<:1 From Being Female& Reproduction, Power, Change Dana Raphael, ed. The Hague& Mouton (1975) pp. 99-108. An Explanation for Matrilocal Residence WILLIAM TULlO DIVALE It has been over a century since the publication of Bachofen's Das Mutter- recht where the issue of matrilineal systems was first raised. The early theorists, McLennan, Tylor, Morgan, and Engels, all argued that the matriliny was a stage prior to the patriliny in the evolution of the family. Evidence for their arguments was based, in part, on scattered matrilineal traits, such as the avunculate, found in many patrilineal societies. With the exception of a few scholars in Germany and the U.S.S.R. this per- spective is no longer held by many anthropologists. This was partly due to the Boasian reaction against the 19th century evolutionists, and partly to the data gathered on peoples of simple complexity which more often indicated patrilocality and bilateral or patrilineal descent. To date no rival hypothesis has been offered to explain the matrilineal survivals nor have we seen satisfactory explanation for the occurrence of matrilocal residence. For instance, the most generally accepted explanation for matrilocality is that the sex which contributes most to basic subsistence will determine post-marital residence practices. Thus if females predominate in sub- sistence activities, residence will be matrilocal. Another theory, postulated by David Aberle and Elman Service, is that tropical rainforest environ- ments are associated with matrilocal-matrilineal peoples. Kathleen Gough has suggested that societies with matrilineal descent are predominantly horticultural rather than agricultural. She argues further that among matrilineal societies, those with matrilocal residence will exhibit low productivity while those with avunculocal residence will exhibit high productivity. While patrilineal societies are almost always P.trilocal, matrilineal '·- ., \ I 1) '( ._::) 1 v ·(' ! . lt, :6 .. 'JxAMTULIODIVALE _.J, f. "l 1.J ,,J f\,.;1. 1'\. q-· I societies exhibit a variety of post-marital residence patterns. It seems likely, however, that matrilineal descent could only develop from a matrilocal residence pattern. Logically we might assume that any ex- planation of why 14 percent of the world's societies have matrilineal descent must ultimately come from an explanantion of the causes of matrilocal residence. The purpose of this paper is to provide such an explanation. METHODOLOGY The theories of matrilocality just described, plus the one to be presented below, were tested cross-culturally on a probability sample of 43 societies (33 patrilocal, 10 matrilocal) and when applicable, on the 1200 societies in the Ethnographic atlas (Murdock 1967). The variables used, which were not already in the Ethnographic atlas, were coded once by myself and once by an independent analyst who did not know the hypotheses to be tested. There is no significant difference in test resultS-when .either _ _9lding i!l used (see Divale 1974a for including page and soufceret'erences ). TESTS OF OTHER THEORIES The hypothesis that residence will be matrilocal when females pre- dominate in subsistence was tested using precoded variables on basic subsistence (gathering, hunting, fishing, herding, and farming) from the Ethnographic atlas (Murdock 1967). However, no association was found (see Figure 1). A recent test by Melvin and Carol Ember {1971), as well as another test by Divale (1974b), also failed to find any relationship be- tween female importance in subsistence and matrilocal residence .. The hypothesis that matrilocality is associated with tropical forest climates was also tested using codings on climate for the first 410 societies in the punched card version of the Atlas (Barry 1967), but no association was found between rainforests and matrilocality. In fact, matrilocal residence was tested against fifteen types of environments and no associa- tion was found. And, as a further check, societies with matrilineal descent were also tested against these environments and only two significant but very small relationships were found- namely that matrilineal descent is negatively correlated with temperate forests and positively correlated with temperate woodlands. Both of these correlations are so low, how-