Methodology

By Geoffrey McCafferty (2008; updated 2013)

This section describes the laboratory and analytical procedures used in the UA-1 pottery analysis. Due to the large number of potsherds recovered, a sampling strategy was developed for selecting a representative subset that could provide the maximum amount of information. The attributes were then defined for the type descriptions and vessel form analysis, and the methodology is described for how these attributes were manipulated to interpret the assemblage.

The UA-1 excavations recovered approximately 1000 bags of potsherds from 671 proveniences defined by unit, level, and feature (Figure 1). On the basis of preliminary sorting, Wolfman (1968:5) estimated the total assemblage at more than 100,000 sherds, but subsequent analysis indicated that the total was at least double that amount. One of the first goals in the analysis, therefore, was to devise a sampling strategy to organize the ceramic assemblage.

Figure  1: Plan of UA-1

Collection units selected for study included primary and secondary contexts suitable for seriation. Thus, analysis focused on excavated levels associated with midden deposits, wells, burials, and floor contact of Structures 1 and 2. Plow-zone and collapsed adobe wall deposits were generally eliminated from the analysis.

A second sampling criterion was the selection of rim sherds for detailed analysis and quantification. Vessel rims are particularly useful for identifying vessel form (Whallon 1968; Smith 1985). Information on form frequencies is important for interpretations of functional variation in ceramic assemblages. Body sherds can also yield information on form, but it is less detailed and more susceptible to bias based on vessel size. In the UA-1 analysis, body sherds were rebagged for possible reconstruction of whole pots, or for future stylistic analysis; unfortunately the sherds from UA-1 were later buried on the UDLA grounds to conserve lab space.

By selecting only rims, it is estimated that the analysis retained approximately 15% of the pottery from the selected contexts, eliminating the 85% that were body sherds (also including bases, supports, and handles). This estimate is based on a sample of total counts made during the initial sorting of rims from body sherds, and is consistent with sherd counts tabulated by Parsons et al. from the Valley of Mexico (1982:Tables 39-55).

Rim sherds were subjected to one final sampling step. During sorting, very small rim sherds were found to be difficult to classify as to form. Even for type designations it was suspected that decorated types, by virtue of having more identifiable characteristics, would be over-represented in relation to undecorated types. To control for this potential bias, rim sherds that measured less than 2 cm on a side (roughly the size of a quarter) were separated out. These were counted as Unidentifiable - too small in the classification. Approximately 25-30% of the rim sherds analyzed fell into the Unidentifiable - too small category.

Quantification of the ceramics was based on sherd counts, but an additional measure was employed for rim sherds from the large trash midden located south of Structure 1. The measurement of degrees of arc (cf. Plog 1985) provided an alternative value that was useful for determining the minimum number of individual vessels for each type and form, and is therefore a means for more accurately estimating the "kitchen tool-kit." Comparisons between the sherd count and degree of arc values provide an opportunity to correct for over-representation of large rim forms as opposed to small rim forms (e.g., comales that break into many pieces in contrast to small-mouth ollas that produce only a few rim sherds from a large vessel).

Another potential of the degree of arc measurement is that it allows an estimate of average sherd size for vessels of similar rim diameter. Assuming that sherd size is affected by the amount of disturbance in a deposit (Bradley and Fulford 1980; Schiffer 1987:267-269), the degree of arc value can provide a means for interpreting the formation processes for particular deposits. Consequently, rim sherds from an area that has been exposed to extensive disturbance (for example, through erosion, trampling, construction, etc.) should have a smaller average degree of arc value than rims from a deposit that was covered quickly and with little such disturbance.

A complementary means for measuring depositional disturbance is the percentage of Unidentifiable - too small sherds from each provenience, where relatively more small rims would be expected in more disturbed levels, or in areas of traffic and regular sweeping. Midden deposits often contain a high frequency of large vessel fragments, while plow-zone levels have a higher frequency of small fragments. For example, the plow-zone deposit (Level 1) of Unit N3/W1 East Balk contained 73% (n=83) sherds in the Unidentifiable --too small category. Considering that very small sherds are often discarded during analysis (Schiffer 1987:269), use as a measure of disturbance and/or site function puts this artifact category to a positive use.

A total of 16,396 rim sherds were analyzed from primary and secondary contexts, usually associated with the two Postclassic structures and related features. Assuming that rims make up about 15% of a typical assemblage, this represents approximately 110,000 potsherds processed. This total represents only a portion of the sherds from the UA-1 site, however, since the unanalyzed plow-zone layers usually contained large quantities of small sherds.

Additionally, 81 complete or reconstructable vessels were recovered at UA-1 (McCafferty 1992a:489-493). Since most of these were missing from the UDLA Archaeology Lab at the time of the analysis, however, they were not included in ceramic tabulations. Over half of the complete or reconstructable vessels (n=47) were found in the large Trash Midden associated with Structure 1, and an additional 13 vessels were discovered beneath the floor of Room 3, Structure 1, perhaps as ritual interments because they were placed near the raised platform altar. Several other complete vessels were found in association with burials.

Ceramic data was coded and entered into a computerized data base, and processed using the SAS statistical program. This information is used to produce frequency tables for the types and subtypes, and for vessel form.