Integrating Functional Analyses of Vessels and Sherds through Models of Ceramic Assemblage Formation
Author(s): Barbara J. Mills
Source: World Archaeology, Vol. 21, No. 1, Ceramic Technology (Jun., 1989), pp. 133-147
Kết hợp những phân tích chức năng của đồ gốm nguyên với mảnh vỡ bằng cách sử dụng những kiểu mẫu hình thành sưu tập gốm
Nghiên cứu hiện nay về đồ gốm khảo cổ học sử dụng cách tiếp cận chức năng được khắc họa có hai mục đích phân tích chính: 1. Nhận dạng vai trò của đồ gốm trong hệ thống kiếm sống và 2. Nhận dạng những sự khác nhau trong các kiểu làng cư trú. Phần lớn các nghiên cứu được dẫn dắt để đạt được mỗi mục đích, nhưng sự liên kết của hai cách tiếp cận thì rất khó vì những kiểu hình thành sưu tập đồ gốm khó xác định. Như một bước tiến tới sự liên kết này, kiểu hình thành sưu tập đồ gốm thể hiện những nét chính của những khác biệt lớn để đánh giá sự khác nhau trong nội dung sưu tập. Sử dụng những tư liệu khảo cổ học dân tộc như thông số đầu vào, sự hình thành sưu tập qua thời gian được tạo hình, và những tác động của những kiểu khác nhau lên nội dung sưu tập sẽ được thảo luận trong bài.
In the last two decades, functional analyses of ceramic artifacts have burgeoned. In addition to providing valuable data on chronology arid systems of production and distribution, ceramic artifacts are now being used to address new questions, such as why do land use systems vary and how can this variation be identified. As a result of this shift in the goals of archaeological research, we are now seeing a shift in the way archaeological remains are classified and interpreted (Binford and Sabloff 1982; Dunnell 1986).
Despite this, relatively little attention has been paid to the development of models of ceramic assemblage formation. This has led to a situation where functional analyses of whole vessels and sherd assemblages are often pursued separately. While the goal of analyses of vessels has been to derive functional types that reflect the use of ceramics in the subsistence system, the goal of assemblage approaches has been to test differences in settlement patterns (Table 1). Advances in each approach over the last decade have
Table 1 Contrasting approaches to the analysis of ceramic function.
been numerous, but there has been relatively little integration of the two. Both approaches are needed to be able to interpret archaeological ceramics. Models of ceramic assemblage formation can provide a framework for this integration.
As a step toward the reintegration of analyses of sherds and vessels, this paper presents a basic model of ceramic assemblage formation. Some of the factors contributing to variation in ceramic assemblages are explored. Since many identifications of ceramic function - whether of vessels or of sherds - are based on technological variables, the study of ceramic technology holds an important role in meshing classificatory and interpretive goals.
Synthesis* a model of ceramic assemblage formation
One way to develop expectations of how vessel classes should vary among sites of different functions is through the development of models of ceramic assemblage formation. These models allow the effects of the interaction of both cultural and natural processes on assemblages to be better delineated. Ideally, models of ceramic assemblage formation must include consideration of both the systems that used and deposited artifacts, as well as the interplay of natural and cultural post-depositional processes acting upon the archaeological record. Here, cultural formation processess are addressed, leaving the consideration of natural formation processes for future investigation.
Several variables may be targeted for their potentially significant impact upon the differential representation of functional classes of ceramic vessels in archaeological assemblages (Table 2). These include: (1) the size of ceramic assemblages in systemic context, (2) the frequency distribution of vessels in each functional class, (3) ceramic breakage rates, (4) site abandonment rates, (5) vessel curation rates, (6) vessel replacement rates, (7) stages of the domestic cycle, and (8) patterns of site reoccupation. These variables have been identified from their use in other models of assemblage formation, including those addressing both ceramic and nonceramic artifacts (Aldenderfer 1981; Ammerman and Feldman 1974; Hatch et aL 1982; Kohler 1978), as well as from the increasingly available ethnographic and ethnoarchaeological literature on ceramics in their behavioral contexts (e.g., Arnold 1985; Birmingham 1975; David 1971, 1972; David and Hennig 1972; Deal 1983; DeBoer 1974, 1983, 1985; DeBoer and Lathrap 1979; Foster 1960; Kramer 1985; Kroster 1974; Longacre 1985; Miller 1985; Nelson 1981, 1985; Nicholson and Patterson 1985; Pastron 1974; Weigand 1969). The present model is used to understand how variability in pottery production, use and discard through time interact to produce differences in the representation of functional classes of ceramics in the archaeological record. It differs from other recent models that have assessed the effects of these variables on estimates of human population size (Kohler 1978), or the chronological sedation of deposits (de Barros 1982; Hatch et aL 1982).
In this article, the interaction of two of the above enumerated variables is modeled: vessel breakage rates (or vessel uselife estimates), and site abandonment rates (also described in the literature as duration of site occupation or occupation span). Vessel
Table 2 Major factors in model of ceramic assemblage formation.
breakage rates and duration of site occupation are the focus of this article because these two variables have been suggested to have important effects on assemblage content. Comparison of effects will be made by discussing the differences that the interplay of these two variables produce in proportions of the general ceramic vessel classes of cooking, serving, and storage. Three slightly different formulae that have been used to calculate vessel breakage rates are discussed (David 1972; de Barros 1982; Schiffer 1975a, 1975b). Variation among these formulae provides a springboard to understanding how variables within the model affect ceramic assemblage content.
Vessel uselife and duration of site occupation
The two variables of ceramic vessel uselife and duration of occupation have been linked in the archaeological literature for nearly thirty years. Baumhoff and Heizer (1959: 308) noted that if one calculated the number of vessels deposited by a single household and if one knew the duration of site occupation, then the number of vessels broken in one year could be determined. Foster (1960: 606) reversed and expanded this equation by suggesting that if one knew breakage rates and the number of vessels per household, then both duration of site occupation and population size could be predicted. Cook (1972) used Foster's estimates of breakage rates and the number of vessels per household along with estimates of duration of occupation from archaeological site reports to estimate its population.
The effects of variable site occupation spans (or duration of occupation) on
assemblage content were modelled by Schiffer (1975b) using the formula shown in Table 3A (see also Schiffer 1975a: 840, 1976: 59-62, 1987: 53-4). He used this formula to discuss possible implications that artifact uselife and duration of occupation may have on assemble content. One of his conclusions was that varying type frequencies could result from sites that were used in similar ways, but for different durations. He called this the "Clarke Effect', since the potential presence of varying type frequencies resulting from the same activities had first been noted by David Clarke in 1972 (Clarke 1972, discussed in Schiffer 1975b: 265). Unlike Clarke's stochastic model of variation, however, Schiffer's model suggested that varying type frequencies may also be produced through deterministic relationships between the variables of artifact uselife and duration of occupation. Given the absence of data on assemblage size and artifact uselife in ethnographic contexts, the combined effect of uselife and duration of occupation was discussed by Schiffer in hypothetical terms, presented as a research design for future investigation.
Using ethnographic data on the Fulani, David (1972) was the first to calculate the absolute effects of ceramic vessel uselife data and duration of site occupation (see Table 3B). His results indicated that the longer the duration of occupation, the greater the differences in the representation of different functional classes in the starting (or systemic) vs archaeological assemblages. David attributed these, differences to the interaction of vessel class uselives and duration of occupation (David 1972: 141; David and Hennig 1972: 20). Using David's formula and ethnographic data from the Shipibo-Conibo, DeBoer (1974: Table 1) noted smaller deviations of the simulated archaeological assemblage from the ethnographically documented starting assemblage.
De Barros (1982: 310), in his simulation of the possible effects of variable site occupation span on seriation results, has concluded that only in cases where the duration of occupation is short and vessel types of different functions have widely variable uselives, will the relationship between these two variables pose a significant problem for ceramic seriation. De Barros thought this situation would rarely arise.
But how short is short and how variable are vessel uselives? This question is particularly relevant to understanding the settlement systems of small-scale agriculturalists. In these cases, occupational histories may have been along relatively small temporal scales. Current archaeological research in many areas of the world, such as the American Southwest, suggest that estimates of duration of occupation based on ceramic type production spans are probably too high (Cordell 1981: 127). Models that consider the systemic effects of variation in duration of occupation are therefore needed and would be widely applicable.
The relationship between vessel uselife and duration of site occupation is more easily addressed now than it was a decade or two ago because of the greater availability of ethnographic data on ceramic uselives. Summary data on vessel uselives are available from several different groups (Table 4) as are the number of vessels in use by functional class (Table 5). For comparative purposes, the vessels used by each ethnographic group were placed into four classes: cooking, serving, storage, and other
(cf. Rice 1987: Table 9.4 for a compilation using slightly different vessel classes). If we consider these data as the range of potential uselife variability, the group with the longest uselife estimates and the group with the shortest provide a way of bracketing possible high and low' values. As Table 4 indicates, the lowest values of vessel uselife have been documented for the Shipibo-Conibo and the highest for the Kalinga. These data provide a means of not only assessing the differences between long and short uselife data, but also for assessing the differences among the formulae presented in the Figure I Plot of frequency of vessel in functional classes by year, using Schiffer's formula and data from the Shipibo-Conibo.
published literature for calculating the frequencies of vessels broken through time.
Calculation of the simulated assemblages at one year intervals was first made using Schiffer's formula (Table 3A) and the starting assemblage size and uselife data reported for the Shipibo-Conibo by DeBoer (1974; DeBoer and Lathrap 1979) and the Dangtalan Kalinga by Longacre (1985; 1979-80 sample). To illustrate how this calculation was made, we may turn to the Shipibo-Conibo data shown in Tables 4 and 5. For this example, we use a mean number of vessels of each functional type per household as: 3.7 cooking vessels, 4.7 serving vessels, 2.8 storage vessels and 2.0 'other' vessels. Since these types have uselives of 1.1, 0.36, 1.5 and 0.75 years, respectively (note that the figure of 0.36 is the median of 0.25 and 0.47 years), the following frequencies of vessels deposited into the archaeological record after 1 year may be calculated as:
No. of cooking vessels - 3.7 * 1 / 1.10 - 3.36
No. of serving vessels - 4.7 * 1 / 0.36 - 13.06
No. of storage vessels = 2.8 * 1 / 1.50 - 1.87
No. of 'other' vessels - 2.0 * 1 / 0.75 = 2.67
For the Shipibo-Conibo example, the frequencies of each vessel type for each subsequent year may be arrived at by substituting the '1' in the above example with the number of years passed in the formation of the simulated assemblage.
Considering first the results from using the short uselife data documented for the Shipibo-Conibo, it is evident that the change in the frequencies of vessels by functional class is linear (Fig.l). Therefore, the relative frequencies of functional classes are Figure 2 Plot of relative frequencies of functional classes as proportion of total assemblage by year, using Schiffer formula and data from the Shipibo-Combo.
constant, and do not change through time, as Figure 2 indicates. The plots for the Kalinga (not illustrated) are identical to those of the Shipibo-Conibo in shape and differ only in the scale of differences between the values for each vessel class.
If the relationship between vessel class frequencies and duration of occupation is basically linear, as the above suggests, why has there been so much discussion of the possible effects? Consideration of the formula presented by David (Table 3B) indicates that there are two differences between it and the Schiffer formula, one of which has an important effect and one that does not. The first and most significant difference is that the entire starting assemblage (represented in the formula by the first 'N0') was added by David to the final calculation of the total number of vessels. This assumes that all of the pots in use at the time the hypothetical site was abandoned were left behind. The results of calculations of the number of vessels deposited in the archaeological record through time using David's formula indicate that the relationship is still roughly linear (Fig.3). The greatest deviation from a straight line is during the first few years, after which the values become stable. Stability in the relative proportions of the different vessel classes appears slightly later for the long uselife Kalinga data than for the Shipibo-Conibo data, but the magnitude of the differences in proportions through time is greater (and more significant) for the Shipibo-Conibo data. The greater magnitude of difference in the Shipibo-Conibo curves is because the systemic assemblage size used by each Shipibo-Conibo household is, on the average, larger than that of Kalinga households. Thus, David's (1972) observation that the representation of vessel classes becomes more disparate through time is true for frequency data only, and the obverse is true for the proportional representation of vessel classes in the assemblage using his Figure 3 Plot of relative frequencies of functional classes as proportion of total assemblage by year, using David's formula and data from the Shipibo-Conibo.
formula; the longer a site is occupied, the more stable the relative frequencies become.
The second difference between the David and Schiffer formulae is indicated by the 'NQ/2' portion of David's formula. The starting assemblages size is divided by two in David's formula because he uses the median uselife for each class (use of the mean would require the same calculation). The effect of this term in the David formula is on the resulting assemblage size only, and the relationship between the artifact frequencies and duration of occupation remains linear. Schiffer (1975b) explicitly assumed a standard deviation of 0 and strictly speaking, he did not need to include an associated probability value. This portion of David's formula is more realistic; as the ethnoarchaeological literature indicates, use-life estimates for a given vessel class and ethnographic group may be highly variable.
A modification of David's formula for calculating the number of pots entering the archaeological record has been used by de Barros (see Table 3C). The difference between the two is that de Barros added one-half rather than the total starting assemblage size to the total number of pots broken. His reason for adding only one-half of the starting assemblage was that it seemed less likely that all of the yet unbroken vessels would be carried away from a site when the site was abandoned (de Barros 1982: 310). The results of calculations using the de Barros formula produces even less of a curve during short durations than the David formula, and therefore the proportions of each vessel class stabilize more quickly. For the short uselife Shipibo-Conibo data, this point of stabilization is reached at approximately four to five years, while the long uselife Kalinga data do not reach this stability until about the tenth year. In both cases, the point of stabilization in vessel class proportions is not reached until approximately a year or two longer than the greatest uselife present in the assemblage.
Any model of assemblage formation must include some way of calculating the number of an artifact class entering the archaeological record. Here, various techniques that have been proposed in the literature have been illustrated, with varying results. Several points may be summarized. First, in its most essential form (Table 3A), the formula for calculating the combined effects of uselife and duration of site occupation produces: (1) linear increases in the frequencies of vessels by class, and (2) no differences in the relative frequencies of vessel classes.
Second, formulae that add the starting assemblage or some fraction thereof (e.g. Table 3B and 3C), will affect both frequency and proportional results, but the latter will eventually stabilize as the starting assemblage size becomes a smaller proportion of the total number of vessels hypothetically entering the archaeological record (DeBoer 1974). The greatest deviations from constant proportions in assemblages are found when the number of pots per household is high and when the proportion of whole pots left behind (i.e. as de facto refuse) is also high. Therefore, the nature of site abandonment, including the rate of off site transport, is an important consideration, but will have a greater effect on sites of short duration than those occupied for longer lengths of time.
The rates of off site transport used in the published formulae of David and de Barros make one critical assumption: that vessels of one class are equally likely to be taken away at the time of site abandonment as vessels of another class. Ethnographic evidence suggests that transport probability of ceramics' may be directly related to weight (DeBoer 1985: 354). Since weight and uselife are also directly related, it seems more likely that all other things being equal, vessels of functional classes with long uselives will be left behind as de facto refuse (see Schiffer 1987: Figure 4.7 for a dramatic illustration of this point). The effect of this variable on assemblage formation will be greater variation in the proportions of each vessel class and a slightly longer period until they stabilize. All other things are rarely equal, however, and in the case of ceramic vessels the amount of remaining uselife and replacement costs have both been suggested to be important variables in the probability of offsite transport (Schiffer 1985: 33-4)
Thus, the variables of assemblage size and rates of curation may combine with the variables of uselife and duration of occupation to have the effect of delaying the point when the stabilization in vessel class proportions occurs. From the ethnoarchaeologicai data examined here, stabilization in vessel class proportions cannot be expected to be reached much sooner than 3-4 years, and could take as long as ten years. It may be concluded from these results that given initial starting assemblages of the same size and type frequencies, sites occupied less than ten years have a reasonable chance of displaying differences in assemblage content, irrespective of differences in site function, and that sites occupied for less than three years will almost _ always show these differences. Conversely, sites occupied for more than ten years, whether for 20 or 100, should theoretically have the same proportions.
It may therefore be reasonable to expect that short duration sites as a group will display low variety (i.e. few numbers of classes) and high variance (i.e. high variability around the mean) in the relative frequencies of vessel classes. In contrast, long occupied sites should display high variety in the number of different classes and relatively stable proportions of each class, even if these sites were used in exactly the same way as sites used for short durations. We may further predict that sites composed of multiple short-term occupations (e.g. three occupations of three years each) will show higher variability than one long occupation (e.g. a single occupation of nine years) because the point of stabilization in proportions may not have been reached in any single occupation. Frequency data from such reoccupied sites would suggest that they were used for long durations, when in fact they are the result of a very different settlement-subsistence system.
Sites occupied for less than ten years are not unreasonable to expect in the archaeological record. Short duration sites may often be found in situations where agriculturalists practise a system of frequent field movement associated with a change in residential site location. These kinds of sites should therefore be expected and models to identify them further developed. If an analysis proceeds by first classifying sites by the presence/absence of architecture or the kinds of features which are present, as many analyses of sherd assemblages do, then short duration residential sites will usually be grouped with long duration residential sites. Instead, if patterning in ceramic assemblages is first assessed, and then compared to other aspects of assemblage and feature content, assemblages produced from short and long duration residential sites are more likely to be separated.
Recent analyses that have assessed assemblage content independent of other typologies have begun to find distinct differences among assemblages associated with architectural sites (Mills 1986; Reid 1982; Whittlesey and Reid 1982). For example, in an analysis of Anasazi ceramics from 562 sites in and around Chaco Canyon, architectural sites with formal trash mounds were found to have narrower ranges, and lower coefficients of variation in proportions of functional attributes than architectural sites without these features (Mills 1986). Yet, these architectural sites with formal trash mounds varied widely in the number of rooms and assemblage size. Since formalization of trash disposal locations might be expected to be the result of longer durations of occupation than sites with no formal trash areas, the narrower range of variation in the relative frequencies of functional classes of ceramics within trash mounds appears to fit with the theoretical expectations for assemblage patterning discussed above. In many areas of the world, the relative frequencies of vessel classes in trash mounds may be the most secure method of identifying what the constant values of the relative frequencies of use classes should be if all other variables are held constant. This assemblage type may then be treated as a baseline and deviations from it investigated.
Critical to assessing patterning in assemblages, however, is the use of functional types that are sensitive to differences in vessel uselife. Several researchers have identified specific ceramic variables that covary with uselife in ethnographic contexts. These include variables of production, use, and reuse, as well as some specific technological properties (Table 6). Among the technological properties are various measures of size, including DeBoer's (1985) quantitative demonstration based on ethnoarchaeological data that vessel weight varies directly with vessel uselife.
Table 6 Correlation of uselife with other factor s of ceramic production, use and technology.
The example discussed here of how uselife and duration, of site occupation interact with each other and with other variables of assemblage formation is meant to illustrate the importance of designing classifications to fit interpretive goals. Further ethnographic data on variables such as curation rates, rates of replacement, and the stage in the life cycle of the domestic group need to be collected and added to the model before differences among assemblages can be attributed solely to differences in the range of activities conducted in a particular location,
In sum, a major question to be asked at this point in functional analyses of ceramics is not only how specifically we can make these identifications (Hally 1986: 291), but also, how these identifications are going to be used for interpretive purposes. In the beginning of this article it was suggested that the interpretive goals of functional analyses of vessels vs those of sherds were becoming dichotomous, rather than complementary. Analyses of ceramic vessel function have been oriented toward discovering finer and finer relationships between subsets of the subsistence system and their correlative vessel properties. In contrast, inter assemblage analyses of sherds have been approached as a means of confirming differences between sites, differences already isolated on the basis of nonceramic criteria, and phrased in terms of settlement variability. This interpretive dichotomy will remain until the formation of the archaeological record is considered in greater detail. By considering the dynamics of assemblage formation, more realistic expectations for how archaeological assemblages should be patterned can be developed and tested against the archaeological record.
Many of the ideas on assemblage formation presented in this paper were formulated through discussions with Chad McDaniei, who greatly improved the logic of my thinking on the subject. Philip J. Arnold III, T. J. Ferguson, Robert D. Leonard, Ben A. Nelson, and Michael Schiffer provided helpful comments and suggestions on earlier versions of this paper; one of which was delivered at the 52nd Annual Meeting of the Society for American Archaeology, in Toronto, Canada. Timothy J. Seaman provided valuable assistance in the production of the figures. Research conducted for this paper was made possible by a grant from the National Science Foundation (Dissertation Improvement Grant No. BNS-8607917), and a Challenge Assistantship from the Office of Graduate Studies, University of New Mexico.
16.x.1988 Department of Anthropology
University of New Mexico Albuquerque, NM 87131
Aldenderfer, Mark S. 1981. Creating assemblages by computer simulation: the development and use of ABSIM. In Simulations in Archaeology (ed. J. A. Sabloff). Albuquerque: University of New Mexico Press, pp. 67-118.
Ammerman, Albert J. and Feldman, Marcus W. 1974. On the 'making' of an assemblage of stone tools. American Antiquity, 39: 610-16.
Arnold, Dean E. 1985. Ceramic Theory and Cultural Process. Cambridge: Cambridge University Press.
Baumhoff, Martin A. and Heizer, Robert F. 1959. Some unexplored possibilities in ceramic analysis. American Antiquity, 15: 308-16.
Binford, Lewis R. and Sabloff, Jeremy A. 1982. Paradigms, systematics, and archaeology. Journal of Anthropological Research, 38: 137-53.
Birmingham, Judy. 1975. Traditional potters of Kathmandu Valley: an ethnoarchaeological study. Man (N.S.) 10: 370-86.
Clarke, David L. 1972. Models and paradigms in contemporary archaeology. In Models in Archaeology (ed. D. L. Clarke). London: Methuen, pp. 1-60.
Cook, Sherburne F. 1972. Can pottery residues be used as an index to population? Contributions of the University of California Archaeological Research Facility, 14: 17-39.
Cor dell, Linda S. 1981. The Wetherill Mesa simulation: a retrospective. In Simulations in Archaeology (ed. J. A. Sabloff). Albuquerque: University of New Mexico Press, pp. 119-42.
David, Nicholas. 1971. The Fulani compound and the archaeologist. World Archaeology, 3(2): 111-31.
146 Barbara J. Mills
David, Nicholas. 1972. On the life span of pottery, type frequencies, and archaeological inference. American Antiquity, 37: 141-2.
David, Nicholas and Hennig, Hilke. 1972. The ethnography of pottery: a Fulani case seen in archaeological perspective. Addison Wesley Module, 21.
Deal, Michael. 1983. Pottery Ethnoarchaeology Among the Tzeltal Maya. Doctoral dissertation, Department of Archaeology, Simon Fraser University.
de Barros, Philip L. F. 1982. The effects of variable site occupation span on the results of frequency sedation. American Antiquity, 47: 291-315.
DeBoer, Warren R. 1974. Ceramic longevity and archaeological interpretation. American Antiquity, 39: 335-43.
DeBoer, Warren R. 1983. The archaeological record as preserved death assemblage. In Archaeological Hammers and Theories (eds A. Keene and J. Moore). New York: Academic Press, pp. 19-36.
DeBoer, Warren R. 1985. Pots and pans do not speak, nor do they lie: the case for occasional reductionism. In Decoding Prehistoric Ceramics (ed. B. A. Nelson). Carbondale: Southern Illinois University Press, pp. 347-57.
DeBoer, Warren R. and Lathrap, Donald W. 1979. The making and breaking of Shipibo-Conibo ceramics. In Ethnoarchaeology: Implications of Ethnography for Archaeology (ed. C. Kramer). New York: Columbia University Press, pp. 102-38.
Dunnell, Robert. 1986. Methodological issues in Americanist artifact classification. In Advances in Archaeological Method and Theory, vol. 9, (ed. M. B. Schiffer). New York: Academic Press, pp. 149-207.
Foster, George M. 1960. Life expectancy of utilitarian pottery in Tzintzuntzan, Michoacan, Mexico. American Antiquity, 25: 606-9.
Hally, David J. 1986. The identification of vessel function: a case study from Northwest Georgia. American Antiquity, 51: 267-95.
Hatch, James W., Whittington, Stephen L. and Dyke, Bennett. 1982. A simulation approach to the measurement of change in ceramic frequency seriation. North American Archaeologist, 3:
Kohler, Timothy A. 1978. Ceramic breakage rate simulation: population size and the Southeastern Chiefdom. Newsletter of Computer Archaeology, 14: 1-18.
Kramer, Carol. 1985. Ceramic ethnoarchaeology. Annual Review of Anthropology, 14: 77-102.
Kroster, Paula Homberger. 1974. Country potters of Veracruz, Mexico: technological survivals and culture change. In Ethnoarchaeology, Archaeological Survey Monograph IV (ed. C. B. Donnan and C. W. Clewlow, Jr). Los Angeles: Institute of Archaeology, University of California, pp. 131-46.
Longacre, William A. 1985. Pottery use-life among the Kalinga, Northern Luzon, the Philippines. In Decoding Prehistoric Ceramics (ed. B. A. Nelson). Carbondale: Southern Illinois University Press, pp. 334-46.
Miller, Daniel. 1985. Artefacts as Categories, A Study of Ceramic Variability in Central India. Cambridge: Cambridge University Press.
Mills, Barbara J. 1986. Regional patterns of ceramic variability in the San Juan Basin: ceramics of the Chaco Additions Inventory Survey. MS on file, Branch of Cultural Research, National Park Service, Southwest Region, Santa Fe.
Nelson, Ben A. 1981. Ethnoarchaeology and paleodemography: a test of Turner and Lofgren's hypothesis. Journal of Anthropological Research, 37: 107-29.
Integrating functional analyses of vessels and sherds 147
Nelson, Ben A. 1985. Ceramic frequency and use life: a Highland Mayan case in cross-cultural perspective. Paper presented to the Advanced Seminar, 'Social and Behavioral Sources of Ceramic Variability, an Ethnoarchaeological Perspective', School of American Research, Santa Fe, New Mexico.
Nicholson, Paul and Patterson, Helen. 1985. Pottery making in Upper Egypt: an ethnoarchaeo¬logical study. World Archaeology, 17: 222-39.
Pastron, A. G. 1974. Preliminary ethnoarchaeological investigations among the Tarahumara. In Ethnoarchaeology, Archaeological Survey Monograph TV (ed. C. B. Donnan and C. W. Clewlow, Jr). Los Angeles: Institute of Archaeology, University of California, pp. 93-116.
Reid, J. Jefferson. 1982. Analytical procedures for interassemblage-settlement system analysis. In Cholla Project Archaeology, vol. 1: Introduction and Special Studies (ed. J. J. Reid). Arizona State Museum Archaeological Series, 161: 193-204.
Rice, Prudence. 1987. Pottery Analysis: A Sourcebook. University of Chicago Press.
Schiffer, Michael B. 1975a. Archaeology as behavioral science. American Antiquity, 11: 836-48.
Schiffer, Michael B. 1975b. The effects of occupation span on site content. In The Cache River Archeological Project: An Experiment in Contract Archeology (ass. M. B. Schiffer and J. H. House). Arkansas Archeological Survey Publications in Archeology, Research Series, 8: 265-9.
Schiffer, Michael B. 1976. Behavioral Archaeology. New York: Academic Press.
Schiffer, Michael B. 1985. Is there a 'Pompeii Premise' in archaeology? Journal of Anthropological Research, 41: 18-41.
Schiffer, Michael B. 1987. Formation Processes of the Archaeological Record. Albuquerque: University of New Mexico Press.
Weigand, Phil C. 1969. Modern Huichol ceramics. Mesoamerican Studies: Research Records of the University Museum Series 69M3A. Carbondale: Southern Illinois University.
Whittlesey, Stephanie M. and Reid, J. Jefferson. 1982. Cholla Project settlement summary. In Cholla Project Archaeology, vol. 1: Introduction and Special Studies (ed. J. J. Reid). Arizona State Museum Archaeological Series, No. 161. pp. 205-216.
Mills, Barbara J.
Integrating functional analyses of vessels and sherds through models of ceramic assemblage formation
Current research on archaeological ceramics that use functional approaches are characterized as having two broad analytical goals: (1) the identification of the role of ceramics in subsistence systems, and (2) the identification of differences in settlement types. Much research has been conducted to meet each goal, but integration of the two approaches has been difficult because models of ceramic assemblage formation have been ill-defined. As a step toward this integration, a model of ceramic assemblage formation is presented outlining the major variables that must be considered in the assessment of differences in assemblage content. Using ethnoarchaeological data as input parameters, assemblage formation through time is modeled, and the effects of different variables on assemblage content are discussed.