MAYA SOCIAL ORGANIZATION

Whatever their other differences, these papers agree on the empirical evidence for Maya social organization. From small rural sites in northwestern Belize, such as those Jon Hageman (2004) discusses, to AnnCorinne Freter's (2004) rural communities in the Copan Valley, to the large sites of Tikal, Caracol, and Chunchucmil treated by Haviland (2001), Marshall Becker (2004), Diane Chase and Arlen Chase (2004), and Scott Hutson and colleagues (2004), all show so-called patio or plaza groups that consist of clustered structures regularly arranged around a central area, often but not always including a distinct ritual structure and set off from neighboring groups with or without walls and adjacent farmland. T. Kam Manahan's (2004) paper, in arguing for marked transformations at post-Collapse Copan, serves nicely as the exception that proves this rule. Successive phases of construction and reconstruction in these compounds suggest occupation by presumably kin-based households that persisted and expanded over time, although, as Hutson and colleagues (2004) suggest, not necessarily without discontinuities or the felt need to reaffirm periodic claims to group status or resources.

A second point of general consensus concerns the dynastic polities that overlay these residential units. Although varying greatly in power and duration from site to site, dynastic rulers claimed local, if not regional, ascendancy, however allied with or dependent on local constituencies or rulers at other sites. Neither unified nor uniform, these Maya elites nevertheless developed common symbols of rulership, hieroglyphic writing, and architectural styles of monumental self-aggrandizement, often centered on veneration of dynastic ancestors and succession from them.

Beyond these two points, differences and debate focus on two questions. First, given the patio or plaza groupings by which the Maya continued to demarcate their built environment, even at very large sites such as Tikal and Caracol, what defined these groups and relations among them? Second, whatever the nature of these groups, how did they relate to the dynastic rulers that rose to prominence at Maya sites?

Regarding the first question, Mayanists have most commonly associated these residential compounds with patrilocal extended families based on ethnographic analogy to living Maya communities. Indeed, it was the ethnographer Evon Z. Vogt (1964:26–28, 1994:390–399) who first proposed a segmentary model of Classic Maya sociopolitical organization. According to such “decentralist” views (Fox and Cook 1996; Fox et al. 1996), descent relations traced from ancestors and ancestral gods would have divided yet integrated Maya society through branching lineages in segmentary opposition to one another under the ruling lineage of the king. Genealogical seniority would account for rank differences, as well as the coexistence of kin-based and state-based groupings. Hageman (2004) most explicitly defines and defends local lineage organization, following African-derived models of segmentary lineages as unilineal, exogamous, corporate, and internally differentiated (Fortes 1953). Freter (2004) cites William Sanders's model of segmentary lineage organization for the Copan polity.

Evidence for nonunilineal (or, at least, nonpatrilineal) descent from the treatment of female forebears in Maya burials and inscriptions, and the general disrepute into which lineage theory has fallen since its heyday between the 1940s and 1960s (Kuper 1982), have prompted recent calls to replace the lineage model with a “house” model (Gillespie 2000a, 2000b; see Carsten and Hugh-Jones 1995). Based on Lévi-Strauss's idea of “house societies,” the house would shift attention away from descent as a defining characteristic of Maya plaza groups and conceive of them instead as a “corporate body [personne moral] holding an estate made up of both material and immaterial wealth, which perpetuates itself through the transmission of its name, its goods, and its titles down a real or imaginary line, considered legitimate as long as this continuity can express itself in the language of kinship [descent] or of affinity [alliance] and, most often, of both” (Lévi-Strauss 1982:174; see Gillespie 2000a:27). In keeping with Lévi-Strauss's (1969 [1949]) earlier opposition to descent theory, perpetuation of the estate itself becomes primary, with descent as only one means of maintaining the corporation instead of the prerequisite for its very existence. Common residence becomes crucial, and non-kin can become attached to the house this way.

The distinction between lineage and house rests on the relative importance of unilineal descent as an organizing principle. Thus, I would argue that the two types differ more in ideology than in actual practice. Both refer to corporate bodies that consist of self-identified groups seen as such by other groups and whose members stand in an “all for one and one for all” relationship to members of other such groups, whatever internal distinctions divide them; they hold privileges and possessions in common but seldom equally; they exist in perpetuity (or, at least, continuity) from one generation to the next; and they depend on kinship (unilineal, bilateral, or cognatic; real or fictive) and marriage for their literal as well as figurative reproduction (Schneider 1965:47–49). Archaeologically, either could map equally well onto Maya plaza groups. Conversely, if these compounds represent extended family households, they could just as easily reflect neither, in part because of ethnographic conflations of filiation, descent, and corporate organization, and in part because of where the theories behind these models differ in the first place.

Ethnographically, the presence of extended family compounds does not automatically translate into corporate descent groups or segmentary lineage systems. An oft cited ethnographic analogy for Classic Maya social organization comes from Vogt's (1969:127–149) description of Tzotzil Maya “localized patrilineages” and sna (Vogt's term, not a Zinacanteco category, for residential clusters of patrilineally related households) in the highland Chiapas municipio (administrative jurisdictions imposed by colonial and national regimes) of Zinacantan. Although Vogt uses the term patrilineage, my reading suggests more the workings of filiation (parent–child links) in continually fissioning extended family households than the enduring, descent-based corporate lineages of African models. Zinacanteco patrilineages consist of related extended family households with a common named ancestor no more than four generations back. Lineage depth remains shallow; collateral spread narrow; and corporate stability ephemeral. Most significantly, and unlike models of unilineal descent groups in west Africa (Fortes 1953), localized lineages do not aggregate at higher genealogical levels beyond living memory, and descent does not serve as an exclusive organizing or juridical principle even within local settlements, since sna and waterhole groups often contain multiple lineages (Vogt 1969:141, 145).

Instead, in contemporary Maya communities, territorial divisions by municipio or cantón supercede kin-based organization. Unlike in some West African kingdoms where corporate descent groups persist politically and jurally under paramount rulers (Fortes 1953:26), Maya communities today do not exist simply at the sufferance of otherwise autonomous kin groups. Kin groups remain subordinated to state-imposed civil and religious offices. This local institutionalization of state power typifies so-called peasant societies subjected to the extractive authority of outside powerholders (Wolf 1966), a condition true for Maya communities since well before the Spanish conquest. I can think of only Momostenango, a K'iche' municipio in highland Guatemala, where Maya lineage heads ever serve as such in a formally recognized community hierarchy but even there only in their capacity as “priest-shamans,” not as autonomous administrative or jural authorities (Tedlock 1992:35).

More telling still, the patrilineal, patrilocal extended family in Zinacantan retains its corporate character only until sons inherit from their fathers or find the wherewithal to build their own houses separate from their fathers', if ideally within the same compound or on land given them by their fathers (Vogt 1969:130–131; see Bossen 1984:86–94, 112–124, for a helpful discussion of generational as well as gendered tensions in Maya extended-family households). When occupying land in the same settlement, households headed by siblings and cousins recognize their common bond through periodic rituals reaffirming their collective claims to local lands and water, but they hold these resources by household, not corporately or exclusively, since such ritual groupings often include unrelated households. Similarly, Vogt (1969:145) writes of patriclans in Zinacantan as exogamous groups sharing the same Maya patronym, but these groups lack named founding ancestors, corporate functions, and economic interests.

In all these ways, localized patrilineages in Zinacantan suggest filiative parent–child links in extended-family households instead of descent as an extended succession of filiative ties that join multiple households into larger, stable corporate groups. Zinacanteco lineages more closely resemble local aggregates of households headed by siblings and collaterals with mutual but not corporate interests. At the same time, the cyclical fissioning of extended-family households need not imply lack of residential continuity, since in many Maya communities one son (often the youngest) takes over the parental household in return for caring for the elderly parents. Archaeologically, residential continuity could persist from one generation to the next through filiative transmission as easily as through descent.

This is not to say, however, that an ideology of descent could not still apply. By itself, the tracing of descent does not create corporate groups, although it can certainly serve to define them. As Schneider (1965:47) points out, “[A] descent rule, rigidly followed, only creates a conceptually distinct category. Its organization as a corporate unit depends upon other factors.” Even in classic unilineal societies such as the Tallensi of West Africa, ritual participation in sacrifices to ancestors serves importantly to activate potential claims that come with shared descent, just as in societies with cognatic descent that give individuals wider potential claims not only in their father's but also in their mother's, father's mother's, and mother's mother's clans, only repeated ritual participation as well as prolonged residence secure rights to ancestral estates (Keesing 1970). In this, Hutson and colleagues (2004) may well be right that architectural modifications to principal patio group structures at Chunchucmil might reflect contested succession to group leadership, not smoothly operating descent rules, but they miss the point that descent rules never work automatically (or alone) in the first place.

Again, Zinacantan also provides an ethnographic example of descent without corporate groups. Zinacantecos venerate ancestral gods (literally “father-mothers”) said to reside in prominent mountain peaks and lesser hills associated with social aggregates from the municipio as a whole down to local household clusters. These ancestral gods, however, mark generic ethnic and territorial attachments, not strict genealogical descent. They dress and speak like Zinacantecos (although they live in non–Maya Ladino-style houses inside their mountains), but they remain unnamed and genealogically unspecified as simply the first Zinacantecos to occupy local lands (Vogt 1969:144, 298–301, 384).

Clearly, Zinacantecos have a descent ideology that they use to differentiate themselves, their lands, and their community from each other and neighboring Maya and non–Maya communities, but this does not correlate with corporately organized descent groups. Conversely, in societies with corporate descent groups, genealogical exceptions to rules of descent need not disprove the rule. As Fortes (1953:33) notes for patrilineal societies that also practice polygyny, “[W]hat is a single lineage in relation to a male founder is divided into segments of a lower order by reference to their respective female founders on the model of the division of a polygynous family into matricentral ‘houses.’”

Filiation, corporate organization, descent ideology, and strict unilineal descent rules clearly never co-vary in all-or-nothing fashion. Filiation, while normally present in one form or another, can occur without becoming descent. While presuming filiation, descent systems such as segmentary lineages may become corporate (or not) in different ways (over property, political alliance, ethnicity), especially at different genealogical levels. And while they can define corporate groups, descent “rules” seldom remain inviolable because filiation by definition works bilaterally through both parents. In other words, occurrence of one of these characteristics never precludes the others, just as their co-occurrence requires explanation, not presumption.

Presumptive co-occurrence of filiation, descent, and corporateness finds its way into the “Copan ancient sian otot” model proposed by Freter for the Copan Valley. Fash (1983) originally used Charles Wisdom's (1940:216–220) description of rural settlements that the Ch'orti' Maya of eastern Guatemala call sian otot (literally “many houses,” aldeas [hamlets] in Spanish) to suggest residential groupings at nearby Copan that perhaps intermarried with one another. Following Sanders, Freter superimposes on this a segmentary lineage organization, with each sian otot containing a “lineal head” who resides in the largest household of the settlement. Ethnographically, however, if Ch'orti' usage should serve as any guide for using the term archaeologically, no evidence supports descent organization for such residential groupings.

Like Vogt, Wisdom (1940:246) writes of “lineage groups,” but he clearly means extended family households with filiation, not descent: “a number of related and mutually dependent households all of whom live either together or in the same neighborhood, and who act as a co-operating group … is a lineage group, all the members of which are related through descent from its oldest surviving parents or through marriage into the family.” At the death of the oldest couple, “the last link which binds the family together is lost, and the family as a unit breaks up. Each household inherits equally of the family property, and each household head becomes the head of his new family group” (Wisdom 1940:249–250). We have here neither segmentary lineages nor even estate-based houses in Lévi-Strauss's sense, only extended family groups with filial transmission that splits inheritance from one generation to the next.

Furthermore, regarding the sian otot itself, Wisdom (1940:219–220) states that it exists as “little more than a rural district in the Indian's mind. … [O]ne aldea has little physical distinction from another and is more a region, or rural locality, than a clearly delimited village. It is a social group, in that its families co-operate among themselves … but family co-operation is not totally confined to the aldea. It has no important cultural distinction from any other aldea within the same municipio.” Like aldeas in other Maya municipios, the sian otot refers primarily to a physical place, a locale to which Maya become attached by incidence of birth and residence. Although such places often do come to serve as the primary locus of family, work, marriage, household, and social cooperation, they are not defined by such activities or relations. They include multiple, unrelated family households; these households hold their lands individually or, in the past, under usufruct from the municipio (Wagley 1941:59–63), and the same families may intermarry in successive generations, but these alliances neither sustain larger corporate descent groups nor coincide with the sian otot in any exclusive or organizational sense.

As with Zinacantan, Ch'orti' household organization and the sian otot provide at best weak analogies for imputing segmentary lineage organization to Classic Maya society. This is not to say that such lineage organization did not exist. It is only to say that analogies from contemporary Maya ethnography cannot be used to support such inferences except by treating attested patterns of filiation as vestiges of now defunct corporate descent groups. Such extrapolation, however, itself becomes an inference that tautologically assumes what it is trying to demonstrate. For instance, where John Fox and Garrett Cook (1996) have turned to ethnohistorical sources to document segmentary lineages among K'iche' Maya, I would agree with Susan Gillespie (2000b:470; see also Hill and Monaghan 1987) that they appear too ready, given the evidence they cite, to equate K'iche' amak and chinamit, both of which they say Colonial dictionaries define as “small rural communities,” with “minimal” or “principle” lineage segments. Furthermore, the descent groups they find in the Popol Vuh clearly pertain to elite, if not ruling, K'iche' lineages whose descent organization might not apply to the larger rural population.

DESCENT AND ALLIANCE AS CULTURAL RULES AND THEORIES OF SOCIETY

The ethnographic distinction between filiation and descent may appear trifling and certainly relative, since in one sense descent is simply a “continuum of filial links,” whether agnatic through males, uterine through females, or cognatic through either (Keesing 1970:768). But the difference relates directly to how anthropologists have conceived of lineages and houses as corporate groups. Before turning to this point, I need to comment on the general problem of rules and practice raised by Hutson and colleagues (2004). (Hageman [2004] also invokes practice theory in opposition to structural functionalist abstractions.) Following Wittgenstein, Heidegger, and Bourdieu, Hutson and colleagues (2004) suggest quite rightly that human beings are not cultural automata blindly following rules, and we cannot explain anything anyone ever does by treating them as such. Nevertheless, I think they overdraw the distinction between rules and practice if they assume that where rules exist they should operate automatically and mechanically, and that, absent this, practice flows solely from an experientially acquired, ever contingent “feel for the game.”

Instead, I would argue rules and practice entail each other in at least two ways. First, although people may never simply “follow a rule,” they can and often do invoke rules to define, defend, or dispute their acts and interactions with one another. Far from hard and fast, such rules express conflicting or alternative ideological claims and plays for power. Even if no one ever follows them to the letter, however, such rules do express publicly recognized values and morals (the “spirit” of the law) that still regulate social behavior by establishing grounds at least for dispute, if not for justifiable sanctions against those caught violating them. Rules also reflect the relative power of different parties to make them and decide which ones apply in what ways to specific situations. Ironically, rules in this sense play an essential role in the social strategizing intrinsic to the Bourdieuian practice that Hutson and colleagues (2004) favor.

Second, rules can refer to descriptive generalizations or expectations—that is, “norms” or conventional “rules of thumb.” Although never determinant or causal, such conventions reflect readily recognizable (if often tacit) regularities, patterns, and mutual understandings that emerge in social interactions and make Bourdieu's notion of practice possible in the first place. Precisely because human beings act by interacting with one another, rules cannot be applied to individual or collective behavior in any absolute, mechanical fashion just as Hutson and colleagues (2004) say, but by the same token interaction cannot occur without some mutual accommodation to conventionally agreed on norms (in the sense of what is “normal”) because otherwise any expectation or interpretation of others' actions would become impossible. It is precisely these kinds of taken-for-granted, experientially learned, socially shared, and materially embodied conventions that constitute culture, and ethnographic description (just as sociality itself) simply cannot do without them (recall Redfield's [1941:132–133] still useful definition of culture as “conventional understandings manifest in act and artifact,” or, more to the point, Bourdieu's [1977:72; italics in the original] own notion of habitus as “durable, transposable dispositions”). Rules may never mechanically account for human interactions, but neither do interactions occur only spontaneously as contingent practices.

Given this, however, ethnographers today would agree that if we write about rules at all, we must treat them in their conventional, not causal, sense, and attend carefully to how they are used, by whom, and with what import. This still leaves unexplained why earlier anthropologists writing on kinship and social organization spent so much time arguing about rules of descent, marriage, and residence (but again, in their conventional rather than causal sense) to the near-exclusion of when, where, how, or by whom these rules got applied. Answering this question requires a closer look at the crux of debate between descent and alliance theory. This in turn reveals that the difference between lineages and houses pertains as much to their rules of descent or the nature of their corporateness as to how anthropologists have used them to argue about the nature of social and moral order in segmentary societies.

Schneider (1965) reminds us that descent and alliance theorists focused on rules, not because they naïvely thought that everyone followed them, but because they saw in rules of descent and alliance contrasting, competing theories of society itself. The fundamental question in this debate centered not on the nature of family, kinship, or household, but on the Durkheimian problem of mechanical (or minimally organic) solidarity: how in societies lacking centralized political institutions did social and moral order emerge between like social segments? Ironically, both descent and alliance theory took unilineal descent groups as their given social segments and debated instead what tied them together.

Descent theory derived ultimately from Maine's (1861) study of ancient law that postulated patriarchy as the earliest social order, with families held together under the authority of the eldest male ascendant. At the death of the patriarch, the families of surviving sons maintained their association, and over time kin relations between these families and their descendants extended to form a broader polity (Kuper 1982:73). Common interests, especially in property and its transmission, led to the formation of corporate groups integrated internally and opposed externally to other groups through descent. Descent became unilineal, theorists argued, because tracing relations only through the father or the mother established the most unambiguous, incontrovertible way of reckoning group membership and rights while also providing a way to relate more distant groups genealogically (Fortes 1953:30; see also Evans-Pritchard 1940).

Descent relations thus came to coincide with “the widest range within which sanctions exist for preventing conflicts and disputes from ending in feud or warfare. The political unit is thought of then as the most inclusive, or maximal, lineage to which a person can belong, and it may be conceptualized as embracing the whole tribal unit” (Fortes 1953:28). For descent theorists, segmentary descent became one and the same thing as society. As the primary idiom of political association and jural adjudication, such genealogical reckonings also clearly shifted over time to reflect “the existing pattern of legal and political relations within and between lineages” (Fortes 1953:28), a further indication of their primary politico-juridical function and importance.

For these theorists, the juridical principle, not the genealogical substance, of descent mattered most. More than simply an extension of filiation, “descent is fundamentally a jural concept. … [W]e see its significance as the connecting link between the external, that is the political and legal, aspect of what we have called unilineal descent groups, and the internal or domestic aspect” (Fortes 1953:30). As this kind of integrative social principle working across different domains of social life, descent precipitated more than discrete components of “social organization.” It reflected a more systematic “social structure” that encompassed “all the different classes of social relations found within a given society” (Fortes 1953:22). More than simply another aspect of culture, social structure referred to “the entire culture of a people handled in a certain frame of reference” (Fortes 1953:21). That frame of reference consisted of corporate descent groups predicated on commonly held property and the solidarity of shared unilineal descent that admitted “no compromising allegiance … outside the group” (Schneider 1965:49).

In contrast to the politico-juridical approach of descent theorists, alliance theorists began with the position Mauss took in The Gift (1990 [1925]) that society resulted not from the pseudo-legal mediation of utilitarian interests by corporate descent groups but from the reciprocity basic to sociality itself. The final chapter of The Gift makes clear Mauss's hope that this more fundamental sense of moral interdependency—a return to the spirit of the gift, not charity—could be revived from former times against the utilitarian, legalistic “tradesman's morality” that he saw increasingly dominating modern life. Accordingly, alliance theorists saw society constituted by the ongoing moral obligations of reciprocity to give, to receive, and to repay by which moral persons (whether individuals or groups) enacted and reaffirmed the social bonds between them. The corporate nature of social segments emerged not a priori through descent but relationally in ongoing relationships with other segments, especially marriage exchanges.

For Lévi-Strauss (1969 [1949]), the primordial instance of exchange—his elementary structure of kinship—involved groups of brothers who exchanged sisters to cement their mutual interdependence as “wife-givers” and “wife-takers.” Lévi-Strauss cited as evidence for his theory widespread “positive marriage rules” that stipulated unions especially with cross-cousins (that is, marriage not necessarily with father's sister's child or mother's brother's child but with an individual from one of the social groups where a same-sex ascendant of ego's own group had married in a previous generation and whom ego would call “cousin” according to the logic of their classificatory kinship terminology). Such positive marriage rules resulted in ongoing alliances between groups through symmetrical or asymmetrical, egalitarian or stratified, restricted or generalized exchange that at once kept groups distinct as “wife-givers” and “wife-takers” yet integrated them into enduring interdependency through the movement of people, goods, and services and the cross-cutting filiative ties among parents, children, and their respective groups.

Descent theorists took issue with alliance theory for ignoring those many societies where no positive marriage rule established lasting alliances between specific kin groups, and marriage remained restricted only by incest taboos. Absent such enduring alliances, descent alone integrated society. Alliance theorists in turn criticized descent theory for trying to downplay affinal relations through analytical sleights of hand such as “complementary filiation” (Fortes 1953:33–35) in which children's claims on the descent group of the parent through whom they did not trace descent became filiative expressions of descent instead of ongoing alliance between groups. Much ink got spilled over whether relations between mother's brother and sister's son in patrilineal West African societies represented mutual claims and obligations indicative of an alliance between their lineages or impositions the sister's son could make on his mother's brother by right of his mother's “residual rights” of descent in her natal lineage (Schneider 1965:50–55).

The intractability of this debate only intensified its vehemence. Rather than simply dismissing each other, descent and alliance theorists had to argue for the relative importance of their own position while trying to explain away the indispensability of the other. As long as social segments remained corporate and exogamous, a point on which both sides agreed, they by definition required filiation (if not descent) as well as affinity. However narrowly or rigidly descent theorists defined lineages by unilineal descent, incest taboos dictated that lineage members find their spouses elsewhere. Although lineages did not necessarily have to look for spouses in the same groups generation after generation, as alliance theory required, marriage did involve extended exchanges of wealth, feasting, and ritual responsibilities for at least as long as the married couple lived (and stayed married), and often for the lives of their children or even grandchildren. At what point did these exchanges also become mutual alliance, not just distaff filiation?

Conversely, for marriage alliances to form and persist over generations, alliance theorists had to admit the existence of enduring groups that could become allied in the first place. Marriages between such groups generation after generation would produce de facto (if not de jure) descent groups with corporate interests at least in mobilizing the necessary personnel and wealth to continue the alliance in the next generation. Indeed, when viewed from the resulting groups instead of the alliances between them, Lévi-Strauss's “wife-givers” and “wife-takers” begin to look suspiciously like corporate kin groups practicing patrilineal descent and patrilocal residence. At what point in generational time did relationally defined groups engaged in horizontal exchanges with one another also become vertically integrated descent groups of wife-givers and wife-takers?

Lévi-Strauss's (1982) house model derives directly from these theoretical differences and difficulties. As the premier proponent of alliance theory, his discussion of houses predictably underplays the importance of descent in maintaining the house's estate and focuses instead on how the house uses shared residence (and actual residences) to fetishize—falsely objectify—inherent instabilities in the relationship between spouses who retain lasting and important rights and obligations in their natal kin groups. “By putting, so to speak, ‘two in one,’ the house accomplishes a sort of inside out topological reversal, it replaces an internal duality with an external unity” (Lévi-Strauss 1982:184–185). The notion of the house as a corporate entity that seeks opportunistically to pre-empt spousal descent group ties serves his theoretical position well by corporatizing (however precariously) alliance itself at the expense of unilineal descent ideologies, corporate descent groups, and even prescriptive marriage rules, a development he then historicizes as the transition from kin-based to state-ordered, class-stratified societies (Lévi-Strauss 1982:186–187).

Even here, however, Lévi-Strauss cannot escape the old debate. The European cases he cites to validate his house model still presume preferential patrilineal descent to which inheritance through female links serves largely as the exception that proves the rule. Furthermore, all of his examples of houses involve estates of nobility, yet he slips unremarked from a discussion of “the house” (1982:172 and passim) or societies “with houses” (Lévi-Strauss 1982:180) to “house societies” as a societal type comparable to those based on descent and alliance (Lévi-Strauss 1982:185). It is not clear, however, whether this type requires all households in the society to be houses or only those of the nobility, to which commoners then become attached. Gillespie (2000a:32) also notes that Lévi-Strauss's published discussions of houses provide “no in-depth examination of the everyday interactions and processes by which the tensions of conflicting rights and obligations—the within-house and between-house relationships—were manifested and negotiated to present some ‘outward face of unity.’” She attributes this to his interest in showing the long-term historical continuity of houses, but I would argue that it more closely reflects his abiding concern with comparative models of general (if not evolutionary) societal types (see also Carsten and Hugh-Jones 1995:18).

Given these many conflations, confusions, and controversies on both ethnographic and theoretical grounds, distinguishing lineages from houses in Classic Maya social organization becomes increasingly problematic. Ethnographically, lineages presume households in which filiation can cross-cut (if not pre-empt) descent, and houses presume continuity of descent (if not strictly unilineal, then cognatic) even if all members of the house are not related by blood or marriage. Theoretically, the structural functionalism of lineage theory has become much too mechanistic and essentialist for those inclined toward more contingent, agent-based modeling of social life, yet the opportunistic corporatism of the house model results in such a generic residential household that even ethnographers have to wonder if “in the end, the problem is not one of discovering which societies are ‘house societies’ but of discovering which ones are not” (Carsten and Hugh-Jones 1995:18).

Maya archaeologists might well give up on such ready-made formulations and reexamine the constituent components of filiation, descent, alliance, and residence that make up these models to decide which elements in what relation would enable them to make the best sense of what they find in the ground. I say this not because I think Classic Maya society was so unusual that theories from elsewhere need never apply. Indeed, Arthur Demarest (1996:823–824) warns darkly against such isolationism. Nevertheless, before we can determine whether theories from elsewhere apply, we must know what we have in hand to compare with them. While models of West African segmentary states, Southeast Asian galactic polities, Indonesian theater states, South Asian regal-ritual cities (Fox et al. 1996), and European court societies (Houston and McAnany 2003:37)—or even African segmentary lineages and medieval European noble houses—provide comparative possibilities, I would suggest that they serve better after the fact once first approximations of the Maya data have been drawn directly from the archaeology, rather than necessarily before the fact in formulating interpretations. Otherwise, a temptation too easily arises to fit findings to whichever descriptive model or theory of the moment looks most appealing, then to take this as both confirmation of its applicability and an explanation of the data. This brings me finally (and briefly) to models and model building.

MODELING THE MUDDLE

In his critique of descent and alliance theory, Schneider (1965:73) concludes that “both the alliance theorists and the descent theorists have a tendency toward the development and propagation of whole-system, over-simple typologies.” Lineages and houses become set assemblages of essentialized features embedded in particular matrices of socioeconomic and political relations that together define mutually (but falsely) exclusive societal types of “descent societies” or “alliance societies” into which given cases must then fit—or form a new type, as Lévi-Strauss would propose with “house societies.” Already for Schneider in 1965, however, “it is too late in the history of the social sciences to think we can go out among societies and, by keeping our eyes open, sort them into their natural classes…. [A] typology is for a problem, it is not for sorting of concrete societies into unchangeable, inherent, inalienable categories” (Schneider 1965:78).

Instead of “total-system” typologies, Schneider (1965:78) says we should build models that address specific questions by taking “a series of relevant elements … rigorously defined as analytic categories and then combined and recombined into various combinations and permutations” to specify how these elements might vary in relation to one another. Such models serve as the starting point for conceptualizing interconnections in our findings, not as an end point for typing what we find. In formulating them, we must also be explicit about what we intend them to model. Are they purely analytical constructs of the anthropologist meant for heuristically ordering data? Are they ideal (if perhaps unconscious) conceptualizations in actors' heads (as Becker suggests about his plaza plans at Tikal)? Or do they represent full-scale cultural reconstructions of “how the system really works” concretely on the ground? Mistaking one kind of model for another, or muddling when we are modeling what, too often results in needless, self-perpetuating polemics.

Part of the trick in such modeling, I would argue, lies simply in remembering the patterned contingencies of social life—with equal emphasis on both pattern and contingency—even when we cannot capture ethnographically or recover archaeologically all the patterns or all the contingencies. Most concretely, we always face what Freter rightly calls the social fabric of interwoven relationships between individuals and groups. The strands of kinship in this fabric consist of the vertical warp of consanguinity and descent and the horizontal woof of affinity and alliance. As in an actual fabric, warp and woof each become so only in relation to the other, just as every intersecting node depends on all the other nodes in the warp and woof to keep the fabric from unraveling.

At the same time, unlike in actual fabric, the same strands and nodes carry different meanings depending on position and perspective. An affinal relation between spouses in one generation becomes a consanguineal relation through their children in the next, just as consanguineal siblings must forge affinal ties with others, whether in the same groups where their ascendants married or in new ones. Similarly, what from the top down looks like descending arrays of individuals linked to each other through apical ancestors becomes, from the bottom up, individual networks of bilateral kin (a kindred or “cognatic kinship”) that reach outward and upward through parents, grandparents, great-grandparents, and so on to an equally wide array of ancestors through whom relations can be traced (Keesing 1970). One array never eliminates the other, but different societies define in different ways when, if at all, which set of such relations will come into play for what set of actors.

Furthermore, residence and work skew the weave. Here again, Freter draws attention to residence not just as residential and ritual structures but also as activity areas that suggest important spaces of interconnection between groups. (I would only question the gendered directionality she associates with such spaces as no more likely than if gendered nouns in Romance languages indicated men's versus women's things.) We can debate forever which first took priority—residence rules or practical activities—but the possibility also remains that such patterns developed in part as Mauss surmised simply to weave society more tightly together and keep the peace. I think here of places like the Trobriand Islands that have matrilineal descent yet virilocal residence where women go to live with their husbands, their daughters leave to live with their husbands, and most sons remain with their wives in their father's village, which may not be the father's matrilineal place, either, but belong to the matrilineage of one of his patrilateral affines. Because almost no one except male matrilineage headmen ever return to live on their own matrilineal lands, Trobrianders expend much time and energy growing and exchanging yams and women's wealth to insure the well-being of their dispersed matrilineages and to bury their own and affinal dead. We could probably never reconstruct all this from settlement patterns and material remains alone, but we can at least begin by always acknowledging the generic (at times purposeful) complexity of social fabrics.

It is into this fabric of cross-cutting, overlapping, interpenetrating patterns of counterposed descent, affinity, cognation, alliance, and residence that social institutions such as lineages or houses get brocaded—that is, like figures in Maya weavings that are directly incorporated into the weave, not merely embroidered on top. In a very real sense, social institutions are never simply cut from whole cloth and dutifully donned. Instead, they emerge as configurations in an existing social fabric of interdependent nodes and intertwined relationships. Thus, we can at best define relative, not absolute, social arrangements, because whichever nodes and relations people in a society may privilege, the others always remain in play, lest the underlying social fabric unravel.

This means that models cannot simply focus on one set of relations without making clear why this set and not some other. Institutions such as lineages or houses do not simply emerge from nowhere but do so under certain circumstances out of particular antecedents for certain ends, and our models must specify as best we can what these might be. Also, models must stipulate the conditions of their applicability because no institution, whether as rule or practice, exists independently or exclusively of other institutions across all individuals and groups all the time in the society.

Such an eye for the intricacies of warp and woof, not just whole cloth, does two things. First, it saves us from interminable arguments about overstated differences between descent and alliance, lineages and houses, kinship and residence, because we must address how they all fit together, not which comes first or counts most in the last instance. Our models can become as flexible in scope and varied in focus as the analytical problems we address because we need not wed model or problem to wholesale typologies. Second, it focuses our attention and explanatory efforts on interconnections within specific cases rather than on global comparisons of different ones. Understanding Maya social organization thus becomes a matter not of finding ethnographic analogies from elsewhere but of thinking through how the different institutions—families, households, temples, palaces, dynasties, estates—brocaded into the social fabric of Classic Maya society each managed to pre-empt the demands of the others long enough and often enough to establish the patterns we see. Comparison and analogy come to apply to pieces of the puzzle, not to the puzzle as a whole.

Indeed, given the problematic nature of distinguishing lineages and houses, an important interconnection that might help clarify the nature of Maya social organization lies in its relation to Maya political organization and political economy: how did plaza groups relate to Maya palaces, Maya kin groups to Maya kingship, and who controlled what? Haviland (2001) places kin groupings in inverse relation to state power and class stratification; Freter (2004) cites the shifting value of and interest in labor from Elliot Abrams's work, and Chase and Chase (2004) suggest that plaza groups controlled production while paramount elite controlled exchange. These seem to me productive ways to give substance to the heterarchy of Classic Maya society in which plaza groups—by definition and however defined—and dynastic kingship literally and figuratively built the Maya world.

RESUMEN

Este comentario examina los problemas etnográficos y teóricos que surgen en torno a la cuestión recién planteada sobre linajes y casas en la sociedad maya clásica. Por un lado, los proponents de los modelos de linajes segmentarios pasan por alto las ambigüedades entre la filiación y la ascendencia, el grupo residencial y el corporativo en las etnografías sobre los mayas que se usan para construir analogías. Por otro lado, los partidarios del modelo de casa de Lévi-Strauss no ven las diferencias relativas y no absolutas entre las teorías de ascendencia y alianza que están a la base de los modelos de linaje y de casa y que hacen difícil, si no impossible, distinguir arqueológicamente entre los dos. Ambos se preocupan por tipos descriptivos y en su mayor parte estáticos y derivados de otras culturas en vez de constuir modelos que definen las partes integrantes de la organización social tales como la filiación, la ascendencia, la alianza, la residencia y luego teorizar cómo los patrones diferentes de relaciones entre estos componentes puedan explicar los agrupamientos institucionales que se encuentran sobre (o en el caso de la arqueología, bajo) el tierra.