Shott raises pertinent issues on the limitations of shovel-test sampling. I agree that alternative methods with more efficient discovery rates should be developed and tested. I argue, however, that shovel testing is the most efficient discovery technique now available for detecting buried cultural remains on a regional scale. In survey contexts where the probability of buried remains is high, shovel testing should remain a primary option.

Aboriginal northern Great Lakes fishing strategies varied with season, target species, and organization of the labor force. The placement of Woodland archaeological sites complied with the structure of these fisheries, but their locations do not reflect prey specificity or one specialized technology. Rather, resource-general locations suggest an essential step in the process of specialization. Flexibility in settlement and social styles existed among prehistoric foragers of the midlatitudes, as did a variety of solutions to food-getting problems. Slow, accretional processes rather than temporally discrete growth processes were responsible for Late Woodland site characteristics in this region. Stable locational-selection patterns are visible through reexamination of the historical data base and through statistical analyses of environmental factors associated with sites at a number of Woodland localities.

Wilshusen has critiqued my recent research on pit structures and pueblos in the prehistoric Southwest and has suggested that the use of pit and surface structures is interrelated in that pit structures in southwestern Colorado become deeper at the same time surface structures become more common. This idea is examined further in my response, in the context of a broad and general argument, while Wilshusen examines pit-structure depth, one particular aspect of the pit-structure-to-pueblo transition.

Martin's suggestion that there is great continuity in subsistence strategy through the Middle and Late Woodland periods of the Northern Great Lakes is rejected. She fails to produce convincing evidence for the use of gill nets during Middle Woodland times and to account for the difference in fish fauna on sites of these two periods. Also addressed here is the possible consequence of economic specialization on population size and fluctuation. It is concluded that unlike Middle Woodland populations, those of the Late Woodland fluctuate rather dramatically. Finally, it is suggested that whatever the cause of the population loss and mechanisms of replacement, these shifts likely have important implications for periodicity in ceramic style change.

In a series of articles Robert McK. Bird systematically has denied that preceramic maize has been discovered in the central Andes. The authors consider Bird's methodology to be unscientific and illustrate this by demonstrating the value of a significant body of archaeological evidence in support of preceramic maize.

Fertility (crude birth rate) was estimated from skeletal and corresponding historical relative mortality ratios for a seventeenth- to eighteenth-century Barbados slave population. The estimates varied widely among themselves according to which data source and mortality ratio was used; they also varied from the actual historical fertility rate. In addition, we have raised logical objections to the use of stable model life tables for inferring nonstable vital rates in archaeological populations. These points are problematic for the broad use of relative mortality to infer relative fertility.

Salt production has been proffered recently as an alternative explanation to irrigation agriculture for the existence of canals and terraces at the Hierve el Agua site in Oaxaca, Mexico. This argument is based in part on new interpretations of the functions of two types of water-control features. Hewitt et al. (1987) consider the so-called "pocitos" to be natural, and the "registros" to be water tanks or collecting basins. Evidence that contradicts these interpretations is presented here. Pocitos, it is argued, were constructed intentionally in order to facilitate manual irrigation, and registros are drop structures built to regulate the flow of water after the arrival of the Spaniards. Salt making might have been carried out to some extent at the site, but only in late prehistoric and historic times. Hierve el Agua was principally an agricultural site throughout most of its occupation.

Cultivation of starchy seeds must be recognized as a complex of variables upon which selection operates and of which population growth is the inherent measure of success. Early weaning through the use of starchy-seed gruels, though recently proposed as the principal factor in the increase of fertility, is insufficient as a prime mover for population growth in west-central Illinois during the Middle and Late Woodland periods.

Doolittle (1989) provides interesting information on pocitos and registros but does not show that they are associated with irrigation agriculture exclusively, nor does he address the main point of our paper — that Hierve el Agua was a salt-producing site. We suggest several criteria for distinguishing salt-production sites from agricultural sites and present some additional observations in support of our interpretation of Hierve el Agua's function.

A recent report by Hull (1987) on the microdebitage analysis of soil samples from a stone-circle site in the Northern Plains indicates the utility of such analyses for the study of use and disposal in lithic-tool-manufacture areas. Its value could have been heightened through greater awareness of recent research on the ethnoarchaeology of tipi use and of site-formation studies in general. Various factors were overlooked in Hull's analysis, including the intentional disposal of refuse away from the place of use, the widespread distribution of secondary refuse as a result of smearing and blending, the distinction between occupation and abandonment refuse, and the effects of rodent disturbance within tipis. The fact that these factors were not considered weakens the applicability of Hull's site-formation model. Research among the Slavey Dene of the western Canadian Subarctic suggests that tipis are better viewed as generalized activity centers, embracing a variety of human activities and events, none of which have strict spatial definition. This helps to explain the weak or nonexistent patterning noted by Hull.

Corruccini et al. apply our method of fertility estimation to a Barbados slave population, find that it yields contradictory and erratic estimates of the crude birth rate (CBR), examine its theoretical basis, and conclude that a pessimistic view of paleodemography in general is warranted. We argue that the sample utilized by Corruccini et al. is not well suited for methodological testing. Misestimates of fertility more likely were caused by the peculiarities of this particular sample than by flaws in our method or weaknesses in its theoretical foundation. Deterministic simulations also demonstrate that our method is theoretically well founded. We cannot, therefore, share Corruccini et al.'s pessimism concerning the feasibility of paleodemographic research.

Holland suggests we have erred by providing a "unicausal" model for increased fertility in prehistoric west-central Illinois. He argues that a fuller multicausal model including coevolutionary effects provides a more accurate depiction of the prehistoric events contributing to increased fertility. We appreciate Holland's call for more elaborate models of prehistoric behaviors and evolutionary interactions between humans and other organisms. Unfortunately, his model is untestable because it includes a feedback loop with no exogenous variables. Because of a lack of outside input, Holland's model is immobile and cannot be used in itself to explain changes in organismal biology or behavior.