Paleoindian Mobility Ranges Predicted by the Distribution of Projectile Points Made of Upper Mercer and Flint Ridge Flint
A thesis submitted to Kent State University in partial fulfillment of the requirements for the degree of Masters of Arts
by
Amanda Nicole Mullett
December, 2009
Thesis written by Amanda Nicole Mullett B.A. Western State College, 2007 M.A. Kent State University, 2009
Approved by
______, Advisor Dr. Mark F. Seeman
______, Chair, Department of Anthropology Dr. Richard Meindl
______, Dean, College of Arts and Sciences Dr. Timothy Moerland
ii
TABLE OF CONTENTS
List of Figures ...... v
List of Tables ...... v
List of Appendices ...... iv
ACKNOWLEDGEMENTS ...... vi
Chapter I. Introduction ...... 1
II. Background ...... 5 The Environment...... 5 Early Paleoindian Complexes: Clovis and Gainey...... 15
III. Models for Early Paleoindian Mobility...... 20 Model 1: The Lithic Centered Model...... 21 Model 2: The Logistical Forager Model...... 23 Previous Research on Mobility...... 24 Tests for Models...... 25
IV. Assumptions and Test Expectations...... 29
V. Materials and Methods...... 33 Materials ...... 33 Methods...... 35
VI. Resharpening Results ...... 40 Statistical Results ...... Graphical Results...... Spatial Results......
VII. Supply Range Results ...... 55
iii
iv
VIII. Discussion, Conclusions, and Suggestions for Future Work ...... 71
Bibliography ...... 79
List of Figures
Figure 1 Lithic Outcrop Map of Ohio 10 Figure 2 Examples of Upper Mercer Flint 11 Figure 3 Examples of Flint Ridge Flint 12 Figure 4 Distance-Based Supply Range: UM 14 Figure 5 Distance-Based Supply Range: FRF 14 Figure 6 Sketches of Clovis Points 15 Figure 7 Sketches of Gainey Points 17 Figure 8 Seeman Supply Range Model 26 Figure 9 Anderson and Hansen Supply Range Model 27 Figure 10 Upper Mercer Length/Distance Bar Graph 42 Figure 11 Upper Mercer Width/Distance Bar Graph 43 Figure 12 Upper Mercer SI/Distance Bar Graph 44 Figure 13 Flint Ridge Flint Length/Distance Bar Graph 46 Figure 14 Flint Ridge Flint Width/Distance Bar Graph 46 Figure 15 Flint Ridge Flint SI/Distance Bar Graph 47 Figure 16 UM Average Length Interpolation 49 Figure 17 UM Average Width Interpolation 50 Figure 18 UM Average SI Interpolation 51 Figure 19 FRF Average Length Interpolation 52 Figure 20 FRF Average Width Interpolation 53 Figure 21 FRF Average SI Interpolation 54 Figure 22 UM Density Distribution 56 Figure 23 UM Supply Range 57 Figure 24 UM Line Profiles 59 Figure 25 FRF Density Distribution 62 Figure 26 FRF Supply Range 63 Figure 27 FRF Line Profiles 65 Figure 28 Combined Density Distribution 69 Figure 29 Combined Supply Range 70
v
List of Tables
Table 1 Upper Mercer Statistical Results 41 Table 2 Flint Ridge Flint Statistical Results 41
List of Appendices
Appendix A Description of Other Flints 87 Appendix B West Virginia Point Data 89
vi
AKNOWLEDGEMENTS
First and foremost, I would like to offer my appreciation for the extensive time, effort, and encouragement that my thesis advisor, Dr. Mark Seeman, offered me throughout my career as a Master’s Student in the Department of Anthropology at Kent State University.
Dr. Seeman spent countless hours reviewing and editing my thesis, as to ensure an excellent product. He also assisted in the process of locating and identifying dozens of projectile points from areas not well represented in my original dataset. I would also like to thank both of the other members that were on my committee, Dr. Kam Manahan (Kent State- Anthropology), and Dr. Jay Lee (Kent State- Geography). The suggestions made by both of these members during my defense were extremely helpful and added quality to the final draft of this thesis.
Next, I would like to thank my friends and colleagues (both inside and outside of the
Anthropology Department). They provided much needed encouragement and support throughout the past few years. I should also thank my family for allowing me to chase my dreams, even if it means being a student until I am thirty! Mom, Dad, Mama, Angie, Krissi,
Quentin, Lily, and Max, you will never know how much you all mean to me, and I owe any success I have to each of you!
vii Chapter I.
Introduction
Homo sapiens sapiens has inhabited North America for over 12,000 years. The first
North Americans entered the continent by crossing the Bering Land Bridge connecting
Siberia and Alaska during the Wisconsinan glaciation. Equipped with the same mental and physical capabilities as other modern human populations, the early colonizers of the western hemisphere spread quickly across the landscape while perfecting the techniques required to be successful in their new environments. The evidence for the exploitation of the Great Lakes
Region (the geographic focus of this thesis) has been dated to about 11,500 years ago
(Anderson 1990; Anderson 1995:9). These first settlers of North America are termed
“Paleoindians”, and North American archaeologists designate the time period between
11,500 BP and about 10,900 BP as the early Paleoindian period (Anderson 1990:4, Morse et al. 1996). Different „culture groups‟ succeed one another throughout this period as evidenced by radiocarbon dates and different assemblage characteristics. “Clovis” populations were the earliest, and they were distributed across the entire North American continent. Depending on location in North America, either Folsom or Gainey groups follow Clovis populations in time. Folsom groups are limited mostly to the Great Plains and the Rocky Mountains
(Haynes1964). Gainey extends across the Great Lakes Region and adjacent areas of the Ohio
1 2 and Mississippi Valleys (Morrow and Morrow 2002:142). Regardless of group classification, early Paleoindians were hunters and gatherers, exploiting a vast landscape filled with a variety of organic and inorganic resources. The questions asked by many archaeologists center on determining how these people went about acquiring these resources. Was their lifestyle and mobility patterning determined largely by the migration of animals, the location of unchanging lithic raw material sources, or something else?
The long-term practice of exploiting two high quality lithic raw materials in Ohio by early Paleoindians provides a useful context for examining questions about the first human inhabitants in North America.
It is the assumption of many archaeologists that lithic supply zones are directly related to the relative mobility of early Paleoindians (Carr 2005:26; Anderson and
Hanson 1988:267; Gardner 1977). As a result, the lithic supply range developed and described in this thesis will be compared to two mobility models. The first model, in which flint outcrops were at the center of early Paleoindian ranges, can be traced at least to the early works of Gardner (1977). The second introduces other natural resources as variables affecting Paleoindian movement across the landscape, and follows more directly the research of Kelly and Todd (1988). The degree to which the Ohio data supports one of these models over the other should clarify a key dimension of the technological organization in these early hunter-gatherer societies.
Within the context of this thesis, a number of assessments were completed to illustrate relationships within the data. More specifically, statistical correlation tests were used to examine the relationship between the location of a projectile point and its quarried source, and also the extent of sharpening on a given point. Bar graphs and line 3 profiles were created to help visualize the relationships between distance and resharpening. Finally, GIS spatial interpolations were created in ArcMap to provide a directional estimate of expected densities and expected resharpening characters for the study area. Results from these tests seem to fit one mobility model better than the other.
Early Paleoindians are thought to be among the most highly mobile hunter- gatherer groups ever (Meltzer 1984:1, Seeman 1994:273). These early inhabitants of
North America exploited migratory big game, and specialized in the production of large, durable bifaces and projectiles (Bradley 1993). Over time, these projectiles were discarded, leaving a trace of the group on the ground. By looking at the extent of reuse and resharpening on these points, as well as the relative density of specific raw materials across the landscape, testable results become available and can be compared to the previously mentioned mobility models. The size and shape of the resulting mobility range allows for conclusions regarding the effects that different resources and landforms have on the distribution of these early groups.
The organization of this thesis is as follows: first, there will follow a chapter
(Chapter II) providing background data on the problem, including a discussion of the environmental context and description of Clovis and Gainey phase characteristics. Next, an explanation of the two mobility models will be provided in Chapter III, along with a discussion on previous mobility research, and a description of tests for the two models.
Following this section will be an explanation of certain assumptions associated with this work, and a detailed description of hypotheses for each mobility model, as well as hypotheses for the relationship between the use of Upper Mercer Flint and Flint Ridge flint (Chapter IV). Next, will be a discussion on the materials and methods used in this 4 thesis (Chapter V), followed by an explanation of the results for resharpening data
(Chapter VI), and a section describing the results of the supply range data (Chapter VII).
Finally, a discussion and conclusions section will bring the results together in order to support or refute the mobility models suggested in the earlier sections.
Chapter II.
Background
Before the mobility models and methodology can be discussed in detail, it is important to understand the background information that bears on early Paleoindian studies in the Ohio region. A brief discussion of the environment, therefore, will include details on the distribution of vegetation, fauna, and lithics that were available at the time of Paleoindian exploitation. Next, a description of the two early Paleoindian culture groups used for this mobility research will be provided. The information will converge to provide a context or background for the models under examination.
The Environment
Environment is a key dimension in understanding early Paleoindian cultural patterns.
Various explanations have been offered in attempts to clarify our understanding of the environment of North America at the end of the Pleistocene. At the time of early
Paleoindians, a rise in temperatures initiated a final retreat of the glaciers from the
Wisconsinan maximum. By 12,000 BP, paleoclimatologists and archaeologists place the ice sheet edge anywhere near northern Michigan. There are differing interpretations regarding the exact distributions for vegetal and faunal resources 12,000-10,000 BP, but is clear regardless of particular investigator that this was a time of dramatic environmental change
5
6
with shifting and difficult-to-predict subsistence resources (Ellis et al.. 1998). The Ohio environment was certainly colder and more unpredictable than today (Ellis et al. 1998). This unpredictability is important to note because of the limits it would have placed on early
Paleoindian organizational characteristics. In contrast to the changing vegetal and faunal distributions, the location of needed lithic materials were static and unchanging, and present day descriptions can be assumed to cover what was there in the past.
Distribution of Vegetation
Suggestions for the distribution of different vegetal types across Ohio are commonly supported by studies of pollen records. These pollen records reveal more information about the regional environment than local habitats. Ellis et al.. (1998) note that the Great Lakes region was most likely comprised of a closed boreal forest. McDonald (1994) notes that Ohio most likely consisted of forests containing spruce and pine, and others agree that the greater region was spotted with spruce parklands (Meltzer 1988). At the Pleistocene-Holocene transition, the regional environment was characterized by frequent variations in vegetation and climate (Shane 1994). More specifically, pollen analyses from areas in Ohio and Indiana imply an oscillation from conifer forests, just before early Paleoindian occupation, to oak dominated forests around 10,000BP toward the end of the Paleoindian period(Shane 1994:7).
Pollen records are difficult to interpret at this period of transformation because of the complexity and speed of changes (Shane 1994:12-13). At the beginning of the period, spruce, pine, latch, and fir pollen rapidly increase. After about 200 years of this trend, an abrupt decrease is noted for both spruce and pine populations. Finally, the decrease is met with local extinction of some spruce forests(Shane 1994).
7
Distribution of Fauna
Two opposing views exist regarding patterns of early Paleoindian faunal exploitation.
On one side, investigators believe that these early hunters were exploiting primarily big game
(Todd et al.. 1990:815; Frison 1998; Bolakrishnan et al.. 2000), while others believe that they adapted to their changing environment by using a generalized forager approach (Meltzer
1993; Byers and Ugan 2005:1624; Dincauze 1993). Evidence for specialized big game exploitation begins with the technology, but is also supported by a few studies that place big game on the landscapes used by early Paleoindian groups (Cannon and Meltzer 2003). In the absence of big game availability, early Paleoindians would certainly resort to foraging the landscape for food (Meltzer 1988; Meltzer and Smith 1986), but the majority of archaeological evidence seems to support the view of a strong orientation towards the hunting of large animals in the Great Lakes region.
At the time of early Paleoindians, several key faunal resources were available. These vertebrates included mammoth, mastodon, caribou, elk-moose, bear, moose, elk, flat-headed peccary, deer, and possibly bison (McDonald 1994, Holman 2001). Faunmap provides evidence for the appearance of Jefferson or woolly mammoth in areas surrounding Ohio at this time. Estimates for extinction of the majority of the Pleistocene megafauna date to around 11,000BP (Morse et al.. 1996). Faunal studies by archaeologists at Nobles Pond indicate a concentrated presence and exploitation of caribou in Ohio and nearby regions
(Seeman et al.. 2008). Storck and Spiess also note the presence of caribou in the Great Lakes region during this period (1994:121). Research at several sites in and surrounding Ohio support the concept of big game availability in the study area at the time of early Paleoindian 8
colonization. If present, it makes sense that they would have been hunted due to their caloric return.
Distribution and Description of Lithic Sources
Despite the assorted views on the vegetal and faunal resources available for early
Paleoindians in Ohio, it can be assumed that high quality lithic outcrops were spatially dependable, and plentiful at the end of the Pleistocene. Such outcrop sources were critical to the regular production of large stone spear points and other tools. Two lithic resources from central Ohio were repeatedly used during early Paleoindian times. The first of these is the black and blue-black Pennsylvanian Upper Mercer flint from Coshocton County (Carlson
1991:213). The second early Paleoindian lithic resource is Flint Ridge flint, which outcrops in Licking County and is associated with the Vanport member of the Allegheny formation
(Carlson 1991:214). Early Paleoindians would utilize other, lower quality raw materials if the situation arose, resulting in the less intensive dispersion of tools made of these other identifiable flints across the study area. Several other flint outcrops exist in Ohio, but were seldom used, and will only be discussed briefly in Appendix A (Figure 1). Other high quality raw material outcrops from areas immediately surrounding Ohio, and are of comparable quality to Upper Mercer and Flint Ridge flint were also utilized with regularity. These include, but are not limited to Wyandotte, Attica, (both from Indiana) and Onondaga (from
New York). These flints also appear throughout the projectile point dataset developed in this thesis as raw material types, and will be described in Appendix A. Again, Upper Mercer and
Flint Ridge flints are expected to be the dominant flint types exploited in the Ohio area during early Paleoindian times (Prufer and Baby 1963). The primary advantage of these two raw materials for mobile hunter-gatherers results from the fact that both of these materials 9
can be obtained in the large, relatively homogeneous pieces. This is a distinct advantage in consistently producing large bifaces with considerable resharpening potential and hence extended use-lives (Tankersley 1989). An historical and physical description of the two flints most directly related to this thesis follows.
Upper Mercer Flint
Early Paleoindians used Upper Mercer flint intensively (Prufer and Baby 1963).
Three hundred and twenty million years ago, this flint was deposited throughout the Upper
Mercer Limestone formation (Stout and Schoenlaub 1945, Luedtke 1992). This limestone formation stretches from Hocking County to Coshocton County in a 70-mile northeasterly belt (Carlson 1991:213). Upper Mercer flint outcrops in multiple forms, including nodules, lenses, thin single beds, and thick, layered sheets (Luedtke 1992). 10
Figure 1. Map illustrating the distribution of chert outcrops in Ohio.
11
The actual flint ranges in color from black and dark grey to light grey and light brown Stout and (Schoenlaub 1945,
Luedtke 1992). Upper Mercer flint varies in quality, and as the quality increases, the material becomes denser and less translucent (Carlson 1991:213). Luster for
Upper Mercer flint ranges from medium to shiny, depending on the outcrop form and location (Leudtke 1992). In many cases,
Upper Mercer flint contains vein- like features composed of siliceous quartz and chalcedony crystals inundated these fractures creating milky white bands (Leudtke
1992)(Figure 2). Evidence for the extraction of Upper Mercer flint by early Paleoindians is supported by the presence of large, circular quarry pits in Coshocton County, and also by the high density of early Paleoindian manufacturing debris in the region (Carlson 1991,
Converse 1973). Pits were dug in order to access the thick, large sheets of Upper Mercer flint. Prufer and Baby (1963) found that 49% of Ohio fluted points for which raw material could be identified was made of Upper Mercer flint.
Flint Ridge flint
Flint Ridge flint is another distinguishable high quality raw material from Ohio. With the majority of outcrops in Licking County, Flint Ridge flint has quickly become one of the most admired raw materials by modern day flint-knappers. Flint Ridge flint was deposited 12
during the middle Pennsylvanian period as part of the Vanport member of the Allegheny formation (Stout and Schoenlaub 1945, Converse 1973). This flint also forms a northeasterly tending belt, extending from Zanesville to Newark, Ohio (Mills 1921). Deposits of Flint
Ridge are predominantly in sheet form, and average at about 1.2 meters thick (Leudtke
1992). The flint bands are dispersed throughout larger grey shale and limestone deposits
(Stout and Schoenlaub
1945). Vibrant colors offset
the flint from the dull
surrounding limestone. In
fact, Flint Ridge exhibits a
variety of coloring,
including purple, blue, pink,
red, green, browns, and
yellows (Figure 3)(Leudtke
1992). Nether’s flint, a
variety of Flint Ridge, is
distinguished by its rigid ribbon or tiger striping (Converse 1973). It appears that one of the only consistent features of
Flint Ridge is its muted luster, and the sparse presence of tiny fossils. Pockets of quartz and chalcedony are also present in Flint Ridge (Converse 1973). The translucence of the raw material ranges from semi-opaque to semi-translucent (Leudtke 1973). The distribution of hundreds of round pits also provides evidence for prehistoric use of this raw material on a large, isolated highland in Licking Co., Ohio known as Flint Ridge (Mills 1921). Converse 13
1973). Historically, early settlers used lower grade Flint Ridge in the formation of millstones.
Flint Ridge flint is the second most common identified raw material utilized by Paleoindians in Ohio (Prufer and Baby 1963).
Conclusions on Lithics
Although Upper Mercer and Flint Ridge flint were used in the manufacture of many projectile points from the study area, by no means did they completely dominate the dataset.
Many other high-quality flint types were used with high intensity, especially in Pennsylvania,
Kentucky, Illinois, and West Virginia. The locations of these other flint outcrops can be noted in Figures 4 and 5, and were used to construct a six-sided Thiessen polygon that illustrate an estimated supply range based on least-cost availability for Upper Mercer or Flint
Ridge flints. This area was calculated by marking the half way points between the Ohio outcrops and other high quality flint outcrops. In other words, hypothetical supply ranges for the Ohio materials were constructed based on a least-cost basis as a starting point for assessing early Paleoindian lithic resources. It is important to note that these polygons for
Upper Mercer and Flint Ridge flint across most of Ohio. Because these were the two most heavily utilized raw materials for early Ohio Paleoindians, they constitute the foci for this thesis investigation of early Paleoindian mobility. 14
Figure 4. Distance-based supply range for Upper Mercer flint.
Figure 5. Distance-based supply range for Flint Ridge flint.
Figure 5. Distance-based supply range for Flint Ridge flint. 15
Paleoindians Complexes: Clovis and Gainey
A final aspect of background to consider for the mobility models in this thesis is the nature of basic early Paleoindian technology, organization, and typology. The Clovis and
Gainey complexes or phases were the main early Paleoindian manifestations in Ohio. A clear distinction between the two early groups is difficult to make. There are multiple similarities
between the groups, and the differences seem to be
temporally based rather than ingrained in cultural
distinctions. In sum, Clovis and Gainey represent a
relatively smooth evolution or continuum from early
to late, and many assemblages or specific diagnostics
appear to occupy points on this continuum rather
than sorting neatly into separate taxons.
Many archaeologists agree that Clovis
populations were in fact the initial colonizers of
North America. Forty years of radiocarbon dating
Clovis sites provides evidence for their existence
from 11,500-10,900 BP (Waters et al. 2007). The
spatial extent of Clovis includes the whole of North
America (Waguespack and Surovell 2003). Clovis
groups are thought to have been highly mobile, low
density, big game hunters. Many Clovis artifacts
have been found over 300km from their raw material source (Barrish 1995). A majority of the kill sites related to Clovis groups provide evidence 16
for a reliance on big game, such as mammoth, mastodon, or bison antiquuus (Webb et al..
1984; Grayson and Meltzer 2003; Frison 1998). Additional Clovis site types include small, temporary camps and artifact caches (Goebel et al.. 2008:1500). High mobility encouraged a small, overbuilt tool kit which included projectile points, blades, heavily ground late stage bifaces, end and side scrapers, drills, bone foreshafts, perforators, and gravers (Haynes 1964,
Haynes 1982, Stanford 1991, Meltzer 2004). The lack of plant processing tools further supports the interpretation that the focus of Clovis groups on the processing of animal products rather than vegetation.
Clovis projectile points can be sorted from other projectiles based on a few main characteristics (Figure 6). Again, it is important to note that early Gainey points bear similarities to later Clovis points, and in these situations, distinctions are difficult to provide.
The following characteristics are noted for ‘perfect’ Clovis points, or those which would automatically fit into the Clovis category. Clovis points are fashioned from large, thin bifaces. Morrow and Morrow (2002:150) note that initially, the preform of a Clovis point is relatively thick when compared to Gainey point preforms. Outre passé techniques were used to remove large flakes that stretched across the center axis of the point’s preform. There is a slight indentation or concavity present at the base of Clovis points. Distally from the base,
Clovis flutes are generally 1/3 to 1/2 the length of the biface. In some cases, multiple flutes were executed on a single side (Morris et al..1999). The lateral margins of Clovis points are generally straight, but in some cases they are slightly convergent toward the base (Morrow and Morrow 2002:147; Morris et al..1999; Frison 1998). Additional flaking may occur post- flute in order to properly thin the base and edges of the points (Morrow and Morrow 17
2002:147). Lateral grinding is also present on Clovis points, and probably facilitates hafting the stone point to a spear shaft or a bone foreshaft.
In contrast to the widespread distribution of Clovis sites, evidence for Gainey groups is limited to areas surrounding the Great Lakes. It is suggested that the Gainey phase directly follows the Clovis phase in the east, and dates to approximately 11,000 BP. Deller and Ellis (1988) more specifically place the phase from 11,000-
10,700BP. It is likely that the environmental change around this time forced Gainey groups to focus on hunting resources that were different from the preceding Clovis populations. The importance of caribou at this time has been particularly emphasized (Seeman 2008). Some archaeologists suggest that a less restricted subsistence may have lead to a slightly smaller mobility range (Andrefsky 1994). Barrish notes that most Gainey artifacts are located within
300km of their raw material source (1995).
Toolkits for Gainey groups are very similar to those for Clovis groups, but they lack the blade technology. Typical tool types for Gainey include projectile points, heavily ground late stage bifaces, end and side scrapers, and gravers
(Gibbon and Ames 1998:307). Again, a focus on animal processing is evident. 18
Gainey points are similar to Clovis points, and general differences are difficult to specify. In comparison with Clovis points, those typed as Gainey are smaller, have relatively long flutes, and retain a deeper basal concavity (Figure 7). Outre passé flaking is not present with Gainey points. A majority of Gainey points show stronger evidence for flakes that extend to the midline, creating a lenticular cross-section (Frison 1998). With the exception of these two factors, the similarities between reduction techniques for Clovis and Gainey points are strong.
Differentiating between Clovis and Gainey assemblages is difficult, if not impossible in some cases. When distinguishing between the two groups, context is important. The presence of the blades always indicates a Clovis assemblage. It is important to note that the use of Ohio landscape and other Midwestern regions seems to differ from the Clovis to the
Gainey phase. Large, aggregate camp sites are typically absent from the archaeological record when speaking about Clovis groups in the Midwest and along the East Coast. These aggregated sites appear with more intensity and frequency as we move temporally into the
Gainey phase, with sites like Nobles Pond, the Gainey type site, and Sandy Springs, Hawk’s
Nest, and Bostrom. When looking at isolated Clovis or Gainey points, it becomes clear that the separate phases are more a part of a multi-dimensional ‘spectrum’, so some points can only be described as more Clovis, or more Gainey than the others. With the introduction of the effects of resharpening depletion on point morphology, Clovis and Gainey points become even less distinguishable from one another (Morrow and Morrow 2002:155). Generally, early
Paleoindian points with shorter flutes and more shallow basal concavities can be typed as
Clovis; and points with longer flutes, and deeper concavities can by typed as Gainey. Either way, the two groups were highly mobile, low in population density, and probably specialized 19
in hunting for subsistence. Paleoindian complexes in the area using Barnes, Crowfield, and
Holcomb points follow these two earliest forms in the regional sequence (Justice 1995:24,
25)
Chapter III.
Models for Early Paleoindian Mobility
Mobility and settlement patterns speak volumes about a given culture’s adaptation.
Binford (1980) was one of the first investigators to show that mobility affects how hunter- gatherers organize themselves for resource exploitation, and his findings were later supported and expanded upon by Kelly (1995). Specifically, hunter-gatherers can be expected to show considerable variability in mobility patterns, taking into consideration both distance separating respective encampments and the frequency of moves between sites (Binford 1980;
Smith 2003). These varying strategies can be found to occupy different points along a spectrum from “foraging” adaptations to “collecting” adaptations. Foragers have high residential mobility, and they move their people to resources (Binford 1980). Conversely, collectors have high logistical mobility, and send foray groups to obtain resources across a considerable area, and then bring them back to stable home bases (Binford 1980). These same basic concepts can be applied to specific archaeological situations in order to best understand the particular patterns exhibited by ancient groups.
It has been argued that early Paleoindians were among the most mobile hunter- gatherer populations ever (Seeman 1994:273). However, there is considerable disagreement regarding exactly the form the high mobility takes with early Paleoindian populations in the
20
21
Great Lakes area. Two explicit models have been formulated and supported with varying degrees of success.
Model One: The Lithic-Centered Mobility Model
Closer to the collector end of the spectrum presented by Binford, the first explanation of early Paleoindian mobility emphasizes the extreme importance of lithic outcrops. It is well accepted that the stone tools used by early Paleoindians required the use of high quality raw materials (Bradley 1993; Dincauze 1993). Further, there is a general lack of evidence supporting the concept of extensive lithic exchange at this time (Ericson and Baugh 1994:4), and consequently, it is possible that the distribution of early Paleoindian tools of particular raw materials correlates with the distribution of early Paleoindian groups. It has also been suggested by many that early Paleoindians were more likely to exploit a single lithic resource, rather than moving among a variety of outcrop sources for resupply (Meltzer 1984).
With these factors in mind, it seems reasonable to conclude that the mobility of early
Paleoindian groups was tethered to single, high quality lithic resources. Multiple researchers have concluded that their own particular datasets show this close spatial correlation of a given group with its outcrop (Smith 1990; Gardner 1983; Carr 2005:11). These ideas also are supported by the higher density of Paleoindian sites near flint sources. When looking at the assemblages at these sites, local (quarried) lithic raw materials tend to dominate in frequency.
These patterns generally occur in areas that are densely packed with other necessary resources (Goodyear 1979). Smith‟s (1991) example of the Wyandotte quarries of south central Indiana fit this description quite well, as they are situated at the intersection of multiple resources (1990). This “lithic-centered forager” model will be tested to see if the 22
early Paleoindian mobility strategies were tightly tied to the Upper Mercer and Flint Ridge flint quarries.
Mobility studies based on lithic distributions are not limited to the early Paleoindian time period. In fact, many archaeologists have evaluated the distribution of prehistoric groups in relation to their major lithic resources. Many, and perhaps most, of these studies have concluded that the proximity to lithic outcrops is central in establishing patterns of mobility and/or band ranges (Smith 1990). Many of these studies reconstruct band ranges with finite boundaries with little or no gradation. Archaeologists have simply drawn boundary lines on maps to approximate the boundaries in question. This seems unrealistically simplistic given observed ethnographic patterns of hunter-gatherer behavior and given the fact that even when restricting discussion to Clovis and Gainey we are dealing with several hundred years of time. Under Model One, groups reliant on a single lithic outcrop are considered to be
„tethered‟ to specific resources. For example, Carr (2005:8) looked at sites surrounding the
Hixton Silicified Sandstone (HSS) outcrops at Silver Mound in Jackson County, Wisconsin, and calculated the percentage of the HSS at each site. He found that 70% of the late
Paleoindian points dropped or discarded near the HSS outcrop were made of HSS. Moving farther afield, Carr noted that the shape of the HSS supply zone was largely determined by the availability of other high quality lithic sources. Thus, the supply zone of HSS stretches further south due to a lack of availability of other high quality raw material outcrops when compared to its distributions to the east and west. Altogether, this lithic supply zone was present in areas up to 150km away (Carr 2005: 23). The locations of other high quality lithic outcrops should have an effect on the distribution and condition of early Paleoindian points made from the two Ohio raw materials studied in this project. 23
Model 2: The Logistical Forager Mobility Model
The second model to be tested by this thesis is what I will call the “logistical forager” model. This model expands on the previously listed lithic-primacy model and includes other variables as determinants of settlement. Variables considered in the past include landscape characteristics, proximity to faunal resources, and proximity to rivers (Gardner 1974, Haynes
1980, Smith 1990). High river terraces provided early Paleoindians with multiple advantages, including access to water and fauna, as well as a larger view-shed (Gardner 1974; Haynes
1980). Smith (1990) notes that the areas with evidence for repeated game presence would also be the places with higher percentages of early Paleoindian sites. Access to other high quality raw materials might have an affect on the distribution of early Paleoindians around a raw material source. Seeman (1994) notes that mobility ranges are largely determined by the accessibility of three things; food, lithic raw materials, and shelter. Fluctuations in time or space in the availability of any one of these three characteristics will lead to a different patterning of groups across the landscape (Seeman 1994:273). Consistency of the lithic assemblage from the Nobles Pond site in Stark County, Ohio suggests a reliance of the group on Flint Ridge and Upper Mercer flints, which outcrop more than 70km from the site. If the logistical forager model is correct, we should expect to see evidence of the effects of these other resource considerations on the size and shape of lithic supply zones that would be considerably different than the expectations for quarry-centered logistical foraging. If the food resources or shelter are not in the area of lithic raw materials, the occupation sites should be somewhere in between each of the exploited resources. Sites like Nobles Pond would seem to provide some support for this second model because this large site is reasonably far from the primary lithic resources. 24
Previous Research on Mobility
Given their great antiquity, the material remains in the archaeological record for early
Paleoindians in eastern North America are usually limited to lithic artifacts. Luckily, many of these artifacts can be linked to a specific outcrop source. In the past and as discussed in an earlier part of this thesis, lithic raw material supply zones have been assumed to act as an indicator of group mobility or home range; with the assumption that most groups will stay tethered to one reliable raw material source and exploit it recurrently (Carr 2005:1; Anderson and Hanson 1988:274; Smith 1990). In addition to delineating supply zones based on the spatial distributions of diagnostic fluted points, the size of these artifacts and their relative degree of resharpening also have been used to distinguish home range characteristics
(Morrow 1997).
There have been many studies attempting to understand the relationship between the degree of resharpening and the distance a tool is from its raw material source at the time of discard. The act of resharpening increases the efficiency of tools that are difficult to make
(Jones 1980). Generally, all Clovis and Gainey points can be assumed to be large tools with relatively standardized proportions at the time of initial manufacture. Changes in the relationship between length and width of a point can therefore be assumed to be largely the result of resharpening as the tool is carried through the history of its use-life (Buchanan
2006). As distance from raw material source increases, evidence for resharpening should also increase as the relative advantage of having a smaller tool increases relative to the cost of making a new one (Morrow 1997). The shape of the fall-off-curve, illustrating the relationship between distance-to-source and degree of resharpening, is also important to consider. Interpretations of this extinction curve by pulses of larger tools some distance from 25
the quarry should be taken as evidence of redistribution, caching, or some other mechanism of moving large points out onto the landscape far distances from the quarry in an undepleted state (Renfrew and Bahn 1991, Brown et al.. 1990).
Most of the previous studies focusing on mobility patterns in the Midwest study lithic assemblages from solitary sites. In these cases, the degree of mobility is determined by calculating the distance of the lithic artifact from its raw material origin. Sites with mainly exotic materials have been interpreted as being produced by groups with high mobility, and sites comprised mainly of local raw materials accommodated groups with low mobility
(Smith 1990; Meltzer 1994). This thesis will take a more comprehensive approach to the problem. Below are listed some expectations for varying patterns of early Paleoindian mobility.
Tests for Mobility Models
Different expectations can be brought to the forefront if we consider previous research on supply zones and mobility ranges in somewhat more detail. This thesis intends to use two methods to test the previously discussed mobility models. Several researchers have prepared estimates for mobility ranges based on the dispersion of lithic artifacts resources about the landscape based on their raw materials (e.g. Carr 2005; Anderson and Hanson
1988). Others have examined the extent of resharpening characteristics in relation to the distance from their raw material outcrop (e.g. Morrow 1997). Both of these concepts will be used to provide a more quantifiable understanding of early Paleoindian behavior.
Extensive resharpening usually indicates the need for extending the use life of stone tools. As the previously mentioned research indicates, high-quality lithics were important to the success of early Paleoindians, and as distance grows from outcrop the need for extending 26
use-life also increases (Keeley 1982, Buchanan 2006). In the lithic-centered model, organic resources are exploited at or near the lithic outcrop, putting less of an emphasis on the conservation of raw materials. In the logistical forager model, the need to access other resources pulls early Paleoindians away from the lithic outcrop. These distant forays require hunter-gatherer groups to make efficient use of their tools, and considerable depletion via resharpening is expected.
The size, shape, and uniformity of supply ranges carry a different set of expectations for each model. With the lithic-centered model, one would expect the density of said raw material to drop evenly, either arithmetically or more likely, logarithmically with distance. 27
Any increase in density at distant locations would lend support to the logistical forager model. Any “hot spots” or areas with longer points and/or distant raw materials far from quarrying locations imply intensive use unexplained by Model 1.
Many, including Gramly, Anderson and Hanson, and Seeman, have provided models of expected supply ranges for lithic sources. Gramly (1988) and Seeman (1994) estimate the supply ranges for early Paleoindian exploitation of several eastern lithic resources (Figure 8). Later,
Anderson and Hanson (Figure 9) estimate the home range for several
Early Archaic groups based on the location of a single lithic resource
(1988:269). It is important to note the large size of each of the ranges posited. These are comparable in size to the 125-150km ranges suggested for Gainey groups
(Stothers 1996). The size and shapes of these ranges are not consistent, and vary considerably from one another. Finally, the dichotomous boundaries indicate that a location on the map was either „in‟ the range, or
„out‟ of the range. Each of these supply ranges was constructed by considering the landscape
(propinquity to rivers, mountain ranges, etc.) and the proximity of the lithic outcrop to alternative raw material sources. A conformal shape would be expected for the lithic- 28
centered model because of the implied irrelevance of other resource locations. The shape expected for the logistical forager model could be difficult to predict, but would be premised on access to resources other than outcrops of high quality stone.
Chapter IV.
Assumptions and Test Expectations
With the establishment of the two mobility models, it is important to note both the assumptions of this project, and also to provide a detailed description of the hypotheses being examined. Any explanation of archaeological data requires a certain number of assumptions, and this exploration of early Paleoindian mobility patterns is no exception. Provided the assumptions are supported by logical expectations, the conclusions for the research should provide useful insights on the problem. Also relevant is the importance of that ‘grey area’ between the different models being tested. Models are generally hypothetical ‘best case scenarios’, and the realistic situation frequently lies somewhere in-between the two extremes.
A set of hypotheses are formulated and explained for the lithic-centered mobility model, as well as the logistical forager model. An additional set of hypotheses are constructed to test the relationship between the exploitation of Upper Mercer flint and the utilization of Flint
Ridge flint.
Many of the assumptions in this research are supported by archaeological evidence.
As stated before, early Paleoindians are assumed to by preference exploit a single, high quality lithic raw material source, all other things being equal. Since the evidence for
29
30
exchange systems during this time is limited, the movement of stone across the landscape should be the result of the movement of specific groups of foragers across the landscape
(Ericson and Baugh 1994, Goodyear 1979). Thus, the discard pattern of early Paleoindian tools should relate strongly to early Paleoindian group mobility.
Hypotheses
The purpose of this research is to determine whether the data suggest a highly tethered, quarry-centric, residential mobility or a loosely tethered, quarry-adjunct, logistical mobility. Both the raw material supply range and resharpening data will be analyzed and compared with the following expectations.
Hypotheses for the Lithic-Centered Model
With its immediate emphasis on the location of the lithic outcrop as the key determinant of early Paleoindian mobility, the lithic-tethered model carries few expectations for either resharpening patterns or supply zone formation across the landscape. In fact, under this model, not much resharpening should be required due to the relatively low cost of resupplying. Therefore, points with evidence for extensive resharpening should be infrequent in the archaeological record. The majority of projectile point finds should be close to their original state, or broken from use, but not extensively retouched. Due to infrequent use of distant areas from the raw material source, correlations of for resharpening characteristics with distance should be weak or inconsequential. The fall-off of size over distance should be characterized by relatively abrupt supply zone boundaries, uninterrupted by large fluctuations in size at distant intervals, a pattern that should be quite visible with histogram or graphic depiction. The raw material distributional data should strongly parallel spatial results for resharpening. Specifically, a smooth fall-off curve from the source is expected. Any 31
deviations indicating larger points well distant from the source and the imposition of raw materials where they are not expected according to a principle of gradual extinction from source supports the alternative view. The boundary for immediate supply zones is difficult to estimate, but for purposes of this thesis is set at 40km following Carr (2005:11). The size and shape of the lithic centered model should be symmetric, graded, and uninterrupted.
Hypotheses for the Logistical Forager Model
The logistical forager model is accompanied by a different set of relationships.
Logistical-forager groups are not tethered to raw material outcrops to the same extent as groups under the lithic-centered model (although this is not meant to imply less of a reliance on a single lithic raw material source). If a group is not tethered to a certain point on the landscape, new areas and resources become accessible. A group that is logistically mobile will travel distances far from the lithic raw material source to obtain resources like food, other raw materials, and possibly mates. For this reason, pristine projectile points have the opportunity to be lost or discarded at a farther distance from the raw material source than with the first model. Logistically mobile groups might be expected to utilize caching to allow for additional and more flexible movement across the landscape, resulting in a density fall- off curve that is not solely reliant on distance. The availability of subsistence resources, in this case, should affect the fall-off curve in addition to the proximity to lithic outcrops. Under this model, it is expected that there should be crests of higher percentages of each flint type where food resources are located, and troughs of lower percentages where resources were not present. If projectiles were lost or misplaced on these expeditions to obtain other resources, we would expect to find larger points in distant areas. Finally, a higher incidence of shorter points can be expected at the outcrop source as a result of convenient discard. 32
Hypotheses for Relationship Between UM and FRF Usage
Statistical tests and spatial patterning also can be used to better understand the variation in the use of Upper Mercer flint compared with that for Flint Ridge flint, two high quality sources that are relatively close to one another, and probably no more than two or three days walk apart. If the two raw materials were utilized in a similar manner, we would expect to find consistencies in their densities across the study area. The evidence for point resharpening should parallel one another. Dramatic differences between them thus should indicate that one was preferred over the other, leading to questions as to why this should be the case for either tethered or highly mobile foragers.
Chapter V.
Materials and Methods
The available dataset contains information on Clovis or Gainey fluted projectile point finds in Ohio and surrounding states (Figure 5). Recorded variables for these points include raw material type, distance of point from its raw material source, and for select points, metric data. These variables provide the basis for visually observing and statistically analyzing the mobility for early Paleoindians. The length and width measurements are presumed to provide insights into the use-life, and the resharpening extent of projectile points. Raw material counts per county are used to look at the relative density of Upper Mercer flint and Flint
Ridge flint across the study area.
Materials
Sources for Data
The projectile point data used in this project came from a variety of sources. Most of the data for Ohio, Indiana, and Kentucky was obtained from Ken Tankersley’s dissertation
(1989). Point data for Pennsylvania was obtained from Fogelman and Lantz (2005). Andy
White provided additional information for the northwestern portion of the supply range in northeastern Indiana (2008pc). Additional information for Belmont County, OH was provided by Steve Durea, and Jeff Carskadden’s survey of 141 Gainey points added data for
33
34
Muskingum County (in press, 2009). Barbra Barrish‟s thesis (1995) was used to add point descriptions for Medina and Stark County in Ohio, and for Jefferson County, in Michigan.
West Virginia projectile points were graciously provided by several collectors and were analyzed by Mark Seeman and myself (Appendix B). In total, over 1,300 points were considered for this research project. Each of the points is documented in the published literature, master‟s theses, or contract reports, with the exception of the West Virginia data and the points from Belmont County, Ohio. The areas with the most information come from
Ohio, southern Indiana, Kentucky, and Pennsylvania. More data from Michigan, West
Virginia, New York, and Ontario would be helpful to solidify northern and southern boundaries for the supply range.
Data Inclusion Requirements
Again, resharpening data and distributional data were used to measure the mobility of early Paleoindians. To be included in this present study, individual points in each data source were carefully evaluated. In order to complete resharpening analyses, information on metrics, point type, point location, and raw material was required. Distributional data had a separate set of inclusion requirements, centering on the numbers of points that could be documented from specific counties in the region.
To be included in the resharpening analysis, projectile points had to first be assigned to either the Clovis or Gainey type, ensuring the use of only early Paleoindian points.
Second, the raw material of a point needed to be identified. If the raw material type was marked as „unknown‟, the point was discarded for this part of the project. An unknown raw material type created obvious issues when determining the distance of a point from its raw material source, which in turn makes any available size data useless for this project. 35
Maximum length and width were measured for many of the projectile points of known raw material type and they were included in the resharpening analysis. A total of 355 points were used in the analysis of length to distance, and a total of 352 points were used in analyzing width to distance, size index over distance, and the relationship between length and width.
Distributional data had a separate set of requirements for projectile point inclusion.
Similar requirements included the necessity for the point to be categorized as either Clovis or
Gainey. In this case, however, a raw material type labeled as „unknown‟ was included, as long as the county level provenience for the point was recorded. It is assumed that Upper
Mercer and Flint Ridge flint are two raw materials that can be distinguished easily from other flint types found in the study area and since the important calculations for the supply range resulted in percent Upper Mercer, percent Flint Ridge flint, or percent “other”, unknown raw materials were simply counted as “not Upper Mercer” or “not Flint Ridge flint”. To be considered in the interpolation, counties needed to have at least five Clovis or Gainey points present to minimize sample bias as a result of small sample size. So, any county with at least five early Paleoindian (ie. Clovis or Gainey) points with raw material notes was used to create a GIS of expected distributions. There were 70 total counties that met the criteria and were used for this analysis.
Methods
Before detailing specifics on the methodology of this project, it is important to note the relevance of sampling techniques in archaeology. Ideally, random samples of data selected from a larger population to ensure an unbiased approach to the question at hand provide the best predictive results. Once such a dataset is complied and analyzed, broad statements can be made about the original population. Unfortunately, random sampling is a 36
difficult process in the realm of early Paleoindian archaeological research since artifacts are especially rare, and most studies in this area are based on additive, available samples. As a result, all recorded points were considered in the initial process of forming a dataset. One possible issue that was considered in this research is the possibility that more projectile points are likely to be found in areas of higher modern development and higher population.
To protect against this factor, percentages were used for each county, assuming that there is an equal likelihood of finding points made of each raw material type.
As stated previously, there are two main datasets being tested to better understand the relevance of the mobility models posited. The first dataset involves resharpening characteristics, and the second evaluates the supply range and density of either Upper Mercer or Flint Ridge across the landscape. Statistical, graphical, and spatial analyses were performed on the resharpening datasets, where only spatial analyses were conducted on the supply range information. The dataset for resharpening studies included 161 points made of either Upper Mercer or Flint Ridge flints. Over 1,000 points met the requirements necessary to be included in the raw material distribution study. The statistical and graphical methods for the resharpening data will be explained first, followed by and explanation of the spatial methods used for evaluating the resharpening characteristics and supply zone patterns.
Statistical Methods
Statistics are often used to document patterned relationships among variables.
Multiple algorithms are suggested for various types of correlation. For this project, correlation was tested between metric measurements and distance. As implied, these tests provide some level of confidence in the relationship between curation (as measured by size) and distance. A non-parametric, two-tailed, Spearman‟s rho correlation test was used to test 37
the data. This is a ranked correlation test, which protects against the oddities and outliers often present in archaeological data. A two-tailed test was used because a positive or negative correlation could not be assumed. It is more difficult for a relationship to be significant with a two-tailed test than at a one-tailed test. The data points were broken into two categories: points made of Upper Mercer flint, and points made of Flint Ridge flint.
Correlation was then tested between three different variables for each category: length and distance, size index and distance, and length and width. The size index is defined as length over width. A total of 125 points provided enough information to be used in the testing of correlation for the Upper Mercer sample. Only 36 points contained the information necessary to test correlation for Flint Ridge flint. The results of this analysis should provide insights into where early Paleoindians generally discarded points and at what use-life stage.
As a fluted point as it gets farther from its raw material source, we want to know if it gets smaller. Do points get shorter as distance increases, longer, or does the length vary in an unpredictable way? Second, the relationship between the size index and distance provides more information on the general shape of points as distance from raw material outcrop increases. If the point has a size index (SI) of 1, the length and the width of the point is equal, implying extensive retouch. If the size index is 3, the point is three times longer than it is wide, implying less retouch and resharpening than in the first case where SI=1. Finally, the relationship between length and width was measured. If there is a positive correlation between length and width (as length increases, width increases), any variation around this
“standard” should imply resharpening.
Graphical Methods 38
The purpose of the graphical analysis of the resharpening data is to quantify systematically the available descriptive information on the length, width, and size index of projectile points at certain distance intervals from the raw material source. In Microsoft
Excel, graphs were created relating average metric measurements to 40km distance intervals.
The bar charts created do not take directionality into account; but they do illustrate variation of metrics in relation to the distance from the lithic outcrop source. This technique was used to maximize the sample size for the relatively small resharpening dataset. Width and length measurements are in millimeters, and distance is always measured in kilometers. Two raw material groups are recorded; first points made of Upper Mercer Flint, and second points made of Flint Ridge flint.
Spatial Methods
GIS is a powerful tool for spatial analysis, and one of the most frequently used function sets for this program is the Spatial Analyst toolset of ArcMap provided by the ESRI corporation. These tools range in application from determining least-cost paths to interpolating surfaces based on known data points. The nature of the resharpening and the distributional data allowed for an interpolation surface to be created based on county level data. It is important to note that the interpolated surface would increase in quality if individual projectile points had locational data more specific than county locational data.
County recordings were used because more specific geographic coordinates for these types of projectile points are hard to come by.
Any early Paleoindian Clovis or Gainey projectile point with both a raw material and a county recording was then input into the GIS database. For resharpening data, the projectile point also would need to have measurements recorded for the length and width. Once the 39
database table was complete, it was joined to spatially referenced county layers retrieved from the ESRI website. The counties in the study area were extracted to ensure a concise and easy to read map. Joining the layers resulted in the addition of projectile point data to county data, creating a categorized, spatially referenced database. Using the select by attribute tool, counties with at least 5 projectile points were then extracted for distributional analysis. The resharpening analysis used two point categories: those made of Upper Mercer, and those made of Flint Ridge flint. The distributional data were categorized into 3 categories: points made from Upper Mercer, points made from Flint Ridge, and points made from Upper
Mercer or Flint Ridge flint.
Once the datasets were extracted for proper interpolation, the county shapefiles were transformed into point layer files. Inverse Distance Weighted (IDW) was then used to interpolate the different analysis surfaces. IDW was chosen because of the heavy emphasis given the characteristics of nearby points, assuming points located close to one another are more likely to have similar results. In total, there were 6 maps created for the resharpening analysis, and 6 maps were created for the distributional analysis.
Chapter VI.
Resharpening Results
The results for various tests evaluating resharpening characteristics address several issues concerning early Paleoindian mobility in the Great Lakes region. As described above, statistical, graphical, and spatial methods were used to explore the resharpening data. The results of statistical tests provide information on the general tendencies of the dataset.
Graphical analyses set up expectations for the use of distances across the study area. Finally, the spatial analysis provides details about specific areas on the landscape.
Statistical Tests
Upper Mercer
The original dataset included 217 projectile points with a raw material type identified as Upper Mercer. Of those 217 points, 125 included the necessary metric measurements required for the resharpening analysis. Table 1 presents the results of the Spearman’s rho correlation tests. There is a negative correlation between length and distance, which is significant at .05. Size index and distance provides a similar negative correlation, but is more highly significant at .01. Points are on the average longer, but proportionally more squat with distance. Finally, the correlation tests imply that there is a positive correlation between the length and width of Upper Mercer points, which is also significant at .01.
40
41
Table 1. Upper Mercer Statistical Results. Correlation Sample Upper Mercer Coefficient Sig. Relationship Size
Length/Distance -0.223 0.12 negative 125
SI/Distance -0.358 0 negative 125
Length/Width 0.38 0 Positive 125
Table 1. Spearman’s rho results for correlation on Upper Mercer points.
Flint Ridge flint
For Flint Ridge flint, the statistical correlation tests were unable to confirm relationships similar to Upper Mercer. Perhaps the small sample size affects results or perhaps the relationships are notably different (there were only 36 Flint Ridge flint points with enough information to be used in the analysis). The strength of the correlation between a point’s length and its width was upheld, with a positive correlation with significance at .01.
Similarities between Upper Mercer correlations and the results for Flint Ridge flint are present in the output tables. A statistically insignificant negative correlation between both length and distance, and size index and distance is present.
Table 2. Flint Ridge Flint Statistical Results. Correlation. Sample Flint Ridge Coefficient Sig. Relationship Size
Length/Distance -0.313 0.063 N/A 36
SI/Distance -0.237 0.177 N/A 36
Length/Width 0.658 0 Positive 36 Table 2. Spearman’s rho results for correlation on Flint Ridge flint points.
42
Graphical Results
Statistical correlation tests provide information about general trends in the dataset.
These trends can only illustrate so much about resharpening characteristics. Graphical analyses can be useful in illustrating non-directional patterns in the data. Each of the following histograms represents estimates of resharpening parameters as a function of distance. The maroon column provides the average length, width, or size index, while the purple columns represent the number of points available at a particular distance interval.
Upper Mercer
Length
Given the non-parametric statistical results, we expect to see that the points get shorter as distance increases. Interestingly, this does not seem to be the case in any clear fashion (Figure 10).
Figure 10. Bar graph depicting length over distance for Upper Mercer points.
43
The average length of a point near the Upper Mercer quarries is about 58mm. This falls slightly before increasing steadily until a distance of 159km is reached, where the average length is longer at 74mm. This is then followed by a second decrease until around 239km, where the average length is 43mm. An increase in length to 73mm is represented at the 280-
319km interval. The five projectile points between 400 and 559km away average at about
51mm in length, which is lower than the Upper Mercer average length of 57mm, but not by much. It is important to note that there is not a smooth fall off for the length of Upper Mercer points over distance.
Width
Most archaeologists believe that the resharpening of early Paleoindian projectile points has little effect on the original width of that point. This is well represented by the bar graph representing average width over distance from raw material source (Figure 11).
Upper Mercer: Average Width Over 40km Distance Intervals
35 30 25
20 N 15 UM Avg. Width 10
Average Width (mm) Width Average 5 0 0-39 40- 80- 120- 160- 200- 240- 280- 320- 360- 400- 440- 480- 520- 79 119 159 199 239 279 319 359 399 439 479 519 559 Distance (km)
Figure 11. Bar graph depicting width over distance for Upper Mercer points.
44
The average width remains fairly constant over the entire study area, varying by only a few mm in either direction. For the Upper Mercer dataset, the average length is 24.5mm. Outliers
(e.g. the average length of 31mm at the 520-559 interval) might be explained by having a larger point to begin with, before any resharpening had occurred. It is interesting to note that some of the most ‘initially large’ points are far from their raw material source.
Size Index
The size index is calculated by dividing the length of a point by its width. For this dataset, the average size index was 2.25. Since the width remained fairly consistent over distance, the size index should be similar to the pattern shown by length. Generally speaking, this does occur, with some differences (Figure 12).
Upper Mercer: Average Size Index Over 40km Distance Intervals
3.5 3 2.5 2 UM Avg. Size Index 1.5 1
Average Size Index Average 0.5 0
0-39 40-79 80-119120-159160-199200-239240-279280-319320-359360-399400-439440-479480-519520-559 Distance (km)
Figure 12. Bar graph depicting size index over distance for Upper Mercer points.
Specifically, There is no drop off in size index present at the second distance interval. This implies that the shorter length represented in the areas 40-79km away are a result of having 45
smaller points in general at this location. There are signs of more retouch (a smaller size index) at areas from 160-279km away, and then from 400-559km away, although in the latter case, the effects of small sample size should be considered.
Flint Ridge flint
The sample size for Flint Ridge flint is small. There are only 36 projectile points analyzed for the length, width, and size-index graphs. Since only 36 points were used in this analysis, outliers may have a relatively strong effect on the average length, width, or size index reported for distance intervals. In order to be consistent, the 40km distance intervals were used once again. It is interesting to note that the average length for points made from
Flint Ridge flint is 61mm, in comparison to the average length of Upper Mercer points of
57mm. Similar to the average width of Upper Mercer points, Flint Ridge flint points have an average width of 25mm. The average size index for Flint Ridge flint points is 2.4, slightly higher than the index for Upper Mercer points.
Length
The graphical patterns of the relationship between length and distance for
Flint Ridge flint is very similar to those for Upper Mercer points (Figure 13). With Flint
Ridge, length increases progressively from 0 to 159km, without the drop in average length from 40-79km shown in the Upper Mercer graph. There is a decrease in length from 160-
239km, followed by an increase in average length in the 240-279km. A decrease in average length is present from intervals 240-319km and 400-439km away. It is obvious that the longest point is present at the 480-519km interval, where the length is 111mm.
46
Flint Ridge Flint: Average Length Over 40km Distance Intervals
120
100
80 N 60 FRF Avg. Length 40
20 Average Length (mm) Length Average 0 0-39 40- 80- 120- 160- 200- 240- 280- 320- 360- 400- 440- 480- 79 119 159 199 239 279 319 359 399 439 479 519 Distance (km)
Figure 13. Bar graph depicting length over distance for Flint Ridge flint points.
Width
The average width of Flint Ridge flint points is about 24mm. The bar graph illustrates the consistency of Flint Ridge flint width (Figure 14). Variations in average width fluctuate within only a few mm of the average width of 25.
Flint Ridge Flint: Average Width Over 40km Distance Intervals
35 30 25 20 N 15 FRF Avg. Width 10
Average Width (mm) Width Average 5 0 0-39 40- 80- 120- 160- 200- 240- 280- 320- 360- 400- 440- 480- 79 119 159 199 239 279 319 359 399 439 479 519 Distance (km)
Figure 14. Bar graph depicting width over distance for Flint Ridge flint points. 47
Size Index
Similar to the pattern presented by Upper Mercer size index graphs, results for Flint
Ridge flint are strongly influenced by the greater variation in the length measurements
(Figure 15).
Flint Ridge Flint: Average Size Index Over 40km Distance Intervals
4.5 4 3.5 3 2.5 FRF Avg. Size Index 2 1.5 1 Average Size Index Size Average 0.5 0
0-39 40-79 80-119 120-159160-199200-239240-279280-319320-359360-399400-439440-479480-519 Distance (km)
Figure 15. Bar graph depicting size index over distance for Flint Ridge flint points.
The size index from 0-39km, and 40-79km are both 2.4. An increase to 2.9 and 3.0 from 80-159km is followed by an abrupt fall off from 160-239km, where the size index is as low as 1.7. An increase to 2.5 at 240-279km is then followed by another decrease from 280-
319km and 400-439km. The single point found within the distance interval 480-519km away is quite large, reporting a size index of 4. It appears that the points with the most evidence of resharpening are 200-239km and 280-319km from the FRF outcrops. 48
These histograms provide non-directional visualizations of where different metrics patterns occur relative to distance. They represent fall-off curves for resharpening attributes , and provide more information on the size of points at different discard distances. More refined spatial analyses can provide additional insights into the meaning of the resharpening data, providing more specific locational information for some of the distributional differences documented above.
Spatial Results
Upper Mercer Resharpening
The following interpolated map is for the projectile points made of Upper Mercer flint and is based on county averages for length, width, and size index (Figure 16). The blue dots on the interpolated maps represent all counties that contain the necessary metrics for
Upper Mercer points. It is important to note that the Upper Mercer averages were selected for the interpolation. These averages could only be calculated for a total of 13 counties. When looking at the spatial distribution of projectile points based on length (Figure 16), it is clear that the longest points are not closest to the location of the Upper Mercer outcrops. Rather, the interpolation shows that the longest fluted points are located 49
Figure 16. Interpolation of average length for fluted points made of Upper Mercer Flint.
in Fairfield (OH), Clermont (OH), and Jefferson (PA) counties. The shorter points form a band across the western border of Pennsylvania. The interpolated map based on width shows that the wider points are also farther away from the raw material outcrops (Figure 17), aligning with the information provided from statistical tests. Overall, the width is drastically more uniform than length, and it is possible that the
50
Figure 17. Interpolation of average width for fluted points made of Upper Mercer Flint.
differences are simply a result of superimposed range differences. The interpolated map for size index appears to be slightly different from the results shown by the statistical or graphical tests (Figure 18). The more narrow points seem to be concentrated to the southwest in Boone (KY) and Pickaway (OH) counties. The shorter and wider points appear to the east, near the border of Pennsylvania. 51
Figure 18. Interpolation of average size index for fluted points made of Upper Mercer Flint.
52
Flint Ridge flint Resharpening
The same three measurements used for Upper Mercer were interpolated for Flint
Ridge flint. Only 12 counties contained the proper data to proceed with interpolation for Flint
Ridge flint. The length distribution implies that the longest points are to the southwest of the
Flint Ridge flint in Boone County, KY (Figure 19). The shorter points are located in the east.
There are also longer points distributed between the southwest and the Flint Ridge outcrops in Licking County. It appears that the Flint Ridge flint projectile points discarded in Licking
County are relatively short.
Figure 19. Interpolation of average length for fluted points made of Flint Ridge flint. 53
The width distribution for Flint Ridge flint points shows wider points farther away from the raw material outcrops, with the exception of Bedford County, PA (Figure 20).
Figure 20. Interpolation of average width for fluted points made of Flint Ridge flint.
By looking at the distribution of size index measurements, it becomes obvious that many Flint Ridge flint points are fairly stubby. The narrowest points appear in Boone County
(KY), and the shorter and wider points are in southwestern Pennsylvania and near the Flint
Ridge flint outcrops (Figure 21). 54
Figure 21. Interpolation of average size index for fluted points made of Flint Ridge flint.
Chapter VII.
Supply Range Results
The distribution of Clovis and Gainey fluted points made of Upper Mercer and Flint
Ridge flints was analyzed in GIS to create interpolated maps. Upper Mercer flint and Flint
Ridge flint were chosen as high quality, Ohio raw materials that were used frequently by
Clovis and Gainey groups. The percentages of fluted points in each county were used as the indicators for interpolation, with a cutoff value of at least five per county set as a minimum for inclusion. The resulting density maps are presented below.
Upper Mercer Flint Density Interpolation
The overall distribution of Upper Mercer flint is well defined for the eastern, western, and southwestern borders of Ohio (Figure 22). The northeastern and southern boundaries are somewhat less clear due to the small sample size from northeast Indiana, Michigan, and West
Virginia. Therefore, I am somewhat less confident in the data from the radii at 0°, 135°, and
315° for the Upper Mercer. It is clear that when moving eastward into Pennsylvania, the decline in Upper Mercer appears to be more abrupt than the drop westward into Indiana.
Moderate drops in density occur in southwest Ohio. Counties directly adjacent to the Upper
Mercer outcrops are more likely to contain higher percentages of the raw material (at least
60%) than those farther away. Other counties high in Upper Mercer flint include Cuyahoga
55
56
Figure 22. Supply range interpolation for Upper Mercer. 57
Figure 23. Supply range interpolation for Upper Mercer >60%. 58
County (OH), and Franklin County (OH). The areas with at least 60% Upper Mercer fluted points are thought of as the main supply range for Upper Mercer flint (Figure 23).
A better understanding of the variability in distribution is provided by Figure 24. The lines stretching 45° and 90° from the Upper Mercer outcrops are “bumpy”, and show a quicker drop off rate for the percentage of Upper Mercer fluted points. Line 180° more closely approximates a smooth logarithmic fall-off. The more westerly radii drawn in Figure
24 show smooth, logarithmic fall-offs or extinctions. The radius stretching into the southwest
(225°) have an abrupt dip, and are followed by gradual fall-off. Moving directly west into
Indiana, the radius at 270° clearly illustrates the fact that Upper Mercer was used more intensively at longer distances in the west than at similar distances to the east along the radius at 90°. Each line profile illustrates the varying nature of this distribution. 59
Figure 24. Below are profile lines depicting extinction curves for Upper Mercer Flint.
Upper Mercer 0º Percent Upper Mercer 45º 58 56 55 54 52 50 50 45 48
46 40 PercentUpper Mercer PercentUpper Mercer 44 42 35
60
Percent Upper Mercer 90º Percent Upper Mercer 135º
60 50 40 30 20
PercentUpper Mercer 10
Percent Upper Mercer 180º Percent Upper Mercer 225º
50 55 40 50 30
45 20 PercentUpper Mercer PercentUpper Mercer 40 10
Percent Upper Mercer 270º Percent Upper Mercer 315º
50 55
40 50
30 45
20 40 PercentUpper Mercer PercentUpper Mercer 10 35 0 30
Flint Ridge flint Density Interpolation
The same processes in ESRI‟s ArcMap were used to generate a density distribution,
or “supply range estimation” for Flint Ridge flint as were used previously for the Upper
Mercer points. After observing the Flint Ridge distribution alongside the Upper Mercer
distribution, it becomes obvious that it is less densely distributed than Upper Mercer flint
(Figure25). The only area with over 50 percent Flint Ridge flint is within Licking County
itself, where the raw material outcrops (Figure 26). At a less distinct array, the Flint Ridge
flint distribution is more difficult to interpret, but it appears to show a similar drop off pattern 61
to the Upper Mercer distribution, especially in the east-west patterning. Jefferson County, PA represents a faint „hot spot‟ for the distribution of Flint Ridge flint. The counties directly surrounding the outcrop have a somewhat heightened percentage of Flint Ridge flint occurrences.
62
Figure 25. Supply range interpolation for Flint Ridge flint. 63
Figure 26. Supply range interpolation for Flint Ridge flint >60%. 64
The datasets used for Upper Mercer and Flint Ridge have similar strengths and weaknesses (Figure 27). Since the same method was used to generate a density distribution of Flint Ridge as Upper Mercer, there exist issues with the radii along 0°, 135°, and 315° from the Flint Ridge quarries due to the lack of near-by observed points. These radii have values that are not fully supported by the data, but rather, are estimates modeled by the interpolation method. The lines stretching into the east and northeast show abrupt fall-off curves and each are uninterrupted by bumps in the predicted percentages of Flint Ridge flint.
The lines extending into the west show a more gradual fall of curve, largely interrupted by extreme changes in percentage. As with the Upper Mercer distribution, there seems to be a different interaction with the landscape to the east from that of the west.
65
Figure 27. Below are profile lines depicting extinction curves for Flint Ridge flint.
Percent Flint Ridge Flint 0º Percent Flint Ridge Flint 45º
52 80
50 70
48 60
46 50
44 40
PercentRidgeFlint Flint PercentRidgeFlint º Flint 42 30 66
Percent Flint Ridge Flint 90º Percent Flint Ridge Flint 135º 60 55 50 50 45 40 40 30 35
20 30
25 PercentRidgeFlint Flint
PercentRidgeFlint Flint 10 Profile Graph Subtitle
Percent Flint Ridge Flint 180º Percent Flint Ridge Flint 225º 50 45 44 40 42 35 30 40 25 38 20
15 PercentRidgeFlint Flint
36 PercentRidgeFlint Flint 10 Profile Graph Subtitle 5
Percent Flint Ridge Flint 270º Percent Flint Ridge Flint 315º 60
50 45
40 40 30
20 35
PercentRidgeFlint Flint 10 PercentRidgeFlint º Flint 30 0
In examining the directional profile radii for both raw materials, it is clear that there is
variation in the rate and form of the extinction curves. Three potential patterns are in
evidence. The first recurrent pattern is a steep drop off, followed by a significant increase at
areas distant from the raw material source (reference Figure 24 at 0°, and 90°; and Figure 27
at 0°).Second, a smooth logarithmic drop off is revealed by some radii (Figure 24 at 45°,
135°, and 180° and Figure 27 at 45°, 90° and 135°). Finally, a third pattern indicates a
certain “prolonged importance” or “plateau effect” of a raw material over distance (Figure 24 67
at 225°, 270°, and 315°, and Figure 27 at 180°, 225°, 270° and 315°). It is important to note the general directionality for each of these trends. The first pattern, where there is a distant section with an increase in volume of a raw material, occurs to the north, and in the case of
Upper Mercer flint, to the east as well. The second, logarithmic drop-off pattern most frequently occurs towards the east and the south. The final pattern occurs only in the western and southern directions. These varying patterns would seem to indicate that several different factors are affecting the spatial distribution of early Paleoindians as measured by raw material usage.
Combined UM and FRF Interpolation
After looking at the relatively small size of inferred supply ranges for Upper Mercer and Flint Ridge flints individually, it is clear that they do not relate well to the previous estimates for early Paleoindian supply ranges at all, either in the Midwest or elsewhere. They also are undersized in comparison to the “least-cost” presented in Figures 4 and 5 in Chapter
2. An interesting pattern is revealed when the percentages per county of Upper Mercer and
Flint Ridge flint are combined (Figure28). When reviewing the results of the interpolation, it becomes apparent that at least 60% to 70% of the projectile points found in many Ohio counties are constructed of either Upper Mercer or Flint Ridge flint (Figure 29), providing a supply range that is much more closely related to those previously proposed. Further, there is still general conformity to the model that Paleoindian groups tended to rely on single-source materials of high quality in a more or less exclusive manner, provided we consider Upper
Mercer and Flint Ridge flint to be a single source. This makes sense at a large scale and given their proximity, and the distances to other preferred materials such as Onondaga, 68
Wyandotte, Attica, and so forth. The fall-off in the frequency of the combined Upper Mercer and Flint Ridge distribution to the west seems much more gradual, and is uninterrupted by sharp decreases or quick increases in the percentage of either flint within particular areas moving in the direction of Wyandotte and Attica. It appears that the extension of the supply range in the southwest is interrupted, creating a quicker drop off than in the northern areas.
There is evidence for interference with the southern boundary near West Virginia as well.
In making reference to the model, the two strongest deviations would appear to be to the east and south. Regarding the former, Neither Upper Mercer nor Flint Ridge flint extend as far east as they should, given the location of the Onondaga and Bald Eagle materials. This makes sense with respect to the difficulty moving east from the Ohio sources due to the notable highlands and difficult-to-cross Appalachian Mountains. To the south, the Kanawha source in particular should be constraining the southward expansion of Upper Mercer and
Flint Ridge flint more than my data would suggest. Here is should be noted that none of the
West Virginia points examined in this investigation were made of Kanawha, although it was the preferred raw material for slightly later Early Archaic groups in the area. 69
Figure 28. Supply range interpolation for either Upper Mercer or Flint Ridge flint. 70
Figure 29. Supply range interpolation for either Upper Mercer or Flint Ridge flint >60%.
Chapter VIII.
Discussion, Conclusions and Suggestions for Future Work
In this thesis, two datasets involving resharpening characteristics and raw material distributions were used to compare two mobility models for early Paleoindians in the lower
Great Lakes region. One I have called the lithic-centered model, and the other the logistical forager model. A non-parametric statistical test, descriptive statistics, and spatial analyses were employed to examine patterning in each dataset. Each analysis provides useful insights into the nature of early Paleoindian mobility. The Upper Mercer flint quarries in Coshocton
Co., Ohio, and Flint Ridge flint in Licking Co., Ohio served as the geographic centers of interest.
The results for tests done on resharpening data seem to show somewhat different patterning when the data are examined at different scales. Initially, a Spearman’s rho two- sided correlation test was run on length, width, and a size index to interpret the effects of tool curation as distance from quarried source increased. Results showed a statistically significant negative correlation between length and distance for the 125 Upper Mercer points. The 36
Flint Ridge fluted points used in this statistical analysis fell short of proving a statistically significant negative correlation, but the results do lean towards a negative relationship between point size and distance from source. The small sample size for Flint Ridge may be
71
72
the reason the results of the test did not reach significance. My summary conclusion based on these results is that early Paleoindian fluted points are shorter relative to their width as the distance from the quarry increases, and in fact heavily resharpened points are sometimes found hundreds of kilometers from the source of a given Ohio raw material. This finding lends support to the lithic-centered model as first posited by Gardner (1977). As distance increases, points become more heavily resharpened as a result of higher replacement cost for a quarry-centered group. Support of the lithic centered model becomes less clear as we move to a finer scale and examine the shape and slope of the extinction curve for projectile point length relative to a fairly constant width using basic histograms and GIS spatial modeling.
Graphical and spatial analyses provide results suggest that although the expectation in that the Spearman’s test is significant for Upper Mercer, the increase in resharpening with distance to source is far from a pattern of smooth extinction. Following from the logarithmic extinction model of distance-to-source as summarized by Hodder and Orton (1976), we would expect to see a logarithmic drop-off for size over distance, but this is not the case. The graphical results actually show clear oscillations or pulses in length over distance intervals.
This fluctuation is also evident in the GIS spatial modeling, where isolated “hot spots” of increased size are present in distant areas from both the Upper Mercer and Flint Ridge outcrops. These patterns lend more support the second model, suggesting that early
Paleoindians were actively moving out across the landscape to exploit non-lithic resources
(presumably mainly food) wherever they encountered them in high densities. These areas with higher densities of larger fluted points are also more consistent with the notion that
Paleoindians would consistently cache fresh tools in locations far from the source with the expectation that such a strategy would allow them to exploit subsistence resources with 73
greater flexibility far from the potential source of lithic resupply. The general advantages of caching tools on the landscape for foraging populations has been discussed elsewhere with the expectation that the cached material might be needed in the future, and this is an advantage for early Paleoindian populations in the Midwest in particular (Seeman 1994:284,
Deller and Ellis 1988). Additional support of the second model also is provided by the spatial analysis of each flint’s raw material density pattern or supply range.
The supply range of Upper Mercer and Flint Ridge flint used by the early
Paleoindians of the Ohio region covered large areas and has been examined in this thesis in several ways. The first a map illustrates the interpolated percentage of each flint type across the study area, which provides suggestions for geographical and cultural boundaries during the early Paleoindian period. These boundaries extend from Pennsylvania to Indiana and from Michigan to northern Kentucky. When observing the rate at which each flint type decreases in density in relation to distance from source, it becomes apparent that there are clearly different patterns to the east and to the west of the outcrop locations. This asymmetry cannot be explained easily by the lithic centered model, and it is concluded here that it is more consistent with the expectations of the logistical forager model. A more detailed discussion as to the cause(s) of these differences is provided below.
Profile radii generated by ArcMap represent the fall-off curves for Upper Mercer and
Flint Ridge flint in multiple directions from the raw materials outcrop. As discussed in the results section, there are three clearly different patterns for these fall-off curves in the data.
The first pattern of extinction shows a pulse of high percentages of either raw material at a considerable distant location from the source. This pattern is characteristic of the radii at 0° and 90° for the Upper Mercer source and 0° for Flint Ridge flint source. This interruption in 74
the fall-off could be explained by positioning caches of “fresh” raw material on the landscape well away from the quarry source in anticipation of future needs. Such a pattern is well documented for some hunter-gatherer populations, and Binford’s (1978) study of the
Nunamuit and their practice of leaving useful materials in caves and other prominent places on the landscape for future use comes immediately to mind. Further, the existence of caches of early Paleoindian bifaces has been documented by numerous investigators prior to the present study (e.g., Deller, Ellis, and Keron 2009). The purpose of caching is also quite consistent with the prospect that at certain times of the year or under certain conditions,
Paleoindian groups that might “normally” center their subsistence practices on a quarry area might foray in a purposeful manner to far-distant places to exploit subsistence resources that occur here in unusual densities. Aggregations of herd animals, particularly caribou, bison, and/or elk might be expected to prompt such patterns of targeted foraging. The best documentation of this pattern in the present study is to the north. One other explanation for this sudden increase in density at a distance from the original raw material source could be the presence of a trade center. This is suggested by Brown et al.. (1990) when looking at a similar fall-off pattern for Mill Creek hoe blades. Because of the lack of sufficient evidence for intensive trade during the early Paleoindian period, it is assumed the first explanation is more likely.
The second pattern of density decline over distance primarily can be seen to the west of the two Ohio quarry sources considered in this thesis. Here the pattern approximates a logarithmic fall-off, but with a notable “plateau effect” of more gradual slope extending up to
175 to 250 kilometers from source. The directionality of this pattern from the quarry is highly similar for Upper Mercer and Flint Ridge. As noted below, it would seem that the 75
cost of moving westerly from the quarries was “cheaper” relative to potential moments to the east. The configurations of local topography and corridors of travel may be important in explaining this pattern. Plateau effects often have been noted in studies of particular goods or commodities, generally with the explanation that within areas close to the source-of- supply potential users or consumers do not overly concern themselves with relatively minor differences in cost; essentially, transport costs close to the source are perceived to be equivalent. Renfrew et al.. (1965), for example, noted a plateau effect in his examination of
Near Eastern obsidian. Topography and cheaper transportation costs near the source may affect such patterning (Hodder and Orton 19xx:118). Hodder and Orton (1976:192, 196-197) suggest that plateau effects also can result from the active management of territorial boundaries or supply sources in single-source situations, as in the case of Dobunnic coins or
Romano-British pottery from the kilns at Oxford and New Forest, respectively. The
“plateaus” observed in the present study, many days travel-time from the quarries, may simply relate back to the “Goodyear hypothesis” and the notion that early Paleoindians will continue to favor a given high-quality source as a corollary of high mobility until the cost of obtaining comparable high quality raw materials, Wyandotte or Attica in this case, are available and relatively low cost. Under such circumstances, replacement may be relatively abrupt.
A third and final extinction pattern for raw material densities over distance is a simple logarithmic fall-off (Hodder and Orton 1976:186) This is a common finding for the utilization of particular raw materials or finished products—as distance increases, the cost to the “consumer” increases logarithmically. Such patterns have been observed elsewhere, for example, with Romano-British Savernake ware (Hodder and Orton 1976:108). In the present 76
study, fall-off from the Ohio quarries approximated this third pattern, along three of the computed radii for Upper Mercer and Flint Ridge, respectively. These are mainly to the east of the quarry locations. It should be noted that the increased costs of travel to the east imposed by the strong relief of the unglaciated Appalachian Plateau and absence of significant river corridors along the Ohio-Pennsylvania border (less than 50 km to the east) may also figure in the generation of this pattern. It should be noted that the Appalachian
Plateau does not appear to be an especially favored environment for early Paleoindian exploitation and fluted points of all types are rare in this physiographic region (Seeman and
Prufer 1982).
The final analyses focused on the notion of a joint “supply range” for Upper Mercer and Flint Ridge flint. When viewing the supply range of either flint type individually, it is clear that they are very small in comparison with previous estimated supply ranges (e.g.
Gramly 1988, Seeman 1994:276). The Upper Mercer range is only about 80x40km, and the
Flint Ridge supply range is limited to Licking County. The distance between the two outcrops is relatively small (only about 40km). On a regional scale, the relative cost of using one of these over the other may not have been very great when it was time to resupply. Thus, if the regional lithic supply zone is assumed to be the area where either is dominant, then it is transformed into a more convincing range (about 200x200km), relative to other estimates of
Paleoindian mobility and supply zones. This combined supply zone still falls off more quickly to the east than the west.
In sum, the combined supply range model for Upper Mercer and Flint Ridge flint seems to be best supported by the available data. Having said this, Upper Mercer still seems to have been the more favored of the two, suggesting that perhaps it may have been closer to 77
other useful resources, or possibly the flint itself was easier to access and more abundant on the landscape, all other things being equal.
In sum, the analyses completed in this thesis seem to support the logistical forager model better than the lithic-centered model. As with most archaeological research, this case study does not fit the model perfectly. With the tests results in mind, and keeping in mind the locations of large sites such as Nobles Pond, Gainey, and Paleo-Crossing, it seems clear that early Paleoindians in the lower Great Lakes region were venturing frequently great distances from their preferred raw material outcrop, and these trips were certainly affected by the landscape and the availability of resources. Distant locations with large points and higher percentages of either flint type also suggest the possibility of caching by these groups, extending their accessibility to the raw material in some cases by more than 100km. Caches of early Paleoindian materials have been identified elsewhere in the Midwest. Early
Paleoindians were mobile hunter-gatherers, focusing on the exploitation of multiple natural resources, and the importance of these two high quality flint types is evidenced by their high percentages across the state.
In order to increase the effectiveness of this project, a few additions should be made.
It is necessary to obtain more information about early Paleoindian fluted points in West
Virginia, northeastern Indiana, and Michigan. If this sample size were larger in these areas, the northern and southern boundaries would be more clearly defined. Once the sample size is expanded, the information and maps provided in this thesis could be compared to information from different time periods (for example late Paleoindian, early Archaic). Much more could be said for mobility patterns if they are viewed relative to one another. The assumption made by most archaeologists is that early Paleoindians had the highest mobility of any group 78
studied in the archaeological record. The mobility range for early Paleoindians has now been quantified, and it is up to future projects to test this model by applying the same methods to other areas and time periods.
Bibliography
Anderson, D.G. 1995a Paleoindian Interaction Networks in the Eastern Woodlands. In Native American Interaction: Multiscalar Analysis and Interpretations in the Eastern Woodlands. M.S. Nassaney and K.E. Sassaman, eds. Pp. 1-26. University of Tennessee Press, Knoxville.
1995b Recent Advances in Paleoindian and Archaic Period Research in the Southeastern United States, Archaeology of Eastern North America 23:145- 176.
Anderson, D.G., and Hanson, G.T. 1988 Early Archaic Settlement in the Southeastern United States: A Case Study from the Savannah River Valley. American Antiquity 53:262-286.
Andrefsky, W. Jr. 1994 Raw-material Availability and the Organization of Technology. American Antiquity 59(1): 22-34.
Balakrishnan, M., Yapp, C.J., Meltzer, D.J., Theler, J.L. 2004 Paleoenvironment of the Folsom Archaeological Site, New Mexico, USA, Approximately 10,500 14C yr B.P. as Inferred From the Stable Isotope Composition of Fossil Land Snail Shells, Quaternary Research 63:31-44.
Barrish, B.L. 1995 The Paleo Crossing Site: Fluted Point Typology and Chronology/ M.A. thesis, Department of Anthropology, Kent State University.
Baugh, T.G. and Ericson, J.E. 1994 Systematics of the Study of Prehistoric Regional Exchange in North America. In Prehistoric Exchange Systems in North America. T.G. Baugh, J.E. Ericson, eds. Pp. 3-23. Plenum Press, New York,.
Binford, L.R. 1978 Nunamuit Ethnoarchaeology. New York.
1980 Willow Smoke and Dogs’ Tails: Hunter-Gatherer Settlement Systems and Archaeological Site Formation. American Antiquity 45:4-20.
79
80
Bradley, B. 1993 Paleo-Indian Flaked Stone Technology in the North American High Plains. In From Kostenki to Clovis: Upper Paleolithic Paleoindian Adaptations. O. Soffer and N.D. Praslov, eds. Pp. 181-208. New York: Plenum Press.
Brown, J.A., Kerber, R.A. and Winters, H.D. 1990 Trade and the Evolution of Exchange Relations at the Beginning of the Mississippian Periods. In The Mississippian Emergence. B.D. Smith, ed. Pp. 251- 274.Washington: Smithsonian Institution Press.
Buchanan, B. 2006 Analysis of Folsom Projectile Point Resharpening Using Quantitative Comparisons of Form and Allometry. Journal of Archaeological Science 33:185-199.
Byers, D.A., and Ugan, A. 2005 Should We Expect Large Game Specialization in the Late Pleistocene? An Optimal Foraging Perspective On Early Paleoindian Prey Choice. Journal of Archaeological Science 32: 1624-1640.
Cannon, M.D., and Meltzer, D.J. 2003 Early Paleoindian Foraging: Examining the Faunal Evidence for Large Mammal Specialization and Regional Variability in Prey Choice. Quaternary Science Reviews 23:1955-1987
Carlson, E.H. 1991 Minerals of Ohio. Columbus: Ohio Department of Natural Resources.
Carr, D. H. 2004 The Organization of Late Paleoindian Litchi Procurement Strategies in Western Wisconsin, Midcontinental Journal of Archaeology 30(1): 3-36.
Carskadden, J. N.d. Some Observations on Early Paleoindian Chert Acquisition and Site Distribution in Muskingum County, Ohio. Unpublished document.
Converse, R.N. 1973 Ohio Flint Types. Columbus: The Archaeological Society of Ohio.
Deller, D.B., and Ellis, C.J. 1988 Early Paleoindian Complexes in Southwestern Ontario. In Late Pleistocene and Early Holocene Paleoecology and Archaeology of Eastern Great Lakes Region R. Laub, N. Miller, and D. Steadman, eds. Pp. 251-293. Buffalo: Bulletin of the Buffalo Society of Natural Sciences.
81
Dincauze, D.F. 1993 Fluted Points in the Eastern Forests In From Kostenki to Clovis: Upper Paleolithic PaleoIndian Adaptations. O. Soffer and N.D. Praslov, eds. Pp. 279- 292. , Plenum Press, New York.
Ellis, C.J., Goodyear, A.C., Morse, D.F., and Tankersley, K.B. 1998 Archaeology of the Pleistocene-Holocene Transition in Eastern North America. Quaternary International 49/50: 151-166.
Fogelman, G.L., and Lantz, S.W. 2005 The Pennsylvania Fluted Point Survey. Fogelman Publishing Company. Frison, G.C. 1998 Paleoindian Large Mammal Hunters on the Plains of North America. Proceedings of the National Academy of Science 95:14576-14583.
Gardner, W.M. 1977 Flint Run Paleo-Indian Complex and its Implications for Eastern North America Prehistory, Annals of the New York Academy of Sciences 288:251-263.
Gibbon, G.E., and Ames, K.M. 1998 Archaeology of Prehistoric Native America: an Encyclopedia. Garland Publishers, New York, pp. 306-310.
Goebel, T., Waters, M.R., O’Rourke, D.H. 2008 The Late Pleistocene Dispersal of Modern Humans in the Americas, Science 319(5869): 1497-1502.
Goodyear, A.C. 1978 A Hypothesis for the use of Cryptocrystalline Raw Materials Among Paleo-indian Groups of North America. Research Manuscript Series No. 156. Institute of Archaeology and Anthropology, University of South Carolina, Columbia.
Gramly, R.M. 1988 Paleo-Indian Sites South of Lake Ontario, Western and Central New York State. In Late Pleistocene and Early Holocene Paleoecology and Archaeology of the Eastern Great Lakes Region. R.S. Laub, N.G. Miller, and D.W. Steadman. Pp. 265-280. Bulletin No. 33. Buffalo Society of Natural Sciences, Buffalo, New York.
Grayson, D.K., and Meltzer, D.J. 2003 Requiem for North American Overkill. Journal of Archaeolgocial Science 30: 585- 593.
82
Haynes, C.V. 1964 Fluted Projectile Points: Their Age and Dispersion. Science 145:1408-1413.
1982 Were Clovis Projenitors in Beringia? In Paleoecology of Beringia. D. Hopkins, J. Mathews, C. Schwger., and S. Young, eds. Pp. 383-398. New York: Academic Press.
Hill, M.G. 1994 Paleoindian Projectile Points from the Vicinity of Silver Mound (47JA21), Jackson County, Wisconsin. Midcontinental Journal of Archaeology 19(2): 223-259.
Holman, J.A. 2001 In Quest of Great Lakes Ice Age Vertebrates. East Lansing: Michigan State University Press.
Hodder I. and Orton C. 1976 New Studies in Archaeology 1: Spatial Analysis in Archaeology. New York. Cambridge University Press.
Jones, P.R. 1980 Experimental Butchery with Modern Stone Tools and its Relevance for Paleolithic Archaeology. World Archaeology 12(2):153-165.
Justice, N.D. 1995 Stone Age Spear and Arrow Points of the Midcontinental Eastern United States. Bloomington: Indiana University Press.
Kagelmacher, M.L. 1999 Ohio Cherts of Archaeological Interest: A Macroscopic and Petrographic Examination and Comparison. MA thesis. Department of Anthropology. Kent State Univeristy.
Keeley, L.H. 1982 Hafting and Retooling: Effects on the Archaeological Record. American Antiquity 47:798-809.
Kelly, R.L. 1995 The Foraging Spectrum: Diversity in Hunter-Gatherer Lifeways. Washington: Smithsonian Institution Press.
Kelly, R.L. and Todd, L.C. 1988 Coming into the Country: Early PaleoIndian Hunting and Mobility, American Antiquity 53:231-244.
83
Luedtke, B.E. 1992 An Archaeologist’s Guide to Chert and Flint. Los Angeles: UCLA Institute of Archaeology.
McDonald, H.G. 1994 The Late Pleistocene Vertebrate Fauna in Ohio: Coinhabitants with Ohio’s Paleoindians. In The First Discovery of America: Archaeological Evidence of the Early Inhabitants of the Ohio Area. W.S. Dancey, ed. Pp. 23-42 Columbus: The Archaeological Council.
Meltzer, D.J. 1983 On Stone Procurement and Settlement Mobility in Eastern Fluted Point Groups. Baywood Publishing Co., Inc.
1988 Late Pleistocene Human Adaptations in Eastern North America, Journal of World Prehistory 2:1-52.
1993 Search for the First Americas. New York. Smithsonian Institute Press. Meltzer, D.J. and Smith, B.D. 1986 Paleoindian and Early Archaic Subsistence Strategies in Eastern North America. In Foraging, Collecting and Harvesting: Archaic Period Subsistence and Settlement in the Eastern Woodlands. S.W. Neusius ed. Pp. 3-31. Carbondale: Southern Illinois University.
Mills, W.C. 1921 Flint Ridge. Ohio Archaeological and Historical Publications.
Moore, C.R. 2008 Raw Material Utilization in Carroll County, Indiana: A Small-Scale Analysis of Diachronic Patterns in the Usage of Attica, Kenneth, and Wyandotte Cherts, Midcontinental Journal of Archaeology Vol. 33:1: 73-106.
Morris, L.L, Seeman, M.F., Summers G.L, Dowd E, Szafranski, C., Barans, P.J., and Nilsson, N.E. 2000 Fluted Points and Bifaces from the Nobles Pond Site: 33ST357, The Ohio Archaeologist 49 (2): 4-12.
Morse, D.F., Anderson, D.G., and Goodyear, A.C. 1996 The Pleistocene-Holocene Transition in the Eastern United States. In Humans at the End of the Ice Age: The Archaeology of the Pleistocene-Holocene Transition. L.G. Straus, B.V. Eriksen, J.M. Erlandson, D.R. Yesner, eds. Pp. 319-334. New York and London: Plenum Press.
84
Morrow, J.E. 1997 End Scraper Morphology and Use-Life: An Approach for Studying Paleoindian Lithic Technology and Mobility. Lithic Technology 22:70-85.
Morrow, J.E., and Morrow, T.A. 2002 Exploring the Clovis-Gainey-Folsom Continuum: Technological and Morphological Variation in Midwestern Fluted Points, in: Folsom Technology and Lifeways (J.E. Clark and M.B. Collins eds.), Lithic Technology, Special Publication No. 4 pp. 141-157.
Prufer, O.H., and Baby, R.S. 1963 Paleo-Indians of Ohio. Columbus: Ohio Historical Society.
Renfrew, C., and Bahn, P. 2000 Archaeology: Theories and Methods. London: Thames and Hudson.
Renfrew, C., Cann, J.R., and Dixon, J.E. 1965 Obsidian in the Aegean, The Annual of the British School at Athens 60:225-247.
Seeman, M.F. 1994 Intercluster Lithic Patterning at Nobles Pond: A Case for “Disembedded” Procurement Among Early Paleoindian. American Antiquity 59(2): 273-288. Seeman M.F. and Prufer O.H. 1982 An Updated Distribution of Ohio Fluted Points. Midcontinental Journal of Archaeology 7:155-170.
Seeman, M.F, Nilsson, N.E., Summers, G.L., Morris, L.L., Barans. P.J., Dowd, E., and Newman, M.E. 2008 Evaluating Protein Residues on Gainey Phase Paleoindian Stone Tools. Journal of Archaeological Science 35:2742-2750.
Shane, L.C.K. 1994 Intensity and Rate of Vegetation and Climatic Change in the Ohio Region Between 14,000 and 9,000 14C YBP. In The First Discovery of America: Archaeological Evidence of the Early Inhabitants of the Ohio Area. W.S. Dancey, ed. Pp. 23-42 Columbus: The Archaeological Council.
Smith, C.S. 2003 Hunter-gatherer Mobility, Storage, and Houses in a Marginal Environment: An Example from the Mid-Holocene of Wyoming. Journal of Anthropological Archaeology 22:162-189.
85
Smith, E.E. 1990 Paleoindian Economy and Settlement Patterns in the Wyandotte Chert Source Area, Unglaciated South- Central Indiana. In Research in Economic Anthropology: Early Paleoindian Economies of Eastern North America, K.B. Tankersley, B.L. Isaac, eds. Pp. 217-258. London: JAI Press Inc.
1991 Paleo-Indian Settlement and Lithic Procurement Patterns in the Karstic Region of South-central Indiana. Ph.D. dissertation. Department of Anthropology. Indiana University.
Stanford, D. 1989 Clovis Origins and Adaptations: An introductory perspective. In Clovis Origins and Adaptations. R. Bonnichsen and K. Turnmire, eds. Pp. 1-13. Corvallis, Center for the Study of the First Americans.
Storck. P.L. and Spiess, A.E. 1994 The Significance of New Faunal Identifications Attributed to an Early Paleoindian (Gainey Complex) Occupation at the Udora Site, Ontario, Canada, American Antiquity 59(1): 121-142.
Stothers, D.M. 1996 Resource Procurement and Band Territories : A Model for Lower Great Lakes Paleoindian and Early Archaic Settlement Systems. Archaeology of Eastern North America 24:173-216.
Stout W. and Schoenlaub, R.A. 1945 The Occurrence of Flint in Ohio. Columbus. Ohio Department of Natural Resources.
Tankersley, K.B. 1990 Late Pleistocene Lithic Exploitation and Human Settlement in the Midwestern United States. Ph.D. dissertation. Department of Anthropology. Indiana University.
Todd, L.C., Hofman, J.L., and Schultz, C.B 1990 Seasonality of the Scottsbluff and Liscomb Bison Bonebeds: Implications for Modeling Paleoindian Subsistence. American Antiquity 55(4): 813-827.
Waguespack, N.M, and Surovell T.A. 2003 Clovis Hunting Strategies, or How to Make Out on Plentiful Resources. American Antiquity 68(2): 333-352.
Waters, M.R., and Stafford, T.W., Jr. 2007 Redefining the Age of Clovis: Implications for the Peopling of the Americas. Science 315 (5815): 1122-1126.
86
Webb, S.D., Milanich, J.T., Alexon, R., and Dunbar, J.S. 1984 A Bison Antiquus Kill Site, Wacissa River, Jefferson County, Florida. American Antiquity 49(2): 384-392. Appendix A
Raw Materials Descriptions
Boggs Outcrop located in Perry County, Ohio. The flint is indented with trace fossils of shells and the actual flint contains shell fragments. Iron ore and carbon heavily affect the appearance of the flint (Stout and Shoenlaub 1945). Boyle This flint is typically light grey in color, and may have a molted cream, orange, or brown tinge. It is not unusual to see impurities in the form of quartz inclusions. The flint looks like a highly cemented welded tuff. Brush Creek From the Brush Creek formation, this flint outcrop is Upper Pennsylvanian in age and comes in the form of nodules or lenses within a limestone matrix. It contains microfossils and is fine to medium grained and has a medium to shiny luster (Luedtke 1992). Fresh samples are mustard in color. Patinated samples are dirty yellow or white. Outcrops are located in Adams County, Ohio. Burlington Flint As part of the Burlington formation, this flint is middle Mississippian in age and is deposited in lenses or nodules. Usually white or pale in color, the samples can be extremely fossiliferous. The outcrops are in Illinois or Missouri (Luedtke 1992). Delaware flint Delaware flint is part of the Columbus limestone formation and varies in quality by location from south central Ohio to Lake Erie. Eroded boulders are quarried and flint layers can be up to four inches thick. Samples are earth colored with grey or dark brown streaks. Patination creates a dusty appearance on the earthy flint (Converse 1973). Hixton Quartzite This raw material is also known as Hixton Silicified Sandstone, and it outcrops in Jackson County, WI, at a location known by many as Silver Mound. The quartz grains are sub-rounded to rounded with flecks of opal and chalcedony (Hill 1994). Hopkinsville Hopkinsville flint outcrops in nodule and tabular forms. Nodules can be spherical and up to 20cm in diameter. Hopkinsville beds may be up to several meters in diameter (Tankersley 1989).
87
88
Harper’s Ferry Quartzite Also known as Roanoke Rapids flint, this flint outcrops at the Virginia/North Carolina border. Indiana Greenstone Also known as Attica flint, this chert outcrops along the Wabash River in nodules that rarely get larger than three or four inches in diameter. It ranges from a dull green with streaks of pale white or grey (Converse 1973). Indiana Hornstone Also known as Wyandotte flint. Part of the Ste. Genevieve formation and deposited during the Middle Mississippian (Leudtke 1992). It outcrops in Harrison and Crawford counties of Indiana and Meade County, Kentucky. Samples are opaque grey, have a shiny luster and are fine grained. Patination leads to a dusty or light grey surface with hints of yellow or tan. No cavities or flaws are present in this nodular flint (Converse). Kanawah Black Flint This is also part of the Upper Mercer limestone formation, coal black, lusterless flint outcrops southeast of Charleston, West Virginia (Leudtke 1992). There is little evidence of the flint being used by any Paleoindians, but is used frequently in West Virginia during the Archaic period. Onondoga Flint From the Onondoga formation and outcropping in upstate New York (Leudtke 1992). The flint deposits date to the lower or middle Devonian and are in the form of nodules. The most frequently used samples are grey to bluish grey. Some samples contain dark grey or tan inclusions (Converse 1973). Pipe Creek This flint outcrops in mid-central Ohio (Prufer et al.. 2001) near the Sandusky Bay area (Ellis et al.. 1991). This lusterless flint varies in color and quality. Samples can be cream, dark grey, maroon, tan, or dull green and tend to be porous and rough in texture. Patination generally results in a yellow cast on the sample. Plum Run Depositional environment is unknown and age is questionable for these outcrops. Aboriginal quarries would have been east of Alliance, Ohio, in Stark County. Flint samples are dull brown with pink, mauve or yellow tinges. When freshly chipped, the flint changes to a glossy grey color (Luedtke 1992). Tennmile Creek Flint Dating to the Middle Devonian, this flint is part of the Tenmile Creek formation. The flint was deposited in the form of nodules or lenses within a limestone matrix in northwestern Ohio. The flint is fairly homogenous or molted pale blue to yellow and contains macrofossils and light colored specks (Leudtke 1992). Zaleski Part of the Zaleski member, dates to the Pennsylvanian, and outcrops in Vinton County, Ohio. This flint is extremely glossy and jet black in color (Converse 1973). With patination, some samples will turn dusty brown or green in color. Deposits are up to 30cm thick (Leudtke 1992).
89