Using National Register criteria, most archaeological resources are significant in that they contain data that will enhance our understanding of past cultural behavior (Criterion d). During the past 15 years, state and federal historic preservation laws have resulted in the completion of nearly 2,000 Phase I, II, and III archaeological surveys in Pennsylvania. The number of prehistoric sites in the Pennsylvania Archaeological Site Survey (PASS) files has grown from 4,500 in 1974 to over 14,000 in 1995. While the database has increased greatly, the Pennsylvania Historical and Museum Commission (PHMC), Bureau for Historic Preservation (BHP),has not had the opportunity to synthesize this material and incorporate the advances in archaeological knowledge into the decision-making process. When considering the needs for archaeological survey in conjunction with state or federal construction projects, the BHP, until recently, utilized generalized models for predicting the location of prehistoric sites.
With the passage of Pennsylvania Senate Bill 879, the PHMC will be responsible for all surveys associated with state-permitted projects. While the funding has yet to be determined for them, it is certain that the PHMC will not be able to conduct the current level of Phase I surveys. In response to a changing compliance environment, the PHMC has developed a plan to prioritize archaeological survey needs in Pennsylvania. We hope this new approach hopefully will increase the effectiveness of Phase I surveys in locating prehistoric sites, which, in turn, will contribute to our understanding of past cultural behavior.
For the past two decades, state and federal construction projects directed archaeological research. Beginning in the 1970s, archaeologists were consulted for the first time on the possible destruction of large numbers of non-floodplain, non-base camp sites. After some discussion (see W. F. Kinsey, III, 1977, One Archaeologist's Dilemma: A Personal View. Pennsylvania Archaeologist 47(1):42-44. Ann Arbor), their unmitigated loss was considered unacceptable by the archaeological community. This has caused archaeologists to look at ecological settings rarely examined, and it is now generally accepted that this approach has produced a far more complete definition of settlement patterns and cultural adaptations.
Although riverine settings have a very high density of prehistoric sites, the most common type of site resulting from compliance surveys is located outside these zones in what is termed the non-riverine or upland setting. Because these sites are located outside the depositional environment of the floodplain, and are often not stratified, they are often multicomponent and have usually been disturbed by plowing. Sometimes defined as "lithic scatters," they are singularly unimpressive, but they are absolutely essential building blocks for settlement pattern research. By mapping the distribution of temporally diagnostic projectile point types or pottery types from both riverine and non-riverine settings, archaeologists have identified the range of ecological zones exploited during a particular phase, and have developed a more complete description of cultural adaptations.
Using this type of data, J. F. Custer (1985, Prehistoric Archaeological Resources of Pennsylvania's Piedmont and Coastal Plain. A Comprehensive State Plan for the Conservation of Archaeological Resources. Historic Preservation Planning Series 1, vol. II. Pennsylvania Historical and Museum Commission, Harrisburg) and Stewart (1980) have used settlement pattern studies to dramatically illustrate basic changes in prehistoric cultural adaptations. These studies have been the major contribution of compliance archaeology and characterize modern archaeological research in the region. However, with some exceptions (i.e., quarry sites, rock shelters, Late Woodland villages or other sites with large numbers of subsurface features), upland sites have limited research potential beyond settlement pattern studies.
We seem to have reached a plateau in our ability to predict the location of significant sites. Improved predictive modeling using systematically traversed transects (Stewart 1980) or satellite imagery (J. F. Custer, and D. C. Kellogg, 1990, Development of a LANDSAT-based Predictive Model for Prehistoric Archaeological Sites in the Upper Conestoga Drainage, Lancaster County, Pennsylvania. Bureau of Historic Preservation, Pennsylvania Historical and Museum Commission, Harrisburg) has met with some success but would require an enormous amount of time and money to implement these strategies widely. A number of regional models have been developed (J. Herbstritt and R. Michael, 1980, Prehistoric Archaeological Site Survey--Pennsylvania Region II. Pennsylvania Historical and Museum Commission, Harrisburg; Custer 1985; Stewart 1980; S. W. Neusius and R. E. Watson, 1991, Testing the Crooked Creek Upland Settlement Predictive Model: 1990-1991. Pennsylvania Historical and Museum Commission, Harrisburg), but they usually require field confirmation of the main variables used for predicting site locations, making them impractical to a State Historic Preservation Office staff reviewing thousands of projects annually using USGS maps.
Sampling the database is an inherent methodology of archaeological research, and it seems appropriate to the resolution of this situation. Given that fewer surveys will be conducted in the future, how will decisions be made as to the location of these surveys? We do not feel that a sampling design is appropriate for sites in riverine settings; a modified sampling procedure is appropriate, however, for upland prehistoric sites.
To identify regions with high-quality data on upland sites, the following variables were considered:
The analysis of these seven variables was conducted by using physiographic zones and watersheds as the geographical sampling units. We focused on identifying the watersheds with the highest values for each variable. Regional averages were established by examining the physiographic zones and, using the watersheds, identifying an acceptable range around the regional average. The regional averages set base-line standards, and these sheds that were significantly below the regional average for a particular variable were not considered to be representative for the phenomena under consideration.
The analysis began by identifying a list of watersheds with relatively high densities of all sites. The site density list was refined to include only regions that included a sample of both upland and riverine sites. From this list we examined each watershed to ensure that it contained a sample of all chronological components. The chronological component list was further refined by examining each watershed for data on patterns of lithic utilization. Finally, we examined each watershed to develop an index of prior systematic surveys. This guaranteed that each watershed had some data derived from actual excavation, systematic surveys, and controlled surface collections.
Thus far, 19 watersheds out of a total of 104 for the state have been identified as having high-quality data on upland prehistoric site locations. Each watershed has an average 280 sites (1 site/4.2 km2) of which approximately 150 are datable. We recognize that the data in the PASS files is biased in a variety of ways and the process outlined above did not result in a "representative sample" of upland sites. We have simply identified the watersheds with the best existing databases on upland settlement patterns. We also recognize that these are representative samples of the existing database and not necessarily representative samples of prehistoric site distributions. However, we assume that in areas of high site densities and high-quality data, there is some redundancy in the database, and this has resulted in the best available approximation of prehistoric reality.
The approach to the development of research priorities described above offers significant advantages over the current use of more generalized predictive models. It represents an explicit statement of upland survey priorities based on survey needs. It will focus surveys in areas where there are few recorded upland sites and/or upland sites with poor-quality data. In regions that have high values for all of the above criteria, the collection and/or survey of additional upland sites will be a low priority, and the BHP will not recommend that Phase I surveys be conducted in upland zones in compliance with historic preservation laws.
The BHP will continue to protect known and significant sites in these regions regardless of their topographic setting, but Phase I surveys will only be recommended for riverine settings where large habitation sites with features and/or stratified sites are expected. In addition, existing models suggest that sites that could contribute to a wide variety of research problems exist in several regions. Such sites, which will be protected, are characterized by an archaeological context offering additional research potential compared to most upland sites. Rock shelters and quarry sites are another site type requiring protection, and these could be located by geologic formation. The BHP will continue to protect and manage the database in these settings even if they otherwise have high levels of upland surveys.
The BHP initiated this program at the end of January, and it is being applied to both state and federal projects. Obviously, at this date, there are no "results" to report. Over the past year we have had several discussions with the archeological community and other state agencies about this project. The latter have universally felt this was a logical idea; i.e., to stop looking in regions where we had high-quality data and save our money for the "really important sites." They also felt this program would reduce the number of confrontations we all had with the private development community. The response from the archaeological community has ranged from "it is about time" to "you do what you have to do" to "this is not logical, it will not appease the developers and it will ruin archaeology in Pennsylvania." There was a widespread concern that this policy hurt historic archaeology because many historic sites are found while looking for prehistoric sites. In response, we will continue to review our historic files for projects in the 19 high-quality watersheds, and we are initiating a program to develop improved models for historic archaeology, especially in these regions, but also statewide.
As a consequence of the reduced level of effort, expenditures of public money for archaeological surveys will decrease. Future archaeological surveys will be directed at filling in crucial gaps in the statewide archaeological database. The ultimate result of this change in approach will be a federal and state program of identification-level compliance surveys that are more cost effective, and will produce more important information given the limited dollars at hand.