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 Five: The Calculation of Prehistoric Demographic Patterns


The Assumption of Residential Contemporaneity
Arguments Against Residential Stability
Arguments for Residential Stability
Hidden House Ruins
Estimating Inhabitants per Structure


There is an essential relationship between demographic and subsistence patterns. At this point, in spite of the attendant difficulties, it is important to attempt a calculation of Tikal population densities at various times. A number of steps are involved.

The Assumption of Residential Contemporaneity

The controversy as to whether or not most ancient Maya house platforms were occupied contemporaneously, at least in Late Classic times has been discussed in detail by Haviland (1963:502, 1969:429, 1970:191). The various arguments for and against a high degree of contemporaneity will now be discussed in the light of the present survey as a step toward the calculation of prehistoric population sizes.

Non-ceramic Evidence for Continuous Occupation

As was pointed out above, the temporal divisions of the Tikal ceramic sequence are not fine enough to prove or disprove continuous occupation by themselves. This has been used as a major argument against continuous occupation by Sanders and Price (1968:165). Non-ceramic evidence has been brought to bear on this issue by William A. Haviland (1969:429, 1970:191). In these publications Haviland brings together from his excavation experience four convincing arguments for the continuous occupation of Late Classic Tikal house ruins.

Haviland's first point depends on the evidence for frequent alteration of structures "during spans of time with no apparent change in ceramics." He presents the example of Str. 6E-26 which was built and then modified five times within the 200 year time span that Imix ceramics were in use. Accordingly, the structure must have undergone an alteration once every thirty years or so, on the average, which coincides with Wauchope's (1938:151) estimate of the "lifespan" of modern Maya houses. "There are other structures which suggest time intervals between alterations of ten to forty years" (Haviland 1970:191).

Secondly, Haviland asserts that "when small structures were alters, portions of older floors and walls commonly remained in use." This situation precludes the possibility of structures having been abandoned for any length of time since component walls and floors would have deteriorated quickly upon exposure to the elements, particularly when thatching began to decay after a long period of use (Haviland 1970). Pole and thatch structures do not last much longer than 30 years in the Peten, even when fairly substantial materials are used. When I visited  Uaxactun for the first time in 1963, the main building of the Carnegie Institution headquarters, built in 1927 or 1928 (see Ricketson and Ricketson 1937, Plate 5b), was still standing. The thatching had holes in it, but it was almost certainly not the original material anyway. When I returned in 1966, the main posts had given way and everything lay in a rapidly decomposing heap covered with advancing vines. The land around the building had evidently been kept clear until a few years before I saw it for the first time. It was at the point of collapse, that destruction of floor surfaces and wall components would have begun in earnest.

Den on Temple V 61a

The third argument of Haviland is that certain middens show evidence of continous deposition. Haviland (1970:191) points out that if "associated architectural developments stood abandoned for any length of time, soil would have accumulated over the midden before further deposition, and would have left stratigraphic evidence of layering within the middens." The rarity of disturbed burials in structure platforms is his final piece of evidence. The intrusion of as many as 11 non overlapping burials into a single platform (Str. 2G-59), at different periods in time "implies remembrance of their location on the part of the Maya over several generations" (Haviland 1970:191). Actually one of these (Bu. 53), but because the later one was intruded to a depth of only .10 m beneath the present mound surface, the earlier one was not touched (Haviland 1963, Fig. 8, p. 562). Haviland's arguments will be assumed to be applicable to the structures investigated in the course of the present survey.

The following percentages of residential contemporaneity for the combined samples of the north and south survey strips are at least possible: Preclassic 29%; Early Classic 98%; Late Classic 93%; Post Classic 2%. These data are tabulated in detail in Table I.

While trends are evident in the comparison of the representational values for Ik, Imix, and the combination of Imix/Eznab and Eznab, absolute values are probably best obtained from "overall" Late Classic regional period figures. The reason for this is the frequently, particularly on the 0 - 6 km portions of both the North and South Survey Strips, ceramic identification could not be carried further than simply Late Classic because of poor preservation. These identifications are shown on Figure 12 and 13 with the parallel hatchure. The exclusion of such identifications from the percentage values for the time periods represented by the Tikal Ceramic Complexes (Ik, Imix, and the combination of Eznab and Fry's Imix/Eznab transition) reduces these values, in my estimation, to below what they should be.

A close look at the information tabulated for Ik, Imix, and Eznab and Imix/Eznab ceramics might at first seem to suggest that much less than 93% of the house mounds were inhabited in Imix times. The "dpp" (definite, probably, and possible) value for Imix in stratum II of the North Brecha Survey Strip given in Table I is 43%. Reference to Figure 12, however, reveals that ceramics from many test pits could be identified as no more that "Late Classic" material and that in fact if these situations are included in the "possible" categories the dpp figure jumps from 43% to 79% . Thus even though the test pits were quite productive, samples were still occasionally too weathered to provide diagnostic sherds for the smaller subdivisions of the main ceramic complexes. All such samples are indicated with a slash symbol in Figures 12 and 13.

Arguments Against Residential Stability

Because of the great length of the time periods represented by sherds attributable even to specific Tikal Ceramic Complexes, it is important to investigate any further arguments bearing on continuity of occupation. Perhaps the best way to begin is with the arguments which have been presented to contradict the assumption of fairly continuous occupation. At least three important ones have been implied in recent literature. They are: 1) the necessity to shift cultivation of  maize, the assumed subsistence base of the ancient Maya (Sanders 1962-3;2) burial practices which involve the abandonment of a residence after the death of an occupant (Thompson 1971; and 3) residential instability as a culture trait of the Maya (Thompson 1971).

Table I - This table presents the percentages of the randomly selected and excavated mound groups which produced ceramics of major ceramic phases for each of the seen strata defined. Percentage figures are given for 1) definite (d), 2) definite and probably (dp), and 3) definite, probable and possible (dpp) identifications. Based on ceramic evaluations made by Robert E. Fry and recorded in the Tikal Ceramics Evaluations volume.

Table I

1 - Fry (1969) defines 5 stratum for the South Brecha Survey Strip. The extra one, (stratum 3) which Fry assigned to the satellite site of Bobal, is not included here. His strata 4 and 5 are relabeled here as III and IV.

Shifting versus Permanent Cultivation

The first of these arguments against the permanence of Maya residence is based on the premise that shifting cultivation would have required a shifting of multiple residence pattern in a moderately populated area because any settlement would soon be surrounded by land in fallow. According to this argument, the Maya farmer, rather than walk a long way to his milpa would simply abandon his residence (perhaps for only a decade or so) and move nearer to the land he was currently cultivating. Though the argument has never been defended in the literature, it appears to have been assumed by Sanders in an attempt to rationalize indicated settlement densities in relation to the land requirements of slash-and-burn agriculture.

If we assume only one-quarter of the mounds to have been occupied simultaneously, then a respectable density of 100 to 200 per square kilometers of population may be calculated. This is within the limits of a system of maize cultivation . . . (Sanders 1962-3:210).

Since the assumption of one-quarter occupancy was based on the further assumption of a slash-and -burn subsistence model, my ability to demonstrate the likelihood of a far more productive kitchen garden model permits me to accept the settlement data at face value.

Burial Practices and Residence Abandonment

A second argument, posed by Thompson (1971), is that the protohistoric pattern of burying the dead beneath the floors of their homes and then abandoning them extends back into Classic times. One of the sources quoted by Thompson, Tomás de la Torre, a Dominican under Bartolomé de las casas in 1545, reports of the Maya that "when someone dies they bury him in his house and immediately lay it in ruins" (Ximenez 1929:31 II 34:3-1). apparently a witness of similar practices, Landa (Tozzer 1941:130) writes, "Usually they abandoned the house and left it deserted after the burials, except when there were a great many persons in it, so that they with their society lost some of the fear which remained in them on account of the death."

If this practice of abandonment was carried out, certainly a large proportion of Maya house sites would have stood empty even if they were eventually reoccupied. Landa's notation of exceptions to this practice should not be taken lightly, however. There are several reasons for doubting that house abandonment was generally practiced by the Classic Maya of Tikal. First of all, only small proportions of the dead were buried beneath the floors. Haviland (1963) reports on 17 burials (Burials 49, 50, 52, 53, 54, 55, 56, 57, 58, 60, 61, 64, 65, 66, 67, 68, and 69) found beneath residential structures he excavated. Not one could be identified as female. Nine could be identified, at least tentatively, as males. thirteen could be identified as adults. Two of the burials appeared to be of youths 12-20 years old (sex indeterminate). In other words, women and children were generally placed elsewhere. Judging from the fact that adult males as well as females are often found in chultuns (Bu. 62) and in pits (Bu. 45, 49) outside structures, it seems that even adult males were not automatically accorded a subfloor burial. Since adult males represent only a fraction of the population, sub-floor burials may have been a relatively unusual occurrence associated with the death of the head of a family or household. The association of structural modifications with these burials (Haviland 1963:369, 500) would seem to support this hypothesis, particularly when it is noted, on the basis of excavated examples, that these modifications are separated by an average of thirty or forty years. Thirty or forty years is perhaps a reasonable estimate of the life expectancy of an ancient Maya individual who survived childhood. Of the 17 Burials listed above, 7 (Bu. 52, 54, 56, 57, 58, 68, and 69) "may reasonably be interpreted as contemporaneous with abandonment of an existing structure . . . and construction of what, in effect was a new structure at the same locus" (Haviland 1963:366). The floors and fill above 6 more (Bu. 49, 50, 53, 55, 64, and 67) were too completely destroyed to allow a judgment on whether or not they were contemporaneous with a sequence of abandonment and new construction. Only one (Bu. 60) (and even this is not certain) seems to postdate the latest construction. Thus it is likely that structures were not abandoned when burials were placed in them as they seem to have been in historic times. This does not mean that total abandonment might not have been preferable; but if, in fact, space was limited, the spiritual dangers involved in moving back into a house in which the dead had been interred could perhaps have been ameliorated by ritual. Such a practice is suggested by a quotation from Lopez de Cogollundo, translated by Thompson (1963:132).

Whenever they make new houses, which is every ten or twelve years, they won't enter or live in them until an old sorcerer comes from a distance of one, two, or three leagues to bless them with his silly charms.

Thompson provides this quotation in the context of a description of "new house rites", he witnessed in San Antonio, British Honduras, in the 1920's. It is very interesting to note that the ritual involved smearing the house posts with raw meat and sprinkling posole inside and outside. The house was then abandoned for two hours so that the hut and the "gods" could take their repast in peace. Such practices are suggestively reminiscent of rituals used to propitiate the dead in Chan Kom, where the bones of the deceased are exhumed and brought back to the house for a ritual meal during the "days of the dead" in early November (Redfield and Villa Rojas 1934:203). New house rituals like those of San Antonio are also practiced in Chan Kom (Redfield and Villa Rojas 1934:146) and Zinacantan. At the latter the ritual involves  the standard "feeding" of the house but is highlighted by the strangely significant burial of a rooster beneath the floor of the "new" house.

The rooster's body is buried in the center of the floor and sprinkled with liquor. The earth is stamped in on top in the same manner used for burying a person in a cemetery" (Vogt 1969:462, emphasis mine)

At this point, in the process of similar rituals in ancient Maya times, the dead may have been buried beneath the floors of their houses. The fact that many sub-floor burials at Tikal were apparently performed during the process of construction seems consistent with this conclusion.

With respect to the archaeological record, it is important to note that Pendergast finds an occupation-burial-re-occupation sequence, without evidence of abandonment at Altun Ha in Balize which matches the Tikal data presented by Haviland (1963).

A general pattern is discernible in the housemound burials, both from this zone and from other areas of the site: When an adult male, probably the head of the household, died, he was buried beneath the house floor and the building was partly razed and covered with a new structure which the bereaved family occupied;. . ." (Pendergast 1970:26).

On the whole, the available archaeological evidence seems rather to work against Thompson's argument that houses and the platforms they stood on were abandoned after the death and burial of an occupant.

Residential Instability

A third argument against the assumption that most Classic Maya house sites were occupied continuously and therefore contemporaneously is based on the "well-known tendency of Maya communities to move from one site to another . . ." (Thompson 1971:215). As Haviland (1972:262) has pointed out, it is risky to draw parallels between the post-Conquest and the Classic Maya eras without careful examination of the dynamics of each situation. Surely there are great differences in economic, social, religious, political, and demographic conditions between these two eras. The "communal and individual volatility" (Thompson 1971:215) of the post-Conquest Maya may to a large degree be attributed to their dissatisfaction with Spanish rule and the absence of a strong native political authority (Haviland 1972). As Haviland points out, the Tikal earthworks may indicate that unsettled times were not unknown to the ancient Maya; but they are also evidence of strong centralized control (cf. Puleston and Callender 1967:48), under which the sort of mobility we see in colonial and modern times was probably impossible. While Thompson sees this mobility as a "characteristic response of the Maya" and thus a stable genetic or cultural trait, demographers seem to view it as a variable linked to economic considerations that are always subject to change.

Dual Residence

While a formal argument for dual residence has not been presented (so far as I am aware), this pattern is still a possibility worthy  of consideration. It could be argued that the Maya maintained an infield-outfield system with a kitchen garden and house near Tikal and a milpa and house in the more sparsely populated intersite area. Such pattern would reduce any population estimate by the number of people assumed to be living in the structures found in the outlying intersite area. I would argue against the likelihood of this situation for a number of reasons. First, there would be less reason for the milpa structures to be build ton elevated platforms around a plaza, as the structures nearer Tikal are, if they were being used to keep part-time residents near their fields. Second, where structures occur together in the intersite area, they show the same "kitchen garden" spacing found in the more densely settled areas (e.g., Str. SE-395/407; 10.8-12.0 km. on the East Survey Strip; 8.5-9.0 km on the West Survey Strip; Strs. NW (N)-222-257, 291, 292; Strs. NE(N)-108/118. Finally, there are not nearly enough structures in the intersite area around Tikal to provide a dual residence for more than a very small proportion of the many thousands of structures that lie within the more densely settled kitchen garden area. Later I will discuss more fully the possibility that outlying areas were under milpa cultivation but that the maize produced on these fields was used primarily by the controlling elite.

Arguments for Residential Stability

Subsistence Stability

Having dealt with the principal arguments against a relatively stable ancient Maya population, we may now consider the arguments for it. Most important is the gradually accumulating evidence for the importance of a more intensive food production system, particularly dooryard or kitchen garden tree-cropping and horticulture (Puleston 1968, Wilkin 1971). The argument basically is that once the slow-growth  orchards and garden plant assemblages had been established at a house site, the decision to move would involve a considerably greater loss of investment than it would for a milpa farmer. The latter can set up a new and perhaps even more productive food producing area in a single year. The careful tree-cropper might spend as much as a decade setting up his slash-and-burn agriculturist. Even if he did decide to move his chances of finding a new site that was better than the one he occupied would be small if population densities were high.

The Burial Data

Haviland's data in favor of the hypothesis that houses continued to be lived in after certain male individuals, presumed to be occupants, had died and been buried beneath the floor has been discussed above. The construction of a new platform over the old one seems to have accompanied this practice often, though not necessarily. Six burials were apparently intruded into the fill of Str. 2G-59-1st without requiring a new construction (Haviland 1963:66). Occasionally, two burials were placed in the fill of a new structure during construction. In the former case it might be suggested that modification s involving replacement of the pole-and-thatch portions of the structure may have been sufficient (Haviland, personal communication). In the latter cases it is also possible that two adult males of the same house or household died at the same time. It is also perhaps worth examining the possibility that the dead were somehow preserved, perhaps by drying, until familial economic factors permitted the construction of a new building. Since many burials from Op. 24 lacked teeth which were not clearly the result of antemortem loss (Bu. 49, 50, 52, 53, 56, 57, 59, 60, 61, 62, 63), it is certainly possible that they fell out during a postmortem-preburial drying stage. Shrinkage of the gums during such a period would have been a major factor in loosening the teeth and would not necessarily involve decay to the point of disarticulation of the limbs if it was carefully done. In line with this possibility, at least one of the above burials (Bu. 62 did  show signs of minor disarticulation and was labeled a two-stage burial by Haviland ( 1936:334). Landa (Tozzer 1941:131) mentions the postmortem cleaning and preservation of heads and the making of wooden statues of "fathers" with fragments of skin attached to cover the holes in which ashes were preserved above ground. Perhaps the bet lead, however, comes from the discovery of 379 teeth from a minimum of 43 people discovered in a deposit along the face of a low platform in the Main Plaza of Yakalche in Belize (Pendergast, Bartley, and Armelagos 1968). Postmortem extraction was indicated by the observation that about 20% of the teeth had been extracted prior to eruption. While all the teeth from this deposit were from young people (19 years of age or less), the postmortem extraction of teeth for ceremonial purposes is unquestionably demonstrated by this fortuitous find. Since numerous caches at Tikal contain human teeth (W. R. Coe, personal communication), the find at Yakalche suggests the possibility that these too are relics of postmortem extraction used for ceremonial purposes. This does not preclude the possibility that some teeth were also removed from living individuals by knocking them out, though Landa (1941) makes no mention of such a practice in his discussion of mutilation.

With respect to the Tikal burials, Haviland finds definite evidence for postmortem removal of teeth in only one case, Burial 63, though he leaves open the possibility that other individuals may have suffered removal and perhaps accidental loss of teeth before burial.

The fact that all remaining teeth [of Bu. 63] had been removed prior to burial is of interest, for this is the only sure example of such practice from Operation 24 burials. However, in many other poorly preserved burials lacking teeth, postmortem removal of the teeth is a possibility, but it is impossible to prove (Haviland 1963:338).

The fact that information on postmortem burial customs of the ancient Maya is of direct relevance to the interpretation of settlement patterns provides a valuable lesson for the importance of a broad approach to archaeological problems. The data as discussed above seem to support the hypothesis that in Classic times, at least, structures were not abandoned after the death of an occupant as they were in historic times.

The Permanence of Residential Architecture

Another factor which bears consideration is the permanence of ancient Maya residential architecture in comparison to that found in historic times. While historic Maya of Yucatan and the Peten sometimes build substructure platforms to level the ground on uneven terrain or hill-slopes, "the Classic Maya seemed to use them in practically every instance" (Wauchope 1938:15). In modern times, substructure platforms are only rarely constructed if the ground is even nearly level. Also lacking in modern situations is the paved plaza courtyard which is almost invariably associated with ancient Maya residences. A common denominator in both these features, substructures and plazas, is the extra time and labor necessary to construct them. How do we explain this difference between ancient and historic residence construction patterns? One possibility is that ancient Maya families, because of their reliance on food producing orchards and kitchen gardens, were able to depend on staying in one place longer than the modern Maya, who subsist on slash-and-burn agriculture, which permits them to move from one place to another fairly easily with a minimal loss of the subsistence investment. The greater efficiency of kitchen garden tree-cropping, in terms of the labor investment required, versus the subsistence yield, would also have given them more time and energy to invest in such activities.

Denny with Ed Shook 1964 at housemound excavation

Thus the greater care and higher expenditure of labor seen in ancient Maya residence construction is taken here to be suggestive of three points: 1) a spatially more stable subsistence system in the past; 2) a greater availability of time and labor for non-subsistence activities; and 3) greater permanence of occupation of structures once they were completed.

Hidden House Ruins

So far all the discussion has proceeded on the basis of the assumption that counts of ancient Maya house ruins are maximal, and our only concern is to determine whether or not they were occupied fairly continuously or not. The possibility that we may be underestimating settlement and population densities must also be examined. Most population estimates in the Maya Lowlands have been based on the assumption that the ancient Maya did not build houses without first constructing the substructure platforms which make settlement mapping so easy over most of this region. But the possibility that a substantial number of ancient house sites without substructures (i.e., "invisible" or "hidden" house mounds) have been overlooked by mappers has been a subject of concern at Tikal for some time. More than a dozen "hidden" house sites have been excavated at Tikal.

Strs. 4F-42, 43, 48 and 49, the first of these to be excavated were simply found in excavations near mapped structures. In 1961 we started using isolated chultuns as leads to hidden structures (e.g., Strs. 2G-61, 62; 6C-60, 61, Sub 1). Later we began looking where there were neither structures nor chultuns. In this way Groups 5C-3, 5F-2, and Str. 7C-62 were found.

Strs. 4F-42, 43; 48, 49; and 2G-61, 62 were the first of these hidden structures to be excavated at Tikal (Haviland, personal communication). They actually did have substructure platforms, but because the platforms were so low they were not visible from the surface.  The discovery of these structures, along with others I excavated in 1963, was guided by the occurrence of chultuns in isolation (i.e., comparatively distant from structures identified on the Tikal Report 11 maps). In every instance where this technique was tried, evidence for structures, in the form of very low platforms, post holes, and heavy scatters of ceramics (including middens) was found. In the vicinity of Chultuns 6C-7 and 9, however, only post holes and ceramics were found, indication either that substructureless houses were utilized or that former platforms had been removed for use as fill elsewhere. In the cast of Str. 5C-57 it seems likely that platform fill was used to fill Ch. 5C-8. In 1966 Bennet Bronson devoted an entire season to testing for "hidden" structures and found evidence of unmapped structures in all five areas he tested.

How much would full knowledge of these "hidden" structures alter our picture of settlement patterns and population density? In all 26 are now known. It is interesting the "without exception, these known 'hidden' structures were abandoned by A.D. 770" (Haviland 1970:192), which is roughly the time when all mapped groups were in use. Since many of the "hidden" structures appear to have been in use only in Early Classic times, their existence may not significantly influence population estimates based on counts of visible mounds occupied during the Late Classic period. They do, however, appear to influence previous population, or at least settlement density estimates for Early Classic times.

While it is impossible to estimate just how many Early Classic 'hidden' structures remain to be discovered at Tikal, this is offset by the fact that not all such structures were in use at one time. However, the indications are that almost as many groups of small structures were in use between A.D. 550 and 650 as by A.D. 770. Therefore, contrary to my statements in earlier articles, it appears that structure density reached a peak by A.D. 550. Then, for just over two centuries, the abandonment of existing structures was offset by construction of new ones, so that overall density remained constant (Haviland 1970:192).

More knowledge of "hidden" structures would be well worth obtaining. I suspect  however, that full knowledge of them would change our basic conception of Classic Maya settlement only a little.  The areas that have produced "invisible" situations less frequently. W. R. Coe (personal communication) reports that the level area between Strs. 5F-9/12 and 5F-13/6, which was cleared by bulldozer in the winter of 1967-68 for a soccer field, revealed no signs of habitation or sherds when he walked over it. This is not surprising, however, since level ridge saddles such as that which occurs here were rarely used as sites for visible construction. Nevertheless, the little excavation that has been carried out between widely spaced, as well as normally spaced groups, requires that any calculations of, at least, Early Classic settlement or population density must be suffixed with a plus for the unknown quantity of "hidden" situations we now know must be there.

Figure 18. "Test-hole" survey technique devised to aid in the location of "hidden" structures through the use of a post-hole digger. On the above site, which was used to test the technique., the structures were not hidden but consisted of fills with low ceramic content.


In view of the great labor investment required for such excavations, what is really needed is a rapid means of testing large "vacant" areas. In 1967 I devised what may be a solution to this problem, but, though I was able to test it, a good opportunity to use it did not arise. The system used by Bronson, which required the excavation of large two meter required test pits to bedrock, was by comparison very time consuming. An area that would require at least two weeks to test by his method could be covered by mine in a single day.

Basically the system requires the use of standard scissor-bladed, vertical-thrust, post-hole diggers as described above for use in the test-pitting survey (Fry 1972). A grid is laid out over the area to be tested, and the points of intersection are excavated to bedrock, if possible. (To avoid confusion with Maya constructions, these holes will be called "test-holes" rather than post-holes.) Experimentation revealed that a grid with five meter intervals was sufficient to pick up an occupation area with comparatively low sherd density in surrounding soils. Most test-holes on a large grid produce no sherds. Use of the post-hole digger for midden search in the ceramic survey shows that even in heavily settled areas, in the immediate vicinity of structures, many holes will produce nothing.

Any Test-hole that produces a sherd or sherds is "checked" by digging four test-holes around it. These "check-holes" are located at the corners of a five-meter square centered on the primary hole that produced the sherds. If the "check-holes" do not produce sherds, it is assumed that the first find was simply a chance find. If on the other had, the "check-holes" produce more sherds, an occupation site may be present.

Figure 19. "Test-hole" survey north of North Earthworks at excavated causeway (Puleston and Callender 1967, Fig. 1, B). Survey was conducted in an attempt to locate an ancient trail possibly headed towards Uaxactun.


The results of a test-hole survey, run over "visible" house ruins with a low ceramic density, as revealed by intensive excavation of this group, is presented in Figure 18. It is clear that the test-hole survey reveals a high frequency of sherds in the area of the ruin in comparison to other areas in the vicinity. As mentioned above, any discovery of significant sherd concentrations revealed by this method must be checked by a regular excavation to confirm the presence of a structure. Such confirmation might result from the discovery of a very low construction beneath the surface post-holes. Such a trench was put in across the "five sherd cluster" and produced no such evidence. In an excavated area of almost seven square meters, only two additional sherds were found. By contrast, more extensive excavation in the "32-sherd cluster" area which revealed ample evidence of construction, artifacts, and greater concentrations of ceramics.

The technique was also used to search for evidence of a trail extending north of the excavated causeway on the North Earthworks. The possibility that ancient trails could be followed through the jungle by use of a technique of this kind deserves further attention. Figure 19 presents the results of our North Earthworks survey which revealed a slightly higher incidence of sherds northwest than northeast of the earthworks causeway. This distribution may result from incidental scatter of sherds along a trail extending northwest in the direction of Uaxactun. Haviland (1972, personal communication) working with Bronson's test-pit data, claims he has evidence for a likely trail running from Str. 3D-126 to Str. 3D-14. At this point the fascinating possibilities of a "trail Survey" conducted y the methods described above can only whet my appetite for further experimentation. I do not consider the efficacy of the technique to be demonstrated.

Estimating Inhabitants per Structure

To return now to the question of calculating aboriginal population densities, a third problem centers on arriving at a figure representative of the number of inhabitants who occupied a single structure. Different sorts of data have been brought to bear on this problem, including calculations of the carrying capacity of the land, as well as ethnohistoric and modern ehtnographic data on the number of people who generally occupy a single dwelling.

Ethnographic Data

In Chan Kom, Redfield and Villa Rojas (1934:91) calculated an average of 5.6 inhabitants per house. This figure is quoted by Wauchope (1938:145) and has been used by Haviland (1965:19; 1969:429; 1970:193) in various calculations of Tikal population size. Recently, he has reduced what he believes to be a "minimal" figure from 5.6 to 5.0 people per house (Haviland 1972:138). He considers this is a safer figure on the basis of reconstructed household statistics for Cozumel where, for reasons that are not discussed, he derives a norm of 3.7 couples (one elder couple and 2.7 young couples) per household.

The assumption of four surviving children per young couple is apparently based on a fortuitous similarity between the population structure of Tikal (as indicated by skeletal remains) and modern Maya populations, studied by Steggerda (1941), where medical care is virtually nil. These modern Maya have an average of three or four children per family. Haviland uses the upper figure for his Tikal calculations because: 1) the ancient Maya did not have Old World infectious diseases to cope with, as modern populations do; 2) the subsistence potential of the Peten is higher than that of Yucatan, where Steggerda's populations were located; and 3) the "efficiency" of organization was probably higher at ancient Maya sites like Tikal in comparison to modern communities (Haviland 1972:137). The latter includes superior food distribution systems which presumably reduced death rates. The combination of one elder couple, 2.7 young couples, and four children per young couple yields a total of 18.2 people per household. With 3.7 couples per household, each of these representing the nucleus of a "nuclear family," there would thus be an average of 4.9 people per nuclear family. This figure is rounded off to five because of the probably higher survival rate for Tikal children as discussed above, and the likelihood that a widow or widower in a society with extended patrilocal families, would move into the house of a married son (Haviland 1972:138). The latter would slightly increase the size of the average "nuclear family," but there is no mention of the lack of statistical significance this would have for the number of people per house since the total number of people and houses would not be affected by such a move. In any event, the figure of five people per house is used for population calculations on the basis of the still unproven assumption that "each individual house was occupied by a single nuclear family consisting of a married couple and their unmarried offspring (Haviland 1972:136).

There are some weak links in this chain of assumptions, data, and logic. The one that worries me most, however, is the conclusion that the average family reared an average of just over four children. A survival rate of this magnitude would imply a population that doubled itself every generation. Assuming a generation to be 20 years a Late Classic population of 30,000 in 600 A.D. would expand to 430,000 by 700 A.D. Even occasional outmigration could not handle such a rate of increase. In fact the archaeological evidence indicates that the population from 500-700 A.D. was fairly stable (Haviland 1970). As an alternative I suggest that Haviland's estimate of children per families too high and that a good deal more of the occupants of each house consisted of unmarried adults, widows, widowers, or childless couples. Haviland make no allowance for these categories.

Another factor in Haviland's reduction of the 5.6 figure appears to have been this own average for household statistics from six 20th century Maya communities in Yucatan (date from Steggerda 19761 and Wauchope 1938) which produce a mean of 4.58 persons per house. An interesting resource which Haviland, however, has not considered is that provided by Villa Rojas (1945).

The least acculturated Maya on the Peninsula of Yucatan are found in east central Quintana Roo (Villa Rojas 1969:244). Redfield (1941:13), who was specifically interested in the acculturative continuum revealed by different Maya communities wrote: "In short, Yucatan, considered as one moves from Merida southeastward into the forest hinterland, presents a sort of social gradient in which Spanish, modern and urban, gives way to the Mayan, archaic and primitive." Of the four  communities on this continuum, studied under the direction of Redfield, two,--Dzitas and Chan Kom, are listed by Haviland. The most and least acculturated, Merida and Tusik, respectively, are not, The monograph on Tusik and other X-Cacal settlements of central Quintana Roo (Villa Rojas 1945) fortunately contains data relevant to the calculation of household size. Data on the nine settlements covered by this study are presented in Table 11. Here we find an average of 6.07 people per residential structure. This is considerably higher than the figure for the more acculturated communities referred to above. These data are more in line with the situation on Cozumel in 1570 as will be indicated below.

Though the ancestors of the X-Cacal must have been affected by introduced European diseases, they were probably less influenced by subsequent social and economic changes that were promulgated in the more accessible portions of Yucatan which produced the village studies by Wauchope and Steggerda. Once the devastating initial epidemics of the new diseases had run their course, it seems more than likely that the people would have returned to former living patterns where it was possible. In total this may have been a period of less than a hundred years. Violent outbreaks of epidemic diseases even during this period, however, were generally separated by years of comparative stability during which times many of the threads of the former way of life could be picked up again. The dramatic die-offs that followed the Conquest were ended by 1600: between 1520 and 1600 the populations of central Mexico declined by 90% (Borah and Cook 1963). A similar pattern pertained to the Maya Lowlands (Thompson 1966). Once this toll had been taken the population curve which had been plummeting began to level out as resistances to the most devastating epidemic diseases became established.

Table II


We know the Spanish were trying to change Maya living patterns where they could get at them by breaking up extended family households (Roys, Scholes, and Adams 1940). Changes introduced by "apostates" who were accepted into the more remote communities were certainly minor in comparison. Still, changes undoubtedly did occur as Spanish influence broke down the extended family pattern in areas of more direct contact.

I would like to suggest, on the basis of the Cozumel and east central Quintana Roo data, that the aboriginal household pattern was one of higher density than is generally found in more acculturated situations. This tentative conclusion is supported by available ethnohistoric evidence, which is presented below.

Ethnohistoric Data

On the basis of his work in the Archives of the Indies in Seville, Hellmuth (1971:23) reports that "one thing which is consistent among all reports on most of the 16th - 17th century Southern Lowlanders is that there were several families per house, not just the nuclear family as found by Sanders." While it may be argued that "house" refers to a household group composed of several residences rather than a single structure it should be pointed out that it does not for the X-Cacal settlements just discussed. Nor does it have that meaning for Landa (1941:86-87) who when he is describing a situation involving small residential structures facing onto a single plaza the term "casa" is obviously used to refer to individual structures rather than the household or the plaza group as a whole (Landa 1941:216). I will assume that Landa's usage was typical and that "house" refers to a single building.

Villagutierre (1933:136) describes the house of a Chol cacique which was occupied by 19 persons. Elsewhere as many as 25 people were "crammed" into a single structure (Villagutierre, 1933:480). Hellmuth (Nov. 1971, personal communication) has informed me of early post-Conquest descriptions of Cholti-Lacandon houses in which sleeping platform were arranged vertically like "plank beds on a troop-ship." Haviland )personal communication) has suggested that these high figures can be attributed to the possibility that the populations of these communities were constantly augmented by "apostates" fleeing into the jungle rom the Spanish. While these communities may constantly have been augmented in this way it seems unlikely to me that such newcomers would have been given space in existing houses instead of being required to build their own.

Avendaño (according to Thompson 1951:390) estimated a population of 2,000 people for Noh Peten (the present island of Flores). Though he was there eight years earlier, Fuensalida's estimate of 200 houses for the island (Cogolludo 1842-45 II:230) permits us to calculate a likely average of 10 persons per house. As G. Cowgill notes (1963:496) this figure is in line with the idea that "there was more than one nuclear family per house  [emphasis mine]. as is common in pre-Conquest Yucatan, as well as Pedro Uc's testimony that in 1573 this was the custom in the Peten town he had been in (Scholes and Roys 1948: Appendix D)."

The Cozumel census of 1570 (Roys, Scholes, and Adams 1940), cited by Haviland (1972:137), describes San Miguel with a population of more than 119 people living in 17 houses, an average of at least seven persons per house and Santa Maria with 183 people in 22 houses, an average of 8.3 persons per house. These figures appear to be more in line with X-Cacal and Noh Peten data than with those of the more acculturated settlements. In spite of the fact that rules of residence may have been changing by 1570 (Haviland 1972:137), Maya notions regarding household size and the limits of personal space requirements may have been fairly stable. The apparent post- and pre-Conquest pattern (Cowgill 1963:496) of more than one nuclear family per house may have considerable antiquity.

While there may appear to be an inconsistency in my arguing for greater stability in who and how many people the Maya permitted to live under one roof, when I have also suggested that important changes have taken place in household burial practices and residential permanency elsewhere, I think that each pattern must be considered separately. I doubt that any anthropologist would argue that all aspects of any culture are equally stable or equally subject to change.

Tabulation of the Survey Data

Survey data relevant to calculations of population density are presented in appendices 2 and 3. The Survey strips were broken into one-quarter km2 blocks (however, blocks 24, 25, 72 and 74 are each one-eighth km2, block 110 is slightly less) which are presented in percentages taken up respectively by tintal (Ab1), escobal (Ab2), areas covered by construction and thus unavailable for cultivation (Ah), and finally, after the subtraction of these figures from the total block area, the land left for cultivation (Af). Descriptions of the tintal and escobal bajo or swamp associations are presented in the section on vegetation. In the initial calculations presented in appendices 2 and 3, it is assumed that these areas did not play a significant role in food production; later, however, in Table 6, figures for cultivable land both with and without them are given.

The Af figures of appendix 2 serve as a basis for the final calculations in Appendix 3. For Appendix 3 the percentage representations for each of the indicated ceramic complexes, shown on Table 1, are applied to the block structure counts. These figures provide an estimate of the number of mounds in use for each time period in each block. It is important to point out here my assumption that the percentage representation of occupation in terms of groups is equivalent to percentage representation in terms of structures. For the calculation of population density, this means we are using the maximum number of structures occupied per group in the calculations in spite of the fact that groups invariably start out with one or two structures, with others being added later. A useful breakdown of this sort of increment, however, will be very difficult, if not impossible, until we have better dating controls that will allow us to place all these developments at separate plaza groups together on a single time scale.

We can note here that the increase in the number of structures per group from Early to Late Classic times, which seems to be characteristic of the central nine km2 of Tikal, appears to be reversed in the intersite area. Apart from the extensive excavations by Green (1970) at Navajuelal, six groups were excavated more or less completely on the South Survey Strip (Strs. SE-211, 212, 213; Se-373, 452: SW-157, Se-382; Sw-158; SW-159, 332: Se-393). Of all of these only the first (Strs. Se-211, 212, 213), at 2.3 km, matches the increment pattern Haviland (1963) found typical of the central nine km2. As these groups fall within what I have defined as "Residential Tikal" this similarity is not surprising.

The other groups, all from beyond 6.5 km, showed evidence of modification and even of the addition of structures in several cases, but within Preclassic and Early Classic times. All five groups showed evidence of Preclassic and Early Classic occupation; not one showed definite evidence of Late Classic occupation despite the comparatively large size of the samples in contrast to the test-pit samples. Of the eight structures in these groups five contained Preclassic ceramics, and all were fairly definitely occupied in Early Classic times. In only one case (Str. SE-373) could even "probably" Late Classic (Ik) ceramics be identified. "Possible" Late Classic ceramic turned up in four others, leaving two without the slightest trace of ceramics from this time period (Strs. Sw-157 and Se-393). This pattern of Early Classic occupation and Late Classic "abandonment" is also reflected, although not so dramatically, in the test-pit data shown in Figures 13, 15, and 17.

In the next series of columns on Appendix 3, the above figures are all reduced by Haviland's figure of 16% , which in a gross fashion allows us to eliminate structures which were "residential adjuncts" and thus not actually lived in (Haviland 1970:93). For the sake of consistency this "correction" is also applied to intersite areas, though the proportion of houses that were dwellings may have been higher.

Of the eight structures in the periphery which were fairly extensively excavated two (16%) are very small and stand near larger structures (Strs. Se-452 and Sw-332). It seems likely that these served as "residential adjuncts" rather than residences' perhaps they are the remains of kitchens. The remaining six structures, while they occasionally show some unusual features, are substantially like what have been considered houses in the central nine km2 of Tikal, in size , shape, and construction. They also show evidence of plastered floors (indicating that they probably were roofed) and associated middens.

The figures thus produced for each time period are then multiplied by 5.4, a calculation of the number of people per house based on Naroll's 1962 floor area technique. The reduction of the structure count by 16%, described in the previous paragraph, is consistent with Naroll's procedure which calls for the measurement of dwelling floor areas only. Cooking and storage areas which are separate from the dwelling are not counted.

Application of Naroll's formula may involve underestimation of occupation density for the ancient Maya. The reason for this is that many important household functions may have been carried out on the plaza which is invariably associated with ancient Maya residences. The utilization of such a feature by this culture might conceivably reduce the roofed floor-space requirement of the average individual. Though we do not have crucial data on floor areas of early post-conquest dwellings of the lowland Maya, ethnohistoric data, as shown above, suggest they were often quite crowded. These data can be misleading, however.


Go Back to Top of Page

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