinuity. This is located in the period between about A.D. 550 and 1250. Possible explanations involving factors external to the system are considered and discounted. An alternative endogenous explanation for settlement pattern change is offered by postulating deterioration in the quality of agricultural land in part of the region. This is then used as a control variable, together with defence, in a model of change derived from Catastrophe Theory. It is argued that, despite problems over data, the resulting folded surface provides a better insight into the original morphogenetic problem than statistical models derived for patterns of continuous development. THE PROBLEM RESEARCH into the historical geography of the southern Mini, a mountainous and arid peninsula of the Pelop6nnisos, Greece (Fig. i), has revealed a major discontinuity in the unfolding of the region's settlement pattern over some five or six thousand years. Human occupation of the region began during Neolithic times (4000-3000 B.C.)' and, though the period is long and the evidence often slight, the steady evolution of a particular settlement pattern can be traced through to full development in the Roman period.2 By the late second century A.D. there were at least 23 settle- ments in the region, compared with perhaps four in the early Bronze Age (300ooo-1900 B.C.), and the distribution shows a concentration in the hill and valley country of Mal6vri in the north-east (Fig. 2). Although some settlement names are found in documents produced later than the second century and more than 50 medieval churches are known, the evidence is insufficient for recon- structing anything like a complete pattern of settlements before the opening decades of the seventeenth century (Fig. 3).3 The next two centuries saw the elaboration of the early seventeenth- century pattern, first by infilling and then by recolonization of Malevri, as the number of settle- ments rose to 129 by about 1830 and a maximum of some 225 later in the nineteenth century.4 Comparison of the late second-century pattern with that of the early seventeenth century reveals elements of both change and stability. Stability is apparent in continued occupation in the vicinities of ancient Kaenepolis, Oetylus, Pyrrhichus and Teuthrone, though there were shifts of site. The dominant impression, however, is one of change. Settlements increased in number from 23 to 61. The total within 0-5 km of the present coastline declined from thirteen at the end of the second century to five in the early seventeenth century. More marked was the disappearance of settlements from Mal6vri and a marked increase on the high and arid marine terraces of the west and south-west. Largely in consequence of these horizontal changes, the height zonation of