SECTION 2.0

PREHISTORIC SETTLEMENT PATTERNS IN UPLAND SETTINGS:
AN ANALYSIS OF SITE DATA IN WATERSHED D
(CONEMUGH RIVER-BLACKLICK CREEK)

Beverly A. Chiarulli

2.1 Introduction and the Physical Setting

Introduction

This section presents an analysis of the distribution of archaeological sites in the Conemaugh-Blacklick watershed. The study area is part of the Lower Allegheny Sub-basin, which includes the lower portion of the Allegheny River from Clinton to Pittsburgh and the entire Kiskiminetas-Conemaugh River system. Also known as Watershed D in the state Pennsylvania Department of Environmental Protection (PADEP) watershed classification, it includes the Conemaugh River - Blacklick Creek drainages. Blacklick Creek, Two Lick Creek and the middle portion of the Conemaugh River are within the watershed (Figure 1). The watershed drainage area consists of 700 square miles (PADEP Watershed Notebook, page 18D). It is in the unglaciated Allegheny Plateau Section of the Appalachian Plateaus Physiographic Province and is located within Cambria, Indiana, Westmoreland, and Somerset counties. Major communities in the drainage include Indiana, Blairsville, and Homer City in Indiana County; Derry, Bolivar and New Florence in Westmoreland County; and Vintondale and Nanty Glo in Cambria County. Ebensburg and Johnstown in Cambria County are located just outside the drainage (Figure 2). Figure 3 shows the arrangement of the 22 townships within the watershed.

Within the drainage, Blacklick and Two Lick Creeks are major tributaries of the Conemaugh River (Figure 4). These tributaries flow south across Indiana County into the Conemaugh. The Conemaugh flows east west across the watershed. Approximately one third of the watershed is south of the Conemaugh; the other two thirds is north of the river. The headwaters of the Conemaugh are in Cambria County on the west side of a ridge that forms a drainage divide between streams which flow into the Atlantic to the east and those like the Conemaugh which flow into the Allegheny River and then flow through the Ohio and Mississippi Rivers to the Gulf of Mexico.

The report is divided into four sections. Following the introduction, the first section consists of an examination of the physical setting of the watershed. The second section, Research Issues, consists of a synthesis of the archaeological literature and previous models that have been used to predict archaeological site locations. The goal of this section is to provide an assessment of what we currently know about the watershed and to identify research questions that have been examined by other researchers. The third section, PASS File Data for the Conemaugh-Blacklick watershed, is an examination of the site file database and the results of previous cultural resource surveys within and adjacent to the watershed. It also includes information from a local individual who is familiar with the distribution of sites in one part of the watershed. He was interviewed to determine whether there are significant unrecorded sites in the watershed. This research was used to summarize research questions for the watershed, develop examples of the types of sites that could be investigated to address these questions, and examine whether these sites could be found in unsurveyed upland areas of the watershed. The fourth section, Site Significance and Survey Priorities, addresses the research potential of upland sites in the watershed and provides recommendations for future surveys in the area.

General Topography

The elevation within the watershed ranges from 654 feet amsl in western Westmoreland County to 2,860 amsl on the crest of Laurel Ridge in Cambria County (Figure 5). The western half of the watershed has topography typical of the dissected unglaciated Allegheny Plateau found throughout southwestern Pennsylvania. The central and eastern portion of the watershed is notable for the appearance of the first and second of a series of north-south oriented parallel ridges, which mark the beginning of the Allegheny Mountain section. The westernmost ridge is Chestnut Ridge, which is located just east of and parallel to Blacklick Creek. Chestnut Ridge marks a change from the more gradual rolling upland topography of the unglaciated Allegheny Plateau to the more mountainous Allegheny Mountain section. The second ridge, Laurel Ridge, also crosses the watershed. A high valley, underlain by an ancient syncline, is situated between the two ridges. The watershed is divided into two distinct environments by Chestnut Ridge. The distribution of prehistoric sites reflects this difference.

Soils (assisted by Marc Henshaw)

Soils in the Conemaugh-Blacklick watershed are typical of much of western Pennsylvania and reflect the underlying bedrock formations of shales, sandstones limestones, and coal. They tend to be acidic except where formed on limestone or calcareous shale and so are generally not conducive to bone preservation in archaeological sites (Figure 6).

Nine soil associations are found within the Conemaugh River-Blacklick Creek Watershed (Table 1; USDA 1968a, 1968b, 1983, and 1985).

Table 1: Soil Types and Associated Archaeological Sites found in the Conemaugh - Blacklick Watershed

Gilpin-Brinkerton-Cavode soils are moderately deep and deep, gently sloping to moderately steep, well-drained and moderately well-drained soils that formed in residual and colluvial material on uplands. This association consists of areas of soils on moderately broad to narrow ridges on hills that are dissected by drainage ways. The Gilpin soils formed in residuum from acid shale and sandstone.

Brinkerton soils are nearly level to sloping soils on concave footslopes and around heads of drainageways. The soils developed in medium-textured colluvium from acid gray shale and siltstone. Cavode soils are on broad nearly level to moderately steep upland ridgetops, sideslopes, and benches. The soils developed in materials weathered from acid gray and yellow slate with some interbedded siltstone and thin bedded sandstone.

Gilpin-Weikert-Ernest soils are medium-textured and moderately coarse-textured soils, located on moderately sloping to steep valley slopes and narrow to broad, rolling ridgetops. Streams are small and numerous. The association is found primarily in the western part of the Indiana County and extends out to the central and north-central parts. In the north-central part, the ridge tops are remnants of an old plateau and in most places are capped by sandstone. Gilpin and Weikert soils each make up 40 percent of this association; Ernest soils and Dekalb soils make up the rest of the association.

Gilpin-Wharton-Ernest soils are moderately deep and deep, gently sloping to moderately steep, well-drained and moderately well-drained soils that formed in the uplands in residual and colluvial material. This association consists of areas of soils on moderately broad to narrow ridges on hills that are dissected by drainage ways. The Gilpin soils formed in residuum from acid shale and sandstone. The Ernest soils have a slowly permeable layer at a depth of 20 to 30 inches. They formed in colluvium from acid shale, siltstone, and sandstone. Ernest soils are on foot slopes of hills and ridges as well as in concave areas on ridge tops. Wharton soils are deep and well drained. They formed in residuum from acid clay, shale and siltstone. Most areas of this association are in cropland and woodland habitats.

Gilpin-Wharton-Weikert soils are moderately deep and deep, gently sloping to moderately steep, well-drained and moderately well-drained soils that formed in residual and colluvial material on uplands. This association consists of areas of soils moderately broad to narrow ridges on hills that are dissected by drainage ways. The Gilpin soils formed in residuum from acid shale and sandstone. Wharton soils are deep and well drained. They formed in residuum from acid clay, shale, and siltstone. Most areas of this association are in cropland and woodland habitats. Weikert soils are on gently sloping to very steep convex dissected uplands formed in weathered residuum from interbedded gray and brown acid shale, siltstone, and fine-grained sandstone. Most is cleared and used for cropland and pasture or is idle. Forested areas are mixed, deciduous hardwoods.

Hazleton-Dekalb-Buchanan. Hazleton soils are found on ridges, hilltops, and upper sideslopes. Slopes are usually convex. The soils developed in material weathered from acid gray, brown, or red sandstone. Most Hazleton soils are in woodland of mixed oaks, maple, cherry and occasional conifers. Some areas have been cleared for pasture and cropland. Dekalb soils are on nearly level to very steep, uplands and ridges. Slopes are usually convex positions. The regolith weathered from gray and brown acid sandstone in places interbedded with shale and graywacke. Most Dekalb soils are in forests of mixed oaks, maple, and some white pine and hemlock. Smaller areas have been cleared for cultivation and pasture. Buchanan soils are on nearly level to very steep terraces and concave portions of mountain footslopes often extending well into the valleys along drainage ways. The Buchanan soils formed in colluvium from acid sandstone, quartzite, metarhyolite, siltstone and shale. Woodland is the major use. Some areas are cleared and used for pasture, small grain, and row crops. Wooded areas are mixed hardwoods of oak, maple and ash.

Leck Kill-Calvin-Klinesville. Leck Kill soils are nearly level to very steep upland soils in convex positions. Leck Kill soils formed in a regolith of residuum or glacial till derived from red shale, siltstone, and sandstone. Steeper areas are dominantly in pasture or forest. Forested areas are mainly mixed oaks, maple, birch, beech, and cherry. Calvin soils are nearly level to very steep. They are on hilltops, hillsides, and sideslopes of ridges. The soils formed in residuum of red noncalcareous shale, siltstone, and sandstone. Stony and steep slopes are largely in woodland of mixed hardwoods, mainly oaks, with some maple and Virginia pine. Klinesville soils are gently sloping to very steep upland soils on convex positions. Klinesville soils formed in weathered reddish shale with some slate, siltstone or fine-grained sandstone. Locally the less sloping areas are used for growing hay and tilled crops. Common trees are chestnut oak, black oak and Virginia pine.

Monongahela-Philo-Atkins are deep, moderately well-drained soils on terraces and floodplains. This association occurs along the larger streams of the area; about 45% of it contains Philo soils. These soils are frequently flooded. About 30% of the association consists of deep, moderately well-drained Monongahela soils. These soils are level to sloping. They have fragipan. About 20% of the association consists of deep, poorly drained Atkins soils that are frequently flooded. The remaining 5% consists of small areas of Tygart and Purdy soils, approximately two thirds of the association is woodland.

Westmoreland-Gilpin-Culleoka is an association that is dissected by many small streams. The topography is rolling and hilly and includes some gently sloping benches, saddles and hilltops. The floodplains and colluvial areas at the foot of the valley slopes are narrow and of small extent.

Gilpin-Brinkerton-Cavode soils are moderately deep and deep, gently sloping to moderately steep, well-drained and moderately well-drained soils that formed in residual and colluvial material on uplands. This association consists of areas of soils moderately broad to narrow ridges on hills that are dissected by drainage ways. The Gilpin soils formed in residuum from acid shale and sandstone. Brinkerton soils are nearly level to sloping soils on concave footslopes and around heads of drainageways. The soils developed in medium-textured colluvium from acid gray shale and siltstone. Cavode soils are on broad nearly level to moderately steep upland ridgetops, sideslopes, and benches. The soils developed in materials weathered from acid gray and yellow slate with some interbedded siltstone and thin-bedded sandstone.

Geology (assisted by Jamie Rich)

As shown on Figure 7 and in Table 2, only a few of the geologic formations in the watershed are associated with prehistoric archaeological sites. These include the Allegheny Group, the Burgoon Sandstone, the Casselman and Glenshaw Formations of the Conemaugh Group, the Monongahela Group, the Mauch Chunk Formation, and the Pottsville Group. Of these, only the Casseleman, Glenshaw and Monongahela formations are associated with significant numbers of sites, although the Pottsville Group is also of interest because it is associated with the two petroglyph sites in the watershed. These formations are described in detail below. The Glenshaw Formation is associated with approximately the same number of sites as might be expected from its aerial extent in the watershed (45% coverage; 46% of the sites). About 10% more sites than might be expected are underlain by formations from the Casselman, which is associated with 31% of the sites. It underlies only 20 percent of the watershed.

Table 2: Geologic Formations and Associated Archaeological Sites in the Conemaugh - Blacklick Watershed

The most interesting distribution is the number of sites associated with the Monongahela formation. This deposit is found underlying only 4% of the watershed; however, 17% of the sites are associated with this deposit. This association is not coincidental since this formation contains deposits of Loyalhanna Chert. The data suggest that the distribution of prehistoric archaeological sites in the watershed is significantly influenced by the use of this material.

Monongahela Group

The Monongahela Group is located at the top of the Pennsylvania system and is divided into the Pittsburgh and Uniontown formations. H.D. Rogers first named this section the "Upper Productive Coal Measures" in 1858 (Shultz 1999). It was defined as the interval from the top of the Waynesburg coal to the bottom of the Pittsburgh coal. Later the group was renamed for the Monongahela River along which many of these strata are well exposed (Shaffner 1958). It is divided into the Pittsburgh and Uniontown formations. Both formations within the group are sedimentary sequences reflecting a low energy depositional environment (Shultz 1999).

The Monongahela Series contains strata of limestone, shale, sandstone, and mineable coal. It is because of the productive coal beds that the formation has been of importance to the economics of western Pennsylvania (Shultz 1999). Prehistorically its importance stemmed from a very different resource. Loyalhanna chert, which was widely used in this drainage, is located in the Uniontown formation near the Benwood limestone (Oshnock 2000). The Benwood limestone is part of the Pittsburgh Formation and is the dominant limestone in the sequence of limestones and mudstones that fall below the Uniontown Formation (Miller 1934).

The prominent outcrops of the Monongahela Group occur along the axis of the Latrobe and Greensburgh synclines, to the west of the Chestnut Ridge anticline. The total thickness of the series ranges from 270-400 ft (Shultz 1999).

Conemaugh Group (including the Casselman and Glenshaw formations)

The Conemaugh Group lies directly below the Monongahela Group. It is defined as the section between the top of the Upper Freeport coal and the bottom of the Pittsburgh coal (Johnson, M. 1923). The section was originally referred to as the lower barren coal measures. It was renamed after the Conemaugh River, along which it outcrops, in 1875 by Oliver Platt (Shaffner 1958). The total thickness of the section varies from about 520 ft in Washington County up to 890 ft in Somerset County to the east (Shultz 1999). The group is dominated by siltstone, claystone, shale, and sandstone. In the study area it can be found flanking the axes of the Latrobe and Greensburg synclines to either side of the Monongahela Group. N.K. Flint divided the Conemaugh Group into the Casselman and Glenshaw formations in 1965 (Shultz 1999).

The Casselman Formation is the upper portion of the Conemaugh Group (Shultz 1999). It has a thickness ranging from 230 ft in the west to 485 ft to the east. The formation has only one marine unit, the Skelley limestone. Above this, the formation is characterized by freshwater deposits consisting of claystone, limestone, sandstone, shale, and coal. The coal has no regional economic importance due to widespread discontinuity. Nomenclature for the Casselman also varies regionally because it is the least studied of the Pennsylvania formations due to a lack of economic rocks or significant fossil zones.

The Glenshaw Formation consists of the lower section of the Conemaugh Group. The top of the formation is defined as the Ames limestone, which lies immediately under the Casselman Formation. The thickness of the Glenshaw ranges form 280 ft in the west to 420 ft in the east (Shultz 1999). It can be distinguished by four widespread marine units, namely the Brush Creek, Pine Creek, Woods Run, and Ames limestones from lowest to highest respectively. All these formations contain both limestone and shale facies. The Glenshaw does not contain significant coal beds.

Drilling for oil and gas wells and development of surface coal mines have been major impacts in this watershed. Figure 8 shows the location of oil and gas fields and well sites in the drainage. As can be seen, most of these are found in the western third of the watershed. Although an enormous number of wells have been drilled, these have traditionally had only a very limited impact on archaeological sites because of the small area of the surface that is affected. However, recent PADEP regulations now require that well sites prevent contamination of groundwater by constructing barricades around the sites. A number of archaeological sites are known to have been affected by this construction in the Conemaugh-Blacklick drainage (Butch Laney, personal communication).

Archaeological sites in this watershed are also threatened by the continued development of surface coal mines in the drainage. Figure 9 shows the extent of coal deposits in the watershed and the distribution of coal mines. The PADEP is the regulatory agency that issues permits for coal mines and historically has not required Phase I surveys to locate archaeological sites in mine permit applications. Unfortunately, some of the most productive coal deposits in the watershed are found in the Monongahela Group geologic formations. As has been discussed above, these formations were also heavily utilized by pre-European cultures.

Table of Contents

Top of Page

2.2 Research Issues

This section of the report examines the archaeological literature to identify general research questions for the study area and to identify approaches that have been used in previous archaeological surveys. The information is limited by the quality of the original data. This review does not attempt to correct or revise the information, but rather to summarize what is known about the watershed. In some cases, surveys and sites located outside the watershed are discussed, because similar information is not available for the watershed.

Use of Chronologic Terms in this Report

Archaeologists use terms such as Late Woodland or Terminal Archaic to define the cultures as well as describe the chronologic placement of those cultures. This report does not attempt to synthesize the prehistoric record of either the Conemaugh-Blacklick watershed or western Pennsylvania. In general, the designations are used in a manner consistent with the usage of the author whose work is being discussed. This does lead to some confusion and inconsistencies in terminology. As William Johnson noted in his review of the draft version of the report, the meaning of a number of terms has changed in the past 70 years and there are terms that currently have different meanings depending on whether they refer to an archaeological phenomenon from northwestern Pennsylvania or southwestern Pennsylvania. Three of these inconsistencies are discussed below.

First, there is a disagreement as to the chronologic placement of bifurcate-base points. Johnson (comments on the draft report) notes that the orthodox placement for bifurcates is in the Early Archaic. In a recent publication, Carr (1998: 80) lists McCorkle, St. Albans, and LeCroy as Middle Archaic types and says this categorization is used in his analyses of the PASS files. This inconsistency in usage is found in the report, and is important because the computerized PASS file database probably also contains this inconsistency.

Second, the Transitional period has been thought to represent an intrusive tradition characterized by the appearance of broadspear point types. Its chronologic placement overlaps the end of the Late Archaic and beginning of the Early Woodland periods especially in the Susquehanna Valley. Johnson (comments on the draft report) suggests that Terminal Archaic is now the preferred term, but since many authors cited here use the earlier term and since the PASS file database uses Transitional as one of its chronologic periods, that term will be used in some cases.

Finally, terminology for the entire Woodland period is confused because prehistoric groups in western Pennsylvania were influenced at different times from different directions outside the region. For example, again as noted by Johnson (comments on the draft report), in the Ohio Valley, the break between Early and Middle Woodland is set at the point where Adena becomes Hopewell (between 100 BC and AD 1), if that change occurs. The break between Middle and Late Woodland is placed at the end of the Hopewell or at AD 350-400. The Lower Upper Ohio Valley seems to reflect this Middle Ohio Valley sequence in which the Late Woodland ends at around AD 1050 and is followed by the Late Prehistoric Monongahela designation. This terminology is different from that used in the Lower Great Lakes, Northeast or Middle Atlantic regions, which use only the Woodland terminology. Archaeologists working in the Middle Ohio Valley have adapted this terminology at different times and so have used the Woodland terminology in many of the earlier publications.

The Middle Ohio Valley terminology, however, is not applicable for the upper Allegheny River valley or the glaciated Allegheny plateau. Johnson places the break between these two nomenclature systems on the drainage divide between the Kiskiminetas River and Crooked Creek. The Conemaugh-Blacklick watershed forms the southern side of that divide. For consistency, when Monongahela or contemporary cultures and sites that post date AD 1050 are discussed in Section 2.2, the Late Prehistoric designation will be used. However, it should be noted that Late Prehistoric is used only for sites in the Ohio River subbasin in the PASS file database, so that tables in Section 2.3 use Late Woodland as a category that includes Late Prehistoric sites. In the PASS files, sites like Johnston and Squirrel Hill are dated to the Late Woodland period even though both are included in the Late Prehistoric Monongahela Johnston Phase.

Previous Archaeological Surveys

Several sources were used to identify research issues for the region. These included several general surveys, as well as archaeological excavations of sites. Previous archaeological investigations in this region began before 1929 with the work of George Fisher and Robert M. Engberg along the Monongahela Valley in Westmoreland County. Systematic investigations were initiated in the 1950s with the Carnegie Museum of Natural History's Upper Ohio Valley Archaeological Survey directed by William Mayer-Oakes (1955), which represented the first systematic attempt to record sites in seventy counties in Pennsylvania, New York, Ohio, West Virginia, and Maryland. Thirty-one sites in Westmoreland County, 3 in Cambria County, and 9 in Indiana County were recorded by Mayer-Oakes (1955).

During this same period, Ralph Solecki surveyed the Conemaugh drainage prior to the construction of the Conemaugh Lake Dam as part of the Smithsonian Institution's River Basin Survey (Solecki 1951). Seven sites were recorded along the Conemaugh during this survey--three on the northern side of the Conemaugh in Indiana County (including 36IN2) and four in Westmoreland County (36WM4, 36WM5, 36WM6, and 36WM10). (These seven sites are also included in Mayer-Oakes' survey.) Of these sites, only the Johnston Site (36IN2) was recommended for excavation. The excavation, conducted by Don Dragoo of the Carnegie Museum, uncovered a Late Prehistoric occupation, identified as a Monongahela village (Dragoo 1953, 1954). More recent research suggests that the village represents an amalgamation of intrusive McFate phase people from the French Creek Valley and local Monongahela populations (Cowin 1981; George 1997; Johnson 1999; Johnson et al. 1979). It is the type-site for George's (1997) Johnston phase. A second Late Prehistoric village from within the watershed was investigated during this same period. The Squirrel Hill Site (36WM35) is located on a terrace of the Conemaugh at the northeastern boundary of Westmoreland County (George 1997; Robson 1958). It is now considered to be another Johnston phase site.

Other excavations of single sites during this period have added to our understanding of the archaeological framework for the Conemaugh-Blacklick watershed. Much of this work focused on the excavation of Late Prehistoric villages. In 1971, Richard George of the Carnegie Museum directed an excavation at the Ryan Site (36WM23), a Late Prehistoric village in Penn Township, Westmoreland County, west of the Conemaugh-Blacklick watershed (George 1974). In the 1970s, Virginia Gerald of Indiana University of Pennsylvania (IUP) excavated the Mary Rinn Site (36IN29), located on Crooked Creek just outside Indiana, Pennsylvania for several seasons with IUP Archaeology Field School students. These excavations have not yet been published. Recently, Neusius and Giles (1999) analyzed a surface collection from the site and presented the results of the analysis at the Society for Pennsylvania Archaeology meetings. Additional excavations were conducted at this site during the summer of 2000 by the IUP archaeological field school. Two other investigations of Late Prehistoric sites have been conducted in this area. The Kimmel Mine site, located at the confluence of Plum Creek and Crooked Creek was investigated by the BHP in the mid 1980s and has yet to be published. The Fishbasket sites, on Redbank Creek in Armstrong and Clarion counties were investigated by Burkett (1999). Both of these latter sets of sites are notable for their locations at the fords of major historic Native American trails across Crooked and Plum Creeks and Redbank Creek, respectively.

Other surveys have included those conducted for CRM projects for gas pipeline construction, industrial parks, the Greensburg Bypass project of the Pennsylvania Turnpike Commission, electrical transmission lines, the Pennsylvania Department of Transportation, housing developments, and surface coal mines.

The PHMC Regional Survey

From 1979 to 1981 one of the PHMC-sponsored regional archaeological site surveys included Westmoreland, Indiana, and Cambria counties in Region VII along with Allegheny, Armstrong, Blair, Clarion, Jefferson, Elk, Cameron, and McKean counties. The Region VII program was conducted by Mark McConaughy and Verna Cowin of the Carnegie Museum of Natural History. More than 300 sites were added to the site files by Cowin in 1980-81. Although her survey concentrated on certain quadrangles, she developed a regional typology of point types and produced a generalized predictive model for the distribution of sites from different chronological periods and environmental settings. Table 3 presents an updated version of Cowin's typology, which includes information from a number of riverine and upland sites from the Conemaugh-Blacklick watershed.

Her other conclusions can be summarized as follows (Cowin 1989, 1991):

1. The majority of sites are located in the uplands adjacent to water, on hilltops, hill slopes, benches, saddles, and bluffs below the highest elevations. Hillslopes are occupied in all time periods, except Paleo-Indian.

2. Floodplains and terraces of both major and minor tributaries were also occupied, especially near stream junctures, including small intermittent runs from springheads. However, during the Late Woodland, fewer floodplain sites are found.

3. Ninety-one percent of the Region VII site sample shows that sites are located on well-drained to moderately well-drained soils.

4. Large villages tend to be located on relatively flat terraces or on saddles or benches with no more than 8 degrees of slope.

5. Medium to large-sized sites are generally located where a variety of habitats can be exploited and where there is access to lithic resources.

Cowin's research found that archaeological sites were located in upland topographic settings such as hillslopes, benches, bluffs, and springheads. Although none of her survey areas were within the Conemaugh- Blacklick watershed, she examined similar topographic areas. Sites are found in similar settings in the Conmaugh-Blacklick watershed. She indicated that she did not have enough information to develop a predictive model for site location, but instead suggested that all of these settings might contain sites and should be examined.

Table of Contents

Top of Page

The Crooked Creek Survey

In the 1980s, IUP Archaeological Services developed a predictive model for prehistoric upland site locations in the Crooked Creek drainage. The initial project was funded by a small PHMC planning and survey grant. Few upland sites had been previously recorded in the drainage. The study was designed to examine whether upland sites were under represented in the site database for western Pennsylvania. The approach used for the Crooked Creek Survey was different from the PHMC regional surveys. The PHMC surveys used inductive approaches in which informants were interviewed to identify locally collected sites in conjunction with surveys of targeted areas. The goal of the survey was to locate as many new sites as possible in all topographic settings. Unlike the earlier investigations, the Crooked Creek survey was an attempt to identify variables that could predict where sites might be located in upland areas. For the Crooked Creek Survey, predictions for the location of prehistoric sites were generated from the archaeological literature and then tested through field investigations.

Table 3: Point Types and Associated Cultural Periods Region VII*

*This table was based on Cowin 1991; however, subsequent research has refined our understanding of the chronological assessments of some of these periods . Changes are based on W. Johnson's comments on the draft version of this report.)

The Crooked Creek Model identified seven environmental variables, which previous investigators had used to reflect environmental variables that affected prehistoric settlement location. The variables included slope, topographic setting, aspect, proximity to Native American paths, proximity to water, general resource productivity (based on soil suitability indices) and general suitability for agricultural production (also based on soil suitability). The same variables are among those used in the PASS data files and are listed on Table 4. For comparison, the slope, aspect, and topographic setting for the upland sites from the Conemaugh-Blacklick site database were included on this table, although these sites were not included in the original study. Interestingly, a number of sites in the Conemaugh-Blacklick watershed are located in what were considered to be undesirable locations, which suggests that our views of what were desirable site locations may not reflect the actual choices made by prehistoric groups in the Conemaugh- Blacklick watershed.

The Elderton U.S.G.S. Quadrangle was used as the study area for the Crooked Creek survey. The quadrangle was divided into 300-meter squares or quadrants. Each of the 1,610 squares on the quadrangle was given a numeric score based on a scoring system for each of the environmental variables listed above. In this system, the highest score was identified as that most favorable to prehistoric occupation. So, for example, slopes of more than 15% were identified as unsuitable for settlement. A square with an average slope of less than 3% received a score of 3 points; a square with more than 15 % slope received zero points. Squares received from 13 points (the highest score) to zero points. A total of 154 squares were identified as lowland areas. Only 7 lowland squares received less than 8 points. The squares identified as upland areas numbered 1,456 and only 11 of the upland squares received 11 or 12 points compared to 73 of the lowland squares. One of the most useful parts of this investigation is the quantification of upland and lowland areas and the development of a method to quantify these environmental settings. The most common upland score was 6. Based on the point scores, quadrants were categorized as to their probability of containing archaeological sites. As an example, the point scores and results for the first year of the project were translated into the following categories:

1. Sites highly probable (10-13 points, 53 upland quadrants, 136 lowland quadrants)- 1 quadrant surveyed, 1 site found

2. Sites probable (8-9 points, 193 upland quadrants, 15 lowland quadrants) - 2 quadrants surveyed, 1 site found

3. Sites possible (6-7 points, 545 upland quadrants, 3 lowland quadrants) - 6 quadrants surveyed, 3 sites found

4. Sites improbable (4-5 points, 532 upland quadrants, 0 lowland quadrants) - 5 quadrants surveyed, 1 site found

5. Sites highly improbable (<3 points, 133 upland quadrants, 0 lowland quadrants) 1 quadrant surveyed, 0 sites found

A sample of each category was selected for survey during this first year using a random numbers table. The squares were surveyed through walkover survey and by the excavation of shovel tests at 15 to 20 meter intervals. The results seemed promising, so a second year of funding was obtained and additional quadrants were surveyed. In addition, since the total quadrant score was based on an average of the scores for that unit, small attractive locations for sites were frequently overlooked.

Table 4: Variables used in the Crooked Creek Survey

These included settings such as drainage divides, springheads, and stream confluences. For the second year of the survey, a variable called secondary setting was added to the model. When an attractive secondary setting was identified in a quadrant, two points were added to the total score. However, the results of the second survey were less clear than those of the first year and the researchers were unable to demonstrate a statistically significant relationship between any of these variables for the prediction of site location. Another problem with this approach is that the combined scores make it difficult for other researchers to determine which if any of the variable categories is effective in predicting site location. The study was not successful in developing a predictive model for the location of upland sites.

The study did demonstrate that archaeological sites were present in the uplands of the Crooked Creek watershed and that these sites could be found through controlled and systematic surveys. The investigation also provided insights into upland site density. A total of 38 quadrants were surveyed during the two seasons. Each covered an area of 300 by 300 meters or 90,000 square meters. The survey covered a total of 3.42 square kilometers and found 11 new prehistoric sites. This is a site density of 1 site per 0.31 square km. In contrast, the high-site-density watersheds identified in the survey priorities policy have densities ranging from 1 site per 1.7 to 3.6 square kilometers. The Crooked Creek sample has an upland site density that is more than five times higher than the most dense of the high-density upland watersheds.

In addition, the Crooked Creek survey is a controlled sample in which a systematic sample of upland settings was investigated. It was not biased toward the "best" upland settings. The entire survey area was investigated using a single survey methodology. Pedestrian survey and shovel test intervals were the same in quadrants with attractive locations as in those on the sides of steep hillsides. Shovel tests were excavated at either 15-meter intervals or at an interval of 20 meters between tests within transects and 10 meters between transects, so that either 400 or 450 shovel tests were excavated in each quadrant.

The survey also found that while most upland sites were small undated lithic scatters, a substantial number could be assigned to a single cultural period (Table 5). Eight of the 13 sites are undated. It is not known if these are "significant", because they were not investigated beyond their initial discovery. Some of the undated sites might be found to contain diagnostic artifacts if examined further.

Table 5: Results of the Crooked Creek Survey

(1) Levanna points are more commonly considered to date to the Middle Woodland
(2) 0 is the lowest ranking indicating sites highly unlikely to occur

At the conclusion of the Crooked Creek study, Neusius and Watson (1991) concluded that although the results of the study were not conclusive, additional testing was needed to determine if it would provide a low cost systematic approach to the investigation of upland areas. Since it was developed to identify possible high probability areas in upland portions of the unglaciated Appalachian Plateau, it was used in the initial stage of investigation of the I-70 to Route 51 Mon-Fayette Transportation Project, conducted in the mid-1990s by Skelly and Loy, Inc. One of the stipulations in the Memorandum of Agreement between the Federal Highway Administration, the Advisory Council on Historic Preservation, and the PHMC was to apply and evaluate the usefulness of model for the identification of archaeological resources as part of the Environmental Impact Statement for the project (Duncan et al. 1996).

For the Mon-Fayette investigation, 214 quadrants were placed over two proposed alignments. Seventeen quadrants were surveyed; these represented a 5% sample of each of the probability categories except the "sites highly improbable" category. This lowest probability category was identified only in quadrants that had been subject to substantial disturbance from surface mines. Only three sites were found, one in each of the three highest ranked categories (ie, sites highly probable, sites probable, and sites possible). The authors concluded that the Crooked Creek model was ineffective as a method for predicting site locations during an alternative alignment analysis (Duncan et al. 1996: 22). Some of the problems that were identified with the model were that the manual calculation of quadrant scores was labor intensive, that the quadrant sizes were so large that they represented multiple microenvironments, and that the cost of investigating a statistically valid sample was prohibitive. These problems relate primarily to the utility of the model in specific phases of transportation planning projects.

However, the actual results of the field survey of the model are comparable to the results from the original Crooked Creek studies. Six archaeological sites were found in the 17 quadrants that were examined. This survey investigated a total area covering 1.53 square kilometers and found a site density of 1 site per 0.26 square kilometers. This density is even higher than that of the Crooked Creek survey, partly because the Mon-Fayette study included non-upland as well as upland areas in the survey. It does confirm the results of the Crooked Creek study-- that when the uplands are systematically surveyed using standardized methodologies, actual site densities are much higher than those projected for the high-site-density watersheds in the survey priorities policy.

The Mon-Fayette GIS Predictive Model

Skelly and Loy then developed a GIS model, which was based on a correlation of known site locations and background environmental variables (Duncan et al. 1998: 28). Although a detailed discussion of the development of the GIS model is beyond the scope of this report, the authors provide some insight into some of the same issues that were encountered in this study of high-site-density watersheds.

First, the baseline archaeological data set was generated from the PASS file database, the Carnegie Museum archaeological site files, and a database on Monongahela sites developed and maintained by William Johnson. The authors found that there were substantial discrepancies between the data sets. The plotted positions of sites differed by as much as 50 to 100 meters in some cases. So only sites that could be located with confidence were used in the model. In general, the authors found that sites recorded as part of a compliance or professional survey were much more likely to contain reliable information.

Second, the development of the model was an extremely complex process. Basic sources for environmental data included the digital elevation model data (DEM) for the three quadrangles; digital stream, river, lake and road data; soil types, springs, and intermittent streams from the county soil surveys; locations of documented historic Native American paths; bedrock geologic formations, and the location of major fifth-order drainage divides. The primary data sets were used to derive a series of 72 variables, each of which was transformed into a gridded data layer in the GIS. Each data set was then used to create secondary variables. For example, elevation was used to create two measures of terrain roughness, two measures of relief, two types of aspect, and two slope variables. A digital view of the terrain was created by a combination of slope, aspect and elevation and was in turn used to create a variable that measured the amount of solar insolation gained by each land parcel at different times of day during the shortest day of the year. The slope data set was used to create variables to measure the difficulty of surface travel to various locations like the nearest water source, trail, vantage points, topographic saddles, drainage divides, and stream confluences. These 72 variables were then statistically evaluated against the existing site database to determine which variables would be of further use in the model. Twenty-six variables survived this process and were used to construct the GIS model (Table 6).

Based on a stepwise regression analysis, six of these variables were dropped from the final GIS model. The accuracy of the model was then analyzed through a Chi-square test of the ranked cells investigated in the Crooked Creek survey evaluation and the Phase I survey of the preferred alignment. The model was found to provide a statistically significant way of measuring the potential of an area to contain archaeological sites.

Table 6: Environmental Variables Used in the Skelly and Loy GIS Model

This model has been discussed in detail because of the potential that GIS models have for combining large amounts of data into new variables that can be used to explain archaeological site distributions. There do seem to be some limitations to this approach, however. As the authors note, this approach is able to calculate the probability that land parcels contain archaeological sites and is useful for planning purposes (Duncan et al. 1998: 67). It is based on existing site locations and so must reflect those in the probability categorizations. Once a large area like a USGS quadrangle has been analyzed in this system, it could be re-evaluated quickly with the modification of a few variables. However, it seems that the initial creation of the GIS would be complicated and time consuming, and at least as labor intensive as more traditional approaches.

Survey of the Kittanning Bypass

Stewart and Kratzer (1989) surveyed the alignments of the proposed Kittanning Bypass and specifically addressed the question of upland site distributions. This project is to the west of the Conemaugh-Blacklick watershed and consisted of a survey of the uplands just east of the Allegheny River. The primary topographic settings in this area are upland flats, which are dissected by deep narrow valleys containing low-order streams. Stewart and Kratzer (1989:20) note that there are few well-drained, habitable landforms within the valley or terrace formations. In this article, the authors first review the available locational data for the general area. This project surveyed 6 transects, each approximately 6 miles long. No archaeological sites had been previously recorded for the area. As part of their investigation, the authors summarized what was known of the locations of various cultural groups in the regions. (The terminology used below follows that used by the authors.)

For the Paleo-Indian period, Stewart and Kratzer (1989:25) suggest "broad valleys of high order streams and extensive upland flats in proximity to surface water are the documented and predicted foci of camps and many activities." They also suggest that the location of sources of high quality lithic material influenced the size and location of Paleo-Indian settlements. While there were no Paleo-Indian sites previously recorded in their survey area, one site, 36Ar37, was located just to the north of the project area, associated with a first-order stream near its juncture with Cowanshanock Creek. Their view is similar to that of Lantz (1985) who reported that Paleo-Indian sites are often associated with first order streams in the unglaciated Allegheny Plateau. While there were no settings in the Kittanning Bypass survey like that of 36Ar37, upland flats associated with first-order streams were identified as potential settings for Paleo-Indian sites. Early Archaic groups are thought to have occupied similar settings.

Stewart and Kratzer (1989) suggest that the Middle Archaic period shows an increase in the number of sites, although this is based on the presence of bifurcate-base points, which they associate with the early part of the period. (Other researchers place bifurcates in the Early Archaic, so would see this increase in that period.) These sites are thought to be located in similar settings to the earlier occupations. In contrast, the Late Archaic cultures are associated with a climatic change to a warmer and drier climate and replacement of hemlock with oak and hickory in the area forests and occupy a wider range of environments (Stewart and Kratzer 1989:24). The authors indicate that Late Archaic cultures can be characterized as "specialized hunters" (Stewart and Kratzer 1989:25) and that they had developed a well-defined schedule for a yearly subsistence round. Although Late Archaic sites are found in all environments, the authors note that Brewerton points representative of the later part of the period are more commonly found in upland settings. Commonly used upland settings are found near the heads of drainage and/or on prominent upland flats overlooking stream valleys. Late Archaic sites are generally interpreted as short-term camps, although more sedentary camps may have been located in the river valleys late in the period and into the Terminal Archaic (Stewart and Kratzer 1989:26)

In the Early Woodland period, Stewart and Kratzer (1989) note the appearance of an additional site type associated with "burial ceremonialism", i.e. burial mounds. However, even with the first appearance of cultivated plants there is no evidence of a major subsistence or settlement shift. Types of sites include base camps in addition to special-purpose and/or transient camps. All of these types might be found in the uplands as well as along the major streams. At many locations, Late Archaic and Early Woodland artifacts are found together, evidence that supports the idea that there was no major change in settlement locations between these two periods.

Stewart and Kratzer (1989) identify this part of the Allegheny drainage as an area influenced by two distinct Middle Woodland traditions. Middle Woodland cultures of the Upper Allegheny River and glaciated plateau show links to Hopewell style traditions, while those of the unglaciated plateau and Upper Ohio River also used mounds for burial locations, but show less evidence of Hopewell influence. Steward and Kratzer (1989:26) report that the area around Kittanning seems to be linked to the Hopewell-related cultures in the Middle Woodland period. They note that these sites are often located in the uplands and that the upland sites may in fact be habitation sites associated with burial mounds located along the rivers.

The Allegheny River Valley was also home to several Late Woodland traditions, including the Mead Island and the Allegheny Valley Iroquois, and was influenced by the Monongahela cultures to the south. For the Kittanning area, Stewart and Kratzer's (1989:27) review of the settlement pattern for the Late Woodland of the Allegheny Valley found that large villages are generally located on the floodplains of the major rivers. The uplands were used as hunting and collecting territories.

Stewart and Kratzer (1989) then identified high probability areas where archaeological sites are expected to occur. These include:

prominent upland flats overlooking stream valleys;

saddles between drainage divides and prominent upland flats;

areas near the head of active drainages;

areas near the head of inactive drainages; and

upland flats adjacent to a first order stream or drainage and in proximity to a stream confluence (Stewart and Kratzer 1989:29).

However, based on a field evaluation of the project area, some of the areas initially identified as having a high potential for containing archaeological sites were re-interpreted as low probability areas. Table 7 lists the environmental settings encountered during the survey, the potential probability that sites might occur, and the results of the survey. Their findings were that even areas that were predicted to have a high probability of containing prehistoric sites did so only in about 20% of the surveyed areas, because the potential location in which sites could occur was much larger than the actual surveyed area.

Table 7: Topographic Setings Investigated During the Kittanning Bypass Survey

Table of Contents

Top of Page

Survey and Excavation in the Meyersdale Bypass Project

More recent investigations associated with the construction of the U.S. 219 Meyersdale Bypass (Boyd et al. 2000) have greatly increased our understanding of the settlement patterns in the lower Casselman River, to the southeast of the Conemaugh-Blacklick watershed. Sixty-eight sites were initially located during the Phase I survey, 21 of these were investigated during the Phase II stage of the project, and Phase III data recovery investigations were conducted at 8 sites. Since the Meyersdale Project was located at the extreme southern edge of the county in a different environmental setting than the Conemaugh-Blacklick watershed the results of the project are only summarized here. As shown in Table 8, there were a number of upland sites investigated in this survey that provide a detailed picture of the prehistoric occupation on the lower Casselman River. The project investigated two alternative right-of-way corridors, each approximately 7 miles long. The investigations provided a transect through a densely populated prehistoric settlement area, although it is not as densely settled as parts of the Conemaugh-Blacklick watershed.

Table 8: Summary of Period of Occupation and Environmental Setting of Sites Investigated in the Meyersdale Project

Boyd et al. (2000) provide definitions for a number of site types that were identified in these investigations. Two site types were identified for the Paleo-Indian and Archaic periods. Base camps (BC) are domestic sites that contain evidence for a number of activities. Short-term camps (STC) are small sites that were visited very briefly with evidence of few activities often related to an activity-specific task. A greater variety of site types appear in the Woodland Period. These include ceremonial sites such as earthworks and mound sites. Hamlets are single- or two-structure sites occupied by groups of related families during multiple seasons or on a year-round basis. Hamlets are expected to have large quantities of lithic and ceramic refuse in pits or middens. Multiseasonal occupations without evidence of horticulture are termed multiseason base camps. Sites that show some but not all of the characteristics of a hamlet are called short-term base camps. Short-term base camps (STBC) are sites that may contain structures but have few ceramics and little or no indication of storage activities. A specialized short-term base camp may be recognized by artifact assemblages that reflect only a limited or specialized range of activities. These may include fishing, hunting or meat processing, as well as lithic reduction, lithic workshop, or lithic procurement sites. Short-term encampments (STE) are ephemeral sites where one or more individuals stopped for a short period of time. Short-term base camps are defined as occupations that lack evidence of multiple seasons of occupation and of the presence of cultigens. Villages are large multiple-structure, multiseasonal, occupations with evidence of a variety of activities.

As a comparison with the current study of sites in the Conemaugh-Blacklick watershed, it is useful to look at the upland site types found for each cultural period in order to identify information that would not be available if these types of sites had not been investigated. All of the Paleo-Indian sites investigated in the project are short-term camps (Table 9). Four of these are in floodplain settings, one is in an upland setting. All of these sites were identified as dating to the Paleo-Indian period based on the recovery of a single diagnostic point. The upland site was identified based on a point found in 1948; only a single flake was found at this location during investigations for the current project (Boyd et al. 2000:43). According to Boyd, the site added little to our understanding of Paleo-Indian settlement in this area other than to confirm that there was some undefined use of upland areas and that this use was not substantially different than Paleo-Indian use of floodplain settings.

Table 9: Distribution of Site Types in Upland and Floodplain Settings on the Meyersdale Project

Two site types were identified for the Middle Archaic period (6500-3000 BC), base camps and short-term camps. Both of the two identified base camps were in the uplands, one in a saddle overlooking the confluence of the Casselman and Elklick Creek (36SO51), the other to the north of and overlooking the Casselman (36SO234). Since 36SO234 was outside the project area, it was only investigated during the Phase I survey. The other site, 36SO51, was subject to a data recovery investigation. Although as shown on Table 8, the site was occupied from the Middle Archaic through the Late Prehistoric periods and might appear to be a typical mixed assemblage site, it produced extremely important information with the recovery of a structure dated to the Middle Archaic. Five features were discovered and dated through radiocarbon analysis. These included postmolds, a hearth, and two earth ovens all of which dated to the Middle Archaic. Interestingly, none of the recovered projectile points were typical Middle Archaic types. The features at the site had been damaged because the area was a plowed agricultural field. Four of the eight short-term camps dating to the Middle Archaic period were located in the uplands. One of these, 36SO106, was located on a gently sloping field above the Martz Rock Shelter sites and was the subject of a Phase III investigation. Eleven Middle Archaic points were recovered from this site, which showed occupation from the Middle Archaic through the Late Woodland periods. Four of the points dated to the early bifurcate point phase of the Middle Archaic; the others were from the later phase. No features were discovered. The site was interpreted as a lithic manufacturing locus during the Middle Archaic and subsequent periods (Boyd et al. 2000:57). The base camps and the lithic manufacturing site were found only in the uplands. For this period, upland site information added significantly to our understanding of the settlement pattern.

The same two site types were also found in the Late Archaic occupations. At the same time there was an increase in the number of sites dated to this period. There were 27 Late Archaic components, although only two base camps were identified, one in the uplands, one on the floodplain. Site 36SO51, which was previously described as a Middle Archaic base camp, was also used as a base camp during the Late Archaic period. Samples from eight features produced Late Archaic radiocarbon dates. The features were identified as four hearths and four pits. Of the 25 short-term camps, most contain a limited range of artifacts and were identified through the recovery of diagnostic point types in multicomponent occupations. Site 36SO106 continued to be used as a tool manufacturing locus. One large Late Archaic feature was excavated at the site. The feature measured 30 sq meters in area and contained over 6,500 flakes, 280 pieces of shatter, 6 bifaces, 4 projectile points, cobble tools including 1 hammerstone, and 1 teshoa and charred wood fragments classified as oak, hickory, American basswood and other unidentifiable hardwoods. The other short-term sites were identified through the recovery of a few points and small number of flakes.

The Terminal Archaic period settlement pattern includes base camps and short-term camps. Both types were found in upland and floodplain settings. There were 19 occupations associated with this period. Generally, sites that contain multiple periods of occupation show a continuation of the type of use; for example, base camps remain base camps. Only Site 36SO241, an upland site, shows a change from a short-term camp to a base camp, although that identification is based only on the recovery of steatite vessel fragments from the site. There are more floodplain sites in this period, but there are 8 upland Terminal Archaic components. Site 36SO51 remains the most intensively utilized base camp. Two features were identified that produced Terminal Archaic dates. Site 36SO106 continued to be used as a resharpening and tool manufacture locus. The other short-term camps contained only low numbers of diagnostic point types and debitage.

The Early Woodland period is dated from 1000 to 100 BC in this study of the Meyersdale project (Boyd et al. 2000). Site types include short-term encampments and short-term base camps. Four short-term base camps were identified with two each situated in the uplands and the floodplain. The two upland sites included 36SO51, at which two features were excavated and dated through radiocarbon analysis, and 36SO106 at which one feature was excavated. Site 36SO106 again shows a continuation of activities including lithic reduction and resharpening seen in the earlier periods of occupation. Nine Early Woodland short-term encampments were also found. Seven of these are upland sites, but contain few distinctive artifacts other than diagnostic projectile points. None of these are identified as single component Early Woodland sites.

The Middle Woodland-Late Woodland occupations are identified as those cultures that occupied the area from 100 BC through AD 900. This period is seen as distinct from the Late Prehistoric cultures, which date from AD. 900-1400. Only three of the six hypothesized Middle Woodland/Late Woodland site types were identified in this project area. These included short-term base camps, short-term encampments, and ceremonial sites. Features containing 29 datable charcoal samples were recovered from five of the sites, although all but 10 of the samples were from 36SO220, a floodplain ceremonial site. Eighteen components dated to this period were found; eight were located in the uplands. The upland sites included 4 short-term base camps and 4 short-term encampments. Two of the upland short-term base camps are the two sites that had been occupied as base camps throughout most of the preceding cultural periods, 36SO51 and 36SO106, which continues to be used as a lithic reduction site. Site 36SO114, which had also been the location of short-term camps and encampments, is again reoccupied in this period as is 36SO223.

Late Prehistoric occupations included village sites, short-term base camps, short-term encampments, and one hamlet. The upland occupations included three short-term base camps, one hamlet, and one short-term encampment. Given the number of upland Late Prehistoric villages in western Pennsylvania, it is somewhat surprising that none were found in this project area. The hamlet, Site 36SO241, contained very few artifacts. Two features were investigated at the site and found to contain a variety of wild plants and fruits including black-eyed susan, elderberry, knotweed, raspberry/blackberry, and a single sunflower seed. One maize cupule was also recovered. Boyd suggests that the 22 Late Prehistoric components form two clusters, one called the Blue Lick Cluster, the other the Elklick Cluster. Each cluster is centered at the confluence of the tributary for which it is named and the Casselman River. Both clusters contain villages, short-term base camps, and short-term camps. The Elklick Cluster also contains an upland hamlet. Both areas were occupied for a similar length of time, from the 10th through the 15th centuries, but the Elklick Cluster had a larger population.

In summary, the investigations of the proposed Meyersdale Bypass have provided us with detailed data on the prehistoric settlement and subsistence systems in a small part of Somerset County. For this study, investigators examined an area covering 5.4 square miles and found 68 archaeological sites. The site density is 1 site per 0.18 kilometers (.07 square miles). If this is an accurate estimate, it appears that the density of sites in the Casselman Valley is significantly higher than that in any of the high-site-density watersheds identified in the survey priorities policy. The site density in the Casselman Valley is significantly higher than the site densities projected for the Crooked Creek Watershed or the area investigated for the Mon-Fayette investigation. Additional studies are needed to confirm these densities and to then find explanations.

Investigations of Other Influences on Site Location

Three other investigations provide insight into factors that may influence the distribution of prehistoric sites in the Conemaugh-Blacklick watershed. These are the location of historic Native American paths, which are presumed to have prehistoric antecedents; the prehistoric exploitation of hollows along the Allegheny Mountains; and the distribution of sources of lithic raw material. Each of these will be briefly described in this section.

Native American Paths

As discussed by Cowin (1981) and Neusius and Neusius (1989), Native American paths are thought to be excellent predictors for the location of archaeological sites. Figure 10 shows the route of the Catawba Path based on Wallace (1965) in the watershed. The Catawba Path ran from Ichsua New York on the upper Allegheny, through Indiana and Uniontown, Pennsylvania, to Morgantown West Virginia, and from there all the way to the Carolinas. It was one of the most important Native American pathways in North America and was known by many names as it passed through Pennsylvania: the Great Catawba War Path, the Iroquois Main Road, the Cherokee Path, and the Tennessee Path. The path, with all of its connections included, extended from Canada to Florida and west into the Mississippi Valley. It was used by Iroquois agents to keep an eye on the international scene and was used by both Iroquois and Cherokee war parties. As Wallace describes the section of route across the Conemaugh-Blacklick watershed:

"The course of the path is not well known and can only be traced where travelers noted its presence where it crossed better-known east-west traders paths. The path can be traced from early maps and oral tradition to have followed this generalized course. …

The path forded Redbank Creek at Brookville, and continued south for a few miles (possibly through Stanton and Worthville), and then veered east to cross Mahoning Creek at what is now Hamilton. The path turned south again to Trade City, Georgeville, and Kellysburg, it crossed Crooked Creek at or near Gaibleton, intersected the Kittaning-Frankstown Path at Shavers Spring (now McElhaney Spring near Indiana State College), passed through Peholand's Town (Homer), crossed Black Lick Creek at what is now Palmerton, veered east to ford the Conemaugh River near Squirrel Hill, and resumed its southwestern course. The path then climbed the ridge overlooking the present Tubmill Reservoir and then passed the site of old Fort Palmer and came into the Indian town of Loyalhanna (Ligonier), where it intersected the Raystown Path.

Fording the Loyalhanna Creek, it ran southwest by way of Pleasant Grove to Stahlstown, and changed to a full west course by present Acme and Lauralville." (Wallace 1965:28; italics added for emphasis)

Myers (1997) has examined the distribution of McFate type sites and artifacts found in association with several Native American paths including the Catawba Path. Although most of these sites are northwest of this study area, two sites he discusses, the Johnston and Squirrel Hill, sites are within the Conemaugh-Blacklick watershed. Site types Myers associates with the paths include open-air campsites, rockshelters, and upland stockaded forts. Myers noted the following associations between paths and site locations. These include:

1. Trail segments are linked by sites in key locations, for example on saddles near drainage divides and stream and river confluences;

2. Temporary sites types like rockshelters and open-air sites are found near the heads of low-order drainages or near confluences. Myers suggests that these drainages provided least-resistance access across divides;

3. Paths frequently follow navigable water routes like the Clarion or Conemaugh Rivers.

Sites in the northern part of the Conemaugh - Blacklick watershed are associated with the Catabwa Path. Other factors, including the abundance of Loyalhanna chert sources, seem to have influenced the location of sites south of the Conemaugh River.

Hollow Exploitation

As discussed in several previous sections, sites are thought to be associated with the heads of drainages or with drainage hollows. Stewart and Kratzer (1989) and Myers (1997) cite these as preferred upland site locations. Stevenson (1982) conducted a survey of five hollows along the Allegheny Front in Centre County. Although the hollows varied in size from 1.11 square kilometers (Williams Run) to 16.23 square kilometers (Dick's Run), sites were found in association with all of the hollows. All had sites at what Stevenson refers to as the entrance to the hollow or the base of the hollow just as it entered the floodplain of a larger stream. In at least one case, the entrance to the Dick's Run hollow, the entrance contained evidence of a long occupational sequence from the Early Archaic through the Late Woodland periods.

Artifact scatters are also found at the top or source area of the hollows. These scatters contain few artifacts and are typically found right at the spring head that forms the drainage or at the convergence of two first-order springs. While it may seem that these light scatters provide little information on the activities conducted at these locations other than that they are present, the Spruce Run Site (36SO153) excavated by Richard George (1984), was found to contain undisturbed evidence of a Late Woodland lithic workshop and associated features.

Lithic Sources

One other important resource that clearly influences prehistoric site distribution is the location of sources for lithic raw material. One type of chert that seems to have strongly influenced the location of archaeological sites in the Conemaugh-Blacklick watershed is Loyalhanna chert. This source and associated workshop sites have been investigated for a number of years by Robert Oshnock, an avocational archaeologist from Westmoreland County (Figure 10). In a recent paper, Oshnock (2000) described quarry areas and workshop sites associated with the Loyalhanna Chert. As he notes, this is the most commonly used local chert in the drainages surrounding the source area including the Conemaugh River, Loyalhanna Creek, and Sewickley Creek. In his paper, Oshnock describes several Loyalhanna sources found on the Derry, Latrobe, and Mammoth 7.5 minute USGS quadrangles. The Derry Quadrangle is partially within the Conemaugh - Blacklick watershed and one of the major sources is located less than 5 miles from the edge of the watershed.

As described in the section on geology above, the origin of Loyalhanna chert is in the Monongahela Group. Most of the quarries exploited floatstone deposits of large blocks at elevations of 1040 to 1100 feet amsl. Sources of Uniontown chert are also found in the Monongahela Group. In his recent paper, Oshnock reports that he has examined the lithics from some 500 sites from within 24 kilometers or 15 miles of the Loyalhanna sources. The predominant type of raw material found at these sites is Loyalhanna chert. All of the sites in this analysis to the north, northeast and northwest of the source are in the Conemaugh- Blacklick watershed. Loyalhanna chert is found in collections from sites as much 42 miles to the east beyond Chestnut Ridge and Laurel Hill. Oshnock (2000) speculates that the headwaters of the Conemaugh itself could have been used as a path to transport chert into the Laurel Highlands, as could the Catawba Path, which crosses the watershed.

The Loyalhanna is considered to be a local chert source and is presumed to have been used in ways that are similar to the patterns of local chert usage throughout the state. However, Oshnock's research provides us with a level of detail about the prehistoric use of this resource that can encourage us to generate new research questions and approaches for the analysis of this material.

Researchers have examined many aspects of lithic reduction in Pennsylvania. Research has focused on defining strategies for tool production by prehistoric knappers, tracking changes in this strategy through time and examining the correlation between different technologies and raw material type. Other studies have looked at curation behavior by prehistoric groups and at changes in reduction strategy with increased distance from a quarry source. Much of this research has been conducted through the analysis of a few sites or quarries that may have been located at some distance from each other. In the Conemaugh-Blacklick watershed, we now know from the sample of recorded sites that we can identify an area where prehistoric groups were attracted to a specific chert resource, so we could design our investigations to concentrate our research on questions of how this resource was used.

A question remains as to whether we have enough sites in this area to investigate the use of the Loyalhanna chert in the existing sample. The Monongahela formation underlies an upland topography. Under the survey priorities model, no additional surveys would be required, so we would add no more sites to his sample. Table 10 presents a summary of the information on known sites that are located on the Monongahela Group geologic formation, the parent formation of the Loyalhanna chert. Of these, nine are single component sites, four are multicomponent sites, and seven lack diagnostics. The single-component sites include no Paleo-Indian or Early Archaic sites, four Middle Archaic sites, one Middle/Late Archaic site, one Late Archaic site, one general Archaic site, one Middle/Late Woodland site, and one general Woodland site. The multicomponent sites contain one Early Woodland component, one Middle Archaic component, two Late Archaic components, two Transitional components, two Early Woodland components, three Middle Woodland components, and two Late Woodland components, in addition to general Archaic and Woodland components. This may not be a large enough sample to be able to answer questions on the use of this resource, since many of these sites may have already been destroyed by the development of surface mines or gas wells.

Table 10: PASS File Sites associated with the Monongahela Group Geologic Formation

Previous Excavations

The research questions developed in the State Plan define the kinds of information that we need to recover from archaeological sites in order to answer a limited set of questions. However, an assumption in the survey priorities model is that most upland sites are multicomponent lithic scatters with little information beyond site age. Is this assumption valid? Is the information to address research issues other than chronology actually available in upland sites? One way to investigate these questions is to examine what has been found in excavated sites, both upland and riverine, to determine if they contain qualitatively different types of information. Using an assumption that published excavation reports in journals such as the Pennsylvania Archaeologist can be used to identify the most important excavations conducted in the region, several students, Barbara Malinky, Emily Griffin, and

Kalynn Yastro, went through back issues of the journal and identified sites from the counties in which the Conemaugh - Blacklick watershed is located. Another source of information on potentially significant archaeological sites is unpublished cultural resource management (CRM) reports.

A review of back issues of the Pennsylvania Archaeologist found descriptions of 22 sites located adjacent to the watershed or in the counties surrounding the watershed. Only two of the 22 excavated site dates to the Archaic Period, six date to the Early or Middle Woodland Periods and the others date to the Late Woodland. Only two of the published sites, the Johnston Site and the Squirrel Hill Site, are located within the watershed. These are not the only sites that have been excavated in the area, but they represent most of the published literature for Indiana, Westmoreland, Somerset and Cambria counties. Table 10 and Figure 11 provide summary information on the sites. Appendix 1 provides a brief summary of the results of each of the excavations.

In addition to the published sites, several others sites listed on Table 11 are described in the "gray literature" or unpublished reports and articles. Four sites in the watershed (36WM277, 36WM278, 36WM282, and 36WM283) were the subject of Phase II investigations as part of the archaeological studies for a surface mining permit area. A data recovery investigation was conducted at 36WM278 as part of this same project. The Mary Rinn Site (36IN29) in Indiana County has been investigated by several researchers from Indiana University of Pennsylvania.

Table 11: A Sample of Excavated Sites from Indiana, Westmoreland, and Somerset Counties (No sites reported from Cambria County)

Types of Data found in the Survey and Excavation Sample

These reports can be used to identify a set of characteristics that many of the sites described above share and that have made them suitable for answering research questions. The characteristics include:

1. Diagnostic artifacts associated with specific temporal occupations, including prehistoric ceramics or other diagnostics associated with Woodland period occupations;

2. Carbonized plant remains, which have produced dates and are associated with diagnostic artifacts in high-quality contexts such as features. These remains can also be subjected to botanical identification to inform the paleo-enviromental reconstruction of the area;

3. Cultural features, which include postmolds that may have been associated with structures and hearths, and may provide information on patterns of social organization;

4. Artifacts associated with a variety of activities including projectile points, which may have been associated with hunting and/or cutting, netsinkers associated with fishing, lithic debitage which resulted from stone tool production, and fire-cracked rock associated with various types of food preparation. Vertical and horizontal patterning suggestive of sites with distinct contemporary activity areas and of changes in resource exploitive strategies through time;

5. Identifiable types of lithic raw material, which can be analyzed to reconstruct prehistoric lithic reduction strategies. Neutron Activation Analysis recently has been shown to be a useful method for identifying chert types and source areas (Chiarulli et al. 2000);

6. Midden or prehistoric trash pits or deposits;

7. Evidence of ceremonialism or ritual activities;

8. Faunal material, including bones and shell, which would aid in the paleo-environmental reconstruction of the site;

9. Human remains.

One expected kind of high-quality information which was not found in the publications examined for this region, but that is know to be available in western Pennsylvania is a series of stratified soil horizons associated with cultural features and diagnostic artifacts.

"Stratified soil horizons" is the one trait that is not present in upland sites, because of the lack of deposition on upland surfaces. Several of the upland sites are single-component occupations or multicomponent occupations in which there are horizontally distinct occupation areas. These sites have also, in most cases been plowed and are currently used for either crops or as pasture. Sites that have not been plowed are often thought to be less disturbed and so repositories of more significant information. However, as shown by these sites, even those that have been plowed may contain substantial quantities of information.

In addition, the archaeological literature has provided us with examples of how even sites that by themselves might not contain enough information to reconstruct prehistoric lifeways can be significant when investigated as part of a localized settlement cluster, as we see in the Meyersdale Bypass investigation. Clearly, the significance of the information contained in any individual archaeological site is only apparent when that information is combined with that from other nearby sites to reconstruct prehistoric chronologies and lifeways.

Research Questions

Archaeological investigations as part of compliance projects are designed to identify, assess, and mitigate sites found eligible for listing on the National Register of Historic Places. Archaeological sites are usually found eligible based on Criterion D: "that they contain information important to the prehistory or history of an area." A guide to the kind of information that may be important for this assessment is found in the Pennsylvania State Plan (Raber 1985), which describes the relevant research questions for each part of the state.

Sites that contain data relevant to any of these topics are likely to be eligible for the National Register. The following provides a brief summary of the relevant questions that were defined in the state plan for each time period for western Pennsylvania. This section summarizes the research questions that are described in the state plan for western Pennsylvania. These were developed from the existing archaeological literature, so are based, in part, on the site information presented above.

Paleo-Indian

The research questions discussed in the State Plan for the Paleo-Indian period relate to specific drainage systems, none of which are part of the Conemaugh-Blacklick watershed. The drainages include: Chartiers Creek in Washington and Allegheny Counties, the Casselman River in Somerset County, along the Little Mahoning and Plum Creek in eastern Armstrong and Indiana counties, Connoquenessing Creek in Butler County, near glaciated features on the Glaciated Allegheny Plateau, near Waterford in Erie County in the French Creek drainage, on the glacial outwash terraces along the Upper Allegheny in Warren County, and along the Lake Erie Escarpment along Conneaut Creek.

However, given the rarity and significance of Paleo-Indian occupation in Pennsylvania, it is likely that any Paleo-Indian site with moderately good integrity would be considered eligible for the National Register.

Archaic

The research problems listed for the Archaic period are less specific and not related to specific geographic areas. Instead, they refer to broad questions of resource exploitation and settlement pattern. The State Plan lists the following topics and types of data that would be appropriate for investigations of the Archaic period and include:

Chronology, which can be investigated through the excavation of stratified floodplain sites to answer questions about the transition from Paleo-Indian to Early Archaic and Late Archaic to Transitional or Early Woodland adaptations

Settlement patterns, from the distributional pattern of Archaic period artifacts in relation to topographic features.

Subsistence strategies, which could be identified with data on paleoenvironmental exploitation to identify Archaic procurement strategies.

Cultural identity, through the definition of regional Archaic artifact typologies. The State Plan recommends that stratified multicomponent sites be excavated to resolve this problem.

There are other questions about the relationship of the Transitional Complex from eastern Pennsylvania to western Pennsylvania Late Archaic cultures that could be addressed through the investigation of sites dated to the Archaic period. A better understanding is needed of this adaptation, which seems to be primarily located along river floodplains.

Early Woodland

The State Plan defines three research questions for the Early Woodland Period. These include the investigation of :

Chronology including the definition of "Forest Notched" points. Excavations to obtain artifact samples and radiocarbon dates are needed to define the period.

Cultural identity, through a distributional study of Adena. Information is needed to distinguish the characteristics of the Meadowood, Lagoon, Forest Notched, and Rossville types.

Subsistence strategies, through the recovery of paleoenvironmental information and radiocarbon dates. Early Woodland sites with well-defined cultural features are needed to answer these questions.

Middle Woodland

Research questions for the Middle Woodland period are similar to those from other periods in that questions of regional variation and subsistence strategies predominate and include investigations of:

Cultural identity through the identification of regional components and the definition of artifact types.

Subsistence strategies to define the transition from Middle Woodland to Late Woodland agricultural populations.

Settlement patterns, through the discovery of habitation sites and other unique site types. Such sites are identified in the regional plan as eligible for the National Register.

Late Woodland/Late Prehistoric

Since these occupations are often large villages with intact features and well-preserved archaeological data, a number of sites have been previously excavated. This past research provides a basis for more sophisticated research questions to be asked about the period, although, because of the frequent presence of human burials, all substantial occupations are probably eligible for the National Register. The research topics identified for this period in the State Plan include investigations of:

Cultural identity, through the collection of data that can be used to better define the distinctions between groups in the Upper Allegheny River drainage relative to the West Branch of the Susquehanna River. Excavations of village sites are needed to identify internal village arrangements, for example, those with circular versus those with long house construction.

Cultural dispersion, indicated by the arrival of Contact period groups from eastern Pennsylvania, which can be answered only by the discovery and excavation of sites from this period.

The research problems identified in the State Plan provide a framework for the evaluation of sites recovered from any particular project. Any site that contains specific data relative to one of these questions is likely to be considered significant and eligible for the National Register. The State Plan also defines site types that are likely to contain significant information. Stratified floodplain sites with evidence of several occupation layers may provide information relevant to several research questions and so are considered to be significant. Even less substantial sites contain information that is important as defined in the State Plan. For several of these periods, the distribution of artifact types by topographic setting is identified as information that would be important for our understanding of settlement patterns and cultural identity. Even multicomponent surface sites may provide information related to this research topic.

There are other factors that affect the determination that a site is eligible for listing on the National Register. One of these is site integrity, which has to do with the physical condition of a site. Integrity refers to the degree to which a site has been disturbed. Even if a site contains relevant data for one of the questions listed above, it may not eligible for the National Register if it contains mixed materials from several cultural periods.

The next section of this report contains an examination of the PASS file data for the Conemaugh-Blacklick watershed. One aspect of the database that will be examined is whether the existing recorded upland sites contain information that can be used to address the research questions described above on chronology, cultural identity, and subsistence strategies.

2.3 The PASS File Database for the Conemaugh-Blacklick Watershed

Table of Contents

Top of Page