NINE

Understanding Households on Their Own Terms

Investigations on Household Sizes, Production, and Longevity at K’axob, Belize

H. HOPE HENDERSON

INTRODUCTION

This chapter applies the ethnographic model by Wilk and Netting (1984) of household economic organization, which predicts how households will internally organize production given differences in household sizes, to an archaeological study of household variability at K’axob, Belize. In presenting this study, I will discuss some of the strengths and weaknesses of defining prehistoric households in terms of what their members do, that is, as the union of the overlapping activity spheres of production, distribution, transmission, reproduction and coresidence (Wilk and Netting 1984). I argue that before archaeologists reconstruct these activity spheres, they should first question how households coordinated basic domestic activities. This more explicit perspective strengthens archaeological reconstructions of household phenomena by directly questioning the relationships among household structures, activities, and adaptability. I expand on these ideas by examining the possible relationships among internal household composition, domestic activities, and household longevity. Finally, this chapter argues that Wilk and Netting’s (1984) household model is similar to current studies of agency and structure (Brumfiel 1992; Dobres and Robb 2000) because it overtly questions the relationship between behavior and structure.

There are three parts to this chapter. First, I discuss some of the challenges of conceptualizing and applying Wilk and Netting’s ethnographic households to archaeological contexts. We need to think more carefully and discuss more explicitly how we reconstruct households with archaeological data. I also discuss the relationship between Wilk and Netting’s ethnographic household and current studies of agency and structure that consider the relationship between choice and behavior at the group level. Second, I summarize the results of an investigation of internal household organization by analyzing variation in household sizes, staple crop production, and wealth differentiation in a sample of ancient lowland Maya households from K’axob, Belize (Henderson 1998, 2003). This study shows that all households in this farming community followed a diverse productive pattern focused on a variety of staple foods but that there were meaningful differences between larger and smaller farming households. Larger households were better able to pool labor and resources to produce an even more diverse array of staple foods. Smaller households, by comparison, focused more of their labor and resources on maize agriculture. Interestingly, the broader economic approach of larger households was more successful. Larger households were wealthier, with more elaborate architecture, and featured more sequential occupations, more than double the number of smaller households. Third, I question why larger households were more prosperous and longer-lived than smaller households. Based on the results of this study, I consider how internal hierarchies and household leadership could have consistently facilitated a diverse pattern of staple crop production and strengthened household longevity at K’axob.

THINKING ABOUT THE ARCHAEOLOGICAL HOUSEHOLD

In their seminal chapter on households, Wilk and Netting encouraged anthropologists to understand households in terms of what they do (see Douglass and Gonlin, this volume). Instead of relying on functional typologies, they argued that anthropologists needed to directly observe the activities that are most commonly carried out and shared by household members (Wilk and Netting 1984:2–6). Investigators should observe and map individual activities to understand how human groups come together to carry out mundane tasks. In an approach taken by many authors in this volume, the product of household analysis, then, is the definition of activity spheres that show where and how individual and group activities coincide. This behavioral focus is amenable to archaeological analysis and is well oriented to archaeological data. However, it is difficult for archaeologists, far removed from observing how people cooperated in the past, to document the relationships among individual activity spheres. Archeologists cannot directly observe the cooperative efforts of people who coordinated different activities, and the lack of this information creates an analytic dilemma for archaeologists who want to reconstruct and analyze households in the past. Unless archeologists begin household studies by questioning how households internally managed life-supporting activities, it is unlikely that we can deduce household boundaries and think about how shared activities would have brought together household members, created friction, or even pulled them apart. As a result, archaeologists run the risk of assuming that the household was an adaptable social unit characterized primarily by unity and shared activities. We assume functionality instead of analyzing it. It is in keeping with Wilk and Netting’s original ethnographic model, which clearly distinguishes between structure and behavior (Wilk and Netting 1984:2), to question the potential relationships among different kinds of internal household organization and household activities (Wilk and Netting 1984:6). What Wilk and Netting originally proposed is that researchers treat household activities and morphological classifications with the same “explicitness” (Wilk and Netting 1984:4). By considering how households coordinated tasks before reconstructing activity spheres, archaeologists are in a better position to achieve a critical evaluation of household structures and activities. In fact, Wilk and Netting recognize the need for historians and archaeologists to start their studies based on their knowledge of the morphological characteristics of households (Wilk and Netting 1984:6).

This means we need to think more carefully about the many possible relationships among different household forms and the long-term success of these small social formations. We should not treat household organization in a deterministic manner or assume household functionality or unity. We need to start our work by explicitly asking about those possible relationships. Fortunately, archaeologists have an analytical advantage in documenting and observing human behavior over long periods of times. By comparing the histories of numerous households, archaeologists are in a stronger position to empirically evaluate household adaptability or functionality by considering whether household organization was related to household longevity. Comparing the lifespans of prehistoric households strengthens this internal focus. By focusing directly and separately on internal household organization, individual histories of residential groups (Hirth 1993), and domestic activities, archaeologists can offer a long-term perspective on household longevity that questions whether group cooperation in carrying out shared tasks was adaptive to the household group or individual household members. This overall evaluation of household longevity would constitute an interesting contribution to anthropological studies of households. Even so, considering these internal relationships does not mean that external factors are secondary, for the long-term success of households is also contingent upon larger social, economic, and political processes.

One example of an archeological study that considers how internal organization influences household activities over time is Widmer and Storey’s analysis of a Teotihuacan apartment compound (Storey and Widmer 1999; Widmer and Storey 1993). They identified a compound-wide cognatic kin group as having been responsible for coordinating and maintaining religious and administrative activities while smaller, extended family groups cooperated in craft production at Tlajinga 33 (Widmer and Storey 1993:103). They also found that shifts in production, from lapidary to pottery, were related to changes in household wealth and mortuary rituals, for through time the number of adult burials with offerings decreased in this household compound. Their reconstruction of household phenomena, which demonstrates the differences between larger and smaller groups within the same residential compound, shows how changes in what households produced was related to household prosperity. In a similar approach, Hendon (1996) argues that archaeologists need to visualize domestic relationships, such as women’s roles in craft production and food preparation, to understand specifically the value of labor and, more generally, household phenomena.

Furthermore, by considering internal household relationships and productive activities, these perspectives question household unity and functional adaptability and enable us to examine households both in terms of structure and agency (Brumfiel 1992:558–559). Studies of agency and structure, though varied (see Brumfiel 2000), are within the spirit of Wilk and Netting’s (1984) original ethnographic model, which explicitly questioned the relationship between structure and behavior and recognized internal variation in the ways household groups come together to coordinate the activities of production, distribution, transmission, reproduction, and coresidence. The approaches of Storey and Widmer (1999) and Hendon (1996) coincide with studies of structure and agency by showing how choice and strategic action, analyzed in these studies as productive activities, can vary depending on gender and economic differences. Moreover, Brumfiel’s call for subject-centered analyses that identify social and economic variables and evaluate them in relation to specific behavioral strategies (Brumfiel 1992:559) is complementary to household analyses. In this case, the internal composition of the household group, rather than the individual, is the “subject” of study. Dobres and Robb (2000:11) suggest a similar approach to Brumfiel (1992) by identifying group agency as the study of the cultural processes by which groups are constructed, negotiated, and transformed. I think there is much similarity between Wilk and Netting’s discussion of households and a more general theoretical search for less mechanistic models of human behavior on the part of archaeologists. All these anthropological approaches question deterministic models of social organization and change.

ARCHAEOLOGICAL HOUSEHOLDS FROM K’AXOB, BELIZE

These interests grow out of a study of prehistoric households in the Maya Lowlands that examined variability in a sample of seventy-two household occupations that dated to between the ninth century BC and the 9th-ninth century AD in the small settlement of K’axob, Belize (Henderson 1998, 2003). The principle objective of this work was to determine whether differences in household size were related to the ways households internally managed labor and organized staple crop production. To do this, I elaborated on Wilk and Netting’s ideas of simple and diverse production to see whether the formation of larger corporate households at K’axob coincided with a more diverse pattern of production consisting of more activities focused around more kinds of resources (Figure 9.1). I also wanted to see whether smaller households followed a simpler form of production, which consisted of a few productive activities focused on a few kinds of resources such as Zea mays (see Figure 9.1). For Wilk and Netting (1984), the ways households schedule major productive activities distinguish simple production from diverse production. Unable to document how households annually ordered specific tasks, I conceptualized household production as the cumulative results of how farmers coped with an array of potential productive alternatives and limitations. Following Wilk and Netting (1984:6), I reasoned that annual productive strategies could vary for any number of reasons and that any decision could also bring about unintended organizational consequences (Henderson 1998). Instead of assuming that a single rationality motivated household production, I broadly questioned whether households internally managed resources and labor in organizationally different ways because of their overall size differences. If this was the case, then over long periods of time the cumulative results of a wide variety of decisions related to staple crop production could result in qualitatively different productive strategies for larger and smaller households. I was more interested in finding average and long-term differences among many households as evidence of multiple household production strategies than in looking for annual or individual differences in production among a few households. In this respect, this study began as an investigation into variation in group-level agency (Dobres and Robb 2000) and a subject-centered analysis (Brumfiel 1992:559) that assumed farming households creatively managed production based on their own internal criteria. Did smaller households tend to focus more on maize agriculture? Did larger households tend to produce a wider array of staple foods?

Differences in household size were reconstructed by comparing the spatial layouts of sixty-two household occupations. In this analysis I differentiated between two size categories: larger, patio-focused corporate households and all other smaller households (Henderson 1998, 2003; see also discussion by Ciolek-Torrello, this volume). Long-term patterns in staple crop consumption and production were reconstructed by comparing the stable bone isotopes (i.e., λ¹³C collagen, λ¹³C apatite, λ¹⁵N collagen) of twenty-five adults from twenty-one different household occupations (Henderson 1998, 2003). Here I summarize the results of that research rather than presenting detailed research methodologies and analyses. I also present in more detail the stratigraphic histories of specific households.

Figure 9.1. Model of household organization and staple crop production

To conceptually link household consumption to household production, I drew on ethnographic research indicating that households produced the staple foods they consumed on daily basis (Netting 1993:18; see also chapters by Gonlin, Neff, and Wiewall, this volume). Thus, I assumed that this was the case for households at K’axob. I realize that not all scholars will agree that that the consumption of staple foods accurately reflects all staples that households habitually produced. Stable bone isotope data, which measure the consumption of carbohydrates and proteins (Ambrose 1993), may not reflect the staple foods that households produced for exchange or tribute (Hastorf 1990). To measure diversity in productive patterns, however, it is not necessary to document all of the staples that households produced. Farming households may or may not produce staples for exchange or tribute, but they have to produce crops to meet their own subsistence needs. Therefore, subsistence crops provide a baseline for measuring diversity in production, and stable bone isotope data directly measure average and composite dietary patterns. For the purpose of this study then, human remains provide the best kind of data to consistently reconstruct and compare dietary patterns between large and small households over three different time periods. A diverse diet in this population is represented by low λ¹³C apatite values, indicating that adults consumed high proportions of plants with a C₃ pathway. In northern Belize a wide variety of staple foods have a C₃ pathway, and presumably, a diet rich in plants with a C₃ pathway contains numerous kinds of staple foods (Bronson 1966; Hammond and Miksicek 1981; Hather and Hammond 1994; Hellmuth 1977; Miksicek 1983, 1991; Wiseman 1983a, 1983b). Most researchers agree, on the other hand, that Zea mays was the principal C₄ crop consumed by Maya populations (Gerry 1993; Gerry and Krueger 1997; Reed 1994; Tykot, van der Merwe, and Hammond 1996; White 1997; White, Wright, and Pendergast 1994). Thus, a diet that included many kinds of plant species would register low δ¹³C apatite values, indicating a high percentage of plants with a C₃ pathway and a low percentage of plants, most likely Zea mays, with a C₄ pathway. Furthermore, the presence or absence of terrestrial, freshwater, or marine resources in the diet, as reflected in δ¹⁵N collagen values, also indicates diversity in consumption and productive activities (Wright and White 1996). This study reconstructs average and long-term consumption of staple foods to generally infer which categories of plants or animal foods (i.e., C₃ plants, C₄ plants, types of protein) were more or less important in household productive strategies.

I envisioned a simple productive pattern as one in which households regularly relied on small work groups that pooled their labor to carry out a few productive tasks focused on a few key resources (see Figure 9.1). Thus, I expected that households would have selectively relied on the cultivation of seasonal crops such as Zea mays and minimized the diversity in the number of staple resources that they habitually exploited. If so, then relatively high δ¹³C apatite values would indicate that individuals consumed a high percentage of Zea mays. Meat may have been an irregular source of protein in adult diets if smaller households dedicated less time and labor to hunting activities. I also would not expect for marine foods to be a staple resource for households following a simple productive strategy. Moreover, I envisioned households practicing simple production as a small and homogeneous social group with little internal differentiation. To the extent that age, gender, or other distinctions were present in these households, I did not expect to find evidence that these divisions significantly changed the ways people pooled labor or staple foods. Thus, at both the household and individual levels, I expected to find little internal variation in stable bone isotope values. Finally, I expected to find evidence of simple production in the many small household remains that were located beneath single mounds (Figure 9.2).¹

Diverse production, unlike simple production, favored an expansion and reorganization of the household labor pool. In diverse production, households organized and participated in more types of productive activities and expanded their staple resource base (see Figure 9.1) (see Douglass and Gonlin, this volume). This economic pattern relied on larger labor pools capable of exploiting a wider variety of raw resources and coordinating more kinds of productive activities. Given the larger labor pool, households following a diverse form of production would have had greater flexibility within the annual agricultural cycle of clearing fields, planting, weeding, and harvesting. The larger corporate households at K’axob could represent such an expansion of the household labor pool. They may have also participated more regularly in hunting terrestrial animals or gathering marine or freshwater resources. I also expected that corporate households, composed of several families, may have developed internal hierarchies that changed the ways that these households allocated and pooled resources as they exploited more staple resources and carried out more kinds of productive activities. As a result, some individuals within these households may have consumed differing proportion of staple foods, either carbohydrates or proteins. If so, then at the individual and household level, stable bone isotope values should be more varied. I expected to find a diverse productive pattern in the larger corporate household remains present beneath basal platform mounds (see Figure 9.2).

Archaeologists document the material remains of residential areas as a basis for defining household phenomena. At K’axob, each household is represented by an occupation, which consists of a single stratigraphic layer, usually the portion of an interior and exterior surface associated with a single residential structure (Henderson 1998, 2003). Residential space at K’axob consisted of structural foundations, such as low plaster floors, that were joined to earthen or plaster exterior surfaces. Excavations habitually uncovered remains of domestic pits, postholes, burials, and cache deposits intruding into structural foundations and exterior surfaces (Henderson 1998, 2003). The seventy-two household occupations identified by this study had these recurrent features, which suggest that these stratigraphic layers were the material remains of habitation spaces (Henderson 1998:73–129). There were three nonresidential occupations documented in this sample, which were special purpose work areas, and these occupations did not have the recurrent features associated with residential spaces (Henderson 1998:42–51). To facilitate household studies at K’axob, I first classified the complex stratigraphic sequences of superimposed residential structures and ambient exterior spaces from twelve excavations into separate household occupations. In this region, the Maya renovated and rebuilt domestic architecture in the same location so that residences were sequentially built on top of one another (Ettlinger 1983; Hammond 1991; Harrison 1990; Levi 1993; McAnany and López 1999; Pyburn 1987; Sullivan 1991). The presence of superimposed residential structures spanning multiple time periods provided the opportunity to sample and identify households as separate stratigraphic layers. Each excavation unit thus identified from three to eleven superimposed residential occupations, making it feasible to document and compare diachronic changes in households at K’axob. This analysis is inspired by the Harris matrix system that groups temporally related features into an analytic unit called “phases” (Hammond 1993:table 9.1; Harris 1989). The seventy-two household occupations that I defined based on separate stratigraphic layers of residential structures and ambient spaces are the equivalent of a “phase” in the Harris matrix system. This method of analysis is also comparative to the concept of “household series” proposed by Hirth (1993).

Figure 9.2. Location of excavation units in basal platform mounds and single mounds at K’axob, Belize

Since this study was focused on long-term patterns, I compared households from three broad time periods: the Middle Formative (800–400 BC), the Late Formative (400 BC–AD 250), and the Classic (AD 250–900). A more precise dating of household occupations, such as the average length of each occupation, was beyond the scope of this study. Instead, the household occupations presented here represent periods when the entire residential area was rebuilt.

VARIATION IN HOUSEHOLD SIZES, STAPLE CROP PRODUCTION, AND WEALTH DIFFERENTIATION

This study documented variation in household sizes, staple crop consumption, and wealth differentiation at K’axob (Table 9.1). Based on these findings, I inferred that household production strategies differed between larger and smaller households. Since these subtle differences in staple crop consumption and household size were also related to wealth differences and longer and shorter household occupations, I also inferred that the ability of some households to expand in size, diversify their resource base, coordinate larger labor pools, and produce a wider variety of staple foods was related to household longevity (see Douglass and Gonlin, and González Fernández, this volume). In earlier works, I have argued that households at K’axob, from the ninth century BC to the ninth century AD, were largely independent and that regional elites had only an indirect effect on the production strategies of farming households (Henderson 1998, 2003).

Investigators working in Mesoamerica have drawn on archaeological, ethnological, and ethnohistorical data to distinguish between the spatial arrangement of larger corporate households and smaller households (Carrasco 1976a, 1976b; Flannery 1976; Flannery and Marcus 1983; Hayden and Cannon 1982; Ringle 1985; Ringle and Andrews 1988; Wilk 1988, 1991; Winter 1974, 1976). These sources provided a basis for this research because they show that corporate households conformed to a specific patio-focused spatial layout and that corporate households were larger than other types of households (Henderson 1998:130–136). For example, Carrasco documents compound households, or cemithualtin, which featured two to four houses around a central patio, in a 1540 colonial census from Molotla, located in Yautepec, a political territory of the Valley of Morelos, which had on average 5.2 married couples and 23.2 people (Carrasco 1976b). The total population range for these corporate households was twelve to thirty-five people.² Similarly, Farriss estimates that prior to the Spanish conquest, larger households in the Maya Lowlands contained twenty to thirty adults and children (Farriss 1984:134). Drawing generally on these ethnohistoric sources, I estimate that larger households at K’axob had twelve to thirty-five people (Carrasco 1976b; Farriss 1984). At K’axob, the clearest category of larger households was the large corporate residence, which included two to six structures joined around a paved plaster patio. These were the largest households at K’axob and they were only present in basal platform mounds. I was able to identify the formation of these larger households in excavation units by the presence of a central plaster patio surface that was joined to residential structures, changes in structure orientation that indicated that several structures were reoriented and built around a central patio, and the repetition of this spatial layout in subsequent household occupations within each residential mound (Henderson 1998, 2003). These three attributes consistently indicated a change in the composite spatial layout and an expansion in the aggregate size of residential areas. Once built, this spatial arrangement was replicated with all subsequent occupations, and I inferred that the formation of these larger households group was lasting.³ In all, I found that 61 percent of all household occupations from excavations in a total of seven basal platform mounds corresponded to the spatial layout of larger corporate households (Henderson 1998, 2003).

Table 9.1. Variability in household occupations at K’axob, Belize

Smaller households at K’axob had a wide range of variation in overall size. For example, I found cases of single residential structures. These simpler and smaller residential areas were common in single mounds and basal platform mounds. I also found cases of intermediate-sized residential areas, which included several informally arranged residential structures. These households differed from the largest households at K’axob because their total area was smaller and because individual structures were not architecturally joined together by a central plaster patio. Even so, in differentiating larger corporate households from all other-sized households at K’axob, I came to realize that large coresidential groups of seven to twelve people, a size range documented for households in the Colonial period for Yucatán (Kurjack 1974:16), were probably common at K’axob. Similarly, based on a 1583 colonial census from Tizimin, Farriss found an average of 9.4 persons per household (Farriss 1984:134). Farming households at K’axob were similar to households in colonial Maya communities that seemed to value large coresidential groups (Wilk 1988; Kurjack 1974:16).

The stratigraphic sequence from excavation Unit 10 is an example of a smaller residence that expanded into a larger household associated with a corporate residence (Figure 9.3).⁴ All five occupations correspond to the Late Formative period (400 BC–AD 250). While the first occupation was a nonresidential work area (Henderson and McAnany 1996), the second occupation represents a smaller household that featured a single apsidal-shaped structure built over an earthen ground surface (see Figure 9.3). The third occupation featured a continuous exterior plaster surface, which I interpreted as evidence of the central patio of a larger corporate residence. The fourth occupation had a variety of domestic features intruding into the central patio of the residence: a portion of a structure, a lip-to-lip dedicatory cache, two adult burials, and a domestic pit filled with fire-cracked rocks. The final, fifth occupation featured a central plaster patio surface and a single adult burial. Thus, after one occupation by a small household, residents here built a larger corporate household that persisted for three more sequential occupations. One adult individual, buried in the central patio of this household’s last occupation (i.e., Zone 23), was included in the random sample of adults for bone isotope analysis. An analysis of stable bone isotopes from this individual registered a λ¹³C apatite value of –10.600 percent, which suggests that Zea mays accounted for approximately 26 percent of his/her carbohydrate diet.

Figure 9.3. Plan view drawings of occupations 2, 3, 4, and 5 from Unit 10, K’axob, Belize

Another example of the transition to a larger corporate household was found in excavation Unit 14, where a series of five smaller households were followed by three occupations of larger corporate households (Figures 9.4 and 9.5). The first four occupations date to the Late Formative period (400 BC–AD 250) and the last three occupations date to the Classic period (AD 250–900). Here, the first occupation consisted of a single posthole and two domestic pits excavated into bedrock. If occupation 1 represents the remains of a residential area, then it was a relatively small household. The next three occupations all consisted of ground-level structures built over earthen surfaces. In particular, Occupation 3 had an unusual number of primary features present in the excavation unit. Four child burials were placed around two small ground-level structures. A multiple burial of a six-year-old and two-year-old was placed in the corner of Structure 2, and a nine-month-old and a one-year-old were interred in the adjoining earthen ground surface. The presence of a midden with animal bones, large sherds of utilitarian pottery, and numerous small sherds was interpreted as a kitchen or food-preparation area of the residence.

Figure 9.4. Plan view drawings of occupations 3, 4, 5, and 6 from Unit 14: two burials from this sequence might reflect the position of a household head; the first example is the burial of a single adult individual in Z35, which is covered by the steps (i.e., Z28) to Structure 6b; second is the example of the burial of a single adult individual on the central axis of Structure 6b, K’axob, Belize

The construction of residential space expanded outward with Occupation 5. Excavations documented a rounded platform structure built over a plaster exterior surface. Unlike previous occupations, there were no domestic pits or middens associated with Occupation 5. Despite the presence of a plaster exterior surface, I did not classify this occupation as a larger corporate household because structure orientation did not change with this occupation. Only with Occupations 6, 7, and 8 was it clear that residential architecture was reoriented to face a central patio area. Residential architecture in these larger corporate households was more elaborate: structure shape was rectangular or square; structural foundations were built as low platforms; and internal divisions within structures indicate a general trend toward more segmented domestic space. Work areas, represented by domestic pits and middens, were not as commonly located inside or next to residential structures. This pattern is common and distinctive in large corporate households, especially those from the Classic period at K’axob (Henderson 1998).

Figure 9.5. Plan view drawings of occupations 7 and 8 from Unit 14, K’axob, Belize

Three adult individuals from Occupations 6, 7, and 8 were randomly selected for stable bone isotope analysis (i.e., Zones 4, 14, and 22). Bone isotope analysis of λ¹³C apatite from the adult interred during Occupation 6 registered a value of –9.200 percent, which suggests that his/her average Zea mays intake was approximately 35 percent. The individual from Occupation 7 had a λ¹³C apatite value of –9.600 percent, suggesting that Zea mays consumption was around 33 percent. Finally, the individual buried during the final occupation had a λ¹³C apatite value of –10.100 percent, which averages to around 29 percent Zea mays consumption.

Stable bone isotope analysis of twenty-five adults from twenty-one different household occupations and dating to the entire occupational sequence of K’axob indicated that all households followed a diverse pattern of production (Henderson 1998, 2003). The mean λ¹³C apatite value of –9.430 ± 0.186 percent at one standard error (n = 23) suggests that on average the adult carbohydrate diet featured 34 percent Zea mays. Root crops (Bronson 1966; Hammond and Miksicek 1981; Hather and Hammond 1994; Hellmuth 1977; Wiseman 1983a), beans, squash, and a wide variety of tropical fruits (Miksicek 1983, 1991; Wiseman 1983a, 1983b) are staple foods with a C₃ pathway documented for this region that probably rounded out the average adult diet. The average λ¹⁵N value for the adult population at K’axob was 9.145 ± 0.144 percent at one standard error (n = 11). The average λ¹³C collagen value was –14.750 ± 0.242 percent at one standard error (n = 10). Taken together, average λ¹³C collagen and λ¹⁵N collagen values suggest that dog, deer, peccary, and turtles were likely sources of protein for farming households at K’axob (Henderson 1998, 2003; Wing 1981; Wing and Scudder 1991; Wright and White 1996). These findings are similar to dietary reconstructions for adult populations from the nearby sites of Cuello and Laminai (Tykot, van der Merwe, and Hammond 1996; White and Schwarcz 1989). These results are also similar to other studies of Maya populations with δ¹³C apatite data, which suggests that Zea mays formed only about 30 to 55 percent of the average carbohydrate diet (Gerry and Krueger 1997:202; Tykot, van der Merwe, and Hammond 1996:359). Finally, temporal analysis of adult diets from the Middle Formative, Late Formative, and Classic periods indicates that production and consumption of staple foods varied little through time. I did not find significant differences in λ¹³C apatite, λ¹³C collagen, or λ¹⁵N collagen values in average adult diets by time period (Henderson 1998, 2003). These results suggested that on average all-sized households followed a relatively diverse pattern of production from the ninth century BC through the ninth century AD. Household decision making and long-term production strategies were not as varied as I had originally imagined.

Even so, I did find evidence of subtle economic differences between larger and smaller households at K’axob, which suggests that household decision-making strategies differed slightly in terms of the number of resources exploited and labor organization. Larger corporate households followed a more diverse pattern of staple crop production than all other-sized households (Figure 9.6). The mean λ¹³C apatite value of –10.050 ± 0.849 percent at one standard error (n = 8) for corporate households suggests that, on average, adults from these households consumed 30 percent Zea mays. The mean λ¹³C apatite value of adults from all other-sized households of –9.100 ± 0.193 percent at one standard error (n = 15) means that, on average, adults from these households consumed 36 percent Zea mays. The 0.950 percent difference between larger and smaller households is significant (t = –2.733, df = 21, p = 0.011). Moreover, the 6 percent average difference in Zea mays consumption is meaningful because it is nearly one-fourth of the average variation documented for the entire Maya region using δ¹³C apatite values (Gerry and Krueger 1997:202; Tykot, van der Merwe, and Hammond 1996:359). This is the first archaeological study to show that individual dietary differences can be explained by examining variation in household labor organization and household size. In this case, a household-level analysis that considers how households internally managed labor and resources helped to understand variation in adult diets and subtle economic differences in the ways larger and smaller households managed resources and labor. Moreover, these findings suggest that the model of the household as a cooperative task group is appropriate for describing past households from K’axob.

Figure 9.6. Comparison of box-and-dot plots for λ¹³C apatite and δ¹⁵N values from adults associated with corporate and smaller households, K’axob, Belize

Were these larger corporate households more internally stratified? I found evidence that suggests differences among individual household members in larger corporate households, for the standard errors of λ¹³C apatite, λ¹³C collagen, and λ¹⁵N collagen values were larger in corporate households than in all other households. For example, the mid-spread of adult λ¹⁵N values in corporate households is more than twice as wide as that for adults from smaller households (see Figure 9.6). This difference means that adult diets in corporate households were more heterogeneous than those in other-sized households. Adults from corporate households were less likely to consume the same proportions of C₃ plants and the same amount or quality of proteins. This study falls short of showing how food was unequally distributed within larger households. Indeed, more research is needed to show how food preparation and redistribution within households may have changed with the formation of larger corporate households. Future investigations could expand on these finding by systematically examining dietary variation in adults from the same household. Ethnohistorical sources from Mesoamerica explain various arrangements that resulted in intrahousehold differentiation (Carrasco 1976b; Evans 1993; Farriss 1984:132–239; Lockhart 1992: 59–93; McAnany 1995; Williams and Harvey 1997:42–48). By combining more detailed dietary analysis with specific models on internal household organization, future research could explain the extent to which internal stratification influenced individual dietary patterns.

An additional source of variation between larger and smaller households was related to length of occupation. Some of the smallest households at K’axob, those found in single-mound locations, followed a less diverse production pattern with each household occupation. There is a moderately strong correlation between length of occupation (X) and λ¹³C apatite values (Y) (r = 0.690, p = 0.021, y = 0.896x—10.950). Length of occupation helped to explain 69 percent of the variation in λ¹³C apatite values in households located beneath single mounds. These findings suggest that some of the smallest households at K’axob had a more restricted productive capacity that made a diverse pattern of staple crop production more difficult to maintain with each sequential occupation. The long-term differences in the production of small households indicate different household productive strategies within the general context of diverse production at K’axob. While larger households were able to further diversify production, smaller households were unable to maintain a diverse productive pattern. McAnany has suggested that these smaller households had less access to agrarian resources during the Classic period (McAnany 1995).

Surprisingly, I also found that subtle differences in staple crop production and household size were related to long-term patterns in wealth differences between larger and smaller households at K’axob. These results suggest that the ways households internally managed labor and resources were related to wealth inequality. For example, larger corporate households on average built low platforms and used sascab construction fill, quarried from bedrock, to construct more elaborate residences. On average, 70 percent of corporate households in this sample used sascab construction fill compared to the 17 percent of other households that had sascab construction fill (X² = 21.20, p = 0.0005, v = 0.63). Similarly 63 percent of corporate households in this sample had platform structures and only 16 percent of other households had platform structures (X² = 12.12, p = 0.0005, v = 0.49). These labor-intensive construction techniques first occurred during the Late Formative period and coincided with the formation of corporate households. Additionally, corporate households featured a longer sequence of occupations. On average, larger corporate households (n = 25) had 5.84 occupations and other-sized households (n = 34) had 2.5 occupations (t = 5.926, df = 57, p = 0.0005). Larger residential groups were better able to socially and economically reproduce themselves through time and to continue constructing their houses over those of their ancestors (McAnany 1995). Smaller households, on the other hand, were less likely to rebuild their residences in the same location. Slight differences in household productive strategies had different long-term consequences for larger and smaller households.

THINKING ABOUT HOUSEHOLD LONGEVITY

What I find interesting about these results is the idea that the ways household members pooled labor and resources to maintain a diverse pattern of staple crop production did not necessarily result in household longevity or prosperity. To understand why larger households at K’axob had more long-term success than smaller households we need to ask what made these groups more resilient and prosperous. Why did larger households have on average double the number of occupations of smaller households? We also need to ask why a more diverse pattern of production was not as tenable for smaller households. These issues are explored and questions raised for future research about household leadership.

As a general observation, larger labor pools gave larger corporate households more economic flexibility. Beginning in the fourth century BC, these larger households were able to participate in a wider range of productive activities. While some scholars have noted that prehistoric farming households were limited by a narrow range of economic opportunities or were inherently conservative regarding production strategies (Hirth 1993; Sahlins 1972), the results of this study suggest that farming households were able to diversify staple crop production. It is important to note, however, that changes in household production were subtle. The production of staple foods was stable in this community from the ninth century BC through the ninth century AD. The formation of larger households beginning in the fourth century BC meant that some households were able to expand an already diverse productive pattern. Other households had more difficulty producing a diverse array of staple plant foods. In this regard, control over larger labor pools gave larger households at K’axob an advantageous flexibility that contributed to their general prosperity and longevity.

The slow formation of larger households at K’axob also lends support to the argument that larger households were able to expand production for their own prosperity. While some larger households began forming during the Late Formative period, they did not become common until the Classic period at K’axob (Henderson 1998, 2003). This gradual pattern differs from that at other lowland Maya Late Formative communities such as Cuello and Komchen (Hammond 1991; Ringle and Andrews 1988), where larger corporate households formed more rapidly. At K’axob, changes in household composition, production, and wealth inequality coincided with the emergence of regional elites at the nearby communities of Nohmul and San Estevan, but the slight nature of these changes suggests that the staple foods that households produced and how they managed production were local affairs and not the prerogative of regional elites. If regional elites had directly controlled staple crop production, I would expect to see more corporate households forming during the Late Formative period and greater changes in the production of staple foods, especially the increased production of staple crops such as maize.

I also suspect that the formation of these larger households involved a social change in the formal composition of the household that contributed to the resiliency of larger households. The architectural pattern of two to six structures physically joined around a central patio area meant that the way households expanded or rebuilt was more structured and restricted. In other households, individual residences were not connected to one another and new residences could always be built closer or farther from existing residential structures. I think that this architectural difference is indicative of differences in the internal leadership and the cohesiveness of larger households. Leadership and authority may have differed from other households in two ways. First, I suspect that household leaders were more successful in coordinating how household members pooled labor, diversified their resource base, and redistributed resources within the group. Second, household leaders probably managed inheritance in such a way that permitted larger coresidential groups to more easily reproduce themselves through time. Household leadership in corporate households was potentially more multifaceted than leadership in smaller households. Future research could expand upon ethnohistorical sources that describe intrahousehold inequality and differentiation to better understand leadership within larger households (Carrasco 1976b; Evans 1993; Farriss 1984:132–239; Lockhart 1992:59–93; McAnany 1995; Williams and Harvey 1997:32–48; see Wiewall’s discussion of ethnohistoric source descriptions of household organization later in this volume).

The formation of larger corporate households also seems to have coincided with a more hierarchal internal structure. Why would a more hierarchal social structure facilitate household prosperity and longevity (Wilk 1989)? This organization, while not equally favoring all household members, may have contributed to household longevity if leaders were successful in generating consensus within the household and in defending the household’s rights and obligations within the community or larger region. For example, household leaders may have wielded more influence within lineages eager to maintain resource rights (McAnany 1995) or in community decisions, such as when to plant crops (Wilk 1991), beginning in the fourth century BC. If this was the case, then a household leader or group of leaders of corporate households may have had privileged positions in mediating rights and obligations within their communities that better responded to the needs and interests of larger households. Farriss (1984:139) has noted that there is no contradiction between a corporate group that shares a set of reciprocal rights and obligations and a hierarchically ordered group that recognizes a central authority that mediates these rights. If this kind of internal differentiation can be demonstrated for larger households at K’axob, then future investigations have the potential to show whether leadership strategies and economic flexibility gave the largest households at K’axob an advantageous position within their communities that smaller households were unable to achieve or perpetuate from one household occupation to the next.

The spatial location of several tombs in larger corporate residences seems to support the idea of internal hierarchies, a household head, or a single high-status individual. These adults are buried beneath the entrance to residences or along the central axis of residences. In the stratigraphic sequence from Unit 14, a tomb of a single adult was capped by three large stones, which served as the steps or entryway into Structure 6 (see Figure 9.4). Here the construction of residential architecture and the tomb of one high-status individual are conjoined. Following this construction episode, another high-status adult individual was buried along the central axis of Structure 6. Both of these burials were incorporated as central elements in residential architecture and featured elaborate burial treatment, which included stone-lined crypts and several ceramic vessels. In Occupation 8, the final residence documented by Unit 14, three burials were placed on the front of a residential structure, but only one of these (Zone 8) featured a partial stone crypt (see Figure 9.5). Moreover, this tomb was later entered and the cranium of the adult individual was removed (Zone 3), a practice associated with ancestor veneration among the lowland Maya (McAnany 1995). These three different burial contexts are what I would expect to find if internal hierarchies were important to larger corporate households at K’axob.

If the ways those larger households managed resources and labor was related to the elaboration of household leadership positions, future research would need to demonstrate the presence or absence of similar internal hierarchies in smaller households. If the shorter occupational histories and reduced capacity of single-mound households to diversify production with each occupation was related to relatively weak internal leadership and a less-unified social group, the tombs of adults should show different spatial patterning and should be less elaborate than those from larger corporate households. Larger horizontal excavations are needed to better understand the spatial location and differentiation of burials in these smaller households.

The results of this research are not conclusive but they suggest that archaeological studies of households can benefit from an internal focus that considers the multiple ways in which households may make decisions. Subtle and gradual changes in household size and production were related at K’axob, even though long-term patterns in household size and staple crop production were stable from the ninth century BC through the ninth century AD. The prosperity of larger households, beginning in the fourth century BC can be understood, in part, as a function of their ability to manage larger labor pools and coordinate a wider variety of productive activities and resources. Future research should build on these results by asking whether these larger corporate households were internally more hierarchal and whether a household head or a privileged group of individuals managed economic and political relationships in a way that enhanced household productivity and longevity. Complementary lines of data on household production, especially the role of craft production (see Henderson 2003), would greatly enhance the information and analysis presented here.

SUMMARY AND CONCLUSIONS

I have argued that Wilk and Netting’s original emphasis on the manifold relationships between household structures and activities is in the spirit of current studies into group-level agency (Dobres and Robb 2000) or subject-centered analyses (Brumfiel 1992); in Wilk and Netting’s work, households are creative and dynamic social formations. I have also argued that meaningful archaeological applications of this model require archaeologists to consider more critically how households coordinated basic tasks before they embark on reconstructing activity spheres. The research agenda on household phenomena should include a direct questioning of the relationships among household structures, activities, and longevity. This perspective enabled me to identify two general household productive strategies with different long-term consequences for larger and smaller farming households at K’axob, Belize.

I realize that some scholars may argue that the long-term patterns and the group-level agency presented here ultimately tell us little about specific practices or household decision makers. Some will not be satisfied with the idea that larger households prospered by producing more staple foods with a C₃ pathway from the fourth century BC onward. For those interested in recovering a sense of individual narratives and action in past societies (i.e., Hodder 2000), productive strategies that are the cumulative result of habitual practices and many annual agricultural cycles may be too far removed from what farmers and households at K’axob actually did at different historical moments to be of much relevance. Nonetheless, questions of agency and practice, recognized by many as a “good thing” for current theory, stem from different theoretical approaches (Brumfiel 2000; Dobres and Robb 2000), and the long-term perspective is justified for several reasons. First, our own assumptions about agency and structure, and individuals and social groups, may delimit to some degree our own ability to recognize unique social arrangements, behaviors, and consequences. I found, for example, that in an economic sense, households at K’axob were less “creative” than I had originally assumed when I proposed simple and diverse production. The small differences found in household size and staple crop production were something of a surprise since I did not expect to find such a conservative pattern of household organization or production. However, once I identified these historical patterns, I was better able to appreciate the subtle variations between smaller and larger households. I could understand both smaller and larger households in terms of the ability of some groups to elaborate on existing cultural practices and the inability of other groups to maintain these cultural practices. This awareness enriched my perspective on household prosperity. Here the concept of prosperity encompasses the ability of some households to further elaborate on traditional agricultural practices and knowledge oriented to a wide variety of plant and animal foods and to pass on the ability to carry out such practices to subsequent household generations. Only the largest households at K’axob attained this level of prosperity. One conclusion is that household studies and theories of agency in general are more fully realized through the reconstruction of long-term patterns.

Second, as Wilk and Netting note, household structures are often “compromises” among different and sometimes “contradictory imperatives” (Wilk and Netting 1984: 20). This observation means that the identification of specific practices, individual intentions, or a group’s rationality may singly never fully explain household phenomena, which is characterized by uncertainty as well as purposeful actions. Thus, if we are to understand households in terms of what they do, we must also consider how household members coped with opposing intentions and needs. This consideration means we will need a robust sense of agency when examining households. More explicit questions concerning the internal composition of households are needed in archaeological reconstructions to better understand the degree to which mutual cooperation and the negotiation of different interests were characteristic of household phenomena. Furthermore, the research presented here, while not conclusive, provides contextual information that could inspire smaller-scale analyses of individual or group practices. A long-term theoretical approach informs rather than excludes smaller scales of analyses.

Finally, I conclude by saying that Wilk and Netting’s notions about studying households in terms of what they do is a forceful proposition that should continue inspiring and challenging archaeologists to investigate household phenomena. Wilk and Netting challenge us to understand households on their own terms. The proposition that households and their members act not only in reference to larger macro-processes and institutions but according to their own particular histories, conflicts, concerns, and aspirations is a generous theoretical proposal for all scholars interested in understanding past households.

NOTES

1. The fields represented in Figure 8.2 are raised fields located in Pulltrouser Swamp that bordered the settlement. Excavation units, which generally measured from sixteen to forty-eight square meters, were positioned within the residential mounds that are depicted here (see Henderson 1998:table 2.1). Thus, this figure indicates the basal platform mounds and single mounds that were selected for excavations.

2. Carrasco labels five households labeled as cemithualtin and eleven households that shared kinship and economic relationships as “compound households.” I suspect that the cemithualtin, which literally translates as “those of one patio,” referred to the specific spatial layout of corporate households identified by this study. These five households had populations of 35, 17, 30, 17, and 39, thus averaging 27.6 individuals per corporate household (Henderson 1998:135).

3. Household sizes may change because of family growth cycles, postmarital residential obligations, differential demographic rates within populations, and even random factors (Hammel 1984). These fluctuations in household sizes and the reasons that larger households formed are not the subject of this study.

4. To conform to the Harris matrix system, each feature at K’axob was given a zone number. Thus, the earthen ground surface from excavation Unit 10 and Occupation II is numbered Zone 11 and is depicted in Figure 8.3 as Z11. Moreover, while occupation numbers are represented in these figures as I, II, and III to conform to the Harris matrix system, in the text they are written as one, two, and three.

REFERENCES CITED