Zimbabwe culture before Mapungubwe : new evidence from Mapela Hill, south-western Zimbabwe

Abstract

Across the globe, the emergence of complex societies excites intense academic debate in archaeology and allied disciplines. Not surprisingly, in southern Africa the traditional assumption that the evolution of socio-political complexity began with ideological transformations from K2 to Mapungubwe between CE1200 and 1220 is clouded in controversy. It is believed that the K22Mapungubwe transitions crystallised class distinction and sacred leadership, thought to be the key elements of the Zimbabwe culture on Mapungubwe Hill long before they emerged anywhere else. From Mapungubwe (CE1220–1290), the Zimbabwe culture was expressed at Great Zimbabwe (CE1300–1450) and eventually Khami (CE1450–1820). However, new fieldwork at Mapela Hill, when coupled with a Bayesian chronology, offers tremendous fresh insights which refute this orthodoxy. Firstly, Mapela possesses enormous prestige stone-walled terraces whose initial construction date from the 11th century CE, almost two hundred years earlier than Mapungubwe. Secondly, the basal levels of the Mapela terraces and hilltop contain e´ lite solid dhaka (adobe) floors associated with K2 pottery and glass beads. Thirdly, with a hilltop and flat area occupation since the 11th century CE, Mapela exhibits evidence of class distinction and sacred leadership earlier than K2 and Mapungubwe, the supposed propagators of the Zimbabwe culture. Fourthly, Mapungubwe material culture only appeared later in the Mapela sequence and therefore post-dates the earliest appearance of stone walling and dhaka floors at the site. Since stone walls, dhaka floors and class distinction are the essence of the Zimbabwe culture, their earlier appearance at Mapela suggests that Mapungubwe can no longer be regarded as the sole cradle of the Zimbabwe culture. This demands not just fresh ways of accounting for the rise of socio-political complexity in southern Africa, but also significant adjustments to existing models.