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 Four: The Presentation of Mapping Data


The East Brecha Survey Strip
The South Brecha Survey Strip
The West Brecha Survey Strip
The North Brecha Survey Strip
The Northwest Border Brecha Survey Strip
The Uaxatun Brecha Survey Strip
The Ceramic Survey


The Ceramic Survey
Having devoted considerable discussion to the spatial distribution of settlement, I would like to turn now to the problem of control of these date in time. Temporal control, insofar as it could be carried out, was based on the ceramic survey conducted by Robert Fry (1969). All ceramic evaluations and the test pit survey itself were carried out by him with the assistance of Guatemalan excavators, of Michael Witter, and occasionally of other members of the sustaining Area Project, including myself.
The objectives of the ceramic survey were eventually twofold: 1) obtaining of large samples of sherds from excavations of structures locate beyond  the limits of the nine km2 map, so that comparison could be made of ceramics from "Central" and "Peripheral" Tikal; and 2) the ceramic dating of the construction and occupation of the structures which had been mapped. These ceramic data, on the basis of evidence and assumptions which are described below, are then used to make tentative calculations of population densities for Late Preclassic, Early Classic, Late Classic, and Postclassic times.
Methods and Assumptions
Essentially, what we are doing here is converting a combination of mapping and ceramic data into population estimates. I will now describe the series of logical steps involved and begin by pointing out that raw data comes from two main sources: 1) the surface mapping of structures; and 2) the ceramic test-pit survey. An explanation of assumptions pertaining to mapping data that have been made in order to convert these data to statements of prehistoric demography follows.

The first assumption here is the 84% of the surface-mapped structures on the Survey Strips represent the remains of residences. This assumption is based on: 1) Haviland's assessment of the excavated sample of 117 small structures from the central nine km2 in which 84% were apparently residences; and 2) the similarity of most structures discovered in the course of the present survey, particularly with respect to size and configuration, to those found in the central nine km2. Limited data from excavations which did not form part of the random sample bear out Haviland's figure in a rough way. In the sample of 13 more completely excavated structures on the South Survey Strip [Strs. SE(S)-211,212,213,373, 452, 382,393,409,410, and SW(S)-157, 158, 159, 332] 9 (69%) appear to be residences. Strs. SE(S)-211, 452, 409, and SW (S)-332 on the basis of their very small size and essentially square plan, are likely to be non-residential structures including kitchens, utility structures (Haviland 1963:485) or oratories (Becker 1971:208). If two more of the latter turn out to be residences under more complete analysis the figure will be 84%. It should be realized that this is a very small and non-random sample, oriented to the investigation of anomalous as well as typical situations, so it is unlikely to be representative. A somewhat better ratio appears in the sample provided by all the structures on the strip between 2.0 and 12.0 km. (except the satellite site of Bobal) where 324 structures out of 372 (87%) appear likely to be residences. This sample is roughly equivalent to Haviland's which excluded the epicentral area and major temple and palace complexes outside this area. If there is an error it seem likely to be on the side of caution than otherwise since special function buildings were probably more prevalent in Central Tikal where the more well-to-do may have been concentrated. The higher proportion of specialists of various kinds who probably inhabited the immediate environs of the site center also probably increased the proportion of non-residential structures used as work areas and for the storage of raw materials and finished products. Though much of specialist's work may have been carried out as "house industries" (i.e., weaving, jade carving, mat making, etc.), other kinds of work that took up a lot of space or tended to produce a "mess," such as flint knapping, wood carving, and the manufacture of large ritual objects, probably required more space.

Corozo palms

The second assumption is the by definition the construction of a "residence" implies occupation. The third assumption is the residences were occupied continuously. A full discussion of this assumption will be found in the next chapter under the heading "Arguments for Residential Stability."

Turning now to the ceramic data, I will begin with a description of how the survey was carried out. The first important step was the selection of test pit locations. This process can be separated into two distinct levels; 1) the selection of plaza groups  to be sampled; and 2) the selection of a test pit site within a particular group. Different techniques and assumptions were involved at each of these levels.

East Brecha 1-2 km

The series of steps outlined by Rootenberg  (1964) for constructing and executing a research design using stratified sampling techniques were followed fairly closely at the first level. This procedure, here I will follow Fry (1969:52-56) fairly closely, involves seven stages: 1) Definition of the universe (Rootenberg 1964:183): the portions of the North and South Survey Strips beyond the limits of the already mapped nine km2 of central Tikal were each used as a universe; 2) Division of the site into strata  (Rootenberg 1964:183): "in order to ensure maximal dispersion of tested groups" (Fry 1969:54) each universe was divided into smaller geographic universes or strata. The breaks were selected on the basis of topographic and cultural boundaries. Generally these also coincided with breaks in settlement density; in all, seven strata were defined (Universes N-I, N-II, N-III, S-I, S-II, S-III, and S-IV). Their limits are indicated in Figures 14 and 15. The portions of the strips which are covered by the central nine km2 were not sampled; 3) Division of the strata into potential sampling units, that is, clusters (Rootenberg 1964:183); plaza groups were used as the basic smapling unit. "Whereever two plaza groups were associated, they were separated if one group shared no buildings with the other. A few extra mounds just off the edges of plazas were also counted as part of the unite" (Fry 1969:55). Isolated, single platforms were also treated as unites; 4) Systematic numbering of the clusters and the recording of their locations on the site map (Rootenberg 1964:183): Plaza groups were numbered to carry out this step in the sampling procedure, but because they included groups from both the NW and NE quadrants on the North Strip and the SW and SE quadrants on the South Strip, they were not used again and are not shown on the maps; 5) Estimation of the percent of the total number of clusters in the site that can be excavated, based upon available manpower and time  (Rootenberg 1964:183: this was impossible to do at the outset but, though we initially underestimated our capabilities we were able to add to the sample for the first stratum later; 6) Calculation of the sampling interval  (Rootenberg 1964:183):  this was impossible to do at the outset but, though we initially underestimated our capabilities we were able to add to the sample for the first stratum later; 6) Calculation of the sampling interval  (Rootenberg 1964:183):  a one-third sample was taken for each stratum, that is, one-third of the total number of groups on each stratum were tested; and 7) Systematic selection of the clusters to be excavated in each stratum  (Rootenberg 1964:184): a random sampling technique was used to choose the numbered sampling units (plaza groups) to be excavated. Between 1965 and 1968 numerous other structures were selected and tested on a non-random basis. Because of our specific interest in dealing with a controlled sample, however, these will not be discussed here. They included non-randomly selected test performed by Fry in Square 1D and by myself on the South Brecha Survey Strip.

Figure 12. North Brecha Survey Strip, 0 - 6 km. Results of ceramic test-pit survey with identification listed as: 1) possible; 2) probable; and 3) definite. The parallel hatchure indicates that identification in terms of the minimal time unites Fry generally used (listed at the left) was not possible.


Figure 12 (cont.) North Brecha Survey Strip, 6 - 12 km. Results of ceramic test-pit survey, with identifications listed as: 1) possible; 2) probable; and 3) definite.


Figure 13. South Brecha Survey Strip, 0 - 6 km. Results of ceramic test-pit survey with identifications listed as: 1) possible; 2) probable; and 3) definite.


Figure 13 (cont.) South Brecha Survey Strip, 6 - 12 km. Results of ceramic test-pit survey with identifications listed as: 1) possible; 2) probable; and 3) definite.


The next step was to produce ceramic data as efficiently as possible for the selected groups. Two related aims were involved here: 1) the production of samples that would produce sherds representative of all the time periods for which a group was occupied; and 2) the production of samples that would be large enough and well enough preserved to permit the ceramic identifications needed.

Excavations of South Survey Strip structures in 1965 revealed that structure fills frequently did not produce usable samples of ceramics. To deal with this problem, a testing technique utilizing scissor-bladed, vertical-thrust, post-hole diggers was employed. This technique was devised by the author while he was carrying out the comparatively fruitless test-pitting surveys of 1965-66 (Puleston 1967). It was tried out in the field for the first time in 1966 with great success and used to search for "hidden" house mounds and trails as well as aid in test-pitting. As a system for locating sherd concentrations around house ruins, it was perfected by Fry and has since been described by him in detail (Fry 1972). As used in the ceramic survey, the technique involved "testing" likely situations along the backs and off the ends of structures where heavy scatters of occupational trash including middens(i.e., piles of such debris with nesting sherds, carbon, etc.) were likely to occur. Often, no sherds would be produced, but if the hole was located in the vicinity of a sizable concentration of ceramics, or even a midden, as many as twenty sherds might be produced. This testing, which generally required no more that four or five post-holes, was a task of a few minutes. These tests were generally located on the edges of the mounds so that the test-hole also penetrated what was fairly clearly collapsed debris from the platform. When a productive hole was located, a test-pit  one to two meters square was placed over it in such a way that collapsed platform fill, as well as the deeper trash deposit were excavated. This procedure greatly increased our chances of picking up earlier ceramic which might have been scraped up and used as fill. This means of locating a test-pit within a group obviously was not random, a fact which increased our margin of error in making paleodemographic calculations for Early Classic and Late Preclassic regional periods. This problem will be discussed in the section on the population of Early Classic Tikal in Chapter VII.

Once the ceramics had been excavated, bagged washed in camp, and rebagged, they were examined by Fry and identified according to the Tikal ceramic complex then represented, or if that was not possible, the major regional period. These identifications were listed in Fry's field notes and on the lot cards, describing the assigned excavation units for each test-pit. In some cases Fry  later re-examined these ceramics in order to check identifications made earlier in the season. Finally this material was brought together in the Philadelphia Tikal Room by Olga Puleston, who typed up the ceramic evaluations for all the suboperations and lots excavated at Tikal (only Operation 20 was excluded) which have provided the basis for data presented here.

Clearly a whole series of assumptions are involved here. At the lowest level we had to assume that Fry's sherds were correctly bagged and labeled. We had to assume that the sherds did not become mixed or separated from their labels during the washing and drying process. After the sherds had been rebagged, they were ready for identification at which time another series of assumptions came into play. We had to assume Fry's identifications could be trusted. His several years of experience with Tikal ceramics while he worked under Patrick Culbert, principal ceramist for Tikal, reassured us in this regard. One weakness which enters here, however, is that ceramic assessments were quantified only in relative terms, not by sherd counts or weights. For example, Op. 131X, 6X3 was reported as follows: Medium lot, Rims and bodies. Tepeu, mainly  Imix, some probable Ik, possible Eznab. This means that in some cases the identification of an occupation must be based on a minimal but undetermined number of sherds. Despite this problem it is assumed that all sherd samples were representative of occupation and that incidental or otherwise non-representative sherds were encountered so rarely as to be insignificant to the overall results of the survey. This assumption was confirmed in the four test-pitted groups where more extensive excavation was carried as described in the last paragraph of this section. We also had to assume that the identifications were made consistently. Here it was important that all  identifications be made by Fry. In spite of this precaution, there is evidence to suggest that he was slightly more cautious on the assignment of "definite" and "probable" identifications on material excavated during early work on the South Survey Strip than later. The effect of this caution may be noted in Table I where the disproportionately low values for definite (d) and definite and probable (dp) identifications for Stratum I on the South Strip may be contrasted to the values for Stratum I on the North Strip. This is particularly noticeable in the comparison of the Early Classic, Imix, and Late Classic categories. This uncertainty appears to have dissipated by the time he got into Stratum II of the South Strip, and thereafter the disparity between the percentages of (d) and (dpp) are considerably smaller.

Figures 14 through 17 are presented as a means of revealing the small difference in the proportional representation of different time periods whether or not "possible" identifications are included. In fact, I suspect that inclusion of the "possible" identifications brings us to a closer approximation of reality than its exclusion. Note, in comparing Figures 15 and 17, that for Stratum S-I there is a higher number of terminal Late Classic occupations than there are for the rest of Late Classic times until the "possible" ceramic identifications are included. The unlikely result shown in Figure 15 is more a reflection of Fry's caution than ceramic reality, and, for this reason, I feel that the risk of misrepresenting the reality of the situation is lower if the possible identifications are included.

Figure 14. Graphic presentation of the proportional representation of occupations for different time periods on each major stratum, based on a combination of definite and probable identifications from the ceramic survey. See the text for an explanation of the purpose of these figures. Percentages are of total "occupations" for each of Fry's strata (the figures in parentheses give the actual count of plaza groups presumably occupied). The areas of the blocks are proportional to these figures. A fine breakdown of these data in terms of 0.25 km2 block structure counts is presented in Figure 24.


Figure 15. South Brecha Survey Strip. Graphic presentation of definite and probable identifications of "occupations" from the ceramic survey. The caption of Figure 14 has further information.


Figure 16. North Brecha Survey Strip. Graphic presentation of  definite, probable, and possible identifications of "occupations" from the ceramic survey. The caption of Figure 14 has further information.


Figure 17. South Brecha Survey Strip. Graphic presentation of definite, probable and possible  identifications of "occupations" from the ceramic survey. The caption of Figure14 has further information.


My next and final assumption is the Fry's notes were correctly transcribed into the ceramic evaluation books I used. Where I have been able to check them, they appear to have been carefully done.

Having dealt with the mapping and ceramic data separately, I would like to discuss their interrelationship in the assumption of occupation. The assumption is the sherds of (a) ceramic complex(es) found in or near a structure are to be taken as evidence of its continuous occupation during the time period(s) when the pottery of the ceramic complex(es) was in use. The assumption applies to sherds found in fill contexts as well as primary occupation debris.

The assumption that ceramics from fill contexts are of local origin is based on a further assumption which is supported, thought not proved, by a series of observations. A corollary assumption is the trash containing pottery fragments was not generally transported as fill from one plaza group to another. The supporting observations are

1) the spacing of plaza groups at some distance from one another, which reduces the likelihood of incidental fill transport;

2) the general availability of soil and limestone, which would have made it unnecessary to go more than a few meters for construction material unless something special was desired (e.g., the black, silty clays, presumably brought up from bajos, which are occasionally found in platform fills);

3) the relatively frequent occurrence in the vicinity of plaza groups of pits and quarries from which considerable quantities of soft limestone, as well as masonry were extracted:

4) the evidence summarized in Chapter V for the permanence of residential occupation, which means that opportunities for the "piracy" of fill materials from other kitchen gardens or structures would rarely arise;

5) the frequent discovery of accumulated ceramic trash behind and off the ends of Late Classic residential structures (Haviland 1963:505), which leads to the assumption that such accumulations also occurred in earlier periods but was occasionally taken up for use as fill;

6) the presence of sherds, stone tools, stone toll fragments, and carbon in platform fills, which are essentially identical to the trash resulting from occupation found in the immediate vicinity of later structures and which essentially confirms that presumed earlier trash deposits were used in platform fills.

Observations of this sort made it plain that complete representation of all occupation periods of a particular group would be more likely if platform fills as well as middens could be sampled. The further observation that the backs of structure platforms collapse outwards, allowing fill materials to spill out over the middens and occupation deposits behind structures suggested to us that test-pits located directly behind structures could provide samples of both kinds of deposit. This approach, of course, worked best when the collapsed fill really did overlie a midden or heavy occupation , and it was with the intention of locating such deposits that the post-hole diggers were used. The technique appeared to be a good one and as might be expected numerous examples of reverse stratigraphy were found. Examples include the test-pits of Suboperations 131 V, 131 Y, 132 B, 132 X, 134 D, 136 D, 136 F, 137 G, 137 R, 137 Y, and 146 F.

In each case the assessment of occupation history indicated by the test-pit results was confirmed by the more extensive excavations. When the more extensive excavations produced sherds not represented in the collections from the test-pits, these were always in quantities that were so small they could not be considered evidence of occupation.

Ops. 138 A, B, C, D, E, which involved thorough evacuation of the group represented by the Op. 134 E test-pit, produced a single Caban scroll foot. No other evidence of Postclassic material was found elsewhere in the excavation or test-pit so it is doubtful that this is representative of Postclassic occupation. Apart from this item the test-pit results were entirely representative of the Ops. 138 A, B. C, D, E material.

In  Ops. 138 I and J which was a follow-up on the Op. 132 P test-pit, a single "very small lot" produced a "Cauac-Cimi" sherd (s) which was not represented in the test-pit material. Again this was too small a quantity in view of the size of the excavation to be considered evidence of occupation.

Finally in Ops. 138 F and H, out of a total of 124 lots (many of them quite large), two to five Preclassic and one probable and four possible "Tepeu" sherds were found. Neither Preclassic nor Late Classic material was produced by the test-pit and I assert that it would have been misleading if they had since they certainly cannot be taken as evidence of occupation. In light of the latter point the possibility does exist that in some case, or cases, a test-pit did come down on a few sherds which were otherwise unique to a particular group and therefore not representative of occupation. The very small quantities involved and the many cubic meters of more extensive excavation it took to produce them, however, suggest that the chances of such an occurrence are quite small. The likelihood that they could affect the overall results of the survey would seem to be even less. Where they could be tested then the test-pits appeared to be excellent indicators of the occupation evidence produced by larger excavations.


As shown in Figures 12 and 13, a total of 90 plaza groups were investigated by the means described above. A total of 43 groups were tested on the South Survey Strip and 47 on the North Survey Strip. The data in the figures is presented so that Fry's identifications are expressed in terms of his designations: "possible," "probable," and "definite." Groups incidentally excavated before and after the random sample was selected are not included in the calculations because of the bias inherent in their selection. In the case of the five test-pits of this category in and just north of Square 1D (Ops. 146 G, H, J, K, L), Fry tended to select the largest and most prominent groups or structures for excavation. Three of them )Ops. 146 G, H, J) were of interest because of their rather atypical size, shape, and lack of obvious association with a plaza. In comparison to the randomly selected sample of ten groups (Ops. 136 A through 136 K in Figure12) from the same area, the five non-randomly selected sites showed poor representation of Preclassic (only one) and Terminal Classic (only two) occupation. The two samples were essentially identical with respect to Early and Late Classic representations. All the non-randomly selected sites produced definite Early Classic ceramics as did the randomly selected ones. Both groups had only one site which did not produce Late Classic material.

Map of the whole of the Tikal Park including the northern defensive earthworks and the survey strips (brecha maps)

The following conclusions are suggested for the areas covered by the Survey:

1) Middle Preclassic occupation site are a) rare, b) widely scattered in peripheral areas, and c) located on higher, well-drained ground.

2) Late Preclassic occupation sites a) may be as much a seven times as frequent as Middle Preclassic occupation site, b) are more common on high ground, and C0 are within five km of the center of Tikal.

3) Early Classic occupation sites a) may be more than three times as frequent as Late Preclassic sites, b) occur on low and less well drained ground as well as in high areas, and c) cluster within five to six km of the center of Tikal and near the satellite sites, Jimbal and Navajuelal.

4) Late Classic occupation sites (excluding those of the Eznab/Imix transition and Eznab) follow the pattern set by Early Classic inhabitants of the Tikal area, with the qualification that certain occupation sites in the intersite area apparently were abandoned. On the North Survey Strip this pattern does not appear to have taken full effect until Imix ceramics started to be used. Abandonment in the vicinity of the satellite sites, Jimbal and Navajuelal, is less marked than in other portions of the Tikal Intersite Area.

5) The time period represented by the utilization of Eznab/Imix transition and Eznab ceramics witnesses a dramatic decline in occupation sites with near abandonment of intersite areas. It would be tempting to suggest a reversion to the Late Preclassic patterns but two related differences forestall this comparison: 1) Late Preclassic occupations are more dispersed (note particularly the 0-6 km section of the North Survey Strips); and 2) satellite sites during the latter period continue to be focal points of occupation, particularly Jimbal at 11-12 km on the North Survey Strip.

6) Postclassic occupation sites are most noteworthy for their scarcity. The two Caban sites discovered, one definite and one possible, were both found on the South Survey Strip near the Laguna Verde Reservoir. Unlike even the scarce Middle Preclassic sites, they do not seem to be concentrated on high ridges, even when we take into account the loci of Postclassic ceramics found within the central nine km2 of Tikal. In both "peripheral" and "central" situations the proximity to water seems a more significant influence on location than drainage and elevation.

Go Back to Top of Page

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