If I have now made clear both the difference in the nature of the basic concepts employed, and the relatively limited way in which I have attempted to apply them, it may be interesting to consider briefly whether any part of the nineteenth-century theories can be salvaged by reconstructing them in terms of the new concepts. It seems to me that Bachofen and his followers (notably MacLennan, J. H. Morgan and, in the present century, Briffauk and Thomson), even if they were wrong in supposing that societies pass through specific phases of development, must be credited with perceiving that there are three main patterns, or model types, of social organization and that these three patterns are associated with (and, in fact, caused by) three distinct patterns of “family” organization. These three patterns are (I) “group” marriage, in which each child feels itself to have a multiplicity of parents, (2) the maternal family, in which each child feels itself primarily under the control of, and related to, its mother, with male influence somewhat indirect, and (3) the paternal family, in which the child feels itself primarily under the control of, and related to, the father, with the mother in a subordinate position. There are, of course, many intermediate forms, notably polyandry and polygamy, where the child feels itself to have a multiplicity of parents of one sex, but only one of the other, but there is no need to go into these various complications here. (27)
The question whether these three family patterns tend to succeed each other in a particular order remains for the moment without a satisfactory answer. The weight of probability seems, at the moment, against their doing so, but the case has still not been adequately investigated; Bachofen and his followers were very properly criticized for adopting a defective method: they took data from “primitive” societies and assumed that because their culture was simple they were in some evolutionary sense “earlier“. We realize now that many of these so-called primitive societies have undergone a long evolution, and cannot be regarded as providing evidence of earlier evolutionary phases. But this is a negative argument; what is needed is a comprehensive study of the order in which these phases have, in fact, succeeded one another in a large number of specific societies. Such a study is, of course, very difficult since primitive societies are costly to study, have few records, and are everywhere being distorted in their development by the impact of Western culture.
However, the present theory provides a tool with which the social evolution of prehistoric periods may be explored rather more readily than at present. For it seems to be consistently true that matrist societies possess mother deities, patrist societies father deities, and intermediate forms deities of both sexes, provided the deity appears in human form at all — while societies in which marriage is on a group basis appear to favour totemistic or animal deities. Fortunately, most societies retain some traditions about the changes which have taken place in the character of their deities, and often there are physical records in the form of cave-paintings, carvings and so forth, from which the nature of their deities can often be inferred. Since various authorities, notably G. R. Levy, in The Gate of Horn, have shown the great antiquity of totemistic and mother religions, it would seem premature to reject the theory that the paternal family, and hence the patrist society, may have made a relatively late appearance in prehistory.