1Although most Mayanists now accept the notion that the large settlements of the Classic lowland Maya were cities, scholars have been slow to analyze the residential areas of these sites using the concepts and methods of urban studies. Instead of employing urban concepts such as residential segregation, spatial zoning, or neighborhood structure, archaeologists have focused on concepts like lineage, house society, and heterarchy. The non-urban focus of Mayanist social analysis may have contributed to the lingering view that the large Maya sites were not quite urban in character.

2In order to advance understanding of the social dynamics of ancient Maya cities, scholars need to apply the kinds of concepts that inform urban research on modern and other historical urban societies. As a step in that direction, I describe several comparative case studies to support the claim that Classic Maya settlement clusters functioned as urban neighborhoods. A number of authors have suggested this interpretation, but without presenting any analysis or discussion (Hammond 1991, p. 266; Marcus 2004; Robin 2003, pp. 330-331; Smith 2010, p. 148). More recently, several studies of Maya urban centers have combined spatial analysis with the results of excavation data to interpret settlement clusters as neighborhoods (Arnauld et al. 2004; Arnauld et al. s. d.; Lemonnier 2011, s. d.). The present paper is intended to advance this line of research by providing a broader comparative framework.

Preindustrial Urban Neighborhoods

4In another work, I define neighborhood as follows: « A neighborhood is a residential zone that has considerable face-to-face interaction and is distinctive on the basis of physical and/or social characteristics » (Smith 2010, p. 139). In most or all premodern cities, neighborhoods are relatively small spatial zones whose creation and maintenance result from social interaction, mutual support, and other bottom-up or generative social processes. They often co-exist with larger residential zones created by municipal or state authorities for administrative purposes. I call these latter units districts. In this paper I focus primarily on social neighborhoods.

5In densely populated cities, neighborhoods are typically small in areal extent. They may correspond to a single street segment, an area around the intersection of two streets, or perhaps a secluded cul-de-sac (a common pattern in precolonial Islamic cities). Neighborhood residents typically know one another by sight and they interact in a variety of ways, often focused on local facilities (Suttles 1972, pp. 55-56). In many (but by no means all) premodern cities, one or more social parameters tend to be specialized at the neighborhood level. The most common of these parameters are ethnicity or race, place of origin, occupation, and social class (Saunier 1994; York et al. 2011). This is another way of saying that neighborhood residents often share one or more social attributes, and these attributes often distinguish the residents of one urban neighborhood from those of another.

6In many premodern (and modern) cities, neighborhoods are created by rural-to-urban chain migration (Greenshields 1980; York et al. 2011). They can also be created when commoners build their houses in the immediate vicinity of the residence of a noble or other important individual (see discussion of Addis Ababa below). In newly-settled urban areas such as shantytowns, neighborhoods are often formed through bottom-up processes of social interaction among neighbors. In the recently formed squatter settlement of George in Lusaka, for example, Schlyter and Schlyter (1979, pp. 50-53) report that neighbors helped one another in numerous ways and they socialized within small areas that can be considered neighborhoods. Alternatively, neighborhoods can be created through external, top-down processes, as in contemporary planned housing developments or ancient cities with planned, walled residential zones such as T’ang-period Chang’an (Heng 1999, pp. 22-29).

7Sometimes civic officials are present in neighborhoods for record-keeping, taxation, organization of labor, or other oversight activities. These may be local residents serving as officials, or they may be state or municipal functionaries imposed from above. Neighborhood members often participate in collective work projects involving maintenance of common facilities, and they typically take part in common ceremonies or festivities at the neighborhood or city level.

8As a result of these various forms of collective activity, day-to-day interaction, and shared social characteristics, neighborhood residents may develop a sense of common identity, and these identities may be expressed in informal or formal labels for neighborhoods (Smail 2000). The neighborhood attributes discussed above have been documented most fully for high-density premodern Old World cities and for contemporary western cities. In the remainder of this paper I argue that these same social processes and conditions characterized settlement clusters in premodern low-density cities, a category that includes the cities of Classic Maya society.

Lowland Maya Settlement Clusters

9Most scholars today agree that the large Maya settlements of the Classic period can be called cities (Arnauld and Michelet 2004; Chase et al. 1990; Ciudad Ruiz et al. 2001). For several decades this classification was resisted by many archaeologists, in large part because the population size and population density of the Maya sites were much lower than other Mesoamerican cities such as Tenochtitlan or Teotihuacan (Sanders and Webster 1988; Willey 1956b). Fletcher (2009; 2011) recently proposed « low density urbanism » as a phenomena that existed in a number of lowland tropical areas before the modern era, including Southeast Asia, India, Africa, and the Maya Lowlands (see note 1). One of the features of many low-density cities is the occurrence of spatial clustering in the distribution of houses, with large open areas between clusters.

[t]he clusters may be thought of as small hamlets, varying in their degree of isolation from one another, and sometimes including an apparently specialized building which may have been used in common by the residents of the cluster. The size of the grouping suggests occupation by a kinship group.

12The next major synthesis of Maya settlement patterns was the volume Lowland Maya settlement patterns (Ashmore 1981a). In her conceptual introduction to the book, Ashmore (1981b) stated that the cluster « is an extended household or aggregate of households and sometimes other units, and may occur as part of a larger settlement or in isolation ». She presented a four-type classification of Maya settlement clusters: informal cluster; homogeneous patio cluster; structure-focused patio cluster; and group-focused patio cluster. The latter type has several variants, shown here in Figure 2.

13Unfortunately, Mayanists made little progress in the social analysis of settlement clusters after 1981. Although Ashmore’s typology provides a solid basis for spatial and social research, few researchers adopted it (although see Ashmore 2007, pp. 54-55). Many scholars use terms like « house cluster » or « residential cluster » without formally defining the term (Fox and Cook 1996; Killion et al. 1989). One strand of research has tried to validate Bullard’s model of the cluster as a kinship group through analogy with modern Maya settlements (Freter 2004; Willey 1981), an interpretation that can be criticized as lacking a strong ethnographic basis (Watanabe 2004). Other scholars have used data on settlement clusters to make speculative arguments about topics such as domination and resistance (Pyburn et al. 1998).

Case Studies: Mesoamerica

15Two contexts within Mesoamerica provide illuminating comparisons for Classic Maya house clusters. The Aztec calpolli is a well-described pre-Spanish social unit whose spatial expressions are quite similar to Classic Maya clusters and minor ceremonial centers. The waterhole group among the contemporary Tzotzil Maya peoples of Highland Chiapas also shows spatial similarities with Classic Maya clusters, although today is it a rural feature, lacking urban expression. These two cases were chosen because they are the archaeological and ethnographic cases of Mesoamerican clustered settlement with the most extensive data.

The Aztec calpolli

16The Nahuatl term calpolli had a variety of meanings in 16th century central Mexico (Reyes García 1996), and it was the topic of a lengthy (and largely unproductive) scholarly debate on the nature of Aztec social organization (reviewed in Offner 1983, pp. 163-175). Here I focus on the dominant meaning of the term: a spatially localized group of households with a common economic basis. Two distinct levels of calpolli are documented in both the historical and archaeological records: a small unit of some ten to twenty households, and a larger unit of 100 to 200 households. The term tlaxilacalli is often used in Nahuatl-language documents for both types of unit, and in documents from Morelos the term chinamitl is often used for the smaller category. The most detailed information on Aztec calpolli is found in a series of early colonial Nahuatl-language census documents from Morelos; these are summarized in several secondary sources (Carrasco 1964; Smith 1993). Major works on the calpolli include Reyes García (1996), Hicks (1982), Offner (1983), and Lockhart (1992, pp. 16-20).

17Five features of the Aztec calpolli are particularly relevant for understanding neighborhoods in low-density cities. First, calpollis existed in both urban and rural contexts. The imperial capital Tenochtitlan was organized in calpolli (Reyes García 1996), which served as neighborhoods. The smaller but more numerous city-state capitals also contained calpolli (Smith 2008, passim). In rural areas, a calpolli was a village. The communities represented in the Morelos census documents include city-state capitals and smaller, rural villages. There is no indication that the organization or social dynamics of calpolli in these settings differed significantly in rural and urban contexts. Second, the members of a calpolli often engaged in common economic activities. In rural contexts (and probably in city-state capitals as well) calpolli members farmed a contiguous area of fields (typically owned by a noble), and in Tenochtitlan craft specialists were organized by calpolli.

18Third, the calpolli was a unit of taxation and administration. Officials known as calpixque resided in the calpolli and organized corvée labor and tax payments (Smith s. d.). Lockhart (1992) calls officials in the smaller calpollis « minor functionaries », whereas the large calpolli typically had a resident noble in charge. Fourth, membership and social organization in the calpolli were structured more by social class and by space or territory than by kinship. In the Morelos documents, the members of a calpolli were integrated through residence and through common relations of subjection or servitude to a noble who owned the land farmed by calpolli members. At the site of Cuexcomate, an Aztec town in western Morelos, a large residential compound on the central plaza was the residence of a noble household, probably the individual who owned the farmland worked by the commoner residents of the site (Smith 1993). According to some documents, calpolli councils oversaw local affairs, including the distribution of land. Fifth, the large calpolli typically contained a temple that provided a focus for common ceremony (Calnek 1976, p. 297).

19In an earlier paper (Smith 1993) I show that the small and large calpolli, as described in the Morelos census documents, correspond rather precisely in size and organization to two levels of settlement clustering at Cuexcomate (Figure 3). The four clusters (drawn informally by hand) correspond to the small calpolli or chinamitl, and the entire site corresponds to the large calpolli. Although the imperial capital Tenochtitlan is often taken as the archetypical Aztec urban settlement, in fact it was completely anomalous in size, density, layout, and organization (Smith 2008). The site of Cuexcomate is far closer to the structure and density of most Aztec cities (the city-state capitals), which can be considered medium-density urban centers. The median population density of Aztec city-state capitals is 50 persons per hectare (ibid., p. 152), which is closer to the density of Classic Maya cities (generally between 5 and 10 persons per hectare, see Culbert et Rice 1990) than it is to the density of Tenochtitlan (over 150 persons/ha).

20Like Classic Maya settlement patterning, Aztec settlement (outside of Tenochtitlan) was spatially clustered. As in the Maya Lowlands, clusters (calpolli) sometimes served as isolated hamlets or villages, and sometimes they served as the spatial building-blocks of dispersed urban centers. In both contexts, the members of clusters engaged in activities and shared the kinds of conditions typical of the residents of urban neighborhoods in areas with high-density cities: they interacted frequently, they worked together on common economic tasks, and they were part of a named social unit with an identity within a larger urban and political structure.

The Waterhole Groups of the Tzotzil Maya

22In the Tzotzil villages around Zinacantan, houses are clustered at three levels (Collier 1975, pp. 79-85; Vogt 1969, pp. 127-149). The smallest cluster is the sna, a small group of households living in close proximity. In most sna, the households are related through kinship ties and joint participation in ceremonies. The next larger level of clustering is the waterhole group, a collection of snas and households that live in proximity to one another. They are organized around naturally occurring waterholes that have both pragmatic and symbolic significance. Figure 4 shows the waterhole groups and snas of the hamlet of Paste. The population of these six waterhole groups varies between 10 and 523 persons, with a mean of 180; the standard deviation is 180 (Vogt 1969, p. 176). The third and largest level of clustering is the hamlet. The hamlet of Paste has 1,086 inhabitants. Hamlets vary from 121 to 1,227 persons, with a mean of 650 and standard deviation of 360 (ibid., p. 162).

23Of these three levels, the waterhole group seems closest to Classic Maya house clusters in its size and scale. Members of the Tzotzil waterhole groups are integrated through social interaction and through activities and institutions in the realms of economics, politics, and religion. Access to water is dependent upon membership in a waterhole group. Members participate in a number of collective labor projects, and an informal governing group of shamans have the authority to fine non-contributors. Waterhole groups maintain a number of cross shrines in and around the group’s land, and periodic ceremonies contribute to social integration.

Case Studies: Africa

24Most pre-European cities in sub-Saharan Africa had low population densities relative to European and Mediterranean urban centers (Coquery-Vidrovitch 1993; Fletcher 1993, 1998; Hull 1976). Houses were typically arranged into clusters that functioned as wards or neighborhoods (Hull ibid., pp. 74-76). The Mossi capital of Ouagadougou, mapped by Binger in the late nineteenth century, is a typical example (Figure 5). Skinner (1974, p. 19) notes that « early travelers reported that precolonial Ouagadougou was a whole complex of villages scattered over 12 kilometers ». This was a typical reaction of Europeans when they saw African low-density cities. When safari hunter Percy Powell-Cotton visited Addis Ababa in 1900, it was a city of 100,000 and capital of a powerful empire (see discussion below). Yet he was not impressed with the size or grandeur of the city: « Dotted across the plain were clusters of huts, many stockaded enclosures – large and small – and several camps, but all very much scattered and more resembling a collection of villages and farmsteads than the capital of a great empire » (Pankhurst 1985, p. 213). One might imagine a similar reaction if residents of Teotihuacan visited Maya cities like Tikal and Copan.

The Nupe efu

26Nadel’s (1942) ethnographic study of the Nupe kingdom in West Africa contains scattered descriptions of a unit – the efu or ward – that resembles the Aztec calpolli in structure and function. Like the Aztec calpolli, the Nupe efu was a basic level of settlement in both rural and urban settings. In the words of Nadel (ibid., p. 34),

[t]he local arrangement of the Nupe village, large or small, is nearly always the same: clusters of compounds scattered comfortably over an area, each « cluster » consisting of a number of walled compounds, « houses » in native terminology, and forming what the Nupe call an efu, or ward. These village wards are separated by stretches of open land, spacious cultivated fields, studded with clumps of trees.

27Figure 6 is Nadel’s sketch map of settlement clusters in the town of Doko (unfortunately this is the most detailed plan available). The 800 inhabitants of Doko were divided into five efu. One of the efu had three clusters, and the others a single cluster each. The capital city, Bida, contained between 25,000 and 30,000 people during Nadel’s fieldwork in the 1920s. Although the lack of scales on Nadel’s maps limits precise calculations, his estimate of the circumference of the city wall (12 miles) produces a density of approximately 10 persons per hectare within Bida. One reason for the low population density was the presence of infield gardens within the city. Nadel (ibid., p. 206) notes that, « Cultivation is also to some extent carried on inside the villages and even inside Bida town ».

28Nadel provides little information about the built environment or spatial organization of Bida. He does not give the number of efu within the city, and information about these wards is scattered throughout his book. Several features of the efu of Bida do stand out, however. First, most of the craft specialists (blacksmiths, other metalsmiths, weavers, bead makers, glass producers) were concentrated within specific efu. Second, foreigners were also concentrated within efu. Such occupational and ethnic clustering or concentration is common (but by no means universal) in urban neighborhoods throughout the world (York et al. 2011). Third, the efu had leaders that Nadel calls « ward-heads » (Nadel 1942, pp. 120,162), but we are told little about these officials. Bida was also divided into larger administrative districts called ekpa (ibid., passim).

Neighborhood clusters in Addis Ababa

29Although Addis Ababa had 100,000 inhabitants and was the capital of a powerful empire that defeated European armies, its built environment around 1900 did not much impress British visitors (see the quotation from Percy Powell-Cotton above). Indeed, it would be difficult to identify Addis Ababa as a city rather than a village from early photographs. Figure 7, for example shows an area of the city close to the royal palace (visible at the top of the hill) in 1897. Although there are large crowds of people, housing is widely scattered and there are significant empty areas.

30Unlike Bida and most other low-density African cities, Addis Ababa was well mapped at its height in 1897 (Gleichen 1971, endpiece). Figure 8 (redrawn from the original published plan) shows the clustering of residences in a pattern that resembles many Classic Maya settlement maps. Commoners were clients of nobles, and their houses were clustered around the compounds of their patrons (Figure 8). Gleichen (ibid., p. 157) described this clustering, and noted that « the importance of the individual [was] measured by the size of the enclosure and the number of smaller huts grouped around it ». Sefer is the Amharic term for these clusters, which are referred to by urban historians as neighborhood in English and quartier in French (Giorghis et Gérard 2007, pp. 42-43). Individual sefer were separated by open space and by roads and streams (ibid., pp. 122-123). Figure 9 shows a road that separated two neighborhoods in early Addis Ababa. The clustering of commoner houses around noble compounds recalls Maya settlement at several levels, including Ashmore’s group-focused patio clusters (Figure 2) and Bullard’s minor ceremonial center.

31Like other tropical low-density cities (Fletcher 2009; 2011), including Bida (see above) and ancient Maya cities (Isendahl 2002), there was significant intensive infield cultivation in Addis Ababa (Gascon 1995, p. 15). This was one of the reasons for its appearance to European visitors as a « collection of villages ». Indeed, Giorghis and Gérard (2007, p. 122) include a subheading in their book on Addis Ababa called « The Rural Character of the City ». There is frustratingly little information about the settlement clusters or sefer at Addis Ababa, although scattered data suggest the presence of the kinds of ethnic and occupational clustering characteristic of urban neighborhoods in other areas (Garretson 2000, pp. 67, 87; Pankhurst 1985, p. 212).

Conclusions

32The argument of this paper can be summarized in three points. First, spatial clusters of houses in three historically-documented traditions of low- and medium-density urbanism (Aztec, Addis Ababa, and Bida) exhibited activities and conditions typical of urban neighborhoods in both premodern and modern cities. These include frequent and regular social interaction among residents; shared social characteristics such as ethnic origin or occupation; an administrative role within the city; a ceremonial focus for social integration; and an indigenous term that European visitors translated as neighborhood or a synonym (e.g., ward, quartier, barrio). In two of the three cases (Aztec and Nupe), the neighborhood clusters were features of settlement in both rural and urban settings. My second point is that modern Tzotzil Maya rural settlement clusters – waterhole groups – exhibit these same spatial and social characteristics. Third, because Classic Maya settlement clusters resemble the historical and ethnographic cases in their spatial form and scale, these comparative data strengthen the interpretations of some Mayanists that they served as urban neighborhoods.

33The recognition that settlement clusters at Classic Maya cities served as urban neighborhoods has several implications. First, this helps archaeologists in their analyses of social patterns within Maya cities and improves our understanding of Maya urban life. The French research outlined above can serve as a model for neighborhood analysis at other sites. Second, the existence of neighborhoods adds strength to the interpretation of Maya settlements as cities, and this aids the comparative analysis of cities around the world. The sample of premodern cities that can be compared is greatly enlarged by the inclusion of the Classic Maya as a fully urban social tradition. Third, the study of Maya neighborhoods contributes to a growing understanding of life in low-density tropical cities (Connah 1987; Fletcher 2009, 2011; Miksic 2000). Maya cities provide examples of a distinctive and successful form of urbanism that flourished for many centuries in a tropical lowland setting. Why were Maya cities successful for such a long period of time? Perhaps a better understanding of neighborhoods and social life will permit this question to be answered in the future.

34Acknowledgments: This paper was written as part of the project, « Urban organization through the ages: neighborhoods, open spaces, and urban life » (http://cities.asu.edu/​), and I thank my colleagues for their support. Several colleagues provided helpful comments on an earlier draft; for this I thank M.-Charlotte Arnauld, Juliana Novic, Cynthia Robin, and Barbara Stark. Maëlle Sergheraert kindly translated the abstract into French. The comments of two anonymous reviewers were very helpful in improving the paper.