What is a kinship system?

To anthropologists of a certain antiquity, this question may seem needless, but the decline in kinship teaching in graduate programmes has produced a lacuna in the transmission of knowledge evident to all who broach such terms as ‘Iroquois’ or ‘avuncolocal’ in the unblinking presence of younger scholars. So, at some risk of stating the obvious, kinship systems comprise four intersecting axes (Fig. 1).

Figure 1.

The four axes of a kinship system.

In practice, the history of study has frequently presumed one or another, or various partial combinations, of these axes as fundamental and sidelined the others. Moreover, this tendency even co-occurs with some national anthropological traditions: e.g. terminology in the United States; descent in Britain; and alliance in France. Some have sedulously argued that the four axes are mutually independent and can only be treated as such if there are to be analytical advances (e.g. Barry Reference Barry2008). But thoroughgoing ethnographic comparison shows their persistent patterns of mutual influence and determination, and while treating each separately may enable discrete conclusions, I contend that more comprehensive explanation requires conjoint treatment of the four axes as much as possible. Of the four, the most definitive feature is typically the terminology or ‘nomenclature’, a structure of terms in a particular language comprising a series of equations, distinctions and reciprocal interplays. Standard ethnography-derived labels for the terminologies (which continue to prevail, notwithstanding efforts to replace them with more neutral descriptors) have themselves come to stand for whole kinship systems (an ‘Iroquois system’, for example). While terminologies are, of course, not available archaeologically, this does not mean reconstructive efforts should simply ignore them, and explanation can benefit by including putative terminologies as hypotheses (see e.g. Whiteley Reference Whiteley, Heitman and Plog2015; and cf. Ensor’s Reference Ensor2013a,b; 2021 proposed material correlates for kinship-system typologies).

Of the six main types of nomenclature, four are ‘classificatory’ in Lewis Henry Morgan’s (Reference Morgan1871) sense, and sociocentric: Iroquois, Crow, Omaha and Hawaiian, which all classify extended kin under the same terms as those in a close family circle. The other two, Eskimo and Sudanese, are contrastively ‘descriptive’ and egocentric, where the links to another relative need specifying from the perspective of a particular individual (‘ego’, or the ‘propositus’) as a node in a network. The classificatory four often coincide with established descent groups, in which the correct term to use for another relative is known not through a descriptive network of genealogical ties, but simply because of his/her ‘clan’ or ‘lineage’ and generation in relation to one’s own, and those of one’s parents et al. (hence the sociocentric aspect). So, Hopi society operates via a Crow terminology, matrilineal descent and matrilocal residence (generally speaking). Everyone belongs to a particular matriclan (e.g. Parrot)—which owns houses, ritual hereditament and some agricultural fields—and beyond that to a more inclusive set of clans (e.g. Parrot, Katsina, Raven, Rabbit and Tobacco in so-called clan-set—a.k.a. phratry—#I in standard Hopi ethnography). If I am Parrot clan, all clans in the set are also my ‘matrilineal’ relatives, for whom I use the same terms as for immediate kin on my mother’s side (despite the absence of known genealogical links among clans in a set). Similarly, my father’s clan (by definition, not my own) is linked to others in a set for whom I use patrilateral (father’s side) kin terms, which are ‘classified’ in the same group as my father, father’s sister, etc. Standardly, kinship positions, or ‘kin-types’, in a matrix are specified by abbreviations. Notations vary, but a common variant uses single letters: P parent; M mother; F father; G sibling; Z sister; B brother; C child; D daughter; S son; E spouse; H husband; W wife. These combine into strings: FBW for father’s brother’s wife, MZS for mother’s sister’s son, and so on. Relationships among kin-types within kin-terminologies exhibit mathematical features, with intricate algebraic properties and generative logics (e.g. Gould Reference Gould2000; Lounsbury Reference Lounsbury and Goodenough1964; Read Reference Read2001). For present purposes, the basic algebraic value of the notational abbreviations serves minimally to distinguish the six major terminology types formally by the equations and distinctions they make among siblings and cousins, as shown in Table 1 (in the table, an = sign means an equation, in contrast to ≠, meaning a distinction; in the kinship diagrams of Figures 2 and 3 below, = means marriage).

Table 1.

Kinship terminologies defined by equations and distinctions of kin-terms used for siblings (‘brother’ [B], ‘sister’ [Z]) and cousins by the speaker (Ego-male or Ego-female).

Figure 2.

The six main types of kinship systems defined by equations and distinctions of kinterms for female siblings and cousins used by male Ego: circles with the same infill (within a single system) represent terminological equations; circles with different infill (within a single system) represent terminological distinctions (e.g. in an ‘Eskimo’ system—typical in modern European societies—Ego uses a different term for his sister than for his cousins, but has a common term for all ‘cousins’ without distinguishing father’s side from mother’s side or whether his link to them is via same-sex or cross-sex relatives of the ascending generation). (After Driver & Massey Reference Driver and Massey1957, diagram 12.)

Figure 3.

Core skewing equations in Crow versus Omaha systems over three generations: Crow [FZ=FZD=FM]: the same kinterm is used for father’s sister, father’s sister’s daughter, and father’s mother; Omaha [MB=MBS=MF]: the same kinterm is used for mother’s brother, mother’s brother’s son, and mother’s father.

Of the four classificatory systems, three—Iroquois (and its Dravidian variant), Crow and Omaha—are further defined by the presence of ‘crossness’: those related via an opposite-sex tie are distinguished from those related via a same-sex tie. The governing equations appear in G⁺¹ (the generation above Ego): M=MZ≠FZ (a mother’s sister is called by the same term as mother, while a father’s sister gets a distinctive term) and F=FB≠MB. These G⁺¹ equations and distinctions entail concomitant patterns in an Iroquois system for Ego’s generation (G⁰): Z=MZD=FBD≠FZD=MBD and B=MZS=FBS≠FZS=MBS (see Table 1); or, siblings are classified with parallel cousins and both are distinguished from cross-cousins. As an Iroquois subtype,Footnote ¹ Dravidian explicitly equates cross-cousin with spouse/sibling-in-law. And here the collocation of terminology with marriage rules comes clearly into view: crossness is ‘a property derived from an alliance structure … it always carries a matrimonial implication’ (Viveiros de Castro Reference Viveiros de Castro, Godelier, Trautmann and Tjon Sie Fat1998, 354).

Such equations and distinctions may be expressed by classic kinship diagrams, as shown in Figure 2 (noting that their symbols—triangle for a male, circle for a female, = for a marriage tie—adhere to the standard format in socio-cultural anthropology since the 1930s: see Wilson Reference Wilson2018—differing from conventions used by archaeogeneticists to represent biological relationships; here, the emphasis is on social descent and genealogical interrelationships, which may or may not correlate with biological descent and ties). To Iroquois crossness, Crow and Omaha add lineal, intergenerational equations, with additional implications for alliance (see below).

As regards the marriage axis of kinship structures, ‘elementary’ alliance, per Lévi-Strauss, predicates intergroup exchange of spouses in direct, symmetrical reciprocity (A⇔B) or indirect reciprocal circulation (A→B→C→D→A). Direct exchange often co-occurs with named exogamous descent groups: notably, moieties (Ravens versus Eagles, Earth versus Sky, Red versus White, etc.) some internally differentiated into constituent unilineal clans and lineages, in some cases also with distinctive (often ‘totemic’) names. Members of one moiety ‘give’ their siblings to the other moiety in exchange for spouses. In Lévi-Strauss’s original formulation, this was ‘sister-exchange’,” but if the gender bias is removed (sibling-exchange) the structural effect remains the same. Some systems, however, prohibit direct exchange and, extending beyond binary divisions, prescribe cycles of marriages among four exchanging groups: women of group A marry men of group B, B women marry C men, C women marry D men, and D women marry A men, completing the circle. This indirect pattern produces a more intricate, though still elementary, system of alliances termed ‘asymmetric matrilateral cross-cousin marriage’ (e.g. Leach Reference Leach1951). Both direct and indirect exchange are framed through dualistic kinship terminologies that assign kinship positions to all members of society, i.e. producing a total kinship matrix that serves as the primary language for the arrangement of social relations. The terminologies commonly termed Iroquois (and Dravidian) specify cousins as ‘kin’ if based on ‘parallel’ (same-sex) connections, but ‘affines’— in-laws or marriageable categories—if connected via ‘cross’ (opposite-sex) ties. A father’s brother’s child or a mother’s sister’s child are ontologically the same as siblings in this schema and off-limits to marriage: indeed, the kin-terms for parallel cousins in a particular language are the same as those for siblings—a ‘brother’ also means a male parallel cousin, for example. But cross-cousins, a father’s sister’s child or a mother’s brother’s child, are not only acceptable but expected, indeed ‘prescribed’ as marriage partners. More than close family ties, Iroquois and Dravidian kin-terms are extended to more distant relatives: i.e. they are ‘classificatory’ in Morgan’s sense, so that one has many ‘siblings’ and many (marriageable) ‘cross-cousins’, not just the first-cousin variety, for whom there may be a de facto practice of avoidance (in Haudenosaunee society, for example, where Morgan first became aware of these systems) in preference for the more wide-ranging socio-political alliances that exchanging with more distant relatives entails.

Crow–Omaha

Descent and alliance theories persisted through the late 1960s at least, although both came increasingly under assault. Lévi-Strauss’s concentration was shifting away from social structure, particularly toward myth. His intention had been to complete a similar comparison of ‘complex systems’ to his 1949 magnum opus on elementary. But he began to realize he would first have to address an intermediate form, ‘semi-complex’, correlative with Crow–Omaha kinship systems (e.g. Lévi-Strauss Reference Lévi-Strauss1966). Crow and Omaha retain some aspects of Iroquois dualism but create different restrictions on direct exchange to the alternative offered by asymmetric prescriptive marriage. Crow–Omaha semi-complex rules disperse alliances away from a closed binary structure, implying an (evolutionary) egress from elementary constraints (e.g. Whiteley & McConvell Reference Whiteley and McConvell2021). Crow and Omaha skew some kin-terms inter-generationally: down the father’s matriline for Crow, the mother’s patriline for Omaha (Fig. 3). A father’s sister (FZ) is called by the same term as a father’s sister’s daughter (FZD), and a father’s sister’s daughter’s daughter (FZDD, not shown on Figure 3) in a Crow system (e.g. for Hopi, ikya, generally translated ‘my aunt’—thus including ‘aunts’ two or three generations younger than Ego); often this extends upwards to include the father’s mother (FM). In an Omaha system the mother’s brother (MB) is called by the same term as the mother’s brother’s son (MBS) and mother’s brother’s son’s son (MBSS, not shown on Figure 3); again, often with an upward extension to include mother’s father (MF). The ‘basic sex’ (Hammel Reference Hammel1965) of a Crow structure is female (i.e. a matriline), while that of an Omaha structure is male (a patriline).

Dravidian, or a similar ‘tetradic’ variant (e.g. Allen Reference Allen, Trautmann and Whiteley2012), has been persuasively argued as a structural proxy for the earliest form of human kinship systems, from which others developed or historically evolved (of course, Dravidian itself was retained in some societies, including kingdoms in South Asia, so the terminology per se does not limit a social system’s scale or evolution: see Trautmann Reference Trautmann1982). From this perspective, while keeping the Iroquois/Dravidian equation of siblings with parallel cousins in G⁰, Crow–Omaha terminologies added intergenerational equations, in effect replacing Footnote ² the prior cross-cousin terms with ‘uncle’, ‘aunt’, ‘nephew’ and ‘niece’ terms, i.e. shifting referents both downward and upward: for Omaha, MBC=MG and FZC=ZC; for Crow, FZC=FG and MBC=C. The transition from Iroquois crossness to Crow–Omaha trends as evolutionarily unidirectional and (perhaps) irreversible (Trautmann & Whiteley Reference Trautmann, Whiteley, Trautmann and Whiteley2012a). But aspects of Iroquois and Crow–Omaha may co-occur: while Crow in terminology, Northwest Coast Haida and Tlingit operate with exogamous moieties reflecting an apparently substrate Iroquois pattern; and geographically they adjoin the Tsimshian-Gitxsan-Nisga’a, who have similar matrilineal, exogamous clans but Iroquois terminology and no Crow skewing (Whiteley & McConvell Reference Whiteley and McConvell2021). Additionally, Kronenfeld (e.g. 2012) has argued, Crow–Omaha skewing is sometimes an ‘overlay’, co-occurring with other terminological forms (like Iroquois, or Cheyenne—a mixed Iroquois-Hawaiian form) in the same society, allowing optative choices of terms in discourse contingent upon social interest. Kronenfeld (e.g. Reference Kronenfeld2009) argues that for West African Fanti, alternative terms are invoked to prioritize competing inheritance privileges.

Kronenfeld’s overlay model does not emphasize the effect on marriage alliances per se, but the addition of Crow–Omaha equations clearly obviates the ‘matrimonial implication’ of Iroquois crossness. As McKinley (Reference McKinley1971b) demonstrated, this obviation of prescriptive (elementary) alliance disperses alliances beyond binary exchange, in the fashion shown for Orayvi Hopi intermarriages among nine exchanging clan-sets (Fig. 4), so marriages over several generations exhibit the ‘aleatory’ pattern seen as characteristic of semi-complex alliance by Lévi-Strauss (Reference Lévi-Strauss1966; Reference Lévi-Strauss1969). Moreover, cross-culturally, the addition of Crow–Omaha equations has been shown to strengthen unilineal descent structures—matrilineal for Crow, patrilineal for Omaha (Dyen & Aberle Reference Dyen and Aberle1974; McConvell Reference McConvell, Trautmann and Whiteley2012; White Reference White1939). In Native North America, Crow and Omaha cases typically co-occur with named totemic unilineal descent groups that corporately own property. Moreover, Crow-matrilineal and Omaha-patrilineal systems correlate very strongly also with post-marital residence rules (patrilocal for Omaha, and especially matrilocal for Crow: McKinley Reference McKinley1971a; Dyen & Aberle Reference Dyen and Aberle1974).

Figure 4.

Hopi semi-complex alliance configured by intermarriages over time among nine matrilineal clan-sets at the Hopi town of Orayvi.

With Crow–Omaha systems, notably via the Hopi Crow case as described by Eggan (Reference Eggan1950), Lévi-Strauss (Reference Lévi-Strauss1966) recognized the alliance pattern was neither prescriptive (elementary) nor proscriptive (complex), but a mixture that came to be termed ‘semi-complex’ (Héritier Reference Héritier1981; Trautmann & Whiteley Reference Trautmann, Whiteley, Trautmann and Whiteley2012a). Alliances tie differentiated kin-groups to each other, but restrictions change with each particular marriage. Among the nine clan-sets of Hopi Orayvi prior to the town’s split in 1906, the networks were labyrinthine (Fig. 4).

Formal proscriptions on one’s own clan-set, father’s clan-set and mother’s father’s clan-set eliminate one-third (structurally speaking) of possible marriage partners. But in this largely endogamous system, a choice among six other sets is in effect ‘prescribed’ as the only available source of spouses. And with each actual marriage, the configuration of prohibitions for the next generation changes, creating the ‘aleatory’ pattern. Perplexed by the sheer number of possible Hopi marriage exchanges (mathematically calculated at ∼300 million) in a society of only a few thousand people, Lévi-Strauss chose to abandon his project on kinship and marriage structures per se. But at this moment when the study of kinship was entering a more general crisis (e.g. Needham Reference Needham and Needham1971; Reference Needham1974), Lévi-Strauss (Reference Lévi-Strauss1975) now steered his interest in alliance toward ‘house societies’, identifying a form that (like Crow–Omaha systems in this regard) was intermediate between kin-based and class-based social formations.

Anthropology’s crisis in kinship studies reached its apogee with Schneider’s Critique (Reference Schneider1984) seeming to mark a final death-blow.Footnote ³ Yet ‘houses’ evaded that fate. In effect, having deflected attention away from their kinship aspects, houses offered an enduring alternative to superseded descent versus alliance debates, escaping the reification of social systems as nested descent groups, and de-emphasizing the structural logic of intergroup alliance. However, as suggested above, if examined outside the cut-and-thrust of discourse pronouncing old kinship’s (premature) death, structural continuities between ‘house societies’ and Crow–Omaha systems—specifiable by particular combinations of terminology, descent, residence and marriage rules—are strikingly apparent.

From the perspective of demography and adaptation, Omaha-patrilineal systems show evidence of outward (‘downstream’) expansion, while Crow-matrilineal systems correlate more with internal concentration and differentiation (Whiteley & McConvell Reference Whiteley and McConvell2021). But discernible spread patterns of skewed kinterms/meanings from one neighbouring ethnolinguistic group to another cut across language boundaries in both Crow and Omaha instances (e.g. for Crow, Iroquoian Cherokee in the Southeast borrowed term-skewing from Muskogean Creek; Haida, a language isolate on the Northwest Coast, borrowed skewing from Athapaskan Tlingit; and Uto-Aztecan Hopi in the Southwest borrowed skewing from Keresan). In this light, the kinship structure is shown as in part an agentive social technology for managing relations, both intra- and inter-societally, notably via intermarriage (i.e. in the examples just cited, among the respective term-skewing neighbours: see Whiteley & McConvell Reference Whiteley and McConvell2021).

Additionally, in the North American Southeast, Northeast, Prairie-Plains and Puebloan Southwest (all with adaptive dependence on intensive horticulture), the ethnographic cases clearly descend from more complex late-prehistoric social formations: Mississippian for the first three, Ancestral Puebloan for the last. Chaco Canyon, the epicentre of Prehispanic archaeological complexity in the Southwest, has been profitably interpreted as a house society (Heitman Reference Heitman and Beck2007; 2015). But comparing the settlement architecture of Pueblo Bonito with Hopi Orayvi, I have also argued (Whiteley Reference Whiteley, Heitman and Plog2015) for the presence of Crow-matrilineal patterns,Footnote ⁴ while binary arrangements suggestive of moieties are evident in some neighbouring sites (e.g. Wijiji). Conversely, my reanalysis of Orayvi social structure (Whiteley Reference Whiteley2008) showed that Hopi social units are often better explained as ‘houses’—i.e. involving affinal and fictive-patrilateral recruitment and inheritance of privileges beyond matrilineal descent—than treated as corporate matri-units alone. As in other instances globally (Gillespie Reference Gillespie2000), the difference between the house and the lineage (especially of Crow–Omaha type) may be more a matter of diachronic versus synchronic perspectives, and a typological oscillation rather than an absolute cleavage.

Northwest Coast houses and Crow systems

The other Native North American Crow–Omaha systems—in California and on the Northwest Coast—also co-occur with adaptive intensification, sedentism, and higher population densities (but without agriculture or pastoralism). Especially on the Northwest Coast, the Crow-matrilineal cases (Haida, Tlingit) are associated with marked social ranking (nobles, commoners, and slaves), with privileges, titles, and crests descending in particular families, often associated with specific nineteenth-century villages comprising large communal houses, some abandoned under colonial pressures including the demographic devastation of introduced diseases.Footnote ⁵ Except for the Crow terminology and matriliny/matrilocality, however, the other social and architectural features noted are identical with those Lévi-Strauss (Reference Lévi-Strauss and Modelski1982) used to designate the ‘house’ in Kwakwa̲ka̲’wakw society, from Franz Boas’s ethnography (Figs 5 & 6).Footnote ⁶ Boas, as Lévi-Strauss (Reference Lévi-Strauss and Modelski1982, 164–6) retailed, shifted position over time. At first, he described Kwakwa̲ka̲’wakw social units as patrilineages that he inferred had evolved from a prior stage of matriliny (then considered a universal trajectory since Bachofen Reference Bachofen1861). But Boas later reversed that inference, i.e. suggesting Kwakwa̲ka̲’wakw society had evolved from an earlier patriliny toward matriliny. Finally, abandoning etic concepts (and their evolutionary implications) altogether, Boas treated the Kwakwa̲ka̲’wakw nuymaym as a unique form deploying dual or ambilineal descent, affinity and non-kinship ties to reproduce itself through time:

Figure 5.

Kwakwa̲ka̲’wakw village of Newitti (Hope Island), British Columbia, 1881. AMNH Special Collections, Photographic Negative #42298. (Photograph: Edward Dossetter.)

Figure 6.

Haida village of Yan (Haida Gwaii), British Columbia, 1881. AMNH Special Collections, Photographic Negative #44310. (Photograph: Edward Dossetter.)

Yet Boas was very much aware of the socio-cultural continuities from the Tlingit in southeast Alaska south to the Coast Salish Lushootseed in Washington State, and argued for a gradient of transitions:

Extensive interchange—especially via the potlatch—was (and remains—even after a colonial law banning it for seven decades, 1885–1951) common among Native Northwest Coast societies. Exchange embraced material goods (Chilkat robes, carved boxes, mountain-goat wool, etc.), prestige foods (eulachon grease, etc.), intermarriages, and encompassed internecine wars through the adoption of enslaved captives, while specific myths and ancestral figures (like Komogwa, the undersea lord of wealth) were widely shared (e.g. Drucker Reference Drucker1955; Lévi-Strauss Reference Lévi-Strauss and Modelski1982). Thus social and ideational chains of interconnection united peoples of different language families into a common ‘culture’. In fact, this was exactly what defined the Northwest Coast as a culture-area, as Boas recognized early on in a manuscript catalogue he was commissioned to produce for the American Museum of Natural History:

The Kwakwa̲ka̲’wakw belonged to ‘the central regions’ (from the penultimate passage quoted) representing the ‘mixed type’ of social organization. The combinatory structure of Kwakwa̲ka̲’wakw social units and inheritance of privileges, together with Boas’s cumulative frustration with how to classify them, directly inspired Lévi-Strauss’s (Reference Lévi-Strauss and Modelski1982, 169–70) house concept, especially via comparison with European feudal houses. But Lévi-Strauss’s definition of houses (see above) surely applies equally to Tlingit, Haida, or Tsimshian—in which matriclans are subdivided into named, localized houses with titles and privileges—just as well as it applies to the Kwakwa̲ka̲’wakw. Moreover, Boas (Reference Boas1940, 360) acknowledged that there had been extensive changes in Kwakwa̲ka̲’wakw society owing to nineteenth-century population losses before he began to work there, and he questioned whether this may have had an impact on the pattern of ambilateral descent and claiming of privileges.

In this regard, there seem palpable similarities with the historic Southwestern Pueblos, among whom much shared culture cuts across linguistic boundaries, and where a gradient of social organization tracks from Crow-matrilineal in the west among the Hopi and Zuni to ambilineal among the Keresans and bilateral among the Tewa (Eggan Reference Eggan1950; Ford Reference Ford and Whiteley2018). Variant earlier attributions of matrilineal versus patrilineal versus bilateral in the history of Tewa ethnology even echo Boas’s Kwakwa̲ka̲’wakw oscillations, and there were comparable processes of population decline and consolidation among the Tewa also (together with long-term superimposition of patriarchal structures by the colonial Spanish state, although nineteenth-century Kwakwa̲ka̲’wakw colonial history is not directly comparable in that regard). As noted, I have argued that Hopi social groups and marriages over time showed matriliny in practice actively intersected by patrilateral ties (including some that are patrilineal), and that inherited privileges are not infrequently held by affines (Whiteley Reference Whiteley2008). Fortes’s (Reference Fortes1969) ‘complementary filiation’ de-emphasized the structural importance of such intersections in a unilineal system, but it is precisely these conjunctures in which the practicalities of history operate within and against—thus often transforming—received ideal-typical (strictly lineal) structures. Prima facie, it seems very likely that Tlingit and Haida Crow-matrilineal systems must have operated with similar interruptions and contraventions of purely matrilineal principles (this exceeds my present scope, but merits further investigation).

Houses and lineages

So when is a house a ‘house’, and when is it not much different from a kinship group—especially of Crow–Omaha semi-complex type? Conversely, when is a descent group (like a Hopi ngyam, ‘clan’) better treated as a house (wungwki, ‘clanhouse’), considering the roles played by affinity, locality and optation in the transmission of its privileges (Whiteley Reference Whiteley2008)? Here Kronenfeld’s argument for Crow–Omaha skewing as optative overlay on an underlying kin-terminology, with alternate strategic deployments, is resonant.Footnote ⁸ If positions, hereditament and titles can switch back and forth among descent and affinal lines, or if descent modalities in fact comprise a suite of alternatives rather than a fixed unitary form, one must ask how different this is, except as a matter of emphasis, from the patterns Boas identified for the Kwakwa̲ka̲’wakw numaym. With the preference in sociocultural anthropology and archaeology for Lévi-Strauss’s houses emerging at the same moment as discontent with lineages reached its height, the shift to an ostensively non-kinship-based concept may be more a matter of epistemological convenience than genuine analytical advance.

In this regard, it is useful to turn to another analyst of Kwakwa̲ka̲’wakw social organization, Irving Goldman (sometimes reckoned as Boas’s ‘last student’). Just before Lévi-Strauss launched his inquiry into the house in the mid-1970s (see e.g. Gillespie Reference Gillespie2000), Goldman (Reference Goldman1975) had meticulously reanalysed many of the same Boas sources Lévi-Strauss depended on. While Goldman’s conclusions on Kwakwa̲ka̲’wakw religion were contested by specialists, his observations on social structure retain significant value. Goldman maintained that lineages were just as important for the Kwakwa̲ka̲’wakw as farther north, but that the definition of lineages as only unilineal and exogamous had prevented Boas from realizing this. Goldman proposed a more inclusive definition for Kwakwa̲ka̲’wakw ‘status lineages’, which operated with ‘in effect a mixed or dual descent’. Lineages remained, he argued, central to Kwakwa̲ka̲’wakw social organization:

That Goldman could reach this conclusion from the same sources used by Lévi-Strauss again suggests seeing a ‘house’ rather than a ‘lineage’ may be a matter of perspective.

With their constitutive amorphousness, houses may have eased the burden of engaging with the four dimensions of how kinship systems operate at a time when the kinship paradigm was undergoing antinomian critique, but the consequence, in my view, has meant a loss of explanatory rigour. Houses and lineages are in practice and diachronically not sufficiently different as to constitute separable analytical units as such, with which to clarify the explanation of middle-range societies, either ethnographically or archaeologically. Conditions for the emergence of houses and Crow–Omaha systems in Native North America appear to reflect an intermediate status in longer-term evolutionary transitions. These include population pressure and social and/or geographic circumscription (per Carneiro’s Reference Carneiro1970 theory of the origin of the state). But in many ways, houses and (Crow–Omaha) lineages appear to represent different perspectives or emphases on similar socio-cultural phenomena, considered diachronically versus synchronically.