HomeBiographyWritingsPhotosContactAncient Maya Settlement Patterns and Environment at Tikal, Guatemala: Implications for Subsistence Models Dennis Edward Puleston A Dissertation in Anthropology, University of Pennsylvania, 1973

Chapter One: The Problem of Ancient Maya Demography


Intersite Settlement Density
Site Orientation of Mapping
The Site Orientation of Excavation
Difficulty of Survey in the Maya Lowland
Quantifying Settlement Data
Trail Surveys
Area Surveys
Site Survey
The Inconsistencies of Subsistence Data
The Sustaining Area Hypothesis


The problems which will receive primary attention here, as already indicated, concern demography, environment, and subsistence. I will not be directly concerned with the definition of "urbanism" or "city," nor will I be concerned with the degree to which Tikal meets these definitions. Though much of the data to be presented here is pertinent to these questions, they also involve consideration of a whole series of additional topics which must await the reports of other workers at Tikal. Of the three items mentioned above, demography, as determined by mapped settlement remains and ceramic data, will be considered first.

A good deal of work has been accomplished with respect to residential settlement in the Maya Lowlands. The most recent summaries have been presented by Willey and Bullard (1965) and Haviland (1966). Both present a review of past work in this field up to the mid-1960's and each refers to sources not covered by the other although Haviland's coverage is much more complete. For the time period they cover, the only important sources I would wish to add are Morley's (1920, Pl. 3) map of the environs of Copan which locates "satellites" up to 12 km. from the center of that site, and Merwin's (1913) survey of the northeast Peten, northwest British Honduras, and adjacent portions of Mexico. Willey and Bullard review the results of Bullard's (1960) survey of the northeast Peten, which will be discussed in more detail below. On the basis of what has been accomplished, they make a number of "inferences" with respect to land use, population density, sociopolitical organization, and the question of Lowland Maya urbanism. With respect to land use, they state that "there is no good evidence that the Maya never placed primary reliance on anything but the milpa or 'slash-and-burn' system of farming." They present as evidence against Ricketson's (Ricketson and Ricketson 1937:10-13) argument that the ancient Maya may have practiced some form of intensive cultivation, the observation that along the Belize River alluvial soils could not have been used for a major crop because of the extent to which they were covered with settlement. In a footnote, they do leave open the possibility that the spaces between households could have been used for "garden crops" and cacao.

Denny on Temple V in 1961

Regarding population density, Willey and Bullard (1965) and Haviland (1966) review some of the problems involved in making population estimate, including: 1) the small size of settlement samples; 2) the difficulty in deciding which structures were dwellings and which served other purposed; 3) the lack of ethnographic data on the number of families occupying modern households for use in estimating the number of occupants per household; and 4) the difficulty in determining the number of households occupied at any one time.

In their discussion of socio-political organization Willey and Bullard return to Bullard's hierarchy of households, clusters, zones, and districts to suggest a hierarchy of power and authority. Willey (1965b) has argued, on the basis of Belize Valley settlement patterns and the discovery of "luxury" items in "house-mound" burials,  that the division between the "urban" upper class and "rural" peasants has been greatly overstated. Bullard's (1960:368-369) data relating to the great number of minor ceremonial centers, the association of household "clusters" with these centers, and the absence of such a direct association between major ceremonial centers and house ruin settlement--Yaxha is given as an example--leads them to support Vogt's (1961) concept of rotating ceremonial offices. The evidence for a higher degree of specialization at major centers, however, forces them to accept a more permanent and probably hereditary system as well.

Regarding the final matter of urbanism, they discuss the current question of whether the major centers were "cities" or not. According to their assessment, Classic Maya centers, on the scale of Tikal, fulfill all the criteria established by Childe (1950) for urbanism except "what is probably the most diagnostic one, sheer size of dwelling aggregates." In contrast, Sanders (1956) has put emphasis on functional, economic criteria by stating that in order to qualify as urban, such communities must have at least 75% non-farmers and show signs of heterogeneous functions and interests in various elements of the population. On the basis of these considerations, they conclude that while the Classic Maya may have been moving towards urbanism and heterogeneity, they did not attain it and can be characterized as a "civilization without cities."

Haviland (1966), in his review of residential settlement studies, covers many of the same points., He places greatest emphasis on the question of the identification of most "small mounds" as houses. He also reviews the relevance of evidence from these structures to questions of social structure, agricultural potential, and the relationship of ceremonial to residential structures. He reviews the Belize River Valley Survey and notes Brainerd's (1956:162) comment that this area with its flood plain and rich alluvial soil may not be very helpful for interpretation of the Petunia and Yucatan. He argues strongly for complex social stratification on the basis of the great variety in the quantity and quality of artifacts found in his excavations of house ruins at Tikal. He further suggests a total Late Classic population of 10,000-11,000 for the central 16 km2 of Tikal which would appear to contradict the findings and conclusions of Willey and Bullard. He argues persuasively that Tikal was surrounded by a large and dense aggregate of dwellings rather than the rural hamlet-like clusters Sanders (1962-63:207) seemed to see on the Tikal site maps. The discussion of population density gets him into the question of the support potential of slash-and-burn cultivation. On the basis of the optimistic arguments of Conklin (1961), Leach (1975), Dumond (1961) and Cowgill (1960, 1962), he concludes rather cautiously that "agriculture in the Peten may be more productive than previously judged" (Haviland 1966:38).

With respect to the implications of residential settlement for social structure, Haviland feels that the lack of obvious dwelling clusters at Tikal argues against the importance of lineages at Tikal. He rejects (Haviland 1967) Vogt's idea of rotating ceremonial offices or cargos primarily on the basis of his analysis of Tikal skeletal material, which reveals a significant difference in stature between tomb and house ruin populations. He does not comment on Willey and Bullard's suggestion that cargos may have functioned in the context of "minor ceremonial centers" or "satellites." Haviland, echoing the proposals of Satterthwaite and Willey, concludes with a call for the kind of field work that would make the comparison of site patterns possible.

Now rather than go on with a summary of all settlement pattern work which has been carried out in the Maya area since 1966, I would like to limit my discussion to studies which relate to what I have defined as intersite areas. As Haviland (1963)  recognized, it is here that the data most significant to overall questions of subsistence and demography will be found.

Intersite Settlement Density

As indicated above, a principal stumbling block in considering ancient Maya demography has been the lack of quantifiable details of settlement density beyond the immediate vicinity of known sites. Apart from the rather unspecific generalizations produced by Bullard's (1960) wide-ranging survey of the northeast Peten, and Sander's (1962-3:211) rough coverage of a 300 km2 area in the Chantalpa for which maps have not yet been published, good comparative data on "intersite" settlement is sadly lacking. The Uaxactun house mound survey (Ricketson and Ricketson 1937), the focal point of discussions of settlement density in the southern Maya Lowlands as a whole until recently, also is of limited value. Though there was no way the fact could have been recognized at the time, it now appears that the area surveyed at Uaxactun lies within the nucleus of settlement that surrounds most major sites and, therefore, is of restricted usefulness for the examination of regional densities. The same limitation applies to more comprehensive surveys at Tikal (Carr and Hazard 1961), Mayapan (Jones 1952, Pollack et al., 1962), Dzibilchaltun (Andrews 1965), the Belize River Valley (Willey et al. 1954), and Seibal (Tourtellot, unpublished site map) have produced much new data.

At Dos Aguadas Bullard (1960:366) sampled an area of 0.5 km2 and obtained a very rough figure for settlement density per unit area. Site maps such as Bullard's maps of Topoxte (Bullard 1970), Altar de Sacrificios (Willey and Smith 1969), and Hellmuth's (1971) preliminary map of Yaxha are specifically excluded here because they make no real attempt to go beyond the limits of major architectural remains. Even in the more comprehensive surveys, however, the lack of intersite settlement data has placed serious limitations on their usefulness for discussions of the problems indicated by Willey (1956) and Haviland (1963:538).

A number of factors have been responsible for this lack, among which are the following.

Site Orientation of Mapping

First and foremost of these is the strength of traditional archaeology's bias towards the "site approach." Archaeology, since it deals with the physical remains of cultural activity, tends to orient investigations to the epicenters of visible or known remains. This tendency is exemplified by the Willey and Phillips (1958:48) use of the "site" as the smallest and most basic unit in a culture area approach to spatial designation. They define "site" as an area "fairly continuously covered by remains of former occupation . . ."

The implication is that a "site" as a significant data-bearing or artifact-producing unit can be given limits that exclude less heavily settled areas. Thus in practice, despite conceptualization of "zonal Settlement patterns" (Sanders 1956:116), archaeological endeavors have rarely extended beyond "site" limits in the Maya Lowlands. When they have, it is often to locate, describe, and test smaller "satellite communities" or "tributary settlements" (Rands `1969:1) without considering elements of settlement that occur between these points. Other surveys of this kind which cover large areas consider only the most obvious remains. These include Merwin's (1913) study of the northeast Peten, the reconnaissance carried out by Ruppert and Denison (1943), the work of Adams and Gatling (1964) in the northeast Peten, the Belize River  alley Survey (Willey, Bullard, Glass, and Gifford 1965), and Sisson's (1970) work in the Chontalpa. Important exceptions to the "site bias" include the trail surveys of Tozzer (1913) and Bullard (1960),  and Sander's (1962-3:211) survey of 300 km2 in the Chontalpa, for which maps are not published. Though Sanders (1962-3:212) reports the existence of "small clusters of 3 to 12 house platforms with no associated ceremonial architecture," I suspect his coverage is spotty and that the smaller groups will be underrepresented in the overall survey results when they become available.

I say this principally on the basis of my own experience at mapping residential settlement under heavy vegetation. From this standpoint 300km2 seems like a very large area to cover completely. Unfortunately the available information gives us no idea of survey techniques or the percentage of this area that could have been covered fairly quickly because it was under savanna.

Thus, in spite of the fact that it is now fashionable to include  smaller residential structures and outbuildings as being part of an ancient Maya "site," there has been little change in the general approach. The vast intersite areas remain underscribed, unexcavated, and essentially unevaluated. The present study does not really represent an exception to this state of affairs, though perhaps I will be able to show that some of the limitations of the site approach have been overcome for the first time. It should be admitted, however, that to overcome these limitations required much greater effort than would have been necessary if we had had full cognizance of the problem when we began.

In the Maya area where "sites" that are studied are often large and jungle-covered, the site approach has posed certain logistical problems that have made it extremely difficult to break out of the old pattern. With the rectangular shape of the traditional site map, greater and greater efforts must be expended to increase the radius coverage for a particular site. At some point, the increment in the effort expended to obtain a slight increase in the "radius" of the map makes continuation financially unrealistic. It is now apparent that at Tikal, four seasons of mapping by several teams was not enough to cover more than a small percentage of what may legitimately be called the site. In the equivalent of about eight man-seasons (1957-Hazard, 1958-Wuram, Levine, and Hazard, 1959-Carr, 1960 Maretinex, Gregerson, and Carr) of work, only nine km2. Hans Gregerson (Carr and Hazard 1961), in 1960, almost doubled the total mapped area of Tikal with a hasty reconnaissance survey that covered a total of seven km2. In spite of this effort, he was able to add a bare 500 meters to the site map "radius." Another 500 meters outwards would have required even greater efforts, with greater inaccuracies since spatial control was based entirely on: 1) aerial photographs without concise landmarks; and 2) paced measurements from the edge of the central nine km2.

The Site Orientation of Excavation

A second reason for the lack of intersite data lies in the questions archaeologists have been asking. The importance of establishing an occupational sequence, requires excavation in localities where activity has been: a) the most continuous; b) the most extensive; and c) the most enduring. The locus of the greatest architectural activity is generally the best place to start looking for such deposits. If the establishment of ceramic sequences can get the archaeologist started in a site area, it also contributes to keeping him there. The very density of information revealed by even small excavations in such an area provides a flood of new and compelling questions that serve to strengthen rather than diminish the already existent bias. In the Maya area, the  availability and challenging complexity of hieroglyphic writing, stone architecture, elaborately furnished tombs, and caches have all posed questions that have tended to focus attention inward on site centers, leaving little in the way of time, energy,  and funds to be directed outward on the environmental and rural human matrix from which the culture derived its energy. This is not to say that the intellectual and physical goals established by the elite at the cultural epicenter of a site such as Tikal were not of great importance. They were of paramount importance for the ways in which they influenced how this energy was obtained and used. I am saying here, however, that a grave imbalance afflicts our knowledge of how and in what ways the influence of the elite was brought to bear on rural populations and the utilization of the available resource base. In other words, the bias I have referred to above has demonstrated what  the ancient Maya accomplished, and we even might roughly quantify this accomplishment in terms of earth and stone moved, the energy required to import obsidian, make plaster [as Erasmus (1965) has attempted to do], carve jade, and son on; but it does not tell us how this was done: how the calories represented by the accomplishments that characterize ancient Maya civilization were extracted from the environment and organized to sustain the magnificent culture, the remains of which have so captured the collective imagination of so many students.

Difficulty of Survey in the Maya Lowland

Another factor which has vitiated the collection of settlement data in intersite areas in the Maya Lowlands is the frustrating density of the vegetation. This has had a significant influence on Lowlands settlement pattern studies. A major determinant in the location of the Belize River Valley Survey at Barton Ramie was the fact that this area had recently been cleared (Willey, et al. 1965:15). Not only has vegetation hampered ground surveys, it has effectively precluded the use of aerial photographs for detailed survey. This problem will be discussed more fully in the chapter on field procedures.

The combination of these factors, including the site orientation of mapping and excavation and the unusual density of the vegetation, has contributed to the slow development of data on residential and intersite settlement data and the consequent inability of scholars to realistically discuss ancient Maya demographic patterns and subsistence, as indicated by Willey (1956:114). The first step has now been taken as a result of the extensive excavations in residential areas undertaken by students working at Tikal (Becker 1971; Haviland 1963). The next logical step into the vast intersite areas, is to be taken here.

Quantifying Settlement Data

Without dealing directly with population estimates for the Maya Lowlands as a whole or in part, general demographic patterns have been the subject of considerable discussion. For the most part this discussion has focused on the question of whether the great centers were cities or not. With their monumental architecture, writing, evidence of social stratification, and so on, the Maya have generally fulfilled Childe's list of characteristics by which cities and civilization (Childe made little distinction between the two) can be recognized. The one major unknown resulting from our lack of knowledge of the totality of residential and intersite areas has been the size of residential population associated with these sites.

As Haviland (1966a:37) has indicated in his review of this entire problem, the size and density of residential populations is intimately linked to the question of subsistence. In the case of the vacant  center models to be found in the Guatemalan highlands, it has been this variable which has received the greatest amount of attention.

Willey (1956:110) has proposed three alternative models for patterns in the distribution of residential settlement:

First there is the one in which (Type A) the ceremonial center is surrounded by dwellings which are so closely spaced that their inhabitants could not have farmed immediately adjacent to them. . . . In the second type (Type B) the ceremonial center is without dwellings, and houses of the sustaining population are scattered singly over a wide surrounding area. The extent of such a sustaining area is unknown . . . . The third idealized type (type C) is similar to B in that there is no appreciable population concentrated in the ceremonial center, but the sustaining populations are spotted through the surrounding country in hamlets or small villages rather than in individual houses.

As already indicated, the lack of quantifiable settlement data from beyond the confines of the major centers has made it impossible to do much more than speculate about the spatial reality of these alternatives.

Trail Surveys

Apart from site surveys such as Ricketson's, which was thought to extend into intersite areas, the only quantifiable data on intersite situations in the Maya Lowlands has come from what can best be called trail surveys. During long treks across the Peten, investigators occasionally took notes on mound and chultuns they passed along the trail. Such observations, when they are reported at all, occasionally appear in the introductory sections of site reports (Maler 1911:5) or within the texts of reconnaissance surveys (Merwin 1913; Andrews 1943; Ruppert and Denison 1943). Useful as these data are for obtaining an impression of settlement patterns and density, they are rarely based on a systematic approach to technique.

In only a few instance has systematically collected trail survey data been presented on maps. The best examples are Tozzer's survey of 1909-10 (1913 p1. 31), Bullard's Northeast Peten Survey (Bullard 1960), and graham's map of North Central Peten (1967:40).

Bullard (1960) observed that : 1) small platforms and chultuns could be found in intersite areas; 2) they could be found in fairly high densities many kilometers from known major ruin areas; 3) well-drained medium-high ridges were preferred for house locations; while 4) steep slopes were generally avoided; 5) they did not occur in major areas; and yet 6) they seemed to occur with greater frequency on high ground near bajos; 7) ribbon-like settlement seemed to occur along the Belize River in British Honduras [though this was never actually demonstrated by survey (Haviland 1966b:592)]; 8) the basic unit was the plaza group with a number of low platforms arranged around a rectangular raised surface; 9) "minor ceremonial centers" were widely scattered; and 10) they often served as focal points for settlement.

There are serious limitations upon these data. The density of the vegetation to left and right of a particular trail, especially in the Peten, makes it virtually impossible to see enough of the country through which one is passing to obtain an adequate settlement sample, and thus characterize settlement for any particular area. On one portion of the South Survey Strip at Tikal, between 3.5 and 4.6 km., it is possible to walk down the trail through one of the most heavily settled areas on the strip without seeing a single platform, simply because the dense jungle undergrowth prevents one from seeing more than a dozen meters to left or right of the trail.

Because of this problem it has been difficult for investigators to offer more than speculation about overall settlement density and distribution, even though they may have traveled back and forth through a particular region for years.

Area Surveys

Observations on intersite settlement patterns and density over an area, as distinct from observations taken along trails, have been very infrequent in the Maya Lowlands. In certain locations, which are still rare in the Peten, sufficient land is cleared of trees for agricultural purposes to make possible a general picture of settlement in a small area. Such clearings may quickly revert to concealing bush or dense growth of ferns, as in the vicinity of Uaxactun, without ever being observed,  much less recorded or even commented upon. There are significant exceptions, however. P. W. Schufeldt (1950:226), a chicle contractor, was much impressed by the number and density of small mounds and chultuns revealed in milpa clearings near Laguna Perdida, northwest of Lake Peten. A map put together by John L. Gatling of the Sun Oil Co. (R. E. W. Adams and Gatling 1964:101) is of interest in that over an area of approximately 1,800 km2 an estimated 60% of major and minor mound group areas have been located. Sander's (1962-3:211)  Chontalpa survey may represent an important contribution, but until his maps are published we will have no more than an impression of what settlement densities and distribution patterns are.

One of the earliest attempts to quantify settlement at a Maya site on the basis of settlement pattern studies was Ricketson's house mound survey at Uaxactun. Portions of a cruciform sample area amounting to 1.94 km2 were surveyed for housemounds by local Guatemalans under the direction of Enrique Schufeldt and F. C. Perkins. Only 953,040 km2 or 49.2% of the area was deemed habitable, and on it a total of 78 structures were located. Allowing an average of five persons per structure with simultaneous occupation of all structures, a density of approximately 430 persons/km2 of "habitable land" is obtained. Ricketson, apparently so that he could be on the "safe" side, reduced his estimates to 25% of this figure. He did so not because of the land requirements of slash-and-burn since he believed agriculture was "intensive," which to him meant maize cultivation with following, but rather because of the difficulties that would have been involved in continual clearing with stone tools (Ricketson and Ricketson 1937:12). Whatever his reasons were, it now seems likely that this reduction by 75% was not warranted, at least on the basis of evidence from Tikal. Haviland (1970:191) has presented a good case for contemporaneous occupation of essentially all residential structures, which appear on the Tikal map, in Late Classic times. It should perhaps also be pointed out at this time that many structures were missed in Ricketson's survey.

It may have been this opinion that gave Ricketson the courage to go on to calculate, on the basis of his survey, a population of 13.3-53.3 million people for the peninsula of Yucatan (Ricketson and Ricketson (1960:367).

Site Survey

This problem of site definition does not seem to exist at Postclassic sites. At Mayapan, with a wall running entirely around the comparatively closely packed residential structures (Willey's Type A), there was little question as to what minimally was to be considered the "site." Jones (1952) mapped approximately 2,100 dwellings, to which A. L. Smith (1962:211) has applied Redfield's average figure of 5.6 persons per house (Redfield and Villa Rojas 1934:91) to obtain an estimated population for the site of 11-12,000 persons. Contemporaneity of all dwellings was assumed, presumably for the upper range of Mayapan's period of occupation which stretches from Katun 13 Ahau (A.D.. 1263-1283) to Katun 8 Ahau (A.D. 1441-1461), when it fell (Pollock 1962:8).

At Dzibilchaltun, without a wall to define site limits, Andrews (1960), with a situation very similar to Ricketson's, went in the opposite direction, claiming the site to be "one of the largest, if not the largest Pre-Columbian city in the New World." The "site," is defined by the limits of densely distributed archaeological remains, covers an area of approximately  50 km2. Andrews' estimates of population on the basis of structure counts range into the hundreds of thousands. Andrews' figures seem inflated to me. On the sections of his map I have worked with (Square Jq), settlement density is lower than we find in the more heavily settled portions of Tikal. It might be pointed out also that even if we accept his figures Dzibilchaltun does not cover even half of what demonstrably falls within the circumference of the North and South Earthworks at Tikal.

At Barton Ramie, as Haviland (1966:33) has suggested, Willey et al. seem to be constructing a model much like Willey's (1956) Type C. In this model, Benque Viejo, a major site on the Belize River with large "place: complexes and "temples," served as a regional center hierarchically superior to smaller "temple" and "palace" mound units within dwelling communities such as Barton Ramie. Benque Viejo would have been dependent upon these communities for construction and maintenance.

Population estimates have been made for Barton Ramie. On the basis of quite limited excavations (see Haviland 1966b:592), it is assumed that the mounds represent the remains of elevated plazuela unites with structures on them. A figure of 7.5 persons per mound is assumed, yielding an approximate total of 2,000 persons for the mapped portion of Barton Ramie during the Spanish Lookout Phase (Willey, Bullard, Glass, and Gifford 1965:576). The 7.5 figure is apparently based on two assumptions: 1) a "biological  family" consisting of five persons was a basic social unit at the site; and 2) there were an average of one and one-half such families on the plazuela mounds. The justification for these assumptions is not presented. Bullard (1964:282) provides a precedent for a figure in this range when he assumes an average of seven individuals per household but here we are left without any explanation whatsoever.

In a survey of eastern Quintana Roo, Sanders (1960:216), on the basis of the assumption that the majority of the structures were residential, claculated a population density for Tulum on the order of 7,000-8,800 persons per square kilometer. Though the area actually involved was quite small and peripheral areas were not examined, density was evidently quite high, suggesting a nucleated site not unlike Mayapan.

Peten forest from the air with the line of a Sacbe or a path through the jungle

Tightly clustered settlement is also reported for the Postclassic site of Topoxte. Between 80 and 100 house platforms were found on the island, which is about 400 by 200 meters (Bullard 1970). Sanders (1962-3:225) estimates a population density on the order of 5,000 persons/km2 but, as at  Tulum, the total population was probably less than a thousand inhabitants. The density of settlement on the island of Noh Peten (the modern Flores) must have been high. Thompson (1951:390) reports, on the basis of a reference I cannot find or  perhaps, a calculation I cannot reconstruct, that Avendaño estimated its population to be 2,000 people. Since the four other "Petens" (islands) were "equal in the number of people, with but little difference," (Avendañ ca. 1910:91,93) and Avendaño's total was 24,000-25,000 for the region as a whole, the mainland population may have been on the order of 14,000-15,000. Avendaño's estimates approximate recent populations. Thje 1964 census yielded a total population of 1,503 people for Flores (Sozo 1970:150) and 26,720 for the entire Department of the Peten (Soza 1970:120).

Considering that the island today is 15 manzanas (Soza 1970:148) or about 0.10 km2 [1 manaza = 1.73 acres (Stadelman 1940:94)] in area, a population of 2,000 yields a density of 20,000 people /km2. This is quite high in caomparison to estimates for other Postclassic sites, including Sander's (1960) estimate of 7,000-8,800/km2 for Tulum and 5,000/km2 for Topoxte (Sanders 1962-3:225).

Crowding is suggested by Avendano's observation that houses were built right down to the shoreline:

I asked them why they had their houses so close to the shore of the lake, and if they had any trouble, with its rising and fallings. They told me that they had no trouble since the lake never rose or fell (Avendano ca. 1910:84).

Haviland (1963:524) has estimated a population of approximately 10,000 persons for the central nine km2 of Tikal in Late Classic times. This estimate is calculated on the basis of sturcture counts taken from the Tikal site map (Carr and Hazard 1961) and his and Becker's excavation of 117 small sturctures at the site.

The Redfield-Villa Rojas figure of 5.6 persons per house was used in the calculations.

Haviland (1970), on the basis of a preliminary assessment of the data to be presented here and the discovery of the earthworks (Puleston and Callender 1967), has redefined "Tikal" and revised his estimate of Late Classic Tikal's population upward to as high as 49,000 people.

The Inconsistencies of Subsistence Data

Having examined some of the problems surrounding the direct calculation of ancient populations on the basis of settlement patterns and density, we will turn now to the subject of subsistence.

The strong link between settlement patterns and subsistence has been pointed out by Sanders (1967:53) (among others) who wrote: "The primary determinant of rural settlement patterns in a peasant society is the agricultural system practiced." We have only to look at the evidence regarding chinampa cultivation in the Valley of Mexico, the hugh mechanized farms in the Midwest of the United States today, the Kraals of East African cattle herders, or the scattered villages of rice cultivators in Southeast Asia to realize the powerful relationship that exists between subsistence and the configurations, size, and density of population in the settlements of food producers. Much evidence can be presented to show that changes in settlement patterns go hand-in-hand with changes in subsistence patterns.

The same influences and relationships certainly existed and still do exist among the Maya. Redfield and Villa Rojas (1934) have presented an interesting model in which they show how the size of the villages of slash-and-burn cultivators in Yucatan is related to the availability of cultivable land within a reasonable walking distance of the town. When the distance becomes too great, or when conflicts over land begin to get too troublesome, (for a time) dual residence may be maintained. This is the centrifugal effect of slash-and-burn agriculture cited by Wolf (1959:60) and others. G. C. Vaillant (1940) held that the ceremonial center and the scattered housemound settlement pattern  of the ancient Maya were a function of the requirements of milpa farming. The actual calculation and comparison of possible yields per unit area and the space available around mapped residential unites presented here reveal that such a conclusion is not possible at Tikal. This is not the only Maya site at which such an inconsistency has been realized. At Barton Ramie, Willey et al. (1965) note that " . . . the one type of soil which might seem to have been suitable for intensive maize cultivation, the alluvial bottomlands of the Belize River, were apparently not used for major crops. Houses were so thickly settled in the alluvial strip that there would have been no room to plant corn" (Emphasis mine).

Inconsistencies of this nature have long been a source of puzzlement and dissatisfaction among Maya archaeologists with the assumption that ancient Maya subsistence was based on slash-and-burn cultivation of maize. In the case of Barton Rami it was concluded that unmapped upland areas further from the river must have been where cultivation was carried out. The incompatibility of space requirements and settlement density has not been the only inconsistency.

Linton (1940:38) was one of the first to recognize soil exhaustion and the need for fallowing as relevant to population estimates for a given area and thus a major problem for the Classic Maya model. The rapid decline in the productivity of soils would have required the people of maximally occupied areas such as Uaxactun, as indicated by Ricketson's data, to move to completely new areas every few years.

Sanders, with full realization of the reality of this dilemma, found himself forced to go along with the unlikely possibility suggested by Linton. "I would assume that the hamlets (around the major centers) were very unstable communities and repeatedly abandoned and reoccupied" (Sanders 1962-3:206).

The improbability of such a model has been proclaimed most strongly by Palerm and Wolf (1957) who feel that "slash-and-burn agriculture . . . could not provide a stable economic basis for the growth and existence of Maya civilization." Wolf (1959:60) later expressed his belief that "the inherent mobility of [a slash-and-burn population's] component units will not lead to a strong state based on tribute and labor drafts." While it can be argued that such a system was apparently functioning at the time of the conquest in Yucatan, the difference in architectural achievements and site longevity at the contrasted sites of Mayapan and Tikal suggest that some more stable arrangement may have been at work in Classic times.

In northern Yucatan the Maya even had access to a significant seafood supplement to their food resources, which Landa (Tozzer 1941:190) tells us they salted and traded inland as far as 20 or 30 leagues.

In spite of his belief in the primary influence of societal development on cultural evolution, Adams (1966:38) states: "It is a truism that complex civilized societies depend upon a subsistence base that is sufficiently intensive and reliable to permit sedentary, nucleated settlements, . . . a circumstance that has implied agriculture."

Slash-and-burn cultivation of maize in the southern Maya Lowlands appears not to meet these qualifications, as will be indicated below.  Wolf (1959:15), following this realization to its logical conclusion, points out that "we have not yet discovered how the Maya could reconcile such quick exhaustion of arable soils with long term, stable city life."

In view of the complexity of arguments for and against the slash-and-burn subsistence base for ancient Maya civilization, a diagram presenting the various alternatives has been appended (Figure 5).

On the basis of this chart it can be seen that many arguments which both sustain and deny the primary role of slash-and-burn cultivation of maize have been advanced over the years. Very few have argued that swidden farming is entirely acceptable without qualifications as a subsistence base for the ancient Maya. Many have preferred to view slash-and -burn agriculture as being ill-suited for the role into which it has been cast and thus ultimately responsible for the collapse of Classic Maya society. Beginning with Orator F. Cook (1909), the limitations or failure of agriculture have been implicated in more explanations of the collapse than any other factor (O. F. Cook 1909, 1921; W. C. Cooke 1931; Morley 1937-8, Vol. 4:334, 1946:72; Ricketson and Ricketson 1937; Lundell 1937:14; Meggers 1954; Sanders 1962-3:114; Sabloff and Willey 1967:328; Willey and Shimkin 1971:13).

Figure 5. A flow chart presentation of the alternative roles slash and burn agriculture may have had in the ancient Maya Lowlands, as they have been proposed in the literature.

Those who have argued that slash-and-burn agriculture could have sustained the rise of this civilization have had to juggle the inconsistencies of indicated population and subsistence potentials. The population variable has been manipulated by suggesting, as Sanders (1962-3) has done, that as few as 25% of the structures at a site were contemporaneously occupied even at the height of Late Classic development. Sanders seems to have obtained this figure from Ricketson's calculations, but the former presents no more justification for it than does the latter. Haviland's (1963, 1970) arguments, which will be discussed in detail in the chapter on the calculation of prehistoric demographic patterns, have already cast doubt on the likelihood of this supposition.

Another example is provided by Brainerd (1956:162) who, apparently by counting groups instead of structures, is able to wrench from the data a picture of low settlement density with the statement:

The survey data showed only 69 house ruins per square mile in an area known to have been inhabited for over 1,000 years.

This represents quite a contrast to Morley's (1923:272) image of the area as being on "of the most densely populated areas . . . in the whole world during the first six centuries of the Christian era." The actual density on the Ricketson survey, in terms of structures, is over 108 structures per square mile; and, as we have already seen, mapping has been woefully inadequate, with perhaps 50% of the visible remains unrecorded or overlooked.

Needless to say, Brainerd, along with Termer (1951:105) has been a strong adherent of the "ceremonial center" model for ancient Maya sites, as his editorial comments in Morley's book indicate: "There is no evidence thus far that the archaeological sites of the classic stage are the ruins of cities and towns; there is on the other hand, much evidence that they were religious centers to which the Maya resorted only for ceremonies" (Morley 1956:261). Thompson (1954) has been the principal proponent and popularizer of this conception of major Maya sites in recent years. It is important to realize that this model of a "ceremonial center" surrounded by evenly dispersed settlement is central to the conceptions and arguments of the "Harvard School" as evidenced in the writings of Willey, Bullard, and others (see Becker  1971). Brainerd felt that this pattern prevailed until Toltec invasions of northern Yucatan. His opinion was based largely on the conviction that 30 people/sq. mi. was the carrying capacity of the land under slash-and-burn agriculture.

Apart from the evidence provided by settlement density, the sheer mass of centers such as Tikal, Uaxactun, Nakum, and others has always suggested to me that substantial populations drawn from the immediate vicinity of such centers must have been involved in their construction.

Again however, without considering the evidence for settlement density, Erasmus (1965) has contested this assumption by demonstrating through experimental studies that mass equivalents of major sites such as Uxmal could have been brought together by relatively few slash-and-burn agriculturalists over a longer period of time. Some of the weaknesses in his argument include assumptions that: 1) the mass of Classic Maya constructions consisted of hastily piled dirt and rock; 2) all the dirt required could have been brought in from an average distance of less than 400 meters from the construction; 3) all the rock could have been brought in from within a radius of 950 meters of major construction by tearing up all the surface rock that was not actually covered by 100,000m2 of inhabited buildings and terraces; 4) this "fill" that made up the bulk of construction did not require the manufacture of lime plaster as an aid to stabilization; 5) the green firewood necessary to burn the lime could have all been obtained within 500 meters of the major constructions; 6) the labor required to construct the numerous (and largely unmapped) "palaces, " "temples," and terraces that surround the main center was insignificant; and 7) the populace did not perform many other massive public works tasks that would have been necessary to maintain the "system" and would have diverted energy from his "pool;" of 48,000 man-days per years, including the construction and maintenance of reservoirs, service as carriers and bearers in trade, and the production of food for the elite.

Temple I on left taken from North Acropolis ca. 1972

Dumond (1961) and Ferdon (1959) have argued that subsistence was not a problem and that slash-and-burn agriculture can produce substantial surpluses, which would have made it possible for the Maya to subsist by this means. In one of the most optimistic views of the potential of slash-and-burn agriculture in the Maya Lowlands that has been presented, Ursula Cowgill (1961:40) estimates that 100-200 persons per mi2 (39-77 persons/km2 ) could be supported by this means in the vicinity of Lake Peten.

She (1962:277) asserts that "on the basis of reported time required for agricultural work, it appears that one man with some assistance from his wife and children, can care for 13 acres under cultivation in any one year. This is enough to supply the maize requirements for 12.6 people and since the reported size of household averaged 5.78 people, this implies that more than half of the total labor supply is potentially available for construction of monumental centers and specialist activities." She also considers and rejects the possibility that the substitution of stone for steel tools would have had any effect on the efficiency of slash-and-burn cultivation.  Hester (1952) demonstrated, in some experiments he did in Yucatan, that second growth on  an abandoned milpa could be cleared in twice the time it took with steel tools if he used limestone fragments to smash it down. In more sophisticated experiments, however, Sarayday and Shimada (1971) have shown, through physiological measurements of the metabolism of people at work, that stone axes require over five times as much energy and six times as much time to accomplish the amount of work that can be done with a steel axe.  Cowgill's conclusion has also received a serious challenge from Reina (1967) who has shown that even with steel tools and largely unlimited land the modern milpero of the Peten is involved in what is essentially a year-round task of food production that produces a deficit more often than a surplus:

. . . it appears that the milpero just manages to 'break even' all his life by expending a maximum amount of physical energy and by planning carefully [Reina 1967:15].

It has also been implied that Haviland's (1970) figure of 5.6 [he has recently revised it down to 5.0 (Haviland 1972)] for the number of occupants per residential house platform is too high. Sanders and Price (1968:163) prefer a figure of 4 and even suggest that each plaza group was occupied by a household of 5.6 members which would reduce the population of the central 9 km2 of Tikal to about 5,000 persons. For reasons that will be discussed in the second half of this dissertation, in the section titled "Estimating the Number of Inhabitants per Structure," I do not believe these figures are warranted. In sum, none of these attempts to rationalize the indicated settlement density and the productivity data of the milpa appears to be tenable.

Given the gradually accumulating arguments for the Contemporaneity of occupation of the majority of residential structures around Tikal, it is becoming clear that it is the basic assumptions within the subsistence model that must now be questioned again. The evidence for residential contemporaneity at Tikal has been recently summarized by Haviland (1970:191) and will be discussed in detail below.

Since the time of Ricketson's (1937)  survey, as more and more attention has been paid to residential settlement, the evidence for the association of substantial populations with the ancient centers has increased rather than decreased. In light of the resultant subsistence "blind," those who maintain their support of the slash-and-burn model are now talking more in terms of the "precarious" balance that existed between "agricultural man and nature in the Southern Maya Lowlands' (Sabloff and Willey 1967:105) or the "delicately balanced socio-economic system" (Willey and Shimkin 1971:13).

In spite of general acceptance of the slash-and-burn model, various alternatives have been proposed which are based on the premise that locally grown maize could not have supported the great centers in the central portions of the southern Maya Lowlands. Meggers (1954), without challenging the assumption that the Maya in this area relied on the cultivation of maize, asserted that it was incapable of sustaining them. In her essentially environmentally deterministic scheme, she classifies the Maya Lowlands as a Type 2 environment of "limited agricultural potential. . . . This means that a culture of the level attained by the Classic Maya could not have developed in the Type 2 environment where the archaeological remains are found, but must have been introduced from elsewhere. Furthermore, since Type 2 environments lack the resources to maintain so high a level of culture, the history of the Maya occupation of the tropical forest should represent a decline or deculturation" (Meggers 1954:817).

While her lumping of the Amazon and the Peten into a single category can be criticized, and her hypothesis that Classic Maya culture occupation of the lowlands represents a history of decline can be disproven (Coe 1957), her statement as to the limitations of slash-and-burn agriculture in this environment seems to remain valid. As an alternative, Wolf (1959:78) has suggested that staples may have been imported from the highlands down into the Peten, but he does not discuss how the problems of bulk transport would have been handled nor yet what would have compensated for this drain on the productivity of the highlands.

Beyond this possibility, a limited number of alternatives to the slash-and-burn subsistence model have been presented. The grounds for such a dramatic shift in traditional thinking have been presented by Wolf:

If we assume that the Maya of the Peten possessed only slash-and-burn cultivation, how do we explain the numerous ceremonial centers of the area? Slash-and-burn cultivation usually implies a scattered population, a population unwilling to pay homage to a center of control. How did the Maya priests accomplish this feat, not once but many times,  not for a brief period but for well over eight hundred years? It is more likely that the Maya, too, possessed some system of intensive  cultivation, supplementary to their slash-and-burn practices, which allowed them to maintain stable centers of control, while at the same time controlling an ever shifting peasantry [Wolf 1959:77].

Wolf (1957:78) offers the possibilities that the Maya may have maintained a system of chinampas in the great bajos around centers like Tikal or perhaps utilized "some other system of cultivation, yet unknown" (Wolf 1959:60). While the question of bajo cultivation remains open, the potentials of root crops have been examined in detail by Bronson (1966), with criticisms presented by Sanders and Price (1968:92) and Cowgill (1971). Basically, the main criticisms are that, in spite of the comparative bulk of root crop yields, their food value is very low (Sanders and Price 1968:92; Puleston 1968:104). Yet in spite of this face, quite dense populations are sustained largely by root crop cultivation in tropical forest areas in southeastern Nigeria. Bronson (1966:255) claims that densities there are as high as 600 persons per mi2  (232 persons/km2). This is contested by Sanders and Price, who state that areas with over 100 people/km2 are "rare, highly localized, . . . supported in part by importation of staples" from other areas, and associated with a highly developed commercial economy such as the cultivation of cacao. While Sanders and Price (1968:93) leave open the possibility that increasing population may have increased the role of root crops among the Classic Maya, particularly in times of shortage, it is clear that they do not feel that this is the key to a subsistence problem which may have faced Maya civilization during its rise.

Another alternative, which I have presented (Puleston 1968), focuses on a food resource little known to scholars but common in the lowland forests and well known to the Maya, the ramon tree (Brosimum alicastrum). The fruit of this tree is highly nutritious, and the tree is tremendously productive . Evidence for suspecting that this food resource may have been of primary importance in Classic Maya subsistence is found in the present-day distribution of trees and the strong possibility that omnipresent chambered chultuns may have been used for their storage (Puleston 1971).

The Sustaining Area Hypothesis

The principle alternate that remains to be examined, the one on which the entire project presented here was predicated, is the sustaining area hypothesis. Our faith in the existence of a significant sustaining area in 1965 when the Tikal Sustaining Area Project started hinged on three factors: 1) our acceptance of the milpa as the principal means of food production by the ancient Maya; 2) our observation that the land available for food production on a maize base in the central 9 sq. km. of Tikal was clearly insufficient to have sustained the indicated Late Classic population; and 3) the presumption that the bulk of that population was not primarily concerned with food production (Haviland 1963, 1965). The problem was clearly recognized and spelled out as a rationale for a Sustaining Area Project at Tikal by William Haviland (1963:528-540). The first point, challenging the milpa principal, has already been discussed in detail in earlier sections. The second point,  lack of sufficient land within central Tikal, will now be discussed in terms of Haviland's work. The third point, ?????????????, will be discussed in the final section on kitchen garden utilization of the ramon.

Of the total 9 km2  covered by the main site map, approximately 12% is comprised of swamp soils [Carr and Hazard (1961:10) indicate a figure of about 14%] with another 27% taken up by structures, plazas, quarries, reservoirs, and causeway (Olson 1969:95). With about 5.5 km2 of upland soils available for cultivation and a minimal estimated Late Classic population of 8,092 (Haviland 1965:19), less than 4,000 m2, or about 0.9 of an acre, would have been available for an assumed average household of 5.6 persons (Wauchope 1938:131; Haviland 1965:16). Population density on the non-bajo uplands would be 1,470 people/km2. We have already seen that Sanders (1962-3:237) has estimated that slash-and-burn agriculture in the Peten will support only 30-60 persons/km2 without deleterious effects on the natural environment. These figures agree fairly closely with U. Cowgill's (1961:40) estimate of 34-77 persons/km2 and are probably reliable. While Haviland's population estimate may be too high, to reduce it by half and assume that all the swamps could be cultivated, does not raise the land available per family to much more than 9,200 mi2, or 2.3 acres per family. Cowgill's (1960:1010; 1961:40; 1962:277) data indicate that as an absolute minimum the "family" must have 20 acres of dependable, arable land.

On this basis, the "dispersed settlement" model (Willey's Type b 1965a) is clearly out of the question, if subsistence is based on "maize as the single major subsistence crop," has been most recently reiterated by Tourtellot and Sabloff (1972:128). At least one element has to be abandoned: maize as the major subsistence crop or self-sustaining dispersed settlement. As will be argued in the conclusions of this presentation, it looks as though both  are precluded in the case of Tikal.

As noted, in 1965, when the Tikal Sustaining Area Project was initiated, we were willing to accept maize as the major subsistence crop. The hypothesis we were working under, which had been constructed by Haviland, assumed that settlement density at Tikal was going to have to drop off somewhere beyond the limits of what had already been mapped. Beyond this drop-off, one would expect to counter-balance the supposed insufficiency of cultivable land in the area closer around Tikal. While dual residence might account for some of the structures found in such an area, there should still be at least 20 acres of good land per house. These less heavily settle uplands, then, would have been Tikal's "bread-basket," where the Tikal population rotated their milpas, utilizing the traditional techniques of slash-and-burn cultivation (Haviland 1963:528).

Go Back to Top of Page

PagesNotesPrefaceIntroductionCh 1Ch 2Ch 3Ch 4Ch 5Ch 6Ch 7Ch 8Ch 9ContactVoivode Multimedia