A recent effort in archaeological critique by Cordell et al. (1987) is undermined by confusion, misunderstanding, and misrepresentation. This comment on their paper illustrates how confusion may arise when the literature of the American Southwest is not read carefully and the causes of variability in the archaeological record are understood incompletely.

A series of 29 radiocarbon dates from 11 sites on Admiralty Island span the last 3,200 calendar years. Although our research corroborates many of the results of de Laguna's (1960) earlier work in the area, we find the Tlingit settlement pattern to be at least 1,600 years old, significantly older than previously believed. Dating of a wooden fish weir demonstrates that mass harvesting of salmon has an antiquity of at least 3,000 years.

The John Rowzée Peyton book is believed to be an eighteenth-century original with some nineteenth-century additions that probably derive from John Lewis Peyton. This does not affect the episode of the mound excavation in Kansas described in the book in question.

The appearance of pithouse settlements in the American Southwest that have multihabitation structures has been considered evidence for the emergence of "village" social organization. The interpretation that village systems are reflected in pithouse architecture rests in great part on the assumption that large sites correspond to large, temporally stable social groups. In this article we examine one of the best known pithouse settlements in the Southwest—Shabik’eschee Village in Chaco Canyon, New Mexico—and argue that the site may represent episodic aggregation of local groups rather than a sedentary occupation by a single coherent social unit.

The comments by Reid and others demonstrate a continued failure to distinguish problems of a general theoretical nature from those concerns relating to specific interpretations of southwestern prehistory. We separate these issues here. We also suggest that confusion arises because there are important philosophical differences between our approaches.

A large corpus of data on plant remains from prehistoric feces yields patterns of prehistoric Four Corners Anasazi diet. A general stability of plant-food consumption continues from at least the Basketmaker III through Pueblo III periods, with corn being the one consistently common fecal remain. Other agricultural plants—both crops and weeds—are important throughout the region. There is temporal and spatial variation in the types and abundance of macroplant remains recovered from these feces. Temporally distinct assemblages from the same study area are more similar than contemporary assemblages from different areas. This indicates the importance of the local resource structure (both natural and human maintained) as a major factor influencing dietary composition, even within a relatively small region such as the Four Corners of the northern Southwest.

Sullivan and Rozen (1985) suggest that inadequate typologies limit the interpretation of chipped-stone debitage. We contend that the methodology of inference rather than typological structure is limiting debitage analyses. Inferences about the meaning of archaeological data must be derived from sources that are independent of the archaeological record.

A prominent subsurface zone (layer) of large stones with diameters greater than 6-7 cm occurs in gravelly soil on colluvial aprons in the Lompoc area of California. The soil is mounded and churned by botta pocket gophers (Thomomys bottae). Sedimentological analyses show that the soil within and above the stone zone—and within the gopher mounds—is relatively homogeneous in fine fraction and forms a biomantle. None of the mounds contained stones with long-axis diameters greater than the maximum diameter of gopher burrows, about 6-7 cm. Larger stones gradually subside and form a stone zone. Both field observations and laboratory tests confirm that gopher bioturbation produces stone zones in coarse gravelly soil. This finding, and similar findings in two other recent studies, have important implications for interpreting archaeological site formation, and for interpreting geologic-pedologic processes inasmuch as artifact layers (and nonartifact layers) in some sites entirely may be due to nonanthropic, nongeologic, postdepositional biological agents.

Archaeological studies concerned with reconstructing activity areas, room functions, and site-formation processes can benefit greatly from analyses of the microartifacts found on and within occupation surfaces. These remains are often primary refuse directly related to activities, and can be used to identify such locations as food-preparation areas, flint-knapping stations, and storage facilities. In addition, certain microartifacts are informative about site-formation processes. For example, the grain-size distribution of charcoal may be indicative of primary vs. secondary refuse, high percentages of corroded and crushed bone from scavenger feces may indicate locations of secondary refuse, and many small sherds could point to heavily trampled areas. A case study from the Iron Age city site of Tel Miqne-Ekron in central Israel demonstrates the use of microarchaeology at a complex sedentary site.

Accelerator radiocarbon dates on shell, charcoal, and human bone samples are an indication that human activity at the Del Mar site dates to 5000 B.P. and possibly from 7000 B.P. The shell and charcoal fragments were dislocated stratigraphically by extensive bioturbation and provide evidence that accelerator dates on extremely small samples should be used with caution. Certain depositional environments will require multiple dating within a site because single-sample dating will not yield definitive ages. Dating of individual amino acids from the Del Mar sphenoid yielded an average age of 4900 ± 40 B.P. and conclusively establishes the age for the Del Mar skull. The previously dated tibia was found to be contaminated with exogenous amino acids, thereby making its age uncertain.

In their article, Cordell et al. (1987:571) assert that the Grasshopper Pueblo skeletal series is an "obviously biased sample." This paper criticizes that position as an incorrect interpretation of the Coale and Demeny (1983) model life tables. An alternative interpretation is offered using a modeling procedure based on the Coale and Demeny regressions and pattern-fitting-by-maximum-likelihood techniques. The present author agrees with Cordell et al.'s theoretical position; uniformitarian assumptions should play an important role in archaeological interpretation, especially where the questions addressed are rooted in human biology. The analysis presented, however, indicates that they overestimate problems with the Grasshopper series, and supports the validity of studies based upon it.

Amick and Mauldin (1989) claim that our approach to describing debitage assemblage variation is unproductive because our typology is free of interpretation. They suggest that our method for assigning meaning to archaeological data is sterile because it is based solely on observations of the archaeological record. Their views seem to be based on inattention to key analytic concepts, an unfamiliarity with the full range of factors affecting lithic assemblage content, and an empirically unsupported position about how knowledge of the past may be obtained from the archaeological record. Experimental studies are useful for developing generalizations about how technological factors may influence debitage assemblage characteristics, but sound description of those characteristics is an essential prerequisite to reliable interpretations.

Shovel-test sampling is the most common site-discovery technique used in vegetated areas, but its effectiveness is questionable. Recent papers in American Antiquity by Nance and Ball and by Lightfoot discuss the technique at length. The papers are welcome additions to the literature, but both contain deficiences that require comment and clarification.

Sullivan and Rozen (1985) attempt to provide an improved method of debitage analysis. We suggest that their article is misleading in a number of ways and fails to address the methodological and theoretical issues under discussion. It is doubtful if their four new debitage categories are "interpretation free," since it now is recognized widely that a theoretical basis necessarily underlies interpretation. Their debitage categories are best viewed as morphological or descriptive groups, with technological correlates which are the expected outcomes of generalized reduction practices. They should not be used in interpretation, since the technological inferences drawn from the four debitage categories are assumed without adequate supporting experimentation on the mechanics of flake fracture. We also reject Sullivan and Rozen"s notion that lithic analysis is best interpreted as a continuum and find the criticisms of so-called stage typologies to be unfounded. Lithic artifacts ultimately should be placed within a product group or step, whenever possible, to facilitate recognition of overall manufacturing practices. Finally, we suggest that the only way to improve the subjective nature of debitage analysis is by example, and we point out where this course is not followed in their study.

Results of strength tests on sand- and shell-tempered ceramics reported by Bronitsky and Hamer (1986) are difficult to interpret because of flawed experimental procedures and have questionable relevance to the archaeological record. Results from an improved procedure suggest shell-tempered ceramics as produced in prehistoric southeastern Missouri were stronger and tougher than sand-tempered counterparts.

Paine's comments do not clarify problems with the data from Grasshopper. We maintain that the age/sex pyramid for the Grasshopper series remains problematical.

Shott's comments on Nance and Ball's article on test-pit sampling are off the mark. His comments reveal that Shott has misread the authors, that he generally has misinterpreted their intent, that he does not understand their research design and therefore does not know how to apply their findings, that he overlooks or avoids numerous important points made by the authors, and that he has his own unique view of archaeological data and survey methods.

Feathers is critical of the selection of materials and the use of commercial clay in our earlier research. The use of shell temper from the Great Lakes to the Gulf of Mexico suggests that the properties of this material were considered useful regardless of the specific clay source. Furthermore, Feathers’s comments indicate the need for further discussion and understanding of the role of simulated use testing of archaeological materials.

Can architecture be treated as an artifact worthy of archaeological analysis? This is a key question that remains unanswered in Gilman's (1987) article on southwestern pit structures and pueblos. The majority of Gilman's article is a cross-cultural ethnographic overview of pit-structure and pueblo use. Gilman takes such a "big-picture" approach that when she finally presents her archaeological data it is insufficient to test her model of architectural change. More importantly, by disregarding temporal changes in prehistoric southwestern pit-structure and pueblo designs, Gilman fails to realize that pit structures and pueblos are architecturally related phenomena during the transition period that is the focus of her research. Pit structures, rather than being independent of pueblos, actually provide the construction dirt with which many of the first pueblos are built.

Ensor and Roemer (1989) claim that our debitage categories inherently are linked to our interpretations (i.e., are not interpretation free), that we have dismissed unfairly the work of lithic analysts who use stage typologies, and that our view of lithic reduction as a continuum is questionable. In responding, we review the descriptive and interpretive limitations of stage approaches, and explore terminological issues that arise from Ensor and Roemer's zealous adherence to stage approaches. Finally, we reiterate our position that approaches like ours, which seek to describe and interpret distinctive assemblages of lithic artifacts, are more likely to yield productive results than those, such as stage approaches, which focus on identifying distinctive artifacts.